STUDIES OF REFORMING REACTIONS WITH CATALYST DEACTIVATION IN TUBULAR REACTORS

Abstract

Dehydrogenation of cyclohexane to benzene under conditions of catalyst deactivation was investigated in an integral reactor under quasiisothermal conditions and at atmospheric pressure. Commercial platinum-onalumina reforming catalyst (Sinclair-Baker RD-150) was used for investiga tion. The operating variables and their range studied are : mole-ratio of hydrogen to cyclohexane (0.91* to k.S), reaction temperature (310-435 0 and liquid-hourly-space-velocity (1.28 to h.77). Reaction product consisted of mainly benzene with some hydrocracked gases. Gas chromatographic analysis of liquid product samples using Carbonax-4000 on Celite column indicated benzene as the only reaction product. Experimental set-up consisted of a feeding section for cyclohexane and hydrogen, vaporizer and preheater section, a reactor section and a liquid product collection section. The flow rate of exit gases was recorded with time. The pressure drop in the reactor increased with time due to coke formation and care was taken to keep the flow rates of the reactants constant by compensating for the increase in pressure drop. The catalyst bed was diluted with glass beads, to maintain it near isothermal using suitable dilution criterion. The kinetic model used in the analysis consists of the following reactions: k1 Dehydrogenation : C£,H12 *" C6H6 + ^2 Hydrocracking: C&H12 k? -> hydrocracked products Deactivation; C&H12 - > catalyst poisons (coke) I I The material balance calculations were carried out considering equilibrium at the exit of condenser and material balance checked well for most of the samples in all the sets for both the cyclohexane and hydrogen feeds. The conversion of cyclohexane into hydrocracked products was calcu lated from the material balance of cyclohexane using time average feed rate. It is assumed that the loss of cyclohexane due to the deactivation reaction is negligible. The kinetic data for undeactivated catalyst was obtained by extra polation of experimental conversion data under deactiving catalyst condition to initial conditions and the same was analysed by assuming first order irreversible kinetics and plug flow behaviour. These assumptions result in the following conversion - space velocity relationship, for initial rate data : XA + io. iitlMj+iO ! k{ Co (i+k) ., T3lK__| ln 0 -(1 +**>•- 3-K •7 CO Under experimental conditions the external heat- and mass-transfer and internal heat-transfer resistances were found to be negligible but the intrapellet mass-transfer resistance was found to be significant. The values of effectiveness factor for the desired dehydrogenation reaction varied between 0.12 and 0.77, and for the duhydroqsracfeitjg reaction it varied between 0.60 to 0.93. With the help of equation 1 and experimental conversion data the apparent rate constants k' and k' were determined. The 2 values of effectiveness factors were estimated from the plot of "nversus 9r{ , where: 1 0 it v S e _ ) 1 1 tan h 3 9 3 0 = v/ e (2) (3) 111 k' p and 02^ = ( 2L )2 __L£ (i,) Se e 2 It is to be noted that 0 TJ can be calculated from the experimental data directly. Knowing the values of k! , k' , "*\* and "*}- , and the intrinsic rate constants k. and k» were calculated. The intrinsic rate constants k. and k_ were observed to depend only on temperature which confirmed the validity of first order irreversible kinetics for dehydrogena tion and hydrocracking of cyclohexane. The values of activation energy and pre-exponential factor were calculated from the regression analysis of In k versus 1/T values for each rate constant, to give k] . exp(3,.57. 12.300, T_7c£_t) w k2 •eXp (,8.97 -IStfS- ) (hr.)(™Latyst) (5) The correlation coefficients for dehydrogenation and hydrocracking rate constants are 0.992 and 0.902 respectively, and indicate good fit of data on Arrhenius plots. It is safe to assume that coke formation on catalyst results in a loss of catalyst activity for only dehydrogenation reaction involving platinum sites and not for hydrocracking reactions. Thus, the rates of dehydrogenation and hydrocracking reactions with catalyst deactivation, are given by : r- s k, -n , a cr (6) '• I H r2 5 k2 ^2 CA (7) The rate of change of activity with time is considered to depend on m power of activity and n power of concentration ratio of cyclohexane to hydrogen as given below : iv -r^-ft • >d'-W (8» In equation (6) and (7) undeactivated catalyst effectiveness factors are used and any change in dehydrogenation activity due to the coke formation Is accounted by activity factor a. Non-linear equations 6 to 8 were solved simultaneously for n -- 1 and assumed values of k, and m, to calculate cyclohexane and benzene mole-fraction at reactor exit, for different times-on-stream using fourth - order Runge-Kutta method on IBM 370/145, UNIVAC 1100 or DEC 2050 computers. The optimal values of krf and m were found for each set by minimizing the variance between calculated and experimental values of benzene mole fraction at the reactor exit for different times -on-stream. The values of k, and m were optimized to within +5 and + 2 percent respectively. The values of m and k^ are in the range of 1.11 to 3.11, and 0.41 to 4.53 (hr)~ respectively. For the solution of differential equations 40 bed increments and 5 minutes time interval was chosen after careful error analysis. The 40 bed increments (bed increment of the order of pellet diameter) gave a computation accuracy of better than 0.05 percent, and 5 minutes time interval resulted in a computation accuracy of better than 0.12 percent, in exit conversion values at the end of five hours time-on-stream. Activity and conversion profiles were calculated for all the sets along the length of the bed for different times-on-stream using optimal values of m and k,, and the intrinsic reaction rate constants k. and k? . A total of 19 sets were analyzed for deactivation parameters. The deactivation rate constant was found to vary only with temperature. The values of activation energy and pre-exponential factor for catalyst deactivation rate constant were found from the linear regression of In krf versus 1/T values, to give - kd * exp (9.475 -^^~ ) (hrf1 0) The value of correlation coefficient, 0.872, for the above equation indicates a good fit of data. A second order polynomial was used to correlate m with Thiele parameter 0j (based on k^ and the constants obtained by regression analysis are as given below: m r0.9502 +0.2623 9y -0.005806 02 (10) For a given set of operating conditions equations 2,3,4,5,9 and 10 were used to calculate the intrinsic dehydrogenation and hydrocracking rate constants, deactivation rate constant, Thiele parameters, effectiveness factors and m, and these values were then used to calculate the conversion and activity profiles along the catalyst bed for different times-on-stream by simultaneous solution of equations 6 to 8 for all the sets. The proposed model gave a good fit with experimental data as is evident from the absolute percentage error variance between the experimental and predicted values of conversion which is 5-5 for benzene and 4.7 for cyclohexane considering all the experimental points. The values of m in the range of 1 to 3 indicate pore-mouth poisoning and confirms simultaneous deactivation.