0% found this document useful (0 votes)
202 views7 pagesHsiehMcNulty SieveValveWeeping
Uploaded by
mcruzuniCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
202 views7 pagesHsiehMcNulty SieveValveWeeping
Uploaded by
mcruzuniCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 7
Dea i rin yt Pte Tt
eps breed ©
cat wl er month
Sioa gs 08
Predict Weeping of
Sieve and Valve Trays ,.
REACTIONS AND SEPARATIONS
Weeping can
significantly affect
tray performance.
Anew correlation
can be applied
to estimate the
weeping rate,
as well as
the weep point
of a sieve tray.
CChangliisioh and
Kenneth JyMetlty
Bor inpsota spe
@Wibrn Te
@®Mn
@& US
trayed tower isa vertical coun-
fercurrent flow channel
designed to achieve a certain
process performance under
some given operating constrains. The per-
formance requirement is usually expressed
as a mass- or heat-transfer duty or separa-
tion specification, while the commonly
encountered operating constraints include
pressure-drop limitation, capacity, and
tumndown-ratio requirements. Proper tray
design ensures that these requirements and
constraints are satisfied by selecting the
‘most suitable tray type and by optimizing
an array of interrelated dimensional para-
‘meters that defines the tray decks, down-
ccomers, outlet weirs, and tray spacings.
‘The design challenges involved provide
the stimulus for research and development
on various aspects of trayed-tower
hhydraulic and efficiency phenomena,
‘One phenomenon that can have a sig-
nificant impact on both the hydraulic and
efficiency performance of a tray is weep-
ing. Hydraulicaly, weeping can affect the
liquid holdup on the tray deck and can
influence the tray pressure drop and other
wet-tray hydraulic characteristics.
Weeping also represents a departure from
the ideal plug flow of liquid across a tray
(for which the theoretical maximum effi-
ciency is achieved),
In an effort t0 establish more reliable
methods to predict the weeping phenome-
na of sieve trays and valve trays, we con-
ducted a research program to collect addi-
tional weeping data, as well as to develop a
new weeping model, Ithas provided useful
results, which we. summarize inthis article,
Theoretical background.
In the open literature (J,2.3), 8 con-
sensus has been reached that weeping
is controlled by the two key parameters:
1. gas F-factor or gas pressure drop
across the tray, hy and
2 the clear liq height on the tay,
However, information on valve-tray
‘weeping performance was found to be
lacking. Furthermore, published models
only allow the prediction of the weep
point — that is, the conditions for
which weeping starts to occur — but
not of the amount of weeping when a
tray is operated within the weeping
region.
In our model development, the fun-
damental two-phase countercurrent
flow limitation (CCFL) theories for
vertical channels, are used. The basic
correlation for CCFL. in vertical tubes
as used by Wallis (4) is:
_ Po
Jon “deze al _
=v
Ser emG=c @
For a straight vertical pipe, Wallis
showed that the characteristic length Z
is simply equal to the pipe diameter. For
complex vertical flow channel — for
instance, a packed bed or an annulus
— the hydraulic diameter of the flow
channel should be used. For flow
channels with a large characteristic
length, such as a 24-in. diameter
empty pipe, the Kutateladze model (5)
assumes that the CCFL phenomenon
should no longer be affected by any
channel dimensions, and the character-
istic length, Z, in Eqs. 1 and 2 should
CHEMICAL ENGNEERING PROGRESS + JULY 180 « 71 REACTIONS AND SEPARATIONS
be replaced by the Laplace
capillary constant:
(918 (p,- Pal (4)
While the Kutateladze
correlation would be pre~
ferred for defining a “system
limit" — for example, the
capacity of an empty tower
or the maximum capacity
that a tray tower can achieve
by increasing the tray spac-
ing, the Wallis correlation is
preferred for defining the
limits of countercurent flow
through trays and packed
beds where the channel
dimensions have a major
influence
‘The Wallis CCFL correla
tion has been successfully
applied to the flooding phe-
nomena of both random and
structured packing (6) and to
weeping from perforated
plates (3,7). Although most
of the perforated-plate CCFL
studies were conducted
using small-scale plate geometries,
the success in using this correlation
suggests that it also should be appli~
able to the correlation of weeping
data taken with large-scale sieve
trays and valve trays.
Experimental work
‘The weeping tests were conducted
with the ait-water system in a 3-ft-
dia, stainless steel tower. General
information on this tower has been
given by Hsieh (6) and McNulty (6).
In this test configuration, the tower
contains three identical trays equally
spaced at 24 in, During a weeping
test, liquid overflow leaving the bot-
{om tray is retumed to the liquid feed
tank via a sealed-bottom downcomer,
while the weepage is collected on the
chimney tray and measured in a
‘weepage collection tank, Fasesan (2)
showed that the weeping rate mea-
sured with tis technique is compara-
ble to that measured with dye-injec~
tion tracing,
Dene eeecoe
ee
For each tray configuration,
weeping data were collected at
five liquid flow rates: 10, 20, 40,
100, and 200 gal/min. For each,
the weeping rates of the bottom
tray were measured at various air
flow rates, ranging from the weep
point to near the dump point. Most
Of the tests were conducted using
an increasing gas flow rate proce-
dure — that i, the gas flow rate
was started at a low setting and
‘was stepped up gradually to the
weep point. Certain tests also were
conducted using a decreasing gas
flow rate approach. Contrary to
the observation of Lemieux (10),
no differences were discerned in
the data measured with these (wo
procedures,
Four sets of trays were tested: @
7.0% open-area sieve tray, a 9.4%
open-area sieve tray, a 20% maxi-
‘mum open-area Type-T valve tray ,
and an 18% maximum open-area
‘Type-A valve tray. Other tray para-
meters are summarized in Table 1
‘72 + lv ton + cHEncaL eNoNEERING pRocRESS
Our sieve-tray weeping data,
along with the data of Lockett (1)
(for %-in, hole diameter, 10.2%,
15%, and 20% open area, 1~3-in.
weir heights), Fasesan (9) (for ¥e-
in. hole diameter, 14.2% open area,
in. weir height), and Thomas (17)
(for L-in. hole diameter, 12.4%
‘open area, 3.in. weir height) const
tute the database for the sieve-tray
weeping correlation, An additional
proprietary hydrocarbon databank
‘with more than 500 data also was
utilized in the subsequent model-
validation efforts. For valve trays,
the current data and that of Fasesan
() (for a Type-T valve tray, 12
valves/f@, 2-in, weir height) were
‘used in the correlation
Data analysis
‘The CCFL model defined in
Bq, 3 was selected to analyze the
weeping data. The flow rate pera-
meters used in Eq. 3 are the super-
ficial hole gas velocity, Voy, and
the superficial hole liquid weeping
velocity, Vp both based on the total
‘open area, Ay, of a tray. These are the
same parameters adopted by Lockett
(2) and Zenz (12). Because the value
of Ay for a sieve tray is clearly
defined, while that for a valve tray is
subject to the percent opening of the
valves (and therefore subject to the
pact ofthe gas flow rate), the sieve
tray weeping data were analyzed
first. The objectives of this analysis
were to verify the validity of Eq 3 in
correlating the sieve-tray weeping
data, to determine the empirical m
and C values in Eq. 3, and to develop
‘a model forthe characteristic length
Zo be used in the CCFL equation.
‘This cortelation was then modi-
fied for application to valve trays,
including the development of a model
for calculating A, and one for Z,
ieve-tray weeping
performance.
One fundamental assumption of
the CCFL model is the linear rela- tionship implied in Fq 3 between the
square 10085 ofthe gas andthe liquid
flow rate parameters. To verity this
assumption, the sieves data were
piloted using the square roots of C,
and C, (Eqs. 9-7), where the effect
of the Characteristic length Z is tem
porary ignored
Ge velotaal ©
CytmC?=c om
Using this “linear square root” rela-
tionship, the 9.4% sieve tray data
taken at I-in, and 2-in, outlet weir
lhcights are shown in Figures f and 2,
respectively. As demonstrated by
these data, the weeping performance
of sieve trays docs follow the linear
square root relationship suggested by
the Wallis CCFL model. The effect
Of liquid flow rate on the weeping
rate is also clearly demonstrated in
the figures. An increase in the liquid
flow rate causes the clear liquid
height on the tray to increase and
results ina higher rate of weeping,
Figure 3 shows the effect of outlet
weir height. Data for the 9.4% sieve
tray operated at 20 gal/min/ft are
shown for thece outlet weir heights,
As the clear liquid height is increased
by a higher outlet weir, the weeping
rate also increases. Therefore, the
clear liquid height, hg, should be
incoxporated into the calculation of
the characteristic length, Z
‘Two other parameters — surface
tension, 6, and hole diameter, Djp
were also considered in developing
the current model. Based on the data
cof Lockett (1), Kreis (5), and the pro-
prietary weeping data obtained for
hydrocarbon systems, no effect of sur-
{ave tension can be discemed for trays
with hole diameters larger than Yin.
In the open literature, the opin-
ions concerning the effect of hole
diameter on weeping at a constant
hole velocity are inconsistent
Based on proprietary data and the
data of the 9.4% sieve tray, we con-
cluded that for sieve trays of indus-
tial interest with a hole diameter
of Yé-in, of larger an increase in
hole diameter at constant hole
velocity will tend to decrease the
weeping rate. As a result, the
effects of clear liquid height and
hole diameter were combined to
form an equation for the catcula-
tion of characteristic length Z:
Z=heDye ®
‘To calculate the clear liquid height,
‘hg. both Colwell’s model (14) and
Garay's model (25) were evaluated,
Colwell’ model was selected for Eg, 8.
Applying Eq. 8, the weeping data
taken with the 9.4% open-area sieve
tray are ploted in Figure 4 using the
dimensionless parameters J,” and J,"
(Eqs. | and 2), As shown, @ reason
ably good correlation is obtained for
all the data taken at three weir
heights (1 in, 2 i,, and 4 in.) over a
20-fold variation in the liquid load-
ing (that is, 5-100 gal/mievft weit),
Based on this set of data, the regres
sion values of m and C for Eq 3 are:
m= 201
c=074
Using Eq. 3, the weep point can
be determined by
U)=C °
‘The data for the 7.0% open-atea
sieve tray are ploted using the same
parameters J” and J,” in Figure 5
‘As shown, this data set also can be
correlated by the same CCFL model
‘The values of m and C are:
3.61
14
This correlation indicates thatthe
‘weep point relationship defined by
we
Stelter
[J+ sont i
anemia
+ Sauna
100 oan weir
¥ logan wee
© wali ir
Figure I. (Above) Weeping data Jor 9.15eapen-area slove
ray with Fin weir height
WM Figure 2. (Right) Weeping data for 2.ASeopen-area sieve
tray with 2 in. weir height.
acd 025
CHEMICAL ENGEIEERNIG PROGRESS © JWLY 882 © 73 REACTIONS AND SEPARATIONS
Lui Rate «20 gaint ie
a
fied
oer]
oat fo Werte ‘i
2 [eaectoon
S oa =
Aaittoaies ia epee =
7 ae i
os soe
(oye
© W Figure 5. (Above) Effect of outlet eight om weeping.
Figure 4 (Right) Weeping data for 9.4%-open-area sieve tray.
Eq. 9 is relatively independent of cluded that the vales of m = 2.01
the variation in the percent open and C = 0.74 should be applicable
area. The weeping rate, as reflected to trays with a 9% or larger open
in the value of m, however, should area. For trays with smaller open
be correlated as a function of the areas, the weeping rate prediction
percentage open area based on these equations will be
‘The data of Lockett (1) and higher than the actual weeping
Fasesan (9) also were correlated by rate, and the m value of 3.61
the same model and yield m and C shout be used.
values comparable to the corre. In conclusion, based on the
sponding values of the 9.4% sieve weeping data taken with large-scale
tray. The weep point data of sieve tays, a new weeping cortela-
‘Thomas (17) for a 12.4% open-area tion in the form of Eq. 3 has been
sieve tray with L-in. dia, holes also developed. This correlation is theo-
is in good agreement with that pre- _retically consistent with the CCFL
dicted by the correlation (Eq. 3, model for vertical channels and can
using m= 2.01 and C= 0.74). be applied to estimate the weeping
Because of the agreement rate, as well as the weep point of a
among the above data sets, we con- sieve tray
performance
Fasesan (9) has shown that the
weeping rate of a 14.3% maximum
‘open area Type-T valve tray is nearly
‘an order of magnitude less than a
corresponding sieve tray with the
same percent open area. This low
‘weeping characteristic of the Type-T
valve tray can be attributed to the
throttling mechanism of the valves,
which automatically reduces the
available tray open area at low gas
flow rates,
‘The low weep ratio of the valve
trays is demonstrated in Figures 6
and 7, where the weeping data of the
20% Type-T valve tray and the 9.4%
Weeping Rate, gaint af Ay
ry
Bas Volos, of Ay
Figure 6. Valve ray has a lower weep rate than a seve
Figure 5. Weeping daa for 7.0%open-area seve tay. tray at aliquid flow rate of 10 galimin(ft of weir
NOTE: KisTER
TA + suri + coe WcALENGREERNG PROGRESS bi
KEMRTS C2019 RATHER THAN
To A b.P5f Dib FERENCE © Type-Fuate ay [|
7] © e4%sievetay
Weeping Rat, galimint? of Arg
ae)
ee o
Bas Velo At Ay
Figure 7. Valve tray also gives a Tower weep rate at aliquid‘ Figure 8. Weeping rate of Type-T valve tray vs. sieve tray.
low rate of $0 galiminf of wei.
sieve tray ate compared at liquid flow rate. The weeping rate will Type-T valve tay, the “hole” velocities
flow rates of 10 and 50 gal/min/ft increase drastically once the gas flow are calculated based on the maximam
weit, respectively. To facilitate the rate is reduced below the weep point. valve open area, Ajay, where
comparison, both the gas flow rate At 10 and 50 gal/min/ft weir, the 7
and the weeping rate are presented measured maximum weeping rates w= CF Py Haup Ry) 144 (10)
based on the total bubbling area. As are 46% and 33% of the liquid feed As shown, compared to a sieve
shown in these two figures the Type- rates, respectively. Because of this tay on the basis of the same C, val-
T valve tray, by the closing of the characteristic of low weeping, Type- ues, the ‘Type-T valve tray has a sub-
valves at low gas flow rates, is able T and Type-A trays can be designed stantially lower weeping rate, as well
to maintain a low weeping rate with- with a high density of valves (that is, asa lower weeping point C, valve.
in its entire weeping region. At 10 the number of valves per unit tray ‘In Figure 9, the weeping rates of
sgal/min/ft weir, the maximum weep- deck area) to achieve a high capacity the 18% maximum open area Type-A.
ing rate is 11% of the total liquid while still maintaining good efficien- valve tray, taken at aliquid flow rate
feed rate; while the rate reduces to cy performance over a wide turn- of 50 gal/min/f, are compared with
only 5.7% of the liquid feed rate at down ratio, the corresponding data for a Type-T
50 gal/min/ft weir. Consequently, the ‘In Figure 8, the same data as pre- tray and a sieve tay. The weeping rate
impact of weeping on the efficiency sented in Figures 6 and 7 are plotted of the Type-A tray is higher than that
‘ofa Type-T valve tray can be kept at using the C, and C, parameters of the Type-T tray, but sil is substan-
4 minimum On the other hand, the defined in Eqs. 5 and 6: For sieve-ray tially lower than that of the seve tay.
weeping rate of the 9.4% sieve tray data, the actual hole velocities are used To correlate the weeping data for
is strongly influenced by the gas in the C, and C, parameters. For the Type-T and Type-A valve trays
0160 =
Sev, =50 glint
1 90", =50 gaint
“ype = 10gatiintt
+ Type 0, =50 gat [3
ons
Sms aes ;
" % (utes
1 Figure 9. Weeping performance comparison Figure 10. Weeping correlation Jor Type valve tay, ——~
Vs + 3.9VP: ovo,
CHEMICAL ENGINEERING PROGRESS © JULY 162) + 75 REACTIONS AND 5.
ET
Nomenclature
Ay “ett hole ae,
pen en a ay
= arimam vive open ara,
empires! contin in CCR.
ctelton|
CG, liquid phase towing tetor deinet
Ing. 6, 05
= se pte loading Fctordtioed in
Ba, 5,08
le dare
equslet hate damete i
salve diameter ia
ale ay Pato defo ia
Fa, 1, imalalve
ave ray act at the begining
fle ve openreion,unhale
revatio constant, Us
ee ig eight i,
ry tay peta di,
ciara Rito vale in
vale itt he dsance
etn the bottom ofa vale and
the tp ofthe ray deck,
mensionles go velocity
imersonies weeping iid
vec
pits! cost in CCFL
‘onlaon
+= vale deny umber of valve?
olmotie gs low ete oa vale,
‘innate
actions opening in he
isumferene of valve
perfil gs veloc in canna
(oat tater, is
pase supe ble velocity,
a
ts
peti veloty io
chanel (ot over) fs
liquid phase upertiial ale
elo,
Laplace cope onstint
haters nin CCFL ode
Pe pasdensiy, hm
PL = li deny, ee
© slice tens
using the CCFL model, equations to
calculate the percent valve opening
ata given gas flow rate are required
to determine the superficial gas
velocity and the superficial weeping
liquid velocity through the valves.
Based on the weeping and pressure
76 » ouwy 92a « chevercaLenaiveenNG PROGRESS
drop data taken for these trays, the
percent opening of the valves can be
calculated as:
Hop = Hyp il Fy2 Fay
(valve open region) (11)
Hyp = (FI Eg)? Hegnit Py