TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

UTLEY:
A. Alkylation-Sulfuric Acid There is not a lot of data on this commercially and not a
lot in the recent literature. There was some work done in
the late 1940’s and early 1960’s that I found. It was reported
Feed that alkylation of branched pentenes yields superior alkyl-
ates to those produced from linear pentenes because hydro-
gen transfer reactions occur to a greater extent with
QUESTION 1:
branched pentenes to greater amounts of trimethylpentane
IS anyone routinely alkylating significant quan- and isopentane, while large amounts of nonanes and tri-
tities (greater than ten percent) of C5 olefins? If so, methylhexane result from alkylation of linear pentenes.
what are the results (yields, octanes, acid con- The octane numbers reported were about 1.5 to 2 num-
sumption, operating problems, etc.)? bers higher for isopentene versus normal pentene. More
specifically, 93 octane for P-methyl 2-butene, and 91 for
WILSON: 2-pentene. Although if you sum up the individual com-
This has been done experimentally in the past with feeds ponent octanes, in the 1961 article, there is essentially no
greater than 10 percent C5’s in the olefin feed. STRATCO is difference in octane between isopentene and normal pen-
currently designing several units that will in fact process tene. Needless to say, I am sure, we will get a better
up to 30 vol % C5 olefins based on the total olefins in the definition of what the octanes are for the C5 olefin com-
feed. They have done quite a bit of laboratory work on this. ponents over the next year or so.
Their findings are that the yields for C5 olefin alkylation The acid consumption was not much different between
are about 1.55 to 1.6 barrels of true alkylate per barrel of the two, with the average being around 1 pound per gallon
olefin. The product octanes are 91 to 93 research, and 89 to of alkylate produced. Additional information in the litera-
91 motor. The acid consumption based on contaminant-free ture can be found in the Journal of Organic Chemistry,
feeds is about 0.6 to 0.9 pounds of acid per gallon of alkylate Volume 6, page 647, 1941; and in PetroChem Engineer,
produced. This is based on an assumption of 98.5 wt % as January, 1962; and PetroChem Engineer, December, 1961.
the fresh acid and 90 wt % spent acid.
They do not really anticipate any operating problems. EILEEN PROHASKA (STRATCO, INC.):
There is some indication that the acid-hydrocarbon emul- We have been doing some testing at our pilot laboratory.
sion in the settler is a little harder to break, and this may Our results show that there are not any significant differ-
require a change in the level of the settler or possibly a ences in octanes between a typical mixed C5 stream and a
larger settler vessel. TAME raffinate type stream. We also looked at the acid
consumption differences between the two. There is some
SCHAUB: reduction with the TAME raffinate. We believe it will be
We do not know of anyone currently doing this, but with enough to offset the effect of any reasonable amount of
the push for reduced gasoline olelinicity, many people are oxygenate carryover. So from an alkylation unit stand-
considering it. One of the concerns about feed con- point, we do not expect there to be any significant differ-,
taminants will be a significantjump in diene content which ences between a mixed C5 stream and a TAME raffinate.
may require selective hydrogenation of the alkylation feed-
stock. Another concern is that Merox extraction efficiency JOHN LYNCH (Petrogistics):
begins to fade on the heavier mercaptans. Between 1943 and 1953, there was about 100,000 barrels
of alkylation capacity in the United States. At that point in
J. G. DEARWATER (Profimatics, Inc.): time, the war program committee basically mandated that
Results from kinetic model runs show that at constant all alkylation units in the country run C5’s to maximize
RVP, for one incremental barrel of butylenes, we get 155 aviation gasoline for the war effort, both in World War II
octane barrels of alkylate. For one incremental barrel of C5 and the Korean War. After the Korean War, all of the C5’s
olefins, we get 1.31 octane barrels of alkylate. Incremental went back to direct gasoline blending. I think what we are
acid consumption is the same. going to find here is that like the 1940’s, the 1990’s are
going to be a period where essentially all of the C5’s will end
up in alkylation with the isoamylene heading over for
QUESTION 2: TAME production.
What are the deviations in octane numbers and
acid consumption between isoamylene feed and WILSON:
normal amylene feed? I would just like to expand on that a little bit. I agree. I

107
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

think that is where we are headed, at least for all of the C5’s
from the catalytic cracker. If you look at the olefms in the
FCC naphtha, a tremendous percentage of them are in that
C5 range. By alkylating them or sending them to TAME
and then alkylating the raffinate, you can substantially
reduce the amount of olefins in the total gasoline pool.

QUESTION 3:
What feed contaminants should be expected with
C5 alkylation? What are the effects of these con-
taminants (including C6 components)?

WILSON:
Primarily the contaminants you get when you send C5’s
to the alkylation unit are increased diolefins and increased
sulfur. You can get some C6 compounds also. All of these
have a tendency to increase the acid consumption It appears that benzene may be the culprit. This may be
somewhat. yet another reason to focus attention on benzene when you
study FCC product recovery unit operation. We have de-
UTLEY: veloped a preliminary design criteria. We recommend that
Feed contaminants expected with C6 olefins are pri- you maintain less than 10 ppm of benzene in your alkyla-
marily the same that are experienced with C4 oleiins, tion unit feed. This means that (with an existing FCC
although the higher carbon number provides a greater debutanizer) you may only be able to process 30% of your
number of combinations of sulfur compounds and dienes. available C5’s in your alkylation unit. Above 30% you may
The sulfur compounds are probably mercaptans and higher have to give consideration to column modifications.
sulfides. The percent sulfur in the C5 will likely be higher
than the C4 feed because of the heavier compounds. The C5 ROBERT DAVIS (R. E. Davis Chemical Corporation):
dienes will consist of a wider variety of species with the I agree totally with the panel, and I would like to raise an
absolute amounts depending on the FCC feed, the reactor issue that is going to rear its ugly head very shortly. As you
operating conditions and fractionation. go into the C5 alkylation, the polymer make that stays in
In addition to these contaminants, C6 + material may be the acid is going to increase. The acid suppliers are up
present depending on how well your debutanizer or de- against a wall on the regeneration of the acid in their
pentanizer operates. The C6 olefins will react with iso- plants. I think that their more serious problem is going to
paraffins present to form a light ASO. The C6 isoparaffins be in the supply and the conversion of the acid back to fresh
will also react much the same way as other isoparaflins acid. The refineries ought to be aware of this, and take all
react, but in this case the products are much heavier, steps necessary to investigate new catalyst systems and
forming again a light ASO. The dienes in the feed and the new methods of handling the amylenes.
C5 sulfur compounds will react much the same way as the
lighter components of these same species. JOHN LYNCH (Petrogistics Inc.):
I would just like to second that motion that was just
STANLEY GUSSOW (Engelhard Corporation): made. There have been different outlooks from the stand-
Engelhard’s selective diolefin hydrogenation technology point of alkylation growth being anywhere from 5 to 30%
also includes C5’s from FCC units as well as C5’s from steam per year. When we look at the alkylation growth, we see
cracker units. In this process, isoprene, pentadienes, cy- that first the incremental C5 olefins will be going to the
clopentadienes are converted to the respective mono- alkylation unit and then propylene will be the incremental
olefins with little saturate production. We can lower the feed. If you are looking at 10% annual growth in alkylation
diolefin content typically to less than 100 ppm. capacity, you are looking at approximately 20% growth in
acid consumption. There is basically twice the acid con-
NORMAN KOLB (The M. W. Kellogg Company): sumption with either propylene or amylenes. So this is an
Kellogg has studied how FCC product recovery section extremely pressing problem and one that must be looked at
operation can be modified to provide C5 feeds to alkylation and resolved shortly.
units. I would like to cite plant experience that supports our
recommendations. As refiners (either intentionally or un-
intentionally) dig into their C5’s through FCC debutanizer QUESTION 4:
adjustments, they have noted unusual settling problems in What steps are used to remove the contaminants
their acid settlers and caustic wash settlers. Some refiners described in question 3?
have identified sulfonated alkylated benzenes in their
caustic wash. Further, when debutanizer operation was UTLEY:
adjusted back to a normal operating mode, sulfonated ben- I do not believe that there is much being done com-
zenes disappeared in their caustic wash, and settling prob- mercially today in the way oftreating the small amounts of
lems disappeared. C5 oletins going to alkylation units. I know there is a lot of

108
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

work going on trying to figure out how to handle this PAINCHAUD:

stream when the amouljts going to alkylation units become We installed an Engelhard butadiene hydrotreater in
greater and greater. We are hopeful that our Merox treater 1980 to treat our sulfuric acid alkylation feed. A paper
with some modifications can handle the increased sulfur entitled, “Feed Stock Hydrogenation Improves Alkylation
content of the C5 stream and the increased variety of sulfur Unit Performance,” was presented at the 1986 NPRA An-
compounds. nual Meeting detailing the results. Very briefly, the results
The dienes will probably be present in high enough were as follows: Butadiene conversion was essentially
concentrations to require selective hydrogenation to mini- lOO%, dropping the stream’s butadiene content from ap-
mize acid consumption. This is not being done com- proximately 0.9 vol % to 100 ppm. This reduced our acid
mercially today, but again there is a lot of work going on to consumption from 1.25 to 0.54 pounds of acid per gallon of
determine how this can be accomplished. The increased alkylate. Alkylate end point dropped from 465 to 415 °F
sulfur levels of the C5 olefms may present some problems Butene isomerization averaged about 45%. This resulted in
for selective hydrogenation. The C6+ material must be an octane increase from 93.7 to 94.6 RON at variable
eliminated through better fractionation, and some under- production rates.
cutting of the C5 olefin may need to be done to leave the
heavier part of the C5 stream in the FCC gasoline to reduce STEGELMAN:
problems with heavier dienes, sulfur compounds, and the Removal of butadiene from alkylation unit feed does
C6 material. result in a lower end point alkylate. Since the butadiene
and alkylation unit feed makes heavy alkylate as well as
WILSON: acid soluble oils, a selective hydrogenation unit usually has
I concur with what Mr. Utley said. The actual dynamics a quick payout due to improved alkylate quality and lower
of the fractionation system on the C5/C6 split in a typical catalyst and chemicals cost. I do not have any data to
FCC gasoline is going to determine that the optimum quantify the lower end point realized. One of our refineries
separation is going to be somewhat less than 100%. I think realized at least a 50% reduction in acid losses from the
that some work has been done that shows that 95% is about rerun section after hydro-isomerization, and the alkylate
where you are going to get an optimum use of the energy octane boost was 1.7 (R+ M)/2. The alkylate octane boost
and reflux ratio. realized with selective hydrogenation, “Hydrisom”, is due
With regard to the sulfur compounds, the extractive type to the butadiene removal and selective isomerization of
treaters will be capable of removing most. I think a range of butene-1 to butene-2.
about 25 to 40 ppm of sulfu,r should be expected in the
mixed C4/ C5 stream. This is somewhat high, and is going to DAVID GRAVES (STRATCO, INC.):
cause some increased acid consumption. However, it seems Earlier this year, STRATCO ran bench scale tests to
to be an acceptable level for an alkylation unit. determine the effect of butadiene on acid consumption,
octane and end point. We tested butadiene levels equiv-
RONALD E. MARRELLI (Phillips 66 Company): alent to 0.5 to 5 wt % of a typical C4 olefin feed. We found
With the increase of the C5 olefins in the feed, what is that butadiene increased acid consumption and end points.
going to happen to the normal pentane that is going to go We did not see any measurable effect on octane. Here are
with it? Is that pentane going to knock down the octane that some of the results on end point: At a level of 0.5 wt %, we
you are trying to increase? How are you going to separate saw an increase of 14 ºF over the base case with no bu-
the pentanes from the C5 olefins? Are you going to have to tadiene. At 1 wt %, we saw an increase of 28 ºF, and at 5
build other towers to separate your normal butane and wt %, we saw an increase of 108 °F
your normal pentanes?
ROBERT DAVIS (R.E. Davis Chemical Corporation):
UTLEY: The decrease in octane comes about by decomposition of
I imagine that normal will go with the alkylate. I do not products that occur in the deisobutanizer due to sulfur
know why it would not. bonding directly to the carbon on butadiene generating
CO2, SO2, free carbon, and free water. You will find that the
WILSON: water make drops on the DIB overhead accumulator when
Certainly if you put it in there, it is going to have those you remove the butadiene from your feed. The polymer-
effects. It is not that bad of an octane material. You also ization reactions that occur in the reactor due to these types
have to address the vapor pressure issues for the pentane of reactions do increase the end point with corresponding
species that are not reacted because they will tend to raise change in the isomer distribution of the alkylate. This
the vapor pressure of the alkylate. There is really not that results in a lowering of octane when you have butadiene.
much normal pentane in FCC gasoline. It is mostly olefins There are no good numbers because it depends on the
and some isopentane. acidity, the hydrogen transfer reactions, and polymer-
ization reactions. So any number will only fit your par-
ticular unit for your particular conditions.
QUESTION 5:
Removing the butadiene from feeds reduces the JOHN LYNCH (Petrogistics Inc.):
amount of acid consumption in the unit, lowers the The discussion has been focused primarily on the C4
end .point and also boosts octane. What are the stream. We see the future going with TAME. When you
increased octane and decreased endpoint values? start running the C5 oletins stream, there may be problems

109
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

with the cyclopentadiene dimerizing on the etherfication quoted before and as quoted in the literature, ranges be-
catalyst. We have not confirmed this yet, but anybody who tween 10 to 30 vol %. For HF, it is reported that production
has worked with cyclopentadiene in the past recognizes it ranges from 15 to 40 vol %. These numbers are for pure
dimerizes readily to dicyclopentadiene. With the cyclopen- compounds, and the presence of feed isopentane will effect
tadiene concentrations in the C5 stream, this can be a real the production of synthetic isopentane because of equi-
problem from the standpoint of catalyst deactivation for librium effects. Additional data can be found in Petro/
etherfication units. If this turns out to be true, then every Chem Engineer, December, 1961, and Petroleum Refiner,
unit involving isoamylene etherfication will need a satu- Volume 37, Number 9, page 316, 1958.
ration unit.
WILSON:
JEAN-PAUL BOITIAUX (IFP Enterprises Inc.): I will communicate the results of some experiments that
In IFP we have several units for selective hydrogenation STRATCO ran on this. They were using a mixed butylene
before alkylation, either sulfuric or hydrofluoric. Typically feed, and then blended it with varying quantities of a
with 0.8 to 1.0% butadiene in feedstock, we achieve selec- mixed C5 type stream. On inspection of the alkylate prod-
tive hydrogenation to less than 100 ppm in both cases, with uct, they showed an increasing level of iC5 in the whole
isomerization of l-butene to 2-butenes 50% for the first case alkylate with increasing C5 olefin charge rate. However,
and up to 80% for the second case. upon correcting the whole product for the iC5 present in the
The alkylate octane number (RON) changes from 93.5 feed, the amount of C5 produced was held relatively con-
without hydrogenation to 94.5 or 96-97 depending on the stant. What this indicates is that the hydrogen transfer
downstream alkylation. The end point is decreased around reaction is at equilibrium with the iC5, that comes in, in the
70 to 80 ºF conditions in the reactor. The actual production is fairly
low.
BRIAN JOHNSON (UOP):
There were several comments about the enhancement of ROBERT DAVIS (R. E. Davis Chemical Corporation):
the octane using selective hydrogenation. We agree that I believe the purpose of this is to check the isopentane,
there is enhancement of the product octane due to isomer- but there is a second reaction that ought to be looked at. The
ization of butene-1 to butene-2 in the feedstock for the HF isopentane that is produced combines with the butylene to
alkylation unit. For a sulfuric acid alkylation unit, the produce 2,2,5-trimethylhexane, a C9 which has a lousy
isomerization can take place in the reactor section of the octane. This is a dynamic reaction rather than a static
unit itself. So, there is some difference in the application of reaction. If you are short of isobutane, you are going to find
SHP in these systems. that you produce more isopentane due to hydrogen transfer
directly from the acid. When you are high on the isobutane,
STANLEY GUSSOW (Engelhard Corporation): that reaction does not occur. You produce more isopentane,
We would agree that there is a continuous debate on the and it will alkylate to the 2,2,5-trimethylhexane, or Ce.
magnitude of the octane increase that you achieve from Take a look at this kind of data and see what you have got.
converting butene-1 to butene-2 in sulfuric acid plants. We
have some data that supports very little octane gain. We
have other data that supports a gain of up to 0.8 (R + M)/2. QUESTION 7:
What operating conditions are required for am-
NORMAN KOLB (The M. W. Kellogg Company): ylene alkylation? (i.e., space velocity, isobutane/
In my opinion there should be no debate on this issue. olefin ratio, isobutane in reacted mix, etc.)
There is no difference between butene-1 and butene-2 feeds
in H2SO4, alkylation units. RANDY SCOTT (STRATCO, INC):
STRATCO designs new alkylation units with an
isobutanelolefin ratio of approximately 9/l, a reactor efIlu-
Process ent content of 68 LV % isobutane, and an olefin liquid
hourly space velocity of approximately 0.3 hr-1. Currently,
all of our new alkylation units are designed to operate
QUESTION 6: using these same guidelines as long as the amylene content
What is the extent of isopentane production when of the alkylation unit feedstock, whether C3/C4/C5 or C4/C5
alkylating C5 olefins? Is there a difference in this combinations, is within the normal limits produced by the
production between HF and H2S04? FCCU.

UTLEY:
Again there is not a lot of commercial data available, but QUESTION 8:
the following data is reported for alkylating C5 olefins. For Due to problems associated with caustic avail-
2-methyl, 2-butene feed, isopentane production was 25 ability and spent caustic disposal, what experience
vol %. For 2-methyl, 1-butene feed, isopentane production do others have using bauxite beds instead of caustic
was 28 vol %. For 2-pentene feed, isopentane production water washing for reactor effluent treating?
was 19 vol %. There is some difference in isopentane pro-
duction when using HF versus sulfuric for alkylation. De- PAINCHAUD:
pending on the C5 oletin used, production of isopentane, as We use bauxite towers in our Kellogg unit. The system

110
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

consists of two towers, one operating and one on standby. removal of both neutral and acid esters, thus avoiding the
Each tower is operated for a one month period during which need for water wash temperature control. Merichem now
the pressure drop rises from approximately 15 to 50 psi. To offers EsterexSh’ technology which uses the FIBER-FILM”’
back wash a tower, water is passed upwards through the contactor in either acid wash service to remove both the
tower and into a pit where it is neutralized with caustic. acid and neutral esters or caustic wash service to remove
The tower is washed until the effluent pH rises to 6.5 which the acid esters and acid carryover effectively, thus mini-
normally takes two to three days. At this point it is drained mizing corrosion of downstream equipment.
and dried with hot natural gas. The bauxite is replaced AQUAFINING”’ water extraction of caustic impurities,
every 9 to 12 months. or alkaline water extraction of the acid impurities, can be
Since we typically operate at 200% of design, we still used for the downstream water wash. The advantages are
experience severe fouling and corrosion in downstream as follows: minimal sized equipment, safety, 100% service
equipment. We have reduced this by adopting a policy of factor, minimal guaranteed caustic phase carryover.
switching towers every month as opposed to operating until Existing conventional treating systems can be retro-
pressure drop limits throughput. fitted with a FIBER FILM”’ contactor, thus saving time
and money.
RATERMAN:
We do not use bauxite for removing trace acid or esters in PHIL BIRK (STRATCO, INC.):
any of our alkylation units. We do however know of a The temperature of the alkaline water wash is important
refinery in the U.S. where they have effectively used baux- to assure that the neutral esters in the net reactor effluent
ite in a four bed unit to treat the acid settler effluent. To stream are thermally decomposed. STRATCO recommends
regenerate the bauxite bed, they wash with hotcondensate, operating the alkaline water wash at 120 °F which is ade-
drain, then purge with hot isobutane before returning the quate for the thermal decomposition of these neutral es-
bed to service. They claim the life of the bauxite is ap- ters. Also, the circulating water should be run at 9 to 11 pH
proximately four years. When a bed is dumped, the bauxite to hydrolyze these neutral esters as well as neutralize any
is disposed of in a non-hazardous landfill. acid carryover.
With proper operation of the alkaline water wash, cor-
WILSON: rosion downstream is minimal, and the pH of the DIB
There is one customer with a STRATCO sulfuric acid overhead water runs between 6.5 and 7. Without proper
plant that also uses bauxite treating. They have good re- heating of the circulating alkaline water, some neutral
sults. They make a high quality alkylate with no down- esters will not thermally decompose until being exposed to
stream corrosion or fouling in the fractionation section. heat in the DIB. Thermal decomposition at that point,
Another less documented, but positive feature, is that the without the presence of caustic, will yield corrosive
removal of water from a reactor effluent provides for a very components.
dry isobutane recycle stream which contributes to a very
low acid consumption at this refinery. ROBERT DAVIS (R. E. Davis Chemical Corporation):
On the temperature of the water wash, the most impor-
JIM MALOTT (STRATCO, INC.): tant thing relative to effluent treating is to maintain the
STRATCO is of the opinion that a fresh acid wash fol- base, whatever you use, at sufficient strength to fully neu-
lowed by bauxite treaters would be the optimum effluent tralize what is coming out of the reactor. At that point, it
treating system. With the acid wash, you should sub- does not matter what your water temperature is. If you
stantially reduce the regeneration cycle of the bauxite cannot fully neutralize, you have to increase the tempera-
beds. We are currently working with refiners and in our ture of the water wash. The key is really in the caustic
laboratory to quantify the benefits of such a treating wash.
system.

QUESTION 10:
QUESTION 9: What methods are used to dry sulfuric acid alkyla-
How important is alkaline water wash temperature tion units prior to start-up? How do you judge when
control in minimizing downstream corrosion? the unit is sufficiently dry?

COOPER: CARON:
In the alkaline water wash, one thing you have to remind Usually all pressure testing is done with nitrogen which
yourself of is the need to hydrolyze alkyl sulfates so they eliminates the potential for large water buildup in the unit.
can be removed from the water wash. We found that the Once hydrocarbon circulation begins, we constantly drain
hydrolysis is very time dependent as well as temperature water from all low points. The refrigeration compressor
dependent. enables most process water to drop out at 40 °F. When we
stop picking up water at low points, we are ready to go.
HERBERT W. WIZIG (Merichem Company):
Merichem’s experience is that water wash temperature is MORGAN:
important when the caustic wash is followed by a water In the past our problems with unit dry out centered on
wash for acid ester removal, since the caustic wash does not moisture removal in the reactors. We now use cloth rags to
remove neutral esters. Acid wash is more effective for physically wipe dry the reactor vessels before closure. We

111
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

then nitrogen purge the unit and drain all low points until the source, and the disposal of the source when the time
no water is found. This usually takes about 12 hours. We comes.
introduce isobutane and follow the normal cool down and
start-up procedure. We have not had any moisture prob- GLENN C. LIOLIOS (STRATCO, INC.):
lems by following these steps, and do not measure moisture On sulfuric acid alkylation units, standard gage glasses
in the unit prior to start-up. equipped with hydrocarbon flushes are adequate for moni-
toringemulsion levels, both in the reactor and in the settler
PAINCHAUD: zone.
We agree that taking care to dry the unit prior to acid
loading is important since an introduction o f w a t e r i n t o t h e HEROS DERGREGORIAN (Giant Refining Company):
acid phase is exothermic and in the extreme case could The panel mentioned that some refiners are using a
cause equipment damage and/or failure. nuclear device, a radioactive device for level indication.
The start-up procedure we use on our Kellogg unit is as Could they reveal which type is that? Is it a UOP or a Texas
follows: (1) Pressure up and purge with nitrogen while Nuclear or what?
draining all low point bleeds. (2) Load isobutane, establish
levels and start the compressor. (3) Establish isobutane UTLEY:
recycle and drain all low point bleeds until no more water is Both of the devices that we looked at are from Texas
drawn. (4) Load acid and establish acid circulation. (5) Nuclear. I do not have a specific model number.
Begin charging feed and start the mixers.
We also take care to bleed water after minor shutdowns MORGAN:
since the possibility exists for water to enter the system We have purchased Texas Nuclear Level Detectors.
through condensers and reboilers if the system is
depressured. BRIAN JOHNSON (UOP):
UOP markets a nuclear density profiler through our
Monirex division in Des Plaines, Illinois.
QUESTION 11:
What means are presently used to determine the
location of acid-emulsion-hydrocarbon interfaces in QUESTION 12:
acid mixers, settlers and storage tanks in HF and
What leads to high sulfur dioxide production in an
H2S04 alkylation units? What are the pro’s and con’s
effluent refrigerated sulfuric acid alkylation unit?
of these methods?
What would be considered a high sulfur dioxide
concentration in the refrigeration system and the
MORGAN:
depropanizer overhead stream? What steps can be
We are currently using displacers to measure HF settler
taken to control corrosion in the depropanizer
acid level and storage drum levels. However, we are cur-
overhead?
rently in the process ofswitching to nuclear level detectors.
Our main goal is to reduce employee exposure and to
reduce the amount of small HF piping. PAINCHAUD:
SO2 is a normal byproduct in the alkylation reaction. The
BRANHAM: conditions that increase SO2 formation are low isobutane
The present system being used to measure levels in our recycle rate or purity, low acid recycle rate or purity, ele-
HF settlers is a source detector radiation level indicator vated temperature, and feed contaminants. SO2 is also
used in such a way that its output is a plot of density versus formed in reboilers by the decomposition of esters and by
vessel height. We also use a gauge glass assembly for the reaction of sulfuric acid with hydrocarbons. These
gauging the fresh acid storage tank. reactions also produce water which can form a corrosive
condition, even if you have a dry cleanup system. To mini-
UTLEY: mize corrosion, you must minimize acid carryover and have
We still use external float columns and tricocks for level a good clean-up system. We prefer bauxite since no water is
measurement and interface determination on our HF unit. introduced. Another option to remove esters from your
We have looked at nuclear devices, but for now we are effluent is a fresh acid wash system. These are offered by
staying with what we have. The pro’s for tricocks are: They Petreco and Merichem. We are currently considering one of
are very simple to use, positive measurement and have these.
been around for a long time. The negatives are a small
amount of hydrocarbon and acid are emitted to the air RATERMAN:
when checked, the operator is exposed to acid, and the In addition to the sources of SO2 that Mr. Painchaud just
tricock valves are a potential source of leakage. mentioned, there is another one which involves the polym-
The pro’s for float columns are: They are well-proven, and erization of butadiene by sulfuric acid which releases SO2.
keep the acid contained. The negatives are: They plug up To keep neutral sulfate esters from reaching the frac-
occasionally, develop leaks, and need calibration regularly. tionator, we use an acid wash followed by an electrostatic
The pro’s for nuclear devices are: no exposure to acid, and precipitator. Based on our experience, we would not rec-
no emissions. The negatives, as we see them, are: the price, ommend using a neutralizing amine in the overhead to
the radioactive source, the regulations that go along with control sulfurous acid corrosion in the column because of

112
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

the acid strength. We seem to have a better use. We use contribute to very large increases in acid consumption at
filming amines up there which seem to work better for us. constant acid purity. Our specification for total oxygen
content is 35 ppm.
DAVID GRAVES (STRATCO, INC.):
We have identified two situations which can lead to STEGELMAN:
higher than normal SO2 production. High acid levels in the Sulfur or oxygenates in the olefin feed will form light acid
acid settlers which results in increased residence time soluble oil which is difficult to remove in the rerun tower
could be a factor. Also injection of high strength acid into without reducing temperature severely. AS0 is often green
lower strength acid settlers could also cause increased in color as it is removed from the rerun column. The main
production. The typical level of SO2 in the refrigeration problem is that its presence will not be readily detected in a
system is 10 to 30 ppm in the depropanizer feed. We have routine acid analysis. It will most likely be weathered off in
seen levels as high 75 to 100 ppm. the test that we use in the laboratory and not appear in the
Regarding corrosion control for those units with the weighted portion of the sample which is being analyzed for
caustic wash on the depropanizer feed, the important vari- AS0 content.
ables are: (1) proper mixer pressure drop of 7 to 10 psi, (2) a
4 to 1 hydrocarbon to caustic circulation ratio, and (3) MORGAN:
caustic strength between 3 and 12 wt %. Data from two of our units show that we see no significant
On dry depropanizers, those units with the caustic wash effects from oxygenate levels less than 50 ppm. Oxygenate
on the propane product, it is very important to assure that levels above this make ASO, as previously discussed, and
the tower is completely dried out before feed start-up. This we have the same problems purging it out of the unit. As
can be accomplished by operating the tower on total reflux long as we can maintain acid purity, however, unit oper-
with a higher than normal bottoms temperature for a ation and alkylate quality has not been affected due to the
period of two days before charging depropanizer feed. high oxygenate levels in the feed.

ROBERT DAVIS (R. E. Davis Chemical Corporation): GESICK:

This question 12 really indicates that the refinery that As already stated, the major effect or oxygenates in the
submitted this is in serious trouble on a mechanical prob- olefin feed is reduced acid purities resulting in high acid
lem. Anything that increases polymerization will increase consumption. Also, since the operating acid purity is lower,
sulfur dioxide production. The formation of the esters, there may be some small increase in octane. The light
depending upon whether they are acid soluble, hydro- compounds formed by the oxygenates can be very difficult
carbon soluble, or water acid soluble, will carry through in to remove from the acid.
the alkylate to the DIB. Decomposition products of those The Meraux Refinery recently experienced a decline in
esters in the DIB generate SO2 in the DIB overhead which operating acid purity which we attributed to oxygenates.
recirculates itself back to the reactor. The SO2 then is At the time this problem occurred, we were injecting meth-
picked up by the refrigeration stream. anol into our gas con chiller unit to remove hydrates. Our
The bauxite tower operation for effluent treating keeps normal practice of internal acid regeneration in the Iso-
the system dry, removes the esters, and results in a system stripper did little to improve the acid purity. We could only
that is much superior in many cases for these types of increase the purity by the use of our external regeneration,
problems. and we also found we had to run the external regeneration
with a lower top temperature to allow the light compounds
to drop out with the CBM and polymer.
B. Alkylation-Hydrofluoric Acid
QUESTION 14:
Feed What is the general experience with oxygenates in
the alkylation feed without an MTBE unit? What
variables in the FCC unit affect this?
QUESTION 13:
What effect do oxygenates in the olefin feed have TIEMAN:
on the operation of the HF alkylation unit? We have seen few instances of oxygen contamination of
HF alkylation units without MTBE processes in the same
UTLEY: complex. We have had several occurrences when oxygen
Generally oxygenates in the alkylation feed will form a contamination has occurred, and it was usually traced to
light AS0 that is difficult to reject from the unit, so the acid alcohol in the saturate butane make-up. On UOP FCC
purity will fall quite rapidly. The low acid purity will likely effluents we have seen no evidence of C3 or C4 oxygenates
cause an increase in the organic fluoride content of the coming out with the olefins.
alkylate normal butane and propane streams. The light
AS0 can be rejected in the rerun column, but the tempera- STEGELMAN:
ture must be reduced substantially to do it. The oxygenate We did not appear to have a problem before the MTBE
produced ASO, the increased fluoride content of the prod- unit came on-stream. However, we have found some acetone
ucts, and the cool temperatures in the rerun column all which does not appear to come from the MTBE unit since it

113
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

is not found in the purchased methanol. This was not Generally most of this oxygen shows up as carbon di-
checked for before the MTBE unit, so we suspect it may oxide, carbon monoxide, carbonyl sulfide, but the reactions
have always been coming from the catalytic cracker, al- in an FCC reactor riser system are very complex, and not
though this sounds unusual. fully understood. A tremendous number of trace com-
pounds can be formed, some of which, if oxygen is present,
MACKEY: will contain oxygen.
We feel the major source of the oxygenates in our alkyl-
ation unit is from the fuel butane stream which we pur-
chase in significant volume from outside sources. We feel Process
that this stream contamination is probably the result of
being treated with alcohols at the production facilities to
control water. We utilize external acid regeneration to QUESTION 15:
control the water content in the acid to below 1%. Histori- Are there easy analytical methods which can be
cally our acid losses have been about 0.16 pounds per barrel used to determine the amount of NaF salts in the
of alkylate overall. We do not feel that we get oxygenates caustic solution used for neutralization of alkylation
through from the FCC unit. unit acid gases?

MORGAN: BRANHAM:
We routinely measure oxygenate levels from our FCC Ashland has found ion chromatography to be an easy tool
units in the propylene/butylene stream between 10 and 50 for the determination of ionic species in water. Ashland has
ppm. These oxygenates are mainly acetone, isopropanol, developed a number of ion chromatography methods for
and sometimes small quantities of ether compounds. We refinery technical services applications. A method for this
have not been able to correlate FCC operating variables to particular application has not been developed, but ion
the oxygenate levels measured. Additional oxygenates in- chromatography should be very amenable to this
troduced into the alkylation feed from the MTBE unit are technique.
DME and methanol. Although both of these affect the
alkylation unit, we watch the methanol level closely since MORGAN:
upsets in the MTBE methanol recovery section can slug We successfully measure sodium fluoride salts in caustic
significant quantities into the alkylation feed stream. solutions with ion chromatography. Another method that
DAVID L. SMITH (ALCOA): we have used that is somewhat easier but less accurate, is to
A number of refiners have reported problems with trace use specific ion electrode. The caustic solution usually has
oxygenates in their feed to the alkylation unit which have to be conditioned to eliminate electrode interaction
then resulted in AS0 problems. An effective method to problems.
remove these trace levels ofoxygenates, whether there is an
upstream MTBE unit or not, is via selective adsorption.
ALCOA has developed a product, Selectsorb CD, that can QUESTION 16:
attain effluent levels on the oxygenates well below 1 ppm. How well is split olefin feed injection working as
compared to predictions for octanes, yields, AS0
KENNETH PECCATIELLO (W. R. Grace & Co.): production?
We have seen some FCCU’s that have acetone showing up
in the effluent. It appears from first glance that it can be STEGELMAN:
correlated with reactor temperature, the amount of riser Phillips has designed several split olefin feed injection
steam going into the unit at the time, and high catalyst systems during the past year. These systems are most often
circulation rates. These high rates will tend to entrain O2 used by a refiner to increase the capacity of his alkylation
from the regenerator.
WILSON:
I will comment a little bit about the FCCU and sources of
oxygen. Most of the oxygen that you see in FCC reactor
products come from air entrainment into the reactor from
one place or another. Places where you can get this are
obviously from the regenerator, and this is primarily affect-
ed by the design of the regenerated catalyst draw off
system, i.e. where it is located with respect to the air
distributor, the velocities going into the standpipe and into
the hopper, and how well it is designed to work on de-
gassing of the catalyst as it is returned to the riser. A few
units use some air in aeration for purges in the regenerated
catalyst system. This also will allow air to get into the
reactor. There are areas where air is used as bleeds. This
will also put a small amount of oxygen into the reactor
system.

114
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

unit typically by 25% or more. A slight increase in alkylate the acid settler are used to prevent most of the entrained
octane may also be achieved by using split olefin feed at the acid from reaching the settler hydrocarbon outlet. Proper
expense of increased ASO make. Their use also allows the sizing of the holes in the trays and providing adequate
total overall isobutane/olefin ratio to be reduced, thereby settling volume or residence time will also help minimize
resulting in energy savings. This is achieved by reducing acid entrainment.
isobutane recycle and taking advantage of the unreacted
isobutane moving up the reactor to create high localized TIEMAN:
isobutane/olefin ratios at the upper sets of injection nozzles. Most of the design criteria have been stated. In addition,
we do incorporate in new units and revamps enhanced
BRANHAM: settling devices in the settler too.
Splitting the olefin feed in a two reactor system has the
effect of doubling the isobutaneloletin ratio at constant HEROS DERGREGORIAN (Giant Refining Company):
isobutane recycle rates. Our data would indicate that unit Which criterion would be more controlling in affecting
performance in terms of octane yield and ASO make follows carryover: velocity or residence time, or both?
closely with the prediction at the resultant isobutane/oletin
ratios. Ashland currently operates three UOP split olefin BRIAN JOHNSON (UOP):
injection HF alkylation units. Ashland’s average yield, UOP uses settling time as the main criteria in the design
based on olefin feed, is 1.83 barrels of alkylate per barrel of of horizontal settlers, and upward velocity as the main
olefin. The average RONC of the alkylate product is 93.1 criteria for the design of vertical settlers.
which does include some propylene derived alkylate. No
comparison of actual ASO production to theoretical has UTLEY:
been made due to the many factors that do contribute to the When carryover does occur in our unit, it is very difficult
ASO production. to tell. Unfortunately, we do not know that it has happened
until after the fact in most cases, when we go in to inspect
MORGAN: our DIB column. Some things you can watch for while the
In one of our units, the Phillips split oletin feed system unit is on-line are pressure fluctuations in your overhead
has increased our capacity within predicted yields, octane, receiver, or in our case, the recycle isobutane pump seal will
and acid settler capacity. The ASO production was slightly smoke a little bit. Also, you can watch for excess acid
higher than predicted. We obtained a 15 to 20% increase in returning from your overhead receiver boots, if your oper-
capacity, and about 0.5 number increase in octane. Our ator picks it up.
other unit with split feed has been difficult to evaluate due
to feed contamination problems.
QUESTION 18:
What is the HF content of ASO from acid re-
QUESTION 17: generation towers? Is the ASO neutralized? How is it
What are the important criteria to prevent acid disposed?
carryover from the settler?
BRANHAM:
GESICK: HF content in the ASO with good regenerator design and
Basically to prevent acid carryover, you need to maintain operation should be less than 5%. The HF content can be
a low operating level in your settler, and keep your acid/ affected by the mechanical condition of the regenerator
hydrocarbon ratios between 1.5/1 and 2/1. internals. Operation at too low of an overhead temperature
will directionally increase the HF content of ASO. The
MORGAN: length of time that the ASO and CBM are allowed to settle
The three most obvious process criteria affecting HF before dumping will also affect the acid content.
carryover are the settler velocity, proper acid purity, and ASO is usually neutralized using liquid KOH as a neu-
the acid settler level. At times we have noticed that low tralizing media. The ASO is usually burned as sup-
isobutane recycle purity also seems to accelerate acid plemental fuel in the alkylation isostripper reboiler heater.
carryover, especially when operating at high rates. Proper
attention to settler internals installation and maintenance MORGAN:
can also reduce carryover problems. We have one unit that runs about 1 to 3% HF in the ASO.
This unit runs very low sulfur, oxygenates, and butadienes
BRANHAM: in the feed. In contrast, another unit that has significant
Carryover criteria from these systems is similar to any problems with feed contamination contains 10 to 20% HF
other liquid-liquid system; i.e., adequate residence time for in the ASO. Our ASO is neutralized with KOH and re-
disengagement and a low velocity to prevent entrainment cycled either to the coker or FCC unit with little to no
of liquid bubbles. Knowing where the acid level is in the downstream operating problems.
settler is also important. Acid strength can also become a
negative factor if allowed to get too low. STEGELMAN:
The HF acid content of ASO from the rerun tower will
STEGELMAN: vary widely depending on optimization of acid feed tem-
Sieve type trays covering the full cross-sectional area of perature, stripping isobutane temperature, and tower

115
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

pressure. It should be possible to reduce HF losses to less We now watch the temperature very close on shutdown and
than 1% in the rerun bottoms when operating conditions start-ups to maintain it close to 300 °F with acid in the unit,
are set to remove only heavy ASO. When operating condi- and have not had any further problems.
tions are changed to facilitate removal of light AS0 or
water, then HF losses to the rerun bottoms can increase BRANHAM:
dramatically. In most cases, cyclical operation of the rerun Most people have experienced some degree of plugging in
tower will be required. their isostripper tower after a shutdown and start-up. We
At Borger, our recent average is relatively high, but it is have found that our problems are usually caused by iron
difficult to get a good reliable sample, and we do not have fluorides which precipitate out due to moisture in the unit,
much confidence in the numbers obtained from the labora- and make their way into the isostripper where they accu-
tory. We do know that our acid consumption is very reason- mulate on the trays. Of course, the way to avoid iron
able. In most instances, the AS0 is neutralized and burned fluorides is to dry the unit more thoroughly before the
in the main fractionator furnace. At Borger, we do not have introduction of the acid to the unit, and also to keep it dry
a furnace, and our neutralized AS0 is put in the topped during normal operation.
crude tank and is fed to the resid HDS unit and on into the During one turnaround, we sandblasted the inside of the
catalytic crackers. isostripper in order to have it x-rayed for possible shell
cracks. Consequently, large amounts of iron fluoride scale
UTLEY: came off from the tower walls and accumulated in the
I agree with Mr. Stegelman. It is very debatable on true isostripper fired reboiler. Upon start-up, the iron fluorides
HF concentration of the ASO. In our case we can measure it formed hydrated fluoride compounds which dissolved in the
just before it enters our DIB reboiler, and we see around 1 hydrocarbons at the higher temperatures and re-
wt % HF in the ASO. We do not neutralize it, and we send it crystallized at the lower temperatures up the tower. These
to our reboiler and burn it as soon as it is produced. crystals plugged the holes on the underside of the bottom
ten trays. This build-up had to be chiseled out.
HEROS DERGREGORIAN (Giant Refining Company): Another possibility is that if soda ash is used to neutral-
In order to optimize the regenerator operation, acid re- ize the unit before maintenance, and this is not flushed out
covery and prevent AS0 returning to the unit, what condi- properly, then you could flush this material into the tower
tions have the panel or audience experienced, bottom tem- during start-up.
perature and also the top temperature in order to optimize
the operation?
General
STEGELMAN:
The actual operating conditions on the rerun tower are QUESTION 20:
much a function of the pressure of the rerun tower. This
What has been the industry experience with seal-
depends on how you return your rerun overhead back to the
less pumps for hydrofluoric acid? How do they com-
settlers. We have found the best way to determine the
optimum conditions is to very carefully study the rerun pare to double mechanical and tandem seals?
operation. Keep good records, and select your own proper
RATERMAN:
operating conditions.
We do not currently use sealless pumps in HF acid
service, but we are investigating their potential use. We
currently use either a single mechanical carbon on silica
QUESTION 19: carbide seal or tandem seals made from the same material.
We have been experiencing plugging problems in We use an acid-free isobutane flush stream to these seals
our isostripper tower trays after shutdowns and start- and usually get about one year of service from them. In
ups. Has anyone else experienced these problems? addition to investigating the use of sealless pumps, we are
What causes the problem and what can be done switching some of our carbon on silica carbide tandem seals
about it? to a silicon carbide on silicon carbide with an ATF barrier
fluid between seals to see if we can improve their
TIEMAN: performance.
Plugging in the isostripper tower after shutdown and
restart can be caused by high initial water levels due to an STEGELMAN:
insufficient dry out. Also, if there has been neutralization Several of our licensees have had very good service with
taking place, and if there are still neutralization deposits, sealless pumps, such as HMD pumps from a United King-
such as ammonium fluoride, there can be plugging because dom manufacturer. Phillips corporate engineering has re-
of inadequate washing before start-up. viewed several U.S. manufacturer pump options and like
the design and construction of the Sundyne can motor
UTLEY: pumps, type HS with monel construction. However, we do
We have experienced the same type of problems, plug- not have any operating experience with these pumps in HF
ging in the bottom section of the DIB. We experience this service.
problem when we let the DIB bottoms temperature get too
cold with acid in the unit. The deposit was a light brown TIEMAN:
material. It was easy to remove, but it did plug the tower up. We do not have any experience with the sealless pump in

116
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

operation, but there is one location where a sealless pump we have also used Tufline valves with good success. At the
has been ordered and will be put in service in the future. present time, all of our isolation valves are Durco valves.
We are working on a dump system and several other modi-
UTLEY: fications, and we have not yet specified the valves, but they
I believe that industry has gained experience with seal- will probably be Durco or Tufline.
less pumps, but there is still a long way to go. There are still
problems with material compatibility. The clearances are
tight, so fouling and plugging are problems. Finally, the QUESTION 22:
head and volume are limited, so applicable services are How are people doing ambient monitoring for
restricted. It is hard to compare the two types of pumps. We hydrofluoric leaks in and around their hydrofluoric
still use a single seal with an alfa centered silicon carbide alkylation unit?
face on both the rotating and stationary face. The station-
ary face is hydro-padded. We get one to three years of MACKEY:
service out of these seals depending on the number of upsets We have installed an atmospheric monitoring system
and power failures that we have. utilizing Gastech HF detectors, model 2321. We have 22 of
We tried double mechanical seals flushed in between the these monitors placed throughout the alkylation unit in
two, and had a lot of trouble with the flush circulating the acid-containing areas. They feed into an alarm board
system. Every time we had trouble with this system we located in the alkylation change room, with a common
backed acid into the flush. This system was on our de- trouble alarm that goes from there to the main control
propanizer charge pumps with the ability to expand to all of room. The alarm board consists of the unit layout indi-
the alkylation pumps. With the trouble we had on the cating the sensor locations. The detectors alarm at a low
depropanizer pump, we were hesitant to expand it to any level of 5 ppm, and a high level of 10 ppm. The system
other pumps. We eventually changed the depropanizer requires a quarterly PM schedule to calibrate the sensors.
charge pump back to a single seal. It seems to work acceptably well. We do get occasionally a
spurious alarm from significant humidity or temperature
CARON: swings. The operators have seen satisfactory results with
At the Sun Sarnia refinery we do not use sealless pumps very minor packing leaks, etc., these alarms picked it right
in HF acid service. We are beginning to install tandem seal up. Overall we are satisfied with the installation.
pumps for this service.
BRANHAM:
I might add that Ashland has field tested several port-
QUESTION 21: able monitors for HF. These include a GMD Systems Auto-
What type of emergency block valves are being stop portable monitor for HF. This basically is a paper tape
used in acid service? How are they being tested? Are type of system and is relative cheap and easy to maintain.
they being installed such that they can be fully There is also an MDA TLD-1 Toxic Gas Detector. The third
stroked? one is a Gastech model 2321, which uses a diffusion electro-
chemical technique. This one is rather expensive.
BRANHAM:
Ashland’s present emergency block valves for acid ser- STEGELMAN:
vice are gate valves that are constructed to UOP HF-1 Phillips has felt that elevated color TV surveillance
rating. These are equipped with remote electric motor cameras and hydrocarbon detectors have been the best
operated actuators. The valves are usually tested weekly. means of providing reliable ambient monitoring for HF
The valves are installed so that they can be fully stroked, leaks in an alkylation unit. This is because an HF leak
but usually the valves travel is restricted to only about would normally be associated with a hydrocarbon release.
50%, as this is viewed as an adequate test. However, with the advances being made in HF detectors
recently, we think that it will not be long before very
MORGAN: reliable detectors will be available that can be used for this
We are using Tufline plug valves for our emergency acid service.
shutoff systems. The valves are installed on both the suc-
tion and discharge of all pumps containing significant WILLIAM SCHLESING (BP Oil Company):
amounts of HF to protect against seal failures. These Mr. Mackey, on the detectors that you are using, at this
valves are remotely operated and tested weekly. Since the point, I assume that they are not tied into any sort of a
pumps are spared, the valves can be fully stroked. We also remote shutdown. Do you have any plans on tying them in,
install emergency shutoff valves on HF containing lines in the future, where they automatically shutdown on an
such as acid transfer lines and reflux pump discharge alarm?
piping. Since these lines are spared or in non-continuous
service, the valves can also be fully stroked when tested. MACKEY:
We went through a very extensive risk management
STEGELMAN: assessment of the unit. The conclusion we reached is, as far
Durco and Pacific valves have been used satisfactorily by as the automatic shutoffs and so forth tied to these detec-
many of our licensees. Some are being installed such that tors, that we did not feel that it was necessary at this stage.
they can be fully stroked on a specific frequency. At Borger, We have the unit manned with two people 24 hours a day,

117
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

and we wanted them to make an initial assessment of the field, Virginia. The report is authored by the above-named
situation and take appropriate action. So we have not group.
coupled these to shutoffs at this time.
SCHAUB:
WILLIAM SCHLESING (BP Oil Company): Our safety services division has done HAZOP studies
I assume that the main concern is to avoid potential false and fault-tree analysis for clients with HF alkylation. They
alarms and an unplanned shutdown. worked together with our environmental services division
who has done HF cloud modeling on computer, using some
MACKEY: of the “what if?” scenarios. We have a lot of experience with
That is right. these kinds of studies by virtue of our long standing work in
the nuclear power industry. We have designed systems for
unit surveillance and water deluge, using the test data
QUESTION 23: generated from the study Mr. Raterman just mentioned. I
What plans do you have to control possible hydro-
will point out for reference the June and July issues of
fluoric acid releases?
Hydrocarbon Processing, 1990. We have installed one of
these water deluge systems at one of our client’s facilities in
BRANHAM: Canada. It includes remotely controlled articulating water
Prevention is the key issue in all cases. Release potential sprays based on field-mounted camera images.
is mitigated by maintaining minimum acid inventory in
the unit at all times. Planned procedures include quick STEGELMAN:
isolation of the source and the moving of acid to a more The following mitigation measures are currently being
secure storage location within the unit. Water sprays would considered as a means of reducing the risk attributed to
be used to contain the HF to the immediate area. releases of HF acid: (a) the installation of water spray
systems, (b) the installation of ambient or environmental
MACKEY: monitoring systems, (c) installation of emergency isolation
We are also involved with two different levels of water or shutoff valves, if they do not already exist, (d) reducing
mitigation for any potential HF leak. We have installed a HF inventory by acid settler compartmentalization, (e)
deluge system over all of the rotating equipment that removal of acid circulation pumps, (f) installation of a rapid
plumbs up a specific shower head that is triggered in the acid dump or deinventory system, and (g) performing site
event that we encounter any kind of HF leak. We also have specific risk analysis to locate any places where you might
installed remotely actuated and operated elevated fire have the chance of a leak.
water monitors that are capable of putting out 1000 gallons I might point out in reducing HF inventory, we have at
per minute of water from each of five locations around the Borger two identical settlers, one with conventional acid
acid containing areas of the plant. cooler bundles, and the other with rod baffle cooler bundles.
Obviously if you have that much water, the containment We can operate about 1.5 to 2 tricocks, lower level in the
also becomes an issue. So both our tire water systems and settler with rod baffle bundles, and obtain the same acid
the drainage systems have been beefed up to accommodate circulation. So this is a means of reducing acid inventory in
the high water flow for a reasonable period of time. We feel the settler.
that the use of water in this kind of quantity, with this sort
of redundancy, is a safe and brute force way to deal with the ELTON C. EUBANKS (Chevron U.S.A. Inc.):
issues that are very prominent these days. We have never Has anyone had a problem with sampling HF with alu-
had to use the system in practice other than to demonstrate minum bombs?
it for many agencies and other refinery interested parties
that have come through our facility to view it. Basically we MACKEY:
feel that it is a functional and viable way to mitigate HF We do not use aluminum. We use solid monel bombs.
releases.
STEGELMAN:
RATERMAN: We also use monel bombs.
Mobil was one of the participants in the Industry Co-
operative HF Mitigation Assessment Program which con- UTLEY:
ducted HF release tests at the DOE test site in Nevada near We do not use aluminum either. We use monel.
Las Vegas in 1988. Based upon our experience with these
tests, we have opted for a two part protection system in the MICHAEL SCHINDLER (Mobil Oil Corporation):
event of a major acid leak. The major line of defense will be Mr. Mackey, when you said that you beefed up your water
an acid evacuation system which will not rely on a pump to handling system in case you have to use the elevated water
transfer the acid to the safe location. This will be backed up monitors, would you expand on that.
by an effective water spray system which will blanket the
entire area. The spray system’s design was based upon the MACKEY:
desert test results which studied the effect of using water Our fire water system is capable of delivering about
sprays to knock down the acid. The results of these tests can 9,000 gallons per minute of water in that quadrant of the
be obtained from the U.S. Department of Commerce refinery. What we did was to dike off the acid containing
National Technical Information Service located in Spring- areas where we feel that most of the water would wind up

118
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

flowing in the event of an emergency. Then we beefed up the STEGELMAN:

sewer system and the neutralization pit where this water Our current design philosophy for an HF acid emergency
would go in order to contain that flow for about 20 to 30 dump system for an alkylation unit is to provide for the
minutes of time. transfer of acid from the unit to remote storage in ten
minutes. The dump system is for both HF acid and hydro-
DANIEL KNEPPER (CENEX): carbon. The system dumps to a vented receiving vessel that
Mr. Mackey, when you say elevated monitors, how high is initially maintained just above atmospheric pressure. A
in the air are they? back pressure controller is provided to keep the HF acid
above its bubble point, thus preventing overloading of the
MACKEY: acid relief neutralizer during emergency transfer of the
They are 30 feet in the air, and are remotely operated by a acid. Gravity transfer is preferred over the use of sealless
joy stick, not dissimilar to a computer game. Our operators pumps. Pressure transfer is the preferred method of mov-
test all live of these monitors out once a week. ing the liquid from the receiving tank back to the alkyla-
tion unit.
DANIEL KNEPPER (CENEX):
What kind of a spray pattern are you using? You are not TIEMAN:
using a fog, so what is the spray pattern? After looking at different systems, we are currently de-
signing, and prefer to transfer by pressure. Although it is
MACKEY: rather expensive, we are going to an unvented drum. The
The monitors are capable of variable patterns. They can reasons are that it is the most simple and reliable system
go from a full fog to a direct stream. We have educated and that we could find.
trained the operators to use the proper pattern as they set
up. We drill them periodically by simulating a release, etc. MACKEY:
We have a video tape available if you would like to get a We are designing an emergency dump system utilizing
copy of it. about a ten minute evacuation period for any one of the
three major acid containing vessels. Our design is focused
on using a pump because of space limitations, but we do like
QUESTION 24: the idea of having a non-pump system by virtue of its
simplicity. Utilizing a pump that is seldom, if ever, used is
What philosophy are refineries using to design a
hydrofluoric acid dump system for the HF Alkylation
the major critical aspect of a dependable system.
Unit? Is the dump system for HF or HF and hydro-
carbon? Does the system dump to a vented or to a
QUESTION 25:
closed receiving vessel? What is the vent pressure if
the acid receiving vessel is vented? Should the HF be What is the present thinking on the maximum
pressure transferred or pumped? allowable temperature for Monel 400 in an acid
regenerator? What is a typical regenerator life span?
RATERMAN:
Our design philosophy is to transfer the total acid inven- GESICK:
tory to a remote vented drum within five minutes. The I do not have any comment on the maximum allowable
drum is vented through a caustic scrubber. The drum pres- temperature for Monel 400, but at Superior our first acid
sure will depend upon a number of operating conditions, regenerator lasted about 16 years. The second one lasted 11
but basically it floats on the scrubber which is not pressure years. The regenerator at Meraux also lasted 11 years. The
controlled, but vented to the flare system. Pressure transfer Superior system runs about a 500 ºF bottoms temperature
is used so that the dumping will not be dependent upon the and a 155 ºF top temperature. The Meraux system runs
operation of a pump. about a 450 °F bottoms temperature and 140 ºF top tem-
perature. The purchase of a substantial corrosion allowance
in the nozzles, particularly the small diameter nozzles, can
serve to extend the life of the regenerator. Long weld necks
or double extra strong should be purchased on anything
that is 4 inches in diameter or under.
Our experience has been that repair welding on Monel
vessels that have been in this service is unpredictable at
best. We elected to purchase a new vessel following our last
turnaround rather than gamble on a successful nozzle
replacement in the vessel. Most of our corrosion in these
vessels is in the bottom sections of the towers. When they
have failed to the point of replacement, it has been in the
bottom section of the tower.
MACKEY:
In the recent four or five years, we have had some fairly
aggressive corrosion in our acid regenerator. Our corrosion
has been primarily in the trayed area of the column, as

119
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

opposed to the bottom. We feel that our general operation of section of the column. I notice that everybody is referring to
the vessel seems to be within normal operating parameters. the problem of the corrosion at high temperature. Can you
We have not been able at this point to identify the exact overcome that by lowering the operating temperature and
cause and mechanism of the corrosion. We are going to use increasing your stripping isobutane to achieve similar
different tray designs in that portion of the vessel and stripping?
anticipate extended vessel life.
STEGELMAN:
BRANHAM: We have found that we can adjust the stripping isobutane
Our regenerators are made of solid plate Monel 400, and slightly and get some drop in acid rerun feed temperature.
the expected life is 15 years. We will be shortly replacing We generally do not move our feed temperature over about
one that does have 20 year service. The nozzles and trays 10 oF during a long period of operation.
are expected to last five to ten years. The typical operating
temperature for our regenerators is usually somewhere HEROS DERGREGORIAN (Giant Refining Company):
around 300 to 350 oF at the bottom. Before we move on to reforming, there is a very impor-
tant subject to be addressed as far as HF alkylation. I would
CARON: like to ask the panel or somebody in the audience to bring
We have experience with two units. In one we limit the me up to date as to the future situation for HF alkylation as
acid regenerator to 300 oF with a typical regenerator life far as the problem of eliminating it in our future design.
span of about ten years. Our other regenerator operates at Secondly, has there been any development work on an
about 295 oF This vessel was replaced in 1965, and has alternate to sulfuric acid alkylation replacing HF?
been in service since with only minimal repairs.
MACKEY:
RATERMAN: I presume that you are alluding to some of the legislative
Most of the corrosion observed by Mobil has been in the efforts with regard to HF processes. In California there is a
acid feed heater of the regenerator which operates at about very aggressive effort now by one of the regulatory agencies
300 oF. At temperatures above 350 oF, the corrosion rates to eliminate hydrofluoric acid units in southern California
for Monel in this service will become very high. To avoid in particular. They are promulgating some rules to poten-
high tube temperatures, saturated steam at pressures less tially legislate out HF units before 1995. Obviously, those
than 100 psi should be used in the heater. In addition, of us that have units out there are very concerned about
pitting of the Monel heater tubes is caused by oxygen at this. We are working with the agencies to attempt to estab-
concentrations as low as 1 to 2 ppm regardless of tempera- lish a more rational basis for making conclusions regarding
ture. Oxygen contamination usually results from using the hazards or the lack of hazards around HF relative to
acid moved by air rather than acid which has been trans- anything else that is used in the refining industry. So, the
ferred by nitrogen, On the average, the Monel heaters in process is under discussion as far as that goes.
this service survive about ten years or more. I believe that you are probably also aware that there is
also a federal piece of legislation involving HF units that is
STEGELMAN: on a more extended time frame. I believe that it is about 12
Metal temperatures above 250 oF cause accelerated cor- or 13 years to phase them out. I think that as an industry, it
rosion in Monel 400. Experience indicates the practical is critical for us to represent the seriousness of mixing
limit for alloy 400 in HF service is about 300 oF We design politics and technology about this unit, or any other, in
for a maximum rerun tower acid feed temperature of 290 oF terms of its viability. One aspect, though that will certainly
Our rerun at Borger has operated for 8.5 years and has had affect this sort of legislation is any kind serious accident or
no general corrosion on the shell. There is some roughness significant release of HF or any other toxic from our busi-
where condensables off the trays have run down the wall. nesses. I think it is particularly critical for everyone in our
There is also some corrosion on the trays. business to take a very aggressive stand about mitigating
potential releases and being sure that the types of systems
TIEMAN: that are being installed to prevent releases are effective and
Our experience and that of our licensees has been similar functional. Additionally, a comprehensive site-specific
with a typical life span in the range of five to ten years. Our hazards assessment is very important.
nominal design life is about ten years. This has been
reached by quite a few licensees with good initial fabri- TIEMAN:
cation quality and operating conditions. As to the second part of your question, UOP and CEPSA,
from Spain, have jointly introduced the commercialization
UTLEY: of a fixed bed, heterogeneous catalyst alkylation process for
I agree with what has been said. We feel that the tem- the alkylation of benzene to form linear alkyl benzene for
perature range for Monel 400 is between 160 and 300 oF detergent production, In that process, though, that reaction
metal temperature. The process temperatures can be is much easier and does not require the activity that the
slightly higher. I do not know what a typical regenerator is, motor fuel application of alkylation does. So, at this time a
but Monel columns should last somewhere around 20 years. solid bed system is not being offered for the motor fuel
application.
HEROS DERGREGORIAN (Giant Refining Company):
We do have extensive corrosion in our regenerators, MARTIN W. PERGA (CENEX):
especially in the area of stripping isobutane in the hottest My comment is relevant to Mr. Mackey’s point on HF

120
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

safety. We are doing some piping rework in our alkylation The range of measurement for both are 0 to 1 ppm full
unit where we were putting in drain valves and we are scale, and possibly 0 to ¾ ppm if you push it. I believe that
going to stress relieve that piping. We had these valves they are reliable, but require good maintenance practices
broken apart prior to stress relieving and the top bonnet of as any on-line monitoring equipment does. Potential prob-
those valves, the bolts were stretched. We measured lems arise in a couple of areas. The sample injected must be
torques holding these valves together from finger-tight, of an accurate consistent quantity. The valve switching
that is, you could loosen them with your finger up to 55 lb-ft must be monitored regularly, and the system must be
torque. The bolting was visibly stretched. This was not a cleaned.
micrometer measurement. I might just caution you that We run at about 0.1 to 0.2 ppm sulfur in our feed and test
these were new valves tested at the factory. It is a reputable it using a laboratory Tracer-Atlas machine. There are two
manufacturer. They were not satisfactory as we were going reasons why we decided to go this route. Consistent accu-
to put them in the plant, rate analysis at this level is a challenge to do with an
on-line instrument and requires the constant attention of
maintenance personnel which we do not have available at
C. Reforming our refinery. Secondly, when we have sulfur excursions,
this instrument has proven invaluable in tracking down
the cause by running samples from different locations in
Feed the plant.

QUESTION 27:
QUESTION 26:
Has anyone obtained plant data relating naphtha
Are on-line sulfur and H2S analyzers available for
reforming severity and feed properties to reformate
monitoring reformer feed and the hydrogen recycle?
aromatic levels and, if so, what are the results?
How reliable are they?

CARON: BRANHAM:
On-line analyzers are available. For sulfur in the feed, Ashland’s data for our fixed bed reformers shows a typi-
there is the Tracer-Atlas; and for H2 S, you can use a cal aromatics content of 47.5 vol % at 91 RON, 48.5% at 92.5
Houston-Atlas. We use the Houston-Atlas in both the RON and 60% at 97.5 RON. The aromatics levels for the
hydrogen recycle and the stabilizer offgas. We particularly CCRs showed 56% at 96 and 62.5% at 99. Naphthenes
like it in the stabilizer offgas. We use these analyzers for varied from 21 to 36% and no correlation was evident,
trending purposes. Any high readings are confirmed by
unit operating variables and/or laboratory analysis of the CARON:
feed. We have results from a start of cycle test run as follows:
Reformate, Feedstock Feedstock Reformate
MACKEY: RON Naphthenes % Aromatics % Aromatics %
Some years ago we had installed a Houston-Atlas H2S 93.6 30.6 12.0 45.0
analyzer on our recycle gas stream, and we continue to 96.4 31.4 12.2 52.3
monitor stabilizer gas vent for H2S content as well. We felt 98.1 31.4 12.5 58.8
that from our experiences that having the on-stream analy- 100.6 31.5 12.9 62.5
zer really did not serve any significant function to outweigh
the maintenance and upkeep of it. So we have reverted to STEGELMAN:
the standard stabilizer offgas monitoring on a once a shift I have looked at one set of data for feed with 17 to 18%
basis. aromatics. The change in delta aromatics is about 0.3% per
°F of reactor inlet temperature. The data were for a 40 °F
RATERMAN: range of inlet temperatures.
We do not have any experience with on-line analyzers for
sulfur in the feed. We do use a Houston-Atlas analyzer to WILSON:
monitor H2S in the recycle gas stream of one of our units. I would just like to cover some data that were published
This analyzer has provided excellent service. However, for by the NPRA in their 1990 gasoline survey. The pre-
the most part, we still use Draeger tubes to monitor the H2S liminary results came out recently. These data are for
in the offgas from the debutanizer. blended pools of reformate, so they do not reflect any one
unit or any one feed. At 97.7 research octane number, the
UTLEY: aromatics were 62.6 vol %. At 100 octane, they were at 68.3
There is equipment available that has been satisfactory vol 1%. There have also been some data published for ben-
in monitoring on-line reformer feed and recycle gas sulfur zene contents. At 98 octane, benzene is between 3.3 and 3.8
content. The equipment I am familiar with is Tracer-Atlas, vol %. This range is mostly a function of the age and
model 722R, for recycle gas, and model 722R./756 for feed operating pressure of the reformer. At 102 octane, they
sulfur content. Both measure total sulfur by converting all were between 4.0 and 5.4 vol %.
sulfur compounds to H2S and then exposing it to lead
acetate paper. The sulfur concentration is calculated by KENNEDY:
measuring the rate of change of color on the paper. Data collected when we were charging a 195 to 350 °F

121
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

boiling range reformer feed, shows that as reforming sever- operated the reformer for five days at a feed sulfur aver-
ity increases, the benzene level in the product increases, as aging approximately 200 ppm. To ease the effect on the
does the total aromatics level, and the 10% point decreases. catalyst, we did lower the octane to 90 rather than keeping
The 90% point does not change significantly. the temperatures up to maintain the higher octane. We also
At a RON of 97.7, the total C5+ aromatics were 57.4 were warned that it would take several regeneration cycles
vol %, the benzene level was 3.8 vol %, and the 10% point to totally remove the sulfur from the catalyst. The reason
was 130 °F? A reduction in severity to a RON of 87.5 pro- this route was deemed acceptable in this unit was that
duced a reformate with 41.5 vol % C5 + aromatics, 2.9 vol % extra regeneration capacity was available, and therefore
benzene, and 10% point of 170 °F. the residual sulfur could be removed. This unit has re-
turned to normal operations.
J. G. DEARWATER (Profimatics, Inc.):
As very rough rules ofthumb, 90 RONC reformate is 60% RATERMAN:
aromatics and 100 RONC reformate is 70% aromatics. We have no available commercial data on the effects of
Three parameters cause the primary shifts in these rules of sulfur excursions to a continuous regenerative reformer.
thumb. We try to keep the sulfur levels in the feed below 0.5 ppm.
a. As WABT is increased, the equilibrium shift towards We would expect that a continuous regenerative reformer
iso-paraffins increases, thus reducing the quantity of would have the ability to quickly recover from a sulfur
aromatics needed for a given RONC. upset due to the fact that the sulfur is a temporary poison to
b. Shifts in reformate IBP can also have an effect. If a the active metal sites. We do have data on a semi-
substantial amount of iC4 is left in a 90 RONC re- regenerative unit where we had 2 ppm feed sulfur and lost 1
formate, the percentage of aromatics in the reformate to 2 vol % of the C5 + yield. We got about a 40% increase in
will be lower than if the reformate has no C4 paraffins. coke for the same 2 ppm sulfur operation.
c. The feed also has a very significant impact. Typically a
hydrocracker feed will produce a reformate with a HEROS DERGREGORIAN (Giant Refining Company):
lower percent aromatics than a virgin feed of the same Does anybody have experience using guard bed as a
distillation and PONA. Higher naphthene and aro- contingency unit for protection in case of an hydrotreater
matic feed gives a higher percentage aromatics in the upset to protect the Platformer?
reformate. Lower endpoint material (300 EP versus
400 °F EP) will result in a slightly higher percent TIEMAN:
aromatics in the reformate. We have one Continuous Regenerative Platformer that I
know of with a sulfur guard bed in front of it. The purpose
there was not to handle upsets in the naphtha hydrotreater,
QUESTION 28: but it was a feed coming from a hydrocracker. The hydro-
What is the maximum sulfur content in the feed for
cracked naphtha could potentially have 2 to 3 ppm sulfur at
a reformer with a continuous regeneration section
end of run catalyst conditions on the hydrocracker. So the
before catalyst activity, unit yields, or coke formation
guard bed was for a different purpose there. I do not know of
is substantially affected?
any CCR Platformer that has a guard bed for protecting
from naphtha hydrotreating upsets.
TIEMAN:
The answer to this question is that we still recommend CHARLES MCCOY (McCoy Consultants):
low sulfur levels in the feed to continuous Platformers. Mr. Tieman, you mentioned that the CCR has the ability
Sulfur is a poison to the platinum function on the catalyst, to get out of a sulfur upset more quickly than a con-
so it affects both the selectivity and the activity of the ventional or semi-regenerative unit. Does not the presence
catalyst. The difference with a continuous regeneration of sulfur interfere with the regeneration function?
unit is that you can get the sulfur off the catalyst with each
cycle through the regenerator, so you can recover quickly TIEMAN:
from a sulfur upset. We still recommend less than 0.5 ppm We have several operating variables in the regenerator.
sulfur in the feedstock. Our experience with the regeneration of UOP CCR Plat-
We have had an experience where we had 2 to 3 ppm of forming catalyst with the severe sulfur upsets has been
sulfur in the feed for a relatively extended period of time. completely successful. We do not end up with sulfate forma-
The refiner held the octane constant. He suffered a 15 to tion or permanent poisoning of the catalyst.
20 °F activity loss, a 1 to 2% C5 + yield loss, and approxi-
mately 1.5 times the normal coke make on the catalyst. R. W. BENNETT (UOP):
On the guard bed question, I know of another CCR unit
BRANHAM: that has a guard bed in front of the Platforming unit, but
We generally follow Mr. Tieman’s recommendations, and when they have a sulfur upset, they bypass the guard bed,
we have not tempted fate by purposely operating above the so that they do not kill it.
prescribed sulfur limit of 0.5 ppm sulfur. We did have one
particularly interesting experience for a short term oper- BASHIR HASSAN ALI (Abu Dhabi National Oil
ation of our CCR Platformer at our Canton refinery. An Company):
exchanger leak required that the naphtha pretreater be We had an actual experience of sulfur breakthrough in
shutdown for several days to make repairs. In all, we the range of 2 to 3 ppm for a few days. We had a lot of

122
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

difficulty in removing the sulfur. The octane decreased, and range was typically 25 to 30 ºF. Operations at lower sever-
it remained below normal for a considerable time. In spite ity, typically 93 to 94 octane, gave end point appreciations
ofcontinued regenerations, we took samples and found that of 20 to 25 ºF
in fact the sulfur had changed to sulfate. We had to resort to
high purity hydrogen during the reduction phase at ele- J. G. DEARWATER (Profimatics, Inc.):
vated temperature of 480 °C for regeneration of the com- We see no significant difference in reformate end point
plete inventory of the three reactors before we regained for 95 and 100 RON. The difference in feed and reformate
activity. end point will vary between 20 and 35 ºF and is a function
of N and A content and the feed cut point.
TIEMAN:
I would like to point that the unit at Abu Dhabi National
Oil Company is not a UOP CCR Platforming unit. Process
MAJED INTABI (SAMAREC):
I would like to revert back to Question 28. In the case of QUESTION 30:
semi-regeneration units where a bimetallic catalyst is In the production of reformulated gasoline, how
used, what would be the maximum allowable sulfur level in are refiners adjusting their reformer operations and/
the feed before a sulfur guard bed would be required? or catalysts? What options are available to reduce the
benzene content of reformate?
TIEMAN:
For semi-regeneration operation, we also recommend TIEMAN:
less than 0.5 ppm sulfur in the feed. If you cannot maintain These issues of how to reduce benzene from the reformate
that, then we would recommend the addition of a sulfur have been discussed previously. I will just reiterate the
guard bed. options that there are. You can prefractionate the feed to
the reformer to eliminate benzene precursors from the feed,
CARON: and then take that light feed to an isomerization unit to
We have a semi-regenerative Magniformer, and we put a increase its octane and to saturate any native benzene that
sulfur trap in front of it. We used to run heavy hydro- is in the light cut.
crackate with 1 to 2 ppm sulfur. We put the sulfur trap in, If that is not sufficient to meet the requirements of
and now have feed sulfur below 0.5 ppm. We have seen a benzene, because of benzene produced in the reformer by
dramatic increase in our cycle length. dealkylation, then you will need to post-fractionate the
reformate to produce a light cut. With that light reformate,
UTLEY: you have basically three options. You can put it to an
We do not have a guard bed, but off of our hydrotreater, isomerization unit which will saturate all of the benzene
we generally operate around 0.1 to 0.2 ppm. adequately and isomerize any of the paraffins that are left.
You can put it to a Sulfolane extraction process to extract a
chemical grade of benzene if you have an outlet for chemical
QUESTION 29: benzene. Thirdly, you can alkylate it. We have a process
What are your actual operating experiences and called Alkymax for alkylating the benzene. This is alkyl-
correlations with 390-400 degrees F end point re- ating it basically with FCC olefins to produce mainly C9
former feed and the corresponding reformate end aromatics. This is acceptable from the standpoint that it
point at both 95 and 100 RON? keeps the benzene in the pool as an aromatic which gives
you good octane. However, if there are eventual restrictions
TIEMAN: on end point of the gasoline, then the viability of this option
The increase in end point going from feed to reformate on will depend on how severe those restrictions in end point
a reformer is going to depend partly on the sharpness of the become.
distillation of the feed to the reformer and on the reformer
operating conditions, particularly octane. For a normal full WILSON:
boiling range naphtha with a 390 to 400 °F end point from a I guess one of the obvious choices to lower aromatics
well operated crude unit, the reformate end point would be would be to decrease the reformer severity. I think that was
approximately 425 °F at 95 research octane and about shown in a discussion on the effect of octane number on
435 ºF at 100 octane. So we would estimate about a 10 ºF aromatics content. Given the fact that in reformulated
increase in end point for a 5 octane number increase. gasoline, there might very well be sufficient octane avail-
able from oxygenates, that is an option. However, you do
BRANHAM: need to consider the fact that in most refineries, the re-
Ashland’s normal target for reformer feed is 400 °F or former is the major hydrogen producer, and we are also
measured by D-86. We typically see around 445-45 ° EP on seeing increased hydrogen demands. So, obviously if you
the whole reformate. lower reformer severity, you will have to think about your
overall hydrogen balance.
CARON:
We routinely ran feedstocks with end points between 390 STEGELMAN:
and 400 °E End point appreciation in the 98 to 100 octane Since much of the reformer feed can be converted into

123
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

benzene one way or another, the only sure way of reducing reformer severity. The other aspect of the reformer is that
the benzene content of the reformate is to fractionate the we are seeing from the catalytic cracking side a push to
benzene out of the reformate product. Once separated, the move up Watson K factor on the feed. When you push the
benzene can be alkylated through commercial processes to Watson K factor up, you also lower the N+A of catalytic
ethyl benzene or cumene or saturated. The first two alter- reforming, this will impact reformer severity targets and
natives will hold the octane the benzene contributes and will also generate more butane.
will add significantly in volume to the gasoline pool. With
reformulated gasoline however, additional aromatics may GEORGE MANTHOS (ABB Lummus Crest Inc.):
find no place in the gasoline pool. With benzene saturation, Our commercially proven ArosatSM process can saturate
the octane drops below the level of the gasoline pool. The benzene and other aromatics without the level of ring
volume does increase with saturation, but the low octane opening and cracking which will take place in an isomer-
will put added pressure on octane to pull the stream back ization unit. It is also well known that the resulting C6
into the pool at the same time removal of the octane from naphthenes from benzene saturation can inhibit isomer-
the aromatics and olefins is being lost in reformulation. ization catalyst activity resulting in lower conversion or
Fortunately, at Borger we have the ability to fractionate higher catalyst volume requirements. Isomerization also
the benzene out of the feed and convert that portion of the has the negative impact of increasing gasoline pool RVP?
benzene into cyclohexane. Our CDTECH etherification technologies for the produc-
tion of MTBE or TAME can replace the lost octane from
PAINCHAUD: aromatic saturation or a lowering of reformer severity.
We agree with the options for benzene reduction de- With respect to the refinery hydrogen balance issue,
scribed by Mr. Tieman. In our particular case we remove when the butylenes for either MTBE or alkylate are
aromatics by extraction at one refinery but not at the other produced via dehydrogenation (such as with our Houdry
two. They are currently planning Isomerization capacity CATGFIN technology) significant hydrogen by-product is
with the benzene destruction option. produced offsetting losses resulting from the lowering of
As an alternative we are considering making a benzene reformer severity.
rich Reformate cut at these two refineries for processing at One final note, we too have a process for benzene alkyl-
our extraction unit in Houston. This will require major ation with the olefins in FCC offgas. It is called the CAT-
revamp of the unit for increased capacity involving equip- STILLS” process.
ment and a solvent change. UOP has briefly looked at the
unit and has estimated that by changing from Tetra- CHARLES MCCOY (McCoy Consultants):
ethylene Glycol to Carom we could increase our capacity by One way to reduce benzene in reformate is to get the
90% using the existing and solve all the hydraulic problems benzene precursors out of the feed by increasing the initial
that would arise with the increased rates. point on the heavy naphtha (reformer feed). The problem is
that this puts some of the lowest octane paraffins into the
KENNEDY: light straight-run, creating a significant octane problem at
The production of reformulated gasoline will probably that end.
require reducing the severity of reformers to minimize
aromatics production and to make up the resulting octane
shortage with a combination of high octane oxygenates and QUESTION 31:
isomerate. The reduction of benzene can be accomplished
What methods are used for disposal of ZnO chlo-
by a combination of reduced reformer severity and the use
ride scrubber catalyst?
of an isomerization unit to saturate benzene and remove the
benzene precursors from the reformate feed.
It should be noted that this operation may result in some UTLEY:
refiners reaching a hydrogen deficiency status. Zinc oxide chloride scrubber catalyst can be handled
through some of the spent catalyst reclamation companies
JOHN LYNCH (Petrogistics Inc.): in business in the U. S. today. Care must be taken in
I would add that besides saturation, alkylation, and handling the catalyst because of the chlorides present, and
Sulfolane extraction, there is going to be a disparity be- shipment should be made in corrosion resistant liners or
tween refineries with regards to the benzene concentration corrosion resistant containers.
in gasoline. From the standpoint of light reformate, re-
fineries that already have in place Sulfolane extraction BRANHAM:
units will be able to trade/purchase light reformate from We had one occasion where we used an old zinc oxide
other refineries, and that way maximize the overall octane material as a chloride scavenger on net reformer hydrogen.
of the U.S. gasoline pool by keeping severities up on the The spent material was determined to be non-hazardous by
reformer. the EP toxicity test, and it was landfilled at an approved
The other issue around the reforming side, and this may non-hazardous landfill site.
not seem believable right now, is that around 1995 or 1996,
not only will hydrogen be a factor, but butane availability
will also be a factor. The normal and isobutane generation QUESTION 32:
from catalytic reforming is going to be a vital issue in What has been the experience with integration of a
determining where you are going to ultimately set the light naphtha isomerization train into an existing

124
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

semi-regenerative reformer? What is the extent of CHARLES MCCOY (McCoy Consultants):

modifications required? If you ever get a leak in that reformer-effluent to
isomerization-feed exchanger, are you not going to have the
UTLEY: world’s biggest runaway?
We have successfully integrated a zeolite-based catalyst
C5-C6 isomerization unit into our semi-regenerative re- UTLEY:
former. We started the complex up in February, 1988. As a We looked at that, and the pressures were such that it
little background, prior to the modification, our reformer would go the other way.
consisted of four reactors. At the No. 1 position we had’a
single downflow reactor. At the No. 2 position we had MAJED INTABI (SAMAREC):
parallel downflow reactors. At the No. 3 position we had a I would like to reverse the question, and ask about the
single radial flow reactor. Parallel reactors at the No. 2 experience in converting from a semi-regenerative re-
position presented some operating and regeneration prob- former to an isomerization unit, ie., the modification re-
lems that we were uncomfortable with. With the avail- quired and the feasibility of this conversion.
ability of more active reforming catalyst, we could remove
one of these reactors from reforming service, thus reducing WILSON:
the amount of catalyst in the reformer and yet maintain the It is certainly feasible. As far as the extent of the modi-
same charge rate as before. fications is concerned, there are probably several licensors
We were hydrotreating the full-range naphtha, making a that can discuss this in some detail.
split downstream of the hydrotreater. We thought we could
pick up 7 or 8 octane numbers in the light straight-run by BRIAN JOHNSON (UOP):
isomerization. Considering these facts, we decided to go Over the last five or six years, during the big increase in
ahead with the isomerization unit, recognizing that if the isomerization activity, we converted semi-regenerative
two were closely integrated, we could minimize capital Platformers to both zeolitic isomerization units and chlo-
expenditures, but we would be giving up some things. rided alumina Penex units. The conversion generally is
The actual modifications required were very simple, and very simple. The stabilizer usually is of adequate size. The
I will highlight the major areas. The reactor is a carbon reactors for catalytic reformers generally operate in the 1 to
steel cold wall reactor. It required no modifications, other 2 space velocity range, which is compatible with isomer-
than a change to the inlet distributor. We added charge ization unit space velocities. Recycle gas compressors are
pumps to take suction from a naphtha-splitter overhead also compatible, if not oversized. For the Penex process, the
receiver. Some care was taken there to ensure water would compressor can even be eliminated by using UOP’s once
not get into the system. Hydrogen was taken directly from through hydrogen technology.
the recycle compressor discharge. We added heat exchange I would say that if not the majority, a very significant
between the isomerization reactor effluent and the isomer- portion of the isomerization business has been revamps of
ization reactor feed. catalytic reformers.
We added another heat exchanger to serve as a final feed
heater. This piece of equipment uses heat from the reformer
reactor effluent and exchanges with the isomerization reac- QUESTION 33:
tor feed for final temperature control. The effluent from the What has been the maximum number of re-
isomerization reactor is commingled with the reformer generation cycles on reforming catalyst? What dic-
reactor effluent in the reformer exchange train where tem- tated catalyst replacement?
peratures and hydraulic loadings were comparable. The
combined eftluents then go to the reformer flash drum and MORGAN:
finally to the reformer stabilizer. All in all, we have been Our semi-regeneration units vary from 10 to 20 cycles.
pretty happy with the operation. The unit has performed To put these numbers in a different perspective, catalyst
well. We have been on-line for 2.5 years now without life averages between 10 and 15 years, although the first
regeneration. The first 18 months we saw some decline in reactor catalyst life usually is shorter. This assumes the
conversion, but the last year we have held steady, about 59 catalyst is dumped and screened on a periodic basis in order
or 60% normal pentane to isopentane conversion. to maintain good flow distribution.
As I mentioned earlier, we realized we were giving up The catalyst is changed out normally on low surface area
some things, so if you are thinking about doing this, these and activity, but noble metals migration and permanent
are some of the things we considered. Monitoring the poison levels have also dictated catalyst change out. On our
reformer operation becomes more difficult. The octane continuous regeneration unit we have achieved 170 cycles
number is now a blend of the two streams. Although the before catalyst replacement. It was changed out due to low
isomerization octane is pretty constant, it is still more surface area and low activity.
difficult to figure out which one is not performing if you do
have problems. Along the same line, the isomerization unit BRANHAM:
adds another source of potential contamination to the re- At our Catlettsburg complex, we now have a load of R-50
former in the form of water, sulfur, or arsenic. catalyst that is on its 10th cycle. It was loaded in November
Some gasoline blending flexibility is taken away with the 1980, and has a total life of 355 barrels of feed per pound of
integration of the two units, and where lead is still added to catalyst. Based on the present activity and stability, this
the gasoline, there is some debit associated with less than catalyst continues to perform at near-new conditions and
optimal use of light straight-run in the leaded blends. the 11th cycle will start before the end of this year.

125
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

that the new catalyst would provide improved yields and

greater stability at higher octane requirements.

MACKEY:
On our continuous regeneration unit we typically find
that the yield losses associated with surface area declines
would justify catalyst replacement when we get down into
the 120 to 130 meters squared per gram of active catalyst
surface area. This typically represents about a 1 to 1.5%
yield loss, by our figures, at our severity. That provides
enough economic justification to replace the catalyst. In
recent years that has been on the order of 100 to 120 cycles
on our CCR unit.

RATERMAN:
In the case of our semi-regenerative units we have had up
to I5 regenerations on a single catalyst. We usually make a
decision to replace the catalyst based on the availability of
At the St. Paul Park refinery, we obtained 445 bbl/lb on either an improved catalyst or the cycle time between
five cycles on UOP R-16G. We then revamped the unit and regenerations becomes too short and uneconomical.
added fresh catalyst to increase the weight by 30%. We then
operated for an additional six cycles, with a total life in- TIEMAN:
cluding the new catalyst of 349.06 barrels per pound. These As Mr. Mackey said, the typical replacement of a CCR
six cycles occurred over a span of five years. This R-16G catalyst will be due to loss of product yield, and in some
catalyst was replaced because the activity and stability had cases activity, when the surface area gets low enough that
deteriorated such that heater limits were reached and cycle it will not maintain chloride balance. That range is broad
length was unacceptably short. This degradation occurred with different CCR units, depending on how severe the
over time and did not appear to be caused by any specific regeneration is and upsets in the regenerator. The question
event. specifically asks about the maximum number of re-
In both cases, comparison of liquid volume yield from 1st generation cycles that we have seen which has been in the
cycle to last did not show a significant deterioration but range of 260 to 285 on a CCR unit.
time between runs was becoming more critical as overall
octane demand was increasing and it was judged better to UTLEY:
go to new catalyst. I agree with what has been said. On a semi-regeneration
unit, we think 10 to 12 regenerations are very good. The
GESICK: things we look at when contemplating a catalyst replace-
The catalyst in our Meraux refinery CCR unit was in ment are the metals and low surface area. If the surface
service for over 10 years, but it had seen less than 100 cycles area is 130 meters squared per gram, we would consider
when we entered the 1990 turnaround. The actual number changing out the catalyst.
was somewhere between 80 and 90 cycles. The catalyst We had a crack in our Unitiner reactor liner, and we
showed good platinum dispersion, no significant metal killed the catalyst in our No. 1 reformer reactor. We had an
contamination, no gamma phase alumina structure, and arsenic level of about 1,100 ppm on the catalyst. That is too
reasonable surface area at about 74% of original surface much.
area. Since we had seen no significant decline in the surface
area for the last two years, and the offset in activity had PAINCHAUD:
remained constant, we elected not to change the catalyst Our last run on our semi-regenerative unit went through
during turnaround. 16 cycles. We used UOP R-50 and regenerated approxi-
Unfortunately, on start-up the moderate surface area mately every six months with an end of run temperature of
catalyst showed a lower coke tolerance than we had antici- 975 ºF. We changed to UOP R-62 in 1988 for economic
pated, and we had regenerator problems that kept us from reasons.
circulating catalyst for several days, and when we got going
the catalyst regenerator at 220 pounds per hour was not MOHAMMED AL-NAGHASH (SAMAREC):
able to catch up on the coke. So we elected to shut down and I have two questions regarding semi-regenerative Plat-
change the catalyst out. The point of this is that as the formers. First, we have experienced a strange phenomenon
surface area goes down, the coke tolerance goes down, and during the once-through wash step of the regeneration.
you may need to change catalyst sooner than you think. Before the carbon burn, we saw a sudden rise in the tem-
perature across the reactors. They went up more than 30 °C
KENNEDY: and dropped back quickly without affecting anything. I
We regenerated a load of bimetallic reformer catalyst 12 would like to ask the panel or the audience, has anybody
times over a period of about 12 years. We changed after our experienced this? If so, what is their comment about it?
run length declined to an unacceptable level, and the attri- Secondly, is the decision to regenerate the semi-
tion of the catalyst spheres became excessive. We also felt regeneration Platformer catalyst based on the end of run

126
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

temperature, or the C5 + yield loss? If the latter, what is the STEGELMAN:

percentage? We use methanol mixed with naphtha and chloride to
allow injection with a single pump. Water injection is
R. W. BENNETT (UOP): typically done with a separate pump. We have had much
As far as the first question, I am not aware of any similar better control of the chloride water balance using meth-
instances, where a temperature spike goes through each of anol. However, we are considering the use of isopropanol
the reactors when you first put air in. before hot weather arrives next summer because it appears
In reply to the second question, I believe generally the we have been losing some of our methanol from our day pot.
consideration is yield decline over the cycle. We all have
anticipated temperature ranges that we expect that yield UTLEY:
decline to occur over, but the bottom line is yield decline. There is no big advantage or disadvantage for using
But then to translate this into the refinery reality, we water injection versus methanol. We use methanol because
generally look at yield decline within a fixed turnaround it is cleaner and mixes well with the chloriding agent we
period to do the regeneration. use. We inject the methanol and chloride via one pump thus
minimizing equipment use. We have some concern about
UTLEY: the cleanliness of water, i.e., the contaminants and min-
In our case it is an economic decision based on loss. And erals that are in the water. I have been told of a couple of
based on today’s crude prices, I think we are going to be occasions where water injection probes have come com-
regenerating much more often. pletely off with mineral deposits. The chloriding agent that
we use and water do not mix well. So separate injection
pumps would be required, doubling the maintenance prob-
QUESTION 34: lems. The obvious advantage of water is that it is not toxic,
In order to control the water/chloride balance in whereas methanol is.
our reformer (Pt/Re catalyst), we are considering
moisture injection. What are the advantages and
disadvantages of straight water injection versus QUESTION 35:
methanol injection? Prior to the reactivation of a skewed Pt/Re catalyst,
we performed a routine sulfur-scrub step using a
KENNEDY: 68% hydrogen stream from another reformer. After
We have examined both methods, and at present we four hours of circulation, analysis of the gas indicated
inject isopropyl alcohol. Water injection has the advantage only 6.3% H2 and 74.3% methane. After hours of a
of simplified handling by operators, but it may be difficult slowly increasing H2 concentration, a decision was
to control the rate on small units. The advantage of alcohol made to abandon the scrub to avoid possible coking
injection is that it is easier to control, since it can be blended of the catalyst. Should electrolytic H2 be used in this
with a naphtha and the chloride component and injected step to avoid any potential for hydrocracking and
using one system. subsequent coke deposition? (TCE was injected dur-
ing this step in an effort to maintain catalyst chloride
BRANHAM: levels.)
We use both water and methanol at different locations
with equal success. The only advantages and dis- KENNEDY:
advantages that we see are the rather obvious ones. Water On the basis that the question is referring to a sulfate
can easily freeze in cold weather and as with all systems removal step which occurs before a final proof burn and
using small diameter pipe or tubing, special care must be oxidation step, we routinely use electrolytic hydrogen. It is,
taken to keep it warm. Methanol has the normal flam- however, according to our catalyst supplier, acceptable to
mability and safety handling problems. Again, we consider use catalytic reformer hydrogen, if the hydrogen purity is
both to be equally acceptable. We have also used isopropyl greater than 75%. We feel that any coke laydown which
alcohol in the past. could occur from using reformer hydrogen will be removed
during the following proof burn.
CARON:
In addition to what Mr. Kennedy said, the advantages of UTLEY:
water over methanol are that it does not cost as much. The I agree with what Mr. Kennedy said. I assumed that the
disadvantages that we see, particularly in our Canadian author of this question meant after the carbon burn step
refinery is that if you use water, you have freeze protection and before the oxidation step, he performed a sulfate re-
requirements. moval step. In doing the sulfate removal step he used a
source of hydrogen that was 68% hydrogen.
PAINCHAUD: A likely cause of the high methane production and low
We agree that methanol and water give you equal per- hydrogen content of the recycle gas was caused by de-
formance in the unit, and the methanol’s advantage is that methylation of heavy hydrocarbons contained in the
it is much easier to control, since you can dilute it with hydrogen source. The high temperatures used for sulfate
hydrocarbons to increase your pumping rate. In our case, removal (950 ºF) would be favorable for demethylation to
we blend both chloride and methanol with hydrocarbons in occur. I do not believe hydrocracking was occurring because
a day pot and use a single injection pump. If any changes in of the high methane content. I believe hydrocracking would
the formulation are necessary, they are done in the day pot. produce hydrocarbon products larger than methane. It

127
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

would be interesting to know what hydrocarbons composed KENNETH N. ABRAHAMS (Star Enterprise):
the remaining 32% of the hydrogen stream. If it was meth- At the prompting of our industrial hygienists, we are
ane or ethane, this phenomena is difficult to explain. A low evaluating alternatives to TCE use for chloride injection.
recycle rate and high temperature, say greater than Has anyone else looked at this? What were your
1,000 °F, would also contribute to demethylation. conclusions?
If this occurred as it is written before the carbon burn
step, it is very difficult to explain because demethylation is R. W. BENNETT (UOP):
a metal function and the metal sites on the catalyst would There are a number of people that are starting to look at
be attenuated by coke on the catalyst. There are a couple of alternatives to l-l-l TCE, as this material seems to be
things I would recommend to avoid problems in this situ- coming under environmental attack as not acceptable. We
ation. Make sure the system is purged well, hydrocarbons are working with the industry to try and find acceptable
in the system under sulfur strip conditions may cause chloride compounds. But the number is diminishing
problems, keep the temperatures under control, under rapidly.
1,000 ºF, maintain a high recycle rate to keep things mov-
ing out of the system and yes we do use electrolytic hydro- RALPH ALM (Husky Oil Operations Ltd.):
gen in this step to avoid problems. Has anyone experienced iron poisoning of their bimetal-
lic catalyst during catalyst regeneration, apparently due to
STANLEY GUSSOW (Engelhard Corporation): iron sulfide migration from the heater tubes? The unit is 23
We have never heard of an experience like this. We find it years old and in high-severity service.
most interesting. We tried to develop a material balance
and could only balance the carbon and hydrogen with the TIEMAN:
addition of naphtha. This implies that there were pockets of We have had occasions when iron sulfide scale has flaked
naphtha around when the conversion was taking place. off of heater tubes and migrated into Platformer reactors.
However, the heat balance calculation that goes along with My only experience with it was that it was more of a
the consumption of hydrogen gives us an adiabatic tem- physical contamination than chemical. It tended to foul the
perature rise of approximately 2,000 degrees F. bed and to plug up center pipes, and that sort of thing. But
I doubt that this temperature was observed, and there- normally the iron does not chemically contaminate the
fore we conclude that something other than hydrocracking catalyst. So removal of the plugging material and maybe
is taking place. But we have no explanation at this time. the first part of the catalyst bed typically will solve the
problem.
JAMES KUMPEL (Texaco Refining & Marketing, Inc.):
We also have observed a phenomenon such as this. We MORGAN:
attribute it to degradation of the TCE injected during this We have experienced that exact same problem in one of
step, as the predominant products were methane and a our units. The problem arose after several runs that were
slight bit of ethane. plagued with sulfur contamination. The iron migration
One additional question on this subject of skewed metals ultimately caused plugging in the first bed. After the cata-
catalyst, can the members of the panel or the audience lyst was replaced, less severe migration persisted for sev-
comment on the necessity or the experiences with periodic eral cycles, After about three regenerations the iron level
acidizing of reformer charge heater tubes to prevent sulfur was high enough to affect the catalyst activity in the first
migration during the carbon burn and oxidation steps in bed. Eventually the sulfur was purged from the unit and
the regeneration? the iron migration problem disappeared.

ABDULLAH AL-DERAIBI (SAMAREC): RALPH ALM (Husky Oil Operations Ltd.):

In our refinery we have two Platformer units and during Was that a bimetallic skewed catalyst?
the regeneration we use the Platformer off-gas in the sul-
fate removal step and the reduction step. But we have had MORGAN:
the same problem, cracking the light ends to 50% hydrogen It is bimetallic but the metals are not skewed.
and 50% methane. During the last regeneration, we used
the hydrogen plant product during the sulfate removal CHARLES MCCOY (McCoy Consultants):
step, and the CO/CO2 was nil. We still had, during the I am not aware of iron being a poison, whether it is
reduction step, the phenomenon of cracking the light ends. bimetallic catalyst or not, but I would remind you that iron
What will be the effect of having the reduction step at low catalyzes the demethylation reactions. The reaction can
hydrogen partial pressure, and is there any significant lead to hot reactors and serious safety problems.
effect of coke laydown alkylate on the catalyst?
J. B. RODDEY (Roddey Engineering Services):
R. W. BENNETT (UOP): I have never seen catalyst unloaded from the first reactor
Reducing the hydrogen partial pressure will definitely of any reformer that did not show iron oxide when the
reduce the cracking reactions; however, there is a mini- catalyst was unloaded after the carbon burn. Usually the
mum hydrogen partial pressure required to get a good catalyst will be a reddish brown color. All reformers have
reduction. We can possibly balance the two to minimize iron migration from corrosion products in upstream
cracking that seems to be occurring on the chloriding equipment. As the iron content goes up in the catalyst, in
compound. the scallops, in the feed effluent exchangers and in the

128
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

heaters, the regeneration becomes more difficult in a fixed- in hydrotreater or hydrocracker reactors. They operate in
bed unit. You will find that as the iron content goes up in the creep range, and the mechanism is creep em-
the first reactor to 2%, regeneration will get very difficult. brittlement. It happens; you have a 50-50 chance probably
of seeing cracking eventually.
Mr. Gesick’s example points out that there are differ-
Mechanical ences from plant to plant. Maybe, as Mr. Raterman said, it
is differences in fabrication. Still, inspection of those points
of high stress, particularly nozzles and support rings or
QUESTION 36: support lugs, on a regular basis is critical. There have been
Is there anything new on age-related cracking of
a couple of nasty fires.
catalytic reformer reactors?
Most of these cracks are not through cracks. They
usually open up wide at the surface and do not go very deep.
RATERMAN: If caught early by regular inspection, they can be repaired
There is really nothing new that we know of in this area. and the reactor returned to service without incident.
What is happening is that we have become more aware of
the poor original welding and inspection of the nozzle QUESTION 37:
attachments that lead to premature failure. As the result
Is anyone using synthetic lubricants on recycle gas
we are now closely inspecting new reactors during the
centrifugal compressors? What benefits have been
fabrication process.
achieved?
GESICK: SCHAUB:
We are not aware of anything new either, but we had two We do not have any direct experience with the use of
essentially identical reactor systems. The Superior system synthetic lubricants in recycle gas compressor service.
and the Meraux system were both built as stacked reactors From what I can see there have been some mixed experi-
built for continuous regeneration services. They were built ences out there in different applications. Some of the excite-
in 1972 by the same shop with the same metallurgy 1¼
ment over these products has begun to diminish. A few
chrome, ½ molybdenum. They went through the same heat years ago a limited survey indicated that at least 25 ma-
treating procedure. You can take the heat treating charts chines were using synthetic lubricants. I refer you to the
and lay them side by side and they look like carbon copies of 1987 NPRA Q&A transcript, page 83. There was discussion
each other. of benefits and problems.
The Superior unit is operated as a semi-regenerative
unit, and the Meraux unit is operated as a CCR. We have JOHN NEIGHBORS (Coastal Chemical Company):
had two occasions where we have had extensive cracking in We are supplying a Union Carbide U-Con product for
the Superior system, and we have not yet had any cracking lubricants for reformer compressors, both centrifugal and
in the Meraux system. Basically, the difference is that the reciprocating. We have had extreme good luck with the
Superior system has gone through several regeneration product. Several refiners are using it in centrifugals. The
cycles, shut down, cool off, and heat up, and the Meraux primary advantage to this product is that it is hydrogen
system has been in steady operation. So the cracking ap- insoluble product, so it is easy to separate. Apparently it
pears to be caused by the cool down and heat up during the has no coke-forming problems, and we believe it will in-
regeneration cycles. crease catalyst life.
KENNEDY: MOHAMMED AL-NAGHASH (SAMAREC):
We have two catalytic reformers, each with three reac- My question concerns using a dry gas seal in a process
tors. One of them is 21 years old, and it is downflow. The compressor. I would like to know the panel or the audi-
other is 15 years old, and it is a radial flow reformer. We ence’s experience in using the dry gas seal in Platformers in
have in the last two years inspected reactors in both units particular.
using angle beam ultrasonic inspection and wet fluorescent
magnetic particle. We did not find any cracks. The metal- TIEMAN:
lurgy on those reactors is 1¼% chrome and ½ molybdenum. I do not think that any of our licensees have tried that
type of seal in Platforming unit service.
R. W. BENNETT (UOP):
For the transcript, the API Operating Practices Com- E. J. SPARROW (Caltex Petroleum Corporation):
mittee has issued a report that is authored by J. Cantwell of Our Philippines reformer has had a dry gas seal working
Caltex, entitled, “Cracking Experience with Chrome, Moly on the compressor for over a year very successfully.
Equipment,” which gives a pretty good overview of what
you ought to be looking at in a reformer reactor. That was BASHIR HASSAN ALI (Abu Dhabi National Oil
presented in May of 1989. Company):
Concerning a comment by Mr. Utley about poison on the
CHARLES MCCOY (McCoy Consultants): catalyst, what is the practice among refiners to prevent this
This subject of reactor cracking has been dear to my from happening or on changing hydrodesulfurization cata-
heart for years. Let us remind people that the cracking lyst before it gets saturated considering the runs between
phenomenon in hot wall reformer reactors is different than turnarounds are quite lengthy?

129
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

UTLEY: seen a lot higher recovery of our benzene and toluene. I still
In our case it was not the fault of the catalyst. We had a think some research needs to be done to handle possible
mechanical problem in the reactor, and we bypassed the upsets in our stripper column, which results in some envi-
feed around the catalyst bed. Are you going 5 years be- ronmental concerns in sending glycol in the water wash
tween dumping and screenings? That is a good question. streams out to the sewer, and also possible increased acid-
One thing that we have found, if you start seeing the ity in our rafIinate stream, which we have seen because of
sulfur break through or some loss in performance in your the new Carom solvent.
reactor, then there is a good chance you have some high
metal loadings on your catalyst, and you should try to get a
sample of your catalyst as soon as you can. QUESTION 39:
In the solvent recovery column overhead ejector
TOM KELLY (ARCO Products Company): system in a Sulfolane extraction unit, we experience
A good way of keeping sulfur out of your unit or iron carry-over of benzene in the ejector effluent receiver.
sulfides off your catalyst, is through the use of a sulfur trap. What is the experience of other refineries?
If you keep sulfur out of the reformer, you will greatly
reduce iron sulfide contamination of the catalyst. BRANHAM:
During normal operations we are able to maintain recov-
ery column vacuum with the vacuum jets off-line. During
D. Aromatics Extraction start-up, we use the vacuum jets, but do not have any
problem with benzene carryover.

QUESTION 38: SCHAUB:

We currently have an old Udex aromatic extrac- This is what you call the nature of the beast. Operation of
tion unit that is no longer operating. Has anyone had the recovery column ejector will educt benzene vapor from
success in converting these units into a more ef- the recovery column receiver. Vapor pressure of the benzene
ficient, cost-effective process? is about 3 to 5 psia at receiver conditions, and it is really
governing the operating pressure of the recovery column.
BRANHAM: The ejector effluent receiver must be designed to decant a
Ashland converted two Udex units to Shell Sulfolane benzene level from the steam condensate within an en-
units in the late sixties. I will defer going into a discussion closed system. But I wonder if the question refers to a
about the comparison between those two particular sol- continuous operation. Also in my experience it is not neces-
vents, but the big addition to the unit was a solvent recov- sary to operate the ejector continuously to maintain the
ery column which operates under vacuum. column vacuum. If that is the case, you should start looking
for air leaks or a condenser performance problem.
SCHAUB:
A few years ago we did detailed engineering on a con-
version for one of our clients who converted a Udex to the QUESTION 40:
Union Carbide Carom process. Conversion provided for What recent developments have been made in
increased efficiency by virtue of the superior selectivity of liquid-liquid extraction towers for better per-
the solvent. They were able to run much lower solvent formance? What is the experience with structured
feed/ratios compared to the Udex. There were no major packings for liquid-liquid extraction?
equipment changes. You are talking mainly piping, pump,
and heat exchange revamp. BRIAN GANYU (UOP):
We would like to respond to the first half of Question 40,
TIEMAN: since we cannot comment on structured packings. The UOP
You are right, Mr. Schaub, only it is UOP Carom process Multiple Upcomer Tray can significantly increase the ca-
now. The UOP Carom aromatics extraction process flow pacity of existing liquid-liquid extractors by direct replace-
and equipment is similar to that of Udex and Tetra units, ment of the trays. In new units, MU Trays can reduce
however, Carom utilizes a superior solvent system that diameter relative to conventional extraction trays.
consists of TTEG and a proprietary co-solvent. It is, there- The MU Tray maximizes the cross-sectional area avail-
fore, ideally suited for Udex or Tetra revamps, and as able for mass transfer and is most often applied in systems
Mr. Schaub said, it can usually be done with minor where the heavy phase has the higher volumetric flow rate.
modifications. Typical applications include extraction processes for ar-
Depending on the extent of the revamp, a Carom process omatics or lube oils. Capacity increases in excess of 30% are
conversion can result in a feed capacity increase ranging possible in a revamp.
from as little as 15% to as high as 90% of the original Fifteen years of successful operating experience have
design, while at the same time reducing energy con- been logged, both in Tetra and Carom aromatics extraction
sumption by 20 to 30% on a BTU/pound extract basis. units. More recently, MU trays have been used in the
retrofit of Sulfolane units.
CHARLES KEMP (Koch Refining Company):
We have one of the new Carom units, and it is true that OSCAR ALDAMA (Coastal Refining & Marketing Inc.):
the new Carom unit has a lot higher selectivity. We have Last year we replaced trays in the extractor column in

130
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

one of our Sulfolane units with rain deck type trays. We it from the wording of this question that it is not. H2S
have been able to produce raffinate with less than 1% physically adsorbs on the alumina surface much the same
aromatics and very little benzene, which is substantially way water does and as such, can be regenerated. Alumina
better than what you are able to achieve with a rotary disk has a low capacity of adsorption for COS, somewhere
type extractor column. around 0.5 wt %, and H2S somewhere around 1 - 2 wt %
depending on the inlet concentration of H2S.
DENNIS BIELINSKI (Glitsch, Inc.): The presence of water will reduce the capacity even
Structured packings are generally less favorable in further because water will absorb preferentially over H2S
liquid-liquid extraction systems due to their high surface If the alumina can be regenerated, and there is a place for
area characteristics. This high surface area tends to coa- the stripped sulfur to go, then it is probably a good service
lesce the droplets into a surface film thereby reducing the for the alumina. If there are not facilities for regeneration,
interfacial area. The structured packing does have an ad- then alumina should probably not be used in this service.
vantage in its ability to minimize backmixing however this In answer to the last part of the question, alumina is a
can also be accomplished more cost effectively with random very good hydrolysis catalyst. If there is water present, and
packing. there probably is because the hydration requirements of
the polymerization catalyst, the alumina will catalyze the
reaction of COS and water to give H2S and CO2. So it is
E. Polymerization quite likely that the COS present will convert to H2S, no
matter what the age of the alumina is.

QUESTION 41:
BRANHAM:
Activated alumina is an excellent COS hydrolysis cata-
Background: The following problems were experi- lyst and is used commercially by Ashland in treating C3
enced on our EADC (Ethyl-Aluminum-Di-Chloride) and C4, for this purpose. Unless the bed has been base
line — Dimersol Catalyst. poisoned, it should be expected to function as a hydrolysis
a) Plugging of vent seal pot line to the caustic catalyst from start-up. Pelleted KOH is a good alternative
washing drum. This line is provided for to activated alumina when only H2S or elemental sulfur
evacuation of the vent seal pot materials to removal is required. Activated carbon has also been shown
the caustic neutralization drum before it is to be effective at lower capacity. We have no data on COS
pumped out. removal for either material.
b) Plugging of overhead 6” line from the vent The first step to evaluate the problem is to identify if H2S
seal pot to the open pit. This line has no is the sulfur species that is the source of the corrosiveness.
flushing facilities attached to it. It is possible to show corrosiveness due to the present of
Have any refiners experienced blockage in their elemental sulfur which is more effectively removed with
EADC lines and what measures have been taken to activated carbon. At Ashland we follow all alumina beds
rectify the problem? Also, we would like to know the with solid KOH scrubbers as guard beds. Therefore we have
type of pumps the refiners are using in EADC han- not attempted to judge the life of the alumina, but instead
dling. Presently, we have a Nikiso reciprocating monitor the KOH.
pump with erratic performance.
DAVID L. SMITH (ALCOA):
HIGGINS: I will elaborate on those last two comments. Indeed
No answer. alumina readily catalyzes the COS reaction, but a route
that a number of folks have gone to in both refining and
petrochemical applications on C3 and C4 streams is to use a
QUESTION 42: selective adsorbent that will both dehydrate the stream as
We are currently using activated alumina to treat well as chemisorb the COS and H2S and any CO2 if present.
propane from our polymerization unit. Every 4 to 5 You do not get a competition between the water and the
months, our propane goes corrosive and we change sulfur compounds. The water is physi-sorbed, whereas the
out the alumina. Are there other desiccants that will sulfur compounds, COS, H2S, CS2, are chemisorbed. So this
last longer? Is there a good way to judge the life of the eliminates the need for a KOH treater.
alumina? Is it possible that the aged desiccant is
allowing COS to convert to H2S or is something else
causing it to go corrosive? QUESTION 43:
last year several participants reported that polym-
UTLEY: erization catalyst could be sold in the dry state to a
I have not used alumina to treat propane from polymer- fertilizer manufacturer. What specifications are re-
ization units, but have used KOH with very few problems. quired on the spent catalyst? How is the catalyst
It seems the KOH treaters run about one year before a shipped?
changeout is needed. The KOH does a good job in removing
H2S. A caustic wash and water wash ahead of the polymer- MORGAN:
ization unit should also reduce problems with COS and We are currently selling our Denver refinery spent
H2S. Alumina in this service should be regenerated. I take polymerization catalyst to a fertilizer manufacturer. The

131
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

cost is 5 to 6 times less than disposing of the material as a

hazardous waste. The main specification is that the cata-
lyst is dry and we ship the catalyst in drums.

BRANHAM:
We have moved some cumene catalyst to an outside
purchaser. The specifications that are generally used are
that the catalyst must be dry and free of volatiles. There are
usually some restrictions on the size of the support bed
material that is allowed to be included. It may be shipped by
truck, rolloff hopper, or super sacks, but not in drums. The
price is based on the percentage of phosphorus and ranges
from a small fee to a small return. There may be a fee for
emptying containers if not shipped in bulk.

PAINCHAUD:
We have also contracted with a fertilizer manufacturer
who prefers bulk shipments, and has a $7/ton surcharge for
shipment in super sacks. We are not presently disposing of ization catalyst to a dry-drilling process. The dry-drilling
our catalyst in this manner. process is faster and eliminates the problem with low pH
water disposal and equipment corrosion. In the dry-drilling
UTLEY: process the catalyst is removed by an eductor system that
I agree with what has been said. I have some friends back discharges directly into a super sack.
in Roosevelt, Utah, at a refinery where I used to work. They They are drilling tubular reactors that have 2 inch tubes
tell me the only specification is that it must be dry. They and are 30 feet long with an Air Tools model 3162, 1 & ½
ship their material in super sacks in a covered truck to inch air motor, with a 1 & 3/4 inch drill bit, exhausting into a
prevent it from getting wet. vacuum system, then dumping directly into super sacks.
After all drilling is completed, the tubes are brushed using
the same air motors with 1 & 3/4 inch wire brushes that are
QUESTION 44: attached. Their turnaround time is about 20 hours from oil
out to oil in.
What methods are used to remove polymerization
and/or cumene unit catalysts from the reactor? Does
BENNY ALLEN (Star Enterprise):
anyone know of firms who offer disposal services
Just a quick question to the three gentlemen that an-
which include actual removal from the reactor?
swered the question. Do you have chambered or tubular
reactors?
BRANHAM:
We have tried a number of removal methods. The sure BRANHAM:
method is to use a waterblast, but this does present a My comments were directed only at chambered bed cu-
number of waste and water disposal problems. Explosive mene reactors.
decompression of a dry purged reactor is often successful for
removing the majority of the catalyst, but some may still
cling to the walls. We have used “Mole Master Company”
and their proprietary air drilling-hammering rig to remove F. lsomerization
large boulders of catalyst and catalyst that has clung to the
reactor’s walls. They do not provide disposal service.
QUESTION 45:
MORGAN: Please comment on special precautions taken in
We steam decompression dump our chamber reactors the shipment and reactor loading of isomerization
into a dust recovery tent. After it is dumped, the tent is catalysts. Specifically, what types of shipping con-
lifted up and the catalyst is put into barrels. This method tainers are required, and what precautions are taken
has all but eliminated dust problems. Our Humber, Eng- during catalyst loading? What techniquesare utilized
land, refinery has tried using a contractor, but found them in handling spent isomerization catalyst?
to be slow and very expensive.
TIEMAN:
PAINCHAUD: I assume this question mainly refers to the chlorided
Our Louisiana refinery drills theirs out dry. As it is being alumina type catalyst, I-8. The UOP I-8 catalyst is shipped
drilled, it is vacuumed into a 20 ton bin with a vacuum in 55 gallon drums which are sealed in double plastic bags
truck. They do not know of a service that will both remove to protect the catalyst from moisture during shipping as
and dispose of dry catalyst. well as during loading. During loading the catalyst is
loaded directly from the drum into the reactor with a
UTLEY: specially designed hopper which avoids contamination by
The people back in Roosevelt tell me they have switched moisture. The catalyst is very sensitive to water so we try to
from a wet-drilling process to remove the spent polymer- keep it away from the humid atmosphere.

132
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

Spent catalyst handling, I think, is not too difficult. We modified activated alumina that absorbs oxygenates and
recommend that the catalyst be dumped into drums which some sulfur. The last bed contains 13-X molecular sieve for
are plastic lined and dry nitrogen purged. If you put the sulfur and trace water removal.
catalyst into a drum with water or with moisture, it will The activated alumina was selected because it has the
corrode the drum. That is the only special handling on same capacity for oxygenates as silica gel and also has some
unloading. capacity to remove sulfur. In our case, additional sulfur
removal was important. If we were not limited on dryer
UTLEY: vessel size, we could have met the same water, oxygenates,
I know this question was probably meant for I-8 catalyst, and sulfur removal objectives with additional molecular
but I thought a comment on the I-7 catalyst might be sieve and silica gel. The oxygenates are monitored weekly
prudent. The zeolite catalyst is shipped in 55 gallon metal by GC analysis.
drums. The drums should not be opened or exposed to the
atmosphere until time for placement into the reactor. The RATERMAN:
zeolite catalyst is a little softer than some of the other The alumina vendors offer specialty adsorbent for oxy-
catalysts that we handle, so care must be taken while genate and sulfur compound removal. One example is
loading the reactor. A catalyst hopper and sock is used to ALCOA’s line of SelectSorb adsorbents. A layer of these
minimize the distance the catalyst is dropped when loading adsorbent can be easily incorporated into the beds of the
into the reactor. Between dumping the drums, the sock feed dryers. We have done this on the liquid feed dryers of
should remain full to minimize breakage when unloading several of our isomerization units. However, we have not
the subsequent drum. No special handling techniques tested these units as of yet to determine how effective this
other than good personnel protection is required for han- added layer is in removing oxygen compounds. We monitor
dling the spent zeolite catalyst. oxygenates in the feed by using a GC mass spectrophy for
specific oxygen compound.
STEGELMAN:
We take special precautions in handling both the fresh TIEMAN:
and spent isomerization catalyst. In dumping the spent I agree with Mr. Raterman on Butamer C4 operation.
catalyst the system is thoroughly nitrogen purged and a Typically, we will recommend a 13-X molecular sieve guard
constant positive nitrogen pressure is kept on the reactor. bed. In some cases where the feeds are more difficult or
All those working around the dump nozzle wear protective contain contaminants we will go with three layers, silica
clothing and fresh air. The catalyst is dumped into 55 gal. gel, and a treated alumina. In Penex units with I-8 catalyst,
steel drums which are purged with CO2. When the drums for C5/C6 feeds, we require hydrotreating of the feedstock
are full, a lid is placed on the drum. These drums are then and that normally will remove all oxygenates that are
shipped to a regenerator. present. In this case though you have to watch out for
The regenerated catalyst is returned in super sacks. The leaking exchangers where you could get oxygenate con-
reactor is loaded with a sock and a distributor plate on the tamination from by-passing of the hydrotreating catalyst.
end of the sock. The catalyst bed is physically leveled with a For C5/C6 Penex units, the feed and makeup gas are both
rake every one to two feet. If entry into the reactor is dried using 4A molecular sieve driers. Analytical methods
required, fresh air is used. The catalyst is brittle and are available to UOP licenses for oxygenates and C4 and
breaks easily so care should be taken not to walk on it or let C5/C6 feeds. We also provide an analysis by using a high
it free-fall. resolution Mass Spectroscopy on a case-by-case basis.

BRANHAM: TED LATHAN (UOP):

I agree with Mr. Tieman’s and Mr. Utley’s comments. I I would like to expand on Mr. Tieman’s comments. UOP
might add that the reactor is kept under nitrogen purge has available for purchase the adsorbent required for use
during the loading. in isomerization feedstock purification.

QUESTION 46: QUESTION 47:

On a C4 or C5/C6 isomerization unit using a chlo- ln a water sensitive isomerization unit, what cri-
rided catalyst, what types of scavenger guard reactors teria are used to determine which equipment to
are available to trap any oxygenated compounds? inspect during turnarounds? This question refers to a
What operating conditions are required to de- unit using l-8 catalyst. Also, what is the accepted
compose them in the hydrotreating section? How are interval length for internal or external inspection?
oxygenates monitored? What types of external inspection are utilized? What
kind of unique problems have been found?
MORGAN:
Our Ponca City butane isomerization unit has been BRANHAM:
plagued for years by oxygenates and sulfur in the normal Ashland has operated and inspected UOP C5/C6 isomer-
butane feed. We have significantly improved the existing ization units for almost 20 years. We do not feel that there
feed dryer performance by installing a three bed system. are any special criteria for inspection procedures or fre-
The first bed uses a 4A molecular sieve designed for bulk quency. We do inspect the reactor internals during each
water removal. The middle bed contains Selexsorb-CD, a catalyst changeout. Also, because of the high degree of

133
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

corrosion and pluggage potential in the stabilizer tower, we MACKEY:

will usually inspect this tower and the overhead system, We do not do internal inspections at catalyst changes,
including the caustic scrubber at each catalyst change. only at the turnaround interval of three years or so. We do
Exchangers and piping in the rest of the unit follow normal external UT and shear wave to check the general condition
refinery inspection guidelines. of the vessels. We think that is satisfactory.

MACKEY:
We inspect the Butamer equipment during our turn- QUESTION 48:
arounds every three years. We have not observed problems What chloriding agents besides carbon tetra-
in the reactor loops at these periodic inspections. We do chloride are people successfully using for l-8 cata-
external ultrasonic inspections, roughly once a year, dur- lysts? What are the purchase specifications and who
ing each reactor catalyst change and use ultrasonic and are the suppliers?
shear wave to generally check for the condition of the
vessels. MACKEY:
In our Butamer unit we have used carbon tetrachloride
TIEMAN: as the chloriding agent for the I-8 catalyst, We understand
I would not add anything to what has already been said. that methylene chloride is an alternative. However, we
have not yet found the supplier who can provide the chemi-
HEROS DERGREGORIAN (Giant Refining Company): cal with the low levels of contaminants that we are seeking.
Our isomerization unit is about four years old and the
unit is performing well. We have not entered the reactor or BRANHAM:
moved the catalyst for screening. Do you think it is time to There are numerous suppliers listed in the OPD Chemi-
do something about it? cal Buyers Directory, Of course the alternatives that come
to mind are methylene chloride, trichloroethane, and
BRIAN JOHNSON (UOP): chloroform. The thing to be cautious about is the presence
The question referred to a zeolitic isomerization unit of stabilizers in these alternatives. The trichloroethane
with regard to when the reactor should be inspected. I usually has 500 ppm butylene oxide added. Chloroform can
think the primary criteria should be that the delta P across contain isoamylene. Ashland is in the process of starting up
the catalyst bed is still normal and within the design a new Butamer unit in Catlettsburg, Kentucky. We will be
guidelines of the unit, I see no reason to go into the reactor using methylene chloride as the chloriding material. We
unless there is evidence of an abnormality. I would wait are fortunate enough to have a supplier in West Virginia at
until the first catalyst regeneration of the unit and then do our location.
the inspection depending on how the regeneration condi-
tions looked. RATERMAN:
We currently use carbon tetrachloride in all of our units
J. B. RODDEY (Roddey Engineering Services): using I-8, but expect to change this to methylene chloride
I believe that what the person is trying to ask is about in at least one unit in the near future. I-8 catalyst is
inspecting. We do not want to go inside the reactor and extremely sensitive to moisture, therefore two of the key
expose the unit to water because of the length of time specifications for any chloriding agent which is to be used
required to dry out. What I hear from the panel is that with this catalyst are the moisture and the oxygen content.
everybody is going in and inspecting at each catalyst We specify the chloriding agent must be available in a
change. Then the question becomes, “How long does it take moisture-free form and it must be free of oxygenates. Often
you to dry the unit out? And do you check for feed effluent manufacturers of chloride compounds will add butylene
exchange leaks and so forth? And if you do, do you hydro- oxide as an inhibitor to prevent the compound from de-
static test?” composing. These oxides readily decompose in the isomer-
It is a big item because you can not dry an isomerization ization unit to form water and deactivate the catalyst.
unit out in a day or two. It takes time to do that and if you
use I-8 catalyst then you are killing catalyst all the time TIEMAN:
you are drying out unless you dry out before you load Chemicals that have nitrogen containing compounds as
catalyst. I do not think we really addressed what the person additives should also be avoided.
was asking. He was asking what is the relationship be-
tween inspecting the unit, exposing the unit to air, dryout
time required, etc. Those are all complicated factors, Could QUESTION 49:
you please comment on that? Does anyone use the isomerization unit to get rid
of benzene in light reformate or light naphtha? Up to
BRANHAM: what percentage may benzene be present? What
On our catalyst changes, we would normally purge with precautions have to be taken in order not to interfere
nitrogen during the loading process to keep moisture out. with the isomerization process?
During the start-up, we bypass the reactors until dry so
there really should not be much moisture that comes in STEGELMAN:
through the catalyst change itself. We have eliminated significant benzene in a light naph-

134
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

tha isomerization unit. The quantity of benzene in the TIEMAN:

isomerization unit feed has been on the order of 2 to 4 vol %. UOP has done pilot plant testing on feedstocks that have
The only way to compensate for the exotherm created by the had up to 15% benzene.
benzene saturation is to quench the stream with cool
hydrogen ahead of the bulk of the remaining isomerization RICHARD D. PETERSEN (Amoco Oil Company):
catalyst. Since the benzene saturation reaction is very fast, I am very concerned about temperature excursions and
the hydrogen quench allows the remaining isomerization runaway reactions when we get up to the 5% benzene level
reactions to take place without the complication of ad- particularly if it brings in a lot of heavier cyclic structures
ditional temperature rise due to the aromatic saturation. and paraffins. They are going to crack and run away.

TIEMAN: BRIAN JOHNSON (UOP):

I agree with that. The isomerization process can be used Our approach in handling high benzene feeds will not
for very high benzene levels. We are working now with only be just to remove the heat of reaction of the benzene
light reformates, fractionated from reformer products. As but to also dilute the reactants in the system so that the
Mr. Stegelman said, the main thing that has to be taken in individual reactors will not see extremely high delta T’s
to consideration is the high exothermic heat of reactions. A Just as a point of reference, the heat of reaction of benzene
new unit today can be offered to process high benzene feeds is about five times that for the hydrocracking reactions. So
or a revamp can be done if the existing unit is not capable of benzene is the primary concern for the control of the exo-
removing the heat. therm. Our approach will be to limit the amount ofbenzene
going to any reactor so that not only is the delta T within a
RATERMAN: restricted range, but also that, at least in some systems,
We have one isomerization unit which, depending on the there would be liquid phase at that point so that some of the
type of crude that it is processing, will have at times up to 5 heat of reaction could be used up by heat of vaporization of
vol % benzene in the feed. The unit was designed for that the product.
kind of benzene content so we can handle the exotherm.
The problem that it causes is that it does raise the average RONALD E. MARRELLI (Phillips 66 Company):
temperature of the reactor causing a less favorable equi- With respect again to the benzene and what the gentle-
librium for the isomerization reactions. man was talking about on delta T, where would you limit
the delta T on the isomerization reaction?
UTLEY: BRIAN JOHNSON (UOP):
I agree with what has been said. We have successfully The delta T will be restricted to something below 100 °F.
run light straight run containing 2.5 to 3 vol % benzene and
4 to 4.5 vol % cyclics to our I-7 unit. We have experienced
large exotherms at concentrations much above these QUESTION 50:
amounts. We have not seen an impact to the process when What impact does fluoride contained in the nor-
running these levels of cyclics. At levels much higher than mal butane feed have on the chlorided alumina bu-
these we feel an intermediate reactor quench is probably tane isomerization catalyst life?
required or maybe two-stage isomerization. In the first
stage saturate the aromatics and then do the isomerization TIEMAN:
reaction in the second stage. Fluoride is a permanent poison for I-8 catalyst. There-
fore, we recommend in Butamer feed that the fluoride
LOUIS SCOTTI (ABB Lummus Crest Inc.): levels be kept below 0.5 wt ppm. This usually accomplished
Mr. Utley partially answered the question. For those by using defluorinators on the HF alkylation unit normal
refiners processing higher concentrations of benzene in the butane stream.
feed, what is the limit of naphthenes that can be handled
without creating problems? STEGELMAN:
On rare occasions when fluoride-containing normal bu-
UTLEY: tane has been processed over a chlorided butane isomer-
Maybe someone from UOP could answer it better, but in ization catalyst, we have experienced noticeable bed loss or
our case at 4 to 4.5 vol %, we do not see much of a problem. loss of catalytic activity. The source of the fluoride has come
exclusively from outside purchased normal butanes. Un-
BRIAN JOHNSON (UOP): fortunately the exact magnitude of the bed loss has not
We have several commercial units operating on reactor been readily quantifiable at the time of the unplanned
feeds with cyclic contents, C6 aromatics, well above 20%. fluorided butane addition.
The isomerization reactions do get more difficult as you get The range of fluoride that has been observed from the top
these heavier and more cyclic feeds but the system has been of the bed of the catalyst samples has ranged from 2 to 6
commercially tested for several years now. wt % based on fresh catalyst weight. At this level the
fluoride is tying up with the chlorided alumina, releasing
RICHARD D. PETERSEN (Amoco Oil Company): either a chloride or an oxygen. Any oxygen release will be
Mr. Tieman, you mentioned high concentrations of ben- quickly converted to water resulting in additional catalyst
zene. but did not mention a number. deactivation.

135
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

G. Etherification first commercialized in New Zealand by Mobil. The re-

maining olefins leaving the MTBE reactor are also simul-
taneously upgraded to gasoline in the MOG unit. By in-
QUESTION 51:
tegrating the MOG with the MTBE unit, you can eliminate
the need for the costly methanol and oxygenate recovery
What are the relative advantages for molecular system, and upgrade the MTBE raffinate to high octane
sieve versus water wash cleanup system for MTBE gasoline.
raffinate? What has been the panel’s experience with
both systems?
STEGELMAN:
Phillips uses a water wash system to recover methanol
CARON:
from the MTBE fractionator overhead material. At one of
The advantage of using water wash is that it is very
our refineries there is no further treatment of the MTBE
cheap. The disadvantage is that it allows concentrations of
raffinate prior to processing in the HF alkylation unit. We
ACN in the methanol to build up to 10 to 100 times the
feel that the water wash system offers greater tolerance for
amounts in the feed. We do not have any experience with
operational upsets of the upstream equipment. We feel this
molecular sieves. is cost competitive with the competing methanol removal
processes. If the refiner chooses to further reduce oxy-
MORGAN:
genates in an alkylation unit feed, Phillips offers a pro-
We just completed an analysis of an absorption recovery
prietary dimethyl ether removal scheme that one of our
system versus water wash at the Ponca City refinery. Our
licensees opted to install very successfully.
analysis included economic, operative, and environmental
factors. In units that run less that 1% methanol in the
raffinate, our study showed that utility and capital costs GEORGE J. MANTHOS (ABB Lummus Crest Inc.):
favored the absorption system. Our MTBE unit typically To remove methanol from MTBE raffinate, we would
runs higher than 1% which made the water wash more design a water wash system which is efficient and results in
attractive. a small amount of methanol in the outlet raffinate stream.
Operationally, we concluded the water wash system is Some plants using our CDTECH technology are operating
able to handle slugs of methanol in the raffinate feed better with 1 ppm of methanol in the raffinate. The raffinate after
than an absorption system. The cyclic nature of an absorb- water wash is suitable for processing in alkylation units.
ent recovery systems tends to increase the control com- Our water wash system has a very high reliability and is
plexity and reliability versus the control requirements of a virtually unaffected by small variations of methanol con-
water wash. We are installing an upstream feed water centration in the feed raffinate. The resulting methanol
wash to minimize ACN and HCN methanol concentrations. water mixture is further processed for methanol recovery.
From an environmental view, the water wash system
certainly does have a potential problem with methanol in TED LATHAN (UOP):
the effluent water. But in our system the effluent water will Since the question is general with respect to cleanup of
normally contain less than 250 ppm of methanol. This low the MTBE raffinate, it should be understood that there are
level of methanol will have no impact in our water treating two distinct molecular sieve processes utilized in the
system. The absorption system does generate some solid cleanup of MTBE raffinate. The Methanol Recovery Proc-
waste and in this day and age we looked at that as an ess is utilized for bulk methanol removal and the Oxy-
important factor. genate Removal Process is used for removal of residual
methanol passing through the Methanol Recovery Process
BRANHAM: or water wash, as the case may be, plus other oxygenates,
Ashland’s MTBE unit at the Catlettsburg complex start- such as dimethyl ether and MTBE, that are not removed by
ed up approximately four months ago and this did include a methanol recovery adsorption or water washing.
molecular sieve cleanup system on the raffinate stream. The Methanol Recovery Process should be evaluated
The experience to date has been good with total oxygenates versus a water wash system specifically for methanol re-
in the effluent being less than 10 ppm. The utility cost for a moval only. The Oxygenate Removal Process serves the
molecular sieve wash was estimated to be approximately purpose of being located downstream of one of these units to
20% below the water extraction process. The capital invest- provide ultra-purification of the MTBE raffinate stream
ment for both systems was comparable. I might mention before further processing in an oxygenate sensitive system.
that the molecular sieve system is a proprietary process and Initial comparisons of the Methanol Recovery Process ver-
does require some licensing and engineering fees which are sus water wash resulted in very favorable capital and
involved in those economics. utility advantages for the Methanol Recovery Process.
Subsequently, some MTBE licensors have optimized the
RATERMAN: design and operation of the water wash system. As a result,
Methanol and oxygenates can be recovered from the the water wash system’s investment cost are lower than the
MTBE effluent by using either water wash or molecular Methanol Recovery Process; yet the Methanol Recovery
sieves. In addition to this, Mobil has developed a catalytic Process still maintains up to a 30% utility advantage.
approach called MOG, Mobile Olefins to Gasoline, which Recent evaluations have shown water wash to be the best
will convert the residual methanol and oxygenates in the economic option.
MTBE raffinate stream to a high octane gasoline. This Briefly speaking to the Oxygen Removal Process, the
technology is an extension of the MTG process which was Oxygen Removal Process provides significant capital and

136
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

operating cost savings compared to dimethyl ether frac- TIEMAN:

tionation, whether it is located downstream of a Methanol I think this comment was made earlier in the session but
Recovery Process or a water wash system. We have Oxy- I will add to what has been said. The selective hydro-
genate Recovery Processes operating downstream of both genation of the diolefins is even important in C5 streams in
Methanol Recovery Processes and water wash systems. producing TAME because it reduces the deactivation on the
Addressing the second part of the question regarding resin. So that is another thing to look at.
experience, the two initial Methanol Recovery units re-
quire significant maintenance attention to the valves lo- DAVID J. MILLER (ARCO Chemical Company):
cated in the Methanol Recovery unit due to their frequent The plant technology that you are talking about is
cycling and the differential pressure the valves are exposed ARCO’s. We have been producing MTBE and TAME simul-
to. This problem has been overcome in the units operating taneously for 2 to 3 years in our own units and in licensed
subsequent to the first two installed by modifying valve units. One can optimize initially to take care of co-
characteristics. Currently there are five Methanol Recov- production parameters, and the pretreatment for this de-
ery Processes operating in the United States and several sign properly takes care of all poisons.
others scheduled for start-up overseas within the next 12
months. JOHN LYNCH (Petrogistics, Inc.):
With respect to the Oxygen Removal Process, it has been From the standpoint of some of the issues around meth-
proven to be a reliable, cost effective system with currently anol separation, the implications to alkylation are com-
over 20 units in either operation, construction, or design. pounded when you start adding the C5’s in with the C4’s.
One critical point here, if you start looking at where we
STEVEN M. FISCHER (Champlin Refining and Chemi- have to go on some of the proposed oxygen requirements,
cals, Inc.): TAME has to be in the picture. From the standpoint of
We have a water wash system on our MTBE raffinate and minimizing oxygenates to particularly the HF units and
it seems to be effective in removing methanol and MTBE, maximizing oxygenate production, one is going to have to
but we get a significant amount of DME through. I would look at separate reaction sections and a unified methanol
like to ask the panel, have they measured DME in their separation section. By utilizing a separate reaction section
water washed raffinate? What levels have they found? for the iso-amylene, you are able to get a cleaner feed to the
alkylation unit.
PAINCHAUD: There are a large number of these oxygenate projects on
We have a dryer downstream of our Water Wash system. the board right now and a lot of them right now do not have
The dryer contains a bed of molecular sieve for water the integration with the amylenes yet. Amylenes are
removal and a bed of ALCOA promoted adsorbent for coming and should be a factor in the design of these units
oxygenate removal. I do not have an analysis of the feed to right now.
the dryers, but I do have the outlet analysis. The stream
contained 12 ppm of DME, 11 ppm of methanol, less than 1 DAVID J. MILLER (ARCO Chemical Company):
ppm water and 10 ppm of MTBE. The oxygenates are We agree with that statement; we have already ad-
higher than expected for which I do not have an explanation dressed the issue commercially. We found that optimized
at this time. water wash, used in conjunction with the UOP molecular
sieve technology, is capable of taking out all of the methanol
MORGAN: and oxygenates that are necessary to meet the HF alkyl-
We do not expect the water wash to remove DME. Cur- ation plant’s requirements. It is being done commercially
rently the DME level runs between 50 and 100 ppm in the in MTBE/TAME units and MTBE units.
MTBE raffinate stream.
GEORGE J. MANTHOS (ABB Lummus Crest Inc.):
CARON: We too have co-produced MTBE/TAME in our CDTECH
We were designed for 400 ppm by weight DME in the units.
raffinate. We typically run at less than 200 ppm. Yes, a C5 stream can be run through an MTBE unit by
adjusting operating conditions, but at a lower capacity. The
conversion will be lower at the same capacity as residence
time required for iso-amylene conversion to TAME is
QUESTION 52: greater than for isobutylene conversion to MTBE. Nor-
Can a C5 stream be run successfully through an mally, a TAME unit has to run at lower pressure and higher
MTBE Unit? If so, what operation changes are re- temperature conditions. Feed pretreatment, such as selec-
quired for TAME production? Is additional pre- tive hydrogenation may be required to reduce the con-
treatment required for feed to the unit? centration of existent and potential gums in the products.
We believe, however, it is not required for the successful
SCHAUB: operation of the CD technology TAME process.
We know of a plant where up to 20 wt % C5, has been
charged to the MTBE unit in a mixed stream without CARON:
significant change to the operating conditions. What they We believe it is possible to run a combined C4 and C5
do is set conditions for optimum isobutylene conversion and stream through an MTBE unit provided the same process
they report a conversion of about 50 to 60% on the amylene. conditions as for C4 are maintained. This will maximize

137
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

MTBE production and will let TAME conversion slide. work that the sulfur injection was required to passify metal
The critical parameters are high RVP of the product surfaces. With changes to the sulfur injection which in-
stream due to C5’s and the yellowing of the product due to clude redundant systems and relocating the injection point,
polymerization of the heavier type of diolefin material. we expect a completely successful restart and long-term
Pretreatment through hydroisom units eliminates diole- operation.
fins in the feed and increases the reactable C5 concentra- In addition to the NPC unit, there are eleven other
tion. You also get the benefit of reduced acid consumption Oleflex units in various stages of design and construction,
in downstream alkylation units. designed to operate on C3, C4, and C3/C4 mixtures. We are
keeping our licensees fully apprised of the start-up situ-
ation in Thailand.
QUESTION 53:
What technologies are being commercialized to UTLEY:
dehydrogenate butanes and/or pentanes to increase Well, in all fairness, I have to mention Air Products
the production of MTBE and alkylate? Catofin process which is an extension of their butadiene
technology.
STEGELMAN:
Phillips steam active reforming process, know by its SCHAUB:
acronym STAR, is included under license along with the I will put in a plug for a fluidized bed process that will be
Phillips MTBE process in Coastal Chemical’s Cheyenne, licensed by Snamprogetti SpA.
Wyoming, MTBE project. It is also to be used in another, yet
unannounced, MTBE project. At Borger, we are about to JOHN LYNCH (Petrogistics, Inc.):
complete the detailed cost estimate to evaluate economics One thing that was not really discussed here was the
on an MTBE process using STAR. pentane issue. Again, it is our viewpoint that there is going
to be a substantial increase in incremental alkylation.
TIEMAN: Even if you look at the second stage of RVP reduction,
UOP offers the new Oleflex technology for the selective potentially going on a 7.7 RVP on gasoline, overall, alkyla-
conversion of propane, isobutane, normal butane, and iso- tion is not going to be constrained by olefin availability, but
pentane to the respective olefins. The first Oleflex unit was isobutane availability.
brought on stream at the NPC refinery in Rayong, Thai- Ultimately, refiners are going to have to purchase
land, in January of this year. This first unit is a high oxygenates to get the 2 wt % oxygen level for gasoline.
severity Oleflex charging propane feed to selectively pro- Considering the butanes balance, normal pentane isomer-
duce a propylene product. ization to isopentane with isopentane dehydrogenation to
There have been two shutdowns on the Oleflex plant iso-amylene for TAME will be a viable process in the
in Thailand. The first one was due to a Reactor 2 center future.
screen mechanical defect, which was caused during in- If you look at the preliminary data from the 1990 NPRA
stallation of the center screen in the reactor and which report, I think one thing that was relatively surprising
resulted in the loss of the catalyst containment in that there was if you add up all the light naphthas in the
reactor. Catalyst then migrated into Reactor 3 which lead gasoline pool right now, that have a 81 (R + M)/2 or less,
to plugging of Reactor 3 and subsequent high pressure drop they represent 11.9% of the gasoline pool. So we have an
and a shutdown. During the ensuing turnaround, we found abundance of pentanes out there that are not going to be
no other problems in the reactor circuit and the reactor able to be moved up to premium unleaded. I think once
circuit was free of coke. everything settles down, we are going to see a lot more
The second shutdown was caused by the loss ofcontrolled activity around pentanes to oxygenates than we see right
sulfur injection to the reactor section, which resulted in now for normal butane to MTBE.
coke formation in selected parts of the reactor section. The
unit was shutdown this time due to a high pressure drop in GEORGE J. MANTHOS (ABB Lummus Crest Inc.):
one of the reactors because of the coke particles plugging Thank you, Mr. Utley for your comment. Several tech-
the center screen. The subsequent turnaround has shown nologies are being commercialized to dehydrogenate bu-
the cause of the loss of sulfur injection and has also con- tanes and/or pentanes. United Catalyst and ABB Lummus
firmed that there has been no further catalyst migration Crest acquired the Houdry Catofin technology from Air
since that first turnaround. Products earlier this year. We have over six operating years
Our conclusions at this point are that we have proven the of experience in commercial units in the dehydrogenation
catalyst activity and selectivity and run at near designed of isobutane for the production of MTBE. In addition, we
throughput, conversion and propylene production during have two repeat customers who will use the Houdry Catofin
the time the unit has been on-stream. We have established technology also for the production of MTBE.
catalyst circulation and full regenerability of the catalyst Pilot plant studies have been performed on the de-
in the CCR regenerator. I did not mention it earlier, but this hydrogenation of pentanes in the past. Based on these
process is UOP’s CCR Platformer. studies, basic engineering has been performed to prepare a
We saw no coking during the first run when the sulfur process design specification for potential commercial oper-
injection was intact. We had anticipated from pilot plant ation of pentane dehydrogenation.

138
TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESS

A. Alkylation-Sulfuric Acid Sulfur leads to increased acid consumption, reduced
alkylate octane numbers, reduced alkylate yield, and in-
creased waste disposal. Commercial experience has shown
that acid consumption is approximately 5 pounds of HF per
Feed pound of sulfur, regardless of the actual sulfur compound.
Three types of sulfur are typically present in each alkyla-
tion feed: hydrogen sulfide, mercaptan, and carbonyl sul-
QUESTION 1: fur. The hydrogen sulfide is typically removed by extraction
What type of feed pretreatment (like selective with regenerable ethanol amine followed by a caustic solu-
hydrogenation and sulfur removal) is required when tion. This ethanol amine extraction will also remove the
alkylating butylenes and amylenes in both H2SO4 and carbonyl sulfides. Mercaptan is removed by washing with
HF alkylation units? regenerable caustic solution. We use the UOP Merox
process.
McCLUNG: When looking at amylene alkylation, the amylene typi-
The usefulness of selective hydrogenation depends on cally contains twice the mercaptans of propylenes and
both process type (sulfuric or hydrofluoric), the value of butylenes combined. Diolefins lead to the production of
incremental octane, and diolefin level in each feed. For polymer in the alkylation unit, leading to increased acid
sulfuric acid base units, economic justification is more consumption. We do not have any means of removing the
dependent on acid cost and diolefin level, with octane ben- diolefin as our alkylation unit feed contains 0.2 to 0.3 wt %,
efit being secondary. For hydrofluoric, acid consumption and UOP’s SHP process usually becomes viable only from a
due to diolefins is important, but isomerization of butene-1 standpoint of reduced acid consumption if the diolefin con-
to butene-2 in C4 alkylation is much more important be- tent is above 0.5 to 1%. As for the diolefin content of
cause butene-2 produces a higher octane number alkylate amylenes, they typically contain 50% more diolefins than
than butene-1. butylenes.
Selective hydrogenation of pentylenes has its primary
benefit in removing of diolefins for both sulfuric and HF HAMILTON:
alkylation. In fact, selective hydrogenation is almost essen- I concur with what has been said. Feed pretreatment
tial because the diolefin levels are so much higher in the requirements become more critical and more severe when
pentylenes. Acid consumption is reduced in both cases with processing amylenes relative to butylenes. This is because
minor value, if any, for isomerization of pentenes. To the the C5 cut will have higher levels of diolefins (ca. l%),
extent that heavy residue caused by dioletins dilutes oc- higher and heavier sulfur compounds, as well as cyclics.
tane, the octane of the alkylate will also be increased. ASO Directionally, these compounds will increase acid soluble
generation can be reduced drastically, or eliminated, with oil formation and, thus, acid consumption/regeneration
the proper operation of selective hydrogenation unit. This requirements in H2SO4 and HF alkylation units.
statement applies both to C4’s and C5’s. Both types of alkylation units will benefit from selective
Both butylenes and amylenes require removal of sulfur hydrogenation of either C4’s or C5’s in terms of reduced acid
compounds down to the 25 to 100 ppm level based on the consumption/regeneration requirements and better prod-
selective hydrogenation process being precious metal in uct quality (lower gum, better color). The utilization of
nature, in order to get the best performance of that unit. selective hydrogenation when processing amylenes is a
These levels are usually accomplished with available caus- practical necessity and is recommended for either H2SO4 or
tic treating technologies. HF units. For butylene processing it is optional. Its incre-
mental benefit will be greater for HF units if the selective
KNEPPER: hydrogenation unit is designed to maximize isomerization
Of course, feed pretreatment is a key to smooth operation of 1-butene to 2-butene. This reaction, which occurs more
of an HF alkylation unit. Oxygenates, sulfur, and diolefins readily in a H2SO4 unit, results in higher octane alkylate.
all decrease the operability and profitability of an HF Both C4 and C5 FCC cuts will need to be treated (e.g.,
alkylation unit. Water acts as an acid contaminant, and can MEROX) for sulfur removal. The C5 cut will be more
only be removed from the acid by regeneration, and even difficult to treat and more stringent processing will be
then a large amount of acid is lost. Large quantities of required. Sulfuric acid alkylation units can generally tol-
water in the acid can lead to severe corrosion within the erate higher levels of sulfur than HF units. A typical
acid section of the unit. Molecular sieve, or aluminum specification for HF is 20 ppm maximum sulfur, a H2SO4
driers, or an oxygenate stripper column are typically used unit can process 25-50 ppm sulfur, albeit at the expense of
to remove water in feed streams. higher acid consumption.

126
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

GRANT:
I would like to add that a table of the effect on acid
consumption by different contaminants was given in the
1987 NPRA Q&A Session under Alkylation question num-
ber 4. With selective hydrogenation, it is critical for the
upstream mercaptan removal extraction process to be ade-
quately sized and properly operated. At one facility, the
design total sulfur content to the hydrogenation section is 5
ppm, but due to the undersized mercaptan extraction unit,
the total sulfur to the hydrogenation is sometimes as high
as 30 ppm. The butadiene content at the reactor inlet is
normally 2400 ppm, and the outlet 10 ppm. Because sulfur
is a temporary poison to the hydrogenation catalyst, total
feed sulfur in the 30 ppm range has resulted in a product
butadiene content as high as 400 ppm.

FREDERICKSON:
We have no special feed pretreatment for our sulfuric acid McCLUNG:
alkylation plants, even the unit alkylating amylenes. Am- No, any water that remains is going to be detrimental to
ylene quality is controlled by closely watching the cut point the alkylation process, but not to the hydrogenation process.
of the feed and the acid consumption in that alkylation
plant. Sulfur removal is typically performed with caustic
and/or Merox treating units. One of our HF alkylation QUESTION 2:
units has a selective hydrogenation unit which went on What will the effect be of adding TAME raffinate to
stream in the first quarter of 1991. This alkylation unit either HzSO4 or HF alkylation units?
feeds butylenes only. Alkylate octane is up 1 to 1.5 num-
bers. ASO production is down 80 to 90%, and the operators HAMILTON:
are very pleased with the performance of that plant. Blending TAME raffinate (i.e., amylenes) into either
sulfuric or HF alkylation units will increase acid con-
JEAN COSYNS (Institut Francais du Petrole): sumption regeneration requirements (owing to increased
From our experience as a hydrogenation process licensor, acid soluble oil formation), and will decrease alkylate oc-
we observe that in case of H2SO4 alkylation on the C4 tane relative to processing of butylenes. With respect to
stream, hydrogenation does not seem to be always required processing the total C5 cut from the FCC, processing TAME
for very low butadiene concentration, for example less than raffinate will result in a reduction in acid consumption if
0.3%. Thus far we have sold seven units for the purpose of oxygenate carryover is controlled. There is also some evi-
butadiene hydrogenation. On the contrary, in the C 5 dence that the reduction in branched isoamylenes will
stream, the diolefin content is much higher, and hydro- reduce the hydrogen transfer reaction. Unlike the octane
genation becomes a must. We have already sold two units advantage associated with processing MTBE raffinate,
for this purpose, and we are expecting a third one. In case of there is no significant increase in alkylate product octane
HF alkylation, butadiene hydrogenation and isomerization under sulfuric alkylation conditions.
of 1-butene into butene-2 are always required for boosting
the alkylate octane number, and, in fact, for this specific BRUCE SCOTT (Bruce Scott, Inc.):
case, we have already sold 15 units. Yesterday we talked about amylene alkylation yesterday
under vapor pressure restrictions. There are a couple of
caveats I want to throw out that really have not been
MICHAEL BUCKO (Merichem Company): mentioned yet. One is the hydrogen transfer reaction Mr.
Merichem caustic treating technology has been widely Hamilton just mentioned. When you feed C5 olefin to an
used in conjunction with alkylation units in order to re- alkylation plant, you will create isopentane. So, the vapor
move H2S and mercaptan sulfur from C3 through C5 hydro- pressure reduction you might expect is not quite there
carbon streams. This is done by utilizing our FIBER- because you do make these light C5’s that end up in the
FILM’” contactor in catalytic caustic extraction and re- alkylate.
generation applications. It enables the refiner to reduce the The other caveat by Mr. Frederickson was alluded to on
amount of acid that is consumed by controlling the total the last question. You have to be very careful how you
sulfur of the treated stream to 10 ppm wt or lower. fractionate these C5’s. All the ugly molecules tend to be on
the heavy end of the fraction, and you can do a lot of good
RAUL OTERO-MARTINEZ (Atlas Processing Company): just by watching the cut point.
On feed preparation, you mentioned the Merox unit for
mercaptans removal. That would be for the sulfurs. After PHILIP BIRK (STRATCO, INC.):
that mercaptan removal, when it goes to the hydrogenation STRATCO recently ran some laboratory tests on sulfuric
process, it was mentioned that any water that would still be acid alkylation of TAME raffinate and also completed a unit
with the feed would be detrimental to the hydrogenation expansion design for a TAME raffinate feed. TAME raffi-
catalyst. Is that correct? nate will yield 91 to 93 research octane and 89 motor octane

127
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

alkylate with an acid consumption of 0.6 to 1.0 pounds per on acid consumption. At one location, as mentioned earlier,
gallon of alkylate. For an alkylation unit already process- we are feeding amylenes to a sulfuric acid alkylation plant.
ing butylenes, adding a TAME raffinate unit, as already We control the upstream distillation of the amylenes to
mentioned, would lower the alkylate octane and raise the hold incremental acid usage to about 1 pound per gallon of
acid consumption, especially if the TAME raffinate con- amylene alkylate. The published values for sulfuric acid
tained high levels of pentadienes and oxygenates. consumption range from 0.6 to 1.0 pound per gallon. Penta-
Acid consumption could be held to 0.6 pounds per gallon dienes in the amylene will increase the acid consumption.
or less if the feed is treated via selective hydrogenation and
oxygenate removal unit. Adding TAME raffinate would PRITZEL:
also increase the RVP of the alkylate as previously men- Mr. Phillip Birk already mentioned this, but research
tioned and decrease the purity of the normal butane prod- information supplied by STRATCO indicates that acid con-
uct. This is due to the high concentration of isopentane sumption for amylenes in a sulfuric acid alkylation unit
typically present in the TAME raffinate. Almost all of that would be in the 0.6 to 1.0 pound of sulfuric acid per gallon of
isopentane ends up in the alkylate and some in the butane alkylate. This consumption could exceed 1 pound per gallon
product. if high levels of diolefins and/or oxygenates were present.
On an HF alkylation unit, the typical acid consumption
for C3-C4 olefins is in the 0.08 to 0.1 pound per barrel of
Process alkylate. For the amylenes, it could be expected to be in the
0.12 to 0.16 pound per barrel of alkylate production.

QUESTION 3:
There have been recent developments in the use of QUESTION 5:
alkylation aids. What methods are being used to What is the proper way to operate a sulfuric acid
monitor the benefits? What are the benefits? unit during an acid run away? What should be done
with the acid?
FREDERICKSON:
Chevron has tried alkylation aids intermittently over the GRANT:
last 30 years. We have not had much success with them. We The first most significant indication of an acid run away
have not tried any of the newly- advertised products be- is a sudden loss in contactor temperature at constant feed
cause the plant operators are not convinced there are any rates. The loss of alkylation reaction will move the unit out
potential benefits that justify trying them. of heat balance, and changes in feed rates to increase
temperature will be totally ineffective. In summary, our
FISCHER: normal procedures call for removing olefin feed from the
We have conducted trials of alkylation additives in the unit and flushing olefin lines with isobutane recycle if
past at two of our refineries. Normal unit monitoring possible; maintain acid recycle and circulation rates, and
methods were used in both cases, and no significant im- maximize fresh acid addition to the affected- system.
provements were observed. We are currently working with Spent acid should be maintained in the acid settler as
a supplier at one of our refineries to conduct a trial on that long as acid settler levels permit and until acid strength
supplier’s alkylation additive offer. They propose using can be increased. Continue to recycle refrigerant to control
statistical process control techniques to measure benefits the heat of polymerization of the dissolved propylene and
from their chemical, and thus will presumably be able to butylene in the acid phase. After 8 to 16 hours of cold
compensate for normal operating variation. polymerization, check the acid phase for heat-up. Sample
the acid, allow to weather, and check for temperature rise.
CARLISLE: If the temperature rise is minimal, then it should be safe to
Betz Process Chemicals is currently marketing an alky- pump the acid to spent tankage.
lation catalyst under the name Alkat XL, with which we
have been having good success. In the laboratory we have NORMAN KOLB (The M. W. Kellogg Company):
shown alkylate yield increases of 5 to 10%, octane number When you find yourself in an acid excursion, it is good to
gains of 0.2 to 0.8, and decreased acid consumption of have historical operating data at your fingertips to facili-
anywhere from 15 to 30%. Along with the material, we do tate problem diagnosis. In addition, it is nice to have a
provide a computerized mathematical modeling SPC pack- quick, low technological tool to verify acidity indication you
age to help determine whether the benefits of the material are getting from other sources (laboratory data, acid
are there indeed, and to help the overall optimization of the analyzers).
unit. Even a low technological analytical tool must be rou-
tinely exercised to insure standard and safe execution of
the test.
QUESTION 4: In my opinion, it is prudent to routinely monitor acid
What is the acid consumption with amylene alkyla- specific gravity for each stage of a multi-stage system using
tion for both H2SO4 and HF alkylation? a hygrometer. This can be done daily by operators. This
determination supplements normal off-unit laboratory
FREDERICKSON: titration of your lowest acidity stage. Historical gravity
We are not alkylating amylenes in any of our HF alkyla- information, along with gravities measured during an ex-
tion plants at the present time, so we have no firsthand data cursion, provide a picture of the extent of your problem.

128
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

A continuous acid analyzer on your discard acid will FREDERICKSON:

provide a precise indication of the timing of events and may This question applies primarily to effluent refrigerated
aid in immediately tracking down the source of a problem. sulfuric acid plants. Since ours are not that design, we have
nothing to offer. Maybe someone in the audience has an
ROBERT E. DAVIS (R. E. Davis Chemical Corporation): answer.
Through the years we have developed a method of track-
ing units that are in the process of an acid run away. I would PAM KEEFER (STRATCO, INC.):
like to enter these into the record. This is the result of about Due to the expense associated with fresh caustic, as well
30 years of experience. I think that it reads, and it should as the disposal cost of this material in certain regions of the
read as a lot of experience going down on the record. The world, STRATCO has researched the viability of utilizing
first intermediate thing to do is to cut the feed out of the three other compounds for removing SO2 from the de-
unit. Maintain the acid circulation at the normal rate. propanizer feed. These compounds include bauxite, zinc
Keep your isobutane recycle at the normal rate, but check oxide, and ammonium hydroxide. Our research indicates
any outside streams for water draw. that all three, and in particular the bauxite and ammonium
In a true acid run away, the DIB overhead accumulator hydroxide, result in SO2 levels in the treated feed similar to
will start to generate considerable amounts of water. This those obtained when caustic is the treating medium.
comes from a side reaction of sulfuric acid with hydro- Further research is needed on our part to determine
carbon in the DIB generating carbon dioxide, sulfur di- exactly the contact time that is necessary as well as the
oxide, free water, free carbon. So the water is really coming amount of each compound that is needed to treat an amount
out of the DIB itself. Do not attempt to do anything to the of contaminated depropanizer feed. To our knowledge there
unit until the water draw ceases on the DIB overhead are no commercial installations that utilize any of these
because it is just continuing to aggravate the problem. compounds I have just mentioned.
While you are doing this, try to isolate the reason, whether
it is bad analysis from the laboratory, water carryover, poor
circulation rates (either on the acid or on the isobutane). General
At this point I would suggest that if you have not already
had in place wooden plugs to pound in the holes that are
about to develop, that you start to carve some. If you have QUESTION 7:
pipe clamps, I would suggest you look for those. The last What are the results of the H2SO4 aerosol testing
thing you will ever do is to cut your refrigeration. I will program that was to be conducted this summer?
cover that for a particular reason. Continue to raise your
acid strength at a normal rate to no more than 10% above GRANT:
normal rate. Try to drive the reactor temperature as low as I do not have any results of this program yet, but I will
you can carry it. The acid corrosion rate is directly related give you an update. The testing was started on September
to the temperature of the acid. 13,199l and is expected to be completed about October 20,
As soon as you have the cessation of water on the DIB 1991. The program includes base line runs on “pure” sul-
overhead, raise your strength back to a normal level. At furic acid, as well as runs on mixtures of sulfuric acid and
that point, you have two choices. If the acid strength has hydrocarbon. The hydrocarbon mixture contains approxi-
been below 80%, you will probably have to discard it. The mately 5% propane, 66% isobutane, 11% normal butane,
catalytic reaction that occurs happens due to formation of and 18% alkylate. The variables being looked at include the
certain materials in the acid. Those are destroyed and acid/hydrocarbon ratio, the release temperature, the re-
converted to the carbon dioxide free carbon and water. We lease pressure, and the orifice size.
have recovered two units this year at 58% acid strength The program is being implemented under the Petroleum
with a loss of one pipe nipple on both units. We tried to Environmental Research Forum, or PERF Currently the
identify if it truly was a run away. The corrosion at two program participants are Amoco, Chevron, Exxon, Mobil,
points should be monitored immediately: the caustic STRATCO, and Texaco. The overall cost of the program is
water-wash train and the depropanizer feed elbow. approximately $475,000. Under the terms of the Par-
Try to maintain your caustic strength in your caustic ticipation Agreement, the results of the study belong to the
wash, and try to maintain your water wash. Watch for project participants for a period of up to five years after the
discoloration of foam at those positions. Do not ever put acid study’s completion. No decision has yet been made by the
at low strength into your spent tank. Polymerization will project participants on publication of the program’s results.
start to occur at around 55 to 60” F, and it will cause a run Participation in the program is open to anyone willing to
away in the tank releasing considerable sulfur dioxide to pay a prorated share of the program costs. Information on
the atmosphere, and a release of the polymer. At that point joining the program can be obtained by contacting Ralph
you can call your painter in to repaint the side of the tank. Cecchetti at Exxon’s Florham Park, New Jersey research
center.

QUESTION 6:
What technologies other than caustic scrubbing, QUESTION 8:
currently exist for treating propane produced off of What is the current status of on-line acid analyzers
H2SO4 alkylation units for SO2 removal? for H2SO4 alkylation units?

129
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

MOYSE: representatives.
Over the last five years, STRATCO has researched and
developed a continuous spent acid analyzer system for use VINCENTE CITARELLA (Exxon Research and Engi-
in sulfuric acid alkylation units. This particular analyzer neering Company):
determines the spent acid strength by measuring gravity of I would like to make a general comment on sulfuric acid
settled acid. The gravity measuring device is mounted alkylation. One of our European affiliates recently brought
externally on the acid piping within a special analyzer on stream a sulfuric acid alkylation unit which produces
cabinet. The advantage of this method is that device is 6,000 barrels per day of alkylate in a single reactor using
non-intrusive, and does not directly contact acid. There- ERE’s alkylation technology. We believe this is the largest
fore, it is not subject to corrosion or fouling. The accuracy capacity alkylation reactor currently in operation. The unit
has been guaranteed to be ± 0.5 wt % for commercial units. has been on stream now for over seven months, and per-
Tests performed in the laboratory have consistently shown formance to date has been excellent. This technology is
the accuracy to be better, down to ± 0.2 wt %. available under license from Exxon Research and Engi-
Currently there are seven units installed: four in the U.S. neering, either directly or through M. W. Kellogg.
and three in Canada. There are several others in fabri-
cation, or under consideration, for various customers EARL KREDER (Monsanto Company):
worldwide. I would like to comment that while not a new process,
there have been significant improvements made to the
GRANT: existing, or the classical, sulfuric spent acid regeneration
Currently one of our refineries utilizes the STRATCO processes. Specifically in the past five years, improvements
model 6000 on-line analyzer off of the last settler to con- have been made to eliminate the lead lined mist precipi-
tinuously monitor spent acid strength. Initially, the pri- tator which has been a general problem from a mainte-
mary problem with the analyzer was maintaining ade- nance standpoint in the process. There is a capital advantage
quate flow rates and temperature through the instrument. to this approach. Also by preheating combustion air, the
By adjusting the thermostat and adding insulation, these total volume of gas through the plant can be reduced which
problems have been eliminated. By properly calibrating reduces the capital cost. There are a number of these plants
the analyzer, it is possible to maintain a continuous acid currently operating and in the design phase.
reading, which on average is within 0.5 percent. Significant improvements have been made in the area of
safety. By replacing brick-lined carbon steel and cast iron
FREDERICKSON: pipe, which have been the industry standard for some 50
We have had little success with on-line acid analyzers. In years, with new austenitic stainless steels, significant im-
our experience they require excessive maintenance and provements in on-stream reliability and safety have
give questionable results. Instead, we have developed an resulted.
automatic titration analyzer that is used for frequent grab When you are talking about very small spent acid re-
samples. These are used twice per shift under normal generation plants, say up to 100 tons of capacity, there has
circumstances, and more frequently if needed. been a study done by Monsanto and Simon-Carves, one of
our European licensees, looking at a modular standardized
DENNIS FINCH (Eppendorf North America, Inc.):
plant, which will further reduce the cost about 20 to 25%.
Eppendorf manufactures a high speed, on-line H2SO4
There was a paper presented on this approach at the
analyzer for this application. This EPAS Series analyzer
STRATCO conference earlier this year in Amsterdam. The
uses reliable FIA (Flow Injection Analysis) techniques to paper is available from Monsanto or STRATCO.
accomplish the analysis. Faster than process titrators and
using far less reagent, the Eppendorf EPAS H2SO4 Analy-
zer analyzes up to six sample streams. Continuous H2SO4
Acid concentration data provides the alkylation unit oper- B. Alkylation-Hydrofluoric Acid
ator the opportunity to maintain high production and con-
sistent quality.

QUESTION 9:
Feed
Are there any new processes that can be used to
regenerate spent H2SO4 acid over the conventional
QUESTION 10:
burning/shift conversion process? What are the capi-
tal costs for the different processes? Are people alkylating C5 olefins in HF units? What
levels are being alkylated? What are the results re-
MOYSE: garding octane, RVP, end point, and ASO yield?
Haldor Topsoe has a new process concept by which capi-
tal as well as operating costs of regenerating spent acid can GRANT:
be reduced by as much as 40%. This concept is now ready for Our operating experience indicates that C5 olefins at
demonstration. No waste water or waste acid is produced. levels above 5% of total unit charge, treated with C3 and C4
The only products from the process are 98% concentrated olefins in conventional amine/caustic systems will cause
pure sulfuric acid, water vapor, CO2, and a small amount of excessive acid soluble oil production and acid consumption.
dry ash. It is a four-step process; it is catalytic; and it The increased diolefin content of the amylene feed in-
utilizes a special falling film condenser which we have creases acid losses through increased ASO make. I would
developed. If you are interested, you can contact one of our like to mention a few references on this topic: the 1987

130
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

NPRA Q&A Session transcript on pages 109 and 110; and alkylating C5 olefins in HF units. Please monitor the fluo-
also two papers presented at the 1991 NPRA Annual Meet- ride content of your finished product. I think you are in for a
ing; AM-91-16, “Integrated Olefin Processing,” by Law- big surprise.
rence Lew, Donald Makovec, and Martyn Pfile of Phillips
Petroleum Company; and AM-91-50, “Ultimate C4/C5 Ole-
fin Processing Scheme for Maximizing Reformulated Gaso- QUESTION 11:
line Production,” by Pam Keefer and Ken Masters of Has anyone replaced all or part of their feed drying
STRATCO, INC. and Jean-Luc Nocca and Jean Cosyns of beds with materials to remove oxygenates from
Institut Francais du Petrole. MTBE operations? If so, is regeneration by any stream
other than butane being considered?
GOLDEN:
Two of our U.S. refineries alkylate C5’s. Normally about KNEPPER:
half or less of the available C5’s are alkylated. Driving We do not have an MTBE operation; however, we have
additional material off of the FCC gasoline results in a oxygenates in our Butamer feed. The oxygenates originate
marked increase in ASO production so that handling of it with the saturate feed coming into the alkylation unit. We
becomes difficult. The increased ASO and polymer produc- are using Alco Selexsorb™ CD alumina in a single Butamer
tion are due to the increased levels of diolefins and C6+ guard dryer that is in series with the feed dryers to keep
material. By alkylating C5’s, the RVP of FCC gasoline is oxygenates out of the Butamer reactors. The regeneration
reduced, bringing about an overall reduction in pool RVP is with butane. We have not quantified the performance of
However, I would second Mr. Scott’s earlier comment that this material in removing oxygenates, however, we lose
the reduction may not be as dramatic as you would suspect. Butamer activity every time this guard dryer is re-
If one third to one half of the available C5’s are alkylated, generated. We will be considering additional Selexsorb in
the end point of the product alkylate will remain within the saturate feed dryers. However, when we do that, the
specification limits. Future regulation of gasoline 90% dis- regenerate will have to go to storage instead of back to the
tillation point, however, could change this. Since the octane olefin feed. Also, I would point out that there is a paper on
of C5 alkylate is not radically different from mixed C5’s in this, “The Adsorption of Oxygenates and Other Con-
our experience, the octane impact on the overall pool is taminants from Isomerization, Alkylation, and MTBE
relatively small. With regard to increasing amylene alky- Feed Streams,” by Coastal Chemical, presented in the
lation, some loss of flexibility in blending premium gaso- AIChE meeting of 1990.
line may be suffered since the total alkylate octane will
drop as more C5’s are added to alkylation feed. WILLIAMS:
Southwestern’s alkylation feed dryers are charged with a
FISCHER: molecular sieve. Our MTBE unit, which is scheduled for
Our experience confirms the previous comments. We are completion during the third quarter of 1992 will include a
alkylating some C5 olefins in one of our HF alkylation units set of adsorbers for removal of oxygenates from alkylation
in the order of about 5% of feed. We have not quantified the feed. The adsorbent system that we have chosen is licensed
results fully yet, but we have seen a drop in octane and an and marketed by UOP
increase in ASO yield.

McCLUNG: Process
To my knowledge, no one is actually processing pen-
tylenes alone in an alkylation unit. I know of one other
refiner besides the ones I have already mentioned who QUESTION 12:
processes C5 olefins in conjunction with C4‘s, but I have no What results are being achieved with staged feed
results for that. injection, especially first stage butylene and second
stage propylene, regarding expected versus actual
PAZMANYI: alkylate yields, octane, acid consumption, ASO pro-
We do not process C5 olefins in our HF alkylation plant, duction, and utility savings? Is there an advantage of
but I have some background information from UOP. They splitting butylenes and amylenes?
told me there have been several applications over the years,
and there has certainly been a lot of study and preparation PRITZEL:
underway for the future C5 amylene alkylation. Regarding I am not aware of anyone operating with segregated
the performance, around 90 to 92 road octane can be ex- stage feed. Plants have been operating for a number of
pected from C5 amylene alkylate with 2 to 5 psi RVP, years with two stages feeding mixed C3-C4 olefins. The
depending on conditions, and whether it is all or only part advantage of two stages is to increase the efficiency of the
(for example, TAME raffinate) of C5 amylamines. Regard- system by optimizing the isobutane to olefin contact in the
ing how much polymer would be made, they expect that the reactor. If the C3-C4 olefins are split, a higher octane alky-
regeneration requirement would be about 1.5 to 2 times late and slightly higher alkylate yield would be, obtained.
more than for C4 olefins. Of course, the impact will be When propylene is charged separately, additional pro-
lessened by being only a portion of the total unit feed. pylenes would react with the excess isobutane via the
hydrogen transfer step resulting in the production of
ROBERT E. DAVIS (R.E. Davis Chemical Corporation): 2,2,4-trimethylpentane and propane. 2,2,4-trimethylpen-
I would like to inject one thing for the people who are tane has an (R+M)/2 octane several numbers higher than

131
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

the typical 92 to 94 octane of alkylate. However, this re- regard. We are told that the corrosion control is quite
quires additional isobutane. You would also have a higher simply a matter of the choice of the correct metallurgy.
acid consumption and additional propane yield.
As far as the splitting of butylenes and amylenes in KNEPPER:
staged feed injection, again I am not aware of anyone doing Our tower uses a monel pipe insert filled with monel
that. However, if that were done, you could expect an Raschig rings. We have tried carbon rings, but they disap-
increase in the hydrogen transfer reaction which would peared after a short time. The teflon rings melted. However,
produce isopentane from the amylenes. The result would be part of the insert uses a teflon seal. The seal holds up
a higher octane alkylate, but also a higher RVP alkylate marginally well, but is 1/4 inch thick. The monel rings also
which, as has been mentioned before, is not really desirable. corroded away after a period of time, but do provide a
reasonable run-length before they are gone. However, the
KNEPPER: monel rings cost approximately $600 per cubic foot.
I concur with Mr. Pritzel. We have a two-stage unit where
we feed a mixed feed to both. In terms of the utilities, we HAMILTON:
have saved quite a little on iso-to-olefin ratio. We run an I have nothing further to add except that we are not
eight iso-to-olefin ratio (molar) right now, but we have been aware of anyone using the teflon trays or the carbon pack-
as low as five. ing. Maybe some of the vendors present have some input on
this.

Mechanical
QUESTION 14:
What methods are being used to agitate covered
QUESTION 13: neutralization pits in HF Alkylation units and has
Is anyone using unique materials such as all teflon there been any problem with vapor evolution from
trays or carbon packing in an HF Acid Regenerator? these covered pits?
What other corrosion control methods are employed?
WILLIAMS:
PAZMANYI: Our neutralization pit is agitated using a Lightning
To the first part, our answer is no. We do not use such mixer. In 1972 we installed a pit blower to move the vapors
materials. To the second part, my comment is that after to the isostripper reboiler for destruction. Since installing
four years of operation, the 3 millimeter internal this blower, we have not had very many problems with
Monel-400 lining of our acid regenerator was corroded and vapor emissions from this pit.
just now we are changing it for a new solid Monel equip-
ment. We tried also to change the regeneration process, and PAZMANYI:
we would like to improve the oxygenate removal from our My answer is also negative. We have never experienced
FCC C4 fraction alkylation feed. vapor or gas evolution from the two neutralization pits in
our HF alkylation unit. The agitation can be done using the
PRITZEL: built-in mixing pumps.
The main problem is to keep the water content of the acid
down. This means frequent monitoring of the acid purity FISCHER:
and water level to prevent acid attack. One of our refineries initially used air agitation, but
subsequently switched to a propeller type mechanical
MOYSE: mixer. This particular pit has a PVC vent pipe on top of the
I have to add a negative comment. We were not able to pit for dispersion of vapors. We also use a small continuous
locate anyone who was using special materials in this nitrogen purge across the vapor space in the pit.

General

QUESTION 15:
What progress is being made on developing a solid
catalyst for the alkylation of light olefins and
isobutane?

McCLUNG:
Unless there is some public announcement at this par-
ticular Q&A Session, I cannot be terribly encouraging on
this subject. What progress is being made is published in
patents, which I review just about monthly. The most pro-
lific in publishing U.S. patents is Mobil. If such a process
had been invented by now, I think that Jim Maiden would

132
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

have told you. He is here, so he can tell you. Mobil’s ap- the quality and representive of the sample. Typically they
proaches in the patents are based on boron trifluoride, are located on the cooling water outlet of the reactor and in
called BF3, and their other approach is zeolitic. However, to the main cooling water return header. Calibration at least
my knowledge there is no commercial application of either once per month is recommended. pH analyzers have also
of these. But again, I suggest that you talk to Mr. Maiden been used, but since there are many variables affecting the
about that. I know that it is the “Holy Grail” of the petro- pH, a fluoride leak would not be detected quickly unless it
leum industry to find this kind of process, and there is a lot was a major leak. Hydrocarbon leak detectors have been
of work being done. I think Mobil is your most reliable used on condensate systems.
source for progress.
WILLIAMS:
PAZMANYI: We have elected to monitor the pH of the cooling water
I have only some literature data. I think there are three exiting the HF Alkylation unit rather than attempting the
types: (1) zeolites, (2) acidic ion exchange resins, and (3) the specific ion route. At the outset, we worked with Rose-
oxide-based super acids. I have not heard of any commercial mount in attempting to analyze for the fluoride ion but, as
applications. Mr. Pritzel has already observed, we encountered sig-
nificant interference. Additionally, there appeared to be
ROBERT E. DAVIS (R. E. Davis Chemical Corporation): problems with the durability of the specific ion probe. We
We have been and are continuing to research this area of therefore decided, just on a cost and reliability basis, to
alkylation. We have had no success in C3-C4 alkylation. select pH measurement. Our standard for this service is the
This is not true in C5 where there are some encouraging Rosemount Model 103. The pH readings are continuously
things. It is not commercial. The work will take at least monitored, recorded, and alarmed via the DCS.
another year before it can be announced.
DENNIS FINCH (Eppendorf North America, Inc.):
MICHAEL HUMBACH (UOP): The Eppendorf EPAS high speed, on-line analyzer hard-
We concur with what has been said. We have a fairly ware will run methods for the analysis of fluoride in cooling
intense R&D effort going on right now in this area. What water, boiler feedwater, condensate and similar streams.
we are finding is that indeed it is going to take a break- The analytical method is determined by the required level
through, not only in catalyst technology, but also in process of detection and the stream’s chemical matrix..
technology. So at this time, we do not have a technology to This analyzer will provide immediate information as to
announce, but we are working on it. breakthrough of condensate polishers or resin beds, allow-
ing fast remedial action. The EPAS can be used to control
blow-down, as well.
QUESTION 16: The EPAS Series analyzers employ an autocalibration
What kind of instruments are being used suc- feature that is operator programmable as to frequency and
cessfully to monitor fluoride ions in cooling water sequence. Calibration frequency is dependent on the
and condensate water both on-line and off-line? method of analysis. Once weekly is typical. Alarms are
What is the testing and calibration frequency? operator programmable, as well.

PAZMANYI:
In our HF alkylation unit, we use Foxboro E99S type QUESTION 17:
fluoride ion analyzers on-line in the cooling water system, How are pH adjustments made for HF Alkylation
in the condensate return system and in the neutralization Unit spent lime before disposal? What reagent is
pit effluent water. The calibration frequency with Orion used?
9409 electrodes is once a month; the checking-out and
testing are done weekly. Last year, the same serial number KNEPPER:
resulted in modified electrodes where the calibration and We now have a KOH system, but previously we used
testing work had to be done more frequently. lime. Our past experience was to trickle a predetermined
We use an off-line Radiometric pH MV measuring in- amount of concentrated sulfuric acid into the sludge, while
strument with fluoride selective electrode, which gives a turning the sludge over with air spargers. The exact
linear millivolt logarithmic concentration characteristic amount of acid required was determined in the laboratory
curve in the full measuring range from (0.1 to 1000 ppm). by titrating a sludge sample to a pH of eleven. We then
Calibration is required every three months. We use a leave the air spargers on overnight and verify the sludge pH
Radiometer Copenhagen pHM 84 pH METER with a prior to disposal. We want to stress that the acid be added at
F-1052 F electrode. a controlled rate as the reaction can be violent if added too
quickly, and this would result in a vapor release.
PRITZEL:
Fluoride ion specific analyzers have been used with suc- WILLIAMS:
cess in some locations on-line on the cooling water. These We adjust the pH of our spent lime by controlling the
analyzers have good sensitivity. On the other hand, they blend of CBM and a neutralization solution containing 4
are subject to interference from the chloride ion, so they percent calcium hydroxide and 2 percent potassium
will not work in salt water service. They also require hydroxide. The CBM flows first into a mixing pit where it is
preventive maintenance since they are very dependent on thoroughly mixed with the neutralization solution; we

133
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

typically correct the pH to between 7 and 9. If we need A 40:1 ratio of water to HF was used as a basis for design.
further corrections, we have the flexibility of adding sack This was developed from the field test conducted by the HF
lime as needed. Once the pH has been corrected, we move mitigation task force led by Lawrence Livermore National
the mixture to a transfer pit for removal and additional Laboratories. In addition to this, our acid circulating
processing via centrifuge. The lime cake produced from the pumps are each equipped with two 25 gallon per minute
centrifuge process is considered a Class I non-hazardous sprinklers that are designed to produce 300 micron droplets
waste, and is disposed in a landfill. The water recovered at 100 psig. These sprinklers are located approximately
from the lime is processed in our wastewater treatment three feet from the pump seals. These acid pumps are also
plant. equipped with double seals, so an inboard seal failure
At one point we did experience poor control in the neu- would result in a flow of buffer fluid into the process.
tralization process. which resulted in high pH material in For containment of HF spills, we have added a retaining
the neutralization and transfer pits. During that period, wall around the perimeter of our new first stage reactor,
our neutralization solution was composed of six percent acid settler, and HF storage vessels. The wall is constructed
potassium hydroxide and six percent calcium hydroxide. of concrete, and is four feet high by approximately six
Reducing the concentrations of these compounds to the inches thick with water stops installed at all joints. The
level mentioned earlier appears to have corrected our pH volume of the containment area is designed to retain a full
control problems. volume of the largest process vessel, which in our case is the
acid storage drum.
DAVID KENNARD (Tetra Technologies, Inc.):
I would like to reiterate a response given yesterday under PAZMANYI:
the topic of recycling solid wastes. Tetra has modified some Our UOP HF alkylation unit was built in 1987 and
of its existing waste acid treatment technology to produce a started up in 1988. The unit design included an automatic
high purity densified calcium fluoride product from ASO isolation system for acid circulating pumps, HF and hydro-
and ARN streams capable of being recycled. Tetra’s Waste carbon detectors, an up/down two-way water curtain
Treatment Division is marketing this process to refiners system around the acid area, HF sensitive paints, a
using HF Alkylation. meteorological station, a change house, emergency booths,
etc.
After some strong complaints from the outside, the fol-
QUESTION 18: lowing was done:
What are the results of the research to modify HF - Emergency telecommunication and a warning system
with an additive to reduce its aerosol forming ten- were established for the two neighboring cities.
dency? How effective is it? - The refinery provided thousands of gas masks for per-
sonal use.
MOYSE: - A foundation was established consisting of a special
We do not have any results at all. As we understand it, a technical committee, responsible for the safety and
lot of work is being done in this area, and we are told that technical level developments in the HF alkylation unit.
the results that have so far been obtained look promising. These developments included reduction of acid inven-
However, we are not able to quantify that. As to the actual tory down to about 25 tons in the system, an emergency
effectiveness, I think you will have to contact your process acid dump system, over-pressurization of the control
licensor because, as we understand it, the results are so far room in case of emergency, isolation of the acid re-
proprietary. generator from the isostripper using remote operated
valve, remote operation of the water curtain, acid un-
loading automatic shutoff valves, a maintenance and
QUESTION 19: operational survey, HAZOP etc.
What HF mitigation measures are being installed These modifications are already in progress (about 90%
by refiners? What level of acceptance by regulatory complete). The acceptance level, after introduction of a new
and community concerns has resulted? policy of “Open Gates,” is improving.

WILLIAMS: KNEPPER:
In conjunction with an expansion of our HF alkylation We have installed remote motor operating valves at the
unit in 1988, Southwestern put in place a number of strategic locations within the concentrated acid section of
measures for reducing HF exposure risk. A water deluge the unit. These include the suction and discharge valves on
system using 13 stationary standard NFPAUL certified the circulating acid pumps, as well as the valves to isolate
monitors was added. The monitors are capable of flowing either one of our dual reactor settler systems. We also have
500 gallons per minute at 100 psig. Deluge monitors were a remote operated valve on our acid unloading line at the
placed at various elevations around the acid storage drums, loading spot which can be shut to isolate any rail car, truck,
reactors, and settlers. Five of the 13 deluge nozzles are or hose problems.
located outside of the acid area. Depending on wind di- Hydrocarbon alarms are located throughout the unit,
rection, these may or may not be utilized during a leak. which we feel will indicate a leak of acid in the unit since
These monitors are adjustable, and the master can be even our most concentrated acid streams contain adequate
activated from the control panel in our central control room hydrocarbon to activate these hydrocarbon alarms. We
or from one of four locations within the unit. have tested these sensors. We feel that the fluoride alarms

134
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

that are currently available will not be reliable in our coupled with community involvement are key issues that
climate. We recently replaced our old vertical mixer settler must be addressed and are not limited to HF.
with a modified settler design which eliminated the old
trayed mixer section. This reduced the required operating FREDERICKSON:
acid inventory by nearly one half on that particular reactor We are using water spray from fixed monitors. We have
system. also established detailed inspection practices and material
The combined acid inventory of both systems was 65 specifications for our plants. Leak detecting monitors are
pounds of HF per barrel per day of alkylate. It is now being scheduled to be installed. So far, we have had no regulatory
operated at 40 pounds of HF per barrel per day. We think we or community problems or concerns about what we are
can approach 25 pounds per barrel per day. The new modi- doing.
fied settler has a boot on the bottom which will provide an
operable acid level range without the full acid volume. The FISCHER:
same project replaced our acid storage drum. Both of these We conducted a fairly extensive risk analysis about four
new vessels were made from low sulfur, vacuum degassed, years ago using an outside consultant. We also subscribe to
normalized steels that are available today. This reduces the the HF mitigation task force work done in recent years. We
possibility of any hydrogen blistering. are now in the process of installing a three level water spray
The project also installed double seals on all of our system made up of close up fire deluge nozzles, a perimeter
concentrated acid pumps, with alarms on the outer seal spray curtain, and several elevated remotely-operated
alkylate flush to indicate any inner seal problems. In terms water cannons. A separate sewer and holding pond will
of regulatory and community acceptance, we are par- collect the runoff water. In addition we will have a remotely
ticipating in the local emergency planning committee. This located dump tank and piping designed to de-inventory the
group is planning for any major problem, not just HF. They unit of HF in less than ten minutes. These systems will be
are installing sirens, and are providing public education on activated separately by the control room operator. We also
how to react in an emergency. have HF detectors at the unit perimeter and have had good
experience with them. In addition to this we have remotely
GRANT: controlled TV monitors at strategic locations throughout
At one of our locations we have already implemented the unit.
hydrocarbon detectors; remote-control TV monitors; re-
mote unit shutdown systems; remote shutdown systems to STEVEN FISCHER (Champlin Refining and Chemicals
isolate smaller sections of the plant in an emergency; Co.):
automatic shutdown systems on the propane and butane For those who are using water mitigation techniques
KOH treaters to protect the equipment from overheating; with these high water ratios (40 to 1), what scenarios are
and HF and hydrocarbon detectors in breathing air intake you using to determine the amount of HF that is released to
which automatically trips the breathing air compressor on which 40 to 1 ratio will apply? What size rupture is being
high level. considered?
Critical isolation valves in HF service are solid monel
plug valves. Critical thermowells have been upgraded to WILLIAMS:
the latest Phillips specifications. All non-essential nozzles A wide variety of release scenarios were examined in the
have been removed from the HF acid reaction circuit. We course of the Hydrofluoric Acid spill experiments con-
have installed a bypass around the acid relief neutralizer ducted in Nevada. Acid release rates varied from about 470
with a rupture disk to route to the flare in case the neutral- gal/m for the vapor cloud dispersion tests, to about 33 gal/m
izer becomes blocked. Work in progress includes; pump seal for one of the water spray mitigation tests. Some of the
fire detection system, water spray deluge systems on all HF graphical data I have seen based on these field tests indicate
inventory vessels and pump seals, and remote controlled that at a volumetric water/acid ratio of about 40, HF scrub-
elevated fire monitors. Planned measures include a rapid bing efficiency approaches 90 percent.
acid dump system and reduction in HF acid inventory. A more complete discussion of the testing program may
be found in the October 17, 1988 edition of the Oil and Gas
GOLDEN: Journal. Additionally, you may wish to refer to a paper by
We are progressing well on a plan that addresses leak D.N. Blewitt, et al., entitled “Conduct of Anhydrous Hydro-
prevention, equipment isolation, leak detection and warn- fluoric Acid Spill Experiments,” International Conference
ing, and leak mitigation. Much of the new installation is on Vapor Cloud Modeling, Cambridge, Mass., November
automated so as not to require intimate operator interface. 2-4, 1987.
Key components of the overall plan are: eliminating piping
and equipment where the risk of a leak may be high; GOLDEN:
installing an automatic vessel isolation and acid dump I know that various orifice sizes have been looked at
system; installing HF detectors, water sprays, and other through the auspices of the HF Mitigation Task Force. I
mitigation and detection equipment; and increasing the think 1/2 and 3/4 inch were two sizes that were looked at,
inventory of personal protective gear, and dispersing the and obviously they set different design criteria. I think that
same over more varied locations. different companies may standardize differently depending
In addition, we train the non-company local emergency on what types of small fittings they have. I believe the
responders through the LEPC in specific techniques to deal thrust has been towards a leak-at-a-bleeder scenario for a
with HF. We further believe that community awareness design basis.

135
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

BRUCE SCOTT (Bruce Scott Inc.): as where the acid goes, that is the big risk, I think, in all of
I have a couple of comments. Responding to the question this. I do not know of any design anyone has that is abso-
on the size of the hole, in my experience most people have lutely failsafe in all scenarios. In your design, it is impor-
been looking at anywhere from a 1 inch or 3/4 inch. I think tant to consider the impact that activating the dump sys-
South Coast Air Quality Management District in Los tem will have on the acid relief/storage vessel. Sizing and
Angeles set up a two inch hole as their standard leak. The pressure rating of the vessel are two critical parameters in
other comment I wanted to make was that the API com- that design. The vessel should be designed to handle a
mittee on HF alkylation has prepared a recommended pressure increase during filling so as to minimize venting
practice, RP751, dealing with HF safety issues. That prac- and relief requirements.
tice is out for ballot now. Balloting ought to be over by the
end of this month or early November. It will cover most of MICHAEL LAUX (Diamond Shamrock, Inc.):
the things you have heard said by the panel. What model HF detectors you are using, Mr. Fischer?

JAMES J. SPEARS (Spears & Associates, Inc.): FISCHER:

I have a general question. Is anyone using magnetic drive The instrument we are using is the TG Series F type HF
sealless pumps for HF service? monitor, manufactured by Bionics Instrument Company of
Japan. The Canadian agent is Aquatronix Inc., 70 Gibson
KNEPPER: Drive, Unit 8, Markham, Ontario.
I am aware that Sundyne is currently selling a pump. It is
not a magnetic drive pump, but it is a completely enclosed
pump that is sealless. They were just in not long ago trying QUESTION 20:
to sell me one, I know that they are being used in California Are there any contamination problems in process-
in some refinery out there. ing the alkylation unit butanes in a C4 isomerization
unit or selling them on the open market as butane
GOLDEN: product? (i.e. pipeline specifications on fluoride
I have a comment on magnetic drive. Maybe some of the concentrations and the status of effective fluoride
vendors can amplify on this. I think the problem with the removal technology to avoid catalyst poisoning.)
magnetic drives so far has been horsepower limitation. You
have to have a rather small alkylation unit for the magnetic FREDERICKSON:
drive to have a good application, If the pumping rates are Alkylation unit butanes are indeed a problem for
fairly high, I do not think the magnetic drives currently can isomerization plants. In the one refinery where Chevron
provide the horsepower required. currently has both an isomerization plant and an HF alky-
lation plant, we send the alkylation plant butanes to a
JAMES J. SPEARS (Spears & Associates, Inc.): KOH treater and then to gasoline blending. We do not
Mr. Golden, what horsepower are you talking about? recycle them to the isomerization plant.
When you mentioned the limits on horsepower, what is the At two other locations we defluorinate the alkylate bu-
typical limit? Magnetic drives go up to about 75 home- tanes with activated alumina to meet sales specifications.
power. So I am wondering what would be the normal re- We are building an isomerization plant at one of these
quirement. (A later inquiry to Dickow, one of the largest locations and are considering recycling the alkylation
manufacturers of magnetic drive pumps, revealed that they plant butanes after defluorination down to 1 to 2 ppm
have pumps up to 200 HP and - depending on the specific fluoride. As for sales, our Warren Petroleum subsidiary
application - potentially higher.) recently set a specification of 1 ppm fluorides in normal
butane that they purchase.
GOLDEN:
Obviously it is going to vary from plant to plant. At our KNEPPER:
plant, the Acid Circulation Pumps are 400 HP, which cur- We do not sell butane, but we do use as a source of
rently exceeds the magnetic drive capability. The Fresh Butamer feed, the side cut butane from the isostripper.
Acid Pump is at the upper end of mag-drive horsepower Feeding isostripper butane to a Butamer unit can be a
range, and therefore is questionable. So far, we have elected problem if adequate feed treatment is not in place. The
to stay with enhanced seals. butane is first treated with two alumina treaters in series
and then dried. We sample weekly the outlet of the first
MOLLA K. ANAM (The UNO-VEN Company): treater and change the alumina when fluoride break-
I have two questions about the rapid acid dump system. through occurs.
(1) When the acid is dumped rapidly, how is the pressure One lesson we have learned is that if the fluorides do
maintained in the dump drum? (2) Is anyone using any break all the way through both treaters they will chew up
remote acid transfer tank? If so, what type of precautions your dryer sieves quite effectively. If we keep a close eye on
are taken for that remote tank? samples, then everything works alright as far as the fluo-
rides are concerned. There is usually something else that
GOLDEN: shuts us down, like oxygenates. UOP says that 1 pound of
Perhaps I did not understand the question, but in our fluoride poisons 100 pounds of catalyst. For selling the
design, the objective is to get rid of the pressure at the point butane, a KOH treater after the alumina treaters is rec-
of the leak as quickly as possible, not to maintain it. As far ommended to drop fluoride levels even lower.

136
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

PRITZEL: unit vessels? What were the locations of cracks on

I do have two other comments. There are contamination the vessels? What percentage of welds are checked?
problems, but, as mentioned, if the proper treating equip- What is the acceptable upper limit of weld hardness?
ment is used you should not have any problems. Also, you
must do routine testing to make sure that you do not have a FISCHER:
high fluoride content or that you are not getting a fluoride Our HF alkylation unit has been operating since 1966.
breakthrough. To date, we have not encountered cracking in any unit
vessels. If we were to encounter weld cracks, the area
DAVID SMITH (ALCOA): inspected would range from a representative sample to
With proper design and operation of the butane de- 100% of the vessel. The maximum acceptable weld hard-
fluorinators on your alkylation unit, you can obtain as low ness for carbon steel is 200 Brinell.
as 1 ppm by weight HF in the effluent. Then, as far as the
feed stream to a Butamer unit, you can, through selective KNEPPER:
adsorbents, molecular sieves, or activated aluminas, re- I do not have an answer on the cracking question. Our
move all those other contaminants. These contaminants biggest problem has been hydrogen blistering. We have
include HF that has come through, oxygenates, or mer- addressed this by using Lukins fineline steel. For weld
captans, or any of the other contaminants of concern. Thus, hardness, we specify a maximum of 200 Brinnell hardness
you can protect that Butamer catalyst. on HF acid service per UOP’s piping specifications. On our
recent construction projects, we have stress-relieved all
new piping and checked the Brinnell hardness of five per-
QUESTION 21: cent of the welds before and after heat treatment to achieve
What has been the experience with cracking of HF the Brinnell hardness specification.

137
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

PAZMANYI: WILLIAMS:
After four years of operating our equipment we did not I also agree with the comments of Mr. McClung. South-
find cracks. There are radial type cracks at the inlet and western relies on stream analyzers for monitoring the
outlet flange surfaces of our acid circulating pumps which moisture content of reformer feed streams. We have stan-
were repaired by deposit welding. The frequency of welds dardized on the Panametrics System One which utilizes an
checking lies between 50 to 100 percent. We feel that 200 aluminum oxide cell. A sample is taken at the feed pump
HB is acceptable as the upper limit of weld hardness. discharge. The slip stream is then routed to the analyzer
and returned to the pump suction. The sample tap is located
BRUCE SCOTT (Bruce Scott, Inc.): on a vertical piping segment.
There are two other things as well as the metallurgy that
will have a strong influence on cracking. A little bit of GRANT:
oxygen with HF promotes cracking and corrosion fero- We also at several of our facilities have in-line Pana-
ciously. There is some evidence that suggests arsenic in the metrics System 3000 moisture probes to determine the
acid will also promote cracking. water content in reformer feed. Also in the laboratory the
method is Karl-Fischer. Another method that one refiner
FRED COLLIER (MAPCO PETROLEUM Inc.): utilizes is a coulometric titration method. The test takes
During a turnaround of our HF Alkylation unit last May, about three to four minutes, but the sample must be col-
we found some severe cracking in our acid settler. The lected under nitrogen blanket.
cracks were found using ultrasonic inspection with 45° and
65° shear waves. One of the cracks, located at a 6° nozzle in FREDERICKSON:
the vessel head, was 3° long and 1° deep. Other cracks were In one refinery we are using an Ondyne analyzer to
found at attachment welds for baffles and trays. All of the measure water content in the reformer liquid feed. The
cracks were repaired and the entire vessel was stress instrument is mounted on a once through slip stream
relieved. around a control valve. If the instrument is given proper
attention, the readings are reliable. In other locations we
rely on monitoring the moisture content of the recycle gas.

C. Reforming DANIELS:
Normally we do not analyze the reformer feed for water.
Rather, we monitor the recycle gas using a DuPont
Feed analyzer.

QUESTION 22: Process

What sampling method is used for analyzing re-
former feed for water content? How is it analyzed?
QUESTION 23:
McCLUNG: When using liquid nitrogen as a coolant, what is
Sampling reformer feed directly through bomb or other the maximum cool-down rate to prevent damage to
vessel for measurement of less than 50 ppm of water is the catalyst and/or unit metallurgy? What are the
virtually impossible. An in situ probe with on stream effects of excessive cool-down rates?
analyzer is the best alternative. Panametric is one manu-
facturer which makes such a device for in situ measure- McCLUNG:
ment of moisture in reformer feed. Liquid nitrogen is commonly used as a supplemental
cooling source, for example, after catalyst beds have de-
MOYSE: creased to 300 to 400° F. I presume that the user who
I agree with Mr. McClung. When I was directly involved submitted this question is using the nitrogen after re-
in refining myself, we used to bubble recycle gas through generation, but before reactivation. Normally a liquid ni-
ethylene glycol and titrate it using Karl-Fischer. We used trogen source is vaporized and the nitrogen heated before
that on a routine basis and also in order to calibrate our being used for cooling. Use of liquid nitrogen directly on
measurement instruments, and also check those. At that catalyst or equipment is not commonly practiced nor is it
time they were Beckman. I am told that some refiners still advisable. Excessive cool-down rates can cause uneven
use this technique on a regular basis. cooling at exchanger pipe connections, damage to pipe
However, with the advent of instruments, I agree with supports or anchors, and a variety of undesirable mechan-
Mr. McClung, Panametric is one. I am told that DuPont also ical contractions. So it is mainly an equipment issue about
has one. That certainly reduces the risk of contamination which to be concerned. The catalyst is less of an issue as
when you are looking for such low levels. I am also told that long as the equipment is cooled at the proper rate.
in the portable analyzer field, there is also an instrument
made by Shaw which I am told is very good. However, it MOYSE:
requires frequent servicing, possibly due to chloride Again I agree with Mr. McClung. We located one refiner
corrosion. in the Houston area who actually uses liquid nitrogen as a

138
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

coolant on a regular basis, but we only have half an answer. tures immediately downstream of the nitrogen injection
They cool down by conventional means until the reactor is point and also at the reactor. Our system features micro-
at about 200 to 250° F. At this point they then utilize liquid processor control with alarms and automatic shutdowns as
nitrogen for the additional cooling. The flow rate is 700,000 maximum suggested cool-down rates are reached. We have
standard cubic feet per hour, and the temperature of that observed cool-down times reduced from three days to under
nitrogen is 35 to 40° F. This procedure works very well. 24 hours with the use of liquid nitrogen. We have the
Obviously due to the safe nature of this procedure, nothing facility to provide various flow rates and a wide range of
is known about the effect of maximizing cool-down rates temperatures and pressures with the equipment that we
and the possible negative effects on either catalyst or own and operate.
metallurgy.
RAUL OTERO-MARTINEZ (Atlas Processing Company):
WILLIAMS: Mr. Williams, on these reactors that you have cooled
We routinely use liquid nitrogen for accelerating reactor down using liquid nitrogen, did they have any cladding? If
cooling operations. Our metallurgy dictates a minimum they did have cladding on them, is still 50 to 75° F per hour
mix temperature of 40° F, and a nominal cool-down rate for cool-down acceptable?
us is 50 to 75° F per hour. As mentioned, the consequences
of excessive cool-down rates can be serious. Very low mix WILLIAMS:
temperatures may result in localized embrittlement, stress The 50 to 75° F per hour cool-down rate is a fairly nom-
cracking of pipelines, or failure of welded joints. inal figure for us. I may not have emphasized this ade-
Also, excessively fast cool-down rates, in addition to the quately, but one of the critical parameters we watch is
problems just mentioned, can lead to catalyst damage due reactor bed temperature differential and overall reactor
to thermal stress. We have found it helpful to monitor temperature differential. For example, on our gas-oil HDS
temperature differentials between individual catalyst beds reactor, we apply the 50 to 75° F per hour cool-down rate
whenever instrumentation is present. If the instrumen- limit, but since we typically maintain a 50° F maximum
tation is not present, then overall reactor temperature bed-to-bed differential and a 150° F overall reactor differ-
differentials can be used as a guideline to controlling cool- ential during the cool-down process, the actual rate at
down rates. which the reactor is cooled tends to fall somewhat short of
50 to 75° F per hour.
CHARLES POWELL (UCISCO - Union Carbide Indus-
trial Services Co.):
I think there might have been a little bit of mis- QUESTION 24: .
understanding here. The use of liquid nitrogen to assist Are new or improved continuous regeneration
catalyst cool-down employs a stainless steel injection quill, reforming catalysts available? Is there any com-
which UCISCO can provide assistance in designing. The mercial experience?
liquid nitrogen is delivered from our mobile nitrogen
pumpers, through a NICOOLSM control skid, and through McCLUNG:
stainless steel flex hoses leading to the injection quill. Rather than exercise crass commercialism here, I am
Redundant control points utilize upstream recycle tem- going to try to be an equal opportunity vendor. Continuous
perature, downstream mix temperature as well as multiple regeneration catalysts are being currently offered by UOP
bed temperatures. This differs from the method where the IFP, Engelhard and Criterion. I will leave to the represent-
nitrogen is heated on the truck prior to cooling the catalyst, atives of these companies in the audience, the statements of
in that the recycle stream is used as the heating source, commercial experience.
significantly reducing the amount of nitrogen required to
cool the catalyst. UCISCO also uses computer modeling to PAZMANYI:
assist with predicted cool-down rates and nitrogen con- Maybe our newest CCR Platformer, which was suc-
sumption. Usually the process is used by refiners where the cessfully started up in September this year, has not the
catalyst must be removed from the reactors, which is on the newest catalyst, but we are fully satisfied with R34
critical path of the turnaround. performance.

JOHN CANCIENNE (Star Enterprise): M. DEAN EDGAR (Criterion Catalyst Company):

Mr. Williams, how much time does this save you on the Yes, we do have a spherical reforming catalyst for the
shutdown by using liquid nitrogen? CCR process. It is called PS10. It has been in commercial
operation since May at one of the Gulf Coast locations. We
WILLIAMS: tried to formulate a catalyst that ‘would offer enhanced
While this experience is on a group of HDS reactors, not surface area stability for repeated regeneration, better
necessarily reformers, we have found that the cooling oper- chloride retention, and reduced attrition losses. So far, the
ation can be accomplished in 10 to 16 hours, versus 24 to 48 commercial experience has borne out our desires.
hours using a more conventional approach.
ROGER PEER (UOP):
UDAY PAREKH (Air Products and Chemicals, Inc.): As an historical note, I would comment that there have
I have just a few words on what Air Products does for this. been a number of catalysts used in the 100 CCR Platform-
application. To prevent excessive cool-down rates, we con- era. We started with R-16 in 1971, which was followed by
trol liquid nitrogen injection rates by monitoring tempera- R-30, R-22, and R-32. The most recent catalyst that Mr.

139
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

Pazmanyi referred to, R-34, was commercialized in 1988. generation due to platinum agglomeration or sulfate
We have 13 units currently running on R-34 and several problems
more to be streamed in the near future. We also have
another formulation in development right now that we plan McCLUNG:
to commercialize on a trial basis by the end of this year. It Our recommendation regarding any kind of exchanger
should be in the market next year. It will have improved cleaning or equipment cleaning really depends on the
performance, in particular increased surface area stability historical performance of the naphtha desulfurization unit
and better chloride retention. in front of the reformer. Engelhard sometimes recommends
chemically cleaning equipment as a minimum prior to the
use of skewed platinum/rhenium catalysts. Some units
QUESTION 25:
which have had a poor history of properly desulfurizing the
feed have actually required mechanically cleaning the
Have refiners found it necessary to chemically
exchangers.
clean or acidize either exchangers or preheat furnace
in changing out to skewed platinum/rhenium PAZMANYI:
catalyst? In the last 15 years, we have changed three semi-
regenerative Platformers to platinum/rhenium catalysts.
DANIELS: We did not acidize either the heat exchangers or the heaters
We have not done so. However, we had maintained a very at this change. Heater decoking was done using a steam/air
low sulfur content in our reformer feed and did not have a mix with tube wall temperatures of 580-600° C
large content of scale built up in the system. We have sulfur (1075-1100° F). Where we had the possibility for nitrogen
traps on the recycle loops of all our units, and we carefully circulation, we did it at 20 bars (290 psi). At the top of the
remove as much sulfur as possible during the regeneration catalyst in the first reactor, we used temporarily a 100
reactivation process. We use a caustic or soda ash scrub on millimeter thick insulation material. After circulating ni-
the recycle hydrogen leading into a regeneration. If we trogen for a few hours, a significant quantity of dust and
know we have a lot of sulfur present, we do a double corrosion products collected in this insulation material. We
reactivation. removed this insulation material before introducing
hydrogen into the unit and we have never experienced delta
FISCHER: P problems in the Platformers.
We have skewed platinum/rhenium catalyst in the ma-
jority of our reformers, and have not chemically cleaned or J. B. RODDEY (Roddey Engineering Services):
acidized either exchangers or preheat furnaces prior to There is no reformer free of corrosion products from
switching to these catalysts. Acid cleaning equipment is a upstream equipment, bar none. So, with time, the ex-
safeguard if there are signs of sulfur laydown prior to changers and the heater will foul with corrosion products
catalyst changeout. Although we did not chemically clean even though there are not H2S upsets in the desulfurizer.
or acidize our reformers prior to switching to a skewed This will cause you problems with catalyst regeneration,
catalyst, since we had not had sulfur related problems, we and the procedures you use obviously will determine
did include a sulfur scrub procedure prior to catalyst whether you can put up with it or not. Eventually you have
changeout. to clean the feed effluent exchangers and the internals of
This involves a high temperature electrolytic hydrogen the reactor.
scrub to remove sulfur as H2S with a temperature of about
950° F. During one sulfur removal step with electrolytic WAYNE CHAMBERS (Chambers Refining Tech-
hydrogen to remove sulfur from the unit, we estimated that nology Inc.):
380 pounds of sulfur was removed from a 30,000 barrel per I have a couple of comments. First of all, when we talked
day reformer following a one-year cycle. about continuous regeneration, there are some units that
have swing reactors and regenerate that way on a con-
FREDERICKSON: tinuous basis. Those units generally have extruded cata-
Most of our reformers switched to skewed platinum/ lyst. If you switch to the new catalyst in those types of beds,
rhenium reforming catalyst a decade ago. We recognized those balls like to run. You have trouble holding them in the
that good sulfur control was essential to take full advan- beds, so you will have to look at that before you make such a
tage of this type of catalyst. However, we did not find it switch.
necessary to chemically clean or acidize exchangers or On the second point, to agree with what was said earlier,
preheat furnaces to achieve satisfactory performance. This if you have a platinum reformer in which you have tolerated
was primarily due to the fact that we were already utilizing higher sulfur feeds, you do also build up sulfur in the
bimetallic catalyst and were controlling sulfur to very low insulation. Some units, in switching to the new bimetallic
levels to get maximum benefit from those catalysts. catalysts, have had problems with the sulfur coming out of
the lining.
HAMILTON:
I would confirm what everyone else has said. However, I
would like to add that even with good pretreater operation, QUESTION 26:
if the system has not been cleaned for many years, it should Does anyone have any information on how ben-
be given strong consideration. Removal of sulfur scale will zene and aromatics levels in the reformate change
prevent disappointingly short cycle lengths following re- with the age of the catalyst?

140
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

REFORMATE AROMATICS CONTENT VS CAT LIFE

Refiner: D.K.V. Hungary

PAZMANYI: question could be interpreted in two ways. One could be

In our Platformer units, we check the Platformate com- related to the total run, and the other one could be on the
position weekly. The aromatic content and the octane num- cycle. So the comments that I have are just qualitative and
ber are the parameters we try to keep constant by raising not quantitative. We believe, as the catalyst in the system
reactor temperatures stepwise. Our experience shows: ages and you raise temperature to maintain activity, there
- 80-180° C (175-360° F) fraction operations resulted in will be a tendency towards more cracking. This will in-
the first cycle increasing aromatics, in the second and crease the concentration of aromatics, especially the lighter
third cycle decreasing aromatics content, but every species. The concentration of benzene in relation to other
time it was within the 52.56% range. (See graph aromatics will also increase slightly as a result of this
above). trend.
- 60-110° C (140-230° F) fraction operations resulted in
fairly constant aromatic content of 40%, and benzene McCLUNG:
content of 16-18%. Speaking for semi-regenerative units, and also using the
- 100-150° C (212-300° F) fraction operations also re- interpretation of over the useful life of the catalyst, I really
sulted in constant Platformate aromatics at 92 Re- would not expect with the same metal dispersion, proper
search Octane Number. In our opinion, there is no chloride level retention, that you would end up with sig-
significant change, except an expected slight decrease nificantly different aromatics distribution or quantity.
in the aromatic production capability of Platforming However, if the catalyst is used past its normal useful life,
bimetallic catalyst in the analyzed period of time. or you do not get good platinum dispersion, or your chloride
level is not maintained as it should be, reactions such as
dealkylation could produce more benzene than the original
MOYSE: catalyst. The dealkylation ofheavier aromatics is a thermal
When we met here Tuesday, we discovered that this effect, not a catalytic one.

141
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

DANIELS: KNEPPER:
My answer is related to age during the cycle. Our labora- I would not say that the chemical we are using will
tory studies are showing that benzene levels in reformate necessarily be called environmentally acceptable, but we
increase by approximately 35% over a 50° F rise. At our are utilizing trichloroethylene as a chlorination medium
severities, we would expect the aromatics to decrease for our reforming catalyst. The trichloroethylene is de-
slightly. Feedstock and unit conditions will, of course, have livered to the unit in bulk skids. It is then transferred via
a bearing on the absolute benzene production. gravity flow to the injection pot as required. A small air-
driven injection pump takes suction from the injection pot
FISCHER: and discharges the trichlorethylene to the inlet of the
We interpreted this question to apply to the overall age of reformer feed effluent exchangers. The bulk delivery sys-
the catalyst throughout its life. Based on our analysis over tem greatly reduces operator exposure to the chloriding
the past five cycles, using a skewed platinum/rhenium agent. In the past, we used individual drums and the
catalyst, both benzene and aromatics levels varied rela- operators had to handle them. If the operators for some
tively independently of catalyst age. We felt that they were reason are required to handle trichloroethylene, they use
more a function of feed quality. Our benzene in the re- suitable gear, such as cartridge respirators, rubber gloves,
formate varied between 2.5 and 4.5 vol % throughout each goggles, etc.
cycle during normal operation.

FREDERICKSON: McCLUNG:
Our data show that for most of the run, aromatics and Engelhard recommends injection of chlorinated hydro-
benzene are constant for a given octane and a given feed. At carbon into the naphtha feed while it is still liquid. This is
higher end of run temperature, however, the data show that usually a closed system. There is good news and bad news
about 10% more benzene is produced. on the environmental area. The good news is that dichloro
compounds are the least offensive; 1,2-dichloropropene,
ART SUCHANEK (Criterion Catalyst Company L.P.): 1,1-dichloropropane, and 1,1-dichloroethane. The bad news
Adding to what Mr. McClung and Mr. Moyse said, cata- is that they are not commercially available.
lyst aging entails higher temperatures as you go towards
the end of run. One of the things that also goes on as you PRITZEL:
operate towards the end of run is the possibility of dropping The available compounds of which we are aware are
your throughput. When you do that, you have to look at perchloroethylene, trichloroethylene, and carbon tetra-
your unit design. You might have some problems with heat chloride. These compounds are injected into the feed.
flux, and you will end up cracking some of the C5 and C6
ring naphthenes. When you do that, your benzene and
aromatics yield will drop significantly. HUNTER:
We have been using 1,1,1-trichloroethane, and we have
JOHN G. DEARWATER (Profimatics, Inc.): primarily been concerned about the industrial hygiene
When processing a full boiling range naphtha at high aspects of its use. We do not use a bulk facility such as Mr.
severity in a typical semi-regenerative unit, the benzene Knepper described, but we have eliminated open transfers
and the reformate will drop slightly from start of run to end to our small feed pots. We are using nitrogen driven drum
of run. You can expect a gross reduction of 0.2 to 0.5% on pumps, and have improved the loading procedures and
volume basis. We typically answer these questions using a operator housekeeping. The feed system is completely
kinetic model of the reforming process that includes a sealed, and we use a diaphragm pump to feed into the
catalyst deactivation function. reformer charge pump suction. Our personnel monitoring
has shown very low levels of exposure to this compound
DAVID PIASECKI (UOP): with this system.
From the perspective of catalyst aging in a CCR Plat-
forming unit, we have tested commercially aged R-32, and
fresh R-32 catalysts in our pilot plants. We do not observe a
relationship between the benzene or aromatics content of
the reformate and the age of the catalyst. These pilot plant
results agree with the commercial operating data which
have demonstrated that the CCR Platforming yield struc-
ture is constant throughout the life of a catalyst load,
provided metal dispersion is maintained and there is no
irreversible poisoning.

QUESTION 27:
What environmentally acceptable compounds are
currently available or being considered as a chlorin-
ation medium for reforming catalysts and how are
they being injected?

142
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

GRANT: FISCHER:
Typically we use a mixture of one part trichloroethane We recently built a semi-regenerative reformer in Puerto
and four parts methanol, mixed with reformate, and inject Rico. Its design was 200 psi average reactor pressure. How-
it upstream of reactor number one. We have also heard that ever, this was a motor fuel reformer.
perchloroethylene is used and is injected at a rate of 1 to 3
ppm of fresh feed well upstream of the reactors for better McCLUNG:
mixing. Based upon OSHA regulations, over an eight hour The lowest pressure semi-regenerative unit that Engel-
exposure period, perchloroethylene has a weighted average hard has licensed operates at 130 psi separator, has a cycle
threshold limit value of 25 ppm, compared to 350 ppm for length of 8 to 12 months, and was designed for 99 octane.
trichloroethane.
MOYSE:
FREDERICKSON: We did a little digging here, too, and what we came up
Most of our reformers currently use trichloroethane. It is with is essentially the same as you have already heard.
used either straight or diluted with reformate, and is There are reformers designed to operate down below 200
pumped into the system from a small measuring pot using a psi. Press&-e, of course, will affect the aromatics yield, and
positive displacement pump. the dehydrocyclization of paraffins, both of which are
favored by ower pressure. With regard to cycle length, you
FISCHER: will probably have to get that from the process licensor, but
Our refineries also predominantly use trichloroethane we believe somewhere in the order of 6-8 months.
mixed with reformate for chloriding the catalyst, with
injection via metering pumps into the reactor train. With PAZMANYI:
regard to environmental considerations, TCE has an ozone We use our Platforming units for BT and TX production.
depletion factor of 0.1. We are evaluating perchloroethane In both units the reactor pressure is in the range of 20-21
for use in chloriding. PCE is a stable high flash material bars (290-305 psi). The average cycle length is 12-15 and
with an ozone depletion factor of only 0.01. 14-18 months, respectively. The average yield for BT pro-
duction is 78 to 82% (benzene 10-12%, toluene 14.15%,
DAVID WATERFIELD (BP Oil Company): xylenes 1.5-2%), and in the case of TX production the
We standardized on 1,1,1-trichloroethane a few years ago overall yield is 81.84% (benzene 1-2%, toluene 15.17%,
based on occupational health considerations. However, just xylenes 24-25%).
recently, we have had comments from various air pollution
authorities throughout the world, particularly in Germany, WILLIAMS:
about the ozone depletion effects of 1,1,1-trichloroethane. The separator pressure target for our BTX semi-
regenerative unit is 240 psig. A typical cycle length for us
QUESTION 28: can be from 8 to 12 months depending upon reactor severity
Methyl chloroform and carbon tetrachloride were and product specification requirements.
among the ozone depleting chemicals added to the
list of substances that are subject to an excise tax to JOHN G. DEARWATER (Profimatics, Inc.):
be collected by the Internal Revenue Service. Does Some refiners are running with the product separator at
any refiner have experience filing for exemptions 150 to 175 psig on BTX reformers. Feed composition, end
from the EPA and the IRS on the basis that CCL4 and point, space velocity, hydrogen partial pressure, and bed
methyl chloroform are converted to HCL? temperatures all significantly impact catalyst deacti-
vation, and have to be considered when estimating cycle
FREDERICKSON: length.
It is our understanding that the widespread use of
chlorinated hydrocarbons such as carbon tetrachloride and EDWARD JUNO (Pennzoil Company):
trichlorethane will be totally phased out by the year 2000. I would like to poll the panel. At what hydrogen recycle to
If the chemicals will be consumed as reactants; i.e., trans- oil ratios were the low pressure semi-regenerative units
formed, we hope to obtain “transformation” allowances. running?
However, we have not yet done anything about this.
DANIELS:
Our hydrogen recycle to oil ratio is 3.0 to 3.5.
QUESTION 29:
What is the minimum reactor pressure with which McCLUNG:
catalytic reformers are designed/operated for BTX The one that I was referring to is in a place of low energy
production with bi-metallic catalyst in semi- cost, so the recycle rate was designed for 11 to 1.
regenerative mode of operation? What will be the
cycle length?
QUESTION 30:
DANIELS: What is the minimum temperature required for
We do not operate with BTX feed, but a C6 to 380° F oxychlorination of reformer catalyst? What is the
naphtha material. Our flash drum is 140 psig, and we get 9 essential difference to the catalyst during oxy-
to 12 month cycles, 100 octane, and a space velocity of about chlorination and normal chloride addition? Should
1.5 WHSV. chloride injection continue or discontinue if HCL

143
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

content in recycle gas goes up due to chloride In reference to water upsets, chloride injection should not
washout? be decreased. The source of the water should be located and
eliminated. The reactor inlet temperature should be de-
FISCHER: creased to minimize coke laydown. Failure to reduce tem-
To begin, the definition of oxychlorination, also known as peratures could result in a substantially shorter catalyst
rejuvenation, is the redispersion of platinum after carbon life. Once HCl levels in the recycle gas begin to decrease,
bum in an oxygen chloride environment. We conduct our the chloride injection should be increased over normal
oxygen chloride soak at an average catalyst bed tempera- levels. The amount of the increase and the duration of the
ture of 950° F. During this step the metals are highly higher injection rate are dependent on the extent of the
mobile and become dispersed on the catalyst. In order to upset. The overall goal is to minimize catalyst damage and
maintain the recommended 1% by weight chloride on the to restore the proper chloride level.
catalyst, a chloriding agent is added to replenish what is
being stripped off with the circulating gas. This maintains McCLUNG:
a balance between the metal and acid sites on the catalyst Regarding the last portion of this question, should chlo-
surface. ride injection be discontinued if the HCl content of the
Following this oxidation step, a reduction process using recycle gas goes up? The answer is generally no. If chlorides
electrolytic hydrogen anchors the metals to the catalyst are obviously being washed off the catalyst, a proper reac-
surface. Chloride addition is resumed after unit start-up to tion is to increase the injection rate.
maintain a proper metal-acid site equilibrium. We base our
chloride injection rates, both during reactivation of the PRITZEL:
catalyst and during normal operation, more upon cal- The only thing that I have to add concerns the minimum
culations of maintaining chloride level based on catalyst temperature. We see a minimum temperature of 950° F.
weight than on HCl content in the recirculating gas.
Usually 1 to 3 ppm by weight of chloride in the feed would M. DEAN EDGAR (Criterion Catalyst Company):
be required to maintain a 1% chloride on the catalyst. I would like to add some clarification on where the tem-
While we routinely monitor HCl content of the recir- perature is measured. A lot of people regard it as the inlet
culating gas, we would not use this information as the sole temperature. Yet what we have seen in commercial oper-
source of telling us if we were properly chlorided. We feel ation is that the outlet temperatures can lag quite a bit
that a close tracking of light end ratios provides a clearer behind the inlet temperatures. You may have a 950° F
picture of the chloride balance. If we detect high HCl in the temperature on your inlet, and you will be down around
recycle gas, we would then check for increased light end 920 to 925° F on the outlet. What we should be trying to
ratios due to more hydrocracking acid site-induced reac- look at is the bed temperature. Normally we recommend a
tions. If a catalyst sample were possible, it would confirm 950° F catalyst bed temperature.
our suspicion of an over-chlorided condition. At this point I have just recently run across some work from a research
we would lower, or perhaps temporarily discontinue, chlo- institute where they have been looking at the most favor-
ride injection until the condition is resolved. able temperatures for platinum redispersion. In that par-
ticular result, it was 1010° F. That is a bit higher than we
DANIELS: normally think of. I believe our 950° F recommendation is
I would like to add that we prefer to keep the catalyst bed probably still the best one for general operation.
temperatures above 900° F during our oxychlorination
step. After a chloride washout, our procedures are that we TONY CHODOROWSKI (Pace Consultants Inc.):
will not put the chloride back in until the system is dried Would the panel like to comment on rhenium dis-
out to less than 50 ppm. tribution? We have heard a lot of comments on platinum
redistribution, but most of the catalysts we are talking
KNEPPER: about are skewed catalysts where the rhenium is also a
I concur regarding the temperatures mentioned. The very important factor.
other part of the question is the difference between oxy-
chlorination and normal chloride injection. Normal chlo- McCLUNG:
ride injection relates to catalyst chloriding and platinum Let me add to what I said a while ago. What Mr. Edgar
dispersion. During normal operation, the catalyst injection said is true about going to higher temperatures even
serves to maintain the proper catalyst chloride level. Oxy- though it is not normally recommended. The redispersion
chlorination serves to adjust the catalyst chloride level with of platinum is really a function of oxygen partial pressure,
HCl and redisperse the platinum with chlorine. time of exposure to it, and temperature. You could actually
The concentration of chloride and chlorine is dictated by fully disperse a catalyst at 900° F if you waited long
the Deacon equilibrium reaction during oxychlorination. enough. So there are a variety of compatible circumstances
The equilibrium relationship is favored by oxygen partial for getting the proper platinum dispersion. To answer the
pressure. Increasing oxygen partial pressure favors the rhenium question, one of the things that has been studied
formation of chlorine. Therefore, maintaining the highest for at least the last 16 years, is what is the state of rhenium
feasible oxygen partial pressure will improve the platinum in a platinum/rhenium catalyst. Some of the earliest work
re-dispersion. In addition, some refiners have injected was done with carbon monoxide adsorption. Marvin John-
chlorine gas during oxychlorination to enhance platinum son of ARCO did the work. Some of the more prolific work
redispersion. lately has been done with even more sophisticated analyt-

144
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

ical techniques. There are at least five different theories on They usually have less than 0.1 ppm. Their guess would be
what the status of rhenium is. Therefore, the answer to Mr. this is due to such leaks. Higher sulfur content can be
Chodorowski’s question depends on which theory you use. caused by HsS recombination at certain conditions.
You would think that we would know more about it, but we Our experience shows that our Platformate contains less
only know more about our ignorance. also than 0.1 ppm of sulfur and our Platformer feed con-
My own personal opinion is that especially with high tains usually 0.2-0.4 ppm. The analysis of the sulfur com-
rhenium catalyst, the mobility of rhenium is influenced at pounds is disturbed by other contaminants, such as
different temperatures by the presence of moisture, and halogens.
that mobility has a relationship to the oxidation state of the
rhenium. So, there is a potential, as best as I understand it, MOYSE:
for actually physically moving the rhenium onto the plati- This is a three-part question, so there is a three-part
num under certain circumstances, which you definitely answer. We talked to several refiners about this, and we
want to avoid. obtained mixed results for leak detection methods. To give
you another perspective, some people tried oil soluble dyes,
CHARLES S. M CCOY (McCoy Consultants): but the results were inconclusive. Some other people con-
From a safety standpoint, let me note that these high firmed what Mr. Williams said. The radioactive tracer
oxygen contents can be dangerous if you still have burning technique is very good. Not only can it tell you if you have a
occurring. Do not try oxychlorination until you have leak, but it can also give you an estimate of the magnitude
burned out all those lumps of coke that are in your catalyst of it. That was very positive.
bed. With regard to anyone seeing sulfur in reformate, the
answer is most definitely yes, anything from a few tenths to
several ppm.
QUESTION 31: For part three, it is not necessarily caused by leaks in the
What methods are effective in detecting leaks in feed effluent exchanger. It could also be due to problems in
feed/effluent exchangers? Our feed normally con- the stripper or the absorber. We have seen this just recently
tains less than 0.3 ppmw sulfur, yet we found on a West Coast refinery.
0.22-0.15 ppmw sulfur in the reformate. Have others On a final point, I would like to mention the difficulty in
experienced sulfur in reformate? At what levels? Was measuring tenths of a ppm of sulfur. We have developed our
it necessarily caused by feed/effluent leaks? own inhouse method which is a modification of the Raney
Nickel method, and it utilizes a calorimetric end point of
WILLIAMS: methylene blue. I would also like to point out that you need
As far as the leak detection aspect of this question is two sets of apparatus: one for the feed, and one for the
concerned, we have had extremely good success using the product.
radio isotope tracer method. There are several companies
in the business of providing this service. The bulk of our McCLUNG:
experience has been with Tru-Tec out of Pasadena, Texas. For vapor or vapor-liquid streams that can be properly
We have been quite pleased with the results. sampled around your feed effluent exchangers, the use of
Concerning the second part of the question, we have not gas chromatography to test for key compounds is effective.
experienced sulfur contamination in reformate, at least to This technique is effective for large leaks using such com-
the extent we have been able to measure. Platinum catalyst pounds as total aromatics, cyclohexane, benzene, or tol-
is very active, and any sulfur contained in the reformer feed uene. When I say large leaks, I mean numbers such as 5 to
should either be deposited on the catalyst or converted to 10%.
hydrogen sulfide. Either a feed effluent exchanger leak or
product contamination in the tank field appear to be the HAMILTON:
most likely sources of sulfur in reformate. In addition to what has been stated so far, another
method would be the use of helium injection into the feed,
PRITZEL: and determination with a mass spectrometer. Careful sam-
Silicon hexafluoride has been used as a tracer. It is pling of the reactor outlet and the effluent out of the ex-
injected upstream of a suspected leaking exchanger on the changer is another method applied. In the foregoing ques-
high pressure side. An estimate is made of the time re- tion though, a feed/effluent exchanger leak in a naphtha
quired for the tracer to reach the low pressure side in the hydrotreater would take a reformer feedstock with greater
event of a leak. The downstream low pressure side is than 0.3 ppm sulfur. A leak in the reformer feed/effluent
sampled to determine if any of the tracer has prematurely exchanger would be indicated by low octane reformate for a
passed through the exchanger to the wrong side. given temperature of operation, low conversion of naph-
I agree with Mr. Williams on the last part of the question. thenes, as mentioned before, especially cyclohexane, and
Sulfur should not be present in the reformate unless the depending upon the size of the leak, a low hydrogen make.
feed effluent exchangers are leaking.
FREDERICKSON:
PAZMANYI: Chevron uses different methods to check for feed effluent
According to the opinion of other refiners from Europe, a exchanger leaks depending on whether the exchangers are
higher content of naphthenes in the effluent will indicate in the first stage (naphtha hydrotreater), or in the second
leakage. The sulfur content in the reformate is very high. stage (reformer). For the reformer, we trace the disap-

145
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

pearance of easily reformed compounds using gas chro- PAZMANYI:

matography. We feel that we can detect leaks down to about We check regularly the methylcyclopentane conversion
one percent bypass. in one of our Platformers because this conversion is charac-
For the naphtha hydrotreaters, we first try looking for teristic of the acidic effect of the catalyst. At the start of run,
sulfur content (after removing the H2 S) at the reactor the conversion reached 92 to 94% and at the end of run, it
outlet and at the effluent exchanger outlet. The alternative was decreased to 82 to 83%. The typical methyl-
to either method is to use radioactive tracer techniques. cyclopentane in the feedstock was about 8 to 9%. We had
During startups we occasionally see sulfur in reformate, some unit trials to check the Platforming operation at
and these incidents are often traced to contamination of higher feedstock methylcyclopentane content. Using the
product with sour stocks, rather than heat exchanger leaks. same operating parameters of reactor temperatures 480,
487, 495° C (900, 910 and 925° F), first reactor inlet pres-
FISCHER: sure 24.5 bar (355 psi), we obtained the following results:
I concur with Mr. Pazmanyi’s comments about the naph- FEED PLATFORMATE
thene content of reformate as a method to determine a leak MeCP% Benzene% MeCP%
in the feed effluent exchangers. Normally, naphthene levels 12.0 17.6 1.3
in the reformate are less than three volume percent. If you 16.8 21.2 1.5
should find levels much above five to six volume percent, it 25.0 26.5 2.5
could indicate a leak. We have not found any sulfur in our 35.0 31.1 4.1
reformate, but there would be a possibility, however, of
having sulfur in the reformate following a sulfur upset in The methylcyclopentane conversion varied between
the unit. This would not necessarily be due to a feed effluent 89-92%. For estimating the selectivity we took into account
exchanger leak. benzene production only from MeCP and cyclohexane. Ac-
cording to this, the selectivity is about 40-44%. The real
selectivity is slightly less, because benzene production
DANIELS:
from the feedstock C6 paraffin content takes place, too.
With regard to leak detection, we have also been suc-
cessful using helium and radioactive tracers.
DANIELS:
We think the published selectivities for normal hexane
JOHN P. MacLEAN (Texaco Refining & Marketing Inc.):
conversion are valid for isohexane. On the selectivity of
With regard to the last part of that question, many years
MCP conversion to benzene, it is difficult to measure be-
ago at a high pressure reformer, we were synthesizing light
cause you need to know how much benzene is being pro-
mercaptans. We had no leak. Some place, either in the last
duced from cyclohexane, C6 paraffins, and the dealkylation
reactor or in the fractionation tower, we were making light
of heavy aromatics. Radioactive tagged MCP could be used
mercaptans. It was sufficient to make our light Platformate
to determine the selectivity to benzene. Others have done
go Doctor positive. The conditions that caused this were
this; we have not. MCP selectivities of 60 to 80 mole percent
never able to be determined. We were never able to figure
were reported by Chevron Research, depending on the
out exactly what was causing it. It seemed to come and go at
system pressure, in the 1976 NPRA Annual Meeting.
random, but it was definitely there, and it was definitely
light mercaptans.
FREDERICKSON:
Our data show that isohexanes have lower selectivity to
JOHN G. DEARWATER (Profimatics, Inc.):
benzene than n-hexane. Selectivity for MCP to benzene is
We would like to point out that it is not unusual for
about 70% at 250 psig.
certain ring sulfur compounds like thiophene to go through
the HDS unit and the reformer virtually unscathed. 0.2
ppm by weight sounds like a high number, but it could be
feed dependent. It is unlikely that the feed effluent ex- QUESTION 33:
changer is the cause of this problem if no other feed effluent For CCR reformers that have changed catalyst
exchanger leak symptoms have been observed. on-line, how long has it taken and what percentage of
the inventory was added for complete changeout?

QUESTION 32: FREDERICKSON:

Are published selectivities for n-hexane con- We have been able to achieve over 300 catalyst cycles in
version also valid for the isohexanes? Also, what are our only CCR. With such long life and the need to inspect
the selectivities of methylcyclopentane to benzene at the reactor screens periodically, about every four years, we
the normal range of semi-regenerative operating have not had an incentive to change out catalyst on-line.
conditions?
ANGELO FURFARO (UOP):
McCLUNG: I would like to preface the answer by first stating that
Isohexanes do not convert with the same selectivity as UOP does not consider the on-line catalyst replacement, or
normal hexane. Selectivities of methylcyclopentane to “changeout on the fly” a substitute for scheduled main-
benzene, as an example, vary from approximately 65% to tenance turnarounds. UOP continues to recommend that
50% at 920° F and two pressures 200 and 350 psi. In other new units be inspected after two years of operation, and
words, at 200 psi, your selectivity would be 65%; at 350 psi, operating units should be shut down at two to four year
it would be 50% at that temperature. intervals for inspection and maintenance. Typically it is at

146
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

DANIELS:
We have been using liquid oxygen for reformer re-
generation for over 20 years. The use of liquid oxygen
allows the regeneration to be carried out at maximum
allowable pressure, shortening the regeneration time and
improving the flow distribution due to higher mass ve-
locities. In regard to safety, your lines must be properly
cleaned and maintained hydrocarbon-free. Another con-
sideration is the type of gaskets you are using. Be certain
you are using teflon gaskets in the oxygen system.

FISCHER:
I concur with Mr. Daniels’ remarks. The benefits of using
liquid oxygen during catalyst regeneration include a nor-
mally good ability to control the carbon burn, and a possible
reduction in the time required to regenerate the catalyst.
one of these turnarounds that the catalyst changeout would Since liquid oxygen can be supplied at a relatively high
occur. pressure, 180 to 240 psi, it may prove advantageous over
If the catalyst, however, has been providing expected low pressure compressed air. The use of liquid oxygen can
performance, replacement during a particular scheduled save significant time over regenerations using low pressure
shutdown is probably not practical, nor necessary. In such air.
an instance, an appropriate time to replace the current Some of the drawbacks involved in using liquid oxygen
catalyst load with fresh catalyst would fall between those include the possibility of losing your burn should you lose
scheduled turnarounds. The basic design of the UOP CCR oxygen supply. Also, deliveries from a reputable supplier
Platforming unit does allow catalyst changeout to take are a must. Additionally, costs associated with transpor-
place while maintaining Platforming unit throughput and tation, storage, and handling must also be addressed. The
operating severity. As a result, a number of refiners with safety precautions for handling liquid oxygen are the same
operating CCR Platforming units, have changed out cata- here as they are anywhere else.
lyst on-line under the supervision of UOP technical advi- Exposure to oil must be avoided in valves, pressure
sors. The time requirement to complete the on-line catalyst gauges, valve packing, etc. Connecting lines are normally
changeout is equal to the time required to complete one full flushed with a solvent to remove any trace amounts of oil,
catalyst regeneration cycle. This is dependent on each dirt, or metal shavings that may be present. Normally a
unit’s design. However, the time for a particular unit will be good oxygen supplier can give a good detailed checklist
comparable to that unit’s current catalyst circulation rate associated with each application. High pressure nitrogen
at the time of the changeout; i.e., longer than design turn- should also be available to purge the unit in an emergency.
over days if the required circulation rate is below unit Other safety precautions that may be overlooked include
design. the use of a probe to inject oxygen should there be a
The amount of catalyst used for on-line catalyst change- potential of coke in the line adjacent to the injection point.
out has ranged from 100 to 103% of the total amount of Additionally, a liquid oxygen leak onto an asphalt surface
catalyst originally loaded in the Platforming reactors and could result in an oxidation induced ignition source.
regeneration section. This amount has provided for a max-
imum five percent back-mixing of the previous catalyst McCLUNG:
load; i.e, the load leaving the reactors, remaining in the Use of liquid oxygen is a common way of accelerating the
new catalyst at the end of the catalyst changeout. UOP burn because it allows an elevated operating pressure
would recommend a minimum of 100% of the original compared to a normal process air system of 60 to 100 psi.
catalyst loading be made available at site plus a small There are really no extraordinary safety precautions nec-
contingency for handling losses, and to strive for that essary compared to using an air source such as a mobile
maximum five percent back-mixing. compressor source. I suggest making certain that the oxy-
Commercially, the amount of back-mixing has ranged gen can be metered and controlled with precision to keep
from two to five percent, with the lower number represen- from having too high catalyst bed temperatures. I know
tative of the CCR Platforming unit which loaded 103% of that one particular vendor has a four cylinder liquid pump
the catalyst inventory. Any refiner contemplating the re- that is staged so you can actually vary both the RPM and
placement of catalyst “on-the-fly” is advised to contact the number of cylinders. As a rule vendors who provide
UOP Platforming Technical Services in Des Plaines, or the liquid oxygen by truck also know the proper safety
refiner’s Continuing Services Representative. precautions.
Though I cannot really quantify this, I believe that the
QUESTION 34: higher mass flow rate that is accomplished during the use
What are the benefits and drawbacks of using of liquid oxygen does provide for a more uniform and
liquid oxygen injection during semi-regenerative complete bum. Therefore you are possibly getting, in sub-
catalyst regenerations? What are some of the im- sequent steps, a better reactivation of the total catalyst.
portant safety precautions necessary? The primary drawback, of course, is cost.

147
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

CHARLES S. M CCOY (McCoy Consultants): QUESTION 36:

A number of operators report that control of oxygen Has anyone debottlenecked a moderate or high
content is more difficult with pure oxygen, particularly pressure semi-regenerative reformer through the use
during the primary bum. Also, remember that the molec- of a plate heat exchanger in reactor charge/effluent
ular weight of the regeneration gas is now 44, not 30. Be service? How successful was it and have there been
sure your compressor can take it. any mechanical or operational problems?

J. B. RODDEY (Boddey Engineering Services): FREDERICKSON:

The use of liquid oxygen implies higher pressures for a We have one refinery that will be installing a plate heat
shorter burn time. When you go to higher pressures, you exchanger in reactor feed-effluent service next year. The
are more apt to have carryover of caustic into the com- piping tie-ins and double containment center pipe are
pressors. That has been my experience. Lower pressure installed and ready. The payout for the new exchanger was
regeneration, while taking longer, is less apt to have carry- based on more throughput and better yields at lower oper-
over of the circulating solution into the compressor. ating pressure.

ANTHONY VICKERS (UOP):

JOHN G. DEARWATER (Protimatics, Inc.): UOP has looked very carefully at the application of the
We agree with most of what the panelists have said. I will Packinox plate type heat exchangers, and we recommend
note one exception that was not mentioned in the area of their application for large, non-fouling services such as
safety. That was basically putting in injection via quills, Platforming feed effluent exchangers. We also specify them
and locating those quills downstream of the recycle com- for similar applications such as xylene isomerization pro-
pressor. Also, all fittings and pipe used for oxygen transport viding the duty is large enough to justify the application.
and injection must be properly cleaned, which was men- Let me list a few of the reasons we are enthusiastic about
tioned. However, we recommend that you have your oxygen the use of these heat exchangers. One Packinox heat ex-
supplier provide tubing lines all the way down to the changer can satisfy the heat exchange needs for a 30,000 to
injection quills. 40,000 barrel per day Platforming unit. This would require
The advantages mentioned with which we agree are the two to three maximum size bundles of the vertical shell in
faster regeneration, higher pressure, thus higher mass tube heat exchangers. Under these circumstances the
flow, better flow distribution in the reactors, which leads to Packinox heat exchanger has an overwhelming cost advan-
a more even catalyst burn. What was not mentioned was tage. The favorable economics this type of heat exchanger
more precise control of oxygen addition, better platinum provides causes UOP to specify a closer temperature when
dispersion because the oxidation step can be performed at we are using this type of heat exchanger.
higher pressure. There are a few other benefits I will touch on here. This
type of heat exchanger is very well suited when you are
looking for very low pressure drop applications and large
Mechanical bundle sizes. With a shell and tube type heat exchanger, we
were forced to compromise between pressure drop and the
unsupported length of tube you get by increasing the baffle
QUESTION 35: spacing. We have been concerned about getting low enough
Has any one experienced stress corrosion in a
pressure drop in these very large shell in tube bundles,
reformer recycle gas trim cooler? What is the reason
while at the same time avoiding vibration of the tubes.
for this type of corrosion?
While some new baffle arrangements allow this problem to
be circumvented they do cause some loss in thermal ef-
ficiency. With the Packinox type of plate exchanger, these
FREDERICKSON: concerns are eliminated, and you are able to run with lower
We have experienced stress corrosion cracking in admi- pressure drops than you would be able to achieve with shell
ralty reactor effluent coolers when the coolers were exposed in tube heat exchangers unless you are willing to adopt
to air during shutdowns. Ammonium chloride deposits baffle arrangements that compromise to the heat transfer.
have been the cause. We have converted to 70130 copper/ We also believe they have an advantage when you have a
nickel or carbon steel bundles to avoid the need to acid wash mixed phase feed such as you have on a Platforming com-
these exchangers. bined feed exchanger. The heat exchanger combines a very
well designed feed injection system with many fewer chan-
McCLUNG: nels, about 1/10 the number of channels to inject the feed
It is really strange because what Mr. Frederickson said into as you would be faced with tubes in the shell in tube
and what I am going to say are the same, but this instance is type heat exchanger. We have monitored very carefully the
not Chevron’s Engelhard knows of an instance where operation of these heat exchangers, and they are per-
admiralty trim coolers experienced extensive stress cor- forming above representation. This is at a time when a
rosion cracking due to the presence of ammonium chloride number of shell and tube heat exchangers are having
in an oxidizing atmosphere. A thorough washing can alle- trouble meeting representations, we believe, because of
viate this problem, but most reformer operators have con- feed distribution.
verted to carbon steel bundles in order to minimize the For revamps, we commonly specify them. They do offer
occurrence of this problem. the advantage of very low pressure drop. They require very

148
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

little plot space, and being able to use a close temperature had success in removing the shroud and using refrac-
approach, minimizes the need to change the feed heater. tory in its place? If so, what type of refractory was
used?
FRANCOIS REVERDY (Packinox):
My comments will be more of a review of the historical DANIELS:
use of the exchangers and what has happened to them. We do not think that alloy is the solution to shroud
Welded plate type heat exchangers manufactured by cracking, but rather the actual shroud design. We would
Packinox have been installed on 20 catalytic reforming not expect any refractory alone would provide satisfactory
units around the world. Another 10 are currently under insulation without a shroud. To solve a continuous shroud
fabrication. Of the 20 in operation, about one half are on problem on a spherical reactor we replaced it in kind with a
semi-regenerative units with design pressures ranging hot wall design.
between 225 and 650 psi. The others are on low pressure
CCR type units. PAZMANYI:
On the medium to high pressure semi-regenerative We use both external and internal insulated reactors.
units, the Packinox has typically been installed as part of a The material of the shell is 1 chrome, 0.5 molybdenum
debottlenecking project, replacing either a train of horizon- steel. For internal insulation protection, we use 13 chrome,
tal shell in tubes or vertical shell in tubes. A combination of 5 nickel ferritic type shroud. One time we changed the
lower pressure drop and increased heat recovery in the feed whole concrete lining, sometimes we have partially re-
effluent heat exchanger has typically been a key to increas- placed it. Twice we found distortion of the shroud in one of
ing unit throughput and reactor yield. our reactors, so we partially replaced this part. The in-
A good example is a case story at a Shell International sulation concrete was not damaged.
refinery with a reformer capacity of 15,000 barrels per day. I also have a question, if you allow me. Our external
The train of 12 horizontal shell in tubes had a total pres- insulated reactors are cladded internally with 18 chrome,
sure drop of 58 psi, and a hot approach of 108° F. The 10 nickel austenitic material. This year we found a lot of
capacity has been increased from 15,000 to 20,000 barrels cracks in big areas in one of our reactors. Crack lengths are
per day. The debottleneck included essentially the replace- about 3 to 25 millimeters, and depths are 1 to 1.5 mil-
ments of the 12 shell in tubes by a single shell Packinox, limeters. What can cause this chloride or polythionic at-
offering 19 psi total delta P, and 47° F hot end approach tack, or something like this?
under the increased throughput condition.
In terms of reliability, the downtime attributed to a BRUCH:
Packinox heat exchanger has historically been minimal. No answer.
From a statistics standpoint, Packinox exchangers have
logged so far over 200,000 operating hours in CRU’s around QUESTION 38:
the world, and a total recorded downtime attributed to the How many CCR units have seen cracking of reac-
Packinox, there is about 200 hours. This, by the way, would tor vessels and feed piping? Where? How extensive?
be the equivalent of eight hours of total downtime for one
year of operation. GRANT:
Finally, and perhaps most telling about the overall value We have experienced cracks in two CCR units. At one
of the plate type heat exchangers for reforming application, refinery, a single crack was found in the bottom head of the
is the endorsement of the Packinox technology by major number one reactor, This was believed to be a parent metal
process licensers and by refining companies. Shell Inter- fault, and not related to process conditions. At another
national, among others, has installed no less than six refinery, numerous cracks were located in the three stacked
Packinox welded plate heat exchangers on catalytic re- CCR reactors. Cracks were found in nozzle welds, vessel
forming units, both on semi-regenerative and CCR type welds, field welds, and a reduction zone weld. Cracks were
units. also found in the heat-affected zones around these welds.
DAVID N. WATERFIELD (BP Oil Company): There were hundreds of cracks in all. One crack in the
I just have one comment on experience. We have one heat-affected zone, near the field weld, measured 66 inches
which has been in operation for five or six years. That was long and 5/8 of an inch deep. Poor fabrication, metallurgy,
put in as part of a revamp to allow higher throughput and and ineffective post-weld heat treatment were determined
lower pressure operation. One concern was catalyst getting to be the cause of the cracking.
into the plates. Someone else mentioned that spherical FREDERICKSON:
catalysts move, and we have had experience of them getting Chevron’s one CCR has been operating for 11 years. We
through holes in center pipes of reactors. So you have to be recently checked for cracks at attachment welds (nozzles
careful to avoid that. and lugs), and did not find any problems. However, in some
The other thing that happened to our particular unit is of our 1-1/4 chrome hot wall semi-regenerative reforming
we managed to put some pretty heavy oil in there and reactors, we have found extensive attachment weld crack-
blocked it up a bit. However, we did recover quite well with ing. This has been reported in detail in previous NPRA
a reformate wash. Q&A sessions.
QUESTION 37: ANGELO FURFARO (UOP):
What alloy can be used to prevent cracking of UOP appreciates the mention of fixed bed reactors also
shrouds on internally insulated reactors? Has anyone since UOP understood the question to mean stacked reac-

149
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

tors. However, of all of UOP’s 99 operating CCR Platform- hydrocarbon suspected to be polycyclic compounds.
ing units, we have received reports of minor to major crack- Production of high purity benzene (maximum per-
ing in about 10% of those units. The first unit of this kind mitted non-aromatics is 50 ppmw) is by solvent
was commissioned in 1971. As already mentioned, the extraction using Sulfolane. Can these polycyclic
locations of the cracks vary from unit to unit. However, compounds affect the solvent selectivity by accumu-
primarily the location has been in the heat-affected zone, lating in lean solvent? Is there any method for quan-
(HAZ) around the large reactor nozzles. Reactor inlets, tifying the amount of heavy hydrocarbons in the
outlets, and manways have generally been affected from solvent? What are the methods employed to remove
the inside. Secondly, the small nozzles, such as unloading heavy hydrocarbons in the solvent?
nozzles and internal attachments have been affected to a
lesser extent. These cracks have been reported also in the GRANT:
HAZ area of the nozzle. We have no recent experience with the Sulfolane process.
The least affected areas appear to be the vessel skirts, However, based upon discussions with UOP the licensor of
internal head attachments, and vessel seams. Only one this process, if the amount of heavy hydrocarbons entering
report of cracking has been made with respect to an inter- with the feed is within specification and relatively stable,
nal reactor head attachments. There have been reports of they will exit in the raffinate stream, and will not accumu-
two vessel seam cracks. late in the lean solvent.
The extent to which these cracks have ranges has been If, on the other hand, the feed heavy hydrocarbons come
very wide, and has already been referenced from the panel. in slugs, they may accumulate in the lean solvent, but will
However, UOP have seen cracks, ranging from various only temporarily affect the solvent selectivity. UOP does
minor cracks that have to be detected only with the most recommend a method to quantify the amount of heavy
modem techniques of today, to those, as already described, hydrocarbons in the lean solvent, and they specify that the
as long as five or six feet in length and more than an inch in lean solvent should have no more than 1000 weight ppm
depth. In general though, grinding and rewelding have hydrocarbon and 0.5 weight percent water.
proved sufficient as a repair technique in conjunction with An original Sulfolane unit design includes a regenerator
proper post-weld heat treatment, (PWHT), temperatures, to remove heavy hydrocarbons. This regenerator operates
and time. under vacuum, and has a steam sparger located at the
The cause of this cracking has generally been termed bottom. The solvent and steam exit the regenerator over-
“creep embrittlement.” This is not considered a descriptive head, while the heavy hydrocarbons remain in the bottom
metallurgical term; however, this is the term that has been to be drained as required.
associated with cracking of low chrome steels. In most
cases, cracks have been traced to stress concentrations at JEAN COSYNS (Institut Francais du Petrole):
changes in the wall thickness, especially at the nozzle to Besides the problem of heavy hydrocarbons, we have also
shell connections, as well as to less than optimal fabrication observed that reformates coming from low pressure re-
procedures such as the use of a minimum ASME recom- formers contain olefins which have to be removed. In some
mended stress relief temperature, as opposed to a higher cases, the clay treatment is no longer efficient for chemical
more optimal temperature. grade aromatics. In order to solve this problem, IFP has
Forgings have on rare occasion been the source of cracks developed a selective hydrogenation process for production
when found to exhibit coarse grain microstructures or high of olefin-free aromatics. This process is cheaper and much
contaminant levels in the metal. As part of UOP’s con- more efficient than the old clay treatment process.
tinuing process to optimize vessel design, UOP has modi-
fied reactor designs to promote improvements in vessel
ANGELO FURFARO (UOP):
fabrication and metallurgy, and to incorporate industry
Much depends on how heavy those compounds are. A
findings resulting from investigations into the causes of the
percentage of heavies could indicate a problem in the up-
cracking.
stream fractionation. It is likely that there is some accumu-
Finally, the last point of the question was regarding
lation of these materials, which we presume to be aromatic,
cracks in reformer feed piping. Commercial feedback is
in the lean solvent, although some will leave with the
very limited, however, these problems have generally been
raffinate. There are always reentry losses of aromatics; i.e.,
due to construction, such as miscalculation of loads. The
the losses from the lean solvent to the top of the extractor, to
problem does not appear to be similar to those found in the the raffinate. It is likely that some equilibrium level would
reforming reactors.
be reached.
CHARLES S. M CCOY (McCoy Consultants): To determine trace hydrocarbons in Sulfolane, in the
Use of the most sophisticated inspection techniques is not lean solvent stream the analysis we would recommend is
needed. A pocket knife works well. Cracks are wide and UOP Method 831. UOP does not have any data by which to
usually shallow, assuming you catch them in a reasonable quantify the impact of these heavies on the solvent selec-
time. tivity, but we do not believe it to be very great. Accumu-
lation to the point where solvent selectivity is adversely
affected has never been reported as a problem.
D. Aromatics Extraction It is interesting to note that units have suffered hot oil
leaks of considerable size have been able to continue to
operate. The heavy material leaves with the raffinate, and
QUESTION 39: some in the solvent regenerator bottoms. In those situ-
Our reformate contains about 0.5 wt % heavy ations where there is a very large ingress ofheavies and the
150
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

unit solvent regenerator cannot keep up, the solvent must Department, has had a Sulfolane antioxidant under de-
then be regenerated ex situ by vacuum distillation. velopment. This product is called UOP SX. Laboratory
research in 1989 demonstrated “proof of principle” of UOP
JENNIFER BLASI (Mobil Oil Corporation): SX.
I have a followup question on solvent use for aromatic A field trial has also provided encouraging, but some-
extraction. I would like to know if the panel has any what inconclusive results. Continued development work is
experience with corrosion problems and corrosion control planned to understand the effectiveness of this new prod-
programs used in Sulfolane units? uct. Until development work is complete UOP SX is being
offered on a limited case-by-case basis. This product is also
PAZMANYI: offered as an adjunct to unit operations where oxygen
Our aromatic extraction plant is not a Sulfolane unit but ingress has been minimized.
we have a Tetra process. We have a corrosion program for
this process. Allow me to also mention that we use satu-
ration reactors in our Platforming units for saturating QUESTION 40:
olefins. They are integrated into the Platforming units, so Does anyone extract aromatics from diesel? Please
we do not have problems with olefins in the aromatic ex- give details.
traction plant.
HAMILTON:
KEN ABRAHAMS (Star Enterprise): We are not aware of anyone using an aromatics or solvent
We do not have any experience on the Sulfolane either. extraction unit to process diesel to reduce the aromatics
We do have corrosion problems that can show up in our content. Licensors, like Texaco and Exxon, have extensive
Tetra extraction unit if you get a lot ofchloride carrythrough. solvent extraction experience with lubricants and gas oils. I
would assume that such processes could be used for diesel,
TONY CHODOROWSKI (Pace Consultants Inc.): but I have no specific information as to the level of residual
Considering corrosion in a Sulfolane unit, I think this is a aromatics possible or the type of aromatics that would
fairly typical problem for some people. Mainly it is due to remain.
the incursion of oxygen into the system. I would like to hear I would like to point out that extraction processes gen-
UOP’s comments on their recent work with a new anti- erally result in yield loss of the product as compared to
oxidant to control the oxygen coming into a Sulfolane unit, hydrogenation routes, which, for aromatics reduction, have
and minimize the oxidation products, and hence the or- demonstrated high selectivity of the yield of diesel boiling
ganic acids that are formed afterwards. range material, and anywhere from two to six percent
volume swell depending on feed characteristics and ar-
ANGELO FURFARO (UOP): omatics conversion required.
In response to the question on corrosion and corrosion
control programs for Sulfolane units asked by Jennifer
Blasi of Mobil Oil, UOP would add that a good “solvent E. Polymerization
conditioning” program will drastically reduce unit cor-
rosion. This consists of limiting oxygen and chloride in-
gress, avoiding excessive reboiler skin temperatures, MEA QUESTION 41:
addition to the system and, most important, proper solvent Is anyone injecting water into the propylene feed
regenerator operation. The degradation of the solvent is of a Catalytic Polymerization unit? At what rate, in
both thermal and oxidative. gallons/bbl feed? How effective is it? What factors
Oxygen ingress is limited by inert blanketing of floating are considered in determining the addition rate?
roof storage tanks which feed the unit and good vacuum
section leak testing. Steam to reboilers should not exceed GRANT:
450° F with bare tubes (500° F for hot oil). A properly sized In one refinery, the water injection rate has varied be-
and well run regenerator should process 1 percent or more tween 0.004 and 0.019 gallons of water per barrel of fresh
of the lean solvent flow. This column removes both heavies feed. It is adjusted based on the free phosphoric oxide
and chlorides, Many units have been revamped to greater content of the spent catalyst at the end of run. The phos-
capacity without adding regenerator capacity. Many of phoric oxide content target is 16 to 20%. The run length is
these operators then find themselves limited in their abil- normally three months. We realize a slight increase in
ity to control solvent quality. The entire lean solvent stream activity with water injection, but it is hard to quantify.
should be filtered to help remove solids which contribute to At another facility, water is injected into the polym-
the corrosion-erosion phenomena. erization reactor feed in order to increase the water content
Oxygen ingress ultimately results in the oxidative de- from the saturation level to approximately 0.1 mole percent
gradation of the Sulfolane solvent, and the resultant for- of the total reactor feed. This equates to about 0.007 gallons
mation of volatile and non-volatile acids. These acids con- of water per barrel of fresh feed. The amount of water
tribute to unit corrosion. In addition, acid polymers injected is adjusted based upon the pressure drop in the
contribute to fouling of the extractor raindeck and other reactors and the perceived activity of the catalyst. Each
Sulfolane unit equipment. reactor has five catalyst beds. If the pressure drop across
Over the past few years UOP’s Refining Chemical De- one bed exceeds 20 psi, water injection is decreased. Water
partment, in conjunction with UOP’s Technical Services injection is also adjusted based on the propylene and butyl-

151
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

ene content. If the total propylene and butylene content The factors in determining the amount of water that
drops to two percent, water injection rate is reduced. If should be in the feed are the reactor temperatures and
propylene and butylene content reaches ten percent, the pressure, and the olefin concentration of the feed.
injection rate is increased.
It should be noted that the manufacturers of solid phos-
phoric acid catalyst specify different hydration re- QUESTION 42:
quirements. Before injecting water or changing water in- Does anybody have any experience with chamber
jection rates, we recommend checking with the catalyst reactors? How do they compare with tubular reac-
manufacturer. tors in the following aspects: a) run length, b) main-
tenance, c) catalyst life, d) safety?
ROBERT BROWN (United Catalysts Inc.):
We find that for most plants, the soluble water in the feed MOYSE:
is typically the appropriate amount for the operations. That The main difference is that the chamber reactors are
water can be adjusted up or down if you have control of the close to adiabatic; whereas the tubular type is isothermal.
water wash or the feed surge drum temperatures. A few Conventional wisdom tells us that should mean it is easier
plants will require additional water injection if for some to control temperature in the tubular type as opposed to the
reason the water is particularly low in the feed. The total chamber where you would probably be using quench. As
water, whether it is the contained water or if it is injected regards to run length and catalyst life, I do not have any
water, as Mr. Grant pointed out, should be determined direct data to give you, but we feel that would be a wash.
by the hydration recommendations of the catalyst For safety, it would appear that the tubular type would be
manufacturer. better, or at least easier to control. However, we are also told

152
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

that there are a lot more chamber then tubular reactors in GRANT:
use. We have used methylene chloride in place of carbon
tetrachloride. Poisoning from trace contaminants does
ROBERT BROWN (United Catalysts Inc.): occur periodically. Our suppliers are now required to for-
Although there are certainly plenty of exceptions, over- ward a certificate with each delivery specifying the quality.
all the catalyst lives are about the same between chamber Contaminants found in off-specification methylene chlo-
and tubular, perhaps a little bit lower in tubular units. The ride were water and methanol. We have no experience with
time onstream, though, is typically much longer for cham- anhydrous HCl.
ber reactors due primarily to lower space velocity and lower
olefin concentration in the feed. Although it is not a point FREDERICKSON:
on the agenda, but was just mentioned, we do not find that We use ethylene dichloride in isomerization units. We
tubular reactors are isothermal. have never identified problems due to contamination of this
material, but we agree it would be a good idea to check for
contamination with oxygen compounds.
F. lsomerization
QUESTION 45:
Many of our isomerization units use steam pre-
QUESTION 43: heaters on the feed stream. We have had problems
What amount of time and what amount of HCL determining when a leak in these exchangers is
have been required in the final drydown of a C4 or poisoning the catalyst. Has anyone had a similar
C 5 isomerization unit revamped from existing experience? Are there any methods of monitoring
equipment? this exchanger for a leak?

KNEPPER: FREDERICKSON:
Using butane we have completed a dryout in three days. We use steam heaters on the feed stream, but maintain
Using hydrogen would take longer. The moisture was the steam at lower pressure than the feed. Obviously, we
monitored using an on-line moisture analyzer. We would find leaks as hydrocarbon in the condensate.
not use any HCl until the unit was dried out.

QUESTION 46:
QUESTION 44:
What results are being achieved on light naphtha
What alternate chloriding agents to CCL4 are being units running mixed C5-C6 feedstocks regarding oc-
successfully used for isomerization catalysts? Is tane increase, conversions and catalyst life? What
poisoning from trace contaminants a problem? Does practical limitations have been experienced regard-
anyone have experience with anhydrous HCL? ing maximum benzene content and C7 + composition?

KNEPPER: HUNTER:
Trichloroethylene is currently being used successfully as We have an isomerization unit which is a UOP TIP
a chloriding agent for our C4 isomerization catalyst. Con- design. It processes a mixed C5-C6 stream to produce an 88
taminants have been a problem in the past. Oxygenated to 89 octane isomerate product. The feed octane is typically
stabilizers found in the trichlorethylene, as well as in in the mid-70’s. The conversion efficiency is maintained at
trichloroethane, can be found in high enough con- approximately 95% which correlates to a 2.5% unconverted
centrations to adversely affect catalyst performance when normal C5-C6 in the product stream. Reactor temperatures
converted to water in the reactor system. We found that and absorber cycle time are controlled to obtain 95% con-
propanediol was being utilized by the manufacturer to version, not to maintain a fixed octane. The isomerization
heat-stabilize the trichloroethylene. The level of the pro- reactor charge C7 + content is controlled to less than three
panediol that was present, and the amount of water percent by fractionation in a dehexanizer tower. This
produced, correlated closely with the short isomerization minimizes cracking which can lead to excessive reactor
catalyst life we were experiencing. We are currently utiliz- temperatures, high product RVP, and high light ends pro-
ing a nonstabilized trichloroethylene material without any duction rates.
adverse effects. At this time, we have no experience with Feed benzene content is typically one percent with a
anhydrous HCl. maximum allowable feed specification of three percent.
Benzene content is limited since it can also cause excessive
McCLUNG: reactor temperatures.
Engelhard continues to recommend the use of carbon
tetrachloride, without inhibitors, normally of reagent McCLUNG:
grade. Technical grade can have trace contaminants or Octane gains of ten to thirteen numbers are reported for
inhibitors which are deleterious to the catalyst. Engelhard Engelhard clients, and this is, of course, on a once-through
has used anhydrous HCl in the activation of a first stage basis on an acid based catalyst. Catalyst lives of up to seven
benzene hydrogenation section for a C5-C6 isomerization years have been reported with cycle lengths of six to eighteen
unit, but not for on stream chlorination. months. Of course this is a regenerable catalyst. The units

153
TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

would indicate that it is more profitable to sell high purity
A. Alkylation-Sulfuric Acid propylene than making alkylation feedstock out of it.

Feed CABRERA:
We see no economic or process benefit of adding the
propane to the splitter upstream of the alkylation unit to
QUESTION 1: take the propane out. An internal depropanizer in the unit
Is t h er e an y b en ef i t r eal i zed b y i n s t al l i n g a is considerably less costly and more cost effective. If the
propane-propylene (P-P) splitter to split propane out capacity of the alkylation is being limited by the amount of
of the P-P feed upstream of an H2SO4 or HF alkyla- propane, I think the place to focus is upstream in the FCCU
tion unit? How would this affect the operating and or gas concentration unit to take propane out before it
investment costs of the alkylation unit? comes to the HF alkylation unit.

TURPIN: FLESHMAN:
Removing the propane in the feed to an HF unit will have This would require very expensive column for a rela-
little effect. The heat duty required to boil up the propane tively small benefit. Propane would go through the unit
and the recycle in the depropanized reboiler will be reduced basically unchanged and can be easily split out of the
as propane is removed. The operating costs involved with product.
this heat that is saved will then be required in the operation ROBERT SHEPARD (SETPOINT, Inc.):
of the splitter. It is quite unlikely that this capital expense The effect of propane in alkylation feed is different for HF
can be justified on an operating cost reduction basis. The alkylation than for sulfuric alkylation. In both cases, there
propane in the feed and recycle acts as inerts in the reactor is an optimum propane level that is consistent with equip-
settlers and may limit recycle. However, if a unit is recycle ment constraints, feedstock costs and alkylate quality.
limited, the capital expenses of the splitter may be In HF alkylation, propane influences the hydrogen
justified. transfer reaction which affects propane yield, isobutane
In a sulfuric unit, removing the propane in the feed will consumption, alkylate yield and alkylate octane, which all
also slightly reduce operating costs. The process advantage increase with decreasing concentration of propane in the
here is the reduction of propane in the recycle will incre- reactor. Control of isobutane recycle propane content is also
mentally increase the iC4 concentration in the reactor important since the reaction cannot distinguish whether
outlets which will improve octane and reduce acid con- the propane entered the reactor with the feed or the recycle.
sumption. The reduction in propane in the recycle may In sulfuric alkylation, propane helps minimize reactor
increase reactor temperatures. Here again, the improve- temperature, but the improved alkylate yield and octane
ment in operation will not justify the capital expense of the are offset by a higher space velocity and increased com-
splitter under most economic conditions. pressor horsepower requirement. When propane removal is
Kinetic model studies will provide a refiner with incre- constrained by the depropanizer, additional propane in the
mental changes in unit operations and economics as pro- feed will reduce the isobutane recycle purity which results
pane is removed from the feed to the unit. in a lower isobutane to olefin ration in the reactor.
MALIK: CABRERA:
Our Corpus Christi refinery has a C3/C4 splitter, not a I think you mean hydrogen transfer reaction not hydro-
propane-propylene splitter. It is able to keep total C3’s in gen shift.
our HF alkylation feed to less than 1%. Without a C3/C4
splitter, our feed would be about 25% C3’s of which 1/3 ROBERT E. DAVIS (R.E. Davis Chemical Corporation):
would be propane. There are significant fractionating and I would be interested in polling the panel as to the types
treating benefits. We run an overall iso-olefin ratio of about of units they have. The experience that I have had indicates
10, of which 7 is attributable to keeping the propane con- that a Kellogg unit or the cascade type unit is much more
tent of the recycle very low, typically 3%. The depropanizer sensitive to propane in the feed than is a STRATCO type
and propane treating system size and cost were minimized. unit. This is due to the lower operating pressure typically in
Also, the acid settler(s) size and cost would be minimized, a cascade unit and the forward flow which causes problems
or if kept the same, better acid settling would result. in getting the settler to stop boiling.
QUINN: QUINN:
I can agree with what Lee Turpin said. The only comment Amoco has several kinds of sulfuric alkylation units:
I would add is that most of the economics that we have seen STRATCO, cascade and jet mixing.

136
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

QUESTION 2: drop in T90 of approximately 10o to 15°F; and

What changes have been observed at the alkyla- drop in end point of approximately 25o.
tion unit when processing MTBE raffinate? (acid con-
s um pti on, RON, MON, T90, EP)? TURPIN:
The effects of processing MTBE raffinate in an alkylation
ARMSTRONG: unit depend upon the type unit being reviewed. In both
Theoretically, removal of the isobutylene from the sul- types of unit, the acid consumption will increase on a per
furic acid unit feed will result in higher octanes, lower acid barrel of feed basis. Installing mole sieves to remove oxy-
consumption and lower alkylate end points. Since the start- genated contaminates will help both processes. An HF unit
up of the MTBE unit in Lake Charles, we have seen our will probably still see some increase in A.S.O. production.
alkylate end point drop 40°F and also have seen a decrease Comparing pilot plant data for feeds of; 1) mixed C4 olefm
in acid consumption. We also saw the expected increase in feed, and 2) isobutene run at constant iso-olefin ratio the
octane. We attribute a large part of these improvements to following conclusions can be drawn:
the fact that the C5’s are rejected into the MTBE product In an H2SO4 unit, the alkylate octanes will increase by
and do not reach the alkylation unit. about one-half to one point. In an HF unit, the alkylate
octanes will decrease by about one-fourth to one-halfpoint.
CARRERA: The change in T90 and end point may not be within the
In the case of the HF alkylation unit, the effect of the loss accuracy of an ASTM D86. From the data I reviewed, the
of the isobutylene is nil since the alkylate octane from end point in a H2SO4 should go down; the amount will be
isobutylene is intermediate to that formed from 1-butene dependent upon the concentration of isobutene in the orig-
and 2-butene. UOP recommends selective hydrogenation of inal feed. The end point of the alkylate from an HF unit
the end butylene rich stream which would isomerize the may go up slightly.
1-butene to 2-butene and would produce an additional The iC5 production per BBL of oletin feed will go down
product octane improvement of the order of 2 to 3 numbers. slightly in both the H2SO4 unit, and in the HF unit.
The acid consumption for a sulfuric acid alkylation unit Our kinetic alkylation model (see results below) is in
would increase by as much as 10% if there were up to 1,000 agreement with these observations, except that with the
parts per million of oxygenate in the feed. In the case of the model we were able to also look at maintaining a constant
HF alkylation unit we would recommend that an oxygenate recycle rate which is probably the way the units would be
removal unit be installed. As my colleague said, the T-90 operated in a refinery.
and end point will decrease in the sulfuric acid unit. In the HF ALKYLATION H2SO4 ALKYLATION
HF alkylation unit, there will be no effect. Constant Constant Constant Constant
Recycle I10 Ratlo Recycle I/O Ratlo

FLESHMAN: Change in RON +0.1 -0.6 +0.8 +0.6

MTBE raffinate has had the isobutylene removed leaving Change in MON +0.1 -0.5 + 0.3 + 0.2
l- and 2-butenes to react. Alkylation produced from iso- Change in T90 -1.0 + 1.0 NONE NONE
butylene is 6 to 7 research octane numbers and 4 to 5 motor Change in EP +2.0 NONE NONE NONE
octane numbers lower than alkylate product from l- and Change in C5 (V%) +0.1 -0.1 -0.2 -0.2
P-butene. It follows that the RON and MON of the alkylate Change in HVY
produced from MTBE raffinate will then be increased. This ALKY (V%) -1.8 +0.7 -0.5 -0.3
octane improvement is a function of isobutylene content
but an increase as high as 1 RON can be expected. Acid QUESTION 3:
consumption in the alkylation unit should be expected to
increase unless proper steps are taken to remove oxygenates Has anyone had trouble with fouling or corrosion
such as MTBE, MTBE byproducts (including dimethyl in an alkylation unit as the result of feeding an iso-
ether), and methanol that may be carried over from the butane stream containing chlorides, particularly the
MTBE unit. The principal product of isobutylene alkyla- isobutane from a butane isomerization unit?
tion is 2,2,4-trimethenyl pentane which boils about 20o to
25°F lower than the primary products from the alkylation QUINN:
of 1 and 2-butenes. Isobutylene also produces the largest Trace hydrochloric acid in the butane feed, and the feed
quantity of Cl0 + material. We would therefore expect a to the DIB tower, is extremely corrosive to the effluent
decrease of the alkylate end point and the alkylate T90 lines, DIB trays, DIB overhead system and the re-
when using MTBE raffinate. This would again depend on frigeration system. Corrosion can be controlled by double
the isobutylene concentration, but each could be expected washing butane with caustic, MEA injection into the DIB
to decrease by approximately 10o to 15o. feed, and constant monitoring of the water wash and the
tower overhead Ph.
LAVERGNE: ARMSTRONG:
Our experience is similar to that of the previous pan- We have not seen any chloride problems in our HF units
elists. At constant olefin space velocity we have observed that are associated with butamers.
the following when processing MTBE raffinate:
increase in research and motor octane of approxi- CARRERA:
mately 0.5; Our experience is consistent with the panel.

137
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

HERBERT W. WIZIG (Merichem Company): amylene alkylation? Have any mechanical or oper-
Merichem’s CHLOREXTMechnology which uses the ating problems been directly related to amylene alky-
FIBER-FILM’” contactor along with an alkaline water lation? What percentage of the heavy mercaptans
wash, can be effectively used to remove hydrochloric acid associated with amylene processing is removed by
and ammonium chloride from this stream as well as from typical feed pretreatment? How are yields and down-
reformate. stream equipment affected by the additional sulfur?

DAVID L. SMITH (ALCOA): CABRERA:

Another potential solution to the chloride problem is The yields for amylenes in an HF alkylation unit are
solid adsorbent beds. Systems have been designed and mostly dependent on the isomer distribution of the feed-
installed to remove the chlorides from the isobutane efflu- stock. The expected yields for two different feedstocks are
ent from isomerization units. as follows:
Typical FCC TAME Raffinate
ROBERT E. DAVIS (R.E. Davis Chemical Corporation): Vol/vol Olefin Isomer Blend Isomer Blend
I want to point out that a lot of people are buying iso- Isobutane consumed 1.35 1.13
butane that has been stored in underground salt domes and Pentane produced 0.45 0.22
there are a couple of people here that have had the experi- C6 + Alkylate 1.57 1.61
ence of getting the water from that into their sulfuric unit
resulting in severe corrosion. The key point I want to focus your attention on is the
thing that may worry you, which is the pentane produced.
The relatively high pentane which is produced, and ranges
Process from .22 to .45, volume produced per volume of olefin used,
is basically a function of the isomeric distribution as we
said earlier. The cause is the hydrogen transfer side reac-
QUESTION 4: tion. This is really a series of reactions which have the
Di s c u s s t h e o p t i m u m o p er at i n g c o n d i t i o n s f o r overall effect of consuming one mole of amylene and two
am y l en e al k y l at i o n v er s u s b u t y l en e al k y l at i o n f o r moles of isobutane to produce one mole of pentane and one
both H2SO4 or HF units. Is it necessary or desirable to mole of iso-octane.
segregate the amylenes and butylenes in the alkyla- The degree of hydrogen transfer is greater for iso-
tion reaction zone? amylenes then for normal amylenes. From the operations
perspective, the primary concern as far as amylene alkyla-
CABRERA: tion is concerned is the acid purity maintenance and the
For the HF alkylation unit the optimum conditions to regeneration capacity. This normally leads to the addition
run would be those that minimize hydrogen transfer; that of selective hydrogenation type processes to remove the
is, lower isobutane to olefin ratio. This will however, also diolefins and get the side benefit of the isomeration of the
cause a lowering of the product octane. The economic bal- 1-butene to 2-butene. The net result is an increase in the
ance consists of the balance between the price of isobutane, research octane number. Proper MEROX treating of the
the RVP of the product, and the lower research octane feed to handle the sulfur is critical. We would be pleased to
number which is produced. Effective feed treatment is the discuss with you the operating strategies for your MEROX
key to successful amylene alkylation. There are no other units to accommodate this case and refer you to a paper we
significant differences in the operating conditions for the published in the 1991 MEROX Symposium. Copies of this
reactor section and the butylenes and amylenes would not paper are available upon request.
be separated.
ARMSTRONG:
TURPIN: In our units, we have observed that about half of the
Work published by STRATCO indicates that in a sulfuric amylene is alkylated to C9’s, with the rest isomerizing to
unit, the segregation of amylenes and butylenes will not make isopentane and promoting the self-alkylation of iso-
significantly change yields or operating costs, assuming butane as Mr. Cabrera described. We have not seen any
the diolefins and cyclic compounds have been removed from mechanical problems in our sulfuric unit that we would
the amylenes. The key here is the saturation of the diolefins attribute to amylenes, and the various technology licensers
and cyclic compounds as required to get this flexibility in can provide you with information on their cleanup
blending amylenes and butylenes. processes.
The economics that I have reviewed show that amylene
alkylation is not profitable, and the only reason we as an MALIK:
industry are considering it is to meet mandated gasoline CITGO’s Corpus Christi refinery has an HF alkylation
specifications. Keeping that in mind, most refiners will unit, rather than a sulfuric acid, but we do have some
probably develop a rather detailed LP program to plan the MEROX treating data. Our feed is MEROX treated at a
minimum quantity of amylenes needed for alkylation to constant caustic circulation rate and we see little or no
meet blending requirements. Whatever minimum amount change in feed total sulfur content with up to 5% amylene
of amylene that has to be processed will be the optimum which is 15% of total C5’s in the reactor feed. HF con-
operation under most economic scenarios. sumption did increase, however. Amylenes consumed
about 70 to 100% more acid per barrel of alkylate produced
than did the butylenes. Some of this increase was due to
QUESTION 5: incrementally having to regenerate externally as our in-
What are typical expected yields associated with ternal regeneration was already loaded.
138
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

QUESTION 6: ing individual C5 olefin isomers with a make-up isobutane

What is the isopentane and normal pentane pro- feed that does not contain any isopentane.
d u c t i o n w h en al k y l at i n g am y l en es w i t h H 2 SO 4 ?
What are these same numbers when HF is used? HERBERT W. WIZIG (Merichem Company):
Merichem Company’s experience in treating C5 olefins
JONES: ahead of C5 alkylation TAME units indicates that the
We have seen numbers between 3 and 5 vol % of both untreated feed will contain primarily ethyl and propyl
isopentane and normal pentane yields using HF alkyla- mercaptans which can be successfully removed using
tion. These numbers vary depending upon feedstock com- Merichem’s THIOLEXTM and REGENTMtechnologies which
position and alkylation operating conditions. Branched utilize the FIBER-FILM’” contactor as the mass transfer
amylenes will hydrogen transfer more than linear am- device.
ylenes and hence lead to higher isopentane yields. Higher
C5 production can be due to the heavy alkylate of C12 T. A. CAVANAUGH (Exxon Research & Engineering
cracking to C5 plus a light alkylate, C7’s. Company):
We have considerable commercial experience in ER&E
TURPIN: sulfuric acid alkylation units running amylenes premixed
Pilot plant results of sulfuric acid alkylation of 1-pentene, with the butylenes. Regarding downstream effects, we see
2-pentene and 2-methylbutene were reported by Lyle Al- that sulfur could be a potential problem since the mer-
bright of Purdue University and Ken Kranz of STRATCO captans in the amylenes tend to be heavier and thus more
in Industrial Engineering Chemistry in 1992. Some infor- difficult to extract.
mation on amylene yields in an HF unit were reported by T. For example, methyl and ethyl mercaptans are prevalent
Hudson and R.S. Logan in “Estimate Alkylate Yield and in the butylene stream, whereas ethyl and propyl mer-
Quality”, September 1975 Hydrocarbon Processing. Other captans are more prevalent in the typical amylene stream.
literature on amylene alkylation include: Masters, Ken- However, you cannot generalize on the amount of residual
neth R., “Amylene Alkylation”, STRATCO Alkylation sulfur after the caustic treating since it depends to some
Seminar, September 1990; McDonald, G. W. G., “FCC Gas extent on the process variables within the caustic treater
Recovery Operations”, STRATCO Alkylation Seminar, itself. In any case, the potentially higher sulfur in the
September 1990. In talking with Mr. Kranz, some of the amylenes will affect acid consumption more than yields.
published yield data are very conservative. I suggest you
get with the vendors directly for yield estimates. NORMAN KOLB (The M.W. Kellogg Company):
Kellogg has recently conducted a series of sulfuric acid
ARMSTRONG: alkylation pilot studies in cooperation with Exxon Re-
We have not seen any normal pentane production in search & Engineering Company. One area of interest is
either HF or sulfuric units, but we have not studied the sulfuric acid alkylation of high normal pentene feeds, that
issue in great detail. is, alkylation of TAME raffinate. A significant issue associ-
ated with alkylation of N— - C5=‘s is saturate pentane
CABRERA: make. A driving force for alkylating C5 =‘s is vapor pres-
Our experience is consistent with that described by Mr. sure reduction. Net pentane make partially cancels the
Jones. For information purposes, the maximum normal vapor pressure advantage associated with conversion of
pentane production that we have seen in our pilot plant C6 = ‘s to C8/C9 + alkylate.
facilities from pure normal pentenes at conditions which Saturate C5 production in H2SO4 alkylation is related to
maximize hydrogen transfer is about 5% of the alkylate oletin type, temperature, and isoparaffin content in the
produced. reaction zone. Isopentane content of total reaction zone
isoparafin has substantial impact on pentane make. Iso-
PAM KEEFER (STRATCO, INC.): pentane competes with isobutane as an alkylation reac-
Previously published yields for sulfuric acid alkylation of tant. When little isopentane is in feeds, substantial iso-
amylenes are a bit conservative and our most recent re- pentane can be made. When a large quantity of isopentane
search indicates typical yields in the range of 2.0 to 2.2 is present, isopentane is (on balance) consumed in an alky-
barrels of alkylate per barrel of C5 olefin. Also, we have lation unit.
found that, unlike the propylenes and butylenes, variations An upper limit for isopentane make with H2SO4 alkyla-
in operating conditions appear to have very little impact on tion is a case where there is no isopentane in the feed. Study
amylene alkylate yields. of this circumstance can provide an end point for iCs make
The only other comment that I wanted to make concerns correlations that are tied to iC5 content of reaction zone
isopentane production while alkylating C5 , olefins in a isopara6in.
sulfuric acid alkylation unit. Under typical alkylation con- Following are H2SO4 alkylation data for an isopentane
ditions with a mixed C6 FCC olefin stream, one can expect free nC5= feed:
isopentane production on the order of 8 to 9 wt % with no
normal pentane production. Operating conditions can Conditions:
cause the production of isopentane to vary anywhere be- Temperature 41°F 68°F
tween 1 to 11 wt %, but typically you will see 8 to 9 wt %. Feed I/O Ratio 6.4 6.4
Research work performed by STRATCO indicates that iso- Rxn Zone LV% iC4 71.9 70.2%
pentane production can be as high as 18 wt % when alkylat- Rxn Zone LV% iCs 3.6% 5.4%

139
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

Rxn Zone LV% iCx 75.5% 75.6% Reactor Effluent N-Pentane vs Time
iC5 LV% Of iCx 4.8% 7.2% Alkytation Run 5.01
C5+ Alkylate Breakdown (Upper Bound For iC5 Make)
iC5 LV% Of C6+ 14.2 20.0
nC5 LV% Of C5+ 0.1 0.3
c6 + LV% Of c5 + 85.7 79.7
TOTAL 100.0 100.0
Typically, nC5 = feedstocks will have associated iso-
pentane, however it is difficult to project a “typical” reac-
tion iC5 percent of total iCx because this is tied to feed
composition, fractionation, and reactor design con-
siderations. “Typical” iC5 make can cover a range from
negative net consumption to the upper bounds cited above.
Normal pentane make in H2SO4 alkylation of nC6 = ‘s is
very small. To study this and the temperature coefficient
associated with nC5 /iC5 make in nC5 = alkylation, we
set-up a continuous flow alkylation system with recycle
acid from a reactor effluent acid/hydrocarbon separator. We
ran an experiment with continuous nC5=/iC4 feed for a FIGURE 2
number of hours at 41°F. We then increased the reactor
temperature to 68°F. We continued running at 68°F for six
hours (based on other experimental work, we knew that
this was just sufficient to lineout the unit). We then re-
turned the system to 41°F. Product compositions at the
end-of-run were the same as start-of-run. We observed an
increase in isopentane make at high temperatures. Normal
pentane make was very low (0.1 LV% of alkylate) at 41°F
Normal pentane make increased at 68°F (still trace at 0.3%
LV% of alkylate).
Figures 1 and 2 illustrate reactor effluent isopentane and
normal pentane concentrations for this experiment.
During this experiment, alkylation acid was con-
tinuously recycled from an acid settler to the reaction zone.
System acid inventory was fixed. Acidity decreased from
92.25 to 91.61 wt % during the run. This is a modest change
in acidity. Effectively, the entire run was made at 92%
acidity.
Work of this type is being done in connection with Exxon
Research & Engineering’s Sulfuric Acid Alkylation Proc-
ess which is available for license.

Reactor Effluent lsopentane vs Time

Alkylation Run 5.01

140
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

QUESTION 7: QUESTION 9: .
In alkylation units that are all electric, what does Do any refiners have pH analyzers on the caustic
one do when there is an emergency shutdown (i.e. wash drum? What are their reliability?
power failure)? What do refiners do with the olefin
feed if the FCCU remains operational? What general ARMSTRONG:
steps should you take to safely secure the unit? We do not have an on-line pH analyzer. We manually pull
a sample from the wash water drum downstream in the
ARMSTRONG: caustic wash drum once per shift. We target for a pH of 6 to
Obviously the first step is to act to insure the safety of the 9 in that wash water drum.
employees in the facility. In summary, this includes stop-
ping the olefin feed and the fresh acid makeup, shutting
STEVE COUSINS (Lion Oil Company):
down all heat input sources, and blocking in the acid oil line
We have used an on stream pH analyzer on our caustic
to prevent the caustic wash from coming back into the
water wash very successfully for 10 years.
reactor or settler. You may also need to reduce the pressure
on the unit to keep the temperature from rising and caus-
ing flaring. When our alkylation unit shuts down, the
olefins are routed into pressurized storage if the FCCU is QUESTION 10:
going to remain on-line. Are there any incentives to convert caustic-water
wash effluent treating systems to acid or alkaline
FLESHMAN: water wash systems? Do they pay out?
We agree with that general procedure. Since cooling from
air fans as well as agitation will be lost, we recommend ARMSTRONG:
blocking the unit in, including shutting off any turbine Changing to a high purity acid wash system could reduce
driven spares. This includes blocking in the feed, the re- corrosion problems caused by sulfate ester formation. The
cycle isobutane pump, and reflux pumps. Shut off electro- ester formation reaction is reversible in a strong acid solu-
static separators and try to establish an alternate seal flush tion. If isobutane is available, this could result in ad-
from turbine driven pumps. Watch the system pressure and ditional alkylate production and regeneration of the acid
vent if residual heat causes the pressure to rise. As the unit catalyst. Typically though, caustic neutralization is suf-
cools and the pressure drops, it may be necessary later to ficient and much less expensive.
maintain pressure with fuel gas to avoid creating a
vacuum. QUINN:
We use caustic and water wash systems on all of our
QUESTION 8: alkylation units. Potential caustic savings have not jus-
Are any refineries adding corrosion inhibitors in tified modifications.
the deisobutanizer (DlB) feed line? Are they bene-
ficial? Is there a concern or experience with fouling HERBERT W. WIZIG (Merichem Company):
o f t h e p ac k i n g i n t h e DIB ? Is t h er e an al t er n at e The acid wash system removes both neutral and acid
injection location which gives better results? esters, whereas the caustic wash only removes acid esters,
thus improving the DIB operation in terms of reduced
QUINN: corrosion and fouling of the trays and reboilers. Ester
We inject MEA into the DIB feed line at most of our removal can be measured by preforming a copper strip
plants to control tower overhead Ph in an acceptable range. analysis on the treated reactor product. Merichem Com-
This has worked very well. We do experience fouling in the pany’s ESTEREX” technology for the acid wash with
DIB reboiler over time and periodically have to clean it. downstream AQUAFININGTM technology for the alkaline
Fouling in the trayed section of the DIB tower has not been water wash can be used to retrofit existing conventional
significant. We do not have pack towers in this service. electrostatic treaters to achieve this conversion without
having to replace the vessels. Another benefit will be that
LAVERGNE: the FIBER-FILM TM system will provide a 100% service
We put a corrosion inhibitor into the feed and overhead factor compared to electrostatic systems which short out
line of the deisobutanizer. With appropriate monitoring of from time to time.
the overhead line corrosion meter, we were able to cut
corrosion from 10 to 1 mils per year. We tend to get some
fouling in the deisobutanizer reboiler from minor acid by JAMES A. ROGERS (Koch Engineering Company, Inc.)
product carryover. The Koch SMVTM type static mixer is an extremely ef-
We use a water wash procedure to wash foulants from the ficient way to contact the alkylate with any of these aque-
reboiler tubes on a weekly basis. The procedure takes about ous streams. The static mixer creates a fine dispersion
20 minutes and upsets the column for about 30 minutes. between the two phases, but consisting of droplets that will
Reboiler steam requirements are reduced by about 30% still separate quite easily. This will greatly decrease the
after this procedure is completed. amount of any “rag layer” generated. This fine dispersion
will allow the maximum amount of mass transfer at pres-
ARMSTRONG: sure drops far less than those for orifice plates or mix
We have not see a need to use a corrosion inhibitor in our valves. Well over twenty systems have utilized this tech-
sulfuric acid unit DIB circuit. nology over the past several years.

141
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

QUESTION 11: TURPIN:

A r e ad v an c ed c o n t r o l s t r at eg i es u s ed w i d el y i n Normal butane balance is a standard function incor-
sulfuric acid alkylation units? If so, where are sig- porated into alkylation unit advanced controls. Profi-
nificant operating improvements seen? What is the matics’ methodology on alkylation unit control, as well as
r et u r n o n i n v es t m en t f o r t h e ad v an c ed c o n t r o l other units, is to provide a base set of control functions
installations? common to all units, and then incorporate specific functions
on an as needed basis. Some of the difficult advanced
ARMSTRONG: control problems, including on-line optimization are ad-
We are in the process of installing advanced controls on dressed using a kinetic model incorporated within the
the sulfuric acid alkylation at our Lake Charles refinery. control scheme.
We will be controlling the reactor temperature and opti-
mizing the debutanizer and DIB column operations. We are ROBERT SHEPARD (SETPOINT, Inc.):
anticipating earnings of about 8 to 10 cents per barrel of As mentioned earlier, the heart of SETPOINT’s alkyla-
unit feed. tion control technology is the stabilization of propane and
n-butane accumulation within the unit. When feed ana-
lyzers are available, this technology does include feed pro-
LAVERGNE: pane and n-butane content as feed forward variables in the
We are currently reviewing deisobutanizer and debu- control.
tanizer process control improvements, specifically, up-
grading to steam flow control based on tray temperatures of
the deisobutanizer, and upgrading to tower bottoms release Mechanical
via LIC to FIC on both tower bottom streams. We are trying
to improve quality assurance on our. alkylate actual vapor
QUESTION 12:
pressure. We estimate benefits to be $ 1 million per year
over current operation. We are also installing wastewater Are there any ways to provide additional surface
Ph control on our alkylate sump to prevent pH excursions area for heat transfer in existing STRATCO Contactors?
in water sent to the wastewater plant.
CABRERA:
UOP and STRATCO are now completing a development
GREG DEARWATER (Profimatics, Inc.): project in which UOP high flux tubes are used in a com-
We have conducted roughly 20 studies on sulfuric acid mercial STRATCO contactor bundle. We expect that the
alkylation units and currently have one application run- results of this project will yield significantly improved heat
ning on-line. This application includes advanced control as transfer coefficients in all contactors in which high flux
well as on-line optimization using a rigorous model of the tubes are installed and therefore will improve in some
process combined with current plant economics. Operating combination their oletin capacity and emulsion tempera-
improvements resulting from advanced control include re- ture. The ongoing development work objective is to accu-
duced acid consumption, reduced loss of isobutane and rately determine what the true performance advantages
normal butane, and improved control of alkylate RVP. are of using the high flux tubes across a wide range of
There are further benefits arising out of on-line opti- process conditions. We expect that conclusive results will
mization including selection of the optimum iso-to-olefin be available before the end of this year.
ratio and DIB pressure. In our experience, these projects
have payouts of less than 18 months. NORMAN KOLB (The M.W. Kellogg Company):
I would like to take the liberty to reword this question as
ROBERT SHEPARD (SETPOINT, Inc.): follows, “How does one circumvent the bottleneck associ-
SETPOINT has performed over 20 advanced process ated with reaction heat transfer across a tube bundle in a
control studies on alkylation units, both sulfuric and HF sulfuric acid alkylation process?” One answer would be to
alkylation. These units varied in size from 3,000 bbl/day up simply eliminate the tube bundle. You can make your
to 19,000 bbl/day. Earnings generated by advanced control initial technology selection or convert to a process that
projects ranged between 15 cents and 35 cents per barrel of involves direct contact heat transfer. A selection for this is
alkylate capacity, and in cases where incremental feed was an auto-refrigerated alkylation process such as that offered
available, earnings as high as 85 cents per barrel of alky- by Exxon Research & Engineering Company.
late were possible. Direct contact heat transfer eliminates the need for a
The heart of SETPOINT’s alkylation control technology large and specialized design tube bundle. Direct contact
is the stabilization of propane and n-butane accumulation heat transfer also reduces reaction zone temperature for a
within the unit. This allows reactor conditions to be ad- given compressor suction pressure and reduces external
justed to either maximize alkylate octane or alkylate mixing energy required for alkylation.
production.
GLENN C. LIOLIOS (STRATCO, INC.):
UDAY HATTIANGADI (Conoco Inc.): In addition to the high flux work that is being done with
One of the problems with alkylation control is forcing a STRATCO and UOP on the pre-1970 STRATCO Con-
n-butane balance. I was wondering if any of the advanced tactors, there is the ability of adding longer tubes to the
control people had feed forward control schemes to enforce a bundle and adding approximately 15% additional tube
n-butane balance on alkylation units? bundle area to the Contactors.

142
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

General study belong to the project participants for a period of up to

5 years after completion of the study. No decision has been
made by the project participants on the earlier publication
QUESTION 13:
of the program results. Participation, which at this point
IS anyone considering the use of remotely operated means co-ownership of the results, is still open to anyone
isolation valves in sulfuric acid alkylation units to willing to pay the prorated share of the program costs
limit the LPG volume that could escape from a line, which was approximately $620,000. Information on joining
pump, or vessel leak? the program can be obtained by contacting Ralph Cecchetti
at Exxon’s Florham Park Research Center.
FLESHMAN:
These valves are routinely being installed here and in
other situations where large volumes of low flash point
material are stored. Requirements for these are often speci-
fied due to HAZOP analysis. Typical break points for re-
B. Alkylation-Hydrofluoric Acid
quiring them are fluids with a flash point less than 100°F
and a volume of 2,000 gallons or more of liquid. Shutoff Feed
valves are typically located at the vessel flange on liquid
outlets and located as necessary to isolate the pumps.
QUESTION 15:
Remotely operated fireproof valves are generally used. In
some cases the valve may be located away from the equip- Is t h er e a r el i ab l e an al y t i c al m et h o d t o d et ec t
ment and manually operated. o x y g en at es i n h y d r o f l u o r i c ac i d al k y l at i o n f eed -
stocks? What are the detection limits?
MALIK:
Although we have an HF alkylation unit, we are in- JONES:
stalling remote isolation valves refinery-wide to isolate We analyze alkylation feed for oxygenates. Oxygenates
LPG volumes. We have prioritized LPG work and inven- can be detected using a gas chromatograph with a flame
tories with a hazard index where higher numbers are given ionization detector. Use of the chromatograph was dis-
to large bodies of hot, low molecular weight LPG’s with cussed at last year’s NPRA Question & Answer Session.
high concentration of HF or H2S. For those locations that For further information, see questions 47 and 48 on pages
are above a certain hazard index, we have a project to 154 and 155 in the 1991 Q & A transcript.
remotely isolate pumps moving these streams. This will
guard against line and pump seal ruptures. MALIK:
Our refinery laboratory uses the Selective Ion Monitor-
QUINN: ing, (SIM) method which uses the mass spectrometer and
This is considered on an equipment by equipment basis will measure oxygenates down to about 5 ppm. ASTM
when large volumes of LPG are involved, and it is not method B4815 is good to about 100 ppm.
limited to alkylation facilities. API RP-750 directs that
there must be a credible scenario to initiate installation of ARMSTRONG:
safeguards and mitigation systems. Concerns associated We feel that the OFID method is the most accurate.
with remote isolation systems have led us to believe that However it is not very applicable in a refinery control lab
deluge systems should be considered before isolation. Re- because of the cost of the equipment and the time it takes to
mote isolation might only be used when there is a credible run the test. We have developed a simpler test at one
failure scenario that is supported either by history of the refinery where we just take a gas bomb, bubble it through
equipment or some unusual case specific circumstance. water, and run a simple GC. The test was designed to check
This approach supports sparing use of isolation systems. for acetone, isopropyl alcohol, and methanol. The accuracy
range is from 300 ppm down to 0.1 ppm. We have also
developed a multicolumn chromatographic technique that
QUESTION 14: has a lower detection limit of about 150 ppm. This detection
Can someone provide an update on the results of level is adequate to tell us if we are going to have a problem
t h e H 2 SO 4 aer o s o l t es t i n g p r o g r am t h at w as c o n - in the sulfuric acid alkylation unit.
ducted last fall?

QUINN:
Process
The testing was started in September of last year and
was completed at the end of October, 1991. Information on QUESTION 16:
the program design was reported at last years NPRA Q&A Wh at i s t h e c u r r en t s t at u s o f b o i l i n g p o i n t d e-
Session. Data analysis and report preparation are now p r es s an t ad d i t i v es f o r h y d r o f l u o r i c ac i d ? Ho w d o
underway. The program was implemented under the Petro- these additives affect unit performance?
leum Environmental Research Forum, PERF. Currently,
the participants are Amoco, BP, Chevron, Exxon, Mobil, WARWICK:
STRATCO, Texaco, and Unocal. Mobil is developing proprietary additive technology to
The results cannot be released at this time because under reduce airborne HF concentrations in the event of an acci-
the terms of the participation agreement, the results of the dental release. Our specific program is, of course, con-

143
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

fidential. We are able to say, however, that we have 2) Putting a conductivity meter on the circulating
screened a large number of additives and are developing sample line to stop the sample flow if significant HF is
one particular system which shows good airborne HF re- detected.
duction in release tests and maintains the performance
benefits of HF alkylation. Mobil is constructing a large- WARWICK:
scale unit to demonstrate this technology. We plan to com- Mobil is developing an on-line test for acid strength.
mercialize this technology to meet the timetable for regu- While results are promising, the system has not yet been
lations on the use of HF in the Los Angeles area. tested under commercial conditions. We plan to test the
system as part of our demonstration of our additive
CABRERA: technology.
I am not trying to redirect the question, but I think the
question really wanted to address the issue of raising the CABRERA:
boiling point rather than depressing it. So I will direct my Our experience is consistent with that described earlier
answer toward the ability of the additive to raise the by Tom Weber.
boiling point of HF.
We announced earlier, in the April meeting of the API
HF alkylation committee, that Texaco and UOP were de- QUESTION 18:
veloping HF additive alkylation technology for use in HF Has an y o n e h ad an y d i s c u s s i o n s w i t h c em en t
alkylation process units. Small scale tests have indicted manufacturers to determine if they can use the spent
that substantial reductions in potential aerosol formation lime from the neutralization pits? Are there any other
can be achieved with no degradation in alkylate quality. uses for this material?
The first test of the additive technology in an unmodified
commercial unit has been completed and the acceptable CRISLER:
performance of the reactor section of the plant with no The fluoride content usually makes spent alkylation
degradation of the alkylate quality has been confirmed. lime unsuitable for use in a rotary kiln. We have just
Control testing in a specifically designed release facility is started processing spent alkylation lime in our biological
now underway to try to quantify and confirm the reduction treatment plant. We experienced some foaming in our DAF
in aerosol formation. Additional long-term tests are when we got a big shot of high Ph material. When we add it
planned for 1993 and we now anticipate, based on our continuously at a low rate, the system seems capable of
current knowledge, that this technology will be offered for handling it.
licensing in 1994.
M. E. PFILE (Phillips Petroleum Company):
As far as other potential uses for this material, if it is
QUESTION 17: calcium fluoride, some refiners have found success in sell-
What is the experience with on-line analyzers for ing this material back to acid manufactures.
hydrofluoric acid strengths and for isobutane recycle
streams?
QUESTION 19:
WEBER: Has an y o n e at t em p t ed t o c h ar g e n eu t r al i zed
We have never found a need for an on-line analyzer polymer to the FCC main fractionator or to FCC feed?
measuring HF strength. Our unit runs 10 to 15% above
design capacity and we find that twice a week sampling and QUINN:
proper acid regenerator operation provides a very steady Amoco has chosen not to charge neutralized ASO to the
acid and water concentration. We did however, try an on- FCCU. If charged to the FCCU feed, ASO would form
line chromatograph for the isobutane recycle stream. The either SiF4 or possibly HF. Presently, ASO is either com-
sample conditioning system for the analyzer was always busted in special furnaces or blended off in heavy fuel oil.
becoming plugged with HF scale and we finally gave up
trying to make it work 6 or 7 years ago. As with HF, we find WARWICK:
that the recycle analysis is very stable and does not require We have done-it in the past with no problems, but are not
continuous analysis so that an analyzer is not really a good currently, as far as I know.
application here. Once a day is plenty for sampling iso-
butane recycle. WEBER:
We also added to fuel oil, or decanted oil in this case, but
MALIK: it is external to the fractionator.
We had to redesign our on-line isobutane recycle and
normal butane GC system to keep the periodic HF excur- ARMSTRONG:
sions in the recycle from harming the GC. While the GC’s - In one refinery, we have been taking the neutralized ASO
are on-line, they are quite useful. To keep them on-line, we back to the gas oil charge tanks for several years and have
are: had no discernable problems. At another plant, we take the
1) Moving the KOH treater directly upstream of the GC, ASO back into the crude slop system, and have seen no
treating only material going into the GC. corrosion problems.

144
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

QUESTION 20: needs replacement. During 1991, for unrelated reasons, we

Are any refineries experiencing increased corro- added a second propane alumina treater. They are run in
sion or fouling downstream of the propane defluori- series and hopefully this will end this type of corrosion.
nators? What is the mechanism for this increased rate
thought to be? Is replacement with monel piping and WEBER:
equipment being considered? We have not had any problems in our propane systems
either and we maintain higher than 400°F. We have a 450°F
ARMSTRONG: inlet temperature to the alumina treaters. However, as
In one unit, we were replacing the defluorinator con- someone else said, we have had problems with our butane
denser every three months. When we established a mini- defluorinators and actually have begun to replace some of
mum operating temperature of 400°F at the defluorinator the air-cooler tubes as well as downstream piping with
outlet, the service life increased to over 12 months. We monel and that has solved the problem.
think this higher temperature increased the rate of propyl
fluoride degradation to propane and HF, and the sub- ROBERT SHEPARD (SETPOINT, Inc.):
sequent HF alumina reaction, to keep the HF in the de- Another source of corrosion downstream of the de-
fluorinator. For any HF that does make it through the fluorinators, which is fairly common knowledge among HF
alumina, we have observed that keeping the outlet tem- alkylation operators, is the strongly acidic solution of HF
perature about 100°F on the last condenser reduces the and water formed by the condensation and hydrolysis of
amount of corrosion prior to the final neutralization vessel. silica tetrafluoride. Silica tetrafluoride is produced in the
We have been gradually upgrading carbon steel piping defluorinators when HF reacts with silica oxide in the
systems and exchanger bundles to monel over the past few alumina.
years, We have also seen this problem in several of our Silica tetrafluoride related corrosion is uncommon since
butane defluorinators. it is easily avoided by purchasing a low silica alumina. A
more common cause is the alumina is spent and requires
CHIANG: replacement. This tends to occur when alumina is changed
Our systems are all carbon steel and we have had no out to fit a maintenance schedule, every 3 or 6 months,
problems with corrosion. rather than a process schedule. The efficiency of the de-
fluorinators is less than 100 percent, meaning that the
CRISLER: treaters need to be changed out before the chemical ad-
Same as Mr. Chiang. We have never seen any corrosion sorption capacity of the alumina is exhausted. The alumina
problems in this part of our system. life appears dependent on space velocity and fluoride level
of the feed to the defluorinator.
MALIK:
Inspection during our 1989 turnaround revealed a groove QUESTION 21:
corroded along the bottom of the propane alumina outlet What recent developments have occurred in solid
line after the cooler. We had only one treater at that time. HF catalyst technologies?
This occurred even though our on stream phenolphthalein
bubbler never indicated acid. Our thinking is that free ARMSTRONG:
water with dissolved HF may have at times flowed along As previously disclosed, Conoco and Neste Oy have joined
the bottom of the pipe and selectively been corroding out with Catalytica to complete the development and com-
this groove. We have begun sending multiple alumina mercialization of Catalytica’s gasoline alkalization process.
treater samples to our lab for HF analysis, with no break- The process is based on a proprietary solid catalyst and
through being indicated. On stream inspection for the proprietary regeneration technology. If successful, the
October, 1992 turnaround again revealed some pipe that Catalytica-Conoco-Neste process would enable refiners to
more economically produce reformulated gasolines without
the use of liquid acids.
High octane has been demonstrated with commercial
feeds in continuous bench scale reactors. The construction
of a seven barrel per day pilot plant has been completed and
commissioning has started at Neste’s research facility
adjacent to its Porvoo refinery in Finland. Start-up is
anticipated before the end of this year. After a successful
completion of the pilot plant operation, the companies plan
to make the technology available for licensing worldwide.
CABRERA:
UOP continues to maintain a substantial ongoing R&D
effort devoted to the development of the solid bed catalyst
motor fuel alkylation process. While we have made some
progress to date, we have not as yet achieved the objective of
developing a process which is competitive with the liquid
acid technologies currently available.

145
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

General 1 year for the H5S sensor. Commercialization of the

technology is expected within 1 year after the pro-
QUESTION 22:
totype development.
Thermal Imagining Systems - It was developed by
Pl eas e d i s c u s s HF v ap o r d et ec t i o n : m ak es an d Amoco with Zeda Corporation. It offers wide area pro-
models of sensors, location in unit, reliability, etc. tection by using thermal cameras and a specialized
computer and software to detect cold HF cloud forma-
WARWICK: tion that is indicative of a release. Mobil has only done a
Mobil is currently examining all of the present HF sen- preliminary review of the technology and has no
sor methods and evaluating emerging sensor technology. comments.
At present we have reviewed the following types of sensors: All of our Mobil HF alkylation units are scheduled to
Point Source Sensors - HF has to enter sensor, like a install high resolution television camera systems to watch
household smoke detector, and is dependent on factors for HF white cloud formation. Also, reactive paints that
such as location, wind, and concentration of HF. change color when HF contacts it are used on key flanges,
ElectroChemical (EC) Types - HF diffuses through a vessels and equipment.
membrane and changes the conductivity of an electro-
lyte to a preset point to alarm. We use Sensidyne WEBER:
sensors in one refinery where 27 have been positioned Ultramar installed the GasTech model 2321 HF detec-
along the unit fence line and in other strategic loca- tion system in its alkylation unit in 1988. The system
tions like acid pumps and acid transfer line flanges. consists of about 20 GasTech electrochemical type HF sen-
They report that they consider the Sensidynes to work sors. All of the sensors are located, obviously, next to
very well but have a high maintenance requirement. equipment with high concentrations of acid. The system
They report almost no false alarms and use setpoints of has different sensors for HF and hydrocarbons and the
3 ppm as low level and 6 ppm as high level. alarm points for HF are set at 6 ppm. We are actually quite
“Shur Shot” Type - It works on the principle of HF pleased with the reliability, especially since the manu-
etching of a silicon oxide chip (mirror) and the light facturer has made some improvements to reduce false
reflectance off the chip. The alarm is sounded when the alarms due to contact with water during wash down, or wet
reflectance diminishes to a predetermined level in a weather.
specified time period. They are designed for cata-
strophic HF release, (i.e., over 100 ppm HF) and are JONES:
qualitative only. In one refinery, we have installed two There are a number of detector manufacturers, such as
of these which were strategically located at the most Gastech, Sensidyne, Draeger, although we have no direct
likely high HF concentration release points. These experience with them. I refer you to page 136 of last year’s
have only recently been installed and no maintenance transcript.
data exists.
Ion Mobility Spectrometry (IMS) Type - It uses a radi- QUINN:
ation source to ionize air samples and detect HE One It was desired that detection capability cover the entire
IMS is in operation at Mobil and is considered a low high-acid areas of the unit. The system that was developed
maintenance sensor with excellent detection capa- uses two different wide-area detection schemes. One is an
bilities (down to 1 ppb). Mobil has also tested the IMS open-path infrared absorption detector unit that alarms
in the laboratory and found it to have very high re- when HF crosses its path; the second is a thermal imaging
liability under varying operating conditions. vision system that will recognize when a cold cloud has
Open Path-It covers a wide area from a single point. It formed. The open-path instrument was adapted from
is dependent on the HF cloud breaking the sensor’s line Wright and Wright, Inc. (Oak Bluffs, MA); the IR cameras
of sight. are manufactured by Xedar Corporation (Boulder, CO); the
Fourier Transform Interferometer (FTIR) Type by vision computer is manufactured by Applied Intelligent
MDA - It detects by reading the IR spectrum to see if Systems, Inc. (Ann Arbor, MI) and Trigon Technologies,
HF is present. Mobil has tested the unit in a refinery Inc. (Nobi, MI) developed the custom software.
setting and determined it is capable of detecting HF
down to very low levels but does not have the service CHIANG:
reliability needed or low cost ($100,000 each) to be Since January 1992, we have installed a dozen GasTech
attractive until it has been further developed. model 2321 sensors, similar to Ultramar. They are located
“Wright & Wright” or “Detectronics” - Mobil has re- throughout the alkylation unit and around the unit vary-
viewed this Amoco developed technology and at this ing from 3, 8, or 12 feet above grade, depending upon the
time feels the technology is still in the developmental location. The detectors are electrochemical type with a
stage and not ready for commercial use. p measurement range of 0 to 15 ppm HF. We have the levels
PERF HF/H2S Sensor - Mobil is actively supporting set at 2 levels, 2 ppm for an internal alarm, and 6 ppm
through PERF the development of a Near-Infrared which notifies the air district.
Diode Laser Sensor that will have capabilities down to
1 ppb and be specific to HF and H2S detection. A
working prototype developed by Southwest Sciences, QUESTION 23:
Inc. is expected within 6 months for the HF sensor and What design philosophies and criteria are being

146
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

c o n s i d er ed f o r h y d r o f l u o r i c ac i d al k y l at i o n p l an t and one valve to sweep the acid from the reactor and settler.
acid dump systems? What have been the operating Our current design uses multiple valves to remove all
experiences with them to date? major acid pools from the unit to this separate dump drum
(again using isobutane recycle) which provides redundancy
WARWICK: (with respect to acid inventory containment) to the acid
The objective of Mobil’s acid evacuation system is to storage drum, and can be located outside the immediate
rapidly transfer in 5 minutes or less, the acid inventory turnaround work area. Both types of systems have been
from the reactor system to an off-site location. The facility reported to function as designed (typically 5 to 10 minutes
is activated by pushing a button in the control center which dump time, but longer in some more complex multiple
will automatically transfer the acid inventory from the reactor unit designs) in dump testing done on hydrocarbon
alkylation unit without any field operator assistance. The and/or acid, but we have not had a crises situation test or
system is also designed to replace the existing fresh acid performance report.
storage drum in order that the acid can also be stored
off-site. CHIANG:
All components of the system have been designed with a The principal criteria in the design of our HF acid dump
high degree of reliability which includes a triple redundant system is the South Coast Air Management District’s Rule
programmable logic controller, parallel data highways be- 1410. The rule specifies that high pressure streams must be
tween the control center and the PLC, and a UPS with dumped in less than 10 minutes and low pressure streams
redundant wiring to power all of the instrumentation. In in less than 30 minutes. Our proposed design would allow
order to maximize the reliability of the AES, the system is evacuation of the acid from our reactor and mixer/settler in
periodically tested while the alkylation unit is on stream. less than 2 minutes, dumping to the fresh acid drum. We
This system has been installed and successfully tested in are still going through the approval process with the
all of Mobil’s locations. AQMD. We have no operating experience.

MALIK: CRISLER:
We have a UOP designed, two reactor/settler system. We We are not looking specifically at rapid dump systems. At
considered installing an off-site dump drum system during our upcoming turnaround at one refinery, we will be in-
our 1992 turnaround. However, design issues such as hy- stalling automatic remote valve closers for our circulating
draulic bump during transfer and pressure versus pump acid pump, fresh acid pumps, isostripper bottoms pumps,
movement were not resolved in time, so this project will be depropanizer charge pumps, and so on. We will also include
restudied for our 1995 turnaround. logic to shut off the pumps when those valves are closed.
During our October 1992 turnaround, we will revamp
our unit to a series acid flow scheme which has significant ROBERT SHEPARD (SETPOINT, Inc.):
acid handling capabilities. For example, if an acid leak Those of you that have installed these rapid dump sys-
develops in one reactor/settler system, all the acid from that tems, are these gravity or pump systems. If they are pumped,
system can be moved to the other reactor/settler system in when you are testing them, are you turning the pump on?
about five minutes. Although this does not solve the LPG
fire problem, it will greatly reduce the rate of HF released WARWICK:
to the atmosphere while the leak/fire is being brought Ours are gravity systems.
under control.

QUINN: C. Reforming
The acid dump systems we are looking at, and actually
we have one in service, entail two different time frames.
Rapid acid dump is the first. This scenario involves rapidly
removing essentially all of the HF acid plus any entrained
Feed
hydrocarbons from the settler, risers, and coolers. Time is
important. This system has been designed to accomplish QUESTION 24:
this in 5 minutes. Again, the system has been completed, What is the experience with sulfur absorbers for
but it has not yet been tested. We intend to test it at the first the removal of trace sulfur compounds from hydro-
scheduled turnaround. t r eat ed n ap h t h as ? Ho w i s ab s o r b er p er f o r m an c e
The second system is acid deinventory. This scenario monitored? What can cause high pressure drop in
involves removing the acid from the unit over a longer time these units? Where is the absorber ideally located -
frame than the period designed for the rapid acid dump. at the HDS unit or at the reformer? Are nickel-based
The intent is to remove the acid outside bathing limits so absorbents giving better performance and reliability
that the turnaround can proceed without HF on the unit. than copper-based absorbents?

CABRERA: JONES:
UOP has designed two types of dump systems; dumps to Sulfur absorbers perform very well in the removal of
the acid storage drum, and dumps to a separate acid inven- trace sulfur compounds from hydrotreated naphthas. The
tory drum or what we call a dump drum. The first type was absorber performance is supervised by analyzing the sulfur
a very simple system design that used the isobutane recycle content of the naphtha hydrotreater stripper bottoms along

147
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

with the feeds to the reformer and isomerization units. Our are familiar with. Both types of guard beds have been very
sulfur absorbers are located after the naphtha hydrotreater effective. The relative merits have been described by other
stripper prior to the naphtha splitter to protect both the panel members.
reformer and isomerization units. Typically, we operate at
about l/2 ppm inlet to trace sulfur on the outlet. WALEED SALAMAH (SAMAREC):
We have had good luck with both nickel-based and Does anyone have any experience with this sulfur guard
copper-based absorbents. In general, the nickel based ab- bed regarding absorption of disulfide and thiophenes?
sorbents will provide a greater sulfur loading and have
been shown to be more effective in removal of mercaptans ROBERT S. BROWN (United Catalysts Inc.):
and sterically hindered sulfur compounds. We have experi- Concerning the question from the floor on disulfides and
enced some flow distribution problems. Parts of the beds thiophenes, the analyses of the sulfur compounds at the
have plugged off while others have not. This causes chan- levels involved are pretty difficult as far as differentiating
neling and premature breakthrough with waste of the sulfur types. However, based on reports and analyses that
absorbent. have been made, we have seen both disulfides and thio-
phenes removed with United Catalysts C28 nickel-based
LAVERGNE: sulfur trap. The point that I wanted to comment on was
We use magnesium oxide sulfur traps on both of our pressure drop. We have seen a couple of units that have
xylene reformers at our Houston refinery. Performance has experienced high pressure drop. We have been able to
been good with a noticeable increase in cycle length on both alleviate high pressure drop by installing a larger size
reformers. The sulfur trap is located after the feed effluent extrudate, a one-eighth inch extrudate. It looks like the
exchangers, but before the first reactor heater. The trap is problem, perhaps in both cases, was design and operation at
monitored by determining the sulfur concentration in the high space velocities.
HDS stripper bottoms and then calculating the lay down of
sulfur on the trap throughout the cycle. The maximum ARAFAH A. AL-MEZAIYEN (SAMAREC):
allowable sulfur lay down, according to vendors, is ap- We keep our treated naphtha sulfur content at 0.2 ppm.
. proximately 15% of the catalyst weight. However, due to the Whenever we start up, we use sweet naphtha, and we send
difficulty in determining an accurate concentration of total the samples to the lab and we got 4 or 6 ppm as the results of
sulfur in HDS stripper bottoms, the sulfur lay down is only the sulfur.
an estimation. We have had some minor problems with
dumping one of the sulfur trap beds due to fusing of the CABRERA:
chips. Let me understand again. You keep it at 0.2 ppm.

ARMSTRONG: ARAFAH A. AL-MEZAIYEN (SAMAREC):

One of our refineries has operated its nickel-based sulfur Yes, in storage. Whenever we start up the reformer, we
absorber for several years with good results. The absorber analyze the same naphtha, we get 4 or 5 or 6 ppm sulfur
is located in the HDS unit. The inlet and outlet streams are content in the feed.
checked for sulfur weekly. This is a very reliable operation.
Another refinery is using a copper-based absorbent. The CABRERA:
copper works well for H2S but is not very effective for If you are sampling in front of the reformer, it is most
mercaptans. It seems to grab the mercaptans initially, but likely sample contamination.
then gradually releases them over the next few days.
Neither of these units has had any pressure drop build-up ARMSTRONG:
problems. It could be that you are picking up some corrosion prod-
ucts, or some sulfide could be complexing with the iron in
CABRERA: your sample piping, and then flushing off when you catch
UOP offers two types of sulfur beds, and Mr. Jones has, I your sample if the piping has been idle.
think, accurately described the merits of both. One is the
ADS-11 system which is nickel-based and the other is a
copper-based system which is associated with the Chevron QUESTION 25:
rheniformer process. UOP commercial experience includes How does the reformer feed distillation affect the
25 ADS-11 Sulfur Guard Beds in various naphtha desul- T90 of the reformate and subsequently the T90 of
furization applications, processing the following streams; reformulated gasoline blends?
hydrotreated straight-run for downstream reforming,
hydrotreated light straight-run naphtha for downstream CABRERA:
isomerization, hydrocracked naphtha routed to down- The feed naphtha distillation as well as the reactor
stream reformers. operating pressure, severity, and feed aromatic content, all
Chevron’s rheniforming experience includes more than have an effect on the product distillation. At constant
25 Sulfur Guard Beds using the SS-D absorbent. The de- octane, pressure and feed aromaticity, increased T50 and
gree of sulfur removal in all cases is typically sufficient to end points lead to increase product T90. Reducing the end
maintain sulfur well below typical reformer maximum feed point of the naphtha feed will lead to reduce reformate T90,
sulfur of 0.5 weight ppm. We have not had any problems but will also impact refinery operations in the following
with pressure drop in any of the sulfur guard beds that we ways: reducing the reformer feed end point provides a

148
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

reduced charge rate to the reformer and a reduction of the fractionation column downstream of the reactors in
reformate and hydrogen yields, upsetting both the gasoline their catalytic reformer. What has been the experi-
pool and the refinery hydrogen balance. ence with regard to the control of fouling and cor-
If additional low end point feed is provided to maintain rosion with the use of these chemicals? Also, have
reformer charge rate, C5 + liquid product and hydrogen there been any problems in downstream aromatic
yields would still decline due to the processing of the lighter extraction units or other downstream fractionation
feeds. towers while using these chemicals?
The heavy naphtha cut removed from the reformer feed
stream must still be processed and/or blended in refinery WEBER:
product streams. The effect of the reformer feed distillation The depropanizer on our Platformer has long been
on the gasoline T90 pool is highly dependent on the indi- plagued by ammonium chloride fouling. Historically, we
vidual gasoline pool in question and the other components have been forced to water wash on-line about once every
which make up the blend. three months and more recently, on a 45 to 60 day interval.
In a typical refinery pool, only the reformate and the FCC About two months ago, we started injecting an amine into
gasoline should have a significant amount of material boil- both the depropanizer feed and the reflux. The amine would
ing above the desired T90 specification of 290 to 320°F. not clean up the fouled column so we eventually had to
Reducing the T90 in the reformate stream or full range water wash anyway, but we have continued with the in-
FCC gasoline will lead to a reduction in the gasoline pool jection with a clean column and have not really reached any
T90 value. conclusions yet.
Although it is possible to reduce the reformate T90 by It seems that the ammonium chloride fouling may still be
fractioning off the tail end of the reformate, this has two occurring so we are not sure that the amine really is going
drawbacks. First, the heavy tail is highly aromatic and to be a success story. However, to be fair, we are nearing
blending the stream into the distillate pool reduces the end-of-run with our Platformer catalyst, so the chloride
distillate product quality. Second, it costs money to first retention is low and trichlorethane makeup is quite high,
reform the heavy tail and then blend it into a product of so the load on the amine is higher than normal.
obviously lower value than gasoline.
Given the problems described above with reducing the QUINN:
reformate T90 by adjusting the naphtha end points and We are using a continuous 3% caustic wash to treat the
assuming that a suitable home can be found for the heavy high pressure separator liquid to the debutanizer in one of
FCC gasoline tail, it may be more practical to address the our facilities. We have been successful in controlling all
issue of lowering the gasoline T90 by first reducing the cut downstream corrosion. Most of our units are using a film-
point of the FCC gasoline. ing and neutralizing amine to treat the overhead in the
The reduction in FCC gasoline end point has the added first fractionation tower. Recently, we initiated a test at one
benefit of reducing the sulfur content of the finished gaso- unit using the Saltplexe chemical to treat the feed and
line which is another important criteria for meeting re- overhead stream on the debutanizer. The test is in its early
formulated gasoline specifications. This is particularly true stages and more time will be needed to fully evaluate.
because of the ascending sulfur profile as a function of the TURPIN:
boiling range in the gasoline which is produced by the I agree with Mr. Webber’s comments on the need for
FCCU. water washing to get chlorides out of the system. That
seems to me to be the best and most cost effective way to get
CHIANG: rid of chlorides. I have seen a number of chemical treatment
I really cannot add much to Mr. Cabrera’s comments programs for corrosion control used in BTX reformer de-
other than the CARB phase 2 specifications of a 300°F 90% pentanizers, but have never associated those chemical
point will force naphtha to our distillate pool and also, do treatment programs with problems in either sulfulane or
not forget, we will be cutting the front end of naphtha for glycol-based solvent extraction units. The chemicals used
benzene control. were volatile and were introduced into the depentanizer
overhead system. Probably only trace amounts went out
WARWICK: the bottom of the tower to the extraction process.
The distillation of the reformer feed has a direct effect on Injecting a compound to complex with the chloride is
the T90 of the reformate; however, the effect of the T90 of going to move the problem in a BTX unit, not solve the
the reformulated blend will depend on the relative amounts problem.
of reformate and other blending components used.
CRISLER:
GEORGE: We have not used any scavenging chemicals for control of
I agree with what has been said already. fouling and corrosion. Control of the feed sulfur to less than
0.2 ppm and nitrogen to less than 0.05 ppm have appeared
to prevent these problems. In the past, we used continuous
water injection at the top of the stabilizer. This appeared to
Process just move the salts around without getting rid of them
causing more problems than it solved.
QUESTION 26: The solution was to slump the column and wash the salts
Many refiners have been using scavenging chemi- out the bottom. Since we quit continuous water injection to
cals for control of fouling and corrosion in the first our stabilizer, we have had no fouling problems.

149
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

HOWARD J. PLATTE (Aqua Process Inc.): insure there is no caustic carryover to the downstream
Our experience with treatments of Saltplexe chemistry equipment?
requires a time period ranging from as short as 3 days to 60
days for complete control and removal of ammonium chlo- QUINN:
ride deposits. The Saltplexe treatment was designed to No, we do not water wash after the dilute caustic wash.
control -and remove deposits of ammonium chloride and
metal salts and prevent corrosion in reformers and other KENNETH N. ABRAHAMS (Star Enterprise):
related processes. I do have a follow-up comment then, on our experience.
Aqua Process has numerous successful applications We have used neutralizing amine. Recently we have had a
covering a wide variety of reformer designs such as UOP, lot of ammonium chloride fouling in our depropanizer
IFP and Amoco. Paybacks we have seen from the program column and a fair amount of corrosion we thought was
range from eight to one down to three to one by improving under-deposit corrosion. We started injecting a neutral-
RVPs or increased distillation efficiencies. To date, we have izing amine to create low melting point salts, and it did
not found any operational problems in downstream ar- solve our fouling problem, but it did not arrest the corrosion
omatics extraction units or in other downstream frac- so we are still considering what to do from here.
tionation towers. In addition, we have observed no adverse
problems regarding quality of products produced in the QUESTION 27:
reformer system or in products produced in downstream Are any novel techniques (other than chemicals)
processes. being used to remove residual chlorides from re-
former stabilizer feed and/or products to: a. prevent
KENNETH N. ABRAHAMS (Star Enterprise): tray fouling in stabilizer from ammonium chloride
Mr. Quinn, you mentioned you use a dilute caustic wash salts? b. prevent corrosion in the downstream gas
on yours. Do you use a water wash after the caustic wash to recovery unit?

150
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

MALIK: recycle gas, also ensure that excessive quantities of chlo-

Citgo installed a water wash system in the No. 5 CCR ride do not slip from the reforming catalyst.
Platformer to remove chlorides from the product hydrogen Even with these precautions, there will still be some slip
and fractionation system charge in March of 1990. Wash of chlorides from the reaction train to the stabilizer. If this
water is injected into the low pressure separator pump chloride slip is causing corrosion and/or fouling problems
suction. The water and reformate are recontacted with the in the stabilizer column, we would recommend installing a
net hydrogen product and sent through a static mixer just fixed bed system containing one of our PURASPEC chemi-
before the recontact cooler. Water rate averages about 4 vol cal absorbents in the feed to the stabilizer. Alternatively, if
% on feed. A neutralizing amine and a water-based cor- the corrosion or fouling problems are in the product
rosion inhibiter are added to the wash water. The water is streams from the stabilizer, it might be more effective to
separated in the high pressure separator and pressured place beds of the PURASPEC absorbent in the individual
into the desulfurizer product separator and reused to wash offgas, LPG or reformate streams.
the reactor feed versus effluent exchanges. ICI Katalco has a number of applications for chloride
After four months of operation, the recontact cooler ex- removal at different points around the stabilizer circuit
changers failed due to hydrogen chloride corrosion. The using its PURASPEC chemical absorbents.
bundles were then upgraded to titanium. Chloride removal
is high and has eliminated the need for water washing the
reformate fractionator columns which previously had been QUESTION 28:
required every 6 to 8 months. What are the panel’s experiences with key indi-
cators used to monitor acid/metal catalyst activity for
ARMSTRONG: catalytic reformers? What key indicators are used at
This issue was discussed at great lengths in the 1987 the end of the cycle?
NPRA Question & Answer Session in question number 30
and the 1989 NPRA Question & Answer Session in ques- WEBER:
tion number 22. Both of those discussions contained brief The key indicators for monitoring acid metal activity can
description of all of the methods that we are aware of, in really be separated into two categories. The first are cata-
addition to some of the things that have been talked about lyst parameters which rely on being able to sample the
here today. At Lake Charles, we have solid catalytic chlo- catalyst as in a CCR unit. The second are Platformer
ride traps on the recycle offgas and on the stabilizer feed process variables.
and both work very well and we have not had any sig- The key catalyst parameters are weight percent chloride
nificant corrosion problems since installation. on the catalyst and hydrogen/platinum ratio. We have a
CCR and have good success measuring both of these, send-
CRISLER: ing samples regularly to UOP. We also monitor catalyst
We do not use any chemicals or alternative treatment for surface area which gradually declines over about 100 re-
the removal of chlorides from the reformer stabilizer feed generation cycles.
and have noticed no downstream corrosion or tray fouling. The Platformer process variables are probably a little
We keep the system dry. UOP does manufacture an ad- more obvious, including C5 + yield, weighted average inlet
sorbent made of ZnO and CaO to be used upstream of the temperature and specific gravity of the recycled gas. All of
stabilizer. those can be meaningful in determining catalyst activity.

MICHAEL S. BUCK0 (Merichem Company): MALIK:

We have developed and commercialized a fiber film proc- Our experience is with continuous reformers which are a
ess to remove hydrochloric acid and ammonium chloride 22,000 bbliday and a 28,000 bbllday UOP designed units.
from the former fixed stabilizer feed streams to prevent We monitor the chloride level on catalyst and try to main-
corrosion in overhead receivers and plugging of trays. This tain 1.1 wt % on the regenerated catalyst three times each
process is unique in that it accomplishes the removal of week to keep the acid activity at the desired levels. Of
chlorides in a single step by utilizing a dilute stream of course, the water level in the system must be controlled to
caustic and a single stage fiber film contactor. By using the reasonable levels of between 15 and 35 ppm. Large in-
contactor, complete removal of chloride is assured as well as creases in chloride injection (EDC) rate is required at the
eliminating the possibility of sodium carryover to the sta- end of the run to maintain the desired chloride level on
bilizer column without the need for water wash. Currently, regenerated catalyst.
two of these units are in operation in Japan with a third
under construction. CRISLER:
To monitor catalyst activity, we watch the C5 + yield and
MARK S. DORAN (ICI Katalco): weighted average inlet temperatures, both corrected for
I would like to make two comments about chloride salts feed PONA, octane of the product, space velocity, and reac-
around the stabilizer. There are two ways to solve salt tor pressure. The corrected C5 + yields provide for the best
problems. The first and obvious one, is to minimize the measure of catalyst performance since lost yields result in
formation of salts in the first place. This is done partly by unfavorable economic conditions. The data allows us to
maximizing the removal of nitrogen at the upstream HDS extrapolate to end-of-run to determine when the economic
unit. Carefully controlling the amount of chloride injected conditions are right for regeneration. We also monitor
into the reaction circuit and the moisture content of the percent of total delta temperature in each reactor and the

151
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

water chloride balance. The water chloride balance is com- lyst may have shifted the selectivity. Thus it will be neces-
pared against the theoretical ratio from UOP correlations. sary to monitor the operation throughout the cycle. Also,
the reactor temperatures are high enough at the end-of-run
GEORGE: to affect selectivity. Consequently, it becomes harder
There are several indicators that can be used under toward the end of the cycle to be confident that the desired
steady state operation to judge the acid-metal balance on acid-metal balance has been maintained.
reforming catalysts:
a. HCL in the recycle gas. TURPIN:
b. Methane and propane yields. The key variables commonly used in monitoring re-
c. Last reactor temperature drop. former catalyst performance are the hydrogen production,
d. Octane response to temperature. light ends yield, specific component conversions, feed rate,
We recommend that the refiner add chloride to the re- feed composition, recycle rate, and reactor temperatures.
former on a continuous basis to maintain the desired acid- Different catalyst vendors will recommend these variables
metal balance. To provide suflicient water for good chloride be recorded in different engineering units or ratios, but the
mobility without causing excessive corrosion or without key variables seem to be rather consistent from one vendor
shifting yield selectivity, we recommend operating with 10 to the next. Identifying the variables to monitor catalyst
to 20 ppm vol H20 in the recycle gas. With 15 ppm H20 in activity is easy in comparison to the next three steps in
the recycle gas, the chloride addition in the feed should be setting up a catalyst monitoring program.
about 0.5 ppm wt. Generally with these levels, when the After identifying which variables need to be monitored,
recycle gas is checked with a Draeger for HCL, a value of a actions must be taken to ensure the measured values are
trace to 0.5 ppm vol is detected. If no HCL is detected, there correct. This takes the coordination of ongoing meter main-
is concern that not enough chloride is being injected or that tenance, correct sampling procedures, and data validation.
the chloride is building up on the catalyst and is not being The third step in setting up a catalyst monitoring program
evenly distributed. This can lead to an acid-metal imbalance. is establishing a base operation to which the measured
Studying the relative yields of methane and propane values can be compared.
under normal steady state operation provides an insight Our recommendation is to use a kinetic model to generate
into the acid-metal balance. Methane production is associ- yields, catalyst decay and operating conditions based on
ated with the metal function. Propane production is associ- actual feed rate, composition and unit severity. Another
ated with the acid function. Some people use the C3/C1 approach is to compare measured data to yields tempera-
ratio. Other people use a ratio called the cracking severity tures, etc., that have been compared with a base feed in
factor which is operation. Once the system is put in place, data must be
Cl
routinely collected and organized into plots or tables, and
(Cl + C2 + C3) then analyzed by a knowledgeable engineer. With today’s
On a weight basis, the ideal value normally is about 0.2. DCS systems, raw data collection, validation, manipulation
Higher values indicate too much methane production and presentation can all be done on-line.
which may suggest a low chloride level. Lower values
indicate too much propane production which may suggest a
CHARLES S. MCCOY (McCoy Consultants):
high chloride level.
In addition to what Mr. George said, methane can also be
In the last reactor, there are both endothermic and exo-
produced in a hot zone of the catalyst bed, the so-called
thermic reactions occurring. Normally, the endothermic
demethalation reaction. In particular, the ratio of methane
reactions’ temperature effect outweighs the exothermic
to ethane (C1/C2 ratio) reflects this. Production of propane
reactions’ temperature effect so that the reactor outlet
and butane is definitely controlled by chloride/water bal-
temperature is lower than the inlet temperature. If the
ance on the catalyst. But the high methane may well
outlet temperature is higher than the inlet temperature,
indicate a safety problem.
cracking is occurring. If the reactor inlet temperature
seems to be too high for the octane level of the reformate,
the cracking may be due to low chloride on the catalyst. The
methane production should be monitored. If the reactor QUESTION 29:
inlet temperature seems to be lower than expected for the What methods are being used to wash compressors
octane level of the reformate, the cracking may be due to during operation to remove chloride deposits? How
high chloride on the catalyst. The propane production successful have they been?
should be monitored.
With the reformer in a lined out condition, all the reactor ARMSTRONG:
inlet temperatures can be increased 5°F. For the octane We do not wash our recycle compressors while they are
range of about 94 to 98, this magnitude of temperature on-line to remove salt deposits. Where salts are a problem,
increase should provide an increase of 1 octane. If the we will water wash before or after regenerations and if
octane response is less than 1 octane, the catalyst may be necessary, we will take a short outage mid-run if so dictated
under chlorided. If the octane response is nearer to 2 or by vibrations. In one unit, we have installed a recycle gas
more, the catalyst may be over chlorided. For operations chloride trap and that has worked very well to keep the
nearer 100 RONC, a temperature increase of 7°F may be salts from accumulating on the compressor. On some other
required for a 1 RONC increase. units, we inject a very small amount of reformate into the
These indicators can be used any time during a run. compressor suction to help keep the salts from accumu-
However, near the end-of-run, coke deposition on the cata- lating there.

152
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

MALIK: role, as you know, has been to satisfy the ever increasing
We use a naphtha wash injected into the suction of the and more demanding requirements to produce hydrogen to
recycle gas compressor. The rate is up to 5 gallons per satisfy the various hydrotreating services due to tighter
minute. This has increased our runs from about 6 months to sulfur specifications, and recently, due to the requirements
about 12 months. The compressor is forced down with high to comply with reformulated gasoline requirements.
vibrations and requires one to two days for off-line water Gasoline reformulation de-emphasizes the octane pro-
washing. duction role of the reforming unit because of the mandated
addition of 2 wt % oxygen to the gasoline pool. The addition
QUINN: of 2 wt % oxygen in the form of MTBE translates to 11% by
We have had a lot of success with continuous on stream volume MTBE with an average blending value for research
washing with sweet heavy naphtha. Lighter naphthas tend octane number of 117. The addition of this large quantity of
to flash from the heat of compression and are less effective. a high octane blending component reduces the octane re-
The naphtha injections with spray nozzle need to be con- quired from the other gasoline pool blending streams mak-
tinuous at about 2 to 3 wt % of the gas flow. Intermediate ing the reformer a clear candidate for reduced operating
washing can result in chloride deposits slufling off un- severity, that is producing lower octane. Unfortunately,
evenly and lead to compressor vibration problems. We are when you reduce the severity of the reformer, this also
using Betz’s GT502 compressor antifoulant at two loca- translates to a reduction in hydrogen yields. However, the
tions. Although expensive, the results appear to be good. refinery hydrogen consumption increases in the re-
formulated gasoline refinery scenario.
WARWICK: Significant reductions in operating pressure are there-
We do not recommend the on stream washing of recycle fore going to be required to increase hydrogen production,
gas compressors. We have, however, effectively washed to balance refinery requirements, and maintain the overall
compressors during shutdowns. refinery hydrogen balance. Maximum pressure reduction
of the existing units may be achieved via low cost effective
WALEED SALAMAH (SAMAREC): revamps to low pressure, low severity type CCR platform-
In our experience with recycle gas, we find some sodium ing applications. Aromatics will not limit the operation
salt deposits on the compressor blades during regeneration. under reformulated gasoline production provisions. The
What type of solvent can be used to clean this type of salts? reduced operating severity, meaning the decreased ar-
omatic content of the reformate, and the dilution of the
TURPIN: reformate with oxygenates in the gasoline pool will satisfy
I assume you are getting the salt deposits during the the total toxic requirements. Further flexibility in terms of
regeneration. aromatics content of the reformate is afforded by the re-
First, use a quill for the water injection to ensure water duction or elimination of benzene from the reformate.
dispersal in the effluent gas. If you increase the water wash We have discussed a few routes of how and what can be
of the gas during the regeneration, you will extract more of done and we will address more issues later when answering
the salts, and that will probably solve your problem. Also, other questions. A 0.5% reduction in benzene is approxi-
check the water makeup/purge rates to ensure the water mately equivalent to a 5% increase in aromatics when
wash has not become concentrated with salts to the point keeping the remaining toxics neutral. The above, I believe,
that it can no longer extract the salts in the gas. As a last demonstrates that the reformer will shift from a swing
resort, you can go to a compressor wash using water or octane producer to a swing hydrogen producer.
water and a soap. If you decide to wash the compressor,
start the wash at the beginning of the regeneration, not ARMSTRONG:
part way through the regeneration. In addition to the items that Mr. Cabrera outlined, the
fact that we will be required to meet annual average pool
ARMSTRONG: specifications will put additional pressure on the operation
We have had some salt deposits that were very sticky and of the reformer and limit its flexibility. There will be new
would not come off the compressor just with simple water constraints placed on the timing for fixed bed unit re-
washing. We added some low sudsing laundry detergent, generations and for maintenance turnarounds on all types
and cleaned the compressor after five or six flushes. of reformers. The specifications must be met on a calendar
year basis, but as we all know, our units do not tend to
follow the calendar with their operating cycles.
QUESTION 30:
What does the panel foresee as the impact of GEORGE:
reformulated gasoline specifications on the oper- Reformulated gasoline appears to affect the operation of
ation of reforming units? the reformer in at least four ways:
Limit on benzene in reformulated gasoline would most
CABRERA: likely result in the feed to the reformer being de-
The role of the catalytic reforming unit has historically hexanized to remove the benzene precursors.
been that of an octane upgrader. The advent of unleaded Concern over the T90 of reformulated gasoline may
gasoline in the 1970’s and the 1980’s placed additional result in the reduction of T90 of the feed to the reformer.
burden on the reforming unit as the primary provider of The use of oxygenates, which has been mentioned, will
high octane blending components. The other less obvious have a high octane blending value which will result in

153
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

the operation of the reformer at lower octane. WEBER:

The use of oxygenates will require less reformate. Thus We use essentially the same procedure as Mr. Malik
the feed rate to the reformer may be reduced. mentioned. The only other point I would like to make is
I would like to add that countering the above possible that when the catalyst is reloaded, it is helpful, although
consequences will be the operations integrated with chemi- not required, to reload it with the same carbon profile, in
cals. In these cases, the aromatics could be extracted. Con- other words, the highest carbon material in the last reactor
sequently, the reformer operation can continue in the same and so forth. That way, the first regeneration cycle will not
fashion as today. Also, the reformer is a primary supply of experience atypical carbon levels when the first reactor
hydrogen, as has been mentioned, with the increased re- charge is eventually circulated to the regenerator.
finery demand for hydrogen dictated by lower sulfur diesel
and gasolines specifications. Pressure will be on the re- CABRERA:
former to supply more hydrogen. In some cases, the need for Just a couple of guidelines on the safety side. It is impor-
hydrogen will be critical enough to require the reformer to tant that the catalyst be preferably below 130°F. The key is
operate at today’s severity or higher, or as mentioned ear- that you do not expose the catalyst stream to oxygen.
lier, operate at lower pressure, and treatment methods Nitrogen, CO2 extinguishers or some other inert gas is
would be employed after reforming on the reformate to required. From a practical perspective, one of the things
meet the gasoline specifications. that I have found very effective in preventing pyrites spark-
ing when you load the catalyst into the drums, is to put dry
TURPIN: ice in the bottom of the drum.
I agree completely with what Mr. Cabrera said earlier.
Reformer units are going to be run for hydrogen production. TURPIN:
Reforming catalyst unloading and storage procedures
WARWICK: must meet three objectives: 1) minimum catalyst breakage,
Reformulated gasoline production will result in lower 2) minimum contact with moisture, and 31 personnel safety.
severity operation of reforming units. Higher feed IBP and Reactor dumping techniques should minimize the free fall
lower feed end point will result in lower throughputs. of catalyst, and the catalyst should be unloaded into the
Octane requirements will be lower due to the presence of containers that will be used for storage and subsequent
oxygenates. Therefore, required pressures and recycles reactor reloading. Keeping moist air away from the cata-
rates will be lower. lyst can be accomplished by purging the containers with a
dry gas. The easiest way to do this is to put chunks of dry ice
in the bottom of the container. This allows a slow purge of
QUESTION 31 the catalyst, and at the same time generates gas to fill the
What precautions are taken during unloading to container as the catalyst cools. If a dry gas is not being put
safeguard reforming catalysts that are intended for in the storage containers, or being generated within the
reuse? containers, wet air will be drawn into the container as the
catalyst and surrounding gas cool. Personnel safety must
MALIK: be addressed when specifying the container purging tech-
When unloading catalysts from our CCR Platformer for nique to be used and the purging of the reactor while the
reuse later, we screen the catalyst as it is unloaded to catalyst is being dumped.
remove any fines and chips, and then store it in catalyst
bins. Some material can be pyrophoric and extra pre- CRISLER:
cautions may be necessary. Cover the drum as soon as the I have nothing to add.
catalyst is unloaded and if necessary, dry ice (solid COz) can
be placed at the bottom of each drum before it is tilled with GEORGE:
catalyst. It is important to keep the catalyst as dry as Reforming catalyst can be unloaded either unre-
possible so unloading is stopped if it rains and any loaded generated or regenerated. The reforming catalyst will be
bins are covered with plastic sheets. Also, it is important to safer to handle if it has already been regenerated. Iron
segregate the so-called heels catalyst in separate storage sulfides found especially in lead reactors are pyrophoric. If
bins so it can be returned for platinum recovery. The high they begin burning on regenerated catalyst, they can ignite
carbon content heels catalyst should not be reused in the the coke on the catalyst which will lead to a hazardous
reactors because of potential problems with high tempera- situation. Regeneration will remove the coke from the
ture and damage to the catalyst regenerator. If the catalyst catalyst and will convert any iron sulfides present in the
is to be stored for prolonged periods, we would place it catalyst to iron oxides which are not pyrophoric.
indoors. Some refiners are using off-site regeneration services for
their reforming catalysts. The catalyst should be cooled to
WARWICK: as near ambient as possible in the reformer. The catalyst
If you plan to reuse dumped reforming catalyst, you should be maintained under inert gas conditions until it is
should burn the coke and cool the catalyst before dumping. regenerated at the off-site facility. Personnel working with
While dumping, avoid free fall to minimize catalyst break- unregenerated reforming catalyst should be protected from
age. Screen the catalyst to remove fines and store in sealed the dust getting on their skin. Plans should be formulated
drums in order to keep it dry. and instituted to handle the catalyst if it should become

154
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

removal steps. Catalyst samples showing greater than 0.1

wt % sulfate either from a previous regeneration or from a
just completed coke burn, will generally require a sulfate
removal step. Also, severe sulfur upsets or a constant ex-
posure of the catalyst to feed sulfur levels greater than 1
ppm during the previous cycles are good indicators that a
sulfate removal step should be performed.
In any case, the presence of sulfate in the catalyst pro-
motes platinum crystallite growth and competes with chlo-
ride for sites on the catalyst causing a severe loss of catalyst
activity and stability. The best procedure to completely
remove sulfate from the catalyst and the plant is to do a hot
hydrogen sweep after the coke bum and short oxidation.
Sulfur present on the catalyst prior to the coke bum is
strongly absorbed on the catalyst surface and effectively
blocked from removal by hydrogen stripping due to the
presence of significant levels of coke. It is only when the
hot. Spraying the catalyst with water is not recommended if carbon has been completely burned off the catalyst and the
the catalyst is to be reused. sulfur converted to sulfate that the hydrogen stripping can
be completely effective and therefore remove the sulfur.
CHARLES S. MCCOY (McCoy Consultants): After the sulfate removal step, an additional reoxidation.
It really is not the catalyst that is pyrophoric; it is the step is recommended to fully restore the platinum to a fresh
finely divided iron. The catalyst, iron free, is not pyrophoric. highly dispersed state.

CABRERA: CRISLER:
That is correct. We use the UOP process.

ARAFAH A. AL-MEZAIYEN (SAMAREC): GEORGE:

Is it possible that the catalyst itself is broken into small While sulfur can be removed from the catalyst in either
pieces after loading two times? the reduced form or the oxidized form, it is easier to remove
in the reduced form. Consequently, there are recommen-
TURPIN: dations to do a hot hydrogen sweep prior to burning coke
Yes. It is not uncommon at all to have some catalyst from the catalyst. As a sulfide, most likely platinum sul-
break, particularly when you are dealing with extrudates. tide, the hydrogen purge can remove the sulfur to very low
If you are dealing with a CCR like Mr. Weber was talking levels at about 950°F.
about, the catalyst is spherical and has a much lower Once sulfur on the catalyst is oxidized to sulfate, it reacts
tendency to break. The amount of breakage and resulting with the catalyst carrier and is tightly bonded. The sulfate
fines production is relatively small with new generation removal procedure converts the sulfate to sulfide which can
catalysts as compared with catalysts 20 years ago. Exces- be removed by hydrogen purging. Limitation of sulfates
sive breakage in a new catalyst is an indication of some removal procedure is the ability to fully convert all the
problem with the base material. sulfates to sulfides. It may take several cycles to get the
sulfates removed.
FLESHMAN: TURPIN:
We would recommend that you screen it when you load it We have a little difference of opinion here. However, I
back into the reactor rather than when you dump it to avoid have a tendency to agree with Mr. George that the best
any fines generation from handling after it has been thing to do is get that sulfur off the catalyst before the
screened. regeneration. In my experience in running units, I have not
had the problem of getting sulfur off before regeneration.
QUESTION 32:
Talking to other people, at times when they have not got the
sulfur off ahead of time, they have been able to do a hydro-
It is usually recommended to have a hot hydrogen gen sulfate strip, but they wish they had stripped it off as a
sweep prior to carbon burn during the regeneration sulfide before regeneration. The best route to take is to
step to protect against forming sulfates in the event keep feed in the unit, get the HDS unit problem corrected,
that the catalyst was poisoned with sulfur. However, and have the catalyst sulfur free prior to shutting down.
one procedure recommends a sulfate removal step
after carbon burning. What is the best way of remov- QUINN:
ing sulfur from reformer catalyst? Does any chemical The only thing that I would like to add is the best way to
reaction take place between reformer catalyst and deal with sulfur excursion problems is to avoid it. Ideally if
sulfur in feed? significant sulfur excursion occurs, reactor temperature
should be lowered significantly in order to minimize the lay
CABRERA: down of coke and sulfur on the catalyst. Temperature levels
The history of the catalyst with respect to sulfur poison- should not be increased until the H2S level in the reformer
ing will, to a large extent, determine the need for sulfate recycle gas returns to typical levels.

155
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

WALEED SALAMAH (SAMAREC): catalysts form the base case against which the performance
I have a question for Mr. Cabrera. Can you give us, of other catalysts with different metal ratios are measured.
please, a figure for the sulfur on the catalyst after sulfate The initial activity and selectivity of equal metal and
removal? skewed metal bimetallic catalysts are essentially the same.
With a metal ratio of 2 to 1 rhenium to platinum, the
CARRERA: stability of the catalyst can be enhanced by a factor of 2 if
Mr. Furfaro, could you please answer Mr. Salamah’s feed sulfur levels are low enough.
question? Thus, it is possible to obtain a cycle length twice as long
with the skewed metal catalyst as with the equal metal
ANGELO FURFARO (UOP): catalyst. The interaction of the platinum and the rhenium
In answer to that question, I agree with Mr. Cabrera as to in the skewed metal catalyst is critical for good per-
procedure. l)pically, during the sulfate removal step, we formance. This interaction affects the ratio that can be
try to monitor the peak sulfur dioxide evolution from the used. This interaction is also affected by sulfur in the
catalyst and then watch that level drop off to under 5 ppm. system. It is generally recommended that a sulfur guard
This works as long as the customer is not resistant to reactor be used with a skewed metal catalyst to obtain the
waiting that long, because it can take awhile. We tend to full potential of the catalyst.
find 10% to 20% of the original sulfur level remaining on Commercially, there are many pounds of both equal
the catalyst at the end of the sulfate removal step. The metal and skewed metal reforming catalyst in operation.
problem is the desire to get quickly back on stream. We find Generally, the equal metal catalyst is more robust and
this step to take anywhere from 24 to, in extreme cases, 60 tolerant of feed quality and operational upsets. If protected
hours just for this sulfate removal step. well from feed quality upsets, and if strict adherence to
procedures is maintained, the skewed metal catalyst can
WALEED SALAMAH (SAMAREC): perform well and provide the expected longer cycle length.
From our data which we have, I think we have a much
higher number than the number you gave, but did you CABRERA:
expect the existence of any type of iron sulfide in the heater The stability advantage with skewed metal catalyst,
piping that emigrate and cause the high sulfur on the such as UOP R62, Chevron Type-F, is very well proven in
catalyst? over 225 commercial operations. On the other hand, newer
skewed type metal products such as R56, are now on the
CABRERA: market and are showing improved activity while still re-
I am not sure it would migrate but certainly some of it taining the stability and yield characteristics of the older
will be evolved. You will remove some and it may show in type of skewed metal catalysts.
your analysis, but I do not think it will migrate to the
catalyst. TURPIN:
The ratio of platinum to rhenium in a bimetallic re-
WALEED SALAMAH (SAMAREC): forming catalyst has very little meaning in and of itself.
This is correct because you already pass through the The platinum, of course, is responsible for the cyclization
carbon and the short oxidation steps where you have high and dehydrogenation reactions that occur in the reactors. A
oxygen content in the gas and high temperature. catalyst with a particular platinum loading is recom-
mended by a catalyst manufacturer based on the unit
ARMSTRONG: design, operating pressure, reactor volumes, feed types and
We have seen some sulfur that migrated out of the fur- operating severity. Rhenium is one of the commonly used
naces onto the catalyst after we thought we had it removed promoters in semi-regenerative reformer operations. It in-
off the catalyst. hibits coke laydown on the catalyst surface. This allows
longer runlength, higher yield over a specified runlength,
CHARLES S. MCCOY (McCoy Consultants): or higher severity operation over the run. The good news is
I remember one case where the catalyst held an incred- the higher the rhenium concentration, the more coke that
ible 3 wt % sulfur at end-of-run. It took several cycles of is inhibited. The bad news is, rhenium is very sensitive to
REDOX to remove it. Without REDOX, it probably would sulfur. If a refiner can control the sulfur in the feed to very
have taken about a week, just circulating hot hydrogen. low levels, the amount of rhenium on the catalyst can be
increased. Controlling the sulfur out of the hydrotreater is
difficult because of the limits and difficulties of the test
QUESTION 33: techniques used in the sulfur analysis. Just as is found on
Wh at i s t h e r el at i o n b et w een t h e p l at i n u m an d any PI control loop, control on a process is dependent upon
rhenium ratios, the activity/stability of the catalyst precision of the feedback. If a refiner is considering putting
and the liquid volume percent yield? Please comment a skewed catalyst in a reformer, do sufficient testing on the
on the commercial performance of skewed and nor- hydrotreater ahead of time to ensure a low sulfur operation
mal bimetallic reforming catalysts. can be maintained.

GEORGE: CEES SMIT (Akzo Chemicals International):

Equal metals loaded bimetallic reforming catalyst have I would like to share with you a commercial experience
platinum and rhenium present in a one to one ratio. These about the difference between a skewed and a balanced

156
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

catalyst. We have a customer who has two, almost equal vided that a tube wall has been cleaned from any scale
reformers; one loaded with the skewed catalyst, the other during unit turnaround.
with a balanced catalyst. They both get essentially the
same feedstock, and it was observed that, indeed, the bal- MALIK:
anced catalyst has in the early part performed superiorly We use an outside infrared survey firm to check our
(1.5%) to the skewed catalyst. But when the cycle pro- furnaces via infrared thermography about every six
gresses, the performance switches and at the end, the value months. The operators check the tubes for hot spots during
of the overall performance is about equal. Given the fact their rounds. We have not had any problems with the coils
that the operation with the balanced catalyst is less critical, outlet temperature being as high as 975°F
most of our European refiners prefer the balanced catalyst.
The performance difference has been graphically shown in WEBER:
a paper, “New Developments in Akzo Reforming Cata- We have replaced virtually all of our knife edge type
lysts”, presented by E. von Broekhoven and l? de Bruin thermocouples with the refractopad type that Mr. Flesh-
during the Akzo Catalysts Symposium, 1991 (page 152, man described. We find them to be a lot more reliable.
figure 3). Moreover, it should be said that the optimal use of
skewed catalyst is in staged catalyst systems in semi-
regenerative units. QUESTION 35:
What is typical expected life for 2-1/4 chrome, 1
MO tubes in reformer service? What criteria is used
Mechanical for tube replacement?

LAVERGNE:
QUESTION 34: The design life for API 530 standards is 100,00 hours or
What is the current practice for measuring furnace 11.4 years. Since we have not operated our reformers at full
t u b e w al l t em p er at u r es i n c at al y t i c r ef o r m i n g design temperature and pressure of the standard, we feel
furnaces? we should get 25 years of service. Our BT reformer has
recently achieved this milestone.
FLESHMAN: Estimation of remaining tube life is very difficult. We
We recommend a thermocouple attachment block of the had an outside consultant perform a remaining tube life
same material as the tube. The block is welded to the tube estimate on our other reformer that has 22 years of oper-
in the manufacturers shop. In the field, the thermocouple ation. They indicated 10 years of remaining life based on
sheath is then inserted into the block and the block is destructive testing of tube samples.
peened over the sheath. If welding to the tube is restricted, The criteria for tube replacement is based on whether the
such as with low chrome alloy tubes requiring post-weld tubes have reached third stage creep.
heat treatment, this is the only method we would normally Practically, we have not allowed more than 3% cir-
use. After this, we then clamp the sheath to the tube in cumferential growth on tubes in hydrogen service. Care
order to limit stress from tube expansion and coil the must be taken to remove all scale prior to taking this type of
sheath four times between the tube and the furnace wall to measurements. We use strapping and ultrasonic thickness
act as a stress relief. In those cases where welding to the testing coupled with visual inspection to detect this.
tube is allowed during erection, we would alternately use a
weld-pad type thermocouple. The thermocouple is welded MALIK:
to a pad with the same material as the tube, and the pad is We operate two CCR reformers which were started up in
then welded on three sides to the tube. The pad would then 1971 and 1983. The older No. 4 Platformer has 9 chrome-l
be covered with insulation and a shield. molybdenum tubes, while the newer No. 5 Platformer has
2 114 chrome-l molybdenum tubes which have been in
CHIANG: service for over 9 years. Tube replacement is determined by
Our operators routinely monitor for hot spots and bal- physical appearance of the tube, such as bulging, cracking,
ance burners appropriately. Our metallurgy is 2 l/4 and sometimes examination of microstructure from in situ
chrome, 1 molybdenum, which is good to roughly 1200°F If metallography. Samples can be cut from the tubes and a
more accurate temperatures are needed, the refractopad metallographic examination or a stress rupture test can be
type temperature indicators that Mr. Fleshman described preformed to determine the condition of the tubes. Typi-
are very accurate. We have not had anyreliability problems cally, the physical appearance of the tubes give the initial
with our tubes. indication that the tubes need to be replaced.

GEORGE: QUINN:
Tube wall temperature measurement is currently prac- The expected life of 2 l/4 chrome molybdenum tubes is
ticed with thermocouples designed for these applications. entirely dependent upon the unit operating conditions. Our
The two critical issues in tube skin temperature measure- metallurgists use API 530 along with the known operating
ment is thermocouple design and routine. I would like to temperatures and pressures, then use the Larson-Miller
refer you to a paper written by James Seebold of Chevron on Parameter (LMP). We feel that these techniques are very
tube skin thermocouples, published in the Chemical Engi- conservative.
neering Progress, December, 1985. Optical parameters are We recently did some destructive stress rupture testing
also used to measure furnace tube wall temperatures pro- on a sample that had been in service for 12 years, and the

157
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

results said that there was still 37 years ofremaining life at more likely the metallurgy specified will meet future
1100°F. needs. As to whether to use 1 1/4 or 2 1/4 chrome reactors,
what is the incremental cost of going to the 2 1/4 chrome
WEBER: and what is the economic impact on the plant if the lower
I have nothing further to add to that. We agree with that. chrome reactors will not be suitable for future operations?
Given those economics, a simple DCF calculation will give
CHIANG: the answer.
I do not have anything to add either. Our tubes have had
very good reliability.
QUESTION 37:
CABRERA: Is c o r r o s i o n c o m m o n l y ex p er i en c ed i n t h e r e-
We agree with the comments of Mr. Lavergne from the f o r m i n g r eac t o r c i r c u i t , p ar t i c u l ar l y i n t h e t r i m
design criteria and retirement practice. We agree with the coolers, condenser and the piping from the reactor
comments of Mr. Quinn as far as what determines the effluent exchangers to the trim coolers? What is the
ultimate life concerning the quality of operation that the cause and how do you control this problem? Should a
furnace is exposed to. m at er i al o t h er t h an c ar b o n s t eel b e u s ed i n t h i s
service?

QUESTION 36: LAVERGNE:

What is the recommended metallurgy for stacked We have not experienced any significant corrosion in the
r ef o r m i n g r eac t o r s ? Is l -1/4 c h r o m e ad eq u at e o r trim coolers, condensers, and or piping during normal ser-
should 2-114 chrome be considered? vice. Our six condensers, four admiralty tube and two
carbon steel bundles, are all in service until regeneration,
CABRERA: reactivation, and H2S scrub activities are initiated. At that
Both 1 l/4 Cr l/2 MO, and 2 l/4 Cr l/2 MO materials can time, we block in the four admiralty exchangers and direct
be used for stacked reforming reactors. Both of these mate- the effluent through the two carbon steel exchangers,
rials, as is usual when one is confronted with material commonly called our “sacrificial” carbon steel bundles. We
selection, have pluses and minuses. Mechanically, 2 l/4 protect the exchangers during this extremely high cor-
chrome has a higher allowable tensile stress than 1 l/4 rosive period by injecting boiler feed water, soda-ash solu-
chrome. But both materials reference the same com- tion, and water soluble corrosion inhibiter, via a spray
pressive stress in the ASME curve. The 2 l/4 chrome tends nozzle assembly and we have had excellent results.
to be slightly more expensive but not drastically so. From a Admiralty was the original design primarily because of
mechanical standpoint, it is highly dependent on other ineffective cooling water treatment when the system was
influences, such as high external wind or seismic loads. built. Admiralty is not practical for use during re-
Metallurgically, 1 l/4 chrome is susceptible to creep em- generation, reactivation, and/or H2S scrub periods because
brittlement. UOP has attempted to offset this problem of its susceptibility to stress cracking in the presence of
through tighter requirements on 1 l/4 chrome steels than ammonia, moisture, and oxygen.
the ASME requires. On the other hand, 2 l/4 chrome is The effluent coolers, those isolated and those used for
susceptible to temper embrittlement. UOP has likewise regeneration, must be water washed to remove ammonium
attempted to mitigate this effect through tighter metal- chloride deposition, thereby minimizing the possibility of
lurgical requirements as well. ammonia chloride carryover into the compressor immedi-
ately after shutdown.
FLESHMAN:
The only thing that I have to add is check the Nelson QUINN:
curves for resistance to hydrogen attack. We have seen some very minor corrosion in the reformer
final coolers and downstream equipment. The key to cor-
MALIK: rosion control, we believe, is keeping the circuit dry.
I want to add that both our Platformer reactors are Hydrogen chloride in the reformer circuit is not corrosive in
constructed of 1 l/4 chrome, half molybdenum steel. We the absence of water.
believe this material to be adequate based upon our experi- Now let me tell you one short horror story. At one loca-
ence. We have had some embrittlement problems with the tion, a filming amine was being added, neat, to the Ultra-
1 l/4 chrome half molybdenum material at our older unit. former effluent just upstream of the effluent air cooler. The
One may want to consider a 2 1/4 chrome, 1 molybdenum subcooled carbon steel spray nozzle corroded and failed. As
for higher resistance to hydrogen damage utilizing the a result, the neat filming amine was injected along the wall
Nelson curves. and corroded a hole in the pipe immediately downstream of
the injection point. The lessons learned are always use
TURPIN: stainless steel, not carbon steel injection quills, and this
UOP is going to recommend the correct metallurgy for a type of chemical, which is in itself corrosive, should never
given reactor application. That is their job, and they do it be injected neat.
quite well. Their selection of a metallurgy is based on the
specification provided by the client and this is where prob- CABRERA:
lems can develop. The more a client can tell the supplier Our experience says that this is not a very general, wide
about the client’s expectations of future operations, the spread problem. Most of the corrosion problems occur as a

158
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

result of regeneration or shortly thereafter. From the UOP specifies that the internal expansion bellows shall be
metallurgical perspective, we think that carbon steel is provided with an internal sleeve attached at the top and
adequate and has proven to be adequate when the recom- open at the bottom (inlet end). The vendor shall include a
mended regeneration procedures have been followed. contingency of 2 inches (50mm) compression in the design
Monitoring the caustic solution for total alkalinity in ad- of the bellows for possible upset conditions that will not
dition to pH helps reduce the possibility of unit corrosion. exceed 100 cycles over the life of the bellows. Upset condi-
The caustic circulation rate and the injection location play a tion considered for this discussion is a compressor failure
very important role in controlling the possible corrosive where the tubes would heat up faster than the shell and
elements. cause compression of the bellows. This contingency is in
addition to the normal elongation/compression values rec-
CHIANG: ommended by the exchanger vendor at normal design
I know the question is pointed at downstream from the conditions.
feed effluent exchangers, but last year we took a good look
at our feed effluent exchangers and upstream to the reactor FLESHMAN:
outlet, and we found hydrogen attack in the carbon-1/2 Under normal operations, these units will expand ap-
molybdenum metallurgy. We replaced our feed effluent proximately l/2 inch. Because of upset conditions, however,
exchanger and our stabilizer reboiler with low chrome we normally design for +/- 2 inches of expansion for
metallurgy. So keep the Nelson curves in mind. approximately 200 cycles.
CRISLER:
While not severe, we have experienced corrosion in the QUESTION 39:
piping from the reactor effluent exchangers and the trim Has an y r ef i n er y h ad ex p er i en c e w i t h i s o l at ed /
coolers. The cause of the corrosion is believed to be high scattered pitting corrosion to the overlay/clad mate-
chloride and water concentrations encountered during oxi- rial of reforming reactors? What could be the cause
dation in our regeneration. We do inject caustic during the and how do you prevent it?
regeneration, but pH is often difficult to control and we
have had problems with the caustic injection nozzle. We do ARMSTRONG:
not believe that corrosion is occurring during the run cycle. Conoco does not clad or overlay reforming reactors. I am
assuming this question was referring to the naphtha HDS
reactors and will make my comments accordingly. We do
QUESTION 38: clad our HDS reactors with 321 or 347 stainless alloys,
In the design of bellows units for large vertical typically over 1 l/4 chrome, l/2 molybdenum steel. We have
countercurrent heat exchangers, what compression not experienced any pitting corrosion, but would suspect
and/or extension allowances should be included to that if you are seeing that, it is a chloride attack during
cover emergency shutdowns, other process upsets downtime or start-up. Prevention would consist of soda-ash
and normal start-up and shutdown conditions? neutralization per the NACE procedure RP0175-85. It is
important that low points and hard to reach spots be
QUINN: thoroughly flushed with the neutralizing solution.
The large vertical countercurrent heat exchangers at one
of Amoco’s Ultraformers are designed considering process FLESHMAN:
upsets, emergency shutdowns, as well as start-up and We agree, we just emphasize that we do not expect to see
shutdown conditions. The normal operating conditions any corrosion during normal operation. However, it can be a
were expected to be 10,000 cycles of concurrent movement problem during regeneration. If the temperature falls
in the lateral direction. The design also included 5,000 below the dew point, corrosion can occur due to the forma-
cycles of nonconcurrent movement, both compression and tion of acidic condensate.
extension. And finally, in upset conditions, the bellows
were designed to take with 100 cycles of nonconcurrent MOHAMMED AL-NAGHASH (SAMAREC):
movement, primarily compression. I would like to ask the panel or the audience if there is
any interval between inspecting the vertical exchanger on
JONES: the Platformer and is there any precaution to be considered
This is a function of the length of the exchanger and for taking these out of service to check them, the blind
expected operating temperature which determines the forward with the lines? What are things that you have to
growth length of the tubes and the shell. Our bellows are consider for the maintenance?
designed for 3/4 inch compression and 3/4 inch extension
with a l/8 inch lateral offset. We have had two cracks in ANGELO FURFARO (UOP):
these bellows in 12 years and we ended up installing a If you are speaking of the vertical combined feed ex-
sleeve in the bellows to control the lateral movement. changers, I was just reviewing that the other day. Our
experience indicates that less than 10% of our units have
CABRERA: been required to pull the bundle for any sort of inspection or
The previous speakers have covered a considerable part maintenance purposes. However, some refiners have been
of our design philosophy. UOP’s practice is summarized known to pull the bottom head, say on every other major
below. inspection, four years or five years, something on that
For vertical combined feed exchangers with the feed on order. Typically, you can see how things look by detaching
the tube side and a tube length of approximately 65 feet, the inlet elbow. As far as safety goes, typical refinery

159
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

procedures, which are getting more strict every year, would

address the safety practice, but typically every inlet and
outlet nozzle is blinded on the vessel that is going to be’
inspected. Local or country inspection regulations may
require more thorough or frequent inspections.
This question came up on a more general form, pres-
surized vessels, in this year’s spring NPRA annual Re-
finery and Petrochemical Plant’ Maintenance Conference
in San Antonio. In that discussion, many refiners stated
they followed the instructions of API 510 inspection code
for scheduling inspection of pressure vessels. That code
states the maximum interval for the inspection of pressure
vessel in corrosive service is half life, or ten years, however,
that inspection can be an internal or on stream inspection.
This code would likely rule in areas where local regulations
are not more stringent. One last note, many refiners do
appear to inspect pressure vessels much more frequently
than every ten years based on experience, equipment his- zene emissions and exposure during unit cleanout
tory, service type, and corporate policy. an d m ai n t en an c e? A r e t h er e an y al t er n at i v es t o
steaming?
WALEED SALAMAH (SAMAREC):
Has anybody experienced damage to the scallops of the MALIK:
reactors and what is your recommended cycle for dumping This question was answered in considerable detail in the
the catalysts and the inspection of these scallops? 1990 NPRA Q&A Transcript on page 146, Question 12. We
essentially use a combination of the procedures mentioned.
CABRERA: This involves pumping out the liquid inventory to the slop
Several of our units have reported damage to scallops. system and steaming out to flare. The entry permit is then
The extent of damage depends specifically on what was the issued by the safety department after checking for benzene
cause. Most repairs are case specific so I would have to levels.
discuss with you the details of what caused the damage and
subsequently address how to fix it.
ANGELO FURFARO (UOP): E. lsomerization
I just wanted to add that typically from an inspection
standpoint, especially on older units, we would recommend
that the reactor stack be inspected at least every three to QUESTION 41:
five years. That should be sufficient to catch any sort of What has been the recent experience of alterna-
changes in mechanical integrity. tives to carbon tetrachloride as a chloriding agent for
In regard to scallops, as Carlos Cabrera stated, scallop isomerization units, e.g. perchlorethylenel
damage has occurred in some units, and individual cir-
cumstances should be reviewed. However, UOP’s scallop CABRERA:
designs have been improved over the years such that scallop There are currently four isomerization units using an
damage has been reduced significantly, and in most in- alternative to carbon tetrachloride. They are using per-
stances, eliminated. chlorethylene. To date, no problems have been reported
For fixed bed units, UOP would recommend that the from any unit that has switched. The longest operation that
catalyst be dumped and screened every two years, or three we have on record is from November 1991. We have recently
regenerations, whichever comes first. approved the use of perchlorethylene as a substitute for
carbon tetrachloride if carbon tetrachloride cannot be
OSCAR ALDAMA (Coastal Refining & Marketing): obtained.
This question is regarding the vertical feed exchangers.
We just recently pulled our exchangers for inspection, and ARMSTRONG:
we thoroughly hydrotested them. We found no evidence of I guess we are one of the four. We have been using
leakage after 14 years of service. When inspecting this type perchlorethylene in a butamer unit for more than a year
of exchangers, the key is to be sure to get your procedure with no apparent problems. We think the real test will be
established so you do not damage the expansion bellows at the ultimate effect on the catalyst runlength, which we
the bottom of the exchangers. have not been able to see any difference in so far.

JONES:
D. Aromatic Extraction I guess we are two of the four. We inject perchlorethylene
not only in our penex unit but in our CCR unit as well.
QUESTION 40: ARMSTRONG:
Wh at p r o c ed u r es ar e u s ed t o c o n t ai n /c o n t r o l b en - I would like to add that you must use a catalyst grade

160
TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

THOMAS E KELLE’IT (CDTECH):

A. Alkylation-Sulfuric Acid The perceived drawback could be carryover of oxygenates
to the alkylation unit, causing higher acid consumption.
The use of a distillation column to remove traces of DME,
Feed water, and methanol from MTBE raffinate has been dem-
onstrated to produce less than 50 ppm oxygenates. With
QUESTION 1:
TAME raffinate, the separation is even easier, due to the
larger difference in the boiling points between DME and
Within the past few years, it has been suggested the G5s.
that a possible method for reducing volatile olefin Proper design and operation of the C5-TAME fraction-
content of gasoline is to process refinery C5 olefin ation will minimize the ether content of the C5 raffinate.
streams in a TAME unit and then utilize the raffinate The combination of these steps can produce a C5 raffinate
as an alkylation feed. Have any refiners used this with less than 5 ppm oxygen. There is also an advantage to
scheme? If so, what are the results? If not, what are producing TAME prior to C5 alkylation. STRATCO has
the perceived drawbacks? indicated that acid consumption is decreased significantly
by removing isoamylenes from the C5s.
LAABS:
Our Louisiana plant has a TAME unit in operation, and
they have charged up to 25% C5 rafflnate to the alkylation QUESTION 2:
unit with no problems. The alkylate produced has a lower What are typical contaminant levels in LPC
RVP, but it results in a higher acid consumption and a lower streams from FCCU’s? Are guard beds used to protect
Research and Motor octane product. They plan to alkylate selective hydrogenation reactor catalyst?
the C5 rafflnate in the summer, when they can take advan-
tage of the lower RVP alkylate to blend low valued products PAULES: -
into the gasoline pool. They have no plans to operate this The maximum contaminant level permitted to one par-
way in the winter. ticular selective hydrogenation process is as follows:
Our Delaware plant is in the process of building a TAME
unit. The project was designed with amylene alkylation in - Total sulfur <40 ppmw
mind, although we probably will be limited by isobutane - H2 S < 5 ppmw
availability. They warn of the situation when there is a - Mercaptans <35 ppmw
significant amount of isobutylene in the TAME unit feed. - Carbonyl Sulfur < 2 ppmw
The MTBE that will be made is much lighter than the - Water Saturates @ 100 °F
TAME, so it will be carried into the C5 raffinate cut. MTBE - Arsenic < 5 ppbw
in alkylation feed will result in a large acid consumption - Sodium < 1 ppmw
penalty. - Basic Nitrogen < 0.5 ppmw
They are also concerned about the economics of amylene - Existent Gum < l0 mg/lOO ml
alkylation. C5 raffinate will have a low C5 olefin con- - Oxygenates < 50 ppmw
centration, so a small alkylate gain can be expected. This We operate a selective hydrogenation unit on a FCCU C4
benefit could be balanced by the additional iso- and stream. This stream is first processed through a typical
normal-pentane that would tie up space in the effluent amine scrubber and a mercaptan extraction unit. We do not
treating and fractionation sections of the alkylation unit. find that any contaminants are a problem in this con-
figuration and do not see a need for a guard bed.
RAJGURU:
We are designing a C5 alkylation unit, as a part of the ROY:
reformulated gasoline project. There are no particular The following contaminants are normally found today in
drawbacks to C5 alkylation per se. Product quality and acid alkylation units: mercaptans, hydrogen sulfide, dienes,
consumption are similar to C3 alkylation. Using a TAME ethylene, and oxygenates like methanol, MTBE, or DME.
raffinate as a feed causes no special problems. Generally, Upstream Merox type units normally reduce mercaptan
amylene alkylate has higher T50 and T90 temperatures, levels to very low values and eliminate hydrogen sulfide in
compared to butylene alkylate, but the distillation point the olefin feed. Ethylene is a problem only when propylene
increases for alkylate from TAME raffinate are less than for is alkylated. The amount of ethylene is a function of the
the straight amylenes. FCCU gas plant fractionation operation.

116
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

The level of diolefins is a function of the FCCU oper- ACREON makes a guard bed material of specially pre-
ations. Today, butadiene levels in the range of 3,000 to pared lead oxide that is effective in removing arsine by
5,000 ppm in the olefin feed are normal. If amylenes are forming a stable lead-arsenic complex.
alkylated, C5 diolefins in the range of 10,000 to 15,000 ppm In a customer’s C3 selective hydrogenation unit, fol-
are seen. Of course, dienes can be eliminated with an lowing installation of an upstream guard bed of ACREON’s
upstream selective hydrogenation unit. MEP191, the selective hydrogenation feed arsine content
When the upstream MTBE unit is equipped with a water was reduced from 150 ppb to below detectable limits, and
wash cleanup system, total oxygenates are usually in the the selective hydrogenation catalyst cycle length increased
range of 400 ppm. Contaminants of all descriptions can also from 20% to 100% of prediction.
be introduced into the unit from the makeup isobutane, This guard bed material is equally effective in removing
particularly if it is imported into the refinery. Thus, it is carbonyl sulfide from 5 ppm to less than 0.1 ppm.
important to analyze the makeup isobutane on a routine
basis. RICHARD GIALELLA (JFP Enterprises Inc.):
Based on information provided by IFP, guard beds to Based on IFP’s commercial applications, FCC LPG
protect selective hydrogenation reactor catalysts are rec- streams can contain 1 to 2 ppm carbonyl sulfide, and de-
ommended when the selective hydrogenation feed contains pending on the type of crude, a few ppb to a few ppm arsine.
arsenic compounds, such as arsines. Normally, this is only Total sulfur can also vary, depending upon the crude proc-
necessary when treating feeds containing significant levels essed, and on upstream processing, from a few ppm to
of C3s. Arsine levels are also a function of the refinery’s several hundred ppm.
crude oil source. Carbonyl sulfide and arsine reduce the operating life of
the selective hydrogenation catalyst. The use of an IFP
LAABS: guard material can eliminate the problems primarily asso-
Our existing ether unit selective hydrogenation reactor ciated with arsine. In addition, the adsorbent will trap
has a guard bed for arsine. We are not sure how well it is some carbonyl sulfide, without impacting arsenic ad-
working, due to the difficulty of the analytical test. This sorption capacity, thus providing extra activity and life for
unit does have a water wash on the feed to remove caustic the catalyst.
and other basic compounds. Careful operation of this wash In C5 streams, sulfur is the principal contaminant. Sul-
is very important, as carryover of water will damage the fur may be decreased by caustic washing, but this results in
catalyst. The feed typically has a 1,3 - butadiene level of additional capital expense and waste streams. IFP has a
2,000 to 4,500 weight ppm (ppmw), and a disulfide content commercially proven sulfur-tolerant catalyst that may
of 15 to 30 ppmw. withstand several hundred ppm of sulfur, and thus avoid
The unit that is being built will also have a feed with less the need for sulfur removal steps.
than 30 ppmw sulfur, which is suitable for selective hydro-
genation. It will not have a guard bed, although a sand
filter will be installed to remove caustic and free water. The QUESTION 3:
filter will reduce the caustic carryover from the Merox unit Many refiners consider selective hydrogenation to
to less than 1 ppmw, which is necessary as sodium is a remove C 4 and C 5 diolefins from their alkylation
permanent poison on the catalyst. feedstocks in order to reduce acid consumption in
their alkylation unit. Will the removal of these C4 or
THOMAS F. KELLETT (CDTECH): C5 diolefins affect alkylate octanes or the alkylate
Guard beds, as has been mentioned, can be used to distillation curve?
remove some of the contaminants, such as mercury and
arsine. One selective hydrogenation process has a distinct JUNO:
advantage, with respect to mercaptans. This process is According to information provided by STRATCO, pilot
called CDHydro’” and it combines selective hydrogenation plant studies show the only effect of diolefins on alkylate
with fractionation. It does not require removal of mer- quality is the D86 end point. Removal of diolefins reduces
captans from the feed, because the mercaptans are con- the end point, but does not affect the other distillation
verted to heavy sulfides which go out with the bottoms points. The amount of end point reduction can be sig-
product. The overhead product can contain less than 1 ppm nificant when starting with high diolefin levels (above
sulfur and less than 50 ppm diolefins. This approach can 5,000 ppm). Removal of diolefins does not affect the alkyl-
reduce the capital cost of the feed preparation system by ate octane.
over 50%.
PAULES:
PAUL W. VANCE (ACREON Catalysts): Processing a C4/C5 cut through a selective hydrogenation
In protecting selective hydrogenation catalyst with unit will tend to decrease acid consumption, decrease al-
guard beds, current catalysts are now relatively insensitive kylate end point, and increase alkylate octane. The alkyl-
to low levels of common sulfur contaminants, such as ate octane is not only improved by the removal of dienes but
hydrogen sulfide and mercaptans. However, feed con- also by the isomerization of 1-butene to 2-butene.
taminants such as carbonyl sulfide and arsine are tem-
porary poisons. Arsine is formed by the decomposition in T. WILLIAMS:
the FCC riser of organo-arsenic compounds in the high We have experience with selective hydrogenation to re-
boiling fractions of the FCC feed. move butadiene on our C4 feed to our HF alkylation unit.

117
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

We have a butadiene concentration of 0.25 to 0.4 vol% in the the market, and it looks very attractive. It has a side benefit
feed and no butadiene in the product. We achieve a 60 to of reducing sulfur concentrations in the amylenes. A TAME
70% conversion of 1-butene to 2-butene. We believe this unit is also a useful feed pretreater, consequently lowering
gives us a 1.5 to 2 octane number increase on our alkylate the acid consumption.
RON. We see very little effect on the distillation, due to the There are no special operating considerations when proc-
small volume of butadiene. essing C5 feeds except that, depending on the amount of
isopentane in the feed and the product specifications, it
RICHARD GIALELLA (IFP Enterprises Inc.): may be necessary to provide a deisopentanizer in the frac-
Installation of a selective hydrogenation unit for pre- tionation section to have a lower RVP alkylate.
treatment of sulfuric acid alkylation feedstock is generally
economic for acid savings at diolefin levels above 0.5%. The LAABS:
removal of C4 and C5 diolefins from feed to both HF and In addition to the comments in Question 1, we have seen
sulfuric acid alkylation units reduces the end point of the more spikes in acid consumption when processing com-
alkylate. Significant octane improvements for HF units can bined butylene/amylene feed. While the exact cause has
be obtained through pretreatment of the feedstock. not been identified, we expect the spikes are caused by
occasional water and oxygenate carryover into the C5 raf-
THOMAS F. KELLETT (CDTECH): iinate, especially MTBE, DME, and methanol.
Dioleflns produce a heavier alkylate. Removal of di- One of our sites performed a study on the impact of
olefins from the feed should decrease the end point of the amylene alkylation. The results did not show a significant
alkylate. CDHydro™ can reduce diolefins to less than 50 impact on the alkylate D86, but it did show an increased
ppm, while conventional fixed-bed technologies typically acid consumption that was traced to the C5 diolefins in the
achieve about 200 ppm. This technology can be built into raffinate.
the MTBE debutanizer at little cost, while eliminating the
need for mercaptans removal or a separate selective hydro- EDWARD G. ELLIS (Exxon Research & Engineering
genation unit. Due to these cost savings, CDHydro™ can be Company):
economical to pretreat alkylation unit feeds with as little as We have considerable experience with amylene alkyla-
0.1% dioleflns. tion, using our stirred, auto-refrigerated, alkylation reac-
tor design. This experience includes both commercial and
HIGGINS: pilot plant operations. It includes 100% amylene feeds and
In a typical feed, do you have a feel for how much the end amylene/butylene feed mixes. In comparison with butylene
point or the 90% point is reduced by removing dioleflns? alkylation, we find that amylene alkylation yields lower
octane, reduced from about 96 RON for butylene alkylate to
ROBERT W. MORSE (ACREON Catalysts): about 90 RON with amylene. We find increased heavy
The removal of diolefins in the feed significantly reduces alkylate production, with the amount in the Cl0+ range
the product 90% point and the end point. For example, a C4 going from about 3% with butylene alkylate, to about 10%
sulfuric plant reduced its alkyate 90% point from 360 °F to with amylene alkylation. We find increased acid con-
330 °F and reduced its end point from 470 ºF to 410 °F. The sumption, which can be double that for butylene alkylation
feed butadiene level was 0.6%. We have seen proportional if it is not controlled. This increase in acid consumption
reductions with lower levels of butadiene in the feed. results from pentadiene and cyclopentene in the feed.
I would like to make another comment concerning the
benefits of pretreating the feed. For plants that are re-
frigeration limited, saturating the butadienes can sig- QUESTION 5:
nificantly increase the capacity of the alkylation unit. When processing amylenes, what material in the
Removing the diolefins substantially reduces the total heat feed is responsible for the higher acid consumption,
load. In our units, the product butadiene is typically 10 ppm i.e., diene content, sulfur compounds, other?
to 50 ppm.
SLOAN:
M. W. Kellogg Company and Exxon have been per-
QUESTION 4: forming pilot plant work with real refinery feeds for over a
What is your commercial experience with C 5 year, with one of the focus points being amylene alkylation.
alkylation? Provide specific comments on alkylate We have data in which a normal refinery C5 cut, that had
D86, feed pretreatment and/or feed preparation re- been selectively hydrotreated to serve as TAME feed, was
quirements, acid consumption and any unusual op- processed in the alkylation pilot plant. This feed contained
erating conditions/problems. residual, diolefin, and cyclic olefin impurities, but virtually
no sulfur. For comparison, a synthetic feed was prepared
RAJGURU: that contained the same list of major components as the
We do not have any commercial experience. However, we refinery feed, but no impurities. It was found that the
are currently designing C5 alkylation units. As has been difference in acid consumption between the two feeds was
said, the alkylate D86 distillation curve is typically higher highly significant and could be calculated from the im-
for C5 alkylate. Selective hydrogenation is desirable for the purity concentration.
feed to reduce acid consumption. Selective hydrogenation, The data will be presented in more complete form in a
concurrent with desulfurization processing is available in joint paper on amylene alkylation that will be given by

118
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

Kellogg and Exxon next March at the NPRA Annual Meet- LAABS:
ing in San Antonio. The alkylation of amylenes will inherently result in
higher acid consumption. However, contaminants in the
RAJGURU: feed will also have a significant effect. Oxygenates such as
The types of contaminants in C5 feeds that consume acid MTBE, DME, and methanol will all contribute to a higher
are basically the same as in C4 alkylation, but there are consumption. The water in the saturated raffinate will also
more species involved. There are oxygenates, dienes, sul- cause problems. Cyclopentene in the feed is expected to
fur, and the miscellaneous like acetonitrile, formaldehyde, consume three or four times more acid than other
acetaldehyde, and propionitrile. The oxygenates are meth- amylenes.
anol, ethanol, TBA, DME, diethyl ether, MTBE, ETBE, Finally, cutting deep into FCC gasoline can result in
TAME. The dienes are 1, 3 butadiene; 3M-1,2 butadiene; more volatile C6s being sent to the alkyIation unit, which
2M-1, 3 butadiene (isoprene); 1,3 pentadiene; 1,4 penta- would also increase acid consumption.
diene; and cyclopentadiene. The sulfur is mercaptans. The
miscellaneous are l-pentyne, acetonitrile, formaldehyde, STEPHEN S. MATHUR (Chevron Research and Tech-
acetaldehyde, and propionitrile. Cyclopentene also con- nology Co.):
sumes some acid (2.5 lb acid/lb cyclopentene). This list One of our refineries is experiencing off-test copper strip
comes from the STRATCO seminar. Obviously, the oxy- corrosion in the light alkylate (aviation gasoline). Has
genates present are a function of upstream processing, but anyone else experienced this?
the dienes are typically present in any FCC depentanizer
overhead. It is also important to limit C6 in the C5 feed LAABS:
because of dienes and benzene in the C6s. We also make aviation gasoline off the light alkylate cut.
For the two years I was at Port Arthur, I do not think we
JUNO: ever failed the copper strip corrosion test.
In most operations today, dienes are the highest con-
tributor to acid consumption from feed contaminants when ROBERT E:DAVIS (R. E. Davis Chemical Corporation):
there is not an upstream SHP or MTBE unit. If there is an I suggest that you go back to your reactor and take a look
MTBE unit, oxygenates can be the largest contributor. at the distribution of the feed nozzles. I do not know which
Sulfur is usually not a problem since most refiners Merox- type of reactor you have, but take a look and see if you still
treat their alkylation unit feed. Following are sulfuric acid have feed nozzles. Quite often this is the case with off-
dilution factors published by STRATCO for various alkyla- specification product. You may be circulating SO2 due to
tion feed contaminants. decomposition side reactions. The SO2 will track, usually,
SULFURIC ACID your refrigeration system, but on occasion you will form
Dilution Factors esters which break down in the rerun tower. I would guess
that you also are having problems with your deisobutanizer
Compound lb Acid/lb Contaminant* right now.
Water 10.6
Ethyl Mercaptan 17.0
Ethyl Disultide 12.8
H2S Approx. 31.8 Process
Methanol 28.3
Ethanol 21.0
QUESTION 6:
t-Butyl Alcohol 10.6
Diethanolamine 26.0 What is the latest experience with alkylation unit
Ethylene 30.6 additives to enhance reaction chemistry? Comment
Dimethyl ether 11.1 on what benefits/penalties you have experienced.
Diethyl ether 11.4
MTBE 17.3 LAABS:
ETBE 16.3 Star and affiliated sites that have tested these additives
1,3 Butadiene 13.4 have had mixed results. Two refineries have tested the Betz
3 Methyl 1,2 Butadiene 13.4 Alkat-XL product, and the results showed little or no ben-
2 Methyl 1,3 Butadiene (iosoprene) 12.3 efit from the additive. However, the test results were sus-
1,3 Pentadiene (piperylene) 13.4 pect in both additives for different reasons. One unit was
1,4 Pentadiene 8.8 running the contactor temperature at 60 °F, which is higher
1 Pentyne 18.9 than that recommended for the additive. On the other unit
there were concerns about the accuracy of the flow meters
*These factors are appropriate for an acid dilution range of involved, and about the process conditions that were used.
98.5-90.0 wt% sulfuric acid. To apply these values to other In the past, one of our refineries tested the ChemLink
acid dilution ranges, the following formula can be used: IPC 1800 product on an alkylation unit of older design, and
they found a significant reduction in acid use. They used
Acid Consumption = (Value from Table) (Discard wt%) the additive continuously for 3 years until they shut down
(10.6) (Fresh wt%-Discard wt%) the unit.

119
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSfNG

MURPHY:
From what we have seen, alkylation units are among the
most challenging units in the refinery on which to measure
the impact of small performance changes. This is due to the
natural variation in the parameters measured, which tends
to hide real operational improvements. Pushing a unit hard
or operating outside normal parameters might give condi-
tions which could be addressed by an alkylation additive.
We have seen and have been able to prove operational
improvements, both in yield and octane improvement and
acid reduction on several units treated with alkylation unit
additives.
R. C. BARLOW (Betz Process Chemicals, Inc.):
I would support what has been said so far. We have seen
acid reduction in the range of 15 to 30%, as well as yield
improvement of 3 to 6% and octane improvement, but a lot
of that improvement can be masked by the variability in not decompose in the caustic wash ahead of the debu-
operation. I would highly recommend that if you are evalu- tanizer. Instead they decompose to sulfuric acid in the
ating an additive, you try to operate at near test run tower itself.
conditions in your evaluation of the additives. The longer As these descend to the reboiler, they oxidize the hydro-
the trial period, the better the ability to document results. carbons, and this causes reboiler fouling. The sulfur di-
Finally, feed quality control is very important. oxide, formed from the reducing sulfuric acid when the
One of the approaches that we are using right now is very, hydrocarbons oxidize, passes up the tower and dissolves in
very vigorous modeling to try to explain some of the differ- the water in the condenser. This makes the water corrosive
ences in operation. We believe it is critical to understanding and leads to condenser corrosion.
variable interaction and will help to avoid long term con- We design for relatively moderate space velocities below
tinuous efficiency questions. about 0.15 and do not experience these problems. We do not
use corrosion inhibitors, or the expensive, awkward acid
QUESTION 7: wash that some are discussing.
It is considered that the appropriate olefin space
velocity for an alkylation unit is 0.25 to 0.5. Does
anyone operate with greater olefin space velocity? QUESTION 8:
What problems are encountered? What is the RVP of amylene alkylate? Is this RVP
controlled more by feed composition or by product
SLOAN: fractionation?
Nothing magic happens at any given olefin space velocity.
As the olefin space velocity increases, the octane number of SLOAN:
the alkylate decreases, the acid consumption increases, Fractionation plays an important role in determining the
and more species are produced that could produce down- RVP of amylene alkylate. The composition of the C5 stream
stream corrosion. from an FCC is not the same for all units. However, in
An increased olefin feed rate also increases the heat of general the saturated C5 s make up about half of this
reaction, which must be removed. This taxes the re- stream, and most of the saturated C5s are isopentane. In
frigeration system and increases the reaction temperature. the alkylation reactor, some isopentane is reacted and some
The increased reaction temperature produces more of the is formed. However, the C5+ from the alkylate reactor is
above undesirable effects. still about half saturated C5s. Thus, a perfect C4/C5 split
will give an RVP of about 10 psi for the C5 portion of the
RAJGURU: alkylate.
We believe that 0.5 is a very high olefin space velocity for Controlling the RVP of amylene alkylate becomes a
sulfuric acid alkylation. Operating at such high space ve- matter of adequate deisopentanization, rather than debu-
locity results in poor alkylate quality. A side effect of the tanization as in conventional C4 alkylation. The feed com-
high space velocity will be high contactor temperatures, position also plays an important role. An FCC C5 stream
since there is a limited amount of bundle surface area in may be half saturated C5s, and a TAME raffinate may be
the contactor. The higher the space velocity, the higher the two-thirds saturated C5s. The alkylate produced from an
heat of reaction to be removed. FCC C5 side stream will have a lower RVP than the alkyl-
ate produced from a TAME raffinate. Amylene alkylation
DAVID S. MCCAFFREY (Exxon Research & Engineer- does lower the RVP of the total gasoline pool by eliminating
ing Company): the unsaturated C5s.
There are at least two problems at high space velocities
and both occur in the deisobutanizer. At high space ve- RAJGURU:
locities, the level of esters in the alkylation acid and the The RVP of amylene alkylate in our current designs is
hydrocarbon product build up as increased amounts are between 3.5 to 4.5 including the isopentane produced in the
generated in the reactor. These esters are neutral and do reaction. Obviously, product fractionation is not con-

120
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

trolling, although our deisopentanizer could be used to frame? Is the reactor a fix bed or some type of
further reduce the RVP if desired. The feed composition has continuous regeneration system?
some effect, and the actual reaction conditions are impor-
tant because they can affect the amount of isopentane JUNO:
produced. Neste Oy, Catalytic Inc. of Mountainview, California,
For one of our recent projects, with a mixed C4/C5 feed, and Conoco Inc. started up a 7 bbl/day pilot plant at Neste’s
product fractionation does control the RVP, but this is Finland Technology Center to demonstrate solid catalyst
because of the large amounts of NC4 and IC5 in the feed. gasoline alkylation technology. Their proprietary process is
being developed to replace hydrofluoric and sulfuric acid
ROY: alkylation units used to produce high octane gasoline.
Based on their pilot plant work, STRATCO has deter- Licensing is to be available in 1994.
mined the RVP of the true alkylate produced in the reaction
zone from a mixed amylene feed at normal operating condi- SLOAN:
tions is approximately 3.5 psi. This true alkylate includes Haldor Topsoe A/S and The M. W. Kellogg Company have
approximately 7 to 9% isopentane produced by the alkyla- been collaborating to develop an alkylation process which is
tion reactions. They have also found that changes in the a substitute for both the current HF and sulfuric acid
composition of the C5 feedstock and operating conditions processes. The super acid catalyst is supported on a solid
affect the amount of isopentane generated, and thus would medium in a unique manner. The reactor system can be
affect the true alkylate RVP accordingly. operated continuously without interruption for catalyst
Probably a larger consideration in amylene alkylate regeneration.
RVP is how the isopentane introduced into the unit with Acid inventory is very low and acid management is
the C5 olefin feed is handled. This isopentane will pass greatly enhanced since the catalyst can be recovered and
through the reaction zone into the unit’s fractionating recycled on the refinery site. The catalyst also has a low
system. Thus, the RVP of the rundown alkylate product, vapor pressure and there is no tendency to form an aerosol.
the whole alkylate, will depend upon the design of the The technology has been successfully tested in a pilot plant
fractionation system and where the isopentane is yielded of ½ bbl/day for over 2 years, including a continuous run of
from the unit. 6 months. The technology is being commercialized and
discussions with potential operating partners are in
progress.
QUESTION 9:
Some refiners operate with acid level below 30 BRIAN JOHNSON (UOP):
inches to avoid flat acid recycle. Is there a minimum As the industry is aware, UOP has been working for a
acid level below which we should not operate? We number of years to develop a solid catalyst alkylation proc-
understand that low acid levels can lead to a hydro- ess. At this point, UOP has developed a truly solid phase,
carbon continuous emulsion in the reactor. Will this regenerable catalyst and is moving into long term stability
cause acid carryover from the acid settler? tests in our pilot plant facilities near Chicago. An initial
engineering package is being developed that will be com-
LAABS: petitive with the capital and operating costs of current
We have one unit whose acid level routinely runs in the technologies.
18 to 24 inch range without problems. We know that if the UOP will present the technology to potential licensees in
level is too low, the emulsion will become too tight, and it 1994 and expects to have a complete commercial design
will not have time to separate. This will result in acid package by 1995. UOP will continue to support the current
carryover from the settler. The actual minimum level would HF technology with ongoing developments such as the
depend on the design of the unit. Texaco-UOP HF additive technology, as well as minimum
acid inventory designs, redundant acid storage designs,
DAVID R. WAGNER (STRATCO, Inc.): and rapid acid dump systems.
The acid/hydrocarbon ratio in the contactor is the best
indicator of the emulsion quality in the contactor. This ratio LAWRENCE A. SMITH (Chemical Research & Licensing
is monitored via the contactor ratio glass. The valves on the Company):
emulsion inlet and outlet should be wide open for maxi- Chemical Research and Licensing has announced a solid
mum acid circulation. The settler acid level should be acid alkylation process based on a “salt” of antimony penta-
adjusted to maintain the acid/hydrocarbon ratio in the fluoride. A 10 bbl/day test unit will be in operation at
contactor in the 50 to 60 vol% range. Chevron Port Arthur before the end of 1993. A one-year
With current designs, a settler level of 18 to 24 inches test program is planned on a variety of commercial
would normally provide a good acid/hydrocarbon ratio. feedstocks.

General B. Alkylation-Hydrofluoric Acid

QUESTION 10: Feed
What is the status of solid catalyst alkylation re-
search? What is the possibility of the development of QUESTION 11:
a commercial process and what is the expected time When operating a hydrofluoric alkylation unit

121
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

downstream of an oxygenate unit, what processing C3 and C4 olefins, we alkylate 1 to 3 vol% C5 olefins when
changes have been found helpful in reducing the light the FCC naphtha RVP is dropped seasonally. Increased
acid soluble oil production due to contaminants gener- sulfur, diolefins, and C6 + contamination together with a
ated in the oxygenate process? somewhat lower iso to olefin ratio results in excessive acid
soluble oil production and acid consumption. We observed
T. WILLIAMS: an octane drop of at least half of a point with no discernable
At one of our facilities, instead of worrying about the change in yield or isobutane consumption.
oxygenates going to the alkylation unit, we decided to remove
them with molecular sieve technology. This has worked out PAULES:
very well. We have, however, had two episodes where we The contaminants of biggest concern are the penta-
essentially flooded the molecular sieve and ended up sending dienes. These compounds dramatically increase acid con-
MTBE to our alkylation unit. This resulted in acid purity sumption and AS0 make. The sulfur compounds in the C5
dropping by 9%. The acid soluble oils (ASO) that formed could cut are also more difficult for a mercaptan extraction unit
only be removed and regenerated externally. The amount of to handle. Unless your unit is specifically designed to do
acid that had to be taken with this ASO, due to its light this, sulfur may be a problem in the alkylation unit. The
nature, was fairly significant due to the similar boiling yields for each amylene isomer are different, but generally
ranges. speaking, true amylene alkylate yield is 1.55 to 1.65 times
olefin charge with a RON of 90 to 92, and a MON of 88 to 90.
PAULES: The isobutane consumption is 1.10 to 1.35 times olefin
I agree with Mr. Williams. We have found the best ap- charge. Synthetic pentane production is 0.20 to 0.50 times
proach is to eliminate the problem at the source, as there is olefin charge.
little you can do once the oxygenates are in the alkylation One mechanical concern is the ability of the unit to
unit. When AS0 is formed as a result of oxygenate con- handle isopentane. Not only is there isopentane in the C5
tamination, you have little recourse other than to drop the charge, but synthetic isopentane production can be be-
temperatures in the rerun tower. This allows the light AS0 tween 20 to 30% of olefin charge. The fractionation section
generated from the oxygenates to exit the unit, but causes of a unit must be capable of removing all the isopentane
extremely high HF losses. while controlling alkylate RVP and preventing a buildup of
isopentane in the isobutane recycle.
BONELLI:
At our Robinson, Illinois refinery, we have operated an HF O’BRIEN:
alkylation unit downstream of a MTBEf/TXME coproduction Feed contaminants for C5 alkylation become a problem
unit since December of 1991. This unit has a molecular sieve when refiners try to recover more and more of the C5
oxygenate recovery unit as part of its flow scheme. One olefins. As you cut deeper into the C5s, you tend to drag
additional comment regarding the effects of oxygenate more of the C6s, dienes, and additional sulfur into the
breakthrough is that we feel that the oxygenate carry- alkylation feed. These materials will increase AS0 produc-
through might be destablizing the acid settling in the mixer tion substantially. The amylene feed will likely include
settler to some degree. However, we cannot quantify the effect some isopentane. The isopentane in the feed, along with
at this time. synthetic isopentane produced in the reaction, can load up
the recycle isobutane loop, and create problems with iso-
O’BRIEN: butane purity. As the isopentane builds up, it begins re-
The only method we have used is a fractionation step to acting somewhat as the isobutane does, but makes a C9
remove DME and methanol from the alkylation unit. Our compound with an octane of only 80.
goal is to limit oxygenates to 35 ppmw.

THOMAS E KELLE’IT (CDTECH): Process

DME, water, and methanol can be removed, as has been
mentioned, by fractionation or by molecular sieve operations.
MTBE, if present in significant quantities, is typically re- QUESTION 13:
moved by molecular sieves from HF alkylation feed because of What is the experience with acid recontactors in a
the required low levels. However, a well designed MTBE hydrofluoric acid alkylation unit? What are the quan-
debutanizer will produce ranate with MTBE levels below 1 titative benefits?
ppm. Diolefins are removed by selective hydrogenation. We
have now integrated this step with the MTBE column using O’BRIEN:
CDHydro™. In Phillip’s experience recontactors have worked very
well in reducing fluorides in most of the product streams.
Our experience indicates a reduction in fluorides of 90% on
QUESTION 12: product streams, except for the alkylate. This results in less
For C5 alkylation, please comment on feed con- frequent purchase of HF scavenger alumina.
taminants, yield structures, mechanical or operating
problems related to this type of operation. PAULES:
I agree. There are claims of up to a 90% reduction in
ROY: fluorides and in some cases, companies have been able to
At our Pasadena plant, along with our normal stream of meet product specifications without additional treating.

122
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

However, many refineries shy away from recontactors be- Wickbold Apparatus. The same method is employed for the
cause of safety concerns as they increase the amount of HF determination of the fluorides in propane. We have estab-
handling equipment in the unit. Also, it is reported that lished that levels down to approximately 10 ppm can be
they are ineffective on high purity propane streams. We reasonably determined.
have operated a recontactor on our main fractionator over-
head stream since the unit was built in 1979 and have no BONELLI:
real history of operating without it. We still treat the In one of our operations, which is a UOP HF alkylation
propane product for trace HF removal. design, our typical fluoride concentration in the normal
butane stream is 150 to 170 ppm by weight. The method
that we use for determining fluorides is the same modified
QUESTION 14: Wickbold method. For refinery operations, we consider 10
Has anyone experienced iron fluoride pre- ppm to be the lower confidence level on which we take
cipitation on coolers downstream of butane and action. We generally do not report values below 1 ppm; we
propane defluorinators? What has been done to alle- just report less than 1.
viate this problem? Since iron fluoride is a corrosion
product, what is the mechanism for this corrosion, T. WILLIAMS:
and where does it occur? Are these corrosion prod- Fluorides in our normal butane draw off the tower typi-
ucts a result of two phase flow (acid-H2O/butane or cally run in the 500 to 600 ppm range. After the alumina
acid-H2O propane) that occurs at low hydrocarbon and KOH treating, this is down to 1 ppm fluoride or less. We
temperatures? use UOP 619 Oxy-hydrogen combustion method and can
detect down to 1 ppm.
O’BRIEN:
We believe that it is possible to have iron fluoride pre- RALPH HARRIS (CITGO Refining & Chemicals, Inc.):
cipitation in coolers. The mechanism is iron fluoride scale Has anybody seen a dramatic increase in organic fluoride
destruction caused by colder than normal product tempera- content of the butanes and alkylate with addition of MTBE
tures. The cold temperatures cause the formation of a or TAME in front of the alkylation unit?
corrosive acid/water phase that can destroy the protective
iron fluoride scale. To alleviate this problem, product tem- BONELLI:
peratures should be kept around 100 °F. We have not. When we put our MTBE unit in service, we
Another likely deposit on the coolers is alumina fines did not know what to expect. Organic fluoride was one of
after recharging a defluorinator. To prevent this, the re- the parameters that we .monitored to see if our fluoride
charged defluorinator should be placed in the lead position balance was much disturbed. It has not been.
for a period of 24 hours to allow any alumina fines to be
deposited on the top of the secondary bed. The defluorinator BRIAN JOHNSON (UOP):
can then be switched to the tail position. The concentration of combined fluorides in the butane
stream coming off of the alkylation unit is not the primary
T. WILLIAMS: consideration. What is also important is how much normal
We run our defluorinators in series as mentioned in the is actually passing through the unit via the olefin and
previous comment. We put the regenerated or reloaded saturate feeds to the unit. If the unit has feedstocks that are
defluorinator back on-line as the second in the series and very low in normal butane, the normal that comes off the
we have experienced high corrosion rates in the low flow unit has a high concentration of combined fluorides since
areas of this second defluorinator. the fluoride production rate is independent of the normal
butane flow. So you need to look at the total moles of
combined fluoride produced rather than just the con-
QUESTION 15: centration in the normal butane stream in order to evaluate
What are typical fluoride concentrations in normal the fluoride production rates for a given unit.
butane streams? What level can be accurately
measured?
QUESTION 16:
O’BRIEN: Velocity in the acid settler is directly related to the
Typical fluoride levels for butane are 200 to 400 ppm. The separation zone (emulsion zone) width in the settler.
level of fluoride in the butane can vary for several reasons. At some velocity, acid carryover will become exces-
Acid strength, reaction conditions, and origin of the butane sive causing high corrosion rates. This velocity cor-
stream are some of the factors affecting fluoride levels. responds to some distance between the hydrocarbon
Organic fluorides can accurately be measured to 1 to 2 ppm. draw and the hydrocarbon/separation zone inter-
face: What is an. acceptable distance for reliable
OSBORN: operation of the unit?
During the second half of 1992, in preparation for a
potential shipment on a pipeline, we did testing of our O’BRIEN:
normal butane stream. Results matched the ranges re- Although the distance between the hydrocarbon draw
ported by Ms. O’Brien, 200 to 400 ppm fluoride. The test and the hydrocarbon separation zone interface may have
used to determine these results is an adaptation of the some effect, the overriding factor is the velocity. After

123
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

exceeding a certain velocity through the acid settler, carry- service. We also use some Durco plug valves. Again, we
over will occur regardless of the height. currently do not fireproof, preferring to leave the flanges
exposed and painted with HF leak detecting paint.
ROY:
The purpose of the HF acid settler is to provide disen-
gaging time between the HF acid and the hydrocarbon General
phases and to reduce organic fluorides in the products.
There are proprietary internals in the settler that facilitate
these two functions. The hydrocarbon velocity in the settler QUESTION 18:
is about 1.5 fpm. What is the most environmentally acceptable dis-
position for acid soluble oils? Where do most refiners
currently dispose of this stream?
Mechanical
PAULES:
We presently neutralize AS0 with caustic and include it
QUESTION 17: in the refinery residual oil pool. The decanted spent sodium
Please discuss the types of emergency isolation fluoride is then reacted with calcium chloride to produce a
valves being used. Are valves/flanges fireproofed in solid calcium fluoride precipitate and a sodium chloride
this service? brine. The solid calcium fluoride can then be disposed of in
a nonhazardous landfill and the sodium chloride brine is
BONELLI: totally harmless to the wastewater plant and the
At our Robinson, Illinois refinery we have recently com- environment.
pleted a substantial addition to our HF alkylation remote
emergency system. The valves that we chose to use are T. WILLIAMS:
Durco 1/4 turn plug valves with teflon liners. They have We reprocess our ASO, after neutralization, in our FCC
teflon seals, no packing. The flanges in this service are not unit feed.
fireproofed.
BONELLI:
O’BRIEN: Following neutralization to below 0.1 wt% HF, the acid
The majority of emergency isolation valves we have seen soluble oils are treated as hazardous waste at our facility
in refineries are ¼ turn, teflon lined valves with pneumatic and transported to an approved incineration facility. At
operators. Some refiners prefer using electric-operated gate Robinson, where we have delayed cokers, we are investi-
valves- with a metal-to-metal seat to reduce the fear of gating the suitability of including this neutralized AS0
teflon melting in a fire situation. into our Mobil Oil Sludge Coking injection.
Phillip’s licensing does not endorse, and the Phillip’s
plants do not practice, fireproofing of valves or flanges, O’BRIEN:
since by covering the bolted areas of a valve, you will limit After caustic treating to remove entrained HF, the AS0
the ability to monitor for leaks in these areas. Actuators can be blended into fuel oil, used as FCC feedstock, or
and other associated equipment are fireproofed. burned in an alkylation unit furnace using a special burn-
er. In the United States, putting AS0 into fuel oil or the
OSBORN: FCC feed tank is the most common disposition. In Europe,
We have installed tufline valves with HYTORK ac- we find people prefer to burn the AS0 in their reboiler
tuators on the outlets of the acid storage drums. The control furnaces.
panel is operated from the field control room. They are not
fireproofed. OSBORN:
In addition to the methods previously mentioned, we are
PAULES: currently injecting it into our coker charge. It has been
We have recently installed 18 air actuated, ¼ turn plug done in the past with no known detrimental effects. It is
valves for automatic isolation of HF systems. With these we important to assure the ASOs are adequately neutralized,
can isolate the acid settler, acid storage drum, acid feed yet do not contain excessive amounts of entrained caustic.
pumps, IC4 recycle, and main fractionator overhead. The
valves are flanged but not fireproofed. The valves are HARRY GILL (Ultramar, Inc.):
pneumatically operated and fail to the last position. We blend our acid soluble oil, after neutralization, with
slurry oil. Slurry oil eventually ends up in the fuel oil,
ROY: although we do not produce any fuel oil from this site. It has
Isolation valves being currently used by our plant are occurred to me that we could consider blending it with
plug valves with pneumatic actuators. We do not currently alkylate produced from the unit. The logic here is that
fireproof valves and flanges in HF acid service because we during internal regeneration, all those sulfur oils go to
believe the potential for leaks or corrosion to go undetected alkylate anyway.
is far greater if not completely visible.
JOE P. CARSON (Mobil Oil Corporation):
T. WILLIAMS: We are currently considering blending AS0 with our
We have used Xomox valves with Matrix actuators in this clarified slurry oil product stream. Some of our initial

124
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

laboratory results have shown that our AS0 Rash point is STEVE HUNKUS (MAPCO PETROLEUM Inc.):
below 100 °F. I was wondering if anyone else had experi- Along the line of this question, has anyone had a successful
ence with low boiling point ASO? Also, what is the typical commercial installation of an alternative or replacement
flash point for ASO? for the time tank to significantly reduce the inventories of
acid and isobutane in that system?
BONELLI:
Because we do not blend it, it is not a parameter that we BRIAN JOHNSON (UOP):
measure. I cannot think of a reason that in our regular We have done revamps of units for low acid inventory and
operation it would really be a concern since we neutralize it as part of that package, the time tanks have been
and dispose of it in an incinerator. eliminated.

QUESTION 19:
C. Reforming
Please comment on your experience in selling
spent fluorinated alumina and/or potassium fluoride
back to suppliers/reclaimers? Feed
O’BRIEN:
Phillips refineries are looking into the option of selling QUESTION 21:
the spent fluorinated alumina, with no actual transactions What is the impact of coker or visbreaker naphtha
to date. Several Phillips licensees are sending their spent on reformer cycle length? Is there anything that can
alumina for reclaiming. We do not produce potassium fluo- be done to improve run length?
ride waste.
PAULES:
The main problems with coker and visbreaker naphtha
QUESTION 20: are their high olefin and nitrogen content. The olefin con-
Are there any refiners actively pursuing alkylation tent should not impose any problems to the reformer itself,
additives to improve the safety aspects of their as all olefins will saturate in the naphtha hydrotreater.
hydrofluoric alkylation unit? However, if a large proportion of these naphthas are fed to a
naphtha hydrotreater, the unit must be equipped to handle
OSBORN: the high reactor exotherms of olefin saturation. After
I am aware of commercial testing being conducted by hydrotreating, the saturated olefins should not affect re-
UOP in concert with a refinery and have discussed the former cycle length. The nitrogen content will necessitate
results of that study. However, to date the information has hydrotreating at higher hydrogen partial pressure than the
not been made public and therefore will need to be pre- typical naphtha hydrotreater with a nickel molybdenum
sented by the parties involved. catalyst.
Some refiners include these naphthas as part of a dis-
O’BRIEN: tillate or FCC feed hydrotreater charge to allow saturation
Since our development of the first HF alkylation units in and denitrification before routing to a typical naphtha
1942, Phillips has engaged in a continuous research pro- hydrotreater in front of a reformer. The ultimate effect on
gram to enhance and optimize the process. Recent efforts reformer cycle length will be determined by the remaining
have focused on process and mechanical improvements to feed nitrogen and olefin content charged to the reformer.
enhance the already low risk and to minimize the pollution
impact of the HF technology. Such efforts have been very PEDERSEN:
successful in sustaining minimal risk. With our goal of We have found that up to 20% visbreaker naphtha in the
continuous improvement,’ these efforts have naturally in- feed does not influence reformer run length noticeably,
cluded catalyst modifications and additives to suppress the provided that you have good hydrotreatment upstream,
volatility of release material. Efforts are continuing in which means bromine number less than 1, sulfur and
development of this process modification. nitrogen less than 0.5, and 1 ppm, respectively.

DALE WILLIAMS (Texaco, Inc.): BONELLI:

Texaco’s El Dorado plant has awarded a contract for the We have only one facility in which we have a delayed
engineering, fabrication, and installation of facilities for coker and that facility has two reformers and naphtha
continuous use of the Texaco-LJCP HF additive technology hydrotreaters. We have made coker naphtha processing
beginning in July, 1994. Release tests of HF and HF ad- trials in each of those naphtha hydrotreaters and experi-
ditive mixtures have confirmed that substantial reductions enced fouling of the feed preheat exchanger train and
in potential aerosol formation can be achieved. reactor bed.
A trial on the number 1 alkylation unit at El Dorado in We feel that when you are hydrotreatingcoker naphthas,
1992, confirmed that use of the additive did not require any your reactor space velocities should be near 1, and the
change in throughput or reaction conditions and did not reactor pressure needs to be 800 to 1,000 psig for 90%
degrade the alkylate quality. This technology will be li- hydrogen purity streams. In addition, a nickel molyb-
censed by UOP and will also be available on a modular denum catalyst is required for that service. For an excellent
basis from the UOP Process Plants and Systems group. view of the impact ofreforming non straight-run naphthas,

125
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

I recommend a paper on the subject that was published by shortened by an increase in catalyst coking rate due to the
UOP. loss of hydrogen recycle rate and higher reactor tempera-
tures, and an operation closer to heater maximum
OSBORN: temperatures.
In addition to the comments by Mr. Bonelli, I think the
best precautions to take are to minimize the end point by ROY:
taking extra care during the switching process to minimize We run coker naphtha, about 8 to 10%, in our CCR feed
fractionator upsets and keep the end point down. It is and we use a silica-based antifoam at the coker. We have
certainly more difficult to reform coker naphtha versus typically seen runs of 1 to 1.5 years on our hydrotreater. We
straight-run naphtha at the same end point. If it is not monitor for silica breakthrough on the hydrotreated
possible to achieve the 800 or so psi hydrotreating, I would catalyst.
strongly recommend that you consider blending coker
naphtha straight to the gasoline pool and offset the low BONELLI:
octane by higher reformer severity. At our location, it is not economical to reform the coker
naphthas given the flow scheme we have, and therefore,
J. B. RODDEY (Roddey Engineering Services, Inc.): they go to the gasoline blend pool.
The corrosion aspect of higher sulfur and higher nitrogen
feed to the naphtha hydrotreater usually results in migra- OSBORN:
tion of corrosion products into the reformer. As the iron In addition to controlling end point, if you feed the naph-
content goes up in any part of the reformer, like the first tha to the reformer, end point is important if you are
catalyst bed, the ability to regenerate the catalyst gets sending it directly to the gasoline pool. At about a 360 °F
more difficult. Fouling of the feed effluent exchangers and end point, you have about a 70 octane, and if you go up to a
in the charge heater is also more common in units that run 400 °F end point, the octane will drop to about a 60 for the
straight-run and coker or visbreaker naphtha simul- same coker charge quality. That is a significant deterioration.
taneously in the same hydrotreater.

CEES SMIT (Akzo Catalysts): QUESTION 22:

Regarding the processing of FCC naphtha, is it common
What are the pros and cons of using nickel catalyst
in the United States to reform this stream or to reform the
in a sulfur sorber rather than copper catalyst?
heart cut? What is the influence on the reformer per-
formance, specifically hydrogen yield?
PAUL W. VANCE (ACREON Catalysts):
BONELLI: For liquid phase sulfur sorbers, ACREON offers both the
We do not do that at our locations. The FCC naphtha is copper and nickel products. Copper provides the least ex-
generally blended as produced for sweetening. pensive initial fill and provides pickup of hydrogen sulfide
and light mercaptans. However, the high surface .area
LEWIS A. FREDERICKSON (Chevron Research and nickel sorber provides approximately 100% greater sulfur
Technology Co.): pickup by weight, greater affinity for more of the types of
I did not hear anybody mention silicon fouling of the sulfur species common in reformer feed such as mercaptans
hydrotreater catalyst. Is anybody using a guard bed? and thiophenes, and represents a more economical cost per
pound of sulfur removed for reformer feed pretreatment.
BONELLI: Both types of products are in use at several refineries.
We do not. I think that most people who operate cokers This nickel product is stabilized with carbon dioxide and
today are very careful with their antifoam injections. I overcomes historical difficulties with other types of non-
know that most of the people, at other companies with stabilized nickel products and can be easily prepared for
whom we have discussed coker operations, are con- service. Nickel products, in general, require a thorough
scientiously minimizing that. It has not been, in most understanding of associated handling requirements and
cases, a problem for them. Although I think we are all still are commonly used by refiners.
looking for a good nonsilicon antifoam for cokers. Disposal of the spent nickel sorber material may also be
much easier than copper, since instead of disposing as a
THOMAS W. KELLY (ARCO): waste, the high nickel sorber often has value to a metal
We do not use silica guard beds but do analyze for silica reclaimer.
when replacing our desulfurization catalyst to help esti-
mate new catalyst life. We expect a 2 year life when run- ANGELO FURFARO (UOP):
ning about 15 to 20 percent coker naphtha in the reformer I agree in general with the previous speaker. I would add
feed. that UOP finds for the copper-based material at least, that
Coker naphtha affects our reformer performance because it is most effective only for the removal of H2S. The nickel
hydrotreated coker naphtha properties are much worse does a better job of removing H2S as well as mercaptans and
than our straight-run reformer feed properties. When disulfides. So if you only have an H2S problem, I would lean
processing a blend of coker and straight-run feeds, we see towards the cheaper copper-based material, but the dis-
lower recycle gas purity and higher reactor temperature posal costs in your area will impact your final decision. On
requirements to make a given octane reformate than if an equivalent disposal cost basis, the copper-based mate-
processing only straight-run. Cycle length is therefore rial is cheaper on a cost per pound of sulfur removed basis.

126
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

MARK S. DORAN (ICI Katalco): PAULES:

Nickel catalysts are more appropriate where the naph- For a typical semiregenerative reformer operating at an
tha feed contains a high degree of complex sorber com- average reactor pressure of 300 psig and 95 RON severity, a
pounds, such as thiophenes. drop of 100 psi average reactor pressure will yield an
Copper catalysts are best suited to applications where increase of approximately 1.0 vol% reformate and 100
the dominant sulfur species are H2S and/or mercaptans. scf/bbl of hydrogen on a rich 65 N + 2A feed, and an
Copper catalysts are much easier to use. They do not need increase of approximately 2.2 vol% reformate and 150
an activation step, do not need to be loaded under inert gas, scf/bbl of hydrogen on a lean 40 N + 2A feed.
and they are not pyrophoric when discharged.
Nickel is classified as a potential carcinogen in some PEDERSEN:
countries, and so requires more stringent handling pro- From our pilot plant we have found that reducing the
cedures. More importantly, if correctly designed, copper- reactor pressure from 375 to 275 psig increases C5 + yield
based systems have a longer life and are a better value to by as much as 5.0 wt%. The corresponding hydrogen yield
the operator when considering cost of catalyst per pound of will increase by approximately 35%. Feed in our test were
sulfur removed. ICI Katalco’s PURASPEC™ range of ab- naphtha from a 40160 mix of Stattjord and Gullfaks crudes.
sorbents includes both copper and nickel sulphur guards Tests were carried out over a severity range from 96 to 102
and had been successfully used on catalytic reformers and RON clear, with a platinum/rhenium catalyst.
isomerization units. ICI Katalco will assist with disposal of
spent material as part of its Catalyst Care Programme. M. EDWARD MORRISON (IFP Enterprises Texas, Inc.):
With a semiregenerative reformer using bimetallic cata-
lyst and operating at 250 to 300 psig (mean reactors pres-
Process sure), the estimated yield improvement at constant octane
after reducing pressure by 100 psi is estimated to be 2.5
wt% for C5 liquid yields, and 250 scf/bbl for hydrogen
QUESTION 23: What gains in reformate and hydro-
yields. This obviously reduces the cycle length of the semi-
gen yields do refiners typically achieve on semi-
regenerative catalyst.
regenerative reformers using bimetallic catalyst
IFP’s DUALFORMING technology solves the problem
when reactor pressure is reduced by 100 psi?
of short cycle length and further increases gasoline and
hydrogen yields. In this particular process, the existing
JUNO: fixed bed reactors operate at low severity so the cycle length
In 1992, we revamped our semiregenerative reformer can be considerably increased. A new reactor and con-
lowering the separator operating pressure from 300 to 170 tinuous regenerator operates at a higher severity. This
psig. Based on pre-revamp and post-audit detailed test provides a low cost alternative to fully continuous oper-
runs, we achieved an increase in C5+ yields of 2.7 liquid ation. In this process the overall yield improvements are 3.5
vol%, and an increase in hydrogen production of 210 scf/bb1. wt% liquid yields, and a hydrogen yield increase of 350 to
Measured research and motor octanes were within 0.2 of 400 scf/bbl.
an octane number and the (R + M)/2 was within 0.1 of an
octane number between the two test runs.
Projected yields from UOP for the R62 catalyst for sep- CHARLES S. MCCOY (Consultant):
Ms. O’Brien, why did you increase pressure when every-
arator pressure reduction from 300 to 200 psig were a 2.1
one else is trying to reduce pressure?
liquid vol% C5+ yield increase and 164 scf/bbl hydrogen
production increase. Projections of catalyst cycle length
between regenerations were exceeded. O’BRIEN:
This data came from several years ago and it was from
O’BRIEN: our Puerto Rico plant. I do not have any other details.
Although somewhat the reverse of the question, in the
case of one of our motor fuel reformers, we increased the
reactor pressure by 140 psig. During the test period the
unit was processing feed with N + A of 60 to 66 vol%, and
operating with severities ranging from 91 to 99 C5+ oc-
tane. The following was observed: hydrogen yields de-
creased 0.19 wt% for the low severity case and 0.3 wt% for
the high severity case; the C5 + liquid yield decreased 1.0
vol% for the low severity case and 0.5 wt% for the high
severity case; and the catalyst stability was increased ap-
proximately 20%.

BONELLI:
At one of our refineries, when we lowered the pressure on
a semiregenerative unit by 80 psig, we realized a gain of 2
to 3 vol% in reformate yield, and an increase in hydrogen
production of 50 to 100 scf/bbl.

127
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

LEE E. TURPIN (Profimatics, Inc.): A semiregenerative type catalyst that has been through
The changes in reformate and hydrogen yields as pres- a number of cycles is less yield sensitive to variations in the
sure is changed vary significantly with feed composition. chloride level than a recent loaded catalyst.
The changes in C5 + and hydrogen yields decrease as the In a low pressure reformer using a platinum-tin type of
quality (N + A) of the feed increases. catalyst, we also find 1.0 wt% to be the optimum chloride
The following table was generated using Profimatics level. When comparing at constant RON level, the relative
kinetic reformer model. C5 + octane was held constant as yields of C1 and C2 versus C3 and C4 are approximately the
well as feed boiling range. same at 250 and 50 psi. However, we found that the relative
yields are slightly dependent on the chloride level. Increas-
ing the chloride level on the catalyst from 0.08 to 1.2
Naphthenes Aromatics Delta C5+ Delta H2 increases the relative yield of LPG versus fuel gas by 10 to
Feed vol% vol% vol%/PSI SCF/BBL/WPSI 15%.
1 30.0 8.0 2.25 167.
J. WILLIAMS:
2 35.0 10.0 1.79 152. Under most circumstances the optimum chloride content
3 40.0 12.0 1.69 148. of reforming catalyst, both semiregenerative and CCR, is
0.9 to 1.1 wt% on catalyst. Higher chloride contents lead to
enhanced hydrocracking reactions and lower chloride con-
To answer Mr. McCoy’s question; one of the reasons I tents to lower activity. Within the range of 0.9 to 1.1 wt%
have seen people raise pressure is because they are meeting chloride, the effect of pressure on methane, ethane, pro-
heater constraints and if they raise the pressure, they can pane and butane yields is much more pronounced than
have a bigger heat sink with the hydrogen and reduce the variations in the chloride content of the catalyst, although
heater constraints on downstream heaters. at higher chloride contents, the relative yields of propane
and butane are higher.
QUESTION 24:
We know that when pressure is reduced in a cata- QUESTION 25:
lytic reformer, the rate of cracking decreases. We There are several reforming catalysts on the mar-
also know directionally that increasing chloride level ket, with loaded densities varying from 32 lbs/cu ft to
on the catalyst will not only increase cracking, but 45 Ibs/cu ft. The manufacturers say the surface area
will also increase isomerization and dehydro- (area/mass, meter squared per gram) is roughly the
cyclization. Is there any advantage then to increasing same from one catalyst to another. Why are low
chloride level to increase paraffin isomerization and density catalysts loaded into reformer units when a
dehydrocyclization in low pressure reformer units? refiner can get roughly 50% more surface area by
Is the delta C5 + yield per 0.1 wt% chloride the same loading a high density catalyst?
at 50 psig reactor pressure as 250 psig? What effect
will this have on the relative yields of methane, J. WILLIAMS:
ethane, propane and butane? The lower density catalysts were developed in the past
when reformer pressures were high and severities low and
PAULES: consequently deactivation low. Today, high density cata-
At any pressure level there is an optimum chloride level. lysts are available which allow operation at relatively low
Any additional increase in chloride above this will do little pressures and high severity, while maintaining acceptable
to promote isomerization and will tend to promote excessive cycle lengths. In some cases, the yield structure, higher
hydrocracking. Operating above this optimum amount will reformate yields, has also improved for the high density
hurt you more at 250 psig than at 50 psig due to the catalysts. A disadvantage, of course, is the higher platinum
increased tendency for hydrocracking at higher pressures. inventory in the case of high density catalysts.

PEDERSEN: LAABS:
The catalyst chloride level is important both for the While there is a benefit to having high surface area in the
initial yields and for the catalyst stability. Increasing the reactor, the platinum/rhenium dispersion in the catalyst
acidity of the catalyst by adding more than 1.0 wt% chlo- determines the activity and the run length. If the low
ride does not increase the selectivity to aromatics or iso- density catalyst provides a satisfactory throughput, ac-
parafins. The opposite effect is actually observed as the tivity, and run length, then the price differential could
degree of hydrocracking increases. support its purchase. However, if the difference in price is
For a 250 psi unit using a platinum/rhenium type of balanced by the need for a higher weight hourly space
catalyst, the chloride level should be close to 1.0 wt%. If the velocity, or by the extra reformate yield that is possible
catalyst is 15% overchlorinated, the initial C5+ yield is with the higher surface area, then the high density catalyst
reduced by 1%. A corresponding 25% reduction of the cycle is the better option.
length is also anticipated. On the other hand, if the catalyst Trying to save money by loading a smaller amount of
is 20% underchlorinated, we do not see any loss in initial high density catalyst is not recommended. Shortloading of
C5 + yield, but the cycle will be reduced by approximately a reactor should not be done without modifications to the
35%. reactor internals.

128
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

WILLIAM J. NOVAK (ACREON Catalysts): In the past we have also used liquid oxygen for re-
We make reforming catalyst with a range of densities as generation. However, we see no incentive for oxygen in-
described in the question. As was discussed, higher density jection since using the high discharge pressure compressor.
catalysts will frequently debottleneck a unit that is limited We do the oxidation step at the same pressure as the
in severity. However, if a unit is operating at relatively low regeneration step. In this step the limit is the outlet pres-
severity, a refiner may consider using a lower density sure of the air compressor. The higher pressure decreases
catalyst simply because it reduces the catalyst expense. the time required for oxidation and results in regenerated
(The reactor available volume can be fully loaded with less catalyst that performs well.
total pounds of catalyst.) Actually, with the trend toward
reformulated gasoline and the increased use of oxygenates, OSBORN:
we see reformer severity coming down a little bit. Therefore I am not aware of any maximum limit on operating
it is very economical, in some cases, for people to actually pressure for a semiregenerative reformer regeneration,
move to a lower density catalyst. except for those set by the compressor and plant mechan-
ical limits. The function of pressure during carbon burn is
BONELLI: well known; that is, it increases the coke burning rate and
Refiners should consider what Mr. Laabs said about not increases the heat removal from the catalyst. Both of these
shortloading reactors without attention to modifying the are due to the increased regeneration gas flow rate associ-
internals of those reactors. There has probably been a great ated with the higher operating pressure.
deal more trouble resulting from shortloading than can Higher pressure also helps during the oxychlorination
ever be recovered by the difference in what one pays for step. One important aspect of the regeneration is the redis-
catalyst. persion of the active metals. Increasing the oxygen partial
pressure favors the formation of Cl2. Chlorine is the chemi-
cal species responsible for metals redispersion.
QUESTION 26:
Is there a recommended maximum operating pres- CHARLES S. MCCOY (Consultant):
sure during the carbon burn stage of a semiregener- Mr. Juno commented about corrosion control. As you go
ation reformer regeneration? Should the oxidation up in pressure all these good things happen that the pan-
step of a regeneration be carried out at the same elists have mentioned. It is generally profitable and at-
pressure as the carbon burn and if so, why? tractive. However, corrosion control becomes more difficult.
Various techniques are used for corrosion control; every one
PEDERSEN: of them is subject to changing chemistry of the reactions
We recommend maximizing pressure, both during car- involved as you go up in regeneration pressure.
bon bum and oxidation. The pressure will normally be
limited by the discharge pressure of the air compressor. By J. B. RODDEY (Roddey Engineering Services, Inc.):
maximizing pressure, you achieve a faster coke burn and As you go up in pressure, my experience has been that
good contact during oxidation due to the higher recycle gas the carryover into the compressor becomes more sig-
flow. nificant. In most regenerations, there is some small
amount of carryover of neutralizing solution into the com-
JUNO: pressor. This occurs whether you have a centrifugal or a
There is no maximum operating pressure at which the reciprocating compressor. We recommend that you wash a
carbon burn stage of a semiregeneration reformer re- centrifugal compressor after or during regeneration in
generation should be performed. The horsepower of the order to remove these salts to avoid damages that can occur
recycle compressor will limit the pressure that you can when salts fly off of a wheel turning at 8,000 rpm. There
achieve. On centrifugal recycle compressors, this is gen- have been many examples of that where excessive com-
erally around 180 psig. pressor damage has resulted because of this carryover. So I
Although I do not have any experience at this pressure, I caution all of you that there is more to be considered than
am told that there are some units on which caustic carry- the time factor and the other benefits that have been men-
over has been experienced at 400 psig on some of the tioned by the panel. That is, carryover; it is a real
regenerations. The oxidation step is not sensitive to pres- phenomena.
sure and is normally carried out at a pressure that is
convenient to the rest of the regeneration procedure. LAABS:
We have always included a step to wash the compressor
O’BRIEN: during regeneration of our semiregenerative platformers.
Over the years we have regenerated reforming catalyst at
pressures from 25 to over 300 psig. Obviously regenerating
at a higher pressure results in higher density of the re- QUESTION 27:
generation gases, thus leading to a shorter regeneration The presence of iron on reformer catalyst suggests
time. For the high pressure regeneration, we use a high corrosion somewhere. Where should we look? Is
pressure oil-free air compressor. We have checked the cata- sulfur the most likely corrosion agent?
lyst properties following the regenerations and have seen
no detrimental effects on the catalyst from the regeneration JUNO:
pressure. The corrosion products that are present in a reformer are

129
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

due to corrosion in upstream equipment. The most common effects also result in a loss of yields. As long as the sulfate is
causes of corrosion can be traced to the desulfurizer strip- present, the platinum crystals can not be properly
per where water and hydrogen sulfide occur in the presence dispersed.
of carbon steel vessels. An intermittent water wash in the The catalyst performance can be restored to fresh cata-
desulfurizer can be the culprit for washing scale and sludge lyst activity and stability through an effective sulfate re-
through the stripper and into the reformer. Another source moval procedure. This involves removal of the sulfate in a
of scale can be the water interface controller on the stripper hydrogen atmosphere with a large presence of chloride.
overhead receiver. If the interface is lost, water is refluxed Chloride competes with the sulfate for sites on the alumina,
down the stripper tower carrying scale and corrosion prod- thus driving the sulfate off the catalyst. The sulfate is
ucts into the reformer. removed as H2S.

PAULES: LAABS:
I agree with the previous comments. One other source of We agree. In addition, we recommend sulfate removal if
the corrosion might be a high moisture content in the there is more than 0.05 to 0.1 wt% sulfate on the catalyst.
recycle gas during regeneration.
RAJGURU:
J. WILLIAMS: I agree with the previous comments on the first part. The
You should look in all parts of the reformer that are in sulfate on the surface of a reforming catalyst prohibits, or
contact with H2S, water, or hydrogen chloride at elevated at least decreases, the rate of noble metal redispersion. Yes,
temperatures, i.e., heat exchangers, furnaces, etc. Sulfur, the catalyst can be restored to close to fresh activity when
probably H2S, is the most likely corroding agent, forming carrying out a sulfate sweep during the hydrogen treat-
iron sulfide. In that way, sulfur is hidden in the system and ment between the carbon burn and the oxychlorination
can migrate to the catalyst when the catalyst is re- step.
generated. Iron sulfide will decompose then into iron oxide Sulfate can be effectively stripped from the catalyst by
and SO2. The latter can react with the alumina from the circulating relatively high purity hydrogen at elevated
catalyst to form stable alumina sulfate. temperature. Chloride addition will improve the rate of
sulfate removal.
T. WILLIAMS:
In addition to the previously mentioned possible cor- J. WILLIAMS:
rosion agents, one other to add is dissolved oxygen from Sulfate formation will make the redispersion of the
tankage if you do not have a good gas blanketing system. platinum more difficult at the time of the oxychlorination
and can cause sintering at the time of reduction because of
CHARLES S. MCCOY (Consultant): the decomposition of the sulfates into water and H2S. The
On the subject of iron and reforming catalyst, iron is also result is a very low activity after regeneration.
a catalyst for the so-called demethylation reaction. I made a We were able to restore catalyst performance by double
dramatic experiment that illustrated this many years ago, oxychlorination with intermediate H2S stripping. A caustic
charging a 50-50 mix of ground-up platforming catalyst wash circulation is necessary in the H2S stripping stage to
and iron fines to a pilot plant. Naphtha was fed at about absorb the formed H2S from the recycle gas and prevent
950 °F; nothing but methane came out the back end. Why recontamination of the system upstream of the catalyst.
we did not blow up the pilot plant and destroy the labora-
tory, I do not know. T. WILLIAMS:
We saw evidence of sulfate formation during a re-
THOMAS W. KELLY (ARCO): generation that followed a sulfur excursion. We did exten-
Another source of iron in reactors is from the oxidation sive hot hydrogen stripping according to UOP procedures.
and reduction steps during the catalyst reactivation Typica1 start-of-run activity was achieved.
process.
THOMAS W. KELLY (ARCO):
I agree that full catalyst activity can be restored if sulfur
QUESTION 28: is completely removed before the chloro-oxidation step.
What is the effect of sulfate formation during the High sulfur on catalyst during this step will cause plati-
regeneration of catalytic reformer catalyst? Can the num crystallite growth to such an extent as to make good
catalyst be restored to fresh catalyst activity and metal redistribution not possible. A study in our process
stability conditions? What are the methods used to lab confirmed that nothing could restore catalyst to new
correct the sulfate problem? activity after it had seen a reactivation while contaminated
with 0.5 wt% sulfur. The catalyst had been through a 12
OSBORN: hour pre-regeneration hydrogen strip to remove sulfur
The existence of sulfate, S04, on catalyst during a re- which was obviously not very effective.
generation has two serious effects. First, the sulfate pro-
motes platinum mobility. In turn, this leads to platinum HIMMAT SINGH (Indian Institute of Petroleum):
agglomeration and loss of catalyst stability. Second, the The Indian Institute of Petroleum (IIP) and Indian Petro-
sulfate hinders the ability of the catalyst to pick up chlo- chemicals Corporation Limited (IPCL) have jointly de-
ride. This in turn leads to loss of catalyst activity. Both veloped a bimetallic reforming catalyst. The catalyst is in

130
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

commercial use in one of the reforming units in our country. plant fuel system. The fuel gas line and gas burner plug-
Due to some plant upsets in this unit, we did experience ging problems have been severe, limiting the firing rates to
sulfate formation which ranged between 0.32 and 1.65 the heaters in the reformer and the naphtha hydrotreater.
wt%. The catalyst in this case was regenerated in two steps We are installing chloride-removing alumina guard beds.
following the procedures outlined by the panel. The ac- Some chloride deposition was noted in the net gas machines
tivity of the regenerated catalyst was close to its original prior to modification of the wash tower internals. We also
value. experienced tray plugging in our platformer debutanizer
and corrosive attack on the platformer debutanizer over-
TERRY TUCKER (MAPCO PETROLEUM Inc.): head line.
We have had success with the sulfate stripping procedure
as recommended by UOP for R62 catalyst. As you are doing MURPHY:
a regeneration burn, the SO2 appears to accumulate in the The burning of offgases can lead to plugging and fouling
reactor bed until near the end of the burn. Therefore, you difficulties. When chlorides are involved, these further
need to test for SO2, via Draeger, just as oxygen break- complicate the problem by adding salts to the organic
through occurs at the end of the burn. deposits. These offgases can be treated with antifoulants
that will inhibit the deposition. Several antifoulants of this
type are available, some specifically for chloride laden
QUESTION 29: gases. The proper treatment will provide protection from
Reforming units produce from the continuous plugging and corrosion that results from a high chloride
catalytic reforming (CCR) section, a continuous environment. Compressor problems caused by offgas can be
stream of offgas rich in chlorides. Normally, this serious. Chemically treated gas can be safely compressed
stream is routed to the fuel gas system to be burned with minimal impacts on the compressor. These gases can
with other refinery fuel gas streams. Has anyone be segregated, but this will only serve to focus the problem
recorded any plugging/fouling problems associated in a different location.
with that type of operation? If yes, which solution has
been adopted (separate burning of offgases, chloride CORNELL SLADE (ACREON Catalysts):
absorption, etc.)? Has scrubbing or fixed bed absorp- ACREON Catalysts recently introduced to the North
tion been used to solve this problem? Have booster American market SAS-357, which is a promoted alumina,
compressor fouling problems been experienced? widely used in Europe and Asia. This product can pick up
between 10 to 15% of its weight in HCI. In the last few
ROY: months the product has been tested and sold to at least two
At our Pasadena CCR unit, heater control valve and major refiners in the U.S. to solve problems such as this.
burner plugging/fouling used to be a common occurrence
when we burned the hydrogen rich offgas without any
treatment. Antifouling additives were tried for a while, but QUESTION 30:
we finally solved the problem by installing a fixed, dry What is the experience of converting semi-
alumina adsorbent bed with very satisfactory results. regenerative catalytic reformers to low pressure?
We still have intermittent ammonium chloride plugging What is the experience with low pressure drop plate
of the booster compressor discharge line. The plan is to type exchangers? What modifications if any are re-
remove the chlorides in the makeup hydrogen through an quired on the recycle compressor?
alumina adsorbent bed.
SLOAN:
PEDERSEN: First I would like to thank PACKINOX for their as-
From start-up of our CCR unit at Mongstad in 1989 until sistance on this one. Lowering the operating pressure of a
last year when we installed a chloride trap, we experienced catalytic reformer at constant capacity has several conse-
heavy corrosion and plugging in our fuel gas systems. The quences on the existing equipment. With the vapor den-
trap is a fixed bed of activated alumina with a sodium sities decreasing, the stream velocities will increase
promotor. It has proven to be very efficient. We have also throughout the process loop, thereby increasing total loop
experienced fouling of the booster compressors and short pressure drop. Unfortunately, the process delta P delivered
time between failure of compressor suction valves, which by the recycle gas compressor will decrease, not increase,
again may be related to chlorides in the offgas. when lowering the operating pressure. However, the com-
pressor can usually be saved, mostly by reducing the pres-
BONELLI: sure drop of the process loop.
At one of our refineries we have experienced chloride The replacement of the existing feed-effluent shell and
fouling problems associated with chloride contamination of tube exchanger train with a low pressure plate type ex-
the net produced hydrogen. The hydrogen, which is changer is a very effective way to reduce the total loop
pumped away by two reciprocating net hydrogen com- pressure drop. In one instance a single shell, plate type
pressors, is first scrubbed with a caustic solution in a wash exchanger was installed to replace a train of 12 shell and
tower to remove the chlorides. However, we operate the unit tubes, with a net reduction in pressure drop of 52 psi.
at throughput rates which result in hydrogen production A second consequence of lower operating pressure is a
above that which the compressors can pump away. This has decrease in overall heat transfer coefficient in the feed
resulted in some unscrubbed hydrogen being burned in the effluent exchanger. Two solutions are available here: either

131
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

modify the heat transfer train, or push the firing of the hauled and the cylinder size was increased from 7.5 to 10
charge heater. With the current low refining margins and inch diameter. These modifications allowed for a hydrogen-
high environmental pressures of today, pushing the heater to-hydrocarbon ratio of 4.5, with a spare compressor for
is probably not a desirable option. backup. Three other compressors were modified and over-
Plate type exchangers have been selected as the feed hauled for booster compressor service. Two condenser
effluent exchanger on over 50 catalytic reformers around shells were changed to allow cooling water service on the
the world, about half of them as part of a revamp either to tube side of the exchanger for maintenance reasons. Sep-
lower the operating pressure, convert to CCR operation, arator liquid pumps were installed to boost the liquid to the
increase throughput capacity, lower operating cost, or a operating pressure of a common stabilizer.
combination of these. Seven of these revamps with a CFE The successful operation of this revamp was described in
welded plate exchanger are located in North America. Question 23. The increase in C5 + yield was primarily at
the reduction of propanes and butanes. This fact was not
T. WILLIAMS: unnoticed considering that during the summer of 1992, the
We converted a semiregeneration unit to low pressure RVP of finished gasoline was approximately 2 RVP lower
with a drop of 100 to 120 psi. Problems encountered were than the previous summer. The process engineering for this
due to high vibration in the vertical feed effluent exchanger revamp was performed by Pennzoil’s process engineering
which resulted in reducing unit charge. No modifications personnel.
were needed to the recycle compressor. We did not install
any plate type exchangers and therefore have no experi- DAVID W. ABLIN (UOP):
ence with them. UOP has been, or is currently, ,involved with the con-
version of 24 semiregenerative reformers to low pressure
RAJGURU: CCR platforming units. Our experience is that the revamp
We have converted two semiregeneration reformers to can be completed with less than 30 days of unit downtime,
medium pressure operation. In one of these we switched as the unit is down to make only the final process tie-ins.
reformer heaters due to a shift in the duty requirements. The recycle gas compressor is usually retained as the basis
The feed-effluent exchanger which was a conventional for these revamps.
multishell exchanger was replaced with a single shell, low Generally speaking, most of the fired heaters, the net gas
pressure drop, vertical shell and tube exchanger. The pres- section, and the fractionation section are all retained in
sure drop saved by this replacement and by the heater shift these revamps, although some equipment modifications
compensated for the reduction in differential pressure of may be required. We agree with the comments made by Mr.
the recycle compressor. Sloan regarding the welded plate Packinox exchangers.
An additional major impact of reducing the pressure is UOP would be happy to help refiners evaluate the poten-
more hydrogen production at a lower pressure. Additional tial benefits of revamping their existing semiregenerative
compression is required by the net gas compressor or by the units to lower pressure CCR units. We now have an engi-
hydrogen user. Low pressure drop, plate type exchangers neering group dedicated to revamp work that can provide a
have had satisfactory operation in this application, and variety of site specific revamp services ranging from feasi-
some of our clients in the United States have started accept- bility studies to complete basic engineering designs.
ing them.
FRANCOIS REVERDY (Packinox, S.A.):
BONELLI: Somebody on the panel talked about a vibration issue
In our case, which we referenced in Question 23 for after reducing pressure drop. One of the advantages of
lowering the pressure on our semiregenerative unit, we did switching to plate type exchangers is that you totally elim-
not require modifications to the recycle gas compressor. We inate any possibility of vibration in your feed effluent
also did not require the use of plate type heat exchangers as exchanger. This is for the very simple reason that the
replacements for our shell and tubes. unsupported span of any single element in the heat trans-
fer bundle is about 1 inch or less. So to get that element to
JUNO: vibrate would be very difficult.
As I stated in a previous question, we revamped a semi- Another advantage of switching to plate type exchangers
regenerative reformer from 300 psi separator pressure to is that you also eliminate the worry about two phase feed
170 psi. In our situation, we had an idle reformer next to the maldistribution in the exchanger. Feed maldistribution is
reformer we were operating. Both reformers had a 5,000 a worry because it will lead to non performance and long
bbl/day nameplate capacity at 300 psi. Our design was to term mechanical problems.
revamp both units to process 3,500 bbl/day at the lower There is no secret: a low pressure drop exchanger has
operating pressure. Detailed test runs were performed on little process delta P available to force a uniform dis-
both units prior to our revamp effort. Two feed versus tribution of the two phase feed across the section of the heat
effluent exchangers were replaced in one unit, and two transfer bundle - and some of those bundle sections are
were modified in the second unit. New separators were quite large today, with 5,000 or more tubes per bundle!
installed as carryover problems were occurring at even the Plate type exchangers have eliminated this worry by
300 psi operation. using a liquid naphtha injection system that relies on the
The recycle compressors were a series of six reciprocating head of the naphtha pump, rather than on the process delta
compressors of which four were in operating condition at P to force a uniform liquid distribution in the heat transfer
the time of the revamp. All the compressors were over- bundle.

132
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

1986, we instituted a chemical aid injected into the com-

pressor suction piping with an atomizing nozzle after a
knockout pot. Since that date we have not experienced
enough loss in compressor efficiency to require a shutdown
for cleaning.

MARK S. DORAN (ICI Katalco):

Fouling by ammonium chloride can be eliminated by
installation of a fixed bed of our chloride absorbent in the
compressor suction line. This results in longer cycle length
and reduced maintenance costs and times. One North
American operator using an ICI Katalco PURASPEC
guard has reduced compressor maintenance costs by over
$100,000 per year.

QUESTION 32:
QUESTION 31: Has anyone come up with a better way to promote
contacting of recycle gas and neutralizing solution
How long an interval are refiners able to run their
during regeneration? Is there something better than
continuous catalytic reformer recycle compressors
full-cone spray nozzles?
between washings to remove NH4CI deposits? What
levels of nitrogen do they have in the reformer feed?
PEDERSEN:
LAABS: The answers are no and no. We have been working with
We typically wash our CCR recycle compressors during these problems over a long time, but have found that a
scheduled maintenance shutdowns every two years. We properly designed nozzle injection system together with
have used a combination of sodium bicarbonate solution close followup during regeneration provides the best
and water for these washes. We control the nitrogen levels protection.
in our feed to less than 0.5 ppm.
J. B. RODDEY (Roddey Engineering Services, Inc.):
BONELLI: Improper mixing in the back end of the unit during
Our compressor typically achieves a 4 year run between regeneration has caused serious corrosion problems in
washings with solutions as Mr. Laabs indicated. Feed to our many units. I get many calls from people with those prob-
reformer is treated to nitrogen contents of about 0.5 ppm by lems. We have developed a technique of analyzing the
weight or less. mixing capability for an installation and we have made a
number of modifications to solve the corrosion problems for
PAULES: refiners.
Typically a continuous catalytic reformer operates on a 3
year cycle. If nitrogen content of the feed is kept less than ANGELO FURFARO (UOP):
0.5 ppm, washing of the recycle compressor between turn- UOP has designed regeneration circuits with a simple
arounds is not necessary. slotted distributor to inject the dilute circulating.caustic for
more than 20 years. The key variables to watch are the
PEDERSEN: caustic circulation rate, which should be a minimum 50% of
We have had our CCR recycle gas compressor in service the unit design charge rate, and the velocity of the injection
since start-up in 1989 without seeing a need to wash. The stream, which should be of 3 to 5 ft/sec exiting the slot in the
nitrogen in the feed is less than 0.5 ppm. distributor. The location of the distributor is key, with
proper location being in the center of the process line in the
RAYMOND A. STEPHENSON (Lyondell-CITGO Re- direction of the gas flow. Proper piping orientation of the
lining Company Ltd.): gas circuit is also critical to good gas contacting. It is also
In our vintage xylene reformer, the compressor has important to monitor both the pH and the total alkalinity
rarely maintained an acceptable efficiency level for 6 which is a more cumbersome test but we believe it is key to
months. However, with a minimum reduction in the feed, keeping track of the availability of buffer in the solution.
we can take one compressor off-line for chemical cleaning, We think the simple slotted distributor can do the job and
all within an 8 hour period. Our other two reformers do not has for a long time.
have the luxury of two compressors. However, these two
stage compressors typically maintain enough efficiency to QUESTION 33:
last until catalyst regeneration is necessary. They are al-
ways washed after the second reactivation period prior to Have there been any notable advances in reformer
catalyst technologies since the widespread use in the
start-up.
mid-80’s of the high rhenium catalysts for higher
With our Unifiner compressor, we rarely achieved a 4
stability operation? What has been the experience
month run before it had to be shutdown and chemically
with the L-zeolite based catalyst systems?
cleaned for removing ammonium chloride deposition. In

133
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

RAJGURU: redox procedures, multiple oxychlorination, etc., when in

One of the overseas refiners that we talked to, uses a fact the whole front end of the plant is full of iron sulfide
catalyst containing tin as a second metal rather than rhe- scale. This person needs to bite the bullet and acid clean the
nium. Getting higher stability operation is a key to using whole feed section of the plant.
tin. This refiner has no experience with the L-Zeolite based
catalyst systems. J. B. RODDEY (Roddey Engineering Services, Inc.):
Proper acid cleaning of the reformer heaters is not easy
J. WILLIAMS: and it is a very expensive procedure if done correctly. All
The advances are more an evolution than a revolution. reformer heaters have paralleled passes. Insuring proper
Reformate yield improvements in the range of 1% have flow through each of the passes is not possible unless
been observed on the newest high density catalysts. radical procedures are used on the heater. To properly
acidize the heater passes each pass must be acidized sep-
PAUL W. VANCE (ACREON Catalysts): arately, one pass at the time. With normal design, this
Advances in ACREON’s reformer catalyst include a requires cutting the tubes for acidizing and then welding
trimetallic catalyst for improved hydrogen and liquid them back together after the acidizing is complete.
yields and high surface area stable catalyst for CCR and When proper procedures have not been used, my experi-
cyclic reformers. ence with acidizing of these heaters has not been good. On
several units where the heaters have been acidized, sub-
sequent start-up resulted in scale migration out of the
QUESTION 34: heater into the scallops or into the catalyst bed. The loose
Before regeneration, a bimetallic catalyst of a scale causing the problem obviously was not removed by
semiregenerative reformer contained the expected the acidizing procedure used.
amount of (~ 0.03 wt%) sulfur. However, during Iron sulfide scales cause serious problems but acidizing
reduction H2S was measured in the recycle gas and may not be the most cost effective alternative.
reactor outlets. After addition to the first reactor of
only 5% of the total required amount of DMDS,
breakthrough was observed on all reactors. Appar- QUESTION 35:
ently, aluminum sulfates have been formed during Has anyone experienced increased corrosion
coke burn. Where is the sulfur coming from? Is it problems at the inlets of the chloriding agent in-
advisable to do another oxychlorination if H2S is jection points during oxychlorination? Was this due
measured in the reduction step? to changing the chloriding agent?

PAULES: T. WILLIAMS:
One of two things has happened. Either sulfur from the On one occasion, a reformer on regeneration experienced
feed has deposited on the catalyst or H2S has been gener- significant corrosion when the chloriding agent was in-
ated from iron sulfides formed as the result of corrosion. If jected directly into the transfer line between the.heater and
the sulfur is from feed sulfur upsets, then a sulfur strip reactor. The problem was solved by installing an injection
prior to the regeneration and oxychlorination step is rec- quill and thus avoiding notable contact between the agent
ommended. If the sulfur is from corrosion products, then and the pipe wall.
the feed piping, exchangers, and heaters should be chemi-
cally cleaned. PAULES:
I agree. Corrosion can occur if you inject the chloride at
J. WILLIAMS: the wrong location or in the wrong manner. Changing the
The most likely source of the sulfur is from the sulfide chloriding agent itself cannot cause corrosion.
scale in the system released during oxychlorination. A
second oxychlorination step would be considered if the H2S
is observed above 5 ppm in the recycle gas during reduction QUESTION 36:
to insure proper metals dispersion. Caustic wash of the At what catalyst surface area is a catalyst replace-
recycle gas after the first oxychlorination will minimize ment recommended? What is the basis for the
sulfur in the system. Refer back to Questions 27 and 28 for change? How, if at all, should the water-chloride
more information on this phenomenon. control policy for semiregenerative reformers
change as available surface area is reduced?
JUNO:
Premature evolution of H2S during the reduction stage of JUNO:
regeneration indicates the presence of sulfates on the cata- With each regeneration, there is a loss of surface area for
lyst. These sulfates on the catalyst form agglomeration on the catalyst. As surface area is reduced, the catalyst has
the platinum. An oxychlorination step should be performed less ability to retain chloride on its surface. After a number
to redistribute the platinum. of regenerations on a semiregeneration unit, the chloride
injection level should be increased to carry a higher chlo-
CHARLES S. MCCOY (Consultant): ride level in the recycle gas to distribute the catalyst chlo-
Mr. Paules mentioned chemical cleaning. I do not know ride to its best activity. If the original chloride level is
how many people have wasted their time and money trying maintained as the catalyst regenerations increase, the

134
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

catalyst will be underchlorided and susceptible to greater T. WILLIAMS:

yield losses following sulfur upsets on the unit. During a past reformer regeneration and catalyst dump
and screen, the resulting catalyst level was insufficient to
OSBORN: cover the open area on the center screen of a radial flow
Basically my comments are the same as Mr. Juno’s reactor. This created a low flow area where a substantial
accumulation of coke formed over the subsequent run cycle.
BONELLI:
Typically for our semiregenerative reformer the start-of- JUNO:
run yields are compared to end-of-run yields immediately In a radial flow reactor, a coke donut ring has been
prior to the regeneration and an economic analysis of the observed at the bottom of the reactor, generally attributed
value of the yield shift gives you the justification point for to the scallops filling up from the bottom with trash and
changeout. scale. This plugs off the flow at the bottom of the reactor,
In our CCR reformer we consider a few more factors, forming a dead area for formation of a coke donut. In
including the chloride retention ability, platinum redis- downflow reactors, an upside-down icicle sometimes has
persion efficiency, and how the yield pattern economics are been observed extending from the bottom of the reactor.
being affected. Generally, our CCR catalyst changeout sur- Improper dense loading of the catalyst or plugging of the
face area is somewhere between 100 to 120 square meters top of the elephant stool are possible sources of this coke
per gram. formation.

PAULES: OSBORN:
A reforming catalyst should be evaluated for replace- We have experienced a donut type coke formation in one
ment at a surface area of 65% of its original value. The of our semiregeneration reformer reactors. The coke
surface area, however, is only one area for determining the growth, in this instance, was attributed to channeling due
catalyst activity. If the platinum is properly dispersed, you to deterioration of the top of the center pipe, and pluggage
may be able to tolerate lower surface area levels. The basis of the scallops near the base of the reactor. The primary
for a catalyst changeout should be a loss in activity re- potential causes include maldistributed hydrocarbon flow
sulting in higher start-of-run temperatures and shorter from either plugged scallops or plugged reactor center pipe,
run lengths. The water-chloride balance should be main- catalyst fines as a result of seal fluidization, or the com-
tained as the surface area of the catalyst declines to main- bination of the two.
tain the same offgas yield. This will necessitate injecting
more chloride as the catalyst ages as chloride retention will CHARLES S. MCCOY (Consultant):
diminish with decreasing surface area. I have seen several of these including the examples the
panel cited. One of the most interesting ones happened
RAJGURU: early in the age of dense loading. A refiner dense loaded a
We have input from an overseas refiner. They say that radial flow reactor about 2/3 of the way up, and then soft
they replace catalyst with new catalyst having a surface loaded the balance. Apparently he built this kind of a donut
area of 210 when the surface area of the old catalyst is ring of coke right at the interface between the dense and
reduced down to 120. However, we make a decision to soft loaded material. I say apparently because they never
change the catalyst as we analyze the items listed below: saw the coke. They went with the burn and high oxygen/
oxychlorination. Later they discovered holes burned
capacity of retaining chlorides,
through the screen on the center pipe in a neat circle where
dispersion efficiency of platinum,
phase change of support alumina, this interface had been.
coke sensitivity, and
extent of metal contamination. THOMAS W. KELLY (ARCO):
We increase the chloride injection as a water-chloride We have seen coke ball formation at the top of radial beds
control as the surface area is reduced. when we have loaded catalyst so high above the holes in the
center pipe that we created a dead zone at the top of the
catalyst.
PAUL W. VANCE (ACREON Catalysts): We have also seen coke buildup because of iron scale
I agree that there are a lot of factors that should be plugging areas of the screen or catalyst bed. We felt sulfur
considered beyond surface area; but in our experience, was our main cause of corrosion and scale buildup. To
surface area on catalyst changeout has ranged from, typi- improve catalyst performance and reduce corrosion, we put
cally, 120 to 150. Higher levels of contamination or other sulfur guard beds in the recycle loop before the first heater
factors and economics may allow justifying a changeout in all of our reformers. In this way, the effects of sulfur are
outside of this range. ACREON’s correlations are used to minimized.
increase chloride addition rates with surface area loss.
RUSSELL WILLIAMS (MAPCO PETROLEUM Inc.):
We have also experienced a coke donut ring at the bottom
QUESTION 37: of the scallops on a radial flow reactor. Based on con-
Have any refiners experienced the formation of a versations with UOP we have come to the conclusion that it
coke donut ring in their semiregenerative reformer was caused by a flow restriction from the retaining ring on
reactors? If so, has the cause been established? the bottom scallop. UOP recommended that we move that

135
TOC/INDEX

LIGHT OIL CATALYTIC PROCESSING

up about one half of a foot. We have not done that yet so I am non-chlorided isomerization catalyst, we increased the di-
not sure if that will fix the problem or not. We did not find version of benzene and benzene precursors to the limit of
our coke ring until after the oxidation step either when we the naphtha prefractionator. Approximately 15 to 20% of
put catalyst from the third reactor into the fourth. We the benzene in the naphtha had been fractionated as feed to
would recommend testing via Draeger for CO or CO2 dur- the isomerization unit. At that time our isomerization
ing the regeneration step to determine if it is there because catalyst was deactivated and we did not see the benzene
it did not show up on the temperature profile. saturation we were expecting. We later repeated the test
run and were able to saturate the benzene from a feed level
of 0.7 to 0.0 wt%. We experienced a 15 ºF temperature rise
on the first bed after which it is quenched with hydrogen.
D. lsomerization Generally 5.0 vol% of benzene in the feed to an isomeriza-
tion unit is considered a maximum without implementing
more complex saturation and quenching schemes. The
QUESTION 38: problems that can occur with increased benzene are tem-
Have refiners experience with diverting straight- perature exotherms, which result in nonoptimum tempera-
run benzene and benzene precursors to isomeriza- ture profiles for isomerization, and cracking and coke
tion plants? How much benzene/benzene precursor laydown from higher boiling range benzene precursors and
material can be included in C5/C6 isomerization feed C7 material.
without problems? What problems are encountered
with higher levels of benzene/benzene precursors? OSBORN:
Our isomerization unit is a TIP unit converted from prior
ROY: reformer service. We have been able to run between 3.0 to
At our Tyler plant, we routinely charge our UOP I-7 5.0 vol% benzene without any problems. The biggest con-
isomerization unit with a hydrotreated, light straight-run, cerns are the large heats of reaction to saturate the benzene
containing 3 to 5% benzene. The unit is a converted re- and the increased hydrogen consumption. If more than this
former and has 3 reactors. We experience a significant amount is expected to be in the feed, then some type of
isotherm, 15 to 35 °F, after the first reactor and use recycled benzene saturate unit may be required.
hydrogen quench to limit excessive isotherms in both the As far as C6 rings, I have heard of up to 20% in the feed.
first and second reactors. Heater outlet temperature is The main problem with this is increased hydrogen con-
adjusted to compensate for the first reactor exotherm. sumption and suppressed isomerization.

T. WILLIAMS:
Our isomerization unit is designed to handle approxi- QUESTION 39:
mately 2.0 wt% benzene. Since benzene hydrogenation will When is the optimum time to change out the l-8
occur within the first 10% of the bed and produces ap- catalyst in Penex reactors? (When activity and delta T
proximately 90,000 BTUs/lb mole, any volumes above this across the beds are zero or before the zero point.)
would need a quench system to substantially control the
temperature. Due to the exothermic reaction and the ele- PAULES:
vated temperature, some of the cyclic material will undergo In the typical Penex unit utilizing UOP I-8 catalyst, the
ring opening reactions and subsequent isomerization. This majority of the total delta T of the system will be in the first
will ultimately result in lower octane on this portion of the reactor due to benzene saturation. Typically the catalyst
feed. will deactivate from feed contaminants in a plug flow
manner. Therefore, the first reactor will completely deacti-
PAULES: vate before the second reactor is affected at all. As the first
Phibro operates two UOP once-through C5 /C6 Penex bed deactivates, the delta T will begin to shift to the second
units. Both units were designed to process all the normal reactor which will negatively impact reaction equilibrium
pentane and the majority of the C6 material present in the and lower product octane.
refinery naphtha pool. This feed stream contains approxi- Therefore, the key to determining the optimum time to
mately 60% of the virgin benzene and 50% of the benzene change the first catalyst bed is to determine when the delta
precursors. The two units were designed to utilize only T has migrated to the second bed to the extent product
reactor feed to cool the reactor effluent. Therefore the units octane is seriously affected. I would look for a 10 to 15 °F
can tolerate only about 5 to 6 vol% benzene in the feed. increase in the second bed, on a sustained basis, as an
Above that, it can become impossible to remove enough indicator of the proper time to replace the first bed. Also,
heat to control reactor delta T. To date, this has been a the second bed should be placed in the lead position after
problem only when processing a few unusual crudes. As far catalyst changeout and the new catalyst bed be put in the
as cyclohexane and methyl cyclopentanc go, there is no lag position.
feed concentration limit as these compounds do not satu-
rate exothermically. However, as the concentration of cyclic BONELLI:
material in the feed increases, the attainable product oc- I would agree that you need to analyze the economics of
tane is lower due to less favorable reaction kinetics. the yield pattern shift as the first bed deactivates. I would
say that the optimum time is certainly not when the delta T
JUNO: across the beds is zero. We typically change the Penex
During a test run at one of our refineries utilizing a catalyst when the isopentane conversion in our lead reactor

136
TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

A. Alkylation - Sulfuric Acid might also be found in the olefin feed. Ethylene will rapidly
deplete acid purity at a rate of 30.6 lb per pound of
Feed contaminant. The sulfuric acid makeup cost at 0.1 vol% of
ethylene in butylene feed works out to about 14 cents per
QUESTION 1. barrel of alkylate.
What level of light ends (hydrogen through ethane) Occasionally, inerts can be found in isobutane feed.
have refiners been able to tolerate in the feed to the One refiner traced his inerts problem back to the use of
alkylation unit? Do refiners with feed selective hydro- fuel gas to transfer purchased isobutane from rail cars to
genation units have product strippers or flash separa- storage.
tors on/y? How does this affect operations? One final thought is that sulfuric acid units are much
more tolerant of light ends than HF units. Light ends in
DEATON: HF alkylation unit feed will cause acid loss and neutrali-
In one C4 olefin hydrogenation unit, a two-stage flash zation problems.
separator system was used to remove light ends prior to
the sulfuric alkylation unit. The high pressure separator SAYLES:
operated at 250 lb pressure, with the separator gas going I agree with the other comments.
to fuel. The low pressure separator operated at 80 lb
pressure, with the separator gas returning to the FCC wet DAVID S. McCAFFREY (Exxon Research & Engineering Company):
gas compressor for olefin and LPG recovery. Hydrogen, Light ends in the feed to a sulfuric acid alkylation plant
methane, and ethane content of the alkylation unit feed act as noncondensables, as has been said, and venting can
typically ran about 0.1, 0.15, and 0.2 vol%, respectively. result in large quantities of isobutane being lost unless the
Alkylation unit operation was not noticeably affected, vent stream is routed to a recovery tower. We usually
except that a vent of noncondensables was maintained recommend putting a stripper on the selective hydrogena-
from the depropanizer overhead receiver. tion unit product to avoid any chance of having to vent
in cases where the isobutane balance is tight. We have seen
McLEAN: some cases where only a flash drum is used, but then a vent
I will address the second part of the question. stream is usually taken in the refrigeration system.
Typically, flash separators only are provided on selective
hydrogenation units upstream of a sulfuric acid alkylation JEAN-LUC NOCCA (IFP Enterprises, Inc.):
unit. We know from experience that all hydrogen can be
Strippers are typically provided upstream of hydroflu- consumed in a selective hydrogenation unit. Therefore, if
oric acid units due to their lower tolerance for feed light you can employ pure hydrogen there will be no light ends
ends. going to the alkylation unit.
The selective hydrogenation units can be designed to
minimize the hydrogen in the product, but the levels of QUESTION 2.
the other light ends are very dependent on the quality of Is an aerosol farmed from a sulfuric acid release? If
the makeup hydrogen. so, is it a safety or environmental concern?

MENEGAZ: SAYLES:
I agree with the previous comments. Most units can This question has great importance in determining the
tolerate 0.1 vol% to 0.2 vol% of ethane in the olefin feed Hazardous Air Pollutant (HAP) status for the Maximum
without venting due to the partial solubility of the ethane Achievable Control Technology (MACT) standards for
in the propane product. On the other hand, hydrogen is listing sulfuric acid. Studies have been conducted to dem-
relatively insoluble in propane and will require venting at onstrate that an aerosol is not stable and does not migrate
around 0.05 vol% in the olefin feed. For this reason, when using different nozzle sizes and nozzle differential
provisions for removing excess hydrogen are usually pro- pressures (STRATCO Engineering and Technology
vided downstream of the feed selective hydrogenation unit. Newsletter, V3, 1993 and NPRA 1994 Annual Meeting
With poor de-ethanizing performance, some ethylene paper # AM-94-18).

140 Light Oil Catalytic Processing

TOC/INDEX

Our practical experience with small sulfuric acid leaks acceptable. Below 62% isobutane in the reactor effluent,
confirms the research data, sulfuric acid aerosol formation or below a four isobutane/olefin ratio, the octane loss and
occurs over a limited distance, 20 ft to 30 ft. acid consumption will increase significantly.
The safety concern when handling concentrated sulfu-
ric acid is related to personnel, environmental, and equip- SAYLES:
ment protection. A release of sulfuric acid is a safety and I agree with his comments.
environmental concern since low pH can cause acid burns
to operators and can produce hazardous waste. Care must BARLOW:
be taken in disposing of contaminated soils, used person- We have observed three units operating with isobutane
nel protective equipment and spent cleanup supplies to to olefin ratios less than 6. Two of those units operate with
ensure proper disposal. We have purchased special absorb- mixed propylene and butylene feeds. In going from less
ents which will contain the acid and maintain the greater than 7, to a ratio of less than 6, we saw a penalty in terms
than 5,000 BTU/lb energy value required for disposal in of volumetric conversion of olefins to alkylate of 5% to
a cement kiln. 10%, octane number dropped in the 0.5 to 1 .O unit range,
and acid consumption increased by 10% to 20%. The
MENEGAZ: third unit we witnessed processed an MTBE raffinate
We agree with the previous panelist’s response. stream with minimal or no propylene or isobutylene in
the feed, so the negative effect was less. We saw conversion
CUNEO: of olefins to alkylate drop by only about 5%. Acid con-
I agree with the previous comments on not forming an sumption increases were fairly nominal, and the octane
aerosol, or having a lingering aerosol. If light hydrocarbons dropped in the 0.3 to 0.5 range.
are released with the acid, a vapor cloud is seen accompa- The practical lower limit for the isobutane to olefin
nied by an acid spray. ratio will be different for every refinery and is set by
A key concern that was not mentioned before was SO2. equipment limitations, operating costs, gasoline pool con-
SO2 may be evolved during the release of sulfuric acid so siderations and the availability of makeup isobutane. Our
precautions need to be taken for the presence of SO2. experience is that most refiners try to stay above 7, but
higher iC4/olefin ratios have incremental benefits in alky-
ROBERT PUSCHINSKY (Allied Signal): late production, acid consumption and octane number.
In addition to SO2 involvement, in acid runaway situ-
ations, tanks, rail cars, and trucks have been known to LAUX:
erupt, also emitting sulfuric acid mist and sulfur trioxide.
The only comment I would add is that if you get much
Of course, those emissions also occur in the spent sulfuric
below the 60% isobutane concentration in the reactor
acid regeneration process and are of concern from a safety
effluent, the potential for an acid runaway increases dra-
point of view. A detailed discussion of sulfuric acid release
matically.
safety and environmental concerns is published in the
“Letters to the Editor” section of the October 1994 edition
MENEGAZ:
of “Today’s Refinery”.
I would like to add that many refiners are re-examining
QUESTION 3.
lower isobutane to olefin ratios to get more capacity from
Are any refiners operating at low isobutane/olefins
their existing units. This is being driven by a stronger
ratio (<6)? What is the impact on alkylate quality?
demand for alkylate volume and low value for octane as a
result of the new RFG standards.
What problems can occur at low ratios? Is there a
practical lower limit?
HEROS DERGREGORIAN (Giant Refining Company):
PORTER: What component in the isobutane recycle is contribut-
We operate two older alkylation units around a six ing towards the lower octane? Are there any particular
isobutane/olefin ratio, and we suffer about a two octane components or any impurity that would cause lower
number loss due to this low ratio plus inefficient mixing octane?
in the units. The isobutane concentration of the reactor
effluent is around 65%. This is a key variable because an PORTER:
increase in feed diluents even at a constant isobutane/ole- The main contaminant in the acid recycle that contrib-
fin ratio will lower the isobutane effluent concentration utes to a lower octane is water. As far as the hydrocarbon
and affect results. streams, any increase in diluents in the feeds or iC4 recycle
So in general, you can operate at a ratio of six or maybe will directionally lower product octane. The most com-
a little less, but you will pay a penalty that has to be mon diluents are propane and n-butane.

1994 NPRA Q&A Session on Refining and Petrochemical Technology 141

TOC/INDEX

LAUX: make alkylate, totally ignoring everything else going

The lower octane results because you do not have the around.
isobutane reacting with the olefin. The olefin will react The lower limit of this is strictly controlled by the
with itself. It will make additional polymer and it will reaction rates. A 4.5 to 1 isobutane/olefin ratio will be the
make a lower octane C8 or higher, olefin. bottom limit before severe polymerization. A ratio of 8 to
1 is desirable. At 10 to 1, you virtually control all but about
HEROS DERGREGORIAN (Giant Refining Company): 5% of the polymerization.
Do propane or normal butane impurities not contrib-
ute? MICHAEL BERLYANT (Bayway Refining Co./Tosco Corp.):
We have run an alkylation unit at 40% excess isobu-
LAUX:
tanelolefin on a regular basis for many years.
Propane and normal butane do not react.
Mechanical
MENEGAZ:
At low ratios, the amount of isobutane in the reactor
QUESTION 4.
effluent, rather than the isobutane/olefin ratio, becomes
What experience is there with plugging of the alkyla-
the critical parameter. So you need to take into account tion reactor distributor?
your propane and other diluents.

KENNETH ABRAHAMS (Star Enterprise): SAYLES:

For the gentleman on the floor, we run an alkylation We have not experienced plugging in the alkylation
unit where we run basically all the normal butane in the reactor. However, solid deposits on the tube side of our
plant through it as well. We have to pay attention to the STRATCO contactor have been observed. These deposits
isobutane/olefin ratio. The normal butane is like dead- are mainly iron sulfate, which was traced to the fresh acid.
wood. It will cost in terms of operating expense, but not The fresh acid contained iron sulfate as a contaminant
in quality or yield penalties. which settled into the contactor head. The sludge level
increased sufficiently to cause high pressure drop and a
ROBERT DAVIS (R. E. Davis Chemical Corporation): shutdown was required to clean the bundle. After discuss-
In answer to the question raised by this gentleman, the ing the quality of our fresh acid with our supplier, exchang-
isobutane has to go into solution and protronate in order ing samples and improving our own testing procedures
to enter into the reaction. The speed of this reaction is through a joint quality control program, the deposit for-
1/200th that of the olefin preparation. Consequently, if mation was eliminated.
you lower the concentration of the isobutane, the reaction
mechanics are simply not there to get the job done. The MENEGAZ:
olefin looks for another home. The isobutane will take its We asked several refiners about this and they said they
time and when it does go into solution, will continue to have not seen this problem either on the feed nozzle or the
acid hydrocarbon distributor in the acid settler.

QUESTION 5.
How does one control the corrosion in the deisobutan-
izer (D/B) tower overhead condensers, debutanizer
overhead condensers and debutanizer reboiler?

FELTROP:
Corrosion control in these areas can be accomplished
by two methods:
Application of an oil-soluble inhibitor program is ef-
fective as long as appropriate diluent and adequate distri-
bution is utilized. Butane solubility for the inhibitor is an
important consideration for effective application.
A second approach is to ensure complete neutralization
of the reactor effluent and minimization of neutralization
products. Optimization of such a system is achieved with
an efficient acid water wash system utilizing electrostatic

142 Light Oil Catalytic Processing

TOC/INDEX

precipitation to separate the phases prior to neutralization. below 7. When this occurs anhvdrous ammonia is injected
This application will reduce the soluble esters as well as into the overhead in addition to the TEA to raise the pH.
the entrained acidity, resulting in very little acidity remain-
ing to be neutralized and fewer esters to be decomposed. BARLOW:
As previously stated, corrosion in the fractionation
MENEGAZ: section of the alkylation unit is typically caused by acid
I agree with the previous comments. I would like to add carryover from the reaction section and/or acid and neu-
that the corrosion on the DIB tower generally indicates tral ester breakdown in the reboilers of the fractionation
that sulfate esters are getting past the effluent treaters and section. The best way to control corrosion is to focus on
decomposing in the tower in the presence of heat. When the reaction zone and reactor effIuent treating sections.
the treaters are designed and operated properly, overhead Maintaining suffcient iC4 /olefin ratios in the reactors,
corrosion should be minimal and two year intervals be- operating with reactor temperatures below 55°F, main-
tween reboiler cleaning should be expected. taining good mixing, and low emulsion and acid levels in
Bauxite is probably the most effective treating system, the settlers will help minimize the root causes of corrosion
since 98% of the esters are removed and it makes for a very in the fractionation section.
dry isobutane recycle. Operation of the effluent treating section at proper
The two other methods are (1) caustic treating the tower mixing, levels, temperatures, and concentrations is critical
feed with a 10° to 15° Baume wash followed by a water to effluent treating efficiencies. If operating conditions
wash, or (2) an acid wash followed by a hot alkaline water dictate operation outside of those recommended, corro-
wash. Acid washing is more effective than caustic washing sion problems in the fractionation section can be effec-
because it removes neutral esters as well as acid esters. About tively controlled through the use of chemical neutralizers.
90% of the esters can be removed this way. With either Neutralizers can be applied to the feed or reflux on the
wash method, a good static mixer is essential to achieve DIB, debutanizer, or depropanizer. The neutralizer used
maximum ester removal. Pressure taps should be provided must be nonvolatile if applied to the DIB so that it does
on either side of the mixer to check performance. not enter the reaction zone with the isobutane recycle.
I would also add that care should be taken to see that Nonvolatile neutralizers will also protect the reboilers in
amines are not recycled back to the contractor. the fractionation section.

PORTER: CUNEO:
I agree with the previous comments, that if reactor I will add that in units with higher levels of propylene
conditions are reasonable with proper treating of the in the olefin feed, we believe there is a higher degree of
reactor effluent, DIB fouling is minimized and overhead formation of neutral esters and employ an acid wash prior
corrosion rates are low. We also watch the pH of the water to the caustic wash to remove them so they do not decom-
in the DIB overhead to be sure that it is above 6. We do pose in the DIB. We have used ammonia in the DIB as a
have one unit in our Pascagoula Refinery that uses a temporary measure to get control of the pH of the over-
filming amine most of the time in the DIB overhead head water in the case of breakthrough.
because we are not able to optimize the above conditions.

SAYLES: General
Higher feed rates and on stream availability are pushing
existing units to the limits, resulting in reduced acid/hy- QUESTION 6.
drocarbon separation due to low settling times. Acid Does anyone use multivariant process control
carryover from the settlers can be caused by low acid schemes to control the refrigeration compressor sys-
strength, high settler level, mechanical failure, or pressure tem, or the DIB isobutane recycle loop in sulfuric
surges and these upsets are the source of DIB tower alkylation units?
overhead system corrosion. The corrosion mechanism is a
function of the acid entrainment and the water content of PORTER:
the DIB feed. Monitoring pH is the most common prac- The alkylation plant at one of our refineries has in-
tice with the injection of an amine for corrosion control. stalled a multivariant process control scheme. Three sepa-
We use Triethanolamine or TEA at a normal dosage rate rate systems were installed on the DIB, debutanizer, and
of 1 ppm to 10 ppm to the DIB feed. The overhead pH the refrigeration loop. We are extremely happy with the
target range is 7 to 7.5. performance, particularly on the DIB and the debutanizer.
Severe acid carryover from the contactors occurs occa- In the past, the DIB would take a very long time to
sionally which causes the overhead pH to rapidly drop stabilize but now the time has been cut drastically. Also,

1994 NPRA Q&A Sesinn on Refining and Petrochemical Technology 143

TOC/INDEX

the specifications on the streams remain steady, with Spent acid tank construction is a cone roof, internally
changes in feed composition. lined and externally steam traced with insulation. The
temperature is maintained between 20°F and 60°F. The
FRANK KLEINSCHROOT (SETPOINT, Inc.): freezing point of 98% fresh acid is about 14°F. In general,
In our systems, the compression system is maximized, as the acid strength is lowered the freezing point will be
subject to compressor speed limitations or minimum re- reduced, although the relationship is not linear. Hydrocar-
actor temperature. Olefin feed is maximized subject to a bons entrained from the holding drum are low vapor
maximum reactor temperature or isobutane availability. A pressure components, which are collected in the spent
multivariable controller is used on the deisobutanizer to tank and removed by an oil skimming system.
regulate recycle, reflux, sidecut and reboiler rates to main- Operationally, acid foaming is a cause of concern in
tain isobutane inventories, alkylate RVP, and reactor is- evaluating the potential for an “explosive” condition in the
obutane to olefin ratio while avoiding propane and tank. When foaming occurs, heat is generated and hydro-
n-butane accumulation in the tower. carbons adsorbed in the spent acid can vaporize. To pre-
vent foaming (as well as corrosion) the spent acid strength
QUESTION 7. is kept above 90% and the acid settler level is maintained
What measures are taken, or facilities installed, to at a sufficient level to ensure hydrocarbons will not be
reduce the risk of having an explosive mixture in spent entrained.
alkytation sulfuric acid tanks? The vapor space of
these tanks contains SO 2 as we// as hydrocarbon. DOUGLAS J. EVANS (Petro-Canada Products):
I wonder if any of the panelists or anybody in this
TREESE: audience, has noticed an upward pressure on sulfuric acid
At our alkylation unit in Los Angeles, spent acid is sent regeneration costs and available capacity with the increas-
to the spent acid surge drum. The spent acid surge drum ing pressure on alkylate capacity? Are refiners considering
floats on the flare through a check valve. From there, the installing packaged sulfuric acid regeneration plants
spent acid is pumped batchwise over to the spent sulfuric within the refineries?
acid storage tank located in our sulfuric acid plant, which
is nearby. HIGGINS:
At the storage tank there is a blanket of CO2 main- No comment.
tained on the tank. We are also fortunate there is a CO2
plant associated with the acid plant. The CO 2 then is
pressure regulated and the tank top is equipped with a
conservation vent, and flame arrestor. Safeguards on the
B. Alkylation - Hydrofluoric Acid
tank include a low pressure alarm, operator surveillance of
the tank frequently (several times during the shift), and a Feed
fuel gas backup to the CO2 blanket system in case we lose
the CO2 inventory. QUESTION 8.
When operating an alkylation unit downstream of an
PORTER: oxygenate unit, whatprocess changes have been found
We believe the key is to ensure that the spent acid is helpful in reducing the light acid soluble oil (ASO)
degassed before going to storage. The spent acid is pre- make due to contaminants generated in the oxygenate
heated and fed to a weathering drum. The vapors released process? Which contaminants cause increased ASO
are caustic treated and heavier hydrocarbons are skimmed make? What has been your experience using desic-
off the middle section. cants designed to remove feed oxygenates?

SAYLES: ARMBRESTER:
Our experience has been that the spent alkylation Dimethylether (DME) is generally the cause of the light
tankage vapor space has tested below the lower explosive ASO. Unlike methanol, which is quickly converted to
limit (LEL). At one location, a blanket gas is used to water in the alkylation reactor, the DME is very stable and
prevent oxygen buildup. can usually be removed from the acid stream only by
To prevent hydrocarbons from reaching the spent acid lowering acid regenerator temperatures and discarding
tank, the spent acid is pumped from the acid settler to a good acid with the ASO-DME mixture.
holding drum where the hydrocarbons are removed. The The most effective way to reduce the production of
drum is vented to the flare system and has a system to light ASO is to minimize oxygenate contamination of the
prevent a vacuum from occurring using fuel gas to equalize alkylation unit feed through proper control of the MTBE
the pressure. process and the alkylation unit feed dryer system.

144 Light Oil Catalytic Processing

TOC/INDEX

Tight control of the methanol to isobutylene ratio in PORTER:

the MTBE unit reactor feed is essential to minimize the At one unit we added a water wash column to remove
formation of oxygenated byproducts such as DME and to DME. It was successful in reducing the ASO production
avoid overloading the methanol recovery and oxygenate to a manageable level. We have not used any desiccants.
removal sections of the MTBE unit.
At our Catlettsburg, Kentucky refinery we have in- THOMAS DAVIS (CENEX):
stalled an on-line oxygenate analyzer in the mixed C4 I would like to ask Mr. Armbrester what type of analyzer
stream leaving the MTBE unit to detect any unusually he was using, specifically what model or brand, for oxy-
high concentrations of oxygenates in the feed stream to genates.
the alkylation unit. We are also considering the addition
ARMBRESTER:
of an oxygenate selective molecular sieve in our alkylation
The oxygenate analyzer we use in the MTBE unit is a
feed dryers. If the feed dryers are used as an oxygenate
Vista Model 3100, manufactured by ABB (Asea Brown
guard, the routing of the dryer regenerant stream becomes
Boveri) Process Analystics of Lewisburg, WV
critical.
In the original design of our alkylation unit the isobu-
tane used to regenerate the dryers was sent back to the Process
alkylation unit feed surge drum and the water was re-
moved via two coalescors and the feed drum water boot. QUESTION 9.
If oxygenates are being absorbed in the alkylation unit feed Our depropanizer feed condensers plug up on the
dryers, then the regeneration stream must be routed to a process side about once per year. What causes this
location where the oxygenates can be recovered. Other- and how can it be corrected? Has anyone eliminated
wise, they will cycle up in the alkylation feed stream until the depropanizer feed condensers and fed the de-
the concentration is high enough to break through the propanizer from the isobutane recycle stream?
dryer beds. We have resolved this problem by routing the
alkylation unit dryer regenerant stream back to the oxy- LAUX:
genate removal section of the MTBE unit. Plugging the depropanizer feed condenser on the proc-
ess side is probably due to iron fluoride. This can be
D’AURIA: confirmed by analyzing the foulant. If that is the problem,
DME is the most common oxygenate contaminant. the solution is to eliminate the source of the corrosion in
Our recommended procedure for removing DME is to the isostripper overhead system. The typical source of
adjust the acid regenerator conditions to reduce the degree corrosion here is excess water in the HE This can be caused
of stripping and allow the DME to exit the bottom of the by running the acid regenerator too hot to allow the
regenerator. The best way to prevent DME, or other incoming water to go out the bottom of the regenerator,
oxygenates, from entering the alkylation unit is to use an or by a wet feed to the unit. In the UOP unit, the acid
ORU unit on the MTBE raffinate stream. To date 11 moisture should be less than 1%. The specification is
ORU units have been installed in this service. typically higher in a Phillips unit, around 2% or a little bit
more.
Karl Fisher analysis of the incoming feeds should show
MENEGAZ:
less than 25 ppm water if distillation drying is being used,
I agree with the previous comments. I would like to add
or 5 ppm or even less if molecular sieve drying is being used.
that maintaining acid purity is the key to avoiding loss of
You can also analyze the depropanizer feed settler acid
octane and avoiding operational problems. Increased is-
boot acid for water to see if the regenerator is recycling too
obutane to olefin ratio may be effective in minimizing side much water. In our HF alkylate at Three Rivers we run the
reactions. acid regenerator overhead temperature at about 140°F to
Regarding contaminants, most refiners target 35 ppm ensure that water is rejected.
to 50 ppm oxygenates in the feed and remove the impurities Eliminating the depropanizer feed condensers and
by molecular sieves or fractionation. Dienes, sulfur, and feeding the depropanizer from the isobutane recycle, as the
nitrogen compounds also contribute to the ASO make. question implies, may simply move the corrosion problem
Alkylation units without MTBE units upstream have to the sidecut condensers.
also experienced contamination problems with oxygen-
ates. These have been traced back to alcohol in purchased MENEGAZ:
isobutane or trace acetone in the FCC alkylation feed. The I agree that the corrosion problem is iron fluoride. At
acetone is thought to come from reactions with entrained one refinery, the corrosion stopped after a butadiene satu-
oxygen from the FCC regenerator. ration unit was added. This eliminated the need for internal

1994 NPRA Q&A Session on Refining and Petrochemical Technology 145

TOC/INDEX

regeneration which eliminated HF contamination in the required. The shorter downtime easily justified the extra
overhead of the isostripper. cost.
Another refinery has the overhead condensers paired Because of this success, we tried using the same vendor
with valving so that one exchanger can be taken off-line at our 1994 Three Rivers turnaround. We used a different
for cleaning. Runs of between one and two years are chemical, though, a sodium hexanate, and it was not as
possible, but the exchangers are cleaned at least once a year successful in removing the iron fluoride from the exchang-
to avoid the risk of not being able to pull the bundles due ers or the isostripper itself. The degree of neutralization
to heavy fluoride salt buildup. The tower is operated to was very good because of procedural changes. The proce-
minimize the overhead temperature and minimize fluo- dure was very short, about 20 hours, and again we were
ride content above the isobutane draw. able to dispose of the solution in the API separator. The
cost was higher than the alternates, $85,000 versus
FELTROP: $50,000 for our single reactor and settler system, but the
I agree with the previous comments. I would emphasize procedure saved four days of downtime which, again,
that appropriate root cause analysis is the way to solve this justified the extra cost. We are now working with the
problem. We should not be focusing unnecessarily on the vendor to improve the chemical’s performance.
condenser section but the focus should be on prevention
of the polymerization that is occurring in this situation. D'AURIA:
We have had comments back from several of our licen-
D'AURIA: sees who have reported satisfactory results usingcitric acid.
The most common source of fouling is iron fluorides, The higher cost of citric acid is offset by reduced down-
and this can be controlled by controlling acid moisture time and lower disposal costs.
content. An additional source of plugging can be olefns
coming in with the saturate butane stream to the isostrip- DEATON:
per, which can then polymerize and foul in the overhead. We used citric acid cleaning for the first time during
With respect to the question of bypassing of the de- our last turnaround and we were very satisfied with the
propanizer feed condenser, this can be done with older results. The estimated time required for citric acid cleaning
isostripper designs which use a vapor dome partial con- was 60 hours less than for hydrochloric acid. However, the
denser on the overhead stream (i.e. no sidecut isobutane cost for citric acid was 50% higher than for hydrochloric.
stream). This bypassing will result in a small amount of From a technical standpoint, citric acid compares favor-
free acid in the isostripper recycle, however. ably to hydrochloric in several areas. Process corrosion is
For modern isostrippers, using a sidecut isobutane far less severe, as the citric acid solution is maintained at a
recycle, bypassing the depropanizer feed condenser will near neutral pH. Process cleaning is as good, if not better,
only move the problem down to the sidecut stream. as the risk of precipitating iron fluoride is eliminated.
Since the pH is near neutral, citric acid solutions can
be heated up to 200° F, which does a better job of removing
Mechanical heavier hydrocarbons and polymers. Disposal of the spent
acid solution is less hazardous because the solution pH is
QUESTION 10.
maintained near neutral. However, citric acid solutions
Has anyone used citric acid for chemically cleaning
contain ammonia, which can create disposal problems
the alkylation units? How did it compare to hydrochlo-
because of ammonia limits in waste water.
ric solutions?

LAUX:
I have seen a citric acid solution used for chemical General
cleaning on an HF alkylation unit five years ago. The
cleaning worked very well. The degree of neutralization QUESTION 11.
was very good. Exchangers bundles that had been stuck Additives have been announced to further reduce the
after previous hydrochloric cleanings came apart very potential hazards in the alkylation process. What has
easily. There was very little residue in the columns or small been the experience of refiners in using such additives
vessels. The residue that did remain in the acid settler was in their commercial units?
washed out with a fire hose. We were able to dispose of
the cleaning solution by slowly draining it into the plant D'AURIA:
API separator. The chemicals are more expensive, though, Texaco and UOP have jointly developed a new HF
but the cleaning duration is two to three days shorter than additive technology called “Alkad”. Allied Signal has also
the hydrochloric procedure because no neutralization is participated in this work as a technology partner. The

146 Light Oil Catalytic Processing

TOC/INDEX

technology is based on the work of Dr. George A. Olah, DEATON:

winner of the 1994 Nobel Prize in Chemistry. This topic was discussed in January 1994 at the API HF
The Alkad technology uses an additive to suppress the Alkylation Roundtable and Seminar on Safe Operation of
HF aerosol formation in the event of a leak from the HF Alkylation units. Reports of success with detectors
alkylation unit. Large scale release tests with this additive seem to vary. Two types of detectors are used. One type is
have demonstrated significant aerosol reductions. In 1992, a “point source” detector for specific ppm determinations.
a trial run was conducted at Texaco’s El Dorado, Kansas These detectors are based either on electrochemical cells or
refinery. This operation demonscrated that the technology the etching of silicon-coated electronic chips.
could successfully produce alkylate, and in fact, the alkylate The second type of detector is an “open path” or “line
quality improved both in octane and distillation. of sight” detector. One type that uses infrared principles
The first commercial unit has now been installed at the is available and another type that uses laser principles is
El Dorado refinery, using a UOP modular design for the under development. The “open path” type detectors
additive recovery and recycle unit. Texaco started up the would be used to detect and respond to larger leaks for
process on September 14th and their early results confirm activation of mitigation and isolation equipment.
the alkylare octane and distillation improvements.
ARMBRESTER:
STANLEY LOMAX (Texaco Refinery): I agree with the previous comments on the use of
As Mr. D’Auria mentioned, we started the injection of acid-detecting paint and video monitoring, and we are also
the additive on September 14th to improve the safety of investigating the use of HF point detectors that would be
our alkylation unit. In addition to the safety aspect, we are installed at the locations most likely to experience a leak.
now producing an alkylate that is more than one and a These analyzers can detect HF in the 0 to 10 ppm range.
half octane numbers higher than our operation prior to Regarding video monitoring, we have that system in
September 14th. That is at the same unit feed rate and place and we also set it up to record each one of the camera
feedstock composition. Along with that, the ASTM dis- signals so that if a leak does occur and the source of the
tillation 90% point and end point are both averaging leak is not identified immediately, we can play back the
about 30°F lower. tape and identify the source of the leak.

HIGGINS: QUESTION 13.

Thank you. I believe a paper on this subject will be What are some of the considerations when designing
presented at the 1995 NPRA Annual Meeting. HF emission mitigation measures?

QUESTION 12. CUNEO:

What methods of HF leak detection are most commonly Several years back, Shell installed extensive water deluge
used? Which methods have been most successful? systems around our HF alkylation unit at Odessa. Any
indication of leaks of either HF or hydrocarbon prompts
BOYCOTT: the board operator to immediately activate the deluge
We have recently installed a video monitoring system button in the control room. They are given instructions
in our alkylation unit. The system is comprised of five on the side of being conservative. Our people are told not
cameras with zoom and pan capabilities. Each camera to go out and investigate a plume observed by the TV
feeds a high resolution color screen in the control room. surveillance. Activate the system first, and then investigate.
The units have been positioned to ensure coverage of all Our process hazards analysis and subsequent dispersion
HF containing equipment. We are currently evaluating modeling of potential worst-case scenarios pointed out the
the addition of hydrocarbon detectors in key areas to need for immediate activation of the water deluge, if we
supplement the cameras. are to be effective in mitigating potential outside spread
Other technologies considered in the selection of this of the release.
system were infrared beam reflectors and wet chemistry As commented in the previous question, we are engi-
HF point source detectors. neering the installation of an open path hydrocarbon
detection system and HF point monitors as well.
CUNEO:
We currently use TV monitors, acid-indicating paint DEATON:
on flanges, and well-trained operators to watch and detect Our first step was to install systems for quick isolation
for HF leaks. We currently are evaluating a system which of equipment such as remote-operated valves and remote-
will employ a perimeter open path infrared detection operated start/stop switches for pump motors. We are now
system for isobutane coupled with ion mobility point HF looking at water deluge systems, and the first step in this
measurements around the acid regeneration facilities. process is to select credible release scenarios and to predict

1994 NPRA Q&A Sesioon on Refining and Petrochemical Technology 147

TOC/INDEX

leakage rates. Water mitigation flow rates are usually rec- developing uses a true solid phase, regenerable catalyst.
ommended at a ratio of 40:1, that is, 40 gpm of water per Currently we are in the process of conducting long-term
1 gpm of acid leak, for a 90% reduction in acid vapor stability evaluations in our pilot plants.
release. Earlier this year we developed a preliminary design
Vapor cloud dispersion models such as HG System and package that indicated that the capital and operating costs
Trace can then be used to predict downwind concentra- for this process would be very competitive with the exist-
tions. There is a joint industry program working to im- ing liquid acid technology.
prove the accuracy of vapor cloud modeling methods. As part of our initiative to increase licensee involvement
Water delivery systems should be designed for quick with our development programs, we have held prelimi-
actuation and for effective delivery of the required water nary discussions with a few customers to provide a status
flow rate to the acid release point. Water runoff, storage, update and explore potential applications of this new
and neutralization also need to be part of the overall process. We expect to have a detailed engineering design
mitigation design package. completed early next year, and, while significant technical
achievements have yet to be accomplished, our goal is to
LAUX: have a full commercial package for the technology avail-
Some additional items that we included were sizing for able by the end of 1995.
300 micron droplets based on literature data, and install-
ing both water curtains and remote monitors to attack the McLEAN:
largest plausible leak scenario. Our system for the water I do not have anything to add to what the other
mitigation also includes a preneutralized water source as panelists have said.
well as remote actuation from the control room and the
field. PORTER:
The rapid acid dump system that we are installing was Chevron is working with Chevron Research and Li-
designed for a pressurized 10 minute dump to a new censing on a process that is suitable to retrofit into an
dedicated vessel. The vessel sizing has double the maxi- existing process reactor system. We started a 10 BPD pilot
mum expected acid volume to allow room for the motive plant in January of this year to demonstrate process per-
fluid, and this system can also be activated both from the formance.
control room as well as from the field.
PHILIP GEREN (Haldor Topsoe, Inc.):
QUESTION 14. In 1988, Haldor Topsoe began a program to develop a
There have been many announcements regarding new solid catalyst alkylation process. Earlier this year the fin-
motor duel akylation technology that claim safety ishing touches were put on the development. The process
improvements over existing sulfuric and HF technolo- employs a small amount of a liquid catalyst deposited
gies. What is the current status of solid catalyst aky- upon the surface of a packed solid support bed.
lation technology? Have any commercial units been We and our partner in commercialization, the M. W.
licensed? Kellogg Company, are now preparing a feasibility study
for a customer of ours who intends to employ the process
ARMBRESTER:
The most developed solid acid processes are in the 0.5
to 10 barrels per day pilot plant stage, with the earliest of
these processes having operated for up to two years now.
The status of each of the processes would be best stated by
the participating firms, but each was briefly discussed in
two recent articles in the Oil and Gas Journal Volume
92(34) of August 22, 1994, pages 49 to 54. The economics
of at least one solid acid process appeared to offer cost
savings over conventional sulfuric acid units. Although
several firms are seeking licensees, we are not aware of any
commercial licenses having been sold.

D’AURIA:
In parallel with our development of the Alkad HF
additive technology, UOP has maintained a long term
research and development program aimed at developing
a solid catalyst alkylation system. The system we are

148 Light Oil Catalytic Processing

TOC/INDEX

in its first commercial scale application. We are not at and acts as a diluent. The loading capacity is a function of
liberty at the moment to discuss our customer. the sulfur compound type, the temperature, and the per-
cent of reduced nickel in the sorbent.
PETER KILNER (Catalytica): Reduced nickel will adsorb H2S to saturation at tem-
Catalytica, Conoco and NESTE are working to com- peratures above 100°F. Adsorption of mercaptans and
plete the development of our solid catalyst alkylation disulfides is low at l00°F, but increases to saturation
process. We are planning to have the technology ready for around 350°F. Thiophenes, particularly sterically-hin-
licensing by sometime next year. We are talking with dered ones like benzothiophene, are only slightly ad-
refining companies about the possibility of joining our sorbed. One supplier specifies the sulfur loading for
cooperation to build the first unit, and we will be disclos- reduced and stabilized 50% nickel catalyst above 325°F,
ing further details of our process at the Oil and Gas as 10 wt% or more for hydrogen sulfide, 10% for mercap-
Journals Alkylation Seminar at the end of this month. tans, 7% for disulfides, and 1% for thiophenes.
Users have achieved different results than what I have
just mentioned. This could be the result of upsets and
C. Reforming variability in the percentages of thiophenes in the naphtha
feed. To my knowledge, no one is routinely measuring
Feed thiophene content at less than 1 ppm total sulfur.

QUESTION 15. PORTER:

How do sulfur species (i.e. H 2 S RSH, thiophenes) We have not used nickel oxide commercially, but our
affect the loading capacity of nickel oxide sulfur laboratory tests generally support the previous comments.
sorbent?
ROBERT W. MORSE (Acreon Catalysts):
BOYCOTT: The surface area of the nickel is also very important in
The pickup of sulfur species on a reduced nickel sulfur determining the adsorption of the sulfur. High surface area
sorbent is dependent on a chemisorption reaction. How- nickel adsorbents have a higher capacity for thiophene.
ever, the molecular size of the sulfur species controls the
mass transfer of the molecules into and through the pores QUESTION 16.
of the material. In this way the mechanism and break- For those refiners using a sulfur guard bed on the
through curve become very similar to those of a pure reformer feed, the very low concentrations across the
sorption system, and molecule size has a direct impact on bed make monitoring the bed performance and opti-
ultimate capacity. As such H2S is the easiest species to pick mizing change-out before the bed is completely spent
up. Thiophenic compounds have been found to give the very difficult. What analytical test methods or other
lowest loading. The published capacity for one of the techniques are being used to ensure that the beds are
commercial chemisorbents is given below. performing properly?

H2S greater than 10 wt% SAYLES:

Mercaptans 10 wt% Typical reformer feed specifications are 0.5 ppmw sul-
Disulfides 7 wt% fur and 0.5 ppmw nitrogen when using bimetalliccatalysts
Thiophenes 1 wt% in semiregenerative units. A sulfur guard reactor will pro-
tect the reforming catalyst provided the inlet H2S concen-
D’AURIA: tration to the guard bed does not cause a sulfur
The question asked about nickel oxide adsorbents. The breakthrough.
active adsorbent actually is reduced nickel. UOP’s nickel We have liquid feed sulfur guard reactors protecting
adsorbent, ADS-llL, is sold in the reduced and surface two different reformers. In both cases, low pressure pre-
stabilized form. The ADS-l1L has a capacity for H2S and treaters are used, less than 600 psi. During normal opera-
mercaptans of about 15 wt% on the adsorbent. The tion, the inlet sulfur concentration is about 140 ppb and
thiophene capacity is much lower, about 2 wt% on the the outlet is 40 ppb. The cycle length can be calculated for
adsorbent. However, even when thiophenes are present your absorber configuration using manufacturer supplied
they do not appear to interfere with the adsorbent’s ability sulfur absorption coefficients. We replace the sulfur guard
to adsorb its full capacity for H2S and mercaptans. bed CuO reactant once per year. The normal run is 440
to 510 bbl of feed per pound CuO. The run is normally
MENEGAZ: limited by pressure drop before sulfur breakthrough.
We agree with Mr. D’Auria’s comment. Sulfur pickup is From our experience, pretreater sulfur breakthrough
done only by the reduced nickel. Nickel oxide is ineffective occurs after unit upsets, mainly power outages. As a rule,

1994 NPRA Q&A Session on Refining and Petrochemical Technology 149

TOC/INDEX

we replace the beds after a major upset as a protection to dumped. Subsequent testing for sulfur levels will not only
the reformer catalyst even if breakthrough has not been define the breakthrough curve, but also indicate which
detected. material may be reloaded. The reloaded material should
Another operational concern is that high levels of ab- be installed at the top of the bed. Care must be taken to
sorbed sulfur onto the guard bed absorbent, with contin- maintain an inert environment.
ued use, could allow the sulfur to desorbe, causing a Another approach would be to install two beds in series
reformer feedstock contamination problem. This means with the required piping to alternate lead and lag posi-
that the observed sulfur breakthrough will occur before tions. By alternating bed position in the train, essentially
the calculated breakthrough. all of the sorbent capacity could be utilized.

PORTER: CUNEO:
We rely on Houston Atlas sulfur analyzers. The analysis Nothing further to add.
takes care, but we believe we can get down to 0.05 ppm
accuracy. As far as detecting when a bed is spent, we believe D’AURIA:
the most straightforward approach is to put liquid sample We have seen all three methods used to monitor sulfur
taps at several elevations on the vessel to monitor the guard bed performance: calculating the sulfur loading
progression of the sulfur removal through the bed. based on feed sulfur, monitoring the outlet sulfur, and
also analyzing the adsorbent. The most common method
TREESE: is monitoring outlet sulfur. As the outlet sulfur begins
I would echo what Mr. Porter has said. We use the approaching 0.2 wt%, we recommend shutting down and
Houston Atlas machine. We use the thermal cracking reloading the bed. The best analytical method for meas-
technique for analyzing our sulfur. In addition to the side uring sulfur in the guard bed outlet stream is ASTM-
taps, I think it is a good idea to calculate sulfur accumu- 4045, which utilizes hydrogenolysis followed by
lated in the bed versus the bed capacity. Tracking that can rateometric calorimetry.
help act as a check and an early warning for when you are
about to expend a bed.
DAMIAN ARCHBOLD (ChemAlert Corporation):
This spring we will be introducing a new technology
McLEAN:
for on-line total sulfur measurements that will give you
The only proven methods are dependent on the ability
the capability of measuring total sulfur at one ppb on-line
to analyze the low sulfur concentrations in and out, and
consistently. This is being tested now at several refineries.
obviously that is difficult to do at such low levels of 0.1
The results will be available in the spring.
ppm to 0.2 ppm, so this can be used as a rough estimate.
It helps to know the capacity of your sulfur guard adsorb-
SONI OYEKAN (Sun Company, Inc.):
ent. Nickel type liquid phase guard beds typically adsorb
between 10 wt% and 20 wt% sulfur and vapor phase One indirect method for checking for the problems on
guards from 15% up. In the case of a vapor phase system, the sulfur bed ahead of the reactor is to check through
only the liquid stripper bottoms need to be analyzed, since catalyst samplers for the level of sulfur on the catalyst in
100% of the sulfur in the recycle loop will eventually be the first reactor. With sulfur guard systems, the sulfur in
removed. that first bed is usually much less than 0.02%. So, if you
Accurate determination of bed removal efficiency may have inefficiencies in the performance of the sulfur guard,
require calculation after the fact, by comparing total sulfur you will notice a rapid increase in the sulfur level in the
loading on the spent material with the total feed sulfur first reactor catalyst samples.
from daily feed analyses.
DOUGLAS J. EVANS (Petro-Canada Products):
BOYCOTT: I was wondering if anybody on the panel might be able
I would like to add a comment on the utilization of the to comment on the affinity of zinc oxide, nickel oxide,
sorbent material. Sulfur guard beds use a chemisorption copper oxide to mercaptan removal, if any.
reaction to fix the sulfur species on the bed material. The
concentration gradient through the bed is very similar to D’AURIA:
that experienced in a pure sorption system. In the case of The affinity for mercaptan removal of the reduced
sulfur chemisorption on reduced nickel, the gradient is nickel adsorbent is very good. The ultimate capacity of
relatively shallow and results in rejection of 20% to 30% ADS-11L, for example, is around 15 wt% at 250°F to
of sorbent capacity as unused at breakthrough. One ap- 350°F. Zinc oxide is used primarily for H2S removal. Its
proach that has been used to reduce the rejection of unused use in reformer applications is limited by the fact that for
capacity is to sequentially label the drums as the bed is every mole of H2S adsorbed, a mole of water is produced.

150 Light Oil Catalytic Processing

TOC/INDEX

MENEGAZ: The liquid feed guard beds are relatively simple and
We understand that the copper material picks up most inexpensive. A typical installation using reduced nickel is
of the hydrogen sulfide and only a very limited amount of a small carbon steel vessel, sized for liquid-hour phase
the mercaptans. velocity (LHSV) of 8 to 12. Normal effluent purity is less
than 0.1 ppm sulfur. There are 50 or more liquid sulfur
QUESTION 17. phase units worldwide.
The benefits of removing sulfur in reformer feed has Several refiners have added liquid guard beds as retrofits
promoted several approaches to a sulfur guard bed to existing units with minimum capital investment. One
system. Both a system on the liquid feed and the refiner is reported to have increased the cycle length from
recycle gas stream are possible. How does the expe- 8 to 23 months after adding a nickel guard bed. Experience
rience on these compare and what unit aspects make
with sulfur adsorbents is discussed on pages 147 to 148 of
one or the other more appropriate? Have there been
the 1992 NPRA Question and Answer Session transcripts.
any problems, difficulties, or limitations with either
approach?
PORTER:
Many years ago, we tried to remove the sulfur in the
McLEAN:
recycle gas and did not like it for the same reason Mr.
Liquid phase sulfur guards treat the reformer feed,
while vapor phase systems treat both the feed and the McLean mentioned. The sulfur in the feed hit the catalyst
recycle gas before they enter the reactor. A guard solely on before the recycle guard bed could take it out. So we
the recycle itself would have marginal value since it could invented the Chevron Sulfur Sorber process, which re-
only take out the H2S in the gas and the catalyst would moves sulfur from the liquid feed. We now use that on all
still be exposed to all the feed sulfur on the first pass. our reformers.
Acreon Catalysts markets adsorbents for both systems.
Liquid phase systems are smaller, operate at lower tem- BOYCOTT:
peratures, and are less expensive to install. They com- I agree with the previous comments. I will add that we
monly use copper- or nickel-based adsorbents. Copper is employ a reduced nickel adsorbent on our liquid feed
effective at removing H2S and mercaptans, while nickel stream.
can also remove some thiophenes. It is also easier to change
out the adsorbent on the run with a liquid phase system. D’AURIA:
Vapor phase systems are more expensive because they The only thing I will add is that the liquid system does
are directly in the reformer loop and operate at higher have the advantage of not having any impact on the
temperatures. They must be isolated during regeneration, reformer recycle gas compressor circuit, whereas the vapor
requiring some additional valving and piping. Since the phase system will add additional pressure drop in the
vapor phase unit treats both the feed and the recycle gas, recycle gas circuit.
it is more effective in responding to upsets which may feed
a large quantity of sulfur to the unit in a short period of MARK DORAN (ICI Katalco):
time. If this upset condition exceeds the ability of the I agree with most of the comments from the panel.
guard to remove on the first pass, thiophenic sulfur will With vapor phase sulfur guards, it is possible to place the
convert to H2S in the reformer and will be removed on
guard on the recycle gas at the low temperature part of the
the second pass. As a result, vapor phase units may be
circuit, and that may well reduce the capital cost of the
preferred for reformers which are subject to such upsets,
figures given by people on the panel.
particularly on more thiophenic feeds.
A liquid feed guard will only remove sulfur from the
Engelhard has presented a paper at the 1989 Annual
Meeting, Paper Number AM-89-45, which compared hydrocarbon feed to the plant. Another point in favor of
these options and offered an economic comparison. vapor phase guards is that they do not increase the hydro-
carbon inventory of the reformer unit. ICI Katalco can
MENEGAZ: provide PURASPEC™ absorbent for both vapor and liq-
I would agree with Mr. McLean’s comments. The vapor uid phase applications. Liquid phase is the most common.
phase process is definitely more effective for removing
sulfur. It is also quite a bit more expensive. I have seen WILLIAM TOREK (Texaco Refining & Marketing, Inc.):
figures quoted of $425,000 for installation in a 10,000 We recently loaded fresh, high nickel, catalyst in a sulfur
barrel per day reformer. Many of these processes are found guard reactor on the run and experienced quite a bit of
in conjunction with units using high-skewed plati- water in the downstream reformer. Can someone com-
num/rhenium catalysts. Such catalysts are extremely intol- ment on the possibilities of water formation resulting from
erant of sulfur and require a higher degree of sulfur removal. the sulfur guard catalyst’s passivation agent?

1994 NPRA Q&A Session on Refining and Petrochemical Technology 151

TOC/INDEX

D'AURIA: experiences with chloride water problems when there is a

It is possible to form significant quantities of water if guard bed in the reactor circuit on the combined feed. It
you use an adsorbent that is sold in the oxidized state, and is more difficult to maintain chloride. A serious water
then reduced in situ. UOP’s ADS-11L, however, is sold in upset will give a large liberation in H2S, which will in turn
the reduced, surface stabilized state. We do not perform nullify the economic benefit for installing the guard bed.
an in situ reduction, and, hence, do not generate any water
on start-up. SONI OYEKAN (Sun Company, Inc.):
I want to address the question of whether the sulfur
RONALD E. MARRELLI (Phillips Petroleum Company): guard is only for upsets. In the case of skewed metal
Is the main purpose of the sulfur guard to prevent catalysts or unbalanced catalysts, where you have high
possible upsets on the hydrotreater from upsetting the rhenium to platinum ratio, you get substantial benefits in
reformer or to prevent any sulfur from getting to the terms of performance if you operate below 0.2 ppmw
reformer? Are there additional benefits in normal opera- sulfur in the feed. So in order to make sure that the sulfur
tion? is quite low in these systems, you need to have the sulfur
guard in line for that particular type of catalyst.
PORTER: You could also have similar benefits for the balanced
Sulfur guard beds are useful to reduce the impact of a catalysts as well, that is, balanced metals catalysts, but you
process upset, but we find they have a great value in have substantial performance benefits for the skewed met-
removing the trace sulfur in normal feed. als if you run in the range of 0.1 ppm to 0.2 ppm sulfur.
That is benefits in terms of octane barrels as well as cycle
MENEGAZ: length.
I know of one incident where a refiner developed a
problem with the upstream hydrotreater with a crack in CORNELL SLADE (ACREON Catalysts):
one of the feed-effluent exchangers, and consequently they Acreon is in a unique position because we can supply
had 4 ppm to 5 ppm in their feed sulfur. They used the two types of vapor phase and two types of liquid phase
vapor phase sulfur guard process and operated for approxi- sulfur guards.
mately 18 months without shutting down or seeing any For complete sulfur removal (i.e., H2S, thiophenes), we
loss of octane. recommend our HRD 264, vapor phase sulfur guard.
Sulfur (as H2S) can be removed by this adsorbent, up to
LAUX:
18% by its weight, which is up to 65 lb/ft3.
We have our sulfur sorber on our hydrocracker naphtha
Our ZN 0406, also used in vapor phase, is a zinc-based
which bypasses the naphtha hydrotreater and is fed di-
product and at 99 lb/ft3 is one of the most dense sulfur
rectly to the reformer. The sulfur sorber provides protec-
guards on the market. It can pick up up to 10% of its
tion against H 2 S from upsets in the hydrocracker
weight.
fractionation system and protection from mercaptans that
For liquid phase applications, the D- 1275 is the prod-
may be formed in the hydrocracker reactor at end of run
uct of choice. It can pick up up to 15% of its weight, and
temperatures.
weighs up to 50 lb/ft3. Nickel is the active ingredient.
HEROS DERGREGORIAN (Giant Refining Company):
Our MD 101 copper product is also for liquid phase
In regards to water formation in the nickel guard bed, applications and can pick up up to 5% of its weight. The
is it possibly high enough for the water to cause unbal- density is 56 lb/ft3.
anced chloride in the reformer?
QUESTION 18.
D'AURIA: What are some of the ways refiners are using N/R
We have not seen any problems with water generation analytical technology to monitor catalytic reformer
on start-up. Theoretically, if you were reducing a high performance?
nickel content, nickel oxide material, a significant amount
of water would be generated. You would have to make sure ARMBRESTER:
this water did not enter the reformer reactor section. As Ashland is using its Infratane® on-line NIR technology
stated earlier, this is not a problem with ADS-11L. to provide octane value, aromatic content, and olefin con-
tent of the reformate product from our low pressure CCR
BOB RODDEY (Roddey Engineering Services, Inc.): reformer at the Catlettsburg refinery. At the present time,
I have not heard anyone say anything about chloride the NIR analyzer reports research, motor and pump octane
water balance, until the previous comment, on units that numbers as operating indicators only, but we have a project
put the sulfur guard in the reactor circuit I have had under way to establish closed-loop octane control by using

152 Light Oil Catalytic Processing

TOC/INDEX

the analyzer’s research octane signal to set the weighted about 10% of the available reformer charge and will allow
average inlet temperature on the reformer reactors. us to reduce the benzene content in reformate by about
In addition to the octane values mentioned previously, 70%, and reduce the overall gasoline pool benzene content
the NIR also provides on-line analysis for total olefins, by 30%. In our case, this amounts to about 500 barrels
total aromatics, benzene, toluene, total xylenes, and ethyl per day less benzene in the gasoline pool. Then, to allocate
benzene. These values can be used to optimize our down- benzene between RFG and conventional gasoline, we plan
stream aromatic extraction process. to pull a side draw from the reformer stabilizer. This side
draw will remove up to 70% of the produced benzene from
CUNEO: the reformate product in a volume as small as 15% of the
We are doing work on characterizing the reformer feed total reformate produced. The benzene content of the side
for possible consideration in feed forward control, as well draw will contain up to 4.5 times the normal benzene level
as for analyzing octane and for benzene. of 1.1 vol% in the whole reformate. The vapor side draw
will be used as blendstock for conventional gasoline.
D’AURIA: The fractionation systems we will employ avoids the
The NIR technique replaces traditional lab or on-line high cost of constructing and operating a solvent extrac-
techniques (such as knock engines, on-line octane analyz- tion unit. Precursor prefractionation can be done with
ers, gas chromatographs) that introduce high maintenance minimal capital investment, but will negatively impact the
costs, slow turnaround, and yet provide limited capabili- gasoline pool octane by diverting 10% of the reformer feed
ties. State-of-the-art NIR analyzers measure properties of as a 60 octane blendstock.
several process streams in the reformer simultaneously on
a real-time basis. UOP recognized the importance of this ARMBRESTER:
technology and recently purchased Guided Wave in order The advantages and disadvantages obviously depend
to offer this capability to our licensees. As of this date, there on each refiner’s processing configuration, as well as their
are 14 Guided Wave analyzers sold for reformer applica- individual situation regarding RFG production and anti-
tions measuring RON, MON, RVP, benzene, or PONA. dumping considerations. If a refiner has an outlet for
These applications are being utilized for both feed forward benzene, either through outside sales or as an in-plant
control based on PONA, or feedback control based on petrochemical feedstock, then the best option may be to
octane. The data from a single NIR analyzer monitoring maximize benzene precursors in the reformer feed and
the feed and reformate streams, coupled with data on then extract the benzene from the reformate.
moisture, chloride and hydrogen content in the recycle gas If an outlet for benzene is not available, then it is
from other on-line analyzers, will provide the basis for probably best to minimize the benzene precursors, primar-
implementing advanced control strategies for the reformer. ily cyclohexane and methylcyclopentane, in the reformer
feed by raising the initial boiling point of this stream.
PETER MIKKELSEN (Petro-Canada Products): These components would then be saturated in the C5-C6
Has anyone had experience with the hand-held NIR isomerization unit, thus reducing the total benzene in the
analyzers for operators to use for control. There are a few gasoline pool. Efficient fractionation will be needed to
of them on the market that are relatively inexpensive and move the benzene precursors into the isomerization unit
we are in fact investigating use of such a piece of equip- feed stream without also sending an excessive amount of
ment. C7S and benzene to the isomerization unit.
One disadvantage of this route is that it reduces the
total amount of reformer feed available, which can nega-
Process tively impact the refinery’s hydrogen balance. If a refiner
has neither an isomerization unit nor aromatic extraction
QUESTION 19. capability, they may be forced to minimize the benzene
In order to control the amount of benzene in future RFG precursors in the reformer feed and blend the entire light
(reformulated gasoline) what are the relative advan- straight run naphtha stream to gasoline.
tages or disadvantages to limiting precursors in re-
former feed versus removing benzene from D’AURIA:
reformate? The major source of benzene in the gasoline pool is the
catalytic reformate that typically contributes 50% to 75%
DEATON: of the benzene in the gasoline pool. Two basic approaches
Clark’s Blue Island refinery in Chicago will market both can reduce net benzene production: minimize its forma-
RFG and conventional gasoline. First, we plan to use tion by removing the precursors charged to the reformer
precursor prefractionation to reduce the benzene content or fractionate a benzene-rich light reformate stream for
of the total gasoline pool. The precursor product will be subsequent benzene conversion or extraction. Preventing

1994 NPRA Q&A Session on Refining and Petrochemical Technology 153

TOC/INDEX

may have to be increased to compensate, which will reduce

the overall gasoline yield. Also, benzene originally present
in the naphtha will not be removed by the fractionation.
And, finally, in some cases this may result in pool benzene
levels below allowable target limits, with the effect of
unnecessary octane and hydrogen debits.
The alternative presented in the paper was the installa-
tion of a reformate splitter and liquid phase catalytic
hydrogenation of the benzene-containing light reformate
stream. This option was evaluated for four different crude
types, with hydrogenation severity adjusted to meet pool
benzene limits while maintaining improved overall gaso-
line yield and hydrogen balance. For the cases evaluated,
significant economic rates of return for the incremental
capital involved were indicated.

PORTER:
I agree with a lot of the previous comments, that the
the formation of benzene rather than converting it after it key decision is whether you want your refinery to make a
is formed makes good engineering sense. However, in benzene product. At our Pascagoula refinery, we are re-
certain cases, practical and economic constraints may limit moving benzene from our reformate, but at another refin-
the ability of refiners to implement this strategy. ery we will try to minimize benzene formation. Once you
When choosing between light reformate processing or decide whether to make a benzene product, then choosing
precursor removal for benzene reduction, refiners must the right option will depend on capital costs and operating
consider three variables: the level of benzene allowed in the expenses between the various alternatives.
regulated pool, the benzene-producing tendency of the
reformer, and the overall refinery flow scheme and crude SAYLES:
source. If reducing benzene below 1 vol% is necessary, the A similar question was asked in the 1990 Question and
precursor removal may not be sufficient. If the reformer Answer Session, question 30.
operating pressure is high, the production of benzene can From our perspective, benzene control to produce RFG
occur even after precursor removal due to dealkylation of or conventional gasoline conforming to the refiner’s 1990
heavier hydrocarbons. Light reformate processing may also baseline requires a comprehensive review of all the gasoline
be necessary if the refinery flow scheme is heavily depend- pool benzene sources. Benzene sources are the FCCU
ent on the reformer as a gasoline pool component, if the naphtha, 25% of the total; light straight run or conden-
crude used is highly cyclic, or if the reformer feed comes sate, 5% of the total; coker naphtha, 5% of the total; and
from a number of other process units in which accurate reformate, 65% of the total benzene in the pool. Limiting
control of benzene precursors is difficult. Finally, if the reformer benzene can have a significant effect and the
refiner has a petrochemical outlet for benzene, extraction methods for limiting benzene are:
of benzene from the reformate may be the most economical 1. Benzene precursors boil from 180°F to 230°F in
choice for reducing the benzene content of gasoline. the HSR feed. Control of the reformer feed IBP to
greater than 220°F to 230°F should remove the
McLEAN: benzene precursors. The increased light naphtha
Engelhard presented a paper at the 1992 NPRA Annual yield from the CDU will lower the gasoline pool
Meeting which addressed this question. That was paper octane/bbl. To recover the octane loss, LSR
number AM-92-51. This answer will summarize the con- isomerization can be used to increase the octane
clusions from that paper. from about 67 to 78 RON. The disadvantage is
The primary advantage for removing benzene precur- shorter isomerization catalyst life and increased
sors from the reformer feed is that in many cases it can be isomerization hydrogen consumption due to the
accomplished with existing fractionation equipment and saturation of benzene.
is therefore on the surface a “no capital required” option. 2. Another approach is removal of the benzene from
It has a number of disadvantages. Valuable hydrogen the light reformate which produces a higher octane
producing compounds are also removed, which of course gasoline pool. Several commercial processes exist
reduces the potential hydrogen yield from the reformer. to remove the benzene. However, these processes
This front end is also low in octane and if it is routed must be carefully reviewed with respect to the
directly to the gasoline pool, then the reformer severity benzene NESHAP regulations, OSHA benzene

154 Light Oil Catalytic Processing

TOC/INDEX

exposure limitations, and disposition of the con- is that? Are the impacts a combination due to pressure and
centrated benzene naphtha produced. The RFG severity of the operation?
regulations have strict prohibitions to limit the
outlets for these concentrated benzene streams. D'AURIA:
3. The third approach is to limit the operating sever- The reformer pressure and octane severity, in addition
ity to minimize benzene production. The produc- to feedstock boiling range, will impact the amount of
tion of benzene is a thermodynamically controlled benzene made through dealkylation of heavier compo-
reaction (temperature/pressure) and a maximum nents. To illustrate this relationship, the table below shows
benzene concentration will be reached depending benzene yields from the reforming of a midcontinent
on the operating conditions. For a semiregenera- naphtha:
tion reformer, there can be a maximum possible Benzene Yield, LV%/FF
benzene increase of 30% above start of run. An-
other way of estimating the benzene rise is a Boiling Range RONC 50 PSIG 300 PSIG
0.0125%/°F increase in WABT. An interesting C6-370°F 98 2.2 2.5
exception is benzene produced by a CCR unit C6-370°F 102 2.7 3.0
which shows no increase in benzene with time. C7-370°F 98 0.5 1.2
C7-370°F 102 0.6 1.5
(UOP R32 fresh versus spent data, NPRA Ques-
tion and Answer Session on Refining and Petro-
chemical Technology, 1991) RICHARD M. GIALELLA (IFP Enterprises Inc.):
At constant severity, lower reactor pressure will Limiting precursors may be effective in decreasing ben-
reduce benzene formation by about 0.0024 wt% zene in gasoline. However, it does nothing to control
benzene per 1 psi reduction in separator pressure. benzene formed by the hydrodealkylation reaction. Also,
Higher pressures will increase aromatics by 0.028 it may severely impact light paraffin isomerization because
vol% aromatics per 1 psi decrease in separator of poorer feedstock. This solution may not be adequate
pressure. (Feed IBP is at 215°F; Gilsdorf, et al, for reformers operating at high pressure and high severity.
“Platforming in the Reformulate Gasoline Era”, Controlling benzene downstream of the reformer permits
1992 NPRA Annual Meeting, AM-92-16) monitoring benzene regardless of reformer feed quality
and octane of the reformate. In addition, processing
Pressure, psig 300 50
downstream of the reformer permits the decrease of ben-
RONC Benzene, wt% Benzene, wt%
zene to low levels in the gasoline pool and takes full
92 0.8 0.2
advantage of the complex model equations in terms of
96 1.0 0.3
toxics reduction, thereby allowing more aromatics ulti-
100 1.2 0.4 mately into the gasoline pool.
IFP offers a full set of processes which allow for the
TREESE: reduction of benzene downstream of the reformer, for
I would, ofcourse, go with the previous comments. We, example, benzene saturation using either a noble or a
being primarily a California refiner, have the additional non-noble metal catalyst followed by isomerization, if so
problem of trying to meet the California Air Resources
desired.
Board (CARB) Regulations, which require less than 1%
benzene in the product.
DAVID N. WATERFIELD (BP Oil London):
We already have dehexanizers ahead of our reformers,
You will be very lucky to get below 1% benzene on a
so most of the precursors are already fractionated out. In
400 lb to 500 lb reformer, even if you remove all the C6
our case, we have to post-fractionate in order to meet the
in the feed. On a low pressure CCR you may get down to
CARB Regulations.
Outside California, people may be able to meet the EPA about one-half percent.
benzene requirements with prefractionation only. Our
thought is that there may be additional fractionation QUESTION 20.
required to meet toxics targets that are upcoming. What methods are being used to reduce plugging and
Refinery rest runs confirmed that prefractionation corrosion in the catalytic reforming stabilizer?
alone, in our case, was not sufficient to meet the benzene
requirements in our reformate. SAYLES:
Similar questions were asked in the 1987 NPRA Ques-
HEROS DERGREGORIAN (Giant Refining Company): tion and Answer Session, question 30 (p. 120) and in the
My question is regarding impact of pressure on the 1989 NPRA Question and Answer Session, question 22
alkylation of high aromatics into benzene. How sensitive (p. 133).

1994 NPRA Q&A Session on Refining and Petrochemical Technology 155

TOC/INDEX

Our semiregenerative stabilizer columns have not ex- This, however, is not a method that addresses the causes
perienced plugging problems in the overhead system. of the pluggage and corrosion.
However, corrosion byproducts in the stabilizer overhead
water boot have been detected. The areas of highest cor- PORTER:
rosion are locations where the overhead vapors reach their We control plugging in our stabilizer, by water washing
dew point. Normally the stabilizer offgas contains 5 ppm the column. This is usually done during regeneration, but
to 7 ppm H2S and NH3 with trace amounts of chlorides some do it on the run.
and water. The corrosion is due to water entrainment in
the stabilizer which hydrates a chloride salt, typically BARLOW:
ammonium chloride and causes localized corrosion. The Most of the fouling and corrosion problems associated
concentrations in the overhead water boot are relatively with reformer stabilizer operation are due to NH4CI depo-
high, 0.5 ppm to 1 ppm Fe and the water rate is very low. sition. When a water wash in a stabilizer is not practical,
Corrosion can be eliminated by keeping the tower bone the use of a chemical solution has proven very successful.
dry which is difficult while maintaining a proper We have seen the use of amine technology to convert the
water/chloride balance in the reactors. ammonium chloride salt into a more mobile salt. In
We believe the corrosion rate is low. However, as a addition to the potential for plugging, the mobile salt can
precaution, a low level of filming amine is used, 1 ppm to be extremely corrosive. This approach is not recom-
5 ppm. Another approach is to use liquid phase scrubbers mended. We have been successful using detergent/disper-
to reduce chloride levels and eliminate plugging. sant packages to reduce salt deposition and keep the salt
We have had an experience with high pressure drop on in the liquid phase.
the feed side of the stabilizer/feed effluent exchanger of
about 90 psi. The high pressure drop was believed to be LAUX:
due to ammonium chloride pluggage. Condensate injec- When we operated only a semiregenerative reformer,
tion into the feed washed the ammonium chlorides from we also were able to simply water wash our stabilizer at
the bundle and reduced the pressure drop to about 25 psi. each regeneration to solve the plugging problem. When
During the condensate wash, the observed stabilizer over-
we started up our continuous reformer in 1991, that
head water phase ammonium concentration increased to
option disappeared.
3,000 ppmw from the normal 5 ppmw to 7 ppmw range
We have good success with injecting a salt dispersant
and the chlorides increased to 2,500 ppmw from the
chemical that drops the ammonium chloride out the
normal 5 ppmw to 7 ppmw range. (The Cl/NH3 ratios
bottom of the stabilizer. We have heard of severe corrosion
are: observed 0.9, calculated 1.0 a good check for ammo-
resulting from using these chemicals, but we did not see
nium chloride salts). A corresponding increase in the iron
any corrosion when we last inspected the stabilizer. That
content from less than 1 ppmw to less than 7 ppmw was
inspection was done 2 years after we started using the
observed. The high concentrations existed for the duration
chemical.
of the water wash and then returned to the normal.
BOYCOTT:
TREESE:
Unocal has been using continuous process improve- I would agree with the comments on ammonium chlo-
ment methods to prevent excursions by improving control ride salts causing the plugging problems. The source of the
of the addition of our chloriding agent. That has resulted problem is in the hydrotreater providing feed pretreat to
in reduced pluggage and reduced corrosion. the reformer. The breakthrough of organic nitrogen com-
pounds causes the formation of ammonia in the reformer
FELTROP: and subsequently ammonium chloride salts. The process-
I agree with the previous comments on elimination of ing of coker naphtha can worsen this problem.
nitrogen from the reformer feed, the effects ofammonium Although we process up to 25% to 30% coker naphtha
chloride, and using the water wash systems. A corrosion in our reformer feed pretreater, we do not experience
inhibitor can also be used to protect the overhead and plugging problems in the stabilizer. We believe that this is
minimize corrosion products in the system. Amine injec- due to careful selection of hydrotreater catalysts to elimi-
tion has also been used by several refiners. If it is used, it nate the nitrogen compounds in our reformer feed.
needs to be carefully monitored because the high molecu-
lar weight salts can deposit in the tower bottoms and CUNEO:
downstream equipment. We would agree with the previous comments, espe-
An analogous plugging problem was seen recently at a cially about the need to set the severity of the naphtha
major refiner and was treated effectively using an antifou- hydrotreater by organic nitrogen rather than by sulfur.
lant that combined a dispersant and a metal deactivator. We try to treat down to less than 0.2 ppm weight.

156 Light Oil Catalytic Processing

TOC/INDEX

Another potential source of ammonia is sulfiding cata- hydrotreater and have not observed any problems with the
lysts during regeneration. We recommend use of a sulfid- reformer feed nitrogen content.
ing agent, such as dimethyldisulfide as opposed to sour
hydrogen from a hydrotreater which can contain some KENNETH D. PETERS (UOP):
ammonia in the vent stream. Further to Mr. Deaton’s comments, we have been speci-
fying calcium oxide/zinc oxide treaters on the feed to the
DEATON: Platforming unit stabilizer column for about four years
For those refiners who run coker naphtha through the now. We have had some success with these treaters, par-
reformer, plugging problems will occur if the naphtha ticularly when the feed has been heated. The colder or
contains high nitrogen content coker naphtha. At Clark’s ambient temperature treaters do not always seem to oper-
Hartford refinery, a 900 psi coker naphtha hydrotreater is ate as well as the treaters that are set up to run with
effective in lowering the nitrogen content from 60 ppm preheated feed in the 200°F to 300°F range.
down to 0.5 ppm prior to blending with virgin naphtha Regarding the ammonium chloride, the primary time
fed to a typical 300 psi naphtha hydrotreater. when enough ammonium chloride is formed to cause
We have also heard that some people have installed zinc operational problems in the Platforming equipment is
oxide and calcium oxide adsorbent beds in the stabilizer when the upstream naphtha hydrotreating unit has an
liquid feed line to remove chlorides. Putting these adsor- upset situation and nitrogen breakthrough occurs. The
bers in a liquid service is relatively new, and the success of increased feed nitrogen then accelerates the ammonium
these systems is not yet available. These beds are much chloride formation. Also, in units where the reformer
more common in gas service and the results are good at catalyst surface area has declined significantly, the catalyst’s
reducing offgas line fouling and corrosion. ability to retain chloride is reduced, and the recycle gas
HCI level increases as more chloride is injected to keep up
DAVID L. SMITH (ALCOA):
catalyst activity. This increased HCI level also contributes
To support Mr. Deaton’s comments, there are selective to increased ammonium chloride formation.
adsorbents that are effective in removing chlorides from
The best method to minimize plugging and corrosion
the liquid reformates to quite low levels, to prevent the
in the stabilizer column is to minimize ammonium chlo-
formation of ammonium chlorides in the stabilizer trays
ride formation with better naphtha hydrotreating opera-
and to eliminate the corrosion problems.
tion to avoid nitrogen breakthrough and better reforming
catalyst surface area management to minimize HCI in the
PAUL FEARNSIDE (Nalco/Exxon Energy Chemicals):
recycle gas.
We have seen success for certain refiners where they pay
particularly close attention to the water wash system on
J. DOUGLAS PERKINS (United Catalysts, Inc.):
the naphtha hydrotreater, particularly if the water wash is
intermittent. We have seen at those times that the feed to We find with the zinc oxide/calcium oxide chloride trap
the reformer becomes extremely wet, carrying the resul- that the amount of ammonia will severely affect the use-
tant ammonia content with it. fulness of this material at lower temperatures, around
100°F. The ammonium chloride coats the material. At
HEROS DERGREGORIAN (Giant Refining Company): higher temperatures, the ammonium chloride decom-
We have been discussing guard beds for sulfur break- poses, i.e., at 230°F to 250°F and higher. So at higher
through to reformer charge. Is anybody doing work on the temperatures it works quite well. At lower temperatures it
use of guard beds for a nitrogen problem? If one is proc- also works, but only if there is no ammonia in the feed.
essing different kinds of feedstock, and the naphtha con-
tains different levels of nitrogen compounds, I would MICHAEL BUCKO (Merichem Company):
expect nitrogen breakthrough causing problems down- Fiber-Film Contactor technology has been proven suc-
stream. I am less concerned about chloride fouling than cessful in eliminating HCI corrosion and both ammonium
on balancing the chloride level on the catalyst. I was and ferric chloride plugging of column internals and the
wondering if anybody has discovered a guard bed for reboiler. Several treating units are currently in operation,
nitrogen removal. most of which are in the Far East.
Secondly, has anybody had experience with Alaskan
crude having a high nitrogen level in the naphtha? MARK DORAN (ICI Katalco):
Fixed beds of chloride absorbent are an attractive solu-
SAYLES: tion to this problem because they offer a low capital cost
We periodically get Alaskan North Slope into our project, compared to a caustic wash system for instance.
refinery and have noticed a slight increase in the feed They also offer a low maintenance/operator attention re-
nitrogen to the hydrotreater. We run about a 900 psi quirement, compared to an antifoulant injection system.

1994 NPRA Q&A Session on Refining and Petrochemical Technology 157

TOC/INDEX

PURASPEC™ absorbents are succesfully installed in such is a more acid-resistant material, but has a slightly lower
applications. water capacity than the RDG-418.
An alternative to using an acid-resistant sieve is to install
QUESTION 21. a nonregenerable chloride trap upstream of the drier.
Does anyone have experience in the use of molecular These recycle gas dryers have been used since the early
sieves on the recyc/e stream for drying during start- 1970’s, and the chloride-resistant adsorbent has been used
up? How much could the dryout time be reduced? Is successfully for over a decade.
there any new technology available for avoiding chlo-
ride attack on the molecular sieve? What other proce- HEROS DERGREGORIAN (Giant Refining Company):
dures or products are being used? We have been using nitrogen drying successfully for
reducing the drying time approximately in half.
LAUX:
We route our recycle gas from our semiregenerative QUESTION 22.
What has been the industry experience with the new
reformer at the Three Rivers refinery through a glycol
high alumina/low platinum CCR reformer catalyst?
contractor and then two parallel molecular sieve beds when
Comment on stability, activity and physical properties.
we start up after regeneration. The dryout time is cut to
two days, which is about 50% of the dryout time without
the dryers. The molecular sieves are loaded with 3A sieve D’AURIA:
and we have not noticed any chloride attack, but they are UOP’s R-132 catalyst was first put in stream in March
only used in this service two or three times per year. of 1992. This catalyst exhibits yields equivalent to our
earlier R-30 series catalyst, with improved activity and a
The contractor and the sieve beds are used as part of a
significant improvement in surface area retention.
cryogenic gas liquids recovery unit when they are not
In March of 1993, we commercialized a second catalyst
being used for regeneration dryout.
in this series, R-l34. This product has the same perform-
ance benefits as the R-132, but with about 23% less
McLEAN:
platinum.
Recycle gas dryers can significantly reduce dryout time. As of September of this year, there are 17 units in
Engelhard makes an adsorbent known as “Sorbead R,” operation with R-130 series catalysts, and there are an
which is specifically designed for recycle gas applications. additional seven refiners that have selected this type of
There are three domestic applications of this material in catalyst for reloads in the near future. The units have been
this service. Sorbead R is more cost effective than molecu- achieving their expected catalyst strength, surface area
lar sieves in this HCI-bearing application. For units with- retention, activity and yield performance. We have been
out recycle dryers, dryout time can be improved by careful very happy with the commercial results.
attention to draining, making sure that there are adequate
low point drains, and that these and the separator are McLEAN:
continuously kept moisture-free until the introduction of
Our affiliate, Acreon Catalysts, markets several new
feed. Ensuring that there are no dead areas during nitrogen generation catalysts for continuous reforming, including
and hydrogen purges, and minimizing the time for chang- E-1000 and CR-201. Both offer improved ultimate life
ing of blinds also help.
versus older catalysts due co decreased rates of surface area
loss. This has been demonstrated commercially on both.
PORTER: The CR-201 has a higher density and therefore has
We have historically used molecular sieves to improve been effective at increasing throughput in bottlenecked
the dryout time. We find that we can improve dryout time units without experiencing any problems in circulation.
by about two to three days using them.
At the Pascagoula refinery, we put a layer of activated PORTER:
alumina in front of the molecular sieve and believe this We have recently cried UOP R-134 and the early results
helps co reduce chloride attack on the molecular sieve. look fine.

D’AURIA: HEROS DERGREGORIAN (Giant Refining Company):

UOP has specified dryers on recycle gas screams in order Does this new high level of alumina have any effect on
to reduce dryout time on start-up. The numbers quoted aromatic production?
by others are similar to our experience.
We offer three different adsorbent products for this D’AURIA:
service: RDG-418, AW-500, and AW-300. The most The R-130 series catalyst has essentially the same yield
commonly used are RDG-418 and AW-500. The AW-500 pattern as R-30 series catalyst, so that at a given octane,

158 Light Oil Catalytic Processing

TOC/INDEX

the aromatic production is the same. For petrochemical a sulfate removal step conducted prior to the oxidation
operations, the higher activity of R- 130 catalysts can be step by circulating electrolytic hydrogen in the presence of
used to increase aromatic production by increasing the a high concentration of hydrochloric acid. Hydrogen sul-
operating severity. fide is formed and then neutralized with a caustic wash.
However, if all the iron sulfide scale in the heat exchangers
STEVE WILLIAMS (Star Enterprise): and heaters has not been removed during the carbon burn,
Has UOP seen any increased coking with the new sulfates will again be formed during the hotter oxidation
R-130 series catalysts? step and thus counteract the total effectiveness of the
sulfate removal step. We therefore recommend hydroblast-
D’AURIA: ing the feed exchangers before the oxidation step to re-
The R-130 series catalyst coke laydown rate is essen- move iron sulfide scale.
tially the same as that for the R-30 series. We have done
extensive pilot plant and commercial tests that have con- McLEAN:
firmed this. I agree with what has been said before. I would add that
the activity loss is due to the effect of sulfates on the
QUESTION 23. alumina base, which adversely affects the platinum redis-
Discuss the presence of iron on reformer catalyst and persion process. This is where the activity loss comes from.
the formation of iron sulfide scale in reformer heaters
and heat exchangers. What effect do these sulfur de- MENEGAZ:
posits have on sulfate formation during regeneration? I agree with the previous comments also. I would add
that we also see most of the iron on the first reformer bed.
CUNEO: In addition, we have also seen a lot of iron sulfide scale on
We have taken core samples from reformer reactor beds. the bottom of the stabilizer, suggesting we had water down
They typically exhibit iron contamination at the top with there at one time or another.
rapidly decreasing concentration as the samples are taken Secondary sources of scale can come from a direct
down into the bed. Very little iron has been found at reduction of carbon steel by hydrogen sulfide at tempera-
depths greater than 12 to 18 inches. tures below 400°F, by oxygen contamination in the naph-
While the iron on the catalyst could be a source of SO2 tha, and by corrosion products in the regeneration
during regeneration, a more likely source is deposits found procedure.
in the feed site of the feed effluent heat exchangers. The Oxygen contamination has been traced back to storage
SO2 is generated when oxygen passes over these deposits in unblanketed tanks, pipeline naphtha, or city water used
and then comes into the reactors and deposits these during upstream in the naphtha hydrotreater. One refiner strips
regeneration. The majority of the SO2 remains in the first his feed with nitrogen to remove the oxygen.
reactor. Regarding the regeneration procedure, we have no
Closely monitoring the sulfate removal allows material comment except that attempting regeneration above 0.5
balancing the sulfur movement in and out of the reactors. wt% sulfur will permanently damage the catalyst.

DEATON: SAYLES:
Iron sulfide scale is formed in reformer reactor feed heat I would like to echo the earlier comment, we have also
exchangers and reactor heaters during normal operations found iron sulfide in our stabilizer bottom, which fouled
due to trace quantities of sulfur not removed in upstream a fired reboiler heater. The fouling required cleaning at
hydrotreating units. Iron on reformer catalysts is usually about an 8-year cycle.
highest on the first reactor. Recent iron analysis ofcatalysts
at one of Clark’s reformers indicated iron contents of 0.36 QUESTION 24.
wt% in Number 1 reactor compared to 0.22 wt% in What are the mechanisms of platinum loss from re-
Number 2 reactor, 0.07 wt% in Number 3, and 0.04 wt% forming catalyst? How much platinum is typically lost
in Number 4 reactor. We recommend rotating catalysts during the life of the catalyst?
between Number 1 and Number 4 reactors every three
years, when the catalyst is unloaded for reactor internal LAUX:
inspection. Actual platinum loss from the reforming catalyst is hard
Iron sulfide scale burns to sulfur dioxide during both to determine because you are dealing with small concen-
the carbon burn and the oxidation step. Sulfur dioxide trations on large volumes of catalysts over several years of
reacts with the metals on the catalyst to form sulfate, which operation. The balances are not as accurate as your audi-
prevents metals dispersion and causes subsequent lack of tors would like to think. Generally, accounting errors are
activity. Sulfates can be removed from the catalyst during the cause of many platinum losses.

1994 NPRA Q&A Session on Refining and Petrochemical Technology 159

TOC/INDEX

The one true loss that does occur is the catalyst that falls ARMBRESTER:
to the ground each time you handle it or open equipment. We agree with several of the previous comments on the
This can actually be very significant. We try to lay plastic mechanisms for losing platinum, both in the CCR and
out to catch as much of the droppings as possible, and have the semiregenerative units. Our losses will typically aver-
picked up as much as a drum during one turnaround. age 2% to 3% over the life of the catalyst, but we will lose
An effective platinum loss does occur as well. If you more than that in the CCR units just due to attrition.
expose your CCR catalyst to high temperatures, over
1200°F, alpha alumina can be formed. This alpha alumina BOYCOTT:
can encapsulate the platinum, making it harder to recover. I agree with the comments on the mobilization of
We have developed some catalyst particles that are nearly platinum as platinum oxychloride during the regeneration
pure alpha alumina. Based on surface areas, over 65% of step. I am not aware of a mechanism by which this
the platinum was encapsulated. The platinum refiners mobilization can remove the platinum from the catalyst
reduce their recovery guarantees to about 98% when the structure itself. We have not been able to document a loss
alpha alumina content exceeds 10%. The guarantee drops of platinum from our catalysts, other than physical attri-
to 97% recovery when the alpha alumina exceeds 25%. tion and loss on recovery.
Laboratory tests have also shown that platinum can be
leached off the catalyst during very severe oxychlorina- D’AURIA:
tion. However, commercial units do not approach these Platinum losses are, for the most part, due to fines
conditions. generation in catalyst handling. Paying attention to cata-
lyst handling is the best way to minimize loss of platinum.
McLEAN: While platinum volatilization is theoretically possible,
We do not know the mechanisms for certain, but we it is extremely unlikely to occur at any reforming condi-
do believe that the losses occur during the regeneration tions, either during operation or regeneration. Our expe-
rience is that questions related to potential volatilization
and activation processes, in the presence of chlorides and
of platinum usually arise as disputes over assay differences
oxygen. We suspect platinum oxychlorides have a finite
with the catalyst reclaimer. We have not seen any signifi-
vapor pressure and therefore are liberated from the catalyst
cant differences reported by our customers for catalysts
in very minute concentrations. We do see evidence of
that we have reprocessed.
measurable ppm levels of both platinum and rhenium in
the circulating washwater used during regeneration and
BOB RODDEY (Roddey Engineering Services, Inc.):
activation.
Through the years refiners’ personnel have backed off
As a rule, less than 5% of the metal will be lost over the
from being involved in catalyst loading with the advent of
life of a semiregenerative type catalyst, and I would agree
companies that specialize in catalyst loading. As a result,
with what Mr. Laux said about the difficulty in balancing there is some sloppy catalyst loading occurring sometimes,
such small numbers. and the representatives of the refinery are not there to see
it. I have seen drums of platinum catalysts on the ground
SAYLES: after a loading. We will be able to help anybody by
Platinum loss is through two mechanisms: supervising the catalyst handler, and that is a good step in
1. Loss of catalyst fines is the major mechanism in minimizing platinum loss.
platinum loss. For semiregeneration units the loss
is from handling during fresh catalyst loading or M. DEAN EDGAR (Criterion Catalyst Company L.P.):
spent catalyst unloading. Our experience has been Occasionally we are asked to help refiners figure out
an overall platinum recovery based on fresh plati- what happens to the platinum between the time the fresh
num addition of about 98 wt%. For CCR units catalyst arrives and the time the spent catalyst is sent out
another loss occurs in addition to loading and for recovery. By far, the two biggest problems we find,
unloading losses; catalyst fines are produced via which the panel mentioned, are: bookkeeping, or keeping
attrition. Reported attrition losses are in the range track of the amount of removed and replaced catalyst when
of 0.1 wt% to 0.2 wt% fresh catalyst. the reactors are dumped and screening losses are replaced,
2. Platinum can be lost if extremely high tempera- and the physical attrition of the catalyst during loading that
tures occur during regeneration or during unit may result in losses downstream of the reactors.
upsets. Platinum can form platinum crystal growth Another item is support material. When the catalyst is
in the temperature range of 900°F to 950°F as a sent to the reclaimer, if you mix the support materials in
result of high sulfur at the start of the regeneration. with the catalyst, even in small amounts, this will act as a
A sulfur removal step is needed to restore the diluent and the percent platinum analyzed will be a lot
catalyst activity at the end of the regeneration. lower in some cases.

160 Light Oil Catalytic Processing

TOC/INDEX

One more item to consider is the analytical procedures ratio, however, the catalyst coking rate will increase by a
used by the catalyst vendor and reclaimer. We sometimes factor of 5 or 6.
see as much as a 10% variation in results. The changes usually required for such a revamp are:
switch to a more stable catalyst,
QUESTION 25. installation of a low pressure drop vertical combined
What are the pros and cons of converting a 400 psig feed exchanger,
semiregenerative reformer to 225 psig? Please dis- addition of a large fourth reactor and heater, and
cuss yield gains, relative coking rates, compressor add a net gas compressor.
and heater revamp and possible charge reduction due The case cited here is an extreme and may not be
to increased pressure drop. practical due to the high coking rates and limits on the
capability of the recycle compressor. In this instance, the
McLEAN: next alternative to consider is conversion to a CCR opera-
The advantage of this reduction in pressure will be a 4 tion. This can be accomplished easily at 225 psig with
vol% to 6 vol% increase in C5+ reformate yield. Hydrogen changes similar to the SR revamp case, except that a new
yield will also increase between 0.4 vol% and 0.6 wt%, reactor stack and regenerator are added, instead of a single
and hydrogen purity in the recycle gas will increase by reactor. This conversion would provide an additional 2
more than 10 vol%. LV% C5+ yield over the semiregeneration revamp (5 to 6
Disadvantages are shorter cycle time and a greater LV% gain total), and could cost as little as 45% of a new
endotherm across the reactors. The net heat of reaction CCR platforming unit. To date, 24 semiregenerative plat-
may increase by up to 40%. Increased furnace duty and formers have converted to CCR operation, 9 of these from
increased heat exchanger surface area may be required to a side-by-side configuration.
meet this extra heat requirement. A third alternative would be the addition of a last
At constant octane, space velocity, and hydrogen to reactor and regenerator, leaving the first three reactors in
hydrocarbon recycle ratio, the cycle length of the lower semiregeneration mode. This is sometimes referred to as a
pressure operation will only be 40% to 50% as long. If this hybrid revamp. We have found that this option has not
is unacceptable, then feed rate can be cut, reformate octane had a good payout because the required investment, while
reduced, catalyst loading increased (if possible), or hydro- lower than a CCR revamp, is significantly higher than a
gen to hydrocarbon ratio increased. semiregeneration revamp, while the yield benefit is similar.
Catalyst loading and throughput will be directly related
to the cycle length. That is, a 50% increase in loading will DEATON:
increase cycle length 50%. The hydrogen to hydrocarbon A similar question regarding mechanical details was
ratio will have to be doubled to maintain the same cycle answered during the 1993 NPRA Question and Answer
length. Finally, octane decrease of 10 to 15 numbers may Session. Two advantages of converting a 400 psi unit to a
be required to hold cycle length constant. Of course, 225 psi unit would be increased yield of about 3 vol%,
combinations of these variables could also be employed. and increased hydrogen production of over 250 scf/bbl.
Reduced pressure at constant feed rate will significantly Four disadvantages ofconverting the unit would be, (1)
increase the pressure drop. Reactor internals may have to the need to regenerate catalyst about twice as often, (2) the
be changed from downflow to radial configuration in order need to revamp the reactor feed effluent exchangers to
to conserve pressure drop. Also, new low pressure drop feed reduce system pressure drop, (3) the need to revamp the
effluent exchangers may also be required to conserve the recycle compressor to operate at lower suction pressure,
overall delta P with the lower pressure operation. and (4) the need to install a booster compressor to supply
produced hydrogen to consuming units. The cost of the
D’AURIA: revamp must be weighed against the value of the yield and
UOP has done many revamps of high pressure semire- hydrogen increases. Assuming average 1993 prices and
generative reformers to lower pressure operations. Every severity of 97 to 98 octane, we estimate the value of the
project has its own unique constraint, but some general yield alone to be twenty-five cents per barrel.
guidelines can be given.
In most cases, a refiner will want to retain his current LEE E. TURPIN (Honeywell Profimatics):
recycle compressor, and maintain feed rate. This can usu- Dropping reactor pressures has been evaluated on sev-
ally be accommodated by allowing the H2 /HC ratio to eral units using Profimatics’ kinetic reformer model REF-
drop as the pressure is decreased. For a decrease in pressure SIM. A few general observations are: (1) a greater C5+ yield
from 400 psig to 225 psig, a semiregenerative unit would increase is observed when dropping from 300 psig to 200
theoretically see an increase in C5+ yield in the range of 3 psig than from 400 psig to 300 psig, and (2) a greater C5+
to 4 LV%, depending on feedstock properties and operat- yield increase is gained when dropping the last reactor than
ing severity. With the decrease in pressure and H2/HC the first reactor. Usually, incremental capital spending to

1994 NPRA Q&A Session on Refining and Petrochemical Technology 161

TOC/INDEX

reduce pressure in the last reactor, preferentially over the in positions by simple repiping, often then allowing both
first and second reactors, yields the highest return on to fit the new revamp requirements. Also, remanifolding
capital investment. The use of a kinetic model allows a of the reactor section heaters to increase the number of
refiner to determine the economic impact of specific passes can relieve a pressure drop bottleneck caused by the
mechanical changes on a unit. With this knowledge, those pressure reduction. If needed, more tubes often can be fit
modifications that provide real economic gains, on a in the firebox at the same time the remanifolding is done.
discounted cash flow basis, can be identified. A charge reduction can often be avoided by adding
catalyst to the unit. The first reactor is scrapped, the
FRANCOIS REVEROY (Packinox, S.A.): subsequent reactors are moved up in position, and a new
One of the options that has not been brought up is that larger last reactor is installed.
you may need to find some additional heat duty at the In the last six months, we have completed projects both
charge heater, and also find some savings in pressure drop. in Australia and the Russian Federation for the revamp
One of the options which is currently looked at by refiners described above, including successful installations of a
is to go to a plate-like feed effluent exchanger. In most Packinox CFE.
cases, the increased heat recovery will eliminate com-
pletely the need to revamp the heater, and in certain cases
where you deal with the older generations of shell and tube Mechanical
exchangers, the savings in pressure drop can also be very
substantial. QUESTION 26.
Has anyone operating a semiregenerative reforming
JEAN-LUC NOCCA (IFP Enterprises, Inc.): unit ever experienced combustion in the heat ex-
IFP has developed two processes to revamp reformers changers downstream of the reactors during the coke
and take full advantage of the lower pressure. The first one burn or subsequent steps in the regeneration? What
is the addition of a continuous reactor and associated causes this? How can this be prevented?
regenerator in the reaction train. You can, for example, at
constant octane, decrease the pressure from 350 psi to 250 LAUX:
psi and increase the cycle. You can gain about 3.50 scf/bbl We have not experienced the problem described. One
in hydrogen, and 6% or 7% gain in yield. potential cause is residual hydrocarbons being swept off
Another option is to add the continuous reactor down- the catalyst and accumulating in low spots. To avoid this
stream of the separator, and to recontact the net hydrogen problem we take extra time to check all of the low spots
produced with the liquid from the separator. The advan- in the recycle loop for residual hydrocarbons and drain
tage of that is that you need no revamp to your existing these low spots while cooling down. We actually have a list
reaction system. With that you can get in the range of 450 of all these low spots in our procedures.
scf/bbl additional hydrogen and 8% additional liquid Our inerting procedure uses two depressuring steps to
yield while also increasing the cycle of the semiregenerative the flare header pressure, followed by repressuring to 100
reactors. The additional advantage of lowering the pres- psig with nitrogen. The third depressuring step goes to
sure is, of course, that you will make less benzene in your vacuum before the final repressuring. We then test the
reformer and that prefractionation will in most cases be nitrogen in the recycle loop for hydrocarbon content. If it
sufficient to meet benzene requirements. is over l%, we depressure and repressure again until the
analysis shows less than 1%.
KENNETH D. PETERS (UOP):
I would like to expand a little bit on the comments from McLEAN:
Mr. D’Auria from the panel and Mr. Reverdy of Packinox. This phenomenon can occur due to combustion of
The UOP Engineering Department has done numerous heavy ends and sulfur compounds. The material will start
revamps with a pressure reduction of this level. Currently, to burn only as sufficient oxygen breaks through from the
we have found that the replacement of the existing com- terminal reactor and temperatures are elevated to the proof
bined feed exchangers with a Packinox welded plate-type burn level.
heat exchanger works out well by reducing the reactor
circuit pressure drop and also reducing the load on both PORTER:
the charge heater and products condenser. Many years ago we experienced combustion in the heat
Typically, the old 400 psig reformer design required a exchanger downstream of the reactors during a start-up.
larger charge heater than first interheater, and dropping At that time, the reformer operated with a lot of H2S in
the pressure significantly increases both the total heat of the recycle gas so the exchangers were heavily fouled with
reaction and the first interheater load. With the addition iron sulfide. We believe the combustion was caused by the
of a Packinox CFE, these first two heaters can be switched ignition of the iron sulfide.

162 Light Oil Catalytic Processing

TOC/INDEX

CUNEO: in reformer recycle gas. In addition, it has two sensors that

We discussed the potential for combustion of iron can provide for continuous and automatic calibration.
sulfide which is generally on the feed side of the reactor Another feature is the analyzer has a rapid sensor drydown
feed to effluent exchangers. In addition to that, we have if it does become saturated by high gas moisture content.
occasionally found the deposits on the effluent side of the This feature allows for accurate and timely moisture moni-
exchanger which were identified as polynuclear aromatics. toring during times of rapid moisture change, such as
We have used a solvent prior to regeneration to remove start-up after regeneration. This analyzer has been proven
these. in a first commercial reforming unit, and there are four
more installations planned in the near future.
BOB RODDEY (Roddey Engineering Services, Inc.): It has been our practice to measure only the recycle gas
Twice during my career, we have had combustion begin moisture on a continuous basis. We feel that the recycle
to take place in the feed effluent exchanger during the gas moisture is the most pertinent variable in the reform-
regeneration oxidation step on a reformer . On both ing operation, and direct measurement of this variable is
occasions, it was sufficient to melt one of the tubes in the the most straightforward way to control the unit’s water
bundle and subsequently get recycle gas bypassing back to chloride balance.
the suction of the compressor. When we shut down and
went in, we found that there were iron sulfide scale and BOYCOTT:
coke in the exchanger. We had gone through the precau- We installed moisture analyzers on both the feed and
tions of making sure that there was no hydrocarbon in the recycle gas streams two years ago. The units have proven
recycle gas, of course, before we started. Nevertheless, it reliable. We have used them primarily for trend checking
was sufficient to start the burn and the heat liberated was of moisture levels, but as we are gaining experience with
enough to burn a hole in the exchanger tubes. The solution the systems we are starting to put more faith in the absolute
to the problem is not to let your heat transfer coefficient numbers.
get low before cleaning the feed effluent exchangers. The systems that we have are based on an aluminum
oxide probe. We have two probes for each installation.
DAVID N. WATERFIELD (BP Oil London): They have been giving us about 6-month run lengths
We had a problem on one of our units about three years between recalibration. While one probe is in service the
ago. It was a semiregenerative unit which had been re- other is sent out for recalibration.
vamped to have a vertical countercurrent exchanger. The The one weakness we have seen with the system is the
conclusion that we came to from our investigations was susceptibility to chloride attack during periods of high
that catalyst had migrated from the last reactor through a moisture level. For this reason we block in the recycle gas
damaged center pipe and lodged against the baffle plates probe during catalyst regenerations. Neglecting to do so
in the exchanger. This occurred around the point where during one regeneration caused us to destroy a probe.
the dew point was reached in the exchanger. We believe
that coke was built around that catalyst and it did not PORTER:
ignite during the carbon burn step, but it burned very We agree with Mr. D’Auria that the recycle gas is the
fiercely during the oxychlorination step and did signifi- best location to measure moisture content. We do not
cant damage to a number of tubes in the exchanger. This measure the feed on a continuous basis. We prefer Ametec
experience shows that it is important to make sure of the moisture analyzers on the recycle gas.
integrity of your center pipe and regularly clean your
exchangers, as some of the other answers have indicated. SAYLES:
Again, we concur with the previous comments and
QUESTION 27. would add that we measure the moisture only in the
Have there been any advances in moisture analyzers recycle gas. We have had an analyzer on the recycle gas the
for reforming units? Does anyone use water analyzers last seven years, with only minor maintenance. Ours is one
on both the liquid feed and the recycle gas? What has of the older designs, which means that during, or just
experience been with these systems? before regeneration and before start-up you have to make
sure that the analyzer is blocked in from the process
D’AURIA: streams. Otherwise, the high moisture content will destroy
UOP has recently commercialized a new type of recycle the cell.
gas moisture analyzer called the “Continuous Moisture
Measurement System”, which overcomes some of the TREESE:
problems experienced with some earlier analyzers in this I will add another vote for recycle gas stream analyzers.
service. This analyzer uses a silicon-based sensor that is We have been relatively happy with the Panametrics ana-
much more resistant to the corrosive agents that are found lyzer. They have made a number of improvements in their

1994 NPRA Q&A Session on Refining and Petrochemical Technology 163

TOC/INDEX

sensors and electronics over the years. They also perform feed/bottoms exchanger could be picking up too much
well at low ppm. heat, and the resultant lower vapor/liquid ratio might be
As a check of the on stream analyzer, we use a Shaw too low for proper water removal. An analysis of the entire
portable analyzer. Shaw uses a different sensor system, so stripper column operation is needed to determine if the
it acts as a confirming cross check. conditions are severe enough to produce a dry reforming
unit feed.
MOHAMMAD AL-SHAHRANI (Saudi Aramco):
I have two questions on moisture streams. What is the QUESTION 26.
best location for sampling the recycle gas, in the suction Our reformer feed sometimes contains up to 3 ppm
or in the discharge of the compressor? Secondly, is there organic nitrogen. We have managed to sustain this
another laboratory testing method? operation with higher chloride rates. Has anyone else
had this problem? What is the experience with high
CUNEO: organic nitrogen in the feed?
We typically sample recycle gas on the suction side of
the compressor and we do off-line analysis of feed mois- McLEAN:
ture content with Karl Fisher. I am aware of one reformer which has operated at 6
ppm feed nitrogen in a 300 psi operation. The biggest
HEROS DERGREGORIAN (Giant Refining Company): problem which was noted was increased compressor foul-
NHT stripper is where you should remove the water ing due to ammonium chloride deposition, which was
which is part of the naphtha hydrotreater sulfur removal discussed in detail in a previous question.
process. From where else would water be coming in the
reformer charge? How could you guarantee that the strip- PORTER:
per is operating properly besides maintaining the proper We do not run any of our reformers at that level of
reflux rate? nitrogen for sustained periods. We would immediately try
to find the source and stop the problem.
BOYCOTT:
We have noted two sources of water in our reformer ARMBRESTER:
feed; improper stripper operation, and cold charge from When high levels of basic nitrogen are present in the
tankage. We use the reflux to feed ratio to control stripper reformer feed, we would expect it to temporarily reduce
operation. As to cold charge, we are currently installing a the catalyst activity by neutralizing some of the acid sites.
line to route cold charge through the stripper. We are The primary effect of the lower catalyst activity would be
doing this to strip water from the stream prior to feeding a reduction in isomerization and/or hydrocracking teac-
it to the reformer. tions in the reformer. The catalyst activity can be main-
tained to some extent by increasing chloride addition, but
DEATON: as mentioned previously, you need to be concerned with
Our guideline for naphtha hydrotreater operation on ammonium chloride salt formation.
the stripper is that the reflux should be 25% of the unit
charge rate, and if your reflux is that rate, you are firing D’AURIA:
enough in the reboiler to remove water. Our specification is less than 0.5 weight ppm nitrogen
in the reformer feed. Nitrogen will act as a temporary acid
LAUX: function poison. Increasing chloride injection will help
One other source of water that we have seen is from marginally, but will also lead to greater catalyst deactiva-
dryer malfunction on our continuous unit. We carried tion and ammonium chloride salt formation. A 10,000
water back into the unit with the catalyst from the bbl/d reforming unit with only 0.5 ppm nitrogen in the
regenerator. feed will produce about 2,000 lb per year of ammonium
chloride, which is going to settle out somewhere in the
PORTER: cold end of the plant. Operation at higher nitrogen levels
When feeding cold feed from tankage, we prefer to feed will produce even greater quantities of salt.
it through the stabilizer to dry it out.
RONALD E. MARRELLI (Phillips Petroleum Company):
KENNETH D. PETERS (UOP): With respect to nitrogen, I know UOP mentioned 0.5
In terms of evaluating the naphtha hydrotreating unit ppm as their maximum specification. What about lower
stripper column for water removal, you really have to look levels of nitrogen? How low a nitrogen level can you have
at the vapor/liquid ratio in the bottom of the column. and still see noticeable effects of nitrogen poisoning on the
Even if you have a relatively high reflux ratio, the catalyst? And are there specific nitrogen compounds that

164 Light Oil Catalytic Processing

TOC/INDEX

will give you more problem in the feed, such as chemical- prevent unscheduled outages. A mechanical availability
treating amines or other corrosion inhibitors that may give analysis, taking into consideration the refinery mainte-
you feed poisoning of your reformer catalyst? nance history, should be used to determine the number of
machines.
D’AURIA: Comparing the options of reciprocating compressor
The 0.5 ppm specification is historical, set by the lower versus centrifugal compressor during a recent major capi-
limit of detection of the analytical method. You should try tal project resulted in the selection of the reciprocating
to operate with no detectable nitrogen in the feed. Any compressor. Although the reciprocating compressor op-
amount of nitrogen coming in is going to remove chloride tion held higher maintenance costs, the overall mainte-
from the catalyst and eventually wind up as a salt in the nance plus capital investment is much lower than for the
cold end of the plant. centrifugal compressor.
With respect to the types of nitrogen species, we have
not seen any real difference in the types. We assume that ARMBRESTER:
any organic nitrogen is going to break down and form Ashland operates two CCR reformers at our
ammonia over the platinum catalyst, and both strip chlo- Catlettsburg refinery, each of which is equipped with a
ride as well as lead to downstream plugging. centrifugal recycle gas compressor and reciprocating net
Our experience is that operating at less than 0.5 ppm gas compressors. In general, centrifugal compressors cost
nitrogen, with our normal chloride maintenance proce- less to maintain and have more reliability than their
dures, allows proper catalyst chloride balance and long run reciprocating counterparts.
lengths. Our experience shows that the maintenance cost for the
reciprocating compressors is more than twice as high as
QUESTION 29. the centrifugal machines, with the centrifugal compressors
We are considering the replacement of our reciprocat- having a mean time between failures that is about three
ing catalytic reformer recycle compressors with one times as long as the reciprocating compressors.
centrifugal recycle compressor. The maintenance In terms of operating cost, the reciprocating compres-
cost is quite high on the reciprocating machines and sors are inherently more efficient than centrifugal com-
extreme problems are experienced during catalyst pressors. When operating at design conditions the
regenerations. What is the difference in operating and reciprocal compressors are 10% to 15% more efficient.
maintenance cost of the reciprocating and centrifugal However, much of this efficiency advantage can be lost if
machines? the compressor is throttled for operation at reduced flow
rates, so the centrifugal compressors offer more opera-
SAYLES: tional flexibility.
We have reciprocating compressors on our reformers at For semiregenerative units, we have found that the
two locations and a centrifugal compressor at one location. reciprocating compressors require extensive valve work to
All three units are semiregenerative reformers allowing the remove salt deposits following regeneration, while the
opportunity for major maintenance during catalyst regen- centrifugal machines usually do not require any additional
eration. work related to the regeneration.
Our experience with the centrifugal recycle compressor
is that a major overhaul would be expected every 5 to 6 CUNEO:
years. Only one compressor is needed due to the excellent I agree with the previous comments. We believe the
on stream time. With the advent of improved rotating availability of reciprocating compressors can be worked
equipment monitoring techniques problems are detected into the 96% to 97% reliability range with good mainte-
before significant damage can accumulate. We believe the nance practices.
major mechanical overhaul cycle can be increased further.
We have not had to take an outage to conduct repairs on CHARLES S. McCOY (Consultant):
the centrifugal compressor between regeneration. Also, lubricated reciprocating compressors will put
For reciprocating compressors the anticipated major some lube oil into the system. It shows up as coke on the
mechanical overhaul cycle should be the same as the first reactor catalyst where you least want to build a high
centrifugal machines, 5 to 6 years. However, our experi- coke level. Nonlubricated compressors solve that problem,
ence has been that the reciprocating machines require but have a few of their own.
major overhauls every 3 to 4 years. If a single machine were
used, valve replacements about every three months and KENNETH D. PETERS (UOP):
other more routine problems could cause unscheduled Several years ago, we shifted our specification for both
outages for repairs. Normally two reciprocating machines the smaller reciprocating recycle gas compressors and for
are required to allow for routine maintenance and to the larger net gas machines away from the nonlubricated

1994 NPRA Q&A Session on Refining and Petrochemical Technology 165

TOC/INDEX

machines to synthetic-lubricated machines. We found the product separator drum and found the demister pad
these synthetic lubrication packages have significantly to be partially dislodged, allowing caustic carryover.
extended the valve life and given us a much higher reliabil-
ity with the reciprocating machines. I refer you to my CUNEO:
response to Question 19 on Page 83 of the 1987 NPRA We have had instances where the compressor was op-
Q&A Sessions book. erating very smoothly on shutdown and the decision was
made not to wash, as Mr. Deaton said, and then the
MAJED A. INTABI (Saudi Aramco): compressor vibrated violently. Those kinds of situations
In the case of a set of reciprocating compressors, the have been corrected in the past with water washing, so we
spare machines allow continuity of operations and on the recommend washing compressors on shutdown if any
run maintenance. On the other hand, when problems are capacity loss or vibration has been measured. We routinely
encountered with the centrifugal compressor, it will have wash after regeneration.
a greater impact on operations and can result in long
duration shutdowns. D’AURIA:
The decision on whether to wash the recycle compres-
QUESTION 30. sor during downtime is based on two factors. One, if you
During in situ catalyst regenerations, we have expe- are going to open the machine and expose it to air for some
rienced salt carryover into our centrifugal compres- maintenance, it should be washed. Two, it should be
sor. Some refiners wash the compressor with a weak washed if there is a probability of ammonium salt deposi-
basic solution at some point prior to start-up but after tion having occurred, either during the normal operation
regeneration. We have heard that some refiners do or during the regeneration. If either of these apply, then
not wash the compressor during this downtime. we would recommend following the washing procedure,
f/ease discuss the best timing for washing the com- as Mr. Deaton just outlined.
pressor and problems associated with this washing The typical reasons for not washing have been either
operation. Why would one not wash the compressor
time constraints, or existing conditions that did not war-
during this time?
rant washing; in other words, no history of ammonium
chloride deposition and no problem with nitrogen in the
DEATON:
feedstock. Also, some people do maintain a continuous
Foaming and carryover of circulating caustic solutions
on-line injection of antifoulant and feel that this will
during regeneration are best minimized by maintaining
remove the likelihood of compressor fouling.
pH between 7.5 and 8. Higher pH tends to cause foaming.
To avoid spikes in pH, makeup solution strength should
MENEGAZ:
be only 5 Baumt. Also, solids in solution cause foaming
We agree with Mr. D’Auria’s comments regarding the
and a good guideline is to flush enough to keep the specific
timing. An excellent description of the water wash pro-
gravity of the solution below 1.07.
cedure can be found in the 1988 NPRA Question &
The best times to wash compressor internals are either
Answer Session transcript. Most refiners to whom we
after the carbon burn or after the reduction step. From a
talked employ a similar procedure. Most do not do a
technical standpoint, the best time to wash is after the
reduction step because this would remove any salt deposits disassembly and cleaning unless planning to service the
incurred during the oxidation step. From a monetary compressor. Most reported excellent results, provided
standpoint, many refiners prefer to wash after the carbon care was taken in thoroughly removing the water from
burn if they are doing other maintenance between the the system. We would also suggest washing the inlet lines
carbon burn and oxidation. to the compressor as well.
For many years, Clark’s Blue Island refinery did not Many refiners also wash on-line with heavy naphtha or
wash their compressor internals for fear of damaging antifoulants to extend run lengths to three years. As with
labyrinth seals and for fear of uneven washing off of large any wash of a high speed machine, it should be started
deposits, which could throw the rotating element out of before you have appreciable amounts of buildup on the
balance. However, during a recent carbon burn, severe blades. Otherwise, you get nonuniform spalhng and vi-
vibration in the compressor caused us to stop in the middle bration problems.
of the carbon burn and wash the compressor internals. We
followed the procedure in the UOP Operating Manual, PORTER:
which consists of five washes with 2 wt% soda ash solu- We do not wash on a routine basis or every regenera-
tion. The compressor was started back up with very low tion, but we do it for a specific reason. Usually it is an
vibration and very high pumping rates. We also inspected increase in vibrations.

166 Light Oil Catalytic Processing

TOC/INDEX

SAYLES: D. Aromatics Extraction

We wash the centrifugal recycle compressor at every
regeneration, but by looking at the critical path we deter- QUESTION 31.
mine that our window of opportunity is while blinds are What are refiners’ experiences with respect to re-
being installed before regeneration. boiler corrosion in aromatic extraction p/ants using
Sulfolane as so/vent? What are the practices followed
CHARLES S. McCOY (Consultant): to prevent corrosion in the reboiler?
Two of you have alluded to the case for washing before
regeneration. I remember a disastrous wreck at one refin- BARLOW:
ery. They did not think they had a vibration problem, but It has been our experience that corrosion in the solvent
with the generation of moisture during the regeneration, circulation section of the sulfolane unit is generally asso-
deposits on the wheels were dislodged. Before the opera- ciated with organic acids formed by degradatiomoxida-
tors could stop the machine, serious damage was done. tion of the sulfolane solvent. Caustic and organic amines
such as morpholine have been used to neutralize these
HEROS DERGREGORIAN (Giant Refining Company): acids and control pH. However, some processors continue
How effective is the antifoulant on preventing com- to experience problems in the reboiler sections, even while
pressor fouling during operation and as well as during or they are using the neutralizers. The problem is generally
after regeneration? blamed on both continuing organic acid attack and ero-
sion due to buildup of insoluble salts in the recirculation
D’AURIA:
loop.
We would recommend use of a proper neutralizing
We have had a number of clients report success with
amine, an amine which forms a liquid salt, that you apply
the injection of antifoulant. They use a hydrotreated
in the overhead system. Also, consider evaluation of some
naphtha or a reformate stream injected at about two pints
of the newer higher temperature organic acid neutralizers.
per million standard cubic feet of recycle gas. Once begun,
this procedure must be maintained on a continuous basis.
CUNEO:
Intermittent injection can result in an imbalance of the
We operate three sulfolane extraction units that predate
rotor due to uneven cleaning of deposits.
the current UOP design. A number of years ago we
experienced significant reboiler fouling and corrosion
ANGELO FURFARO (UOP):
problems. It was attributed to buildup across the salts in
I want to add one point to Mr. D’Auria’s comments.
the system resulting from the ingress of oxygen and chlo-
You should always start the injection on a clean ma-
rides. We had to focus on source control of the oxygen
chine/rotor. The machine should be clean so that starting
chloride in order to eliminate the problem.
this injection does not cause existing deposits to come off,
which can damage the rotor blades. D’AURIA:
There are three mechanisms for corrosion at the Sulfo-
MOHAMMAD AL-SHAHRANI (Saudi Aramco): lane extraction unit reboilers: acidic attack, erosive attack,
Does anyone keep the compressor on-line during the and chloride stress corrosion cracking. Historically, these
reduction step? types of corrosion occur under different circumstances.
Acidic attack occurs when the Sulfolane solvent inven-
PORTER: tory is exposed to excessive heating (greater than 350°F)
No, we shut the compressor down. and/or oxygen; the Sulfolane solvent undergoes a thermal
decomposition, which is accelerated at higher tempera-
ANGELO FURFARO (UOP): tures, and dramatically increased in the presence of oxy-
During the reduction, UOP utilizes the compressor. gen. The key to preventing acidic attack is to prevent
Other reforming licensers may not. Others may perform overheating at reboilers and prevent oxygen contamina-
a hydrogen or hydrogen/nitrogen gas-through reduction. tion. However, if the solvent becomes acidic, it can be
Since UOP specifies use of the compressor, we would not corrected through proper operation of solvent regenera-
add any more hydrocarbon to the system while circulating tion facilities and effective neutralization with MEA.
that gas than normal, so UOP would not be using the (Typical desirable pH ranges are greater than 7 for circu-
compressor wash ar that point in time. lation water streams and 5 to 6 for circulating solvent;
solvent acid number should be less than 0.05 mg
LAUX: KOH/g.)
We do use our compressors to circulate hydrogen dur- Erosive attack originates from circulation of solids
ing the reduction step. (such as pipe scale, corrosion debris, polymeric byproducts

1994 NPRA Q&A Session on Refining and Petrochemical Technology 167

TOC/INDEX

of solvent decomposition and other foreign matter) which Some miscellaneous failures also occurred, attributed to
are not properly removed from the unit. The erosive attack erosion. A particulate filter system was added to eliminate
typically occurs at the extractive stripper column reboiler, erosion failure.
due to the higher velocities involved to limit the tube skin Finally, any refiner with ongoing corrosion problems
temperatures. The best countermeasures are effective in the sulfolane unit should be aware that air leaks into
chemical cleaning of the extraction section prior to initial the towers which operate under vacuum is the single
start-up and periodically at turnarounds (if necessary), and largest source of corrosion in a sulfolane unit. Large air
effective operation ofsolvent regeneration, so as to remove leaks are readily obvious because the solvent will turn red
the solids from the circulating solvent. UOP is studying due to iron in the solution. Small air leaks are best
alternate metallurgies to improve erosion resistance in this detected by regularly testing the sulfolane solvent for acid
service. neutralization number, expressed as mg of KOH per gram
Chloride stress corrosion cracking is caused by chloride of solution.
contamination in the feed. UOP recommends that the
feed contain less than 0.1 ppmw total chloride in order to ARMBRESTER:
avoid this type of corrosion. For units which have in- We also use MEA addition to maintain a low acid
creased likelihood for chloride contamination, UOP has number and the guideline that we use is 0.05 mg KOH
begun to recommend use of Duplex 2205 metallurgy per gram of solution, maximum.
(22% Cr, 5% Ni, 3% MO, annealed such that it is equal
parts austenitic and ferritic steel). QUESTION 32.
Are there any activity and selectivity comparisons of
DEATON: the commercially available clays to minimize unde-
During five years associated with a sulfolane unit, I did sirable reactions such as disproportion tran-
not experience corrosion in any of the reboilers in contact salkylafion? What types of clays and operating
with sulfolane. All of these reboilers were carbon steel and conditions reduce these undesirable reactions in
they were either stab-in reboilers or kettle reboilers. All of paraxylene, cumene and BTX processes?
them used 450 psi steam and the steam was on the tube
side. The sulfolane solvent in this unit was maintained in
ARMBRESTER:
good condition by operating the solvent regenerator con-
At our Catlettsburg petrochemical complex, we use
tinuously and with the addition of MEA.
Engelhard’s F-24 clay in the clay treaters for both the
At another refinery, corrosion was a problem in both
cumene unit and the sulfolane BTX unit. In the sulfolane
the stripper reboiler and the solvent regenerator reboiler.
unit, the typical start of run temperature is 310°F and the
The predominant mode of failure was process-side corro-
inlet temperature is increased as the clay ages. The typical
sion near the steam inlet of the vertical thermosiphon
end of run inlet temperature is 390°F. It is critical that the
reboilers. All of these reboilers used 250 psi steam, and
operating pressure be kept high enough to ensure that the
steam was on the shell side. Corrosion was eliminated by
BTX stream remains in the liquid phase during this
replacing the carbon steel bundles with 304 stainless steel
and by operating the solvent regenerator continuously. operation.
Poor clay life has been a problem in these treaters as a
result of low space velocities and relatively high olefin
content, but selectivity has not been an issue.
In the cumene unit, benzene production in the clay
treater is a problem as a result of disproportionation of
cumene to benzene and di-isopropyl benzene. These
treaters are run at much lower temperatures than the BTX
clay treaters, with typical start of run inlet temperature of
80oF, because we have seen the benzene production in-
crease at higher temperatures or reduced liquid hourly
space velocity.
We have tested Engelhard’s F-44 and F-54 clays in this
treater and we were able to operate at higher temperatures,
but the clay life appeared to be shorter. At the 3.5 liquid
hourly space velocity we run, the F-24 appears to give us
the best service. If we were consistently running below 3.0
liquid hourly space velocity, one of the more selective clays
would be reconsidered.

168 Light Oil Catalytic Processing

TOC/INDEX

D’AURIA: with different techniques, and I would encourage anyone

We have not seen any quantitative comparisons of the who has specific problems related to this to contact Catalyst
various clays. For benzene or cumene treating services, Technology directly.
where generally there is a lower olefin content, the impact
of yield loss is not significant. Therefore, most operators QUESTION 34.
will select the highest activity clays for these services as they In a chamber type reactor, high pressure drop and bed
typically give the longest run lengths. channeling are problems that shorten catalyst run
For xylene service, however, where olefins can be higher, length. What mechanical and/or process steps can be
the higher activity clays tend to have greater losses of taken to reduce these problems?
product due to transalkylation, unless the clay tempera-
ture can be reduced. Therefore, in these services we tend DEATON:
to recommend lower activity clays such as Sud Chemie’s Catalyst manufacturers have always stressed the impor-
Tonsil CO, in order to minimize ring loss. tance of feed moisture control. Other factors to consider
are (a) catalyst bed inlet temperature philosophy, (b) novel
catalyst loading ideas, (c) quench nozzle design, and (d)
E. Polymerization thermowell design.
The normal practice with catalyst bed inlet tempera-
QUESTION 33. tures has been to stagger the inlet temperatures in ascend-
What techniques have been used other than ing order. However, I understand that some refiners are
hydroblasting for removing catalyst from tubular now experimenting with running a constant inlet tem-
reactors? perature profile, with hotter inlet temperatures on the first
three beds, in an effort to decrease the heat of reaction in
D’AURIA: the last two beds where higher temperatures cause more
The other catalyst unloading technique that has been polymer formation.
successfully used is dry drilling of the tubular reactors. In the area of novel catalyst loading ideas, one catalyst
Government environmental regulations are making it manufacturer offers blends of catalysts and charcoal. Two
more and more difficult to dispose of the low pH, high super sacks of these blends are put at the top of each
phosphorus water produced in the conventional hy- catalyst bed. This provides an inert component of similar
droblasting. density and size in the top of each bed to prevent high
Dry drilling, coupled with, a well designed catalyst reaction rates, thus preventing excess polymer formation
conveying and separation system, can provide a very clean in the top of the bed.
and efficient operation and the catalyst can then easily be Quench nozzle design is important in order to provide
collected in super sacks and sent to disposal. As an exam- uniform radial temperature profile even during unit turn-
ple, one refiner who has a tubular reactor with 160 tubes downs. I had good experience redesigning for lower charge
of 2-inch ID, 30 feet long, required only eight to nine rates using Bete Fog nozzles.
hours from the beginning of the reactor unheading to the As far as thermowell designs, the optimum setup is to
completing of the reactor closure after catalyst loading have radial temperature readings across the top and bot-
when using this dry drilling technique. Drilling times are tom of each bed to help in adjusting quench rates to
approximately 3 minutes per tube for the drilling itself, minimize radial temperature differences.
and about 1.5 minutes per tube for wire brushing.
The only reported problem we have heard with dry D’AURIA:
drilling is drill bit sticking. This can result if there are some We have seen refiners try both mechanical and process
soft, gummy deposits encountered in the tubes. Typically, steps. As Mr. Deaton just mentioned, mixing an inert
there are three to five tubes that have these deposits, and material such as charcoal in with the catalyst has been
these are hydroblasted out after the rest of the catalyst has successfully used to minimize pressure drop buildup. In
been unloaded. addition, refiners have sometimes short loaded the bottom
bed of catalyst in order to give them less pressure drop
SAYLES: buildup and prolong the catalyst life.
I agree with Mr. D’Auria. On the process side, the primary knob you have to turn
is proper catalyst hydration, and this is really the most
DAVID WOOD (CRI International, Inc.): important parameter in prolonging catalyst life.
I agree with the panel comments on dry drilling for Other process variables that can prolong run length are
removing catalyst from tubular reactors. I would add that maintaining the proper severity ofoperation by maintain-
our affiliated company, Catalyst Technology, has consider- ing the combined feed olefin concentrations within the
able experience for removing catalyst from tubular reactors recommended levels, controlling catalyst temperature

1994 NPRA Q&A Session on Refining and Petrochemical Technology 169

TOC/INDEX

profile by propet use of interbed quench, and minimizing the reactor inlet, and thus help dissipate the heat of
the reactant residence time over the catalysts. We normally reaction.
recommend a 0.9 minimum space velocity.
ARMBRESTER:
MENEGAZ: Theoretically, any strong acid catalyst with the appro-
I agree with Mr. D’Auria’s comments. Limiting the priate acid strength and distribution will be effective for
temperature rise per bed to 35°F has worked well for one polymerization or aromatic alkylation. Several systems,
refiner. If the olefin content of feed is above 30%, using including zeolites, have been identified for this applica-
recycle diluent should be considered. tion. However, we are not aware of any significant use of
Regarding the temperatures, multiple thermocouples these materials, probably due to the relatively low cost of
are recommended since channeling makes individual tem- purchasing and disposing of the solid phosphoric acid
perature readings unreliable. catalysts, and their years of effective performance.
Catalyst breakage during loading is generally not sub- All the alternative catalysts of which we are aware are
stantial enough to cause problems. In one odd case, recip- significantly mote expensive. The major driving force to
rocating pumps caused high pressure drop by introducing change may be disposal costs. If an in situ regenerable
air into the process, forming coke. catalyst were available, future landfill liabilities may pro-
vide incentive for their use even at a high per barrel
SAYLES: purchased catalyst cost.
I would like to add that we have seen situations where
pressure drop has developed in one reactor and not in MENEGAZ:
another. We have skewed the feed rates to the individual Except for Dimersol units, almost all polymerization
reactors and tried to balance the pressure drop. gasoline units today use solid phosphoric acid. There was
We typically get 2,500 hours per catalyst load and a liquid phosphoric acid process developed by Chevron,
attribute the run length mainly to processing only C3 which involves spraying the acid periodically on quartz
olefins and keeping the heavier materials out of the unit. particles, but I do not believe anyone is using that today.
Also, Chevron had a bulk polymerization process in-
QUESTION 35. volving phosphoric acid. I do not believe that went past
Are there any other catalysts beside solid phosphoric the pilot plant stage.
acid catalysts that can be used commercially for poly- Several suppliers are looking at zeolite-based catalysts.
merization? Cumene synthesis is successfully being done that way. In
the past, alumina chloride and solid copper pyrophos-
DEATON: phate have been tried.
I am aware of two technologies licensed by IFP. These
are the Dimersol process and the Polynaphtha process. ROBERT ALLSMILLER (United Catalysts, Inc.):
The Dimersol process utilizes a liquid soluble nickel- To follow up on Mr. Armbrester’s comment about
based catalyst injected into the feed at a rate of a few zeolite polymerization, our parent company, Sud-Chemie,
hundred ppm. The reactor is a simple residence time has developed in conjunction with a South African Com-
vessel operating at 120°F and 300 psi pressure. Heat of pany, a process using zeolites for the conversion of olefins
reaction is controlled with a cooled pumparound system. to diesel. This process has now been operated commer-
The reactor effluent contains catalyst and the reaction in cially for well over a year and is available for licensing.
the effluent is stopped with the injection of caustic. The The comment about the cost of the catalyst is abso-
caustic wash is usually neutralized or used in FCC flue lutely correct. Zeolites do cost quite a bit more than solid
gas scrubbers. Propane product meets HD-5 specifica- phosphoric acid catalyst but zeolite catalysts last longer.
tions.
The polynaphtha process utilizes an acidic silica alu- MARK DORAN (ICI Katalco):
mina catalyst in a fixed bed reactor. There is no chemical One of our customers using a PURASPEC™ sulfur
injection used to maintain catalyst activity such as in a guard process on the liquid feed to their C5/C6 isomeriza-
reformer. The process is relatively low pressure at about tion unit has seen a significant improvement in perform-
150 psi pressure and reactor temperatures are less than ance of their isomerization unit. Installation of the guard
those in phosphoric acid catalyzed reactors. Multiple bed has increased the life of the isomerization catalyst by
reactors in series are used and the heat of reaction is 800%. The increased protection to sulfur ingress offered
controlled by cooling between reactors. LPG recovered in by the PURASPEC™ unit has also allowed more flexible
the fractionation column is recycled to the reactor to raise operation of the upstream hydrotreater leading to its cycle
conversion and to decrease the concentration of olefin in lengths increasing more than 100%.

170 Light Oil Catalytic Processing

TOC/INDEX

F. lsomerization a permanent poison for this catalyst. It is considered a

more severe poison than water because part of the deacti-
QUESTION 36. vation mechanism will release more water, which further
Have refiners processed defluorinated alkylation bu- deactivates the Butamer catalyst.
tane through a butane isomerization unit? if so what HF itself will be picked up by the feed dryers but will
has been the commercial experience? Has catalyst deactivate the sieves and lead to water breakthrough.
life been affected?
QUESTION 37.
PORTER: What has been the experience with temperature run-
Yes, we have processed defluorinated alkylation butane. away in a butane isomerization reactor during start-
Fluoride, of course, is a serious poison for isomerization up? How do you prevent these runaways?
catalysts, and so you need to closely monitor the operation
of the defluorinators. We consider 1 ppm fluoride the
PORTER:
maximum specification. Also, any residual HF in the
Most of our butane isomerization experience has been
isomerization feed will react with the molecular sieves, and
with recycle hydrogen units. There have been no problems
require them to be recharged more frequently. If you are
not careful, you can have water breakthrough, which will with temperature excursions. However, one of our once
also poison the isomerization catalyst. through units did have a runaway situation on restart after
One side note is that we have had catalyst poisoning a short shutdown.
from fluorides in purchased butane that was received in It is believed the hydrocarbon was not adequately
rail cars. purged from the reactor on shutdown due to the low flow
rate of hydrogen during the sweep step and the hydrocar-
LAUX: bons began to crack during the outage. When the hydro-
I agree with Mr. Porter’s comments. An actual experi- gen flow was reestablished with the catalyst still hot, the
ence we had involved not changing out the KOH treater hydrocarbons were promptly saturated, releasing heat.
downstream of the alumina treaters quickly enough. We For this specific situation, the heat input to the process
took four days to change it out instead of one. When you should have been stopped and the catalyst cooled below
are pushing your alumina lives to the end of the run, that 250°F using feed or recycled isomerate. Recycled isomerate
will allow enough fluoride to get to the molecular sieves. is the best choice, since it is more likely to be non-reactive.
This actually destroys the sieve, as Mr. Porter mentioned.
We deactivated the first reactor catalyst in one month after D’AURIA:
poisoning the sieve in the isomerization unit. Temperature excursions have occurred on several Bu-
tamer units. The cause of these excursions has either been
BOYCOTT: long hydrocarbon residence time in the reactor at high
The feed stream to our butamer is the normal side draw reactor temperatures, or, in a couple of cases, mistakenly
off of the alkylation unit isostripper. We operate two charging feed with high olefin content to the unit. To
defluorination alumina beds in series on this stream. The avoid excessive residence time at high temperatures, we
defluorinators have provided protection against HF attack have revised our operating procedures so that the unit is
on both the mole sieve driers and the isomerization cata- brought to full feed rate before raising temperature to
lyst. We are getting six to nine months operation on the
reaction conditions.
lead bed.
QUESTION 38.
D’AURIA:
What methods are being employed to reduce dryout
It is common practice to treat the normal butane stream
from an HF alkylation unit prior to feeding it to a butane time in the reactor and fractionator systems at start-up
isomerization unit, such as Butamer. We specify the two in a butane isomerization unit?
bed system that Mr. Boycott just mentioned. We use a
special high purity grade of alumina, and operate at ele- D’AURIA:
vated temperatures. Typically, this is then followed by a One of the time-consuming steps for the start-up of the
KOH treater to guard against any fluoride breakthrough Butamer unit is the removal of iron oxide from the equip-
and also to reduce the moisture content of the butane ment. Many of our customers have done chemical clean-
stream. ing of the reactor section as well as the stabilizer section
Organic fluorides, if they do get through this pretreat prior to the precommissioning phase. Once the iron oxide
system, will not be removed by the Butamer feed dryers is removed by this chemical cleaning, the drydown proce-
but will convert to HF over Butamer catalyst. The HF is dure proceeds very quickly.

1994 NPRA Q&A Session on Refining and Petrochemical Technology 171

TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

A. Alkylation-Sulfuric Acid at the 85% acid level, you are walking on a knife edge and
you can experience severe problems during an upset.
Feed
FRONDORF:
QUESTION 1. We had an opportunity about a year ago to brush up
Are there technologies available to remove selec- on our acid runaway operating experience. We experi-
tive/y or to hydrogenate cyclopentenes? enced a runaway resulting from loss of spent acid flow
from the settler. It appeared to take about 24 hours before
ROSS: the runaway actually started. The onset was detected, as
When treating petrochemical streams from an ethylene Mr. Johns mentioned, by a drop of approximately 30°F,
plant, for example, hydrogenation is in fact a means of from 55°F down to as low as 25°F, in contactor tempera-
removing the high levels of cyclopentenes. However, ole- ture. This occurred over a period ofapproximately 2 hours.
fins are saturated. For alkylation feed preparation, the Our acid strength was probably on the order of 92% when
preferred method of controlling cyclopentenes is through
the incident began, and we measured a strength of 40%
fractionation in the depentanizer, because the amount of
afterwards.
cyclopentene is rather low in the FCC C5 stream and
fractionation can reduce the content in the olefin stream
to less than a percent or so. Of course, the acid consump- Process
tion in the alkylation unit will increase. However, selective
hydrogenation of the diolefins is clearly a good idea as the QUESTION 3.
first step, but attempts to hydrogenate the cyclopentenes What is encountered when the fresh acid wash is
will likely also saturate a significant amount of the amyle- bypassed? When do these phenomena occur? Has
nes and is not currently practiced as far as I know. anyone seen a rag layer develop in the acid wash
drum? What causes it to be formed, and, it not re-
JOHNS: moved, how is operation affected?
For FCC gasoline, you would normally see the cy-
clopentenes left in the gasoline rather than as alkylation ABRAHAMS:
feed. Texaco is developing a process called Isotex for When the fresh acid wash is bypassed, alkyl sulfates end
pentene isomerization and this will, of course, increase up in the alkaline wash drum. This causes the pH to drop
TAME production. and the caustic consumption to increase. We have not
detected a rag layer in the acid wash drum. On one of our
PATRICK HOH (HRI, Inc.): alkylation units, we do not have a fresh acid wash. On the
IFP has a licensed process to hydrogenate selectively others we do.
more than 97 wt% cyclopentenes to cyclopentane. Cy-
clopentene separation can be carried out in a fractionator.
VICENTE A. CITARELLA (Exxon Research & Engineering):
The acid wash is recommended to remove neutral esters
QUESTION 2.
that may be present in the reactor product. These esters
How much drop in contactor temperature has been
pass through the caustic wash system and decompose to
seen during an acid runaway? About how long does it
take to go into a runaway, assuming a fixed acid sulfuric acid and SO 2 in the deisobutanizer. The sulfuric
depletion rate, i.e., starting at 85 wt%? acid cokes on the hotter tubes of the reboiler causing
fouling, and the SO 2 rises in the tower and causes corro-
JOHNS: sion in the wet overhead system. We have not experienced
The acid runaway reaction has a lower heat of reaction. these symptoms with the Exxon Research and Engineering
And therefore, the affected contactor will drop below the stirred, autorefrigerated reactor. We believe that this is due
other contactors in temperature and then carry on to the to the conditions that we use for our reactor designs, such
others, etc. If left unchecked, the temperature can reach as as low space velocity. We do not provide an acid wash
low as 30°F. If you are already operating in this mode and system in our design.

124 Light Oil Catalytic Processing

TOC/INDEX

QUESTION 4. have internal regeneration facilities. The improvements

Are there any new methods for on-line monitoring unit also increase isobutane use, so the benefits are lower if you
acid concentration? Has anyone successfully used a are isobutane limited.
Coriolis effect mass flow meter to correlate with unit
acid concentration? EMANUEL:
We have also introduced the co-catalyst at our Port
FRONDORF: Arthur refinery within the last 6 months. While using the
We continue to experiment with Coriolis meters in co-catalyst, we have seen that acid consumption has de-
trials sponsored by the manufacturers. Our initial test creased by approximately 10%, and there appears to be a
failed due to an extremely short life on the sensor tube. very slight increase in our alkylate yields. But we have not
We attributed this to high velocities in the tube, which been able to really quantify the yields or octane benefits at
resulted in erosion and early failure. Both stainless and this time.
hasteloy have failed. We are just starting another test with
a much larger tube, operating at a lower velocity. We have FRONDORF:
been able to derive a loose correlation between concentra- We have also tested the Betz product over a period of
tion imputed from density and acid concentration by time. I can give you some vendor numbers and then where
titration. The correlation is perhaps not suitable for precise we fit in. Yield improvements are reported anywhere from
control, but certainly capable of indicating trends. 3% to 7%, acid consumption improvements anywhere
from 10% to 27%, and octane improvements anywhere
JOHNS: from 0.3 to 0.7 numbers. For the trials that we have run
We have not yet used a Coriolis meter, but we are over a period of time, our results have fallen within those
looking at it. Two of the plants have STRATCO on-line ranges. The system is set up to model your unit using the
acid monitors. One plant likes them and the other does neural network type of modeling effort, so there is a lot of
not. There is another new technique that we are consider- data that needs to be collected. We have worked with Betz
ing trying in the near future which may provide better
over the past 2 years and have run two 60-day trials in that
accuracy.
period of time. Even with all that data and the modeling
effort, it is still difficult to measure improvements due to
JODY OVERSHINER (Micro Motion, Inc.):
the scatter and the changes on the unit. It would appear to
We are presently using Micro Motion meters in six
us that the benefits will be most important to those units
refineries for monitoring acid concentration in the alkyla-
that are run what I would call severely, meaning outside of
tion units. With the Micro Motion meter we are measur-
the new design recommendations from STRATCO, and
ing density and making a correlation between density and
where you will have more opportunities for improvements.
concentration.
There are some requirements to consider, such as ve-
QUESTION 6.
locity and the percentage of hydrocarbons present in the
streams. But it has been Micro Motion’s experience that What benefits have been observed by operating in
once these are taken into account, the meter does trend “Emulsion Recycle” mode? Has anyone been able to
and tracks laboratory data very well. quantify these benefits?

QUESTION 5. JOHNS:
Alkylation unit co-catalysts are being used more fre- This emulsion recycle mode has been a way of life for
quently. What yield, octane, and acid consumption the Texaco sulfuric units for over 30 years. We obviously
benefits have been seen? believe in the benefits. At this point, I cannot tell you any
delta benefits since we have operated this way for so long.
ABRAHAMS: But we believe that they are real and that is the way we
One of our plants is currently evaluating Betz’s ALKAT- operate.
XLco-catalyst. Preliminary results of actual use on the unit
indicate the following: yield has improved 3% to 4%, acid ABRAHAMS:
consumption has been reduced by 19%, rhe octane num- We have not quantified the benefits ofemulsion recycle
ber has increased 0.7 numbers, and the T90/95 and end at our one site that was started up originally in this
point have decreased some. Other than the acid consump- operation. STRATCO does report the benefits of lower
tion, the results require the use of modeling techniques to end point and higher alkylate octane, since the lower acid
quantify the benefits. For this reason, we have been ac- residence time decreases polymerization. We are preparing
tively pursuing trials at our two sites that incur outside to modify another of our alkylation units for emulsion
regeneration costs and have been less energetic where we recycle by lowering and relocating baffles in the acid

1995 NPRA Q&A Session on Refining and Petrochemical Technology 125

TOC/INDEX

settlers. We should be able to get some comparisons when lower, to about 85% to 86%. All of these make the
that is complete. equipment sizes larger.
Pentane effects — When alkylating amylene, these
EMANUEL: pentanes coming in with the feed will add extra load
Our Port Arthur refinery incorporated the emulsion to the alkylate debutanizer since the target split is
recycle approximately 1 year ago. Upon review of the data between the normal butane and isopentane, and also
over this time period, we have been able to show that our because of iC5 will have a very slight tendency to
acid consumption decreased by approximately 5% to 10%. compete with iC4 for alkylation reaction. To over-
come this, the iso-to-olefin ratio may need to be
GREGORY MULLINS (Marathon Oil Company): increased a small degree.
We went to emulsion recycle in November and we have Alkylate product — The alkylate octane from the
documented in excess of a 20% increase in throughput. mixed olefin feed will be lower than the alkylate from
That is one of the benefits. You also need to consider that Cd= feed only case. Also, because there are pentanes
the unit has less acid in it and it is more susceptible to in the alkylate product, the vapor pressure will be
other upsets such as acid runaways. higher. In our design we saw that with amylene in
the feed, alkylate vapor pressure is about 2 psi (in
QUESTION 7. RVP) higher than straight butene alkylate. To over-
What are the implications of commercially alkylating
come the vapor pressure problem, an alkylate depen-
tanizer tower may be needed.
C5=s and C3=s (with C4=s present also)? Are there any
unusual, adverse, or favorable interactions?
MORGAN:
Mr. Shen has mentioned most of the comments I would
SHEN:
have made. I will add, however, that propylene currently
This question talks about the interactions between the
has a commercial value much greater than the alkylate that
propylene and the amylene in alkylation. The main inter-
could be manufactured from the propylene. This makes it
action is a result of the fact that propylene alkylation
more favorable to fractionate and sell that product. Also,
requires a higher acid strength, while amylene alkylation
the amylenes can be blended to a very nice octane value
prefers the lower acid strength. If you try to combine them, on their own in the gasoline pool, unless you are faced with
they will have octane and acid consumption penalties, but reformulated gasoline specifications.
this would require separate olefin storage, feed cooling,
water removal, iC4 recycle control, and separated reactors. ABRAHAMS:
This could be a significant amount of extra capital and We have had similar results. We did fairly extensive tests
may not be economical. Currently we are designing a at one of our plants on a feedstock that was 20% amylene
sulfuric acid unit; it reacts propylene, butenes, and and 80% butylene feed. We saw octane drop 2 numbers,
amylene. It is roughly a 36/39/25 split. There are a few end point go up 15°F to 20°F, and acid consumption
issues that caught our attention. increase by 60% for the incremental feed. The impurities
Upstream fractionation — The upstream depentan were a problem for us. We got a larger, hard-to-break rag
izer needs to provide a sharp fractionation. Any large layer in the acid settler, probably from higher dienes and
amount of cyclic compounds in the back end of the foaming problems in the alkaline water wash, which was
alkylation feed will create the problems of polymeri- attributed to aromatics. We also experienced a decrease in
zation and high acid consumption. The Cd/C5 mix the isobutane recycle purity, which Mr. Shen pointed out.
stream needs to be treated for sulfur/mercaptan
removal and through a selective hydrogenation step DEADY:
to saturate dienes. We know of one refinery that operated two alkylation
Propane effects — When alkylating propylene, pro- units. One was an HF unit, and the other was a sulfuric
pane that comes in with the feed charge may con- acid unit. They both ran the units base loaded with
centrate in a number of process areas. In the propylene and butylene and flexed them with the amyle-
compressor loop, higher propane concentration will nes. When they did this, the alkylate quality remainedvery
increase the compressor discharge pressure. It also good. In fact, they thought that the RON and MON
lowers the purity of recycle iC4 from the refrigera- values actually increased slightly. However, they did see
tion unit. Another place is the DIB tower overhead. increased acid consumption at the sulfuric acid unit,
Higher propane concentration there will increase probably due to the presence of CG plus contamination of
the tower operating pressure, and it will lower the the amylene stream. Fortunately, the refinery had spent
alpha ratio and the fractionation efficiency. The acid regeneration capability and they were able to handle
recycle iC4 purity from the DIB tower will also be this increase.

126 Light Oil Catalytic Processing

TOC/INDEX

QUESTION 8. should partition to the water phase in the overhead accu-

With oxygenate carryover to the alkylation unit, we mulator to prevent them from recycling back to the reactor
believe that the water-content of the spent acid is section. Reboiler fouling in the alkylation unit is typically
higher. What analytical methods are available for the result of corrosion problems in the unit. Corrosion
determining the water content? What are typical ana- products that resulted from acids and esters in the frac-
lytical results? Have you noted any impact on unit tionation feed migrate down the column. Corrosion prod-
operation other than increased acid consumption? ucts can also come in with a fractionator feed from
upstream equipment. The corrosion products become
ABRAHAMS: bound to the reboiler with polymer and ester decomposi-
At one of our plants, we tested our spent acid for water tion products. While most refiners have attempted a rou-
content a few times using a modified Carl Fisher method. tine water washing of the reboiler on-line, this tends to be
These analyses did not show excessive oxygenate carryover. a short-term solution, and must be continued until a
In the spent acid samples, we have measured between 2.5 reboiler can no longer perform as required.
wt% and 5.3 wt% water. My understanding is that
STRATCO can provide a procedure for testing this if you POTSCAVAGE:
are interested. Instead of giving any more of a detailed explanation,
for those of you who would like more information on the
QUESTION 9. causes and the best solutions of these problems, I refer you
What has been the industry experience with corrosion to the National Association of Corrosion Engineers’ Ref-
and fouling of the depropanizer, debutanizer, or O/B COR database.
towers? Could you address cause and effect?
SHEN:
MORGAN: The only thing I have to add is that when you push the
Corrosion in the deisobutanizer and debutanizer on the alkylation unit over 110% of capacity, and the mixing
alkylation unit is becoming a more prevalent problem for device upstream of the alkaline water wash makes a very
refineries. Alkylation units are typically running above small droplet, the settling drum may not have enough
110% of the design capacity to meet the refiner’s need for residence time for a good separation. In that case, you
alkylate in the gasoline pool. The increased loading on the probably have to consider redesigning the in-line mixing
reactor section creates more side reaction products: acid device to give it enough contact, but not a verystrongshear
and neutral esters and acid carryover from low settling force to create the small droplets.
times.
Acid and esters are the cause of corrosion in the frac- ABRAHAMS:
tionators. Poor mixing and lower residence times in the We had fouling from sodium salts in our depropanizer
reactor effluent treating section allow more of these com- when the caustic drum strength got too high. We fixed
ponents to enter the fractionation section. The result is that by putting in a continuous caustic injection system
increased corrosion activity. to keep the strength between 3% and 5%.
Corrosion problems in these fractionators can be effec-
tively controlled through the use of chemical neutralizers. FRONDORF:
Neutralizers can be applied to the feed or the reflux of these I do not have anything to add on the main process
columns. The neutralizer used in the deisobutanizer must corrosion, but I wanted to point out that at our Lake
be nonvolatile at the overhead conditions of the column Charles, Louisiana Refinery we bring in a saturated C3/C4
so that it does not enter the reaction section with the stream from outside the alkylation unit for further frac-
isobutane recycle. tionation. We bring that stream straight to our depropan-
Nonvolatile neutralizers will also protect the reboilers izer tower, which is operated in the dry mode. We need to
in the fractionation section. Corrosion in the depropan- watch that stream carefully as far as keeping salt drying
izer on the alkylation unit is a result of SO2 and SO3 operation under control. We have had problems with salt
liberation in the fractionation section. SO2 and SO3 end type plugging in the line going to the depropanizer at
up in the refrigeration loop where they can react with restricted points. We also have had at least one incident of
water to form acids. Proper operation and sizing of the corrosion products in the depropanizer where we think we
depropanizer feed treaters can minimize corrosion. Units did get a little moisture into the depropanizer section.
without depropanizer feed treaters are at the greatest risk
for severe corrosion. One alkylation unit does not operate ARTHUR.W. DeZEEUW (Otto H. York Company, Inc.):
their depropanizer due to this corrosion problem. I would like to ask the panel if anyone has had experi-
Chemical neutralizers are used to mitigate corrosion in ence with use of packed coalescers for dewatering feed
the depropanizer. Neutralizers used in the depropanizer streams. The Otto H. York Company offers permanent,

1995 NPRA Q&A Session on Refining and Petrochemical Technology 127

TOC/INDEX

fixed bed coalescer units capable of reaching very low free ABRAHAMS:
water/caustic levels - less than 10 ppm. We do not have exactly the facilities described here, i.e.,
not having a depropanizer. We do have some experience
HIGGINS: with venting gas from our C4 alkylation units, though.
No experience. When the vent gas combines with wet streams, corrosion
occurs very quickly. I have been told that the corrosion
VICENTE A. CITARELLA
A (Exxon Research & Engineering Company): occurs in about 6 months.
As I indicated in my comments on Question 3, we have In another location, we found out the hard way that
not experienced corrosion and fouling in the deisobutan- SO2 from this kind of a stream will react with H2S in a
izer or debutanizer towers with the Exxon Research and gas collection system and plug up a lot of pipe with
Engineering stirred, autorefrigerated reactor. elemental sulfur. We now caustic scrub that vent gas before
With respect to the depropanizer, the only thing to add sending it co the refinery header and gas plant.
is that usually the depropanizer feed is caustic and water
washed to remove SO2 and possible trace carryover. Caus- QUESTION 12.
tic stress corrosion cracking can occur if excessive caustic Will increased amylene feeds to sulfuric acid alkyta-
carryover is experienced. To prevent it, post weld heat tion result in a shortfall in acid regeneration capacity?
treating is recommended for hot operating equipment Are refiners considering construction of on-site regen-
that can be exposed to caustics. eration plants?

ABRAHAMS:
General The response to this, of course, can be expected to vary
from place to place with the increased trend toward
QUESTION 10. amylene in alkylation feed. Our spent acid rate has in-
What have been the typical /eve/s of total sulfur in creased, as was discussed in Question 7. However, our sites
alkylate from sulfuric acid units as measured by the report that they have enough access to regeneration capac-
ASTM D2622 X-ray method? Have any refiners with ity. We have adequate on-site facilities in our Delaware
caustic and water wash neutralization sections expe- refinery. We tore down acid regeneration facilities in our
rienced excursions in a/&y/ate sulfur? If so, what Port Arthur, Texas plant a few years back in favor ofoff-site
corrective actions were successful? regeneration. Our Louisiana refinery has had an excellent
relationship with the DuPont regeneration facility next
JOHNS: door since the refinery was built.
One of our sulfuric units routinely monitors the sulfur
content of the alkylation blend tank by the x-ray sulfur JOHN W. WARD
D (Haldor Topsoe):
method described. Typical sulfur content ranges from 13 Haldor Topsoe is currently talking with several U.S.
ppm to 45 ppm weight with an average of 23 ppm over a and Canadian refiners regarding the construction of an
60-day period. This unit also has a caustic and water wash on-site Haldor Topsoe WSA plant for regenerating their
neutralization section but has not experienced any excur- spent sulfuric acid. Depending on the refiner’s volume and
sions in alkylate sulfur attributed to this neutralization cost of acid, we are finding payouts in the l- to 3-year
section. range. Refiners can further benefit by designing the WSA
However, there have been excursions in alkylate sulfur.
The unit charges a mixed butane stream from the up-
stream units directly to the DIB tower to recover isobu-
tanes. The upset conditions on upstream units result in
H2 S contamination of the DIB tower, which leads to
sulfur excursions in the alkylate and, obviously, high acid
consumption. The sulfur concentrations from these upsets
have at times exceeded 100 ppm, and at times have caused
near acid runaway situations.

QUESTION 11.
Do any refiners with C4 sulfuric acid alkylation that do
not have a purge depropanizer, employ a vapor purge
from the refrigerant receiver rather than a liquid
purge? Since this vapor can contain SO 2. what dispo-
sitions for it are deemed acceptable?

128 Light Oil Catalytic Processing

TOC/INDEX

unit to allow the simultaneous processing of H2S gas. The JEAN-LUC NOCCA
A (IFP Enterprise, Inc./HRI, Inc.):
WSA unit can then serve as a backup to the existing Claus The main increase in acid consumption linked with
plant and allow the refiner to process higher sulfur crudes processing C5 corresponds with the C5 dienes. Conversion
without building new sulfur recovery facilities. of the dienes to corresponding olefms increases isoamylene
yield by about 5% to 10%, depending upon the amount
BRUCE BOLDUC (BOC Gases): of dienes that you have in your feed. IFP presently has
One option to consider for additional regeneration eight operating units treating C5 for either alkylation or
capaciry is oxygen enrichment. BOC has quite a lot of TAME feed.
experience in the safe supply of oxygen for sulfuric acid
regeneration facilities. The process benefits are quite simi- QUESTION 14.
lar to oxygen enrichment for Claus where oxygen replaces What are the primary considerations in selecting a
the inert nitrogen. We have been able to demonstrate replacement catalyst for a selective hydrogenation
capacity increases of 20% to 30% safely and successfully. unit in HF alkylation feed pretreatment application?

SOLIS:
B. Alkylation-Hydrofluoric Acid Selective hydrogenation (SHP) in this application has
two main functions. First, it hydrogenates the diolefin,
Feed mainly butadiene, to reduce acid consumption and poly-
mer formation in the alkylation process. Second, it en-
hances the octane of the alkylate by isomerizing 1-butene
QUESTION 13.
to 2-butene. As either of these functions decline, the need
When processing a full range amylene stream through
to replace the SHP catalyst becomes evident. The choice
the HF unit, what techniques are available to reduce
of catalyst for this service must be based on the quality of
the impact of contaminants such as oxygenates, di-
the feed to the SHP unit.
enes, and sulfur compounds on acid consumption and
Today most of the SHP units use palladium at 0.3 wt%
polymer byproduct formation?
or nickel 30 wt% catalyst. There are two main differences:
the activity and the poisons resistance. Both are a conse-
HANSEN:
quence of the different metal phases involved, as well as
We process a TAME unit raffinate stream in the HF their metal concentration levels.
alkylation unit when economics dictate. The feed contami- Therefore, the catalyst selection is mainly dictated by
nants of concern are dienes, oxygenates, sulfur, and heavy the type and concentration ofcontaminant in the feed and
olefins. Dienes are removed in a selective hydrogenation in the hydrogen make-up. Some of the main features of
section of the TAME unit. Methanol content is controlled these two types of catalysts, based on pilot plant work that
by proper operation of the methanol recovery water wash we have done with both, are listed on the following table.
system in the TAME unit. DME is removed by routing the
stream with the normal MTBE unit raffinate feed stream Pd/Al 2O3 Ni/Al2O3
through a fractionator for DME removal. Sulfur levels are Conversion (%) 1C4= 2C4= 85
low enough for good HF alkylation unit operation when Resistance to:
combined with the normal unit feed. Heavy olefins are H2S+COS+RSH Inhibitor Temporal Poison
removed by fractionation in the TAME unit.
Organic S Inhibitor Inhibitor
CO Inhibitor Inhibitor
PARKER:
Cl Inhibitor Poison
The only thing I have to add to Mr. Hansen’s comments
Pb-Si Inhibitor Poison
is that suitable pretreatment of the sulfur can be done by
several proven caustic processes that have been available I will add that nitrogen compounds as Nitriles will also
for a long time. For diolefin removal, the Phillips Hy- inhibit the reaction in both catalysts. A general conclusion
drisom process can handle a full range C5 cut. Our Borger, we can make is that nickel-based catalyst can handle levels
Texas refinery does process C5 olefins, but they are the top typically observed in FCC C4 streams, and palladium
cut, not a full range cut. catalyst has been typically used in sulfur-free feedstock.

DAVID L. SMITH
H (Alcoa Industrial Chemicals): FRONDORF:
Another process option to remove the oxygenates and We agree with Mr. Solis. The Corpus Christi SHP unit
sulfur compounds from the amylene stream is via selective has been on-line for a year, and we have not yet faced a
adsorption, either at the tail end of your TAME unit, or need for catalyst replacement. We expect that the criteria
at the feed to your HF alkylation unit. for determining catalyst replacement will depend on

1995 NPRA Q&A Session on Refining and Petrochemical Technology 127

TOC/INDEX

2-butene/l-butene selectivity versus mono olefin satura- ing pressure and the fractionator design need to be com-
tion, contaminant sensitivity, and cost. patible with the type of feed being processed. We do feed
C3 and C4 streams through our unit without difficulty.
HANSEN:
I would like to add that excess hydrogen control and/or KARL LIN (Alcoa Industrial Chemicals):
removal is very important in maintaining good acid con- I wanted to announce that Alcoa has developed an
sumption in the downstream alkylation unit. We feel that arsine removal catalyst. Selectsorb AS can be used to
noncondensibles in the alkylation unit feed have been a remove arsine from the FCC C3 cut. We are now under-
significant source of acid losses in our unit. going field trials.

QUESTION 15. JEAN-LUC NOCCA (IFP Enterprise, Inc./HRI, Inc.):

What type of problems can be expected when process- With regard to C3/C4 mixed feed, when you try to push
ing a mixed C3/C4 stream in an alkylation feed selec- isomerization from 1-butene to 2-butene, you tend to lose
tive hydrogenation unit? a small amount of propylene. With regard to contamina-
tion, you may have some COS and H2 S going to the
SHEN: C3/C4 feed. Therefore, it is very important to have a sulfur
There are basically two types of selective hydrogenation tolerant catalyst.
units, the conventional fixed bed system and the new Also, there may be arsenic in C3 /C4 feed, and we
catalytic distillation system. We have designed a number of provide a commercially proven trap material for arsenic.
fixed bed systems for C3/C4 mixed streams. Currently, we Arsenic is not a permanent poison. It is only a temporary
are working on the C4/C5 catalytic distillation system in an poison, and complete activity can be restored through
FCC depentanizer tower. As it is designed, it will reduce adequate procedures.
the C3/C4 dienes from the 1100 ppm range to 30 ppm.
Both types of units are sensitive to metals, as has already
been mentioned. Since the C 3 make from the FCCU
Process
contains arsine, it is a permanent poison for the catalyst.
It is better to fractionate out the C3 to bypass this unit, or
QUESTION 16.
put in a guard bed to remove the arsine.
Both types of units are very sensitive to plugging by the What has been the recent operating experience with
heavy ends, so for the fixed bed system, the upstream HF alkylation additives to improve the vaporpressure
fractionation is quite important. A hot hydrogen stream characteristics of the circulating acid sfream?
sometimes can be used to de-gum the catalyst bed. For the
catalytic distillation system, a steady reflux stream in the JOHNS:
depentanizer tower could help the situation. Also, for the ALKAD operations are continuing at Texaco’s El
catalytic distillation system the tower upsets should be Dorado, Kansas HF alkylation unit. Additive operations
minimized because every time heavy ends hit the catalyst since September 1994 have confirmed that the alkylate
bed there is a chance of some metals contact with the octane and distillation have improved compared to opera-
catalysts. tions without the additive with similar rates and C3, C4,
Another thing to emphasize is the selectivity between C5 feedstock compositions. Inspection of the unit during
1-butene to 2-butene. For the HF unit, 2-butene is more the April 1995 turnaround confirmed that there were no
favorable than 1-butene; it produces better alkylate oc- significant changes in corrosion or fouling compared to
tanes. It is something worth considering. prior inspections.

SOLIS: PARKER:
SHP units using mixed C3 /C4 as feedstock for HF Phillips and Mobil have been jointly developing a
alkylation have been in service for many years. The opera- reduced vapor pressure liquid catalyst alkylation process,
tion of the unit is nearly identical to that with C4, with REVAP. Phillips is currently completing detailed design to
the exception that as the isomerization of l-butene to convert the HF unit at its Woods Cross, Utah refinery to
2-butene is pushed, C3 olefin retention decreases. Typi- the BEVAP technology. My understanding is that Mobil
cally, from 1% to 3% of olefin is lost during this operation. is also in the detailed design phase to convert the HF unit
at its Torrance, California refinery. Adetailed paper on this
PARKER: REVAP process was presented at the API Operating Prac-
I am only familiar with the Phillips Hydrisom process tices Symposium on May 9, 1995. This paper includes cost
which can handle both C3 and C4 feeds. This process and yield information along with the general description
includes a downstream stripping fractionator. The operat- of the design changes needed.

130 Light Oil Catalytic Processing

TOC/INDEX

QUESTION 17. sulfuric alkylation units to practice the Topsoe process.

Do any of the solid acid catalyst systems under devel- The process employs a liquid super acid that is dispersed
opment show significant economic advantage versus upon the surface of a particulated solid substrate in a fixed
H 2 S0 4 or HF alkylation? If so, when will they be bed reactor.
commercialized?
W. ROBERT EPPERLY (Catalytica Advanced Technologies):
PARKER: As reported earlier, Catalytica, Conoco, and Neste have
I already discussed the liquid catalyst process we are operated a 7 bbl/d pilot plant for over 5000 hours. Progress
now developing. During the research process, we looked has been made toward our target of a 3¢ to 4¢ per gallon
at solid catalyst and concluded that a liquid catalyst system advantage, based on bench scale results. Further work is
was superior. Major problems that we observed with solid needed over the next 2 years to reach our targets and to
catalyst systems were short run lengths between regenera- have the technology ready for licensing. We are having
tions, the inability to regain activity on the catalyst after discussions with several companies, under confidentiality,
regeneration, and high capital investment and operating
to join our partnership.
costs. Costs are high because you need large volume solid
beds, and heat removal equipment for the highly exother-
FRANK HIMES (UOP):
mic reaction is expensive. You also need regeneration and
solids handling equipment. UOP is continuing to develop the solid catalyst system
As for comparing the operating cost, the catalyst cost previously announced to the industry in this and similar
at our Phillips unit at Sweeny is running about 5e per forums. The material is a solid phase regenerable catalyst
barrel of alkylate. Our acid losses are about 0.08 lb per and UOP is in the process of confirming product quality,
barrel of alkylate. We are currently running about 3 years yield, catalyst stability, and similar parameters in our pilot
between turnaround intervals and are considering moving plants. Earlier this year, an initial detailed design was
to 4 years based on the condition of the unit. I would add completed and costs were developed by a major contrac-
that we have pretreatment in front of that unit and it seems tor. Based on these estimates, the technology is competi-
to make a big difference. tive with current alkylation technology. While certain
technical achievements remain to be accomplished, UOP
SOLIS: expects to be in a position to support a commercial design
We are following very closely the work that several by early 1996.
groups are carrying out for the development of processes
using a solid catalyst for motor gasoline alkylation. I would
like to say that this question already has an answer for other
types of petrochemical alkylation. CEPSA, through its
affiliate Petresa Canada, has started up the first Linear
Alkylbeneze Plant (LAB), where benzene is alkylated with
Cl0 to Cl2 olefins in a solid bed catalyst reactor system.
The solid bed alkylation unit has a capacity of 75,000 tons
per year of LAB and has been installed at Becancour,
Canada. The plant, started up in May 1995, is operating
today at design capacity.

PHILIP GEREN (Haldor Topsoe, Inc.):

The fixed bed alkylation process which is being com-
mercialized by Haldor Topsoe, appears to have about a
$1.10 to $3 per barrel of alkylate economic advantage
when compared to sulfuric acid alkylation. In answer to
the second part of the question, the status of commerciali-
zation is that there presently are about 20 companies,
under secrecy, evaluating the technology for specific appli-
cations. Three of these could be considered to be hot
prospects. And we have reason to believe that it may be
possible to sign the first license agreement in this calendar
year or early in 1996.
The three hottest prospects are for retrofit ofequipment
to existing catalytic polymerization, HF alkylation and

1995 NPRA Q&A Session on Refining and Petrochemical Technology 131

TOC/INDEX

DAVID PARNELL (TPA, Inc.): netic. That means that sometimes the float will stick to the
What happens to the quality of the alkylate from these casing and lose movement. One refinery in northern
solid catalyst processes? Do the octane values stay basically Europe started an HF reactor unit in the summer with the
the same as they are with the liquid acids, or do the octane Monel magnetic indicator. It worked well in the summer,
values get better or worse? Also, what happens to other but during the winter they lost the level reading, so they
properties of the alkylate? What is the expected life of the had to heat trace the instrument to make it functional
catalyst or cycle time for the regenerable type catalysts? again.

PHILIP GEREN (Haldor Topsoe, Inc.): KELLER:

As the reactions are the same, the octane quality is In general level applications, the magnetic level column
approximately the same as with sulfuric and HF acid. has become the hardware of choice not only in the alkyla-
Yield is a function of how much or how little acid soluble tion unit but throughout the refinery. Sight glass cleaning,
oil byproduct is made. And in a fixed bed process with
leaking, and blowouts have obviously been eliminated
short residence time during which the olefins are within
where the magnetic devices are used. Single point, multi
a reactive zone of a fixed bed, the ASO production is
point, and complete level control can easily be
about one-fifth of that which is expected in a sulfuric acid
adapted/added to the magnetic level columns.
plant.
On the negative side, as with most intrusive level
FRANK HIMES (UOP): devices, rag layers, interface levels and varying specific
We are targeting similar octane and yield to the liquid gravity can cause major problems. It is very important to
catalyst systems, and we think we will be able to make that. properly specify the float’s specific gravity and material of
construction.

Mechanical KENNETH G. TASKER (HRI, Inc.):

In support of Mr. Keller’s comments, some of HRI’s
QUESTION 18. licensees use magnetic level indicators with a marked
What has been the industry’s experience with the degree ofsuccess. Our experience with these indicators has
metal type magnetic level indicators? primarily been in the aromatics plant areas. These services
are very clean in nature.
PARKER:
Our Borger, Texas refinery has had magnetic level de- JEFFREY FITT (BP Oil Company):
vices on the settlers, recontactors and depropanizer boot One of my experienced operators, D. Travaglini, would
for a number of years. According to a source there with move the set point of the float controller numerous times,
whom I spoke, they have been very dependable. They have up and down, over his shift to make sure the float was not
had minor problems with the float sticking and hanging. frozen. This practice would also keep the rod free of
That kind of problem with the float hanging is typical of fluoride deposits over the 15% to 20% range that he would
what has been reported in our HF seminars. My only move the controller. Several of our refineries have taken
comment is that operational influences (such as water level this principle and automated it on the process computer.
in the unit, neutralization during turnarounds, or if some
Throughout the day the computer changes the level con-
people are extending the run life) can lead to more scale
troller set point, and an alarm sounds if the acid level does
buildup in the unit that may impact the effectiveness of
not correspond to the set point. We have also verified these
these devices.
acid boot levels using ultrasonic thickness (UT) meters
This is somewhat off the subject, but we have had
and found very good level correlation with these devices.
success with nuclear level devices for most of those same
services, and from our seminars almost all the experiences
DON A. HUSTEO (UOP):
with the nuclear devices have been very successful.
UOP sells a Density Profile System (DPS), which is
SHEN: generally used in settlers for tracking the interfaces be-
With the magnetic level indicator for an HF unit, tween acid, emulsion and hydrocarbon. The density meas-
wherever it has contact with the acid, Monel material urement is based on nuclear backscatter technology. We
needs to be used to avoid fluoride stress corrosion crack- have had a few customers who have installed this in acid
ing. At normal temperature, Monel is not magnetic. The storage drums and acid dump drums. The advantage is
indicator flows up and down freely with the liquid level. that the sensor rides up and down inside of a well in the
However, we have been told that when the temperature vessel so it does not contact acid. Therefore, there is no
drops and approaches 40°F, the Monel becomes paramag- sticking of the sensor assembly due to iron fluorides.

132 Light Oil Catalytic Processing

TOC/INDEX

General these provide food for thought. I also think that capital
costs are going to vary widely for different plants and
QUESTION 19. different companies, etc.
Discuss the costs associated with the following ideas Conversion of HF to sulfuric is not an easy job, and we
in regard to managing the hazards of hydrofluoric acid expect the costs to run one-half to two-thirds the cost of
release: a new sulfuric acid alkylation unit, or about $2,500 per
a) convert the reaction section to sulfuric acid; stream day barrel. A complete mitigation system, water
b) install complete mitigation and rapid dump technology; mitigation, rapid dump, water runoff, containment, cam-
c) install HF vapor pressure suppressant technology. eras, detectors, etc., has been reported to be as high as $20
million to $30 million for a 15,000 bbl/d unit or about
PARKER: $1500 to $2000 per stream day barrel. HF acid vapor
On part B of the question, Sweeny did install Phillips’ suppressant technology depends on the technology, the
rapid acid transfer technology. This is on a 15,500 bbl/d unit configuration, the equipment limits, etc. But we
unit. The records for the project indicated that the cost think it would be in the $500 to $1000 per stream day
was about $2.1 million for this project. They had pre- barrel range.
viously installed water curtain mitigation technology
when the unit was built, so that cost was not included. My KELLER:
understanding is that this cost is somewhat lower than we We, likewise, have only done rough estimates. All the
have been seeing typically on revamp units, and it may numbers are based on our existing 10,000 bbl/d HF unit.
have been helped out by the fact that the unit was relatively We think that in order to convert it to sulfuric acid
new and may not have required as many modifications as technology, we would have to demolish the reactor sec-
some older units. tion, the C3 and C4 alumina treaters and the C4 KOH
On Part C, preliminary cost numbers, from both the treater. We would revamp the HF stripper, regenerator,
paper to which I referred and from our experiences with and neutralization equipment at a cost of $80 million in
our design at Woods Cross, show a capital expenditure of California. We have already done the series acid flow for
somewhere between $3 million and $4 million to be rapid acid transfer at a cost of $3.5 million. The water
typical to convert to the REVAP technology. You will also spray has been upgraded to API standard for $2.5 million.
have higher operating costs with the new process. And if we need to put in an outside battery limits rapid
dump technology, it is another $3.5 million. The acid
SHEN: vapor pressure suppression, we think, will be $15 million
Our Long Beach office is currently doing engineering to $20 million after we pay royalties.
work for Mobil in Torrance, California for their 21,000
bbl/d alkylation unit. As Mr. Parker mentioned in his BOB PUSCHINSKY (Bob Puschinsky, Inc.):
answer to question 16, there are cost data available in the I have observed mitigation costs of $4 million to $8
API paper. million for the units that I have been working with, which
The REVAP process has been tested and shows that the are sized up to 10,000 bbl/d. I have heard of $2 million
additive reduces airborne HF migration downwind in the mitigation costs for a grassroots unit.
event of an accidental acid catalyst release. Test data show
that, typically, HF airborne reduction with this additive is
QUESTION 20.
expected to be between 60% and 90%. The actual reduc-
What is the typical mode and cost of disposal for spent
tion depends on the release conditions and the amount of
caustic from relief gas scrubbing and polymer (i.e.,
additive added to the acid.
ASO) neutralization in HF alkylation operation?
The additive is a non-volatile, non-odorous, low toxic-
ity material. It is compatible with carbon steel and Monel
SOLIS:
alloy, which are commonly used in the conventional HF
process. The published data shows that this additive has a All acid vents and reliefvalves in our HF alkylation unit
small but positive influence on the alkylate octanes. are piped to an acid relief system. The acid relief gases flow
Oxygenates were also added to the feedstock for the through a header to a knockout drum where any liquid
demonstration units. The results proved the process is carryover separates from the gases. The relief gases flow
tolerable to MTBE raffinate type feed impurities. from this drum to a scrubber where they are neutralized
with caustic. As caustic we use K(OH).
FRONDORF: HF+KOH KF+H2O
I caution that the numbers I have from our Corpus
Christi personnel are only general guidelines. We have not On the other side, the polymer is also treated with
made detailed budget estimates on all of these choices, but KOH in a neutralizer drum.

1995 NPRA Q&A Session on Refining and Petrochemical Technology 133

TOC/INDEX

We have a low KOH makeup requirement for these two PARKER:

services because our system includes a spent KOH regen- I know there are companies working with our plant
erator using (CaOH) 2 lime by following the following trying to figure out a way to reprocess the calcium fluoride
reaction: material, converting it back to HE There are a few bugs
to work out, but within the time frame you are talking
2KF + Ca(OH)2 2 KOH + CaF2
about, we might have a workable process.
The CaF2 settled at the bottom of the regenerator from
where we send it for disposal to be used as filling material. RICK LEICHT (National Cooperative Refinery Association):
Our overall consumption is 0.4 lb of KOH per ton of Are any of the vapor suppression additives affecting the
alkylate and 0.25 lb of Ca(OH)2 per ton of alkylate. disposal of the sludge?

PARKER: PARKER:
We license these plants around the world, and disposal Basically, the additive is kept inside the unit, so you
options are really dependent upon the plant location and really would not be looking at disposal during normal
permitting restrictions. At Sweeny, we have a fairly strict operation. Worst case, it will be present at very small levels
permit and are probably near the top of what you have to in any sludge generated, and it has a low toxicity and
spend to dispose of the material. We initially routed our should not create a disposal problem.
liquid effluent directly to our waste water treater, and we
did get some fallout into our waste water treater system
that we had to treat as primary sludge. Since then, we went C. Reforming
to the filter press operation that I described earlier. Total
cost for that operation is about $5.00 per barrel ofeffluent Feed
coming out of the calcium fluoride mix tank. That cost
includes rental on the filter press, labor, chemicals and QUESTION 21.
supplies, transportation, and disposal costs.
Are refiners experiencing corrosion or fouling prob-
lems in the vertical combined feed exchanger (VCFE)
HANSEN:
in the reforming unit? What measures are being taken
I described our caustic soda-based calcium fluoride to resolve this, and which cleaning methods are being
system this morning. In addition, I would like to mention
used? How is the fouling different from feed to product
that the typical chemical consumption for sodium hydrox-
side?
ide is in the range of 2 lb to 2.5 lb per pound of HE and
the calcium chloride consumption is in the range of 3 lb
POTSCAVAGE:
to 3.5 lb per pound of HE
By virtue of the vertical design of the exchangers men-
tioned in the question, these exchangers have far fewer
KELLER:
problems than the traditional horizontal exchangers. In
See Environmental Control and Energy Conservation
fact, we have never seen bundles in a vertical design have
Section Question 27 for more details.
fouling problems. But as you will hear from my colleagues,
The neutralization section of our HF alkylation unit is
occasionally there will be problems in the heads.
a standard UOP recirculating KOH system. The spent
KOH is regenerated with lime. Precipitated calcium fluo- If fouling does occur on the feed side, it can be organic
ride is filtered using a rotary vacuum filter. The recovered or inorganic in nature. If it is organic, it is the result of
KOH solution is recycled back to the recirculating system polymerization of the feed, which is more likely if the feed
and the calcium fluoride solids are trucked to a permitted went through intermediate storage or was purchased. If
landfill as hazardous (corrosive) waste. We have some costs this is the case, possible solutions are oxygen stripping,
for the filtering operation. The filtering costs us $263 per feed filtration, avoiding intermediate storage or tank blan-
ton of filter cake. To truck it away from the refinery is keting. If the foulant is inorganic in nature, it is most likely
another $107. The landfill charges $48 a ton and the corrosion products. And clearly the best way to approach
disposal taxes are $32 a ton. The grand total is $450 a ton that is to go back upstream and find out where the
for the filter cake. corrosion is taking place.
On the product side, there could be fouling due to salts,
JOSEPH PICCARI (Allied Signal, Inc.): in which case a water wash is the simplest approach to use.
Can anyone address state or federal regulatory issues If the foulant is iron sulfide, then there is most likely a
that might have a direct impact on the disposal of solid corrosion problem in place, and the best approaches are a
fluoride waste within the next 5 years? metallurgy change or some form of chemical treatment.

134 Light Oil Catalytic Processing

TOC/INDEX

SOLIS: corrosion. We probably will not be able to look at this

We have had one fouling experience in our vertical again for another 4 to 5 years.
combined feed exchanger (VCFE) on the feed side (tubes)
during regeneration caused by caustic foam entrainment. FRONDORF:
We resolved the problem with a hot water wash before Between the. Corpus Christi refinery and the Lake
start-up and after regeneration. Charles refinery, we have three CCR units, each one
Also, it has been reported by our licenser (UOP), that equipped with vertical feed effluent exchangers. At Corpus
tube side fouling can occur during normal operation due Christi, neither of those units has experienced corrosion
to feed contamination, typically overuse or misuse of cor- or demonstrated a need for pulling the exchanger bundles
rosion inhibitor. High rates of inhibitor injection, coupled to date. One of our units has been in service for 20 years.
with an inhibitor that is not highly hydrocarbon soluble, We do plan to replace the bundles on the oldest unit
results in the inhibitor dropping down the stripper and during the 1996 turnaround due to some pluggage that
moving on to the reforming unit. The inhibitor can then we had noted in a number of tubes during the 1992
combine with corrosion products and deposit on the VCFE turnaround. It is likely that this pluggage is related to
tubes in the areas where vaporization of the liquid feeds corrosion, although we have not definitely identified that.
occurs. Inhibitors are typically high molecular weight ma- The Lake Charles CCR has four vertical tube bundles
terials, e.g., C9+, and should be injected in the overhead with 13 years of service. They have not been pulled, but
line and not in the reflux of the hydrotreater stripper. we have plugged a small number of tubes in one bundle
Although unlikely, shell side fouling could result from that failed the nitrogen test on the last turnaround. Our
ammonium salt deposition. Typically the exchanger tem- current plan is to pull that bundle and conduct a more
perature is not cool enough for the salts to deposit on the detailed inspection at the next turnaround before deciding
shell side of the VCFE; rather the salts deposit further on a course of action. We have not seen evidence of
downstream. At high severities, trace amounts of PNAs significant fouling or loss of performance to date.
(polynuclear aromatics) could be generated at the re-
former, and it is possible that PNAs could deposit on the JOHNS:
shell side of the VCFE. We have three units with these type exchangers. We
Removing the bundle for cleaning purposes must be have experiences similar to Mr. Frondorf’s regarding cor-
avoided unless absolutely necessary. Pulling the VCFE rosion and fouling. It has not been severe. It has been quite
tube bundle can be very difficult. Removing and re-install- adequate over a 15-year period on most of those units.
ing the bellows is a difficult task. In addition, setting the However, one unit did have a crack in the bellows, which
bundle horizontally will, in most instances, cause damage, was corrected on a turnaround.
such as tube-to-tube sheet leaks. We thought we had a leak on another unit. We tried to
determine on stream if it was a leak, but could not do it
MORGAN: with sampling. I caution that you will be frustrated if you
Our CCR unit was constructed in 1978 and since that try to sample and check octane or some parameter that
time we have had 17 years of monitoring for corrosion in will help you determine a leak. We did x-ray testing and
the heads. We have not seen any corrosion. However, we found a leak in one of the exchangers and a possible leak
did see a deposit through the flange face, which we sub- in the other. When we shut down, we found that the one
sequently entered the exchanger to remove, and during that had the positive leak had a very small crack in the
that process we destroyed the expansion joint. It is very bellows. The other one that showed up so-so had an
difficult to get into the vertical exchangers to clean them internal flange leak separation, which probably was leak-
without doing some damage. This dead spot at the bottom ing from day one, 4 years prior to shutdown.
of the head on the tube side was a hard scale that seemed Most of these exchangers did not experience significant
to be almost welded to the metal surface. After cleaning fouling. There have been problems with caustic deposits
it, we found only about 2 gallons of deposits there, and and ammonium chloride corrosion reported on other
we had a strong feeling that these deposits were really a units.
product of construction and not necessarily of operations.
KELLER:
EMANUEL: Our CCR platformer is 15 years old. We have had light
We have two vertical exchangers in which we have to heavy fouling on the feed side (tube side), depending
found corrosion on the bottom heads where residue and on the vertical exchanger elevation, with the middle sec-
scale collect during the run. In April of this year, we went tion being the worst. Cleaning has been moderately suc-
in and sandblasted to remove all the hard deposits, and cessful using high pressure water blasting (up to 10,000
used thermal fusion spray coating with a TP321 stainless psi). Generally, the feed side has tight scale, while the
steel powder to see if this would help this area as far as product side has loose/friable scale.

1995 NPRA Q&A Session on Refining and Petrochemical Technology 135

TOC/INDEX

We have experienced significant corrosion (10 mpy) on ANGELO FURFARO (UOP):

the feed side of the exchanger. The corrosion is possibly a The panel has presented a lot of very positive informa-
result of salts and moisture in the feed/hydrogen mixture tion. Since platforming “should be,” and in general is, a
entering the exchanger. The corrosion is confined to the clean service, there “should not” be exchanger fouling.
first 10 ft of the vertical bundle and appears to stop at a UOP’s experience with vertical combined feed exchangers,
temperature where free water no longer exists. The bundle especially in CCR platforming process unit service, bears
has been replaced twice (after 8 years and 5 years ofservice) that out. Our experience list shows that in over 1000
due to this feed side corrosion. operating years of CCR platforming process experience,
We would like to ask the audience if anyone has con- there have only been 3 recorded instances of feed side
sidered installing either a sacrificial 10 foot exchanger ( it fouling.
is always the bottom 10 ft of the big exchanger that is These refiners have cleaned, but not pulled, the ex-
corroded) or a liquid feed guard drier? changers due to tube side fouling. Tube side fouling has
been due to feed contamination, typically overuse or
GARY KUCHCINSKII (Amoco Oil): misuse of a corrosion inhibitor, or from sending feed from
We know that ammonium chloride deposition is highly unblanketed intermediate storage to the platforming proc-
dependent on temperature. Could those of you who gave ess unit. High rates of inhibitor injection, coupled with an
us all the information about fouling on your exchangers inhibitor that is not highly hydrocarbon-soluble, result in
give us an approximate effluent temperature coming out the inhibitor dropping down the stripper and moving on
of those exchangers? to the reforming unit. The inhibitor can then combine
with corrosion products and deposit on the VCFE tubes
JOHNS: in the areas where vaporization of the liquid feed occurs.
I cannot cite a temperature right now, but let me say Inhibitors are typically high molecular weight materials,
that the ammonium chloride corrosion was from a deposit i.e., C9 +, and should be injected in the net overhead line
in which the ammonium chloride was not washed out and not in the column reflux. UOP recommends a filming
before exposure to air upon shutdown. amine, Unicor C, at a rate of about 2 ppm to 4 ppm based
on the net overhead flow. Refiners should also ensure
THIERRY A. SOURP
P (Packinox): proper hydrotreating water injection rates, i;e., 3 LV% to
Fouling problems are, overall, very significantly re- 5 LV%, are employed in that unit’s CFE section, to
duced when using Packinox welded plate heat exchangers. minimize corrosion and corrosion products being sent to
The lower tendency to foul is inherent to corrugated plate the hydrotreater stripper and eventually the platforming
designs, as the turbulence level is much higher between process unit.
two corrugated plates than inside a straight tube at the
Although unlikely, shell side fouling could result from
same stream bulk velocities.
PNA (polynuclear aromatics) and/or ammonium chloride
In addition, by carefully addressing liquid feed distri-
salt deposition. Typically the exchanger temperatures are
bution and vertical entrainment, a well-designed plate
not cool enough for the salts to deposit on the shell side
exchanger will also eliminate liquid phase stagnation, a
of the VCFE; rather the salts deposit further downstream,
common source of feed side fouling in the field. (A low
i.e., in the reformate stabilizer, fin fans, and recycle gas
gas/liquid ratio associated with a poor level of turbulence
compressor. It is possible PNAs could deposit on the shell
may lead to phase separation and non-entrainment; the
side. Trace amounts of PNAs could be generated from the
liquid cannot move upward in the tube, it begins to
reformer, typically at high severity.
stagnate and eventually to “cook” on the heated tube wall,
There have been two reported cases of ammonium
creating fouling in the low to middle section of the tube).
Proper feed distribution inside a Packinox is ensured by a chloride salts deposition in Packinox exchangers (reactor
unique injection of the liquid phase directly into the heat effluent), which eventually required cleaning. In both
transfer bundle. Vertical entrainment is guaranteed by cases, feed nitrogen was typically greater than 0.5 ppm.
maintaining the stream velocities above certain plate-pat- Cleaning was required after approximately 1½ to 2 years
tern-related velocity thresholds. The turbulence level will on-line and was successful.
then keep gas and liquid phases homogeneously mixed Pulling the VCFE tube bundle can be very difficult.
throughout the lengths of the plates. Removing and re-installing the bellows is also a difficult
For corrosion, our plates being of stainless steel, the task. Care should be taken not to damage the exchanger
only precaution that we take on semiregenerative type when pulling or installing the tube bundle. In addition,
reformers is to make sure that the regeneration procedures setting the bundle in the horizontal will, in most instances,
cannot lead to the simultaneous presence of water, chlo- cause damage such as tube to tubesheet leaks. UOP does
ride, and oxygen. not recommend removing the bundle unless absolutely

136 Light Oil Catalytic Processing

TOC/INDEX

necessary. However, as it does occasionally become neces- SHEN:

sary, the following has been provided to refiners: This question has four parts. The first part addresses
UOP would recommend that you remove the bundle the prefractionator location. The location for the prefrac-
for a complete evaluation of the shell, floating head, tionator depends on the naphtha processing configuration
floating head expansion bellows, bottom shell cover, and you have at the refinery. If you have an isomerization unit,
tubes. UOP has, over the years, seen corrosion problems the most common place for the naphtha splitter is down-
to the heat affected zone (HAZ) of the bellows attachment stream of the full range naphtha hydrotreater. The over-
weld, corrosion to the dead area of the bottom sheet cover, head stream can be fed into the isomerization unit. One
and a problem with a leak at the bottom tubesheet. UOP thing that I will point out is this isomerization unit might
feels the bottom tubesheet case was caused by tube vibra- need some modification to handle the exothermic heat
tion. This resulted in the tubes breaking away from the caused by the benzene saturation. If a refinery does not
rolls and, over a period of years, creating a larger tube hole have an isomerization unit, the naphtha splitter tower
diameter in the tubesheet. could be placed upstream of the naphtha hydrotreater. The
Care should be taken when removing the tube bun- overhead stream can be caustic-treated, then blended into
dles. Damage to the tubes has occurred when the bundle the gasoline pool.
has been laid in the horizontal. Some refiners have had The second part of the question is about the prefrac-
tionation efficiency. The separation of benzene precursors
success leaving the bundle hanging in the vertical for
is not as easy as many people expect. In one design we used
cleaning and inspection. Shell inspection is performed
an existing tower with 40 trays. We kept the C7 and the
after cleaning and, in some instances, included a wet
heavier in the overhead to be 1.5 LV% or less. The results
fluorescent magnetic particle examination of the welds.
showed that only a portion of the benzene precursors go
UOP is not aware of cracking being found in the 1¼ Cr to the tower overhead. The numbers showed 57% of the
section. Corrosion to the shell has normally not occurred, benzene in the feed went to the overhead, 59% of the
or has been minimal. methocyclepentane went overhead, and only 30% of the
cyclohexane went overhead.
J.B. ROOOEY (Roddey Engineering Services, Inc.): The third part of the question is about pump upset
The vertical exchangers that are on many of the reform- protection. Most of the units we encountered do not have
ers have the feed on the tube side and in a horizontal a surge drum after the naphtha splitter tower. We believe
exchanger; usually the feed side is on the shell side. I think that a reasonable design of the liquid residence time should
the experience of the industry is that the vertical exchang- be sufficient for this case.
ers do not have as much of a tendency to foul when the The fourth part addresses the correlation between the
feed is on the tube side. However, all reformers have benzene precursor concentration and the reformate ben-
corrosion products in the feed. These corrosion products zene content. I have numbers here, and we know that
come from iron sulfide deposits in the stripper and in the reformate benzene concentration is strongly a function of
naphtha/desulfurizer. They usually deposit in the feed the catalyst acidity and the operation severity. For a 300
effluent exchanger and on the tube side, in the case of a pound semiregenerative unit at 90 severity with zero ben-
vertical exchanger. Many of these have fouled severely. As zene precursors in the feed, the benzene content of the
some of the panelists have already said, some of the tubes reformate is from 0.28% to 0.41% depending on the
have completely plugged. They can be cleaned by lancing acidity of the catalyst. When operating at 95 severity, it
from the top, and I guess that is the only way to clean ranges from 0.46% to 0.61%. When operating at 100
them. But it is not necessary to pull the exchanger. Usually, severity, it ranges from 0.67% to 0.97%. Those are liquid
the fouling is on the tube side so it can be lanced in place. volume percent of benzene.

QUESTION 22. ABRAHAMS:

Some refiners are prefractionating reformer feed (re-
One of our plants prefractionated benzene precursors
from their feed. While they were doing that they found
moving benzene precursors) to reduce benzene in gaso-
answers to the four-part question. One, the naphtha split-
line. Please cite your experience with the following:
ter tower is located at the crude unit. The simulation of
a) where are the prefractionating towers located;
that tower shows tray efficiencies of about 75%, although
b) what fractionation efficiencies are typical; they did not really report fractionation efficiencies on
c) how is the reformer charge pump protected from pre- individual components.
fractionator upsets; Upsets are reduced to the units with intermediate tank-
d) how well does the benzene precursor concentration in age before the naphtha pretreater. Testing has shown some
the reformer feed correlate with reformate benzene correlation between feed precursor concentration or oper-
content? ating severity with reformate benzene content; but the

1995 NPRA Q&A Session on Refining and Petrochemical Technology 137

TOC/INDEX

data are scattered. As Mr. Shen mentioned, it involves an article that appeared in Hydrocarbon Processing in June
several factors. of 1993 that quantifies benzene yield as C6 is cut out of
the feed.
KELLER:
We built a new naphtha hydrotreater (NHT) a year ago
for the federal reformulated gasoline specifications. We Process
located the prefractionating column between the naphtha
hydrotreater and the reformer. The benzene precursors are QUESTION 23.
taken overhead as a hydrotreated light naphtha blending What has been the industry experience in the removal
stock. of the “heel” catalyst from CCR reformer catalyst for
Our prefractionator design is 95% C7 recovery in the re-use:
bottoms and 95% cyclohexane recovery in the overhead. a ) Was “density grading” successfully used to produce a
A performance test showed 99.7% of the C7 in the bot- “light” fraction forreuse? How did the unit perform after
toms and 95.3% of the cyclohexane in the overhead. the “light” fraction was used as makeup to the unit?
Most refineries run direct from the NHT to the re- b) Has anyone successfully ex situ regenerated the “heel”
former. A reformer feed drum is not required. There are catalyst split into light and heavy fractions, and used
no mechanical or automatic devices to protect the re- either one for makeup to the unit?
former charge pumps. In our case, the prefractionator
bottoms feeds a rerun column where the overhead be- EMANUEL:
comes reformer feed and the bottoms goes to jet fuel. We We completed a turnaround in April of this year where
rely on operating alarms, specifically high and low accu- we generated approximately 35,000 lb of reformer catalyst
mulator level alarms, at the rerun column to alert the contaminated by heel catalyst. We took it out and had it
operators. density graded by CRI of Louisiana. Of that amount,
A good correlation exists between benzene precursors approximately 31,000 lb was recovered as a light fraction.
in the reformer feed and benzene in the reformate. The The light fraction was returned to the reactors and no
reformate benzene runs 0.2 LV% to 0.3 LV% above the operational problems were noted on start-up. We did not
total C6s in the feed. This increase can be explained by do ex situ regeneration in this process. There was an article
cracking reactions in the reforming process. in the September 18, 1995, issue of Oil & Gas Journal,
which was co-authored by one of our process engineers. It
GIUSEPPE MOSCA (Glitsch, Inc.): has a more detailed account of this procedure.
I would like to draw your attention to a revamp we did
at a major refinery in Italy. The straight-run naphtha FRONDORF:
splitter was retrayed using Glitsch high capacity trays. We have had at least two occasions that I can think of
Because the only purpose of the revamp was to improve where we unloaded and subsequently reused the CCR
the fractionation efficiency, by using our Nye TM Trays, we catalyst. In the one case at our Corpus Christi, Texas plant
were able to install about 25% more trays than were we cascaded the catalyst load from one of our two CCRs
existing by reducing the tray spacing. The results were very down to the other. In the process of dumping the catalyst,
good: the C6- in the bottom went from about 3% to less we tried to separate out the heel catalyst visually first, and
than 1%. Unfortunately we do not have any data on then sent the remainder of the load down to the other
benzene into the reformate. But, what we know is that reformer. We introduced it by going through the regen-
after this revamp the refinery was able to produce the eration section while purging the existing catalyst out at
gasoline with less than 1% benzene. the same time. It made for a very interesting regeneration
for a considerable period of time. But we did get through
LEE E. TURPIN (Honeywell Profimatics, Inc.): it. It was a long process and we had to watch the regenera-
When significant quantities of MCP and cyclohexane tion carefully.
are in the reformer feed, one can reasonably predict ben- We have done a similar thing at Lake Charles, Louisiana
zene yields from severity and volume percent of precursors where we tried to separate out the heel catalyst visually as
in the feed. However, when you dehexanize the feed, the best we could and then went back in on a dual burn. We
C6 paraffin that is produced via cracking is readily cyclized were in a dual burn period for a long time. Again, we got
and then dehydrogenated to benzene. The amount of C6 through it.
paraffin that is produced via cracking decreases as pressure Corpus Christi made an attempt at density grading of
decreases, but the percent of that C6 paraffin that is the heel catalyst in early 1993. Their reformer had taken
converted to benzene increases. In my opinion, the only an unscheduled shutdown and at that time the catalyst
way benzene yields can be accurately predicted with a inventory was only about 1 year old. So there was a strong
dehexanized reformer feed is with a kinetic model. I wrote incentive to recover as much of the catalyst as we could.

138 Light Oil Catalytic Processing

TOC/INDEX

And we expected we might recover 40% to 50% of the The catalyst itself is able to flow when unloaded, i.e., it is
heel catalyst as light usable material. not fused together.
The process was unproven at the time, but the eco- One additional comment: as noted in earlier responses
nomic savings were such that we felt it was worth it. We to this question, UOP has guided a few of our licensees
worked closely with the catalyst handler. Roughly the last through the restart of either inadvertent, or intentional,
8% to 10% of the catalyst showed signs of heel contami- reuse of heel catalyst. The amount of catalyst contami-
nation, so we routed it to the density grading. We worked nated with heel catalyst is dependent upon the unloading
for quite a long time but could not make a density procedures used. Operation with heel catalyst mixed into
separation of lights and heavies. In fairness to the situ- the catalyst inventory can be arduous and time consum-
ation, with the catalyst as new as it was, there may not have ing, but it can be done without negative consequences.
been as much carbon separation built up over time as The main concern is to assure that the catalyst is processed
would be with a little longer run. We understand, as Mr. through the regeneration section without causing damage.
Emanuel has mentioned, that several other clients have The heel catalyst has high levels of carbon, 20 wt% to 50
used the technique successfully. wt%, and requires a number ofcatalyst regeneration cycles
to reduce this carbon level to normal, i.e., approximately
KELLER: 5%. This requires good operations monitoring of both the
We were one of the ones that used it successfully. In reaction and regeneration sections.
December 1993 we density separated one load of heel The CRI approach is viable and has had good success
catalyst into one light fraction that was less than 0.2 wt% of late. Communication with UOP during the unit reload
carbon, and one heavy fraction that was greater than 30 and restart is recommended.
wt% carbon. We have been using the light fraction as
makeup to the CCR since January 1994 with no adverse RIG ZIMA (Ashland Petroleum Company):
effects. In February 1994, we took our CCR down to correct
We have not ex situ regenerated heel catalyst. We sent a reactor section mechanical problem. We unloaded about
the high carbon (heavy) fraction to a platinum recovery
110,000 lb of catalyst. After the free flowing catalyst was
facility because tests showed that it would be difficult to
unloaded, we went in and manually unloaded the heel
burn off the carbon without overheating the catalyst.
catalyst. This heel catalyst was a mixture of high carbon
catalyst, which was 20 wt% to 50 wt% carbon, and low
CHARLES S. MCCOY (McCoy Consultants):
carbon catalyst, which had a carbon content of less than
How hot has the heel catalyst been? This material has
2 wt%. We sent the mix out to CRI for density grading.
been in a dead zone and not only has experienced a lot of
It was then separated into three grades. They returned
coking, but must also have seen a lot of heat. Has anyone
72% of the heel mixture as a low carbon portion. This had
done any laboratory work to determine the temperature
an average carbon content of 1.2 wt%. The midgrade and
to which it has been exposed?
the last grade, the high carbon catalyst, were also returned.
We reloaded the highest carbon catalyst back into the
ANGELO FURFORO (UOP):
reactor’s dead space first, then the midgrade into the
I am not sure whether Mr. McCoy is referring to
remaining dead space. The low carbon catalyst was loaded
temperatures the catalyst experiences in the reaction sec-
tion, or the regeneration section. Please clarify. into the last reactor, followed by the remaining 110,000
lb of catalyst.
CHARLES S. MCCOY (McCoy Consultants): Separation and reuse of this heel material and the low
While it is in the reactor building up this high coke carbon portion saved us about $185,000 in fresh catalyst
level, it must be very hot. How hot does it get? Have you costs. It also helped minimize the risk of high tempera-
tried thermogravimetric analysis or anything like that to tures in the regenerator and damage to catalyst pills. No
look at the temperature history? problems were experienced on start-up or in subsequent
operation.
ANGELO FURFORO (UOP):
There are limited commercial data concerning the tem- QUESTION 24.
peratures the heel catalyst experiences during normal proc- Have continuous reforming operators found it eco-
essing in the reaction section. However, both the limited nomical to separate the “whole pills” from the dust
data and UOP studies of vapor flow around the catalyst collector fines and reuse as makeup to the unit? If so,
in this area show that the catalyst is not hotter than the what separation method was used? Were the whole
rest of the catalyst bed. The flow around the catalyst pills pills regenerated ex situ before being added back to
is extremely good. It is not a low flow region by any means. the unit and if so, how?

1995 NPRA Q&A Session on Refining and Petrochemical Technology 139

TOC/INDEX

KELLER: Second, the “whole pill” fraction recovered will typi-

No, it is not economical for us to recover whole pills cally contain 4 wt% to 5 wt% coke. If this material is used
from the fines. We operate at the UOP design elutriation as makeup catalyst, it will return to the regenerator with
gas rate. Our CCR fines loss is about 8 lb/day. Chips and approximately 8 wt% to 10 wt% coke. Without an adjust-
whole pills makeup 25 wt% of the fines. Our potential menr in regenerator conditions, this catalyst will be dam-
incentive to recover whole pills is less than $15 a day before aged when it passes into the high oxygen areas of the
handling and regeneration costs. regenerator. This can result in phase-damaged catalyst,
which is very soft and inactive. The result is usually a
EMANUEL: significant increase in fines (which means more catalyst
We do not separate the whole pills and return as makeup will be required) with faster screen plugging and
makeup. more catalyst transfer problems.
However, in response to Mr. Frondorf’s suggestion, if
FRONDORF: the refiner has invested in the hardware to make the proper
Our Corpus Christi refinery has been recovering whole separations, UOP recommends that the whole pills be
catalyst pills from fines for many years. We do find this screened several times to ensure no smaller diameter par-
recovery to be economical on the whole pill percentage, ticle material (chips or dust) remains. UOP then recom-
typically 10% to 15% of the amount of fines retained mends that these whole pills be sealed in air-tight
before shipment back for recovery. A lot depends on the warehouse storage until the next regenerator screen clean-
ability to seal and store the fines in a dry condition and ing. During screen cleaning the recovered whole pills can
the ability of the screen operator to effectively make the be added back to the disengaging hopper manway so that
cut between whole catalyst and chips. We purchased a the catalyst coke is burned before the catalyst returns to
small portable screener a number of years ago to do this, the reactors. UOP believes this is preferable to ex situ
so we have already spent that small amount of capital to carbon burning the recovered whole pills since the condi-
do that. In our case, we do not regenerate the recovered tions are sure to be properly controlled in the regenerator.
catalyst prior to reuse, although this is desirable. We load
this catalyst directly into the surge drum. QUESTION 25.
Have refiners used spent molecular sieves for chlo-
RIC ZIMA (Ashland Petroleum Company):
ride removal? How do they compare to alumina?
We hold our CCR catalyst sweepings and fines until we
have between 20 and 30 drums then we send them down
JOHNS:
to CRI to undergo a density grading. We have found it
We have used alumina for chloride removal on reformer
economical, and we do reuse the whole pill catalyst. The
production hydrogen. It is my understanding that the
dust is sent out for platinum recovery. The whole pill
capacity of the alumina and the molecular sieve are about
catalyst is regenerated ex situ and returned to makeup
inventory for reuse. the same, but the disposal cost for alumina is much
cheaper. Both will generate some organic chlorides, but I
ANGELO FURFARO (UOP): understand there are some new adsorbents out there that
UOP commercial feedback from licensees is mixed. will not do so.
Most return the dust collector material for metals recovery,
and several have tried separating the whole pills from the JAMES SOUTHWORTH (UOP):
fines. UOP recommends that all material (fines, chips, and Molecular sieves have been used as HCI scavengers in
whole pills) collected from the dust collector(s) be re- reforming service for many years. Therefore, it should be
turned for metals recovery. possible to reuse spent drying/purifying molecular sieves
UOP is aware that a number of refiners choose to screen for HCI removal service. UOP would like to recommend
the material collected from the dust collector(s) and use that if a refiner is considering this, then they have the used
the “whole pill” fraction as makeup catalyst. There are two material checked for remaining adsorption performance.
risks associated with screening this material and adding We would suggest that the material be tested for capacity
the “whole pill” fraction of the dust to the unit as makeup and x-ray crystallinity; otherwise the installation of an
catalyst. First, the screened “whole pill” fraction usually untested material with limited remaining capacity could
still contains chips and dust, which lead to regenerator result in a short adsorbent life in HCI removal service.
screen plugging, reactor centerpipe plugging, and/or cata-
lyst transfer problems. On more than a few occasions, this MARK S. DORAN (ICI Katalco):
practice has been identified as a leading cause for such ICI Katalco can provide chloride adsorbents that do
problems. not form organic chloride byproducts.

140 Light Oil Catalytic Processing

TOC/INDEX

QUESTION 26. CHARLES McCOY (McCoy Consultants):

Discuss operating experience with stabilizer tower Mr. Frondorf, what do you do with this 18% to 20%
side draws designed for benzene removal. What is the benzene stream once you have it? What is it good for?
concentration of benzene in the side draw? How is the
side draw stream used? FRONDORF:
We feed that to our aromatics extraction unit.
HANSEN:
QUESTION 27.
Valero chose to install a separate reformate splitter in
There have been notable advances in semiregenera-
which a benzene-rich side draw is taken. The overhead
tive reforming catalyst technology. What has been
and bottoms products are recombined and routed to
refiners’ commercial experience in terms of
either blending or sales, while the side draw is sold to
a) activity,
others for benzene recovery. The typical benzene recovery
b) stability,
is 90% with a side draw purity of 30%. Adjustments to
c) yields,
both purity and recovery are made based on economic d) length of runs,
conditions. e) performance after several regenerations, and
f) use of catalyst tailored to reactorposition and conditions?
FRONDORF:
We debutanize or depentanize our CCR reformate and JOHNS:
our one fixed bed reformate first before splitting out a There are some new catalysts available that have im-
benzene-rich side draw in a second tower. I interpreted the proved yields at, hopefully, similar cycle lengths, depend-
question as whether anybody was making a side draw ing on which reactor you put them in and the reaction
directly off the debutanizer or depentanizer, and we do not mechanism that you ask the catalyst to perform. With the
do that. Typical results have been on the order of 3% to exception of maybe RFG mode of operation, it has been
5% benzene by volume in the splitter tower feed, with side our experience that if you have a good semiregenerative
draw streams ranging from 12% to 30% with a more catalyst that is working, you probably want to keep it.
typical amount of around 18% to 20% benzene in the side Hydrogen in the RFG mode may present a problem for
draw and minimal benzene in the bottoms. you. And that is where you may want to look at the new
catalysts.
SHEN:
Most of the design cases we have seen do not produce SOLIS:
a good concentration ofbenzene heart cut from a stabilizer We have in our Gibraltar Refinery a semiregenerative
tower side draw, as many people expected. The reason is reformer that has been revamped to increase capacity from
that if you put the side draw point above the stabilizer feed 25,000 bbl/d to 35,000 bbl/d. As you can imagine, we have
nozzle, the concentration of C5 s in the side draw will pushed the capacity to our optimum level for an overall
dilute the benzene concentration. The amount of C3/C4 refinery view, but we would like to improve the present
will also be too high, making it unsuitable for use as isomer performance of the unit in terms of hydrogen production,
or benzene saturation unit feed, unless you change the octane number, and aromatic production, especially level
ofaromatization of C8. We have an aromaticextraction and
stabilizer into a depentanizer operation.
fractionation system downstream of this unit. We have to
If you place the side draw below the stabilizer feed
pass C8 products through the solvent extraction unit to
nozzle, then the amount of C7 and heavier can dilute the
remove all C8 nOnarOmatic components.
benzene concentration. Even vapor side draw still cannot
For these reasons, we are tracking very closely the
overcome the C7+ situation. development of new reformer catalysts, looking for a
There are just not enough trays in the stabilizer to potential replacement for the existing bimetallic catalyst
perform the fractionation most refiners like to see. It is not (Pt and Re). We are testing in the pilot plant the perform-
an easy fractionation. Ideally, you would like to operate it ance of a new catalyst system. We are simulating in our
at low pressure to get good alpha ratios. The stabilizer experiments placing the last generation high activity (Pt-
cower pressure is too high for that. Ge) catalysts in the first and second reactor, and a high
The more common design is in two modes, i.e., using stable conventional bimetallic catalyst in the third reactor.
a separate tower to do the job. Mr. Hansen mentioned one Initial results are very attractive in terms of improving
design case with a side draw. In the other design case aromatization and hydrogen production. We expect that
people use a second tower and make a combination of C5 with this new catalyst approach we may be able to send
and C6 overhead and send it to the isomerization or C8 aromatics directly to fractionation without passing
benzene saturation unit. through the aromatic solvent extraction unit.

1995 NPRA Q & A Session on Refining and Petrochemical Technology 141

TOC/INDEX

VAN IDERSTINE: temperatures. Tail reactor operation results in similar yield

We have a 10,000 bbl/d semiregenerative unit, and in for R-72 and Pt/Re. However, with R-56, UOP’s maxi-
the next month we are going to be changing out the mum activity Pt/Re catalyst, the higher coke tolerance in
catalyst in our lead two reactors to one of the recently the tail position results in higher activity/stability of the
developed catalysts in hopes of higher activity, selectivity, SL system. R-72 full load stability is somewhat less than
and hydrogen yield. In addition to that, over the next year the staged load stability and as a result, UOP recommends
we are looking at doing a couple of other things to try to the staged R-72 system as superior in all cases to a full load
maximize our C5 + yield. Right now, our product separator of R-72. The two refiners that have regenerated this system
runs at about 385 psig, and we are investigating what we have reported second cycle performance equal to the first
have to do to drop the system pressure by about 100 psig. cycle performance. C5+ yields have been improved by 1
Also, we are considering decreasing our cycle time between LV% to 3 LV% above Pt/Re, while relative hydrogen yield
regenerations. Analysis of our situation suggests that re- has been improved on the order of 10% or more.
ducing the period between regenerations to 12 months Regeneration conditions are very similar to standard
from our current 24 months would be economical for UOP platinum/rhenium regeneration procedures.
improved yields.
M. DEAN EDGAR (Criterion Catalyst Company L.P.):
ROBERT MORSE (ACREON Catalysts): We have another system that uses two catalysts. PR9
ACREON has RG582, which is a trimetalic catalyst. It catalyst is an equal metals, platinum/rhenium, catalyst
is a balanced platinum/rhenium catalyst with a third with a modified base. It minimizes paraffin cracking, so
metal. It is now in operation in five units with another two we find we get about a 1% increase in yield of C5+, and
expected on stream soon. It has been regenerated in three maintain the hydrogen that has been produced. As good
of those units. The regeneration procedures are essentially as that is, we also recognize, as many of you do, the
the same as for balanced platinum rhenium catalyst. It chemistry involved in the reforming process. We find that
shows equivalent activity and equivalent or slightly higher if we tailor a special catalyst for the lead reactors we can
stability than balanced platinum rhenium catalyst. The maximize the rings in the product.
yield improvement on C5+ has been from 1% to 3% and With a PS7 and PR9 catalyst combination, we get
it has shown higher hydrogen yield and purity. The per- about a 2 vol% increase in our C5+ yield, 100 scf/bbl to
formance of the second and third cycles have been equiva- 200 scf/bbl higher hydrogen make, and, of interest for
lent to the performance of the first cycle. people who are recovering aromatics, a boost in xylenes.
This combination has been in use for about 2 years in
LIZ ALLEN (Criterion Catalyst Company, U.K.): cyclic operation. It has undergone hundreds of regenera-
In February 1995, we installed our skewed platinum tions, and we find this system to be very stable. The two
rhenium catalyst PR28 in a semiregeneration unit in catalysts are quite compatible as far as regeneration proce-
northern Europe. Since start-up, the catalyst has shown an dures and chloriding.
activity advantage of about 10°F compared to the previous
catalyst system which was also a platinum rhenium skewed JAMES D. WEITH (Unocal Corporation):
formulation. Previously, the best cycle achieved was about With the tailored catalysts, do they all reach end-of-run
9 months, but with PR28 in the unit, we are about 9 at the same time, or do you have a catalyst in one reactor
months into the operation and we have only now reached requiring regeneration before the catalysts in the other
the start-of-run temperature for the previous catalyst sys-
tem. With this we feel we are well on course to achieve
nearly double the cycle length that was possible with the
previous catalyst.

ANGELO D. FURFARO (UOP):

The patented R-72 Staged Loading (SL) scheme, as
presented in the 1995 NPRA paper number AM-95-30,
represents the highest activity, stability, and yield of any
semiregenerative system in the marketplace today. With
R-72 SL the use of catalyst is tailored to reactor position.
Optimum performance results from using the R-72 cata-
lyst in the lead reactors and R-56 in the back. As shown
in the paper, pilot plant modeling of lead reactor condi-
tions illustrates that the R-72 catalyst achieves higher yield
and higher octane than Pt/Re catalyst at equivalent reactor

142 Light Oil Catalytic Processing

TOC/INDEX

reactors? Also, is the regeneration procedure the same for KEN LOUDER (KTI Corporation):
the individual tailored catalysts in the various reactors, or The Zeoforming forming process uses the L-Zeolite
is a separate regeneration procedure required for each catalyst. Looking at motor reformate, we have found that
reactor? higher boiling range feeds will lay down more coke than
competitive reforming catalysts. The C5+ yields are about
ANGELO FURFARO (UOP): 20% lower, but the C3+ yields are about 22% higher than
UOP’s R-72 Staged Loading has been loaded in four conventional reforming catalysts when operating in the 85
semiregeneration units. The carbon profile at end-of-cycle to 98 octane range. Our information indicates it will
is consistent with that of platinum/rhenium catalyst. tolerate up to 1000 ppm sulfur. As far as we know, no
There are no unusual carbon profiles across the reactors. alumina-based units have been converted to use Zeolite
catalysts for light naphtha reforming. In summary, it seems
QUESTION 28. to me that if you want C3s and C4s and possibly hydrogen
L-Zeolite catalyst has been recommended for light at the expense of reformate, that is your catalyst. We do
naphtha reforming. Please cite commercial experi- not see that much application in the United States.
ence on the following:
a ) What is the process sensitivity to feed boiling range? DAVID L. SMITH (Alto Industrial Chemicals):
b) How do yields compare to alumina-based reforming On the question of the contaminants in the feed naph-
catalysts? tha stream, solid adsorbents can be effective for removing
C) What are recommended feed contaminant limits for water, oxygenates, sulfur compounds, and some nitrogen-
sulfur nitrogen, water, etc. ? based compounds from the feed naphtha. Refineria Isla in
Curacao is in the process of installing a bed of activated
4 Have alumina-based units been converted to use these
alumina and Selexsorb CD selective adsorbent to remove
catalysts?
water and phenols from the feed naphtha stream to their
catalytic reformer.
SOLIS:
We have under design and construction a 12,000 bbl/d
ANGELO FURFARO (UOP):
light naphtha reformer. All the information that we will
To reply to Mr. Louder’s comments, the L-zeolite cata-
supply here is based on pilot plant and basic engineering
lysts under discussion, such as RZ-100, show totally dif-
work. The feed consists of two sources. We split the
hydrotreated naphtha coming from the existing hydro- ferent characteristics from those used in the Zeoforming
treater into two cuts. We plan to send the C8+ cut to the process. The most significant advantage RZ-100 has, rela-
existing semiregenerative reformer, and the C6 and C7 will tive to alumina-based catalyst, is for processing hexane and
be sent to the new light naphtha reformer. The other part heptanes. Selectivities to aromatics are factors of 2 to 4
of the light naphtha feed is raffinate coming from the times greater for RZ-100 than state-of-the-art alumina-
existing aromatic solvent extraction unit. With this new based reforming catalysts. For heavier feeds with end
reformer, we expect to produce basically benzene, hydro- points in the 340°F to 360°F range, selectivities are equal
gen, and toluene. It is really a hydrogen production unit; to or better than state-of-the-art CCR. Even greater ad-
we plan to produce 61 tons of hydrogen per day. Also, we vantage is observed relative to semiregenerative reforming.
will produce 170,000 tons of benzene and 88,000 tons of The feed contaminant limits are similar to standard re-
toluene per year. For our specific feedstock conditions, if forming. The sensitivity of this type of catalyst to sulfur is
we compare the yields in hydrogen, benzene, and toluene well known.
with a standard semiregenerative reformer, hydrogen pro- UOP has a platforming catalyst, R- 15, which is used in
duction from the light naphtha unit will be 4% hydrogen various locations around the world where LPG has a very
versus 1% with the standard reformer, the benzene yields high value. It is a zeolitic catalyst designed to make a lot
will be 38% versus 7%, and toluene 20% versus 15%. For of C 3 /C 4 at the expense of reformate. The reformate
hydrocarbons higher than C7, the yields are similar be- octane is of gasoline quality, while providing significant
tween the L zeolite and alumina-based catalyst. The feed quantities of LPG for local consumption.
must be sulfur- and water-free; a specific guard has to be
provided to meet these requirements. QUESTION 29.
Generally speaking, it is feasible to retrofit an alumina Has anyone converted a semiregenerative reformer to
reformer to a light naphtha unit. The most important step continuous reforming? What hardware had to be
is to carry out a very specific and stringent cleaning added or rep/aced? Has anyone added a continuous
operation to ensure the total removal of all sulfur and regenerated final reactor as a compromise to full unit
sulfur compounds that could be deposited in piping ves- conversion? How does performance compare to a
sels or any other equipment of the unit. continuous regeneration unit?

1995 NPRA Q & A Session on Refining and Petrochemical Technology 143

TOC/INDEX

A new system, named Dualforming Plus, uses the same

new equipment, but is installed downstream of the sepa-
rator. This effectively decouples the existing reaction train
from the new one. Therefore, it minimizes downtime,
typically by about a week. You have no problem with
furnace retubing and so forth. It allows you to upgrade
two reformers at the same time. Typically, comparing a
semiregenerative unit and a new modern low pressure
reformer with the Dualforming Plus, you get 60% of these
benefits for only 30% of the cost. For example, a major oil
refiner in Europe found that the total cost for a 25,000
bbl/d grassroots, low pressure reformer was $100 million
a year, while the Dualforming Plus was only $30 million.

QUESTION 30.
We have excess reforming capacity and want to add
MORGAN: butane to the reformer feed to isomerize n-butane to
Nine refiners have converted their side-by-side reactor isobutane:
semiregenerative units to UOP CCR platforming units. a) What ratio of iC4-to-nC4 can be achieved?
The downtime required to convert a unit is approximately b) Is there any adverse impact on the naphtha octane and
30 to 35 days. The revamp costs can range from 45% to yields?
75% of the cost of a new unit. The benefits are: increased c) Is there anything special we can do to enhance iC4
reformate yield of approximately 5 LV%; increased hydro- equilibrium?
gen yield ofapproximately 360 scf/bbl; decreased RVP and
benzene production, due to higher catalyst selectivity at ABRAHAMS:
reduced operating pressures; and increased unit reliability, Test runs were performed at one of our refineries to
since the CCR platform unit operates 360 days per year investigate this idea. A small stream of normal butane was
as a result of the continuous catalyst regeneration. Four fed to the platformer. Unfortunately it proved difficult to
more UOP conversions are in design and construction, find evidence of the extent of isomerization at such high
and these benefits mentioned assume that no revamps to reactor temperatures and throughput. Pilot plant results
heaters were made and that the RONC and throughput from Refinery Process Services, and data we have received
were held constant. The recycled gas compressor through from UOP provide conflicting results. RPS has told us of
the unit was held constant and not decreased. 20% conversion of normal to isobutane. UOP reports that
no conversion was seen. Until more information is avail-
KEN LOUDER (KTI Corporation): able, we are not planning to continue field tests.
KTI designed and engineered the addition of a con- As for part C, as far as I know, the main thing is that
tinuous regenerated fifth final reactor using IFP technol- the temperature in reforming is too high to support that.
ogy. The modifications included the fifth reactor and You can consider revamping an idle unit to isomerization,
heater, a sulfur guard bed, reactor effluent feed preheat or you can create a parallel reactor system and piggyback
exchanger, some feed filters, and a chilling system for it off your recycle system and recovery section.
better C5+ recovery and higher hydrogen purity. The unit
was designed for a 2-year run in the existing semiregen- JOHNS:
erative reactors operating at 98 RON. It is a 150 psi unit I confirm what Mr. Abrahams says. The equilibrium of
processing 30,000 bbl/d of 365°F end point feed. normal to isobutane at 900°F is very poor. The other
problem you may have if you try to inject a lot of butane
JEAN-LUC NOCCA (IFP, Enterprised/HRI, Inc.): into the reformer feed is difficulty controlling the unit
IFP proposes two types of hybrid systems that are with the LPG mixed with the reformate.
combinations between semiregenerative units and con-
tinuous reforming. The first type, called Dualforming, SHEN:
involves the installation of a new reactor and heater with I agree that the temperature is not favorable for C4
a new regenerator, which is installed in series with the equilibrium. In addition, if nC4 is added into reformer
existing reaction train. This results in an increase in octane, feed, a couple of things might happen besides the change
an increase in cycle length, and an increase in yield. We in reactor yield. First, additional C4 will add additional
currently have four units in operation using this system load to the debutanizer or the stabilizer tower. It might
accumulating over 100,000 bbl/d of feed. have negative impact on your reformate RVP.

144 Light Oil Catalytic Processing

TOC/INDEX

Second, if you put a lot of nC4 into the reformer feed, guard catalyst. They did not detect it until the reformer
the recycle gas hydrogen concentration might change. was upset with sulfur breakthrough. What easy, eco-
Conduct a test computer run to qualify that. We put 8% nomical, and reliable methods are in use to avoid fhis
of the nC4 into the feed and found that for a semiregen- problem? How often do refiners review these data to
erative unit at 300 psig pressure, the recycle gas hydrogen assure this situation does not occur? In answering this
concentration dropped from 89% to 86%. question, please give specific sampling schedules
To take advantage of the excess capacity of the reformer, and analyzer/fest types.
we believe that it may be worthwhile to investigate the
option of converting one of the reformer reactors into an MORGAN:
isomerization reactor and having an integrated isomer/re- It has been my experience with sulfur technology that
forming unit. the adsorber does work well in removing trace quantities
of sulfur. It holds the sulfur, and it tends to be forgotten
SOLIS: about through time because it does not create any particu-
In addition to what has been said, related with unfa- lar problems. When there are unit upsets, such as a power
vorable isomerate equilibrium at reforming operating failure with an already loaded sulfur adsorber bed, the
remperature, I would like to point out that butane isomeri- adsorber beds have been known to break through and slug
zation may have some kinetic constraints because of the the reformer with a large quantity of sulfur and create
higher coke level on a reformer versus an isomerization major operating problems on the reformer. In my opinion,
catalyst. As a result, the thermodynamic equilibrium will this situation can best be cured by eliminating the sulfur
be difficult to achieve. Further addition of butane to the adsorber and instead choosing to remove the sulfur in the
reformer, keeping constant the amount of naphtha, will hydrotreater where it can be controlled.
increase the naphtha space velocity, which will have a
detrimental effect on the reformer yield and octane. ABRAHAMS:
Finally, if excess reforming capacity is available to allow
Operators at our plants have used lead acetate tests to
an additional amount of butane, it will be necessary to
look for upsets when they are really in trouble. Otherwise,
consider the debutanizer capacity.
we use laboratory testing between 1 and 3 times a day to
look for a sulfur breakthrough. Our laboratory uses a
LEE E. TURPIN (Honeywell Profimatics, Inc.):
Houston Atlas analyzer. While we have not used it on
Adding nC4, to virgin naphtha reformer feed will mar-
reformer feed, we have installed a Houston Atlas analyzer
ginally impact reformer operations. Generally speaking,
on-line on an isomerization unit to check for feed quality.
the nC4 will approach an equilibrium with about 45% of
the mixed butane being iC4 . The equilibrium value is
JOHNS:
temperature sensitive, and the approach to equilibrium is
space velocity sensitive. Adding the nC4 will marginally I believe one of the most reliable methods in sulfur
reduce the space velocity, reducing the equilibrium ap- adsorber technology is to have 2 vessels of sulfur adsorbent
proach. But this is counteracted by an increase in space in alternate lead/lag position operation. You can also put
velocity, which will require higher reactor temperature, a sample point halfway down the vessel and change out
which in turn will marginally increase equilibrium value. the catalyst when breakthrough is detected out of the first
The bottom line is, I do not think you will be able to see vessel or at the halfway point.
a difference in the iC4 to-nC4 ratio when small amounts I have recent experience with a couple of adsorber
of C4 are added to a virgin naphtha feed. operations. One is on a heavy hydrocrackate stream. The
I used a kinetic model to review the change in unit other is off the stripper of a coker naphtha hydrotreater.
operation and found that if I added 6% normal butane to We use the Houston Atlas sulfur analyzer, and we are able
the feed, the reactor average bed temperatures would to detect very low levels of sulfur; we typically have seen
increase by about 0.3°E The recycled molecular weight 0.5 ppm to 1 ppm coming in. Typically, with a particular
would increase by about 0.35. Because of the increased adsorbent, we will get a couple of years of life. There have
recycled molecular weight, the average weighted average been times when that sneaked up on us, and occasions
inlet temperature is actually decreased by about 1°F. where sulfur has come through without good detection.
I want to tell you about a little testing tip that you can
QUESTION 31. learn from experienced operators. You ask them to test the
We are aware of a refiner that used a “sulfur sorber” gas or liquid coming out and they will stick a draeger tube
technology downstream of the naphtha hydrofreater in and let it bleed through for a long period of time. If
(upstream of a reformer). Hydrotreater unit upset there is sulfur present, over time it will show up. This is
conditions resulted in a high level of sulfur leakins something that may not be very quantitative, but it sure
onto the sulfur sorber, completely saturating the sulfur will tell you if you are getting some shots of sulfur .

1995 NPRA Q&A Session on Refining and Petrochemical Technology 145

TOC/INDEX

MARK S. DORAN (ICI Katalco): the high quality metal seated ball valves were expected to
In regard to the monitoring, I would like to make three be better than the original plug valves. However, the results
separate points. The first is we would recommend daily showed no marked improvements. Later we had success
monitoring by mass balance of the weight of sulfur re- with a two-motion sliding gate valve, which swings across
moved onto the sulfur guard bed. If you couple this with the opening and then pushes up against the seat for a tight
information from the adsorbent supplier about the capac- seal. This valve replaced one of the plug valves in the
ity of the particular material for sulfur, then clearly you typical double block and bleed assembly, Because of the
can predict when the end of bed life will be so that it will extensive piping changes required, only two sliding gate
not come as an unpleasant surprise. valves were installed (one at reactor outlet and one at
As part of that, it is worth noting that on the sulfur regenerator outlet). This change, along with more plug
adsorbent itself, the capacity for sulfur is affected by factors valve repair and testing efforts at the local Hills McCana
such as the operating temperature, the space velocity and service center, has greatly improved our reliability. We are
even the type of sulfur species to be removed. Therefore, now happy with hydrogen isolation valve performance.
I really do recommend that people operating these beds (Background: The two motion valve is made by the Ever-
speak closely with the suppliers so that they get the most lasting Valve Company.)
reliable data.
The second point I would like to add is that to achieve ABRAHAMS:
a reliable mass balance, you do need a reliable and accurate In one location we have been working closely with a
low level sulfur analyzer. We recommend one of the ana- local repair shop to fine-tune factory recommended repair
lyzers from Houston Atlas. And the final point, to check procedures. This effort has been successful in reducing the
that your prediction of the end of life of the bed is correct, number of valve failures. From January 1994 to April 1994
it is possible to follow the sulfur adsorption mass transfer there were 7 failures, and since that time there have been
profile as it moves down the length of the bed. This can only two. The unit presently uses Kamyr Neles ball valves.
be done either by having a process stream sample point at The only new valves tested have been an Argus valve and
the mid-point of the vessel, or, alternatively, looking for a Fisher 1032 valve. The Argus valve was installed in 1990.
H2S in the offgas from the reformer stabilizer. You will Its actuator failed; the valve itself was fine. It was replaced
find that H2 S concentrates there to a far higher extent with. a Fisher valve. The Argus valve has been recondi-
than, for example, on the reformer recycled gas. Again,
tioned, and we are going to test it on the spent catalyst side
you have a very quick and easy way of predicting when the
at the next opportunity.
mass transfers zone for sulfur is starting to sneak out of the
bottom of the vessel.
EMANUEL:
At Fina, packing leaks seem to be our most serious
Mechanical
problem with ball valves in catalyst and hydrogen service.
To improve packing performance and provide better load
QUESTION 32.
distribution on the packing, the packing gland was modi-
What has been done to improve ball valve perform-
ance in continuous reformer service? What are your
fied from a two-bolt to a four-bolt design. In addition, a
a) repairprocedures? b) new valve selection criteria?
monthly valve inspection maintenance program was initi-
ated to maintain proper packing loads and valve opera-
FRONDORF: tion. All of our valve repairs are presently performed by a
We currently utilize Hills McCana ball valves in all valve manufacturer approved facility. To date, we seem to
three of our CCRs at the Lake Charles and Corpus Christi be gaining ground, and we feel we have made significant
refineries. We are in the process of changing to Argus improvement with this four bolt design.
valves consisting of arguloy ball and seats with 316 stain-
less steel bodies. We have changed a number of valves over PATRICK HOH (HRI, Inc):
the past year at both refineries with excellent results to For the IFP continuous reforming unit, two special ball
date. The repair of the Argus valves will be handled by a valves were developed by Omri and IFP, also by Calabry
local shop certified by Argus. and IFP. These ball valves are very reliable and have
excellent service records. The valve manufacturer includes
KELLER: the repair procedure with these valves. The advantages of
We are happy with the V-Ball catalyst circulation valves. these ball valves are that they are gas tight in closed
Our concerns were with the hydrogen isolation valves, position and they also require minimum instrument air
which are prone to leaks. We conducted a series of tests on pressure to operate. Regarding the new valve selection
various ball valves. We also increased our repair and moni- criteria, the valves should be suitable for the intended
toring efforts to track the results of the changes. Initially operating conditions and they should be durable.

146 Light Oil Catalytic Processing

TOC/INDEX

ANGELO FURFARO (UOP): SHEN:

The current UOP specifications require only full port I agree with the previous comments. One small advan-
ball valves to minimize obstructions to the flow of catalyst tage we found for dense loading in the radial reactor is,
when open. The many approved vendors use various com- since the catalyst is more compact during the loading
binations of hard coatings and hard materials of construc- process, the phenomenon of the catalyst “sagging” close
tion for the ball and seats to increase the life. An alternate to the end of run does not occur. This cuts down the
rotating disc valve type has also been successfully used. chance of bypassing the catalyst bed.
For UOP CCR platforming licensees, if the valve speci-
fications are out of date, more than a few years old, you ANGELO FURFARO (UOP):
may simply contact the UOP District Service or Des UOP recommends dense loading extruded catalyst,
Plaines Technical Service representative for an updated independent of reactor type. Spherical catalyst may also
supplier list. UOP has increased the number of approved be dense loaded, but with a smaller increase over bulk
suppliers, to 6 or 7, providing a good selection of source, density or sock loaded density. The following table shows
support, and availability. the expected density increase for dense loading.
% Increase Over % Increase Over
QUESTION 33. Bulk Density Sock Loading
What are the catalyst loading recommendations for
Spherical Catalyst 3 to 5 ~ 10
reformers?
Extruded Catalyst 8 to 10 ~ 15
a) Compare dense loading versus sock loading for down-
flow and radial reactors and spherical and extruded Dense loading results in more pounds ofcatalyst loaded
catalyst; and thus a longer run length. Comparing a dense loaded
b) What density differences are expected for spherical and extrudate to sock loaded sphere, the pressure drop in-
extruded catalysts for different loading methods? creases by about 2% to 6%, and the vapor maldistribution
C) What are the process advantages, if any, for dense decreases approximately 4% to 10%.
loading in terms of flow distribution and/or reactor
activity? RICHARD QUINN
N (Catalyst Technology, Inc.):
d) How much is reactor pressure drop increased by dense Based on our experience in loading reformers, we
loading? would recommend dense loading. Or, if you can construct
a suspended loading platform, you can then use the sock
SOLIS: method. Dense extrudate loading for reformers will give
On our four reforming units all reactors are radial. We you 5% to 20% density increases over the compacted bulk
have experienced both types of catalyst loading. We have density. For spherical catalysts you will get 1% to 2%,
operated with dense loading with extruded catalyst and mainly because of some breakage.
sock loading with spherical catalyst. Generally speaking, The process advantages for dense loading are that if you
dense loading results in 7% more catalyst by weight. For run your loader correctly, you will have an even load with
a given octane number, we start the run with lower reactor no convex or concave surfaces, and you should get much
temperature in the case of dense loading because the space better flow distribution. The pressure drop would be
velocity is lower. We have experienced a run length with correlated with the density gain.
this type of loading from 5% to 10% longer. This advan-
tage must be compared with the capital cost of increasing
platinum inventory. But frankly speaking, with the variety
of crude oil that we normally run, it is difficult to appre-
ciate the advantages of these two different types of catalyst
loading. Finally, in radial reactors we do not have the same
differences in reactor pressure drop between dense and
sock loading.

MORGAN:
Dense loading results in more pounds of catalyst loaded
and thus a longer run length. Comparing a dense loaded
extrudate to a sock loaded sphere, the pressure drop in-
creases about 2% to 6%, and the vapor maldistribution
decreases 4% to 10%.

1995 NPRA Q&A Session on Refining and Petrochemical Technology 147

TOC/INDEX

QUESTION 34. We have also noticed that sometimes we find chlorides

What is industry experience with fired heater reboiler in our TEG unit. We measure chlorides in the solvent even
return line isolation to protect against depressuring the though we are using carbon and sock filters. We have
entire stabilizer column in the event of a tube failure correlated some of our reboiler fouling problems with the
(automatic isolation valves, check valves, other)? chlorides. We do a daily check on our solvent pH, and if
we experience a pH drop, then we add MEA.
EMANUEL:
We continue to use standard check valves on our re- SOLIS:
boiler return lines for isolation. Of course, one concern is We have had two Shell sulfolane units operating since
that with the extended run times that we are now experi- 1969. We have experienced corrosion and solvent degra-
encing, there are fewer chances to inspect these critical dation when air has entered into the extraction and solvent
check valves. recovery sections. The oxygen promotes sulfolane degra-
dation to sulfur compounds, causing solids that will create
JOHNS: fouling in exchanger and mechanical seals, poor opera-
I know of good experiences with the use of remotely tion, etc. By avoiding air entry into the vacuum areas of
operated isolation valves in this service. In a case where we the unit, we have not experienced any more corrosion or
had a fire on the reboiler, the isolation valve was success- solvent degradation problems in this area. But we still have
fully closed remotely on the reboiler return, minimizing problems in the water stripping system. In one of the units
damage to the heater. we have completely replaced the water stripper and the
water stripper reboiler.
SOLIS:
My experience is different. In the hazop studies we have EMANUEL:
conducted which referred specifically to fractionation of We operate a sulfolane unit, and when we have solvent
light products, our decision has always been to cut auto- degradation we experience an acceleration of fouling in
matically the line between the bottom of the column and the towers. But the major problem area is corrosion in our
the reboiler. The main reason is that any problems with stripper reboiler and recovery column reboilers.
mechanical seals of pumps or leaks in equipment located
on the ground level could create a greater problem. Our KENNETH G. TASKER (HRI, Inc.):
hazop policy, following the insurance company’s and the As stated, corrosion in the solvent extraction unit is due
safety authorities’ recommendation, is not to go with an to the buildup of acidic compounds formed by the decom-
approach of total isolation of the lines coming and going position and degradation of sulfolane. Formation of these
from a reboiler to a tower. At very least, a relief provision compounds is accelerated at temperatures greater than
should be installed in the transfer line. 18O°C, or 356°F, and the presence of oxygen, of course,
also accelerates degradation. Generally hot spots are the
primary cause of thermal degradation. HRI designs the
D. Aromatic Extraction unit to minimize this effect by utilizing medium pressure
steam in reboilers. Since part of the process involves a
QUESTION 35. solvent recovery tower and a solvent regenerator that
What problems are encountered in solvent extraction operate on the vacuum, oxygen ingress is always a possi-
units with corrosion and solvent degradation? bility and a concern. Therefore, a very tight system is
required. Avoiding hot spots and fully minimizing oxygen
PARKER: ingress, plus using a continuous solvent regenerator to
We operate two aromatic extraction units at Sweeny. remove any higher boiling acidic degraded products, and
One uses tetra ethylene glycol (TEG) as a solvent and the having a good pH control system for the circulating
other uses sulfolane. Normally, the sulfolane unit runs in solvent and the water streams, ensure virtually corrosion-
C7/C8 service and the TEG unit runs in C6/C7 service. free operations. Based on our experience, our systems work
Typically, our problems have been in the reboilers. This is very well, with some of our clients rejecting less than 50
the stripper column reboiler at the TEG unit. On the to 100 gallons per year of degraded solvent. In a few
sulfolane unit, we have a recovery column in addition to instances, periods in excess of 6 months have been
the stripper, and have had corrosion on both reboilers. achieved without any solvent rejection.
The primary source of contamination is typically oxy-
gen. This causes an acid type corrosion in the unit. Our LEE E. TURPIN (Honeywell Profimatics, Inc.):
sulfolane unit runs at a vacuum on the backend recovery I like the way this problem is worded because it links
system, so we have a lot of problems with air leakage if we “corrosion” and “solid degradation.” In my experience in
do not keep on top of it. running a sulfolane unit, we found the heaviest corrosion

148 Light Oil Catalytic Processing

TOC/INDEX

occurred under areas of degraded solvent deposits. I agree liquid/liquid extraction with extractive distillation. Ex-
that preventive measures are the best solution, but that tractive distillation by itself is superior only in situations
does not solve al1 the problems because people make where a high purity benzene is the only product required.
errors. We installed a full flow cartridge filter for use when The solubilities and so on that Mr. Morgan mentioned
problems started. This enabled us to remove degraded make up basic chemistry. With the sulfolate extraction
solvent particles before they deposited. Having clean sol- process operating with reformate and pyrolysis gas feed
vent also seemed to make it easier to get the solvent streams, benzene and toluene purities as high as 99.999%
regenerator system working properly. have been consistently achieved in our units. And xylene
purities in excess of 99.6% have been consistently achieved
QUESTION 36. along with extremely high recoveries.
Are there any comparisons available on aromatic
extraction technology, specifically extractive distilla-
tion versus liquid/liquid extraction? E. lsomerization
MORGAN: QUESTION 37.
The use of extractive distillation alone versus the com- Has anyone operated both regenerable and nonregen-
bination of a liquid/liquid extraction and extractive distil- erable alumina-based isomerization catalysts? What
lation is determined by the boiling range of the feed and is the advantage in regenerating isomerization cata-
the purity recovery requirements of the application. To- lyst?
day’s product requirements and market conditions typi-
cally dictate the need for very high purity BTX products, MORGAN:
as well as high recoveries. Yes, I have experience with both the regenerable and
In liquid/liquid extraction, the relative solvent selectiv- nonregenerable isomerization catalysts. My experience
ity between hydrocarbon groups is in the following order:
with the regenerable isomerization catalyst has not been
aromatics first, then naphthenes, and then paraffins.
good. In fact, regenerations have failed in every case,
Within a particular family of hydrocarbons, lower carbon
although we did get some run length from a regenerated
number species are favored over those of higher carbon
isomerization catalyst. The problem seems to be with the
number. Therefore, the heavy nonaromatics can be sepa-
complexities of dryer designs and regeneration system
rated easily.
designs and the general sensitivity of this type of catalyst.
In extractive distillation, the relative solvent selectivity
The regenerable unit that I had experience on was con-
between the groups remains in the same order: aromatics,
verted to a nonregenerable unit and has performed very
naphthenes, and paraffins. However, the higher carbon
well since that time. I imagine there are people who may
number species are favored over those of a lower carbon
have different opinions or even different experiences.
number within a family of hydrocarbons. Therefore, the
lighter nonaromatics are easily rejected.
FRANK HIMES
S (UOP):
Extractive distillation by itself may be acceptable for
extraction of a single aromatic component, such as ben- Many non-UOP units that used regenerable chlorided
zene alone, but it may be limited in the purity and recovery isomerization catalysts have either been shut down or
that can be achieved for a wider boiling range material converted to utilize UOP I-8 catalyst. It is our under-
unless extensive prefractionation and postfractionation standing that the regenerations were generally problem-
facilities are provided. For these stringent applications, atic, requiring long shutdown periods and questionable
liquid/liquid extraction in combination with extractive recovery of catalyst activity. There are UOP zeolitic
distillation has had wider commercial appeal. isomerization units that are regenerable and have had good
It is by the synergy between liquid/liquid extraction and regenerations; however, these are not as high conversions
extractive distillation that one is able to effectively reject as I-8 units.
nonaromatics across a wide boiling range, simultaneously
realizing very high purities for each of the benzene, tolu- JEAN-LUC NOCCA
A (IFP Enterprises/HRI, Inc.):
ene, and xylene products, while at the same time achieving We provide a regenerable catalyst, and I do not know
very high recoveries of aromatics. of any units where this catalyst has been replaced. The
regenerable catalyst offers the benefit of not replacing the
KENNETH B. TASKER
R (HRI, Inc.): catalyst in the case of contaminated feedstock. As you al1
I am in total agreement with Mr. Morgan. Liquid/liq- know, these catalysts are very sensitive to a number of
uid extraction is superior to extractive distillation when contaminants. We know, for example, of a company that
high purity benzene and toluene products, or BTX prod- regenerated this catalyst four times, which led to substan-
ucts, are required. Sulfolane extraction combines both tial savings in catalyst cost.

1995 NPRA Q & A Session on Refining and Petrochemical Technology 149

TOC/INDEX

QUESTION 38. (during a general refinery turnaround), and we noticed no

lsomerization unit offgas scrubbers are usually made new corrosion.
of carbon steel. Do these vessels provide satisfactory
service without subsequent c/adding or coating with FRANK HIMES (UOP):
more corrosion-resistant materials? As of today, there are over 100 UOP isomerization units
with carbon steel scrubbers in acidic gas scrubbing service
KELLER: and only a few have seen it necessary to go with other than
The original carbon steel unlined vessel did not give carbon steel to keep corrosion rates reasonable. I think the
satisfactory service. After a little over 3 years of service, the main difference from other similar applications of acid
vessel had corroded as much as 100 mils at the nozzle to base neutralization is that the neutralization requirement
shell welds and showed severe spot corrosion at the “toe” here is highly predictable, because there is normally precise
of the shell weld seams. The vessel was repaired and control of the rate of acid formation in the process. There
internally coated with epoxy. Since the application of the is normally a constant and a very carefully metered injec-
coating (and subsequent maintenance of the coating every tion rate. There are some exceptions, such as with an
2 to 4 years), the vessel has required only minor repairs. injection pump that drifts away from its normal rate.

FRONDORF: QUESTION 39.

We also did not have a favorable result. After about 2 What level of C 7 S is acceptable in the feed to an
isomerization unit, and does the higher level cause
years of service, the bottom of the scrubber was weld
coking? What is the effect on catalyst fife?
overlaid with type 309L stainless up through the transition
cone of the vessel. The recommendation from the licenser
VAN IDERSTINE:
was Monel overlay, but we chose the stainless route based
We run the feed through our isomerization unit at 1%
on the recommendation of our metallurgist. An inspec-
to 2% C7S, which is quite a low level. We have only on a
tion of the vessel this past week has shown that the weld
rare occasion, in an upset condition, experienced high
overlay has held up well, but that the region immediately
levels, up to around 9% or 10%. During that period of
above it is in need of overlay as well. When we finish this
time, we experienced reduction in the octane of the isom-
overlay, this will leave us with overlay up through the
erate, notably more gas production of the stabilizer, and
bottom pack section above the secondary caustic distribu-
increased RVP in the isomerate.
tor.
As for the effect of the higher C7s on catalyst life, we
really cannot speak of that because normally we control to
VAN IDERSTINE:
such a low level.
We have had good experience with our carbon steel
offgas scrubber that was fabricated in 1986. There is no SOLIS:
coating or cladding in the scrubber, and we have only We have extensive data related to this subject because
experienced minor shell pitting. We believe that the caus- we have developed our own isomerization catalyst that is
tic distributor at the top of the vessel protects the shell and commercialized by Siid Chemie - UCI. First of all, I
the packed section of the tower from contact with the would like to distinguish between C6/C5 mixture isomeri-
offgas. The only corrosion ofsignificance that we have had zation and only C5 isomerization. C7 content is much
is on our Monel inlet gas sparger, which we had to replace more tolerant in the C5/C6 because the conditions for C5
in 1993 after only 6 years of service. Notably, it is impor- isomerization are more stringent. For C5 and C6 feed, we
tant to keep the caustic temperature below 150°F to do not have problems until 5% of C7. In the case of C5,
prevent caustic embrittlement. we do not feed more than 2% because higher levels will
create enough cracking to temporarily deactivate the cata-
IBRAHIM M. AL-HATOW (National Oil Distribution): lyst, requiring a catalyst regeneration in a shorter time
We experienced severe corrosion and pitting within the period. Nevertheless, the catalyst recovers its original ac-
first few months after the commissioning of the isomeri- tivity.
zation unit. We investigated the problem and found out The following chart plots the impacts of C7 component
that the reason was the drop in caustic soda concentration. in C5 isomerization in terms of gas production.
On two occasions the concentration was below 1 during
the changeout, which resulted in subjecting the scrubber ABRAHAMS:
to acidic medium. We recommend to operations to change The unit design of our isomerization unit calls for less
out spent caustic soda before its concentration comes than 1.5 wt% C7 + material feeding the nonprecious metal
down to 3%, to be sure the medium is always alkaline. The catalyst in our Union Carbide designed Hysom unit.
scrubber was checked last week after 2 years of operations However, experience has shown this to be a mid-run

150 Light Oil Catalytic Processing

TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

A. Alkylation-Sulfuric Acid Most of my activities center around our hydrocracking

alliance with Mobil and Akzo Nobel, MAK Hydro-
Process cracking, and our alliance with ER&E for sulfuric acid
alkylation.
Question 1. I am thrilled to be part of this panel and part of this
What are typical levels of oxygenates in Alkylation Q&A session. My hope is that all of us will go away from
Unit Feed from an MTBE unit using a standard water this Q&A with something of value to help us in our jobs
wash for oxygenate removal? Is there any commercial and help our companies improve their bottom lines.
experience with an on-line moisture analyzer up- Our information for oxygenate content of raffinate
stream of the alkylation unit? from an MTBE water extraction step is 1200 ppm to 1600
ppm oxygenates. That is in addition to the water of
JOHNSON: saturation. The raffinate is then used as olefin feed to a
UOP would expect around 1000 ppm oxygenate to sulfuric acid alkylation unit without further water wash.
come through a simple water wash with the load to the On entering the alkylation unit, the feed is chilled by heat
water wash somewhat dependent on the MTBE unit exchange, and a coalescer is used to remove free water. We
design. When followed by a depropanizer and an oxygen- would not expect to see a moisture analyzer on this stream
ate removal unit, the oxygenates can be reduced to less when used as a sulfuric alkylation feed stream.
than about 1 ppm. In HF alkylation or in the Topsoe fixed-bed alkylation
For an HF alkylation unit the typical design would process, which has been jointly developed by Kellogg and
allow up to 10 ppm oxygenates. Dimethyl ether is the Haldor Topsoe, oxygenates as well as dissolved water are
species that gets through. This will go overhead in the removed from feed in driers using molecular sieves or
isostripper and also in the acid regenerator on the alkyla- alumina as adsorbents. These driers utilize moisture ana-
tion unit and tend to build up in the acid. So DME is the lyzers to determine the end of the adsorption cycle and to
problematic species. initiate bed regeneration.

KIMBRELL: HUNKUS:
We have measured between 50 ppm and 200 ppm We also believe the typical levels of oxygenates in olefin
dimethyl ether and water. We track this with an on-line feed downstream of a typical MTBE unit run between
analyzer. We have a coalescer downstream of the oxygenate 500 ppm and 1000 ppm. I also have a friend who ran his
unit to try to prevent any water carryover into the alkyla- unit at lower levels, including Methanol at less than 200
tion unit. ppm; MTBE at less than 50 ppm; and DME at less than
100 ppm.
ARNDT: We feel a moisture analyzer on most typical units is a
We feed MTBE raffinate to alkylation plants at our 3 lower priority than good design and operating practice. A
major domestic refineries. However, we rarely analyze typical unit has a large recycle isobutane stream that
these raffinates for oxygenate contents. In Richmond spot combines with the olefin feed and is cooled by the reactor
samples have showed dimethyl ether concentrations of effluent. The total combined feed gets cooled from 90°F
1000 to 2000 ppm. For the second part of this question, to 100°F down to 45°F to 60°F, usually producing free
we do not have moisture analyzers on alkylation feeds. We water. Our opinion is that the feed coalescer should be
do have moisture analyzers on catalytic reformer feed from designed to remove over 98% of the free water.
tankage. It has worked quite well for years. Even if you totally dry the olefin feed before it combines
with the recycle isobutane (for most sulfuric units, this also
GENTRY: is a saturated stream), as it is cooled down to 50°F to 60°F,
I will take this opportunity to introduce myself. My it will still contain some free water.
name is Art Gentry. I am the Product Director for Petro- We believe it is always a good idea to periodically
leum Refining for the M. W. Kellogg Technology Com- analyze the combined feed in and out of the feed coalescer
pany. In that capacity, I am engaged in managing and to verify that the coalescer is operating correctly. On-line
developing proprietary refining technologies for Kellogg. moisture analyzers are hard to maintain and can lead to

160 Light Oil Catalytic Processing

TOC/INDEX

serious operating problems if relied upon too heavily GENTRY:

without applying additional process understanding and In the Exxon, stirred, autorefrigerated alkylation reac-
operational awareness. tor design, all of the recycle isobutane from the deisobu-
tanizer is mixed with the olefin feed upstream of the
R.E. (Ed) PALMER (Mustang Engineering, Inc.): reactor. This premixed stream is then cooled, chilled to
I have some data I was involved with a couple of years 40°F, and passed through a coalescer to remove free water
back. Methanol is typically 20 ppm to 40 ppm, dimethyl before being split and equally injected into the reaction
ether 200 ppm to 300 ppm and MTBE 50 ppm to 70 zones. The recycle refrigerant is not mixed with the olefin
ppm. This is downstream of standard raffinate water wash. feed, but injected directly into the first reaction zone.
Commercial experience during the early years of operation
Question 2. conclusively showed a large alkylate octane benefit for the
What are the benefits of premixing a portion of the premixing configuration. Thus, premixing has been the
unit’s isobutane recycle with the olefin feed being standard configuration practiced.
injected into the reaction zones of an autorefrigerated For a grassroots design, the typical isobutane-to-olefin
unit? What target isobutane-to-olefin ratio at the in- ratio at the premixed feed point is about 4-to-1 by volume.
jection zone is desired? When the recycle refrigerant stream is included, the overall
isobutane-to-olefin ratio external to the reactor is about
KOOIMAN: 8-to-1 by volume. Due to the nature of the autorefriger-
I am here with information from a couple of places. My ated design, where isobutane refrigerant directly vaporizes
thanks go out to the folks at Koch who have been very from the reactor, the definition of isobutane-to-olefin
helpful in answering some of these questions and the folks ratio can be somewhat confusing. Usually, we talk about
at Murphy. Koch has a refinery at Pine Bend, Minnesota the concentration of isobutane in the reactor product
and another one at Corpus Christi, Texas and Murphy Oil stream going to the deisobutanizer which is generally in
has a small refinery in Superior, Wisconsin and another the range 50 vol% to 60 vol%.
one in Meraux, Louisiana. I appreciate the opportunity to
be here. JACKSON:
We tested premixing with the theory that it would The benefits associated with premixing recycle iC4 with
improve the reaction by lowering the reaction temperature olefin feed prior to injecting it into an autorefrigerated
at the initial acid hydrocarbon contact point upstream of alkylation unit are two fold: increased local iC4-to-olefin
the impeller. Also, we wanted to test whether premixing ratio at the point of injection resulting in all the known
would improve the reaction by giving the emulsion for- benefits that carries, and increased mixing due to higher
mation a head start upstream of the impeller. velocity of the feed mixture.
We experimented with premixing the effluent with the No quantitative information on the absolute benefit is
olefin feed stream that had already been combined, but it available, but STRATCO states that the benefits follow
had not, at that point, been run through a mixer upstream the old adage “more is better” when referring to iC4-to-
of the contactor. We used an in-line static mixer to mix all olefin ratio. The normal design limit is pressure drop
the hydrocarbon feeds prior to injection into the contac- across the olefin feed system. Recycle iC4 until it starts to
tor, but we were not able to see any increases in either the affect the fresh feed rate to the unit then stop.
octane, endpoint or yield, and we did not see any improve- The recycle is simply attached to the olefin feed line
ment in the acid consumption. without any special sparger.
There was some erosion, leaks and fouling of the piping
system, and this was probably due to the fact that the Question 3.
effluent recycle streams were untreated and the piping Does anyone have an effective procedure for on-line
arrangement was prone to excessive turbulence. That sys- water washing of the DIB? If so, what are the general
tem has since been removed. guidelines and how long does the procedure take?
What are the effects on the unit operation while the
LEMMON: DIB is being water washed?
Tosco operates an autorefrigerated alkylation plant at
their Bayway refinery, and they report that premixing of LEMMON:
isobutane recycle with olefin feed can allow an increase in The DIB tower at Tosco’s Avon refinery has experienced
unit rate, particularly when the autorefrigerated unit is periods of gradual tray fouling. The fouling, which was
isobutane circuit hydraulically limited. If you shift some assumed to be caused by neutral ester degradation prod-
of the isobutane recycle in the olefin feed and you can ucts, caused the tower to flood prematurely and limited
maintain rate and still hit your targeted isobutane-to-ole- heat transfer from the bottom reboiler. On-line water
fin ratio of 1-to-4. washing successfully removed some of the foulant and

1996 NPRA Q & A Session on Refining and Petrochemical Technology 161

TOC/INDEX

reduced tower flooding. Bottom reboiler fouling was not

as effectively removed, and the reboiler eventually had to
be pulled and mechanically cleaned.
The procedure used involved:
Reducing unit charge rate to minimize wet alkylate
production.
Diverting the DIB alkylate bottoms product to a
dedicated tank equipped with vapor recovery.
Reducing DIB bottoms temperature to minimize
water revaporization (250°F to 270°F). Do not allow
alkylate vapor pressure to exceed storage tank safety
targets.
Injecting raw water at approximately 8% of total
DIB feed rate.
Monitoring overhead accumulator closely and
draining water to prevent water recycle to contac-
tors. Watch out for water in DIB side draw normal FEARNSIDE:
butane. I will address some fixes to help mitigate the root causes
The DIB tower also experienced two incidents of rapid of the internal DIB fouling and corrosion, and also the
fouling due to reaction condition upsets. Both times reboiler fouling. One of the rules of thumb that I use is to
on-line water washing failed to restore tower performance maintain the caustic water wash drum temperature in
and the tower had to be opened for mechanical cleaning. excess of 120°F. That helps maximize the removal of the
Aggressive “on-line” water washing, and “on-line” hot acid esters. Also, maintaining a Total Dissolve Solids
water washing with the aid of a surfactant were both (TDS) of that water wash at 6000 or less, helps minimize
employed unsuccessfully to remove the foulant. Use of the any acid carryover.
surfactant increased the amount of black polymeric mate- What we have found over the years is that if you pay
rial removed but did not remove the restriction which was close attention to the cleanup of the hydrocarbon before
causing the tower to flood. Entering the tower revealed it gets to the DIB, you can stop a lot of the internal
that the foulant was lodged in down comers near the corrosion and fouling. Also, for certain refiners we have
tower’s side reboiler draw trays. started up a high detergent filmer injection internally. This
has helped keep the salts moving through the tower and
KIMBRELL: also mitigate any corrosion byproducts which ultimately
We have water washed our DIB twice successfully. We end up in the reboiler and cause the fouling of that reboiler.
have not really had any problems with it. Our system is Certain case histories indicate refiners previously having
set up where the bottoms from the DIB go to a debutan- to pull reboilers every 3 to 4 months to clean them when
izer. So our concern with butanes in the alkylate is not as they were maximizing throughput. They have now been
great, perhaps, as others. We have a procedure similar to on-line in excess of 6 years, not having to water wash or
that described by Mr. Lemmon. clean the reboiler.

ARNDT: Question 4.
Two of our alkylation units are equipped with simple Has anyone experienced repeated problems with the
DIB’s (no butane side draw). We have water washed them refrigerant recycle pumps (cavitation, etc.)? It so,
both, one all too routinely. Essentially we reduce the what was the cause and what was done to eliminate
temperatures to the low 200’s and route the downstream the problems?
debutanizer overhead to off-test (will be high vapor pres-
sure). We operate the debutanizer carefully to make sure KIMBRELL:
our alkylate vapor pressure is okay. We inject clean water My name is Mike Kimbrell. I am associated with the
to either the mixed butane feed at the top of the column Engineering and Technology Center of ARCO Products
or in with the reflux. A typical rate would be 40 gal/min Company here in Anaheim.
for a 22,000 bbl/d alkylation unit. We watch and drain We have had some problems with our refrigerant recycle
the low points throughout the whole procedure. We also pumps. It appears that the fluid has been actually “freezing
check our product streams to make sure we know where up” on the discharge side of the pumps. It is our belief that
the water is going. Water is continued for about 2 hours. the suction trap has a leak in the weir, and it is allowing
The procedures have been very effective and we do not acid to make its way from the net effluent side over to the
notice any effects in the reactor section. refrigerant recycle side. At the lower temperatures, the acid

162 Light Oil Catalytic Processing

TOC/INDEX

is just freezing up. We have been able to raise the level on off line, pulling the strainer and finding a slushy type of
the refrigerant recycle side of the weir and since then have ice, or whatever you call it, to be the source of the problem.
not had any further problems. The plant determined this was related to increased quan-
tities of propylene in the feed, particularly at higher feed
KOOIMAN: rates to the depropanizer.
We had a lot of problems with DIB charge pump There also had been some previous problems with
cavitation when we upgraded to a vertical Sundyne with cracking in some welds in the depropanizer feed line, so it
dual ports. The cavitation, of course, would cause loss of was known that there were some problems with carryover
flow and the suction trap level would go up and trip off from the depropanizer feed system caustic wash. They
the refrigeration compressor. We found that there was not were operating the flash drums at 15°F to 18°F and
sufficient net positive suction head available when the running about 3% to 5% propane in refrigerant recycle.
compressor suction pressure would drop. In fact, we They were able solve this problem by raising the refrigerant
determined that the liquid was boiling in the suction line. compressor suction pressure to get up over about 20°F in
So we raised the suction trap approximately 30 ft to get the flash drum when running an increased propylene rate
the better liquid column head, and that has cured the to the unit.
problem.
We learned that the proper pump application is critical
to stable operation in this application. The Sundyne Mechanical
pumps require that we maintain flows within a minimum
and maximum range. So now there is a minimum flow Question 5.
spillback and a maximum flow restriction orifice to ensure What mechanical or process considerations are most
that the pump stays on the curve in the region of stable often investigated and/or incorporated to decrease
operation. unit operating costs? Improve mechanical reliability?

LEMMON: LEMMON:
We had a similar experience to the one reported by At this point I would like to introduce myself, my name
ARCO where the alkylation plant would be running is Chaz Lemmon. I work at Tosco’s Avon, California
normally, the refrigerant recycle pump flow capacity refinery, where I have been employed continuously for the
would drop off over several days and eventually we would last 18 years, all of this time the refinery has been under
switch to the spare pump. Then we would test the dam- Tosco’s ownership. The Avon refinery is located near
aged pump and find that it was performing just fine the San Francisco. Tosco currently operates two other refiner-
next day. So we began to speculate that we were freezing ies, the Ferndale refinery in Washington state and the
some sort of material in the pump impeller. The thing that Bayway refinery in New Jersey. Recently press releases have
was unique in our situation is that it only occurred when indicated that the fourth Tosco refinery in Trainer, Penn-
we were running a high percentage of propylene in the sylvania might restart operations in 1997.
feed. It did not occur when we were running all butylene All of Tosco’s refineries are fuel oriented. The Avon
and this kind of mystified us. We also have a suction trap refinery is the most complex featuring all the cracking
with the weir in it. We raised refrigerant side levels. We units commonly employed. The Bayway refinery is our
still experienced the problem when we were running largest refinery and is home to the world’s largest FCCU.
propylene. The Ferndale Refinery is located in one of the most
We did obtain some material from the pump. It was a beautiful areas in which I have ever visited an operating
very purply organic looking material. Others indicated refinery. The three refineries together feature a wide spec-
that possibly it could have been propyl sulfates and from trum of refining technology. We have all types of reform-
that we have concluded that when we were running high ers, all types of alkylation plants, and a lot of refining
concentrations of propylene feeds under reaction condi- history as all the plants have been in operation for many
tions perhaps we were making propyl sulfates. We also years. I would like to take this time to thank all the Tosco
found the same material plugging the taps of our refriger- employees who helped me prepare for these sessions.
ant recycle flow meters. Has anyone else experienced Question No. 5 addresses mechanical and process con-
problems at their alkylation plant when charging higher siderations we have investigated to improve reliability and
than normal ratios of propylene, i.e., propylene in excess reduce operating cost. I would like to start by quoting the
of one-third of the olefin charge? party line on alkylation plants: “maintaining favorable
reaction conditions cannot be emphasized enough when
R.E. (Ed) PALMER (Mustang Engineering, Inc.): alkylation plant reliability is being discussed.” All the typi-
I had some experience with a plant that had this exact cal reliability problems, like fouling of towers and plugging
problem. This was actually manifested by taking the pump of pumps, are promoted by poor reaction conditions.

1996 NPRA Q & A Session on Refining and Petrochemical Technology 163

TOC/INDEX

Isobutane-to-olefin ratios should be maintained at your The 76 Products Company is what you formerly recognize
licensor’s recommended targets. You must realize that as as the Refining and Marketing Division of Unocal 76.
you lower isobutane-to-olefin ratios you are doing so at a Our refining system includes four refineries, two in Los
penalty to reliability. Angeles, one in the Santa Maria area and the fourth near
It is essential that you monitor and maintain the right San Francisco at which I work.
acid-to-hydrocarbon ratio. We have been paying very close In addition to the issues that Mr. Lemmon mentioned,
attention to this ratio in recent years. we focus work in several process areas for improvement.
Most alkylation plants are refrigeration capacity lim- First, the higher fresh acid strength (up to a 99.5% maxi-
ited. Refrigeration condensers need to be maximized in mum) will reduce the water in the unit and improve octane
their performance. I highly recommend that you pay close while reducing corrosion.
attention to your cooling tower water treatment program Second, improve the feed quality by reducing the dienes
and the refrigerant condenser exchanger water side tube (same as you would reduce heavier feed content) this
velocity. reduces acid consumption, and thus your operating costs.
Regarding air cooled refrigeration condensers, if that is And third, controlling feed quality to minimize the
what you are using in your plant, you might want to inerts (nC4) is helpful in promoting effective contacting.
inspect them and see if they need to be cleaned on the
finned air side of the tubes. We clean certain coolers once DiCAMILLO:
a year. It is also important to make sure your refrigeration My name is Dave DiCamillo. I am a Technical Service
loop propane concentration is at optimum levels. At our Coordinator with Criterion Catalyst Company in Hous-
effluent refrigerated plant, we target 15% propane in the ton, Texas. We are a leading international supplier of
refrigerant loop. hydrotreating, reforming and styrene catalysts. I have to
Unique to Tosco, we have had success in utilizing admit that as a young boy when I dreamed of going to
packaged absorption refrigeration systems to supplement Hollywood and being on stage, this is not exactly what I
the main refrigeration system. These units are using low had in mind.
pressure steam to generate chilled water which we use to Mr. Lemmon and Ms. Barker covered things well. A
cool the feed to the reactors. This supplements the original few things they might have missed are: if you minimize
alkylation plant refrigeration system. acid inventory, you can avoid stagnant zones. And perhaps
It is important to realize that chillers do result in a a controversial item is running compressors at as low a
decrease in refrigerant recycle flow and, therefore, they suction pressure as possible.
hurt the isobutane-to-olefin ratio. We have had success in
improving our acid management by the installation of a Question 6.
spent acid density meter. We use a Coriollus type flow What is the commercial experience with mechanical
meter to obtain the density data. The flow meter has given seal technology applied to alkylation reactors? Are
us some acid temperature information which we found they widely used? What is the expected life of the
interesting, and I recommend that alkylation plant opera- seal? Do they reliably meet VOC emissions limits?
tors begin to observe their spent acid temperature.
We have also recently installed a feed forward calculated ARNDT:
acid demand system, which lets the operators know in Contactor seals have been a trouble spot for us, like
advance where their acid demand is going. This has paid everyone else. In the old days with the crane pusher type
dividends in smoothing out acid strength upsets and seals, seal life was about 4 months. Newer versions in-
reduced acid consumption. I am sure other refiners have creased seal life to about 8 months. At that point, one of
observed that plant operators tend to ramp the acid up our refineries worked closely with Borg Warner to develop
quickly when they go below target and then very slowly the double liquid cartridge seal. The first one was installed
drop it back down. That is a common human response. 6 years ago in one of our alkylation plants. Seal lives
But if you can tell them in advance where the acid demand increased to between 2 to 4 years. We are slowly changing
is going, you will find that they line that out very nicely. out all our existing contactors. Our 2 new West Coast
With regard to plant reliability, I strongly recommend alkylation plants with a total of 22 large contactors are
that all small bore piping in an alkylation plant be of alloy equipped with the double liquid seals. The units have been
construction up to the first block valve. When you have on stream for a year. The seals are monitored monthly and
an acid runaway situation, you will really appreciate hav- they are holding: i.e., no emissions detected yet.
ing the extra integrity of alloy thin wall small bore piping.
BARKER:
BARKER: Our experience is that the standard seals prove to be the
I am Ellen Barker representing the 76 Products Com- most reliable and they do not present any problems with
pany which is a wholly owned subsidiary of Unocal 76. emissions. We tried a double seal design a few years ago

164 Light Oil Catalytic Processing

TOC/INDEX

which had the seal pot and barrier fluid. But after several One of our units uses gate valves specified as A351-
seal and bearing failures, we went back to the standard seal CN7M trim alloy 20, with ANSI, raised face smooth
design which is proving to be very reliable. flanges API 600. So far, the experience has been good, but
they have only been running for around a year.
GENTRY: STRATCO recommends Alloy 20 block valves in sul-
I can comment on the experience with mechanical seals furic acid or sulfuric emulsion service. Larger valves,
on mixers which are part of the ER&E stirred, autorefrig- greater than 2 in., often have a carbon body with Alloy 20
eration reactor system. Double mechanical seals used in internals to minimize the cost of the valve.
this application have given excellent service. This is be- Ball valves and plug valves are also used in isolation
cause the seal application is a very mild one. service, as on HF units. The main problem reported
First, the reactor pressure is very low, typically less than appears to be the actual position of the valve relative to the
10 psig. Second, the shaft rotational speed is very low, position indicator due to internal misalignments.
significantly lower than the high speed impellers used in STRATCO also has reports of the Teflon sleeve turning in
other reactor designs. Finally, the seals are in the vapor the valve body and causing flow restrictions. Some small
space, which means that even if a seal leak did occur, it is valves have also been reported to not close tightly in
a gas leak. service. This could be due to the valve remaining con-
If a seal failure should occur in the ER&E reactor, that stantly fully open and may require exercising occasionally
mixer can be taken out of service without shutting down to cure this problem. This is also covered in question 15
the alkylation plant. Additionally, the mixers are designed in the HF Alkylation section.
so that the seal cartridge can be changed on the run. This Carbon steel block valves have been shown to perform
is possible because of the low operating pressures. ER&E adequately and have adequate service life in very low flow
recently had occasion to survey plants that use their tech- lines; that is in lines with fluids less than 2 fps with no
nology for experience with this capability. The answer turbulence. Alloy 20 is used in lines with high velocity
came back that although the operators recognized that and/or high turbulence.
they had the capability to change out the seals on the run, If plug valves are selected, then on gravity flow lines,
they have not had occasion to do so because of the e.g., on the emulsion lines between the contactors and the
excellent seal service. settlers, STRATCO recommends that the plug valves be
expanded to the same open area as that of the line. This is
KIMBRELL: not normally the case unless otherwise specified.
We have had mechanical seals on our contactors for Alloy 20 referenced above is an austenitic stainless steel
some time. Our average seal life is about 18 months. We designed to be very resistant to hot sulfuric acid which
would like it to be longer and we are working with the readily attacks 316 stainless steel. It is a high Ni (32% to
seal vendor to try to extend that, but we do not have any 35%), chromium (19% to 21%), with smaller quantities
results yet. of manganese (2%), molybdenum (2% to 3%) and cop-
per (3% to 4%), and very low carbon content (0.05%).
The Alloy 20 commercial term usually used is Alloy 20
LEMMON:
Cb-3 for piping and ASTM SA-351 grade CN7M for
Tosco’s Bayway Refinery reports that they use a dura castings.
metallic model BRT 4000 mechanical seal and they are
getting 18 months of life out of it. They also report that KIMBRELL:
they have problems meeting VOC seal emission limits at I agree with those comments. If the acid is high velocity,
times. we recommend using an Alloy 20 valve. In the lower
velocities, we have used carbon steel valves with 316
Question 7. stainless trim. We do have some experience with the Teflon
What type of isolation valve is recommended for al- lined plug valves, and it has actually been quite good. We
kylation service? Is Alloy 20 construction required? did have a problem with those type valves in the alkaline
Do Teflon lined plug valves work well? water wash where the Teflon got scored. But other than
that, our experience is quite good.
JACKSON:
As you can probably hear from my voice, I am the token LEMMON:
Limey on the panel. I introduced myself in the Heavy Oil I am going to address the operating experience side of
section. this question. I have personal operating experience with
BP has two sites that have H2SO4 alkylation units; one carbon steel body valves equipped with 13 chrome seats,
is an old M. W. Kellogg sulphuric acid unit and the other carbon steel valves with Alloy 20 seats, valves of complete
is a new STRATCO unit. Alloy 20 construction, and the plug valves with Teflon

1996 NPRA Q & A Session on Refining and Petrochemical Technology 165

TOC/INDEX

lining. They have all leaked when I did not want them to. service, a single well contains both the source and the
If you are looking for the magic leak-proof valve, I am not detector and operates on a back scatter radiation principle.
aware of one. The Teflon lined valves look more reliable For sulfuric acid alkylation units, the system would be
to me. They have worked well, but they do introduce an configured for a two well operation. The high density of
additional possible failure mode. We had one instance the sulfuric acid requires operation in a transmission mode
where the Teflon liner shifted, blocking some of the flow with the source in one well and the detector in the second
path, and another instance where the valve hand wheel well. The main advantages of the UOP density profile
operator was misaligned during maintenance and the system are that it does not contact the acid and the low
operators were inadvertently confused and operated the strength radiation source is contained inside a well inside
valve in an improper way. the vessel being monitored.
The Bayway refinery says they use 316 stainless valves
in most applications and only use Alloy 20 when there are KIMBRELL:
chlorides present. We installed two probes in each of our acid settlers. The
probes were made by Agar to measure the emulsion den-
ARNDT: sity. They have provided fairly good service, but we really
We prefer Teflon-lined plug valves with Alloy 20 trim. have not had a problem with emulsion carryover, so we
If the velocities are above 3 fps to 5 fps, we use an Alloy have not validated them.
20 body.
We had some installation “opportunities” with these BARKER:
valves, at our new Richmond alkylation plant, but we are We rely on ratio glasses in the contractor reactor and
over these and the valves are performing well. overall sight glasses in the settler.

BARKER:
We use gate valves for the isolation valves in the alky- General
lation plant, solid Alloy 20 on the smaller 2 in. valves; the
Question 9.
larger valves are carbon steel with all Alloy 20 trim.
Has anyone noticed an increase in the corrosion rate
in the reaction section when an upstream oxygenate
Question 8.
unit was brought on-line? If so, what is the mechanism
Acid emulsion level is difficult to reliably measure in that is causing the corrosion and are there particular
our alkylation plant. Other than sight glasses, are areas where the corrosion takes place?
there any reliable level detector technologies in use
commercially? JACKSON:
The only sulfuric acid alkylation unit BP has with
JOHNSON: upstream MTBE only started up in 1994 and was started
I am Brian Johnson representing UOP. I have spent my up from new with an MTBE unit upstream of the alkyla-
entire career, which will soon be 28 years with UOP. I have tion unit.
worked primarily in the Technical Service Department. STRATCO has had three licensees relate experiences of
Early in my career I spent approximately 6 years in the increased corrosion in the reactor section on installation
Caribbean area where I had the opportunity to meet of upstream oxygenate units. They are also aware of 2 new
another of our panelists, Mr. Terry Smith, and our mod- units that have experienced higher than usual corrosion.
erator, Mr. Terry Higgins. These units have only been run on MTBE raffinate. No
After that assignment, I worked in Des Plaines in the quantitative information has been made available to date,
Platforming Technical Service group as a process special- but increased corrosion has been seen on the contactor
ist, my most recent assignment is in Light and Heavy spider assemblies, impellers, hydraulic head straightening
Ends in Technical Service where I have responsibility for vanes, feed nozzles and tube bundles as seen in the exhibits
managing a group of process specialists who handle which follow. The first shows the contactor shell assembly
isomerization, alkylation, catalytic condensation, cu- side view, the second is the end view, and the third is the
mene and phenol synthesis, and some of the heavy oil hydraulic assembly head. The third and fourth exhibits
technologies. show the tube bundle heat exchanger. All of the internals
Regarding the question; the interfaces between acid, shown suffer from corrosion.
emulsion and hydrocarbon are being detected in acid The obvious control to minimize corrosion to the
settlers using UOP’s density profile system. This device alkylation unit is to minimize upsets and poor operation
uses a radioactive isotope in closed wells to detect changes and consequential carryover of oxygenates into the
in the density of the surrounding fluids. In HF alkylation alkylation unit from the upstream oxygenate removal

166 Light Oil Catalytic Processing

TOC/INDEX

equipment. The type and complexity of the upstream LEMMON:

equipment makes a big difference; water extraction re- We did not notice any increase in alkylation plant
moves the methanol, but does not touch the dimethyl corrosion after our MTBE plant came on-line 2 years ago.
ether (DME); water extraction removes the methanol, We have seen the type of corrosion that Mr. Jackson was
followed by stripping or molecular sieves which should talking about throughout the 30 year history of our plant
remove all oxygenates (DME) and some or all of the water. and would not attribute it to MTBE operations.
This latter setup can be considered a “best practice” when
it comes to equipment design and trying to minimize the ARNDT:
corrosion on the alkylation unit equipment. Our first MTBE plant started up in 1992. The raffinate
The mechanism that is believed to cause the corrosion is fed to a sulfuric acid alkylation unit. This alkylation unit
of the alkylation equipment has been attributed to higher is closely monitored for corrosion. We have not seen any
water content in the spent acid (as with all type of alkyla- increase in corrosion rate over the last 4 years which we
tion units, H2SO4 and HF). The sulfuric acid will convert can attribute to oxygenates in the feed.
any oxygenates to water and the process of corrosion goes
on. What is not known for sure is how much of the quoted J.L. RAINA (Indian Oil Corporation Ltd.):
increased corrosion is due to water content in steady-state I would like to know in general what are the reasons for
operation and how much is due to periodic upsets from the formation of oxygenates. Is this due to ingress of
the upstream units. oxygen into the system? What steps are being taken to
minimize that? Oxygenation would not normally take
KOOIMAN: place unless there are sources of oxygen available.
We put an MTBE unit on-line and after the MTBE
unit went on-line, we really did not see any increases in JACKSON:
the corrosion rates. However, our corrosion rates were not The belief is that the corrosion in the alkylation unit is
as low as we wanted them to begin with. due to the MTBE being carried forward. So it is the

1996 NPRA Q&A Session on Refining and Petrochemical Technology 167

TOC/INDEX

upstream operation of the MTBE unit that causes the • FCCU DeC4 and is responsible for the DeC4 opera-
problem. The acid and the MTBE react together and you tion and product targets, C5 in overheads and RVP
get water forming. The increased water concentration in of FRCCG.
the acid leads to an increase in corrosion. So it is the old • Alkylation feed controller and is responsible for the
chestnut of less water less corrosion. And what you do to control of the iC4-to-C4= ratio, feed rate, inventory
keep them out is you put in some sort of mitigation, to control of the C4 storage and C4 splitter and the iC4
prevent the MTBE carrying forward into the alkylation internal inventory.
unit. • C4 regasser control and is responsible for the control
of the level in the regasser based upon demand.
An optimizer will run above the three controllers and
B. Alkylation-Hydrofluoric Acid pass directives and targets to maximize the overall profit
(objective function) of the C4 streams.
Process The steps we would recommend before any refiner
launches into having someone install MVPC on their
Question 10. alkylation unit are:
Please describe any control strategies, including • Know where your iC4 is coming from and the range
“multivariable controls”, used to manage the isobu- of how much.
tane inventory in an alkylation unit. • Know where your C4= is coming from and the range
of how much.
JACKSON: • Understand the C4 disposal options you have.
We have a variety of iC4 management strategies: from • Know your economics associated with each disposal
simple good practice where the site identifies the location route and the break points.
of LPG to their alkylation units, with minimum and • Know where you have ullage flexibility and disposal
maximum olefins, iC4 contents, based on historic sam- flexibility and the priorities associated with each.
pling to full blown MVPC systems. From simple good • Understand the effects associated with all iC4 loss
practice and a knowledge of your unit and refinery, the and control strategies; e.g., reduce the iC4 loss in the
unit process engineer and operators can react based on alkylate and the nC4 in the recycle increase; etc.
experience and knowledge of their routing options. For • Understand your refinery’s iC4 and C4= production
example: for situations where you are iC4 limited: flexibility; e.g., ZSM-5 and FCC riser outlet tem-
• iC4 content of butane splitter bottoms should be less perature effects.
than 2%. • Understand FCC DeC4 tower operation and all
• iC4 content of alkylation main fractionator side cut other LPG towers and the limits on the gasoline
should be less than 10%. RVP, etc.
• Straight run C3/C4 splitter columns should have iC4 • Deep understanding of your unit’s objective; e.g.,
in C3 product stream less than 1 %. maximum iC4-to-C4= ratio for given alkylate make
For situations where you are in excess iC4: or steady iC4-to-C4= ratio and maximum through-
• Increase ZSM-5 addition rate to your FCC. put, etc.
• Stop FCC olefins to storage and route to alkylation Once this is understood by operations and the local
unit. process engineer, the refinery can start thinking through
• Minimize C4= to any other polymerization unit. its MVPC strategy and the opportunities associated with
• Increase iC4 recycle to straight run LPG storage to the control and optimization.
build iC4 content. (You could increase the C3= to
the alkylation unit [C3= to the alkylation consumes DAVIS:
20% more iC4 than C4=alkylation], but less attrac- Good morning. I am Tom Davis with the CENEX
tive if you have a C3= Catalytic Polymerization Unit refinery. I have been there for 14 years. The refinery is
(CPU) on site; downgrading C3 to fuel gas). located in Southern Montana and is the sole refinery flying
This is good basic skilled operation of the alkylation the CENEX flag. We currently charge 47,000 barrels a day
plant. of mostly Canadian sour crude.
Two of our units have gone to full blown multivariable Our control scheme is a lot simpler than what Mr.
process control (MVPC) systems, but have a number of Jackson talked about. CENEX normally runs their UOP
separate MVPC systems controlling the overall economics HF alkylation unit at maximum rates, so the typical
and hence iC4 management of the alkylation unit. control strategy is to maintain as high an isobutane con-
A third is currently installing three separate MVPC centration as possible and monitor acid strength. Multi-
systems with an optimizer siting above the three control- variable controls, as such, include an on-line GC on the
lers. The three controllers are on: depropanizer bottom stream, a charge flow controller on

168 Light Oil Catalytic Processing

TOC/INDEX

the butane isomerization unit that is usually at maximum do not use ASO color to indicate the source of the ASO
rates, and a sphere of isobutane on flow control that precursors. That said, all the refineries I spoke with knew
supplements the isomerization unit as required. the general rules and colors. Our leading alkylation expert
created the following picture, so let Barney’s rainbow guide
CARL KLIESCH
H (Treiber Controls, Inc.): you as your ASO contaminants.
Treiber Controls has implemented control and optimi-
zation of an alkylation unit. The application has been
running continuously and profitably for over 4 years. Our Barney says...
Optimal Predictive Control (OPC) software is a multivari- “Let the rainbow guide you on your ASO Contaminants”
able predictive constraint controller. An OPC controller
on the deisobutanizer column controls alkylate RVP, iC4
in butane, and nC4 in isobutane recycle, by manipulating
reboiler steam, butane draw flow and reflux. A second
OPC controller maintains the level in the isobutane recy-
cle drum and controls the iC4-to-olefin ratio by adjusting
iC4 makeup and recycle rate. The long time constant of
the drum and the complicated dynamics associated with
the recycle circuit make it difficult for the operator co
maintain the level without resorting co major changes in
makeup flow which upset the process. The predictive
capability of OPC handles this problem smoothly. There
are also OPC controllers for temperature control of the The refineries that have ASO disposal problems,
contactors, acid level control and control of the depropan- whether intermittently or continually, do ‘know their
izer. colors’. The guides we use are:
Our Closed-loop Reconciliation and Optimization Color Feed Contaminant
(CRO) software performs an economic optimization us-
Yellow Sulphur
ing a first-principles chemical engineering model of the
entire plant. New values for up to 25 set points are Red Oxygenates
calculated and sent to the controllers every 45 min. to 60 Purple Dienes
min. Among these set points are the iC4 in butane target Brown Caustic
for the OPC controller and the olefin feed rate. This results
Green Amine
in an economically optimal retention of iC4 in the unit
and use of olefin feed to exploit all available iC4. Refer-
ASO is sampled and a white popsicle stick is used to
ence: Treiber, S.; McLeod, R.S; Boyle, T.J.; Powley, G.; and
dip in the sample. The color of the ASO gives the operator
Lee, S., “Closed-loop Plant Wide Optimization”, CPPA
and process engineer an indication of the source of the
Control Systems, 1992 Conference, Whistler, British Co-
contamination causing the ASO formation.
lumbia, September 29 - October 1, 1992.
Yellow indicates sulfur breakthrough and that the
LPG treaters need some attention. One of our refin-
Question 11.
eries uses a 20 ppm sulfur in C4= feed co the alkyla-
How are color charts used on acid soluble oils to
monitor the presence of feed contaminants? What
tion unit as an indication that the treaters needcloser
information can be provided and how can it be used?
attention. As sulfur tends to form “light” ASO, it
can be quickly removed from the system by drop-
KOOIMAN: ping the acid rerun feed temperature to around
Although we do not do it yet, it sounds to me like it is 275°F (135°C). This drops the light ASO out of the
a pretty good idea. We have seen a darker brown in the acid stream via the tower bottom. If it is a real serious
polymer as the amount of sulfur in the feed goes up. It problem, we would probably drop the stripping
seems to me that we could use some kind of a chart to temperature from 350°F to 340°F (175°C to 170°C)
provide feedback to the folks responsible for keeping the to get it out even quicker and accept the resulting
sulfur out. increase in HF losses from the rerun base from the
lower stripping temperature.
JACKSON: Red indicates breakthrough from the upstream
It depends on which refinery you talk to and if they MTBE unit. It could also be due to water carryover,
have an ASO disposal problem. The sites that do not have especially if you do not have an MTBE unit
an ASO disposal problem and it is not a unit constraint upstream of the alkylation unit. Water is almost

1996 NPRA Q&A Session on Refining and Petrochemical Technology 169

TOC/INDEX

always due to poor performance of the feed dryers non-destructive testing, welded pipe heat treating require-
due to incorrect operating conditions on the regen- ments, valve replacement schedules, flange repair and
eration cycle or sometimes from problems with the inspection schedules, gasket and stud requirements.
plc logic. You will always loose a lot of acid when
Materials:
trying to remove the constant boiling point azeot-
rope of water and HF. The feed and stripping tem- Carbon Steel to have less than 0.2 wt.% sum of Cr, Ni, &
peratures both have to be dropped. Cu. (If possible).
• Purple means you have to talk with the FCC/RCC Inspection Frequency:
operators and see if they have been running higher Paint — roughly every 2 years (T/A frequency) — exter-
regenerator temperatures than usual or changed nal every 5 years maximum; UT every 5 years maximum
feedstocks, etc. The rerun tower would then be or 1⁄2 life.
operated with a maximum feed temperature of ap- New vessels are WFMT inspected prior to installation.
proximately 285°F (140°C) and maximum strip- Existing vessels are not WFMT inspected unless there is a
ping temperature of 350°F (175°C) and the heavy compelling reason to suspect cracking.
ASO will readily drop out. Large Piping:
• Brown indicates that you have to look at the separa- All butt welds in HF service are PWHT.
tion of your LPG treaters and the LPG/caustic sepa- Valves:
ration. A useful tip given to me by a young engineer Most valves are replaced in less than 10 years due to
is to first check that all the caustic was drained from operability problems. If valves are opened for repair, they
the ASO surge drum boot prior to looking at the are inspected.
ASO color. There have been a few instances of
Flanges:
personnel spending a long time looking at samples
Flanges are informally inspected and repaired if opened.
of caustic thinking that it is ASO. Brown ASO
indicates that the upstream treaters could also be Gaskets and Studs:
carrying over caustic (or amine) in the olefin feed. Use only B7M studs.
The first port of call is then the feed surge drum Use solid Teflon inner ring on gaskets.
where the boot is drained looking for caustic (or
amine). The rerun tower is somewhere between the ARNDT:
light and heavy ASO conditions. Concurrent with the development of API 751, we made
• Green, the last one, is caused by amine carryover an extensive review of our Salt Lake HF alkylation plant.
from the upstream treaters. The same checks and From that review we established inspection frequencies,
procedures are made as for caustic generated ASO based on the plant data, that have turned out for the most
(brown ASO). part to match API 751. There are a few places where we
As you expect, the ASO is never created due to one call for longer frequencies and some for shorter frequencies
particular incident, so the color is a blend. The ASO color as justified by plant data.
is only one tool in assessing possible causes of feed con- One of the outcomes of our review was the develop-
tamination, trending the units operation down to 6 min. ment of a good tagging system for flanges, valves and small
averages, talking with the operators (on the alkylation and bore piping. We have carefully grouped these tags so we
upstream), and knowing the upstream constraints and get a better picture of system problems. We now are getting
incidents, will all help the engineer minimize the ASO more insight to more global problems and can now take
production. Knowing the specific unit and looking at the more system-wide remedies.
ASO over a number of operating “incidents” will enable
you to react quickly and accurately. JACKSON:
All the BP Oil Alkylation refineries have their own
Question 12. inspection procedures based on the industry’s, and our
How are refiners interpreting the API 751 inspection own, best practices. All know API 751 inspection recom-
requirements? What difficulties have been encoun- mendations and guidelines and have reviewed it in detail
tered, such as small bore pipe inspections? and in-line with their own procedures.
API 751 defines small bore piping as having diameters
DAVIS: less than 2 in. and the key areas are summarized below:
Below is a detailed list of the CENEX API 751 practices • All threaded joints to be disassembled, cleaned, in-
that have been incorporated into maintenance, inspection spected and replaced where necessary. Radiograph
and new vessel purchase. These practices define materials of joints is an acceptable alternate.
and the limits of certain contaminants, inspection • Welded joints to be radiographed after construction
frequencies, painting frequencies, new and existing vessel and repair. Weld hardness should be checked.

170 Light Oil Catalytic Processing

TOC/INDEX

• Bleeder and vents to be radiographed. Question 13.

These are only the key aspects and API 751 should be At what acid settler velocities does acid entrainment
referred to directly for more detailed information. become a serious problem? Can acid entrainment be
Most of our sites, and one in particular, have gone traced by monitoring the acid return from the main
through an exercise of identifying all dead legs and small fractionator and depropanizer accumulators? How
bore piping items and either eliminating them or mini- much acid entrainment is considered normal?
mizing their length. At every shutdown, every 2 years, the
remaining dead legs, threaded fittings and welds are all HUNKUS:
radiographed. The short lengths of piping remaining are Hello, my name is Steve Hunkus and I am with
checked ultrasonically along their lengths and special at- MAPCO Petroleum. I wanted to take the time to welcome
tention is given to any bends. The number of tricocks on you all here and encourage you to ask questions and
the acid drum has been halved from 30 to 15. The level participate not only from the floor but avail yourself of all
bridles have been removed wherever possible. the experience that is in this room. I know that these
Question and Answer Sessions have been one of the
Any and all flanges are broken every shutdown and the
highlights of my career in the industry that I love.
gasket replaced. This also saves time at start-up.
I also wanted to caution everyone that Mr. Jackson is
Some BP sites have instigated a number of predictive going to introduce himself in the FCC session because
and probability programs based on historic data to assist that’s where he is going to answer the most questions with
with identifying areas that need review. One refinery uses the longest answers, and he feels that his answers are the
TISCHUC and PCMS. TISCHUC considers operational best prepared there. So you better be ready.
criticality assessment to: My team at MAPCO’s Mid-South operation is respon-
• Prioritize inspection and maintenance activities; sible for Safety and Training. I have been chairman of the
• Select inspection methods and frequencies; Safety Committee there for about 4 years, for Environ-
• Control inspection waivers and concessions; mental efforts, for our Quality Control Laboratory, for Oil
• Select items of equipment for reliability improve- Movements, for Engineering, Maintenance and Opera-
ment; tions. I also have been fortunate to have a core role in our
• Determine maintenance schemes and frequency; and organizational and people development efforts, both in
• Demonstrate regulatory compliance. the Mid-South business unit, and enterprise-wide. We
It uses operating temperatures, pressures, process con- have a refinery in North Pole, Alaska where the Refinery
taminants and metallurgy and predicts corrosion and Manager resembles the most famous resident up there. We
erosion rates. It uses API RP14E for flows in pipes. It run about 140,000 bbl/d there and one of the their most
identifies high critical systems. notable features is that they use kerosene for a cooling
PCMS is Plant Condition Monitoring System and medium instead of cooling water. Our Mid-South plant
tracks the condition of the system as the name implies. It in Memphis, Tennessee runs about 107,000 bbl/d now.
inputs material condition, ultrasonic data, and gives the We are trying to inch that up every day.
short and long term condition of the systems. I was born on a farm in Ohio, and I will not go into
Both work on steady-state data and do not take into too much more detail about that. I went to Michigan State
for Chemical Engineering and one year while I was there
account maloperation of plant upsets. PCMS has a sensi-
we did beat Notre Dame, Terry. I just thought I would
tivity module that can be used to look at abnormal opera-
point that out.
tions and situations. These have proven particularly useful I started off my career in head office with Shell Oil
on alkylation units. Company in the reforming, hydrotreating and hydroproc-
essing areas and moved into the FCC and resid processing
LEMMON: areas. I was also affiliated with the research effort out at
Tosco’s operating HF alkylation plant is a Phillips Westhollow, where Shell Development Company is lo-
design plant, and it is being inspected consistent with API cated. Then I went to work for an independent refining
751 recommendations. When the API 751 recommenda- and marketing company, United Refining, in Warren,
tions were published, we increased the frequency of valve Pennsylvania, where I learned how to do a lot with very
and pump inspections to comply. All small bore piping little technical support and less sophisticated resources
was originally installed with pipe threaded connections than I had at Shell. I also got quite a bit of bankruptcy and
to enable frequent replacements, and at this facility all lobbying experience at United. Next, I was involved in the
small bore piping is completely replaced on a 4 year start-up and expansion of 3 small businesses which
schedule, typically during turnarounds. Additionally, evolved into about $10+ million worth of sales in 3 years,
15% to 20% of the unit’s small bore piping is radio- primarily fabricating equipment for the refining, process-
graphed each year. ing and petrochemical industries. So I am a bit closet

1996 NPRA Q & A Session on Refining and Petrochemical Technology 171

TOC/INDEX

mechanical engineer. I then joined MAPCO, where I have JOHNSON:

had various roles, including Process and Project Engineer- UOP has licensees who have intentionally, or other-
ing, Project Management, Operation Superintendent, wise, put C5’s into their Selective Hydrogenation Proc-
Operation Manager and now General Manager. esses (SHP) to add C5’s to the alkylation Unit feed and
Regarding the question, we have seen carryover in the to do so without any added dienes. The SHP conditions
1.8 fpm to 1.9 fpm range at various iso recycle and feed were adjusted only based on the diene load. The net
combinations in our 10 ft vertical acid settler. UOP impact was a reduction in the total C4 feed rate due to
guidelines indicate this is also their upper range experi- the addition of C5’s. Butene-2 to Butene-1 ratio was not
ence for stable operation, and I can certainly verify that. impacted appreciably.
About 1.5 fpm is the point where they recommend
normal upper limits for design. Acid condition, tempera- JEAN-LUC NOCCA (HRI, Inc.):
ture, and physical condition of the equipment also affect You can feed C5 two ways. You can increase the amount
the point of entrainment. Acid entrainment, if not cor- of C5 in the C4 by modifying the DeC4 operation and have
rected, will almost immediately result in an opportunity a selective hydrogenation process running on the mixed
to acidize and inspect your fractionation equipment, C4/C5. There is also the other case where you send C5
evidenced by tray plugging in the isostripper below the directly to the alkylation unit through a depentanizer. We
feed tray, resulting in high C5 material in the iso recycle have nine units running like this. The advantage is obvi-
stream. You may not be able to get a temperature differ- ously the diene reduction. If you take a typical acid con-
ential across the top section of your tower. That would be sumption for a C5 feed, you can pay out the SHP in less
another indicator for you. than 2 years. The benefit in terms of ASO yield are difficult
Yes, you may need to use solubility charts or calcula- to quantify because the SHP typically runs all the time.
tions to verify flow rates. You can watch it build in the site
glass and check accumulation rates if you need to verify Question 15.
flows.
What is the recommended frequency and preferred
You really need to consult your licensor about the safe
approach on exercising and testing key block valves
operating parameters like this. They will have a lot of
and control valves in the HF alkylation unit? If a
experience to share with you. My experience has been that
program has been developed, what is the basis for
any significant acid entrainment results in unacceptable
developing the prioritization system?
unit performance.
HUNKUS:
JOHNSON:
UOP’s guidelines for our licensees are, 1.5 fpm upward We bring in a dedicated operator scheduled to exercise
HCBN velocity in a vertical settler and 1.8 fpm for the valves about every month, and check all the other
revamps with normal acid strengths. We would expect that safety equipment on a rotating basis. We feel that
the weight percent HF be kept above 88 wt% to avoid any DURCO and XOMOX valves, which are Teflon lined,
entrainment. We have had a variety of settler designs over need to be cycled to prevent cold flow creep in the liner
the years and have seen entrainment primarily at low acid which could prevent or retard proper operation when
strengths in vertical settlers. needed. Rotating this duty among operators also helps
build confidence and a workman like approach to these
DAVIS: activities during potential emergency situations. Proper
I can only add that at the moment we have a 2 UOP training is very important since these inspections require
design vertical mixer settler reactor system. Our velocities knowledge and experience to be properly done.
are around 1 fpm to 1.25 fpm, and we have no carryover The priority is for the acid isolation valves to be re-
problems. It sounds like a good idea to monitor acid boot motely operated. Other priorities are backup valves or
flows but perhaps the metering might be a little more hydrocarbon blocks that are also exercised. We can run the
complicated than you thought. There are solubility charts remote operated valves locally from the control room on
that UOP puts out that will give you an idea of how much the west side of the unit and also from outside the east
you should be carrying over on a normal basis. perimeter of the unit. We also have HF detectors and
hydrocarbon detectors on a monthly PM schedule. Other
Question 14. valves are moved once per month at a minimum. Valves
Is anyone sending pentenes to an alkylation feed and flanges are also inspected on a monthly or quarterly
hydrotreater, and, if so, what are the operating con- basis under our Leak Detection and Repair Program
ditions? What are the effects of hydrotreating on (LDAR). These activities are based upon either regulated
yield, product quality, ASO production and acid intervals or operations/maintenance teams experience to
consumption? maintain “no-fault” operation.

172 Light Oil Catalytic Processing

TOC/INDEX

JACKSON: well within the feed design capacity, might have contrib-
Our best practice refinery moves all ball valves in acid uted to its excellent reliability.
service once per month. The valve is moved a quarter turn
to move it just off its seat. Operation supervisors consider HUNKUS:
this job absolutely invaluable and feel it should be ex- We found that by trying to make a 5 year cycle and
panded to include all the valves around the C4 splitter. It running our unit at full capacity, we were incurring a lot
is an effort, as it requires a lot of discipline to ensure this of unscheduled downtime, a real problem especially dur-
substantial job is adhered to. ing the summer when we have to haul excess butanes to
Some of our refineries have tried to stroke their control off-site storage. We have moved to a 2 year cycle of planned
valves too, but care is obviously needed not to severely outages to reduce total downtime and especially to mini-
upset the unit’s operation. One of our refineries tried to mize any outage time during low vapor pressure gasoline
instigate a program to stroke the control valves, but upset season.
the plant and are now very reluctant to try this again. We also wanted to assure ourselves of unit reliability
As bad as leaving valves in the fully open or fully closed from an olefin compliance standpoint in our gasoline
position, is leaving valves operating in a tight control pool. This philosophy has paid off in less total downtime
range. Some services on our alkylation units are so stable and confidence in year-end unit operation when we are
that their operating band is very narrow (± 5%), such that trying to plan our gasoline blending.
they will not be able to fully shut the valve. One particular
refinery has installed valve exercises that keep the valve ARNDT:
moving slowly outside their normal operating range. This At our Salt Lake facility, we recently achieved a 4 year
was installed on the debutanizer accumulator acid boot run. The hope is to duplicate that again. They do have to
and moved the acid level set point from 30% to 70%. come down for mini pit stops to take care of some of the
Locally it is called a “wobulator”. PSV’s. The plans are to spare these valves so that they can
keep the unit up for all 4 years. After the 4 year run, the
DAVIS:
unit cleaned up well and looked good even though 3
exchanger shells had to be cut open.
CENEX has a motor operated valve isolation system
Another Chevron refinery (ex-Chevron now) achieved
around critical vessels and pumps on the HF alkylation
close to a 4 year run. They reached the opposite conclusion.
unit. These valves provide remote isolation in the event of
a catastrophic failure. The valves are normally partially
Question 17.
stroked once a month. Several are in routine service such At what acid purity should feed be pulled from the HF
as regenerator bottoms pump and the acid transfer pump acid unit?
and are used several times per week. We have no program
with control valves. JOHNSON:
UOP’s standard recommendation is to pull the feed out
Question 16. when the concentration of HF in the circulating acid has
What operating and maintenance practices have re- dropped to below 78%. Factors such as acid inventory, the
finers found necessary in extending turnaround cycles ability to add fresh acid, known elimination of the incom-
to 4 to 5 years? What conditions have been found when ing contaminants, whether there is settling zone carryover,
equipment was opened? and past experiences, can all be factored into the refiner’s
actual guideline.
LEMMON:
Tosco’s HF alkylation plant at the Ferndale refinery KOOIMAN:
currently operates on a targeted four year turnaround We are pretty conservative on this. In one refinery,
cycle. In order to accomplish this, they installed redundant though we have never been in the situation, if we get down
critical service relief valves to allow on-line servicing. They to 80%, then we do something whether it be with the acid
installed Monel skid bars underneath heat exchangers to or start pulling the feed back. Before we initiated this, they
prevent fluoride scale binding of bundles, and on-line got down to 75% to 77% without serious problems. That
inspection intensity has also been increased. They really was quite a while ago.
have not had any difficulty reaching 4 years, and they At another refinery, we actually pulled the feed out at
believe they could possibly run longer. There has been a 82% because after being hit with something, we were still
bit more fouling observed upon entry, but nothing that not recovering after 24 hours. We felt it was best just to go
would preclude a longer run length. In talking with some ahead and make a move so we could start recovering.
of the panelists whose units run at or above design capaci- One other thing I would like to add on this is that the
ties, the fact that our unit operates at 80% of name plates, end point is real helpful to us as well. That gives us a little

1996 NPRA Q & A Session on Refining and Petrochemical Technology 173

TOC/INDEX

STEVEN M. FISCHER (CITGO Petroleum Corporation):

I have a general question in the processing area that was
not included this year. Could someone from the panel or
floor comment on any recent developments in new alky-
lation technology?

GENTRY:
As I mentioned earlier, Kellogg, along with Haldor
Topsoe, is developing a new alkylation technology. Mr.
Phil Geren of Haldor Topsoe is here and is prepared to
comment on our status.

PHILIP GEREN (Haldor Topsoe, Inc.):

We are very close to a first project. This year, a U.S.
major oil company signed an agreement with us expressing
its intent to build the first plant to practice the Topsoe
bit quicker response. We see that as the acid purity de-
Fixed-Bed Alkylation technology. The project go or no go
creases our end point goes up.
decision was to be based on a 3 month feasibility study,
and that study was completed in September. It resulted in
LEMMON:
a recommendation to proceed with the project.
Our plant’s posted target is 82%. Although they once
However, we learned about 10 days ago that the capital
experienced 78% and recovered with no apparent problem.
budget of the company has been cut in response to extraor-
dinarily low refining margins, and the alkylation project
HUNKUS:
has been canceled. We are again actively seeking the
We reduce feed about 50% when we reach 83% acid
pioneer user for the process. We are optimistic because 43
strength and we cut feed out when we reach about 81%,
other companies are presently studying applications for
which has only happened on a few occasions. I will also
mention that we watch end point and octane as early the process under secrecy agreements.
warning signs for problems with the acid strength.
JOHN SHECKLER (UOP):
JACKSON: UOP has had a significant R&D effort underway to
Most of our refineries pull olefin feed to the alkylation develop a new motor fuel alkylation process that will
unit when the acid concentration reaches 80% or less. compete with the existing liquid acid technologies, both
Some locations pull their feed sooner (82% is typical). HF and sulfuric. This effort is nearly complete and UOP
Below 80% the acid strength can plummet. The feed is has begun commercial discussions with a number of
also pulled when the water content reaches between 2% refiners.
and 4%, depending on the refinery. The main aspect of The UOP solid catalyst alkylation technology uses a
the water content is that the corrosion rate begins to take solid catalytic material rather than a liquid or a hybrid
off at higher water levels. One refinery has intermittent system. We continue to confirm product quality, yield,
ASO problems that can require them to put the acid on catalyst stability and other similar parameters in our pilot
regeneration, typically when the ASO is around the 4% plant facilities. The catalyst performance and product
level. quality are both competitive with liquid acid catalyst
Clearly good alkylation olefin feed and acid analysis is currently used for motor fuel alkylation.
paramount in keeping ahead of the game. Moisture ana- UOP has also completed an engineering design for this
lyzers are also important and one used on some of the catalyst system and the capital costs are competitive with
group’s alkylation units is a Panametric capacitance device. comparable alkylation technology. The process is being
Like all instruments, it works well with attention. developed to operate at moderate conditions without
requiring exotic metallurgy. Although we cannot elaborate
DAVIS: on the unit or reactor design at this point, we do not
CENEX has seen acid strengths as low as 65%. Nor- anticipate any significant deviations from practices found
mally, olefin feed is not pulled but reduced such that the in a modern refinery. UOP is committed to bringing this
acid regenerator can catch up and/or fresh acid may be technology to the market and we have initiated customer
added. Please remember that CENEX has a two reactor discussions for the initial commercial installation. As you
system so that 65% acid strength reflects only one of the may know, UOP has recently commercialized a new proc-
reactors. ess in the detergent alkylation market based on a solid

174 Light Oil Catalytic Processing

TOC/INDEX

catalyst and we are confident that we will experience HUNKUS:

similar success in the motor fuel alkylation market. We are about ready to begin field work on a revamp of
our existing HF alkylation unit. We plan to include the
LLOYD WINGER (Conoco Inc.): ability to test cycle the dump system without actually
Back to the existing alkylation processes, is there a dumping the acid. Our present system includes remote
recommended optimal water content for the circulating operated plug valves with backup manual valves for final
acid? block-in. We can presently partially cycle our existing
remote operated valves. We plan to use this same strategy
JACKSON: in our rapid deinventory system so that the remote oper-
We would recommend you target somewhere around ated valves can be stroked on-line as a system, or individu-
1%. If you go any lower than that, then you affect octane. ally, giving the operators the flexibility to use the system
If you go any higher, then you get higher corrosion. for normal deinventory just to assure that the valves still
operate.
RONALD E. MARRELLI (Phillips Petroleum):
At the Phillips’ refinery in Woods Cross, Utah, we are Question 19.
in the process of revamping our existing alkylation unit to What acidizing techniques and solutions facilitate
use a new process developed jointly by Phillips and Mobil, bundle extraction from exchangers?
to reduce HF acid volatility. The new alkylation process is
called ReVAP, reduced acid volatility process. It is under DAVIS:
construction at this time and should be completed by the CENEX uses a 5 wt% to 10 wt% solution of sulfamic
first quarter of next year. acid circulated with an acidizing skid for approximately
24 hours. Temperatures are held at 130°F to 140°F. Main-
DON A. HUSTED (UOP): tenance flushes and neutralizes according to the manufac-
Does anybody use an on-line analyzer for acid strength turer’s procedures and hopes the bundle extracts from the
or is this all laboratory analyses? And, do you believe that shell. Note: This procedure is used for maintenance pur-
an on-line analyzer would be beneficial? poses only; it is not used for acidization of an entire
alkylation unit in preparation for a total shutdown.
DAVIS:
We do not have an on-line analyzer. Ours is done HUNKUS:
strictly by the laboratory. It would sure be handy to at least We generally use HCl or citric acid. We always establish
have a backup device to be able to read on-line. water circulation and acid inhibitor and do any preheating
required before the acid is added.
JACKSON: In designing your acidization program, it is important
We do not have an on-line analyzer that measures the to establish well thought out circulation loops and high
acid directly, but we do recommend a moisture measure- enough flow rates, and to provide proper ventilation of
ment. Panametrix is the type of instrument that we use, system high points in the circulating tank to the flare
and it has been very successful. header for H2S and other gases that may evolve during the
acidization process. We use well marked P&ID’s with
HUNKUS: numbered, coordinated points. I also feel it is important
We do not have an on-line analyzer either. We rely on to establish calculated volumes, chemical requirements for
the “Barney” method, looking at the ASOs, the endpoints each loop, and minimal required circulation rates. I also
of the gasoline and octane, to give us early looks. And the insist on identifying and listing all required equipment on
only moisture measurement we are comfortable with is the the purchase order.
Karl Fisher titration which we have the ability to do We sometimes neutralize our equipment to treat any
around the clock. possible remaining HF before acidization, and generally we
neutralize our acidized equipment before breaking to at-
mosphere. The acidization fluid is then put to a tank for
Mechanical further treatment, if required. It is important to remember
that ammonia will form salts that can require extra time
Question 18. and effort if you are trying to extract bundles, trays, or
Discuss routine test systems developed to assess the pump assemblies. We have also used “solvent/cleaning
reliability of rapid de-inventory systems. How are agent” systems with some success, and there are several
actuators and valves tested without actually activating vendors who will work with you to design a custom pro-
the “ de-inventory “ system? gram. Our results were somewhat mixed, but the deposits

1996 NPRA Q & A Session on Refining and Petrochemical Technology 175

TOC/INDEX

appeared dryer and more powdery, and probably made is the potential presence of pockets of acid trapped in the
bundles easier to pull and clean. scale.
Low flow rates, cold solutions, dead zones, and partially From a recent shutdown, one of our alkylation units
vapor-blocked equipment are some of the most common removed over 600 lbs (300 kg) of iron from an acid wash
reasons for poor results. We feel that having someone work of the main fractionator after a 12 to 18 month run.
with the crews on the night shift, both from an effective- Scale is apparent in the whole of the acid circuit. A
ness standpoint as well as an environmental quality con- classic location for scale build-up is on relief valve (RV)
trol and safety measure, helps ensure desired results. Also, lines. Any lines that are sloping down to the RV will
once you start acidizing, do not stop until you are done experience build-up of scale. All such lines were spare RVs
unless you want to “renew” equipment. taking inlets off the main RV inlet line resulting in dead
legs. Basically, any dead legs will potentially result in scale
KOOIMAN: formation. Some sites have found the whole inlet to the
We use dilute sulfuric acid to remove the fluoride build RVs blocked with scale. A number of our refineries have
up followed by a neutralizing solution of water and soda resolved these problems by running new pieces of piping
ash. But I do not have comparison on how well that works with the slope back to the main line, away from the RV
versus other methods. itself.
Scale build-up results in potentially inaccurate flow
Question 20. meters. NRVs typically will begin to pass after 6 months
How common are fluoride scale buildup problems on of operation.
check valves and other components of an emergency To avoid scale, our experts suggest that refineries keep
acid evacuation system? How do you avoid these acid strength high and water content low, around 1%. Any
conditions? Please discuss the appropriate design lower will result in octane loss. Design away problems by
criteria and operating practices. minimizing dead legs and length of dead leg runs. Know
your unit.
JACKSON:
Fluoride scale build-up is a big problem. To understand MO VADEKAR (CHEM TECH Consulting):
the problem with scale and where it is most likely to occur, I want to relate an experience I had recently regarding
you need to understand the corrosion process. As we know alkylation plant corrosion. We were trying to find the
the water in the olefin feed eventually accumulates in the reason for unexpected corrosion in an alloy and carbon
acid. The carryover of more or less water means more or steel line. The metal inspector at this refinery said a curious
less corrosion. To make it worse, operating conditions can thing to me. He said that today, alloy and even carbon steel
result in a lot of water and only trace quantities of acid in equipment made from newer (alloy) steel product contain
different parts of the plant. For example, the top of the trace impurities from more and more scrap being proc-
main fractionator and the acid regenerator generally have essed in the steel making process. His suspicion was that
high water and low acid ratios. As a consequence, the this unexpected corrosion in straight sections of piping
tower internals often suffer from severe corrosion during was related to such newer (alloy) steel construction. Con-
operation. ventional wisdom will tell you that such corrosion should
One of our experts describes the corrosion problem as not occur, but it did happen. He believed that trace
a lightning strike — it rarely occurs in the same place twice impurities which are not detected remain in the new
and even when it does the corrosion rate is often signifi- (alloy) steel and are probably the causes of corrosion. My
cantly different. question to the panel is, has anyone experienced anything
Enough said on corrosion. The iron fluoride scale that like that?
results from the corrosion tends to result in blocking of
small bore piping and internal, such as orifice tappings; it DAVIS:
tends to stop valves seating, and it tends to “freeze” non I thought that UOP put out some impurity require-
return valves (NRVs). ments on carbon steel in alkylation units that had to do
The lines tend to stay “clean” of scale in relatively high with limit requirements on chrome, nickel and copper in
velocity areas. Sites have seen scale layers about 3⁄4 in. (6 carbon steel. That is all I am aware of.
mm) thick on trays and on vessel walls and as deep as up
to 12 in. (300 mm) in drums. FRANK HIMES (UOP):
Scale will show itself as hard and soft. On some of the Yes, based on the various NACE papers on the subject
hard scale, the development of the scale shows as growth and other information, UOP has implemented a limit on
rings, like on a tree trunk, that is invisible to inspection carbon steel piping in hydrofluoric acid service such that
until it is tapped with a hammer and it shatters. The soft the Chromium, Nickel, and Copper combined total less
scale appears porous. The obvious danger with this scale than 0.2 wt%. This is being specified for any current project

176 Light Oil Catalytic Processing

TOC/INDEX

work. We will continue to monitor industry findings in this You then vent via the next section at a low point. Repeat
area and may revise this specification in the future. this until the oxygen level is less than 0.5%. Do the same
for the second section and vent into the next logical section
RONALD E. MARRELLI (Phillips Petroleum): until the oxygen level is less than 0.5% and so on. The first
Has anyone on the panel or in the audience seen section will take 3 to 4 cycles and the remaining sections
nitrogen compounds come from the FCC over to the will take less due to higher volumes of nitrogen.
alkylation unit and cause non-condensables and flaring on When the whole unit is free of oxygen, hydrocarbons
the unit? can be brought into the front end and the unit brought
up to operating condition. The water is still looked for at
HIGGINS: all low points. We have found the depropanizer proves to
No response. be the most troublesome and wettest. At the hydrocarbon
circulating and water draining stage we find the water
drops off from around 2 gallons of water per day to no free
General water after less than 2 days.
The hot nitrogen dryout is basically the same as the air
Question 21. dryout, except you use nitrogen and you do not need the
What are the relative merits of isobutane circulation pressure/depressure sections to purge the unit. This is an
and hot nitrogen dry-outs after turnaround concerning advantage, but looking for low point leaks is a safety
time, energy cost, and quality of dry-out? concern to be considered, and the nitrogen use is higher.
The cost of the air dryer and air compressor, assuming
JACKSON: you have to hire, them is between $5,000 to $10,000. The
There have been a number of procedures developed to nitrogen purge is the same as when you iC4 dry. The
dry out the alkylation unit after a shutdown. The typical nitrogen dryout does not require the compressor and
procedure used throughout the industry is the hot isobu- dryer, but does require more nitrogen. The costs between
tane (iC4) dry out; circulating the iC4 and draining at the the two gas dryout systems is fairly similar. Both gas dryout
low points. Other procedures have been developed that all procedures will save between 2 and 4 days on start-up. At
try to shorten the start-up duration. Two that have been
the cost of between $25,000 and $100,000 per day lost
used successfully within BP are hot nitrogen and hot air
revenue when the alkylation unit is down, it clearly pays
dryouts.
to minimize start-up time. Some refineries do not have
The hot air dry out has been successfully used at a
olefins feed when the unit is ready for start-up earlier,
number of our sites and has saved around 2 to 4 days on
because the FCCU is still down. But you can then schedule
the typical iC4. Basically, you can either use a hot air
dryout method using ambient quality air or use a dry air to take longer for your alkylation unit shutdown, and this
dryout technique. We have found that by hiring an indus- will result in a small shutdown crew with less overtime,
trial air conditioning unit; chilling the air and knocking etc., all realizing a lower shutdown cost.
out the water before compressing it and adding it to the
unit is the best method. This is done by breaking the unit KOOIMAN:
down into 6 or 7 logical systems, adding the air at a high That sounds like a pretty good process. Our experience
point and draining at the low points. Using air also allows has been that the isobutane takes longer. The nitrogen does
you to leak test the unit after a shutdown before hydrocar- not get the system quite as dry.
bons and acid are introduced into the unit.
This point is worth expanding further. We generally ARNDT:
acid clean our alkylation units during a shutdown. You Traditionally, we used isobutane circulation. It would
should, and we do, break every flange to check the flange take a good 2 days and the acid moisture level after start-up
face and reinstate the flange. With somewhere between was high (3% to 5%). We would get measurable metal loss
1000 and 2000 flanges on the unit you should then leak the first couple of months of operation.
test the systems using tape, bubbles, and good procedures. We did a hot nitrogen dryout for the last start-up and
Even with experienced personnel, we typically find 2 to 4 dryout only took 1 day. The acid moisture level was only
leaking flanges. With air, this is easy and safe to do. Once 1% after start-up. We are cold or hot nitrogen dryout.
the unit is leak tested, the unit should be pressure/depres-
sured via the low point drains until no water is seen. No RENE LUCENA (Corpoven Petroleos de Venezuela):
water meaning only drops of water are apparent. The I want to ask the panel if they think a chemical clean
system is then nitrogen purged. This is done by isolating out of the alkylation plant during the shutdown for main-
each logical section and starting with the front of the unit tenance is necessary to preserve the metallurgy of the
and filling the first section and pressuring to about 30 psig. plant?

1996 NPRA Q & A Session on Refining and Petrochemical Technology 177

TOC/INDEX

JACKSON: refinery profitability wherever we can utilize Nalco/Exxon

Yes, we would recommend an acid clean. To do some products and services.
of the inspections and to get to some of the pipes, we With that, onto the question. Can amine inhibitors
would actually like to get in there. We have never been increase the ammonia content? Sure, if they are not ap-
limited in terms of time, and the cost associated with plied properly. Good control of the inhibitors upstream
cleaning is small compared to the information and the should contribute no more than 0.01 ppm to 0.05 ppm
confidence you have in the apparatus on the operational of nitrogen in the reformer feed, well below the detection
security afterwards. So, yes, we would recommend you do limit of most laboratories and the analyses they are per-
an acid clean. forming.
Moving on downstream to the debutanizer and the
fouling problems associated with that, it all comes down
C. Reforming to the particular economics that the refiner is living with.
Do you water wash the debutanizer tower and loose 12-24
Process hours of saleable product every time? Do you have an on
stream continuous antifoulant product or program which
Question 22.
is very effective? Or do you go the route where you can
intermittently clean the tower using a high detergent
Discuss ammonia chloride fouling issues. Can up-
filmer without resorting to the traditional feed reduction
stream amine inhibitor programs contribute to feed
and tower slumping while water washing when the pres-
nitrogen? Are debutanizer feed chloride treaters
sure drop or distillation capabilities are not performing up
widely used? Are debutanizer antifoulant programs
to the particular refiner’s needs at the time.
effective?
So there are three different options there. You need to
look at all three and consider the economics of each option
BARKER:
and see which is the best for your particular needs.
We have experienced several shutdowns of debutanizer
and depropanizer systems to water wash the ammonium ARNDT:
chloride deposits. In addition, we have also had to fre- I would like to address this one on the chloride issues.
quently clean the debutanizer feed exchangers. Our re- We have one refinery that has chlorides everywhere and
former stabilizer overhead gas and liquid feed to this the others are not far behind.
debutanizer and since our new chloride traps were put into From a reformer, we see problems with 3 species of
service on each stream, we have had no further problems chlorides: ammonium chloride, HCl, and organic chlo-
with plugging or developing pressure drop. In the area of rides. We have fouled the recycle compressors, corroded
upstream amine inhibitors, we do not use them. But we the tops of stabilizer or debutanizer columns, and cor-
do suspect that they would break down in the reforming roded overhead systems in benzene recovery columns way
reactors to produce more ammonia which would then downstream. We have also seen a lot of tray plugging and
react with the chlorides to produce more salts. low point damage in downstream Gas Recovery Units
(deethanizers, depropanizers, and debutanizers), and, of
DiCAMILLO: course, fuel gas system plugging.
Upstream inhibitors will add nitrogen, but we would The easiest chloride species to remove is HCl. The
expect that the naphtha hydrotreater would take those out choice is alumina absorbents or caustic wash. We have
before the feed reaches the reformer. Antifoulant programs taken the more conservative approach — caustic wash —
would probably not be useful in the debutanizer because in most of our units. This is largely due to the problems
ammonium chlorides are water soluble and you would not of getting accurate HCl analyses at the very low ppm levels
need to go with an antifoulant in that instance. needed to monitor absorption beds.
The ammonium chloride issue is best mitigated by
FEARNSIDE: limiting nitrogen levels in your reformer feed. We like to
I am Paul Fearnside. I work for Nalco/Exxon Energy see less than 0.2 ppm which is about the detection limit.
Chemicals. My background includes 18 years in refineries, This means you need to operate your naphtha hydro-
the first 8 with UOP, the second 8 with Nalco and the last treater for both HDS and HDN. There are a few feeds
2 years with Nalco/Exxon. It has been a very interesting 2 that bypass the naphtha hydrotreaters, usually coming
years for us bringing together two different companies, come from hydrocrackers. Every once in a while you can
two different cultures, and turning it into one truly glo- get surprised by the amount of nitrogen in a hydrocracked
balized firm. I thought when I left UOP my worldwide naphtha. Usually this is not nitrogen dipping out the
travels were over, but they have come back. My duties are reactor but may be organic nitrogen compounds in your
basically worldwide tech service with the goal of increasing reactor effluent wash water.

178 Light Oil Catalytic Processing

TOC/INDEX

Regarding the organic chloride issue, we have only seen typically we do ours bi-weekly or monthly. We do not have
this in a CCR application. We tried a solid bed absorber chloride treaters because we do not have the need to have
on the reformer separator liquid before going to the debu- chloride treaters or any antifoulant programs.
tanizer with very little success. The adsorbents, and we
have tried quite a few, would load only up to 1 wt% RICHARD M. NASH (Shell Oil Products Company):
chlorides before breakthrough. We tried injecting amine I am aware of a Shell refinery outside the United States
inhibitors into the debutanizer column itself but only that uses an alumina-based absorbent on the feed to the
succeeded in plugging and corroding the column trays. stabilizer column with very good results. The main prob-
lem with operating it is that it fouls with ammonium
FEARNSIDE: chloride salts and builds up pressure drop. They have 2
I would caution you on using any type of neutralizing beds which they normally operate in series. When the lead
amines in the DIB column. That will foul and will cause bed plugs, they purge the salts from the bed using a steam
substantial additional corrosion. So please do not use a purge making sure they never pass through the dew point.
neutralizer type amine in your DIB column. Also, on
They can then put it back in service and usually then
another note we have found that quite often simple ad-
switch the order of the beds. That has worked very well
justments to the upstream hydrotreater water wash and
paying attention to level controls and things like that will for them and solved some very frequent fouling problems
eliminate a lot of carryover of nitrogen into your reformer. of their stabilizer and the need for frequent washes of the
So, it is a very simple thing to start with if you are trying trays and the overhead system.
to track down sources of nitrogen.
CHARLES L. MORGAN (LaGloria Oil and Gas Company):
KIMBRELL: We have been tracing chloride fouling of the debutan-
I would agree that operating the naphtha hydrotreater izer feed bottom exchangers for some time. We are at the
to minimize the amount of nitrogen breakthrough into point where we feel there is a relationship between the
the reformer really minimizes the ammonium chloride amount of coker naphtha that we are processing in the
fouling. We also try to control coker naphtha endpoint, naphtha hydrotreater, the duration of the catalyst being
because that is where a fair amount of nitrogen comes on-line in the naphtha hydrotreater, and some contami-
from. We do not have any experience with debutanizer nation of silicones into the naphtha hydrotreater, deacti-
antifoulants, because we typically run a year between vating the denitrification activity of the catalyst. I was
washing our debutanizers. wondering if the panelist would have any comments
regarding that and any light could be shed on it?
KOOIMAN:
Our chemical supplier was a little sensitive to the issue ARNDT:
on the upstream amine inhibitor programs and gave me We have a lot of experience with that, especially with
some information on the typical inhibitors. His informa- the silicon loading. We have two naphtha hydrotreaters
tion said that the nitrogen content is less than about 1% each feeding 30% to 40% coker naphtha. We have in-
and the application rate is 5 ppm to 20 ppm. So if I am stalled prereactors on both units to help control silicon
calculating that right, then it would not even show up in fouling and gum fouling of the hot feed/effluent exchang-
Mr. Arndt’s analysis. ers, furnaces, and main reactors. The prereactors are situ-
ated in the preheat exchange train and operate in the 350
SMITH:
to 400°F range. They pickup about one half of incoming
I am Terry Smith with Saudi Aramco. I have been with
silicon. Luckily we have well-aged naphtha hydrotreaters
Saudi Aramco for about 21 years now. Saudi Aramco
operates 5 domestic refineries in Kingdom. They are fuel with 2 or 3 main reactors in series. We can monitor silicon
refineries for local products, with over a million barrels a poisoning in our main reactors by matching delta T of the
day capacity. We also have 2 joint venture refineries with first reactor in series. We load catalysts that hold maximum
Mobil and Shell, a little over 600,000 bbl/d refining amounts of silicon. It is vital to keep active catalyst in the
capacity. We will also bring to bear some gas plant experi- reactors. The naphtha hydrotreater must remove the ni-
ence we have with our large gas plants throughout the trogen down to 0.2 ppm to alleviate NH4Cl problems in
Kingdom. the reformer.
On this question, really it is all of the above. We agree
that chloride fouling is not a big problem for us because RENE LUCENA (Corpoven Petroleos de Venezuela):
the Arabian light crudes that we run have low concentra- Have some of you have experienced scale and plugging
tions of nitrogen. We believe the water washing programs due to iron and coke in the first reactor during a noncon-
should be suited to the depositions that you see and tinuous reformer regeneration?

1996 NPRA Q & A Session on Refining and Petrochemical Technology 179

TOC/INDEX

ARNDT: So as we go up in pressures to decrease the burn time,

We have seen both scale and plugging due to iron and the chances for carryover increase. Another thing that
coke in the first reactor. I have seen the catalyst in the first causes the carryover is letting the specific gravity of the
reactor with more than 10 wt% iron. There are two solution get too high, the neutralization solution that is,
sources; upstream units/tankage, and in-plant corrosion during the regeneration.
products. Iron from in plant corrosion is usually a result So one should control the total solids and the pH very
of neutralization solution entrained during regeneration. carefully and not get outside the guidelines that are known
Most of the time, we replace first reactor catalyst every for entrainment. If you can stay within those parameters,
dump and screen. For the majority of our units, the first then you can avoid carryover and plugging during the
reactor only holds 10% of the total catalyst load so it is regeneration.
not painful to do this.
As for carbon, again we see it from two sources, one from Question 23.
heavy feed upsets, the other from recycle compressor seal Does any vendor regenerate and rejuvenate semire-
oil. You need to watch your seal oil consumption especially generative reforming catalysts? This would eliminate
during the lower pressure operation of regeneration. the in situ regeneration/rejuvenation of these cata-
lysts and thus save unit start-up time and possibly
FEARNSIDE: improve catalyst activity by having lower catalyst
We have seen a few customers come back to us with this particle temperatures during regeneration. Also, the
problem. For one, it was attributed or traced back to the critical in situ oxidation/reduction steps could be
fact that they had gotten hit with a slug of organic chlo- eliminated which could remove extensive time and
rides which had drastically increased the corrosion on the operating errors during this critical catalyst activation
reactor effluent side of the hydrotreater. All that iron then step and also reduce corrosion on equipment in the
had carried through and plugged the reformer bed. semiregenerative unit.
In other cases we had they have traced it to naphtha
that has come out of an old tank where the bottoms had
SMITH:
not been cleaned properly. A lot of the iron and corrosion
Commercial regeneration facilities in North America
byproducts were sucked in and caused the problem.
and Europe claim the ability to regenerate and rejuvenate
reforming catalyst. Our evaluation of these facilities to
HUNKUS:
This relates back to the plugging of the first bed with determine the actual ability did not proceed due to the
caustic and iron. Recently we decided that our low pres- poor economics of the venture for our facility.
sure separator, which is fine during normal operation, Each of our two fixed bed reformers undergoes normal
allows entrainment of the neutralization solution during regeneration every 9 months and a dump and screen
regeneration, probably due to the pressure and higher gas regeneration every 2 years. The time saving benefits of the
molecular weight. We are modifying our unit for our next ex-situ regeneration could have only been realized during
regeneration to go through the high pressure separator a dump and screen regeneration. The use of ex-situ regen-
during regenerations to knock that entrainment out. erated catalyst would have extended the 6 to 7 days
duration of normal regenerations. Additionally, to be fea-
SMITH: sible, a complete spare catalyst batch (approximately
Back to the previous question, we have experienced coke 150,000 lb for us) including precious metals would have
ball formation in the bottom of the first reactor. It has to been required. So the cost of spare catalyst, the shipping
do with running beyond the severity limits of the original costs to and from the regeneration facilities and the limited
design and also the center pipe distribution design was not use spare catalyst could not offset the time saving benefits
done as well as it could have been. It created dead spots in for the dump and screen regenerations. So in summary, it
the bottom of the radial flow reactor and so we had coke was not economical for our case.
ball formation which eventually holed through the center
pipe. We had catalyst migration through the furnaces. HUNKUS:
For many years, vendors have offered ex-situ regenera-
J. B. RODDEY (Roddey Engineering Services, Inc.): tion services for catalytic reforming catalysts, including
Regarding Mr. Hunkus’s comment about entrainment CCR, semiregenerative catalytic reforming, and other no-
during regeneration, as pressures go up in the regenera- ble metal based catalysts. One vendor also utilizes their
tion, the speed of burning coke increases. It decreases the density grading technology to remove high carbon CCR
time for the regeneration. However, separation of the platformer heel catalyst from the lower carbon, reusable
solution from the gas at higher pressures is more difficult platformer catalyst. These services are available on a turn-
than it is at lower pressures. around or an “at leisure” basis. An Oil & Gas Journal

180 Light Oil Catalytic Processing

TOC/INDEX

article, published in the September 18, 1995, describes the a turnaround basis or in conjunction with a spare charge
techniques and costs associated with this. replacement. In this last case, one batch can be regenerated
I agree with a lot of Mr. Smith’s comments. You may be to serve as a spare batch for several comparable units to
able to save some time if you just happen to have a spare minimize the expense of the catalyst.
catalyst load for relatively small incremental cost, which is
not very likely. More importantly, you also get to compare RONALD E. MARRELLI (Phillips Petroleum):
your in-situ regeneration to a controlled third party regen- We have used ex-situ regeneration for about the last 15
eration. This option also allows complete sampling of your years, sending the reforming catalyst out to various vendors
catalyst charge for operational adjustment, performance (Engelhard and Tricat) to have the carbon burned. This has
monitoring, and replacement evaluation. been very successful at our facility. It allows us time to work
You probably want to dump and screen your catalyst on our compressors, do work on the reactors, repair scallops
every third regeneration anyway and a major turnaround and screens. Our unit is relatively small with about 40,000
is a good time to consider this. You certainly should dump lb. of catalyst. We can get it in and out in a week.
and screen if you see any signs of maldistribution, hot spots,
tails, lingering CO2 generations, or other symptoms. GARY A. STEPHENS (TRICAT, Inc.):
TRICAT routinely regenerates semiregenerative re-
JOHNSON: forming catalysts in North America and occasionally re-
UOP is aware of several companies that provide ex-situ generates CCR catalyst on both a turnaround and non
regeneration of reforming catalyst. We basically agree with time-critical basis. Our regeneration technology ensures a
the previous comments, but I would like to add that to complete carbon burn on each particle. The configuration
our knowledge, the ex-situ regeneration only does the of our plant also allows us to regenerate the catalyst fines,
carbon burn and screening of the catalyst. The catalyst still as well as the catalyst which provides our customers with
requires that the carbon burn be followed by an oxida- additional saving on precious metals recovery costs and
tion/chlorination step and reduction step to properly re- ensures very low catalyst losses. Our plant is not equipped
condition the catalyst which would then have to be done to perform oxychlorination or reduction steps. These steps
in-situ. are performed in situ following catalyst loading.

KIMBRELL: LEE E. TURPIN (Honeywell Hi-Spec Solutions):

I agree with those comments as well. We have considered If the oxidation/chlorination step was to be done exter-
doing this, and the only time it seems like it might make nally, it probably would be done at lower pressures, than
sense is if we were going to dump in the screen catalyst you would use in an actual operating plant. Could the
anyway. But we have not done ex situ regeneration. panel comment on the effect of doing those steps at lower
pressure?
EDWIN A. CARLISLE (Eurecat U.S., Incorporated):
All of the Eurecat facilities can currently provide carbon HUNKUS:
removal and screening of the reforming catalyst. The Just off the cuff, the only comment I would make is
Eurecat S.A. facility in La Voulte, France, can also perform that I think you would have to be more careful that you
the oxychlorination treatment. This step is performed in did not get excessive particle temperatures because you do
controlled conditions of temperature and partial pressure not have as much mass to carry away the heat generated,
of oxygen, water, and chlorinating agents to redisperse the and the burn and redistribution would both take longer.
platinum phase, just as it is done in continuous catalytic
reformers. Eurecat S.A. can also provide the subsequent J. B. RODDEY (Roddey Engineering Services, Inc.):
step of hydrogen reduction. The oxychlorination step is the platinum redistribution
A batch of spent reforming catalyst has recently been step. Like any other chemical reactions, it has a rate and
successfully regenerated. The carbon was burned and the the rate is proportional to the concentration of oxygen.
oxychlorination step was performed. This catalyst per- So, as the concentration of oxygen increases, and the
formed well when utilized by the refinery. Separately, other pressure would increase that concentration, then the re-
batches of fresh reforming catalyst have been pre-reduced quired time to redistribute the platinum decreases. So, yes,
in the Eurecat equipment. The quality control features the pressure would affect the redistribution of the plati-
monitored include surface area, chloride content, and num, though only from a time function.
platinum dispersion (by H2/O2 titration), in addition to
the typical physical characteristics. ANGELO P. FURFARO (UOP):
This off-site treatment offers the benefit of reduced Mr. Roddey is correct that the platinum redispersion
reforming unit down time as well as reducing inherent step is rate dependent upon oxygen concentration, mean-
corrosion problems. This processing can be done either on ing that the moles per hour are important. This is the case

1996 NPRA Q & A Session on Refining and Petrochemical Technology 181

TOC/INDEX

whether oxygen is delivered via higher pressure or higher We believe you must also use judgment and verify
concentration. Good dispersion at low pressure, but high calculations to make sure your instruments do not mislead
oxygen concentration, is apparent from the fact that UOP your optimization efforts. Always hand calculate your
has more than 120 CCR Platforming Process units in additions at least once yourself to check that you are in the
operation utilizing atmospheric regeneration. The tech- right ballpark. By this I mean order of magnitude verifi-
nology provides for proper oxychlorination conditions, cation against accepted standards.
and thus good platinum redispersion at low pressure We would recommend that you tend to run a little
conditions. under chlorided and a little on the dry side unless you
In fixed bed operations, increasing the rate of platinum really believe you have the data to convince yourself that
dispersion is usually accomplished via higher regeneration you see a real benefit for a different operation. The key to
pressures and/or higher oxygen concentration during the a good analyzer program is dedicated support, verification,
oxychlorination step. Longer time at the refiners standard and buy-in by the operations and maintenance team. If
oxidation conditions can also improve platinum disper- the instruments suddenly predict a sharp change or rates
sion. Prior to increasing the oxygen concentration during
drift say 10%, the operators should feel comfortable with
the oxidation step of the catalyst regeneration, refiners
backup chart type systems and ask questions. You should
should review the capacity and capability of the recycle gas
machines and attendant seal gas systems with their com- not have to make sudden or frequent adjustments of your
pressor vendor. In addition, refiners with cold wall reactors chloride and water addition rates. Small changes will take
should remain conservative with regard to increasing oxy- months to equilibrate. I feel the most likely tendency of
gen concentration due to concerns over possible internals new engineers is to over control, over chloride and operate
damage via combustion of pockets of coke which could too wet resulting in less than optimum run length and
exist behind the reactor liner. yield performance. I know I did. Do not let the low
start-of-run temperature fool you.
Question 24.
Is there any noted success in continuous measure- JACKSON:
ment of recycle hydrogen chloride composition, with I agree with Mr. Davis and Mr. Hunkus. We use
respect to catalyst cycle optimization? moisture analyzers. We are not aware of anybody that uses
chloride analyzers. Exactly what Mr. Hunkus said, setup
DAVIS: your water chloride rates and leave well enough alone.
CENEX uses color indicator tubes manufactured by The feedback from our refineries and the reformer
Drägor to obtain recycle hydrogen chloride content once experts within BP would not recommend that the refiner
a shift. This is not a continuous reading, but suffices to continuously measure the chloride level in the recycle gas
monitor chloride injection. The chloride content com- stream. There are better areas to focus attention that will
bined with the continuous moisture analyzer allows lead to optimal catalyst cycle times.
CENEX to maintain proper water chloride balance which We recommend that the water content of the recycle
is important for cycle length. gas is continuously measured and the chloride level is
tested intermittently, once a day, by the support engineer.
DiCAMILLO: The water and injected chlorides for a given operation
We are not aware of anyone using a continuous chloride equilibrate on the reformer catalyst. Slugs of water, nor-
monitor. I agree with Mr. Davis. Frequent analysis, either mally from the upstream hydrotreaters, are the usual
on a shift or a daily basis, will give you early detection of
upsets that affect the chloride balance.
possible upsets which will help prolong your life in the
reformer. First and paramount is to ensure the upstream opera-
tion of the hydrotreat unit is stable. Once this is assured,
HUNKUS:
the water and chloride dosage rates are known and set up.
Yes, I know a major that has had very good results Once set up, leave well enough alone and allow the
optimizing cycles, surface chemistry and selectivity with a chlorides to equilibrate on the catalyst. Do not fiddle! A
properly maintained system, including moisture analyz- support engineer that continually changes the dosage rate
ers, on-line catalyst sampling, rapid process monitoring of the chlorides or the water will end up chasing his own
and elaborate kinetic modeling. For a small independent tail. The system needs time to equilibrate. Either too much
refiner, this seems like a lot of effort for results that can be or too little chloride on the catalyst will shorten the life of
more easily and reliably achieved with traditional methods the catalyst cycle.
for traditional reformers operating within normal limits. So, stabilize the upstream hydrotreater, continually
But this can be very important when defining new areas monitor the water, Draeger the chloride level, and leave
of operation or exploring unusual feedstock. well enough alone.

182 Light Oil Catalytic Processing

TOC/INDEX

JOHNSON: octane crunch returned, we could go back to 5 to 51⁄2 days,

A continuous chloride analyzer has been installed and and achieve good regeneration with good dry outs.
is monitoring HCl in the recycled gas on a semiregenera- The key to minimizing regeneration times are:
tive reformer in the Southwestern U.S. It was placed in 1. Dual front, high pressure burns (150 psi to 200 psi).
service at the end of 1995. The analyzer has been reliable 2. Use double blocks and bleeds to allow blinding
and tracks HCl over a range of up to 4 ppm. The analyzer without depressuring.
response to moisture upsets has also been rapid. It is 3. A recycle gas drier is a must (reduces dry out to 24
interesting that sudden increases in the chloride reading hours).
tends to be a good indicator of the water increase in the 4. First class, reliable Oa, moisture, and pH analyzers.
recycled gas. 5. Operator knowledge, commitment, and coordi-
nation.
SMITH:
I agree that continuous monitoring is not necessary for BARKER:
good chloride control. Good chloride control can be Utilizing parallel burn facilities, reduced our “oil to oil”
achieved as mentioned by: 1) using Drägor tube measure- cycles to 6 to 7 days with a 5 to 6 day dry out period. We
ments which we do 3 times weekly; 2) we also control by also are octane long, so there has been no economic
collecting on stream catalyst samples approximately once advantage to really to reduce this cycle time any further.
every 4 weeks; 3) continuous measurement of the recycled Each of the regenerations does include a water wash of the
gas moisture using an on stream analyzer and; 4) careful stabilizer and we have eliminated the sulfur removal steps
monitoring of any shifts of C5+ to LPG production. By since we have upstream sulfur absorbers.
using these techniques, Ras Tanura Refinery has been able
to maintain chloride on catalysts at 0.92% with the stand- DAVIS:
ard deviation of 0.05%. Chloride is injected into the feed Using Mr. Arndt’s definition, a fast regeneration for
charge at about 0.6 ppm to 0.9 ppm while methanol CENEX is about 4 days of actual regeneration and about
injection is based on maintaining about 5 ppm of water 3 to 4 days of dryout time.
in the recycled gas. We believe there is no yield benefits in
a tighter control of chloride than we practice. DiCAMILLO:
There are a few other tricks you can use to speed up the
Question 25. time from oil out to oil in. The steps themselves are pretty
For regeneration only shutdowns, on semiregenera- standard, but there are some tricks. If you run at higher
tive reformers, what is the average time between “oil pressure during the regeneration, it will speed up the burn.
out” and “oil-in? What steps do you normally include You can use liquid nitrogen cooling where appropriate. We
in your standard regeneration (compressor have already mentioned parallel burns. One thing that has
washes,...)? What changes have you made, or are you not been mentioned yet is to anticipate the next step in
considering, to reduce the “oil to oil” time? your process. If you anticipate and think ahead and have
everything ready for the next step, you can save a lot of
ARNDT: dead time in going from one step to the next step. We
First we have to define regeneration time. The question found a lot of time savings just doing things like that.
here says oil out to oil in. I will redefine regeneration time
to mean oil out to dry, high severity operation. Any times HUNKUS:
I quote will be based on this definition. I like to include My definition of oil to oil is oil to oil, with all due respect
dry out since it can be very time consuming. We went to my distinguished colleague. My experience is mostly in
through 3 separate eras in Chevron on regeneration time. the 6 to 8 day range. I have also been part of a team where
There was the dawn of history when it would take us 10 to we found we could consistently go oil out to oil in, in just
14 days. In the late 1980s or early 1990s, in what I call the over 5 days. This requires properly maintained equipment
octane crunch time when we were running many of our and around the clock technical support on-site. For the dry
units at cycle times of 8 to 12 weeks (high feed rates, high out, we typically run 4 to 5 days. I also will add that if you
octane, and bad feed), we got regeneration times down to do not have a moisture analyzer, you can watch the amps
4 to 4 1⁄2 days. We did have a high incidence of poor on the recycle compressor. This gives you a pretty sharp
regenerations and when you have a run length of 8 to 12 indication of when your system suddenly dries out and
weeks and you have a poor regeneration you can get into a hydrogen purity jumps up. Your flow meters should also
real snowball where you are regenerating all the time. respond to the change in molecular weight.
We have now taken more care and with oxygenate use, After blinding and evacuation steps, we always include
we are no longer in an octane crunch. Our regeneration a compressor wash and feel this is very important. Once
times are up in the 6 to 7 day range. I think that if the we failed to do a good wash and we paid for it with high

1996 NPRA Q & A Session on Refining and Petrochemical Technology 183

TOC/INDEX

vibrations that risked major damage to the recycle com- LEMMON:

pressor and added several days while we took the machine Our experiences are very similar to those of the other
apart to look at it. We generally do some feed/effluent panelists. We have gone to higher pressure regeneration
cleaning in the naphtha hydrotreater, wash some towers with two stage burn. Our semiregeneration unit is only
and do PM on the reformer complex major equipment regenerated about once a year, so we take our time and do
such as the lube oil skids and the distillate compressor it carefully. We take about 6 days for the regeneration step
system along with the standard UOP regeneration. and about 5 days to dry out.
We have done all the normal things like multi point air
injection, moving to higher pressure, going with a larger SMITH:
compressor motor, using high pressure nitrogen for dryout Our typical oil in oil out is 6 to 7 days, with another 3
after oxidation, larger vacuum jets and spectacle blinds. days for dry out. Our standard regeneration includes a hot
We also have two turnaround coordinators lead the regen- strip, the carbon burn (oxidation) step, rejuvenation, re-
eration and work around-the-clock to ensure the best duction and pre-sulfiding. Compressor washing is only
done if performance drop indicates the need which we
coordination among all support resources, including con-
have not done for the last 5 years.
tractors. These coordinators, who are generally our top
The most significant recent change to reduce time and
operators, lead the planning of all activities before the improve regeneration quality is to allow up to 5% oxygen
regeneration with the support team involved. I have also during the secondary carbon burn if exotherms and oxy-
seen greater than 300 psi separator pressure and liquid gen consumption tests permit. Oxygen content during
oxygen and liquid chlorine used to further reduce regen- the primary burn is still rigorously controlled at 0.8%
eration time. This type of operation requires special exper- maximum.
tise and increased safety equipment and awareness. Previously, secondary burn was constrained to 2%
maximum oxygen due to concerns about localized coke
KIMBRELL: ball temperature excursions. Additionally, the standard
Depending on how much maintenance needs to be UOP hot circulation reduction has been implemented
done, we go typically 7 to 9 days oil out to oil in, and then instead of the previous stagnant procedure.
we take another few days (3 or so) to dry out. We do not
normally wash the compressor during regenerations. De- KHALID AL-REHAILI (Saudi Aramco Refinery):
pending on how much sulfur is laid down on the catalyst Do you not think that the amount of catalyst is playing
during the run, we may elect to regenerate twice. This a role in the regeneration duration?
double regeneration does an excellent job of removing
sulfur from the system. SMITH:
Certainly, the amount of carbon that you have on the
KOOIMAN: catalyst makes a difference. That would probably be the
My experience is similar to that of Mr. Hunkus and Mr. limiting factor.
Kimbrell, about the same times for oil out and oil in and
dry out. The only thing I will add is that we found it TOM KELLY (ARCO Products Company):
worthwhile to monitor the piping at the caustic injection The more catalyst you have the longer it takes to
point. We had some trouble in the past with that. Also a regenerate. But regeneration time is also dependent upon
good electrical preventive maintenance program is impor- the recycle rate, and, of course your pressure.
I have one comment about short coke burn times after
tant. It pushed us towards doing preventive maintenance
a full cycle. If there is not much coke on the catalyst to
on the recycle compressor.
burn off, you may have something else deactivating your
Regarding higher pressure my experience, especially catalyst. So before bragging about short regeneration time,
working with Mr. Arndt in the past, is that higher pressure consider that you may have a hidden problem.
is generally a faster regeneration. But we found that we
were compressor limited in one case. We got above the ARNDT:
pressure that would exceed the recycle compressor amper- We achieve our short regeneration times with high
age limits and the operators would start closing off on the coke-on-catalyst. We usually reach EOR with last reactor
butterfly valve on the compressor inlet. Looking at the carbon levels of 18% to 20%. At times we have exceeded
curves we found that when the suction valve was partially these levels at EOR but do see a real yield penalty.
closed, the loss in recycle exceeded the gain of the higher
pressure. The fastest burn was achieved at a pressure that J.L. RAINA (Indian Oil Corporation Ltd.):
allowed maximum compressor amperage with the com- On the dryout there are two phases. One is the first
pressor inlet butterfly valve wide open. reduction up to a certain level, maybe 100 ppm or so.

184 Light Oil Catalytic Processing

TOC/INDEX

Another is for going from 100 ppm below. Most of the

time is taken in going from 100 ppm and below.

HUNKUS:
I agree with you, and it is very important not to go to
high severity until you do get the moisture down. One of
the changes that we have made in the last 5 years or so, is
go to maximum charge rate at low severity to speed up that
dry out time. In the past, we restricted the charge rate back
so that they could keep the temperatures down and go to
premium production immediately. What we do now is go
to maximum charge rates and hold the severity down. We
do that with inventory management up front. We are not
forced into making high severity right after regeneration.

FRANK ALLEN (Conoco Inc.): unit. We think a combination of catalyst change out and
My question goes back to the speed or the shortness of mechanical modifications resulted in the fines reduction.
the regeneration, and Mr. Arndt can probably answer this I really interpreted this as two questions. The second
best. What hydrogen-to-hydrocarbon ratios are you run- part of the question is what advances in reforming catalyst
ning on the run? This goes back to the compressor size, technologies have been made in the last 5 years? I inter-
oxygen limits, and inert circulation rate. What is your preted that to mean both in CCR and fixed bed reforming.
typical hydrogen-to-hydrocarbon ratio on the run that A couple of the more obvious ones are increasing surface
you can get these kind of quick burns? area of the catalyst to extend catalyst life, running with
staged catalysts, changes in promoter metals to make
ARNDT: improvements in yield and stability, improved selectivity
of the catalyst and use of L-zeolite supports for C6 and C9
We were running down as low as 3 to 3½. That helped
conversion.
give us those 8 to 12 week runs. But our feed rate was way
up. At design feed rate, I think the compressor was designed
JOHN D. COOK (Acreon Catalysts):
for more like 5 to 6. Obviously, we do not run at design
Several refiners have converted to our new generation
feed rate anymore, but the compressor is good-sized.
CCR catalyst (AR series and CR series) to provide excel-
lent selectivity, stability, low fines make, and most impor-
Question 26. tantly, improved surface area stability. This catalyst
What has been your operating experience with the provides relief from the prior generation catalyst limits
newer generation of CCR catalyst aluminas? Has any- allowing them to achieve higher throughputs by over 15%
one documented a decrease in fines generation? to 20%. This is extremely important today to increase
What advances in reforming catalyst technologies hydrogen production for treating and upgrading units for
have been made in the last 5 years? increased refining margins.
Acreon/Precatalyse also supply a unique tri-metallic
JOHNSON: catalyst for fix bed reformers. This catalyst provides im-
The fines make for the R132 series and the R30 series proved hydrogen and liquid yield for reforming units.
catalysts have been very similar. This is based on a statis- Over 13 units are currently in operation and these selec-
tical analysis comparing the operation of over 50 CCR tivity improvements have been shown over repeated regen-
platforming units operating with the R130 series catalyst eration cycles, as well as through several severe sulfur
versus the R32, R34 series catalysts. The benefit of the upsets. Our trimetallic RG-582 offers a now proven solu-
R130 series is a better retention of the surface area in the tion to improve yields without compromises in cycle
CCR service. length, ease of regeneration or tolerance to upsets.

DiCAMILLO: Question 27.

We would attribute fines generation both to catalyst What are the advantages/disadvantages of adding
quality and mechanical problems on the unit. In some cases continuous regeneration to the test reactor of an ex-
we have noticed fines reductions of 10% to 20% on CCR isting fixed bed reformer versus replacing the entire
units that have switched to Criterion catalyst. We will not fixed bed unit with a CCR unit? Is the performance of
give all the credit to the catalyst there. We also made the hybrid unit close enough to the CCR to make this
recommendations on mechanical modifications to the option economically competitive?

1996 NPRA Q&A Session on Refining and Petrochemical Technology 185

TOC/INDEX

JOHNSON: reactors. Therefore, you increase the yields and increase

Many customers have compared the costs and benefits hydrogen production.
of converting their fixed bed unit to either a hybrid or a There are some disadvantages to the hybrid system in
CCR platforming operation. There was a paper presented that, of course, the yields are slightly lower than for a new
at the 1989 NPRA Annual Meeting (AM-96-47), “Con- low pressure continuous unit. In terms of a delta yield fixed
version of Fixed Bed Reformers to UOP CCR Platforming bed unit — new LP unit — you get about 60% to 70% of
Technology” by M. W. Golem, N. L. Gilsdorf and D. E. the benefit with the hybrid system. However, it only cost
Felch that provides reference to the subject. about one-third of what the new system would cost. I think
The main disadvantages of hybrid technology relative you have to look at both yields and economics.
to CCR technology are: The other advantage of the hybrid system, when you
Lower on stream factor. The hybrids still retain a install the new reactor with the regeneration system down-
semiregenerative reactor in front necessitating re- stream of the separator, is that because this system is
generation every 1 to 2 years. completely independent from the existing reaction train,
Higher operating pressure. Pressure reduction in a you can install it without shutting down the existing unit.
hybrid system can be less because you have actually The tie in takes only about a week. You have no limitation
added pressure dropped to the reactor circuit if a due to the existing recycle compressor.
heater and reactor were added.
Lower yields. The yield structure of the hybrid cata- LEE E. TURPIN (Honeywell Hi-Spec Solutions):
lyst system is basically equivalent to the platinum I have done several studies for our clients using our
semiregenerative catalyst. So the benefit of the CCR kinetic model (REF-SIM), and it appears to me that the
platforming platinum tin system is not achieved. yield benefit of putting in a hybrid system is about 60%
This benefit amounts to about a 1.5% to 2% yield to 80% of what can be achieved by going to a CCR, just
advantage at equivalent pressure. The reason for this due to the pressure reduction on the last reactor. This
is because the yield structure is fixed by the less variability, of course, depends on how far the last reactor
selective platinum in the front reactors. pressure is being dropped. This does not take into consid-
Additionally, hybrid units will exhibit a yield decline eration the effect of changing the catalyst in the lead
across the semiregenerative cycle compared to more con- reactors from a platinum-rhenium to a platinum-tin as
stant yields of UOP CCR. We have completed 28 revamps noted by Mr. Johnson.
of fixed bed reforming units to CCR platforming, with 23
Question 28.
in operation and 5 in design and construction. Half of
Are the new generation CCR catalysts significantly
these units were conversions of side-by-side reactors and
improving aromatics yields?
the other 14 were conversions of units that had already
employed the stack reactor operating in the semiregenera-
DiCAMILLO:
tive mode.
There are new generation CCR catalysts available, but
so far they have not demonstrated improvements in aro-
KIMBRELL:
matic yields versus the older catalysts. We do however,
We have looked at this option to debottleneck our
have a catalyst in fixed bed service, PR9, that has given
reformers, but the conversion costs are about half of a new
increased xylene yields.
CCR, and the yields were not as good as with a new CCR.
We elected not to do it. We felt it was not economical,
especially with the aromatics constraints in gasoline now.

DiCAMILLO:
I would agree with those comments.

JEAN-LUC NOCCA (HRI, Inc./IFP):

I would like to comment on the hybrid system for
reforming. There are two ways you can implement a
hybrid system, either by adding a reactor in series with the
existing reactors (DUALFORMING), or adding a new
reactor downstream from the separator (DUALFORM-
ING PLUS). In all cases, you put a platinum tin catalyst
in the new reactor linked to a continuous regeneration
system, and you can lower the pressure of the existing

186 Light Oil Catalytic Processing

TOC/INDEX

JOHNSON: a nitrogen blanket system. We do not experience any

UOP does not expect any difference in the aromatic problem such as exchanger or reactor fouling when utiliz-
production for the platinum tin catalysts like R-32, R-34 ing this sweet naphtha.
and R-132, R-134. These catalysts do, however, show an
improved aromatics yield over the platinum rhenium Question 30.
catalyst systems. Are there any procedures available for chemical
cleaning the shell side of the feed/effluent exchangers
Question 29. in catalytic reformers?
What is the maximum length of time one can keep
desulfurized naphtha in a tank for processing in cata- SMITH:
lytic reformer if the bromine index is in the range of Our experience is that the major cause of effluent
50 to 100? exchanger fouling deposition is a sublimation product
from the reactors to the lower temperature effluent ex-
KIMBRELL: changers. In our case, deposition of these compounds
Even though the bromine index of 50 to 100 is very occurs primarily during high severity operations onto our
low and should not be a problem storing in tankage, stripper reboiler shell which is heated by the reactor efflu-
oxygen contamination of the hydrocarbon stored in tank- ent downstream from the other feed/effluent exchangers.
age can come from either blanket gas or from leakage These organic foulants are generally polynuclear aromatics
around the seals. Once the oxygen dissolves in the hydro- with many coke like characteristics. As a result, they would
carbon and that material is brought back into the reformer, be very difficult to chemically remove.
it can react with more olefinic feeds in the unit and cause Additionally, the suspected carcinogen PNA com-
plugging. We have actually done some of that. We have pound “coronene” has been identified as a component of
also stored untreated coker naphtha in tankage for as long our deposits. Therefore, all resulting wastes from cleaning
as a month. The tank was blanketed. We would not these exchangers has special handling and disposal require-
necessarily recommend that, but it has been done. ments. Given the nature of the deposition, we are forced
to mechanically clean the exchangers which also mini-
JOHNSON: mizes the amount of waste that is sent to our Class I
The only thing I have to add to that is on test methods. disposal site. At our current lower severity operation, this
Stored naphtha can be checked for stability using ASTM type of fouling in our effluent exchangers has not been a
D381, existent gum and motor gasoline and aviation fuels problem.
by the jet evaporation test. A more precise method to
predict future stability is to use ASTM D873, oxidation JOHNSON:
stability by potential residue method. This involves run- My comments will regard pulling combined feed ex-
ning the naphtha for 4 hours and then evaporating the changer bundles primarily from our vertical bundles in
stressed fuel by ASTM D381. A residue of less than 5 CCR service. In this case, the feed is on the tube side. It
mg/100 ml will correlate about 85% of the time with a 1 is not at all common to pull the combined feed exchanger
year ambient storage. bundle unless there is cause. Tube side fouling has been
due to feed contamination typically over use or misuse of
DAVIS: the corrosion inhibitor. Care should be taken when remov-
All I can really add is our own experience at CENEX. ing the tube bundles. Damage to tubes has occurred when
CENEX has a cone roof tank that stores reformer charge the bundle has been laid in the horizontal position.
for up to a year before we use it. I too suggest looking at
a gum stability test to give an idea of how long you can KOOIMAN:
store your feedstock. I have been told that you can use an acid wash with a
dispersant for inorganics or hot oil flush with organic
DiCAMILLO: dispersant for organic deposition. I would probably get
Just as a further point of clarification on bromine index with a reputable chemical vendor if I was confronted with
versus bromine number, a bromine number of 1 equals a that.
bromine index of 1000. So we would agree that an index
of 50 to 100 is very low and it should be stable. DiCAMILLO:
Cleaning heat exchangers cannot be avoided in a re-
HASSAN A. MAGHAZEL (Saudi Aramco): former. Having said that, there are chemical procedures
We have one sweet naphtha storage tank. Our former available. The feeling seems to be that they do not work
cycle length is 1 year. The sweet naphtha is kept in the tank very well. The main problem is that the chemicals will
for long periods without being used. Our storage tank has loosen up the scale but not dissolve it or remove it

1996 NPRA Q & A Session on Refining and Petrochemical Technology 187

TOC/INDEX

completely. When the refiner starts the unit back up, that could cause more hydrocracking reactions to occur in the
scale carries over into the first reactor and plugs up the reactors.
reactor. The other possibility is salt accumulation in the
strainer. It could be sodium chloride from the neutralized
J. B. RODDEY (Roddey Engineering Services, Inc.): products of the regeneration. The separator should also be
We can remember when bimetallic catalyst first came checked to see if there is entrainment or carryover.
out. The big concern was sulfur because we knew those
catalysts were more sensitive than the all platinum catalysts ARNDT:
that we had been working with in earlier years. So a I have a dumb answer. We have taken them out, so they
number of units were acidized including the feed/ effluent do not plug. In fact, we do not even know if they would
exchange to remove scale rather than pull and clean by plug.
hydroblasting. Mr. DiCamillo’s comments were experi-
enced in many of these units. In other words, the scale DiCAMILLO:
would just simply loosen up and go into the first reactor I will take the opposite approach. An obvious observa-
and then foul the first reactor’s scallops in the case of a tion here is that if your strainer is plugging, add a parallel
radial flow or foul the top of a bed in the case of a down strainer and you can switch back and forth and solve the
flow. problem that way.

R.E. (Ed) PALMER (Mustang Engineering, Inc.): Question 32.

One client we have has vertical combined feed exchang- Has anyone experienced shots of sulfur containing
ers which have been in service about 25 years. They clean feedstock to their CRU? What are the relative econom-
them about every 3 years after performance falls off to ics of regeneration of the catalyst, hot hydrogen strip,
roughly 60% to 70% of design. I have another client that or allowing the sulfur to be stripped off during opera-
tion? What procedures are used in operation of a CRU
has had their units in service about 16 or 17 years. They
system contaminated with sulfur?
have never cleaned them and again these are vertical
combined feed exchangers. It does limit their ability in
HUNKUS:
terms of charge rate and octane.
We have experienced large and small shots of sulfur into
our reformer on initial start-up after regeneration, and
Question 31.
with the naphtha hydrotreater stripper upsets during
During regeneration and after start-up when recycle
power outages. We have also suffered sulfur breakthrough
gas drier is taken in-line, compressor suction strainer
with the loss of hydrogen makeup and purity in the
frequently gets choked. Are there any new types of
hydrogen generation unit. We have even mistakenly
strainers available for quick change over?
charged sour feed into our reformer for about 10 hours
while trying to run sweet feed for a quick naphtha hydro-
JOHNSON: treater exchanger cleaning.
When a semiregenerative reformer starts up after cata- If you see the problem very quickly or are lucky as we
lyst regeneration, the recycle gas is wet. It is recommended were and have or get your reactor temperatures down low
that the operating severity of the unit be maintained at a enough, about 890°F to 900°F, you will see almost no
low level until the water concentration in the recycle gas effect on unit performance. The sulfur will strip off sur-
is sufficiently reduced. This process can take approxi- prisingly quick. If you hit the unit so hard it will not
mately 5 days to accomplish. The time off-line can be generate hydrogen, then you are regenerating. Talk to your
reduced to about 2 days by incorporating a start-up dryer licensor; UOP has always been very helpful for us. We have
in the recycle gas circuit immediately following the catalyst seen as much as a 70% decline in off gas and delta Ts have
regeneration. dropped about the same amount. We experienced a near
We do not have any information with regard to new total recovery in about 31⁄2 days. That is with our current
types of strainers. However, I will point out that if a cycle on R62 which is a year next week with about a 5°
strainer is used, a fine mesh strainer should not be used. loss in start-of-run (SOR) temperatures.
UOP would recommend that if a strainer is in the com- The basic procedure is to strip sulfur off by cutting
pressor section at all, it would be a course mesh strainer. reactor temperatures to 890° to 900° F, keep charge rates
The second point is, what is plugging the strainer? If and pressures as high as possible, gas as much hydrogen as
it is absorbent from the dryer, the dryer will need to be possible and watch the H2S in the off gas and in the
inspected to identify the cause of migration of the absorb- reformate. I have seen more than 20 ppm in recycle gas
ent. If the absorbent actually is leaking from the dryer and with no detectable long term damage to the cycle perform-
could get into the reactors, this is of some concern. It ance. If your operators leave the temperatures up and coat

188 Light Oil Catalytic Processing

TOC/INDEX

over the platinum with carbon, the hot hydrogen strip may JAMES W. JONES (Turner Mason & Company):
buy you a week or so to get ready, but you are still going Is there any reformer operator or process licensor that
to have to regenerate. has an operating procedure that when you lose your recycle
compressor does not call for the immediate pulling of the
KIMBRELL: feed along with shutting down the heaters?
Yes, I agree with most of those comments. We will also
say that the condition that the catalyst was in to begin HIGGINS:
with, before you have this sulfur upset, will also determine No response.
how well a catalyst comes back. So if you are near a start
of run to begin with, it has a better chance of recovering. KHALID AL-REHAILI (Saudi Aramco):
If you are near the end of run, it probably will not recover. My question is for Mr. Johnson from UOP. The semire-
We continue to feed the unit to try to purge the sulfur generative reformer units using UOP catalysts have
out of the system, similar to what Mr. Hunkus does. We changed to staged catalysts. What is the driving force for
have also installed sulfur guard beds prior to the first this change? Is it yield improvement or higher octane
heater in our recycle loop, and that helps us recover from requirements?
a sulfur upset.
ANGELO FURFARO (UOP):
LEMMON: Refiners have been changing to UOP’s R-72/R-56
We also occasionally have the unexpected pleasure of staged loading system for its high C5 plus yield perform-
finding sulfur in the reformer feed. Our procedures are ance, as well as its long catalyst cycles. The staged loading
similar to what has been said before. The most important system is an optimal combination of UOP’s high yield
thing is to cut temperatures as quickly as possible. We do R-72 catalyst, with the excellent stability of our skewed
operate all three types of reformers. They each respond a platinum/rhenium R-56 Platforming catalyst. For more
little bit differently. The CCR reformer, with its high information contact your UOP Customer Sales or Service
hydrogen yield, seems to recover the quickest and is least Representative.
sensitive to sulfur. In the past we have just cut 10° or 20°
and ridden it out. We do maintain charge rate at a high
rate to purge sulfur more quickly. In a CCR reformer you D. Aromatics Extraction
will form some sulfates in the regeneration step, so it may
take several cycles to get all of the H2S out of the system. Question 33.
For the swing reactors in our Powerformers, they do the What new solvents are available for improving aro-
same thing, cut temperature. They may delay regeneration matics extraction? What improvements are they sup-
to allow more of the sulfur to strip off. Our semiregenera- posed to provide, and has operating experience
tion units are operated similar to what the other panelists verified these improvements?
have said.
DiCAMILLO:
ARNDT: A solvent I am familiar with is NMP. Its advantages are
We keep the unit on stream to strip out the sulfur it is non-flammable, it improves separation, and it has a
before regeneration. On a few occasions, we have com- better temperature tolerance. I understand there may be
missioned the recycle dryer to help remove sulfur from some co-solvent systems on the market, but I do not have
the system-worked well. So if you have a dryer consider
using it to remove sulfur.

DiCAMILLO:
The type of catalyst you have in the unit and the
magnitude of the sulfur excursion will have some influence
on what you decide to do. A preferred order of corrective
action would be to continue with the clean feed and take
the octane drop if you can tolerate it. You could perhaps
increase your chloride injection to compensate. The next
thing we would try would be a hot hydrogen sweep. A last
resort would be a regeneration. I would like to reiterate
what Mr. Lemmon says. In the case of a sulfur excursion,
you might need to add an extra sulfate removal step during
regeneration, so that would extend the regeneration time.

1996 NPRA Q&A Session on Refining and Petrochemical Technology 189

TOC/INDEX

any feedback on them. I am sure this is an appropriate FEARNSIDE:

vendor question and we will get some floor response on I agree with what both Mr. Hunkus and Mr. Johnson
this. have said. We usually recommend that you mitigate any
oxygen ingression. If you cannot find it straight away, we
JOHNSON: recommend using oxygen scavengers until the problem
UOP offers Carom as an upgrade alternative for the can be found and corrected. Another thing that a lot of
Udex and Tetra operations. The main benefits being ca- people tend to forget is the thermal degradation products
pacity, purity and reduced utilities. For a new unit, we if the fractionator reboiler skin temperatures get too hot.
would expect a 15% capital cost savings for Carom versus
Sulfolane. To date, all operating units have been revamps, KENNETH G. TASKER (HRI, Inc.):
however, the first new Carom unit is expected to start up I can endorse everything the panel has said. Tempera-
next spring. ture and oxygen are the main ingredients in forming the
degradation products. Keeping the oxygen out is the
Question 34. number one consideration, along with keeping the tem-
In aromatics extraction using sulfolane, how are you perature within a limitation of about 180°C to 185°C. We
handling the degradation products? have one unit that has gone a complete year with no
degradation product at all. We also have on-line regenera-
HUNKUS: tion and on-line removal of degradation products. For
In general, we believe you should first focus on the best those not so tight units, we have seen a rejection rate of
operation of your process to minimize production of the up to about a barrel a month. On naphtha cracking
degradation products. We have seen very little of the complexes, the degradation product is generally disposed
degradation products that were initially predicted from of with the C9+ aromatic stream into the fuel oil. So that
our regeneration unit. We generate less than 100 gallons is a nice easy disposal. The alternative is, as reported, to
per year of this material, which we dispose of as hazardous get rid of it as a hazardous waste. Generally speaking, we
waste. are not talking significant quantities.
What we have seen is a very fine black particulate which
we filter out of the Sulfolane stream and also from the BILL MENHEERE (Nova Chemicals (Canada) Ltd.):
aromatic stream. These particles are indicative of corrosion In a Tetra unit, is anyone using a non-silicon based
which has occurred at the flash region of our reboilers on antifoam with success?
the extract stripper and the recovery column. This corro-
sion is not evident anywhere else in the unit. HIGGINS:
We process raw virgin naphtha with its inherent higher No response from the panel.
levels of contaminants which is considered unusual and
difficult by the licensor. We also operate in the tempera-
ture range conducive to the formation of degradation E. Polymerization
products, but we believe we also run a tight unit, prevent-
ing the introduction of oxygen to the system. We do feed Question 35.
our extraction unit directly from an upstream crude unit We have recently lost the only vendor who treated our
as opposed to having an intermediate feed tank. This will spent Dimersol catalyst. Does anyone have any sug-
definitely reduce the potential for oxygen addition to the gestions for disposal. If in-house neutralization is
system. We currently do not add any oxygen scavenger. done, what is done with the alumina sol, and how is
it handled once formed? Any advice from the catalyst
JOHNSON: companies on handling?
In our designs, we are handling the degradation prod-
ucts in sulfolane by removing heavy components via dis- KOOIMAN:
tillation in the solvent regenerator. We add MEA to the We have a polymerization plant at Pine Bend and
unit to neutralize acids that are formed. We provide filtra- another one in Corpus Christi. Any sludge that is left in
tion in the designs on the lean solvent circuit to remove the vessel we will take out during turnaround as hazardous
some of the solvent formed by the degradation of the waste. We found there is not a lot of information in the
solvent and the solids created by the corrosion of equip- industry on polymerization plants. I would appreciate any
ment. We have upgraded the metallurgy of the hottest contacts in the industry.
surface exposed to the solvent in the recovery column
reboiler to reduce the corrosion there. Every 4 to 6 PATRICK HOH (HRI, Inc.):
months, the regenerator must be emptied and the material There is a company in Louisiana that has routinely
discarded as waste. taken the Dimersol spent catalyst for disposal. If the

190 Light Oil Catalytic Processing

TOC/INDEX

Dimersol licensee has a wet gas scrubber in the FCCU, ARNDT:

HRI suggests that this spent catalyst can be routed there The main problems have been with chloride migration
for disposal. If Dimersol catalyst is treated by neutraliza- to downstream equipment, usually because we have not
tion and filtration steps, the filtrate can be put to the done a good job scrubbing. This occurs because we have
refinery waste effluent stream. The filter cake can either not paid attention to, or in a couple of occasions have
be sent to the landfill or to the hazardous disposal site shortcut UOP’s design and screwed up the design on our
depends on local government restrictions. own.
Another problem area is electric heater reliability. We
have knocked them out due to moisture ingress. Nitrogen
F. Isomerization purging of the energized components has gotten us out
of that.
Question 36.
What has been the industry experience with UOP’s BARKER:
I-12 isomerization catalyst? How has it affected sulfur One of our isomerization units has very few mechanical
tolerance, run length, activity, etc.? problems, but large corrosion problems associated with
our net gas scrubber. It is a Venturi scrubber and has never
JOHNSON: performed to design. The heaviest corrosion that we find
UOP has loaded four units with the I-12 catalyst. All is on the welds in the heat affected zones and we are still
four units have operated well. One of the four units has battling that issue.
done a sulfur strip operation and the catalyst has recov- On a second isomerization unit, corrosion issues are
ered. Catalyst activity and stability is similar to I-8. At this extremely minor, but we have had numerous mechanical
point in time, all of the units loaded are still in their first reliability issues. Initially, our automatic on-off gate valves
catalyst cycle. were very unreliable. The issue turned out to be low air
pressure to the solenoid valves used to activate the pistons.
LEMMON: We initially were operating with this air supply pressure at
One of the four units Mr. Johnson is referring to is 100 to 110 pounds and increased the reliability of the
located at our Bayway Refinery. UOP I-12 catalyst was valves quite a bit by merely increasing the air pressure up
loaded into both the Bayway units’ butamer reactors. to 115 pounds.
Following start-up, they did observe some rapid catalyst A second major problem we have is with our recycle
deactivation, or what appeared to be deactivation, in the hydrogen compressor mechanical seals which have very
first reactor. The suspected causes of the deterioration were short life spans. They lasted only 4 to 6 months after initial
oxygenates breaking through the molecular sieve beds or start-up. Seals have been modified and now have a longer
possibly reactor flow maldistribution. No unusual catalyst life, but different compressor problems. Currently, the
deterioration was noticed in the second reactor after it was compressor loses 1 to 2 drums of seal oil a day to the seal
brought on-line. Sulfur in the reactor feed has always been pots.
maintained at levels below 0.5 ppm. A third problem was with our original feed dryer which
had numerous level switch failures and poor dryer water
Question 37. removal efficiency. A new, larger capacity feed dryer was
What reliability problems are refiners having with installed and the current operation is satisfactory.
their isomerization unit (i.e. corrosion, leaking cool- And last, power disturbances have shut down our ID
ers, etc.)? Have these problems been identified, and fans quite a few times. This triggers an entire unit shut-
what solutions have been developed for them? down. We are currently tackling shutdown logic to modify
the latter logic and resolve this issue.
FEARNSIDE:
The biggest problem that we have run across is related HUNKUS:
to the feed dryers during the regeneration phase as the Since we operate a “baby” TIP isomerization unit, we
chlorides come off the molecular sieves, especially during do not have the corrosion and other problems associated
the initial heat up and dry off of water with the chlorides. with a PENEX unit, but we also do not get the higher
The downstream piping, exchangers, and associated octane or lower operating costs. We have had very few
equipment has been corroding quite rapidly. problems with our isomerization unit, as is my under-
What we have recommended that people do is inject a standing of the general industry practice. Our isomeriza-
dilution steam or install a dilution steam line downstream tion unit typically runs too much heavy feed and
of the feed dryers, and at the same time inject an appro- produces a lot of cracked material from C7 feedstock
priate neutralizer into that steam to handle the chloride components, sometimes C7+. We have also had problems
load that initially comes off. with our cycle sequencer, but this was just a control

1996 NPRA Q & A Session on Refining and Petrochemical Technology 191

TOC/INDEX

instrument application problem. The problems are mini- ARNDT:

mized greatly by not skimping on the switching valves. This is a tricky question because there are five parts. So
High quality valves more than pay for themselves with I am going to march through them one at a time. We have
reduced maintenance and better on stream factors. 3 Butamers in the U.S. We have seen molecular sieve life
I know of a refiner that lined the PENEX off gas as high as 300 to 400 cycles and as low as 100 to 150 cycles.
scrubber with KYNAR. This plastic-like liner eliminated It is all a function of what the contaminate levels, especially
corrosion problems there as long as the liner was kept chlorides or fluorides (fluorides from HF alkylation bu-
intact. Mechanically introduced holidays or inadvertent tanes). Chlorides attack the molecular sieve. One pound
steaming reduced the effectiveness of this system. of chlorides will kill five pounds of sieve.
We have had to regenerate on somewhat reduced cycles, Regarding running with sulfur, we have run units for
mainly due to heavier feedstock introduction, but the extended periods of time with 2 ppm to 3 ppm sulfur and
main problems have been mechanical in nature. We have suffered no ill consequences. We do not have sulfur ana-
also had chronic problems with sticking remote operated lyzers on our units.
valves which we solved with automatic oilers in the pack- Finally, we use either dried makeup hydrogen or dried
ing and improved instrument air quality with nitrogen feed for regeneration. We usually hold the regeneration for
provided as a backup. We also push our cooling ability very 2 to 4 hours at 550°F to 575°F.
hard for a phosphate treatment program during the sum-
mer which runs well beyond typical temperature/cycle DAVIS:
ranges. Bypassing the hottest coolers and using extra dis- CENEX draws a normal butane rich side cut stream
persant also helped. from the main fractionator in the HF alkylation unit. The
side cut is defluorinated and then dried with molecular
KOOIMAN: sieve before charging it to the butane isomerization unit
We had some problems with the steam reboiler on the which is a UOP butamer unit.
stabilizer, but I think that was due to the steam side Typical performance is 150 cycles per load, lasting
problems. about 1500 bbl/lb. Since the stream is drawn from the
Our isomerization unit was charged with a stream that alkylation unit, CENEX is concerned about fluoride
was unhydrotreated, so we had rapid activity loss as sulfur breakthrough rather than the potential for sulfur and
accumulated. We were able to make it operable by running chlorides. Regeneration conditions are total time of one
only streams from sweet crudes. When those streams were day, 18 hours is a hot regeneration at 500°F, 6 hours of
not available, we would circulate hydrogen which re- cold cool down at 110°F. The regenerate is a slip stream
moved some sulfur and partially reactivated the catalyst. of the isom bottoms product.

Question 38. GENTRY:

When molecular sieves are used in butane isomeriza- We are familiar with a butane isomerization unit, in
tion feed dryer service, what sieve life has been operation almost 4 years, with 650 to 660 cycles having
achieved in terms of number of cycles and bbl/lb? been achieved over 46 months. This corresponds to 7.7
What has been the performance with respect to sulfur bbl of feed per pound of molecular sieves. Although the
removal? Has anyone operated with a sulfur analyzer unit was specified for a 20 ppmw feed sulfur level, in
at the dryer outlet? What chloride levels can be toler- actual operation sulfur in the feed was nil. There is no
ated in the butane feed to the dryers? What streams sulfur analyzer at the dryer outlet. The maximum chlo-
are used for regeneration, and what are the regenera- ride level allowed is 1 ppmw as hydrogen chloride. Dried
tion conditions? C4 feed is used to regenerate the molecular sieve in this
unit, although Kellogg has used an isobutane stream in
LEMMON: other units. The regeneration stream conditions in this
We have recently installed molecular sieves in butane unit are 315°C and 1000 kPa. The regeneration stream is
isomerization service. They have been operating for a composed of 1 wt% C3- and 1 wt% C5. Isobutane varies
period of 7 months and to date have shown no signs of from 20 wt% to 40 wt% and the balance is normal
deterioration. The regeneration cycle is every other day. butane.
Sulfur levels after the molecular sieves have remained
below 0.5 ppm. We have successfully operated an on-line Question 39.
sulfur analyzer for the entire time. Isomerate from the Has anyone experienced catalyst poisoning of Bu-
stabilizer bottoms is used for molecular sieve regeneration. tamer or Penex Units due to CO or CO2? What was the
I do not have any information on chlorine levels. feed source? How was the problem solved?

192 Light Oil Catalytic Processing

TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

A. Alkylation-Sulfuric Acid course, is to replace the coalescer, which we are planning
on doing next year.
1. Feed
THIERRY DARDE (Pall Corporation):
Question 1. Since it has been mentioned that there is a need for a
Does anyone have problems with ice formation in the good liquid-liquid coalescer upstream of the alkylation
contactor feed nozzle piping where the ‘‘wet’’ iso recycle plant, I would like to mention that Pall Corporation has
joins the cold effluent recycle? a viable liquid-liquid coalescer which can deliver below
15 ppm free water downstream, which in this case would
MALEK: protect the alkylation plant very efficiently.
We are aware of people who have had freeze problems
with their recycle iso-olefin feed control valve or their
refrigerant recycle control valve. The cause has usually STEVE MATHUR (Chevron):
been traced back to poor coalescing of the recycle isobu- You could actually see freezing in the lines if you are
tane and the orefin feed. You can check the effectiveness water washing the DIB a lot, because the water tends to
of your coalescer by analyzing the water content on the carry with the iso recycle when it meets the feed junction
inlet versus the outlet. I believe that STRATCO designs of the refrigerant recycle, which could be at 17°F or 20°F.
for 98% removal of free water in their coalescer designs. The situation is worsened if the coaleser is not functioning
You should also check your orefin feed, makeup isobutane, well. The actual freeze point happens right where the cold
and recycle isobutane for water content. You may be refrigerant recycle meets the isobutane recycle (including
getting large amounts of water that impact the feed the feed). This can cause the flow control valve in the
coalescer and overwhelm its ability to withdraw water iso plus feed line to fully open to compensate for the loss
away fast enough. The isobutane recycle can be a problem of flow due to partial freezing.
if you have low residence time in your overhead accumula-
tor, which can impact effectiveness of the water boot. Question 2.
What instruments have refiners used for the custody
BINFORD: transfer of spent and fresh sulfuric acid for the alkyl-
First, I would like to thank STRATCO for helping ation unit?
provide answers to several of these questions, especially
those that dealt with the design and operation issues, and CAIN:
Allen Moser of STRATCO deserves special recognition We have our acid regenerated by Dupont, which has
for his help here. a plant adjacent to our Louisiana plant. They have an in-
Ice formation in the STRATCO contactor reactor feed line magnetic type flow meter manufactured by Foxboro,
nozzle piping can occur when the temperature of the and we agree with them that it is highly accurate and
combined wet isobutane recycle and the cold effluent very reliable.
recycle is less than 32°F. Typically, this condition results
only when the propane content in the refrigeration loop
HAHN:
is about 20%. For this to occur, the refrigeration compres-
sor is usually oversized to allow higher discharge pressures Although we do not use them, we know of several
for total condensing. refiners that do use mass meters for acid transfers. More
common for us is to use tank gauges or traditional meters.
FOSTER: STRATCO does recommend mass meters for custody
I agree with what Mr. Malek and Mr. Binford said. transfers.
We are planning to replace our coalescer in the next year.
Until that time, we are working around the problem that MALEK:
we are having. We do not have a temperature indicator The bulk of the installations that we have seen use
here, but a calculation exists for the operator group so weigh scales, either truck weigh scales or rail weigh scales.
that they can approximate this temperature on the run. Usually, the refinery will have truck weigh scales. The
They use this calculation to maintain the feed nozzle at rail weigh scales are a little bit less common, but they are
around 34° by bypassing a slipstream around the feed and readily available. It is probably the most effective way to
effluent exchanger to heat the stream. A better solution, of reliably measure this material.

130 Light Oil Catalytic Processing

TOC/INDEX

FUSSELL:
At our Port Arthur refinery, we bring our sulfuric acid
into the plant, as well as out of the plant, in rail cars to
Olin. With the spent acid, we test the concentration, we
find out what the concentration is in the acid, and then
we take a strapping on the rail car. On the fresh acid, we
use the bill of lading and we strap the rail car.

GREG HALLAUER (Micro Motion, Inc.):

Micro Motion’s mass flow meters are successfully being
used for custody transfer of spent and fresh sulfuric acid.
Meter sizing and material selection are critical to ensure
the safety and longevity of the system.
Acid velocity should be maintained below 2 fps to 3
fps. Various metallurgies are available, including Tefzel-
lined meters.
Density and temperature, measured by the same meter, as Mr. Malek indicates, is pure enough, then it is conceiv-
are used to monitor concentration during transfer. ably possible to have zero gas at the reactor outlet, and
hence, zero going to the alkylation unit. But in the real
Question 3.
world, we do approach almost stoichiometric proportion
What levels of light ends (hydrogen through ethane) in
of gas. But, there is always a little bit there.
the alkylation feed have refiners been able to tolerate?
Most of the units require only a flash separator. Use
Do refiners with feed selective hydrogenation units have
of a stripper, of course, will ensure that the maximum
product strippers, or flash separators only?
amount of light ends are stripped out, and that will be
a function of your incoming hydrogen. You need to try
MALEK: to ensure that as much of the light ends as possible is
Most sulfuric acid units can handle small amounts of stripped before the feed goes to the alkylation unit.
C2 and lighter material. We have reviewed units that have
run as high as 3 liquid vol% and 4 liquid vol%. This HENKE:
material will get purged via the refrigerant compressor The amount of C2 that can be tolerated in the alkylation
receiver pressure control valve. feed depends on the design of your particular alkylation
Many selective hydrogenation units only have flash unit, and how much propane/propylene is in the feed.
separators, which is a potential source of light ends into The light ends will eventually end up in the depropanizer
the sulfuric acid units. A typical unit that is using real section. If there is not much propane/propylene in the
high quality or high purity hydrogen will have very little feed, the C2 will significantly pressure up the depropan-
inerts coming through with the treated material, because izer section.
the hydrogen consumption is very near stoichiometric in We operate two UOP HF units with selective hydro-
this unit. We would recommend sending the vent stream treaters. Both units are equipped with product strippers.
from the flash separator of these hydrogenation units to Both units operate on butane/butylene feed. The only
the FCC wet gas compressor rather than the fuel system. propane/propylene that enters the unit is through varia-
This will recover the olefins in this stream. This stream tion in upstream fractionation. The C2 in the alkylate
is very small compared to the total flow of the wet gas feed must be kept very low. We target less than 0.02 liquid
compressor. vol% in order to avoid venting from the depropanizer
You may consider sending the refrigerant compressor overhead receiver.
receiver pressure-controlled vent to the FCCU, as well. We also operate a STRATCO sulfuric acid unit. This
This stream will contain some trace amounts of SO2 unit operates with 15% propane/propylene in the feed.
and SO3. Several years ago, we ran for a short period of time with
a selective hydrotreater upstream of the sulfuric acid unit.
TASKER: The only light ends removal was through two simple
In general, dealing with the second part of the question, flashes. The operation was runnable, but the light ends
IFP and HRI design the selective hydrogenation units, did result in high pressures on the compressor discharge
which we refer to as SHU’s. We design them to avoid receivers, which restricted compressor capacity. The selec-
excess hydrogen in the reactor. With no excess, the pro- tive hydrotreater operation was discontinued for numer-
duction of light ends is minimized. If the hydrogen stream, ous reasons. If we had continued with that operation, the

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 131
TOC/INDEX

economics would have justified the additional provisions proper coalescing of the feed has a major impact on the
for light ends removal. water content in the circulating acid.

STEVE MATHUR (Chevron): PETERSON:

One of the things that we have found necessary in HF In one of our refineries, we have a dimersol unit that
alkylation units is to have a condenser in the vent line processes most of the propylene. But, at times, we can
on top of the depropanizer overhead drum, because if put propylene, or the rest of the propylene, to the alkyl-
you do not condense the HF out of there, you are going ation unit, so we have a common feed drier for the pro-
to blow a lot of acid out with the noncondensables, and pylene.
that could actually increase the acid consumption rate In another refinery, we have a fractionation dehydrator
substantially. The situation gets a lot worse if you have downstream of the MDEA and Merox treating system
a lot of noncondensables (C2 etc.), and also, if you are which dries propylene along with the butylene.
not operating the HF stripper correctly so that the ethane, When only alkylating butylene, we may not need to
instead of coming out with the propane product, goes run this dehydrator tower in front of the system. But,
overhead and pressures up the unit. because of the severely hydrotreated feed to the FCCU,
we do not have to run the Merox system. MDEA is not
2. Process perfect at getting all of the H2S, so this partially condens-
ing tower acts as an H2S back end stripper to help prevent
Question 4. sulfur consumption of the HF.
What is the recommended water content of circulating
acid in HF and H2SO4 alkylation units? What are the best BINFORD:
ways to control this? For units operating with full mixed butylene feedstocks,
STRATCO recommends typical water content of 2% to
ROMAN: 3%. Typical water content is 3% to 5% for units with
At Ashland, we operate three HF alkylation units. In an MTBE raffinate feed. Water content of the feed is
all of them, we try to maintain 0.5% to 0.8% water in controlled by monitoring the efficiency of the feed
the circulating acid. We try to maintain the olefin feed coalescer through tracking the water knockout rate. Oth-
moisture at less than 5 ppm. Higher moisture levels can erwise, water content is controlled by reducing water in
cause problems via two different mechanisms. Carbon the fresh acid or in the olefin feed.
steel in the unit is normally protected by dense, hard, Water precursors, such as methanol, need to be con-
adherent iron fluoride scale. Dilute acid will convert this trolled as well. Besides decomposing and forming water
iron to iron fluoride tetrahydrate, which has a specific in the reaction zone, methanol is a cosolvent which allows
volume of 3.5 times the original scale. This expansion more water to get past the feed coalescer.
causes the scale to fall off the underlying metal. Continued
exposure will convert the scale to iron fluoride octahy- Question 5.
drate, which is actually soluble in hydrofluoric acid. Is anyone using alkylation co-catalysts? What are the
Another problem is the formation of azeotropes in the benefits?
stripper. The settler effluent contains several hundred ppm
of dissolved, concentrated HF acid. This acid will be FUSSELL:
driven off as it flows down the stripper, until a maximum In our STRATCO alkylation unit, we have been using
boiling azeotrope forms at 36% HF in water. This azeo- the Betz Alkat product for approximately 2 years. We
trope is extremely corrosive, and if the concentration is have seen a 10% to 15% decrease in our sulfuric acid
high enough, it will enter the side reboilers and cause consumption. We knew we were making progress there
failure. We have recently begun to use seamless tubes on when Olin gave us a call. We have also seen a 2% to 5%
our side reboilers to avoid the possibility of this type of increase in the alkylate production off of the unit.
corrosion at the tube seam welds where there is always We have not seen any significant octane gain that we
some residual hardness. have been able to quantify. That is part of what is adver-
tised, but we were not able to see that.
MALEK: We have calculated about a 200% return on investment
As a general guideline for operating HF units, you with the use of the Alkat.
need to keep the water content in an HF acid below 1 We have also noticed that it has made our operation
wt%. This is done by external acid regeneration. a lot more stable than the operation was before. We receive
For a sulfuric acid unit, a good operation will be 3 our feed to our alkylation unit from Huntsman, and it
wt% to 4 wt% maximum water. As discussed previously, varies quite a bit, and we used to have a lot of operational

132 Light Oil Catalytic Processing

TOC/INDEX

problems. The Alkat seems to have stabilized the unit improvement. It is probably justified when you weigh the
for us. additional cost of more isobutane versus the catalyst cost.
We are getting ready to do a trial using the Nalco/ The numbers that I have reviewed indicate that the use
Exxon product. We are not doing this because we have of these catalysts is cost-effective.
a problem with the Betz product; we just feel like doing
some experimenting. Question 6.
What experiences, if any, have people had using capaci-
BINFORD: tance probes to determine the settler acid level and the
The numbers that Ms. Fussell quoted are pretty typical acid-to-hydrocarbon ratio in the reactor?
of what we see with our cocatalyst product. We have
evaluated it in 21 commercial alkylation units, which FOSTER:
equates to about half of the alkylation units in the United We currently have capacitance probes on four settlers
States. Typical performance improvements quantified are and DP cells on two others.
3% to 7% improvement in unit yield; 10% to 27% We have had good experience with the capacitance
reduction in unit acid consumption; and a 0.3 to 0.7 probes. They more accurately identify where the top layer
number increase in alkylate octanes. That small value of foam exists, and they show a good trend of the acid level.
octane increase can be very valuable, but as Ms. Fussell
mentioned, it may be very hard to see in the unit, and Question 7.
hard to quantify. Reduction of undesirable side reactions We have had occasional foaming in our alkaline water
in the unit, as a result of using the co-catalyst, has also wash system. Could it be caused by poor make-up water
reduced corrosion in the downstream DIB tower and quality (TDS) or aromatics in the feed or something else?
reduced alkylate end point. Units alkylating higher levels What type of makeup water do you use (BFW, conden-
of propylene in their feedstocks or mixed propylene-butyl- sate, process water), and what is the quality? If the
ene-amylene feeds, and generally those units operating at problem is aromatics, at what level does it occur?
more severe conditions, exhibit the upper end of the
improvements I quoted before. BINFORD:
We are evaluating a new product that is specifically STRATCO has noted, in their experience, that the
targeted at higher acid reduction numbers. In our pilot alkaline water wash will have a tight emulsion when a
plant, we have seen acid reduction in the 25% to 35% surfactant enters the water wash with the olefin feed. Seal
range. We do have a commercial test on-going, and oil and lube oil leaks into the process can also cause tight
another just about ready to start. emulsions. Aromatics, ethylene, or ethyl mercaptan can
all cause a detergent or soap formation in the alkaline
CAIN: water wash at levels as low as 50 ppm in the olefin
We have conducted several trials in two of our plants feed stream.
with alkylation aids. We have been unable to quantify To prevent the formation of solids in the alkaline water
the benefits. So, we are still working with them, and we wash, they recommend that the makeup water contain
will try to get those quantified. less than 20 ppm hardness as magnesium plus calcium
carbonate. Continuous water makeup is required to main-
FOSTER: tain a conductivity in the circulating water in the range
We have done several trials of BetzDearborn’s Alkat of 5000 to 8000 micro moles per centimeter.
XL. We have seen a reduction in acid consumption of
about 9%, and an increase in alkylate yield of 3%. We CAIN:
are not sure that these results are 100% statistically sign)fi- We use clarified water as makeup water, with a total
cant, however. After the final Alkat XL trial, we intend hardness of less than 80 ppm in two plants.
to conduct a trial of the Alkat AR. I think that Mr. Binford
was talking about Alkat AR with the higher reduction in FOSTER:
acid consumption, somewhere between 25% and 30%. Mr. Binford mentioned the 5000 ppm to 8000 ppm
range for TDS. Our utility water happens to run about
MALEK: 4000 ppm, so it is a very good quality and we use util-
The data that I have reviewed on the various catalyst ity water.
systems do indicate that there is some improvement. One Condensate can also be used as makeup water, but the
refiner told me that he thought that a lot of it might hotter it is, the more it tends to absorb hydrocarbon, so
have been attributable to the fact that his engineers were it causes the TOC to go up in the water purge stream.
looking at the unit a lot closer while they were evaluating We also use an antifoam by Baker Petrolite, it is IPC
the use of the catalyst. But I really do believe there is an 5120. We use it at about 2 ppm in the caustic wash

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 133
TOC/INDEX

drum in order to control the amount of foaming in the check the source of your isobutane. It is probably coming
neutralization section. from the butane isomerization unit stabilizer. This prob-
lem is usually due to an over-injection of chloride in
HENKE: the butane isomerization unit, and to improper stabilizer
We have not experienced foaming problems due to operation. Get the licenser to review your chloride injec-
makeup water quality or aromatics in the feed. We did tion rate and make sure that you are refluxing the stabilizer
have a foaming event that we attributed to high caustic adequately, and thereby reboiling and stripping this tower
concentration in the alkaline water wash vessel. This con- properly. This tower is not an area where you should be
dition was corrected by repairing the makeup water flow trying to save reflux energy.
meter and reestablishing an adequate makeup water flow
rate. We use softened well water as makeup water to the BINFORD:
alkaline water wash. The TDS of this stream is approxi- I agree with all of Mr. Malek’s comments on the chlo-
mately 1000. ride issue. While I was researching this, I did gather a
lot of information on corrosion in the overhead of the
MALEK: depropanizer and in the debutanizer.
You would have to have a fairly large amount of aromat- Corrosion in fractionators of alkylation units is becom-
ics in your feed, probably greater than 2% to 3%, to see ing a more prevalent problem for most refineries. Many
a foaming problem. If you are using reformate to start U.S. alkylation units are running above 110 % of design
up your unit, or as a seal flush, you may have a problem capacity to meet the refiner’s need for alkylate in the
during those times. However, you should also see emul- gasoline pool. The increased loading on the reaction sec-
sion problems in your acid settler emulsion level gauges. tion creates more side reaction products, acids and neutral
esters, and acid carryover. Acids and esters carrying
HERBERT W. WIZIG (Merichem Company): through from the reactor are the most common cause of
For over 10 years, Merichem has operated a Fiber corrosion in the fractionators. Poor mixing and lower
Film娃 water wash system in Aquafining SM service which residence times in the reactor effluent treating section
is downstream of a conventional caustic wash on the allow more of these components to enter the fractionation
alkylation reactor product. We have seen extremely good section. Corrosion in the DIB comes from not aufficiently
results with this service. Merichem recommends that the neutralizing the alkyl sulfates in the water/caustic wash
makeup water to the water wash be steam condensate, upstream. The alkyl sulfates may break down to form tars
boiler feed water, demineralized water or equivalent. The and SO2. The wash needs to be at least 120°F to thermally
total dissolved solids not exceed 100 ppm weight as cal- decompose the dialkyl sulfates.
cium carbonate. Corrosion in the depropanizer on the alkylation unit
is the result of SO2 and SO3 liberation in the fractionation
STEVE MATHUR (Chevron): section. SO2 and SO3 end up in the refrigeration loop
We have found that if you have high TDS in the where they can react with water to form acids. Proper
overall alkaline water wash, higher than 10,000 ppm, you operation and sizing of the depropanizer feed treaters
will tend to carry the alkaline water with the reactor can minimize corrosion. Units without depropanizer feed
effluent to the DIB. treaters are at the greatest risk for severe corrosion. One
Hardness is another matter, however. With hardness, alkylation unit does not operate its depropanizer due to
what you do is foul up your alkaline water exchanger. If this corrosion problem.
the hardness is very high, within three weeks you will Corrosion problems in these fractionators can be effec-
end up cleaning that exchanger on a continual basis. The tively controlled through the use of chemical neutralizers.
only long term solution is to use either condensate or soft Neutralizers can be applied to the feed or the reflux of
water as makeup water to the alkaline water wash. these columns. The neutralizer used in the deisobutanizer
must be nonvolatile at the overhead conditions of this
Question 8. column so that it does not enter the reaction section with
What effects are seen in the DIB operation when chlorides the isobutane recycle. Nonvolatile neutralizers will also
come in with make-up isobutane? Has anyone experi- protect the reboilers in the fractionation section. Neutral-
enced severe tray fouling in the DIB and-or pluggage in izers used in the depropanizer should partition to the
the feed/isobutane recycle coalescer as a result of it? water phase in the overhead accumulator to prevent them
from recycling back to the reaction section.
MALEK: Reboiler fouling in the alkylation unit is typically the
If chlorides are present in your isobutane, you can be result of corrosion problems in the unit. The corrosion
assured you are going to have problems in the DIB col- products can migrate down the column and concentrate
umn. In order to make sure that you do not have chlorides, in the reboilers. Corrosion products can also come in

134 Light Oil Catalytic Processing

TOC/INDEX

with the fractionator feed from upstream equipment. The BINFORD:

corrosion products become bound to the reboiler with Troubleshooting the acid wash is typically done on an
polymers and ester decomposition products. While most indirect basis. The inlet and outlet acid strengths should
refiners have a routine of water washing the reboiler on- be monitored on a weekly basis to see how much non-
line, this tends to be a short term solution (the bundles acidic material is absorbed. There is a test method for
get less clean each time). Routine water washing must be measuring free acid in the effluent from the electrostatic
continued until the reboiler can no longer perform as precipitator. The method involves centrifuging a weath-
required. Chemical antifoulants can be applied that will ered sample and volumetrically determining the amount
help prevent polymer formation and keep the corrosion of acid that separates from the hydrocarbon.
products dispersed. In most cases, we have been able to
economically extend reboiler run lengths to the desired
turnaround schedule with a properly designed antifoul- HERBERT W. WIZIG (Merichem Company):
ant program. Merichem offers EsterexSM technology, which uses fresh
sulfuric acid along with a Fiber Film娃 contactor to
remove both the acid and neutral esters which cause foul-
CAIN:
ing in the downstream DIB. Since the Fiber Film娃 con-
We have had fouling problems in our alkylation DIB tactor has no moving parts or electrodes that short out
side and bottom reboilers. with emulsions and corrode in the acid environment,
We are currently using a Nalco/Exxon dispersant with virtually 100% on stream time between turnarounds is
good results. achieved, along with minimal acid carryover.
Question 9.
What are your experiences with, and the causes of foam- Question 11.
ing in spent acid tanks or acid trucks/railcars? Discuss experience with running above 0.6 olefin space
velocity in sulfuric acid alkylation units. Specifically
CAIN: comment on reaction zone temperature limitations, cor-
At Port Arthur, we do not have any problems with rosion problems, alkylate yields and acid consumption
foaming in our spent acid tanks or trucks. Foaming usually effects.
occurs due to the presence of hydrocarbons in the
spent acid. HAHN:
Yes, we operate both of our alkylation units above 0.6
FUSSELL: space velocity. We suffer a little because of that. Our
We have had occasions where we have had foaming corrosion rates are somewhat high. Our temperatures are
problems in the spent acid railcars. We have determined a maximum of 65°F, and we routinely run at 65°F for a
that this is caused by alkylate in the spent acid, so we large part of the year.
have several steps that we go through to try to get all of Our worst corrosion occurs in our lowest strength acid
the alkylate out. system. That final contactor operates at 88-89% sulfuric
acid. We have experienced contactor tube leaks as fre-
Question 10. quently as every 4 months to 12 months. A lot of this is
How do you troubleshoot the acid wash electrostatic due to a very localized corrosion, sort of like an impinge-
precipitator to ensure that it is working efficiently? Does
ment effect. We are designing a splash plate to protect
anyone test for acid in the effluent? What procedure
that nozzle entry point.
is used?
Our octane numbers are somewhat lower than what
may be desirable. But overall, our alkylate production is
HENKE: much higher, and we accept these less than perfect condi-
The operation of our acid wash electrostatic precipita- tions in order to maximize our alkylate production.
tor has been basically trouble free. The operators monitor
performance by watching level, temperature, amps, and
volts. Acid carryover is maintained below 10 ppm. We FOSTER:
have found that, in our unit, operating at a lower level We have had some similar experiences. We have two
improves overall performance of the electrostatic precipi- of our contactors that operate at this space velocity. Cur-
tator. rently, we have not seen any corrosion problems. Our
We do not test for acid in the hydrocarbon effluent C10Ⳮ molecules are about two times what they normally
on a routine basis. We have obtained semi-quantitative would be, and we see a slight increase in the 90% point
data on the acid in the effluent by direct sampling of the and in the endpoint. Acid consumption does not seem
hydrocarbon, followed by titration in the lab. to have been affected.

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 135
TOC/INDEX

HENKE: quality is improved by removing the side reaction prod-

Operation at or above 0.6 olefin space velocity is profit- ucts.
able in our case, provided that we maintain contactor
temperatures below 65°F. We have extensive DIB feed BINFORD:
pretreating that includes a fresh acid wash system. This STRATCO notes that it is generally not cost-effective
tends to minimize the corrosive impact of the esters to switch an existing unit to bauxite treating, unless the
formed at the high space velocities. Our acid consumption existing treating system is wholly inadequate and has to
is higher and our alkylate yield is slightly lower. I also be replaced away.
refer you to page 120 of the 1993 NPRA Q&A Transcript, In bauxite treating, the bauxite is regenerated (dried)
where there is a substantial amount of information on by recirculating an isobutane stream. This typically does
this topic. not affect the isobutane consumption appreciably. How-
ever, there is some capital investment required for the
FUSSELL: equipment to condense, vaporize and pump the isobutane
I would just like to say that about 5 or 6 years ago, through the circuit. An alternative is a once through sweep
we did operate our alkylation unit at maximum charge of natural gas, if the natural gas is dry enough.
rate. That was our whole goal in life—to put as much
charge through the unit as we possibly could. We were Question 13.
running very warm on our contactor temperatures. We Discuss what corrosion has been experienced in sulfuric
were using over 1 lb of sulfuric acid per gallon of alkylate. acid or HF alkylation alkylate splitter tower overhead sys-
That was, in our opinion, the right way to run the unit. tems.
We brought KBC in to do one of their famous yield and
energy surveys. They told us that we probably did not HENKE:
want to run the unit that way. We have no significant corrosion in our alkylate split-
We did a series of test runs and a lot of economics ters or aviation gasoline towers.
and determined that yes, they were right. That was one
thing we were not doing properly. We totally changed
our philosophy on running the unit. We cut the rate PETERSON:
back, got the contactor temperatures down, and we cut We have a yes and no answer. In one of the refineries,
our acid consumption in half. We definitely got phone we have not seen any problems in the HF alkylate splitter
calls from Olin at that point. The octane on our product overhead. This system has a fired reboiler.
went up quite a bit. Our alkylate yield per barrel of feed In another system, we have had some problems with
went up. We were actually making almost as much alkylate the fluorides apparently breaking down (it is an HF unit).
at the lower charge rates as we were at the higher charge The chrome parts of valves in the overhead system have
rates. So overall, that was definitely the right move for had some problems. This tower is reboiled with an AGO
us to make. pumparound.

Question 12. FOSTER:

Is there much industry experience with using bauxite for We use a corrosion inhibitor and a neutralizing amine
effluent treating? What benefits and costs can be in the overhead system to maintain a pH of about 7 and
expected from switching to bauxite? Since our refinery in the overhead water boot, and our iron levels are down
is short on isobutane, can anything else be used for dry- below 10 ppm. That normally keeps us down to a corro-
out during bauxite regeneration? sion rate of about 2 mils. Under normal operating condi-
tions, the corrosion rate is very low. We have experienced
PETERSON: some upset conditions where we have eaten up overhead
We have a bauxite treater downstream of a reactor in condensers pretty quickly.
one of our refineries. It removes any residual acid and
neutral esters from the stream prior to the fractionation DAVID S. McCAFFREY (Exxon Research & Engineering Co.):
in the deisobutanizer. There is very little tray corrosion In sulfuric acid alkylation splitter towers, corrosion is
and very little fouling in the deisobutanizer. No chemicals due to the breakdown of neutral esters releasing SO2. The
are needed in the overhead system, since all the bad mate- SO2 in the overhead system will combine with free water
rial is taken out upstream. The two bundles and reactors and give you a corrosive environment. In a modern ER&E
have a service life of over 25 years. auto-refrigerated sulfuric acid process, neutral esters are
A further benefit is that the isobutane recycle stream not formed to any significant extent by proper setting of
is dry, which decreases the acid consumption. Alkylate operating conditions, and thus the problem is precluded.

136 Light Oil Catalytic Processing

TOC/INDEX

STEVE MATHER (Chevron): based on the loss in ignition analysis, with the remainder
In HF alkylation plants, if you have corrosion in the being inorganics comprised of calcium and magnesium
overhead, there is a likelihood of the reboiler steam leaking carbonate. Plugging in the alkaline water wash heater is
through. If the water is reaching the overhead with the usually due to hardness in the makeup water, and thus
fluorides, it will cause low pH. It could be a very small the deposits.
leak that may not be really evident, but we have seen it.
FOSTER:
Question 14. We have found that fouling in the alkaline water wash
Has anyone experienced plugging problems on the flash heater is almost always preceded by an upset in the reactor
drum side of the suction trap/flash drum? If so, what section. If the iso/olefin ratio falls below 7 or if the spent
was the cause and/or solution? Has anyone documented acid strength falls below 88%, we tend to see polymeriza-
a case of hydrate formation? tion in the reaction section, and polymer collection in
the alkaline water wash heater. Our analysis of the material
MALEK:
in the fouled exchanger shows it to be about 60% inor-
If you are getting plugging in the flash drum side of ganic (mostly iron) and 40% organic (mostly C 4 -C 6
the suction trap/flash drum, it can be caused by ice or organic acids, C3-C9 aldehydes, and C6-C8 substituted
hydrates. This happens when water gets into the bottom alkanes).
of the depropanizer. Check the depropanizer operation
and make sure that you are not getting any carryover
HERBERT W. WIZIG (Merichem Company):
from the depropanizer feed treating section, and that you
Merichem recently replaced a conventional caustic
are running the depropanizer effectively.
wash for a major European refiner with a Fiber Film娃
BINFORD:
caustic wash ahead of the sulfuric acid alkylation unit
When a small amount of water is returned to the flash depropanizer. The unit has been on stream less than 1
drum in the depropanizer bottoms, the water will freeze year, but the benefits have been significantly reduced
at the temperatures and pressures in the flash drum, result- caustic carryover resulting in improved life of the trays
ing in plugging of the suction screens for the pumps. The and the reboiler.
solution is to increase the overhead temperature of the
depropanizer—which increases water removal poten- 3. Mechanical
tial—or to increase the pressure in the flash drum such Question 16.
that freezing does not occur. What is the experience on erosion/corrosion of nozzle
tips and candles in olefin feed distributors of alkylation
CAIN:
units? What materials and/or coatings have refiners used
We have had the same problem with water leaking
to prevent this damage? Our unit is equipped with an
into the system from an exchanger.
Alloy 20 spray nozzle on each candle with a tungsten
Question 15. carbide insert. The candle is a carbon steelpipe nipple
Has anyone experienced fouling or plugging in the alka- with a ‘‘greycoat’’ appled. Our unit is fitted with ‘‘Light-
line water wash heater or depropanizer feed heater? If ning’’ turbine type mixers.
so, what was the cause and/or solution? Was any of
the material that fouled the exchanger collected and CAIN:
analyzed? If so, what was the composition? In our Port Arthur plant during our last inspection,
we noticed some slight cracks on the feed nozzles and the
FUSSELL: spider vanes. The impellers were checked, but they showed
We have recently commissioned a caustic wash system no signs of erosion or corrosion.
upstream of the depropanizer on our alkylation unit. It
is actually upstream of the feed heat exchanger prior to 4. General
the depropanizer. We have had a lot of plugging problems
in the heat exchanger as well as in the depropanizer. We Question 17.
are slowly but surely troubleshooting our way out of these What are the panelists’ thoughts on preneutralizing the
problems. We have determined that the plugging has been spray mitigation water? Has anyone used sodium silicate
caused by the caustic salts. in this service?

BINFORD: ROMAN:
Some analysis provided by STRATCO, based on their We think the best approach is to allow the runoff
experience, is that the foulant is typically 50% organics, flow to a nearby impoundment area where it can be

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 137
TOC/INDEX

subsequently neutralized with lime. If there is no local Higher molecular weight sulfur compounds can also
impoundment area adjacent to the unit, we would recom- come through if your Merox system has some problems.
mend installing a pump pit and pumping the overflow Isobutane consumption increases not only from the alkyl-
into a nearby diked area, again, for subsequent neutraliza- ation of the olefin, but also because of the significant
tion. Pre-neutralization would be our last choice due to hydrogen transfer that occurs with amylene. Again, this
its high cost and problems associated with spray nozzle is all a balance between what you really want to do with
plugging. RVP reduction and these other operating costs. The iso-
pentane and normal pentane are going to end up in the
ROBERT E. DAVIS (R.E. Davis Chemical Corporation): gasoline pool anyway. But maybe you would rather have
Years ago, I worked for a company that manufactured them go there with the FCC gasoline as opposed to the
HF. One of the products used for any spills was sodium alkylate, depending on how you are using your alkylate
silicate. It forms a sodium silica fluoride, which is totally for octane optimization.
insoluble and can be swept up and disposed of.
SAMUELS:
Question 18. I like some of the things Mr. Peterson said here. With
What is the current ‘‘best’’ practice for washing the resid- amylenes, of course, you usually have a large amount of
ual H2SO4 from an alkylation unit after draining for turn- dienes which, if not removed, will form light ASO in the
around? Is a neutralizing circulation used, or is deluging alkylating unit. Also, the C5 mercaptans are harder to
with water to the sewer a common practice? If a circulat- treat in a standard Merox unit. Selective hydrogenation,
ing basic water wash is used, what is the typical flow operated to remove sulfur, can have the added benefits
diagram for it, and how is the pH of the water monitored of saturating diolefins and converting 1 butene to 2-
and controlled? butene to improve alkylate quality. So there are some
advantages there. It should also be noted that to get the
CAIN: amylenes, a lot of saturated pentane must typically be
We inject 50 baume caustic into the wash, carefully taken. Normal pentane lowers alkylate octane while iso-
time the wash, check pH to determine whether to add pentane increases the vapor pressure. Also, for refiners
more caustic or stop, and then check to make sure that who have ether units, if you have not prefractionated,
the pH is in a safe and legal range to drain. We dump you are going to make TAME and dilute your ether with
from one settler to another to conserve the caustic. unconverted C5’s.

FUSSELL: STEVE MATHUR (Chevron):

We also circulate caustic through our three contactor A long time ago we did a test alkylating large amounts
systems. We have a 25 wt% caustic solution. We take of amylenes in an HF unit, and were astonished to find
the waste caustic to our waste treatment facility. that the hydrogen transfer of the C5 olefins could be as
high as 65%. In other words, 65% of the C5 olefins was
going to make isopentane instead of alkylate. So the point
B. Alkylation-Hydrofluoric Acid of alkylating was almost disappearing due to the high
vapor pressure of isopentane.
1. Feed
2. Process
Question 19.
What problems are caused by alkylating amylenes? What Question 20.
contaminants cause the different problems, and how can What has been your experience with HF and HC sensors
the problems be mitigated? for vapor detection? Please give your overall philosophy
of selecting different technologies and extent of success/
PETERSON: reliability with each type and make of sensor.
There are several concerns with the alkylation amy-
lenes. It depends on your balance of RVP reduction needs ROMAN:
and your operating costs associated with the amylene Ashland has had good success with our current sensors.
processing. The inclusion of heavier hydrocarbons with There are 23 phototype sensors scattered throughout the
the amylenes sometimes can occur, particularly where you HF unit in our largest plant, and there are 4 quasi-mass
are really running your FCC debutanizer very hard, as spectrometers that sample ambient air around the pumps.
we run ours. If our GC shows a component called ‘‘C6Ⳮ,’’ The phototype sensors made by Molecular Analytic
we know we have gone too far. We think it is probably bounce a light beam off a mirror. When the mirror
something like cyclopentenes and other heavier items. becomes distorted with acid, the light pattern changes

138 Light Oil Catalytic Processing

TOC/INDEX

equipment having to be modified from the original design.

In addition, the early data indicate a possible reduction
in HF consumption, and we expect to have additional
data over the coming weeks to further prove the viability
of this new process.
Question 22.
For HF alkylation systems, what instruments are refiners
using to indicate acid regenerator column bottoms lev-
els? We currently use a temperature indicator with
some success.

PETERSON:
We are still using the temperature indication method
in the Ardmore Refinery to control level in our HF regen-
erator. Our Alma Refinery, however, is a little more
advanced than we are on this. They have a radiation
and alarms. The quasi-mass spectrometer pulls ambient system aimed diagonally across the level control region
air into a cleaning and drying system, and actually analyzes of their regenerator, and they have seen some very encoura-
the air. This instrument can detect Zero ppm to 10 ppm ging results from this. They do not overdraw too low, so
of hydrofluoric acid. The philosophy in selecting the sen- they are able to save some acid. They can also stabilize the
sors was to get inexpensive, proven sensors for general operation, because they know where the level is, instead
monitoring, and get a more sophisticated instrument for of waiting for a temperature to indicate-almost like a
monitoring areas which are more likely to experience batch system.
a leak.
SAMUELS:
Question 21.
Now that some of the new vapor suppression additive
In Robinson’s HF alkylation unit, we successfully oper-
systems have started up, what has been the experience
ate a Texas Nuclear fixed source detector level indicator
with their performance?
at the bottom of the acid regenerator. Temperature indica-
tors above and below this instrument act as a backup
HAHN:
system.
In July, 1997, Texaco and UOP completed the latest HAHN:
field test on a new, more cost effective additive called At E1 Dorado’s HF alkylation unit, they also found
Alkaid, at Texaco’s El Dorado refinery. This test was very that temperatures and DP cells were not very reliable.
encouraging. The main objective was to determine the They also have installed nuclear detection devices, and
increased operating costs, if any. There were no substantial have found them to be much more reliable.
increases in operating costs as far as they could determine.
The flash chamber testing of the unit circulating acid Question 23.
sample taken during the additive operation shows substan- For HF alkylation systems, what instruments are refiners
tial reductions in aerosol formation, in comparison to the using to indicate acid settler levels other than sight
circulating acid without the additive operation. glasses and DP cells?
We are very encouraged by these test results, and intend
to conduct a more extensive field test this fall. After a SAMUELS:
successful field testing period, we intend to go commercial At Marathon’s Garyville Phillips HF unit, we use an
with this product next year. OHMART nuclear level indicator to monitor the acid
level in the settler. The instrument continuously measures
LARRY W. SHOEMAKER (Phillips Petroleum Company): the density in the settler and physically travels from the
Phillips started up their reduced volatility alkylation top to the bottom of the vessel about every 1.5 hours.
process (ReVAP) at the Woods Cross Refinery near Salt The instrument plots the density versus level in the vessel.
Lake City, Utah, in early August. Since that time, the The difference in density between the acid and the hydro-
process has worked extremely well. The alkylate product carbons is very distinct making this a very useful level
quality and yields have been observed to be equal to or indicator. To date, the instrument itself has been very
maybe slightly better than that seen with traditional HF reliable, but has been somewhat subject to mechanical
catalysts. The operability and the reliability of the unit problems, due to the physical nature of the instrument,
have been good, with only one minor piece of mechanical e.g., the detector moving up and down.

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 139
TOC/INDEX

At Robinson’s UOP HF alkylation unit, we also use SAMUELS:

an OHMART nuclear detector profiler for the mixer Fresh HF acid pump seal failure is a classic problem
settler and for the acid storage drum. On the HF acid in HF units. Even with much effort over the years, Mara-
boot levels, we have used a combination of capacitance thon has not been able to solve this problem. We are
probes, as Mr. Foster has described, and a displacer unit. currently evaluating two upgrades for this service. First,
We have actually removed the sight glasses from these sealless magnetically coupled HF acid pumps were
boots for safety reasons. installed in Garyville. At this point our success with these
The dual technology combination has proven effective. pumps has been marginal, and I probably could not rec-
The capacitance probe does not suffer from hangups asso- ommend this design based on Garyville’s experience. The
ciated with the displacer. But the capacitance probe has initial capital cost is high compared to traditional designs,
had problems reading emulsions during start-ups—prob- the failure rate has been higher than expected, and the
lems that the displacer has not had. We have also com- repair costs are high, in the $10,000 to $13,000 per
pleted a trial installation of a nuclear guage designed to event range.
be mounted in a level bridle, to supplement a displacer, An alternative, new technology has been developed and
with success. The capacitance probes are preferred, how- installed in Robinson. This technology is the application
ever, due to substantially lower initial cost. of an eductor system to educt fresh acid from the storage
drum into the mixer settler. A flash acid eductor trial was
PETERSON: successfully completed on November 15, 1997. Obvi-
We currently have nuclear level detection in our acid ously, balancing the system flows and pressures is key to
settlers. We are looking at some other technologies. We operating the eductor successfully. We plan to design and
do not know the results, yet. install a 150 8pm eductor for unit start-ups.
Just to add a little bit about the eductor design, it
ROMAN: educts HF using a slipstream of 225 8pm of cooled, high
We currently use OHMART gauges, which were pur- pressure (325 psi’’) iso recycle from upstream of the HF
chased directly from OHMART, and which use a cesium reactor as the motive fluid into the eductor, and it draws
nuclear source. This is a two-tube system. One tube is a 25 8pm from the acid storage drum, which operates at
source on a pulley, and the other contains a Geiger detec- 50 psi’’. The fresh acid and isobutane mixture (250 8pm)
tor. These systems are now considered obsolete, and we is then discharged into the mixer settler at 210 psi’’. The
are evaluating replacing them with newer back-scatter eductor is inherently sealless. There are no moving parts,
technology, which uses a single tube and a much so it is more reliable than a conventional pump. The
smaller source. installed capital cost of the eductor is about half the cost
We have also installed several Pro-Mag units, which of a sealless magnetically coupled pump.
are magnetically coupled to external systems that use a Finally, I just want to mention that in our emulsion
monel tube and an internal float. We have had problems pump, or circulating acid pump, in the UOP unit, we
with some of the monel tubes due to the low and variable do use a tandem seal, and have had very good success
Curie point of monel. It can become magnetic if the with that, and have no plans to change that design.
Curie point happens to be above the settler temperature,
rendering it useless.
K-Ray also makes a back-scatter system, but we have
C. Reforming
no experience with it.
1. Process
BRIAN JOHNSON (UOP):
UOP offers a density profile system that uses a low Question 25.
strength radioactive technology to indicate the settled What is the operating experience of reforming units being
acid, the emulsion and the settled hydrocarbon levels. controlled by advanced controls and optimizers? What
The instrument has been widely used in the industry in the operating parameters are utilized for control to optimize
current density profile system version and other density performance?
profile versions for many years.
CAIN:
3. Mechanical In our Louisiana plant, our reformer has advanced
controls, including DMC. It has been in place for almost
Question 24. a year now, with good results. The controls maintain
How successful have magnetic drive pumps been in HF inferential and analyzer variables on reformate octane,
service? How do they compare with other sealless RVP, and light end cuts by manipulating reactor tempera-
pumps? tures, the tower operating pressures and temperatures, etc.

140 Light Oil Catalytic Processing

TOC/INDEX

We also have purchased and are in the process of installing unit. We used our multivariable predictive control soft-
an Applied Automation FTNIR analyzer. We hope to ware, OPC (Optimal Predictive Control), to control reac-
have that installed by year end. It will model RON, tor weighted-average reactor temperature and inlet tem-
benzene, aromatics, and RVP. Also, in our Port Arthur perature profile by manipulating furnace outlet tempera-
plant, they have been using advanced control since 1985, tures. Furnace tube temperatures are constraint variables:
and are in the process of starting up an on-line optimizer. the controller will attempt to meet the desired temperature
profile through the reactors while respecting maximum
limits on tubeskin temperatures. For optimization, like
PETERSON: the alkylation model I discussed yesterday, we used our
One of our refineries introduced advanced controls in CRO (Closed-loop Reconciliation and Optimization)
the semiregenerative reformer about 10 years ago, and software to perform an economic optimization using a
they have worked quite well. They do inferential octane first-principles, non-linear chemical engineering model of
controls by looking at the WABT, the feed rate, and other the entire plant. The model is in simultaneous equation
variables. They minimize the hydrogen recycle ratio in form, so the same model is used for a full reconciliation
both the reformer side and at the hydrotreater side. They and parameter update, followed by an economic optimiza-
minimize pressure in the reactor section, all, of course, tion. What is unique about this reformer application is
relative to the constraints of the catalyst system. that the model starts at the crude unit atmospheric tower
In the stabilizer system, pressure is minimized to match and includes the hydrotreater and the naphtha splitter.
a predetermined allowable off-gas rate. Reflux ratios and The current crude slate is known, so the optimizer rigor-
pass balancing controls on the towers are used. The big ously calculates the naphthenic and paraffinic content of
the reformer feed and calculates the change in weighted
key to making this successful is to make sure the operators average temperature required to maintain a target octane.
understand what the systems are doing. Operators tend The crude slate changes every 3 days and the octane target
to really get concerned when they see an automatic system changes twice weekly. The optimizer also computes and
start changing the pressure on a tower. implements the optimal feed 95% point on the atmo-
spheric column, and the optimal feed 5% point on the
ROMAN: naphtha splitter. New values are calculated for up to 25
Ashland has had good experience with reformer setpoints and sent to the controllers every 45 to 60
advanced controls. General advanced control design calls minutes. The application has shown high uptimes and
had a payout in the range of 6 to 9 months
for control of the reformate octane by resetting reactor
weight average inlet temperatures against both charge and LEE E. TURPIN (Honeywell Hi-Spec Solutions):
interheater constraints. A bed profile controller maintains Advanced process controls on catalytic reformer units
an operator-entered specification and temperature profiles have been very effective in stabilizing unit operations and
across the bed. Reactor liquid hourly space velocity is pushing constraints such as feed rate, equipment limits
used as a feed forward parameter in reformate octane and carbon make. We have installed on-line closed-loop
predictive controllers. Most success has been found in optimization on several units, including multiple reformer
those systems using Ashland’s NIR on-line technology to simultaneous operations.
measure reformate octane. Some reformers rely upon a I would like to point out that the flat profile of profit
calculated octane value, which is updated with laboratory versus operating conditions may result in an optimizer
values. Optimization parameters include temperature pro- eventually downgrading to just a constraint pusher, some-
file across the beds, final reformate octane, and the amount thing that needs to be watched very closely. I would
of naphtha, which bypasses the reformer to achieve an like to note another problem: we see some optimization
problems have been put on reformers where the unit is
overall octane demand, as dictated by gasoline blending. varying octane as an independent variable, which is a local
The quality of the feed being introduced into the optimization, but it does not fit into the refinery’s general
reformers is also an optimization parameter, which is, of operating plan and results in a non-optimization of the
course, controlled by the split at the dehexanizers or naph- plant. The summer edition of Hydrocarbon Technology
tha splitters upstream of the units. Quarterly contains an article I wrote that reviews the
All of our CCR units have on-line equations which different operating parameters for optimization of reform-
are used to calculate coke on catalyst, which is based upon ers, and includes some economic data on the impact of
an oxygen balance around the regenerator. moving independent variables.
PARIMAL KANTI BISWAS (Indian Oil Corporation Limited):
H. C. KLIESCH (Treiber Controls Inc.): For a semiregenerative type of reformer, with the oil
We have implemented advanced control and on-line, plate heat exchanger, how are feed and effluent leaks
closed-loop unit optimization on a catalytic reforming detected?

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 141
TOC/INDEX

HENKE: Typical unit operation requires conversion of the

In our refineries, we commonly use radioactive tracer octane or aromatics target specified by the refiner’s Eco-
technology to detect leaks. I suggest that you can contact nomics and Scheduling Group into a weighted average
the people at Tru-Tec in LaPorte, Texas. They provide inlet temperature (WAIT). Using a nonlinear model of
innovative leak detection solutions. the reaction system, UOP as developed a model that
provides that target over a wide range of operating condi-
FUSSELL: tions. Since the platforming process unit is often critical to
If you are talking about the feed effluent exchanger, the refiner’s ability to produce gasoline and/or aromatics,
we look for isohexane in the product. If it makes it that maximum throughput can be achieved by pushing unit
far, we probably have a leak. constraints typically to maximize unit charge rate. The
constraints may include furnace tubewall temperatures,
ROMAN: feed and recycle gas reactor hydraulic limits in CCR units
We generally use sulphur hexafluoride and inject it and coke production (either to match burn capacity of
into the feed, and then look for the peaks, using gas the CCR or hit a run-length target on a semiregenera-
chromatography to find that leak. If you see just one tive unit).
peak, or if you see two peaks, you know you have a leak. Coke production is strongly influenced by the hydro-
gen to hydrocarbon ratio, but is difficult to control because
SAMIR HALAWANI (Saudi Aramco): of the time horizons involved compared to other process
I would like to forward this question to Peterson. We variables and the fact that only the coke production rate
have a semiregenerative UOP reformer, which runs on a can be controlled. Once coke is produced, it can only be
pneumatic control. My question is, what is required to removed by combustion.
have an optimization on the reactors, or on the octane, Process units typically fall into two classes regarding
as you mentioned? optimization potential. constrained optimum or global
optimum. For a standalone reformer, operation to con-
PETERSON: straints is typically the only optimization potential. Since
The refinery that has the advanced controls is on a the constraints can be identified by either unit parameters
DCS; it is a Bailey system. The Ardmore refinery has not (valve positions, temperatures, etc.) or calculated variables
done advanced controls on a Bailey system, and we have (the tubewall temperature, pinning or coking), the use of
not attempted anything on pneumatics. We are in the an MPC that can push constraints is typically sufficient
process of converting the rest of our refinery over to to realize maximum potential from the unit. However,
distributive control, and we are also in the process of when the reformer is part of a ‘‘naphtha complex,’’ which
going back in some of these units and looking at advanced might include a naphtha splitter and an isomerization
controls. But at this point, we have done nothing with unit (such as UOP’s Penex process unit), there are oppor-
pneumatics. tunities for trade-offs in feed rates and compositions that
can only be evaluated by more complex optimization
TERRY D. TUCKER (MAPCO PETROLEUM, Inc.): calculations. To make these calculations, rigorous, kinetic
Helium can be used to detect tube leaks in feed/effluent models of the units and detailed simulation of separations
exchangers while they are in service. Helium is injected and other unit operations are necessary. This form of
on the tube (feed) side of the exchanger. A gas chromato- optimization is significantly more complex in develop-
graph is connected to a bleeder on the shell side, outlet, ment and execution, but offers sizable benefits to the
of the effluent exchanger. After the helium has been refiner.
injected into the feed side of the exchanger, the gas chro- I agree with Mr. Roman with regard to the use of NIR
matograph will indicate multiple spikes of helium if any technology, except that we would use UOP Guided Wave
of the exchanger’s tubes are leaking. analyzers for this aspect of the information feedback sys-
tem. NIR is still considered a ‘‘secondary method,’’ as
ANGELO FURFARO (UOP): opposed to octane number determination via an engine
The summation of the panel’s responses is a good place which is a ‘‘primary method.’’ That being said, NIR is
to start. UOP has been involved with many APC projects being used for routine on-line control and analysis func-
in recent years. The implementation of APC in reforming tions today in refineries around the world as it can rapidly
projects incorporates the use of UOP’s proprietary reform- and accurately mimic (‘‘statistically equivalent’’) many of
ing models to provide setpoints and constraints for the the analyses now required and performed by primary
advanced control layer, typically a multivariable predictive methods on gasoline blends.
controller (MPC). More than 30 projects for both fixed- Two ASTM practices/methods have been developed
bed and CCR platforming process units have been com- covering the use of NIR analyzers and their calibration.
pleted. The first is ASTM E1655 which covers the chemometric

142 Light Oil Catalytic Processing

TOC/INDEX

calibration of an NIR analyzer. The second is ASTM D6 Question 27.

122 which has just passed ballot and will be published What process variables can improve the hydrogen purity
soon. This standard covers the validation of an NIR ‘‘sys- from a CCR reformer contactor? What process add-ons
tem’’, in other words ‘‘assuring that the analyzer system’s have been used to further increase purity for high purity
results are statistically equivalent to the primary method.’’ uses, such as hydrocrackers and resid hydrotreaters?

Question 26. PETERSON:

What has been the industry experience using density We use an absorption of the net offgas from the
grading to recover CCR catalyst that was contaminated reformer with refrigerated low pressure separator liquid.
with ‘‘heel’’ catalyst? Has anyone attempted to ex situ The primary purpose of installing the system was to
regenerate ‘‘heel’’ catalyst for reuse? recover the LPGs that used to go to fuel gas or to the
hydrotreaters. Because the purposes is LPG recovery, we
take the debutanizer or stabilizer overhead gas and recycle
FUSSELL: it back in front of this absorber to recover what would
In 1995, we used this process with CRI here in Louisi- have been lost in that stream. If you get too cold or too
ana. I would refer you to the 1995 NPRA Q&A Transcript high in pressure, you can set up an ethane recycle, where
where Mr. Dale Emmanuel provided a very comprehen- you absorb it in the front end and reject it out the back
sive answer. end. It just comes around and around and the debutanizer,
in our case, would have been undersized to handle this.
Another caution: if this hydrogen is going to a unit
HENKE:
downstream that has a centrifugal recycle compressor, it
We have no direct experience with density grading. may need a few heavy ends just to build enough pressure
However, we have been told that this process is well from the density of the gas. With gas too pure, we have
proven and has been used successfully by several rehmers. had some problems toward end of run in a hydrotreater
We have completed an ex-situ regeneration of our heel with high reactor delta Ps holding back recycle ratio.
catalyst. The carbon content on the heel catalyst was The hydrogen purity upgrade is on the order of 3%
between 40 wt% and 50 wt%. After inspection of the to 6%.
catalyst, we determined that it was not suitable for reuse.
We question the viability of regenerating and reusing heel MALEK:
catalyst with this much carbon. The CCR unit makes a higher yield of hydrogen com-
pared to a semiregeneration unit. The hydrogen tends to
CAIN: be of a higher purity, because there are a lot less cracked
When dumping the reactor, we separate out the last gases associated with the hydrogen. These gases can be
50 drums or so. We send them to CRI for density grading, further upgraded downstream, to whatever purity level is
and the heel catalyst will be sent out for metal recovery, required in your hydrocrackers, using PSA systems.
while the good catalyst is returned to the unit and BARLOEWEN:
reloaded. My answer is with regard to which process variable
can improve H2 purity from a CCR reformer. The only
PAUL FISHER (TRICAT, Inc.): significant handle you have is pressure. Reducing pressure
TRICAT has performed three off-site regenerations of improves purity, and that has been the whole thrust of
CCR catalyst containing ‘‘heel’’ catalyst that was returned CCR technology. However, in a modern CCR, you do
to the CCR unit for reuse. In each case, carbon levels not have very much room to move down in pressure.
were reduced to below 0.4 wt% with excellent surface That answer may apply more to an old, fixed-bed, semire-
area retention. We do not believe that density grading is generative reforming unit.
necessary. Each catalyst lot contained ‘‘heel’’ catalyst with Question 28.
over 40 wt% carbon that would have been separated What are typical targets for moisture and chloride in the
through density separation. An ex situ regeneration of all recycle gas for semiregenerative, cyclic, and CCR units?
of the catalyst from the reactor saved the expense of density Are these targets dependent on catalyst type or age?
grading, and all the catalyst was returned to service in Now are chloride/water levels controlled, and what is
the CCR. The regeneration in TRICAT’s unique ebullat- the typical variability?
ing bed reactors returned a greater amount of CCR catalyst
from the reactor section, including heel catalyst, than TASKER:
would have been available after density grading, and saved We generally recommend the same water and chloride
the cost of purchasing fresh catalyst for makeup. on the catalyst for both the semiregenerative and CCR

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 143
TOC/INDEX

processes. We try to maintain the water level to about making water and deactivating the isomerization catalyst.
15 ppmw to 25 ppmw, and the HCL levels between 1 This is not easy for us to find.
ppmw to 2 ppmw. The ratio of water to chlorine is You can try to Drager tube for CO, but hydrogen in
therefore usually in the range of 10:1 to 25:1. The real the system interferes in the test. So we have actually had
target is to maintain a chloride level on the catalyst to send samples off to places that can find the CO in the
between 0.9 wt% and 1.1 wt% of chlorine. recycle gas.
On the question of type and age, we do not generally
differentiate between catalysts. But there is a difference ANGELO FURFARO (UOP):
on age, since the chlorine retention ability reduces with For semiregenerative operation, UOP generally recom-
age. Therefore, we should maintain a higher chloride level mends effective feed stripping (to less than 1 ppmw H2O)
with a lower ratio of about 20 to 25 for fresh catalyst and injection of 4 ppmw H2O into the feed. This practice
and a 10 to 15 ratio on the other aged catalysts. The normally results in 15 mol ppm to 20 mol ppm H2O in the
targets in the recycle gas stream are dependent upon the recycle gas. Some moisture is recommended to facilitate
type and the age of a catalyst being used. Chloride reten- chloride distribution, but more than 30 mol ppm in the
tion is determined by two main factors: the type of impuri- recycle gas has adverse selectivity and stability effects. The
ties that are in the alumina used to make the catalyst, chloride (HCl) content of the recycle gas depends on
and, of course, the surface area of the catalyst. catalyst type and age, but generally varies from a trace to
For control, it is important to be able to sample cor- 1 mol ppm to 2 mol ppm. The chloride injection rate is
rectly. This calls for a sampling device, or a sample thief, varied according to catalyst type, reactor temperatures,
which can be used to extract samples. Then, the chloride performance indications, experience, and analytical results
levels on the catalyst can be measured, and the operations if made possible by catalyst sampling devices.
adjusted. If there is no sampling device, then the control Cyclic units tend to have high and variable moisture
becomes more of a black art. as a result of freshly regenerated catalyst being returned
As an alternative to sampling the catalyst, normally the to service at intervals. There is rarely if ever a need to
water and chloride levels are controlled by measuring the inject H2O into a cyclic unit. The variability of recycle
water and the HCL in the recycle gas and adding the gas moisture content and varying amounts of chloride
appropriate amount of alcohol and/or chloriding agent. being released from freshly regenerated reactors makes
To achieve a constant water to chloride balance, it is chloride control very difficult and the recycle gas HCl
important to keep a constant moisture level in the feed. concentration highly variable. The catalyst age and type
If the moisture level in the feed varies, then you will also affect the recycle gas HCl concentration. UOP does
always be chasing that water to chloride balance. not design cyclic units.
There continues to be difficulty in measuring or in For CCR platforming units, there is no need to inject
getting instrumentation for proper analysis of the water H2O or chloride into the reactor section, except for periods
content. However, we have to be very careful with the of extended regenerator downtime. The regenerated cata-
instruments requiring re-calibration. With the CCR pro- lyst chloride content is controlled by injection into the
cess, it is not required that a sample has to be taken from regeneration tower to a target value which depends on
the reactor, since a sample can be taken at the regenerator. catalyst type. The spent catalyst chloride content is slightly
This sample device can be mounted on the 4 in. dump lower due to loss of chloride in the reactors. The amount
line. of chloride lost is a function of catalyst residence time in
the reactors, the recycle gas moisture content, and the
PETERSON: reactor temperatures, but is normally relatively small. Nor-
I would agree with the comments and the general mally, there is sufficient moisture from the regeneration
ranges of control and the means of control. I would stress so that no water injection is necessary. Recycle gas HCl
that taking a sample of the catalyst be used to calibrate your concentration depends on catalyst type, catalyst age and
licenser’s recommendations as to correlating moisture and recycle gas moisture content, and can vary from less than
chloride content in recycle gas. It is far better to know 1 mol ppm to as high as 8 mol ppm or 9 mol ppm if
what you really have than to be thinking that you are the catalyst surface area is very low and/or the recycle gas
doing okay. moisture is abnormally high. UOP’s surface areas stable
We had also another case where we had poor stripping catalysts help to limit chloride loss.
of the carbon oxides from the CCR catalyst coming back
into the system. The carbon oxides were apparently under- Y.K. KUCHHAL (Indian Institute of Petroleum):
going the shift reaction and making some water in the The water content of recycle gas for S-R bimetallic
reformer, which was making moisture control more diffi- catalyst is targeted in the range of 15 ppmv to 20 ppmv.
cult there. This carbon monoxide was also getting over For this value of moisture, chloride dosing may be done
to our isomerization unit where it was shifting there and intermittently to maintain chlorine content in the catalyst

144 Light Oil Catalytic Processing

TOC/INDEX

at around 1.0 wt%. For recycle gas moisture of 20 ppmv exchangers. I am unaware of a safer or better method at
to 30 ppmv the chloride in the feed is kept at 1 ppmw. this time.
It is further increased to 2 ppmw for recycle gas moisture
of 30 ppmv to 45 ppmv. For higher moisture content in PETERSON:
recycle gas, the reactor inlet temperatures are reduced as We normally wait for a breakthrough in our sulfur
per the advice of the catalyst supplier. When the moisture guard bed. We detect this either with laboratory analysis
content increases, it becomes necessary to look into strip- of the splitter bottoms, the heavy naphtha, which is not
per operation. always real precise at these low sulphur levels that you
In order to meet a minimum target of 15 ppmv mois- see out of a sulphur guard bed. A more reliable test, but
ture in recycle gas, it may be desirable to inject water with a little time lag, is to look for the H2S in our debutan-
with feedstock (3 ppmw to 5 ppmw). Water injection is izer offgas. This gets magnified in the Ardmore Refinery,
done after thoroughly confirming the moisture content because we recycle this gas back to the reactor effluent
of the recycle gas. system for LPG recovery.
As far as catalyst sampling down the bed in the sulfur
SAMIR HALAWANI (Saudi Aramco): guard bed goes, we had installed catalyst samplers designed
What is the latest technology in moisture analyzers on after the method used for the vapor phase reformer catalyst
a recycle gas for a semiregenerative reformer? samplers. This did not work very well on the liquid system,
so we have abandoned use of those.
CAIN: As for chloride traps, we have one on the LPG from
Dupont 560 the reformer going down to the saturated gas fraction-
Sensitivity: 0.5 ppm by vol. ation. That was put in to help control corrosion in the
Accuracy: Ⳳ10% of reading or 2 ppm, whichever is line to the saturated gas and in the saturated gas unit itself.
greater, on all ranges. We probably have grossly oversized this vessel, because it
Repeatability: Ⳳ3% of reading or 0.5 ppm, whichever has run about 3 years, and it was designed for 9 months.
is greater. But the reason for that is probably the difficulty of actually
Calibration: Has a built in moisture generator for cali- pinning down how much HCl is in an LPG stream. It
bration. is not real easy to do that with Drager tubes.
Note: Very accurate, and dependable.
Range: Normal running zero ppm to 50 ppm moisture. TASKER:
Used for normal running on the CRU because of the Well, as you heard, and based on our experience, most
low concentration of moisture in the gas. operators do wait for breakthrough to replace their sulfur
Ranarex Moisture Analyzer Model XMA-72 1 B and chloride guard beds. There are other techniques which
No specifications in the manual on accuracy or repeat- are involved, including monitoring the pickup and so
ability. on, but those methods tend to be somewhat inaccurate.
Calibration: using a separate moisture generator source. Acreon Catalysts has helped to resolve this problem by
Range: 0 ppm to 1000 ppm moisture. allowing companies to have a spare charge of material
The larger range allows us to utilize this meter during on-site at no charge until it is placed into service.
startup. Also, the reading on this meter is usually 5 ppm
to 6 ppm lower than the Dupont meter when the concen-
BARLOEWEN:
tration falls down to normal operations. Plan to replace
this one with a Meeko moisture analyzer. I agree with the comment that most people, in fact all
people that I know, go to breakthrough. I would echo
Question 29. Mr. Peterson’s comment that the debutanizer offgas is a
Do operators wait until breakthrough on their sulfur and real sensitive indicator when you are starting to see sulfur
chloride guard beds to decide when to change out the breakthrough. It is magnified, and you will see a detectable
beds? Is there a better method? change there faster than anywhere else.

HENKE: CAIN:
Our strategy with chloride guard beds is to wait for In Port Arthur, we have two guard reactors in series.
breakthrough. We recently began to monitor both organic Once we spot a breakthrough in the lead reactor, we
and inorganic chlorides using Sensidyne tubes (these are switch to the lag reactor, then change the catalyst out in
model #14L for hydrogen chloride, 0.2 ppm to 40 ppm). the lead. We monitor the Delta T across the second guard
We found that Draeger tubes only detect the inorganic bed as the first line of defense against breakthrough. If
chlorides. This is an important distinction since the the Delta T becomes too large, it is time to change out
organic chlorides also contribute to fouling in hydrotreater the lead guard bed. Furthermore, if the nitrogen in the

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 145
TOC/INDEX

second guard bed effluent becomes high, we know we

need to do an emergency changeout of both beds.
FLETCHER:
One of our customers has a strategy which is slightly
different than what has been described so far. With respect
to the sulphur guard, the cost of guard catalyst is minimal
compared to its benefit. An estimate of bed loading is
therefore made based upon the inlet sulfur loading and
feed rates. Change out is done prior to the calculated end
of run. With respect to the chloride guard, it is less critical,
which allows for testing of the breakthrough before chang-
ing the catalyst.
FUSSELL:
In Port Arthur, we do not have a sulfur guard bed.
We do have chloride beds on our gas and our liquid ZIYAD AZZOUZ (Saudi Aramco):
streams on our reformer, and we do run them until we Mr. Cain, regarding nitrogen in the feed, because this
see breakthrough. is a difficult product, how do you know if your test
is effective?
Question 30.
In the reformer recycle hydrogen gas there is a fairly CAIN:
high quantity of heavy hydrocarbons (we have measured The N2 test is ASTM D4629. We test the Naptha
over 2%). How does this affect the reformer catalyst in hydrotreater product before it is charged to the platformer.
terms of coking and conversion? Our specification is greater than 0.5 ppm. We then adjust
the reactor temperature accordingly to remove N2 and
ROMAN: sulfur.
It is important to know the composition of the heavy
hydrocarbons. If the material is aromatic, there should THIERRY DARDE (Pall Corporation):
be very little effect. The hydrogen to hydrocarbon ratio Very recently, one of our customers in Europe asked
will decrease slightly, and this will tend to lower conver- us to do a survey because he was experiencing recycle
sion to aromatics and also increase coke laydown. How- gas compressor breakdown due to liquid carryover. We
ever, if the heavy material contains appreciable amounts installed a test system made up of a high efficiency liquid
of orefins, as can occur in low pressure units, they can gas coalescer. We installed this downstream of the knock-
undergo Freidel-Crafts alkylation with the aromatics and out pot and upstream of the compressor. We measured
form a heavy green oil. This normally occurs when the indeed between 1% and, on peak condition, up to 6%
aluminum bed used to remove the chlorides ages and of liquid hydrocarbon carryover with the recycle gas. The
becomes saturated with aluminum chloride. It then acts good news was that our liquid gas coalescer was able to
as a Freidel-Crafts catalyst, and the result is severe valve remove all that liquid carryover down to a level of below
fouling in downstream compressors. 0.1 ppm of liquid in a gas.
PETERSON: R. RAJAMANI (Mangalore Refinery and Petrochemicals Ltd.):
We have an old reformer, which we have been con- We have a slightly different problem, that is the con-
stantly trying to push more and more through. We have densation of reformate downstream of the chloride treater
a fairly high velocity and a fairly small low pressure separa- in the hydrogen at inlet to the PSA. So, we would like to
tor. We do see a little bit of full range material coming back know how, in cold countries, you avoid the condensation
around. This concerns us, because we have reciprocating there which ultimately plugs the catalyst and reduces effi-
recycle compressors, and they do not like to pump very ciency?
much liquid. We have had to look at the design of the
piping system into these compressors, and put a little trap HIGGINS:
pot on the one that has the worst piping to help catch No reponse.
the liquids to keep us from having compressor problems.
HEROS DERGREGORIAN (Valero Refining & Marketing
CAIN: Company):
We operate in a 0.7 to 1.7 range without any bad My question concerns the problem of an accurate ana-
effects. lyzer to detect the moisture level in the recycle gas in

146 Light Oil Catalytic Processing

TOC/INDEX

order to control the chloride and water balance. It is quite SAMUELS:

common, and I want to ask the panel and the audience In regard to the gentleman who asked the question
what other method is available to check the approximate about low nitrogen level detection in the reformer feed,
water level in the recycle gas in order to adjust either the we recently got into this topic because of a Penex unit
chloride or water level to reach the balance? As you know, where any kind of nitrogen level is undesirable. You defi-
you might see a high chloride level in the recycle gas. nitely want to target below a 0.5 ppm range, because of
Does this mean that your chloride level has increased or the destruction of the catalyst. In a reformer, of course,
is it because a high water level has leeched out chloride you want to stay below 0.5 ppm, because it is the limiting
from catalyst? Unless you know the approximate quantity reactant in salt formation.
of water in the recycle gas, the action taken could be We found that detection below about 2 ppm was not
either adding additional chloride or reducing water. accurate. As a matter of fact, we involved our research
center. We continue to do the nitrogen test, but it is more
HAHN: for looking for blatant nitrogen problems than actually
On water detection in the recycle gas, we have tried controlling nitrogen in a 0.5 ppm or below range. We just
moisture on-line analyzers but they have not been very do not think it is accurate in that range. After checking, we
effective for us. So, we have resorted back to an old use the ASTM D4629 test method but have not been able
technique, called the dew point analyzer, which is just a to achieve the results described by Mr. Johnson of UOP.
stainless steel cup. You put dry ice in it, and then you
pass the gas over it and watch for the dewpoint change FOSTER:
with a thermometer. That works fairly well, once you We found a GC specifically for very low levels of
become proficient in the technique. We use that to moni- nitrogen. It is manufactured by J.Y. (Jobin Yvon Emission)
tor our moisture level. Corporation, and it is a spectrophotometer method utiliz-
ing ICP (Inductive Coupled Plasma) with a nitrogen
channel.
TASKER:
We hope to be able to measure it down to the 100
Yes, I would support that wholeheartedly. The old ppb range and lower with that.
technique often works best, and it is actually quite accu-
rate. You should poll licensees to find out what success BRIAN JOHNSON (UOP):
they have had with the moisture analyzers. If you do have We have typically used a chemluminescence method
a problem with them, go back to the old technique. It for determination of total nitrogen in light naphthas for
is a really good one. isomerization units. The method is ASTM D4629. In
that procedure, the limit of detectability is listed as 0.3
JOSECK: ppmw. Our lab has been able to push that level down to
We do have moisture analyzers. We use them more 0.1 ppm by adjustment of sample size. I suggest that those
for detection of spikes in the feed or upset conditions. of you who have not used this method will find it to be
We usually add alcohol as water and control it based on a fairly easy method to analyze for nitrogen in the feed.
feed, and find that much more reliable than the ana-
lyzer itself. Question 31.
Where in the reforming process is the precipitation of
ammonium chloride a problem? What effect does the
HEROS DERGREGORIAN (Valero Refining & Marketing reactor section operating pressure have on the chloride
Company): deposition problem?
Is anybody using analysis of recycle gas, methane to
heavier ratio, to estimate the water level? FOSTER:
We have had a lot of problems with ammonium chlo-
PETERSON: ride deposition since going to 100% Venezuelan crude
We have used that as a trending tool. We have looked operations. Ammonium chloride precipitation tempera-
at the methane to methane plus ethane and propane ratio, ture is directly proportional to the product of the partial
and if we see that ratio going up, then we are probably pressures of hydrogen chloride and ammonia. Depending
drier than we want to be. If we see that ratio going down, on your reformer feed nitrogen content, these salts can
we could be wetter than we want to be. We generally use plate out anywhere between the feed/effluent exchangers
this in areas where we do not do a lot of moisture injection and the recycle compressor, and includes your reform
and control, like some semiregeneration units where we stabilizer, as well.
do not have a lot of upsets and do not change octanes This can occur at 200°F for low concentrations of
all that much. But it has been helpful in the past to find ammonia and HCl to as high as 260°F for higher concen-
some problems. trations. In order to determine where in your system the

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 147
TOC/INDEX

deposition can occur, you need the sublimation curve for of chlorides being carried out of the reduction zone into
ammonium chloride. This is based on the product, of the reactor and on through the system. I believe UOP
course, as I mentioned a minute ago, of the ammonia did modify their design to correct that problem.
and HCl partial pressures. UOP has a good graph of this.
These partial pressures can be reasonably estimated from BARLOEWEN:
knowing the nitrogen content in the reformer feed and I just wanted to put in a pitch to keep the chloride
the HCl concentration in the hydrogen recycle. You check in the reformer via traps, and do not export it to the
where this sublimation condition coincides with the tem- hydrotreater where it is going to cause plugging and corro-
perature conditions in the reformer, then you can deter- sion, as well.
mine where the precipitation will take place. If you are
increasing the nitrogen content of your feed, or if you CAIN:
suffer a nitrogen upset or a chloride upset, this precipita- At our Louisiana plant, our problems show up in our
tion will tend to occur in the cooler regions of your absorber de-ethanizer tower, so we take an opportunity
reformer. to wash our tower whenever we can. At Port Arthur,
As far as pressure is concerned, lower pressure can move the depropanizer is occasionally affected by deposition of
the deposition to the cooler part of the system, and higher ammonium chloride, and we have a caustic wash tower
pressure can move the deposition to a hotter part of the to remove the chlorides in this propane stream.
system. It all depends on where the least painful place to
deposit your salts is. Most of us do not have that much JOSECK:
room to move on pressure, and yield considerations usu- We have noticed problems downstream of the semire-
ally dictate where we run our pressures. generative reformer with precipitation of ammonium
chloride in the stabilizer shortly after start-up and mid-
FUSSELL: cycle. The salts were attributed to high nitrogen in the
We have experienced some ammonium chloride salt feed. The ammonium chloride was removed by injecting
plugging in our refinery on the reformer. The first place water into the stabilizer feed—a very simple procedure.
we have seen this is in our depentanizer on our reformer. We have ammonium chlorides in the recycle during regen-
We periodically have to do an on-line water wash to clear eration which are washed out after the regeneration pro-
this up. It is pretty obvious when we need to do this; the cess. We have not encountered ammonium chloride
tower starts acting up. So we wash it and everything lines deposits in the reformer during the high nitrogen in the
out fairly well. feed associated with salts in the stabilizer. We have ammo-
We have also had problems in our downstream hydro- nium chloride in one of two hydrogen booster compres-
processing units. Specifically, we have a toluene dispropor- sors. The ammonium chlorides are predominantly in the
tionation unit, and the compressor on that unit has a booster compressor operating at a lower cylinder jacket
suction screen. The compressor quit operating—basically temperature. Ammonium chloride problems in the
quit working. We pulled the screen and discovered a good reformer are a nitrogen problem. There are usually enough
layer of salt on that screen. We went back to our reformer chlorides available in the system to react with all the
and discovered that we had some blocked valves around nitrogen in the feed. The temperature at which ammo-
our chloride bed that were partially open, and that we nium chloride sublimes is a function of the concentration
were actually bypassing material around it. Once we got of HCl and ammonia. It is also a function of the water
those blocked in, we no longer had that problem. We in the vapor phase. If the salt is present on the pipe
also periodically feed reformer hydrogen as fuel gas to upstream of the compressor, and if the mole fraction of
our gas turbine. We have also seen some salt deposits on water increases in the recycle, the salt can and does move.
the filter on the fuel gas upstream of that gas turbine. It can vaporize and then form again downstream. As
I believe that the cause for the salt is, as Mr. Foster operating pressure increases, the temperature at which
mentioned, in large part due to the nitrogen in the feed sublimation occurs also increases, thus making the salt
to the reformer. We do not analyze for this nitrogen, so problem worse. In our case, the sublimation is occurring
I would be very interested in hearing about how his in a booster compressor operating at a low temperature.
analyzer works out. I know that we run more nitrogen
in our feed than UOP would recommend, and that is PETERSON:
something that I am going to continue to work on, as Our biggest problem is in the debutanizer stabilizer,
far as getting that nitrogen down. and we occasionally have to do a water wash. On the
As far as the chloride side of it, I believe that the causes water wash, we reduce the temperature so that the brine
of the chloride carrying through on a CCR are more can leave the bottom of the tower. Otherwise, we just
related to the design of the unit. Our particular unit was may rearrange where the salt is if we do not let the salts
designed in the mid-to late 70s, and it does have a problem go out the bottom. When you are done with this, you

148 Light Oil Catalytic Processing

TOC/INDEX

want to make sure that you get all of the water out of C5Ⳮ yield benefits have been on average 2 vol% to 4
your overhead receiver, so that you do not have new vol% points over the incumbent catalyst. The hydrogen
HCl coming up and forming a very low pH and highly yield benefits have, on average, been about 150 scf/bbl
corrosive situation in the receiver. to 300 scf/bbl above the performance of the incumbent
We would agree that good denitrification is important. catalyst.
We do not have a real good naphtha hydrotreater. We For lower pressure operation with this catalyst, the
have a 300 psi separator, so we have to be quite careful more the difference is enhanced. In other words, there is
in what crudes we buy and how much nitrogen is in the a greater yield benefit. However, this could be masked
naphtha portion. by more catalyst deactivation by coke which occurs more
rapidly, of course, at the lower pressure. The least that
ROMAN: could be said about the unit operating pressure is that it
We occasionally get problems on our recycle compres- has not affected the RG-582 catalyst yield benefits.
sors due to deposition of ammonium chloride salts. We
will inject a small amount of naphtha on the run to try FLETCHER:
to wash them out or redistribute them. On occasion, we One customer reports recently loading UOP R-72 and
have even used a couple of gallons of water for a wash. having seen a 1.5% increase in liquid volume yield. They
Obviously, that has to be done through a quil so that it feel the catalyst has lived up to its promise.
is properly dispersed.
ANGELO FURFARO (UOP):
HERBERT W. WIZIG (Merichem Company): UOP offers high yield catalyst systems for both CCR
SM
Merichem offers Chlorex technology with a Fiber platforming and semiregenerative platforming units. In
Film娃 contactor to remove hydrochloric acid and ammo- CCR platforming, one unit is operating with UOP’s R-
nium chloride upstream of the stabilizer by means of an 170 series high yield catalyst. This catalyst met perfor-
alkaline water wash. We have four units operating on mance guarantees and matches predicted performance.
CCRs in Japan. The results have been a significant There are now 27 platforming units operating with UOP’s
decreased fouling of the stabilizer trays and the reboiler. R-72 staged loading catalyst system which provides a yield
advantage without sacrificing stability. The expected yield
THIERRY DARDE (Pall Corporation): advantage has been observed commercially. For either
We have installed several of our very high efficiency catalyst system, there is no significant effect of pressure
liquid gas coalescers on the recycle gas line on reformers, on the yield advantage within typical operating ranges.
both in the U.S. and in Canada. Installing these high
efficiency liquid gas coalescers results in dramatically MICHEL ROY (Petro Canada Products):
improved reliability of the compressor by removing the We have been using RG582 for 2 years now. It has
ammonium chloride. lived up to expectations. We have seen the higher hydro-
gen yield. Also, I would like to comment on the robustness
Y.K. KUCHHAL (Indian Institute of Petroleum): of the catalyst. We operated it severely and it still survived.
Ammonium chloride in the recycle gas can deposit It has gone through major sulfur and humidity upsets
in coolers, separators, stabilizers cold trays and recycle and its activity recovered quickly.
compressor creating fouling, under deposit corrosion or Catalyst loading was done using a proprietary catalyst
mechanical problem. It is worth remembering that 0.5 dispenser supplied by the manufacturer. Yield predictions
ppmw of organic nitrogen in the feed leads to approxi- were conditional to this type of loading.
mately 2 tons per year of NH4Cl for a million ton plan. We are presently in our second cycle with RG582.
The regeneration procedure for this catalyst has been
Question 32. simplified compared to our past experiences. The catalyst
Have the high yield catalysts lived up to their expecta- was regenerated successfully and is again performing
tions? What yield improvements have been realized? within the yield projections. These results were obtained
Does the yield improvement decrease at lower pres- despite periods of higher than normal N2 in our feed and
sures? a 15% increase in unit charge. Compared to selectivity
projections, we have seen an increase for certain aromatics
TASKER:
which has benefitted our operation.
There is a publication on the RG-582 promoted bime-
tallic catalyst. This catalyst has been very successful. The Question 33.
publication is by ProCatalyse and Acreon. What are the problems associated with reducing the
This particular publication summarizes the overall pressure in either a CCR or cyclic reformer, and do the
yields obtained. The high yield 582 catalyst has met per- yield benefits justify the lower pressure operation for a
formance expectations in all of the units to date. The gasoline or an aromatics unit?

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 149
TOC/INDEX

MALEK: percentage of aromatics in the feed to the sulfolane unit

This question deals with the problems in reducing when we would prefer to feed the higher aromatic material
pressures in catalytic reformers, specifically a CCR or from the 50 psig CCR.
cyclic reformer. The overwhelming success of UOP’s CCR
process dramatically illustrates the benefit of operating at TASKER:
low pressures in reforming. Economic and other reasons I reflect some of Mr. Malek’s comments, except I am
have resulted in UOP settling in, it appears, at a separator promoting the IFP technology. With the lower CCR
pressure in the range of 45 psi to 60 psi on the latest pressure, there is no real operating problem, but the rate
generation of CCRs. If you are trying to operate at lower of coke make, of course, will rise. As long as the regenerator
pressures than that, you are going to probably encounter is large enough, the increased coke make is no problem.
some problems that are associated with dehydrogenation There is an increase in hydrogen production, and of
reactions. You will see olefins showing up in your products course, more reformate, typically about a 1% gain in
while operating at very low pressures. As you lower the production of the reformate. However, from an economic
pressure, if you have the ability to do that with your standpoint, lower pressure will result in an investment
compressor, you are going to lower hydrogen to oil recycle cost increase, due to the size of the recycle and net gas
ratio. This will accelerate coking and dehydrogenation compressors, the size of the lines, size of the furnaces, etc.
reactions. From operating cost viewpoint, there will be an increase
One refiner reported to me that he did a little experi- in power consumption of the recycle and net gas compres-
mentation on his own. He had excess capacity in his sors. There will be limitations, as have been pointed out,
reformer CCR recycle compressor, and he began lowering for ultra-low pressure operation and, typically, that is the
the pressure. He saw a continued increase in liquid yields recycle gas compressor. At ultra low pressure units, that
and hydrogen production. When he got down to about are presently designed for catalyst turnovers in 3 to 4
30 psi pressure, however, he got into a situation where days, (100 regenerations per year), reducing the pressure
olefin and diolefin formation was so rapid that he was will require a high regeneration frequency with an eventual
fouling the compressor and experiencing vibration. He detrimental effect on the catalyst’s ultimate life.
brought the pressure back up. At low operating pressures and high severity, when you
are running about 100 RON, the olefin and diolefin
PETERSON: contents in the reformate do significantly increase. Reduc-
One barrier we have reached trying to lower pressure ing the average reactor pressure from 10 bars to 3.5 bars,
in our unit, which we have generally been trying to do, or let us say, 150 psi down to about 50 psi, virtually
is some harmonics in our vertical feed/effluent exchanger. doubles the amount of olefin and diolefin contents of
Apparently, we get into a flow regime where, at the lower the product.
pressures, the velocities are just right and we get some The problems that you get are typically the following.
tube vibration. We can kind of play the exchanger. We For gasoline production, there will be greater instability
can get the vibration from different places in the exchanger of the gasoline and increase in additive that will be
as we raise and lower the pressure. Obviously, wearing a required. For aromatics production, there will be an
hole in the tube in the exchanger is not a good idea, so increase, as already mentioned, and the need of clay treat-
we run the pressure as low as we can, but just above ment, which gives rise to clay life and clay disposal prob-
where we can hear the noise. It is a noise that kind of lems.
sneaks up on you. It is not like a bang or an alarm. Your For this last point, we do have the Arofining process
operators may have to be warned to listen for obscure licensed by IFP and HRI, which is the selective hydrogena-
noises. tion of orefins and diolefins in the reformate. We feel
that that is a good solution.
ROMAN: For the other points on the yield benefits resulting
At our Cattletsburg refinery, we have two CCR units, from further pressure reduction, we do not consider those
one that runs at 110 psig, and a second unit that runs to be attractive enough compared to the associated prob-
at 50 psig. We normally run the lower pressure CCR at lems which have already been highlighted.
higher severity to take advantage of the better yields. We
see a rapid decline in clay treater life in our sulfolane unit FUSSELL:
when we switch the feed from the 110 psig unit to the On our UOP CCR, we have plenty of room left in
50 psig unit due to olefins. This is a particular problem, our regenerator. Several years ago, we took a hard look
since this clay treater on our sulfolane unit is undersized, at running lower pressure. Our unit currently runs at
and operates at a 1.8 liquid hourly space velocity. We approximately 110 psi on the product separator. We have
would normally design these clay treaters for a 0.5 liquid two 100% booster compressors that have not been known
hourly space velocity. This forces us to actually lower the for their reliability over the years. In order to run a lower

150 Light Oil Catalytic Processing

TOC/INDEX

pressure, we were going to have to run these two machines 2. Mechanical

in parallel. After doing a very thorough economic evalua-
tion, we decided that the reduced reliability associated Question 34.
with running both machines 100% of the time would What methods have been used to control salt build-up
pretty much overcome all the positive benefits of the in the recycle compressor, other than washing it? How
increased yields and increased hydrogen production. effective have they been?

ANGELO FURFARO (UOP): FOSTER:

The panelists have provided some excellent informa- For control of salt build up in the recycle compressor,
tion with which UOP generally agrees. Regarding Ms. we have tried several methods with some success:
Fussell’s comments on the net gas compressors, UOP has ● liquid hydrocarbon injection at the compressor suc-
reviewed this situation with customers, and many with tion;
the same concern have decided to add a third reciprocating ● dispersant injection at the compressor suction; and
compressor in parallel to handle the additional net hydro- ● installing demister mesh at the compressor suction.
gen production that results with the higher reformate These have all worked to some degree, but none with
yields and/or higher severity operation following the raving success. I can recommend any of those, but before
revamp. Based on more modern compressor designs, the you do any of them, be sure to wash your compressor,
reliability of reciprocating machines is now in the 90Ⳮ% because you do not want to start dispersion injection into
range, reducing maintenance concerns. a running compressor that is not clean. You may flake
Revamps must be designed to accommodate the con- off pieces of salt with significant problems afterward.
straints put forth by the customer. In general, you try to These are all treating symptoms of a problem. The
save as much heater as possible, as well as the recycle problem is that if nitrogen gets into your system, and if
compressor. UOP has performed dozens of process unit you maintain a HCl concentration in the reformer, you
revamps, either to upgrade existing older CCR platform- are going to form ammonium chloride salts somewhere,
ing process units, or to convert semi-regenerative reform- and they will deposit somewhere. The best method is to
ers to the CCR platforming process. UOP has found that minimize nitrogen entering the reformer section.
most concerns can be overcome, and performance and Since we went to a higher nitrogen feed content, we
financial objectives met, through detailed studies that have had to revisit our HDS reactor and operate it more
include good communications with the customer. as an HDN reactor. One example of a discovery that we
As noted by the panel, lower pressure operation found was that over the years, we had neglected to run
improves catalyst selectivity, increasing yields, but com- it at as a high a pressure as we were capable of running
pressor limitations coupled with lower operating pressure it, because as an HDS reactor, it was not that big a deal.
will decrease the available hydrogen recycle (H2/HC ratio), As an HDN reactor, it was. So we found that we had
thus increasing coke make. The effect of the improved some room to move upward in our hydrotreating pressure,
selectivity on the heat of reaction needs to be accommo- and we were able to raise the pressure 50 psi, with a
dated. The increased coke production due to lower pres- few very minor modifications. Our goal, ultimately, is to
sure and H2/HC needs to be accommodated. The effect control total nitrogen content to the reformer down below
on gas/liquid separation also needs to be considered. UOP 250 ppb.
has a lot of experience in revamping platforming and CCR
platforming units to lower pressure and in revamping
regeneration sections to higher capacity.

STUART GRAY (Criterion Catalyst Company):

Criterion has recently commercialized a new CCR cata-
lyst, PS-40, which is generally targeted for first generation
units with coke burning capacity limits. However, it can
also show benefits in more recent generation units due
to its increased stability. In pilot plant tests, the catalyst
has shown a 30% to 50% reduction in coke on catalyst.
Refiners will be able to use this to either reduce pressure
and increase yield, bringing the coke back up to the
precatalyst changeout level, increase feed rates or process
different low grade feedstocks. The catalyst is a modified
platinum/tin version of PS-20.

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 151
TOC/INDEX

I talked a minute ago about the method for analysis CAIN:

and UOP did, too. The overall thing I would advise At Star, salt build-up on the recycle compressor has
you to do is to start looking at your hydrotreating as a not been a big problem for us. We normally water wash
denitrification operation. That means looking at your with hot condensate whenever we are down.
catalyst, as well. There are catalysts on the market now
that do a much better job of denitrofication than what THIERRY DARDE (Pall Corporation):
you may have used in a reformer hydrotreater in the past. I guess at this point I should just reiterate the statement
I made earlier. Our high efficiency liquid gas coalescer
BARLOEWEN: can really protect compressors, and remove any trace of
I would reinforce the statement about catalysts. The ammonium chloride and then suppress any extensive
California refineries have been dealing with high nitrogen maintenance.
feedstocks in the reformer hydrotreaters for many years.
Much of the rest of the country has tended to use naphtha KEVIN FERRAL (CITGO Refining & Chemicals):
hydrotreaters as a place to reuse spent catalyst in a second There is one thing to be aware of, too, when you use
or third service. As you shift your feed slate, it is important dispersants in washing systems to move the salts. When
to put in a high activity, fresh, Nickel Molybdenum cata- you do your MOC review, check the metallurgy of your
lyst. You will be surprised at how much better you can do. equipment downstream, because we did experience metal-
lurgical problems and severe corrosion problems with
injecting various inhibitors to try to prevent salt deposi-
BINFORD: tion.
I agree with the previous comments about trying to
treat the cause and not the symptom, and instead getting Question 35.
at the root cause. For those refiners who are still having For continuous catalyst regeneration systems using
problems with fouling, dispersant-based antifoulant pro- stainless steel instead of Inconel, have there been prob-
grams can be very effective. It is very important that you lems with stress corrosion cracking? What precautions
start those up on a reasonably clean system. Once the need to be taken?
system gets too fouled, you just stand a real risk, especially
on a centrifugal machine, of creating imbalances that can ROMAN:
tear a compressor apart. We did not use 18-8 austenitic stainless steel materials
I did want to note a couple of things that are very in any of our regenerators since they are susceptible to
important to designing a program that is going to be stress corrosion cracking in hot chloride environments.
successful. Most of the programs that are unsuccessful, I The presence of chlorides can be particularly harmful
think, are because of poor thought going into them. For when the unit is down and free water might be present.
any dispersant program going into a compressor, good Increasing the nickel content above 32% increases the
atomization of the product is a key to success. Using a resistance to chloride stress corrosion cracking and is rec-
pneumatic atomizing nozzle, with recycling a small ommended. Monel, Inconel and Incoloy are all accept-
amount of the gas from the discharge of the machine, is able. We use Inconel 600 material and have not had
one way to go. This would be applicable to reciprocating any problems.
compressors, especially where you can not tolerate any
liquid. Obviously, the entire system needs to be well Question 36.
designed and built to withstand constant vibration and What is the typical life of the vertical combined feed
potential metal fatigue. The last thing you want is some exchanger (not Packinox) in reforming service? What
small piping around that compressor that is going to type and frequency of inspections and maintenance are
break loose. If you can tolerate some liquid (a centrifugal recommended?
machine), using a light naphtha as a carrier through an
atomizing nozzle can be very effective as a means to CAIN:
disperse the product. In this case, you have some naphtha At Port Arthur, we inspected ours after a 10 year run,
carrier going through the machine, which will also help and it did not show any problems. At our Delaware City
carry the dispersed salts out of the compressor. The choice plant, we also did not have any problems. Typically, the
of the product itself is very important. It must be compati- bottom baffle and distributor are inspected during every
ble not only with the carrier, but also must have good turnaround, which is approximately every 4 years.
light hydrocarbon compatibility and good salt dispersing
capability. These are a few of the things that you really FUSSELL:
need to look at closely if you are going to consider a At our Port Arthur refinery, we have two parallel verti-
chemical treatment alternative. cal feed/effluent exchangers on our CCR. They have been

152 Light Oil Catalytic Processing

TOC/INDEX

in service for 12 years. In 1995, we had a major turnaround We do not have the GC detection method. We use a
on the unit, and we strongly considered pulling these beta particle generator and a little ammeter that measures
exchangers. We actually put together a pretty detailed the electron absorption by the sulfur hexafluoride tracer
plan on how to pull them. But as we got closer and closer in the system. This system had also been used in one of
to the day of the turnaround, we decided against pulling our refineries to find reactor bypass in a reformer. We
them for several reasons. Number one, they would take were able to determine that, yes, we had to get in these
up space, and the equipment required to pull them would reactors and seal the tops of the beds.
also take up space required for other activities during the Our turnaround cycles on the exchangers themselves
turnaround. We made phone calls to UOP and to other are in the 3 year to 7 year range.
refiners and determined that very few people have actually
pulled these exchangers. ROMAN:
So, instead of doing that, we did a pretty extensive Our experience is the same. These are the exchangers
inspection. We did acoustic emissions; we also did ultra- that never get pulled. We have not had any leaks. They
sonics. We removed the insulation and did ultrasonic do not foul.
testing in order to determine that the exchangers were fit One further item on the leak detection, we have used
for service. We have not seen any heat transfer loss on these all three methods: radioactive tracers, helium, and sulfur
exchangers, so we have never really had any requirement to hexafluoride. We stopped using radioactive detection due
pull them for cleaning reasons. to the handling problems. If you have the capabilities, the
sulfur hexafluoride is much more sensitive than helium.
HAHN:
We also have two vertical feed exchangers in service. SAMUELS:
One is 8 years old in our service, and it was a second- In our Robinson refinery, we have not pulled the
hand to us, so it has several prior years service in addition exchangers in their 16 years of service. However, due to
to our service. Another unit has an exchanger that is over loss of heat transfer prior to a spring turnaround this year,
10 years old now. The only problem we experienced was we did try to chemically clean the shell side of these
one leaking baffle, about 2 years ago. Now we inspect exchangers without pulling them. We used a potassium
the bellows about every 5 years during turnaround. There permagenate solution and an acidizing wash. Coming
are manufacturers recommendations for how to pull these back on-line, we actually lost heat transfer. We believe
bundles. We studied these but we decided that we could the problem was due to lack of velocity of the circulating
not justify it, either. We have not seen any heat loss either. solution, and we think we moved some of the deposits
around. So we do, in fact, plan to pull those bundles in
MALEK: the near future. The way we intend to do this is to actually
Our experience with these exchangers with our clients build a superstructure at the unit site and suspend the
has been very good, also. As a matter of fact, one mainte- bundles for cleaning, rather than laying them down.
nance supervisor recently told me that his exchanger had Since the Q&A session, we have pulled and cleaned
already outlasted four refinery managers. these bundles. The primary cause of fouling was tube-
side pluggage of heavy hydrocarbons. The tubes had been
PETERSON: hydroblasted without dropping the tubs and tube-side
We have 17 years of operation on our feed/effluent internals. This is a must for hydroblasting the tubes.
exchanger. We have had three leaks of the expansion Hydrocarbon on both tube and shell sides was poly nuclear
bellows at the bottom end. The first time one was found, aromatics, with a higher molecular weight on the shell
we were successful at weld repair for a while, but that side than on the tube side.
weld repair later failed again. The last time that we had
the exchanger open, we replaced the bellows. We have R. RAJAMANI (Mangalore Refinery and Petrochemicals Ltd.):
never pulled the exchanger to look at the inside of the Did this question address only the non-Packinox
shell or to look at the outside of the tubes. We did do exchangers?
acoustical emissions testing to help us determine that
the shell did not have any active indications (to use an ANGELO FURFARO (UPO):
inspector term). The CCR platforming process unit at Mangalore has
We had some heat transfer loss this last time, and we a Packinox welded plate exchanger. Those exchangers
were overjoyed when we took the dollar plate off the top have, in CCR platforming service, shown excellent non-
to see that we had a bunch of tubes plugged. We were fouling operation.
able to wash that out and get all of our heat transfer back. However, in regard to this specific question, the life
As far as looking for leaks on-line, Mr. Roman pre- of VCFE’s is similar to that of other process equipment.
viously mentioned using the sulfur hexafluoride method. Typical life can extend well past 15 years, but that can

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 153
TOC/INDEX

be shortened by internal vibration (especially in early some testing on the spent Sulfolane to check the chloride
vendor designs), improper maintenance, corrosion, and content, and then I can compare notes with Mr. Cain.
so on. Inspection generally consists of pulling the shell
cover, dropping the floating head and dye checking the JASON NOE (UOP):
bellows. Exchanger component metal thickness readings Traditionally, UOP has removed chloride from the
are also taken. In fixed bed reforming service, corrosion solvent systems by passing a slipstream of solvent through
can occur at the bottom inside surface of the shell cover single stage vacuum distillation or single stage steam strip-
(a dead area) if this area is not regularly drained during pping. More accurately, we allow mostly chloride salts
regeneration and unit restart. Also, there is another poten- to accumulate in the bottoms material of this type of
tial area for corrosion on the outside diameter of the tubes equipment which is manually removed intermittently.
at the very bottom, just above the floating head tubesheet. UOP recommends the injection of monoethylamine
You can see this without pulling the bundle. UOP does (MEA) to neutralize the acid contaminants that can form
not require the pulling of this bundle unless necessitated in extraction units as a result of the breakdown of the sol-
by process reasons (fouling), or dictated by the regulatory vents.
agencies, company policy or a real need to do so. If the Some operators have used ion exchange to remove ionic
bundle is pulled, proper support of this piece is mandatory species in general from the process units. This appears to
to avoid damage when handling or placing the bundle in be a very efficient method for those who have difficulty
the horizontal position. maintaining a clean solvent stream, but this practice is
not the norm for the majority of process units.
D. Aromatic Extraction A slipstream processing of solvent through activated
carbon has also been pursued by some operators, but this
method, also, does not appear to be widely-practiced, and
Question 37. would tend to produce more solid waste than the previous
What methods are available for removing chloride from two strategies.
a solvent system?

CAIN: E. Polymerization
In our Delaware City plant, we operate a solvent vac-
uum regenerator on a continuous basis to remove chlo-
Question 38.
rides from the solvent. We target approximately 100 ppm
Where is Dimersol catalyst injected, and what ratio of
chloride level in the solvent.
alumina to nickel do you maintain?
TASKER:
In HRI’s sulfolane aromatic extraction process, we do TASKER:
have a continuous slipstream to solvent regeneration. The The Dimersol process uses two different types of cata-
amount that you put through is a function of just how lysts. One catalyst is injected into the feed line and the
well you are controlling your pH and oxygen ingress into other is routed to the recycle stream before entering
the recovery tower. the reactor.
In one of our licensed processes in Japan, we have run The ratio of these catalysts is somewhat proprietary
18 months with water white sulfolane, which is most information, but the range is typically in the order of 12
unusual. It is an extremely tight unit. The two things to 16 on a molar basis.
that affect solvent degradation most are the build-up of The injection points have been changed. Injection is
acidic compounds due to high temperature operation and after the pumparound system at the coldest point of the
the oxygen ingress. You get acidic compounds forming loop. The nickel ‘‘soap’’ is injected to either the suction
quite readily when you have oxygen leaking in. It is often or the pumparound pump or into the feed. The nickel
very difficult to detect the oxygen ingress except by looking ‘‘soap’’ is usually where the highest concentration of pro-
at the degraded solvent that you pull from your continuous pylene in the feed occurs. We have found that quill injec-
on-line regenerator. tion is the best method for mixing. Previous use of static
We generally advocate removal of a small quantity mixers has been dropped from our designs.
every few weeks, like up to 5 gallons, but this is directly
a function of how well you are operating the unit. We PETERSON:
are not specifically looking at the chloride content as such. Our EADC is injected in the air cooled reaction recycle
We are just maintaining good pH control in a carbon or pumparound cooler with the outlet returned back to
steel system, and we see no corrosion. So one thing, based the reactor. Our aluminum to nickel ratio is currently
on the question, that I am going to do is explore doing being run within the range that Mr. Tasker mentioned.

154 Light Oil Catalytic Processing

TOC/INDEX

Question 39. minutes every shift through the pumpout line to the
Has anyone had problems with forming sludge in Dimer- caustic treating section.
sol pumparound coolers? If so, how have you con-
trolled it? STEVE MATHUR (Chevron):
What we have found is that sludge is a way of life with
TASKER: a Dimersol circulation loop. I suggest that if you are
Some of our licensees have problems with sludge in the seeing a lot of sludge, and your catalyst consumption is
Dimersol pumparound cooler. IFP and HRI recommends high, especially the EADC, then you probably have a
purging some of the reactor bottom stream to the down- problem with the molecular sieve driers. Check the regen-
stream neutralization section, occasionally, to minimize eration cycle, the state of molecular sieves, the regeneration
the sludge deposits. temperature, etc. That is where we found our problem,
In efforts to improve the overall situation, we are now and were able to reduce the catalyst consumption by half.
proposing a slightly different and new philosophy. We
now look at having the process fluids in the tubes for Question 40.
high velocity. However, this may not meet the client’s How frequently do you check for contaminants (ammonia,
requirements and therefore, we may have water on the water, diolefins) in Dimersol feed? What test methods
tube side in the U.S.A. This is a client choice. are used?
In addition, the design and type of the pumparound
cooler is important. It is important to avoid the dead HENKE:
areas which occur in the box headers in air coolers, since We have an on-line water analyzer that checks moisture
this is where sludge can accumulate. Our experience is in the discharge of our feed dryers. This analyzer can be
that sludge comes from impurities in the C3 feed. The switched to sample the dryer inlet as well. We only check
main contributors are water and dienes and, therefore, for ammonia occasionally with Draeger tubes, and have
driers can be justified. Levels of contaminates which can only had one or two minor upsets with ammonia in the
cause problems are about 5 ppmw with diene and the last several years. Generally, if the water is kept out, the
saturation level of water, which is about 1000 ppm in ammonia will not be a problem.
the C3 feed. So, the solvent needs to be dry. Therefore, We have not detected diolefins in our Dimersol feed.
in start-up procedures, a lot of emphasis is placed on While diolefins are difficult to detect and measure, we
ensuring that the unit is as dry as possible before operation. have found gas chromatography to be a reliable method.
Corrosion has been a problem in the units, as well When testing for contaminants in the laboratory, we
as erosion. use ion chromatography for ammonia, and Karl-Fischer
titration for water.
PETERSON:
We are one of the units that has a problem with sludge PETERSON:
formation. We see it in our recycle cooler, or pumparound We probably do not analyze as often as we should for
cooler. This is an air cooler, so we do have the material some of these contaminants. We do try to control the
on the tube side of the exchanger. About once a year, we amount of diolefins that get into the system by running
have to shut the unit down and clean the cooler, the lines, our C3/C4 splitter in such a way that quite a bit of the
and the reactor. We determine the need to shut down propylene goes down the tower, so that we do not have
by a decrease in recycle flow and a decrease in the cooling a very good chance of getting any of the diolefins going
capacity of the system. up the tower into the Dimersol. We do have driers on
A basket strainer has been installed in front of the the system, and we try to keep the operation of the
cooler, and I do not know how much of a change this depropanizer or the splitter, the drier, the Merox system,
has made to the situation. The cleaning is done by hydro- and the LPG aiming system running very steady, and just
blasting, and because the system is intolerant of water, try to mind our business there to keep the dirt out of
we do a hot nitrogen purge or strip of the system to the system.
get all residual moisture out of the system before we
reintroduce the feed and catalyst. CAIN:
We have on-line analyzers for water and diolefins. We
CAIN: use a Panametrics analyzer for water and a GC for the dio-
We monitor our feed quality for diolefins, sulfur, and lefins.
moisture. We regenerate our molecular sieve driers every
24 hours, and we have lowered our reactor temperature TASKER:
from 125°F to 110°F, which seems to have helped. We I agree with everything that has been said. You can
also blow down our reactors, about 50 bbl/hr, for 5 use an ASTM method to check for ammonia. IFP and

1997 50th Anniversary NPRA Q & A Session on Refining and Petrochemical Technology 155
TOC/INDEX

HRI recommend testing the feed for contaminants on a attractive revamp opportunity for improving the perfor-
daily basis. For the moisture, the DuPont and Panametric mance of existing oncethrough zeolitic units and TIP
type have been quite good for measuring the moisture in units.
the Dimersol feed. For diolefins, we recommend the IFP
D2-77 method to test the diolefins. Question 42.
Have any refiners experienced excessive yield loss in
isomerization units? To what has this loss been attrib-
F. Isomerization uted?

Question 41. PETERSON:

Have the new high activity zeolite and chlorided isomer- The biggest impact that we have seen on yield loss in
ization catalysts proven to be a significant improvement isomerization units has been when we go into a Clean
over their predecessors? Air Act benzene destruction or benzene prevention mode,
where we try to include all of the benzene precursors (i.e.
C6 compounds) into the isomerization unit instead of
FLETCHER:
down into the reformer. Our naphtha splitter was not set
Two new high activity chloride catalysts produced by
up for a perfect splitting between C6 and C7, so we get
Akzo Nobel, named AT-2 and AT-2G, have proven com-
a considerable amount of heptane type material into the
mercially to be more active than existing amorphous cata-
feed. A considerable amount of this apparently cracks the
lysts. Presently, a 13,000 bbl/d C4 isomerization unit in
propane, and this propane is lost to the fuel system, since
Dubai has been running very satisfactorily on AT-2 for
we do not have an LPG recovery system on the stabilizer
more than 2-1/2 years with an estimated activity improve-
offgas. We looked at putting in an LPG system, just to
ment of approximately 40%. This leads to higher isobu-
collect this propane, but with the caustic wash to control
tane formation due to the improved thermodynamics at
the HCl, we have not been able to justify it. Occasionally,
the lower reactor temperature, or to higher throughputs
when we feed some natural gasoline that has a little bit
with the same yield.
of normal butane in it through our naphtha hydrotreater,
For C5 and C6 isomerization, two refineries, one in the
and on through the isomerization unit, we get enough
U.S. and one in Europe, have chosen to use AT-2G. Both
butanes that, coupled with the propane, may make recov-
are reaching their targets while using less catalyst than
ery justifiable. We are studying this at this time.
before. Other ways to exploit the catalyst improvement
include: l) increasing isomerization capacity with the same Question 43.
product quality, or 2) treating more severe feeds like How are refiners addressing salt deposition and corro-
benzene. In the latter case, the heat formation penalty, sion in their isomerization units?
which is caused by treating more benzene, is compensated
for by the lower operating temperature of the catalyst.
PETERSON:
We think we have seen ammonium chloride deposition
FUSSELL: in the reactors that has led to higher pressure drop in the
We are currently using the UCI Isopar catalyst in our system. The salt ties back, as discussed previously, to the
TIP unit. We loaded it about 11⁄2 years ago, and we have naphtha hydrotreating section, and with the nitrogen in
been very happy with it. It is still holding close to start- the feed. We have a 300 psi separator naphtha hydrotreater
of-run activity after 500 days on stream. Conversion is with space velocity of about 12, and are running some
also higher than with the catalyst we previously had in fairly high sulphur crude. We cannot really go to the
the reactor. Another plus is that the catalyst is more nickel-moly catalyst in this case, as Mr. Barloewen had
moisture tolerant—with excursions up to 800 ppm with suggested. So we have had to limit the amount of high
only a temporary loss in catalyst activity. nitrogen crudes that we run. We have been fairly successful
at keeping the ammonia levels down in the reformer and
BRIAN JOHNSON (UOP): in the isomerization unit since then.
UOP has recently commercialized the LPI 100 catalyst. As for corrosion in the downstream part of the unit
This catalyst is based on a novel sulfated metal oxide in the offgas scrubber from the stabilizer, we used to have
formulation. The catalyst is approximately 150°F more a lot of leaks in the caustic system. What we now do is
active than the industry standard zeolites. This means test the caustic for free caustic on a daily basis. In the
that the operating temperature for this catalyst is in the past, however, at times we more than spent it. We went
350°F to 400°F range. It is also fully regenerable, and to an operator pH test of the system on their rounds
tolerant to such contaminants as water. The new catalyst through the unit every 4 hours. We are not trying to run
is used in the Par-Isom process, and represents a very to a 7 pH or anything like that. If they see anything less

156 Light Oil Catalytic Processing

TOC/INDEX

IV. LIGHT OIL CATALYTIC PROCESSING

A. Alkylation-Sulfuric Acid published by the refiner in the July issue of Hydrocar-
bon Processing.

Feed Process
Question 1.
Question 2.
What analytical methods are recommended for determin-
Has anyone used the new alkylation aids/additives (e.g.
ing the water content of the alkylation unit feed and spent
Alkat AR) to reduce acid consumption? If so, what was
sulfuric acid? Please report typical levels of water in
the acid reduction observed as compared to the vendor’s
these streams.
predicted reduction? Any problems?
ZETLMEISL:
Karl Fisher titration is used to determine the water HIGGINS:
content of the sulfuric acid and the alkylation feed. Typical No panel response.
water levels in the sulfuric acid are 2.5-3%. Typical levels
in the feed are 500-1,000 ppm. The goal obviously is JAMES D. WEITH (Fluor Daniel, Inc.):
to minimize the amount of water in both the acid and When I was formerly with an operating company we
the feed. tried this. Because the feed to the Alkylation Unit varied
so much due to the variable operation of the FCC, the
TURPIN: only thing we could check for was statistical significance.
There should not be any free water content in the We compared two 60 day operating periods with and
olefin feed to a sulfuric acid alkylation unit going to the without the additive. While the data appeared favorable
coalescer, and should be no more than 10-15 ppm coming for the additive, we could not detect any significant
out of the coalescer. If you can get a sample, without improvement by its use. Therefore, we could not tell if
contaminating it with water in the process, the water in it was noise from the data that made it look favorable for
the feed can be measured using a Karl Fisher titration. If the additive, or the additive itself. Because of that we
you really feel you have a problem with water in the feed, concluded that the additive was too expensive to warrant
the installation of an in-line analyzer (Panameterics or using. There were no problems with the additive when it
DuPont for example) is probably the best way of monitor- was in use, however; and its supplier was very cooperative.
ing. The coalescer performance can be checked against
hydrocarbon saturation tables. Fresh acid is going to be HAROLD J. EGGERT (BetzDearborn):
typically 0.5 to 2% water. The spent acid is typically This new chemistry is about 18 months old. We have
about 3-4% water on a straight b-b feed when the fresh had nine field trials and currently there are six ongoing
acid is about 1.5% water, and lower if the fresh acid applications. The acid savings we have seen on in-plant
strength is higher. This can be measured using a modified evaluations run in the 5-20% range, depending on the
Karl Fisher technique. When small amounts of ethers are unit. The chemistry works, the critical question is whether
present, the water is going to go up—by maybe 1-2%. it is economically viable. It has to be judged on a unit-
About 10 lbs of acid must be purged for every pound of by-unit basis.
water that enters the reactors. Incidentally, water dilution
lowers the catalyst (acid) activity 3-5 times as fast as Question 3.
hydrocarbon diluents. We have experienced corrosion problems in floating roof
Lyle Albright talked about water and other feed impuri- tanks storing aviation alkylate. The corrosion takes place
ties in his series of articles that appeared in the Oil and in the wetted area above the roof apparently from contact
Gas Journal starting November 12, 1990. A second good with the Gulf Coast ambient moisture. We run NACE
reference is; Petroleum Refiner, Vol. 37, No. 9, September spindle tests on the gasoline, and have never shown a
1958, pg. 324. problem. The bottom of the tank never shows signs of
corrosion, yet we have to replace the tank shell every
MONIQUE STREFF (Micro Motion, Inc.): 12 years due to internal corrosion. We have theorized
Micro Motion’s meter is being used to successfully esters and S02 as the culprits, but we don’t have a way
measure spent acid concentration. This has resulted in a to quantify the corrosivity of the gasoline, so that we can
considerable reduction in acid consumption. A paper was effectively treat it. Any ideas?

122 Light Oil Catalytic Processing

TOC/INDEX

ZETLMEISL: to regenerate it. Small amounts of nitrogen are probably

The first thing the questioner is looking for is a better fixed to nitrate in the process.
way to quantify the corrosivity of the gasoline. Rather After several cycles of regeneration, the nitrates build
than rely on the NACE spindle test alone, we also run up to detectable levels. The upper limit of nitrates is about
ASTM D-665, which introduces saltwater and is run for 50 ppm and according to STRATCO, nitrates at this
a longer time. This test is more severe than the spindle level are not a problem.
test and sometimes we see corrosion with it, even when
we do not see corrosion in the spindle test. Question 5.
On the rest of the question, the questioner is looking What is a good procedure for water washing the alkylation
for speculation about the corrosion of this floating roof unit deisobutanizer tower?
tank, and speculations are fun. So I have two speculations,
JACKSON:
if you do not mind.
A typical DIB water wash procedure was reported in
My first suggestion is to investigate the possible pres-
the 1996 Q&A Transcript, page 161 by Mr. Charles
ence of bacteria. It might be that the bacteria are killed
Lemon, the Tech Service Manager at Tosco.
off in the bottoms, but not on the walls.
The fouling in the tower is caused by neutral ester
Another speculation, which is a bit more of a reach,
degradation products, which can be cleaned up using
is that traces of diesters from the sulfuric acid might
online water wash procedures.
absorb onto the walls and hydrolyze when exposed to
● Reduce charge rate to minimize alkylate production
moisture. This situation would produce localized low pH
● Divert the DIB bottoms to a dedicated tank that
conditions on the walls. Explaining why the bottom does
is set-up with vapor recovery
not also corrode is a bit more difficult. It may be that
● Reduce DIB bottoms temperature to minimize
more water gets down into the bottom and dilutes the
water revaporization (250-270°F) while watching
acid so that you have a more moderate pH in the bottom
Alkylate Rvp
of the tank.
● Inject raw water (TDS ⬍ 6000 ppm) sometimes
RADCLIFFE: with a surfactant into the top of the column ⬇ 8%
This question presumably relates to a sulfuric acid of the total DIB feedrate.
alkylation unit. However, I do have some related experi- ● Monitor overhead accumulator closely and drain
ence for a HF alkylation unit. What happened was we water at lowpoints to prevent recycling water to the
experienced corrosion of the floor plates in the alkylate Rx section. Also watch the N-C4 side draw for water.
rundown tank. To counteract this we kept a heel of caustic ● Procedure takes approximately 2 hours.
in the tank at all times, which we then tested weekly
RASBOLD:
for strength.
Regarding testing for the corrosivity, what we discov- We have successfully on-line water washed our Toledo
ered was that if you ran a copper strip corrosion test, Alky DIB tower several times in order to remove accumu-
which is the normal way we checked the alkylate rundown, lations of salts and corrosion products. Condensate was
you would get a negative result. But as soon as you put the washing medium and was injected into the tower via
some water with it, you would see the copper strip go the butane stream. The BB feed rate was reduced from
off right away. The issue is the alkylate is so dry that there a normal 1400 barrels per hour to 300 barrels per hour.
is no reading; the acid is not reactive without the water. The butane feed rate was reduced from 300 to 100 barrels
per hour. The condensate injection rate was maintained
ROBERT E. DAVIS (R.E. Davis Chemical Corporation): at about 50 barrels per hour. The bottoms tower tempera-
This may have a bearing. Even the acid manufacturers ture was reduced to 260°F. We continued these conditions
face this problem with their spent acid tanks. The release for 15 hours while monitoring the effluent water in the
of carbon dioxide off of the alkylation reaction is certainly alkylate for pH and visual color. Once completed, we
possible. Water vapor and CO2 form carbonic acid at the returned the tower to normal operating conditions.
interface. That is one of the areas of corrosion it can Results of a typical wash was an increase of 30°F bottoms
enter. The other is the sulfur dioxide diesters or the par- temperature and a reduction in alkylate RVP of four num-
tially reacted products that escape the caustic water wash. bers.
Question 4. Question 6.
Has anyone determined a source of nitrates in spent acid? Has anyone used an iron supplement like ferrous sulfate
in the fresh acid to prevent corrosion (Paper No. 494:
ZETLMEISL: Corrosion Abatement in Sulfuric Acid Alkylation Unit Hori-
The most likely source of nitrates is from the regenera- zontal Contactors, Corrosion 97, NACE 1997)? If so, have
tion of the sulfuric acid. The spent acid is burnt at 1000°C you noticed any side effects?

1998 NPRA Q & A Session on Refining and Petrochemical Technology 123

TOC/INDEX

ZETLMEISL: WALDRON:
This question and the paper that is referenced in it At Beaumont, we are taking a little bit different
refer to the fact that saturating the acid with ferrous sulfate approach to acid control, which we plan to close loop by
removes one of the key driving forces for breaking up the the first quarter of 1999. For two months, we have been
passivating surface on carbon steel in concentrated sulfuric using online calculations based on operating conditions
acid. The corrosion mechanism in sulfuric acid is mass and feed contaminant levels to set fresh acid flow targets.
transfer limited under turbulent flow conditions. The This is a predictive type of control instead of a reactive
corrosion rate is determined by the difference in concen- control using a spent acid analyzer. The results to this
tration of dissolved iron sulfate at the corroding steel point have been very positive. We have been meeting acid
surface versus its concentration in the bulk solution. spending targets much more consistently previously.
I do not have any direct experience with the practice,
but I contacted someone who was associated with the J. BRIAN HART (Aspen Technology):
work reported in the paper, and he knew of two locations I wanted to followup on what Mr. Turpin mentioned
that were using the practice. about closed loop advanced control on this problem.
Aspen Tech has successfully implemented advance control
using a DMC plus controller to control spent acid
ROBERT E. DAVIS (R.E. Davis Chemical Corporation): strength.
A couple of refiners got into serious trouble with the
fresh acid tank having a heel of iron sulfate which had Question 8.
dropped to the bottom. As they pulled from the tank, How often should a refiner monitor the pH of the water
the sulfates were increased in the acid feed to the reactor. from the feed coalescer? What is the normal range of pH?
This generated foaming in the settler as the particles tend
to ebulate. The root cause was excess iron sulfate. LAMB:
STRATCO would suggest that the frequency of drain-
Question 7. ing the water from the coalescer, should be three times
Is anyone using an advanced control scheme to control a week. The target pH is 6 to 7 with an MTBE raffinate
spent acid strengths? alkylation feed and 8 to 9 pH if a mercaptan treating
system is upstream of the alkylation unit.
TURPIN: Question 9.
For years managing acid strength has been a control Has anyone ever experienced oscillations in the acid
objective. In fact, American Oil tried to do this on the settler acid level? How about with contactor reactor dP
first alkylation on-line optimization project in 1961. The and amperage? If so, what was the frequency of oscilla-
problem is part handling the dynamics, and part in meter- tion and the cause of the problem?
ing acid strength. The acid strength metering problem
has been solved. Stratco has sold 18 of their acid analyzers, HIGGINS:
and there are other companies that have tried similar type No panel response.
instrument applications. The dynamics is tough because
of the 16-20 hours to steady state. This can be handled ROBERT E. DAVIS (R. E. Davis Chemical Corporation):
through multivariable control. I know of only a few suc- A couple of the designs of the multistaged reactors
cessful closed loop application of acid strength control. I have inherent in them a design flaw with which the intake
think the most recent reported application of open loop to the compressor is directly across from a PDIC. The
control was in the article ‘‘Reduce acid Usage on Alkyl- oscillation that was experienced in these plants was over-
ation Unit’’ that appear in the July 1998 issue of Hydrocar- come by extending that line considerably further. When
bon Processing. The reported acid savings were in excess the PDIC opens up to control delta P across the zone,
of 1gpm on a 9000 bpd unit. I am quite confident that the compressor takes suction on that line and it drops
had the analyzer been integrated into a multivariable con- the level.
troller the savings would have increased by another 25%. As you go forward, the preceding zone dumps to that,
I would like to add, controlling acid strength is only so you setup the oscillation. This, of course, changes the
one part of a larger optimization issue. The second part flow rate to the settler. So you will get a variation of the
of the problem is, knowing the optimum acid strength flowrate to the settler. The person that is experiencing
at which to operate. Just controlling acid strength, is like this should take a look at the engineering design. The
selectively shooting off a toe as opposed to firing a random PDICs should be set at a minimum of 1.4 pounds differen-
shot at your foot. There are big bucks to be mined in tial pressure per zone. If they can operate at 1.6, I think
these process operations. they will be a lot better off.

124 Light Oil Catalytic Processing

TOC/INDEX

Mechanical A/S New Fixed Bed Alkylation technology. Amoco has

determined that the FBA technology is not economical
Question 10. at this time for the Yorktown refinery. Despite the decision
What commercial experience have refiners had with the not to proceed with the alkylation unit at Yorktown,
various types of dual mechanical seals in sulfuric/HF Amoco remains interested in the development of the FBA
acid service? What is a typical run length for this type technology and may be interested in using that in the
of seal system? Are run lengths long enough to justify future.
the additional expense over a single seal system? Haldor-Topsoe A/S remains committed to commer-
cialization of the FBA technology and is actively pursuing
SMITH: a number of potential projects around the world.’’
Valero’s HF alky unit has always had dual seals in this
service and the seal configuration utilizes a tandem seal
design. This design does not afford any increase in primary BENNETT:
seal life over a single seal configuration. UOP has developed the AlkyleneSM technology to elimi-
What tandem design does afford is approximately a nate the hazards associated with liquid acids such as the
100% safety factor. The secondary seal provides a com- unmitigated aerosol production, and public risk due to
plete containment upon the failure of the primary seal. acid transportation. The AlkyleneSM technology is a solid
This allows an orderly shutdown, isolation of the pump catalyst process and is offered for commercial application.
and timely maintenance scheduling with no exposure of We have continued to discuss this technology with refiners
personnel to the process liquid. The tandem design is also in the U.S. and around the world. We have done a detailed
emissions free when maintained according to guidelines. design of the reactor section and completed the process
flow. We have done some estimated costs of the system
BENNETT: and have shown that the AlkyleneSM process to be competi-
We have been designing HF alkylation units with dou- tive with existing liquid acid technologies. Alkylate quality
ble seal pumps for the last ten years and we have had seal is comparable and catalyst consumption is low. We believe
lives of up to three years reported to us. this to be the alkylation technology of the future.
JACKSON:
Many refiners have converted the STRATCO alkyl- MARTYN E. PFILE (Phillips Petroleum Company):
ation contactor reactors to a double liquid cartridge Between January and March of this year, two new
mechanical seal replacing the original single seals or the liquid phase technology units were placed onstream, i.e.,
older two piece double seal design. There seems to be commercialized reduced volatility alkylation in Mobil,
good success with this seal arrangement with run times Torrance, California and Phillips, Salt Lake City, Utah.
exceeding three to four years compared to the 12 to 18 This process can also permit shipments of additized HF
month life of the previous seals. catalyst as well and is the subject of NPRA papers issued
in 1997-1998.
General Since that time, the process has exceeded Phillips’
Question 11. expectations, particularly in the area of reducing HF and
What progress has been made in commercializing new caustic consumption associated with acid regeneration.
technology such as solid catalyst bed alkylation for the
alkylation of olefins? Question 12.
Is anyone aware of any new materials on the market
LAMB: that are highly resistant to corrosion/erosion attack from
Amoco and Haldor-Topsoe have been working for sulfuric acid and/or low pH water?
some time to develop Haldor-Topsoe’s FBA technology
for possible application at the Amoco Yorktown refinery.
Raytheon Engineers and Constructors as well as others WALDRON:
were short-listed to bid on this project. Shortly after the We are not aware of any new material to resist corrosion
announced acquisition of Amoco by BP, this project was erosion in this service. Mobil Beaumont Refinery uses
terminated and there was an announcement by Amoco alloy 20 in services, which are dilute acid or highly erosive.
dated September 8, 1998, as follows: We have used NI resist and CD-4MUCU in erosive
‘‘Amoco has decided that it will not build an alkylation services for rotating equipment. We have also successfully
unit at the Yorktown refinery. This would have been the used 316L stainless steel for acid concentrations not
first commercial alkylation unit to use Haldor-Topsoe exceeding five percent at ambient temperatures. This alloy

1998 NPRA Q & A Session on Refining and Petrochemical Technology 125

TOC/INDEX

may be useful for handling low pH water as referred to BRIAN JOHNSON (UOP):
in the question. UOP have been involved throughout the development
of API 751. We consider that API 751 and the more
general API 750 (Management of Process Hazards) are
B. Alkylation-Hydrofluoric Acid excellent resources for use as basic refining practice for
our licensees. This is particularly true for those whose
Process experience does not indicate specific differences from the
recommended practices provided within API 751. It
Question 13. should be noted by all that numerous references within
Is it the panel’s opinion that API RP-751 (HF Alkylation API 751 suggest or even call for modifications to be made,
safety) should be adopted as a basic refining practice? based on specific refiner records/experience.
Question 14.
GENTRY:
With less forgiving phosphate cooling water programs,
I originally was not sure if this question addressed the
what effective early detection systems are used for the
current version or the expected update of API 751, but
early identification of HF acid leaks? (Salt formation and
I will give you our opinion. Our current practice is to
pluggage can be significant if leaks are not corrected
follow our licensor recommended practices with enhance-
immediately.)
ments from the existing version of 751. Thus, we currently
do not hold 751 as an absolute basic refinery practice.
We understand our licensor will adopt the new version TRAEGER:
of 751 at which time we will probably move to adopt it Last summer, we experienced a leak in our HF alky
also. We have not seen the final version of the new 751, reactor bundle. As a result, we are now checking the pH
but it is our understanding that there was a significant of the cooling water downstream of HF acid containing
effort to resolve the differences between the licensors and bundles routinely every shift. The pH controller for the
thus, providing the field operators the opportunity for cooling tower basin was a local controller and we have
this to become an accepted recommended practice. For now brought that signal into the DCS where it is alarmed.
now the decision, I believe, is still out. Also, two years ago we purchased a fluoride analyzer
for use on various HF alky product streams. The analyzer
TRAEGER:
has proven to be very sensitive and accurate. Since the leak,
I guess I read the question a little different and I we have started analyzing the cooling water for fluoride
struggled with the definition of basic refining practice content. Although we have not had any leaks since starting
and what they were getting at. But my response is that the fluoride test, we have been able to see a shift in the
in many cases we already utilize API recommended prac- baseline fluoride levels when we switch on and off city
tices as basic refining practices. Industry has put into place water for our source of cooling tower makeup.
many of the practices, which are included in RP 751. On One additional possibility would be the daily monitor-
the other hand, RP’s are general in nature and not written ing of calcium levels compared to those predicted by
in enough detail to be considered design standards for cooling tower cycles. Lower than expected calcium levels
all units. could indicate a precipitation of calcium fluoride as a
result of a HF leak.
TURPIN:
SMITH:
API RP-751 was first issued in 1993, and is presently
going through the balloting for its first 5-year review. As At Valero, we monitor the fluoride concentration in
with all of the API recommended practices and proce- the circulating water. This is run in our laboratory with
dures, the contents are a function of the experiences of a specific ion electrode twice a week, not continuously.
the members of the committee. I think this RP is well For a major failure, an online continuous pH analyzer
written and the recommendations are quite solid. Usually, will tell you very quickly if you have got a problem.
as these practices age, that is go through more and more
revisions as the committee members are rotated, the RPs Mechanical
become broader and of more value.
The revised RP-751 is an excellent starting point for Question 15.
building a basic refining practice. However, each refiner What is your normal maintenance cycle on the HF or
needs to incorporate their own experiences in developing sulfuric alkylation unit reactors? Describe the areas
a basic refining practice customized to meet their special where corrosion/erosion are major concerns and what
needs and situation. repair methods are typically used?

126 Light Oil Catalytic Processing

TOC/INDEX

RADCLIFFE: unit. This has been reported in the past, but I will mention
The biggest improvement that I believe can be made in it again. We limit the amount of trace impurities of
maintenance cycles in HF alkylation units is in removing chromium, nickel and copper to a maximum of .2 wt%
water from the feed. At Milford Haven, we replaced the in everything we can. It is not always possible to meet
feed dryer activated alumina desicant with molecular these specifications, but any material with higher levels
sieves. That has allowed the maintenance cycle to be than that is flagged and we monitor it closely for corro-
changed from an 18 to 24 month cycle to a 48 month sion rates.
cycle. In addition, the amount of the scale deposition and
corrosion has been much reduced.
JACKSON:
Typical maintenance cycles of four plus years is being
BENNETT:
achieved. On a STRATCO unit, the items that tend to
For a HF unit, we recommend cleaning the bundle
require maintenance are the hydraulic head impeller, the
out and the shell at each turnaround so the bundle can
DP nozzle insert, the feed nozzles and the mechanical
be easily pulled and verified to be cleaned prior to res-
seal. Typical item such as wear rings, hydraulic head
treaming. We consider the corrosion rates on a low water
weldment and tube bundles will operate in the 8-10 year
content, low temperature acid system, such as this to
range. Repair methods for the impellers DP nozzles, feed/
be very low. In many HF units, reactors are original
mixing nozzles and the mechanical seals are simply to
equipment, even though they have long service lives.
replace with new parts. Mechanical seals can often be
rebuilt and used as a backup as required. Weld repairs of
MASTRACCI:
the impellers are almost always not a good idea as the
A 3-year turnaround schedule is used for the HF Alky
weld buildup corrodes and erodes away very quickly.
Unit at the Corpus Christi refinery. Areas of corrosion
As far as other items mentioned, the wear ring is typi-
concern include those where stress concentrations occur,
cally replaced when it is worn beyond recommended toler-
specifically, in the weld heat affected zones and nozzle
ance and the tube bundles are rebuilt when about 15%
attachment welds. We have completed several repairs
of the tubes have been plugged due to leakage. The shell
where the area is cleaned (sandblasted), overlayed using
cones are removed in the field and replaced with a com-
a carbon steel electrode and finally post weld heat treated.
plete new cone assembly when the actual wall thickness
approaches the minimum required thickness for design
SMITH:
pressure.
At Valero in Corpus Christi, our cycle on the HF alky
The hydraulic head weldment over time will corrode/
unit is every three years, which corresponds with the HOC
erode and repairs consist of weld overlay and remachining
unit turnaround. We have not experienced any corrosion
the head back to original specifications. Another solution
or erosion in the unit reactors. You should always look for
to handling the high wear areas in the design is to upgrade
hydrogen blistering when one is inspecting the reactors.
the metallurgy of the components to alloy 20 or possibly
Hastelloy C. Improved service life can easily justify the
TRAEGER:
additional associated costs.
Next spring we will be completing our first three year
run on the alky unit. Prior to this, we generally ran two
Question 16.
year cycles. Our practice was to pull bundles and inspect
What changes have been made to improve the life of
reactors every other turnaround. So on a four year inspec-
OW exchangers in trace HF acid service like isostripper
tion cycle, we will most likely pull and inspect reactor
overhead condensers and sidecut trim coolers?
bundles every turnaround on the three year cycle. We
operate two reactors. The unit was built in the 1960s.
The second reactor was installed in 1991. TRAEGER:
In the history of the unit, we have only had one leak In the case of the isostripper overheard condenser, we
in the reactor bundle itself. The only other area in the have a complete exchanger spare bundle and shell. At
reactor where we have ever seen any problems has been turnaround the exchangers are swapped out and dirty
related to the feed distributors. Over time we have lost exchange is retubed in preparation for the next run. The
some metal in the distributors and they are due for replace- main cooling water flow to the isostripper overhead con-
ment this coming turnaround, but they have never caused denser comes from the outlet of the reactor cooling water
us any serious problems. circuit. We have put in a connection to introduce cold
The only other item I might add that is more general cooling water makeup directly to the isostripper overhead
in nature than the HF alky unit is in the past few years condenser to help maintain operating temperatures. I real-
we have made some changes in the specifications for the ize that these steps are mostly treating symptoms and not
metallurgy for all our new equipment in the alkylation necessarily the causes, but it is what we do to get from

1998 NPRA Q & A Session on Refining and Petrochemical Technology 127

TOC/INDEX

turnaround to turnaround without leaks in that overhead range. For general platinum, platinum/rhenium and plati-
condenser and we have been fairly successful at that. num/tin reforming catalysts, no protection is needed for
As mentioned earlier, the main means of eliminating H2S and mercaptans in this range. These are normally
this corrosion is watching your water content in the unit. considered reversible sulfur adsorption poisoning. Proper
As an additional note, the sidecut trim cooler bundle operation of the hydrotreater is the most cost-effective
continues to be a problem for us. At this point, we do not strategy.
have any good answers on how to reduce that corrosion. Thiophene is a little different, as it has a much higher
initial toxicity and may have a higher tendency towards
WALDRON: irreversible sulfur adsorption. If thiophene is truly a prob-
Along that line for our recommendations for chronic lem that can not be handled with normal operation of
problem exchangers are to ensure that the tube roles have the feed hydrotreater, a high temperature guard bed on
been properly reduced. Typical wall reduction should the feed is warranted. (Catalytic Naphtha Reforming,
range from 5-10% of wall thickness. Also ensure that Chapter 10, Antos, Aitani and Parera, Marcel Dekker Inc.)
the bundles are properly handled during cleaning cycles.
Rigging which stresses the tube sheet area should be ALLEN:
avoided. Tube sheet rams to assist bundle at reassertion Most hydrotreated reformer feeds contain only H2S
should be minimized. A thorough plan should be adopted and mercaptans. The H2S should be removed from the
which minimizes overall bundle handling. Although not reformer feed in the naphtha hydrotreater stripper. If the
endorsed by some licensors, seal welding tube ends has stripper is not doing its job, check the reflux ratio and
proven successful for some owner operators. tower temperatures. The mercaptans generally are the
result of recombination reactions in the equipment down-
BENNETT:
stream of the naphtha hydrotreater. They are formed
To pick up on Mr. Waldron’s point, we have changed through the reaction of trace olefins in the product with
our designs so that we strength weld the tubes to the tube H 2 S. Thiophene is more common when processing
sheet metal and that seems to help the matter. cracked feeds. All of these species can be removed onto
guard beds of non-regenerable absorbents—such as the
PURASPEC range from ICI Katalco—who gave me sig-
SCOTT STRATTON (Chevron Products Company): nificant input on the answer. For this combination of
Before we leave the area of alkylation, I was wondering sulfur compounds, the vapor phase sulfur guard may be
if any of the panel members or any of the audience has more effective than the liquid phase sulfur guard.
experienced a problem similar to what we are experienc- Trace thiophene is significantly more expensive to
ing. We have STRACTO double seal contactors and the remove than trace H2S and mercaptans using sulfur guard
problem is characterized by a loud rumbling noise in technology. As well as requiring a nickel based absorbent
the contactor. About two days later we had a contactor it only loads to 1 or 2% wt sulfur on the absorbent.
head failure. Typically H 2 S and RSH can be removed onto other
cheaper absorbent materials, e.g., copper, to levels of up
HIGGINS: to more than 20% wt.
No panel experience. Alternative, the naphtha hydrotreater might be
upgraded. Removal of thiophenic sulfur needs a NiMo
type hydrotreating catalyst operated at moderate pressure.
C. Reforming The exact payback of a project to install vapor or liquid
phase sulfur-guards or upgrade the performance of the
Feed naphtha hydrotreater depends on many factors including
the amount of sulfur present and the sensitivity of the
Question 17. reforming catalyst to sulfur.
What is the most cost-effective strategy to protect cata-
lytic reformer catalyst subjected to trace levels of Thio- JACKSON:
phene, in addition to H 2 S and mercaptans in the At a refinery in the midwest the reformer feed is pro-
naphtha feed? cessed with isomerization unit feed in the naphtha hydro-
treater. Since the LHSV is quite high at 12-15, we always
TURPIN: provide high catalyst activity through selection of excellent
Normally we think of trace measurements being a posi- catalyst, frequent change outs (every 18-24 months), and
tive test less than the first increment of precision on the reactor bed screening to prevent channeling and bypassing
analyzer being used. With today’s analyzers, I think trace of the feed. Feed effluent exchangers are seal welded at
amounts of sulfur now mean in the less than 1⁄4 ppm the tube sheet. There is a water wash of the reactor effluent

128 Light Oil Catalytic Processing

TOC/INDEX

with high quality water, (no oxidizing contaminants such ALLEN:

as oxygen) prevents stripper plugging. Close control of PONA determinations should be made depending
the stripping parameters is maintained to always assure upon the frequency of crude type changes. For reformers
adequate boil up and frequent operator stripper bottoms in refineries where the crude source is constantly changing,
analysis are conducted for sulfur compounds, such as a the PONA should be determined at a minimum of every
doctor test for mercaptans and Drager tubes for H2S. two weeks. A weekly determination would probably be
A sulfur guide bed is utilized between the stripper and better, however. The PONA data can be used together
the naphtha splitter. The naphtha splitter overhead is with the operating conditions and yields to evaluate the
Drager tube tested frequently for H2S to detect any strip- unit performance. In many cases, this is the method used
per or sulfur guard bed bypass problems, feed bottom to fine tune the water chloride balance, to obtain the best
exchanger leaks or sulfur guard bed spending. Sulfur guard yield and selectivity.
bed pressure drops are also checked frequently to detect In almost all cases, the money that you spend on these
for maldistribution, bypassing or vaporization. analyses will be recovered in improved yields. Nearly all
Recycled gas and debutanizer off gas are Drager tube correlations for a reformer require information about the
tested by the operator every round for H2S. Some of the composition of the feedstock. There are other methods
above steps are preventative and some are detection of such as gravity and distillation to approximate the compo-
problems before the problem gets too severe. sition, but this adds another variable to evaluating unit
performance.
MASTRACCI: RASBOLD:
A properly operated reformer naphtha desulfurization The frequency of running PONAs varies at each Sun
unit will remove sulfides (mercaptans) from the reformer location, from once per day to once per week. However,
feed. Any remaining trace levels of sulfides can be removed the utilization of this data is consistent. All locations use
in the liquid phase by a high nickel catalyst guard bed. this information as an indicator for feed quality, and
This bed, however, will not remove thiophene. To some therefore as a tool for establishing unit octane response
extent depending on the cut point used and the refinery and catalyst end-of-run predictions. No daily operating
objectives, thiophene can be controlled by distillation at conditions on the unit are set by these numbers at any
the naphtha HDS stabilizer since it boils at 183°F. of our locations.

SMITH:
SONI OYEKAN (Amoco Oil Corporation):
At Valero, we run PONAs three times a week on the
Ms. Allen referred to the fact that vapor phase systems combined reformer feed and use it to track catalyst activ-
can do this job in terms of removing thiophenic com- ity. LP models contain vectors to adjust yields and octane
pounds. Vapor phase sulfur guard systems have been used based on feed N plus A’s. LP models are also used to
successfully with bimetallic catalysts, such as platinum develop monthly operating plans and to evaluate refinery
rhenium types, and including high rhenium catalysts. feedstocks. We do not watch them on a daily basis. We
Basically, this system is installed with the reactors in the set a target RON and then maximize the rate to the CCR
reformer loop and essentially gets the thiophene out by limits of our reformer.
converting thiophenic sulfur compounds to H2S which
is then removed from the recycle gas. In catalytic reformers TURPIN:
using vapor phase sulfur guards, we consistently determine The typical frequency for running component analysis
that the sulfur, and in fact the H2S, in the recycle is of the feed and reformate is once per week for unit moni-
negligible. Often the recycle gas hydrogen sulfide is close toring for those units that are being monitored by a process
to zero. engineer. There are lots of units that nobody is watching
A paper (AM-89-45) by Ronald McClung, Rino from a catalyst performance perspective, and those units
Kramer and Soni Oyekan was presented at the 1989 have rather infrequent feed and product composition anal-
NPRA Annual Meeting. That paper compared liquid and ysis performed.
vapor phase sulfur removal processes. For a copy of the Typical reformer monitoring is for shifts in catalyst
paper, please contact the Acreon Company in Hous- selectivity and catalyst deactivation. The short term advan-
ton, Texas. tage to unit monitoring is to insure the reactions that are
occurring are those reactions you intend to incur. The
Question 18. long-range objective in unit reformer monitoring is to
What is the typical frequency for running PONAs on maximize application of the catalyst.
reformer feed? How are the results used to improve In refineries where advanced process controllers and
operations/economics? on-line optimizers have been installed, and where feed

1998 NPRA Q & A Session on Refining and Petrochemical Technology 129

TOC/INDEX

changes are frequent, feed analyses are being done several required installation of high capacity trays but it was
times a week. re-used.

Process BENNETT:
We have done this 28 times, 14 times on units that
Question 19. had side-by-side reactors and 14 times on units that had
In replacing an existing semiregen reformer with a CCR, stacked reactors. In most cases, we try, as the other panel-
has anyone seen good economics for using parts of the ists pointed out, to make use of the existing equipment.
old equipment in the construction of a modern CCR? If Generally, the existing equipment that can be reused are
so, which parts of the old unit can be incorporated into the recycle compressor, the product separator and the
the new unit economically? debutanizer and the existing reactor stack.

WALDRON:
LAMB:
We have reused several major parts in unit conversions.
Raytheon Engineers and Constructors worked on two
The economics depend upon the operating conditions of
units for Mobil and agree with what Mr. Waldron has
the old unit versus the new unit. You should evaluate
said. The other disadvantage of using part of the existing
reusing the feed system, feed effluent heat exchangers,
unit is typically the optimum pressure level you end up
heaters, recycle compressor, off gas compressor and prod-
with is different than what it would be for an all new unit.
uct stabilizers.
So typically, the result is going to be maybe somewhere
Advantages include less site preparation, reuse of exist-
between a 200-pound semiregenerative operation that you
ing utilities and infrastructure. Disadvantages include
may have today compared to a CCR platformer operation
more complex turnaround planning and possibly longer
of possibly 40 to 50 psig.
unit downtimes.
Question 20.
TURPIN: Has anyone had experience with the new semiregen or
The off-site equipment; storage tanks, feed lines and CCR catalysts specifically formulated to increase aromat-
pumps, and rundown equipment can all be used. Cold ics production? Has experience matched the vendor’s
side equipment such as the feed pump and control valve yield prediction?
may be useable as well as down stream fractionation equip-
ment. Depending on recent possible upgrades that have
occurred to the semiregen unit, the DCS system and on- TURPIN:
stream analyzers may be reuseable. Several utility systems I think the question refers to the catalyst that is formu-
as well as electrical switch gear may also be re-useable. lated to increase BTX production. The total aromatics
yield does not shift significantly, the shift is towards a
higher yield of the lighter aromatics. I have seen this type
ALLEN:
of catalyst used in several plants, and the incremental
In going from an old, high pressure semiregenerative
benzene yield has been substantially, (a few percent of
reformer to a modern, low pressure CCR reformer, there
feed) more than what I would have expected over a con-
is normally little of the old unit that is worth salvaging.
ventional motor fuels catalyst. I do not know of any
The reactors will need to be changed, the recycle compres-
refiner that did not eventually make the vender’s yields
sor replaced and the feed/effluent heat exchangers also
predictions.
replaced. In addition, the furnaces will have to be
upgraded, if possible, to compensate for the higher reactor
temperature. ALLEN:
However, our partners at Fluor Daniel have seen one Criterion recently introduced our PR9 and PR29 cata-
revamp of a semiregen reformer to a CCR where a fair lysts which were specifically formulated to suppress crack-
amount of the equipment was re-used. The revamp basis ing which reduces the overall hydrogen yield and the
was to double the feed rate to the unit. Three of the four C5Ⳮ plus yield. This formulation also protects against
fired heaters were revamped with new tubes and burners dealkylation of higher aromatics which occurs predomi-
and only one new heater was required. The lower hydro- nately in the last reactor where the average reactor temper-
gen-to-oil mole ratio associated with low pressure reform- atures are higher.
ing operations allowed the recycle compressor and turbine We have seen for the aromatics production there has
driver to be revamped. The product separator was re-used been an increase of about a half a percent in the xylene
with minor modifications. Due to the higher liquid yield production and an overall yield increase of about 1 vol%.
of the lower pressure reformer, the debutanizer column That is about 1.2 wt%.

130 Light Oil Catalytic Processing

TOC/INDEX

BENNETT: RASBOLD:
In the semiregen area, our R-72 stage loading system A very large coke and catalyst ball (approximately 800
is now loaded in more than 30 units and generally gets cubic feet in size) developed in the last reactor of a semire-
about 1% C5Ⳮ liquid yield benefit over the previous gen reformer at our Philadelphia refinery. The coke ball
operations. In CCR PlatformingSM units, our R-174 cata- developed during an operational period of only six weeks.
lyst has been in commercial operation more than two The formation of this ball in so short a time period was
years. That catalyst also shows about 1% higher C5Ⳮ likely due to a combination of events; 1) low liquid hourly
liquid yield than the previous catalyst. space velocity of 0.5 to 0.6 which is somewhat lower than
our normal, 2) possible flow distribution problems which
ABDULLAH AL-MALEKI (Saudi Aramco): was evidenced by fines production and higher than normal
Has anybody had experience in achieving 100% naph- reactor pressure drops, and 3) low sulfur feed that could
thene conversion in a regenerator platform unit? promote metal catalyzed coking. Our unifined naphtha
was running .05 ppm sulfur before the sulfur guard.
TURPIN: Therefore, the reformer feed contained little if any sulfur.
Because of the equilibrium reaction rates, it is not There is actually no way of saying how much each
possible to get to 100% naphthene conversion. factor contributed. We did observe filamentous coke in
the ball with electron microscopy. We do agree that fila-
mentous coke is due to metal catalyzed coking. Since low
ABDULLAH AL-MALEKI (Saudi Aramco): sulfur in the feed could lead to metal catalyzed coking,
What is the maximum conversion of naphthene? we have discontinued the use of our sulfur guard bed
prior to the reformer.
TURPIN:
The maximum naphthene conversion is going to vary Question 22.
depending on the pressure, temperature and space velocity What are the typical catalyst attrition rates in a CCR unit
of the reformer unit. Typically on a 400 psig unit, you and what are the causes of high attrition? What are
might get as low as 1.5% naphthene remaining. If you typical platinum losses during maintenance activities
go down to 200 psig, you might get in the ballpark of and catalyst reclamation?
1% remaining naphthenes. At 0.75% to 1%, you run into
equilibrium considerations and can not go any further. WALDRON:
Typical attrition rates vary between 0.5 and .1 wt%
JOHN COOK (Acreon Catalysts): of the catalyst circulation rate. This rate tends to increase
Acreon Catalysts has two different catalytic solutions with catalyst age. Several problems can cause high attrition
that can remove the excess sulfur compounds that slip rate due to catalyst impact or grinding. They include too
through the naphtha hydrotreater. If the amount of thio- high or too low of lift gas velocities, rough edges in the
phenes is limited to trace amounts, we suggest using our catalyst circulation loop, mechanical damage to reactor
liquid phase sulfurguard catalyst (D-1275). If the quantity internals, high pressure drop across catalyst transfer lines
of thiophenes is over a trace, we suggest our vapor phase or high pressure drop across the reactor. Also, regenerator
sulfurguard catalyst (ACG-369). temperature runaway can weaken a catalyst and thus
Acreon Catalyst is making IFP AR-405 CCR catalyst. increase fines make.
This catalyst was developed to increase aromatic produc- Platinum remains on the catalyst. So platinum losses
tion. The places where this catalyst has been installed during maintenance should depend on care and good
have met operator expectations. housekeeping to avoid catalyst loss. Catalyst reclamation
should recover at least 95% of the platinum on used
Question 21. catalyst. Of course, this depends upon high quality analyti-
Is it possible to get the sulfur content of a reformer feed cal data for good platinum accounting.
low enough to have carburization (growth of filament
type carbon) in the unit? SMITH:
On Valero’s CCR reformer, our first catalyst load ran
WALDRON: for five years and did very well. Then on the next genera-
We have experienced filament type coke in a CCR tion catalyst, we have had a problem with fines. This was
reactor operating at less than .15 ppm sulfur in pretreated loaded in 1996. When we loaded the catalyst it appeared
feed. The coke growth occurred on the vessel walls. Over ‘‘dustier’’ than it did before, but that is supposition.
time this coke growth can cause scallop and other reactor We have had a fines problem, but it may be self-
internal damage and can result in a significant impact on induced. In 1996 we changed the gap on our number
reactor performance. one lift engager, cutting it down to reach a high enough

1998 NPRA Q & A Session on Refining and Petrochemical Technology 131

TOC/INDEX

velocity to lift the catalyst. We also changed the diameter ALLEN:

of the lift pipe from 2.5 to 3 inches. Right now the CCR product gas and liquid streams can be extremely
velocity is 2.5 times what it was, around the outer anullus. corrosive, particularly as the reforming catalyst ages and
At the bottom of number four reactor, an isolation chloride slippage increases. Therefore, it is highly recom-
valve, which was pinched down on for a period of time, mended that chloride removal systems using beds of non-
may have added to the fines production. regenerable absorbants be specified for these streams.
On the regenerator side, we may have had incorrect I had some significant input from Synetex on this
O2 concentrations in the burn zone. We ran the O2 as answer as follows. Traditionally, the absorbent of choice
high 1.6%. We did now see high temperatures per se in the was alumina based because such materials are cheap on
regenerator however, we did see holes in the inner screen. a dollar per pound absorbent basis. In recent years, newer
We may be burning at elevated temperatures in the generation products, which are based on chemical absorb-
chlorine zone, but this may be caused by the screen plug- ants, such as the pure respect range from Synetex, have
ging with poor distribution of air in that area. been used more widely for two reasons.
As far as platinum losses, we went to great lengths First, their higher capacity means longer guard bed
about a year ago trying to define platinum loss, but trying lives and lower costs per pound of chloride removed.
to track this accurately is difficult, even with good proce- Secondly, because they are generally alkyline in nature
dures for handling the fines. We did notice though, that and chemically react with the chloride in the process
fines tended to have a higher platinum concentration than stream; they do not create an acidic surface.
the hole pills. Experience has shown that such acidic surfaces as can
be found with the alumina products promote the forma-
ALLEN: tion of organic chlorides in the process stream by reaction
Most of the causes of high attrition have been covered. of HCL with trace olefins. These organic chlorides are
I want to share what we have seen is typical with reported not removed by the guard bed and therefore slip into
platinum losses. Comparing fresh and reclaimed catalyst, the exit stream. For net gas streams, this means organic
we usually see about 8-10% loss. But I did want to make chlorides pass to the downstream units such as hydrotreat-
sure that people remember that reforming catalysts pick up ers where they are reconverted to HCL. Then, of course,
water. Make sure that, for an apples-to-apples comparison, all of the problems associated with ammonium chloride
the weights are compared on a dry basis so that you do formation and fouling can occur.
not make the mistake of thinking that water converts The phenomenon of organic chloride synthesis across
to platinum. alumina guard beds has only become apparent in recent
BENNETT: years with the development of the necessary analytical
With the wide variety of designs in operation, we have techniques. A more complete look at this subject is pro-
seen a decreasing number of fines as we changed the vided in the 1998 NPRA Annual Meeting paper AM-98-
regenerator design. The CycleMax娃 regeneration section 22 entitled, ‘‘Third Generation High Capacity Chloride
design is the third generation, and some new units of this Removal from Catalyst Reformer Product Streams’’.
generation experience very low fines make with an average Chloride removal guard beds can be and are used to
of 0.02 wt% of the catalyst circulation rate. As you heard protect stabilizer columns from corrosion. Guard beds are
from the other panelists, the commercial reality changes simple to operate and need little attention. By contrast,
from unit to unit, based on the circumstances of the injection of chemical additives or neutralizer needs closer
individual unit. attention and analysis to avoid excessive chemical costs.
In terms of platinum recovery, generally our experience The exact cost effectiveness will depend on the specific
with the reclaimers is they will guarantee 99.0-99.5% location. The absorbent vendor should be able to advise
recovery of the spent catalyst assay. on operating costs and vessel installation costs.
Concerning housekeeping issues with the CCR Plat-
formingSM units, you have to be a little more diligent than SMITH:
you do on a semiregen unit because of the fines, which We have no liquid stream going through the chloride
have to be both analyzed and assayed correctly and then adsorbent. We only have a gas stream and we have no
recovered. Lastly, when there are reactor section dump- stabilizer corrosion problems. That is probably because,
ings, no matter what type unit, the flow bin integrity gets we do not have a stabilizer downstream, we have a
to be a pretty serious issue. PSA unit.
Question 23. However, just as a note, we use as a calcium oxide
What has been the experience with various chloride zinc blend of adsorbent. We chose this over the normal
adsorbents on CCR net liquid and gas streams? Is this alumina based adsorbent because of less spooge produc-
a cost effective method to address stabilizer corrosion? tion potential. We have cleaned up the adsorbent by

132 Light Oil Catalytic Processing

TOC/INDEX

steaming it in place and we have done that three or four TURPIN:

times with success. Typically, it is done with Draeger or some other type
of gas tubes. I am sure you can buy an on-line analyzer
HERBERT W. WIZIG (Merichem Chemicals & Refining Services): if you wanted to. The critical thing is to determine the
Merichem recommends treating the reformer stabilizer compound you are looking for and use an appropriate
feed with an alkyline water wash to remove both hydro- test technique.
chloric acid and ammonium chloride down to nondetect-
able amounts. HIGGINS:
What are others doing for end of life activated alumina
ANGELO FURFARO (UOP): in chloride scrubbers for reformer hydrogen?
Net Gas Streams: There is a wide variation in the
performance of various adsorbent products (aluminas and
non-aluminas) in net gas treating for trace HCl removal. RADCLIFFE:
HCl capacity testing alone cannot be used to predict I would imagine it is possible to recycle the alumina
commercial performance of the products. By-product for- from the absorber to an aluminium plant. Certainly that
mation from the chloride loaded adsorbent and fouling is the case with fluorine absorbers. So I would imagine
can have a significant impact on commercial operation. that they can be sent back to an aluminium refinery.
UOP’s newest product for this service, Molsiv 9139A, has
been specifically developed to have low catalytic activity to WALDRON:
form by-products and maximum chloride capacity. In I think maybe it is not exactly the answer to that, but the
side by side testing in a commercial CCR Platforming problem we have seen coming out of alumina absorbers on
unit net gas stream, the chloride capacity of Molsiv 9139A off gas is that sometimes they will trick you with organic
was twice that of a competitive promoted alumina product chlorides coming out. You think you doing well on the
and over three times that of a competitive non-alumina beds, but somehow they are producing organic chlorides,
product. which go on downstream and create problems.
LPG Streams: Similar to net gas streams, treatment of
LPG streams for trace HCl removal requires an adsorbent ANGELO P. FURFARO (UOP LLC):
that has minimal catalytic activity to form by-products A few refiners have reported success using Gas Tech
when chloride loaded. UOP has developed a promoted tubes to determine the presence of chlorides downstream
alumina, Molsiv CLR-011, specifically for this LPG ser- of the alumina bed where they did not have such success
vice. with the Draeger tubes. You might try that as an inexpen-
Reformate Streams: Stabilized reformate typically does sive alternative to determine when the alumina is actually
not contain sufficient chlorides to require treatment for spent and organics are starting to break through.
chloride removal.
Stabilizer Corrosion: Corrosion is typically not a prob-
lem in a reforming unit stabilizer unless there is a external DAVID L. SMITH (Alcoa Industrial Chemicals):
stream containing water that is also being fed to the Another inexpensive detection method for the organic
column. Corrosion in downstream units processing the chlorides are Kitagawa tubes out of Japan. Refiners have
stabilizer off gas or net overhead liquid (LPG) due to found that, just as Mr. Furfaro indicated, the Dräger tubes
trace levels of HCl can be a problem if these streams are do not properly measure organic chlorides. Sometimes the
mixed with streams that contain water. This problem can gas tech tubes do and sometimes they do not. The Kita-
be resolved by treating the stabilizer off gas stream and gawa tubes have provided reproducible results.
net overhead liquid stream to remove HCl. Various CCR units around the world using the alumina
Fouling of the top trays and the overhead condenser based selective adsorbents for chloride removal have found
of a stabilizer column with ammonium chloride salts can in approximately the last 25% of its effective life is where
be a problem if there is excessive nitrogen in the feed to you see the upsurge in organic chloride and gum polymer
the reforming unit. Again, corrosion problems do not (‘‘spooge’’) formation. If you do a premature recharge,
normally result from this fouling unless there is an external you can minimize those adverse effects.
stream containing water that is also being fed to the
column. Mechanical
HIGGINS: Question 24.
How do you monitor organic chlorides in a hydrogen Has anyone experienced upward movement of the center-
system? What techniques and equipment can be used for pipe in a cold wall reactor? What steps have helped in
online indication? minimizing the problem?

1998 NPRA Q & A Session on Refining and Petrochemical Technology 133

TOC/INDEX

BENNETT: JACKSON:
There have been several reports of the upward move- Raised Face Flanges with spiral wound gaskets are more
ments of centerpipes in cold wall reactors. The apparent commonly used (since about 1980) to replace RTJ flanges.
cause is numerous thermal cycles of the metal liner in RTJs are an expensive alternative, but have a good perfor-
conjunction with the catalyst bed pressures exerted caused mance record. The sealing ring in these joints is often a
by thermal expansion. The catalyst bed basically holds hardened 17-4 pH stainless steel to ensure a tight seal
the centerpipe. The liner moves down and the centerpipe against the hydrogen.
stays where it is. So it essentially walks up. Typically, these rings are replaced whenever it is neces-
The satisfactory engineering solution to prevent this sary to separate the flanges. Some success has been experi-
in a cold wall reactor has not been demonstrated to our enced using Ultrasonics to detect groove ring cracking.
knowledge. In some cases, these cold wall reactors have Acoustic Emission is also another technique that can be
been replaced with hot wall reactors, partially in order to successfully used to screen RTJ flanges for possible areas
resolve the centerpipe movement problem and at the same of concern. Once a particular concern has been identified,
time to improve vapor distribution within the reactor. then the joint needs to be opened and Dye-Penetrant
In order to minimize catalyst containment loss caused tested for cracks.
by centerpipe movement, the seal of graded alumina balls
can be used at the bottom of the centerpipe to keep the GENTRY:
catalyst from going through. RTJ cracking has not been totally isolated to chloride
Another way you can get some sense of what is going or polythionic acid gas cracking, but stress cracking related
on is to put a position indicator at the top of the centerpipe to the RTJ design. Most of the problems that we encoun-
through the coverdeck and record the distance below the tered were related to the design flanges that fell outside
manway in the original position. Now when you open the normal API standard flanges, which stopped at about
the reactor, you can check the distance to monitor the 12 inch, 1,500 psi. Most of these flanges were usually
relative elevation change of the centerpipe. associated with vessels.
It seems that during the design phase of the vessel, the
TURPIN: manufacturer minimized the material bulk of the flange,
The upward movement of the center-pipe occurs in i.e., the pipe wall inside the ring diameter. Thus, cracking
both hot and cold wall reactors. The problem occurs when was always from the ring groove to the inside of the pipe.
the reactor is cooled too quickly and the center-pipe is This type of failure should not cause a catastrophic
not thermally contracting at the same rate as the exterior failure, but usually a leak. Inspection and detection had
walls and catalyst bed. If the system is cooled too quickly, been limited to turnarounds on these flanges, which were
and the center-pipe is the last piece of the system to cool usually open for maintenance during catalyst changes.
and thus contract, the center-pipe hangs in the process As an alternative, raised face flanges can be used, but
and results in an upward movement. are usually more susceptible to leaks during temperature
swings on the flanges or the reactors. All this actually
J.B. RODDEY (Roddey Engineering Services):
leads to is the conclusion that either type flanges can be
There has been a number of cases of cold wall reactors
successful and damage on ring type joints can usually be
with the centerpipe lifting up. There is another mecha-
eliminated or at least minimized if the correct assembly
nism that can cause this other than the one mentioned
and torquing techniques are used.
by Mr. Bennett. That is caused by flow behind the liner
or a crack in the liner and warping the liner.
TURPIN:
Another incident that happened mostly in Europe, but
Stress corrosion cracking from chlorides and a special
some places in the USA, is that some material in the
form of stress corrosion cracking from polythionic acid
refractory caused coking which caused a buildup at the
were quite common in reformers years ago when austenitic
bottom under the centerpipe. This physically lifted the
stainless steels were used. I suspect that if the unit in
centerpipe.
question has RTJ flanges, it is probably an older unit
Question 25. and this may be the source of the question. Suitable
In the majority of hydrocracker and reformer reactor cir- precautions must be taken during shutdowns to avoid
cuit piping systems, stainless steel RTJ flanges are being these problems.
used. But these RTJ flanges are prone to cracking due to If the RTJ flanges themselves are cracking, going to
chloride corrosion or polythionic acid attack. Generally, raised face flanges is not going to solve the problems.
these flanges are not subjected to inspection at every The problem is most likely solved in changing operating
shutdown. In such cases, what are industry’s experiences procedures or in the correct application of metallurgy.
and practices to inspect and detect these cracks? Can In the temperature and pressure range of most current
these flanges be substituted by raised face flanges? reformer operations, a piping specification can be applied

134 Light Oil Catalytic Processing

TOC/INDEX

that allows raised faced flanges with the appropriate gas- downstream of this exchanger. In many cases this may
kets. This is not true in the case of hydrocrackers. In be the cause of your fouling problems.
either case, I would think it would be terribly expensive
to replace RTJ flanges with raised face flanges. BENNETT:
As Mr. Jackson pointed out, this is not an insignificant
task. Fortunately, most of these vertical exchangers now
WALDRON:
are more in CCR PlatformingSM application where they
We have seen cracking in RTJ flanges and rings in all
do not foul as readily and where the feed is on the tube
different kinds of materials. In general, we use raised
side. Therefore if there is a feed (or tube side) problem,
face flanges up to class 1500 with some gasket problems.
it can be cleaned in situ.
Another thing we have used with some success is the
If you are considering pulling a bundle in the vertical
clamp type connectors similar to gray lock.
exchanger, you need to get the vendor involved and proba-
bly also the licensor. You clearly need to look at the
Question 26.
supporting structure as well and the nearby plot area for
How, and how often are refiners cleaning the tube and
a place to ‘‘rest’’ the exchanger.
shell sides of their vertical reformer feed/effluent
exchangers?
RASBOLD:
Our vertical feed/effluent exchangers in our semiregen
JACKSON: motor reformer unit in Tulsa were installed in the mid-
Generally, refiners try to avoid cleaning these vertical 70s. They were never pulled or cleaned during the first
exchangers. Some refiners have tried to chemically clean ten years of service. Subsequent problems with internal
these exchangers using potassium permanganate with little leaks stemming from acoustical vibration caused these
or no success. In some situations, after chemically clean- bundles to be pulled four times for various repair and
ing, heat transfer worsened. Typically these units are not redesign applications over the following ten years.
pulled to be cleaned at the exchanger bundle rack. This involves removing the entire assembly from the
At least one refiner cleans their exchangers by dropping structure via a crane, putting it on the back of a tractor
the bottom tube sheet and plate, and then blasting with trailer rig and taking it to the exchanger manufacturer.
10,000 psig water, using a flexible wand, up through the During the inspections of these two bundles, we found
bottom of the tubes. no evidence of substantial scaling deposits within the tubes
I do want to relate one particular successful situation. or the shells.
The exchanger was removed after a run length of approxi-
mately 15 years. A steel structure to hold the bundles was SMITH:
designed by the engineering department. The bundles At Valero, we have cleaned the tube side of our CCR
were lifted and set down onto the structure. Scaffolding reformer feed/effluent exchanger twice since coming
was built around the steel structure with levels every ten online in 1992 (in 1995 and 1996). We do plan to clean
feet or so. A 3/16⬙ hose was used with 7,000 psig water again in January. It should be noted that we shutdown
to go completely through each of the tubes which were in 1996 to clean these exchangers because of the lost heat
approximately 65 feet in length. After some of the tubes exchange efficiency.
had been cleaned, a Boroscope was used to insure that Our tube side is the feed stream. When it was cleaned,
the tubes were getting cleaned properly. On a final inspec- we found material on the tubes which was water soluble.
tion, the tubes were filled with water and some were still I believe UOP is working diligently on identifying the
holding water. At that time, a 1/4⬙ hose was used with material. Several tube roll leaks were also found during
7,000 psig water to go back through the tubes once again a nitrogen tightness test. The tubes were re-rolled.
to clean the ID. All the tubes were successfully cleared We have no chemical injection upstream in our NHT
and held no water, except for two out of the 5,200 tubes. at our debutanizer. The reason I bring that up is that I
The tube OD was cleaned using a short wand approxi- understand some claim that chemical injection at this
mately 24⬙ in length and hydro-blasted several times again point may contribute to the fouling on the feed tube side.
using 7,000 psig water.
Results indicated the temperature profile was greatly TRAEGER:
improved, with an overall recovery of 20% of the We have a couple of sets of these bundles that were
exchanger duty. Other benefits obtained resulted in record installed in the early 1970s, two vertical bundles in our
throughputs on the unit while using less fuel gas in the semiregen reformer. They have never been pulled since
charge heater. then. We have cleaned them a couple of times on the
As a general comment, investigate your regeneration tube sides. We pulled the top head and water washed
procedures and make sure you are adding the chloride from above with the wand. A couple of times during the

1998 NPRA Q & A Session on Refining and Petrochemical Technology 135

TOC/INDEX

lifetime of the bundles, we have experienced a heavy tar BENNETT:

like buildup on the shell sides on the bottom of both We just surveyed our licensees and 60 of them
exchangers. These were both cleaned by circulating a hot responded. The average period of time between turn-
kerosene through the shell sides of the exchangers. There arounds was 3.2 years. That included about a dozen units
is a good discussion of this in the 1997 NPRA Q&A Tran- in the Far East that have to shutdown every year for
script. regulatory requirements. Ten percent of the units
exceeded five year cycles between turnarounds. We are
WALEED A. SALAMAH (Saudi Aramco): getting to the point now on these units with the ability
Regarding these exchangers, did you consider them as to change the catalyst on the fly, that the PlatformingSM
pressure vessels, which have to be inspected at certain unit is generally turned around on some other mainte-
intervals for cracks on the shell side? nance schedule.

BENNETT: GENTRY:
I believe it is our experience that we have been able We have experienced turnaround intervals ranging
to convince people it is not a pressure vessel and does from 3 to 5.5 years. In our experience, we have had more
not need to be treated accordingly. problems associated with our naphtha hydrotreater versus
the CCR. Problems with the naphtha hydrotreater include
TRAEGER: exchangers, heaters and compressor fouling. Factors asso-
We have pressure tested them with nitrogen, but have ciated with the CCR, which have driven turnaround inter-
not pulled them to inspect them. vals, are usually heater repairs or reactor issues.
ANGELO P. FURFARO (UOP LLC):
There is also a discussion in the 1995 and 1996 Q&A SMITH:
Transcripts on this very subject which are worth reviewing. As stated earlier, our first run went five years before
our first turnaround on our reformer CCR. We expect
V. K. KAPOOR (Indian Institute of Petroleum):
3-4 year cycle life on the catalyst.
Catalytic Reforming is a clean operation and fouling Operating cycles are not only set by the catalyst cycle
problems are not frequent in vertical reformer feed/efflu- life, but also the catalyst condition, (i.e., surface area, the
ent exchanger. It may however be observed due to deposi- activity, etc.). Other reasons for shutdowns are fouling
tion of organic or inorganic materials either on the feed of vertical exchangers, recycle compressor fouling and
side in the tube or effluent side in the shell. The feed fines pluggage in the CCR.
side fouling might be due to iron scales, corrosion inhibi-
tors, the feed processed from the unblanketed tank or D. Aromatic Extraction
caustic foam entrained during regeneration. Inhibitor may
pass from stripper to reformer, combine with corrosion
products and deposit at specific locations in vertical Question 28.
exchanger where liquid vaporization takes place. Plugged Does anyone have experience with using O2 scavengers
tubes can be cleaned from top by lancer and high pressure in solvent extraction processes (e.g. UDEX units)? If yes,
water blasting. Shell side fouling could be due to polynu- are they hydrocarbon based scavengers or amine based
clear aromatics. Attempts may be made to clean it by scavengers, usually used in BFW treatment? Are oxygen
circulation of reformed naphtha with a temporary arrange- stripping or feed segregation more cost effective solu-
ment for circulation. To pull and reinstall the bundles is tions?
a very difficult job. Horizontal placement of the pulled
tubes bundle could cause tube to tube sheet leakage. I
am aware of a case, where a refiner has not pulled out ZETLMEISL:
bundles after 14 hours of operation and has followed above We have had considerable success using oxygen scaven-
discussed cleaning procedure. The opportunity could be gers in furfural units. The question refers to UDEX, but
at the time of turnaround or regeneration and interval of the same principles should apply. We use proprietary
about 2 to 3 years. formulations of hydrocarbon based oxygen scavengers.
These chemicals are more akin to fuel stability antioxi-
Question 27. dants than to boiler feed water chemicals. Since UDEX
How long have refiners operated CCR reformers between uses glycol type solvents, UDEX might be more tolerant
maintenance shutdowns? What factors are considered in to water formulations, but I have no direct experience.
setting operating cycles? What operating and mechanical Excluding the oxygen ingress in the first place is, of course,
problems can cause units to be shutdown? the preferred course of action.

136 Light Oil Catalytic Processing

TOC/INDEX

BENNETT: drastic reduction in plant throughput. Particularly severe

Oxygen scavengers might be suitable for dissolved oxy- pitting and stress corrosion of the reboiler tube bundle
gen in the feed, but the oxygen that enters through leaks in the Solvent Recovery Column (SRC) of the unit was
in the vacuum system is likely to degrade the solvent before noticed which led to leakage of heating oil into the system.
it comes in contact with traces of the oxygen scavenger. A comprehensive analysis of all streams connected with
We have had a bad experience with boiler treatment this unit was undertaken. These investigations revealed
type oxygen scavengers in at least one operation. We that the solvent contains appreciable concentration of
question whether the scavengers will be effective against chloride (150-190 ppm) and total acidity (0.1-0.2 mg
oxygenated compounds such as peroxides that decompose KOH/g) along with suspended solids and gummy material
at the conditions in the process. The best approach is, as (0.15 wt%). The high level of total acidity (indicative
Mr. Zetlmeisl said, a vigorous program to eliminate the of oxidative degradation arising from O2 ingress and/or
oxygen ingress. thermal degradation of sulpholane) remained even after
the 5 t/a solvent regenerator was placed on stream.
Question 29. The presence of significant levels of chloride was sur-
Discuss/review experience with extraction unit corrosion prising since chloride content of feed was controlled
sources other than oxygen intrusion. Specifically address within specification limits of 0.1 ppm. The source of
chloride attack. Does anyone process their extraction chloride contamination was traced to seawater seepage in
unit water to remove corrosion products, such as salts, the overhead condensers of SRC column. Chloride ion
chlorides, etc.? is known to cause pitting corrosion, which is indicated
by breakdown of passivity. This is believed to arise by
WALDRON: either of three possible mechanisms: adsorption of chlo-
Corrosion is a common problem on both UDEX and ride ions, penetration of chloride ions or mechanical break
Sulfolane units. Problems are most severe in the, lean down of passivating film. Chloride ion contamination
solvent reboiler tubes. Upgraded metallurgy can slow the appears to be the prime cause of pitting corrosion observed
reboiler corrosion, but you must ultimately control the in the reboiler tube bundles.
contaminants that cause the problems. The two main An adsorption based solvent clean-up system has been
sources of contamination are oxygen from ingress and developed. This process uses a proprietary adsorbent and
chloride typically from the reformate feed. Chlorides often can reduce chloride to less than 1 ppm and total acidity
get the most attention because they are easy to measure. less than 0.05 mg KOH/g. A skid-mounted unit with in-
However, as mentioned before, oxygen is vital as well. situ regeneration of adsorbent, is under design and is
To talk specifically about chlorides, reformate contains expected to be commissioned by mid 1999.
trace chlorides anywhere from .5 to 3 ppm. Variation is
due to reformer operation, catalyst age, chloride guard Question 30.
bed age, etc. What are typical energy consumptions, per pound of
Chlorides build up in the extraction unit for two rea- extract produced, for the various solvent extraction sys-
sons. First, the circulating solvent has a high concentration tems?
of water and efficiently extracts the chlorides from the
feed. Second, the solvent water is recirculated in near TURPIN:
infinite closed loop. The standard designs only allow In addition to the different energy costs between aro-
minuscule purge. However, chlorides, organic acids and matics extraction units, several process differences also
other byproducts can be removed by various means of exist. These differences include; percent aromatics that
solvent regeneration and water rejection. are recovered, solvent costs, capital costs of a unit, purity
of recovered aromatics, royalty fees, corrosiveness, solvent
BENNETT: thermal stability and toxicity of the solvent.
The only thing I have to add to that is that we are The total of nine different extraction type processes
aware of one UDEX licensee that put their own patent are shown on the following table. You can see that for
ionic exchange system on the circulating water to improve the first four, the steam usage is in the same ballpark.
the situation. Electricity for the Distapex process is significantly lower.
This is a selective process where you select one particular
M. K. KHANNA (Indian Institute of Petroleum): aromatic that you want to recover. I do not have cooling
An Aromatic Recovery Unit (ARU) processing catalytic water requirements on the Morphylane or Octenar pro-
reformate in an Indian refinery has been facing corrosion cesses.
and fouling problems. The extraction solvent used is sul- On the third table, there are a couple of interesting
pholane. The extractor is a sieve tray column and fouling things to point out. First, the unit labeled UDEX is using
of the sieve trays has resulted in unstable operation and the original solvent MEG rather than what you typically

1998 NPRA Q & A Session on Refining and Petrochemical Technology 137

TOC/INDEX

see in a revamped unit which uses DEG. The other thing BRIAN JOHNSON (UOP):
to look at is the cooling water numbers for the Sulfolane Following are some operating guidelines we would
process. The cooling water numbers are not reported on recommend:
a consistent basis. For example, on the Sulfolane unit, A. Increase LHSV—Operating fresh catalyst at low
the amount of cooling water is what you use if you have space velocity can increase the quantity of heavier poly-
a very high use of fin fans. mers in the poly gasoline so most refiners try to maintain
Energy Consumptions For Various Solvent Extraction Systems
the space velocity of the reactor within design range as
soon as possible after a restart. A baseline is established
Process Steam Electricity Cooling Water from which future runs can be based on. The key is to
*/mlb Arom. lb/ kWh/ gal/ keep the most active catalyst at the design space velocity
(and in certain situations much higher) and to slowly
Arosolvan 800 4.3 2500 decrease the LHSV as the catalyst ages. The lower the
Carom 700 4.6 3800 space velocity, the higher the residence time for the hydro-
carbon in the reactor, and hence the possibility for longer
Distapex 900 2.5 3100 chained olefins to form.
Morphylane 900 4.5 B. Decrease Reactor Temperatures—Reducing reac-
tor outlet temperature will result in improved selectivity
Octenar 800 3.5 to lower molecular weight polymers and less cracking
Sulfolane 820 4.7 400 of heavies. It should also be noted that cooler outlet
temperatures can also result in cooler sections of the cata-
Tetra 700 4.7 4000 lyst bed being less productive and being less fully utilized,
Tetra II 700 3.9 3100 therefore may reduce the conversion per pass.
C. Decrease Olefin Concentration to the Reactor(s)—
Udex 1300 7.6 900 Reducing olefin concentration in the combined feed by
increasing paraffin recycle is an alternative, but fraction-
ation restrictions may be a constraint.
MASTRACCI:
D. Minimize Conversion—The unit should be oper-
The New Jersey refinery processes a nominal 55% ated based on the minimum conversion that is acceptable.
aromatic BTX reformate in the sulfolane unit. Average Operating the reactors at high severity as indicated by
energy consumption for this unit, which has an external high conversion causes higher molecular weight oligomers
solvent regenerator, is 630 BTU/lb of extract produced.
to be formed.
The majority of the energy is steam usage and electrical
Another approach is to load the reactor(s) with catalyst
consumption is considered negligible.
that has more void space between the pellets and allows
for more room for coke growth before reaching pressure
S. M. NANOTI (Indian Institute of Petroleum): drop limitations. UOP has been offering different catalyst
Utility Requirements Per Metric Ton of Extract for grades for our Catalytic Condensation units apart from
Solvent Extraction Technologies our typical SPA-1 and SPA-2 catalysts that are designed
Utilities Sulpholane Arosolvan Morphylex Formex Gtc.* Morphy-lane* Tetra for maximum production of high value-added product
Electric 9.1 11.0 8.0 12.0 8.0 10.0 0.4
Heat Requirement 250 400 450* 370 325 230 295 or to maximize feed utilization. The UOP Higher Olefin
(Mcal)
Cooling Water (Cubic 7.0 30.0 27.0 4.5 47
(HO) series of catalysts were developed for the operating
Meter). ( t⳱15°C) conditions associated with the production of higher ole-
Basis: 1. Extraction Section Only (Excluding fins, such as nonene (C9⳱) and tetramer.
Fractionation) We believe that the best approach is to try both opera-
2. Reformate Containing ca 60 % Aromatics tional and catalyst changes to optimize desired selectivity
* Extractive Distillation Processes and the catalyst life as well as the economics of the unit.

E. Polymerization F. Isomerization

Question 31. Question 32.

Has anyone optimized Poly operating conditions to max- What are the current pressure limitations when consider-
imize C7-C9 olefin production in Poly gasoline? If so, ing conversion of a low pressure naphtha reformer to an
what operating changes were made? isomerization plant?

138 Light Oil Catalytic Processing

TOC/INDEX

BENNETT: on whether it is sulfur free or with sulfur. We have applica-

Most isomerization units run at moderate pressure. tions up to about 4% benzene right now.
There are two variables that are impacted by the pressure
of the unit, feedstock composition and feedstock contami- Question 34.
nants. If the feed is heavy and contains a high level of What has been the experience with the new higher activ-
C6 rings, higher pressure becomes an advantage. If the ity pentane/hexane isomerization catalysts?
feed contains sulfur, higher pressure is also advantageous.
Most reactor sections of TIPSM units operate in the 250- BENNETT:
300 psi range. PenexSM units operate between 300 and UOP has developed two new high activity isomeriza-
450 psi, but most of them in the higher level. Lower tion catalysts. One is LPI-100. A high activity catalyst is
pressures could be tolerated, but need to be done on a applied to existing zeolite isom units as well as new isom
case-by-case basis. units. LPI-100 was commercialized in 1996 and is running
in three commercial units. The performance of LPI-100
MAUREEN GILBERT (Kellogg Brown and Root): is better than other catalysts available.
Our CKS ISOM process can be applied over the same I-80 is the most recent development for the PenexSM
type of pressure ranges including the low pressure range ButamerSM market and that also is higher activity catalyst
found for naphtha reformer conversions. We have a zeo- than its predecessors. It was commercialized in May of
lytic catalyst, which has been applied to pressures down this year and there are currently three units running with
to around 300 psi. this operation. It is performing at expectations in all
the units.
Question 33.
What is the maximum temperature rise you have oper- RADCLIFFE:
ated with over an extended period of time on a pentane/ I have a report from an isomerization plant, which has
hexane isomerization unit? What maximum benzene con- operated since 1992 with UOPI18 isomerization catalyst.
centration in the feed have you run? They have reported no problems.

RADCLIFFE: CHRIS JOHNSON (Akzo Nobel Chemicals):

Maximum temperature rise across the isomerization For the most part, we have seen refiners quite interested
reactor obviously has significant implications when con- in taking advantage of the higher activity catalyst such as
sidering new gasoline specifications, particularly the auto our AT2G for pentane and hexane isomerization by short
oil 2000 and the benzene limits that exist in Europe. loading the reactor. In this manner, they achieve compara-
I contacted my colleagues in Milford Haven and they ble performance to what they have been obtaining in the
report running with up to 5% benzene in feed for periods past while saving up to 25% on their catalyst and platinum
of up to four days. A 40°C temperature rise has been costs. Despite having fewer pounds of catalyst in the
observed. This could be further increased subject to the reactor, tolerance to sulfur poisoning as well as water
limit on reactor outlet temperature. poisoning remain quite good and often times better than
in the past.
In one instance, the refiner is able to process more
BENNETT:
severe feeds, that being benzene, naphthene and C7 con-
We have a guideline of 100°F maximum rise. We know tent, while maintaining the previous performance. We
of a few units that have operated at 125°F rise. The know of a few refiners who will use this catalyst to comply
maximum amount of benzene in the feed for these with future benzene legislation. Furthermore, increase life-
instances is 5-6%. There are two problems with the high time with the new catalyst appeals to customers that suffer
delta T. One is that the combined feed exchange system from oxygenate and water poisoning.
can return that heat to you much quicker than you would
like, thus raising the reactor inlet temperature. Secondly, Question 35.
we have seen a couple instances of high reactor outlet Have refiners experienced leaks through the valves
temperatures leading to excursions. around the feed and hydrogen driers? What types of
valves or preventative maintenance procedures have
MAUREEN GILBERT (Kellogg Brown and Root): helped to provide tight shutoff?
With our CKS ISOM process, we typically have a low
delta T (approximately 5-10°C) and it depends on the WALDRON:
C7 content in the feed. There is a maximum temperature, During the late 1980s, we were using 8⬙, 300 psi rated
about 285°C. We have various designs we can apply to gate valves in drier service. We would change these valves
the reactor to take benzene contents up to 15% depending out on the average of once every three months due to

1998 NPRA Q & A Session on Refining and Petrochemical Technology 139

TOC/INDEX

IV. LIGHT OIL PROCESSING

A. Alkylation to an absolute minimum. This heavy material dramatically
increases acid consumption in a sulfuric acid unit and
1. Feed produces significant quantities of acid soluble oil (ASO)
in a HF unit. They also did see a slight decrease in alkylate
Question 1 octane quality with the inclusion of amylenes.
Are any refiners outside California alkylating amylenes?
If so, with what results? Do you have any plans to do so
in the future, and what are the reasons? JACOB:
One facility I am familiar with tried alkylating the
STYNES: remaining amylenes twice on a trial basis. There are no
When we started up our HF alky unit in Borger, plans to do this in the near future. The Deisobutanizer
Texas in 1942, we alkylated amylenes, and we have been tower fractionation suffered. The acid consumption was
routinely alkylating amylenes since that time. As far as high—between C3’s and C4 processing acid consumption.
industry experience, we know of a lot of examples. Alkylating C5 olefins consumed a lot of iC4’s. This facility
Amylenes have ranged from 11 to 30% of the olefins is usually short of iC4 and long on C3 olefins, and can be
fed. Amylenes make a product octane, which is compara- acid limited, they prefer to alkylate C3’s with its lower
ble to, or slightly higher, than propylene alkylate. The iC4 consumption for incremental feed. There is little desire
reason that we continue to alkylate amylenes outside of to pursue this again.
California is it reduces the amount of olefins and RVP A second refiner is alkylating amylene on an ongoing
in our gasoline pool. basis. This facility was designed to process C3, C4 and C5
One warning, if you have not alkylated amylenes before olefins. The amount of C5 alkylated is dependent on the
and you start bringing them overhead in your FCC debut- requirements for premium gasoline. By alkylating C5’s,
anizer, a substantial amount of heavies may come over the octane of the alkylate drops 2 numbers.
with the amylenes and that makes a lot of ASO. You
have to make a fairly tight cut if you are going to bring Question 2
amylenes into your alky unit. What is the observed production of isopentane in the
alkylation unit when processing amylenes? How has that
VERRENKAMP: production been reliably measured?
We do it at a few of our sites, primarily as Mr. Stynes
said, for RVP control. We will continue to do so whenever
the economics are obviously favorable. Range is about 10 STYNES:
to 15% on olefin feed, a little lower than what Mr. Stynes It is worth mentioning that the selection committee
said Phillips practice. We usually feed from FCC naphtha combined the sulfuric acid alkylation and HF alkylation
splitters so that we do get quite a tight cut. At one location questions this year. All my comments are about HF alkyl-
where we process a significant amount of these amylenes, ation.
the hydro isomerization unit upstream of the alky was There are many different numbers floating around on
able to cope with the pentadienes. The result was no how much synthetic isopentane you are going to make
appreciable drop in alkylate octane. if you alkylate amylenes, and some of them are really
high. I want to talk about where some of these numbers
GATTE: came from before I get on to commercial experience.
My comments are consistent with those of the other I have seen numbers as high as 0.5 to 0.8 barrels of
panelists. We conducted a straw poll of our FCC catalyst iC5 per barrel of amylene charged. Of course, we would
customers to determine the extent of amylene alkylation not be telling you how wonderful it is to run amylene in
that is currently being practiced in the US. While it does our feed if that was our commercial experience. In my
not appear to be a very common practice, we found several experience those numbers come from research studies that
refineries that currently feed a fraction of their amylenes have been done, specifically some by Phillips, that used
to their alkylation units, both in HF and sulfuric acid pure amylene feeds with very atypical acid compositions
operations. One refinery reported good success in process- and atypical operating conditions.
ing amylenes. However, as Mr. Stynes stated, they cau- Our commercial experience is much lower than that.
tioned that care needs to be taken when processing this Depending on the situation in your unit, I would suggest
stream to ensure that the amount of C6Ⳮ material is held a number between 0.1 and 0.25 barrels of isopentane per

1999 NPRA Q & A Session on Refining and Petrochemical Technology 159

TOC/INDEX

BRIAN JOHNSON (UOP LLC):

I will confirm the wide variation in isopentane produc-
tion. Our information indicates that incremental isopen-
tane production in the range of 12 to 67 vol% of the
incremental amylenes in the feed have been reported for
commercial operations. Reliable measurement is difficult
because the amylenes are generally limited to a minor
portion of the feed. Pilot plant operations on pure amylene
feed have shown that the hydrogen transfer reaction can
occur to an extent sufficient to explain the increased
isopentane production. The hydrogen transfer reaction is
expected to result in increased alkylate yield and octane
at the expense of increased isobutane consumption. The
greater isopentane production offsets some of the volatility
reduction gained by alkylating amylenes. Limitation of
feed contaminants is the key to successful processing of
amylenes. In addition to increased levels of sulfur and
barrel of amylene charged is about right for a commercial nitrogen, the amylene fraction can contain diolefins,
application. cyclopentenes and C6 plus materials, which will all increase
One of the questions that was asked is how do you tar production and reduce the octane. Increased attention
determine this. You do make isopentane even if you are to upstream treating and fractionation is recommended.
running a C3/C4 feed. So we subtract the amount of Selective hydrogenation may be justified to reduce the
isopentane we make without amylene in the feed, from load on the acid regeneration facilities.
the amount we are making with amylene in the feed and
call that the synthetic isopentane production. Dividing
that by the barrels of amylene, we come up with the 2. Process
numbers that I am quoting. Question 3
Here are some hints on how to minimize the amount Are there any additives or methods that have been used
of synthetic isopentane. If you have isopentane in your to either increase the octane of propylene alkylate or
feed or your recycled isobutane, you will depress the decrease the acid usage when processing propylene or
formation of isopentane. If you have heavies in the feed amylene feeds?
or in your recycle isobutane, you will increase the amount
of isopentane you make. KRISHNA:
High acid strengths make more isopentane due to I would like to share some experience we have had
cracking reactions, so keep your acid strength below 90% with the Betz Alkat additive. We started experimenting
to minimize the amount of synthetic isopentane. with this additive about three or four years ago when I
was involved in that business. We did a test run in a
VERRENKAMP: small alky plant for about three months. As most people
Good response. The only thing I would add to that know, before you test an additive like that, you need to
is that in sulfuric acid alkylation plants the observed pro- get your base case lined out and understand the process
duction is less than what you get for the HF acids. It is well so that you know what the differences are when you
somewhat less than a half of the HF acid alky plants. try a new additive. We did that. And, of course, developing
a model that is associated with this Alkat process is also
KRISHNA: key. We did that, too.
We have operated both sulfuric and HF plants, We did a two or three month test run, got results back
although we have only one HF plant left in our system. that clearly indicated that acid consumption went down.
I will answer the question for both types of processes. At that time there was enough data to suggest that it just
For sulfuric, we see about 20% hydrogen transfer, i.e., about paid for the chemical.
about 20% of the amylene gets converted to isopentane. Since then I understand that we have done a year long
We see this in both of our California refineries where we test in another small alky plant in the system, and we
operate two world scale alky plants, sulfuric alky plants have now shown convincingly that we get about a 12%
feeding C3s, C4s and C5 feeds. reduction in acid consumption. At that point, obviously
On the HF, we did a detailed test run many years ago it will more than pay for the chemical.
on a plant that we do not own anymore. The test run Based on that successful result, we are now testing it
report indicated a number as high as 60% isopentane. in one of the large alky plants on the West Coast, one

160 Light Oil Processing

TOC/INDEX

of these plants that I mentioned that feeds C3, C4 and to do the same thing. Since then I understand that we
C5 olefins. It is too soon to tell what the results are, but now have a Coriolis meter supplied by Micro Motion.
hopefully we can report in six months or a year or so. This has been installed in one of our alky plants. I hear
it is working really well. It does not require any special
STYNES: sample conditioning. In fact, the results are so good that
We do not know of any additives that will enhance the operators actually believe those data more than they
octane or decrease acid consumption when processing believe the offline titration. So we think that this may be
propylene or amylene in your HF alky feed. a cheaper, more effective way to go, and we have put our
near NIR analyzer on the shelf for now.
VERRENKAMP:
Like Chevron we have done trials with the Alkat addi- PARK:
tive at about three of our sites. The results were mixed, We operate a sulfuric type alkylation unit by
and any improvements were not universally acknowl- STRATCO and we also operate sulfuric unit by Mon-
edged. They may have been actually due to better unit santo. We have a total of seven analyzers. One is in the
monitoring as a result of the trial. Our feeling is that this alkylation unit, but this unit is already successfully started
type of additive would help a bad operation. up without an analyzer. But this year we installed the
one, but this one does not work well right now because
STEVE MATHUR (Chevron): of hydraulic problems. But it will be fixed within this
When processing propylenes mixed with C4 olefins, year. Other analyzers in the SAR unit work well in normal
acid consumption depends on if you have mixed C4 olefins operation situation, but unfortunately during startup
or C4 raffinate from the MTBE plant. Your acid consump- period they did not work anymore because of impurity
tion will be lower for mixed C4’s due to propyl sulfates and also this analyzer is very sensitive to temperature.
reacting with the isobutylene in the mixed C4s case as Therefore, we keep the temperature around 81°C. The
opposed to an MTBE raffinate that has little isobutylene. analyzer type is Foxboro’s 8701, ITEC (Intelligent Elec-
The biggest difference that we see with C3 olefins and trodeless Conductivity Transmitter) type.
C4 mixed vs. MTBE raffinate is a 40 to 50 degree increase
in end point of the alkylate. When processing C5 olefins,
it is critical to to not dig too deep into the C5 cut so that PROOPS:
cyclopentenes are minimized. Cyclopentenes consume 8 Our Pine Bend refinery also has a Micro Motion Cori-
to 10 times the normal pentene acid consumption. The olis mass flow meter in spent sulfuric acid service and we
acid consumption is about the same for mixed C5’s or are installing a second. The meter runs quite well. Flowing
TAME raffinate but the T90 and the EP do go up substan- density is correlated to the laboratory titration to create
tially if you are alkylating TAME raffinate versus mixed the on-line acid strength estimation. The correlation drifts
C5 olefins. over time as the operation of the unit changes, so we do
routinely update it.
Question 4
What has been the experience with online acid analyzers
to determine circulating acid strength? What has been STYNES:
the reliability and accuracy? What type of sample prepa- Phillips has an online FTIR analyzer at our Sweeny
ration system is required? refinery. It has been there since May of 1998. It is being
marketed by ABB. The analyzer measures concentrations
KRISHNA: of HF, water and ASO in the system acid. It was calibrated
We have experimented with a couple of different ana- with gravimetric blends, so it is very accurate. In fact, it
lyzers for measuring online acid strength. A few years ago is more accurate than the laboratory titration for HF. I
we developed a proprietary method for using NIR for have a long praise letter here from the operations personnel
acid strength. We have a patent on it. We tested it com- at the Sweeney refinery. They really like it. There have
mercially for several months. It worked really well. There been no plugging or fouling problems and maintenance
were some sample conditioning issues. One of the issues requirements have been minimal. No special training or
was that the probe was sensitive to the flowing stream additional personnel are needed to maintain it. It uses a
and if you had any kind of bubbles—we used to call closed loop sample system, which passes a small slip stream
them micro bubbles—-they affected the probe and the of the acid through an optical cell. The sample piping
fiber optics associated with that. and cell are located in the HF area. The light source and
But I think we solved all those technical problems. It electronics are in a shelter outside of the HF area, and
was a success, but our sense was that it was a fairly expen- the light is transmitted with fiber optics. It has been very
sive analyzer and we were looking for some other ways reliable and accurate.

1999 NPRA Q & A Session on Refining and Petrochemical Technology 161

TOC/INDEX

ROBERT LUDOLPH (Sunoco, Inc.):

At our Toledo refinery we have had excellent experience
with an on-line spent acid analyzer that was installed
on our sulfuric alkylation unit. The technology that was
chosen was the Micro Motion Coriolis meter that has
been mentioned by a couple of the panelists. We have
placed it on our spent acid line. We have had no mainte-
nance costs associated with this loop for two and a half
years. The only maintenance issue that arisen from time
to time is the buildup of polymer residue on the tubes
which will cause the meter to drift. We correct this drift
with a lab bias update program. Our corrected meter
readings versus the lab have been quite accurate and we
have been pleased with the results. Usually our spent acid
concentration readings are plus or minus 0.2% relative Figure 1
to the lab measurements. The beauty of the system is that
there are no sampling lines or filtering requirements for
the meter. More information regarding this application system that can remove HF from the unit in less than
can be found in the July 1998 edition of Hydrocarbon Pro- 10 minutes. Figure IV-1 is a photo of the activated system.
cessing. We also built a model for this particular unit and put
it in a wind tunnel to establish the dispersion effects. I
Question 5 will emphasis again, the requirements at other units may
What are the best practices with regards to HF mitigation? be quite different. Incidentally, when we did wet test this
particular unit for the first time, the refinery located up
the road phoned our site wanting to know where their
VERRENKAMP: cooling water had disappeared to!
I guess the first comment is the best practice is not to
have a release. Now, the way to achieve this, in our DAVIDSON:
opinion, is to ensure equipment reliability through posi- In addition to all of that, we have 3,000 gallon a
tive material identification, ironclad inspection processes, minute remotely controlled water cannons. We keep these
operating within established equipment limits and adher- directed at areas of the unit with the highest acid invento-
ing to those limits and conducting root cause failure ries. We have unit surveillance with color video monitors.
analyses and followup if failures do actually occur. I mention color because I think in the early days black
Maintaining HF detection and painting all flanges in and white was supposed to be easier to see a HF leak,
HF service are very important, and we like to ensure that but we have not seen that to be the case.
the operator rounds include periodic examination of HF
containing equipment, including instrument lines. This KRISHNA:
allows you to pick up small leaks before they become I think Mr. Verrenkamp has covered it pretty thor-
large releases. oughly. We focused on water monitors, rapid de-invento-
For more capital intensive mitigation techniques, the rying and emergency block valves. At least one plant we
best practice is to install a system appropriate to your were associated with did extensive modeling also, wind
specific circumstances. Factors such as credible release tunnel type of modeling.
scenarios, HF inventory, prevailing atmospheric condi-
tions, HF cloud dispersion modeling, location of other STYNES:
refinery process units, location of the surrounding com- I agree strongly with Mr. Verrenkamp, the first thing
munity, population centers, etc, play in the design of the you want to look at is how not to have a release. The
mitigation system. only thing I will add is two of our refineries do have
At one of the locations where we have quite a high water curtains and water cannons around their HF areas,
HF inventory, the HF alky is close to other process units and the rest of what I have as other mitigation has been
and it is close to a town. In this case, we have installed mentioned. Of course, what we think is the state of the
substantial HF detection capability, a water spray system art is the Reduced Volatility Alkylation Process (ReVAP威)
which is automatically activated by the HF detectors and that we jointly developed with Mobil. In the unlikely
which can deliver a 40:1 ratio of water to HF for our event of an acid leak, that will reduce your airborne
credible release scenarios and a HF inventory rapid dump HF by 90%.

162 Light Oil Processing

TOC/INDEX

WILLIAM SCHLESING (The Pace Consultants): VERRENKAMP:

Previously, I was supervisor for a HF alky plant at BP’s We did attempt to control water and acid at one of
Marcus Hook, PA facility. An additional benefit of color our HF alkylation units in the early nineties. We did this
monitors we saw was that it gave us a third man during by cycling up the feed dryers to higher outlet water content
routine maintenance and routine work on the unit result- and by injecting steam into the feed downstream of the
ing in an extra measure of safety. We could watch from dryers. It was difficult to see any benefit from this practice,
a distance what was going on and monitor the operation and some believe that corrosion increased in certain areas
and make sure it was done safely. of the unit. We have not done any control of water content
Secondly, with the water deluge, the other thing you since then. Vendors never recommend doing this. They
have to consider is where does all the water go. You have will, however, recommend to ease off on the acid regenera-
to plan in advance in terms of diking, equipment and tor operation to allow the water to come up slightly, if the
unit elevations, anticipated water runoff directions and water content gets below say 0.5% in the circulating acid.
personnel safety and access in the event you ever have to
use the deluge system. DAVIDSON:
We would be pretty nervous about injecting water, so
WILLIAM JOHNS (Texaco): we do not do it. But we do try to control it by not being
We have a successful practice in one plant. It involves overly aggressive in our feed drying. In fact, at one time
an additive and the technology is called the Alkad Process. we were thinking about putting in mol sieves but decided
Appropriate information for the transcript has been sup- against it just for that reason. So we generally are staying
plied by Brian Johnson (UOP) shown as an answer to at about 0.5% water in acid.
Question 13 in this section.
Question 6 STYNES:
In sulfuric acid alkylation, what is your commercial expe- In an HF alkylation unit, the best way to try to control
rience with drying the olefin feed and recycle isobutane water is with your rerun tower, but we do have one
streams? In addition to reduced acid consumption, what refinery that regularly runs at very low water levels in
other benefits are there? Is there a minimum water con- their HF, 0.3 to 0.8 wt%. They were able to purchase a
tent in the acid to maintain activity? wet propylene stream, injecting that downstream of the
butane dryer. It seemed to work okay but they could not
VERRENKAMP: purchase the material for very long, and therefore did not
We have coalescers in feed and recycle isobutane get any long term experience or any corrosion data.
streams at some of our sulfuric alkylation units. The extent
that we are willing to spend money to remove water is Question 8
really dictated by the cost of sulfuric acid to these sites. Has anyone dried iC4 recycle by sidecutting the deisobu-
At one our sites we have our own sulfuric acid generation tanizer column, and if so, what benefit was observed?
plant, so acid consumption here is not as great a concern.
Corrosion can also be a consideration since the corrosi- VERRENKAMP:
tivity of sulfuric acid will increase as the water content We have a grassroots HF alkylation deisobutanizer
increases. One additional benefit of feed coalescers is that scheduled for construction at one of our sites next year.
it will allow for early detection of slugs of water coming We are modifying the more traditional design by taking
in the feed thus reducing the chance of having an acid a vapor side draw of iC4, albeit only about four trays
runaway due to water. down from the feed (Figure IV-2).
The reason is predominantly for purity of the recycle
PROOPS: stream, but some benefit will be realized through reduced
Pine Bend has a coalescer downstream of the feed/ corrosion of the cooling exchangers on this service. The
effluent exchangers. We do a mediocre job of removing tower overheads containing residual water, HF, propane
water, as we are running the coalescer beyond design rates. and some iC4 will be fed to the depropanizer. There was
We typically see 3 to 6 wt% water in the spent acid. Most some debate about adding a reflux to this system, but the
of the deviation comes from the spent acid strength. benefits would probably only be marginal. Given the
accuracy of HF modeling on the rigorous simulation tools,
Question 7 it would be difficult to quantify accurately. It is a UOP
For HF alkylation units, is anyone controlling (restricting) design incidentally.
feed drying, or using water addition, to control the water
content of the acid? If so, what controls are used to Question 9
prevent getting too much water into the unit, and how Does anyone use fuming sulfuric acid makeup to offset
is corrosion controlled where the water and acid mix? dilution of the acid by water?

1999 NPRA Q & A Session on Refining and Petrochemical Technology 163

TOC/INDEX

HF release to atmosphere; otherwise, you are going to create

C3 a hazard for refinery personnel.
iC4 De propaniser
Feed Question 10
Are there any new developments to reduce acid consump-
tion and improve alkylate properties, such as octane and
distillation characteristics for HF and sulfuric units?
iC4
4 Recycle PARK:
We operate sulfuric acid type Alkylation unit and the
feedstock to the unit is the C4 stream from MTBE unit.
We have installed CD-Hydro at the top of the CD
nC4 column of MTBE unit to reduce acid consumption caused
50 Storage by 1,3 butadiene in the Alkylation unit. After installation,
we can reduce acid consumption from 66 to 32 ton/day
and the 1,3 butadiene content 6,900 to 80 wppm of the
alkylation unit feed.
Alkylate STYNES:
Storage We have seen some alkylate quality improvements
Figure 2. Deisobutaniser using ReVAP威. The commercial unit we have operating
at our Woods Cross refinery saw half an octane number
gain, and a decrease in alkylate T90 and end point. We
also have slightly lower acid consumption using ReVAP威
PROOPS: than with traditional HF operation.
I got this answer from Koch Sulfur Products Com-
pany—a major provider of sulfuric acid regeneration ser- KRISHNA:
vices. They said no one does this as far as they know. I have a couple of things to mention, both of them
Theoretically you are mixing 102 to 105% equivalent on sulfuric acid plants and where there is a need or a desire
acid strength with spent acid to generate higher strength to process propylenes, butalenes, and amylenes together.
circulating acid. The spent acid would be in the low 90’s The first one has to do with segregating C3 and C5
strength with very low water. Eventually you do have to feeds. STRATCO has published a detailed paper on this
send the spent out to a regeneration unit because of a couple of years ago at the NPRA annual meeting. Essen-
entrained hydrocarbon. Also, keep in mind EPA has ruled tially one of our California refinery’s new alky plant has
the only acceptable method to handle spent acid with been designed to separate the propylene from the amylene
hydrocarbons is thermal decomposition. feed. This is to keep negative interaction between those
Concerns with this scheme would be: two feed components that ends up producing some C10
1) oleum is substantially higher environmental risk plus molecules. We think that segregation is worth proba-
than 99% acid; bly about a number to a number and a half in octane.
2) if not properly controlled, you risk having substan- The other comment is that in processing amylenes we
tial SO3 vapors go out the top of the settler into have learned that the threshold for runaway is perhaps
the compression system; lower for C5 feeds, perhaps even lower than 86%. This
3) you would likely need to modify or add a separate recognition has allowed us to spend down to lower acid
fresh acid storage system, and strengths than we would otherwise, which of course
4) oleum is generally more expensive than 99% fresh reduces on acid consumption.
acid per acid equivalent.
BRIAN JOHNSON (UOP LLC)
STEVE MATHUR (Chevron): For HF units, generally the biggest boost in perfor-
I just wanted to bring the refiners’ point of view on mance is associated with the installation of a SHP unit,
this. The fact that SO3 is there means that you have more selective hydrogenation process, which can have a major
than 100% strength acid (oleum) and you do not have effect on acid consumption where the diolefin concentra-
the water that is needed for the ionization of the alkyl- tion is moderately high. Additionally, the SHP unit gives
ation reaction. a significant increase in the product octane because the
The other issue is that the tanks would have to have SHP catalyst isomerizes the butene-1 to butene-2 which
a system where the SO3 gas can be treated in case of produces a higher octane alkylate.

164 Light Oil Processing

TOC/INDEX

STEVE MATHUR (Chevron): have performed this particular revamp on three or four
In addition to comments from the panel, the acid sites without any problems at all.
consumption essentially is a function of the C3, C4, C5
olefin type and their composition. One thing I wanted BRIAN JOHNSON (UOP LLC):
to mention is that there is a different optimum spent acid UOP has several versions of staged olefin feed or series
strength for different combinations of C3, C4 and C5s. olefin feed in our HF units whereby there may be two
For example, you could have a 40-60/ C3-C4 stream and reactor sections receiving the isobutane recycle flow in
your spent acid strength optimum is more like 91%. You series with the olefin split to each at the hydrocarbon
could have a C5 stream, and your optimum could be 88%. inlet to the reactor itself. The olefin is injected through
So you could be saving the acid depending on if you have multiple feed nozzles as the emulsion passes through the
knowledge about where your optimum is. reactor. There more than 20 series flow isobutane units
and almost all UOP units have the multiple feed injectors.
Question 11 In the case of split olefin feed series iso recycle, the
In either HF or sulfuric units, has staged olefin feed main advantage for revamp considerations is that the
been commercially practiced? What were the benefits alkylate throughput can be increased without replacing
and detriments? major fractionation towers by using the fractionated iso-
butane twice as it passes through the reactor section.
STYNES:
Question 12
For HF acid, Phillips uses the staged olefin feed injec-
Our sulfuric acid alkylation unit has seen increased ero-
tion at our refineries. In addition there are about 20 other
sion and corrosion in recent years that we believe is due
licensee units using it with some currently in design. Of
to oxygenate carryover from our upstream MTBE unit.
course, the basic concept is that staged olefin results in
Has anyone else seen this same thing? What amount of
a higher isobutane to olefin ratio at the point of olefin
oxygenate have you measured? Have you seen an
injection, so you have to look at your particular unit to
increase in the water content of your spent acid? What
see what the advantages might be. If you are at a low
are you doing to solve this problem?
isobutane to olefin ratio currently, like the alky unit I am
most familiar with, you can expect to see some real benefits
in terms of higher octane. You can have as much as two KRISHNA:
octane number boost, with one octane being more typical. The olefins processing schemes in our two California
You will also see lower ASO yield and higher synthetic refineries include both MTBE and TAME plants, at least
alkylate yield. for now. These large alky plants, I mentioned earlier,
If you are running at higher isobutane to olefin ratios, process all the C3 olefins, the normal butenes and the
maybe you are in the flat part of the octane curve, then normal amylenes. Whenever we have had upsets in the
there are a couple of different ways you can take advantage MTBE or TAME plants, we have seen significant carry-
of staged olefin injection. Since you have a higher iC4 to over of methanol, as much as 2,500 ppm or so in the
olefin ratio, you can increase your feed rate, and maintain alky feed. These are really spikes and we really do not see
the same alkylate properties. So, it is a good way to a big problem with corrosion in the contactor tubes or
debottleneck your unit. Or if you are happy with your shell because of that. On the other hand, a constant
feedrate and happy with your octane, you could decrease carryover of MTBE or methanol due to a poor feed
your circulating isobutane rate and save some energy while
maintaining alkylate properties.

VERRENKAMP:
We practice this at several of our units. The economics
that generally drive us to do this, certainly on our HF
units, is increased throughput without replacement of the
isostriper tower. For the sulfuric acid units, the staged
olefin feed approach can be taken in cascade reactors.
We had some initial concerns originally in implement-
ing these systems, particularly on the HF units, because
we were worried about the field welding on HF containing
equipment. Rigorous weld procedures had to be estab-
lished and followed to ensure that the integrity of this
equipment was not compromised. Having said that, we

1999 NPRA Q & A Session on Refining and Petrochemical Technology 165

TOC/INDEX

coalescer operation, for example, could lead to corro- do improve and there was no change in the amount of
sion problems. ASO produced.
The other interesting report I received from one of
those two refineries is that whenever the MTBE and BRIAN JOHNSON (UOP LLC):
TAME plants are down but the selective hydro plant is A five month trial of UOP’s AlkadSM additive technol-
still running, we find that the acid consumption in the ogy was completed in July 1998. The trial was conducted
alkylation plant actually goes down. This is in spite of using the refinery’s HF alkylation unit to assess operating
the fact that you have isobutylene and isoamylene now costs, additive recovery and operability associated with a
coming to the alky plant. We still see a reduction in acid new, more cost effective additive. Continuous recycle of
consumption, which to us means that we do have some the additive/acid complex was maintained during the trial.
carryover of ethers when the oxygenate plants are running, The unit operated well and the results indicated that
that are consuming acid. We think it is dimethyl-ether, the operating cost with the additive did not substantially
probably in the 1,000 to 1,500 ppm range, that is being change from operation without. Product quality, yield,
carried over. The acid consumption from the carryover and soluble oil production were consistent with the base
is high enough that it apparently offsets the incremental case operation without the additive. AlkadSM technology
consumption from isobutylene and isoamylene when the is available for license from UOP.
oxy plants are down.

VERRENKAMP: 3. Mechanical
I agree with Mr. Krishna. In our experience, most cases
Question 14
of increased erosion or corrosion in the alkys is usually
What advances have been made in contactor mechanical
due to increased temperatures and velocities that result
seals to decrease emissions and increase reliability?
from pushing the unit harder to meet the octane blending
What type of seals are you using and what benefits have
requirements rather than any oxygenate carryover.
you seen?
HAROLD EGGERT (BetzDearborn Hydrocarbon): Question 15
We have been involved in several refineries that had We recently opened our contactor during a scheduled
corrosion issues due to water carryover. What we found turnaround and found substantial erosion and corrosion.
in those cases was adding a filmer or corrosion inhibitor In addition, we have noted reduced Mean Time Between
to the DIB overheard actually increased water carryover. Failure on the seals. What have you done to improve
If corrosion up there is a problem, the use of a neutralizer reliability and performance?
is effective, but if you use a filmer, it tends emulsify the
DIB overhead, drags water back to the front end of the
JACOB:
unit where it is combined with the olefin feed and the
filmer can actually coat the coalescer elements not allowing Two of our clients use double mechanical seals. Each
them to do their job. So if you are still feeding a filmer, uses a double mechanical seal with API plan 53. This is
you may want to consider improving post-treatment and an emission free design. They have an oil fill system that
removing that for corrosion protection. utilizes check valves and a hand pump, which allows filling
of the barrier tank without shutting down the contactor.
Question 13 On one unit, this seal has been in service since unit
How many HF alkylation units are using ‘‘reduced acid start-up and it has been operating for six years on the
volatility’’ additives? Have any changes been observed second unit.
in the operating conditions and/or the product quality, One of our clients has seen erosion on the head and
yield and acid soluble oil production? corrosion on the impeller. They changed the material of
the impeller to alloy 20.
STYNES:
There are two currently operating plants using the PROOPS:
Phillips ReVAP威 technology, Mobil’s Torrance refinery We have a single mechanical cartridge seal design (BW/
in California and Phillips’ Woods Cross refinery in Utah. IP seal provided by STRATCO -part number 60X1C28).
We also have a licensee located in the LA basin who will These have a mean time between failure of about two
be starting up their unit in mid-2000. There have been years. Recently we have included contactor seals in our
some minor operating changes for the acid regeneration LDAR program, so this may drive us to a Plan 53 if the
tower. However, the core operating conditions for the fugitives are not consistently low enough. One advance
reactor, fractionator and product treatment are we have made is to go away from Teflon sleeve bearing
unchanged. And as earlier noted, the product properties in favor of Grafoil.

166 Light Oil Processing

TOC/INDEX

We have coated our high erosion areas, including the mechanical seal. The seal faces are alpha silicon carbide.
hydraulic heads, with Belzona liquid metal. This has been The seal static elastomers are Aflas and the dynamic elasto-
in service for over three years, depending on the specific mers are Kalrez. All the metal is monel except for the
contactor. There was some lamination in one contactor springs and set screws which are Hastalloy C. We are
that we found last fall (caused by original surface prepara- generally getting 24 plus months out of these, except for
tion deficiencies), but it has all been reparable. the depropanizer reflux pumps, and I do not know why
other than the fact that they are self flushed.
Question 16
How vulnerable is the HF pump seal? Are there any fixes BRIAN JOHNSON (UOP LLC):
that can significantly improve the long term performance UOP units have acid circulating pumps and we do not
of this seal? consider the pump seals for these pumps to have a high
level of vulnerability, particularly with the modern dual
PROOPS: mechanical seal system we recommend. Our licensees have
We experience 1-3 years mean time between failure in reported good seal lives on the acid circulating pump in
our Corpus Christi HF alky unit acid circulation pumps. recent years with the dual seal systems specified by UOP.
These are bellows type monel tandem seals, API Plan 32 Historically this service is moderate enough in pumping
with isobutane flush vented to the flare. We are in the conditions that even single mechanical seals with a backup
process of changing these to pusher-type seals. Other auxiliary packing have given reliable service with blow
pumps in HF acid service, such as HF transfer pumps, out protection provided by venting the area between the
have similar mean time between failure. seals to a closed header that goes to the neutralization
scrubber. One of the main causes for the better seal life
STYNES: performance reported recently is the wider use of silicon
For a Phillips HF unit, we do not have an acid circula- carbide in the seal faces.
tion pump on the contactors so what I am talking about In addition, on all units designed by UOP since the
here is the acid rerun feed pump, or if you have one, the early 80’s, remote operated block valves have been pro-
rerun overhead product pump. The secret to long life on vided for the acid circulating pumps to end any potential
these seals is to keep the HF away from the seals. HF release very quickly.
attack is bad enough, but what usually happens in our
experience is you get some fluoride buildup around the WILLIAM SCHLESING (The Pace Consultants):
seal face which eventually pushes the seal faces apart, and Again, with regards to my former life with BP, if you
that is when you get your failure. can keep the seal from failing, that is the ideal situation.
The secret is good seal flush. If you maintain a good But as a backup to that, having local HF detection moni-
clean, preferably HF-free, seal flush material you will have tors by the seal in the event of a leak is valuable for early
a good long life on your seals. Not everyone has an HF- detection and response. Secondly, again with regards to
free material, so look around in your unit for the stream remote video monitors, one of the places that we kept
that has the lowest amount of HF, and that is usually our video monitor trained on (when we were not using
the best solution. The other thing that I will say about it to scan the area) was on those pump seals. In that
that is you need to keep the seal flush going even when manner it again gave us that remotely located third person
the pump is down. When we do all this, we get around that could immediately identify a leak from the safety of
two years mean time between failures on our pumps. the control room.
VERRENKAMP: VERRENKAMP:
I would just like to echo Mr. Styne’s sentiments on I will just add that those two items are still being
maintaining a good seal flush. That is absolutely critical practiced at BP today.
in reliability on these services. One of the other main
causes for improved performance throughout the industry Question 17
recently is the more widely used silicon carbide in the What types of problems are experienced when feeding
seal faces. That seems to have helped quite dramatically. C5s to an alkylation feed selective hydrogenation unit?
We have also installed a number of sealless pumps in
HF offloading and rerun tower services. Our results so KRISHNA:
far have been variable. The couple of things I will mention have to do with
catalyst poisons. This is based on our experience.
DAVIDSON: The selective hydrogenation catalyst is particularly sen-
We also have a Phillips unit, and all of our pumps sitive to sulfur, which is a temporary poison and lowers
use a cartridge type, stationary design, tandem, balanced, activity. You really have to pay attention to the sulfur

1999 NPRA Q & A Session on Refining and Petrochemical Technology 167

TOC/INDEX

rate (e.g. PONA, higher/variable end point, nitrogen, or

catalyst poisons)?

VERRENKAMP:
Our experience varies on the coking rate of coker naph-
tha but in general a factor of between two and three times
the coking rate for equivalent straight run naphtha is
typically used. This factor only applies when the end point
is consistently held below 340-350°F. With higher end
point, the coking rate is extremely difficult to predict.
The presence of direct coke precursor is in the heavy end
of the coke and naphtha is believed to be the primary
factor in the higher coking rate. Part of the cause for the
higher coking rate is the higher concentration of the C5
ring naphthenes versus C6 ring naphthenes in the coke
specification for the particular catalyst you are using. I and naphtha relative to the straight run.
understand that UOP has a new catalyst that will tolerate Now, this all assumes that the feed to the platforming
350 ppm sulfur. unit has been properly hydrotreated to saturate all the
Also, chlorides in the makeup hydrogen will do the olefins and to remove sulfur and organic nitrogen down
same thing, temporary poison, again. For C3 and C4 units to less than half a weight ppm. However, since coke
arsine, even ppb quantities, is really bad. It wipes out the and naphtha can contain perhaps 20 times the sulfur
catalyst, and it is a permanent poison. If you do have concentration and 50 times the nitrogen concentration
even small amounts of arsine in the feed, then you have as straight run, proper hydrotreating usually cannot be
to install a lead oxide bed, as we had to do in one of achieved in a naphtha hydrotreating unit originally
our refineries. designed for only straight run naphtha. Obviously the
presence of any additional sulfur and nitrogen in the
VERRENKAMP: reformer feed will increase coking even further.
We are quite wary of sulfur coming into the feeds when
we process amylenes particularly to the hydroisomeration KRISHNA:
unit, because the upstream LPG MEROX systems have We see relative coking rates for coker naphtha from 2
a lot of trouble removing all the sulfur when you are to 4 times that for virgin naphtha. I think, as mentioned,
processing C5’s through them. We run to a maximum it is really an end point effect that causes the variability.
sulfur in feed specification to these units of 20 ppm. If you have a 350-355°F end point coker naphtha, maybe
STEVE MATHUR (Chevron): the relative coking is twice as much as virgin naphtha,
Another critical issue with C5 selective hydrogenation but if you go to a 400 end point coker naphtha, then it
processes is whether your reactor is upflow or downflow. is much, much higher than that. So I think that is why you
We have both types. We have an upflow at Richmond see the range in the variability on the relative coking rates.
and a downflow at El Segundo. Whereas both of them Question 19
work perfectly fine, I understand that if the diolefin levels Does anyone process hydrotreated FCC naphtha in a
go up substantially (⬎1.0%) in a downflow reactor you reformer? What is driving you to do this? What is the
could actually get into substantial hydrogen distribution impact on cycle length and product quality?
problems because you could actually have the hydrogen
disappear in the middle of the reactor resulting in a single
DARDEN:
phase and diolefin breakthrough. Also, if you have too
A little background on the reforming setup at LCR
much hydrogen, you could actually start saturating the
will probably help understand why we do some of the
olefins instead and end up losing yield. Consequently, it
things we do at the refinery. We have two reformers, one
is better to go with an upflow design.
that processes BT feed and one that processes a heavy
end for either inclusion in the motor fuel pool or to
B. Reforming produce mixed xylenes for the Udex unit. That having
been said, we do process a heart cut off of the heavy fluid
1. Feed gasoline that goes back to the Magnaformer, which is the
Question 18 motor fuel unit that feeds the Udex system. The reason
What is the relative coking rate for coker naphtha feed we do that is pretty simple—money. On a normal basis
versus virgin naphtha? What drives the higher coked xylene pricing, and the paraxylene / orthoxylene spreads

168 Light Oil Processing

TOC/INDEX

to gasoline are high enough that it outweighs the contribu- differences between the alkylcyclopentane and the alkyl-
tion of that FCCU gasoline stream in the gasoline pool. cyclohexane yields are fairly small. Nitrogen contents in
We do not see any debit to reformer life for processing the FCC naphtha are also typically higher than the virgin
that stream because the NⳭA characteristic of it is usually feeds leaving the naphtha hydrotreater and this may be
better than the feeds that we normally run. We typically a challenge to achieve the target feed levels to the reformer.
see about a 16 to 18 month cycle on the Magnaformer and
we make about a 98 octane product from this reformer. ANGELO FURFARO (UOP LLC):
Gasoline pool economics on a normal basis do not drive FCC naphtha is processed in several platforming units.
us to put the heart cut stream to gasoline. Typically this The incentive to process FCC naphtha in a reforming
stream boiling range is 230—380°F on a D86 distillation. unit is a net increase in the gasoline pool RONC. For
most refiners, however, FCC naphtha is not a viable
KRISHNA: feedstock for the platforming unit since the small RONC
I mentioned yesterday that in one of our California gain for the gasoline pool usually does not justify the
refineries we process FCC naphtha. I think the question costs involved in hydrotreating and reforming the FCC
had to do with the desulfurization of FCC naphtha. So naphtha. The heart cut of FCC naphtha (170-300°F) is
in this case we do feed FCC gasoline to a naphtha hydro- most suitable for processing in the reformer, however
treater. It is about 15% of the total feed to that NHT, this material must first be properly hydrotreated to meet
and it then goes on to our reformers. We do not see an reforming unit feed specifications. At typical conditions
impact on cycle life or product quality in the reformer required to adequately hydrotreat FCC naphtha, any ole-
as long as the FCC behaves itself. fins will be hydrogenated into paraffins and approximately
1 LV-% of the aromatics will be hydrogenated to naph-
thenes. FCC naphtha typically contains more nitrogen
PARK:
than straight run naphtha, therefore adequate nitrogen
Normally we do not process this kind of naphtha in removal would determine required hydrotreating severity.
the reformer, but during initial startup of our FCC unit The impact of processing FCC naphtha on reformer cycle
we process some hydrotreated naphtha in our reformer. length can be severe, in some cases doubling, or even
We have a guideline of maximum portion of these naphtha more than doubling, the coke laydown rate compared to
under 20% in terms of considering catalyst poisoning straight run naphtha.
materials, particularly nitrogen and olefin contents. The subject of processing FCC naphtha was discussed
Our FCC process is equipped with ARDS. The FCC in ‘‘Catalytic Reforming of Heart Cut FCC Naphthas’’
naphtha quality was nearly identical to normal naphtha presented at the 1985 NPRA Annual Meeting (paper
except for a slightly high sulfur content. We found no AM-85-56).
indication of low performance or other detrimental impact
on catalyst and cycle length.
JAMES JONES (Turner Mason & Company):
But, since the FCC naphtha contains relatively high
naphthene content, the operating condition and heater We are familiar with a refinery in the Pacific Northwest
load was more severe than normal situation. Semi regener- that was doing this as recently as 1997 to maximize low
ation unit was operated at the reduced charge rate (70%). sulfur blendstocks for CARB gasoline. They already had
the additional reforming capacity in place and believed
this approach to be more economical than making new
VERRENKAMP: investments just to hydrotreat the FCC gasoline by itself.
We do this at several sites. Our incentive has typically We agree this practice desirable when existing capacity
been to recover aromatics for chemical sales instead of in naphtha hydrotreating and reforming units is available
gasoline blending. We do anticipate increasing reforming and you have some fractionation equipment that can
of FCC naphtha in the future to recover octane after make the heart-cut FCC naphtha. In situations where
hydrotreating to meet gasoline sulfur restrictions. Some idle equipment or capacity is available, reforming heart-
of the reformers have processed up to about 20% FCC cut FCC naphtha to producing low sulfur gasoline is likely
naphtha. About 10% is typical for the sites that do it. to be more economic than the alternative of investing new
The FCC naphtha has a higher NⳭ2A content than the capital in, and paying relatively large royalties for, a grass-
virgin feed, so the impact on cycle life is dependent on roots selective FCC hydrotreating unit.
the relative NⳭ2A content and the target octane. In
addition, FCC naphthas tend to have higher alkylcyclo- Question 20
pentane contents at the same NⳭ2A. In general, the Do refiners with BTX extraction facilities change reformer
alkylcyclopentanes have poorer reformate yield and higher feed boiling range in response to changes in benzene-
coking tendency than their alkylcyclohexane counterparts. xylene prices? How much can yields be shifted within unit
However, with a middle cut of the FCC naphtha, the constraints? Are gasoline blending constraints limiting?

1999 NPRA Q & A Session on Refining and Petrochemical Technology 169

TOC/INDEX

DARDEN: limited range within which to modify the extent of aro-

On a typical basis, we do not see a lot of reason to matic dealkylation.
change the cut to produce benzene or toluene. We have In principle, the feed boiling range could be adjusted
a pretty fixed demand off our system so we take the to vary the product composition in such a manner as to
purchased naphtha and the naphtha that we produce in maximize the value of the aromatics produced. It could
house through an LEF tower to about a 290°F overhead be beneficial to selectively exclude precursors of benzene
end point to feed that material to the BT reformer. We or xylenes to maximize production of the other if there
do not swing that cut very often. We do swing the bottom is more C6 to C9 range naphtha available than can be fed
portion that goes to the magnaformer as the xylene pricing to the aromatics complex. Other processing options for
changes and as the kerosene economics change. the fraction of the BTX naphtha that is excluded and the
effect on the gasoline pool would have to be considered.
PARK: There could well be constraints that prohibit shifting
We have reformate from two CCRs and a semi regener- benzene production to the gasoline pool in order to max-
ation reformer. We have experience of changing feed ini- imize paraxylene.
tial boiling range up to 82°C from normal IBP of 75°C The complexity of aromatics complexes can vary sub-
for the purpose of mainly enhancing light straight run stantially. There may be, by design and unit synergy, a
naphtha yield from CDU process not for BTX improve- lot of flexibility with which to vary the product slate
ment. We call this the LSR swing cut operation depending regardless of the distribution of aromatics from the Plat-
upon economic benefits including ethylene plant bound- forming unit. There are processes to alkylate, dealkylate,
ary. The operation consequence as expected provided us isomerize, and disproportionate and/or transalkylate aro-
a few benefits; reformate yield was slightly improved about matics such that the production of the more desired aro-
1.0 vol% from 80.5 to 81.5 vol%. Xylene yield was notice- matics can be increased from the distribution produced
ably changed about 2.0 vol% from 24.4 to 26.4 vol%. in the Platforming unit. A8’s can be selectively converted
Benzene and toluene yield was also enhanced about 0.5 to any of the desired para-, meta- or ortho-xylene isomers
vol%. We judge total BTX yield would be flexible depend- by separating the desired component and reestablishing
ing upon BTX precursor content in naphtha feedstock. the equilibrium distribution of the remaining A 8 ’s.
The main process constraint was heater because of higher Depending on the processing units installed and the cata-
required heat of reaction resulted from higher boiling lytic systems employed, the paraxylene/benzene ratio pro-
range. So, in some cases, reduction of charge capacity duced by an aromatics complex can vary between 0.6
would be required. and 4.0.
The driving force to maximize production of the more
PROOPS:
desired aromatics in the platforming unit is decreased
Our Corpus Christi refinery, which is a major aromatics when such processes are available. Instead, maximization
producer, generally does not do this. We vary toluene of total aromatics that can be converted to benzene and
dealkylation charge in response to benzene prices. We xylene is favored. The optimization of product value from
have also reduced toluene disproportionation charge when a particular naphtha in an aromatics complex involves
toluene value in gasoline exceeds benzene and xylene mar- consideration beyond that of the naphtha distillation.
ket prices.
LEE TURPIN (Aspen Technology):
ANGELO FURFARO (UOP LLC): What is not stated in the question but rather implied
The benzene yield from a Platforming unit is maxim- is the feed rate also needs to be an independent variable
ized by a feed initial boiling point (IBP) of approximately if feed boiling range is to be considered an independent
180°F and minimized by a feed IBP of approximately variable in the optimization of a chemicals reformer opera-
220°F or higher to eliminate direct benzene precursors. tion. In addition, the unit must be able to handle signifi-
A feed End Point (EP) of approximately 300°F would cant variations in coke laydown, and the plant must have
minimize xylene production and an EP of approximately a home for the naphtha cut not processed through the
340°F would maximize xylene production in the Plat- reformer. For refiners, varying feed rate means looking
forming unit. Higher feed EP would increase the produc- at alternate feed streams or purchased naphtha as incre-
tion of C9Ⳮ aromatics (A9Ⳮ). A9’s may be desired as mental feed. In a petrochemical plant, this means buying
xylene precursors if a means to transalkylate A9’s with more base or additional naphtha. An operation with vary-
toluene is available. Even without benzene precursors ing feed rates, feed boiling ranges is an ideal situation for
there will be some benzene production due to dealkylation on-line kinetic model based optimization. I recently did
of other aromatics. Aromatic dealkylation is minimized a study on a plant that could vary feed rates, and feed
by low-pressure operation. Existing units only have a boiling, and found the potential benefits to be very, very

170 Light Oil Processing

TOC/INDEX

substantial when looked at through a cycle of aromatics reactors have inherently lower possibility of these prob-
price fluctuation. lems than fixed bed type reactor.
Usually we reflect several monitoring points to check
2. Process these problems such as mainly changes in product yield
and reactor pressure drop and temperature decrease for
Question 21 each reactor. Particularly, catalyst loading is very impor-
What data are used to detect bed damage, plugging, and tant. Special caution should be taken to prevent catalyst
maldistribution in radial flow reactors? fluidization on the top of catalyst bed section during
PROOPS:
service cycle. We guess catalyst fluidization on the top
For initial troubleshooting, we measure reactor pressure section is the main root cause of screen plugging and
drops (with DP cells, not local gauges), reactor delta T’s, reactor pressure drop.
and long term trends in percent of total delta T for each
reactor. Continuous monitoring of yields, catalyst activity Y. K. KUCHHAL (Indian Institute of Petroleum):
and deltas versus predicted performance also tell us if In radial flow reactors, a decrease in delta T across the
there is a problem. If something looks suspicious, we call reactor, decrease in octane number and an increase in the
Tru-Tec. naphthene content of the reformate may indicate bed
Bed damage, plugging and maldistribution in radial damage, plugging or maldistribution of the feed.
flow reactors can be identified and characterized by the
use of radioisotopes. Inert radioactive gases, such as argon- WILLIAM MIXON (TruTec Services):
41 or krypton-85, can be injected with the feed and In addition to the diagnosis of maldistribution and
monitored externally with radiation detectors located at pluggage and damage in radial flow reactor catalyst beds,
the inlet, outlet and at various heights on the reactor to we have also done successful testing to identify fluidization
trace and quantify the flow path of the feed. Normally of catalyst beds. This is done by simply using a stationary
feed would travel through the catalyst from the outer source and detector placed above the catalyst bed while
annulus into the inner pipe and exit the reactor. When rates are increased. We can identify at which point the
plugging, damage or maldistribution occurs, the detectors bed becomes fluidized by monitoring for density changes.
respond to the timing and concentration at the various
locations, indicating poor flow paths. Question 22
Gamma scans can also identify catalyst loss or reactor Has anyone measured chloride content, or more specifi-
damage. The scans at different chords measure the densi- cally organic chloride content, of reformer stabilizer
ties of the bed. Voids or pluggage can be distinguished feed? (a) Have you measured chlorides upstream and
from the normal bed by comparing the density differences downstream of a stabilizer feed chloride guard bed? What
along the chords. was the chloride removal capacity? (b) Have you experi-
enced organic chlorides in the stabilizer bottoms?
STYNES: (reforming) (c) Have you used chemical agents to tie up
We also watch delta P on the reactors. Another tech- free chlorides in the stabilizer feed to reduce the chloride
nique is to watch the delta T versus what would be pre- in the stabilizer overhead? Have there been problems
dicted for that part of the run. Also, although this is after as a result of having the chloride going with the stabi-
the fact, you can watch your regeneration burn plots. If lizer bottoms?
you get a big tail, then you have a distribution problem
or bed damage.
BILLS:
KRISHNA: For the purposes of clarity, I will answer 24 and 22
I think the previous panelists covered it pretty well. through point (a) together.
We also look for high pressure drop, lowered endotherm This question has particular relevance for a CCR or
in the reactors and early breakthrough of oxygen during moving bed type reformer. The process conditions are
regeneration. We also use gamma scans to look at the such that the product olefin content (C4 or C5) may be
catalyst bed height and see if there is any indication of between 4% and 7%, and the system chloride content
fluidization or actual transfer of catalyst from reactor to may be 3 or 4 ppmv. Primarily, the chloride compounds
reactor. formed in this system will appear in the stabilizer feed as
either HCl or ammonia chloride, organic chlorides will
PARK: form at ppm levels.
We have never suffered big problems with radial flow Measurement of the ammonium chloride and HCl
reactor internals, but have just experienced minor prob- content of the stabilizer feed is easily accomplished, and
lems related internal screen damage. We guess radial flow may be done in the local lab. However, measurement of

1999 NPRA Q & A Session on Refining and Petrochemical Technology 171

TOC/INDEX

organic chlorides is a more difficult process requiring was being picked up by recontacting the reformate with
sophisticated analytical equipment and techniques. the net gas downstream of the net gas alumina chloride
Several methods may be used to remove chlorides from removal bed. Since then we have changed that setup and
the stabilizer feed including chemical agents, alkaline was- no longer have that arrangement.
hes, or alumina bed absorption. Alumina bed absorption
may be the most common as well as a possible catalyst VERRENKAMP:
for the formation organic chlorides. The alumina absorbs We agree with Ms. Bills as well. The only thing I
HCl, however it does not absorb the ammonium chloride would add is that we have actually tried chemical agents
or the organic chlorides. As the alumina absorbs more to tie up the chlorides. It was basically unsuccessful for
and more HCl it becomes acidic and eventually a catalyst us. The salt stuck to the bottom of the tower and caused
for the reaction between HCl and the C4/C5 olefins present a reboiler fouling issue. I have actually heard of other
in the stabilizer feed. As this process continues, the engi- cases where problems occurred in the feed bottoms
neer may falsely presume, based on inlet and outlet HCl exchangers and in the reformate storage tanks.
measurements, that alumina is continuing to absorb HCl
from the stabilizer feed. However, chlorides are actually BINH LE (Merichem Company):
‘bypassing’ the alumina to the stabilizer undetected— Most of our licensees of the CHLOREX unit measure
now as organic chlorides. the chloride before and after the removal process. Nor-
Theoretically these chlorides should be destroyed in the mally they are looking for only inorganic chloride.
stabilizer reboiler, exiting the stabilizer overhead, however, Recently we have seen some problems with the decrease
these chlorides may polymerize, in which case the stabilizer in performance, and we/clients start to investigate and
reboiler temperature is insufficient to destroy the long look for organic chloride. We detected a substantial
chain molecules. These polymerized chlorides may be amount of organic chloride in the reformer stabilized feed.
found in the stabilizer bottoms equipment. Note that The performance of the removal would be dependent
some of these polymers can be quite carcinogenic and on the type of organic chlorides. If they are inorganic
some consideration should be given to mitigation and/or chlorides (i.e. NaCl , NH4Cl), we normally see 95% or
clean up techniques if vessel entry is required. better removal by our CHLOREX process. Organic chlo-
Formation of organic chlorides can be remedied by rides mostly would go through the system and end up in
choosing the correct absorbent and/or limiting contact the downstream stabilizer. In the reboiler they will convert
time between the stabilizer feed and the chloride guard. to HCl and HCl would end up in the overhead. They
In addition, the chloride guard (alumina) should be have seen a lot of corrosion in the downstream equipment
changed on a set maximum time schedule or based on and piping of the overhead. Merichem continues to work
total pounds of chloride absorbed. Most alumina chloride with our licensees on finding ways to minimize the forma-
guards can absorb between 6% to 12% by weight. With tion and the removal of organic chlorides.
regard to the other chloride removal options, various oper-
ators have used chemical agents to absorb the chlorides ANGELO FURFARO (UOP LLC):
with mixed results. Ms. Bills did an excellent job covering the critical
points. Below I have included my answer to the above
KRISHNA: questions, but first want to make one quick point regard-
That is a pretty complete answer. We agree with Ms. ing Mr. Krishna’s comments. If you are having fouling
Bills; we think you could form organic chlorides. The or corrosion problems and are carrying through the system
alumina bed could react HCL with hydrocarbon that is chlorides that should not be there, it will be difficult to
absorbed on the alumina, to form organic chlorides. As measure the organic chlorides at these low levels. Chevron
she mentioned, we think it is probably more of an issue was actually recontacting a chloride treated steam with a
for CCR reformers because they tend to have higher stream that had not been chloride treated. This recontact-
olefins in the feed. ing picked up trace amounts of chloride and carried them
Having said that, the stabilizer bottoms runs hot downstream causing problems.
enough that we think it probably decomposes the organic There have been very limited measurements of organic
chlorides so you do not see it all the time. I think the chlorides. Of the data available, the ‘‘total’’ chloride of
decomposition temperature is maybe 300-325°F or so. the stabilizer feed is about 1 wt ppm. With such low
One other incident that is probably worth mentioning concentrations it has not been possible to distinguish
is that a few years ago, in our CCR reformer, we had between chloride types. Using the spent chloride analysis
a significant problem with chlorides, both organic and of units with stabilizer feed treaters, the stabilizer feed
inorganic. We did a lot of work to try and find out what ‘‘total’’ chloride appears to be typically in the 1-3 wt
was going on and did detect some organic chloride in the ppm range. One unit, with higher chlorides, was able to
stabilizer bottoms. I think it turned out that the chloride measure organic chloride at the stabilizer feed. This unit

172 Light Oil Processing

TOC/INDEX

also had a liquid chloride treater in place. The organic such chemical agents usually involved an aqueous
chloride content was ⬃ 2 wt ppm and total chloride solution, which can cause other corrosion concerns.
was 3 wt ppm. The organic chloride concentration was Also, some refiners have tried chelating agents, but
relatively constant in and out of the liquid treater, as well in some instances problems have been reported in
as the stabilizer bottoms. It is expected that the organic the bottoms exchangers, downstream units, and in
chlorides would not be removed with their adsorbent reformate tanks.
material and that these species would leave the unit with
the reformate.
a. Have you measured chloride upstream and down- J. B. RODDEY (Roddey Engineering):
stream of a stabilizer feed chloride guard bed? What I was under the impression to make an organic chloride
was the chloride removal capacity? you had to have chlorine gas and an olefin or an aromatic.
What I just heard from several people is that you can
Yes, there have been limited analyses performed. actually react with the HCL directly with the olefin I
It is expected that a new guard bed material will guess or an aromatic. There is a reaction called a Deacon
remove 100% of the HCl. This is consistent with reaction whereby in the presence of water at temperatures
the analyses available. As the bed ages the removal above 700°F you can form chlorine gas. Now, are we
efficiency will decrease and typically some HCL sure that the reaction is HCL with the olefin or due to
will be detected in the stabilizer overhead gas water in the recycle gas? We had trace quantities of chlo-
stream. The presence of HCl in the overhead is rine gas and the chlorine gas is actually the agent that
not linear with guard bed age. In most liquid guard reacts with the olefin or the aromatic.
bed materials there is little or no removal of
organic chlorides.
BRIAN JOHNSON (UOP LLC):
b. Have you experienced organic chlorides in the sta- We have a lot of experience with very dry systems in
bilizer bottoms? (reforming) isomerization where this same reaction has occurred and
The data available indicates chloride in the stabi- been identified. I do not think it is the water that we are
lizer bottoms. There is very limited data on this talking about here. It is HCL and an olefin that recom-
stream and the ‘‘typical’’ concentrations, which are bines even at fairly low temperatures in these systems.
less then 1 wt-ppm, do not allow distinctions
between total chloride and organic chlorides. Any Question 23
chlorides measured in the stabilizer bottoms should Which brand and model of ‘‘sniffer’’ tube do you use for
be organic chloride, however, since HCl should measuring chlorides in reformer recycle gas? Have you
leave the column overhead. Reports of chlorides tried others? Over what range of chloride concentrations
in certain downstream units seem to confirm the do you believe this tube is ‘‘best’’? Are there any recent
presence of organic chlorides in the stabilizer bot- advancements in field measurement technologies, such
toms. as solid state sensors?
Commercial Note: UOP commercialized in 1999
PCL-100 in the stabilizer feed stream that has BILLS:
proven to have significantly longer life prior to At our plant, for one reformer we use the sensidyne
breakthrough than other adsorbents used in this tube and for one reformer we use the draeger tube. That
service. The new adsorbent is removing both HCl has more to do with operator preference. We went through
and organic chlorides, thus eliminating chlorides a lot of trouble when we were trying to switch over
from the stabilizer for extended periods of time. between these two. We actually found that the draeger
As of this writing, the first loading has been on- tube tends to be more accurate when measuring, but
stream for seven months, more than twice as within lines of tolerance we found that the colormetric
long as any previous product used. PCL-100 is
method is only Ⳳ25% accurate, and that was kind of a
available through UOP LLC.
blow to those of us who had been really relying on those
c. Have you used chemical agents to tie up free chlo- tubes. So either one is good enough, and their accuracy
rides in the stabilizer feed to reduce the chloride is not that great anyway.
in the stabilizer overhead? Have there been prob-
lems as a result of having the chloride going with
DARDEN:
the stabilizer bottoms?
I agree. We use both types in our reformers. The
Yes, there has been very limited use of such chemi- operators like the sensidyne because it is a single pull
cals, with varying degrees of success. The use of versus 10 pulls on the Dräger tube.

1999 NPRA Q & A Session on Refining and Petrochemical Technology 173

TOC/INDEX

ANGELO FURFARO (UOP LLC):

Based on equilibrium data, the conditions necessary
to form organic chlorides do exist in the reforming unit.
Organic chlorides form as the product of olefins and HCl.
The reaction rates for organic chloride formation are a
function of temperature. The temperatures present in the
reactors are too high to allow organic chloride formation
and the temperatures present after the product condenser
are so low that the reaction rates are near zero. Some
organic chlorides are expected to form as the reactor efflu-
ent is cooled in the CFE and product condenser. In
general, more severe operations (higher temperature and
lower pressure) favor higher olefin concentrations and so
the concentration of organic chlorides would be expected
to increase.
It is also known that the concentrations of light organic
chlorides in the net gas can increase dramatically across
a net gas chloride treater using activated alumina that is
‘‘spent’’ and highly loaded with chloride. This is because
the alumina becomes acidic and catalyzes the formation
DAVIDSON:
of organic chlorides from trace light olefins and HCL
We use the Dräger tubes in the 1 to 10 ppm range. present in the net gas. In some cases, virtually all of the
We have also tried Gas Tech, which we use for sulfur HCL entering the treater was being converted to organic
but not for chlorides. chloride and passed through the treater in that form. For
this reason, UOP advocates replacing the alumina based
PROOPS: on a calculated chloride loading rather than depending
We use both Dräger (CH29501 1a) and Gastech (Sen- entirely on HCL measurements at the outlet of the treater.
sidyne 14L) tubes for chloride measurement in reformer
recycle gas. We use Gastech tubes for detecting break- Question 25
through of the chloride absorbers because they can detect What criteria are used to determine when to ‘‘dump and
organic chlorides. screen’’ catalyst for semi-regen reformers (e.g., time,
number of run cycles, performance, number of tempera-
ANGELO FURFARO (UOP LLC): ture cycles, other)?
The two most commonly used hydrogen chloride
detector tubes are manufactured by Dräger and by Sensi- DAVIDSON:
dyne/Gastec. The Dräger tube ‘‘Hydrogen Chloride 1/ We will go 9 to 10 regenerations before screening, or
a’’ has a standard range of measurement of 1 to 10 ppm three years, whichever comes first.
using 10 strokes, however with 5 strokes the range extends
to 2 to 20 ppm, and one may be able to detect down to 0.5 KRISHNA:
ppm HCl using 20 strokes. The Gastec tube ‘‘Hydrogen We recommend dumping and screening every three
Chloride Low Range 14L’’ has a standard range of mea- regenerations or two years, whichever comes first. We
surement of 2 to 20 ppm using 1 stroke, however with also look for how far the center pipe has ratcheted out
5 strokes the range is 0.4 to 4 ppm. Through a range of of the socket and how close we are to passing catalyst
experience using various ‘‘sniffer’’ tubes, we have found from reactor to reactor.
the Dräger Multi Gas Detector to be the easiest apparatus Also, I have a couple of added thoughts from our
to use and read. reforming expert. Because of the radial reactor design,
maldistribution due to fines is very hard to pick up by
Question 24 pressure drop alone. The only indicator that has a track
Is it possible to form organic chlorides in a catalytic record at all of showing this is a long afterburn during
reformer, under reasonable operating conditions? regeneration, a poor secondary burn and/or failed coke
ball test. Unfortunately, all this comes too late in order
BILLS: to schedule the right equipment and to allow for the
The answer is ‘‘yes’’ under normal conditions, particu- necessary time during the regeneration window. The typi-
larly in CCR or moving bed type units. Please refer to cal damage that is done due to fines comes primarily from
Question 22 for a more complete answer. high temperature coke balls. Reactor scallops, center pipes,

174 Light Oil Processing

TOC/INDEX

and in the worst case reactor walls have been damaged in the reactor and melting the ring off the side of the
because of this. reactor. So a very small delta T when you are going up
in oxygen is a dangerous thing.
PARK:
Our regeneration frequency is based on a one year cycle Question 26
because of heater limitation. Second, another objective of What analytical methods are currently being used to
dumping and screening is the makeup of fresh catalyst measure sulfur and nitrogen levels of less than 1 ppm
for the sake of restoration of catalyst activity. So, we in reformer feed? How reliable are these methods?
usually perform the catalyst regeneration and the dumping
and screen work simultaneously every year. Through this STYNES:
work, we maintain catalyst activity up to nearly identical We are using a pyrochemiluminescent technique by
performance. Antek to measure very low parts per million, even parts
per billion, of nitrogen, and we also like Houston Atlas
STYNES: which gives good repeatable answers below the 1 ppm
We dump and screen every three years, or two run range for sulfur.
cycles.
DARDEN:
DARDEN: We also use the Antek instrument for both sulfur and
We dump and screen every other cycle on both of the nitrogen. The technology behind that test method is
reformers. One item that would cause us to dump and ASTM D-5453.
screen earlier would be a reactor that has a particularly
bad run. This could manifest itself by a high delta P, low DAVIDSON:
activity, etc. We would then will dump and screen that We use two methods for determining sulfur. One is
reactor during the next regeneration. During the carbon the 5453 analyzer. It shows us if there is sulfur in the
burn step, if we see an unusually long tail at the end of sample, and we think it is probably accurate to within
the burn, we will dump and screen also. 0.3 ppm. We also check the hydrogen recycle stream and
the stabilizer overhead stream for H2S with the Gas Tech
Y. K. KUCHHAL (Indian Institute of Petroleum): tubes, and we find this is a better method than looking
For semi regenerative reformers, generally dump and for sulfur in the reformer feed.
screening of the catalyst is being carried out during any
indication of bed damage, plugging, maldistribution or 3. Mechanical
increase in DP across the reactor bed. To obtain good
performance, it is advisable that the first reactor is screened Question 27
after every one successive regeneration. However, for other An important aspect of continuous reforming is ball valve
reactors, catalyst screening can take place if any of the reliability. What is the commercial experience with the
above indication is as and when observed. available valves, and maintenance intervals? What is
your ‘‘best practice’’ approach to this equipment?
J. B. RODDEY (Roddey Engineering):
Everybody alluded to a coke ball during regeneration, KRISHNA:
and obviously that is something that happens in a Let me start by mentioning the types of valves we have
reformer. That is one of the reasons that we recommend in relation to the terminology used by UOP on CCRs.
dumping and screening to avoid those coke balls. We have McCanna ball valves for what UOP calls ‘‘B’’
No one mentioned that if you go up on oxygen after valves, and Fischer ball valves for the ‘‘V’’ valves.
you finish the carbon burn and when you go into the Both of these ball valves seem to withstand the service
oxidation period, sometimes it is very deceptive. You can to an acceptable degree. The one problem with these
see a very minor increase in delta T of less that 10°F valves is catalyst dust-induced attrition in the packing
because most of the gas is bypassing the coke ball and areas which results in leakage and the potential for fires
very little of the gas is actually going into the coke ball. around the packing area. Most of this dust can be traced
So although a lot of heat is being generated in a very back to the vent valves or the ‘‘G’’ valves, which are
small area, you can still get hot enough to melt a hole in Mason-Neles valves in our case. These vent valves will
the center pipe. start leaking and allow catalyst dust to migrate into the
We are aware of a very recent occurrence of that, and ball valve packing area, which then leads to the leakage
on one occasion about 8 or 9 years ago, a 10°F delta T I mentioned earlier.
when raising oxygen for the oxidation period resulted in Our best practice approach in this case would be to
melting the bottom two feet off of almost every scallop monitor the leakage rate of the vent valves and replace

1999 NPRA Q & A Session on Refining and Petrochemical Technology 175

TOC/INDEX

the seat and plug as soon as possible. These valves are drop 35%. This reduced the average reactor pressure about
easy to change out. It does require a short down time on 6 psi and improved reformer yield slightly.
the CCR regenerator, and for those who cannot afford Pine Bend replaced a pair of vertical combined feed
that, that would be a problem. effluent exchangers with a single Packinox exchanger. This
Because the damage of the leakage from these valves debottlenecked the reactor circuit heaters and coolers and
is often slow and not very detectable, most refiners will allowed a 15% increase in capacity. We found it critical
put off replacement until the leakage rate actually affects to reliably filter the feed ahead of the Packinox to prevent
the lifting pattern or the lock hopper sequence. Usually plugging of the exchanger spray bars, which have 3
by this time the associated ball valves have sustained dam- mm holes.
age also.
We try to keep a complete vent valve assembly ready STYNES:
for replacement in order to minimize the down time of Phillips has revamped some reformers with vertical
the regenerator. It has also been our experience that if rod baffle heat exchangers and also used them in new
the vent valves are changed out soon after they are detected construction for feed/effluent service. We have had great
to be leaking, the ball valve life is greatly extended. We success due to the low shell side pressure drop and positive
also plan and do leak test on all 8 ball valves during the tube vibration protection.
plant’s scheduled turnaround every four years.
DARDEN:
PARK: We installed a Packinox on our Magnaformer several
As for our experience in UOP CCR reformer, G-type years back. Initially we had a lot of problems with the
ball valves were the troublemakers due to the damage of spray distributor plugging. We did make some modifica-
the inside Teflon ring mainly caused by catalyst fines. So, tions and since that time we have had a very reliable
we replaced most of these valves with other ball valves. operation. The Packinox replaced a lot of exchangers in
To avoid this plugging problem, we want you to con- the preheat train that were historic leakers, so safety was
sider a recommendation of operating guidelines, which the main driving force for us to replace the exchangers.
is to have frequent checks on catalyst fines removal system. We did see an increase in hot end approach temperature,
Insufficient fines removal may result in valve plugging typically about 40-50°F now versus 110°F with the other
problems including regenerator screen plugging. shell and tube exchangers. The pressure drop through the
According to our unique experience, lower temperature preheat train has also been reduced significantly from 50
operation in regenerator section would be very helpful. pounds down to about 20 pounds across the Packinox.
High temperature exposure during the regeneration usu-
ally could affect on the strength of catalyst support mate- JACOB:
rial. Lower temperature is the better. Our operating prac- Twisted tube design has proven to be very cost effective
tice is limiting the highest temperature in regenerator on a reformer revamp we were involved in. In this revamp,
section to under of 520°C. the existing shell could be reused, which required no
piping modifications. This design has a very low pressure
PROOPS: drop of the shell side and improved heat transfer effective-
We replaced all of our 1982 vintage Rockwell ball ness when compared with a conventional shell and
valves with Neles-Jamesbury valves within several years tube exchanger.
of startup. These proved much more reliable. We are still
not satisfied with the performance of these valves, and DONALD WALKER (Petro-Canada):
are now installing 4⬙ Argus ball valves (model FK76M) As part of a debottleneck of a semi-regen reformer we
in services with the highest failure rates (BV numbers 3, replaced the feed effluent exchangers with a Packinox
5, 54 and 56). We are very happy with the performance plate exchanger. The improved heat recovery and reduced
of the Argus valves. pressure drop allowed us to increase the capacity of the
unit by about 25% on a constant octane basis. Since
Question 28 commissioning, we have regenerated the unit 5 times and
What is the experience with Twisted Tube, Rod Baffle, have not experienced any operation issues or fouling.
Packinox and other exchangers for revamps and new
units on catalytic reforming units? ANGELO FURFARO (UOP LLC):
Packinox welded plate-type heat exchangers have
PROOPS: proven to be the most efficient and economical choice
In one of our reformer trim coolers, we replaced a for the combined feed exchanger service on Platforming
standard bundle with a twisted tube bundle, which units. Over 115 Packinox exchangers have been specified
decreased separator temperature and lowered pressure for this service and 77 of these are already in operation.

176 Light Oil Processing

TOC/INDEX

The advantages of the Packinox exchanger are lighter

weight, a compact design, and high energy recovery. One
Packinox exchanger can now handle up to 60,000 BPSD
capacity, while a traditional ‘‘Texas Tower’’ type shell-
and-tube (vertical single pass) exchanger is limited to
approximately 15,000 BSPD throughput before parallel
exchangers have to be provided. As a result, the total
installation cost of the Packinox exchanger is lower than
the shell-and-tube type exchanger of comparable capacity.
For the past 10 years, almost all new UOP CCR Platform-
ing process units have used Packinox-type exchangers.
For revamp applications, Packinox exchangers are used
to replace the existing shell-and-tube exchanger to reduce
the pressure drop and to increase the heat recovery, often
avoiding costly revamping of the charge heater and prod-
ucts coolers. In some revamp situations, the shell-and-
tube type ‘‘Texas Tower’’ exchangers are added in parallel Also I must correct what was said a minute ago. I was
to the existing exchangers to increase throughput. at BP Amoco’s Texas City refinery earlier this week, and
More than 25 Packinox exchangers are also in operation they have confirmed reaching 70,000 barrels per day in
in other refining and petrochemical services, including their Packinox. I believe that it is a record flow rate for
naphtha hydrotreating. a single shell heat exchanger.
Rod baffle-type combined feed exchangers have been The 81 units in service have cumulatively logged 410
included at client request in the design of four new CCR years of operation. Almost exactly half of those years are
Platforming units over the last 18 years. Three of the in the continuous regeneration modes, the other half in
four units’ exchangers have experienced significant cross semi-regenerative mode. I have done a quick calculation
leakage problems, requiring the first unit to replace the of the overall out of service rates due to shortcomings of
tube bundle twice, the second unit to replace the bundle Packinox, and I come up with 0.13% or an average of
once, and the third unit to plug several leaking tubes. 12 hours per year.
This first unit is currently replacing the Rod baffle Most interventions on Packinox exchangers have been
exchanger with a Packinox exchanger. The third unit is to clean liquid feed injectors and upstream liquid feed
in the process of replacing their two parallel bundles with filters as somebody had mentioned earlier. We have also
new, improved rod baffle-type bundles. The fourth unit had to clean out salt deposits, construction debris, iron
has been on-stream since 1996, and no leakage has sulfide deposits, migrated catalysts and just about anything
been reported. that can come through process lines at one point of
Twisted tube heat exchangers are a relatively new appli- another.
cation in platforming units, and the operating data is still We have had to perform repairs on 9 bundles, usually
limited. The twisted tube design has, to our knowledge, in the form of rewelds or channel plugging. Finally, we
only been used for one bundle replacement revamp proj- have replaced one bundle on site and replaced 2 first
ect. The advantage of the twisted tube exchanger in a generation exchangers with our current third generation
revamp situation is that it allows the exchanger to process design. The third generation was introduced in the early
more throughput while retaining the existing exchanger 1990s and constitutes the bulk of our references.
shell, thereby eliminating the need to modify the structure To complete the picture, I must say that two Packinox
and the piping. The first revamp installation was imple- exchangers have been idled due to refinery closure.
mented in late 1998, and the thermal performance Now, what should you do to ensure a good operation
improved by approximately 15%. of your Packinox in catalytic reforming?
● First, perform a good review of operating procedures
FRANCOIS REVERDY (Packinox, Inc.): and a good training of the operators. We have a
As of June of this year, which is the last time I took training crew that does that quite effectively.
a count, there were 118 Packinox combined feed effluent ● If you retrofit an older Reformer, a good HAZOP
heat exchangers in catalytic reforming service around the review makes sense. In particular, if the last reactor
world plus another 50 some in other services. Of those is quite old and its outlet screen questionable, we
118, there were 81 in operation and another 37 at different recommend installing a back up screen inside the
stages of fabrication and commissioning. The operating effluent transfer line between reactor outlet and
range goes from about 6,000 barrels per day to over Packinox inlet. Also, take time to ensure that your
65,000 barrels per day. liquid feed filters are of good quality and do not

1999 NPRA Q & A Session on Refining and Petrochemical Technology 177

TOC/INDEX

allow mischievous bypass—We have often seen this. ● Earlier we were taking feed from the storage tanks.
Some clever operators have even found a way to Now the feed to the unit comes directly from the
back flush their liquid spray bars on line, with only upstream units. This eliminates the chances of any
15 minutes production loss. oxygen ingress in the feed thus reducing the chances
● Finally, verify your nitrogen content in the feed to of sulfolane degradation.
make sure that you will not reach crystallization ● The temperature of the heating media was reduced
temperature inside the plate pack, because salts as the degradation of the sulfolane is faster at higher
would otherwise deposit over time. temperature.
● Use of micro-filter in the circulating solvent system,

C. Aromatics Extraction thereby removing the degraded material on a contin-

uous basis.
● The problem of the corrosion was found to be very
Question 29 much acute in the solvent recovery column reboiler
What methods have been used to cost effectively remove and the connected piping. Hence the metallurgy of
chlorides from an aromatic extraction solvent? the re-boiler tubes has been changed from carbon
steel to stainless steel and a portion of the shell also
STYNES: had been cladded with stainless steel material.
Chlorides do not represent a problem in our extraction ● As far as corrosion inhibitors are concerned, we are
unit. That is because we remove the chloride salts from still using MEA to control the pH value of the
the solvent by semi-batch distillation of a slip stream off sulfolane. No other chemical could be tried so far.
the solvent circulation circuit. If any other suitable chemical has been tried else-
where, the same can be tried in our units also and
KRISHNA: accordingly feedback will be shared.
I understand that ion exchange worked quite well in One of the research institutes in India has done a pilot
one case where there was a chloride contamination of the plant study on sulfolane regeneration wherein a slip stream
solvent. The solvent was tetraethylene glycol, and the of the circulating sulfolane is continuously regenerated
levels were reduced from 1,000 ppm or so down to 50 by passing it through a packed bed for removing the
ppm. degraded portion of sulfolane. The process is not yet
commercialized.
Question 30
What is the expected corrosion rate for equipment in a ANGELO FURFARO (UOP LLC):
Sulfolane circulation system with carbon steel vessels? The expected corrosion rate in a sulfolane circulation
Which vessels have experienced the highest corrosion system with carbon steel vessels is low. However, when
rate? What has been the experience with corrosion inhibi- oxygen is allowed to contaminate the system the corrosion
tors other than MEA? rate can become very high at certain locations in the plant.
If an operator were to plan to use a feed known to be
PARK: contaminated with oxygen, then the rate of corrosion
We do not have any serious corrosion problem in the could be expected to be high, 100 mils per year or more
Sulfolane circulation system, but stripper bottom corro- at the worst locations in the unit. Such corrosion rates
sion is relatively high. The expected corrosion rate for are generally unacceptable.
the stripper bottom shell is 0.4 mm per year. We have Typically corrosion is linked to units with significant
used only MEA as the corrosion inhibitor. solvent degradation and high velocities in the equipment
as a result of running significantly over capacity. Locations
STYNES: of high corrosion are also concentrated near the areas
We have the same experience, very low corrosion rates of highest temperature and where solids tend to collect.
and we have only used MEA. The highest corrosion rates Reboilers of the stripper column and recovery column
we do have are in the overhead vapor condensers and reflux are targets for corrosion due to the surface temperatures
accumulators and those can run up to 5 mils per year. and the tendency of the circulating material to have a
concentration of solids. Associated piping is also a com-
G. G. TANDON (Bharat Petroleum Corporation Limited): mon location for corrosion.
We have two aromatics extraction units in our refinery. MEA is usually used as a neutralizing agent more than
One is for production of benzene and toluene and the as a corrosion inhibitor. Some operators have replaced
other one is for the production of food grade hexane. MEA with DEA due to handling concerns and some have
The sulfolane corrosion problem has been resolved to used a mixture of the two. Our information indicates that
great extent in our units by using the following techniques. these materials work in an equivalent fashion. There have

178 Light Oil Processing

TOC/INDEX

been experiments using inorganic neutralizing agents but KRISHNA:

the concern about adding to the circulating solid load This composition of this poly gas is from one of our
tends to steer operators away from these agents. refineries. It is 98% olefins, 132 bromine number, and
specific gravity of about .734. The ASTM distillation
D. Polymerization initial boiling point is around 99, 50%, 272, and end
point is about 309°F. The carbon number distribution
in wt%: C6 minus is 3.9, C7 is 2.4, C8 is 7.1, C9 is 79.9
Question 31
and C10 plus is 6.7.
For polymer gasoline units, what is the typical product
component distribution? What types and concentrations
PROOPS:
of ‘‘by-products’’ are typical?
According to Oil and Gas Journal’s annual refining
DAVIDSON: survey, there are 25 units operating in the U.S., with a
The standard answer is the product composition is total capacity of 65 MBPD.
feed dependent, of course. I actually do have some differ-
ent feed product distributions here from propylene up HIGGINS:
through butylenes. Based on the 1989 NPRA gasoline survey, published
Polymer gasoline composition is influenced by feeds- in 1991, the following polymer gasoline quality was
tock and operating conditions. Typical component distri- reported, which provides some indication of component
butions for polymer gasolines produced from propylene, distribution.
butylene, and mixed C3/C4 feedstocks are summarized
below. The olefin content of polymer gasoline is about Parameter Value Parameter Value
98%. 10% 149°F Gravity 68.0°API
30% 223°F RVP 9.1 psi
Feed Propylene C3 /C4 Butylene 50% 236°F Olefins 94.6 vol%
70% 296°F
Gasoline Composition, wt%
90% 346°F
C6 6 4 1
C7 3 26 3
C8 5 22 76
C9 54 20 2 E. Isomerization
C10Ⳮ 32 28 18
RON 93.5 95.5 99 Question 32
MON 81.5 82 84 Has anyone experienced a runaway on a butane or pen-
tane/hexane isomerization unit? What caused it, and how
Our poly plant produces a lighter type of gasoline as severe was it? Is process configuration a contributor
compared to FCC gasoline (e.g., recycle versus hydrogen once through)? What mea-
Typical distillations are: sures were instituted to prevent a recurrence?
10% 100°F
JACOB:
50% 150°F
90% 330°F A runaway was experienced in our butane isomerization
unit, caused by a complex series of events within the
Introduction of any feed contaminants can cause the refinery. The root cause of the runaway was determined
following products to be produced across the catalyst: to be a leaking feed control valve.
Scenario: Operations experienced a loss of DCS console
1) H2S Ⳮ C3 ⳱ 䊳 C–C–C–S–H due to failure of a back-up power supply system. A circula-
Hydrogen Olefin Mercaptan tion procedure was invoked due to the loss of the control
Sulfide
console. This included bypassing the reactor section and
2) CH3SH Ⳮ C3 ⳱ 䊳 CH3S–C3H7 routing feed directly to the stabilizer. The feed control
Mercaptan Olefin Thioether valve was not blocked in, resulting HC reaching the isom
H catalyst. The indication of the excursion was evident in
3) HC ⬅ CH Ⳮ O \
CH3 C ⳱ O
O2
CH3COOH
2
catalyst
䊳 䊳 the accelerated temperature rise in the center TI’s of the
Acetylene Oxygen Acetaldehyde Acetic Acid lead reactor. A 1 degree/minute rise was noted by opera-
tions personnel. Temperature of center TI’s reached a
4) HC ⬅ CCH2 Ⳮ O2 䊳 CH3CH2COOH
catalyst maximum temperature of 900oF. Attempts to manually
Methylacetylene Oxygen Propionic Acid depressure the unit were delayed due to a plug in a 2⬙

1999 NPRA Q & A Session on Refining and Petrochemical Technology 179

TOC/INDEX

manual relief line. The plug released after the valve was
open for a few minutes, and the unit depressured. A
temperature excursion was never seen in the lag reactor.
Due the circumstances under which this event took
place, no relationship was seen between the process config-
uration and the event.
The circulation procedure was modified to block in
the feed control valve whenever the reactor feed is bypassed
to the stabilizer.

KRISHNA:
We have seen runaway on both the butane and pentane-
hexane unit. Both incidents were on hydrogen once
through units. In the case of the butamer unit we believe
the unit was hit with heavy ends on a relatively new
catalyst at low feed rates. We think hydrocracking of the low space velocity. So we watch that fairly carefully. Also,
heavy end led to the runaway. the shift towards bezene destruction in the isomerization
On the pentane-hexane unit the incident was the result units has caused us some problems. Poor naphtha splitter
of cutting feed back to the reactor after a brief outage operations or upsets upstream may cause higher levels of
without cooling the reactor to below 150°F. Of course, the C7 plus material in the feed.
we have now reinforced the procedures and the training With regard to measures to prevent a reoccurrence,
so that operators do cool the reactor before they reintro- we have instituted minimum gas and liquid flow rate
duce feed. guidelines to stay above the LHSV minimum targets plus
an emergency response hierarchy as a function of the
MAYO: maximum bed temperature. If quench is unavailable, the
Akzo Nobel supplies amorphous catalysts for both of two most important considerations during the excursion
these services and we are not aware of any exotherms when are pressure reduction, so you can reduce the vessel stress
using our materials. Where we have heard of exotherms as well as slowing down the hydrocracking reaction rate,
occurring, it is primarily a result of feed properties, i.e. and to increase liquid feed rate. The gas rate is generally
very reactive feeds, high olefin or benzene content and minimized or stopped altogether. One other thing that
as Mr. Krishna pointed out, high end point materials. we do is bypass the feed effluent exchangers. That helps
Flow maldistribution and very low space velocity appear to mitigate the heat a little further.
to be a common element as well. If you do have to
drop your space velocity below half the design rate, we DAVIDSON:
recommend recycling the stabilizer bottoms to keep the We also experienced a temperature runaway one time
mass flux up. Also, in our experience it is uncommon to on a pentane/hexane isom after a FCC startup when the
have these types of problems with units that recycle FCC gasoline was inadvertently slopped for longer than
hydrogen. normal. We attribute the runaway to that excess olefin
coming back at us. The reactor temperature red lined at
STYNES: 1000°F. The good news is there was no damage other
We have had a couple of runaways on a pentane isom than some paint scorching.
where we suspected ring opening of cyclopentane. We
could not validate that, but since we did not find any BRIAN JOHNSON (UOP LLC):
aromatics or olefins in the feed at that time, we implicated I think the panel has discussed a good diversity of types
the cyclopentane. We have had instances of high olefin of units. They have zeolitic units, amorphous units, and
levels in butane isom units that caused temperature spikes C4, C5, and C6 units. All of these types of units have
and have had spikes due to aromatics in a C5/C6 isom experienced temperature excursions for various types of
unit. We have a maximum spec on benzene in our C5/ catalysts. It is not necessarily only related to operating
C6 isom feed to avoid that. temperature, with the amorphous units running well
below the temperatures of the zeolitic units.
VERRENKAMP: We have found that the primary root cause of tempera-
We have experienced this on pentane/hexane units but ture excursions is having a reactor that is too hot for the
I would not call them runaways. We like to call them amount of feed that is being processed through it. One
excursions. The causes, as Mr. Stynes point pointed out classic example was given by the panel, and this has hap-
before, are usually a poor flow distribution as a result of pened in other instances, where a feed control valve was

180 Light Oil Processing

TOC/INDEX

blocked in but another line that was bypassing that control Question 34
valve allowed a small amount of hydrocarbon to enter What control variable and limit (e.g., maximum exoth-
the reactor. Maldistribution occurs in the warm reactor erm) is used in applications where benzene is saturated
and the hydrocarbon starts to react and forms a local hot in an isomerization unit?
spot, which then can initiate further reaction with the
hydrocarbon that is coming in the reactor. Operator train- VERRENKAMP:
ing is very important, which helps assure that the proce- Most of this data comes from UOP design. Benzene
dures are accurately followed. is present in varying levels in virtually all isom unit feed-
stocks. They tend to boil with the C6 paraffins and slightly
Question 33 below the heavies which is the cylcohexane. Fractionation
What has been the commercial experience with the newer capability naturally determines the amount of C7s in the
high activity isomerization catalysts? feedstocks. So with units operating with benzene levels
of less than about 5% usually there are no modifications
MAYO: necessary on the unit. At about 5%, the reactor delta T
Akzo Nobel has recently introduced improved amor- is approximately 100°F. That is about the maximum delta
phous isomerization catalysts, AT-2 and AT-2G. We have T that UOP likes to go to without further careful monitor-
observed, in C5/C6, units about one full octane number ing of the heat removal capability of the exchanger train
improvement and for C4 units a 2 to 4 number improve- and the temperature control system. Depending on the
ment in the isobutane ratio at equal yields. Our customers LSHV and the catalyst activity, care needs to be taken
have taken advantage of this higher activity by increasing that the high exotherm produced by the benzene in the
feed rate or severity or in some cases even short loading feed does not push the reactor temperature above the
the reactor to save costs with no debit in performance. optimum equilibrium temperature for the reactions.
We tend to avoid using benzene feed as the controlling
parameter at BP Amoco. A major issue has been control-
STYNES: ling feed C7 plus content without online analyzers while
We have I-8 in our butane isom and I-7 in our pentane trying to make low benzene content gasoline products.
isom unit. Compared to previous catalysts, we have slight
increases in product ratio at much lower temperatures.
KRISHNA:
Our feeds are extremely clean, low water, sulfur and fluo-
We monitor the exotherm across the top active bed as
rides. So we have not really tested the ability to handle
well as the outlet temperature. We have one refinery where
contaminants.
the feed benzene content is about 6% and the maximum
exotherm is at about 100°F. We have another plant where
BRIAN JOHNSON (UOP LLC): the benzene content is higher, 9 or 10%, and there we
UOP commercialized LPI-100 catalyst in 1996. This control to about 90°F. We also recycle isomerate to control
is a high activity catalyst that is a replacement for the exotherms by dilution.
high temperature zyolitic catalysts. LPI-100 has been Now, before somebody points out that it is not prudent
installed in five different units and two other units are to recycle isomerate because it is bad for isom catalytic
now in construction. activity, I just want to point out that the driver there is
UOP commercialized I-80 catalyst in 1998. I-80 is really to remove the benzene in order to meet the CARB
the highest activity Isomerization catalyst commercially Phase II specifications and not so much octane improve-
available. Since that time 11 units have been loaded with ment through isomerization. It is sort of a California thing.
I-80 in both C4 service as well as C5/C6 service.
In several of the C5/C6 units, the feedstock is a very MAYO:
difficult one to process with heavies in the range of 25 Just to add, we have observed maximum exotherms of
to 30%. The I-80 catalyst has also undergone several 120°F in streams with 5 to 6 wt% benzene. That is on
sulfur contamination incidents and has recovered from a C5/C6 unit with amorphous catalyst.
the upset well. Even with these difficult feeds, the catalyst
stability has been very good. In fact, no I-80 has been
STYNES:
changed out of any unit to date.
I guess we are the guys who take the conservative
approach. In our one C5/C6 isom unit, we have a feed
PROOPS: splitter and we control the amount of benzene in the feed
We have loaded I-80, and are quite happy with it. to keep the delta T between 40 and 50°F in the top bed
Unfortunately we found out that it is not any more sulfur of the reactor. That correlates with running 2 to 3%
tolerant than I-8 was. benzene in the feed.

1999 NPRA Q & A Session on Refining and Petrochemical Technology 181

GASOLINE PROCESSES

Alkylation

Question 31
What is the current status of solid bed alkylation technology? For those working on this
technology, how does it compare in yield, octane, capital and operating costs with current
alkylation processes?

John Gieseman (ABB Lummus Global)

ABB Lummus Global, Albemarle Catalysts, and Neste Oil have successfully operated a
demonstration unit for nearly three years. The yield, octane and other key parameters are all at
or better than the conventional liquid technologies. The ISBL capital to build these units is on
par with HF units and significantly lower than Sulfuric units. Recent improvements to the
catalyst have reduced operating costs to levels lower than the liquid technologies. Because of
the inherent advantages of the solid catalyst, reliability is expected to be better than the existing
acid technologies as well.

Question 33
API Recommended Practice 751 calls for all joints in HF acid service to be inspected at least
once every ten years. How do you comply with this recommended practice? Have you used non-
destructive testing techniques to avoid breaking and re-making flanges?

Kurt A. Detrick and Robert Dolejs (UOP LLC)

To start, “A ll joints” could refer to flanged, threaded, and welded joints, but the question is
specific to flanged joints.

Flanged joint inspections are covered in paragraph 3.4.3.4 in Section 3.4 Inspection of
Commissioned HF Unit Equipment in API-751. It states that “the inspection frequency of
flanged joints should consider the respective corrosion rate in conjunction with the calculated
sealing surface requirements. In the absence of inspection data, all flanges in Main Acid service
should be inspected every ten years, and all flanges in Trace Acid service every fifteen years.
For flanges in the same process circuit, the findings from one flange inspection may be applied
to other flanges of equal size and age.”

With that said, compliance is maintained through a proactive flange inspection program that
completes scheduled inspection on time. This program identifies all piping circuits and pressure
vessels in main and trace acid service. Each flange is given an identifier in the program. The
inspection planning process creates a schedule for mandated inspections and identifies
preparation and access requirements. This information is used in turnaround scope development
and maintenance planning. Cost of the program can then be tracked. Program compliance is
demonstrated by tracking and documenting inspections. Updating inspection plans is also
critical to the success of the program to maintain compliance with API-751 and adjusting
frequency as needed.

11
With the frequency of inspection established the task of scheduling needs to be addressed. For
example, if turnarounds are typically scheduled every 3 years, then one-third (1/3) of the Main
Acid service flanges would be inspected every turnaround along with one-fifth (1/5) of the Trace
Acid service flanges.

As far as using nondestructive testing to avoid breaking flanges we have heard that there is an
ultrasonic based technology commercially available to detect flange corrosion while a unit is on-
stream, but we have no actual experience with this technology. At this time, visual inspection of
the opened flange is the only technique we have experience with for inspection of the seating
surfaces of the flanged joint.

Although not addressed in the question, but something we feel that needs to be discussed here is
the inspection of small bore non-butt welded piping. This is specifically covered in paragraph
3.4.3.2 in API-751. It states that “threaded joint fatigue as well as corrosion product
accumulation in inactive branch connections require more frequent assessment. Inspection
programs should also include NDT provisions to examine and ensure the integrity of these small
bore non-butt welded piping circuits. A representative sampling of these joints should be
radiographically examined every 5 years. Owner-operator data should be used to increase or
decrease the inspection interval within the limits of the applicable standards. Profile radiographs
can be used to determine the condition of threaded joints including thread engagement and seal
weld coverage of exposed threads in seal-welded joints. Disassembly of non-seal welded joints is
an alternative to radiography.”

Ethanol

Question 37
Discuss how ethanol blending requirements have impacted refining operations including: 1)
blend formulations; 2) octane balance; 3) driveability specifications; 4) lab testing and
procedures; and 5) reformer severities.

William H. Keesom (UOP LLC)

For those directly involved in ethanol blending, the following is well known, but as a general
overview of approaches being taken, the following information is offered:

The question of ethanol blending should be considered for two different situations: as a
replacement for MTBE in RFG; as a new blendstock in conventional gasoline. When ethanol
replaces MTBE in RFG there is less impact on gasoline and refinery operations than when
ethanol is a new blendstock in conventional gasoline.

The Energy Act of 2005 requires an increase in biofuel use. The act requires a gradual increase
in biofuel use from 4 billion gallons in 2006 to over 7 billion gallons in 2012. These numbers
translate to around 260 MBPD of biofuel in 2006 and 470 MBPD of biofuel in 2012. Most of
this biofuel will be ethanol.

MTBE is being phased out of US gasoline because of bans on using MTBE and product liability
concerns from blending MTBE in gasoline. Most MTBE is used in RFG, which comprises

12
around 30-35% of the gasoline consumed in the US. The amount of ethanol needed to replace
MTBE in RFG is around 300 MBPD. With the removal of the oxygen requirement in RFG as a
result of the Energy Act, ethanol is not required for RFG. However, ethanol is a logical
replacement for MTBE. It has similar properties to MTBE and using ethanol in RFG avoids
major change in RFG blending when MTBE is removed. While the major outlet for ethanol is
currently RFG, the provisions of the Energy Act call for increased biofuel (ethanol) use which
will require blending more ethanol into conventional gasoline. Market size and logistical costs
will likely determine which conventional gasoline areas will blend the additional ethanol needed
to meet the requirements of the Energy Act.

Ethanol containing gasoline cannot be shipped in conventional gasoline distribution systems

because these systems often contain water and cause separation of ethanol from gasoline. As a
result, ethanol is generally blended into gasoline at the final terminal before transport to the
service stations and is not blended into gasoline at the refinery. Terminal blending of ethanol
requires investment in ethanol unloading, storage, and blending facilities.

The impact of ethanol blending should be considered in two situations:

Ethanol as a replacement for MTBE in RFG:

• Blend formulation: There will be a small impact on volume – MTBE typically is used at
around 11 vol% and ethanol at 10 vol%. Because ethanol adds around 1 – 1.3 psi to a
gasoline blend it may be necessary to remove some light components to meet RFG VOC
limits. There is no RVP waiver for ethanol in RFG.
• Octane: Ethanol has slightly higher octane than MTBE but the impact on gasoline octane
from replacing MTBE with ethanol is small.
• Reformer severity: Because ethanol and MTBE have similar octanes, expect a small
impact on reformer severity when ethanol replaces MTBE in RFG. Driveability: ethanol
is reported to somewhat worsen the driveability of gasoline. A new driveability formula
under consideration by ASTM includes a calculation of the impact of ethanol on
driveability.

Ethanol as a new blendstock in conventional gasoline:

• Blend formulation: Ethanol will likely be blended at 10 vol%. If it is a new blendstock
in conventional gasoline, this level of ethanol use will have significant impact on
gasoline blending. Gasoline volume will increase. As a result of its higher octane and
different distillation profile there will be some change in gasoline blend recipes when
ethanol is used. Because conventional gasoline is allowed a 1 psi waiver when blending
ethanol, there will be little or no need to blend around the RVP added by ethanol.
• Octane: Ethanol will add around 2 to 3 more octane numbers which will allow refiners
to blend lower octane streams.
• Reformer severity: Refiners will be able to reduce reformer severity by a few numbers
when ethanol is a new blendstock in gasoline. Expect reformate yield to increase but
hydrogen yield to decrease.
• With the RVP waiver and less stringent RVP specs, this may be the more attractive
place to blend ethanol.

13
GASOLINE PROCESSES

Process Safety

Question 50
What is the proper firefighting media to use when putting out a fire when both spent sulfuric acid
and heavy hydrocarbon are present (e.g. in a spent acid tank or a diked area which has a layer
of hydrocarbon floating on the spent acid)?

Ed Lowe (Pasadena Refining Systems)

When question was posed to several people, the predominant response recommended using a 3%
Alcohol-resistant aqueous film forming foam (AR-AFFF). Aqueous film forming foams (AFFF)
are water-based foams, frequently containing sodium alkyl sulfate, and/or perfluoro telomer as
surfactants. They have the ability to spread over the surface of hydrocarbon-based liquids.
Alcohol-resistant aqueous film forming foams (AR-AFFF) are foams with the properties of AFFF
foams with additives to make the foam resistant to the action of alcohols.

In general discussions with Gasoline Panel members, it was disclosed that actual use of AFFF
foam did not work well in real-OLIHVLWXDWLRQVVLPLODUWRWKHTXHVWLRQ¶VVFHQDULR7KHILUHNHSW
lighting off because it was difficult to establish a foam blanket. Panelists recommended
considering use of carbon dioxide.

There is a type of foam being used in Europe that should be considered for this particular
application. The foam, which is a vapor suppressing foam, is intended for use on spills of acid
materials.

70
Question 51

Reforming unit stabilizer column top trays and overhead condensers can experience fouling with
ammonium chloride salts which are commonly removed by on-line water washing of the column
overhead. What practices do you employ to reduce the risk of rapid corrosion and the potential
failure associated with this fouling and subsequent water washing procedure?

Pedro Fernandez (Jacobs Consultancy)

Ammonium chloride salts deposition is a common problem in many naphtha hydrotreating and
naphtha reforming units. Problem areas include reactor products condenser areas, compressor
inlet areas and the overhead section of the product stabilizer.

Ammonium chloride salts form because of the presence of ammonia and chlorides in the reactor
effluent stream. These compounds combine and precipitate as the streams are cooled and the
effective concentration of the compounds increases in the gas phase.

The temperature at which these salts will start depositing will depend on the ammonia and HCl
FRQFHQWUDWLRQEXWLQJHQHUDOLWLVSRVVLEOHWRVHHVDOWGHSRVLWVDVWHPSHUDWXUHVGURSEHORZÛ)

Once formed, the ammonium chloride salts are hygroscopic, and readily absorb water. Once in
contact with water, they form a very corrosive acidic solution, which if not managed properly
will lead to extensive under-deposit corrosion in the areas where salts have deposited.

Good management of ammonium chloride deposits and corrosion starts with preventing the
formation of compound. The two problem compounds are ammonia and chlorides.

Chloride is a necessary component of the reforming catalysis. It helps maintain surface

GLVSHUVLRQRIWKHNH\PHWDOIXQFWLRQVLQWKHFDWDO\VWDQGWKHUHIRUHFDQ¶WEHHOLPLQDWHG

Ammonia is formed from the hydro-denitrification of organic compounds present in the feed.
Proper pretreatment of the feed is the most important strategy to minimize the formation of the
ammonia chloride salts. In general, reforming licensors and catalyst vendors recommend that the
feed to the reforming unit should contain less than 0.5 ppmw nitrogen. We consider this to be
the minimum standard of care. Many operators target values of less than 0.2 ppmw nitrogen in
the reformer feed; we would encourage people to push their hydrotreating severity to achieve at
least this level.

While it is relatively easy to meet very low levels of nitrogen with most conventional straight run
naphthas, the story is different when processing thermally cracked materials, naphthas derived
from heavy oil processing units, or from very high nitrogen content crudes.

One can calculate how much ammonium chloride will be formed in the reactor based on the
amount of nitrogen in the feed. As a reference point, a naphtha reforming unit processing 20,000

71
BPD of naphtha with a nitrogen content of 0.5 ppmw will produce about 4,700 lb of salts (almost
one hundred 50 lb sacks). This volume of salts will deposit and accumulate somewhere in the
unit if unmanaged.

Several techniques have been used to mitigate or eliminate the formation and depositing of salts
once the nitrogen has entered the unit:

The most effective technique²other than removing the nitrogen in the feed²is the Installation
of chloride guard systems on the product separator liquid upstream of the stabilizer. By
eliminating the chloride component, formation of the salts is eliminated.

Two types of systems have been used in this service: one is a scavenging solid adsorbent bed;
several vendors including UOP, Sud-Chemie, and others offer materials that can be applied in
this service. A second alternative is an alkaline (caustic) wash of this stream; Merichem has
installed an alkaline wash-contactor system in a couple of units. Units with both types of
systems report good results.

Some refiners try to adjust processing conditions to minimize the salt formation in the stabilizer.
In our opinion, however, as long as the ammonia and the chloride are present, these changes only
move the problem elsewhere²typically to the overhead condenser.

If the salts are formed and the refinery is forced to wash them out, the water washing technique
used must include adequate precautions to avoid leaving wet salt deposits behind, since as noted
earlier this could lead to severe under-deposit corrosion. The on-line washing technique that we
are familiar with and has been used at several refiners involves using a large amount of water
and slumping the tower during the process:

- Inject a large amount of water in the stabilizer reflux line and reduce the reboiler
temperature so that the water comes out with the bottom product;
- Monitor the difference in the water conductivity between the inlet and outlet to determine
when there are no more salts in the column;
- Reduce the water rate and increase the reboiler temperature so the water comes out on the
overhead drum, wash the overhead condenser and receiver (again using the change in
water conductivity as an indication of when the wash has been completed);
- Be very diligent in draining all water boots and low point drains.

Tina Drumheller, Frontier Refining

Primary locations for plugging in reformer stabilizer systems are the feed to the stabilizer tower
and feed/bottoms exchangers.

Typically, separator level is the first indicator of plugging. When plugging occurs, water
injection is usHGWRPLWLJDWHSUHVVXUHGURSLVVXHV:DWHULVKDUGSLSHGYLD´OLQHWRWKHWRZHU

72
feed, and a small amount of water is used to relieve plugging in the line. Caustic injection/water
wash upstream of the feed coalescer and neutralizer injection in the overhead are used to mitigate
corrosion. If a significant tower plugging problem is encountered, water wash is utilized. The
tower is slumped and large volumes of water are used to flush the salts in the tower/overhead
system.

Paul Fearnside (Nalco Energy Services)

Nalco recommends a procedure similar to what is being done successfully to mitigate salt fouling
in the FCCU main fractionator (see question # 16). It is possible to severely reduce the frequency
of the water washes by using a salt dispersant. The salt dispersant also has the added benefit of
UHPRYLQJPRVWRIWKH³ZHW³FRUURVLYHVDOWVWKDWXVXDOO\UHPDLQDIWHUWKHLQWHUQDOZDWHU washing
procedure is completed.

Malcolm Sharpe (Merichem Chemicals & Refinery Services LLC)

Merichem has been licensing CHLOREXSM Technology since 1990 and currently has four (4)
licensed units in operation. CHLOREXSM XQLWVXWLOL]H0HULFKHP¶VSURSULHWDU\),%(5-FILMTM
Contactor to non-dispersively contact stabilizer feed with an alkaline water solution to
effectively remove HCl. Since the FIBER-FILMTM Contactor is non-dispersive there is no need
to separate droplets out from the continuous hydrocarbon phase and as a result carryover and
carryunder (and their associated drawbacks) are not an issue. Instead the aqueous phase
(alkaline water in this case but also commonly caustic, amine and wash water) adheres to the
fibers while hydrocarbon flows in between the wetted packed fibers. The result is the creation of
immense interfacial surface area providing superior mass transfer without dispersion. For more
information, an article reprint from Hydrocarbon Technology International and entitled
³(OLPLQDWLQJ&KORULGH&RUURVLRQDQG)RXOLQJRI5HIRUPHU6WDELOL]HU&ROXPQV´FDQEHIRXQGRQ
0HULFKHP¶VZHEVLWHwww.merichem.com) under the Refining Technology/Product Literature
tab.

Question 52
Have you found highly condensed aromatics (i.e. red oil) around the reforming unit, especially
around heat exchangers and/or valve leaks? What safety precautions do you recommend for
handling this material?

Jay Ross (Axens NA)

The PNA material formed is in the coronene (C24H12) or ovalene (C32H14) family of fused
aromatics with a characteristic red/orange color. It generally is found in cold spots in the process

73
and can lead to plugging and pressure drop problems. The material is only really soluble in
benzene and hence cleaning is by mechanical means with the waste treated as hazardous.

Others have commented on finding this material around the recycle compressor and the stripper
reboiler when heated by reactor effluent (1996 Q&A). Operation under severe conditions of
high temperature, high EP feed and low H/HC promotes formation of this toxic material.

Coronene: C24H12, Tb=525°C, Tmp=440°C

Alkylation

Question 53
In a hydrofluoric acid alkylation unit, what can you do to prevent plugging in the acid-soluble
oil caustic neutralizer?

Ed Lowe (Pasadena Refining Systems)

Up until 2004 PRSI (then Crown Central) used a caustic (sodium hydroxide, NaOH) solution to
neutralize acid soluble oil (ASO) and experienced the same problems as suggested by this
question (i.e. ± plugging in the ASO neutralizer). When the switch was made to using a
potassium hydroxide (KOH) solution, the plugging problems were significantly reduced.

Currently, we use a KOH solution starting at strength 30% w/v. The KOH solution is
recirculated in the neutralizer unit until it is depleted to 1% w/v strength. When the 1% w/v
strength is reached [the strength is monitored by periodic in field testing by operators], the
neutralizer recirculating solution is dumped to a holding tank, and the neutralizer is recharged
with 30% w/v KOH solution.

Originally Crown Central had a spent KOH regeneration system process on site. By reacting the
spent KOH solution with a lime (CaOH2)) solution, KOH was produced along with the relatively
insoluble calcium fluoride salt (CaF2). The calcium fluoride was separated from the KOH
solution and returned to the ASO neutralization process. The calcium fluoride salt was sold.

74
On site regeneration of the KOH now has been discontinued. Instead the spent KOH solution is
regenerated off-site and returned to PRSI for use.

Tina Drumheller, Frontier Refining

Two of the common sources of plugging in acid-soluble oil (ASO) caustic neutralizers include
scale resultant of corrosion and salt formation due to neutralization. Of the two, salt formation
has been the more problematic.

Typically, one of two types of caustic is used in ASO neutralization: sodium hydroxide or
potassium hydroxide. The basic chemistry for both is straight forward acid-base neutralization

NaOH + HF -> NaF + H2O

KOH + HF -> KF + H2O

Once salt formation exceeds solubility limits in solution, the salt crystals will tend to
agglomerate as they settle out of solution.

The following are two means of preventing plugging

1. Maintain adequate circulation to prevent solids settling out.

2. For units with static mixers, ensure mixer design is adequate to ensure good mixing of
caustic with ASO at the injection location.

If a plugging problem due to salt formation occurs, it can be mitigated by either water wash or
steam out. If steam is used, stress cracking can occur with stainless steel in the presence of
caustic and high temperature.

,Q)URQWLHU¶VH[SHULHQFHVROXWLRQVRIERWK1D2+DQd 5% NaOH have been used. With

either solution system plugging at the mixer occurred frequently. Steam had to be used several
times to relieve plugging, resulting in accelerated corrosion rates.

Conversion to 10% KOH resulted in prevention of plugging episodes with no other system
modifications.

The increased operating expense associated with potassium hydroxide use was offset by
installation of a caustic regeneration system.

75
Question 54
In a sulfuric acid alkylation unit, what can you do to minimize foaming and/or plugging in
caustic wash or water wash systems?

Scott Alvis (Merichem Chemicals & Refinery Services LLC)

With conventional dispersive contacting devices such as static mixers, a conventional answer
would be to monitor caustic strength. If strength is too high, precipitation can occur which leads
to plugging. Reducing caustic strength requires a corresponding increase in circulation rate
which leads to foaming. The best way to avoid both problems is to use a non-dispersive
contacting device such as the FIBER-FILMTM Contactor. Because caustic and water adhere to
the fibers, there is greater surface area available for mass transfer without sacrificing the
concentration gradient driving force. By avoiding dispersions, there is less tendency to form
emulsions which also minimizes carryover and carryunder issues. In many cases, the FIBER-
FILMTM Contactor can be easily retrofitted onto existing separator vessels. For more information
on the FIBER-FILMTM Contactor, please visit www.merichem.com.

Question 56
In a sulfuric acid alkylation unit, there have been problems keeping the acid wash electrostatic
precipitator (EP) operational. What steps do you recommend to improve the reliability of the
EP?

Scott Alvis (Merichem Chemicals & Refinery Services LLC)

The sole purpose of the EP is to separate a dispersed phase from a continuous phase as a result of
using dispersive contacting devices such as a mix valve. The surest way to avoid EP operational
issues is to eliminate the EP altogether by using a non-dispersive FIBER-FILMTM Contactor.
Instead of dispersing the aqueous phase into the hydrocarbon phase as small droplets that are
difficult to separate, the aqueous phase acid adheres to the fibers while hydrocarbon flows in
between the wetted packed fibers. The result is the creation of immense interfacial surface area
providing superior mass transfer without dispersion, thus avoiding the need for an EP to reduce
acid carryover. For more information on the FIBER-FILMTM Contactor, please visit
www.merichem.com.

76
Question 59
For a hydrofluoric acid (HF) alkylation unit, what instrumentation do you recommend for
controlling HF acid levels throughout the unit?

Ed Lowe (Pasadena Refining Systems)

Generally, the use of nuclear level indications/controllers are gaining wide acceptance in the
industry. Several refiners have used nuclear levels in this service. External displacement level
transmitters or interface level sensors are also acceptable. You should discuss this issue with
\RXUXQLW¶VWHFKQRORJ\OLFHQVRU

Tina Drumheller, Frontier Refining

Frontier has used differential pressure (dp) cells, magnetic float, and nuclear level measurement
in HF alkylation units.

Both the dp cell and magnetic float level measurement became a reliability concern due to either
instrument failure or plugging requiring instrument removal for cleaning.

Of concern in use of both types of instrumentation was the potential for exposure due to
mechanical failure and maintenance requirements.

Nuclear level measurement has been successfully used in the following HF alkylation unit
locations:

Acid settler

Depropanizer overhead boot

Acid stripper reboiler

Flare drum

Benefits of nuclear instrumentation include

1. Decreased risk of personnel exposure to HF

2. More robust level indication during start-up
3. Increased reliability

The main operational concern with nuclear level indication is susceptibility to x-ray. Procedures
have been used to address the loss of nuclear level indication during periods when x-ray
inspection is conducted. For example, automatic level control is set to manual during x-ray
inspection that could interfere with nuclear level measurement.

77
Paul Hewitt (TRACERCO)
Due to the nature of the highly corrosive contents of a refinery alkylation unit careful
consideration must be given to selection of an instrument to measure liquid level and interface
between two immiscible phases.

The instrument selected must be reliable, as maintenance free as possible and ideally none
contact with the fluids to minimize potential leak points on a process vessel.

Nuclear gauge technology is one measurement technique that meets all of the above criteria and
is commonly used in alkylation process level control applications. The technology provides real
time data, no moving or wetted parts leading to reduced maintenance and high reliability, is
externally mounted and intrinsically safe / explosion proof.

Liquid Level Measurement

In the case of liquid level measurement a nuclear gauge is designed with a small radioactive
source on one side of a vessel that is housed in a shielded container. The container is designed
such that it emits a fan beam of electromagnetic radiation across the vertical length of
measurement range required. On the opposite side of the vessel a sensitive detector is positioned
within the fan beam. In the case of a level gauge, the detector will be positioned in a vertical
orientation across the span of level range required. If a level alarm is used this will be positioned
in a horizontal position at a specific vertical height. Figure 1 shows a typical arrangement for a
level gauge

Figure 1 Ȃ Typical Level gauge Arrangement

When the vessel is empty the electromagnetic radiation penetrates the vessel wall, across the
inside of the vessel and through the other wall providing a small radiation signal at the detector.

78
In the case of proportional level measurement, as the level of liquid in the vessel rises above the
minimum, it reduces the amount of radiation reaching the detector over an increasing length of
the vertical range. This causes a decrease in detector count rate. Using calibration data recorded
within the instrument the detector response is directly proportional to liquid level within the
vessel. The detector response is converted to a 4-20 mA and / or digital signal and relayed to a
control system.

Alarms allow low level or high level liquid positions to be detected. In the case of a high level
alarm the detector will measure a specific radiation signal when vapor is present between the
vertical position of the radiation source and the detector position. As the liquid level reaches this
vertical point the radiation is attenuated to background radiation levels. At this point, the detector
can be used to send a warning signal to the control room (High Level Alarm) or actually trip a
process activity. In the case of a low level this acts in the opposite manner. When liquid level
covers the radiation signal between source and detector the radiation reaching the detector is at
natural background levels. When the liquid level falls below this point there is a rapid increase in
radiation reaching the detector. In much the same way as the high level system, the detector will
send a warning signal to the control room (low Level Alarm) or actually trip the process activity.

In the case of nuclear gauge alarms it is very easy to test the systems on a regular basis. In the
case of high level alarm testing the operator simply has to shut off the radiation using the shutter
located on the shield. With zero radiation emitted from the shield, the alarm will act as if liquid
level has reached the alarm set point and will trip. In the case of a low level alarm a small
radiation test source can be manually positioned close to the detector. The increase in radiation
signal acts in the same manner as if liquid level has fallen in the vessel and a low level condition
has occurred. Due to the very sensitive detectors used, the test radioactive source is very small
and will cause no safety concerns to process operators.

Finally, in order to minimize radiation source use, it is common within an instrument design to
use one source of radiation and position a proportional level gauge across the measurement span
together with critical alarm detectors at alarm and trip points. All of the detectors utilize the same
radiation source.

Interface Level Measurement

Nucleonic interface gauges use a small radioactive source that is mounted internally through D´
QR]]OH XVLQJ D VHDOHG ´ SLSH WKDW ORRNV VLPLODU WR D WKHUPRZHOO  7KH UDGLRDFWLYH VRXUFH LV
transferred from the shield external to the vessel to the end of the pipe inside the vessel. An
external radiation detector is located on the outside of the vessel close by the internal pipe. The
source is designed so that it can be pulled through the pipe and back into the shield. This allows
the vessel to be entered with no safety concerns. Figure 2 shows a typical arrangement.

79
 Detector

Figure 2 ± Typical Interface gauge Arrangement

In the case of a proportional interface level measurement the radiation signal passes through the
´ SLSH ZDOO OLTXLG DQG YHVVHO ZDOO EHIRUH UHDFKLQJ WKH GHWHFWRU :KHQ WKH SDWK EHWZHHQ WKH
source and detector is filled with the lower density liquid (alkylate) the signal intensity at the
external detector is relatively high. When the higher density phase increases in height (HF
solution) the radiation intensity at the detector is reduced. Through calibration of the system
using the typical densities of both fluids involved, the interface position between the two
immiscible phases within the vessel can be measured and controlled.

In a similar manner to liquid level measurement, detectors can also be positioned at specific
vertical heights across the interface range to act as alarms. The detectors are tuned to alarm or
trip at specific radiation signal strength. In the case of a high level alarm, radiation will be
relatively high due to the less dense phase presence. If lower phase increases or an emulsion
band moves upwards through the vessel the detector radiation signal will reduce and a signal will
be sent to the control room warning of the condition. In the case of a low level alarm the reverse
conditions would apply with an increase in radiation signal due to lower density liquid presence
lower in the vessel.

80
Isomerization

Question 60
How do you detect leaks in an isomerization unit's steam charge heater? Have you been able to
detect a leak before a significant portion of the catalyst bed was deactivated?

Pedro Fernandez (Jacobs Consultancy)

This question is related to the performance of any light naphtha isomerization units operating
with a platinum on chlorinated alumina base catalyst. A common problem in these units is the
deactivation of the catalyst because of the ingress of water or any other oxygenated compound
that is transformed into water in the reaction section of the unit.

This is a familiar processing scheme: where the feed and make-up hydrogen gas are first dried
and then mixed and preheated before being sent to the charge heater up-stream of the reactor.
%HFDXVH WKH UHDFWRU LQOHW WHPSHUDWXUHV DUH ORZ W\SLFDOO\ EHORZ Û) PDQ\ XQLWV KDYH EHHQ
designed with a steam-exchanged charge heater. The use of steam as a heating medium saves
money and simplifies operations, but it opens the possibility of a cross exchanger leak that would
permanently deactivate the catalyst.

An old rule of thumb was that 1 pound of water would kill about 100 pounds of catalyst. While
different catalyst manufacturers may use different numbers, as a general rule this means:

10,000 Barrels of feed @ 225 lb/barrel @ 1 ppm of H20 = 2.25 lb of water

or 225 lb of dead catalyst.

This number is important because it puts in perspective our general distrust in the use of
analyzers as a safeguard to protect the isomerization unit catalysts. Small moisture changes lead
to rapid deactivation rates.

In this respect, we believe that minimizing the risks and exposure to charge heater steam leaks
should start with the proper design of the charge heater. We are very familiar with UOP design
practices for these exchangers, which include:

1) Designing the steam side pressure lower than the process side pressure.
2) Ensuring that the charge heater tubes are welded to the tube sheet.

We are not aware of any refiner relying on a moisture analyzer as a leak detection mechanism.
We believe this is a practical approach: unless the leak is caused by a major mechanical failure,
leaks from these properly designed exchangers will result in a very small change in moisture
content. Under this scenario, the simple reading from the moisture analyzer does not give the

81
operator sufficient information to decide whether the leak is real. It may be just a bad
instrument, a problem in the up-stream driers or a real leak.

One option to putting a moisture detector on the process side would be to put a hydrocarbon
detector on the condensate coming out of the exchanger. A leak in the exchanger should register
as a real spike in the hydrocarbon content reading.

Question 61
Have you found that you needed to install a methanator upstream of a chlorided catalyst
isomerization unit to remove carbon monoxide (CO) from the feed? What is the source of the CO
and how much of a difference has the addition of the methanator made to catalyst life? What is
the expected payout for the cost of the methanator?

Jay Ross (Axens NA)

In general oxygenates, including CO and CO2 should be avoided in the hydrogen stream fed to
an isomerization unit. CO and CO2 are often considered a poison due to possible water
formation via methanation. As such, the CO+CO2 content in the hydrogen is limited to 10 ppm
± a value well within steam reforming unit capabilities.

However, conditions in the isomerization unit will not promote significant methanation and a
reversible inhibition by carbonyl formation is the main concern.

In one isomerization unit, an upset in hydrogen source led to high CO breakthrough and a
deactivation wave passed through the lead benzene saturation reactor and then the isomerization
reactors. The benzene reactor dT (delta temperature) dropped sharply as the Pt was temporarily
poisoned by the CO. This led to transients in the heat exchange system and benzene passed to
the isomerization reactors. As the hydrogen purity recovered, the unit activity recovered.

82
 Bed dT

Benzene Reactor
Lead Isom
Lag Isom

time

Although some methanation could occur leading to water formation and permanent catalyst
damage, the normal operating temperature in an isomerization reactor is too low for this to occur
to a significant extent.

Pedro Fernandez (Jacobs Consultancy)

This question is also related to the performance of light naphtha isomerization units operating
with a platinum on chlorinated alumina base catalyst. In this case the culprits for the
oxygen/water ingress are carbon monoxide and carbon dioxide that ingress into the unit as
contaminants to the make-up hydrogen.

Low concentrations of CO are common in many of the available refinery hydrogen streams.
Typical values are sufficiently low²in the 0 to 10 ppm range²that the catalyst deactivation
caused by these streams will be barely noticeable. In this respect it is useful to look at some
numbers:

10,000 BPD feed @ 300 SCF/B hydrogen make-up with 10 ppm CO =

0.08 lb of water equivalent,

or 8 lb of dead catalyst.

There are multiple potential sources of CO in the refinery hydrogen gas:

Hydrogen from steam reforming plants contains CO and CO2. In older hydrogen plants,
CO2 is removed by solvent extraction and the CO is removed in a methanation reactor.
The hydrogen product from a methanation reactor typically has less than 10 ppm CO +
CO2, so this product should not be a problem for isomerization units.

83
Sid Dunn (Baker Petrolite)
Baker Petrolite is fully committed to helping hydrocarbon processors effectively and efficiently deal with the
various HSE complications Legionella populations can create. We have developed a position paper on how we
engage with our customers to help them comply with industry guidelines as appropriate. We are happy to share this
paper with any hydrocarbon processor facing such potential issues. Specifically, we have particular expertise in the
application of certain biocide treatments cited by industry guidelines as appropriate for the control of Legionella
numbers.

GASOLINE PROCESSES
Safety

Question 70
How frequently do you have fires on reformer reactor flanges? What bolting techniques and gasket types do you use
to prevent leaks? What other fixes, such as flange resurfacing, have you employed? Do you use steam rings as a
temporary fix?

Rick Grubb (Chevron USA)

We have experienced several fires in the past but recently this has been greatly reduced. We’ve developed a Best
Practice for gasket selection and bolting techniques. Following these Best Practice Guidelines has been successful.
Areas were we’ve seen problems in the past we have installed steam rings and employed techniques such as chicken
wire to make detecting a leak easier.

We use Kamprofile gaskets for any raised face flange over 24 inches. These gaskets are robust, easy to
handle, provide a very reliable seal and the gasket core can be reused after replacement of the facing materials. A
modified star pattern bolting technique is employed where 4 nuts are torqued to 100% in a star pattern and the rest
are torqued to 100% in a clock-wise pattern for two passes on each nut. After insulators are finished with the piping
someone needs to verify that the flanges (studs) are not insulated to ensure the studs maintain their integrity.

Flange surface finish should be 125 RMS to 250 RMS, which is a pretty standard finish. (Pipe flanges are
typically 250 RMS.) A high quality graphite material must be used for the Kamprofile gasket covering. We specify
APX-2 graphite as it has the highest oxidation temperature. Stud loads should be set to obtain a 20,000 to 25,000
psi gasket stress. Steam rings should never be needed if the right gasket is used and assembled correctly. They
should only be used in emergencies and never as a temporary fix.

Question 71
Is your company planning to install modified HF acid capability (usage of a volatility suppressing additive)? What
are the incentives for doing this? What alternatives have you considered?

Javier Quintana (Valero Energy)

Vapor suppression additives represent only one possible feature for consideration as part of a comprehensive HF
acid Process Safety Management (PSM) program. Industry recommendations to take into consideration are outlined
in API RP 751 – “Safe Operation of HF Acid Alkylation Units”. An effective program will comprise
recommendations emphasizing leak prevention and monitoring along with mitigation systems to be used in case of a
leak event. There are various elements recommended under API 751 for consideration:

• HAZOP assessment of existing unit equipment & controls

– per API RP 750 – Management of Process Hazards
• Management of Change program to address controls, instrumentation, metallurgy, procedures, relief system
• Thorough emergency response and control plan
• Regular and thorough unit audits to address:
– Unit specific & industry incidents and risk exposure in the unit
– Inspection, maintenance & training records
– Mechanical & procedural changes since previous audit
 – Testing & maintenance of detection, monitoring, control systems
– Testing & maintenance of mitigation systems in case of a leak
– Procedure compliance, understanding via observation, interviews
• Technology developments that further reduce accident risk
• Regular and rigorous equipment inspections to confirm integrity of unit equipment, especially acid
containing equipment
• Use of correct metallurgy – per NACE 5A171
– Confirmation with Positive Material Identification program, especially in case of repairs or
changes to equipment, piping
• Tracking mechanism to ensure resolution, close-out of identified issues – design tracking system to
facilitate next audit
• Reliable level instrumentation systems less prone to fouling, such as nuclear, radar, ultrasonic or magnetic
• Minimum acid inventory
• Mitigation systems

Mitigation systems can include active and/or passive elements, as well as tell-tale components, such as HF detectors
(point, open path or imaging systems, deployed as deemed necessary in the unit risk envelope) and HF-sensitive
paint on flanges, pump seals, etc. Active mitigation systems will often include water deluge and/or water curtain
systems, remote video monitoring and remote activation / isolation systems, and a rapid dump system to contain the
acid inventory and limit the impact and quantity of any release event. Passive systems may include barriers and
catch pans to contain any acid release, minimum acid inventory control and staging in the plant equipment, or vapor
suppression additives.

Each facility should determine their location-specific risk factors, and assess the appropriate combination of active
and/or passive mitigation systems needed to minimize risks involved in operating the unit. Any PSM program
should also be verified to be in compliance with relevant local, state, and/or federal regulations.

Alkylation

Question 72
What feed contaminants can lead to metal corrosion in both sulfuric acid and HF alkylation units? What operating
conditions promote corrosion? What do you do to reduce corrosion and/or remove contaminants?

Rick Grubb (Chevron USA)

I’ve consulted others within Chevron and this answer has been provided by Gary Ash from Pascagoula and Steve
Mathur from Corporate. The biggest corrosion problem in the reactor section of the alky unit is acid. Feed
contaminants don't really cause significant corrosion by themselves, but they can potentially make the acid more
corrosive. Water is the worst and corrosion rates can increase very dramatically if water contamination results in
diluting acid below 92% strength. Alcohols (present in MTBE unit raffinate) can also increase acid corrosion. Most
other feed contaminants can spend acid but will not greatly accelerate corrosion at normal acid strengths.
The above being said a good deal can be done to minimize sulfuric acid corrosion by controlling
temperature and velocity. For acid services we generally limit temperatures to <100 F (max) and try to minimize
velocity and turbulence in acid piping by using long radius elbows, carefully controlling the use of throttling valves
or flow orifices, etc. Also critical is to avoid potential dead legs in pipe where acid could potentially settle out and
then sit. And our most important control is metallurgy. In general we do not use carbon steel pipe for acid service.

Allen Kaiser (Delek Refining)

There are several good practices that all refiners should follow in the operation of their alkylation units – 1) closely
monitor the acid strength and water content in the case of HF; 2) for HF units, carefully monitor the acid
regeneration operations to ensure both good rejection of acid-soluble oils and retention of acid; 3) for sulfuric units,
maintain good operations in the effluent treating section (acid/caustic wash followed by a 120ºF+ water wash); 4)
monitor the equipment and piping that is susceptible to excessive corrosion with a good mechanical integrity
program. NACE papers 04645 “The Effect of Operating Conditions on Corrosion in HF Alkylation Units” and
07570 “The Top Ten Corrosion Issues Affecting HF Alkylation Units” contain good references for metal corrosion
in HF alkylation units (with some applicable items for sulfuric units).
Question 73
Does alkylate volume yield decrease if T90 increases? Have you quantified the costs and benefits of reducing T90 by
changing reaction conditions?

Allen Kaiser (Delek Refining)

Each of the variables that a refiner can directly and easily effect changes to in the alkylation unit has a particular
impact on the alkylate distillation end point (EP) and 90% boiling point (T90), but few have a direct impact on the
T90 itself without indirectly affecting it by creating more or less polymer material. Most refiners are already
running near an optimal point to minimize polymer production in their units – which directionally improves alkylate
yield and reduces alkylate T90. Monitoring the primary reaction variables and performing a cost/benefit analysis on
significant changes to those variables, with some limited test runs to determine particular equipment and processing
limitations, can likely provide a clear indication on any gains to be made.

Question #73 – Please include

in answer book

Alkylate Yield & T90

Increasing Parameter T90 Yield RON
Iso/Olefin p n n
Acid/Hydrocarbon p n n
Acid Strength p n p
Iso Recycle Purity p n n
C5= and/or C4== Feed Composition n p p
Feed Contaminants n p p
Reaction Temperature n p p
Reaction Time n p p

Gasoline Q&A
 Question #73 – Please include
in answer book

Alkylate Yield & T90

Rough guidelines – situational dependent!
4 6 8 Iso/Olefin 14 16 18
0.9 1 1.5 Acid/Hydrocarbon 2 2.5 3
80 83 86 HF Acid Strength 90 93 96
84 86 88 Sulfuric Acid Strength 92 94 98
75 80 85 Iso Recycle Purity 95 98 100
0 C5= Feed Composition 5 7 10
0 C4== Feed Composition 0.5 2 4
60 70 80 HF Reaction Temperature 90 100 110
35 40 50 Sulfuric Reaction Temperature 60 70 80
Deviations from “center” range lead to increasing problems / constraints Gasoline Q&A

Kurt Detrick/Joe Zmich (UOP LLC)

Olefin feed composition probably has the greatest effect on T90 – especially the C5 olefin content.

For a given olefin feed composition, changing the process variables within the normal range of operation would
typically not have a strong effect on T90 and the effect on the alkylate volume yield will also be limited. If the T90
increases the specific gravity of the alkylate would also be expected to increase and since the alkylate weight yield
would remain about the same, the volume yield would be expected to be slightly lower. If the T90 is increased by
decreasing the isobutene/olefin ratio, then the octane of the alkylate would be expected to decrease – this effect
could be significant.

UOP has not quantified the costs and benefits of reducing T90 by changing reaction conditions.

Question 74
Have you experienced a shortage of KOH supply for your HF alkylation unit? Are you concerned about KOH
availability? What are your alternatives if KOH is unavailable?

Question 75
The butane stream from a catalytic polymerization (cat poly) unit which contains 69% isobutane, 14% butylenes,
and 17% normal butane would appear to be an excellent alkylation unit feedstock, especially if isobutane is in short
supply. In the case where the cat poly unit uses solid phosphoric acid (SPA) catalyst, what are the consequences of
having trace amounts of phosphoric acid in the alkylation feed? Do you have a rule of thumb for estimating the
increase in acid consumption based on phosphoric acid concentration in the feed? Are you aware of any refiner
which has fed a cat poly butane stream to an alkylation unit?

Kurt Detrick/Joe Zmich (UOP LLC)

UOP does not have experience with traces of phosphoric acid coming into the HF Alkylation unit. However, we
strongly recommend avoiding this possibility because in most cases, combinations of two or more strong mineral
acids result in a much more aggressive corrosion situation than either of the individual acids alone.
UOP is aware of at least one refiner that feeds a Cat Poly product stream to the HF Alky unit. The product stream
goes through a water wash, sand tower (coalescer), and a UOP Merox™unit along with the olefin feed from the
FCC unit before going to the Alky unit.

Question 76
The regeneration of feed dryers/sulfur guard beds on butane isomerization units generates a butane slop stream.
Will processing this butane slop stream in an HF or sulfuric acid alkylation unit cause any problems? If so, what
else may be done with this slop stream?

Javier Quintana (Valero Energy)

The spent drier regenerant stream from a butane isomerization unit contains a substantial amount of butanes which
are attractive to recover. The composition of these butanes depends on the regenerant source, and can be primarily
n-butane, i-butane, or a mix of the two. However, this stream also contains the various impurities which the drier
adsorption cycle managed to accumulate and remove from the isomerization unit charge stream. These impurities
include water, sulfur and oxygenate compounds which are primarily nC4 co-boilers, which must be considered in
determining optimal routing of this stream. These impurities will also be contained at relatively high concentrations
in the regenerant stream; only excess water is removed to any extent in the regeneration system via the regenerant
coalescer, but the final regenerant slop stream is saturated when leaving the isomerization unit.

If the spent regenerant is sent to an alkylation unit for recovery of the butanes, then the alkylation unit feed driers
will remove water, but not necessarily the sulfur or oxygenate impurities, generally with three consequences:
ƒ Sulfur compounds and oxygenates will consume acid, but are not necessarily completely converted in the
alkylation unit.
ƒ Acid consumption byproducts will increase load on the acid regenerator
ƒ The unconverted portions of sulfur and oxygenates will typically end up the nC4 stream from the
fractionation section, recycling back to the isomerization unit feed

Acid consumption levels due to these impurities can be significant, in the range of 8-12 pounds of acid per pound of
contaminant in sulfuric acid units. Oxygenates can particularly problematic in HF units, where lighter boiling range
contaminants are made, often reducing the initial boiling point of the acid soluble oil stream by more than 100°F,
forcing a reduction of temperature in the acid regenerator, thereby increasing physical losses of acid.

The following sketch illustrates the impact that such unconverted sulfur and oxygenate species will have, in that
they will add to the level of impurities that were introduced via the fresh the alky butane feed, resulting in a “wind-
up” of sulfur and oxygenate levels in isomerization unit feed, possibly breaking through the feed driers and
deactivating isomerization catalyst.

iC4Recycle

D
Alky
C4Feed I
Unit
B Isom Isom
ASO nC4 Driers Unit

SpentIsom
Regenerant Alkylate
The tolerance of a particular unit to this wind-up effect depends in large part on the amount of sulfur in the refinery
fresh butane stream to the alkylation-isomerization section and the available slack capacity in the isomerization unit
feed driers. If there is little slack capacity in the driers, then any such wind-up presents risk of break-through and
catalyst deactivation.

Alternatives to sending the spent regenerant to the alkylation unit involve finding a location where either the
contaminant species are converted to a greater extent than in the alkylation unit, resulting in a lower extent of wind-
up, or where the residual impurities do not recycle at all to the alkylation or isomerization units, thereby eliminating
the wind-up effect. The ideal solution is usually to co-process the spent regenerant in a hydrotreating unit (generally
the naphtha treater is used), with the treated refinery butanes pooled and sent to the alkylation / isomerization
section.

Depending on the refinery fuel balance, some dispose of the spent regenerant by vaporizing the stream into the fuel
system, although the sulfur level of the stream may present a concern with SOx levels in heater flue gases. Another
alternative is to send the stream to a FCC butanes sweetening unit, although sulfur removal occurs to a lesser extent
than in the hydrotreater.

Question 77
What is your experience with cooling water exchangers in an HF alkylation unit? How long do you go between
cleanings? Do you have a special water treatment program for cooling towers dedicated to the alkylation unit?

Allen Kaiser (Delek Refining)

Cooling water exchangers in an HF alkylation plant are no more problematic than any other exchanger in the plant
when considering the water side of the bundle. In the HF unit where I have experience, the HF coolers were
generally able to maintain acceptable performance for 2-3 years. Nearly all of the leaks in the water cooler
exchangers I’ve had to deal with were due to process side corrosion. Both alkylation units that I have direct
experience with have their own dedicated cooling towers. The type of treatment program for an alkylation cooling
tower is not necessarily significantly different than any other tower in the refinery. Rapid pin-pointing of the
leaking exchanger is critical to minimizing corrosion damage to other exchangers and to the cooling water
distribution piping.

Kurt Detrick/Joe Zmich (UOP LLC)

We have seen corrosion in cooling water exchangers in HF alkylation units on both the process and water side.

Typically the corrosion rate on the process side of the reactor bundles and the acid cooler bundles is relatively low at
normal operating conditions. Excessive temperature or high %water in the circulating acid can cause corrosion and
plugging on the process side of these exchangers.

Refiners have reported higher corrosion rates on the process side of water-cooled condensers – especially if the
water content of the circulating acid is high.

Corrosion on the cooling water side of the exchangers is often due to low cooling water flow rate. The low water
flow rate can lead to fouling of the tubes due to biological growth or calcium scaling – both of which can lead to
under-deposit corrosion. Inadequate treatment or blowdown of the cooling water can also cause corrosion on the
water side of these exchangers.

HF acid leaks into the circulating cooling water can cause tube failures in multiple exchangers if the leak is not
detected quickly. UOP recommends conducting both fluoride and hydrocarbon tests on cooling water tower to
check for tube leaks due to the presence of LPG and HF acid.

UOP strongly recommends a separate cooling water tower for the HF Alky unit, but I do not believe that we
require it in our typical new unit specifications.
There is new unit activity for HF units. Current permitting requirements do not prohibit installation of new HF Alky
units. However, in many cases, it is necessary to evaluate the risk of installing an HF Alkylation unit, and to show
the permitting authorities that this risk is acceptable. Probably the best way to do this is to perform a Quantitative
Risk Assessment (QRA) for the HF Alky unit.

Question 78
For HF alkylation units, have you changed your criteria for materials given the low availability of low carbon/non-
recycled steel? Are you heat treating welds? Can you control Brinell Hardness with welding procedures? For small
bore pipe, do you recommend using flanges or threaded pipe?

Allen Kaiser (Delek Refining)

I would not recommend a materials specification change for HF services. API RP 751 does recommend post-weld
heat treating (PWHT) of all welds in HF service. The upper recommendation on Brinell Hardness (BH) is 200 BH.
The overall hardness of a weld can be (at least partially) controlled by weld procedures. Most weld procedures are
not sufficient, in and of themselves, to guarantee repeatability in meeting the 200 BH spec. Small bore piping
should also use socket-weld flanged connections. NACE papers 03651 “Specification for Carbon Steel Materials
for HF Acid Alkylation Units” and 5A171 “Materials for Receiving, Handling, and Storing Hydrofluoric Acid” both
contain good references for materials selection in HF alkylation units.

Blending

Question 79
It has been reported that di-isobutylene (isooctene) causes a stability problem when blended in gasoline. Do you
have experience blending di-isobutylene in gasoline and, if so, were there stability or other problems?

Rick Grubb (Chevron USA)

I’ve consulted others within Chevron and this answer is based on information provided by Shenggao Liu and Dave
Kohler. There is no intrinsic, chemical reason to expect this olefinic stream to be more susceptible to stability issues
than other highly olefinic streams. As with other olefins, isooctene isomers are subject to easier oxidation than the
corresponding saturated C8 isoparaffins/isooctanes. If unchecked, the oxidation to hydroperoxides may promote
further molecular condensation to form higher molecular weight species which may lead to insoluble gum residues
in the fuel. This tendency can be mitigated effectively by the addition of a low level of appropriate antioxidant
chemical, either phenylenediamine or hindered phenol type chemistries. As with the addition of antioxidants to
other process streams, it is always most effective to add the antioxidant as soon as the product exits the process unit
in the rundown line to storage.
Chevron has installed two units with the ion exchange resin type catalyst technology for isooctene
production. These were a conversions of former MTBE plants. To the best of our knowledge, the blending of their
isooctene has not caused any measurable decrease in gasoline stability, at least as assessed by existent or potential
gum contents.
We also have experience with two solid phosphoric acid distributed on a solid support type oligomerization
processes. The olefin oligomer streams cause no stability issues with the refinery’s motor gasolines and would be
stabilized with a minimum amount of an appropriate antioxidant.
In addition we have experience with Dimersol type of olefin oligomerization units. They dimerize
propylene into highly branched isohexenes which may be blended to premium gasolines as high as 25 volume %.
We have seen no stability issues blending this material. In any case, these Chevron gasolines would contain at least
some minimum level of an appropriate antioxidant.
Depending upon the source of the C4 olefins, the feed stream may also contain measurable amounts of
diolefins, such as 1,3-butadiene, or, if the distillation range is broad enough perhaps even isoprene (2-methyl-1,3-
butadiene). Conjugated diolefins such as these two species are generally much easier to peroxidize than the
corresponding monoolefins of the same carbon number. This tendency can be mitigated effectively by the addition
of a low level of appropriate antioxidant chemical.
Again, depending upon the source of the C4 olefins, the feed stream may also contain measurable amounts
of reactive sulfur compounds such as ethyl- or propyl mercaptan. These mercaptan species are notorious for being
easy to peroxidize. This tendency can be mitigated effectively by the addition of a low level of appropriate
antioxidant chemical.
Javier Quintana (Valero Energy)
Valero Energy has several iso-octene units in operation, and has experienced no issues with stability of the iso-
octene product or the gasoline into which it is blended. The key is to inject the stabilizing antioxidant chemical into
the product rundown line at the appropriate dosage rate. This is consistent with A stability test, such as ASTM
D525, on the iso-octene product can be used to optimize the dosage rate for a particular antioxidant. Your chemical
treatment vendor should be consulted for specific product and dosage recommendations.

Isomerization

Question 80
Where in the isomerization reactor catalyst bed does the hydrogenation of benzene (exothermic) occur? How does
this affect the other isomerization reactions? What concentration of benzene in the isomerization feed is acceptable?

Richard Rosin/Joe Zmich (UOP LLC)

Benzene saturation occurs at the inlet of the reactor. In order to assure maximum octane and yield from a light
naphtha isomerization reactor it is necessary to saturate benzene. Benzene will adsorb on acid sites making these
sites unavailable for paraffin isomerization. If benzene is not immediately saturated in the front of the reactor, the
benzene will tie up acid sites effectively increasing the space velocity for paraffin isomerization reactions.

The main effect of benzene in a light naphtha isomerization reactor is thermodynamic rather than kinetic. The very
high heat of saturation of benzene will increase the temperature of the reactants. The maximum achievable
concentration of high-octane iso-paraffins is limited by equilibrium. The equilibrium level of iso-paraffin decreases
with increasing temperature. The effect of benzene on isomerization reactions is to raise the temperature to a point
where the shift in thermodynamic equilibrium lowers the maximum attainable octane. Multiple reactors, with inter-
reactor cooling, may be required to process feedstock with benzene levels greater than 3 lv%.

For most purposes the maximum allowable concentration of benzene in isomerization feed is 5 vol%. Higher feed
levels of benzene are possible with a benzene saturation reactor in front of the isomerization reactor(s), e.g., UOP
Penex PlusTM process.

Javier Quintana (Valero Energy)

Valero Energy has experience with processing isomerization feeds as high as 5 lv% benzene content, with a
substantial resulting exotherm in the lead reactor. The inter-reactor cooling capacity of the unit, and the resulting
temperature that can be maintained at the lag reactor inlet, will determine the isomerate product quality that can be
achieved relative to the same feed with lower benzene levels.

Question 81
Has the optimum feed for light naphtha isomerization units changed given that: 1) ethanol blending reduces the
octane value of other blendstocks; 2) the demand for premium gasoline is down; and 3) ethanol blending increases
RVP compliance costs? Are you removing pentane from the isomerization unit feed stream or shutting down the
unit? Or, are the units still valuable for isomerizing normal hexane and saturating benzene?

Allen Kaiser (Delek Refining)

Each refiner will make one of the following three moves when introducing ethanol into the gasoline blend pool –
shut the isomerization unit down, change the isomerization feed makeup by dropping most hexanes into the reformer
feed, or keeping the isomerization unit up at previous rates and feed composition, but off-setting the increase in pool
octane by lowering reformer (or other unit) octane severity. The specifics of that choice depend on refinery site
economics and regulations.
In the first case, the refiner’s isomerate stream can be offset by the ethanol introduction and the cost of importing the
ethanol is lower than the cost of processing the isomerization feed. In the second case, moving the hexanes into the
reformer feed likely provides better overall gasoline pool RON and RVP balance and is favored by economics. The
isomerization of the pentane molecules is still likely favored due to the increase in octane. In the third case, the
refiner is probably operating the isomerization unit for benzene control and has no other palatable options for
isomerization feed.

Javier Quintana (Valero Energy)

The Renewable Fuels Mandate has impacted isomerization unit economics due to reduced octane requirements to
produce CBOB (Conventional Blendstock for Oxygenate Blending) gasoline streams, increased vapor pressure
constraints, as RVP specifications must be met after ethanol splash blending at the distribution rack, and flat
gasoline demand from US refineries after accounting for volume impact of ethanol blending and finished gasoline
imports.

While the economics for isomerization unit operation may vary with each refinery’s unique situation, in general the
impact expected is to render C5 isomerization economics marginal to negative, with C6 economics slightly better
than C5 economics due to the difference in RVP of the products. However, most refineries do not separate the C5
naphtha fraction from the C6 fraction, and therefore, additional equipment and operating cost may be necessary to
selectively process a C6 cut – this may be difficult to justify. This is especially the case when the C6 isomerization
function is provided by the naphtha reforming unit. Furthermore, if the isomerization unit is shut down due to
economics, the light naphtha stream will likely contain some benzene that was previously saturated in the
isomerization reactor, which may lead to a reduction of the light naphtha / heavy naphtha cutpoint as part of a
refinery’s MSAT II compliance strategy. This situation is likely to occur if a post-treat strategy is adopted on the
reformer, thereby containing all benzene to be treated in a light reformate stream. This approach can spare the
isomerization unit to be deployed for saturation only, so as to achieve compliance without aggravating the tighter
RVP constraint that will prevail in a CBOB blending environment.

Question 82
In light of coming benzene regulations, are you using (or planning to use) the isomerization unit for benzene
conversion? How does this affect isomerization catalyst performance and unit operation? How does this affect the
gasoline pool? How does benzene saturation in an isomerization unit compare to a dedicated benzene saturation
reactor/catalyst?

Richard Rosin/Joe Zmich (UOP LLC)

This question relates well with question #80. Light naphtha isomerization units are extremely flexible in their ability
to handle increasing amounts of benzene in the feed. A UOP PenexTM process unit operating with <5 vol% benzene
in a straight run light naphtha feed can in most cases be easily revamped to a Penex-PlusTM process unit capable of
handling over 30 vol % benzene in the feed.

Saturation of benzene with isomerization of light paraffins in a new isomerization unit leads to an octane gain as
opposed to an octane loss in a stand-alone benzene saturation unit.

Allen Kaiser (Delek Refining)

There are several options for dealing with the benzene production in each of our facilities – precursor reduction,
post-treatment “destruction”, and extraction.

First, extraction relies on having an outlet for the benzene stream. If this outlet is not available, or the
economics do not support the additional capital costs of this solution path, then it will not be a fit for the refinery.
Second, post-treatment destruction via a saturation unit is likely to be a popular choice if the refinery either does not
have assets to reuse, revamp, or capacity available in existing units. The final options are pre-treating and post-
treating in an isomerization unit. The unit must be able to handle the designated feed stream, the unit yield shift
must fit into the refinery gasoline blend pool, and the scheme must meet the benzene reduction requirements.