Modelling and Simulation of an Industrial

Fluid Catalytic Cracking Unit

SIMONA RADU*, DRAGOS CIUPARU

Petroleum and Gas University of Ploieºti, 39 Bucureºti Bvd., 100680, Ploieºti, Romania

To meet ever-changing market dynamics, industrial processes must adapt to new structures of the market
demand and environmental restrictions. Using the model provided by Aspen HYSYS and adjusting the kinetic
parameters, we aimed to simulate the functioning of an industrial plant as such to obtain results as close as
possible to the values observed during operation. The adjusted model allows optimization of the industrial
plant operation by extending the range of feeds or catalysts, and also by changing the plant geometry leading
to higher yields of desired reaction products, which allows maximization of plant profitability.
Keywords: industrial FCC unit, kinetic model, Aspen HYSYS, simulation

Fluid catalytic cracking (FCC) is a versatile and often and a certain catalyst [4, 7]. A 6-lumps model was proposed
profitable unit of any modern refinery, which is used to in 1998 [8], and a similar one few years later (2002) [9].
convert heavy distillates and residual raw materials in These simple models describing the kinetics of cracking
lighter and more valuable products. The FCC unit has the reactions of 3, 4, 5 and 6 specific chemical lumps have the
greatest contribution to the production of the gasoline pool, disadvantage that some of them can not forecast separately
but is also a significant provider of raw materials for the different specific catalytic cracking products.
petrochemical industry, such as light olefins. Thus, the FCC More complex models using a larger number of
unit has a major contribution to the overall profitability of components can also be found in the literature. The most
the refinery [1]. popular one is the model with 10 components proposed in
To meet the ever changing requirements of the market, 1976 [10]. An upgraded version of this model was proposed
the plant operators often have to make adjustments in the in 1987 [11]. This last model has 12 components, two
operation parameters of industrial plants in order to improve gaseous components in addition to the original Jacobi
yields and increase efficiency of the catalytic cracking unit. model.
Major modifications, however, must be taken into account There are even more complicated models, which have
only after a concerted and successful effort that involves advantages over previous models containing fewer
understanding the reactions chemistry, feed characteristics components, namely that only one group of kinetic
and performance of the equipment. In this respect, using a constants can be used for various feeds, and most of the
rigorous simulation models is of crucial importance [2]. reaction products can be estimated separately. As
Coupling hydrodynamic and kinetic models allow disadvantages, these models have a larger number of
calculating, more or less detailed, the performance of an kinetic parameters to be estimated and an increased
industrial plant. Modelling is therefore important to assess complexity of the possible numerical solutions, because
the main limiting phenomena and offer the opportunity to each component needs one more differential equation in
optimize reactor performance by adjusting reactor design the mathematical model [12].
and operating conditions so as to limit the steps to The most recent models published in the literature were
maximize the performance of the reactor. Independent on 8 lumps [13] and 9 lumps models [14], and a model with
the level of detail involved in the modeling, it is essential to 12 lumps [15]. However, besides the models published in
describe all relevant phenomena that must be accounted the literature, there are models that lay at the foundation of
for, which is not necessarily an easy task [3]. commercial software packages that simulate industrial
Regarding fluid catalytic cracking modeling, the processes, as the Aspen Engineering Suite. Aspen’s HYSYS
literature reports are divided into two categories, namely existing kinetic model for catalytic cracking is based on a
kinetic, and unit-level models, respectively. Kinetic models series of sub-models that can simulate the operation of
focus on chemical reactions taking place inside the riser the entire cracking unit. The thermal balance is satisfied
or reactor section of the FCC unit and attempt to quantify both for riser and regenerator. The model is complete and
the feed as a mixture of chemical entities to describe the includes the feed supply, the riser – reactor system, stripper,
rate of reaction from one chemical entity to another. In regenerator, valves for feed vaporization and cyclones [16].
contrast, unit-level models contain several sub-models to The simplified scheme is suitable for obtaining quantitative
take into account the integrated nature of a modern estimates of fractionated reaction products from the
catalytic cracking unit. A basic unit-level model contains reactor effluent. The model also includes several coke
sub-models for riser / reactor, regenerator and catalyst formation pathways and a distribution scheme for the feed
transfer sections [2]. sulfur and nitrogen into the reaction products.
There are many models in the literature that reproduce The riser conversion kinetics is derived from the Mobil
the kinetics of reactions occurring in the riser/reactor ten-lump mechanism [16]. The Aspen FCC model has
system of the catalytic cracking unit. Simple models, with expanded the number of reactant/product lumps to 21,
only few components, are suitable for specific simulations and changed the functionality of several key lumps. All
where the kinetic model is developed for a specific feed reactions are based on the well-understood first order

*email: simradu@yahoo.com

REV. CHIM. (Bucharest) ♦ 65 ♦ No. 1 ♦ 2014 http://www.revistadechimie.ro 113

 Fig.1. Schematic for 21 lump reaction
paths

Table 1
TWENTY-ONE LUMP MODEL [16]

kinetics and are considered to occur in vapour phase. The match the specific product compositions observed in the
kinetic expressions are integrated along the entire length industrial unit (provided that information is available),
of the riser, and are dependent on the catalyst bulk density, additional tuning of kinetic parameters for the reaction rates
the coke on catalyst, MAT activity, the basic nitrogen, and of paraffinic and aromatic hydrocarbons must also be
metals content. The MAT activity and the basic nitrogen performed.
are provided by external model sources and affect the riser This model divides the reactants and the reaction
kinetics uniformly. The catalyst bulk density and the coke products into lumped aggregates of material classified by
on catalyst are also integrated along the entire riser length, chemical nature and boiling point range. These lumps are
and are functions of pressure drop, coke formation, and similar to pseudo-components, but are also based on their
molar expansion. molecular structure in addition to the boiling range for
As mentioned above, reactor kinetics is based on the typical pseudo-component breakdowns. The molecular
21-lump kinetic model. The reaction pathways depicted in structures selected are based on likely reaction pathways
figure 1 represent paraffinic cracking, naphthenic ring and understood mechanisms known to exist in fluid
opening, alkyl side chain cracking, ring condensation, catalytic cracking chemistry. The table below summarizes
kinetic coke formation from typical condensation the lumps used in the Aspen model.
reactions, and metals coke formation due to The riser – regenerator system is presented in figure 2.
dehydrogenation. The reaction paths have been logically The riser – reactor system is the most important part of
grouped to make yield parameterization more convenient. the catalytic cracking unit as the cracking reactions take
Thus, all the pathways leading to gas formation constitute place in the riser. At the bottom of the riser, the gas oil feed
one class, the gasoline pathways form other class, and so comes in contact with the hot catalyst coming from the
on. Following this approach, with only a small number of regenerator and, consequently, vaporizes. As a result, the
yield measurements during unit operation, the kinetic rate cracking reactions start and the density of the feed
parameters for more than fifty reaction pathways can be decrease causing an increase in the velocity of the vapour
easily tuned to match the yields of the industrial unit. To phase. The increasing gas phase velocity accelerates the
velocity of the catalyst and the riser behaves as a transport
114 http://www.revistadechimie.ro REV. CHIM. (Bucharest) ♦ 65♦ No.1 ♦ 2014
 reactions proceed along the riser and in the reactor. The
coke produced in the stripper is determined from the
catalyst-to-oil ratio and the stripper performance curves.
The effluent from the riser splits into a solid part that
proceeds to the stripping zone and a vapor phase,
containing most of the reaction products and unreacted
feed molecules, which heads up towards the cyclones.
The solid part consists of the catalyst, the kinetic coke,
and a fraction of vapour hydrocarbons adsorbed on the
catalyst. Further cracking of the adsorbed hydrocarbons
on the catalyst surface occurs in the dense bed while some
of this material along with the stripping steam, proceeds
to the cyclones where it mixes with vapour effluents from
the riser.
Heat balances are performed at each mixing point of
the above coupled system consisting of cyclones, dilute
phase, and dense bed. These balances yield different
Fig. 2. Reactor-regeneration unit temperatures for each point in the system: riser outlet
(cyclone inlet), dense bed, and the reactor plenum
bed reactor. The coke by product of the cracking reactions chamber.
is deposited on the catalyst surface leading to a decreasing The reactor dense bed model is a differential-algebraic
activity as the catalyst moves toward the exit of the riser. model taking into account a single catalytic cracking
At the riser exit, the deactivated catalyst is separated from reaction for the low concentration of hydrocarbons
the vapour hydrocarbon products through specially adsorbed at the catalyst surface. It also performs the
designed riser termination devices and sent, through the pressure drop calculation along the height of the catalyst
striper, to the regenerator for burning off the coke from its bed, which can be set using pressure measurements in
surface and restore its catalytic activity. The vapour the industrial plant, or is to be specified directly in the model.
products are sent to the main fractionator for product In the latter case, the pressure drop is calculated.
separation and recovery [18]. The model for the reactor dilute phase (reactor plenum
The riser model consists of balance equations chamber) performs the mass balance, heat balance,
establishing the distribution of the reaction products, key pressure drop, reaction conversion and product distribution
point temperatures, pressure drop calculations, and coke calculations. The dilute phase model represents the
distribution. The riser model is a segment of the fluidized reaction volume that exists between the outlet of the riser
riser that models the kinetics in the riser and includes the and the inlet to the reactor cyclones [16].
geometry of the riser for hydraulic and volume effects. It Similar to the reactor model, the one for the regenerator
takes the hydrocarbon feed after the nozzle exit and consists of sub-models for the dense bed, the dilute phase,
combines it with the regenerated catalyst to take the and the cyclones. In each of these sub-models heat
material to the reactor. The chemistry in the risers is balance, material balance and pressure drop calculations
endothermic and uses the heat generated in the are performed. The regenerator dense bed kinetic model
regenerator for the chemical transformations. This process performs calculations for bubbling bed, heterogeneous
is tracked along the length of the riser and is observed in combustion of coke, and both for the homogeneous and
the temperature profiles. In these profiles, the temperature heterogeneous conversion of CO to CO2. The regenerator
of the hydrocarbon – catalyst mixture gradually drops from cyclone model performs a two-phase, loading-based
the entry zone to the riser exit into the reactor. These pressure drop calculation per cyclone. It completely returns
temperature drops are used in the models to determine the catalyst to the regenerator dense bed. This approach
catalyst flow rates [17]. sets up a recycle of the catalyst that can alter the steady-
The reactor model consists of three primary sub-models. state level of coke on the regenerated catalyst and the
As the hydrocarbon mixture enters the reactor vessel, a dense-bed temperature [16].
process of disengagement of the vapour hydrocarbons and As discussed above, the user has access to a number of
catalyst particles begins. Cyclone models are the final stage kinetic parameters of these models that can be adjusted
of this disengagement at the top of the reactor. Material as such that, by simulation, the model will reflect more
entering the cyclone models arrives there from the reactor accurately the behaviour of the industrial experimental unit.
free-board area. This area is represented by a model that To this end, in this contribution we aim at using the Aspen
sends material, primarily catalyst, to the dense bed of the HYSYS model and, by tuning the kinetic parameters of the
reactor. From there the material enters the stripping zone model, to obtain in simulations results as close as possible
where steam is used to remove as much of the remaining to those observed in an industrial experimental plant. The
hydrocarbons as possible from the catalyst before it enters resulting tuned model paves the way to optimize the
the spent catalyst transfer line [16]. functioning of the industrial plant with respect to product
Coke sources are separated into five distinct categories: yields and operating parameters under different market
kinetic coke, metals coke, coke resulting from the conditions.
Conradson carbon of the feed, coke produced by non-
vaporized feed, and coke formation controlled by the
stripping performance. The Conradson carbon coke and Experimental part
non-vaporized feed coke are assumed to be physical types In the FCC model available in the HYSYSTM module of
and are therefore deposited on the catalyst at the entrance the ASPEN Engineering Suite package, the feedstock
of the riser, prior to any cracking or coking reactions. Kinetic characteristics and the catalyst properties used in an
coke and metals coke are both determined from kinetic industrial plant were introduced along with the geometrical
expressions and are deposited on the catalyst gradually as characteristics of the equipment. The feed consists of a
mixture of vacuum distillate and coker gas oil having the
REV. CHIM. (Bucharest) ♦ 65 ♦ No. 1 ♦ 2014 http://www.revistadechimie.ro 115
 Table 2 Table 4
VACUUM DISTILLATE PROPERTIES PROPERTIES OF THE EQUILIBRIUM CATALYST USED IN THE
EXPERIMENTAL INDUSTRIAL UNIT

temperature and the gasoline, light cycled oil (LCO) and

coke product yields to test the model adequacy to the
industrial plant performance.
Using the values of the operating parameters collected
from the experimental industrial unit during test data
Table 3 acquisition, simulations were run initially with the intrinsic
COKER GAS OIL PROPERTIES values of the kinetic parameters of the model. In order to
assess the adequacy of the model using the intrinsic values
of the kinetic parameters, the values obtained by simulation
for the yields of the main reaction products and for the
regenerator temperature were plot in figures 3 - 6 against
the experimental values collected during test data
acquisition from the experimental industrial unit.
As shown in figures 3 to 6, simulation of the catalytic
cracking unit using the intrinsic values of the kinetic
parameters of the model leads to significantly different
results compared to those obtained experimentally in the
industrial plant. This indicates the need for model
adjustment prior to obtaining an accurate simulation model
that would be able to reproduce the functioning of the
experimental industrial unit.
Results and discussions
Using the intrinsic values of model parameters, the
deviation of the yields obtained by simulation were
significant compared to the actual values obtained
experimentally in the industrial plan. The most significant
deviations were observed for the LCO and coke yields. In
order to quantify the deviation of results obtained in
simulation from their experimental values the comparison
was made using as a benchmark the model accuracy
percentage error, E, and root mean square deviation, σ,
characteristics presented in tables 2 and 3, while the calculated with the following relations:
properties of the equilibrium commercial catalyst used in
the industrial unit are provided in table 4.
Subsequently, several sets of test data were recorded
from the experimental industrial plant – in operation within
a Romanian refinery – together with the operating
parameters used during the test data collection. Data sets
were collected from the industrial catalytic cracking plant
for a period of 7 months using the same catalyst and the
same feedstock. The test data sets collected cover feed
characteristics, riser and regenerator temperatures, feed Using the accuracy error, for the LCO yields calculated
flow rate, riser and regenerator pressures, as well as main with the intrinsic values of the kinetic parameters of the
reaction products’ yields (gasoline, diesel, coke, propylene, model were as high as 35 %, with most of the experimental
and gas). yields higher than those obtained in simulation. At the same
Simulating the industrial plant using the actual time, the regenerator temperature values resulting from
geometrical characteristics, feed stock properties in Tables the simulation were 15 to 20° C higher than those observed
2 and 3, catalyst characteristics in table 4 and the working experimentally. Because also the gasoline experimental
riser temperature, the stripping steam pressure and yields have higher values than the calculated ones we
temperature, regenerator bulk density, and the pressure conclude that a faster conversion of LCO to gasoline and
regime, the model provides estimated values for different of gasoline to gas is needed in order to improve the model
parameters, among which we have chosen the regenerator adequacy.

116 http://www.revistadechimie.ro REV. CHIM. (Bucharest) ♦ 65♦ No.1 ♦ 2014

 Fig. 3. Initial model response compared with real data on Fig. 5. Initial model response compared with real data on LCO
regenerator temperature

Fig. 6. Initial model response compared with real data on coke

Fig. 4. Initial model response compared with real data
on gasoline yield
5.48 to 0.98%. Similarly, the LCO errors dropped from 13.47
The user has a number of kinetic parameters of the to 1.64 %.
model that could be adjusted so that the model can give However, the Fraction Concarbon to Coke parameter is
estimates that faithfully reflect the behaviour of the also important as a lower value causes an increase of the
industrial plant. Because of the observations above we LCO and coke yield errors, but a better correlation for the
chose to modify several kinetic parameters. Their values gasoline yield and the regenerator temperature. The
were modified in successive steps and the simulation regenerator temperature is correlated with the amount of
results were compared with test data from the industrial coke formed. Therefore, by reducing the regenerator
experimental unit. temperature will eventually result in a lower coke yield,
We have started by adjusting the Basic Nitrogen while maintaining a constant temperature in the reactor.
Deactivation for G Pathways, Pre-exponential Factor for Also, changing the flow of dilute phase (Regenerator Dilute
Gasoline Cracking, Ea/R for Gasoline Cracking, Distillate Phase Flow Parameter) leads to values closer to those
Naphthenic Cracking to G Lump, Distillate Aromatic observed in the experimental industrial unit.
Sidechain Cracking to G Lump, and the Distillate Aromatic Regarding the LCO, it is known that getting a smaller
Ring Cracking to G Lump. Adjusting the parameters of the yield of LCO is always associated with increased coke
Distillate Cracking to C lump group, leads to changes in and gas yields. Thus, changing the values of the C group
regenerator temperature and gasoline yields. Even if the parameters, which are responsible for the formation of
errors on the regenerator temperature slightly increases, gases, adjusts the LCO yield values closer to the actual
this adjustment has a positive effect as it decreases the values obtained in the real catalytic cracking plant. The
gasoline yield error from 4.77 to 0.21 %, which translates parameters in the C group are the Conversion of H2 to C2
to a total error reduction by 95.6 %. Yields to Liquid Volume Basis, Distillate Paraffin Cracking
For the decrease of the coke yield error the Basic to C Lump, Distillate Naphthenic Cracking to C Lump,
Nitrogen Deactivation for Coking Pathways, Fraction Distillate Aromatic Sidechain Cracking to C Lump, and
Concarbon to Coke, CRC Response Intercept, CRC Light Gas Delumping to Hydrogen. Distillate paraffin
Response Slope, Global Catalyst Deactivation Factor, and cracking to C lump refers to the natural log of the pre-
the Catalyst Activity / Coking Factor parameters were exponential factor for the first order cracking of light paraffin
further adjusted. The CRC Response Intercept and the CRC (L-P) to light gases (C-lump). Distillate naphthenic cracking
Response Slope are two parameters that tune simulation to C lump represents the natural log of the pre-exponential
responses to the presence of coke on the catalyst, primarily factor for the first order cracking of light naphthenic (L-N)
for partial burning units, while the Global Catalyst to light gases (C-lump). Distillate aromatic sidechain
Deactivation Factor, and the Catalyst Activity / Coking Factor cracking to C lump is a parameter representing the natural
parameters in the MAT Tuning category adjust the effects log of the pre-exponential factor for the first order cracking
of the equilibrium catalyst activity in simulation influencing of light aromatic sidechains (L-As) to light gases (C-lump).
yields and make-up rate [17]. Following several trials, by Finally, the C lump is used to calculate the light gases for
modifying the CRC Response and MAT Tuning group methane through the pentanes, and is based on a
parameters, we succeeded to obtain much smaller values correlation using the C lump produced in the kinetic paths
of the coke errors, the final values decreasing from an initial and the composition of the feed. Within the kinetic system,

REV. CHIM. (Bucharest) ♦ 65 ♦ No. 1 ♦ 2014 http://www.revistadechimie.ro 117

 Table 5
DIFFERENCES BETWEEN THE INITIAL AND
THE FINAL VALUES OF KINETIC
PARAMETERS OF THE MODEL

Fig. 9. Theoretical and experimental values correlation

Fig. 7. Theoretical and experimental values correlation for the for gasoline
regenerator temperature

Fig. 10. Theoretical and experimental values correlation for

bottoms
Fig. 8. Theoretical and experimental values correlation for LCO
values of the kinetic parameters in table 5 were considered
satisfactory, as simulating the functioning of the industrial
the C lump component is divided directly into ten individual unit using the conditions of the test data sets, the results
light-gas components and coke [17]. predicted by the model accurately reflected the behaviour
The adjusted parameter values with errors less than 5 % of industrial plant within a maximum error value of ± 4.88
and the smallest mean square deviations were chosen for %.
the tuned model. Following successive attempts, the
118 http://www.revistadechimie.ro REV. CHIM. (Bucharest) ♦ 65♦ No.1 ♦ 2014
 Fig. 12. Theoretical and experimental values correlation for diesel
Fig. 11. Theoretical and experimental values correlation for coke
and bottoms
To illustrate the accuracy of the tuned model, values of
the main variables of the process predicted by the model functioning in a Romanian refinery. The tuned model has
were plotted against the values of the same variables been proven convergent for a wide range of operating
measured experimentally under the same conditions in parameters, and accurate within 4.88 % of overall error for
the industrial plant. Figures 7-11 show the regenerator the investigated range of parameters for which we had
temperature and yields of gasoline, diesel and coke, experimental values for comparison. The adjusted model
respectively, as obtained from simulation and can be further used to optimize the operation of the
experimentally. industrial plant.
Simulations were performed using the model that gave
the best response for gasoline, LCO and coke yields, and References
for the regenerator temperature, respectively. The model 1. SHISHKOVA1, I., STRATIEV, D., PECHLIVANOV, D., 45th International
gave responses that remained in the parameters of Petroleum Conference, June 13, Bratislava, Slovak Republic, 2011.
minimum deviation for a longer period of time. Comparing 2. PASHIKANTI, K., LIU, Y. A., Energy Fuels, 25, 5298-5319, 2011.
simulation results obtained using this model with 3. GAUTHIER, T.A., International Journal of Chemical Reactor
experimental results from the experimental industrial plant Engineering, Vol. 7, Article A22, 2009.
we obtained the results in figures 7-11. 4. WEEKMAN JR., V.W, I&EC Process Design and Development, Vol. 7,
A good correlation is observed with respect to the No. 1, pp. 90-95, 1968.
regenerator temperature and the yield of gasoline and coke. 5. BLASETTI, A., DE LASA, H., Industrial & Engineering Chemistry
However, the LCO yield does not show a similar correlation. Research, Vol. 36, Issue 8, 3223-3229, 1997.
A possible explanation is that, in the data sets obtained 6. PITAULT, I., NEVICATO, D. FORESSIER, M., BERNARD, J-R. , Chem.
from the refinery, there were not available information with Eng. Sci, vol. 49, no. 24A, 4249-4262, 1994.
respect to the amount of heavier distillates, while the 7. CERQUEIRA, H.S., BISCAIA JR., E.C., SOUSA-AGUIAR, E.F., Applied
simulation software provides data only in terms of LCO. Catalysis A: General, Vol. 164, Issues 1–2, 35-45, 23 December 1997.
Taking into account that for the feed we only had daily 8. ROªCA, P., Teza de doctorat, Ploieºti, 1998.
average compositions (constant ratio of vacuum distillate 9. ANCHEYTA, J., SOTELO, R, Rev. de la Sociedad Quimica de Mexico,
to coker gas oil is not always constant) and considering vol.46, 1, 38-42, 2002.
that heavy oil is included in the slurry residue, we considered 10. JACOB, S. M.; GROSS, B.; VOLTZ, S. E.; WEEKMAN, V. W., JR. A,
more relevant a graphical representation of the sum of AIChE J, 22, 701, 1976.
diesel and bottoms yields resulting from simulation versus 11. OLIVEIRA, L.L., Disertacao de Mastrado, COPPE/UFRJ, Rio de
the real data collected from the industrial unit. As Janeiro, RJ, Brasil, 1987.
expected, the data correlates significantly better, as it can 12. SOUZA, J.A., VARGAS J.V.C., VON MEIEN, O.F., MARTIGONI, W.P.,
be observed in figure 12. The calculated error is 4.36 %. Engenharia Termica (Thermal Engineering), vol.6, no.1, 19-25, June
Using this model that accurately reflects the industrial 2007.
plant operation it is possible to simulate different scenarios 13. HAGELBERG, P., EILOS, I., HILTUNEN, J., LIPIÄINEN, K., NIEMI,
for industrial plant operation. A first attempt was to V.M.,. AITTAMAA, J., KRAUSE, A.O.I., Applied Catalysis A: General,
investigate the model response to changes in the feed flow 223, 73–84, 2002.
rate and the reactor temperature compared to the real FCC 14. HONGJUN, Y., CHUNMING, X., JINSEN, G., ZHICHANG, L., PINXIANG,
unit data. Yields for the major reaction products, such as Y., Catalysis Communications, 7, 554–558, 2006.
gasoline, diesel and coke were estimated through 15. WU, F., WEN, G H., LUO, S., Acta Petrolei Sinica (Petroleum
simulation for temperatures between 520 and 540 °C, and Processing Section), V24 (5), 540-540, 2008.
several feed flow rates. 16. *** Aspen FCC 2006.5, User’s guide, Aspen Technology, Burlington,
MA, SUA, 2007.
Conclusions 17. *** Aspen HYSYS Refining CatCracker, Operations Guide, Version
We have successfully adjusted the intrinsic Aspen Number: V7.2, Aspen Technology, Inc., Burlington, USA, July 2010.
HYSYS fluid catalytic cracking model to accurately reflect 18. GUPTA, R.K., KUMARY, V., SRIVASTAVAZ, V.K., International Journal
the performance of an industrial catalytic cracking unit of Chemical Reactor Engineering, Volume 8, Review R6, 2010.

Manuscript received: 23.04.2013

REV. CHIM. (Bucharest) ♦ 65 ♦ No. 1 ♦ 2014 http://www.revistadechimie.ro 119