ie050891p

8/2/2019 ie050891p

1/8

Kinetic Modeling and Simulation of the Selective Hydrogenation of

the C3-Cut of a Thermal Cracking Unit

Bo Wang and Gilbert F. Froment*

Department of Chemical Engineering, Texas A & M University, College Station, Texas 77843-3122

The kinetics of the gas phase selective hydrogenation of methyl acetylene (MA) and propadiene(PD) over a Pd/alumina catalyst were investigated in a fixed bed tubular reactor at temperatures60-80 C and a pressure of 20 bar. Hougen-Watson type kinetic equations were derived. Theformation of higher oligomers slowly deactivated the catalyst. The effect of the deactivatingagent on the rates of the main reactions as well as on the deactivating agent formation itselfwas expressed in terms of a deactivation function multiplying the corresponding rates at zerodeactivation. Finally, the kinetic model was plugged into the reactor model to simulate anindustrial adiabatic reactor.

1. Introduction

Propylene is an important building block in thechemical industry. It is used to make two widely appliedpolymers (polypropylene and polyurethane) as well as

a range of chemical intermediates. Thermal cracking ofhydrocarbons is the major source of propylene produc-tion. The C3-cut obtained by the steam cracking ofethane, propane, and naphtha typically contains morethan 90% propylene and up to 6% methyl acetylene(MA) and propadiene (PD). The MAPD content in thepropylene stream depends on the type of feed and onthe severity of the steam cracking. MAPD has to beremoved from the C3-cut in order to obtain propylenethat meets the specifications for the subsequent polym-erization. Selective hydrogenation of the C3 -cut is theestablished route for this. If properly conducted, it alsobenefits the overall propylene production of the plant.

MAPD selective hydrogenation has been a topic ofconsiderable research over the past several decades.

Most of the work concentrated on catalyst development.Catalysts investigated in the selective hydrogenation ofMAPD ranged from the initial nickel sulfide1 andnickel-tungsten sulfide2 to other supported metal cata-lysts, such as nickel, copper, zinc, palladium, platinum,or mixtures thereof, dispersed on silica or alumina.3-5

Palladium is the most active and selective metal for thehydrogenation of alkynes to the corresponding olefins.Presently, all industrial catalysts used in selectivehydrogenation are bimetallic catalysts with a palladiumactive phase supported on alumina and promoted byanother metal. These bimetallic catalysts promote selec-tive hydrogenation more efficiently and more economi-cally than monometallic ones.

Both gas phase and liquid phase hydrogenation ofMAPD are in use in industry. Liquid phase selectivehydrogenation has been developed recently. It is be-lieved that the liquid phase operation offers someadvantage over the gas phase process in operating costas well as catalyst life, due to the removal of producedoligomers from the catalyst surface by the liquid flow.Industrially, the hydrogenation may take place eitherbefore (front end) or after (tail end) the fractionation

unit which is used to remove H2, CH4, and CO from theC3-cut. The essential feature of the tail end selectivehydrogenation is the relatively pure feed and theaccurate control of the feed composition.6

Relatively little work has been performed on the

kinetics of the selective hydrogenation of the C3-cut.Villora et al. reported the kinetics of gas phase selectivehydrogenation of a C3-cut as well as the C2-C3 mixtureat atmospheric pressure.7,8 The kinetics were expressedin terms of empirical power law models. Green oilformation was not reported. Uygur et al. investigatedthe kinetics of liquid phase selective hydrogenation ofa C3-cut.9 A Hougen-Watson model was used to modelthe reaction kinetics. The experiments were performedin the presence of internal and external mass transferresistance, but these limitations were accounted for inthe derivation of the rate equation.

The selective hydrogenation of a C3-cut is veryexothermic, and CO is added to the feed when the

catalyst is fresh, to moderate its activity. The catalystslowly deactivates, because of a deactivating agentformed by oligomerization side reactions. Control of theselectivity in MAPD selective hydrogenation is veryimportant because propylene can be consecutively hy-drogenated to form propane if operating conditions suchas the H2/MAPD molar ratio or the temperature are notcontrolled properly. In this case, the overall selectivityfor propylene may be negative, meaning a loss ofvaluable propylene. In addition, this is a dynamicprocess due to the catalyst deactivation associated withgreen oil formation. The deactivation of the catalyst withtime imposes a continuous adaptation of the operatingconditions to maintain the optimal selectivity. Thedeactivated catalyst needs to be periodically regener-

ated. Operating variables must be optimized taking intoaccount selectivity, run length, and regeneration costin order to obtain a maximum profit.

Kinetic equations for the hydrogenation as well as forthe green oil formation are valuable tools for theimprovement of the process performance. So far no workhas been done to determine the kinetics of the greenoil formation and the catalyst deactivation.

The purpose of this work was to develop a kineticmodel for this process. The operating conditions weresimilar to an industrial tail end gas phase hydrogena-tion process. The kinetic model was of the Hougen-

* To whom correspondence should be addressed. E-mail:[email protected].

9860 Ind. Eng. Chem. Res. 2005, 44, 9860-9867

10.1021/ie050891p CCC: $30.25 2005 American Chemical SocietyPublished on Web 11/02/2005

8/2/2019 ie050891p

2/8

Watson type, i.e., accounting for the adsorption ofpropylene, MA, PD, and green oil. The deactivation of

the catalyst is accounted for by linking the activity tothe content of the deactivating agent.10

2. Experimental Section

2.1. Feed Mixtures. The C3 mixture, similar incomposition to that of the industrial C3-cut, was pur-chased from Scott Specialty Gases. The composition ofthe C3 mixture was as follows: 1% PD, 2% MA, 96.8%C3H6, 0.17% C3H8, and 0.03% CH4. The purity of theH2 and N2 gas was 99.99%.

2.2. Catalyst. The catalyst used was C31-1 suppliedby SUD-CHEMIE. C31-1 is a Pd-based hydrogenationcatalyst. It contains 0.03 wt % Pd with < 0.1 wt % otherheavy metals as promoter, coated as a 200 m layer on

the 2-4 mm -Al2O3 core. Prior to reaction, the catalystwas activated by the removal of water from the catalystand the subsequent reduction of palladium oxide on thesupport to palladium black. The removal of water wascarried out by circulating nitrogen through the reactorat 150 C for 2 h. The reduction was conducted byfeeding a quantity of hydrogen diluted with nine vol-umes of nitrogen at 150 C for 2 h. After reduction, thecatalyst bed was cooled to ambient temperature bycontinuing the nitrogen purging to ensure a hydrogen-free atmosphere in the reactor prior to start-up of thereaction.

For fixed bed catalytic reactors, the ratio of the tubediameter to the particle diameter, dt/dp, should be atleast 10 to approximate plug flow. The commercialC31-1 catalyst is a sphere with diameter between 2 and4 mm, and the reactor i.d. is limited to 0.45 in. forreasons of heat transfer area per weight of catalyst. Theratio requirement is not satisfied. Therefore, the egg-shell catalyst was crushed, sieved to different sizefractions, and analyzed by an inductively coupled plasmaspectrometer. It was shown that the Pd content of thevarious sieve fractions was not identical. To obtainparticles representative of the commercial catalyst, butalso satisfying the requirements of a kinetic study withbench scale equipment, the different sieve fractions weremixed and repelletized again, evidently to a size muchsmaller than the original. The hydrogenation was thentested with catalyst pellets of this type but having

different sizes. It was found that the internal mass andheat transfer resistance were negligible for particle sizes

smaller than 0.20 mm, and these were retained for thekinetic study. Also, the operating conditions were chosenby calculation to avoid external mass and heat transferresistance.

2.3. Reactor Unit. Figure 1 shows the experimentalsetup. A reliable and flexible monitoring and supervi-sory control unit was developed by combining FieldPointdistributed I/O modules and a PC that used a LabVIEWapplication. The unit monitored and controlled processvariables such as feed flow rate, total pressure, andtemperature.

The reactor feed consisted of four streams, i.e., H2,C2H6, the C3 mixture, and N2. Ethane was used as aninternal standard. The flow rates of H2, N2, and C2H6

were controlled by a gas mass flow controller (BrooksInstrument 5850i). The flow rate of the pressurized C 3mixture was controlled by a liquid mass flow controller(Brooks Instrument 5881). The C3 mixture was vapor-ized in the mixing chamber and, then, mixed with othergas streams inside this chamber before flowing to thereactor. The reactor was heated by an oil bath. Thereactor temperature profile was monitored by a slidingthermocouple, mounted inside the reactor along thereactor axis. The reactor pressure was controlled at theexit by a back pressure regulator. In the kinetic study,the temperature and pressure were varied in the rangeof 60-90 C and 16-20 bar, respectively. Under theseoperating conditions, the C3 mixture was gaseous.

The analytical section consisted of two on-line gaschromatographs (GCs) for the analysis of both the feedand the effluent of the reactor. A flame ionizationdetector (FID) GC with a 15 m 0.53 mm 15 m HPPLOT/Al2O3 capillary column was used to analyze C3and green oil (up to C12). A thermal conductivity detector(TCD) GC was equipped with 5A (6 ft 1/8 in. o.d.) and13X (6 ft 1/8 in. o.d.) packed columns to analyze H2.5A was used as a precolumn to trap hydrocarbons andneeded to be periodically regenerated. An internalstandard method was used. As the relative responsefactor for each component with respect to the internalstandard was known, each identified peak could becalculated in an absolute way, so as to give the correctamount of each component in the exit flow.

Figure 1. Experimental setup for the selective hydrogenation of the C3-cut from a thermal cracking unit.

Ind. Eng. Chem. Res., Vol. 44, No. 26, 2005 9861

8/2/2019 ie050891p

3/8

Both CH4 and C2H6 were tested as internal standards.It was found that C2H6 was far better than CH4 inregard to accuracy and repeatability.

3. Experimental Results

3.1. Definitions of Conversions and Selectivities.Conversion of MA

Conversion of PD

Conversion of MA and PD into propylene

Conversion of MA and PD into propane (PN)

Conversion of MA and PD into green oil (GO)

Selectivity for products

3.2. Experimental Operation and Observations.The MA and PD conversions declined from the start ofthe run onward due to catalyst deactivation, as il-lustrated in Figure 2. The kinetic modeling of theprocess, therefore, involved two aspects: the kineticsof the main reactions and the kinetics of deactivation.It was decided to partition the problem and to use theinitial data for the first aspect and the complete dataset collected in a run, as shown in Figure 2, for thesecond. Each run was carried out with fresh catalyst to

avoid having to deal with different dispersions of themetal compound of the catalyst caused by regeneration.A rigorously standardized catalyst pretreatment pro-duced samples with identical initial activity. To studythe deactivation under sufficiently varied conditions, therun length was chosen to cover a 10-20% decrease inconversion. The reactor effluent was analyzed at timeintervals of 1 h. Each run lasted for 10-20 h, dependingupon the operating condition, i.e., upon the deactivationrate of the catalyst. The amount of deactivating agentwas determined at the end of the run by combustion.Each conversion measured at the exit was used in theparameter estimation. The catalyst deactivation sig-nificantly changed the amount of heat generated. Tomaintain a constant temperature profile during the run,the oil bath temperature needed to be adjusted fre-quently. A set of 21 runs was performed, covering a widerange of temperatures, H2/MAPD molar ratios, and runlengths, so as to gather information on the deactivation.Typical examples illustrating the effects of temperature;space time, W/F0MAPD; and the molar ratio of H2/MAPDare given in Figures 2 and 3. The run length isexpressed in terms of the total amount of MAPD fedper kg of catalyst, F0MAPD/W.

4. Reaction Scheme

In the hydrogenation process over Pd/-Al2O3 catalyst,two kinds of active sites, i.e., metal sites and acid sites,

Figure 2. Evolution of the conversions as functions of the run length, expressed as F0MAPD/W, at 60 C and W/F0MAPD ) 18.29 kgh/kmol. H2/MAPD molar ratio: (A) 1.5; (B) 1.1.

Figure 3. Evolution of the conversions as functions of the run length for H2/MAPD ) 1.5 (molar ratio) and W/F0MAPD ) 4.06 kg h/kmol:(C) 60 C; (D) 80 C.

XMA )moles of MA in - moles of MA out

moles of MA in

100

XPD )moles of PD in - moles of PD out

moles of PD in 100

XPP )moles of PP out - moles of PP in

moles of MAPD in 100

XPN )

moles of PN out - moles of PN in (impurity in C3)moles of MAPD in

100

XGO )moles of GO out

moles of MAPD in 100

selectivity )moles of product i formedmoles of MAPD converted

100

i refers to PP, PN, or GO (1)

9862 Ind. Eng. Chem. Res., Vol. 44, No. 26, 2005

8/2/2019 ie050891p

4/8

are involved. The metal sites catalyze hydrogenationreactions, but the acidity of the alumina catalyzesoligomerization of the olefins, thus producing the so-called green oil. Three possible reaction schemes, whichdiffer only in the mechanism of the green oil formation,with a carbon number below 12 were derived. In Scheme1, green oil is formed out of MA and PD and hydroge-nated afterward. In Scheme 2, green oil is formed outof propylene only. In Scheme 3, green oil is formed notonly out of MA and PD but also out of propylene.

5. Kinetic Modeling

5.1. Rate Equations for Hydrogenation and Oli-gomerization. In all three schemes, MA and PD arehydrogenated over Pd active sites to form propylene,which is subsequently hydrogenated to form propane.When the surface reaction between dissociatively ad-sorbed H2 and MA or PD is the rate determining step,the hydrogenation rates on Pd can be written, in theabsence of deactivation, as the following:

The denominator appearing in the reaction ratesr10, r20, and r30 is [1 + (KH2PH2)1/2 + KMAPMA + KPDPPD+ KPPPPP + KPNPPN]3.

Acid sites are responsible for the formation of greenoil. The reaction rates for the formation of green oil fromMA and PD in Scheme 1 are as follows:

The denominator in the above reaction rates is [1 +

(KH2PH2)1/2

+ KMAPMA + KPDPPD + KC6PC6 + KC9PC9]3

.The reaction rates of the green oil formation frompropylene in Scheme 2 are written

with ) [1 + KPPPPP + KC6PC6 + KC9PC9 + KC12PC12]2.In Scheme 1, the equations defining the net rate of

formation for the various components are

In Scheme 2, the equations defining the net rate offormation for the various components are

Scheme 3 is actually the combination of Schemes 1 and2. In Scheme 3, the equations defining the net rate offormation for the various components are

5.2. Parameter Estimation. In the absence ofdeactivation, the set of steady-state continuity equationsfor the components in a plug flow reactor can be written

Scheme 1. Reaction Scheme with the Formationof Oligomers out of MA and PD

Scheme 2. Reaction Scheme with the Formationof Oligomers out of Propylene

Scheme 3. Reaction Scheme with the Formationof Oligomers out of MA, PD, and Propylene

r10 ) kMA

0KMAKH2(PMAPH2- PPP/K1)/

r20 ) kPD

0KPDKH2(PPDPH2 - PPP/K2)/

r30 ) kPP

0KPPKH2(PPPPH2- PPN/K3)/ (2)

rMA-C60 ) kMA-C6

0PMA2PH2

2/

rMA-C90 ) kMA-C9

0PMAPC6PH2/

rPD-C60 ) kPD-C6

0PPD2PH2

2/

rPD-C90 ) kPD-C9

0PPDPC6PH2/ (3)

rPP-C60 ) kPP-C6

0KPP2PPP

2/

rPP-C90 ) kPP-C9

0KC6KPPPC6PPP/ (4)

RMA ) r1 + 2r4 + r5

RPD ) r2 + 2r6 + r7

RPP ) r1 + r2 - r3

RPN ) r3

RC6 ) r4 + r6 - r5 - r7

RC9 ) r5 + r7 (5)

RMA ) r1

RPD ) r2

RPP ) r1 + r2 - r3 - 2r4 - r5

RPN ) r3

RC6 ) r4 - r5

RC9 ) r5 (6)

RMA ) r1 + 2r6 + r7

RPD ) r2 + 2r8 + r9

RPP ) r1 + r2- r3 - 2r4 - r5

RPN ) r3

RC6 ) r4 + r6 + r8 - r5 - r7 - r9

RC9 ) r5 + r7 + r9 (7)

dXi

dW/F0MAPD) Ri i ) 1, 2, ..., 6 (8)


8/2/2019 ie050891p

5/8

The integral method was used for the kinetic analysis.The above set of differential equations was numericallyintegrated by a fourth-order Runge-Kutta method. Theparameter estimation was based on the Marquardtalgorithm for multiple responses, minimizing the objec-tive function (Xiexp - Xi)2. The number of parametersin the above three models amounts to 28, 24, and 38,respectively. The parameter estimation was performedon all initial data simultaneously at all temperaturesby directly substituting the temperature dependence of

all the parameters into the corresponding continuityequations. To facilitate the parameter estimation, thefollowing reparametrization was carried out, where Tmwas the average temperature of the experiments.

The discrimination between rival models was basedon the requirement for all parameters to be positive andsignificant and on the residual sum of squares as a test

for the fit of the data. The model based on Scheme 1fits the data best. The sum of squares of residues forthe three models is 0.08, 0.13, and 0.12, respectively,so that model 1 was retained for the main reactions inthe absence of deactivation. Figure 4 compares theexperimental conversions obtained at 70 C with thevalues simulated using model 1.

6. Catalyst Deactivation

6.1. Observations. Due to catalyst deactivation, theconversion at the outlet decreased with the run length.The catalyst deactivated faster when the content in theexit flow of oligomers, called green oil, was high. Thedeactivation rate declined with the run length. Each runcovered at least a 10-20% decrease of the conversionof MA and PD, so as to permit an accurate modeling ofthe deactivation. Each run lasted for 10-20 h, depend-ing on the operating conditions and, therefore, on thedeactivation rate of the catalyst. The reactor effluentwas analyzed at intervals of 1 h. The deactivating agentcontent was determined at the end of the run bycombustion. Typical results at 80 C are shown in Figure5.

Figure 4. Comparison of experimental and simulated conversions as functions of the space time at 70 C: () experimental; (solid line)simulated.

k ) A exp(-ERT) ) A* exp[-ER (

1T-

1Tm)]

K) A exp( HRT) ) A* exp[H

R(1T-

1Tm)] (9)


8/2/2019 ie050891p

6/8

The formation of the deactivating agent is affectedby such factors as conversion; process time, expressedasFMAPD/W; and the H2/MAPD molar ratio. The higherthe conversion and the longer the process time, the

higher the content of the deactivating agent. Moreover,as seen in Figure 5, the deactivating agent content at aH2/MAPD molar ratio of 1.1 is higher than that at 1.8,even at a shorter process time. A high H2/MAPD molarratio reduces the formation of the deactivating agent.

6.2. Kinetic Modeling of Deactivation. The GCanalysis detected green oil up to C12 in the reactoreffluent. Therefore, the deactivation was ascribed toeven heavier oligomers, with carbon numbers exceeding12 and represented here by C. These higher oligomerswould not elute from the reactor under the operatingconditions of this study and, in other words, would beirreversibly adsorbed on the catalyst sites and act asdeactivating agents.

A C15 oligomer can be formed in various ways out of

lower oligomers and propylene:

The initial rate of formation of the deactivating agentcan be written

Because the amount of C9 and C12 is so small, rC0 wasreduced to k1PC6PPP3. The effect of the deactivatingagent on the main reaction as well as on its own

formation was expressed in terms of a deactivationfunction,i, multiplying the corresponding rates at zerodeactivation. Froment studied the kinetics of catalystdeactivation in hydrocarbon transformation reactionsand justified the use of an exponential deactivationfunction.11,12 An exponential function in terms of thecontent of the real deactivating agent, C, not of time,was used. That establishes a link between the deactiva-tion and the operating conditions but requires a rateequation for the formation of the deactivating agent.

Referring to Scheme 1, the rate equations for the mainreaction can be written in the presence of deactivation:

The rate of formation of the deactivating agent (11)becomes

In the above equation, CC is written in terms of theamount of carbon per unit weight of catalyst. Theamount of the deactivating agent was determined by

combustion at the end of the run, as if it were coke.When the catalyst deactivates at a point in the

reactor, the conversion at that point is affected. Con-sequently, the conversion profile is modified with timeand the reactor is operating in non-steady-state condi-tions. The continuity equations for component i and theevolution with time of the content of the deactivatingagent of the catalyst, CC, can be written

If the deactivation is not too fast, the term Fi/t can beneglected.

6.3. Estimation of the Deactivation Parameters.It follows from Scheme 1 that accounting for thedeactivation introduces 10 deactivation parameters forthe hydrogenations and oligomerizations: Ri (i ) 1, 2,..., 7) in (12) and RC in (13) as well as the frequencyfactor AC and the activation energy EC in the rateequation for the formation of the deactivating agent (11).The content of the deactivating agent is available at theend of a run, so that the objective function includes notonly the residual sum of squares of each componentresponse during the run but also that of the deactivatingagent at the end of each run. The number of runs usedfor the estimation of the deactivation parametersamounts to 21 with each response of MA, PD, PP, PN,C6, and C9 taken every hour and C taken at the end ofthe run (Figure 5). The values of the kinetic parametersderived from the estimation based upon the initial ratedata were fixed in the present estimation. The plots ofthe experimental and simulated values as a function ofspace time at different temperatures accounting for thedeactivation are similar to those of Figure 4 and arenot shown here. The model provides a good fit of theexperimental data.

7. Simulation of an Industrial HydrogenationUnit

The selective hydrogenation process to be simulatedoperates in the gas phase and is carried out in a fixedbed adiabatic reactor. The reaction is very exothermic,and CO is added industrially to the feed when thecatalyst is fresh, to moderate its activity. The catalystis slowly deactivated in this process, due to oligomerformation with Cn >12. To compensate for the loss ofcatalyst activity and maintain satisfactory yields, thereactor temperature needs to be increased correspond-ingly, and sometimes H2/MAPD and space time alsoneed to be adjusted during the run. When the catalysthas lost too much of its activity, it needs to be regener-ated by careful combustion to avoid excessive temper-atures that could damage the catalyst.

Figure 5. Dependence of the content of the deactivating agentof the catalyst on conversion. The amount of MAPD fed perkilogram of catalyst (shown inside the figure) and the molar ratioH2/MAPD were measured at 80 C. H2/MAPD ) 1.5, except for apoint at F0MAPD/W) 57.4, for which the ratio ) 1.1, and a pointat F0MAPD/W) 82, for which the ratio ) 1.8.

rC) rC0 exp(-RCCC) )

AC exp(-EC/RT)PC6PPP3 exp(-RCCC) (13)

{Fi

t+

FtMm

Fg

Fi

Z)

FtMmFBRiFg

CC

t) rC

(14)

C6 + 3PPC9 + 2PPC12 + PP

f C (10)

rC0 ) k1PC6PPP

3 + k2PC9PPP2 + k3PC12PPP (11)

ri ) ri01 ) ri

0 exp(-RiCC) i ) 1, 2, ..., 7 (12)


8/2/2019 ie050891p

7/8

The effect of the addition of CO at the start of therun was accounted for by multiplying the rate equationsfor the hydrogenation by a factor exp(-COPCO). The COis different for the various reactions so that there arethree different CO values, i.e., 1, 2, and 3, whichcorrespond to MAPD hydrogenation, propylene hydro-genation, and GO formation reaction, respectively (seeScheme 4).

Because internal diffusion limitations are negligible

with the eggshell catalyst a pseudohomogeneous modelwith plug flow was used to simulate the industrialadiabatic reactor. An energy equation needed to beadded to (14)

with the initial and boundary conditions:

The continuity equations and the energy equationwere integrated simultaneously with respect to t (runlength) and Z (reactor length). The history of the unitover 6360 h was accounted for in the calculation of theamount of the deactivating agent, CC, using the rateequation for its formation and accounting for its deac-tivation.

The calculated conversions and selectivities are plot-ted in Figure 6 as a function of bed length. Theagreement with the industrial exit conversions is excel-lent. It can be seen that MA and PD are alreadycompletely converted after 10% of the total bed length.The industrial reactor is oversized in order to compen-

sate for catalyst deactivation. It is also observed thatthe conversion into propylene reaches a maximum atabout 10% of the bed length, beyond which it decreasesbecause of hydrogenation into propane. The oversizingsignificantly penalizes the propylene production of thehydrogenation process.

8. Conclusions

A relatively detailed reaction scheme was found to be

necessary for the description of the essential featuresof the gas phase selective hydrogenation of the C3-cutof a thermal cracking unit. It considered not only thehydrogenation reactions on metal sites but also theoligomerization on acid sites leading to green oil, i.e.,oligomers with varying C number. The kinetic modelexplicitly accounted for the adsorption of MA, PD,propylene, propane, and green oil components and alsofor the deactivation of the catalyst, by linking theactivity to the true deactivating agent, instead of justto time. Such an approach requires a reaction schemeand a rate equation for the formation of the deactivatingagent. Because C6-C12 oligomers eluted from the reac-tor, higher oligomers with a carbon number of 15 or

higher were taken to be responsible for the deactivationthrough irreversible adsorption on the acid sites.The kinetic model was plugged into a pseudohomo-

geneous reactor model to simulate the industrial adia-batic reactor at a given point in time. For a commercialprocess this cannot be done without accounting for thehistory of the plant. The effect of both catalyst deactiva-tion and feeding CO at the start of the run had to beincluded to produce a good fit of the industrial data. Thesimulation also revealed certain consequences of thechoices made in the design stage of the reactor, associ-ated with the deactivation of the catalyst and a policyof long operating cycles between catalyst regeneration.

Acknowledgment

The authors are grateful to Dr. R. G. Anthony forstimulating discussions and to Dr. J. Grootjans and Dr.D. Mertens for providing valuable advice and data. B.W.is grateful to Total Petrochemicals for a postdoctoralfellowship.

List of Symbols

MA ) methylacetylenePD ) propadienePP ) propylenePN ) propaneGO ) green oil

Ai ) frequency factor for elementary reaction step i, kmol/

(kg cat. s)AC ) frequency factor for formation of the deactivatingagent, kmol/(kg cat. bar4 s)

cpi ) specific heat of fluid i, kJ/(kg K)Ci ) molar concentration of component i, kmol/m3

CC ) deactivating agent content of the catalyst, kg C/kgcat.

dt ) internal tube diameter, mdp ) particle diameter, mW/F0MAPD ) space time, kg cat. h/kmol

Ei ) activation energy for elementary reaction step i, kJ/kmol

EC ) activation energy for formation of the deactivatingagent, kJ/kmol

F0MAPD ) molar feed rate of reactants MA and PD, kmol/h

Figure 6. Conversions of MAPD: selectivities for PP, PN, andGO as functions of the catalyst bed length in an industrialadiabatic reactor after 6360 h. Points are industrial results.Process conditions: PD content of feed (wt %), 0.855%; MA contentof feed (wt %), 1.319%; H2/MAPD molar ratio, 2.37; inlet T, 331.3K; exit T (simulated), 383.8 K; exit T (industrial), 378 K.

Scheme 4. Definition of Parameters Expressingthe Effect of CO on the Various Reactions

i)1

N

Ficpi

T

Z+

i)1

N

CicpiT

t) [

j)1

R

rj(-H)j]FB (15)

CC) 0, at t ) 0, all Z

Fi) F

i

0 and T) T0, at Z ) 0, all t, i ) 1, 2, ..., 6


8/2/2019 ie050891p

8/8

ie050891p

Documents

Transcript of ie050891p