C!wm?ical Enginrcrlng Sciencr. Vol. 45, No. 8. pp. 2269-2275. 1990. Mw)9-2509/90 $3.00 + 0.

00
Printed in Great Britain. h 1990 Pergamon Press ply

COUNTERCURRENT- TWO-PHASE FLOW

FIXED BED CATALYTIC REACTORS

P. TRAMBOUZE - Iastltut Franfals du P&role

BP 3 - 69390 VERNAISON (France)

ABSTRACT

While a very large number of publications have been devoted to fiied bed catalytic reactors with cocurrent two-phase flow, this
cannot be said of fixed bed catalytic reactors in which the two fluid phases are in countercurrent flow. Very few publications have
dealtwithtbistypeofreactor. Inthefollowingpages,wewouldlike~opoin~ou~~epotentialadvanlagesofthesecountcrcurrentsys-
terns. These advantages can be especially important, for example in reversible reactions or those inhibited by one of the reaction pro-
ducts. The theoretical bases for making a detailed comparison of the two variants, for various reaction systems are first discussed.
Then are presented a number of comparative examples of some typical cases: exothetmic and non-exothennic reactions. rever-
sible reaction, parallel reactions inhibited by one of the products. These reaction systems represent clearly-identified actual cases,
such as the hydrogenation of aromatics, the hydrotreating of petroleum cuts. the synthesis of MTBE. etc.. The last part attempts to
draw certain conclusions from the comparisons made, without ignoring the technological aspects and the practical constraints.

KEYWORDS
Fixed bed. two-phase flow, countercurrent

Catalytic reactors employing two or more fluid phases have been the subject of a large number of studies, and many industrial ap-
plications use this typeof reactor. Several variants are conceivable, according towhether thecatalyst is used in a fixed bed, moving
bed, fluid&d bed, or in the form of a slurry. Moreover, both phases may circulate in different ways, generally limited to three stan-
dard arrangements: downflow cocurrent. upflow cocurrent and countercurrent. Among these alternatives. the arrangement most
often used, and which has attracted the largest number of investigations, is certainly the trickle-bed, in which the catalyst is in a fixed
bed. while the fluid phases (usually gas and liquid) are in cocunznt downflow (Ng, 1987).

Intbisarticle, weshallconsiderararelyinvestigatedarrangement,inwhichthecatalystisinafixedbed,butwiththefluidphasesin
CountercUrrent flow. This typeof reactor, while occasionally mentionnedin theliteratureas feasible, has actually been little studied,
and always in a very specific context. Similarly, while a few industrial applications have heen mentioned, they always represent
specialcases. Tothebestofourknowledge,theonlyarticlesinwhichcomparisonsaremadebetw~nacountercurrentarrangement
with the standard trickle-bed are due to Mills et nl. and Goto et al. (Mills, 1984 - Goto 1978 a,b).

Industrial applications include two examples of fued beds with countezcment flow. The AR0SATproces.s ofammatics hydro-
genation using a countercurrent fixed bed to maximize the hydrogenation reaction. which is favored by lower temperature and hi-
gher hydrogen partial pressure (Reilly, 1973). The production of MTDE by the ‘reactive distillation’, in which the reaction takes
place simultaneously with the separation of the MTBE produced. which further favors theconversion normally limited by the che-
mical equilibrium (Smith. 1987).

We would like to show here that the system using a fixed bed with two phases in countercurrent flow must be considered moregene-
rally as apotential competitor of the trickle-bed. Following a briefreview of the balance equations representing the countercurrent
fixedbcdandtrickle-bedsystems, weshallapplythem to anumberoftypicalexamples,for which thecomparison between the two
arrangements tends to favor the countercurrent technique.

2269
2270 P. TRAMB~UZE B13

1 MASS AND HEAT BALANCES

Let us consider a reaction system in which the stoichiometry is written in the general form:

Zclij.+=O (i= l.N j = l.R)

These reactions take place at the surface of a solid catalyst in the form of grains or packing elements (Raschig rings, Berl saddles).
We shall assume that the different reactions are notlimited by Inca-particle diffusion. The kinetic equations, however, may differ
according to the reactionsconcerned. Fluid phases1 and II may consist of mixtures of liquid plus gas or two immiscible liquids. In
every case, phase I is the denser, hence liquid phase, always flowing downward.
Weshallconsiderthemassandheat~ansferbetweenfluidphasesIandII, with thesametransfersbelweenfluidphasesandcatalysc
particles assumed to be rapid and hence ignored. Accordingly, the catalyst temperature at which the reactions occur is assumed to
beequal tothatofthedenserphase(I=liquid). Similarly,theconcenaationsusedinIhekineticequationsarethoseofphaseI. The
mass and heat balanceequations for each of the components in phase I and phase II are written as follows, by referring toFigure 1:

dz + Ni.S - (~ij.rj) E.S = 0 (1) (2)

d PiI . C,) TI d (ZFilI . C,) TB

+~jAHjE.S+H.S=O (3) fH.S=O (4)
dz dz

The terms of mass transfer (Ni) and heat uansfer (I-I) are expressed per unit volume of reactor. Based on the film theory:
Ni= ki’4(CiI_CiI*) (5) H = h.a (TI - TII) (6)

In equations (2) and (4). the signs-and + are used according to whether the flow directions of phases I and II are cocurrentor coun-
tercurrent In writing tie balanceequations.equations (1) to (4). we have ignored backmixing. Additional lcrmscould be inuodu-
ted into these equations to take them into account. as the axial dispersion coefficients can be reasonably wellestimated. Note also
that the reactors are assumed tobe adiabatic, in other words without any heat exchange with the exterior. To calculate the perfor-
mance of one type or the other, it is clearly necessary to evaluate the different parameters involved in the foregoing equa-
tions.

(a) Characteristic values of the chemical system

Thestoichiometriccoefficients~jcharactizetbereaction system. The effectofthetempemtureon thereactionratesrjwillbeac-
counted for using Arrhenius law (activation energy E$ The enthalpies of reaction must also be known to introduce them into the
heat balances, and also into the chemical kinetic equations inthe event that certain reactions are reversible. Since thereactions are
catalytic. the reaction rates areexpressed per unit volumeof catalyst. This is why the volume fraction E of thereactoroccupied by
thecatalyst is a factorof the chemical kinetic term. This parameter E differs significantly according to the type of arrangement:
E = 0.65 for a trickle-bed, whereas it may be much lower for countercurrent flow (E = 0.3 to 0.35 for a Raschig ring or Berl saddle
packing).

(b) Characteristic values of mass and heat transfers

TheparameteIsatheinterfacialarea. ki, themasstransfercoefficients. and h. the heatPansfercoefficientcanbeestimated with va-
riable accuracy from existing correlations. We have considered both liquid/liquid and gas/liquid systems. Yet it is important to
keep in mind that gas/liquid systems are mainly concerned in practice. The examples that we shall cons&r relate exclusively to
gas/liquidsystems. Inprincipe. however,liquid/liquidsystemscannotbeexcluded. It mustalsobepointedoutthat theliteratureis
themostabundantforgas/liquidsystemswithrespecttotransferparameters. Whencocurrentandcountercurrentsystemsarecom-
pared, the interfacial area adisplays the widest differences. Thus for a tickle-bed using spherical grains 1.5 10 2 mm in diameter,
thegatiiquid interfacial area reaches values between 1000 and 1,500 m-l , whereas in a countercu~ent system using packings over
5 mm innominaIsize.thevalueofaisalways lowerthan500m-1 (Puranik, 1974-Reichelt. 1974). As tothetransfercoefticie~~sk~
and h, it can be considered as a fist approximation chat their values depend more on Lhe vansferrcd compounds and on Ihe types of
phase present than on the fluid flow direction. Accordingly when comparing cocurrent and countercurrent arrangements for tie
samechemicalsystem,weshallimplicitly assumethatthetransfercoefficientsaceidentical. Notethat, while Ihenumberofmeasu-
rements and correlations concerning mass transfer coefficients found in the literature is relatively large, this cannot be said of the
heat transfer coefficients h. The Chilton-Colbum analogy is usually applied to calculate h from ki (Cheremisinoff. 1986 - Morsi,
1984,1989).
B13 Countercurrent two-phase Row fixed bed catalytic reactors 2271

Equation (5) includes the values Cs*, concentration of compcmnd Ai in phase I, which is in equilibrium with phase II, in which the
concentrationofthesamecompoundAiisC~ TocalculateCiI * , it is necessary to have a law which accounts for the equilibria bet-
ween phases. The simplest of these laws is Henry’s law. The Kicoefficients are often used for gas/liquid systems, to relate the molar
h-actionsxiandyi, intheliquidphaseandvaporphaserespeetively. Henceyi=$.xp Thisisthetypeofequation thatwe shallusein
our comparative calculations for specific reaction systems.
In comparing the cccurrent and countercurrent rurangements. we have shown that themost significant differences are varied: in the
balanceequations,in which thesignsofcertain termsarereversed,or concerning someimportantparameters,such as&andA,whosc
numerical values may differ by a factor of 2. The major difference will certainly be found when integrating equations (1) to (4). In
fact,theboundaryconditionstobeadoptedarefundamentallydifferent. Foracocurrentsystem.theboundarycondidonsaregiven
forz=O. TheintegrationofthisdifferentialsystemwithinitialboundaryconditionsiseasilyachievedusingastandardRunge-Kutla
procedure. The integration is stopped either for a reactor length L or for a given conversion of reactant Ai
The situation is completely different for acountercurrent system. The. boundary conditions are then distributed at the two ends ol
the reactor (Figure 1):

for z = 0: FiI = Fir (0) TI = TI (0) (7) forz=L: F~JJ = FiB Q TB = TB (L) (8)

Methods areavailable for solving this type of problem, such as the ‘shooting’ method or the use of finite differences. Nevertheless,
given the coupling between the heat and mass balances, the differential system is far from linear, so that conventional methods are
notreallyeffective. Themethodthatwefoundmostreliableconsistsofadoublealtemateintegration(Franks, 1972). Starting wilh
z = 0, a first integration is made for the equations relative to phase I, assuming that the variables relative to phase II are known. The
values obtained for thevariables relative to phase1 are stored for eat h integration step. The equations relative tc phase 11are then in-
tegratedfrom z=L,bystoringthevaluesofthevariablesforeachstep. Thesealtemateintegtationsare thenrcpcated inonedirection
andthenintheotheruntil thedesiredaccuracyisobtained. Itisclear thatthisprocedureofaltemateintegrationsrequircsmuchlon-
ger computer time than required to calculate a once-through cocurrent arrangement.

2 COMPARATIVE EXAMPLES

Weappliedtheequationspresentedabovetoanumberofreactionsystems which we felt tobe especially sensitive to thephasc flow

direction. We accordingly selected the following cases: Simple exothennic reaction; reversible reaction; reactions in parallel,
oneofthesereactionsproducingan inhibitingcompound. Thischoicemayappeararbitrary andiscertainiynotcomplete. ItmereIy
provides examples to highlight the potential value of acountercutrentanangement. Thecriterion that we selected to make our com-
parisons is the conversion of one of the reactants, taken as a reference, divided by the quantity of catalyst employed, or the specific
conversion.

Accordingly. for a cocurrent system:

Fil (0) + Fin (0) FiI (L) - Fin (L)

Specific Conversion (CSPCO) = (9)
L.S.c

By a similar calculation for a countercurrent system:

F, (0) + F, (L) - F, (0) - F, (L)
Specifii Conversion (CSPCC) = (10)
LS.E

The adoption ofthis criterion helps to account for substantial differences in the valueof e according to the type of arrangement. Other
comparative criteria can obviously be adopted.

2.1 Simple exothermic reaction

Inasimplercaction. ~j~=O,thechangeinthesystemc~bech~cterizedbyasinglep~~eter,themol~extent <‘,bythestan-

dard equation:

T=T~- j& (II)

L i
2272 P. TRAMBOUZE B13

In a cocurrent arrangement, both phases I and II enter at temperature TO and exitat an approximately equivalent temperature T. The
denominator term sl& is accordingly expressed as follows:
(12)
Eli= ~iIciI + ~~Iin

By contrast, in a countercurrent system, phase II enters for z = L at temperature TO, and leaves the reactor for z = 0 at a temperature
close to TO. The denominator term is thus reduced to BsEfi.

Let p denotes the ratio of total heat capacities of phases I and II.
Asanumericalexample.ifwetake: AT,=40@andp=2.6 thisleadsto ATcc =56”c. Assumingan activation energy of about 20
kcal/mol, the ratio of the specific conversions, assuming To = 3OOcIc, is: CSF’CUCSF’CO = 1.22.
The countercurrent reactor therefore requires a smaller amount of catalyst toobtain the same conversion. Yet care must be taken not
toexceedtheopemtingstabilitylimitof thereactor,leading toexcessivetemperaturesthatcouldbeverydetrimental tocatalyststabi-
lity.

2.2 Reversible reaction

Let us now consider a reversible reaction of the type: A 1 + A2 -A3 + A4 of w hich the heat of reac tion is sufficiently low to be ignored
(isothermal system).
The expression of the reaction rate is assumed to be conventional:

c31 . c41

Kl!

Itisalsoassumedthatthereaction prod~cts(A3andA~)arequitedifferent withrespecttotheirvolatilities. A4 isfoundpreferentially

in the vapor phase, while most of A3 remains in the liquid phase. This applies, for example. to the synthesis of methyl tertio butyl
ether (MTBE) (Gicquel. 1983) and ethyl tertio butyl ether @TBE), for which the reactive distillation technique has been proposed
and is used industrially. Other reactions can effectively be conducted by the same technique (Mikitenko. 1986). The results of the
comparison that we made between cocurrent and countercurrent systems are given in Table 1.

2.3 Reversible exothermic reaction

One example of a reaction of this type is the hydrogenation of heavy aromatics in a petroleum cut. The reaction to be carried out is
written as follows:
Ar-H+ZH20R-H AH = - 30 kcal/mol E = 20 kcal/mol
(Al) (A2) (Ag)
On a conventional catalyst, the chemical kinetics can be expressed as: r = k , C21(Clr - CIE) where

c31
C lE=
KE. @Hd2

We carried out various comparative calculations corresponding to a hydrocarbon cut with the following molar flows :
aromatics 6 naphtenes 1 paraffins 3-13

The hydrogen was introduced in gaseous form and under pressure (100 bar), at the rate of 5.1 or 10 mol/mol hydrocarbon, or about
450or88OVl. Byvaryingtheva1uesassumedfork,KEandTg,anumberofcomparisonsweremadebetweencocurrentandcounter-
current systems. Note that, for a cocurrent system, the specific conversion, as defined by equation (9). reaches a peak at a certain
conversion rate. We took this maximum value as a reference for our comparison, which is given in Table 2.

2.4 Two paraIle1 exothermic reactions, one producing an inhibiting compound for both reactions

The situation occurs when a hydrocarbon cut containing sulfur

derivatives is treated with hydrogen. The following reactions are si-
multaneous (L-e Page. 1987 - Schuit. 1973):

Ar-SH + 3H2 + AR-H + H2S ArH+2H2 N R-H

CA4> (A,) (Al) (A5) (A1) (A2) (A3)
B13 Countercurrent two-phase flow fixed bed catalytic reactors 2273

The kinetics of thesereactions wereanalyzedusing several hydrolreating catalysts. Inhibition caused by H2S was usually identified.
The kinetic equations can accordingly be written:
-El/W
x4 -pH2c
q = k1
1 +B, -x5

Thecoefficients~Iancl~reflecttheintensityofinhibitionbyH2S. Itcanbeassumed.forexample,that~I=4and~=1OO.ElandE2
havesimilarvalues: El = 17 kcal/mol.andE2= 2Okcal/mol. Theenthalpiesofbothreactions are approximately thesarne: AH= - 30
kcallrnol.
We also made a comparison between cocurrent and wuntercurrent operation. based on the following feed composition:
FlI=6 F31= 1 F41 = 0.5 incrts (paraffmic hydrocarbons) = 2.5 - 10
The results of the cocurrent/countercmt comparison are given in Table 3.

3 DISCUSSION AND CONCLUSIONS

In our presentation of the comparison between cocurrent and countercurrent two-phase flow catalytic reactors, we intentionally res-
tricted ourselves to a relatively simple model, but one taking account of the essential factors to ensure a valid comparison. It is clear
that the nature of the chemical system concerned, whether from the kinetic or thermodynamic standpoint, is of considerable impor-
tance. Our model can take account of any type of reactionsystem. The effect of the flow dire&on is es.sentially felt on transfers bet-
weenphases. whclherbymasstransferorbyheattransfer. Thisis why weintroducedthesetransferequationsintoour model. On the
other hand. we did ry)t consider transfer between fluids and panicles and intra-particle diffusion. These transfers are in fact affecicd
only slightly. if at all. by the fluid flow direction.

Ourexampleswereobviouslyselectedu,demonseateasignif~anteffectofthephaseflowdirection. Yettheexamplesarenotfar-fet-
chedbut.fartothecontrary,typicalcasesofwidespreadindustrialoperations. As weemphasized. thedifferencebetwcenthetwosys-
temsrelatestotheshuctureof thefixedbedemployed. In thecocurrentsystem,currentpractice is to use spherical or cylindricalcata-
lyst particles. between 1 and 5 mm in diameter. These particles cannot be usedin countercurrent systems. A recent Mobil Oil patent
concerning a countercurrent hydrocracking operation clearly stipulates that the catalyst grains must be over 5 mm in diameter (Pa-
tent, 1988). Even with these particles, the serious drawback of the countercUITent system is its lack of flexibility with respect to the
flowrateofthefluidphasesandtheriskofflooding. Thepatentmentioncdaboveaimsprimarilyto controlthecccurenceofflooding
bymeansofaliquidleveldetectoratthetopofthecatalystbed. Tocontendwiththisdiificulty, webclievethatthemostrationalsolu-
tionis~useaspecially-shapedcatalyst.asroutinelydoneinabsorptioncolumns(RaschigringsorBerlsaddlesetc). Thisisimplicit-
lyas~umedinourcalculations. Thisleads totheadoptionofmuchlowerv~uesforeandRinthecountercurrentsystem thanintheco-
currentsystem. Thus.eveniftheadvantagecalculatedonthebasisofspecificconversionpe~unitcatal,itisdiminis-
bed if the comparison is made on the basis of reactor volume. The preparation of catalysts in the form of Rasc hig rings or Berl saddles
may as10 appear as an added difficulty. This shaping is possible for certain catalysts (some already exist) but undoubtedly cannot be
extended to all types of catalyst.

(nconclusion.wefeelthat.forcertainreactionsystems.thevalueofacoun~urrentarrangement maybeimportantenoughtojustify
the creative effortmade to overcome the difficulties that could arise in the use of equipment in countercurrent operation. This demon-
siration is in fact the aim of this paper.
2274 P. TRAMBOUZE B13

NOTATION
Sl- interfacial contact area per unit volume ( m2/m3)
C+, CIII - concentration of compound Ai in phase I or II ( mol/m3)
C+. CD - molar heat capacity of compound Ai in phase I or II (average value) (cWmo1.K)
Ej - activation energy for the reaction j ( cal/mol.)
FiI, FaI - molar flow rate of compound Ai in phase I (mol/s.)
k - reaction rate constant
ki - mass transfer coefficient between phases I and II (m/s.)
Ki - vapour liquid equilibrium coefficient
KE - chemical equilibrium constant

h - heat transfer coefficient between phases I and II (caI/m3.K)

L - total length of the catalytic bed (m)
S - cross section area of the catalytic bed ( m2)
TI, TII - temperature of phase I or II K
Xi - molar fraction of compound Ai in liquid phase
yi - molar fraction of compoind Ai in gaseous phase

Table I. Reversible Isothermal Reaction (k = 1; KE = 1)

COCURRENT COUNTERCURRENT

G=30 CSPCO = 6.62 6.04 CSPCC = 18.1 7.6

L=30 CONVERSION(l) = 16.4 % 22.9 8 CONVERSION( 1) = 15.2 % 21.8 %

Table 2. Aromatics Hydrogenation (k = IO; KE = 0.0025)

COCURRENT COUNTERCURRENT

G=lOO CSPCO(MAx) = 2.31 CSPC = 3.67 3.92

L= 20 CONVERSION(I) = 61.6 % CONVERSION(l) = 73.5 % 59.7 %

G=Sl CSPCO(MAX) = 5.47 CSPC = 7.33

L= 12 CONVERSION(I) = 66.3 % CONVERSION( 1) = 59.01 %

Table 3. Two Pan-ale1 Reactions Inhibited by One Product (kl = 6; k2 = 5; KE = 0.025)

COCURRENT COUNTERCURRENT

G=lCO CSPCO = I.25 CSPCC = 1.9

L= 20 CONVERSION(l) = 35 % CONVERSION( 1) = 34.5 %

G=51 CSPCO(MAx) = 3.49 3.05 CSPCC = 4.14

L= 12 CONVERSION(l) = 78.5 % 50.5 % CONVERSION( 1) = 50.6
B13 Countercurrent two-phase flow fixed bed catalytic reactors 2275

REFERENCES

Cheremisin0ff.N.P. (1986),Handbookofhearandmass~~er. VolumeZ. MassTransferandreactordesign. GulfPublishingCo.

Houston.
Franks. R.G.E. (1972). Modeling andsimulation in chemical engineering. Wiley. Interscience. New York.
Gicquel, A. and N. Torck (1983). Synthesis of MTBE catalyzed by ion-exchange re.sin. Journ. of Catalysis, fi. 9-18.
Goto. S. and J.M. Smith (1978 a). Performance of slurry and trickle-bed reactors. AI.C.h.E. Jour. 24. n02, 286-293.
Goto, S. and J.M. Smith (1978 b). Analysis of three-phase packed bed reactors. A.I.ChE. Jour. a, n02, 294-302.
Le Page J.F. and ~011. (1987). Applied heterogeneous catalysis. Editions Technip. Paris.
Mikitenko P. (1986). La distillation r&ctive : ptincipe. applications et perspectives. Pelrole et Techniques=, 34-38.
Mills,P.L..E.G.Beaudry andM.P.Dudukovic(1984). Compat-isonandpredictionofreactorperforrnsnceforpackedbeds with IWO-
phase flows. I. Chem. E. Symposium. Series n”87,527-534.
Morsi. B.I. and al. (1984). Hydrodynamics and gas-liquid-&id interfacial parameters of cocurrent downward two-phase flow in
trickle-beds. Chem. Eng. Commun. a, 267-293.
Morsi, B-1. (1989). Mass transfer coefficients in a trickle-bed reactor. The Chem. Eng. Jour. 41.41-48.
NG, K.M. and C.F. Chu (1987). Trickle-bed reactors. Chem. Eng. Progr. (nov. 1987), 55-63.
Patent (1988). U.S. patent4.757.281. Mobil Oil Co. issued 5/7/88.
Puranik, S.S. and A. Vogelpohl. (1974). Effective interfacial area in irrigated packed towers. Chem. Eng. Science, 29, 501-
507.
Reichelt, W. and E. Blass (1974). Die Berechnung von FtillMrperapparaten bei Gegensuom von Gas und Flitssigkeit. Chemie. Ing.
Technik, 2,171-222.
Reilly.J.W., M.C. Sze,U. Saranto andU.Schmidt (1973). Aromaticreductionprocess is commercialized. The Oil andGasJourn.
(Sept. 17.1973). 66-68.
Schuit, G.C.A. and B.C. Gates (1973). Chemistry and engineering of catalytic hydrodesulfurization. A.I.Ch.E. Journ. 1p, 4 17-
i38.
Smith, L.A., E.M. Jones and D. Hcam (1987). Working with catalytic distillation. Paper presented atA.I.Ch.E. 1987 SummerNa~.
Ueeting. (Aug. 1987).

F F
Fi II it

BED

cocuRRENr
‘r-f
CXXNEFClJRRENl
F
i II

Figure 1 : Basis of the comparaison of the two types of reactor