Heterogeneous Approach to the Catalytic Cracking of Vacuum Gas Oil

Sep 17, 2008 - Abarasi Hart , Amjad Shah , Gary Leeke , Malcolm Greaves , and Joseph Wood. Industrial & Engineering Chemistry Research 2013 52 (44), ...
2 downloads 0 Views 387KB Size
Ind. Eng. Chem. Res. 2008, 47, 7631–7641

7631

Heterogeneous Approach to the Catalytic Cracking of Vacuum Gas Oil Mustafa Al-Sabawi,† Jesu´s A. Atias,‡ and Hugo de Lasa*,† Chemical Reactor Engineering Centre, Faculty of Engineering, UniVersity of Western Ontario, London, Ontario, Canada N6A 5B9, and Engineering & Process Sciences, The Dow Chemical Company, Freeport, Texas 77541

This study examines the roles of intracrystallite diffusion and reaction phenomena during the catalytic cracking of vacuum gas oil. Catalytic cracking experiments on FCC-type catalysts were performed in a fluidized benchscale CREC riser simulator. This reactor was operated under close-to-industrial FCC conditions in terms of temperature, reaction time, partial pressures of reactant and products, and catalyst-to-oil ratio. The activity and selectivity of two USY zeolite catalysts, with very similar properties but varying zeolite crystallite sizes, were determined. A five-lump kinetic model describing the catalytic cracking of gas oil to light cycle oil, gasoline, light gases, and coke and accounting for diffusional constraints experienced by hydrocarbons while evolving in the zeolite pore network was considered. The results show that the catalyst with the smaller crystallites provided higher activity and selectivity toward desirable intermediate products (gasoline with low aromatics) and lower selectivity for terminal products (coke), indicating that diffusion plays a significant role in catalytic cracking. Diffusivity and kinetic parameters, including modified Thiele modulus and effectiveness factor, were established to determine the effects of crystallite size and temperature on the operating regime of the catalyst. It was found that, in the 510-530 °C range, the overall cracking rate is controlled by the highly temperature-sensitive intracrystalline gas oil transport, whereas in the 550-570 °C range, the overall cracking rate is dominated by a mildly temperature-sensitive intrinsic cracking rate. 1. Introduction The focus of many petroleum refineries has recently been shifted toward processing heavy petroleum feedstocks using the fluid catalytic cracking (FCC) process. In this respect, improved catalyst technology is needed to convert low-value heavy hydrocarbons to more valuable gasoline, middle distillates, and light olefins. This is of particular importance as the existence of bulky molecules in heavy feedstocks presents a major challenge to FCC processes because of possible hindered transport of hydrocarbon species within the zeolite pore network.1 Modern commercial FCC catalyst particles consist of a matrix containing USY zeolite crystallites where most of the activity resides.2 The matrix has large pores (macropores) and usually exhibits some activity to behave as a precracking agent. The macropores in the matrix also act as passageways for molecules to diffuse from the surrounding environment into the particle interior, thereby reaching the zeolite particles. The USY zeolites are considered to be the major component of currently used FCC commercial catalysts, as they provide the activity and selectivity of the catalysts. These zeolites consist of supercages that can inscribe 12.4-Å-diameter spheres. Rings, composed of 12 oxygen atoms and having a free diameter of 7.4 Å, act as openings into these supercages.3 Each supercage cavity is connected to four other cavities that are, in turn, connected to three-dimensional cavities to form a highly porous structure.4 It is within this pore structure that the locus of the adsorption capacity and catalytic activity reside for many reactions. Once reactant hydrocarbons reach a zeolite crystallite, they diffuse into its interior until they adsorb on the active sites and undergo chemical transformation. Generally, the pore sizes of USY zeolite crystallites are on the order of molecular dimen* To whom correspondence should be addressed. Tel.: +519-6612144. Fax: +519-850-2931. E-mail: [email protected]. † University of Western Ontario. ‡ The Dow Chemical Company.

sions, and therefore, molecules inside the micropores cannot escape the potential attraction of the zeolite pore walls. This means that molecules in the zeolite micropores exist mainly in an adsorbed state, rather than in free form.5 Hence, it is imperative to examine diffusion and reaction phenomena in the zeolite pore network when studying catalytic cracking of hydrocarbons. In the case of heavy feedstocks such as vacuum gas oil, many bulky molecules exist, amplifying the effect of diffusion in the catalytic cracking reaction. The adsorption of hydrocarbons on the active sites of the zeolite crystallites is a necessary step for the catalytic cracking reaction to occur.6 The extent to which hydrocarbons are adsorbed depends significantly on the reactor operating conditions, the feedstock being converted, and the catalyst properties. However, adsorption/desorption between the mobile phase and the adsorbed concentration on the framework often occurs rapidly compared to the time required for the overall transport, i.e., diffusion, and as a result, adsorption takes place at closeto-equilibrium conditions.7 Regarding catalytic cracking kinetics, a frequent approach adopted is that of lumped pseudohomogeneous models without specific descriptions of effective diffusivity and reaction phenomena. Consideration of these physical steps, accompanying the catalytic conversion intrinsic kinetics, has a major impact on sound overall kinetic rates as well as relevant hydrodynamic calculations in the riser unit.8,9 Given the above facts, the objective of the present work was to contribute to the modeling of vacuum gas oil catalytic cracking rates by incorporating the intracrystallite diffusion and intrinsic reaction steps and to study the effects of diffusion on the overall cracking of gas oil and the product selectivity. To accomplish this, data were collected from the fluidized riser simulator at the Chemical Reactor Engineering Centre (CREC) of the University of Western Ontario, a novel unit invented by de Lasa10 that overcomes the technical problems of the standard microactivity test (MAT). The CREC riser simulator is an ideal

10.1021/ie701745k CCC: $40.75  2008 American Chemical Society Published on Web 09/17/2008

7632 Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008

Figure 1. USY zeolite crystallites present in the FCC catalyst. F ) 825 kg of crystallities/m3, ε ) 0.51 m3 of voids in crystallite/m3 of crystallite, USY zeolite content ) 30 wt %.

tool for catalytic cracking reaction studies, allowing the development of experiments under conditions essentially identical to those of commercial units, given (a) close control of catalyst and gas-phase reaction times; (b) short contact times within the range of a few seconds similar to those encountered in industrial riser reactors; and (c) accurate control of temperature, reaction time, and partial pressures. To secure the value of the catalytic cracking of vacuum gas oil, modeling studies were developed under relevant FCC process conditions in terms of partial pressures of gas oil, temperatures (510-570 °C), contact times (3-7 s, for both the hydrocarbons and the catalyst), and catalyst/gas oil mass ratios (5), using well-fluidized catalysts. 2. Transient Diffusion of Hydrocarbon Species within a Reactive Zeolitic Catalyst In this study, the selected matrix for the FCC catalyst was silica, making it fully unreactive and thus providing an effective approach to assess zeolite performance without matrix interference. Consider the case of a well-mixed batch reactor (CREC riser simulator) in which zeolite-based catalysts are brought into contact with a gaseous phase containing a component species, i, that diffuses into the particle (of diameter dp) within which the zeolite crystals are embedded (Figure 1). There are three steps in the diffusion process: (1) First, component i in the bulk gas phase has to diffuse across the stagnant layer surrounding the particle. The stagnant “film” thickness is determined by the fluid-particle hydrodynamics. The higher particle Reynolds numbers obtained in the CREC riser simulator lead to low film diffusion and, consequently, negligible influence of this step in the overall reaction rate.11 (2) Next, component i diffuses into the macropores (or mesopores) of the inert matrix. Knudsen/ molecular diffusion dominates this transport process, and because this type of diffusion is relatively fast, it does not affect the overall rate. (3) Finally, component i diffuses into the zeolite crystallites. This diffusion process is known as intracrystalline, configurational, or micropore diffusion. This study also covers the possibility of surface diffusion as the diffusive motion of adsorbed species over the walls of the pores. To proceed with transient reaction modeling, one can consider a simplified spherical crystallite geometry of radius R. Thus, the transient uptake within the zeolite crystallites with chemical transformation of component species i can be described by a mass balance on the differential crystallite shell as (KiFc + εs)

(

)

Di,eff ∂ 2 ∂Ci,in ∂Ci,in )- 2 r + Fcrri ∂t ∂r r ∂r

(1)

where ri is the rate of consumption/formation of species i, ε is the crystallite voidage, Ki is the adsorption constant for species i and Fcr is the density of the USY zeolite crystallite. The described derivation assumes that the system is under isothermal conditions and that the diffusion phenomena within the pore structure can be represented by an overall effective diffusivity, Di,eff. It is also assumed that the adsorption is at equilibrium. Thus, adsorption is much faster than diffusion and reaction, so that equilibrium exists at every local position inside the crystallite. Furthermore, a number of boundary conditions are required for solving eq 1, as follows: (a) instantaneous vaporization (at t ) 0, Ci,ex ) Ci,ex|t)0), (b) symmetric concentration profiles inside the crystallites (at r ) 0, ∂ Ci,in/∂r ) 0), and (c) negligible transport limitations around the 60-µm catalyst particles and inside the inert matrix (at r ) R, Ci,in ) Ci,ex) Conditions a and c are fully satisfied under the operating conditions of the CREC riser simulator, whereas condition b can be validated given the crystallite shape with all three dimensions being of comparable magnitude.12 The catalytic cracking of vacuum gas oil in the mini-fluidized CREC riser simulator is an unsteady-state reaction process. If chemical reactions are assumed to take place in the zeolite crystallites of the FCC catalyst only, given the chemically inert matrix, a material balance of reactant species i in the outer surface of an assumed spherical zeolite crystallite can be written as -

(

∂Ci,in V dCi,ex ) Di,eff Wcr dt ∂r

| ) r)Rcr

3 RcrFcr

(2)

with Ci,ex and Ci,in representing the gas oil concentration outside and inside the crystallite, respectively. In eq 2, V represents the volume of the riser simulator, Wcr is the mass of zeolite crystallites loaded in the reactor, and Rcr is the crystallite radius. When reaction occurs simultaneously with mass transfer within a porous structure, a concentration gradient can be established by eq 1. Because interior surfaces are thus exposed to lower reactant concentrations than surfaces near the outer crystallite area, the overall reaction rate throughout the catalyst particle under isothermal conditions is less than it would be if there were no mass-transfer limitations. Under these conditions, the apparent activation energy, catalyst selectivity, and other important observed characteristics of a catalyst become dependent on the pore structure of the catalyst and the effective diffusivity of the reactants and products. Thus, one can envision

Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 7633

that FCC catalysts might display, under a certain set of operating conditions, reduced reactivity due to transport process limitations. A classical approach to assess the significance of transport processes under isothermal conditions is the use of an effectiveness factor, η. The effectiveness factor is defined as the ratio of the actual reaction rate to the rate that would occur if all of the surface throughout the inside of the catalyst particle were exposed to reactant of the same concentration and temperature as exist at the outside surface of the particle13 1 rmean rmean η) ) ri|r)R k C n|

( )

4 3 πR 3

i i,in r)R

)

hi′ )

∫ ∫ ∫ r dV i

(4⁄3)πR3

ri|r)R

(3)

where ki is the intrinsic kinetic constant, rmean is the observed rate of consumption/formation, and n is the reaction order. The numerical solution to the system of differential eqs 1 and 2 was obtained by discretizing the time and position coordinates and applying a three-level finite-difference scheme. Then, the concentration profile in the crystallite at any time, t, was determined and integrated over r to obtain the effectiveness factor from eq 3. Numerical values for various diffusion and kinetic parameters used in this simulation are presented in following sections. From a macroscopic standpoint, the CREC riser simulator reactor can be considered as a fluidized batch reactor. This is a reasonable consideration given that the zeolitic catalyst is exposed to a well-mixed fluid environment in the CREC riser simulator.11 As well, it can be assumed that the fluid dynamic conditions and catalyst particles lead to a situation where the layer surrounding the catalyst particles does not limit the overall reaction rate. Thus, combining eqs 1-3, the design equation for the catalytic cracking of species i in the well-mixed mini-fluidized CREC riser simulator can be represented by the following species balance equation V dCi,ex ηssri ) Wcr dt

(4)

where ri is the rate of production/consumption of species i per unit mass of zeolite. Because this reactor can be considered a well-mixed unit and because intercrystalline transport of reactant species is not the limiting step in this reactive system, Ci,ex can be assumed to be uniform throughout the reactor volume at any specific reaction time. Solving eqs 1 and 2 numerically shows that the accumulation term on the left-hand side of eq 1 is negligible for reactions with reaction times longer than 3 s. This consideration was verified to be valid by Al-Khattaf and de Lasa1 for reaction times greater than 2 s. Under these conditions, one can adopt the quasi-steady-state approximation and define a quasi-steadystate effectiveness factor, ηss, as ηss )

tanh(hi′) hi′

(5)

The effectiveness factor, ηss, requires the definition of a modified Thiele modulus, hi′, that depends on the zeolite crystallite size, geometry, and apparent density, as well as the reactant diffusivity and the intrinsic rate constant. Thus, the modified Thiele modulus is defined as hi′ )

1 aext



(n + 1) FcrφintkiCi,ex 2 Di,eff

where φint is the catalyst activity decay function and aext is the crystallite specific external surface area. It is important to note that aext in eq 6 can be defined using the characteristic dimension of the zeolite crystallite, R, because practically all of the cracking reactions take place in the zeolite crystallite with negligible influence of the matrix.14 Therefore, approximating the crystallite geometry to be equivalent to a sphere, so that aext ) 6/R, eq 6 becomes

n-1

(6)

R 6



3 Fcrφint(k1 + k2 + k3 + k4)Ci,ex 2 Di,eff

(7)

It is also expected that diffusion of gas oil in USY zeolite falls under the configurational regime, which is an activated process whose dependence on temperature can be represented by the Erying equation:15

[ ( )]

Di,eff ) Di,0 exp

-Ei,D 1 1 R T T0

(8)

where Di,0 is the pre-exponential factor for diffusion and Ei,D is the activation energy for diffusion. Although extensive work has been reported on diffusion in zeolites, there is uncertainty in the specific values of the effective diffusivities of hydrocarbons, because in most cases, these values are obtained by extrapolation.8 Moreover, Ruthven and Kaul16 recognized that the effective diffusivities are obtained from model calculations that exclude the presence of chemical reactions. Thus, it becomes imperative to assess the values of the effective diffusivities by introducing a model that combines the effect of hydrocarbon diffusion in the zeolite pore network under reaction conditions. By taking this approach, an estimation of both intrinsic kinetic and diffusion parameters can be accomplished. 3. Kinetic Modeling: Diffusion and Reaction Phenomena The development of kinetic models for the catalytic cracking of vacuum gas oil is essential in the prediction of the behavior of commercial FCC units. The complexity of kinetic models can vary depending on the phenomena being studied and, thus, the number of parameters being considered. To accurately depict the behavior of commercial FCC units, diffusion of hydrocarbon molecules inside the zeolite pore network, and reaction are considered in this present study. In developing a kinetic model, a lumping strategy can be considered to be an effective approach with various chemical species grouped according to their boiling point, because thousands of molecules are involved in the gas oil cracking reaction scheme. In this respect, a five-lump kinetic model consisting of gas oil, light cycle oil (LCO), gasoline, light gases, and coke was developed. The reaction scheme of the proposed model is described in Figure 2, and the lumps, which are classified on the basis of boiling point ranges, are reported in Table 1. The model does not consider secondary cracking reactions of products into coke, as it was found in a previous study by Al-Sabawi et al.17 that the kinetic constants for these reactions are orders of magnitude smaller than those for the primary reaction of gas oil into coke. Similar observations have been reported in the literature by Ancheyta-Juarez et al.18 and Oliveira and Biscaia.19 As discussed in the previous section, the catalytic cracking of vacuum gas oil in the mini-fluidized CREC riser simulator is an unsteady-state reaction process represented by eq 3. Thus, under quasi-steady-state conditions and upon application of eq 4, the disappearance of gas oil (A) in the well-mixed mini-

7634 Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008

CA )

yAWhc MWAV

(11)

where yA is the mass fraction of gas oil; Whc is the total mass of hydrocarbons injected into the riser simulator; and MWA is the molecular weight of gas oil, which is equal to 397 kg/kmol. Thus, the consumption of gas oil in the riser simulator can be evaluated using the equation Figure 2. Five-lump reaction network for the catalytic cracking of vacuum gas oil on FCC catalysts.

(

)

Whc V dyA ) ηssφint (k + k2 + k3 + k4)yA2 Wcr dt MWAV 1

Furthermore, each intrinsic kinetic constant, ki, can be postulated to change with the reactor temperature, T, following an Arrhenius-type equation

[ ( )]

Table 1. Boiling Point Ranges of Proposed Lumps lump

range of hydrocarbons

boiling point range (°C) at 1 atm

light gases gasoline light cycle oil gas oil

C20

342.7

ki ) ki0 exp

fluidized CREC riser simulator can be represented by the following species balance equation V dCA ) ηssrA Wcr dt

(9)

where rA is the rate of the consumption of vacuum gas oil in the absence of diffusion control and CA represents the concentration of gas oil in the gas phase. Catalytic cracking of gas oil is expected to be strongly affected by gas oil diffusion through the zeolite pore network, with the effectiveness factor term, ηss, being an important parameter. On the other hand, consumption of the light gases, gasoline, and LCO lump is not expected to be hindered by diffusion, and therefore, the effectiveness factors for these lumps are assumed to be equal to 1. This simplification was validated by our research group using model compounds in experiments performed under similar conditions.8,12,20 For instance, chemical species with molecular critical diameters in the range of gasoline (e.g., 1,2,4-TMB) and LCO (e.g., 1,3,5-TIPB) are not expected to display diffusional limitations under the temperatures, pressures, and gas-phase concentrations of this study. Following the lumping scheme of Figure 2, the rate of gas oil consumption can be expressed as -

V dCA ) ηssφint(k1 + k2 + k3 + k4)CA2 Wcr dt

(12)

(10)

where k1-k4 are the intrinsic rate constants considered in the reaction network of Figure 2. The second order assigned to gas oil cracking is a typical assumption given the different reactivities of the various lumps forming gas oil. This can be explained by the changing reactivity of the gas oil molecules, where, at low conversions, the most reactive molecules crack more readily and, as the conversion increases, the reactivity of the feed molecules decreases.21 On the other hand, reactions involving LCO and gasoline cracking are considered to be firstorder reactions, because the kinetic order of cracking single molecules is unity.22 Moreover, in the CREC riser simulator, which is a batch reactor unit of constant volume, the concentration of gas oil, CA, at any time can be related to the mass fraction, yA, by the equation

-Ei 1 1 R T T0

(13)

where Ei represents the energy of activation; ki0 is the preexponential factor; and T0 is the centering temperature defined as the average temperature used in the reaction experiments, which is 540 °C. It is well-known that the deposition of coke on the catalyst surface decreases the catalyst activity, because coke covers the active sites of the catalyst. The catalyst activity decay function, φint, can be used to relate catalyst activity to the coke concentration on the catalyst, as suggested by Froment and Bischoff.23 Thus, φint can be evaluated from the expression φint ) exp(-λXE′)

(14)

where λ is the deactivation parameter for gas oil cracking and XE′ represents the mass of coke produced per mass of gas oil injected. Similarly, species balances can be established for the other lumps, with these balances containing formation and/or consumption terms. The species balances for the various lumps are as follows: Light cycle oil (B): V dCB ) φint(νBηssk1CA2 - k5CB) Wcr dt

[ (

(15)

)( )

Whc MWB V dyB ) φint νBηss k y 2 - k5yB Wcr dt MWAV MWA 1 A

]

(16)

Gasoline (C):

(

νC V dCC ) φint νCηssk2CA2 + k5CB - k6CC Wcr dt νB

[ (

)( )

)

(17)

Whc MWC V dyC ) φint νCηss k y 2+ Wcr dt MWAV MWA 2 A νC MWC k y - k6yC (18) νB MWB 5 B

]

( )

Light gases (D):

(

) ( ) ]

νD V dCD ) φint νDηssk3CA2 + k6CC Wcr dt νC

[ (

)( )

(19)

νD MWD Whc MWD V dyD ) φint νDηss k3yA2 + ky Wcr dt MWAV MWA νC MWC 6 C (20) Coke (E):

Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 7635

dXE ) φint(νEηssMWE k4CA2) dt

[ (

(21)

)( ) ]

′ Whc MWE V dXE ) φint νEηss ky 2 Wcr dt MWAV MWA 4 A

(22)

In the above equations, νi represents the stoichiometric coefficient for lump i, and XE is the mass of coke produced per unit mass of crystallites. Note that ∑yi + XE′ ) 1. Note that the molecular weights of the product lumps are 46, 114, 226, and 800 kg/kmol for light gases, gasoline, LCO, and coke, respectively. 4. Experimentation Thermal and catalytic cracking experiments were performed in the CREC riser simulator using vacuum gas oil with a molecular weight of 397 g/mol. The analytical data for this feedstock are reported in Table 2. The CREC riser simulator is a laboratory-scale unit, invented by de Lasa.10 This simulator is an experimental reactor that operates isothermally and at constant volume of the reaction mixture and enables injected gas oil feedstock to vaporize and, in the case of catalytic cracking reactions, come into contact and mix with fluidized catalyst throughout a predetermined time span. Well-mixed conditions are assumed to exist in the reactor as a result of the high gas recirculation rate, as reported by Ginsburg et al.11 There is also limited thermal cracking, with 2-5 wt % conversions at the highest temperature level (570 °C). A schematic diagram of this 54-cm3 unit and its components can be seen in Figure 3. The main reactor consists of upper and lower shells that allow the catalyst to be loaded and unloaded easily into a basket. This basket, which is located in the lower shell of the reactor, is bound by two grids that trap the catalyst and restrict its mobility within the basket. Note that, for thermal cracking runs, this catalyst basket is left empty. An annular space is created between the inner walls of the bottom shell and the outer portion of the basket. This space allows for the recirculation of chemical species in the reactor by the rotation of an impeller positioned above the catalyst basket. A metallic gasket is used to provide a seal between the chambers in the upper and lower shells. A packing gland assembly with a cooling jacket supports and seals the impeller shaft. As the impeller rotates, it creates a low-pressure region just below its blades. As a result, gas introduced into the bottom Table 2. Properties of Vacuum Gas Oil Feedstock property

value

API total sulfur (wppm) total nitrogen (wppm) refractive index at 67 °C aniline point (°C) carbon residue (wt %) carbon-to-hydrogen ratio (wt/wt) viscosity (cSt at 100 °C) simulated distillation (°C) initial boiling point (IBP) 10 wt % 20 wt % 30 wt % 40 wt % 50 wt % 60 wt % 70 wt % 80 wt % 90 wt % final boiling point (FBP)

21.7 8700 1600 1.4924 79.4 0.6 84.7/12.4 7.32 211.9 320.3 354.9 382.4 408.0 428.9 449.7 473.7 501.6 536.6 632.7

Figure 3. Schematic diagram of the CREC riser simulator.

Figure 4. Typical pressure profile in the CREC riser simulator during gas oil thermal cracking and catalytic cracking over Y-zeolite catalyst at 500 °C, 7 s, and C/O ) 5.

shell of the reactor is induced to flow upward through the catalyst chamber. Thus, the impeller provides a fluidized bed of catalyst particles, as well as intense gas mixing inside the reactor. The cracking reaction mechanism employed by the riser simulator mimics the cracking process that takes place in largescale riser reactors by matching the relative pressures of the chemical species, the gas-solid contact regime temperature, and reaction time. The riser simulator operates in conjunction with a four-port valve, which enables the connection and isolation of the 54cm3 reactor and the 1120-cm3 vacuum box, and a six-port valve with a sampling loop, which allows samples of reaction products to be collected. Pressure transducers are installed in both the reactor and the vacuum box chambers to monitor the progress of each cracking experiment. A pressure profile in the CREC riser simulator for the thermal and catalytic cracking of vacuum gas oil is shown in Figure 4. This figure shows that, prior to the injection of the feedstock into the reactor, the pressure of the reactor is 14.7 psia (1 atm), whereas the vacuum box is maintained at very low pressure (∼1 psia). To maintain this difference in pressure, the reactor and vacuum box are isolated by closing the fourport valve. At the time of injection of the feedstock into the reactor, the reactant rapidly vaporizes, causing an abrupt increase in pressure (A-B). Another stage follows the reactant vaporization whereby the gaseous gas oil experiences cracking into different hydrocarbon products, causing an expansion in the system. As a result, a less profound pressure increase can be

7636 Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 Table 3. Propertiesa of the FCC Catalysts with Large and Small Crystal Sizes

zeolite contentb unit cell size BET surface area external surface area crystallite size crystallite density

large-crystallite CAT-LC

small-crystallite CAT-SC

29% 24.28 Å 197 m2/g 20 m2/g 0.9 µm 825 kg/m3

31% 24.28 Å 169 m2/g 25 m2/g 0.4 µm 825 kg/m3

a Properties reported for USY zeolites after being pelletized and exchanged with ammonium nitrate. b In terms of area (BET) for zeolites, matrix, and catalysts before steaming.

seen in the pressure profile (B-C). Once the preset reaction time is completed, the four-port valve is automatically switched to connect the reactor and the vacuum box. The initial large difference in pressure between these two chambers causes the evacuation of the reaction products from the reactor into the vacuum box. This evacuation, which occurs instantaneously because of the significant differences in pressure and volume between the reactor and vacuum box, leads to a sudden drop in the reactor pressure and a consequent rapid pressure stabilization in both chambers (C-D). Note that any further cracking of products present in the vacuum box is prevented because of the low temperature at which the vacuum box is held (340 °C). A six-port valve with a sampling loop is connected to the vacuum box. This allows for the collection of a sample of the reaction products present in the vacuum box. The collected sample of reaction products is then sent to an Agilent Technologies 6890N gas chromatography unit to be quantified. The gas chromatography-mass spectrometry (GC-MS) unit uses a flame ionization detector (FID) and a 25-m HP-1 capillary column of cross-linked methyl silicone with an outer diameter of 0.20 mm and an inner diameter of 0.33 µm. A specific temperature program is incorporated into the GC instrument that allows quantification of the reaction products coming from the sampling loop of the riser simulator setup. The column temperature was programmed to be at -30 °C for 3 min, to then increase at a rate of 15 °C/min to 235 °C for 1 min, and finally to increase at a rate of 40 °C/min to 320 °C for 4 min. The injector and detector temperatures during the experiments were set to 320 °C. The signals from the GC unit were quantified by Chemstation computer software, which generates and displays peaks of different areas and heights, with each peak representing a specific compound. Functions in the Chemstation software allow for the determination of the retention times, area counts, and area percentages per peak. The compound/peak identification process was carried out using an MS detector and the same capillary column and temperature program as used in the GC quantification process. This detector was used to identify hydrocarbon species in each product lump, including benzene and other aromatic compounds found in gasoline. Two standard FCC catalysts were used in the gas oil catalytic cracking experiments, characterized by Tonetto et al.14 These two catalysts were prepared in the same fashion and, thus, had almost identical properties and characteristics (reported in Table 3), with the exception of the Y-zeolite crystallite size. One of the catalysts had large 0.9-µm-diameter crystallites (referred to as CAT-LC in this study) and the other had small 0.4-µmdiameter crystallites (CAT-SC). With the only major difference between the two catalysts being the crystallite size, the quantitative evaluation of diffusional constraints could be carried out. The unit cell size was measured by X-ray diffraction following method ASTM D-3942-80, and the surface area was determined using the BET method.

Figure 5. Conversion of vacuum gas oil over CAT-LC at different residence times. Experiments were conducted using a catalyst-gas oil ratio of 5 at temperatures of (]) 510, (0) 530, (4) 550, and (O) 570 °C.17

Figure 6. Conversion of vacuum gas oil over CAT-SC at different residence times. Experiments were conducted using a catalyst/gas oil ratio of 5 at temperatures of (]) 510, (0) 530, (4) 550, and (O) 570 °C.17

Experiments involving the thermal and catalytic cracking of gas oil were carried out at four temperatures of 510, 530, 550, and 570 °C and three reaction times of 3, 5, and 7 s. The catalytic cracking experiments were performed using a consistent catalyst/gas oil ratio of 5 (weight of catalyst ) 0.81 g, weight of injected gas oil ) 0.162 g). 5. Results 5.1. Catalyst Activity Results: Gas Oil Conversion. Upon analysis of the gas oil conversion at typical catalytic cracking conditions using the two catalysts, CAT-LC and CAT-SC, it was observed that the overall gas oil conversion to the four product lumps increased with both temperature and reaction time, as seen in Figures 5 and 6. For instance, experiments conducted at 3 s using CAT-LC resulted in gas oil conversions of 27, 42, 50, and 56 wt % for temperatures of 510, 530, 550, and 570 °C, respectively. Similar trends were observed using CAT-SC at the same reaction time, with the gas oil conversions being 32, 47, 55, and 59.5 wt % for temperatures of 510, 530, 550, and 570 °C, respectively. Typical errors for the gas oil conversion reached (2.5 wt %. Examination of the gas oil conversion on USY catalysts indicates that the overall rate of conversion is dependent on pore diffusion resistance within the crystallite network. The gas oil conversions obtained from catalytic cracking runs using the two catalysts, CAT-LC and CAT-SC, were compared (Figure 7) to determine the influence of internal mass transfer on the calculation of the overall reaction rate. In this figure, experimental points that fall on or are close to the solid line indicate

Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 7637

Figure 7. Vacuum gas oil conversion obtained using CAT-LC versus CATSC. Experiments were conducted at temperatures of (]) 510, (0) 530, (4) 550, and (O) 570 °C using a catalyst/gas oil ratio of 5. The lowest point of each symbol refers to a reaction time of 3 s, and the middle and highest points refer to 5 and 7 s, respectively.

that CAT-LC and CAT-SC provided similar gas oil conversions, whereas points that are farther from the solid line indicate much higher conversions attained with CAT-SC. Differences in gas oil conversions between the two catalysts imply that conversion is a function of the crystallite size and, therefore, that the reaction rate is influenced by intracrystalline diffusion of reactant molecules. On the other hand, similarities in gas oil conversion indicate that the conversion is not significantly affected by the crystallite size, so that the reaction rate is less influenced by intracrystalline diffusion. In this respect, it can be noticed that the differences in gas oil conversion using CAT-LC and CAT-SC, which are initially large at lower temperatures and reaction times, become smaller at conditions that correspond to the values of higher conversions (i.e., higher temperatures and reaction times). For instance, the gas oil conversions obtained using CAT-LC and CAT-SC at 510 °C and 7 s were 44.5 and 55 wt %, respectively. At 570 °C and 7 s, however, CAT-LC and CAT-SC provided similar conversions of 72 and 73 wt %, respectively. The similarity in gas oil conversions obtained with the two catalysts under specific operating conditions (i.e., at higher temperatures of 570 °C) strongly suggests that catalytic cracking of gas oil is less influenced by crystallite size and that diffusion limitations of gas oil molecules in the zeolite pore structure do not play a significant role in influencing the overall catalytic cracking rate. On the other hand, at lower temperatures (i.e., at 510 °C), one can observe substantial differences between gas oil conversion using CAT-SC and CAT-LC, and these differences can be attributed to intracrystallite diffusional effects. It is under these conditions that the intracrystallite diffusion path, along which reactant molecules have to be transported before reaching an active site, becomes a critical parameter. 5.2. Catalyst Selectivity Results: Yields of Products. The distributions of the four product lumps (LCO, gasoline, light gases, and coke) were calculated using the 13 kinetic (Ei, ki0, λ) and 2 diffusion (D0 and ED) parameters reported by Al-Sabawi et al.17 and solving the set of eqs 12, 16, 18, 20, and 22. The values of the parameters calculated using data obtained with CAT-LC were as follows: (a) k10 ) 1.17 × 10-2 ((2.3%), k20 ) 4.76 × 10-3 ((2.8%), k30 ) 4.22 × 10-3

((3.3%), k40 ) 1.49 × 10-3 ((4.0%) [where k10-k40 are in units of m6/(kgcrystallite mol s)]; (b) k50 ) 5.45 × 10-2 ((10.3%), k60 ) 3.24 × 10-2 ((13.3%) [where k50 and k60 are in units of m3/(kgcrystallite s)]; (c) E10 ) 106.3 ((12.5%), E20 ) 108.0 ((5.5%), E30 ) 122.7 ((6.0%), E40 ) 1.49 × 10-3 ((5.1%), E50 ) 94.24 ((23.4%), E60 ) 86.24 ((32.4%) (where E10-E60 are in kJ/mol); (d) D0 ) 9.58 × 10-13 ((3.5%) m2/s, ED ) 237.7 ((2.7%) kJ/mol; and (e) λ ) 26.10 ((4.5%). These parameters were established using a two-step statistical nonlinear regression process, with degrees of freedom ranging between 100 and 109. It is important to note that these parameter values are very similar to those calculated using data obtained with CAT-SC. In fact, the differences in the calculated parameters for the two catalysts were determined to be statistically insignificant. This similarity is expected because the two catalysts have similar acidities and the same structural properties, as reported by Tonetto et al.14 The yield distributions of the light gases, gasoline, and coke lumps consistently increased with reaction time, whereas the LCO yield initially increased but leveled off with the progression of the reaction, until reaching a maximum, as shown by AlSabawi et al.17 This maximum level of LCO yield can be explained considering the competition between LCO formation reactions and the overcracking of LCO into gasoline. The product yields attained at low reaction temperature (510 °C) using CAT-LC and CAT-SC were different, especially for the intermediate products LCO and gasoline. For instance, at a reaction time of 7 s, CAT-LC and CAT-SC yielded 17 and 20 wt % LCO, 14 and 18.3 wt % gasoline, and 11 and 14 wt % light gases, respectively. It should be stressed that these observed differences are due to the gas oil being catalytically cracked at conditions under which diffusion limitations are significant (lower temperatures). As expected, a catalytic cracking process controlled by intracrystalline diffusion of reactant molecules is favored by the reduction of the crystallite size. As a result, the smaller-crystallite zeolite provides higher accessibility to its active sites with an increased capacity for the cracking of gas oil molecules, which leads to higher overall gas oil conversions. From the above-described results, it can also be observed that, not only does the smaller-crystallite zeolite (CAT-SC) provide larger overall VGO conversions than the large-crystallite zeolite (CAT-LC), but it also provides higher selectivity of desired products (i.e., gasoline) while reducing the selectivity of undesired product species (i.e., coke). For instance, at 510 °C and 7 s, the CAT-SC and CAT-LC catalysts displayed gasoline selectivelies of 28.7% and 25.2%, respectively. Differences in gasoline yields and selectivities between the two catalysts studied can be traced to the net effect of competing reactions: (a) gasoline formation reactions, cracking of both VGO and LCO, and (b) gasoline overcracking reactions. Gasoline formation reactions involve cracking of larger molecules, such gas oil and LCO, into smaller hydrocarbon molecules, with gas oil cracking being strongly influenced by intracrystalline transport at lower temperatures. On the other hand, gasoline overcracking reactions, which involve relatively smaller gasoline molecules, are not influenced by crystallite size. Thus, with smaller zeolite crystallites (CAT-SC), the gasoline formation reactions are increased, whereas the overcracking reactions remain essentially unaffected, with the net effect being higher gasoline yields and selectivities. In this respect, the gasoline yields obtained in this study are, in the context of their changes with crystallite size, consistent with data reported in the technical literature by Camblor et al.,24 Maselli and Peters,25 and Rajagopalan et al.26

7638 Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008

Figure 8. Amount of benzene in gasoline for a range of gas oil conversions obtained using (b) CAT-LC and (9) CAT-SC.

Figure 9. Total aromatics in gasoline for a range of gas oil conversions obtained using (b) CAT-LC and (9) CAT-SC.

The yields and selectivities of light gases are higher for CATSC than in CAT-LC. This difference can be attributed to a larger influence, in the case of CAT-SC, of gas oil cracking (primary reaction). Also, the CAT-SC catalyst produces more gasoline than the CAT-LC catalyst, thereby contributing to higher levels of light gases via gasoline overcracking. For instance, the selectivities of light gases obtained at 510 °C and 7 s using CAT-SC and CAT-LC were 24.9% and 21.3%, respectively. The coke selectivities obtained using the two types of catalysts varied considerably, with the selectivity obtained using CATSC being lower than that obtained using CAT-LC. For instance, the difference in coke selectivity at 7 s is 1.5%. This difference in the amounts of coke formed using the two catalysts can be assigned to the diffusional constraints that exist in the zeolite pore networks. Because the CAT-SC catalyst has shorter diffusional pathways from the active sites to the crystallite outer surface, there is less opportunity for coke condensation reactions to take place. On the other hand, in the CAT-LC catalyst, with longer diffusional pathways, there is an increased chance for the larger gas oil molecules to remain in the catalyst pore network to form coke. These trends are in agreement with previous literature data presented by Maselli and Peters25 and Rajagopalan et al.26 Thus, it is verified in the present study that crystallite size plays a very important role in FCC catalysts for both conversion and selectivity, especially at conditions that fall under the diffusion-controlled regime. Reducing the USY zeolite crystallite size decreases the diffusional path, which causes an increase in the vacuum gas oil cracking rate, as well as a reduction in the tendency of gasoline and other intermediate products to overcrack. This leads to the production of more desirable products, such as gasoline, while minimizing the production of catalyst deactivating coke and heavy hydrocarbon fractions, such as light cycle oil. 5.3. Catalyst Selectivity Results: Benzene and Aromatic Content in Gasoline. Recent regulations imposed on gasoline quality have caused major efforts by refineries to produce environmentally friendly gasoline, especially with reduced aromatic and benzene contents. These compounds are carcinogens that are harmful to both the environment and public health. Because the quantity of aromatics, specifically benzene, in the gasoline lump must be limited, it is essential to determine which of the two catalysts yields lower aromatic and benzene levels. Figure 8 compares the percentages of benzene present in the gasoline lump obtained using the two catalysts. It is apparent that, at any particular conversion, the amount of benzene formed using the CAT-SC catalyst is lower than the amount formed using the CAT-LC catalyst. For instance, at a gas oil conversion of 40%, the benzene contents in gasoline were found to be 2.63 and 2.85 wt % for CAT-SC and CAT-LC, respectively. At a

higher conversion of 70%, the benzene yields for CAT-SC and CAT-LC were 2.85 and 3.33 wt %, respectively. Thus, it can be noticed that the difference in benzene yield between the two catalysts is even more significant at higher overall conversions. This difference can be explained by the constraints that exist in the zeolite pore network, which cause diffusional limitations. As alkyl-aromatic compounds diffuse through the pores of the crystals, there is an increased likelihood that these chemical species will undergo overcracking reactions in the CAT-LC catalyst than in the CAT-SC catalyst because the crystallites of CAT-SC have shorter diffusional pathways. In other words, the smaller amount of time spent by alkyl-aromatic compounds in the catalyst pore network led to a lower likelihood that they would be overcracked. Thus, the production of benzene (a terminal product that is unable to undergo further cracking) through overcracking reactions was minimized. Figure 9 reports the amounts of total aromatics present in the gasoline lump obtained for the two catalysts. The trend is similar to that for benzene: At any particular conversion, the total amount of aromatics formed using the CAT-LC catalyst is higher than the amount formed using the CAT-SC catalyst. For example, at a gas oil conversion of 60%, the total aromatic content in gasoline was found to be 21.7 wt % for CAT-LC compared to 20.3 wt % for CAT-SC. To further assess the effects of zeolite crystallite size on gasoline quality, the total aromatics obtained using the two catalysts can also be compared with another family of compounds present in gasoline, such as olefins. This can be accomplished by calculating the olefin/aromatic (O/A) ratio. The significance of comparing these two families of compounds comes from the fact that most of the olefin species are formed via primary cracking reactions of paraffin molecules and that most aromatics are the products of olefins undergoing secondary cyclization reactions. In other words, olefins can be regarded as intermediate products formed from paraffins that can continue to react to produce aromatic species. Figure 10 summarizes the O/A ratios in gasoline obtained at various reaction conditions. These findings show that the O/A ratio increases as crystallite size decreases. For instance, for experiments conducted at 3 s, the O/A ratio consistently increased with temperature from 0.55 at 510 °C to 0.65 at 570 °C with the CAT-LC catalyst and from 0.62 at 510 °C to 0.71 at 570 °C with the CAT-SC catalyst. Similar trends in O/A ratios were obtained at reaction times of 5 and 7 s. These results show that using smaller zeolite crystallites enhances the formation of olefin species during the reaction and reduces the production of aromatic compounds. Thus, not only do the smaller crystallites minimize benzene formation, but they also reduce the overall aromatic content of gasoline. The explanation for the higher total aromatic content and lower O/A ratios obtained with larger crystallites is the potential

Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 7639

Figure 10. Olefin/aromatic ratio at various reaction temperatures for (9) CAT-LC and (b) CAT-SC. Residence time ) 3 s. Figure 12. Effectiveness factor versus modified Thiele modulus obtained for the catalytic conversion of gas oil using (9) CAT-LC and (0) CAT-SC at 530 °C. Note: The lower, middle, and upper points in each of the two sets correspond to experiments with reaction times of 3, 5, and 7 s, respectively.

Figure 11. Effectiveness factor versus modified Thiele modulus obtained for the catalytic conversion of gas oil using (9) CAT-LC and (0) CAT-SC at 510 °C. Note: The lower, middle, and upper points in each of the two sets correspond to experiments with reaction times of 3, 5, and 7 s, respectively.

diffusional limitations in the zeolite pore network. When a long hydrocarbon molecule in the gas oil feed, such as a paraffin, undergoes primary reactions involving the cleavage of carboncarbon bonds, radicals are produced. These radicals are very unstable and usually are stabilized when they link with free hydrogen atoms in the system. However, because of the scarcity of free hydrogen, the carbon radicals are forced to form double bonds with their neighboring carbon atoms. As a result, olefins are formed. Upon leaving the active sites in the catalyst, these olefins can undergo secondary reactions, such as cyclization, on other sites to form aromatic rings. This tendency is minimized, however, in catalysts with smaller crystallites because of the shorter diffusional paths that exist in their pore networks. Therefore, fewer aromatics are formed from olefins in CAT-SC than in CAT-LC, and this accounts for the lower total aromatic contents and higher O/A ratios obtained with CAT-SC. 6. Thiele Modulus and Effectiveness Factor The effectiveness factor, ηss, and the Thiele modulus, h′, for the catalytic conversion of gas oil can be evaluated using eqs 5 and 7, respectively, and the kinetic and diffusivity parameters adopted from Al-Sabawi et al.17 The Thiele modulus can be obtained using the calculated intrinsic kinetic and diffusion parameters. Although the Thiele modulus is clearly a function of the experimental temperature and the mole fraction of gas oil, it can also be noted that this parameter is a function of the intrinsic deactivation function, which is related to the mole fraction of coke produced as described in eq 22. Figures 11-14 show the changes in the effectiveness factor with the Thiele modulus obtained for the catalytic conversion of gas oil using

Figure 13. Effectiveness factor versus modified Thiele modulus obtained for the catalytic conversion of gas oil using (9) CAT-LC and (0) CAT-SC at 550 °C. Note: The lower, middle, and upper points in each of the two sets correspond to experiments with reaction times of 3, 5, and 7 s, respectively.

CAT-LC and CAT-SC at various reaction times and at temperatures of 510, 530, 550, and 570 °C, respectively. Analysis of Figure 11 shows low effectiveness factor values for both types of catalysts, especially at lower reaction times (3 and 5 s). For instance, the effectiveness factors for CAT-LC and CAT-SC are 0.418 and 0.759 at 3 s, 0.592 and 0.883 at 5 s, and 0.635 and 0.910 at 7 s, respectively. These values indicate that there is a major influence of intracrystalline diffusion on the catalytic cracking of gas oil at a reaction temperature of 510 °C. This observation is not as apparent in Figure 11, as diffusional constraints in the zeolite pore network become less significant with temperature. Thus, it is evident that gas oil cracking reactions at 510 °C are, in fact, in the diffusion-controlled regime. Furthermore, in Figures 12 and 13, it can be noticed that the effectiveness factors obtained at 3 s for CAT-LC are less than 1: 0.591 and 0.732 at 530 and 550 °C, respectively. Therefore, it can be concluded that the system is also in the diffusion-controlled regime at these two temperatures at a reaction time of 3 s, a finding that can also be confirmed by the difference in gas oil conversion in Figure 7 at these conditions. It is important to note that all other reaction conditions give rise to effectiveness factor values that are close to 1 and, thus, fall under the reaction-controlled regime. Analysis of the results also shows the key role played by the reaction time in the calculation of the effectiveness factor. It was found that the effectiveness factor increases with time. In Figure 12, for instance, the effectiveness factor obtained at 530 °C using CAT-LC was 0.591 at 3 s versus 0.833 at 7 s. This can be explained by the progressive catalyst deactivation that

7640 Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008

Nomenclature

Figure 14. Effectiveness factor versus modified Thiele modulus obtained for the catalytic conversion of gas oil using (9) CAT-LC and (0) CAT-SC at 570 °C. Note: The lower, middle, and upper points in each of the two sets correspond to experiments with reaction times of 3, 5, and 7 s, respectively.

occurs with time. As the deactivation of the catalyst increases, the intrinsic reaction rate decreases without affecting the intrapellet mass transfer. This causes a reduction in the Thiele modulus and a corresponding increase in the effectiveness factor. Figures 11-14 also show that the effectiveness factor evaluated for CAT-SC is higher than that evaluated for CATLC at each set of reaction conditions. These findings support the fact that diffusional constraints are not as evident in CATSC as they are in CAT-LC, because CAT-SC has zeolite crystallites that are smaller in size. Because there are less diffusional limitations in the zeolite pore network in the case of CAT-SC, higher gas oil conversions at low reaction temperatures are expected than when using CAT-LC, an observation that also supports the findings reported in Figure 7. 7. Conclusions The contributions of the present article can be summarized as follows: (1) Intracrystalline diffusion plays an important role in FCC catalysts; a rigorous model is required to describe this phenomena. (2) The overall gas oil cracking rate over USY zeolite-based catalyst is controlled by the highly temperature-sensitive intracrystalline gas oil transport in the 510-530 °C range. (3) The overall gas oil cracking rate is dominated by a mildly temperature-sensitive intrinsic cracking rate in the 550-570 °C range. (4) Reducing the zeolite crystallite size leads to the production of more desirable products, such as gasoline, while minimizing the production of catalyst-deactivating coke and heavy hydrocarbon fractions, such as light cycle oil. (5) Lower yields of aromatics, specifically benzene, in the gasoline fraction are obtained with the smaller zeolite crystallites because of the limited diffusional constraints that exist in the zeolite pore network, leading to more environmentally friendly gasoline. (6) The calculated modified Thiele modulus and effectiveness factor show the effects of crystallite size and temperature on the operating regime of the catalyst. Acknowledgment The authors are very appreciative of the Natural Sciences and Engineering Research Council of Canada (NSERC) for providing financial support through a Canada Graduate Scholarship (CGS) to M.A.-S. The authors also acknowledge the contribution of Imperial Oil Inc.

aext ) specific external surface area (m-1) Ci ) concentration of species/lump i in the vapor phase (mol/m3) Ci,ex ) concentration of species/lump i outside the USY crystallite (mol/m3) Ci,in ) concentration of species/lump i within the pore network of the USY crystallite (mol/m3) Di,eff ) effective diffusivity of species i (m2/s) Di ) intracrystalline diffusivity of species i (m2/s) Di,s ) surface diffusivity of species i (m2/s) Di,0 ) pre-exponential factor for diffusion of species i (m2/s) Ei,D ) activation energy for diffusion of species i (kJ/mol) Ei ) intrinsic activation energy for species/lump i reaction (kJ/ mol) h′ ) modified Thiele modulus ki ) intrinsic kinetic constant [m6/(kgcrystallite mol s) or m3/(kgcrystallite s)] ki0 ) pre-exponential factor for reaction [m6/(kgcrystallite mol s) or m3/(kgcrystallites)] Ki ) adsorption constant for species i [m3/(kg of catalyst)] L ) USY zeolite crystal size (µm) MWi ) molecular weight of lump i (kg/mol) mcat ) mass of catalyst (kg) mi,cat ) mass of species i adsorbed on the catalyst at the moment of complete vaporization (mol) n ) reaction order Pcatalytic ) total pressure at the moment of vaporization in a catalytic i run (psia) Pthermal ) total pressure at the moment of vaporization in a thermal i run (psia) pi ) partial pressure of species i (kPa) R ) radius of zeolite crystallite (m) or universal gas constant [J/(K mol)] r ) radial coordinate (m) ri ) kinetic rate of consumption/formation of species/lump i [mol/ (kgcrystallite s)] t ) reaction time (s) T ) reactor temperature (K) T0 ) average temperature of experiments (813.15 K) V ) volume of the riser simulator (m3) Wcr ) mass of catalyst (kg) Whc ) total mass of hydrocarbons inside the riser (kg) yi ) mass fraction of lump i in the vapor phase yi,pred ) mass fraction of lump i in the vapor phase predicted by kinetic model XE ) mass of coke produced per mass of crystallite (kgcoke/kgcrystallite) XE′ ) mass of coke produced per mass of gas oil injected (kgcoke/ kggas oil) Greek Letters ε ) USY crystallite porosity ∆Hi ) heat of adsorption of species i (kJ/mol) φint ) intrinsic catalyst activity decay function ηss ) effectiveness factor Fcr ) density of the USY zeolite crystallite (kg/m3) νi ) stoichiometric coefficient for lump i calculated as MWA/MWi λ ) deactivation parameter (of the Reactant Conversion model12) Subscripts A ) gas oil lump B ) light cycle oil lump C ) gasoline lump D ) light gases lump E ) coke 1,2,4-TMB ) 1,2,4-trimethylbenzene

Ind. Eng. Chem. Res., Vol. 47, No. 20, 2008 7641 1,3,5-TMB ) 1,3,5-trimethylbenzene AbbreViations CAT-LC ) catalyst prepared with large zeolite crystallites CAT-SC ) catalyst prepared with small zeolite crystallites CFL ) confidence limit C/O ) catalyst-to-oil ratio (kgcatalyst/kggas oil) LCO ) light cycle oil VGO ) vacuum gas oil

Literature Cited (1) Al-Khattaf, S.; de Lasa, H. The Role of Diffusion in Alkyl Benzenes Catalytic Cracking. Appl. Catal., A 2002, 226, 139–153. (2) Sadegbeigi, R. Fluid Catalytic Cracking: Design, Operation, and Troubleshooting of FCC Facilities; Gulf Publishing Company: Houston, TX, 2000. (3) Bhatia, S. Zeolite Catalysis: Principles and Applications; CRC Press: Boca Raton, FL, 1990. (4) Atias, J.; de Lasa, H. Adsorption and Catalytic Reaction over FCC Catalysts in a Novel Fluidized CREC Riser Simulator. Chem. Eng. Sci. 2004, 59, 5663–5669. (5) Do, D. D. Adsorption Analysis: Equilibria and Kinetics; Imperial College Press: London, U.K., 1998. (6) Hershkowitz, F.; Madiara, P. Simultaneous Measurements of Adsorption, Reaction and Coke Using a Pulsed Microbalance Reactor. Ind. Eng. Chem. Res. 1993, 32, 2969–2974. (7) Weisz, P. B. Molecular Diffusion in Microporous Materials: Formalisms and Mechanisms. Ind. Eng. Chem. Res. 1995, 34, 2692–2699. (8) Atias, J. A.; de Lasa, H. Adsorption, Diffusion and Reaction Phenomenon on FCC Catalysts in the CREC Riser Simulator. Ind. Eng. Chem. Res. 2004, 43, 4709–4720. (9) Martignoni, W.; de Lasa, H. Modelling Industrial Scale FCC Riser UnitssA Heterogeneous Model for Adsorption and Reaction Effects. In Fluidization VIII: Proceedings of the Eighth Engineering Foundation Conference on Fluidization, May 14-19, 1995, Tours, France; Large, J.F., Laguerie, C., Eds.; AIChE: New York, 1995; Vol. 1, pp 507-514. (10) de Lasa, H. I. Riser Simulator for Catalytic Cracking Studies. U.S. Patent 5,102,628, 1991. (11) Ginsburg, J. M.; Pekediz, A.; de Lasa, H. I. The CREC Fluidized Riser Simulator: Characterization of Mixing Patterns. Int. J. Chem. React. Eng. 2003, 1 (A-52), 1–12. (12) Al-Khattaf, S.; Atias, J. A.; Jarosch, K.; de Lasa, H. I. Diffusion and Catalytic Cracking of 1,3,5 Tri-iso-propyl-benzene in FCC Catalysts. Chem. Eng. Sci. 2002, 57, 4909–4920.

(13) Satterfield, C. N.; Sherwood, T. K. The Role of Diffusion in Catalysis; Addison-Wesley: Boston, 1963. (14) Tonetto, G.; Atias, J. A.; de Lasa, H. FCC Catalysts with Different Zeolite Crystallite Sizes: Acidity, Structural Properties and Reactivity. Appl. Catal., A 2004, 270, 9–25. (15) Ruthven, D. M. Principles of Adsorption and Adsorption Processes; John Wiley: New York, 1984. (16) Ruthven, D. M.; Kaul, B. K. Adsorption of n-Hexane and Intermediate Molecular Weight Aromatic Hydrocarbons on LaY Zeolite. Ind. Eng. Chem. Res. 1996, 35, 2060–2064. (17) Al-Sabawi, M.; Atias, J. A.; de Lasa, H. Kinetic Modeling of Catalytic Cracking of Gas Oil Feedstocks: Reaction and Diffusion Phenomena. Ind. Eng. Chem. Res. 2006, 45, 1583–1593. (18) Ancheyta-Juarez, J.; Lopez-Isunza, F.; Aguilar-Rodriguez, E. 5-Lump Kinetic Model for Gas Oil Catalytic Cracking. Appl. Catal. A: Gen. 1999, 177, 227–235. (19) Oliveira, L.; Biscaia, E. Catalytic Cracking Kinetic Models. Parameter Estimation and Model Evaluation. Ind. Eng. Chem. Res. 1989, 28, 264–271. (20) Atias, J. A.; Tonetto, G.; de Lasa, H. I. Catalytic Conversion of 1,2,4-Trimethylbenzene in a CREC Riser Simulator. A Heterogeneous Model with Adsorption and Reaction Phenomena. Ind. Eng. Chem. Res. 2003, 42, 4162–4173. (21) Hagelberg, P.; Eilos, I.; Hiltunen, J.; Lipia¨inene, K.; Niemi, V. M.; Aittamaa, J.; Krause, A. O. I. Kinetics of Catalytic Cracking with Short Contact Times. Appl. Catal., A 2002, 223, 73–84. (22) Van Landeghem, F.; Nevicato, D.; Pitault, I.; Forissier, M.; Turlier, P.; Derouin, C.; Bernard, J. R. Fluid Catalytic Cracking: Modelling of an Industrial Riser. Appl. Catal., A 1996, 138, 381–405. (23) Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Design, 2nd ed.; Wiley: New York, 1979. (24) Camblor, M. A.; Corma, A.; Martinez, F.; Mocholi, F. A.; Perez Pariente, J. Catalytic Cracking of Gas Oil. Benefits in Activity and Selectivity of Small Y-Zeolite Crystallites Stabilized by a Higher Siliconto-Aluminum Ratio by Synthesis. Appl. Catal., A 1989, 55, 65–74. (25) Maselli, J.; Peters, A. Preparation and Properties of Fluid Catalytic Catalysts for Residual Oil Conversion. Catal. ReV.-Sci. Eng. 1984, 26, 525–559. (26) Rajagopalan, K.; Peters, A. W.; Edwards, G. C. Influence of Zeolite Particle Size on Selectivity During Fluid Catalytic Cracking. Appl. Catal., A 1986, 23, 69–80.

ReceiVed for reView December 21, 2007 ReVised manuscript receiVed March 19, 2008 Accepted July 15, 2008 IE701745K