Energy & Fuels 1994,8, 131-135
131
Kinetic Studies for Catalytic Cracking of Heavy Oil from Waste Plastics over REY Zeolite Ahmad Rahman Songip,? Takao Masuda, Hiroshi Kuwahara,t and Kenji Hashimoto* Department of Chemical Engineering, Kyoto University, Kyoto 606, Japan Received July 21, 1993. Revised Manuscript Received September 13, 1 9 9 P
A kinetic model was developed to represent the catalytic cracking of heavy oil from waste plastics by rare-earth metal exchanged Y-type (REY) zeolite to produce gasoline. The influences of reaction conditions on the product distributions were previously reported by us. On the basis of these results, a reaction pathway was proposed and a set of differential equations was developed. The kinetic parameters were determined by nonlinear least-squares regression of the experimental data. These parameters were found to be proportional to the amount of strong acid sites of the used catalysts. The calculated values of the product distribution were found to be in good agreement with the experimental data. 1. Introduction
The increasing amount of waste plastics which causes serious pollution problems is a cheap and abundant source of chemicals and energy. The chemical recycling method, which converts waste plastics to useful hydrocarbons, was recognized as an ideal approach.' Waste plastic is mostly generated from households and small-to-medium size enterprises. Hence, the following chemical recycling process is considered to be the most promising approach in view of lower capital and transportation costs and fewer engineering constrains:2 first waste plastics are thermally cracked in a pyrolysis plant built in each collection station. Then, the produced heavy oil is transported to a centrally located cracking plant and is converted to gasoline. We have reported that a rare-earth metal exchanged Y-type (REY) zeolite catalyst was an effective catalyst for the catalytic cracking of the heavy The influences of reaction conditions and catalytic properties of REY zeolite on the yields of products and on the gasoline quality have also been p r e ~ e n t e d . ~ The main objective of this work is to propose a reaction pathway for the catalytic cracking reaction of heavy oil and to develop a kinetic model for the cracking reaction, based on our previously reported resultsa3 A set of differential equations has been developed for representing mass balances of reaction products. Kinetic parameters in these equations were evaluated by nonlinear least-squares regression of the experimental data to the numerical solutions of the differential equations. The obtained activation energies have been compared to other reported values. Also, the effects of reaction temperature, On leave from the Department of Chemical Engineering, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia. t Permanent address: Sanwa Kako Co. Ltd., Kyoto 613,Japan. e Abstract published in Advance ACS Abstracts, October 15,1993. (1)Chem. Eng. 1992,July, 30. (2)Songip, A. R.; Masuda, T.; Kuwahara, H.; Hashimoto, K. Appl. Catal. B: Enuiron. 1993,2, 153-164. (3)Songip,A.R.; Masuda, T.;Kuwahara, H.; Hashimoto, K. Production of High-Quality Gasoline by Catalytic Cracking over Rare-Earth Metal Exchanged Y-Type Zeolites of Heavy Oil from Waste Plastics, following paper in this issue.
Table I. Physical and Chemical Properties of Fresh REY Zeolites' catalyst Si/ Al crystal size (pm) amount of total acid sites4 (mol.(kg cat.)-1) amount of strong acid sitesb (mol.(kg cat.)-')
REY-1 REY-2 REY-3 REY-4 4.8 4.8 4.8 4.8 0.1 1.0 0.1 0.1 2.99 2.78 2.91 2.44 0.79
0.57
0.66
0.70
Based on the total amount of ammonia desorbed in the TPD experiment. b Based on the amount of ammonia desorbed above 573 K in the TPD experiment. (I
catalyst crystal size, and amount of strong acid sites of the used zeolite catalysts on the activity of the reaction were analyzed. 2. Experimental Section 2.1 Catalyst. Four types of REY zeolites (Si/Al= 4.8) having different crystal sizes and acidic properties were employed. The physical and chemical properties of these zeolites are summarized in Table I. REY-1 was supplied by Tosoh Co. Ltd., Japan. REY2, REY-3, and REY-4 were prepared from Na-Y-type zeolite by ion-exchange method using asolution of rare-earth metal chloride. The crystal sizes of the catalyst were measured by scanning electron microscopy (S-510, Hitachi). 2.2 Feed Oil. The feed oil was prepared by pyrolyzing polyethylene plastics at 723 K. The oil was then distilled at 473 K to remove lighter hydrocarbons and to obtain a residual oil having a heavy oil content of 95 % The yield of this residual oil was about 70-80 % This oil was used as the feed oil. 2.3 Catalytic Cracking Reaction. Details of the experimental apparatus and procedures, analytical methods, and lumping of reaction products were similar to those in the previous study.2sS The catalytic cracking reaction of heavy oil obtained by pyrolysis of waste polyethylene was conducted under the followingconditions: time factor (=WIF: ratio of mass of catalyst, W, to mass flow rate of feed oil, F)of 0.2-3 (kg cat.).(kg oil)-l.h and a reaction temperature of 573-723 K. Since all the experiments were terminated after 3 h on stream, the effect of aging on the catalyst was not examined. A catalyst with a crystal size of 0.1 pm was mainly used, in order tominimize the resistance to mass transfer within the zeolite crystallite. The lumping of reaction products are as follows: gaseous compounds (carbon number of 1-4: Cl-C4), gasoline fraction (C5-Cll), heavy oil (above C12), and c ~ k e . ~The J cokedcatalyst was burned off using a thermal gravimetric microbalance (GT-
.
.
0887-062419412508-0131$04.50/0 0 1994 American Chemical Society
Songip et al.
132 Energy & Fuels, Vol. 8, No. 1, 1994
a
100
g
80
B
I
-; ; a 60
"
Gasoline
.-.P
83
1
20 0 0
1
2
3
4
Time factor, W/F [kg-catkg-oil.'?h]
c
100
g
80
s
Y
.8
60
! 8
40
0
1
2
3
4
Time factor, W/F [kg-cat.kg-oil".h]
Y
'-
20 0 0
1
2
3
4
Time factor, W/F [kg-cat.kg-oil-'.h] Time factor, W/F [kg-cat.kg-oiI-'.h] Figure 1. Kinetic runs performed using a catalyst with a crystal size of 0.1 pm (REY-1): (a) 573 K, (b) 623 K, (c) 673 K, and (d) 723 K.
Shimadzu) at 773 K in an air stream. The amount of coke loading was obained by the weight loss of the catalyst. 2.4 Measurement of Acidic Properties. The catalytic activity reduced rapidly at the beginning of the reaction and finally approached a constant value after about 3 h on stream.2 Since the steady-state value of activity was used, the acidic properties of the catalysts used for the reaction during the 3 h on stream were employed. The acidic properties of catalysts (fresh and used) were measured from the temperature-programmeddesorption(TPD) spectraof ammonia. However,the conventional TPD experiment employing a thermal conductivity detector4 cannot be used to measure the acidic properties of the used catalysts. This is because volatile materials were desorbed and deposited on the detector during the TPD experiment,leading to rapid reduction in the sensitivityof the detector. Therefore,a thermal gravimetric analyzer (TGA) was used for the fresh and used catalysts in the TPD experiment. The usual temperature range of 373-873 K in the TPD experiment was employed for the fresh catalysts.' Whereas the temperature range was 373-683 K for the used catalysts, since a large amount of adsorbed ammonia reacted with coke on the catalyst at atemperature higher than 683 K. Hence,the amount of acid sites of the used catalysts was underestimated by about 10% more than that of the fresh catalysts. The total amount of desorbed ammonia was regarded as the total amount of acid sites. The TPD spectra of HZSM-5and mordenite show two peaks which are usually separated at about 573 K.2 The peak at a higher temperature range correspondsto ammoniadeaorbed from acid sitesof strong acid strength,whereas that in the lower temperature range correspondsto acid sites of weak acid strength. Therefore,the amountof ammonia desorbed above 573 K was used as the amount of acid sites with strong acid strength.6 3. Results and Discussion 3.1 Reaction Pathway. Figure la-d shows the typical relationships between product distributions and time factor, WIF, at different temperatures.3 As the WIFvalue 31,
(4) Hashimoto,K.; Masuda, T.;Mori, T. Stud. Surf. Sci. Catal. 1986, 28,503-510. (6)Mori, N.; Nishiyama, S.;Tsuruya, S.; Masai, M. AppZ. Catal. 1991, 74,,37-52.
I
G A S E I N E
Figure 2. Reaction pathway proposed in this work.
increased, heavy oil was cracked to produce gasoline and gaseous products. Also, the gasoline product subsequently underwent further cracking to yield gaseous products. Hence, the gasoline yield showed a maximum value which appeared in the lower WIF value at higher reaction temperature. Coke was gradually increased and was considered to be produced from gasoline and heavy oil. Figure 2 illustrates a possible reaction pathway for explaining the product distributions shown in Figure la-d. Weekmane proposed a model of a three-lump reaction pathway for the cracking of gasoil of petroleum. The three lumps were gasoil, gasoline, and the sum of gas and coke. However, gaseous products reported in this work were mostly olefins, not paraffins.2 Olefins are useful intermediates in many chemical reactions, such as methanol to gasoline (MTG) reaction.' Furthermore, since the amount of coke deposition on an REY zeolite is significant as compared to HZSM-5 zeolite and silica alumina? coke should be taken separately as an independent product to facilitate precise analysis of catalyst deactivation caused by coke deposition. These results will be presented in future publications. Hence, it is desirable to improve Weekman's model by separating coke and gaseous products. An improvement of Weekman's model was proposed by Oliveira and Biscaia.8 Their model, however, omitted the (6) Weekman, V. W. Ind. Eng. Chem. Process Des. Deu. 1969,8 (3), 385. (7) Clarence, D. C.; Silvestri, A. J. J. Catal. 1977,47, 249-259.
Energy & Fuels, Vol. 8, No.1, 1994 133
Catalytic Cracking of Heavy Oil reaction from heavy oil to coke. In this paper, the proposed reaction pathway separately takes into account the heavy oil, gasoline, gas, and coke lumps and is considered to represent the product distributions shown in Figure la-d. 3.2 Kinetic Model. The experimental conditions were set up to ensure that the heat and mass transports limiations across the film were negligible.2 Also, limitations due to the intraparticle diffusion was assumed to be insignificant. The mass balance equation of the ith component can be written as follows:
0
1
2
3
Time factor, W/F [kg-catkg-oil'! hl
Ui
-d=W ri (i = A, B, C, D) where Fi is the mass flow rate of the ith lump [kgh-'I, W is the mass of catalyst [kgl, ri is the production rate of the ith lump per unit mass of catalyst [kg-(kg cat.).h-l], and suffixes A, B, C, and D refers to heavy oil, gasoline, gas, and coke lumps, respectively. Weekman6 reported that the cracking of heavy oil followed second-order kinetics with respect to the weight fraction of heavy oil, whereas that of gasoline was expressed by first-order kinetics with respect to gasoline. Since each lump had distribution of carbon numbers, e.g. C5-Cll for gasoline, concentrations based on mass [kgm-3] were employed in this work. On the basis of the proposed reaction pathway in Figure 2, the following differential equations were obtained using eq 1:
MA = -(kl dW
0
1
2
3
Time factor, W/F [kg-cat.kg-oil".h]
Figure 3. Second-order test for the cracking of heavy oil lump: (a) zeolite with a catalyst crystal size of 0.1 pm (REY-l), (b) zeolite with a catalyst crystal size of 1.0 pm (REY-2).
+ k, + Fz3)C2 (3) (4)
The conversion of heavy oil, XAwas defined by where suffix 0 refers to the reactor inlet. The integration of eq 7 gives
(5) where Ci represents the mass concentration of the ith lump. The effect of volume changes during the reaction was negligible, because the flow rate of the inert carrier gas was large and the mass flow rate of feed oil was small. Hence, the total mass concentrations at the inlet of the reactor and inside the reactor were assumed to be equal. Thus, the mass concentration of the ith lump, Ci, can be written under conditions of constant temperature and pressure by Ci = Cj i (i = A, B,C,D)
(6)
where COis the mass concentration of feed oil at the reactor inlet and fi is the weight fraction of the ith product lump which is equal to the ratio of Fi to the mass flow rate of total hydrocarbons, namely F. Equation 6 was substituted into eqs 2-5, yielding (7)
(8) Oliveira, L. L.; Biscaia, E. C., Jr. Ind. Eng. Chem. Res. 1989, 28, 264-271.
Figure 3, parts a and b, shows the plots of [XA/(l xA)]/c02fA0against the time factor, which were obtained at a constant COvalue of 0.1 kg.m3. The [XA/(l- XA)I/ COf'AO value was well correlated to WIF, suggesting that eq 7 was valid. The validities of other reaction kinetics (eqs 8-10) were discussed in later sections. 3.3 Estimation of Kinetic Parameters. The kinetic parameters (kl-k5) in eqs 7-10 were evaluated by nodinem least-squares regression of the experimental data. This analysis was carried out by using SPEEDUP, a comprehensive modeling and simulation package, developed by Aspen Technology, Inc. The program used NL2SOL, which is an adaptive nonlinear least-squares a l g ~ r i t h m . ~ The effect of the number of experimental data points on the magnitude of the evaluated parameters and the product distributions of the reactions calculated by using the evaluated parameters was examined. A total of 16 experiments at different time factors, WIF, at 673 K were conducted. The kinetic parameters were evaluated by using all of the 16 and 5 selected data points. The differences between the evaluated parameters from 5 and 16 data points were found to negligibly small, as were the (9) Dennis,J. E., Jr.; Gay, D. M.; Welsch, R. E. ACM Trans. Math. Softur. 1981, 7 (3), 369-383.
134 Energy & Fuels, Vol. 8, No. 1, 1994
Songip et a1.
0
0
weak acid amount
0.5 1 1.5 Amount of acid sites [moLkg-']
2
Figure 5. Correlation between the amounts of strong, weak, and total acid sites of used catalystsand rate constants (kl)using catalyst with crystal size of 0.1 p m (REY-1)and W/F= 0.75 (kg cat.)-(kgoil)-l.h.
T''xlO3 [K-'1
Figure 4. Arrhenius plots of kinetic parameters. Table 11. Effect of Zeolite Crystal Sizes on Rate Constants crvstalsize [uml kl kz k2 ki kk 570 90 0.95 4.8 0.1 0.085 455 75 0.78 3.9 1.0 0.069
differences in the product distributions calculated from those parameters. Hence, only five experimentalruns were conducted at other reaction temperatures of 573,623, and 723 K. Substituting the optimized kinetic parameters into eqs 7-10, the product distributions were numerically calculated and are represented by curves in Figure la-d. The correlation coefficients between the calculated and experimental values at all reaction temperatures were above 0.99 (correlation coefficient of 1means perfect fit). Hence, the calculated values are in good agreement with the experimental data and the kinetic parameters are found to be well evaluated. 3.4 Effect of Catalyst Crystal Size. The magnitude of the diffusivity within usual catalyst pellet, such as silica alumina, is about 10-6 mZ/s,lOand its value is almost equal to that of intercrystalline diffusion. Whereas the intracrystalline diffusivity of faujasite zeolite including X- and Y-type zeolites is about 10-10m2/s.11Hence, the dominant resistance to mass transfer was considered to be the resistance within the zeolite crystallite. The reaction was performed using REY zeolites with different crystal sizes. The effect of catalyst crystal size on the kinetic parameters are summarized in Table 11. The larger the crystal size, the smaller the rate constants. Using the differences between zeolite catalyst with crystal sizes of 0.1 and 1.0 pm, the effectiveness factors were estimated to be 1.0 and 0.8 for catalyst crystal size of 0.1 and 1.0 pm, respectively. The estimation method is described in the Appendix. Hence, the reaction proceeds over the zeolite catalyst with crystal size of 0.1 pm under rate-control condition. 3.5 Effect of Temperature. Figure 4 shows the plots of Arrhenius type using the evaluated kinetic parameters. (10) Smith, J. M. Chemical Engineering Kinetics; 2nd ed.; McGrawHill: Auckland, 1970. (11) Ruthven, D. M. Principles of Adsorption and Adsorption Processes; John Wiley & Sons: New York, 1984.
Table 111. Activation Energies Using a Catalyst with a Crystal Size of 0.1 ficm (REY-1) rate rate constants E, (kJ-mol) constants E. (kJ-mol) ki 50.7 kr 35.1 kz 75.5 ks 42.1 k3 18.5
The data are found to lie on a straight line for each parameter. The slopes of these straight lines give the activation energies, which are listed in Table 111. The activation energy for the reaction of gas formation from heavy oil (k2) is 75.5 kJ-mol-l and is comparable with other data for gasification reactions: 58.6 kJ0mol-l in case of CaX catalyst12 and 61.5 kJ-mol-l in case of silica alumina13for gasification of polymer waste, and 75 kJemo1-l for the reaction of ga~0il.l~ Thus from the above discussion, the proposed reaction kinetics were consideredto be valid. The difference of 24.8 kJ-mol-l between activation energy of gaseous formation (kz) and that of gasoline formation (kl) explains the fact that the selectivity of gaseous products increased, while that of gasoline decreased with an increase in temperature, especially above 673 K.3 3.6 Correlation between Strong Acid Amounts and Activity. Figure 5 shows the typical relation to the kl value to the amounts of strong, weak, and total acid sites of the catalysts used for the reaction during the 3 h on stream. The k l value was found to be well proportional to the amount of strong acid sites, but not to the amounts of weak and total acid sites. The dependencies of other rate constants on the amount of strong acid sites of the used catalysts were examined, as shown in Figure 6, parta a and b. Values of rate constants increase linearly with the amount of strong acid sites. Hence, strong acid sites are the active sites for the catalytic cracking reaction, and a similar result was observed in our previous work.3 4. Conclusion
1.The catalytic crackingof heavy oil from waste plastics was conducted using rare-earth metal exchanged Y-type (REY) zeolite under conditions of a time factor, W/F, of 0.2-3 (kg cat.).(kg oil)-l.h and a reaction temperature of 573-723 K. (12) Ayame, A.; Uemichi, Y.; Yoshida, T.;Kanoh, H. J. Jpn. Petrol. Zmt. 1979,22 ( 5 ) , 280-287. (13) Uemichi,Y.;Ayame,A.;Yoshida,T.;Kanoh,H.Ibid. 1980,23(1),
.-.
.?.LA.? ""
(14)Weekman, V. W., Jr.; Nace, D. M. AIChEJ. 1970,16 (3),397-404.
Energy & Fuels, Vol. 8,No. 1, 1994 136
Catalytic Cracking of Heaoy Oil
. , . , . ,
a120,
,
.
.
E. = activation energy, kJ.mol-l fi = mass fraction of ith lump ( = F i n F = mass flow rate of feed oil, kph-l Fi = mass flow rate of ith lump, kg-h-l kl, kg, k~ = intrinsic second-order rate constant, ma.(kg.(kg
I 5
cat.)-h)-'
kd, k6 = intrinsic first order rate constant, m8.(kg cat.)-h)-l ri = production reaction rate of ith lump, kp((kg cat.)-h)-l W = mass of catalyst, kg-cat XI = conversion of heavy oil lump (= 1 - F ~ F A o ) 0
0.1
0.2
0.3
0.4
0.5
Subscripts A = heavy oil lump
Amount of strong acid sites [mol.kg-']
B = gasoline lump C = gaseous products lump D = coke lump i = ith product lump 0 = reactor inlet
Appendix: Estimation of Effectiveness Factor
0
0.1
0.3
0.2
0.4
0.5
Amount of strong acid sites [mol.kg-']
Figure 6. Correlation between the amount of strong acid sites of used catalysts and rate constants using a catalyst with a crystal and W / F = 0.75 (kg cat.).(kg oil)-l.h: (a) size of 0.1 pm (REY-1) kz, ks, (b) kr, ks. Table IV. Effemtivenerr Factor, 4 B
cwstal size (um) 0.1 1.0
kl 1 0.78
k2
kn
1 0.83
1 0.82
kr
1 0.81
When overall reaction rates, roa,land roa8,are measured respectively for two catalysts with different crystal sizes, 2 R 1 and 2R2, the following relations hold 42/41 = R,/R, (A-1) q 2 h = 'oa,zl'oba,1 (A-2) where 4 is the modified Thiele Modulus and q is the effectiveness factor. When the reaction kinetics is of first order, eq A-2 can be rewritten as t z h = kotm,$kobs,l (A-3) where
ka 1 0.81
2. A kinetic model was proposed to represent the catalytic cracking reaction, where heavy oil, gasoline, gas, and coke were separately taken into account as products. The kinetic parameters appeared in the proposed model were evaluated by nonlinear least-squares regression method. The proposed kinetic model was found to well represent the experimental results. The activation energies obtained in this work were in good agreement with those reported by other investigators. 3. The analysis of the effectiveness fador showed that the resistance to pore diffusion was negligibly small, especially for the REY zeolite catalyst having a crystal size smaller than 0.1 pm. 4. The rate constant for each step in the catalytic cracking reaction was found to be well proportional to the amount of strong acid sites of the used catalysts.
Nomenclature Ci = mass concentration of ith lump, kg.m4 Co = maas concentration of feed oil a t the reactor inlet, kg-m-9
(A-4) and koa is the observed overall rate constant calculated in this work. The effectiveness factors for first-order kinetics ( k , and ks) can be estimated by the following procedures: (i) assume a value of 91, (ii) calculate q~ from eq A-3, (iii) determine 42 using eq A-4, (iv) calculate $1 from eq A-1, and (v) determine q1 using eq A-4. These procedures were repeated until the initial and calculated values of q1 agree. The conversion of heavy oil over zeolite catalysts with crystal sizes of 0.1 and 1.0pm obeyed second-order kinetics (eq 121, within experimental error, as shown in Figure 3, parts a and b. Assuming that the concentrations of heavy oil on the outer surface of the crystallite of the two zeolite catalysts were about the same, eqs A-1 and A-3 could be approximately applied for second-order kinetics. The effectiveness factors for kl, kz, and kg were estimated by the same procedures as for first-order kinetics, but using the chart of 4 vs q that was proposed by Bischoff.16 The effectiveness factors thus obtained are listed in Table IV. (15)Biechoff, K.B.AZChE J. 1965, II (2), 351-355.
