Catalyst Evaluation and Kinetic Study for Ethylene Production

524 Ind Eng. Chem. Res 1989,28, 524-530Catalyst Evaluation and Kinetic Study for Ethylene ProductionAngeliki A. Lemonidou* and Iacovos A. VasalosDepartment of Chemical Engineering and Chemical Process Engineering Research Institute, P.O. Box 1951 7,University City, 54006 Thessaloniki, GreeceEugene J. Hirschberg and Ralph J. BertolaciniAmoco Research Center, Naperuille, Illinois 60566The need to produce olefins via catalytic cracking has attracted increased attention in recent years.In this paper results are presented with catalysts composed of two or more of the following metaloxides: CaO, A1203, SrO, MgO, Ti02, Mn02, Zr02, and K20. A fixed bed reactor system was usedto evaluate the catalysts with n-hexane as feedstock. The best results were obtained with calciumaluminate based catalysts. The higher olefins yields obtained with the new catalysts compared withcr-Al2O3 are explained by the higher conversion obtained at the same process conditions. The kineticsof the pyrolysis of rz-hexane was also studied. The overall decomposition reaction is first order bothfor thermal cracking and for cracking in the presence of catalyst 12Ca0.7A1203. The kinetic parametersestimated in the presence of the catalyst are how = 4.57 X lo6 cm3/(g.s) and E, = 26 130cal /mol.Literature ReviewThe most commonly used method for manufacturinglight olefins is introducing streams rich in paraffinic hydrocarbons,such as ethane, propane, and naphtha, intoexternally heated tubes. In this thermal decompositionprocess, the cracking reactions are controlled by varyingthe composition of the feed materials, the reaction temperature,and the residence time in the reaction tube.This method has some drawbacks, such as the hightemperature required for cracking reactions, the depositionof coke in the tubes, and especially low selectivity inethylene and other desired products (Nowak and Gunschell,1983).To overcome these problems, a serious effort started afew years ago to introduce catalysts in the process. Variouscatalytic processes for the preparation of olefins have beenreported, but these processes have not yet been appliedcommercially because of various difficulties.Wrisberg et al. (1973) and Andersen et al. (1975) proposedas a catalyst a mixture of zirconium and hafniumoxides associated with active alumina to increase themechanical strength. The main advantage of this type ofcatalyst is the minimization of the coke production, whichwas achieved by adding to the catalyst a small amount ofalkali metals or alkaline-earth oxides, especially KzO.The oxides of zirconium and hafnium were used ascarriers in the catalyst proposed by Colombos et al.(1978a,b). They used as an active compound MnOz oncarrier ZrOz or HfOp. The feedstock was atmosphericresidue, and the ethylene yield was relatively good.According to the studies of Pop et al. (19791, a bifunctionalsynthetic mordenite having the formula (yH-zM.uNa)0.A1203.Si02 (M = Cu, Ag, Co) was used, with verygood yields of ethylene at relatively low temperatures. Inthis patent there is no mention of coke yields, perhaps highbecause of the type of catalyst used.A different type of catalyst was proposed by Yegiazarovet al. (1978), consisting of In203 on pumice. The yield ofethylene was high and the coke deposition negligible. Thehigh price of Inz03 is a discouraging factor for this approach.Adelson et al. (1979) suggested KVO, on pumiceas a catalyst for the production of ethylene, with promisingresults.Senes et al. (1972) proposed a catalyst with low porosityand containing at least one oxide of the metals of thegroup, formed by the rare earths and antimony, in smallamounts in association with a mixture of refractory oxideswith magnesium oxide and zirconium or aluminum oxide.The results were rather discouraging.The most important research has been done at ToyoEngineering Corp. where Tomita et al. (1973, 1976) suggesteda catalyst characterized by high ethylene yield andlow coke deposition. The catalyst consists of refractorycalcium aluminate with one oxide of the group of calcium,beryllium, magnesium, or strontium oxide. All the catalyststhey proposed were of low porosity and surface areaand of high crystallinity. The best results were obtainedby a simple catalyst proposed by Kikuchi et al. (1985) with51.46% CaO and 47.73% A1203 The crystal phase consistsof a mixture of the aluminates Ca12A114033 and Ca&,06.The yield in ethylene varies between 34 and 39 wt YOdepending on the temperature, pressure, steam-to-oil ratio,and contact time. The disadvantage of this process is thehigh yield of carbon dioxide.In trying to assess the work of previous investigators,one quickly concludes that there is no limitation to thenumber or type of oxides proposed. However, the use ofalumina oxide as a carrier or as an active component iscommon. There is also general agreement that the catalystshould be a mixture of at least two oxides, usually refractory.Most of the catalysts disclosed have low porosity, highcrushing strength, and very low surface area, except forthe mordenite-type catalysts.There is generally limited literature information regardingthe kinetic mechanism with these catalysts. Tomitaet al. (1973) suggest that alkaline-earth metal oxidescontrol the dehydrogenation reaction of the feed hydrocarbonsto prevent carbon deposition. They also suggestthat aluminum oxide promotes the reaction between hydrocarbonsand steam.Based in the literature review, it is therefore concludedthat the process of cracking hydrocarbons to produce lowerolefins in the presence of catalysts is still under developmentand much work remains to be done.Experimental SectionCatalyst Preparation and Characterization. On thebasis of the literature search, several catalyst samples wereprepared using a mixture of at least two metal oxides. Theaim was to produce an identifiable crystal structure. Itwas realized that mechanical properties like crushing0888-5~85/S9/2628-0524$01.5Q/Q C 1989 American Chemical Society

Ind. Eng. Chem. Res., Vol. 28, No. 5,1989 525calciningcatalyst CaO Alsos SrO MgO In,03 Ti02 MnO, ZrO, K,O pumice temp, "C mz/g crystal structure101 1 11300 0.1 CaAl*O.102 3 11300 0.31n Ca3AI2O8103 12 71300 0.24" Ca12A1,,03sCa,AI~O~104 1 21300 0.3 CaA1,07105 1 61300 2.1 CaAl,,OU201 3 2 11300 0.1 CarAlLh202203204301401402501502503a Multipoint BET surface area.grams/mole of oxide.1 11 1Table 11. Correlation between Calcination Temperature,Surface Area. and Coke Productionsurfacecrystal calcination area, coke yield,catalyst structure temp, "C m2/g wt % on feed1300 CB~&,O~. 1300 3.05 0.11Ca,A1,0,1100 Ca12A111033. 1100 5.20 0.60Ca3Al,06850 Ca0.A1,03 850 20.00 4.50600 CaCO, 600 100.00 4.85area,1300 0.7 CaiAIiOB1300 0.1 Ca,lAl,10,rCa3AlzO~1300 0.4 SrAI20,.Sr3AI2O61300 1.9 MgAlO,13' 800 7.2 In&800 4.3 CaIn04~In,031 950 3.5 KIMgTiOle9506500.64.3TiO,zrozstrength should be considered at a later stage.The procedure for preparing the catalyst samples wasas follows: A fixed quantity of each oxide (in nitrate orcarbonate form) was placed in a ball mill for 1 h. Distilledwater with a small amount of an organic binding agent wasthen added to the mixture. The amount of water is animportant factor in determining the plasticity of the sample.After the sample is extruded to the desired size, thecatalyst is first dried and then heated to 600 "C to convertnitrates to oxides. The crystal structure is then formedby calcination for 24 h at a high temperature. The calcinationtemperature is selected from the phase diagram,while the calcination time is chosen high enough to ensurethe formation of the crystal structure.Standard analytical techniques were used for the characterizationof all catalyst samples prepared. To identifythe crystal structure formed after the calcination procedure,the samples were tested in a Miniflex-Rigaku diffractometerwith monochromatized Cu Kal radiation anda Ni filter. Crystal phases identified are shown in TableI.Measurements of the surface area of the samples usingthe BET single-point technique are also shown in TableI.A second series of catalysts containing the same metaloxides was prepared by calcination at different temperatures.The properties of these catalysts are listed in Table11.All catalysts were compared with a-A1208 prepared fromr-A120, by calcination at 1350 'C for 17 h.Experimental Apparatus and Testing Procedure.A schematic diagram of the experimental unit is shown inFigure 1. The unit is conveniently divided into threesections: feed preparation, reactor section and liquidcollection, gas analysis.n-Hexane was used as the standard feed for all tests.The feed is picked up with nitrogen, and it is then introducedinto the upper reactor section where it is mixed withFigure 1. Schematic diagram of the experimental unit for theevaluation of catalysts.steam. The water, after measuring its flow, is vaporizedin a furnace, and it is then heated to the desired temperaturewith a wall heater.A stainless steel reactor (1.25-cm i.d.) was used. Astainless steel screen located in the center of the reactorholds the catalyst. A three-zone furnace was used tocontrol the reactor temperature, which was monitored atthree locations. The temperature within the reactor wasmeasured by three Chromel-Alumel thermocouples locatedin the feed inlet. in the catalvst bed, and in the lower partof the reactor.The oroducts at the reactor exit are first cooled in anice-cooied liquid receiver where the steam and heavy hydrocarbonsare condensed. The gases then flow to a gascollectingsystem where the volume is measured by waterdisplacement. The gases are analyzed in two gas chromatographs.A Varian 3700 with a thermal conductivitydetector and a Carbosphere column is used to analyze forH2, NP, CHI, CO, and COP.Hydrocarbons (paraffins and olefins with up to sixcarbon atoms) are analyzed in a HP 5710A gas chromatographequipped with a FID detector and a Porapak Qcolumn. n-Hexane unreacted is partly collected as liquidin the receiver with water and partly collected as gas inthe gas product collecting system.Experimental Conditions. All catalysts were evaluatedat the following conditions: n-hexane flow rate,0.465-0.498 g/min; steam flow rate, 1.7 g/min; nitrogenflow rate, 3&32 mL/min; reaction temperature, 782-786"C; space velocity, It10 h"; run time, 75 s; catalyst weight,3 g; and catalyst diameter, 0.159 cm.

526 Ind. Eng. Chem. Res., Vol. 28, No. 5, 1989An effort was made to maintain the temperature constantduring the runs. This was attained with the threezonefurnace equipped with temperature-indicating controllers,PID. Plug flow conditions were tested with asuitable tracer response analysis.A series of runs were carried out to establish the kineticparameters for the decomposition of n-hexane to lighterproducts. Feed flow rate, steam flow rate, and reactiontemperature were varied within the limits of the experimentalunit. The range of the experimental conditions wasas follows: n-hexane flow rate, 0.23-0.86 g/min; steam flowrate, 0.92-3.3 g/min; nitrogen flow rate, 30 mL/min; reactiontemperature, 730-780 "C; space velocity, 4.6-17.2h-l; run time, 75 s; catalyst weight, 3 g; and catalyst diameter,0.159 cm.The reaction temperature mentioned above refers to thetemperature measured during the reaction in the catalystzone.Composition EffectsCatalyst samples with different chemical compositionswere tested under the same experimental conditions. Theresults of this testing are reported in Table 111. Evaluationof the samples was based not only on the yield of thedesired products (CzH4, C3H6) but also on the undesiredproducts (CO, C02, CHI). Two selectivity ratios wereproposed, one expressing the ratio of CzH4 and C3H6 yieldsto COz and CO yields and the other expressing the ratioof CzH4 and C3H6 yields to C02, CO, and CHI yields. Thehigher the two selectivity ratios, the more suitable thecatalyst for the process.The main conclusions from these results are as follows:1. Contrary to what has been reported in the literatureby Colombos et al. (1978a,b), the selectivity ratios of theoxides of Mn, Ti, and Zr are low. It should be mentionedthat the Mn02.TiOz catalyst shows a moderate activity,whereas the MnOz.ZrOz sample shows a low activity,mainly due to the high carbon oxides yield.2. Inz03 on pumice shows moderate selectivity in promotingthe desired products. The lower selectivity of thiscatalyst compared with the one reported by Yegiazarov etal. (1978) is probably due to the higher space velocity usedin this study, 9 h-l.3. The results of combinations of metal oxides of CaOand/or Al,03 with SrO are not successful because of thelow selectivity ratios. The cause of this low selectivity isthe relatively high carbon oxides yield especially withsamples 3Ca0-Sr0.2Al2O3 and Sr0.Al2O3.4. The best results were obtained with catalysts containingCaO and Alz03, depending on the molar proportionsof the oxides. X-ray diffraction analysis of thesesamples indicated that the formation of a certain crystalphase depends mainly on the molar proportions of thecomponent oxides. Selectivity ratios vary with the molarproportions of CaO and A1203. The mixture of CaO andAlz03 at a molar ratio of 12:7 shows the highest selectivityratios of all catalyst samples prepared. Kikuchi et al.(1985) proposed this type of catalyst, 12Ca0.7A1203, forthe production of ethylene, with very promising results.The X-ray diffraction pattern of this sample shows theformation of two crystal phases, one of Ca12A1,,033 as amajor component and the other of Ca3Al2O6 as a minorcomponent. It was difficult to reproduce samples withpure Ca12A114033 crystal phase. The presence of the twophases has also been reported by Kikuchi et al. (1985).a-Alumina samples tested under the same experimentalconditions showed relatively low selectivity ratios mainlydue to the low ethylene and propylene yields. Crackingof n-hexane in an empty reactor shows low selectivity dueI40 ,P Alumina- 600-Catalyst0 EX-Catalyst I 1!00-Catalystl 0 !300-Cata:yst." ,770 780 190 K OTemperature IC1Figure 2. Ethylene yield in the presence of catalyst samples, calcinedat different temperatures and in the presence of a-alumina.Iw3 1z: 61 ..10 ,A AluminaA 600-CatalystI 850-Catalyst I 1100-Catalyst0 1300-CatalystA1=..A ~2 1> I A-zII1 1" 1770 i00 :90 EO3Temperature IC)Figure 3. Carbon oxides yield in the presence of catalyst samplescalcined at different temperatures and in the presence of a-alumina.T310E 42P Aluminao 0 850-Catalyst1300-CatalystA 600-CatalystI 1100-Catalyst0 .L0Aa A n00770 780 790 EO0Temperature (ClFigure 4. Coke yield in the presence of catalyst samples, calcinedat different temperatures and in the presence of a-alumina.to the lower olefins yields. It is worth noting that somecatalyst samples have lower selectivity than the relatively"inert" a-Alz03 and even lower than that of thermalcracking (empty reactor) because of the high carbon oxideyields.Calcium Aluminate Based CatalystsRealizing that a new catalytic process for producingethylene should minimize the formation of coke on catalyst,a catalyst composed of 44 wt % CaO and 56 wt %AlzO, was calcined at the following temperatures: 600,850,1100, and 1300 "C. Testing was carried out at the conditionsreported in the previous section.A comparison of the ethylene, COz + CO, and cokeyields for the four different calcination temperatures is

Table 111. Selectivity Ratios of Various OxidesInd. Eng. Chem. Res., Vol. 28, No. 5, 1989 527n-CsC2H4, C3Hs9 CHI, CO + COP, C2H4 C3Hs/ CZH4 + C3H6/ masscatalyst composition wt % wt 70 wt % wt% co + co:, CO + COP + CH, balanceCaO.Al,OI 34.7 18.6 6.412Ca0.?Ai203Ca0.2A1203Ca0.6A12033Ca0.Sr0.2Al2O33Ca0-3Sr0.2Al2O312Ca0.Sr0.8A1203Sr0.A1203MgO.Al203InPO3.pumiceCa0.1n203KZO.MgO.7TiOzMnO2.TiO2ZrOz.MnOza-A1203empty reactor34.8 19.033.3 18.932.7 15.630.3 17.831.4 13.633.6 16.829.8 14.332.7 17.331.5 15.832.0 16.728.7 15.830.0 15.930.3 17.222.2 13.516.9 10.86.06.24.55.25.35.15.66.15.95.35.35.15.55.94.42.71.71.94.22.811.43.46.42.31.810.71.91.94.42.12.019.931.626.311.216.73.914.86.921.726.54.523.323.610.717.313.75.86.96.35.55.92.65.93.65.96.03.06.36.44.74.54.394.394.295.792.197.394.895.792.397.296.797.795.296.292.396.997.5100A Aluminam 12Ca0 7A1203 Catalyst0 CaO 261203 CatalystTable IV. Properties of Catalystssample 9034-168A 9034-170Achem composition 12Ca0.7A1,03 Ca0.2AlZO3calcination temp, "C 1200 1200calcination time, h 24 24particle diameter, cm 0.159 0.159surface area, m2/g 2.08 4.45XRD pattern Ca12A114033.Ca3A1z06 CaAl,06A Alumina o Ca0.2Al203 c a t v'" 1 12Ca0.7A1203 Catalystj.", 1775 185 795 80sTemperature IC)Figure 5. Effect of temperature on the n-hexane conversion in thepresence of a-alumina and calcium aluminate catalysts.shown in Figures 2-4. For comparison purposes, theresults are also shown for a-A1203. In the preparation ofthe sample of a-A1203, an effort was made to reproducea solid sample with the same pellet size and surface areaas the sample of CaO/A1203 calcined at 1300 "C. It is notclaimed, with this work, that a-A1203 is completely inactive.Adelson et al. (1979) proved that a-alumina and evenquartz show some catalytic action compared with an emptyreactor.From a careful study of Figures 2-4, the following conclusionscan be drawn:1. The ethylene yield (Figure 2) with CaO/A1203 ishigher than with a-A1203. However, there is no discernibledifference between samples calcined at different temperatures.2. Yields of C02 and CO are considerably lower forsamples calcined at 1300 "C (Figure 3).3. High calcination temperature leads to a substantialreduction in coke yield. As shown in Figure 4, catalystscalcined at 600 and 850 OC produce 4-6 wt % coke on freshfeed, while samples calcined at 1100 and 1300 "C produce0.5-0.9 and 0.1-0.2 wt % coke, respectively.X-ray diffraction clearly showed the absence of crystalstructures at low calcining temperatures. It is believed thatthe low surface area of samples calcined at high temperaturesis the primary reason for the low coke yield (Table11). Therefore, it is essential to calcine the catalyst samplesat high temperatures so as to minimize the coke production.The need to continue the evaluation of calcium aluminatebased catalysts with samples of specific geometrynecessitated the preparation of larger batches of catalyst." j75 185 795 E05Temperature (ClFigure 6. Effect of temperature on ethylene yield in the presenceof a-alumina and calcium aluminate catalysts.11I IA Alumina12CaO.7A1203 CatalystD CaO 211203 Catalyst1 P 132 om*4P 0-I1 ,715 785Temperature (CI795 805Figure 7. Effect of temperature on carbon oxides yield in thepresence of a-alumina and calcium aluminate catalysts.at Amoco Research Center. The properties of catalystsprepared in this manner are reported in Table IV.The conversion of n-hexane and the yields of ethyleneand carbon oxides are shown in Figures 5, 6, and 7, correspondingly.X-ray diffraction confirmed the existenceof Ca3A1206 and Ca12A117033 in sample 9034-168A. The

528 Ind. Eng. Chem. Res., Vol. 28, No. 5, 1989Table V. Effect of Hydrogen Treatment on 12Ca0 7AI2O3Catalyst Performancetype of catalyst untreated treated with H,test conditionstemp, "C 780 780feed flow rate, mL/min 0.68 0.68H20 flow rate, mL/min 1.7 1.7catalyst wt, g 3 3feed conversn, wt % 50.76 40.24product yields, wt % (on feed)HZ 0.58 0.55CH4 5.10 4.28C2H6 1.16 0.89C2H4 21.51 17.08COHB 0.38 0.25C3Hs 13.06 10.24C4H8 7.21 5.56C4H6 1.41 1.11COZ 1.14 0.83co 0.18 0.27feed recovery, wt % 96.70 97.80results shown in Figures 5-7 confirmed again the higheryield of ethylene with these catalysts compared with a-Al,O,.To explain the increase in the feedstock conversionoccurring in the presence of calcium aluminate, a catalystsample was treated with hydrogen in a thermal gravimetricbalance (Du Pont Series 99 thermal analyzer). The catalystweight was reduced by 1.5% between 780 and 800 "C.The reduced catalyst sample was subsequently testedat 780 "C. The results of testing catalyst samples treatedand untreated with hydrogen are reported in Table V. Adecrease in the feedstock conversion was observed whenthe catalyst was treated with hydrogen.It is therefore believed that the catalytic action is dueto the presence of active oxygen of the calcium aluminatecatalyst. This oxygen is leaving the system in the formof water after the catalyst was treated with hydrogen.The presence of active oxygen in calcium aluminates wasidentified and studied in detail by Imlach et al. (1971) andby Brisi and Lucco-Borlera (1983). This active oxygenpromotes nonselectively all the cracking reactions, as isconcluded from the same selectivity ratios obtained inthermal and catalytic cracking. This action of oxygen wasconfirmed by many research groups involved in methaneoxidative coupling (It0 et al., 1985; Keller and Bhasin, 1982;Hinsen and Baerns, 1983). They claim that oxygen in thecatalysts (alkaline, or alkaline-earth oxides with very lowsurface area) promotes the formation of methyl radicalsat the catalyst surface and hence the overall cracking reactionsof hydrocarbons in the gas phase.Pyrolysis ModelA simplified first-order reaction pyrolysis model wasused to fit the kinetic data. It was found that the reactorbehaved as an integral nonisothermal reactor, and forcalculation purposes it was divided into three zones asindicated in Figure 8. The following assumptions weremade:1. Each reactor zone behaves in a plug flow mode.2. For estimation purposes, each zone can be treatedat constant temperature.3. Because of the high dilution of steam, there is noappreciable volume expansions.Based on the above assumptions, the following equationsdescribe the conversion of n-hexane for each reactor zone:(-r~)~ dV, = FAO,~ dXA, i = 1-3 (1)(-rAIr = K(TJCAL (2)1 - XA = (1 - xAl)(1 - x A2)(1 - X.43)(3)'Ieam Feed 7 n I TCTC 'lTcFigure 8. Reactor configuration.where (-rA)i is the rate of disappearance of n-hexane, K(TJis the rate constant at temperature Ti, Ck is the n-hexaneconcentration, FAO,I is the molar flow rate of the reactantin the inlet of each reactor zone, VI is the volume of eachreactor zone, Xk is the n-hexane conversion at each reactorzone, and XA is the total conversion at the exit of thereactor.In reactor sections where no solids are present, the rateconstant is equal to the thermal cracking rate constant K,.For the middle reaction section where solids are present,we assume that two parallel paths of cracking reactionsoccur: in the bulk between the particles and on the particlesurface.The design equations for this case are the same as (1)and (21, but the rate constant is equal toK = K,t + K,(1 - €)pp(4)where K, and K, are the rate constants for thermal andcatalytic reactions, respectively, t is the void fraction ofthe catalytic bed, and pp is the particle density.The assumption of the two parallel paths in describingthe decomposition reactions in the presence of particlesis valid because there are no selectivity differences betweencracking in the presence of particles (a-alumina,12Ca0-7Al2O3 catalyst) and thermal cracking (empty reactor).Product selectivity is defined as grams of eachproduct per 100 g of n-hexane converted (Figures 9-11).By use of (1)-(3) and the experimental conversion forkinetic runs in an empty reactor, the Arrhenius parametersof the overall thermal decomposition rate of n-hexane wereobtained by using the algorithm of Levenberg-Marquadt.The estimated values of the two parameters KO, and E,for thermal cracking runs were used also for the estimationof the values of the kinetic parameters describing thephenomenon in the presence of solids (alumina, catalyst).The following numerical values were obtained:

Ind. Eng. Chem. Res., Vol. 28, No. 5, 1989 5290 Thermal cracking=” I I I I I I0 20 40 60 eo 100Conversion (wt4)Figure 9. Ethylene selectivity as a function of feed conversion.zY ““1Alumina0 Thermal crackingI0 12Ca0 7A1203 Catalystfrequency factor, activation energy,cracking cm3/ (es) cal/molthermal 0.206 X lo9 (s-l) 35680alumina 2.16 X lo8 36 500catalytic 4.6 X lo6 26 100The results of the application of the Levenberg-Marquadtalgorithm in all the kinetic runs are reported in thesupplementary material. The adequacy of the first-orderreaction model proposed is shown in Figure 12. It isevident that the calculated n-hexane conversion by thesimplified kinetic model proposed for thermal, alumina,and catalytic cracking is in good agreement with the experimentalvalues.To our knowledge, there are no literature data concerningthe values of the n-hexane kinetic parameters inthe presence of a-A1203 and 12Ca0-7Al2O3 catalyst. Theestimated values of the thermal cracking kinetic parametersare somewhat lower than those reported in the literaturefor the decomposition of n-hexane (Zdonik et al.,1970; Illes et al., 1973; Ebert et al., 1983). The experimentaldata used for the estimation of the kinetic parameterscover a wide range of conversion. As it is known, thereaction rate constant decreases with increasing conversion.This is due to the self-inhibition effect as described byLeathard and Purnell (1970). For the data collected athigh conversion, the self-inhibition effect has to be takeninto account. The dependence of the rate constant onconversion is expressed byO / I I I I0 20 40 60 BO 100Conversion IwtX)Figure 10. Propylene selectivity as a function of feed conversion.40P Alumina0 Thermal cracking12Ca0 7A1203 Catalyst“ I0I20 40 6Coniersion IwtkPIBOI100Figure 11. Butylene selectivity as a function of feed conversion.100“ I0P Alumina (3 12Ca0 761203 Catalyst0 Thermal cracking20 40 60 EO 100Observed ConversionIwt%lFigure 12. Relationship between the predicted feed conversion bythe first-order kinetic model and the experimental conversion.@where K, is the rate constant at very low conversions(

~ ..~ -530 Ind. Eng. Chem. Res., Vol. 28, No. 5, 1989,OOT . ~1 Alumina 12Ca0.7:1203 Catalyst 4I 0 Thermal cracking IKay = frequency factor for thermal decomposition,KO, = frequency factor for catalytic decomposition, cm3/ (gs)KO, = frequency factor for zero conversion, sw1E, = activation energy for thermal decomposition, cal/molE, = activation energy for catalytic decomposition, cal/molE, = activation energy for zero conversion, cal/molGreek Symbolst = void fraction of the catalytic bedpp = particle density, g/cm3N = inhibition coefficient1- - -- ..7--- -1~2040 LO LO :ODObserved Conve^sionIwt%lFigure 13. Relationship between the predicted feed conversion andthe experimental for the model taking into account the inhibitioneffect.calcination temperature of 1300 "C.Samples calcined at low temperature favor the formationof coke. Therefore, high calcination temperature is necessaryto minimize coke production.Comparison of product yields in the presence of a-A1203with those of calcium aluminate catalysts leads to higherolefins yield in the presence of 12Ca0*7Al2O3, mainly dueto the higher degree of hexane conversion.No selectivity differences exist between calcium aluminatecatalyst and a-A1203 at the same conversion level.The overall thermal decomposition of rz-C6H,, is well describedby the first-order reaction model. In the presenceof a-alumina and of catalyst 12Ca0.7A1203, the overallcracking reactions are also described by first-order reactionmodels.The inhibition effect of the products on the overall kineticrate constant was also studied, and it was found tobe significant for the thermal cracking and for the crackingin the presence of a-alumina and 12Ca0.7A1203 catalyst.AcknowledgmentA. A. Lemonidou and I. A. Vasalos thank the Secretariatof Research and Technology for the financial support ofthis research.Nomenclature(-rA)i = rate of disappearance, mol/(s-cm3)CAi = reactant concentration, mol/cm3FAo,i = initial molar flow of reactant in the inlet of each reactorzone, mol/sVi = volume of each reactor zone, cm3XAi = conversion at the end of each zone, wt %XA = conversion at the exit of the reactor, wt %K = rate constant, s-'K, = thermal cracking reaction rate constant, s-'K, = catalytic cracking reaction rate constant, cm3/(g.s)ISupplementary Material Available: Tables 1-111 reportingthe estimated values of conversions for each reactorzone, the residuals of each run, and the residual sum of squaresfor the thermal cracking kinetic runs and for kinetic runs inthe presence of a-A1203 and 12Ca0.7&O3 catalysts (3 pages).Ordering information is given on any current masthead page.Literature CitedAdelson, S. V.; Vorontsova, T. A.; Melnikova, S. A. Neftekhimiya1979, 19, 577.Andersen, K. J.; Fischer, F.; Rostrup-Nielsen, J.; Wrisberg, J. USPatent 3,872,179, 1975.Brisi, C.; Lucco-Borlera, M. Cement0 1983, 3, 155.Colombos, A. J.; McNeice, D.; Wood, D. C. US Patent 4,087,350,1978a.Colombos, A. J.; McNeice, D.; Wood, D. C. US Patent 4,111,793,1978h.Ebert, K. H.; Ederer, H. J.; Isharn, G. Znt. J. Chem. Kinet. 1983, 15,475.Hinsen, W.; Baerns, M. Chem.-Ztg. 1983, 107, 223.Illes, V.; Welther, K.; Pleszkats, I. Acta Chim. Hung. 1973, 78, 357.Imlach, J. A.; Dent Glasser, L. S.; Glasser, F. P. Cem. Concr. Res.1971, I, 57.Ito, T.; Wang, J.-X.; Lin, C.-H.; Lunsford, J. H. J. Am. Chem. SOC.1985, 107, 5062.Keller, G. E.; Bhasin, M. M. J. Catal. 1982, 73, 9.Kikuchi, K.; Tomita, T.; Sakamoto, T.; Ishida, K. Chem. Eng. Prog.1985, 81 (No. 6), 54.Leathard, D. A,; Purnell, J. H. Ann. Rev. Phys. Chem. 1970,21, 197.Murata, M.; Saito, S. J. Chem. Eng. Jpn. 1975, 8, 1, 39.Nowak, S.; Gunschell, H. Pyrolysis of Petroleum Liquids: Naphthasto Crude. In Pyrolysis: Theory and Industrial Practice; Albright,L. F., Crynes, B. L., Concoran, W. H., Eds.; Academic: New York,1983; p 319.Pop, G.; Ivanus, G.; Boteanu, S.; Tomi, P.; Pop, E. US Patent 4,172,816, 1979.Senes, M.; Lhonore, P.; Quihel, J. US Patent 3,644,557, 1972.Tomita, T.; Kikuchi, K.; Sakamoto, T. US Patent 3,767,567, 1973.Tomita, T.; Kikuchi, K.; Sakamoto, T. US Patent 3,969,542, 1976.Wrisberg, J.; Andersen, K. J.; Mogensen, E. US Patent 3,725,495,1973.Yegiazarov, G. Iu.; Cherches, B. Kh.; Krutko, N. P.; Pauskin, I. M.Neftekhimiya 1978, 18, 237.Zdonik, S. B.; Green, E. J.; Hallee, L. P. Manufacturing Ethylene;The Petroleum Publishing Co.: Tulsa, OK, 1970; p 27.Received for review July 27, 1987Revised manuscript received March 4, 1988Accepted January 13, 1989