Вы находитесь на странице: 1из 10

Please cite this article in press as: Lopes, G. C., et al.

Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.


Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
Computers and Chemical Engineering xxx (2011) xxxxxx
Contents lists available at ScienceDirect
Computers and Chemical Engineering
j our nal homepage: www. el sevi er . com/ l ocat e/ compchemeng
Three-dimensional modeling of uid catalytic cracking industrial riser
ow and reactions
G.C. Lopes
a
, L.M. Rosa
a
, M. Mori
a,
, J.R. Nunhez
a
, W.P. Martignoni
b
a
University of Campinas, School of Chemical Engineering, 500 Albert Einstein Ave, 13083-970 Campinas, SP, Brazil
b
PETROBRAS/AB-RE/TR/OT, 65 Repblica do Chile Ave, 20031-912 Rio de Janeiro, RJ, Brazil
a r t i c l e i n f o
Article history:
Received 17 May 2010
Received in revised form 18 October 2010
Accepted 30 December 2010
Available online xxx
Keywords:
Numerical simulation
Gassolids ow
Nonuniformity
FCC
Riser
Cracking reactions
a b s t r a c t
Athree-dimensional and two-phase owmodel to predict the dynamic behavior of a uid catalytic crack-
ing (FCC) industrial reactor was developed in this work. The study took into account heat transfer and
chemical reactions. A four-lump model was proposed to represent the catalytic cracking reactions in
which the heavy oil (gas oil) is converted into gasoline and light hydrocarbon gases. Gas acceleration
inside the reactor due to molar expansion and a model to describe undesirable catalyst deactivation
by coke deposition on its surface were also considered. An Eulerian description of the phases was used
to represent the two-phase ow. A commercial CFD code (Ansys CFX version 11.0) was used to obtain
the numerical data. Appropriate functions were implemented inside the CFX code to model the het-
erogeneous kinetics and catalyst deactivation. Results show nonuniform tendencies inside the reactor,
emphasizing the importance of using three-dimensional models in FCC process predictions.
2011 Elsevier Ltd. All rights reserved.
1. Introduction
FCC units are used in most reneries worldwide, since this pro-
cess converts high molecular weight gas oils or residuum charge
stocks into lighter hydrocarbon products inside a riser reactor in
a few seconds. In an attempt to improve this protable opera-
tion, studies on uid catalytic cracking processes have increased.
Two main lines of research are found in studies about FCC. One of
them takes into account chemical reactions and related phenom-
ena, while not considering some important uid dynamics aspects.
The other set of studies aims to understand the gassolid hydro-
dynamic behavior inside these reactors without taking account the
reactions involved.
Many complex reactions occur inside the reactor during the FCC
process. Toprovide approximations tothe stoichiometry andkinet-
ics of these reactions, a widely used technique is to describe the
complex mixtures of hydrocarbons by aggregating large numbers
of chemical compounds into a unique compound(lump) represent-
ing the whole set of components. This produces a small number of
representative pseudo-components. Many studies have focused on
obtaining estimates for the kinetic parameters of lump models. The
studies conductedby Blasetti andde Lasa (1997), Farag, Blasetti and
de Lasa (1994), Jurez, Isunza, Rodrguez and Mayorga (1997), Lee,

Corresponding author. Tel.: +55 19 3521 3963; fax: +55 19 3521 3910.
E-mail address: mori@feq.unicamp.br (M. Mori).
Chen, Huang, Pan (1989) and Pitault, Forissier and Bernard (1995)
used a four-lump model to represent the cracking of hydrocarbons.
In this work the four-lump model was also used.
Many studies involving simulations of the FCC processes and
focusing on reactor optimization have recently been developed.
The research of Ali H, Rohani S, Corriou (1997), Han and Chung
(2001a,b), Nayak, Joshi and Ranade (2005) and Ahari, Farshi and
Forsat (2008), for example, simulated the dynamic behavior of
the riser, considering the reactions and associated phenomena as
they were occurring in plug-ow reactors. These predictions used
simple one-dimensional mass, energy and chemical species bal-
ances. Although the modeling is simple, these studies obtained
good results for the gas oil conversion and product yields at the
reactor outlet and some of them could reproduce some hydrody-
namic axial aspects.
Experimental studies have also shown that when a vertical gas
riser transports solid particles, they are distributed nonuniformly
over the cross section. Incirculating uidizedbeds, for example, the
ow has a dilute central solids region with high velocities for both
uid and solids and a high solids concentration near the walls. This
type of radial ow is dened as core-annulus. Furthermore, in FCC
operations the bed is axially divided into dense and dilute regions,
since cracking reactions expand the gas phase and accelerate the
solid.
Zhang, Tung andJohnsson(1991) conductedexperimental stud-
ies andshowedthat theradial solidvolumefractionproledepends
only on the value of the cross-section-average-volume fraction,
0098-1354/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compchemeng.2010.12.014
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
2 G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx
Nomenclature
C
i
molar concentration of component i [kmol m
3
]
C
d
drag coefcient [-]
C
C
constant of elasticity modulus function [Pa]
C

constant, 0.09
C
c,1
constant, 1.44
C
c,2
constant, 1.92
d particle diameter [m]
activation energy [J mol
1
]
g gravitational acceleration [m
2
s
1
]
C elasticity modulus [Pa]
H static enthalpy [J mol
1
]
k kinetic constant of reaction [m
3
kmol
1
s
1
] or tur-
bulent kinetic energy [m
2
s
2
]
k
0
pre-exponential factor [m
3
kmol
1
s
1
]
K
c
deactivation constant [kg
cat
kmol
1
]
Nu Nusselt number [-]
p static pressure [Pa]
P
k
shear production of turbulence [Pa s
1
]
Pr Prandtl number [-]
q
1
specic coke concentration [kmol kg
1
cat
]
R reaction rate [kmol m
3
s
1
] or universal gas con-
stant [J mol
1
K
1
]
Rc Reynolds number [-]
1 static temperature [K]
u velocity vector [ms
1
]
Greek letters
interphase momentum transfer [kgm
3
s
1
]
volume fraction [-]
c turbulence dissipation rate [m
2
s
3
]
catalyst decay function [-]
interphase heat transfer coefcient [Wm
2
K
1
]
I diffusivity [kgm
1
s
1
]
z thermal conductivity [Wm
1
K
1
]
molecular viscosity [Pa s]
density [kgm
3
]
o
k
constant, 1.00
o
c
constant, 1.30
Subscripts
g gas phase
lum laminar
R reaction
s solid phase
turb turbulent
irrespective of operational conditions, solids properties and bed
diameters, and developed a correlation to estimate this variable.
However the experiments were performed at low velocities and
solids ux, which are not within the range used in industrial appli-
cations.
The solids ow direction in the wall region is often reported
to be downwards in low-ux risers operating under uxes of less
than 200kgm
2
s
1
and supercial gas velocities up to 6.5m]s.
However Prssinen and Zhu (2001) observed that for high-ux
applications, the ow direction is normally upwards. This makes
the data obtained under low-ux conditions less useful. They also
noticed that under the low-ux (100kgm
2
s
1
) condition, the
radial distribution of particle velocities is more uniform and less
sensitive to changes in axial position than that under a high-
ux (300kgm
2
s
1
) condition at the same gas velocity. They also
observed that with the increase in gas velocity, the ow develop-
ment became faster.
Most of the experimental studies on riser hydrodynamics were
developedfor pilot plants withdiameters nolarger than300mm. In
order to determine the inuence of reactor scale on the gassolids
owhydrodynamics, Landeghemet al. (1996) studied the FCC pro-
cess for two different scales: they performed a cold setup in a pilot
plant and obtained several samples in industrial plants. Although
they did not report the catalyst concentration proles measured in
the commercial plants, Landeghemet al. (1996) observed that they
where similar to those obtained in the pilot plant. This indicates
that the core-annulus structure also exists in large beds (diameter
ca. 1 m).
The studies cited above are considerably complete in terms of
gassolids ows. However, their description of the hydrodynamic
characteristics of the owis not suitable for modeling an FCC riser,
since cracking reactions are not considered.
An important contribution to the study of the hydrodynamic
effects in the FCC process was developed by Deng, Wei, Liu and
Jin (2002). They simulated a commercial plant using three differ-
ent models: plug ow, one-dimensional and two-dimensional, and
compared these results with the data obtained from a commer-
cial riser at Shengli Petrochemical Company. They observed that
the two-dimensional model provideda better approximationof the
results for gas oil conversion and the yields for the species to the
industrial data, with an error of about 1wt.%. On the other hand,
when the plug ow model was used, the conversion of gas oil and
the gasoline yield were 13 and 16wt.% higher, respectively, show-
ing that the considerationof radial nonuniformity is necessary. This
difference justies further research.
This work extends these contributions to a three-dimensional
model which takes into account radial dispersion. This is an impor-
tant consideration in FCC reactor studies, since a shorter and more
uniform catalyst distribution in the riser reactor could potentially
produce a better reaction performance, depending on the interac-
tion between the phases. Therefore it is fundamental to know the
hydrodynamic behavior of this process.
Computational uid dynamics tools have been widely used in
chemical process engineering to predict and analyze the dynamic
behavior of gassolids ows. Since a typical FCC unit converts large
amounts of feedstock into valuable products, such studies can opti-
mize equipment performance withconsiderable economic benets
to a renery.
A mathematical model to simulate the gas-particle turbulent
ow in a three-dimensional riser reactor is presented in this
work. The uid dynamic two-uid model is further extended to
predict hydrocarbon formation and its consumption according
to the four-lump catalytic cracking kinetics, considering catalyst
deactivation.
2. Mathematical model
The transient two-uid model used in this work considers a
three-dimensional gassolids ow, including heat transfer and
chemical reactions. It is assumed that the feed enters the riser
vaporized; then the gas velocity is increased due to the molar
expansion caused by cracking reactions. The model uid dynamic
equations presented in this section were obtained from the ANSYS
CFX-Solver Theory Guide (2009) and the catalytic cracking kinetic
model was taken from the work of Farag et al. (1994).
2.1. Gassolids ow model
The two-phase ow model uses an Eulerian description of both
phases, so the gases and the dispersed solid particles are treated as
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx 3
an interpenetrating continuum. The governing conservation equa-
tions are the following:
2.1.1. Continuity equations
The gas and solid phase continuity equations are given respec-
tively by

t
(
g

g
) + (
g

g
u
g
) = 0 (1)

t
(
s

s
) + (
s

s
u
s
) = 0 (2)
where stands for the volume fraction, for the density and u
for the velocity of eachphase. The subscripts g ands indicate the gas
andparticulate phases, respectively. No mass transfer is considered
to occur between the phases.
2.1.2. Momentum equations
The gas and solid phase momentumequation may be expressed
as

t
(
g

g
u
g
) + (
g

g
u
g
u
g
) =
_

g
_
u
g
+(u
g
)
1
_
+
g

g
g
g
p +(u
s
u
g
) (3)

t
(
s

s
u
s
) + (
s

s
u
s
u
s
) =
_

s
_
u
s
+(u
s
)
1
_
+
s

s
g
s
C
s
+(u
g
u
s
) (4)
where is the viscosity, g the acceleration of gravity, p the pres-
sure, C the modulus of elasticity and the interphase momentum
transfer, which in this work is modeled using the Gidaspow drag
model, which combines the Wen Yu correlation with the Ergun
equation. For dense regions (
s
: 0.2), is given by
= 150

2
s

g
d
2
s
+
7
4
|u
s
u
g
|
s

g
d
s
(5)
and for dilute regions (
s
= 0.2), it is given by
=
3
4
C
d
|u
s
u
g
|
s

g
d
s
(6)
where d
s
is the particle diameter and C
d
is the drag coefcient.
For solidparticles, at large Reynolds numbers (Rc : 1000, sufcient
for inertial effects to dominate viscous effects) the drag coefcient
is independent of the Reynolds number:
C
d
=
0.44

1.65
s
(7)
When the Reynolds number is less than 1000, both viscous and
inertial effects are important. Hence, the drag coefcient is deter-
mined from an experimental correlation:
C
d
=
1

1.65
s
24
Rc
(1 +0.15Rc
0.687
) (8)
The modulus of elasticity (C) that is used to predict the solids
pressure is given by Gidaspow (1994):
C = exp[C
C
(
s

s,mux
)] (9)
where
s,mux
is dened as the maximumvolume fraction for the
granular phase. For monodispersed spheres, the packing limit is
about 0.63.
2.1.3. Turbulence equations
The gas-phase effective viscosity was predicted as a sum of the
molecular viscosity and an eddy contribution:

g
=
lum,g
+
turb,g
(10)
In the present work the turbulent viscosity was modeled using
the k-epsilon model. In this model the eddy viscosity is given as the
ratio between the turbulent kinetic energy (k) and its dissipation
rate (c):

turb,g
= C

g
k
2
c
(11)
where C

is a constant. The values of k and c come directly


fromthe differential transport equations for the kinetic energy and
turbulence dissipation rate as follows:

t
(g k) + (g ug k) =
__

lum,g
+

turb,g
o
k
_
k
_
+P
k
g c (12)

t
(g c) + (g ug c) =
__

lum,g
+

turb,g
oc
_
c
_
+
c
k
_
C
c,1
P
k
C
c,2
g c
_
(13)
where o
k
, o
c
, C
c,1
and C
c,2
are constants and P
k
is the turbulence
production, which is modeled using
P
k
=
turb,g
u
g

_
u
g
+(u
g
)
1
_
(14)
2.2. Heat transfer model
The heat transfer between phases has an important role in the
catalytic cracking of the hydrocarbons, since the hot regenerated
catalyst in contact with the gas oil feed should provide enough heat
for liquidvaporizationandendothermic cracking reactions. The gas
and solid energy equations are expressed respectively as

t
(g g Hg ) + (g g ug Hg ) = (g zg 1g ) +(1s 1g ) +g g

r
^Hr
Cr
t
(15)

t
(
s

s
H
s
) + (
s

s
u
s
H
s
) = (
s
z
s
1
s
) +(1
g
1
s
) (16)
where H, 1 and z are respectively the static enthalpy, the tem-
perature and the thermal conductivity of each phase; the last term
in Eq. (15) represents the variations in energy due to the endother-
mic reactions; and is the interphase heat transfer coefcient,
which is modeled using the following correlation:
=
Nuz
d
s
(17)
where d
s
is the mean diameter of the dispersed phase and Nu
is the Nusselt number. For a particle moving in an incompressible
Newtonian uid, the Nusselt number is a function of the Reynolds
and Prandtl numbers of the particle. The Nussel number used in
this work is estimated by the RanzMarshall correlation:
Nu = 2 +0.6Rc
0.5
Pr
0.3
(18)
2.3. Catalytic cracking kinetic model
The variation in chemical species in the gas phase is modeled as
follows:

t
(
g

g
C
g,i
) + (
g

g
u
g
C
g,i
) = (
g
I
i
C
g,i
) +

R
i
(19)
where C
g,i
is the concentration of specie i in the gas phase, I its
diffusivity in the phase and

R
i
the consumption or formation of this
specie due to the cracking reactions.
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
4 G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx
Fig. 1. Kinetic scheme of the four-lump reaction model.
Table 1
Denition of lumps Farag et al., 1994.
Lump Number of carbons
Gas oil C
13
and higher
Gasoline C5C
12
Light gases C
1
C
4
This work used a four-lump model to take into account the
catalytic cracking reactions. The representative reactions of this
kinetic model are shown in Fig. 1 and the denition of the lumps
according to the number of carbons in the molecules is based on
the work of Farag et al. (1994) and is presented in Table 1.
The general rate equation for reaction r is given by
R
i,r
= k
r
C
n
i
(20)
It is assumed that the cracking of gas oil is a second-order reac-
tion (n = 2) and the cracking of gasoline is a rst-order reaction
(n = 1).
The dependence of kinetic constants on temperature is given by
Arrhenius equations according to
k
r
= k
0
r
exp
_


r
R1
_
(21)
The parameter , appearing in Eq. (20), is the deactivation func-
tion which is related to the deposition of coke on the catalyst
surface. It is expressed by
= exp(K
c
q
1
) (22)
where K
c
is the deactivation constant estimated by Farag et al.
(1994) as a function of catalyst type (for FCC10 catalyst, K
c
=
8633 kg
cat
]kmol, according to Farag et al. (1994)) and q
1
is the
specic coke concentration given by
q
1
=
C
CQ

s
(23)
As cited previously there are many studies in which kinetic con-
stants for the four-lump model were estimated (Blasetti & de Lasa,
1997; Farag et al., 1994; Jurez et al., 1997; Lee et al., 1989; Pitault
et al., 1995). This workbuilds ontheworkof Faraget al. (1994), since
they estimated individually both the kinetic data and the adsorp-
tion parameters for different types of catalysts. This enables the
development of future work in which the adsorption effects can be
studied. As shown by Martignoni and de Lasa (2001) the adsorption
of hydrocarbons in the catalyst pores can signicantly affect the
solids and gas velocities and the residence time, and consequently
the intrinsic reaction kinetics, increasing the reaction yield.
Thevalues usedfor thekinetic constants werethoseobtainedfor
FCC10 catalyst (sample free of metal traps, nickel and vanadium).
Farag et al. (1994) concluded that overcracking of the gasoline
formed was negligible because the kinetic constant values for the
cracking of gasoline to produce light gases and coke are very close
to zero. Since the activation energies and heats of reaction are not
found in their work, this work used those reported by Jurez et al.
(1997) and Han and Chung (2001b). All these values are listed in
Table 2.
The kinetic constants, given in Table 2, were evaluated at 550

C
and are dependent on the amount of solids. In order to predict
these values at any temperature and catalyst concentration, the
pre-exponential factor of the Arrhenius equation was isolated from
Eq. (21) and multiplied by the local concentration of solids. Then
the kinetic constants were evaluated as follows:
k
r
(1,
s
) = k
r,550

C
(
s

s) exp
_

r
R
_
1
1

1
550

C
__
(24)
In order to consider the heterogeneous and endothermic kinet-
ics and catalyst deactivation, it was necessary to implement Eqs.
(20) to (24) in the CFX code.
3. Simulation
The commercial code ANSYS-CFX 11.0 was used to obtain the
numerical data for the model of this work. Appropriate func-
tions were implemented to couple the four-lump catalytic cracking
kinetic model and the catalyst deactivation with the solver. CFX
applies the nite volume methodtodiscretize the conservationand
closure equations. Inthis method, the calculationdomainis divided
into a nite number of control volumes in which discrete variables
are calculated. CFX uses shape functions to evaluate spatial deriva-
tives for all the diffusion and the pressure gradient terms. Transient
terms were approximated using a modied second-order back-
ward Euler scheme. The convective terms were discretized through
an upwind difference scheme. Once determined, the set of alge-
braic approximate equations was solved using a coupled approach,
which solves the hydrodynamic variables in a single matrix.
Three-dimensional predictions werecarriedout inorder toeval-
uate heat transfer, chemical reaction and the effects of geometry,
especially in the entrance area of the riser reactor. The geometry of
the riser reactor considered is illustrated in Fig. 2.
It is very important that CFD results are mesh independent.
Five mesh densities with 370,000, 520,000, 710,000, 900,000 and
1,200,000 control volumes were tested using the conditions inCase
1, giveninTable5. Themass fractionof gasolineat theoutlet andthe
pressure drop of the system are used to monitor numerical errors,
since they denote a representative variable of the goal of the simu-
lation and a global variable, respectively. As can be seen in Table 3,
there is no signicant difference between the results for the mass
fraction of gasoline. In contrast, the pressure drop increases with
the increase in mesh size up to 710,000 control volumes, where
pressure stabilizes. Thus this mesh proved to be adequate and it
was used to simulate the process.
A nonuniform grid is used in order to guarantee that smaller
control volumes are present where variable gradients are steeper.
To maintain an aspect ratio of less than 20, the length of the control
volumes along the axial direction can not be very large. Details of
the mesh in the inlet and outlet regions are shown in Fig. 3. The
distribution of the control volumes along the radius of the reactor
is also shown.
Table 2
Kinetic constants, activation energies and heats of reaction.
Reaction r kr |
550

C
(m
6
kmol
1
kg
1
cat
s
1
) r (J mol
1
) ^Hr (kJ kg
1
)
Gas oil gasoline 20.4 57,360 195
Gas oil light gases 7.8 52,750 670
Gas oil coke 3.0 31,820 745
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx 5
Fig. 2. Geometry of the riser analyzed (inlets and outlet).
Table 3
Monitoring of the variables for different mesh densities.
370,000 520,000 710,000 900,000 1,200,000
Mass fraction of gasoline 0.466 0.464 0.466 0.468 0.465
Pressure drop (Pa) 6316 6395 6488 6503 6484
Fig. 3. Details of the mesh used for the industrial riser.
Vaporized gas oil mixed with water vapor is injected into the
base of the reactor and a side entrance is used for feeding in the hot
catalyst 7wt.% of the total water vapor is fed in with the catalyst
as shown in Fig. 2. The nonslip condition at the walls was used
for both phases.
The dimensions of the industrial riser simulated in this work
are the same as those given in the work of Ali et al. (1997) and are
shown in Table 4. In order to compare the simulated results with
sets of experimental and commercial runs, two cases with different
operating conditions, shown in Table 5, were performed. The phys-
ical properties of the reactive species and the catalyst were taken
fromthe work of Martignoni and de Lasa (2001) and Landeghemet
al. (1996) and are listed in Table 6.
Table 4
Dimensions of the simulated industrial riser.
Part Riser Secondary entrance
Length (m) 34.2 2.0
Diameter (m) 0.8 0.6
Angle of inclination 45

Height from the principal inlet (m) 2.0


The time step used for the solution was 10
3
seconds. This value
provided a Courant number less than one for both cases, which
ensures that the simulation results are independent of the time
step chosen. The convergence criterion for advancing in time was
that the RMS residuals were less than 10
4
. The simulations were
solved using a parallel code with nine partitions. About ten days
of calculation were necessary to predict ten seconds of reactive
ow for each simulation on computers provided with Xeon 3GHz
dual-core processors.
4. Results and discussion
The time-averaged results presented in this section were
obtained by simulating unsteady-state cases over a period of time
long enough to show that the variables had a cyclic behavior.
Simulateddata fromCase 1onthe conversionof reactive species
at the riser outlet were compared with experimental data reported
in the work of Farag et al. (1994), from which the kinetic constants
used in these simulations were taken. Farag et al. (1994) used an
isothermal microreactor at 783K and a catalyst-to-oil ratio of 6.
Because of the contrast in scale and type of reactor, the bound-
ary conditions used in the simulation were similar, but not the
same as those applied in the experiments. As can be seen in Table
7, despite the differences in reactor scale and conditions, there is
qualitative agreement between simulated results and experimen-
tal data, showing that the kinetic parameters used can be adequate
to describe the catalytic cracking of gas oil.
Table 5
Operating conditions used in the simulations.
Parameter Case 1 Case 2
Mass ux of feed oil (kgm
2
s
1
) 20 40
Catalyst-to-oil ratio (kg
cat
]kg
gasoil
) 7 7
Steam (wt.%) 3 3
Feed temperature (K) 600 500
Catalyst inlet temperature (K) 900 900
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
6 G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx
Table 6
Physical properties of reactive species and catalyst.
Specie Gas oil Gasoline Light gases Coke Catalyst
Density (kgm
3
) 6 1.5 0.8 1400 1400
Specic heat (J kg
1
K
1
) 2420 2420 2420 1090 1090
Thermal conductivity (Wm
1
K
1
) 0.025 0.025 0.025 0.045 0.045
Molecular weight (kgkmol
1
) 400 100 50 400
Particle diameter (m) 65
Table 7
Simulation results compared with experimental data.
Experimental (Farag et al., 1994) Simulation of case 1 Relative deviation (%)
Unconverted gas oil (wt.%) 30.9 26.6 13.9
Gasoline yield (wt.%) 49.0 47.8 2.4
Light gases yield (wt.%) 14.8 18.2 23.0
Coke yield (wt.%) 5.3 7.1 34.0
Table 8
Characteristics and operating conditions of commercial runs (Derouin et al., 1997).
FCC riser
Internal diameter (m) 0.7 1
Height (m) 30
Solids
Mean diameter (m) 60 70
Mass ow rate (kgm
2
s
1
) 300 600
Gas
Mean velocity (m/s) 415
Derouin, Nevicato, Forissier, Wild and Bernard (1997) reported
on samples taken in a commercial run along the riser center line.
The operating conditions used in the simulation of Case 1 are in the
range given in their work (Table 8). As shown in Fig. 4, the agree-
ment betweenthese samples withsimulatedgas oil conversionand
gasoline yield at the riser center line is good. The dependence of
the kinetic constants on catalyst concentration and temperature
induces the formation of peaks of conversion close to the reactor
bottom, since this regionhas hightemperatures anda large amount
of solids.
The model was alsoveriedby comparing the simulationresults
fromCase 2withthe data foundinthe workof Ali et al. (1997). Their
datarefers toanindustrial riser withaheight of 33mandadiameter
of 0.8m, operating under the same conditions as those simulated
in Case 2 (Table 5). A comparison of the model results with the
published commercial data obtained at the riser outlet is shown in
Table 9. It is important to emphasize that the computational model
requires detailed information on the feedstock and the catalyst as
well as the design of the industrial reactor, which is seldom avail-
able in the published data on industrial reactors. Therefore it was
necessary to make some assumptions to allowfor the process sim-
Fig. 4. Predicted model results and the plant data obtained by Derouin et al. (1997)
along the riser center line.
ulation. Despite the assumptions, the simulated results show good
agreement with the plant data.
When the cracking reactions occur, large molecules are broken
into smaller ones causing a volumetric expansion of the gas phase
and hence an increase in its velocity. This expansion also causes an
increase in catalyst velocity due to the drag force, decreasing the
solids volume fraction and separating the bed into a dense region
close to the bottom and a dilute region in the rest of the riser. This
set of events can be observed in Fig. 5, where average values for the
gas-phase density, gas-phase velocity and catalyst volume fraction
in cross sections along the reactor height are shown.
Fig. 5. Radial average gas density, gas velocity and solids volume fraction.
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx 7
Table 9
Comparison of predicted model results and the plant data reported by Ali et al. (1997).
Plant data (Ali et al., 1997) Simulation of case 2 Relative deviation (%)
Gasoline yield (wt.%) 43.9 44.0 0.2
Coke yield (wt.%) 5.8 6.6 13.8
Temperature (K) 795 780 1.9
The gas-phase temperature prole in an axial plane of the riser,
for Case 2, are showninFig. 6. The temperature is about 500 Kat the
bottomand increases when brought into contact with hot particles
that are injected at a temperature of 900 K. Rapid heat transfer
occurs, and because of the endothermic behavior of the catalytic
reactions, the temperature of the mixture decreases.
The inlet zone of the riser is the most complex part of the reac-
tor; it is here where intense turbulence and ow inhomogeneities
can result in high temperature and concentration gradients, lead-
ing to undesirable thermal cracking and catalyst deactivation. In
this region the solid particles and the gas phase are brought into
contact. Consequently heat is transferred between the phases and
the cracking reactions are initiated. As can be seen in Fig. 7, which
shows the catalyst volume fraction proles for Cases 1 and 2 in
internal planes, when the solid particles enter the riser, they slide
along the lateral pipe and fall into the main duct. The gas enter-
ing the reactor bottom has to cross the barrier formed by the solid
particles. This obstacle is more easily transposed in Case 2, since
the mass ux of feed oil is higher in this case and there is enough
kinetic energy to readily drag the solids.
It is known that the core-annulus structure exists in FCC risers.
In Fig. 8 the distribution of the catalyst in cross-sectional planes at
different heights is shownfor Cases 1and2. Inspectionof this gure
reveals that, for both cases, close to the bottom, at a height of 5m,
there is a higher concentration of solids and the ow eld is highly
nonhomogeneous. These inhomogeneities tend to decrease as the
gas phase accelerates andthe owbecomes more diluted. The core-
annulus structure is observed for both cases; however part of the
solids remains in the central region of the reactor. This pattern is
less pronounced when the supercial gas velocity increases with
the expansion of the bed due to the development of the reactions
along the riser height.
In Fig. 9, the solids volume fraction and gas-phase velocity pro-
les in a plane normal to the catalyst entrance are shown for Case
2. When the gas encounters the barrier formed by the solid phase
Fig. 6. Gas-phase temperature prole.
it moves out, carrying some of the catalyst particles which accu-
mulate on the walls. Another fraction of the solids remains in the
central region of the reactor, highlighting the patterns seen in Fig.
8. The increase in gas-phase velocity can also be observed in Fig.
9. This is caused initially by the restriction of the ow by the solid
phase and later by the cracking reactions that rapidly increase the
number of moles, expanding the bed.
Gas-phase velocity stream lines for Cases 1 and 2 are shown in
Fig. 10. Swirl areas canbe seeninthe entrance regionfor bothcases,
indicating that strong turbulence is present there. As can be seen in
Fig. 7. Catalyst volume fraction contour plots for (A) Case 1 and (B) Case 2.
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
8 G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx
Fig. 8. Cross-section proles of catalyst volume fraction.
Fig. 9. Proles of catalyst volume fraction and gas-phase velocity.
Fig. 10. Gas-phase velocity stream lines for (A) Case 1 and (B) Case 2.
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx 9
Fig. 11. Turbulence kinetic energy contour plots for (A) Case 1 and (B) Case 2.
Fig. 12. Relative difference in temperature contour plots for (A) Case 1 and (B) Case 2.
Fig. 11, high values of turbulence kinetic energy are observed in the
entrance region. In Case 2 the highest values of turbulence kinetic
energyareseenintheareawherethegas phaseaccelerates, whilein
Case 1 there are also high values in the region where the phases are
in contact with each other, indicating that the interaction between
the phases has an important role in turbulence production in this
case.
The relative difference in the local temperature of the solid
and gas phases is shown in Fig. 12. As can be seen, the difference
between the temperature of the phases in Case 1 falls to values
close to zero at low heights, while in Case 2 this process takes
longer. These results indicate that the high values of turbulence
kinetic energy at lower heights in Case 1 causes a better mixing of
the phases and, consequently, a more efcient heat exchange.
5. Conclusions
This work presented an extended model for an FCC industrial
riser reactor. Coke deposition on the catalyst pores and its deacti-
vation rate were included in the mathematical model. The species
concentrations predicted in the simulations using the four-lump
model showed good agreement with sets of published experimen-
tal and commercial data.
The results obtained with this model highlight the importance
of three-dimensional simulations to correctly predict geometric
effects, specially in the feed injection area where the gas oil meets
the hot catalyst, causing strong momentumandtemperature gradi-
ents. This region requires special attention because it is the section
where the cracking reactions are initiated.
Results also indicate that the gas and solids feed owrates inu-
ence the ow pattern. A different interaction between the phases
changes the way in which heat is transferred and consequently the
yield of the reactions. Thus, it is important to know the nonunifor-
mities of the ow in order to correctly predict the reaction yield.
If a high yield of gasoline is desired, it is necessary to maintain a
more uniformradial dispersion. The three-dimensional model pro-
vides a more correct solids distribution, giving a better prediction
of reactor performance.
In conclusion, the model presented here is useful for the opti-
mization and operation of FCC risers. Future work will also include
Please cite this article in press as: Lopes, G. C., et al. Three-dimensional modeling of uid catalytic cracking industrial riser ow and reactions.
Computers and Chemical Engineering (2011), doi:10.1016/j.compchemeng.2010.12.014
ARTICLE IN PRESS
GModel
CACE-4170; No. of Pages 10
10 G.C. Lopes et al. / Computers and Chemical Engineering xxx (2011) xxxxxx
a more sophisticated kinetic scheme containing more lumps and a
model to predict gasoil vaporization as well as a new geometry
conguration including several injection inlets for the feeding of
gasoil droplets.
Acknowledgments
The authors are grateful for the nancial support of PETROBRAS
for this research.
References
Ahari, J. S., Farshi, A., &Forsat, K. (2008). Amathematical modelingof the riser reactor
in industrial FCC unit. Petroleum and Coal, 50(2), 1524.
Ali, H., Rohani, S., & Corriou, J. P. (1997). Modelling and control of a riser type
uid catalytic cracking (FCC) unit. Chemical Engineering Research and Design, 75,
401412.
Ansys Inc. (US). ANSYS CFX-Solver theory guide. Release 12.0. Canonsburg, PA
(2009).
Blasetti, A., & de Lasa, H. (1997). FCC riser unit operated in the heat-transfer model:
Kinetic modeling. Industrial and Engineering Chemistry Research, 36, 32233229.
Deng, R., Wei, F., Liu, T., & Jin, Y. (2002). Radial behavior in riser and downer during
the FCC process. Chemical Engineering and Processing, 41, 259266.
Derouin, C., Nevicato, D., Forissier, M., Wild, G., &Bernard, J. R. (1997). Hydrodynam-
ics of riser units and their impact on FCC operation. Industrial and Engineering
Chemistry Research, 36, 45044515.
Farag, H., Blasetti, A., & de Lasa, H. (1994). Catalytic cracking with FCCT loaded
with tin metal traps. Adsorption constants for gas oil, gasoline, and light gases.
Industrial and Engineering Chemistry Research, 33, 31313140.
Gidaspow, D. (1994). Multiphase ow and uidization: Continuum and kinetic theory
descriptions. Academic Press, Inc.
Han, I. S., & Chung, C. B. (2001a). Dynamic modeling and simulation of a uidized
catalytic cracking process Part I. Process modeling. Chemical Engineering Science,
56, 19511971.
Han, I. S., & Chung, C. B. (2001b). Dynamic modeling and simulation of a uidized
catalytic cracking process Part II Property estimation and simulation. Chemical
Engineering Science, 56, 19731990.
Jurez, J. A., Isunza, F. L., Rodrguez, E. A., & Mayorga, J. C. M. (1997). A strategy for
kinetic parameter estimation in the uid catalytic cracking process. Industrial
and Engineering Chemistry Research, 36, 51705174.
Landeghem, F. V., Nevicato, D., Pitault, I., Forissier, M., Turlier, P., Derouin, C., &
Bernard, J. R. (1996). Fluid catalytic cracking: Modelling of an industrial riser.
Applied Catalysis A-General, 138, 381405.
Lee, L. S., Chen, Y. W., Huang, T. N., & Pan, W. Y. (1989). Four-lump kinetic model
for uid catalytic cracking process. Canadian Journal of Chemical Engineering, 67,
615619.
Martignoni, W., & de Lasa, H. I. (2001). Heterogeneous reaction model for FCC riser
units. Chemical Engineering Science, 56, 605612.
Nayak, S. V., Joshi, S. L., & Ranade, V. V. (2005). Modeling of vaporization and crack-
ing of liquid oil injected in a gassolid riser. Chemical Engineering Science, 60,
60496066.
Prssinen, J. H., & Zhu, J. X. (2001). Particle velocity and ow development in a long
and high-ux circulating uidized bed riser. Chemical Engineering Science, 56,
52955303.
Pitault, I., Forissier, M., & Bernard, J. R. (1995). Dtermination de constantes cin-
tiques du craquage catalytique par la modlisation du test de microactivit
(MAT). Canadian Journal of Chemical Engineering, 73, 498504.
Zhang, W., Tung, Y., & Johnsson, F. (1991). Radial voidage proles in fast
uidized beds of different diameters. Chemical Engineering Science, 46,
30453052.

Вам также может понравиться