Studies on the Kinetics of Heavy Oil Catalytic Pyrolysis

Transcription

1 60 Ind. Eng. Chem. Res. 00, 4, Studies on the Kinetics of Heavy Oil Catalytic Pyrolysis Meng Xiang-hai,* Xu Chun-ming, Li Li, and Gao Jin-sen State Key Laboratory of Heavy Oil Processing, University of Petroleum, Beijing, China 049 For Daqing atmospheric residue catalytic pyrolysis over a LCM-5 catalyst, the effects of the reaction temperature, residence time, and weight ratios of catalyst to oil and steam to oil on product distribution were studied in a confined fluidized-bed reactor. The results show that the effects of these factors differ from each other, while the reaction temperature is the most important one. Ethylene yield increases with the reaction temperature, while the yields of propylene, butylene, and overall light olefins pass through maxima. A new five-lump kinetic model is developed for heavy oil catalytic pyrolysis, and a simple parameter estimation method is proposed in this paper. Rate constants and model parameters are estimated with the proposed method and experimental data. The simulated results illustrate that the five-lump kinetic model can well predict the yields of gasoline and diesel, gaseous alkanes, and light olefins.. Introduction Catalytic pyrolysis is a process of cracking petroleum to produce light olefins at high temperature over a catalyst. Compared with conventional steam pyrolysis, catalytic pyrolysis can reduce energy cost and flexibly adjust product distribution. There has been great interest in the research of heavy oil catalytic pyrolysis in China since the 980s. The Research Institute of Petroleum Processing and Luoyang Petrochemical Engineering Corp. have developed some catalytic pyrolysis technologies separately, such as deep catalytic cracking (DCC,, catalytic pyrolysis process (CPP,,4 and heavy oil contact cracking process (HCC, 5,6 etc. Theoretically, high temperature, short residence time, and large weight ratios of catalyst to oil and steam to oil are good to heavy oil catalytic pyrolysis for light olefin production. For different catalysts, these factors have different effects on the product distribution. Heavy oil catalytic pyrolysis involves a complicated reaction system. Its kinetics and corresponding models should be investigated so as to deeply study its reaction mechanism, to provide elemental parameter values, to simulate an industrial process, and to optimize operating conditions for industrial units. The conventional methods to estimate parameters are usually the Gauss- Newton method, the Marquardt method, and their modified methods. 7-9 In this paper, the effects of the reaction temperature, residence time, and weight ratios of catalyst to oil and steam to oil on light olefin yields and the product distribution of Daqing atmospheric residue catalytic pyrolysis are investigated over a LCM-5 catalyst (a catalyst for HCC technology in a confined fluidized-bed reactor. Then, a new five-lump kinetic model is developed, and model parameters are estimated.. Experimental Section.. Feedstock and Catalyst in the Experiment. In this paper, a Daqing atmospheric residue and LCM-5 are used as the feedstock and catalyst, respectively. Their main properties are given in Tables and... Apparatus. In experiments of heavy oil catalytic pyrolysis, a confined fluidized-bed reactor was used, the Table. Properties of Daqing Atmospheric Residue parameter value parameter value density (9 K, g/cm 0.9 group analysis, wt % viscosity ( K, Pa s 0. saturates 59. BMCI 6 aromatics 9. H/C atom ratio.79 resin and asphaltene.7 carbon residue, wt % 4. Table. Properties of LCM-5 parameter value parameter value microactivity index active component, wt % 9.4 packing density, g/cm.0 particle size distribution, wt % pore volume, cm /g µm.0 surface area, m /g µm 8.8 abrasion index, wt % µm 56. chemical content, wt % >80 µm 4. Al O 7. Fe O 0.50 schematic diagram of which is shown in Figure. It was comprised of five sections: oil and steam input mechanisms, a reaction zone, temperature control, a product separation, and a collection system. Distilled water and feedstock were kept in separate vessels and pumped by different pumps. A variable amount of distilled water was pumped into the furnace to form steam, and then the steam is mixed with the feedstock pumped by another pump simultaneously at the outlet of a constanttemperature box. The mixture was heated to approximately 77 K in a preheater before it entered the reactor, where catalytic pyrolysis reactions took place. Gaseous cracked products and steam were condensed and separated into two parts of cracked gas and liquid, which was then stewed and separated into water and cracked liquid... Operating Conditions. The ranges of operating conditions for the main catalytic pyrolysis runs are given in Table..4. Analytical Methods. Cracked gas was analyzed by a HP6890 gas chromatograph first, the data were disposed by a Chem Station, and then the volume percentage of components in cracked gas was obtained. The cracked liquid was analyzed first by a simulate distillation gas chromatograph to get the weight percentage of gasoline, diesel, and heavy oil, and then chromatograph-mass spectrograph was applied to ana- 0.0/ie004l CCC: $ American Chemical Society Published on Web 0//00 转载

2 Ind. Eng. Chem. Res., Vol. 4, No. 4, Figure. Schematic diagram of the experimental setup:, constant-temperature box;, steam furnace;, feedstock; 4, electronic balance; 5, oil pump; 6, water tank; 7, water pump; 8, preheater; 9, reactor furnace; 0, thermocouple;, reactor;, inlet and outlet of the catalyst;, filter; 4, condenser; 5, collecting bottle for liquid products; 6, gas-collect tank; 7, beaker; 8, gas sample bag. Table. Operating Conditions for Main Pyrolysis Runs parameter value reaction temperature, K 9-0 water inflow, ml/min catalyst weight, g 0-60 catalyst to oil weight ratio 0-0 steam to oil weight ratio residence time, s.0-.5 lyze the detailed components in cracked liquid and their relative percentage. The coke was analyzed by a coke analyzer, exploited by State Key Laboratory of Heavy Oil Processing, University of Petroleum, Beijing, China.. Experimental Investigations on Pyrolysis Reactions For Daqing atmospheric residue catalytic pyrolysis over LCM-5, the components of cracked gas and liquid were analyzed in detail, and the influences of the reaction temperature, residence time, and weight ratios of catalyst to oil and steam to oil on light olefin yields and product distribution were studied... Influence of the Reaction Temperature. The effectiveness of the reaction temperature was studied on conversion, selectivity, light olefin yields, and product distribution in the temperature range of 9-99 K. For these sets of runs, the weight ratios of catalyst to oil and steam to oil were kept fixed at 6 and 0.67, respectively, and the residence time was varied between.8 and.0 s. Effect of the Reaction Temperature on Conversion and Selectivity. Feed conversion was defined as the sum of the yields of cracked gas, gasoline, diesel, and coke. The selectivity was defined as the mass of light olefins formed per unit mass of feedstock cracked. The influences of the reaction temperature on conversion and selectivity are shown in Figure. The conversion of feedstock is above 90% and increases slightly with increasing reaction temperature, whereas the selectivity of light olefins shows a maximum. Figure. Effects of the temperature on conversion and selectivity. Effect of the Reaction Temperature on Yields of Cracked Gas, Liquid, and Coke. The effect of the reaction temperature on the yields of cracked gas, gasoline, diesel, and coke is shown in Figure. As can be seen, the yield of cracked gas increases significantly with temperature but almost holds the line when the temperature is over 96 K. On the contrary, the yield of gasoline and diesel decreases greatly with temperature and then does not show an appreciable change with temperature over 96 K. The yield of coke is comparatively very low and increases gradually with temperature. Cracked gas was analyzed in detail, and the results show that, in the reaction temperature range of K, the yield of total cracked gas is very high, exceeding 65 wt % of feedstock, and increases significantly with an increase of the reaction temperature. In cracked gas, the yield of the overall light olefins occupies wt % of total cracked gas. What is more, the ethylene yield is much bigger than the propylene yield, and the propylene yield is much bigger than the butylene yield.

3 604 Ind. Eng. Chem. Res., Vol. 4, No. 4, 00 Figure. Effect of the temperature on cracked gas, gasoline, diesel oil, and coke yields. Figure 5. Effect of the residence time on light olefins. Figure 4. Effect of the reaction temperature on light olefins. Figure 6. olefins. Effect of the weight ratio of catalyst to oil on light Cracked liquid was analyzed in detail, and the results indicate that the degree of catalytic pyrolysis is very high because cracked liquid is mainly composed of aromatics, including alkylbenzene, alkylnaphthalene, alkylindene, and alkylanthracene, etc. The percentage of chromatogram area of overall aromatics is about 80%. The yield of alkylbenzene decreases with temperature, while that of alkylnaphthalene increases, which shows that the pyrolysis degree becomes deeper and the condensation degree of cracked liquid increases with an increase of the temperature. Effect of the Reaction Temperature on Yields of Light Olefins. The yield variation of light olefins with temperature is shown in Figure 4. The yield of ethylene increases with temperature, while the yields of propylene, butylene, and overall light olefins pass through maxima. Under the same operating conditions, the ethylene yield is significantly higher than those of propylene and butylene. Besides, the gaps between the yield of ethylene and the yields of propylene and butylene become larger with increasing temperature. In practice, catalytic pyrolysis involves a catalytic pyrolysis reaction on catalyst and thermal pyrolysis reactions in catalyst void. Ethylene is mainly from the thermal pyrolysis reaction that follows the free-radical mechanism. As the reaction temperature increases, a proportion of the thermal pyrolysis reaction becomes higher, and for that reason, the ethylene yield increases gradually and the increasing velocity of the ethylene yield is much bigger than those of the propylene and butylene yields at lower reaction temperature. Propylene and butylene are not the final products, which can undergo secondary reactions, and the degree of secondary reactions becomes high with increasing temperature. Therefore, the yields of propylene and butylene show maxima... Influence of the Residence Time. Here, the residence time refers to the contact time of oil vapor with catalyst. The effect of the residence time on light olefin yields was studied in the range of.0-.5 s, keeping the reaction temperature, catalyst to oil weight ratio, and steam to oil weight ratio constant at 97 K, 6, and 0.5, respectively. As Figure 5 shows, the yield of overall light olefins decreases gradually with the residence time, while the yields of ethylene, propylene, and butylene decrease very slightly. The results show that the yield of total light olefins does not increase by prolonging the residence time, and catalytic pyrolysis of heavy oil is suitable for operating at short residence time. As residence time is prolonged, the yields of ethylene, propylene, and butylene will decrease because of secondary reactions... Effect of the Weight Ratio of Catalyst to Oil. Under almost the same experimental conditions (the reaction temperature and steam to oil weight ratio were kept constant at 97 K and 0.6, respectively, and the residence time was about.8 s, the effectiveness of the catalyst to oil weight ratio was studied on the yields of light olefins in the range of 0-0. The variation of light olefins with the catalyst to oil weight ratio is shown in Figure 6. As can be seen, the yields of ethylene and overall light olefins increase slightly with increasing catalyst to oil weight ratio, while the yields of propylene and butylene almost do not change. Catalytic pyrolysis

4 Ind. Eng. Chem. Res., Vol. 4, No. 4, Figure 8. Five-lump kinetic model. (4 To produce much light olefins, heavy oil catalytic pyrolysis is usually conducted under such conditions as high reaction temperature, short residence time, and large weight ratios of catalyst to oil and steam to oil. Figure 7. Effect of the weight ratio of steam to oil on light olefins. reactions cannot take place without acidity centers on catalysts. However, the microactivity index of the LCM-5 catalyst was only, and it indicates that acidity centers on the catalyst are of a fat lot. Thus, the catalyst s primary function is to offer energy and location. Here, the pyrolysis reaction is mainly thermal pyrolysis reactions, following the free-radical mechanism and creating much ethylene. When the catalyst is enough to offer energy and location for hydrocarbon pyrolysis, the effectiveness of the catalyst will become unimportant. Increasing the catalyst to oil weight ratio means high reaction rigor and high pyrolysis degree; besides, ethylene is close to the final product and, therefore, its yield increases slightly..4. Effect of the Weight Ratio of Steam to Oil. Under almost the same experimental conditions (the reaction temperature was kept constant at 97 K, the catalyst to oil weight ratio was kept fixed at 6, and the residence time was about.9 s, the paper studied the effectiveness of the steam to oil weight ratio on the yields of light olefins in the range of The variation of light olefin yields with the steam to oil weight ratio is shown in Figure 7. The yield of overall light olefins increases gradually with increasing steam to oil weight ratio, while the yields of ethylene, propylene, and butylene increase very slightly. Steam has the capacity of preventing coking on catalyst, which is in favor of pyrolysis reactions of hydrocarbons, and, consequently, the yields of light olefins increase with the steam to oil weight ratio..5. Characteristics of Heavy Oil Catalytic Pyrolysis. From the investigations of Daqing atmospheric residue catalytic pyrolysis, some features of heavy oil catalytic pyrolysis can be concluded as follows: ( Heavy oil catalytic pyrolysis follows both the carbonium ion mechanism and the free-radical mechanism. ( The yield of cracked gas of catalytic pyrolysis is very high. Besides, the ethylene content in cracked gas and the aromatics content in cracked liquid are also very high. ( Because the reaction temperature of catalytic pyrolysis is very high, some gaseous products with high molecular weight can undergo secondary cracking reactions. Besides, olefins can convert into alkanes by hydrogen-transforming reactions and can also convert into gasoline and diesel by polymerization and aromatization reactions. 4. Development of the Lumping Kinetic Model 4.. Development of the Physical Model of Heavy Oil Catalytic Pyrolysis. As for heavy oil catalytic pyrolysis over LCM-5, a five-lump kinetic model is developed in this paper. The physical model is shown in Figure 8. Features of the five-lump kinetic model are as follows: ( For fixed catalytic pyrolysis system, the feedstock can be considered as one lump. The feedstock would not be lumped according to group analysis, and accordingly the application scope of the model will be limited. However, the model is very simple; moreover, it is suitable for the present status that catalytic pyrolysis usually uses a paraffin atmospheric residue as the feedstock in China. ( Gaseous products can be divided into two lumping species: alkanes and olefins. Light olefins are the aimed products of catalytic pyrolysis; therefore, they need to be considered as one lump. Besides light olefins, there are a large amount of unaimed gaseous products, such as hydrogen, methane, ethane, propane, butane, etc. These unaimed gaseous products should also be considered as one lump. ( Because the reaction temperature of catalytic pyrolysis is very high, such gaseous products as propane and butane could convert into light olefins by secondary cracking reactions. Consequently, the reaction between alkane lumping species and olefin lumping species should be considered. (4 Coke is considered as one pump, despite its low yield. Similar to catalytic cracking, catalytic pyrolysis technologies are apt to utilize the operating manner of circulation fluidization for catalyst, and that heat balance is an important content for designation. Coke is very significant to heat balance for the pyrolysis system, so it needs to be considered as one lump. 4.. Development and Solution of Lumping Kinetic Models. The reaction network is shown in Figure 8, and the reaction of lump I generating lump J could be expressed by the following mathematical model: k ij I 98 v ij J ( To simplify the mathematical models of heavy oil catalytic pyrolysis, the following assumptions have been made: ( Heavy oil catalytic pyrolysis belongs to gas-solid phases catalytic reactions, and chemical reactions are the controlled step, while axial dispersion in the reactor can be neglected.

5 606 Ind. Eng. Chem. Res., Vol. 4, No. 4, 00 ( A confined fluidized-bed reactor can be considered as an isothermal plug-flow reactor. ( For heavy oil catalytic pyrolysis, reaction expression ( means irreversible reactions of the second-order. However, for other reactions, expression ( means irreversible reactions of the first-order. With the coking deactivation of catalysts considered, equations of the five-lump kinetic model about heavy oil catalytic pyrolysis can be deduced from the continuity equation and reaction rate equation as follows: dc dx - L u g (k + k + k 4 + k 5 C φ ( dc dx L u g [v k C - (k + k 4 + k 5 C ]φ ( dc dx L u g (v k C + v k C - k 4 C φ (4 dc 4 dx L u g (v 4 k 4 C + v 4 k 4 C + v 4 k 4 C φ (5 dc 5 dx L u g (v 5 k 5 C + v 5 k 5 C φ (6 Catalyst deactivation models of catalytic pyrolysis can use those of catalytic cracking as a reference and can be described by time-on-stream theory. The equation of catalyst deactivation models is as follows: φ e -Rt c (7 Because eqs -6 are associated with each other and are highly nonlinear, exact analytical solutions are not possible and then a numerical method for solving nonlinear differential equations must be employed. 0 Instantaneous concentrations of lumping species have been simplified in this paper, and then the approximate analytical solutions of eqs -6 can be obtained. The expression of the feedstock concentration on the outlet of the reactor can be obtained directly from eq : C 0 C f + L k u C 0 e -Rt c g (8 where k k + k + k 4 + k 5. If eq is divided by eq, we can obtain dc dc - k k C C -v k k (9 which has initial conditions X 0, C C 0, and C 0, where k k + k 4 + k 5. Equation 9 is a first-order nonhomogeneous linear differential equation, which has the following general solution: C exp( - k k C [ - ν k k exp( k k C dc + constant ] (0 For such indefinite integrals as e /x dx, its analytical solution is not possible, so only its integral value can be obtained. In virtue of infinite series, e x can be expanded into power series: In like manner, exp[(k /k (/C ] can also be expanded into power series; neglecting the high-order items and only taking the first six items, then we obtain Therefore e x + x + x! + x xn (! n! exp( k k C + k k C + ( k 4( k k k 4 C + 6( k exp( k k C dc { + k k C + ( k 6( k k C + k k ln(c + ( k 4 k k k C + C + 4 0( k 5 C ( 5 k C + C + 4( k k 4 C + 4 0( k k 5 C 5} dc k ( - C 6( + k k ( - C + 4( k -C + 0( k 5 k -4C 4 ( When eq is substituted into eq 0, with initial conditions, the concentration of the gasoline and diesel lump on the outlet of the reactor can be expressed as follows: C f exp( - k k C f v k k [ (-C f - k ln C k f + ( k k + C f ( k k C + f 7( k k 4 C + f 480( k k 5 C f 4] ( - exp - k k C f v k k [ (-C 0 - k ln C k 0 + ( k k + C 0 ( k k C + 0 7( k k 4 C ( k k 5 (4 C 0 4] Equation 4 divided by eq will give the concentration of the gaseous alkane lump in terms of C and C : dc - k 4 dc k C C -v k - v k C k k C (5 If the instantaneous concentration variation of C is neglected and C is substituted with the average concentration C f /, according to the solution of eq 9, the expression of the gaseous alkane lump on the outlet

6 of the reactor can be obtained: C f exp( - k 4 k C f{ v k k [ (-C f - k 4 ln C k f + ( k 4 k + C f ( k 4 k C + f 7( k 4 k 4 C + f 480( k 4 k 5 C f 4] + v k C f k [ + k 4 C f k C + f 6( k 4 k C + f 4( k 4 k C + 4 f 0( k 4 k 4 C + 5 f 70( k 4 k 5 C f 6]} ( - exp - k 4 k C 0{ v k k [ (-C 0 - k 4 ln C k 0 + ( k 4 k + C 0 ( k 4 k C + 0 7( k 4 k 4 C ( k 4 k 5 C 0 4] + v k C f k [ + C 0 k 4 k C + 0 6( k 4 k C + 0 4( k 4 k C ( k 4 k 4 C ( k 4 k 5 (6 If eq 5 is divided by eq, we obtain C 0 6]} dc 4 - v 4 k 4 - v 4 k 4 C - v 4 k 4 C dc k k C k C (7 When the instantaneous concentration variation of C and C is neglected and C and C are substituted with the average concentration C f / and C f /, the concentration of the light olefin lump on the outlet of the reactor can be expressed as follows: C 4f ( v 4 k 4 + v 4 k 4 C f + v 4 k 4 C f k k C 0 C f (C 0 - C f (8 Equation 6 divided by eq will give the concentration of the coke lump in terms of C and C : dc 5 dc - v 5 k 5 k - v 5 k 5 C k C (9 In like manner of the solution of eq 7, the expression of the coke lump on the outlet of the reactor can be obtained as C 5f ( v 5 k 5 k + v 5 k 5 C f k C 0 C f (C 0 - C f (0 In the above expressions, v ij M i /M j. The final analytical solutions of the five-lump kinetic model equations (-(6 are expressions (8, (4, (6, (8, and ( Solutions of Model Parameters. According to the final analytical solution expressions of the five-lump kinetic model equations of heavy oil catalytic pyrolysis, the Matlab program to seek model parameters was compiled by a least-squares method. On the basis of experimental data, the lump concentrations on the Ind. Eng. Chem. Res., Vol. 4, No. 4, Table 4. Rate Constants for Catalytic Pyrolysis of Daqing Atmospheric Residue a reaction temperature, K R, s k, (mol/kg of gas - s k, (mol/kg of gas - s k 4, (mol/kg of gas - s k 5, (mol/kg of gas - s k, s k 4, s k 5, s k 4, s a Negative values for k 4 indicate that the reaction in the lump model is reversed. Table 5. Frequency Factors and Activation Energies frequency factor activation energy k 0, (mol/kg of gas - s E, kj mol k 0, (mol/kg of gas - s E, kj mol k 40, (mol/kg of gas - s E 4, kj mol k 50, (mol/kg of gas - s - 8. E 5, kj mol k 0, s E, kj mol k 40, s E 4, kj mol k 50, s E 5, kj mol k 40, s E 4, kj mol -.64 outlet of the reactor at 97 and 99 K were calculated. With the calculated data and Matlab program, rate constants of reactions between lumps in the model were obtained and are listed in Table 4. Then frequency factors and activation energies were calculated and listed in Table 5 on the basis of rate constants and Arrhenius expression (. k k 0 exp(-e/rt ( It can be seen from the data in Table 4 that the values of catalyst deactivation constants are very small, which means that catalyst deactivation has a very slight influence on CPP. Under experimental conditions of heavy oil catalytic pyrolysis, the macroreaction orientation between gasoline and diesel and light olefins is that light olefins generate gasoline and diesel. The aromatics content in gasoline and diesel of heavy oil catalytic pyrolysis is very high, and aromatics cannot generate light olefins by ring-opening reactions, whereas light olefins can yield gasoline and diesel by polymerization reactions, cyclization reactions, and aromatization reactions. From the data given in Tables 4 and 5, we know that reaction rate constants of gaseous alkanes forming light olefins are very large and the activation energy is very small. This illustrates that gaseous alkanes easily form light olefins by secondary cracking reactions. This also explains the phenomenon that, in gaseous alkanes, the methane and ethane contents are very high, whereas the propane and butane contents are very low Effect Test of the Five-Lump Kinetic Model. The correlations of experimental yields and modelcalculated yields of gasoline and diesel, gaseous alkanes, light olefins, and coke at 97 and 99 K are shown in Figures 9 and 0, respectively. It is clearly illustrated that the model-calculated values and the experimental values have good correlations. With the prolonging of the residence time, the yields of gasoline and diesel, gaseous alkanes, and light olefins increase quickly at residence times of less than 0.5 s but increase very

7 608 Ind. Eng. Chem. Res., Vol. 4, No. 4, 00 Figure 9. Predict results of a five-lump model at 97 K. Figure 0. Predict results of a five-lump model at 99 K. slowly at residence times bigger than 0.5 s. The yields of gaseous olefins and light olefins reach maxima in the residence time range of s and then decrease slightly. The coke yield increases monotonically with the residence time. It is clear that a short residence time is in favor of producing more light olefins, and the coke yield is very low at that time. 5. Conclusions Such operating conditions as the reaction temperature, residence time, and weight ratios of catalyst to oil and steam to oil have different influences on light olefin yields and the product distribution of heavy oil catalytic pyrolysis, and the reaction temperature is the most significant one. The ethylene yield is much higher than

8 Ind. Eng. Chem. Res., Vol. 4, No. 4, the yields of propylene and butylene. Under optimized operating conditions, the ethylene weight yield can reach 5%, and the weight yield of the overall light olefins can exceed 50% at the highest. On the basis of the intensive investigation on the pyrolysis reactions for heavy oil catalytic pyrolysis, a new five-lump kinetic model has been developed. Concentration expressions of lumping species at the outlet of the reactor have been obtained, and a simple parameter estimation method has been proposed in this paper, which holds less strictness for initial values of model parameters. Rate constants and model parameters have been estimated with the method and experimental data. The simulated results have been tested, and the fivelump model can well predict the yields of gasoline and diesel, light alkanes, and light olefins. The simulated results have shown that a short residence time is good for the production of light olefins. The applicability and the differences from conventional methods are required for further investigation. Nomenclature I, J lumping species I and J k ij rate constant for the reaction of lump I to lump J E ij activation energy for the reaction of lump I to lump J, kj mol - k ij0 frequency factor for the reaction of lump I to lump J v ij chemical measurement coefficient for the reaction of lump I to lump J C i, C i0, C if mole concentration of lump I instantaneously and on the inlet and outlet of the reactor, mol/kg of gas X nondimensional reactor length L reactor length, m u g average velocity of hydrocarbons and steam in the reactor, m s - φ deactivation function of catalysts Rdeactivation constant of catalysts tc residence time of catalysts, s y i weight yield of lump I, wt % M i molecular weight of lump I, kg mol - R gas constant, J mol - K - T reaction temperature, K Literature Cited ( Xie, C. G. Commercial Application of Deep Catalytic Cracking Catalysts for Production of Light Olefins. Petrochem. Technol. 997, 6 (, ( Zhou, P. L. Deep Catalytic Cracking (DCC Technology. Petrochem. Technol. 997, 6 (8, ( Xie, C. G. Studies on Catalytic Pyrolysis Process for Ethylene Production and Its Reaction Mechanism. Pet. Process. Petrochem. 000, (7, (4 Xie, C. G.; Wang, X. Q.; Guo, Z. X.; Wei, Q. CPP Technology for Olefin Production and Its Commercial Trial. Pet. Process. Petrochem. 00, (, 7-0. (5 Sha, Y. X.; Cui, Z. Q.; Wang, L. Y.; Meng, F. D.; Wang, G. L. A New Process for Ethylene ProductionsHeavy Oil Contact Cracking Process. Pet. Process. Petrochem. 995, 6 (6, 9-4. (6 Sha, Y. X.; Cui, Z. Q.; Wang, M. D.; Wang, G. L.; Zhang, J. Olefin Production by Heavy-Oil Contact Cracking. Petrochem. Technol. 999, 8 (9, (7 Xi, S. L. Nonlinear Optimization Methods; Beijing, 99. (8 Ding, F. C.; Zhou, Z. J.; Li, X.; Wang, Z. W.; Zheng, S. H. Methods Predicting Parameters of FCC 5 Lumping Dynamic Model. Pet. Refin. Eng. 00, (4, (9 Luo, X. L.; Li, R. L. Estimation of Rate Constants for Complex Chemical Reactions. Acta Pet. Sin. (Pet. Process. Sect. 996, (, (0 Lee, L. S.; Chen, Y. W.; Huang, T. N.; Pan, W. R. Four- Lump Kinetic Model for Fluid Catalytic Cracking Process. Can. J. Chem. Eng. 989, 67, Received for review May 9, 00 Revised manuscript received September 5, 00 Accepted September 7, 00 IE004L