Anda di halaman 1dari 9

Applied Catalysis A: General 177 (1999) 227235

5-Lump kinetic model for gas oil catalytic cracking


Jorge Ancheyta-Juareza,b,*, Felipe Lopez-Isunzac, Enrique Aguilar-Rodrgueza,b
a

Instituto Mexicano del Petroleo, Eje Central Lazaro Cardenas 152, Mexico 07730, DF, Mexico b Instituto Politecnico Nacional, ESIQIE, Mexico 07738, DF, Mexico c Universidad Autonoma Metropolitana-Iztapalapa, Mexico 09340, DF, Mexico Received 1 May 1998; received in revised form 20 July 1998; accepted 20 July 1998

Abstract A new 5-lump kinetic model is proposed to describe the gas oil catalytic cracking (FCC) process. The model contains eight kinetic constants, including one for catalyst deactivation, taking into account LPG (combined C3C4), dry gas (C2 and lighter) and coke yields separately from other lumps (unconverted gas oil and gasoline). Apparent activation energies were determined from experiments obtained in a microactivity reactor (MAT) at temperatures: 4808C, 5008C and 5208C; for a catalyst-to-oil ratio of 5 using vacuum gas oil and equilibrium catalyst, both recovered from an industrial FCC unit. Product yields predicted by this model show good agreement with experimental data. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Gas oil catalytic cracking; Lump kinetic model

1. Introduction The gas oil catalytic cracking (FCC) is one of the most important processes in a modern renery oriented towards maximum production of high octane gasoline. The complexity of gas oil mixtures makes it extremely difcult to characterize and describe its kinetics at a molecular level. Therefore, the description of the complex reactions, which occur in this process, has been studied by lumping large numbers of chemical compounds [1]. Many kinetic models for the FCC process have been proposed [25]. Some authors included heavy products as lumps [68]. Others separate the components
*Corresponding author. Address: Instituto Mexicano del Petro leo, Eje Central Lazaro Cardenas 152, Mexico 07730, DF, Mexico. Fax: +52-5-368-9371; e-mail: jancheyta@hotmail.com

by the chemical type [9,10,14]. However, all these models involve the gaseous products as a lump and in some cases the gases are also lumped together with the coke yield. The prediction of coke yield separately from other lumps becomes very important to perform heat integration studies and to design and simulate air blowers and the FCC regenerator and reactor. Conversely, models which describe the gaseous products in detail [11,12] have many kinetic parameters (>20) to be estimated. This is an important consideration for evaluation studies when laboratory data available for determining rate constants are usually limited. In the present work we propose a new 5-lump kinetic model for catalytic cracking which splits the light gas-lump into dry gas and LPG. This separation is very important because the key FCC products can be predicted separately.

0926-860X/99/$ see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S0926-860X(98)00262-2

228

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

2. Cracking kinetic model It is necessary to use kinetic models with some detailed product distribution in order to predict the behavior of commercial FCC units [13]. However, the more lumps a model includes, intrinsically more kinetic parameters that need to be estimated and, consequently, more experimental information is required. Thus, taking into account that the major industrial products in FCC process are gasoline, LPG, dry gas and coke and considering the observations that some authors, who found that the probability of the cracking reaction from gasoline to coke [7,15] and gases to coke [8,16] can be neglected since the kinetic constants for these reactions were many orders of magnitude smaller than for the others, the model shown in Fig. 1 could be used to describe the gas oil catalytic cracking. This model considers ve lumps (unconverted gas oil, gasoline, LPG, dry gas and coke) and it has only seven kinetic constants and one for catalyst deactivation to be estimated. The main advantage of this model over the previous ones is that it can predict the coke formation, which supplies the heat required for the heating and vaporization of the feedstock and to perform the endothermic reactions; LPG, which contains important hydrocarbons (C and C ) used together with iso3 4 butane as alkylation and MTBE feeds; and dry gas, which is used as fuel gas in renery. These three products (coke, LPG and dry gas) can be predicted independently from other lumps with the proposed model. Another advantage is that the proposed model has less kinetic parameters to be estimated compared with

other 5-lump kinetic models reported in the literature [7,8,10]. The introduction of heavy feeds to FCC units, like atmospheric residuum, increases the amount of coke and gaseous products, particularly hydrogen content in dry gas, from the dehydrogenation activity of nickel and vanadium, causing limitations in the plant operation in terms of gas compression or regeneration airblowing capacity, in addition to the negative effect on gasoline yield and catalyst deactivation. Thus, the prediction of dry gas and coke yields becomes very important to design and simulate the gas compressor and air blower. One limitation is that the kinetic model does not consider products heavier than gas oil, such as light cycle oil and heavy cycle oil. These two products are taken together as the unconverted gas oil in the proposed model. Another limitation is that the parameters of the kinetic model depend on the feedstock and catalyst properties. 3. Differential fixed-bed reactor model 3.1. Kinetic expressions For each reaction, a kinetic expression (ri) was formulated as a function of product yields (yi), deactivation function (0) and kinetic constants (ki). Gas oil cracking was considered as a second order reaction and gasoline and LPG as rst order [2]. The use of rst order reaction for cracking of LPG has been discussed in the literature [17]. This is because the most reactive molecules of the feedstock lump disappear rst and the remaining molecules have a lower kinetic constant when conversion increases [17]. The exponential law was assumed for catalyst decay (0) which depends on the time-on-stream (tc). A non-selective deactivation model, which is based on the hypothesis that 0 is the same for all reactions, was used in this study in order to simplify the overall kinetic model and parameter estimation [18]. A selective deactivation model would approximate more the reality and would give slightly better results, however it would need more accurate deactivation kinetic data [19]. Based on these assumptions, the reaction rates of the proposed model are Gas oil X r1 k1 k2 k3 k4 y2 0 1 (1)

Fig. 1. Proposed 5-lump kinetic model. LPG: C3'sC4's. Dry gas: C2 and lighter.

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

229

Gasoline X Gas LP X Dry gas X Coke X

r2 k1 y2 k5 y2 k6 y2 0 1 r3 k2 y2 1 k5 y2 k7 y3 0 r4 k3 y2 k6 y2 k7 y3 0 1

(2) (3) (4) (5) (6)

r5

k4 y2 0 1 0 ekd tc

Decay function X

The minimization of the objective function, based on the sum of squared errors between experimental and calculated yields, was applied to nd the best set of kinetic parameters. This objective function was solved using the least squares criterion with a nonlinear regression procedure based on Marquardt's algorithm [24]. 4. Experimental 4.1. Materials The feed, vacuum gas oil, and commercial equilibrium catalyst taken from the circulating inventory of a catalytic cracking plant used in this study, were recovered from an industrial FCC unit (Tula Renery in Mexico). The equilibrium catalyst was previously de-coked at 5808C during 3 h. The characterization of the feed and catalyst are reported in Tables 1 and 2 respectively. 4.2. Microactivity runs The MAT (microactivity test) technique, a normalized ASTM procedure for a standard feedstock which allows to change easily the reaction conditions, was used for kinetic measurements. The xed-bed tubular plug ow reactor, reactor oven, oil injection and
Table 1 Feedstock properties (vacuum gas oil recovered from Tula Refinery in Mexico) API gravity Aniline point (8C) Molecular weight KUOP Sulfur (wt%) Nitrogen (ppmw) Refractive index at 208C Nickel (ppmw) Vanadium (ppmw) ASTM distillation (8C) IBP 30 vol% 50 vol% 70 vol% 90 vol% FBP 25.5 79 352 11.86 1.98 877 1.5085 0.1 0.8 235 409 433 467 517 568

3.2. Transport effects In order to ensure that the data were collected in the true kinetic regime and transport effects were insignicant, the following criteria were examined and satised [20,21] (L, catalyst bed length; dp, particle diameter; Pe, Peclet number; Rep, Reynolds number based on particle diameter; x, fractional conversion; n, number of moles): L 20n 1 ln b dp Pe 1x Pe b Pemin where Pe Pemin 0X087 Re0X23 p 333 vsX 0X0015 (7) (8)

56 vsX 0X033   L dp

(9) (10)

1 8n ln 1x

The Thiele modulus (0s) presented values lower than 0.3, therefore the isothermal effectiveness factor () is essentially equal to unity. These results agree with those reported by Hari et al. [22] using a similar MAT unit and operating conditions. 3.3. Mass balance The kinetic model was incorporated into an isothermal plug ow reactor model. Axial dispersion and internal diffusion in the differential reactor are neglected. The following mass balance, solved with the RungeKutta method, was used to evaluate the product yields (yi) from a set of kinetic constants for each temperature [23] (z, reactor length; WHSV, weight hourly space velocity; &L, liquid density; &C, catalyst density):   dyi 1 &L (11) ri dz WHSV &C

230

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

Table 2 Industrial equilibrium catalyst properties Physical properties Superficial area (m2/g) Average bulk density (kg/m3) Chemical properties Al2O3 (wt%) Re2O3 (wt%) Na (wt%) Fe (wt%) Ni (ppmw) V (ppmw) Cu (ppmw) Ca (wt%)
a

155 890

brine volume. Varying the time of injection, and consequently the gas oil rate, establishes a range of catalyst time-on-stream values to provide different space-velocities. 4.3. Analysis of products

34.8 1.5 0.4 0.51 503 2200 20 0.17

Initial coke content.

products recovery system is described by ASTM D3907-92 method [25]. Experimental runs were performed at reaction temperatures of 4808C, 5008C and 5208C and constant catalyst-to-oil (C/O) ratio of 5. In each experiment, a new portion of equilibrium catalyst was used. Preheated feed (0.80.005 g) is injected using a syringe bomb through a 4 g bed of catalyst maintained at the required cracking temperature. Liquid product from the reactor is collected in an ice-cooled receiver. The uncondensed gaseous products pass through the liquid products and are collected in a brine solution. The gas volume is determined by displacement of the

Gaseous products (C1C6) were analyzed by gas chromatography. Liquid products were analyzed by the chromatographic simulated distillation procedure described by ASTM D-2887 method. The carbon content of the catalyst was determined after reaction by combustion using an infrared analysis of the produced CO2. The fractions of gasoline, light cycle oil and heavy cycle oil were dened by the cut points at C52208C, 2203428C and 3428C, respectively. The product yields were calculated as weight percent of the reactant. Mass balances were performed for each run in the range 1005% and the conversion in weight percent was evaluated as the sum of C5gasoline, LPG, dry gas and coke, representing 100% minus light and heavy cycle oil yields. 5. Results and discussion Experimental data of product yield proles for gasoline, LPG, dry gas and coke at 4808C, 5008C

Fig. 2. Gasoline yield vs. space velocity. Catalyst-to-oil ratio of 5 and reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C.

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

231

Fig. 3. LPG yield vs. space velocity. Catalyst-to-oil ratio of 5 and reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C.

Fig. 4. Dry gas yield vs. space velocity. Catalyst-to-oil ratio of 5 and reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C.

and 5208C as a function of space velocity (WHSV) are presented in Figs. 25 respectively. Conversion as a function of reaction temperature is shown in Fig. 6. Fig. 2 shows that gasoline yield is always increasing, it means that the reaction is below the overcracking. However, the last two points (WHSV: 6.1 and 7.6 h1) indicate the beginning of the overcracking, as can be seen by the tendency of the gasoline yield

curves, which seem to have a maximum near the small values of space velocity (WHSV<6 h1). The smaller space velocity increases the contact time and favors gas oil conversion and gasoline yield (Figs. 2 and 6). Table 3 shows the estimated values of apparent activation energies and kinetic constants at 5008C for each reaction lump involved, which are within the range of those reported in the literature (520 kcal/ mol) [26]. The 95% probability intervals show the EA's

232

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

Fig. 5. Coke yield vs. space velocity. Catalyst-to-oil ratio of 5 and reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C.

Fig. 6. Conversion vs. space velocity. Catalyst-to-oil ratio of 5 and reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C.

to be estimated quite precisely as can be seen in Table 3. The knowledge of the apparent activation energies is very necessary to apply the model to industrial reactor simulation where the catalyst temperature changes from 7508C to 5308C. Fig. 7 shows a comparison between experimental and predicted product yields for gasoline, LPG, dry gas and coke at reaction temperature 4808C using the

kinetic model proposed in the present work. Fig. 8 shows the experimental and calculated gasoline yields in the range of reaction temperature 4805208C. It can be observed from these gures that the 5-lump kinetic model predicts sufciently well the experimental data. Although lumping, in a sense, can be considered semi-empirical, the reliability of the estimated kinetic parameters and predicted values is considerably improved compared with those obtained by other

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235 Table 3 Apparent activation energies determined in the range of reaction temperature of 4805208C and kinetic constants at 5008C for each reaction lump (k1, k2, k3, and k4 in wt frac1 h1; k5, k6 and k7 in h1) Kinetic parameter k1 k2 k3 k4 k5 k6 k7 Kinetic constant at 5008C 0.1942 0.0348 0.0001 0.0140 0.0061 0.0032 0.0200 Activation energy, EA (kcal/mol) 13.690.02 12.460.05 11.800.08 7.630.15 17.410.19 10.180.38 9.490.49

233

correlations for FCC product yields predictions compared with theoretical methods can be found in [28]. FCC reactor simulation should be used to validate the proposed kinetic model and also to predict how the effects of reaction conditions affect the product yields in order to evaluate the predictive ability of the model outside the range of the data used to generate the kinetic parameters. 6. Conclusions A new 5-lump kinetic model for gas oil catalytic cracking is proposed. The model includes unconverted gas oil, gasoline, LPG, dry gas and coke as lumps and has seven kinetic constants and one for catalyst deactivation. Experimental data obtained in a microactivity reactor at temperatures of 4808C, 5008C and 5208C, and catalyst-to-oil ratio of 5, were used to estimate apparent activation energies for each involved reaction. Product yields predicted by this model show a good

methods, e.g. curve tting and use of polynomials. The proposed kinetic model gives accurate predictions of product yields in FCC process with average deviations less than 2%. Linear correlations reported by Pope and Ng [27] give high deviations with respect to our experimental data (1520%). A complete discussion of the disadvantages of the use of empirical

Fig. 7. Predicted (lines) and experimental (symbols) product yields for (&) gasoline, (&) LPG, () dry gas and (*) coke at reaction temperature of 4808C and catalyst-to-oil ratio of 5.

234

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235

Fig. 8. Predicted (lines) and experimental (symbols) gasoline yields at reaction temperatures of (&) 4808C, () 5008C, and (*) 5208C and catalyst-to-oil ratio of 5.

agreement with experimental data with average deviations less than 2%. The advantage of this model over the previous ones is its ability to predict the LPG and dry gas yields separately from other lumps. This is very important to analyze the effects over important components of the FCC units, such as compressors, air blowers and regenerator. Acknowledgements The authors wish to thank Instituto Mexicano del Petroleo, CONACyT and the British Council for their nancial support. References
[1] F.J. Krambeck, Kinetics and Thermodynamic Lumping of Multicomponent Mixtures. Elsevier, Amsterdam, 1991, pp. 111129.

[2] F.H. Blanding, Ind. Eng. Chem. 45 (1953) 1186. [3] V.M. Weekman, Ind. Eng. Chem. Prod. Res. Dev. 7 (1968) 9095. [4] L.C. Yen, R.E. Wrench, A.S. Ong, Katalistics' Eighth Annual Fluid Cat Cracking Symposium, vol. 7, Budapest, Hungra, June 1987, pp. 17. [5] L.S. Lee, Y.W. Chen, T.N. Huang, W.Y. Pan, Can. J. Chem. Eng. 67 (1989) 615619. [6] T. Takatsuka, S. Sato, Y. Morimoto, H. Hashimoto, Ind. Chem. Eng. 27 (1987) 107116. [7] J. Corella, E. Frances, Fluid Catalytic Cracking, ACS Symp. Ser. 452 (1991) 165182. [8] R. Maya, F. Lopez-Isunza, Avances en Ing. Quim. 14 (1993) 3943. [9] S.M. Jacob, B. Gross, S.E. Voltz, V.W. Weekman, AIChE J. 22 (1976) 701713. [10] M. Larocca, S. Ng, H. de Lasa, Ind. Eng. Chem. Res. 29 (1990) 171180. [11] T.M. John, B.W. Wojciechowsky, J. Catal. 37 (1975) 240 250. [12] A. Corma, J. Juan, J. Martos, J.M. Soriano, in: Proceedings of the Eighth International Congress on Catalysis, vol. II, Berlin, 1984, pp. 293304. [13] B.W. Wojciechowski, A. Corma, Catalytic Cracking Catalysts, Chemistry and Kinetics, Marcel Dekker, New York, 1986.

J. Ancheyta-Juarez et al. / Applied Catalysis A: General 177 (1999) 227235 [14] P.G. Coxson, K.B. Bischoff, Ind. Eng. Chem. Res. 26 (1987) 12391248. [15] L.L. Oliveira, E.C. Biscaia, Ind. Eng. Chem. Res. 28 (1989) 264271. [16] I. Pitault, M. Forissier, J.R. Bernard, Can. J. Chem. Eng. 73 (1995) 498504. [17] F. Van Landeghem, D. Nevicato, I. Pitault, M. Forissier, P. Turlier, C. Derouin, J.R. Bernard, Appl. Catal. A 138 (1996) 381405. [18] J. Corella, M. Garca-Dopico, E. Frances, AIChE 1994 Spring National Meeting, 1721 April 1994. [19] J. Corella, F.G. Morales, M. Provost, A. Espinosa, in: D.N. Saraf, D. Kunzsu (Eds.), Recent Advances in Chemical Engineering, McGraw-Hill, New Delhi, 1988, pp. 192210.

235

[20] D. Mears, Ind. Eng. Chem. Proc. Des. Dev. 10 (1971) 541. [21] F.M. Dautzenberg, Characterization and Catalyst Development, Am. Chem. Soc. 11 (1989) 99119. [22] C. Hari, K.S. Balaraman, A.R. Balakrishnan, Chem. Eng. Technol. 18 (1995) 364369. [23] B.C. Gates, J.R. Katzer, G.C. Schuit, Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979 p. 92. [24] D.W. Marquardt, J. Soc. Ind. Appl. Math. 2 (1963) 431441. [25] ASTM Standard Methods. ASTM D-3907-92 Method (1992). [26] U.A. Sedran, Catal. Rev.-Sci. Eng. 36 (1994) 405431. [27] A.E. Pope, S.H. Ng, Fuel 69 (1990) 539546. [28] D. Wallenstein, U. Alkemade, Appl. Catal. A 137 (1996) 3754.

Anda mungkin juga menyukai