You are on page 1of 14

Powder Technology 246 (2013) 303316

Contents lists available at SciVerse ScienceDirect

Powder Technology
journal homepage: www.elsevier.com/locate/powtec

CFD simulation of cylindrical spouted beds by the kinetic theory of


granular ow
Seyyed Hossein Hosseini a,, Goodarz Ahmadi b, Martin Olazar c
a
b
c

Department of Chemical Engineering, University of Ilam, Ilam 69315-516, Iran


Department of Mechanical and Aeronautical Engineering, Clarkson University, Potsdam, NY 13699-5725, USA
Department of Chemical Engineering, University of the Basque Country, Bilbao, Spain

a r t i c l e

i n f o

Article history:
Received 8 November 2012
Received in revised form 30 April 2013
Accepted 4 May 2013
Available online 17 May 2013
Keywords:
Cylindrical spouted bed
CFD
Hydrodynamics
Two-Fluid Model
Drag function
Solid viscosity

a b s t r a c t
The hydrodynamics of a cylindrical spouted bed was studied using a EulerianEulerian Two-Fluid Model (TFM)
including the kinetic theory of granular ows. A series of simulations was performed; and the inuences of
the drag model, solid shear viscosity model, discretization scheme, as well as, transport equation for granular
temperature were studied. The CFD results showed that different drag and solid shear viscosity models led to
signicant differences in the model prediction for the dilute region of the bed. The representative unit cell
(RUC) drag model and the Syamlal et al. [1] viscosity model were found to be in close quantitative agreement
with the experimental observations. In terms of the solid ow pattern in the spout and fountain zones, it was
found that an algebraic equation for granular temperature with the appropriate coefcient of restitution provided reasonable results at considerable computational economy compared with the full transport equation. It was
shown that the discretization scheme signicantly affects the computational model predictions; therefore, the
computational modeling scheme should be optimized. The TFM model was also used to predict particle velocity
proles and voidage distribution in the spout and fountain regions. The simulation results were fairly consistent
with the experimental data in a wide range of gas ow rates.
2013 Elsevier B.V. All rights reserved.

1. Introduction
Spouted beds are well known for their ability to handle coarse particles corresponding to Geldart group D, because they provide good
mixing and circulation patterns when beds made up of large and irregular particles are used [2]. Today, spouted beds are used in various industrial processes, such as drying, coating, granulation, black liquor,
polymerizations, pyrolysis and gasication because of their efciency
in providing effective gascoarse particle contact [310].
A spouted bed typically has three distinct regions: a central spout,
annulus, and a fountain region. The volume fraction of particles varies
from almost zero in the spout region to its maximum packing limit in
the annulus region, which covers a complex recirculation pattern and
a range of diverse gasparticle interactions.
In order to make more effective use of cylindrical spouted beds in
industrial applications, detailed knowledge of gas and particle hydrodynamics is needed. Although new approaches to the experimental investigation of spouted beds have provided much needed information,
computer simulation techniques have also evolved into useful tools
for obtaining detailed information of the instantaneous ow behavior
in spouted beds.

Corresponding author. Tel.: +98 913 7944470.


E-mail address: s.h.hosseini@mail.ilam.ac.ir (S.H. Hosseini).
0032-5910/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.powtec.2013.05.016

With the major advances of computational methods, computational


uid dynamics (CFD) has been used for modeling multiphase ows to
reduce design time and cost. CFD has now emerged as an effective tool
in the simulation of gassolid ows, especially in uidized and spouted
beds.
The approaches that are used in the simulation of multiphase
ows are the Two-Fluid Model (TFM) and the discrete element method
(DEM). The DEM approach is based on a Lagrangian trajectory analysis
and, therefore, is a more natural way to simulate the gassolid ow. The
trajectory analysis approach, however, becomes computationally more
demanding as the number of particles increases. In the TFM approach,
the gas and the particulate phase are described as interpenetrating
continua [11]. The popularity of this approach compared to the DEM
stems from relatively smaller CPU and memory resource requirements.
As a result, the TFM approach is more commonly used for modeling of
practical spouting ows and is also used in the current study.
Another approach for understanding the details of gassolid ows,
which has been used by a number of researchers [1218] in recent years,
is direct numerical simulation (DNS). While DNS provides an indispensable ability to investigate the micro-scale details of gassolid ows and
provide a test bed for constitutive laws for higher-level models such as
two-uid models, it is computationally expensive. Thus, the applications
of the DNS approach have been limited to low Reynolds number ows
and idealized geometries. That is, using DNS for typical industrial-scale
or even laboratory-scale gassolid systems has been out of reach due

304

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

to the required extensive computational resources. More recently, several hybrid methods, such as the smoothed particle hydrodynamics (SPH)
method and lattice Boltzmann scheme, for solving the two-uid model
of particleuid uidizations were developed [1921].
A successful TFM depends on the proper modeling of all possible
intra- and inter-phase interactions, such as gassolid drag, in addition
to collision and frictional contact between particles and between
particles and the wall [2224]. In this context, by introducing the concepts of solid pressure and viscosity, the kinetic theory of granular
ow has been established, which has been used to compute the solid
stresses. A summary of the particle interactions in CFD simulation of
gassolid uidization systems is provided in the subsequent sections.
Several researchers, assuming two-dimensional and/or axial symmetry conditions, have simulated cylindrical spouted beds with CFD
tools, using the combined kinetic theory of granular ow (KTGF) and
TFM. Among these publications, there have been some efforts on the selection of optimal modeling parameters. The drag function, which has
the primary effect on the hydrodynamic interactions of the phases in
spouted beds, has been studied in detail. Du et al. [25] used the TFM
model of the FLUENT commercial code to simulate the cylindrical
spouted bed studied experimentally by He et al. [26,27]. Du et al. [25] investigated the effect of the drag model on the prediction of spouted bed,
and found that the drag models suggested by Syamlal and O'Brien [28],
Arastoopour et al. [29] and Gidaspow [30] provide qualitatively satisfactory ow patterns, whereas the best agreement with the experimental
data is obtained by the Gidaspow [30] model with switch function. Furthermore, Hosseini et al. [31] studied the effects of different drag models
on the CFD results of a cylindrical spouted bed with a non-porous draft
tube, and found that the WenYu drag function was most appropriate.
Recently, Hosseini et al. [32] simulated a two-dimensional conical
spouted bed via a EulerianEulerian approach and showed that the
drag model of Dalla Valle [33] was the optimal choice for their simulated
case. These earlier works found that the drag model signicantly
impacts the model prediction for the solid phase ow.
The restitution coefcient, which is a measure of inelastic particle
particle collisions, is another key parameter for computing the solid
phase properties. Du et al. [34] studied the effect of restitution coefcient
and maximum packing on the CFD simulation of a spouted bed. Hosseini
et al. [32] investigated the effects of restitution coefcient on the hydrodynamics of a two-dimensional conical spouted; and, consistent with the
ndings of Du et al. [34], they found that the restitution coefcient significantly affects the granular temperature (particle uctuating motion).
The other essential parameter in a CFD simulation of gassolid systems is the particlewall interaction. Zhang and Yu [35] found that computer model simulation for uidization depends on the boundary
conditions used. In particular, the free- and no-slip boundary conditions
for particulate phase at the wall led to different types of slugs in the slugging uidized beds. Li et al. [36,37] investigated the inuence of the wall
boundary condition on the predicted ow hydrodynamics of gassolid
uidized beds and bubbling uidized beds. They found that the
solid-phase wall boundary condition has a substantial impact on the
model prediction. Almuttahar and Taghipour [38] showed that for the
ow of uid catalytic cracking (FCC) particles in a high concentration circulating uidized bed, the use of a small specularity coefcient wall
boundary condition, rather than the no-slip wall boundary condition,
leads to model predictions that are in better agreement with the experimental data. More recently, Lan et al. [24] investigated the hydrodynamics of spouted beds using the TFM with different values of the specularity
coefcient and restitution coefcient for solid-phase wall boundary conditions. In the present study, the solid-phase wall boundary conditions
proposed by Lan et al. [24], including the effects of specularity coefcient,
as well as, restitution coefcient are adopted and used.
The solid particles in the annulus region undergo frictional contact
with multiple neighbors due to the high solid volume fraction. Consequently, the frictional stress must be included in the CFD model. Huilin
et al. [39] and Shuyan et al. [40] used the kinetic-frictional constitutive

model for the solids to simulate the gassolid ow pattern in spouted


beds by CFD code K-FIX. Their results are consistent with the experimental data obtained by He et al. [26,27] and San Jos et al. [41].
Hosseini et al. [32] analyzed the sensitivity of the CFD simulation
results of a spouted bed to the variation of physical and modeling
parameters including the frictional stress models. They concluded that
the frictional stresses are of particular signicance in the simulation of
conical spouted beds. Accordingly, a suitable frictional stress model is
required not only in the annulus, but also in the spout zone (a dilute
particulate region) for predicting the spouted bed hydrodynamics.
Van Wachem et al. [42] provided a summary of the governing equations and closure models commonly used in CFD simulation of gassolid
uidization systems. Their quantitative comparisons showed that the
CFD models in bubbling uidized beds are not sensitive to the use of
different solid stress models or radial distribution functions. They also
concluded that gravity and drag force are dominant for the majority of
the ows and that frictional stresses become signicant for very dense
ows. van Wachem et al. [42] and Hosseini et al. [43] showed that
by using an algebraic expression, instead of the full equation, for the
granular temperature for CFD simulation in dense gassolid systems
in bubbling and slugging uidized beds, the computational time is
reduced without losing accuracy.
Almost all the modeling results of a cylindrical spouted bed show an
underprediction of particle velocity in the spout region. From the CFD results of Du et al. [25,34], it is found that their model underpredicts the axial
and radial distribution of particle velocity, as well as voidage at different
bed levels. Nevertheless, their model has been widely used by numerous
researchers for simulating cylindrical and conical spouted beds. Zhonghua
and Mujumdar [44] studied gasparticle ow behavior in a cylindrical
spouted bed as well as in a rectangular spoutuid bed containing spherical particles. Their model is quite similar to that proposed by Du et al.
[24,34]. They used FLUENT code and compared their results with the experimental data of He et al. [26,27] and Link et al. [45], respectively, for cylindrical and rectangular beds. The overall ow patterns within the
cylindrical spouted bed, as well as bubble formation inside the rectangular
spoutuid bed, are well predicted by the model. The model, however,
underpredicts the particle velocity in the spout portion of the bed.
Wang et al. [46] simulated the cylindrical spouted bed used by He et
al. [26] and reported that the simulated axial particle velocities signicantly underestimate the experimental data at every axial level. They
adjusted (scaled down) the experimental data of axial particle velocity
and compared the adjusted experimental data with the simulation
results. A good agreement was found by their adjustment.
Duarte et al. [47] simulated a cylindrical spouted bed by using a parabolic gas velocity prole, generated by the CFD model, which was at
the same fountain height as in the experimental study of He et al.
[27], and reported suitable results in terms of axial particle velocity at
different levels.
Thus, this present study is focused on providing an understanding
of the patterns of gassolid ows in the dilute spout and fountain
regions of cylindricalconical spouted beds. In particular, the effects
of certain modeling parameters, including solid viscosity, restitution
coefcient, drag models, and the use of algebraic and/or full granular
temperature equations, are analyzed. Inuence of the discretization
scheme on the CFD results is also investigated. Finally, the proposed
model is evaluated for different gas inlet velocities. Moreover, the
quantitative effects of certain parameters included in the drag and
solid shear viscosity models are also studied.
2. Numerical method
2.1. Hydrodynamic model
The governing equations and the associated constitutive models of
the EulerianEulerian TFM that are used in the simulation of cylindrical spouted beds with a conical base are summarized in this section.

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

The continuity equation for the qth phase without any mass transfer
between the phases is given by

305

Solid pressure:
2






q q q q q 0
t

P s s s s 2s 1 es s g 0 s :

10

q,

respectively, are the volume fraction, density and


where q, q and
velocity of the qth phase. The conservation of momentum for the gas
and solid phases is given as

The solid shear viscosities as given by Gidaspow et al. [48] and


Syamlal et al. [1] are
s

p 
 1=2
2
10ds s s
4

1 s g 0 1 es
s s ds g 0 1 es s
961 es g 0

5
5
11

Gas phase:






g g g g g g g g P g
t
 
g g gK g s g s

p 

 1=2
d s
4

2
s s ds g0 1 es s
s s s
1 1 es 3es 1 s g 0 :
5
5

63es

12
The frictional viscosity as given by Schaeffer [49] is

Solid gas phase:








s s s s s PP s s
t s s s
 
s s g K g s g s

s; fr

P s sin
p :
2 I 2D

13

The diffusivity of granular temperature is given as:


where s = 1 g.
The transport equation for granular temperature, s (uctuation
kinetic energy of particles), is given as

 


3
s s s s s s s ps I s : s
2 t
ks s s gs :

ks

r
p 
2
150ds s s
6
s
2
:
1 s g 0 1 es 2s ds s g 0 1 es
3841 es g0

5
14

Solid bulk viscosity is given as:


s

Assuming that the granular uctuation energy is in local equilibrium


and its production and dissipation are balanced locally, the convection
and diffusive terms can then be ignored. Under this assumption,
Eq. (4) may be expressed as



0 ps I s : v s s :

Solid and gas phase stress tensors:

g0

!2:5

s; max

s; max

Several values of the restitution coefcient, es, are tested in the


present simulations.
The transfer of kinetic energy between phases is expressed as:
gs 3K gs s :




3 s g g s g 2:65
g
for g > 0:8
K gs;WenYu C D
ds
4


0:687 
24
for Re b 1000
1 0:15 g Res
C D g Res
C D 0:44
Re > 1000

for


2
s g s g
s g
1:75
for g 0:8:
K gs;Ergun 150
ds
g d2s

16

Symbol

Description

Experimental
run

Computer
run

s(kg/m3)
g(kg/m3)
s,max
Ums(m/s)
Ugs(m/s)

(kg/m3) Solid density


Gas density
Maximum solid volume fraction
Minimum spouting velocity
Gas supercial velocity

Same
Same
0.59

Same

E
ew
Ds(mm)
Dt(mm)
H(mm)
H0(mm)
s

ds(mm)
g(Pa s)

Particleparticle restitution coefcient


Particlewall restitution coefcient
Diameter of the spout gas inlet
Diameter of the bed
Vessel height
Static bed depth
Loose packing volume fraction
Internal friction angle of particles
Particle diameter
Gas viscosity

2503 kg/m3
1.225
Not reported
0.54
0.54, 0.594,
0.648, 0.702
Not reported
Not reported
19
152
1400
325
0.588
Not reported
1.41
1.7894 10 5

Collisional energy dissipation:




2
12 1es g 0
2 3=2
p
s
s s s :
ds

In this paper, the drag models from Gidaspow [30] and Du Plessis
[50] are investigated. The drag function suggested by Gidaspow [30]
is expressed as:

Table 1
The physical and numerical parameters.

Radial distribution function:


s

15

The constitutive models used are







2
s s s s s T s s s s I
3

 2

g g g g g T g I :
3

 2
4

s s ds g 0 1 es s :

0.90.96
0.9
Same
Same
1000
Same
Same
28
Same
Same

306

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

Fig. 2. Comparison of solid shear viscosity models and drag functions.

Fig. 1. Schematic of (a) geometry, and (b) grids of the spouted bed simulated.

To avoid the discontinuity of the two equations for dense and low
concentration conditions, a switch function is introduced that provides
a rapid transition of the drag expression from dense regime to dilute
regime. That is,
arctan150  1:750:2 s 
gs
0:5:

bed. More details on the RUC model are found elsewhere [50]. Santos
et al. [52] used an RUC drag function without the frictional term in a
CFD model for the hydrodynamic study of a spouted bed with conical
base. They overpredicted the axial particle velocity for different bed
levels. However, they did obtain a suitable fountain height using this
drag model. The RUC model is expressed as

Kgs ARUC

Kgs

g d2s

17

BRUC




g s s g
ds

19

where

Thus, the momentum exchange coefcient can be expressed as




1gs Kgs;Ergun gs Kgs;WenYu :



s 1 g g

18

The representative unit cell (RUC) model, proposed by Du Plessis


[50], is similar to the Ergun equation [51], and its formulation is
based on the pressure drop through porous media.
The RUC drag model is used to provide physical meaning to the
semi-empirical coefcients of 150 and 1.75 used in the Ergun equation,
and it uses analytically derived parameters. This drag model is applicable over the entire porosity range and steady laminar ow regime and is
well suited as a drag model in numerical computations. According to
several researchers, the RUC model provides reasonable predictions
in the two-uid modeling of gassolid uidization systems such as
spouted beds, jet uidized beds, and bubbling uidized beds [5254].
It is a most promising drag model for TFMs because no empirical coefcients are involved and it allows for the introduction structured adaptations for improvement, based on the physical conditions in the bed.
Another positive point of this model is that the same model is used
over the entire range of voidages and Reynolds numbers found in a

ARUC

BRUC

3g
; g 0:99
26:8
H e 1Het 1H e 2
785:0;
g > 0:99

8
< 26:8
:

1H e 2
2:25;

20

; g > 0:01
g 0:01

and

1=3
H et 1 g

2=3
H e 1 g
:

21

Turbulent gas uctuations in the spout and fountain regions may


affect gassolid ow behavior, but there is no consensus on the best
turbulence model for the CFD simulation of spouted beds or whether
turbulent uctuation effects should be accounted for. Only Du [55]
found that the dispersed turbulence model is a better choice than
its per-phase counterpart for simulating ow behavior in spouted

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

beds. Therefore, the dispersed turbulence model is adopted in this


study, where turbulence predictions for the gas phase are obtained
by the standard k model.

Table 2
Fountain heights, maximum particle velocities along the bed axis for different solid
shear viscosity models.
Items

2.2. Simulation conditions


The experimental data of He et al. [26,27] for a cylindrical spouted
bed with conical base are used to validate the present model. They
used a ber optic probe to measure particle velocities as well as voidages
in a spouted bed. A detailed description of the experimental setup and
the spouted bed studied was reported by He et al. [26,27]. The corresponding physical and numerical parameters selected for the present
simulation are listed in Table 1. The ow in the freeboard above the
bed is assumed to be fully developed and, to fulll this condition, the
vertical dimension of the bed must be sufciently high. Hence, in the
simulation carried out, a bed height of 1.0 m is deemed to be sufcient.
2.3. Solution procedure
A commercial grid-generation tool, GAMBIT 2.2, is used to create
the two-dimensional geometry and computational grid. In addition,
the CFD code FLUENT 6.3 is used to simulate the hydrodynamics
of the cylindrical spouted bed studied by He et al. [26,27]. The set of
governing equations described in Section 2.1 is solved by a nite
control volume technique.
The phase-coupled PC-SIMPLE algorithm is used for the pressure
velocity coupling. Cammarata et al. [56] investigated the consistency of
two- and three-dimensional simulations. They recommend using the
two-dimensional/axisymmetric models for reducing the computational
time when the gassolid ow behavior is close to two-dimensional
or when axisymmetric and three-dimensional out-of-plane motion are
negligible. In this study, axial symmetry is assumed, and an axisymmetric
computational model is used in the simulation. A second-order upwind

Fig. 3. Contour plots of solid volume fraction in the bed for U = 1.2Ums, es = 0.9, the
algebraic equation for granular temperature, the Gidaspow drag model [30] and (a): the
Gidaspow et al. [48] viscosity model, and (b): the Syamlal et al. [1] viscosity model.

307

Experimental Predicted results


data
Viscosity model of Viscosity model
Gidaspow et al.
of Syamlal et al.
[48]
[1]

Maximum particle velocities 5.67


along the bed axis, m/s
Fountain heights, m
0.25

3.19

3.95

0.105

0.165

discretization scheme is used for momentum, turbulence kinetic energy


and turbulence dissipation rate equations; and a rst-order upwind
scheme is only used for the volume fraction term. Transient simulations
are performed with a constant time step of 0.0001 s with 30 iterations
per time step. A convergence criterion of 103 for each scaled residual
component is used for the relative error between two iterations. A grid
resolution of 36 180 was used in this study, which is the same as that
of Lan et al. [24]. The bed and grid structures for the computational
domain are illustrated, respectively, in Fig. 1a and b.

2.4. Initial and boundary conditions


The Dirichlet boundary conditions at the bottom of the bed (nozzle)
are used to specify a uniform gas inlet velocity. The gas enters through
the nozzle in the axial direction. Thus, the inlet gas velocity is given as
ux,0 = 0 and uy,0 = U.
As mentioned before, the outow condition with zero velocity
gradients is specied at the top of the freeboard. At the axis of the
spouted bed, the velocity gradients for both phases and the granular
temperature gradient along the radial direction are zero. A no-slip
boundary condition at the lateral bed wall is assumed for the gas
phase. The particle normal velocity is set to zero at the wall. The Johnson

Fig. 4. Contour plots of solid volume fraction for U = 1.2Ums, the RUC drag model, the
viscosity model of Syamlal et al. [1], the algebraic equation for granular temperature
and different values of es.

308

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

3. Results and discussion

lower values compared to those in the Gidaspow et al. [48] viscosity


model at low solid concentrations. Note that lower shear viscosity
values are calculated for the spout region, where the solid volume fraction is less than 0.25. Furthermore, a decrease in solid viscosity leads to
a decrease in the resistance of solid particles against their upward
movement. Therefore, the Syamlal et al. [1] shear viscosity model will
lead to lower resistance in the spout region. Fig. 2 shows both models
lead to the same results in the dense region of the gas-uidized beds.

3.1. Quantitative analysis of drag and solid viscosity models

3.2. Effect of solid viscosity on CFD model predictions

As noted before, the Gidaspow [30] model is used in the present


simulations of gassolid uidization systems. The drag models given
by Eqs. (16)(21) are used in the analysis. Fig. 2 compares the variation of drag functions with solid volume fraction. Here, a relative
interstitial velocity of 10.0 m/s (which is typical in the spout region)
and a Reynolds number of 953 are assumed. Fig. 2 shows that in the
dilute region (solid volume fraction b 0.25) the RUC drag model
leads to much higher values compared with those shown in
Gidaspow [30]. Eqs. (20) and (21) clearly show the coefcients ARUC
and BRUC in the RUC model are a function of volume fraction, and
ARUC increases sharply with an increase in the gas volume fraction.
This in turn leads to the higher values of drag predicted by the RUC
model in the dilute region, which includes the spout zone. Recently,
Ren et al. [58] used CFD-DEM to show that the drag force is signicantly higher in the spout region than in the annulus and fountain
regions. Given the signicance of drag force in the spout zone, the
RUC drag function is expected to improve the simulation results
that are underpredicted by the model of Du et al. [25,34].
Another constitutive parameter that affects the particle phase
behavior in the spout zone is the solid shear viscosity. Fig. 2 also
shows a comparison of the expressions suggested by Syamlal et al.
[1] and Gidaspow et al. [48] for solid shear viscosity as a function of
the solid volume fraction under the conditions used in the simulation,
for e = 0.9. It is seen that both models yield similar values for the
solid shear viscosity at high solid volume fractions (s > 0.3). The
Syamlal et al. [1] shear viscosity model, however, leads to signicantly

As noted before, the algebraic expression given by Eq. (5) is used


for evaluating the granular temperature and to compute the solid viscosity that is used in the solid phase momentum equation. As noted in
Section 3.1, the two expressions for solid viscosity given by Eqs. (11)
and (12) behave differently in the dilute zone. These viscosity models,
that are available in the FLUENT code, are used in the simulation and
their effects are evaluated. Fig. 3 shows the predicted contour plots of
solid volume fraction in the bed for e = 0.9, U = 1.2Ums (0.648 m/s)
and with the use of the Gidaspow drag model [30]. The model predictions for the Syamlal et al. [1] and Gidaspow et al. [48] solid shear viscosities are shown in Fig. 3. The experimental and computational
results of fountain heights and maximum particle velocities along
the bed axis are listed in Table 2. Table 2 shows that the CFD model
underpredicts the fountain height and axial particle velocity for
both solid viscosity models when the Gidaspow [30] drag model
is used. However, there are signicant differences in the values of
fountain height and axial particle velocity when either solid viscosity
model is used. Predictions of the Syamlal et al. [1] solid viscosity
model (Fig. 3 and Table 2) are closer to the experimental results
than those of the Gidaspow et al. [48] solid viscosity model. However,
both viscosity models predict typical ow patterns for spouted beds
recognized by the three regions, namely, spout, annulus and fountain.
Fig. 3 clearly shows that the solid volume fraction is low in the spout
and high in the annulus for both viscosity models. Thus, unlike dense
gas-uidized beds, use of appropriate solid shear viscosity for the
dilute region is important for the CFD simulation of spouted beds. It
is found that compared to the Gidaspow et al. [48] viscosity model,
the Syamlal et al. [1] solid viscosity model gives lower values of viscosity and higher values of solid particle velocity, gas volume fraction
in the spout zone and fountain height. Therefore, the solid viscosity
model signicantly impacts the CFD results for spouted beds, especially
in the dilute regions.

and Jackson [57] wall boundary condition is used for the tangential
velocity, and granular temperature of the solid phase at the wall, with
an assumed specularity coefcient of 0.05. Initially, the particle concentration in the spouted bed is specied, and gas velocity inside the
spouted bed is set to zero. The particle concentration in the freeboard
region is also set to zero.

3.3. Effect of restitution coefcient on CFD model predictions

Fig. 5. Contour plots of solid volume fraction for U = 1.2Ums, the RUC drag model, the
viscosity model of Syamlal et al. [1], the full transport equation for granular temperature
and different values of es.

The kinetic theory of granular materials provides explicit constitutive


models that account for energy dissipation due to inelastic particle
particle collisions by means of the restitution coefcient. Sensitivity of
model predictions to variation of the coefcient of restitution has been
reported previously for several uidization systems. Accordingly, as
the coefcient of restitution increases, the granular temperature, solid
pressure and solid viscosity increase [34]. Unfortunately, an accurate
measurement of the restitution coefcient is difcult because its value
depends on particle's material, size, shape and roughness. Hosseini et
al. [32] investigated the effect of the value of the restitution coefcient
in the range of 0.8 to 0.97 in a CFD simulation of conical spouted beds.
Fig. 4 shows the contour plots of the solid volume fraction predicted by
the present model using the algebraic equation for granular temperature,
the RUC model with U = 1.2Ums, and different values of the particle
particle restitution coefcients. The results show that an increase in the
value of the restitution coefcient leads to a decrease in the fountain
height. The CFD model conrms the ndings of Du et al. [34] concerning
the effects of the restitution coefcient on axial particle velocity, granular
temperature and voidage distributions at different bed levels, which are
not reported for the sake of brevity.

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

Lan et al. [24] suggested that the fountain height is a key parameter
that can be used to determine the accuracy of the numerical models for
describing the hydrodynamic behavior of spouted beds. A value of 0.90
for the restitution coefcient leads to a full accordance between the
CFD results and the experimental data [26] in terms of fountain height
(fountain height of 0.25 m for both experimental and computational
results). It should be noted that such a method is used for selecting
the restitution coefcient when evaluating the complete transport
equation for granular temperature in the model.
Fig. 5 shows contour plots of solid volume fraction as predicted by
the model using the full transport equation for granular temperature
and RUC model with U = 1.2Ums and three values of the particle
particle restitution coefcients (e = 0.94, 0.96 and 0.98). The results
reveal that an increase in the value of the restitution coefcient leads
to a decrease in the fountain height and an increase of solid concentration in the spout and fountain zones. From Fig. 13, it is seen that
a value of 0.96 for the restitution coefcient as suggested by Lan et
al. [24], leads to a predicted fountain height which is the same as
the experimental value when the full transport equation for granular
temperature is used.
3.4. Effect of drag function on the CFD model predictions
Here, the algebraic expression given by Eq. (5) for granular temperature, the solid shear viscosity of Syamlal et al. [1], e = 0.9 and
U = 1.2Ums are used. The effects of changing the drag model are analyzed. The drag model of Gidaspow including a switch function and the
RUC drag model are implemented by means of user-dened functions
(UDF) (which are written in C-code and compiled in FLUENT).
Fig. 6 shows the effect of the drag model on model predictions for
the solid volume fraction and axial particle velocity in the spouted
bed. It is seen that the RUC model predicts axial particle velocity
with a maximum value of 5.5 m/s near the nozzle on the bed axis,
which is close to the experimental value of 5.67 m/s [27]. Furthermore, as previously mentioned, the RUC model and e = 0.90 lead to

309

the best agreement with the experimental data in terms of fountain


height, whereas Gidaspow's drag model underpredicts the fountain
height considerably. That is, the use of the RUC drag model signicantly improves the CFD results.
Finally, Fig. 6 shows that the RUC drag model predicts particle ow
behavior better than the Gidaspow model, especially in the spout and
fountain regions. The signicant difference in the prediction of axial
solid velocity in the spout zone by these drag functions conrms the
validity of the quantitative analysis of drag models carried out in
Section 3.1. Therefore, the RUC drag model is used in the remaining
simulations.
3.5. Effect of granular temperature equations
The algebraic expression for calculating the granular temperature
was obtained from the energy equation by Lun et al. [59] by assuming
that granular energy is dissipated locally. That is, the convection and
diffusion terms are ignored; thus the generation and dissipation are
balanced, which leads to an algebraic equation for the granular temperature. Faster convergence and more stable solution are attained
by using the algebraic equation for the granular temperature, with
the calculation time being reduced by approximately 20%.
Based on the Syamlal et al. [1] viscosity model, simulations are
performed using the full transport equation for granular temperature
and the algebraic expression approximation. As mentioned above, in
order to achieve the same fountain height with these approaches, restitution coefcients of 0.96 and 0.9 should be used for the full transport
equation and the algebraic expression, respectively. Fig. 7a shows the
predicted solid volume fraction contours for different gas velocities
when the algebraic granular temperature is used. Fig. 7b shows the
same contours for the case when the full transport equation is solved.
The result obtained using the Gidaspow et al. [48] viscosity model is
also evaluated when the full transport equation is solved, and contours
are shown in Fig. 7c for comparison. Both methods for calculating
granular temperature provide three distinct regions: a central spout,

Fig. 6. Contour plots of solid volume fraction in the bed for U = 1.2Ums, es = 0.9, the algebraic equation for granular temperature, the Syamlal et al. [1] viscosity model and (a): the
RUC drag model [50], and (b): the Gidaspow drag model [30]. Contour plots of axial solid velocity in the bed for U = 1.2Ums, es = 0.9, the algebraic equation for granular temperature, the
Syamlal et al. [1] viscosity model and (c): the RUC drag model [50], and (d): the Gidaspow drag model [30].

310

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

Fig. 7. Contour plots of solid volume fraction for (a): the algebraic granular temperature, the RUC drag model and the Syamlal et al. [1] viscosity model at different gas velocities,
(b): the full transport equation, the RUC drag model and the Syamlal et al. [1] viscosity model at different gas velocities, and (c): the full transport equation, the RUC drag model and
the Gidaspow et al. [48] viscosity model at U = 1.2Ums.

annulus, and a fountain. As conrmed from Fig. 4, a common method for


decreasing fountain height is to increase the restitution coefcient. For a
xed coefcient of restitution, use of the full transport equation for
granular temperature leads to signicantly higher fountain height compared to that of the model that uses the algebraic equation. Fig. 7 also
shows considerable differences in the predicted values of solid volume
fraction in the fountain and spout zones calculated by using the transport equation and the algebraic equation. That is, the CFD results for
the solid concentration in the fountain and spout zones given by the
transport equation are signicantly lower than those corresponding to
the algebraic equation. The predicted particle ow patterns in the core
and periphery region of the fountain are also quite different.
Fig. 7c shows the solid volume fraction contours using the full
transport equation for granular temperature and the Gidaspow et al.
[48] viscosity model with e = 0.9. Here the simulation signicantly
underpredicts the fountain height and particle velocity along the
bed axis, even for low values of a restitution coefcient of 0.9.
Furthermore, the solid shear viscosity of Gidaspow et al. [48] overpredicts the solid concentration in the spout zone and, therefore,
underpredicts the axial particle velocity in that region. However, the
general ow pattern in this case is reasonably well predicted. From
these discussions, it is concluded that the solid viscosity model is
a critical parameter for CFD simulation of spouted beds for both
cases: when an algebraic expression for granular temperature is considered or the full transport equation for the granular temperature is
used.
When an algebraic expression for the granular temperature is
used, Fig. 7a shows that there is a narrow, denser zone surrounding
the axis near the top of the spout and at the fountain center. That is,
the use of the algebraic expression for the granular temperature
leads to a much higher solid concentration along the axis in the fountain zone, compared with that obtained with the use of the transport
equation for the granular temperature. As observed in Fig. 7a and b,
an increase in the gas ow rate leads to a decrease in local solid concentration (an increase in local voidage) over the entire spout region,

which is consistent with the trend reported by He et al. [26,27].


Despite the concentration differences, both methods lead to a similar
fountain height for the same gas velocity, which is consistent with the
experimental data.

Fig. 8. Contour plots of solid volume fraction in the bed for U = 1.2Ums, es = 0.9, the
algebraic equation for granular temperature, the Syamlal et al. [1] viscosity model and
(a): proposed discretization scheme in this study, and (b) the rst order discretization
scheme.

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

3.6. Effect of discretization scheme


Discretization is a process by which the governing partial differential equations are converted to algebraic equations for numerical solution. As noted before, we used the second-order upwind scheme for the
discretization of the momentum equation, and transport equations
for turbulence kinetic energy and turbulence dissipation rate. The
rst-order upwind scheme was used only for the volume fraction
term. Such a choice has been already used by several authors including
Hosseini et al. [61], Bettega et al. [62], and Behjat et al. [63] for CFD simulation of uidization systems. Fig. 8 shows a comparison between the
selected schemes in the present study and the rst-order discretization
for all terms which was recently used by Lan et al. [24]. Here for
the present study, the viscosity model of Syamlal et al. [1], algebraic
granular temperature, restitution coefcient of 0.9, and the RUC drag
model for U = 1.2Ums are used. This gure shows that with the use
of a second-order discretization scheme, the fountain height is reasonably predicted. Moreover, the maximum value for the axial particle velocity near the nozzle on the bed axis is 4.25, 5.5 and 5.67 m/s,
respectively, for the rst-order upwind scheme, second-order scheme

311

used in the current work, and the experimental data [27]. Clearly, the
rst-order scheme underpredicts the fountain height and axial particle
velocity. The second-order scheme, however, predicts results that are
much closer to the experimental data. It is perhaps worth mentioning
that for simulation of a conical spouted bed, Wang [64] showed that
the modeling results of the second-order scheme for the momentum
equation are in fair agreement with the experimental ndings,
compared to the rst-order scheme. Therefore, the use of secondorder discretization scheme for CFD simulation of spouted bed is
recommended.
3.7. Investigation of several hydrodynamic parameters in the spout and
fountain regions
Fig. 9 shows the model predictions for the radial voidage distribution
at different fountain levels for U = 1.1Ums, U = 1.2Ums and U = 1.3Ums.
Here, the viscosity model of Syamlal et al. [1], algebraic granular temperature, a restitution coefcient of 0.9, and the RUC drag model are used.
The experimental data of He [60] are reproduced in this gure for comparison. In Fig. 9, ZF is the vertical coordinate in the fountain measured

Fig. 9. Experimental and computational results for the radial voidage distribution at different fountain levels and different gas velocities.

312

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

from the bed surface. It is observed from both computational results and
experimental data that the solid volume fraction is higher in the core region and gradually decreases as the radial distance from the axis increases. The solid concentration in the core region also decreases as the
fountain level increases, except near the top of the fountain, where the
trend is reversed. According to Fig. 9, the solid concentration in the fountain zone is sensitive to gas velocity; that is, the concentration decreases
as spouting gas velocity increases. The CFD results of the full transport
equation related to solid concentration in the fountain zone are not
shown in Fig. 9b to avoid making the gure too crowded. Fig. 7b, however, shows that the full transport equation for granular temperature with
the viscosity model of Syamlal et al. [1] underpredicts the solid concentration in the fountain zone.
The particle dense zone is most noticeable for U = 1.1Ums with a
uniform value in the core region of the fountain.
Fig. 10 compares the experimental prole of the radial voidage
distribution at different levels in the spout for U = 1.2Ums (Fig. 10a)
as reported by He [60], the proles predicted by using the algebraic
equation (Fig. 10b), and the full transport equation (Fig. 10c) for granular temperature in the CFD model. It is seen that as the spout level and
radial distance from the spout axis increase, the local voidage decreases.
Both methods for calculating granular temperature, but with different e
values, provide trends that are similar to the experimental data. That is,
both the algebraic and the transport equations for granular temperature

together with the Syamlal et al. [1] viscosity model lead to reasonable
results in the spout region of cylindrical spouted beds when the appropriate coefcient of restitution is used.
Fig. 11 compares the simulated radial proles of axial particle
velocities at different levels in the spout for various gas ow rates of
U = 1.1Ums, U = 1.2Ums and U = 1.3Ums with the experimental
data of He [60]. Here, both the algebraic and full transport equations
for the granular temperature with the restitution coefcients of 0.96
and 0.9, for the full transport equation and the algebraic expression
respectively, are used in the CFD simulations. This gure shows that
the particle velocities peak at the bed axis for all bed levels and
decrease to zero at the spoutannulus interface. Fig. 11 shows that the
present model's simulation results for axial particle velocity proles at
different bed levels are consistent with the experimental data, whereas
only qualitative agreements were reported in the earlier two-uid
models of [15,24,34,36].
Model predictions for the particle velocities along the bed axis are
shown in Fig. 12 and are compared with the experimental data of He
[60]. It is seen that the algebraic granular temperature model with
e = 0.9 and the full transport equation for granular temperature
with e = 0.96 predict results in the spout zone which are in reasonable agreement with the experimental data of He [60]. The simulation
results show that the particles in cylindrical spouted beds are rapidly
accelerated to their maximum velocity at a height near the inlet zone;

Fig. 10. The radial voidage distribution at different levels in the spout region for U = 1.2Ums (a): the experimental data, (b): predicted results of the CFD model including the
algebraic equation, and (c): predicted results of the CFD model including the full transport equation.

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

313

Fig. 11. The experimental and simulated radial distributions of axial particle velocity at different levels in the spout for various gas ow rates.

then they gradually decelerate, and their velocity drops sharply to


zero in the fountain region. This trend is greatly different from that
observed in conical spouted beds. Hosseini et al. [32] showed that in

the conical spouted beds the particles are quickly accelerated near
the inlet zone, and then the particle velocity prole stays roughly
at for a conical spouted bed.

314

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

higher than the experimental data, there are qualitative agreements.


(The differences observed could be due to the selected frictional
stresses.) The spout diameter is the key parameter in spouted
beds, and Fig. 13 shows that the present model provides a reasonable prediction of spout diameter and the hydrodynamics of cylindrical spouted beds.
4. Conclusions

Fig. 12. The experimental and computed particle velocities along the bed axis for
various gas ow rates.

Finally, Fig. 13 compares the simulated spout diameters calculated using the algebraic equation for the granular temperature and
the Syamlal et al. [1] viscosity model for the gas ow rates of
1.1Ums and 1.3Ums. The spout diameter is determined by tracing
particle velocity over the entire spout and annulus regions and
detecting the point where particle velocities pass through zero.
That is, the corresponding radial positions at different bed levels
are the boundary between the spout and annulus regions. The experimental data of He [60] are reproduced in Fig. 13 for comparison.
As can be seen in this gure, the spout diameter increases with bed
level in all cases, and although the simulated results are slightly

Fig. 13. The experimental and simulated spout diameters calculated by the algebraic
equation for the granular temperature and the Syamlal et al. [1] viscosity model at
two gas ow rates of 1.1Ums and 1.3Ums.

The hydrodynamics of a cylindrical spouted bed was studied using


a EulerianEulerian Two-Fluid Model (TFM) incorporating the kinetic
theory of granular ow. The parameters for modeling the drag function
and the solid shear viscosity, as well as the granular temperature, have
been investigated in order to optimize the CFD model predictions. The
simulation results showed the signicant inuence of the expressions
for drag and the solid shear viscosity on the model predictions, especially
in the dilute region of the bed.
Comparison of the results with the experimental data suggests
that the representative unit cell (RUC) drag model and the Syamlal et
al. [1] viscosity model are most suitable for prediction of gas and solid
ow patterns in cylindrical spouted beds, particularly, for solid axial
velocity and voidage in the spout zone. In addition, it was shown that
the algebraic equation for granular temperature with the appropriate
coefcient of restitution can accurately predict the solid ow pattern
in the spout and fountain zones at considerable computational economy compared to the use of full transport equation for granular temperature. It was shown that the simulated axial solid velocity, voidage,
spout diameter, solid concentration in the fountain zone and axial
solid velocity along the bed axis for various gas ow rates are in close
agreement with the available experimental data. The numerical results
also showed that the discretization scheme signicantly affects the
computational model prediction for the fountain height and axial
particle velocity. Analysis of the effect of using different frictional stress
models is left for a future study.
Nomenclature
parameter in the RUC model []
ARUC
parameter in the RUC model []
BRUC
drag coefcient []
CD
particle diameter [m]
ds
diameter of the spout gas inlet [mm]
Ds
diameter of the bed
Dt
restitution coefcient []
es
g
acceleration due to gravity [m/s2]
radial distribution coefcient []
g0
H
vessel height [mm]
Static bed depth [mm]
H0
I
stress tensor []
second invariant of the deviatoric stress tensor []
I2D
diffusion coefcient for granular energy [kg/m s]
ks
gas/solid momentum exchange coefcient [kg/m3 s]
Kgs
Kgs,Ergun gas/solid momentum exchange coefcient by Ergun equation [kg/m3 s]
Kgs,WenYu gas/solid momentum exchange coefcient calculated by
WenYu equation [kg/m3 s]
P
pressure []
solid pressure []
Ps
r
radial coordinate []
R
radius []
Reynolds number []
Res
t
time []
U
supercial gas velocity [m/s]
minimum spouting velocity [m/s]
Ums
velocity [m/s]
vi
z
height coordinate measured from the distributor [m]
uctuating particle velocity of the particulate phase [m/s]
v s

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

Greek letters
volume fraction []
i
the collisional dissipation of energy [kg/s3 m]
s
granular temperature [m2/s2]
s
solid bulk viscosity [kg/m s]
s
shear viscosity [kg/m s]
i
density [kg/m3]
i
i
stress tensor [Pa]

angle of internal friction [deg]


transfer rate of kinetic energy [kg/s3 m]
gs

Subscripts
fr
friction
g
gas
i
general index
ms
minimum spouting
p
particle
q
phase type (solid or gas)
s
solids
T
stress tensor

References
[1] M. Syamlal, W. Rogers, T.J. O'Brien, MFIX Documentation: Volume 1, Theory Guide,
National Technical Information Service, Springeld, VA, 1993. (DOE/METC-9411004,
NTIS/DE9400087).
[2] K.B. Mathur, P.E. Gishler, A technique for contacting gases with coarse solid particles,
AICHE Journal 1 (1955) 157164.
[3] R.G. Szafran, A. Kmiec, CFD modeling of heat and mass transfer in a spouted bed
dryer, Industrial and Engineering Chemistry Research 43 (2004) 11131124.
[4] H. Ichikawa, M. Arimoto, Y. Fukumori, Design of microcapsules with hydrogel as a
membrane component and their preparation by spouted bed, Powder Technology
130 (2003) 189192.
[5] L.A.P. Freitas, J.T. Freire, Experimental study on the dynamics of a spouted bed
with particle feed through the base, Brazilian Journal of Chemical Engineering
14 (1997) 269280.
[6] K. Jono, H. Ichikawa, M. Miyamoto, Y. Fukumori, A review of particulate design for
pharmaceutical powders and their production by spouted bed coating, Powder
Technology 113 (2000) 269277.
[7] M. Olazar, R. Aguado, J.L. Snchez, R. Bilbao, J. Arauzo, Thermal processing of
straw black liquor in uidized and spouted bed, Energy & Fuels 16 (2002)
14171424.
[8] M. Olazar, M.J. San Jos, G. Zabala, J. Bilbao, New reactor in jet spouted bed regime
for catalytic polymerizations, Chemical Engineering Science 49 (1994) 45794588.
[9] M. Olazar, R. Aguado, A. Barona, J. Bilbao, Pyrolysis of sawdust in a conical spouted
bed reactor with a HZSM-5 catalyst, AICHE Journal 46 (2000) 10251033.
[10] Z. Deng, R. Xiao, B. Jin, H. Huang, L. Shen, Q. Song, Q. Li, Computational uid
dynamics modeling of coal gasication in a pressurized spoutuid bed, Energy
& Fuels 22 (2008) 15601569.
[11] G. Ahmadi, M. Farshad, On the continuum theory of soliduid mixture a
superimposed model of equipresent constituents, Indian Journal of Technology
12 (1974) 195198.
[12] J.J. Monaghan, A. Kocharya, SPH simulation of multi-phase ow, Computer Physics
Communications 87 (1995) 225235.
[13] J. Ma, W. Ge, Q. Xiong, J. Wang, J. Li, Direct numerical simulation of particle
clustering in gassolid ow with a macro-scale particle method, Chemical Engineering
Science 64 (2009) 4351.
[14] Q. Xiong, B. Li, F. Chen, J. Ma, W. Ge, J. Li, Direct numerical simulation of sub-grid
structures in gassolid owGPU implementation of macro-scale pseudo-particle
modeling, Chemical Engineering Science 65 (2010) 53565365.
[15] L. Wang, G. Zhou, X. Wang, Q. Xiong, W. Ge, Direct numerical simulation of particle
uid systems by combining time-driven hard-sphere model and lattice Boltzmann
method, Particuology 8 (2010) 379382.
[16] Q. Xiong, B. Li, G. Zhou, X. Fang, J. Xu, J. Wang, X. He, X. Wang, L. Wang, W. Ge, J. Li,
Large-scale DNS of gassolid ows on mole-8.5, Chemical Engineering Science 71
(2012) 422430.
[17] Q. Xiong, B. Li, J. Xu, X. Wang, L. Wang, W. Ge, Efcient 3D DNS of gassolid ows
on Fermi GPGPU, Computers and Fluids 70 (2012) 8694.
[18] H. Nasr, G. Ahmadi, J.B. Mclaughlin, A DNS study of effects of particleparticle
collisions and two-way coupling on particle deposition and phasic uctuations,
Journal of Fluid Mechanics 640 (2009) 507536.
[19] K. Sankaranarayanan, S. Sundaresan, Lattice Boltzmann simulation of two-uid model
equations, Industrial and Engineering Chemistry Research 47 (2008) 91659173.
[20] S. Melchionna, Incorporation of smooth spherical bodies in the lattice Boltzmann
method, Journal of Computational Physics 230 (2011) 39663976.

315

[21] Q. Xiong, L. Deng, W. Wang, W. Ge, SPH method for two-uid modeling of particle
uid uidization, Chemical Engineering Science 66 (2011) 18591865.
[22] S. Abu-Zaid, G. Ahmadi, A simple kinetic model for rapid granular ows including frictional losses, ASCE Journal of Engineering Mechanics 116 (1990)
379389.
[23] G. Ahmadi, D. Ma, A thermodynamical formulation for dispersed multiphase
turbulent ows, part I: basic theory, International Journal of Multiphase Flow
16 (1990) 323340.
[24] X. Lan, C. Xu, J. Gao, M. Al-Dahhan, Inuence of solid-phase wall boundary condition
on CFD simulation of spouted beds, Chemical Engineering Science 69 (2012) 419430.
[25] W. Du, X.J. Bao, J. Xu, W.S. Wei, Computational uid dynamics (CFD) modeling of
spouted bed: assessment of drag coefcient correlations, Chemical Engineering
Science 61 (2006) 14011420.
[26] Y.L. He, C.J. Lim, J.R. Grace, J.X. Zhu, S.Z. Qin, Measurements of voidage proles in
spouted beds, Canadian Journal of Chemical Engineering 72 (1994) 229234.
[27] Y.L. He, C.J. Lim, J.R. Grace, Particle velocity proles and solid ow patterns in
spouted beds, Canadian Journal of Chemical Engineering 72 (1994) 561568.
[28] M. Syamlal, T.J. O'Brien, Simulation of granular layer inversion in liquid uidized
beds, International Journal of Multiphase Flow 14 (1988) 473481.
[29] H. Arastoopour, P. Pakdel, M. Adewumi, Hydrodynamic analysis of dilute gassolids
ow in a vertical pipe, Powder Technology 62 (1990) 163170.
[30] D. Gidaspow, Multiphase Flow and Fluidization, Academic Press, San Diego, 1994.
[31] S.H. Hosseini, M. Zivdar, R. Rahimi, CFD simulation of gassolid ow in a spouted
bed with a non-porous draft tube, Chemical Engineering and Processing 48
(2009) 15391548.
[32] S.H. Hosseini, G. Ahmadi, B.S. Razavi, W. Zhong, Computational uid dynamic
simulation of hydrodynamic behavior in a two-dimensional conical spouted
bed, Energy & Fuels 24 (2010) 60866098.
[33] J.M. Dalla Valle, Micromeritics, Pitman, London, 1948.
[34] W. Du, X.J. Bao, J. Xu, W.S. Wei, Computational uid dynamics (CFD) modeling of
spouted bed: inuence of frictional stress, maximum packing limit and coefcient
of restitution of particles, Chemical Engineering Science 61 (2006) 45584570.
[35] S.J. Zhang, A.B. Yu, Computational investigation of slugging behaviour in
gas-uidised beds, Powder Technology 123 (23) (2002) 147165.
[36] T. Li, Y. Zhang, J.R. Grace, X. Bi, Numerical investigation of gas mixing in gassolid
uidized beds, AICHE Journal 56 (2010) 22802296.
[37] T. Li, J.R. Grace, X. Bi, Study of wall boundary condition in numerical simulations
of bubbling uidized beds, Powder Technology 203 (2010) 447457.
[38] A. Almuttahar, F. Taghipour, Computational uid dynamics of high density circulating
uidized bed riser: study of modeling parameters, Powder Technology 185 (1) (2008)
1123.
[39] L. Huilin, H. Yurong, L. Wentie, D. Jianmin, D. Gidaspow, J. Bouillard, Computer
simulations of gassolid ow in spouted beds using kineticfrictional stress
model of granular ow, Chemical Engineering Science 59 (2004) 865878.
[40] W. Shuyan, L. Xiang, L. Huilin, Y. Long, S. Dan, H. Yurong, D. Yonglong, Numerical
simulations of ow behavior of gas and particles in spouted beds using
frictional-kinetic stresses model, Powder Technology 196 (2009) 184193.
[41] M.J. San Jos, M. Olazar, S. Alvarez, M.A. Izquierdo, J. Bilbao, Solid cross-ow into
the spout and particle trajectories in conical spouted beds, Chemical Engineering
Science 53 (1998) 35613570.
[42] B.G.M. van Wachem, J.C. Schouten, R. Krishna, C.M. van den Bleek, J.L. Sinclair,
Comparative analysis of CFD models of dense gassolid systems, AICHE Journal
47 (2001) 10351051.
[43] S.H. Hosseini, W. Zhong, M.N. Esfahany, L. Pourjafar, S. Azizi, CFD simulation of the
bubbling and slugging gassolid uidized beds, Journal of Fluids Engineering
(ASME) 132 (2010) 4130141311.
[44] W. Zhonghua, A.S. Mujumdar, CFD modeling of the gasparticle ow behavior in
spouted beds, Powder Technology 183 (2008) 260272.
[45] J. Link, C. Zeilstra, N. Deen, H. Kuipers, Validation of a discrete particle model in a
2D spout uid bed using non intrusive optical measuring techniques, Canadian
Journal of Chemical Engineering 82 (2004) 3036.
[46] Z. Wang, H.T. Bi, C.J. Lim, CFD simulation of spouted beds using a pressure source
term, Industrial and Engineering Chemistry Research 49 (2010) 50535060.
[47] C.R. Duarte, V.V. Murata, M.A.S. Barrozo, A study of the uid dynamics of the spouted
bed using CFD, Brazilian Journal of Chemical Engineering 22 (2005) 263270.
[48] D. Gidaspow, R. Bezburuah, J. Ding, Hydrodynamics of circulating uidized beds.
Kinetic theory approach, uidization VII, Proceedings of the Seventh Engineering
Foundation Conference on Fluidization, 1992, pp. 7582.
[49] G. Schaeffer, Instability in the evolution equations describing incompressible
granular ow, Journal of Difference Equations 66 (1987) 1950.
[50] J.P. Du Plessis, Analytical quantication of coefcients in the Ergun equation for
uid friction in a packed bed, Transport in Porous Media 16 (1994) 189207.
[51] S. Ergun, Fluid ow through packed columns, Chemical Engineering Progress 48
(1952) 8994.
[52] D.A. Santos, G.C. Alves, C.R. Duarte, M.A.S. Barrozo, Disturbances in the hydrodynamic behavior of a spouted bed caused by an optical ber probe: experimental
and CFD study, Industrial and Engineering Chemistry Research 51 (2012)
38013810.
[53] B.M. Halvorsen, J.P. Du Plessis, S. Woudberg, The performance of drag models on
ow behaviour in the CFD simulation of a uidized bed, in: M. Rahman, C.A.
Brebbia (Eds.), Proceedings of the Sixth International Conference on Advances
in Fluid Mechanics (AFM VI), Skiathos, Greece, 810 May (2006), Advances in
Fluid Mechanics VI, vol. 52, WIT Press, UK, 2006, pp. 312.
[54] E. Esmaili, N. Mahinpey, Adjustment of drag coefcient correlations in three
dimensional CFD simulation of gassolid bubbling uidized bed, Advances in
Engineering Software 42 (2011) 375386.

316

S.H. Hosseini et al. / Powder Technology 246 (2013) 303316

[55] W. Du, Computational Fluid Dynamics (CFD) modeling and scaling up studies of
spouted beds, China University of Petroleum, Beijing, China, 2006.
[56] L. Cammarata, P. Lettieri, G.D.M. Micale, D. Colman, 2D and 3D CFD simulations of
bubbling uidized beds using EulerianEulerian models, International Journal of
Chemical Reactor Engineering 1 (2003) 4855.
[57] P.C. Johnson, R. Jackson, Frictionalcollisional constitutive relations for granular
materials with application to plane shearing, Journal of Fluid Mechanics 176
(1987) 6793.
[58] B. Ren, W. Zhong, B. Jin, Z. Yuan, Y. Lu, Computational Fluid Dynamics (CFD)
Discrete Element Method (DEM) simulation of gassolid turbulent ow in a cylindrical
spouted bed with a conical base, Energy & Fuels 25 (2011) 40954105.
[59] C.K.K. Lun, S.B. Savage, D.J. Jeffrey, N. Chepurniy, Kinetic theories for granular
ow: inelastic particles in Couette ow and slightly inelastic particles in a general
ow eld, Journal of Fluid Mechanics 140 (1984) 223256.

[60] Y.L. He, Hydrodynamic and Scale-up Studies of Spouted Beds, University of British
Columbia, 1995 , (Ph.D. Thesis).
[61] S.H. Hosseini, R. Rahimi, M. Zivdar, A. Samimi, CFD simulation of gassolid bubbling
uidized bed containing FCC particles, Korean Journal of Chemical Engineering 26
(2009) 14051413.
[62] R. Bettega, C.A. da Rosa, R.G. Correa, J.T. Freire, Fluid dynamic study of a semicylindrical
spouted bed: evaluation of the shear stress effects in the at wall region using computational uid dynamics, Industrial and Engineering Chemistry Research 48 (2009)
1118111188.
[63] Y. Behjat, S. Shahhosseini, M. Ahmadi Marvast, Modeling gas oil spray coalescence
and vaporization in gas solid riser reactor, International Communications in Heat
and Mass Transfer 37 (2010) 935943.
[64] Z. Wang, Experimental Studies and CFD Simulations of Conical Spouted Bed
Hydrodynamics, University of British Columbia, 2006 , (Ph.D. Thesis).

You might also like