Numerical Simulations of Hydrodynamic Behaviors in Conical Spouted Beds

Z.G. Wang, H.T. Bi and C.J. Lim

Fluidization Research Centre
Department of Chemical and Biological Engineering
University of British Columbia
Vancouver, BC, V6T 1Z4, Canada
Abstract
By comparison with experimental data on the radial distribution of the static pressure
and vertical particle velocity in conical spouted beds, numerical simulations show that, among
all factors investigated, the axial solid phase source term has the most significant influence on
static pressure profiles, followed by the restitution coefficient, while other factors almost have
no effect. Apart from the solid bulk viscosity, almost all other factors affect the radial
distribution of the axial particle velocity, although the influence of the axial solid phase source
term is less significant. For complex systems such as conical spouted beds, the new approach
proposed in this work by properly selecting the axial solid phase source term shows a great
potential to improve the CFD simulation of spouted beds.
Keywords: Conical spouted bed; Hydrodynamics; Numerical simulation; Multiphase model
1. Introduction
Due to the vigorous solids motion and intimate gas-solids contact, conical spouted
beds have been commonly used for drying suspensions, solutions and pasty materials (Pham,
1983; Markowski, 1992; Passos et al., 1997, 1998; Reyes et al., 1998). Many other
applications have also been under research and development, such as catalytic partial
oxidation of methane to produce synthesis gas (Marnasidou et al., 1999), incineration of
waste-materials, coating of medical tablets (Kucharski & Kmiec, 1983), coal gasification and
liquefaction (Uemaki & Tsuji, 1986), pyrolysis of sawdust or mixtures of wood residues
(Aguado et al., 2000a, 2000b; Olazar et al., 2000a, 2000b, 2001).
Comparing with other multiphase systems, such as fixed beds or packed beds,
fluidized beds, circulating fluidized beds (co-current upflow) and co-current downflow fluidized
beds (downer), numerical simulations of spouted beds, especially conical spouted beds, have
received less attention and reported results are still insufficient and controversial. Duarte et al.
(2005) simulated a cylindrical spouted bed as described by He ea al. (1994a, 1994b). In their
simulation, the parabolic velocity profile which is typical for the laminar flow was used as the
turbulent inlet boundary condition, and the average velocity used was higher than the actual
operating condition. By using empirical correlations for granular viscosity (originally obtained
from experiments for FCC particles) and elasticity modulus, Huilin et al. (2001) and He et al.
(2004) simulated experiments of He et al. (1994a, 1994b). Huilin et al. (2001) also simulated
experiments for a conical spouted bed operated by San Jose et al. (1998). Based on their
descriptions, the uniform velocity profile was applied for the turbulent inlet boundary condition.
Taking the kinetic theory approach for granular flow, Lu et al. (2004) simulated experiments of
He et al. (1994a, 1994b) for cylindrical spouted beds and experiments of San Jose et al.
(1998) for a conical spouted bed, with the uniform velocity profile applied for the turbulent inlet
boundary condition. Kawaguchi et al. (2000) simulated experiments of He et al. (1994b) too. In
their simulation, particle motion was treated discretely by solving Newtons equation of motion
for each particle, the discrete element method (DEM) was employed to model the interaction
between particles, with a uniform velocity profile applied to the inlet.
It is well known that, for a cylindrical multiphase system, as the fluid velocity is
increased, the pressure drop across the bed increases before the minimum fluidization velocity
is reached, and the multiphase system is operated as a packed bed or fixed bed with the
pressure drop of the bed being well described by the Ergun equation. Once the minimum
fluidization velocity has been achieved, particles are in a dynamically suspended condition,
and the pressure drop across the bed will remain constant within a wide range of fluid velocity,
being equal to the effective weight of the bed per unit area. Because of the difference of the
suspended state between a packed bed and a fluidized bed, approaches used to simulate
packed beds and fluidized beds are quite different. For example, there are two options in
commercial software FLUENT specially designed for simulating packed beds.
For spouted beds, the bed structure is quite different from conventional fluidized beds
and packed beds. At stable spouting, a spouted bed consists of three regions, a spout in the
center, a fountain above the bed surface and an annulus between the spout and the wall. The
spout and the fountain are similar to fluidized beds with particles dynamically suspended, while
the annulus region is more like a packed bed or moving bed. At partial spouting, there are only
two distinct regions, an internal spout that is similar to a fluidized bed and the surrounding
packed particle region which is similar to a packed bed. Thus, original codes used for
simulating fluidized beds or packed beds in commercial softwares are unsuitable for the
simulation of spouted beds. Furthermore, at stable spouting, it was found that the maximum
ratio of the spouting to the fluidization pressure drop is about 0.64-0.75 for cylindrical spouted
beds (Mathur and Epstein, 1974); for conical spouted beds, the ratio of the spouting to the
fluidization pressure drop is about 0.4 based on our recent experimental data as well as those
from Mukhlenov and Groshtein (1964, 1965), well below the pressure drop under fluidized
conditions. This suggests that particles are not in fully suspended state even at stable
spouting, the inclined wall as well as the base of the bed exert some kind of supporting forces
on particles in the upper region. From this point of view, for spouted beds, the general gravity
term in the vertical momentum equation for the granular phase in commercial software for
simulating fluidized beds seems to be not applicable, and thus needs to be modified.
For conical spouted beds, because of the existence of the inclined wall, the radial
momentum equation for the granular phase should also be modified. For cylindrical spouted
beds, if the static bed height is higher enough, the conical base takes only a small part of the
whole bed, and the modification of the radial momentum equation can be neglected. Whereas,
considering the complexity of the flow structure of conical spouted beds, we only focus on the
vertical momentum equation for granular phase in the annulus in the current simulation.
Several studies have been reported on the effect of factors on simulation results of
gas-solids fluidized beds. Goldschmidt et al. (2001) reported that the hydrodynamics of dense
gas-fluidized beds strongly depends on the amount of energy dissipated due to particle
collisions or the restitution coefficient. McKeen and Pugsley (2003) simulated a freely bubbling
bed of fluid catalytic cracking (FCC) catalyst using four kinds of drag models, and found that
the same degree of bed expansion was achieved using different drag equations. Therefore,
they concluded that simply choosing a different drag model could not improve predictions, and
a modified drag model by introducing a scale factor had to be applied in their simulations in
order to bring their simulation result close to their experimental data. They also investigated
the effect of the mesh size on simulation results. Yang et al. (2003) simulated a circulating
fluidized bed using both the Gidaspow drag model and a modified Gidaspow drag model
based on a new voidage function derived from their EMMS (the Energy-Minimization Multi-
Scale) approach. It was found that the dynamic formation and dissolution of clusters could be
captured using their modified drag model, with the simulated outlet solid flux and voidage
profile in both radial and axial directions being in reasonable agreement with experimental
results. Lu et al. (2004) investigated the effect of the restitution coefficient and the frictional
viscosity in their simulation of San Jose et al. (1998) experimental data. It was shown that both
the restitution coefficient and the frictional viscosity had effects on the radial voidage
distribution to some extent.
The objective of this study is to investigate the effects of those factors, especially the
solids phase source term, on CFD simulation results of a conical spouted bed at stable
spouting state. In the simulation, the fluidized beds approach is used with user-defined
functions (UDFs) to modify different gravity terms in the annulus region, different fluid inlet
profiles and different fluid-solid exchange coefficients. Experimental data on the distribution of
the static pressure, vertical solid velocity, and solid fraction are then used to validate simulation
results.
2. Governing Equations
Currently there are two approaches for the numerical calculation of multiphase flows:
the Euler-Lagrange approach and the Euler-Euler approach.
In the Euler-Lagrange approach, the fluid phase is treated as a continuum by solving
the time averaged Navier-Stokes equations, while the dispersed phase is solved by tracking a
large number of particles (or bubbles, droplets) through the calculated flow field. The dispersed
phase can exchange momentum, mass, and energy with the fluid phase. A fundamental
assumption made in this approach is that the dispersed second phase occupies a low volume
fraction.
In the Euler-Euler approach, the different phases are treated mathematically as
interpenetrating continua. Since the volume of a phase cannot be occupied by the other
phases, the concept of phasic volume fraction is introduced. These volume fractions are
assumed to be continuous functions of space and time and their sum is equal to one.
In FLUENT, three different Euler-Euler multiphase models are available. For granular
flows, such as flows in risers, fluidized beds and other suspension systems, the Eulerian
multiphase model is always the first choice, so does our simulation on gas-solid conical
spouted beds.
Based on the general description of the Eulerian multiphase model, by simplification,
following governing equations can be derived.
Assumptions:
No mass transfer between spouting gas and glass bead particles;
External body force, lift force, as well as virtual mass force are ignored;
Density of each phase is a constant.
Continuity equation for phase q (both fluid phase l and solid phase s):
0 ) ( ) ( = +

v
t
q
q q
(1)
where
vq
is the velocity of phase q;
q
is the volume fraction of phase q, and the following
condition holds.
1
1
=
=
n
q
q

(2)
where n is the total number of phases, and n=2 in our simulation.
Conservation equation of momentum:
For the fluid phase l:
) ( ) ( ) (
v v
K
g P
v v v
t
l s
sl
l
l
l
l
l l l
l
l l
l
+ + + = +

(3)
For the solid phase s:
S v v
K
g
P
P
v v v
t
s s l
ls
s
s
s
s s
s s s
s
s s
s
+ + + + = +

) ( ) ( ) (

(4)
where
l
is the density of the fluid phase, P is the static pressure shared by all phases,

l
is
the fluid phase stress-strain tensor, g is the gravitational acceleration, K
ls
=K
sl
is the
momentum exchange coefficient between fluid phase l and solid phase s,
s
is the density of
the particle,

s
is the solid phase stress-strain tensor, P
s
is the solid pressure,
Ss
is the solid
phase source term.
The stress-strain tensor for phase q:
I
v v v
q q
q q
T
q q q
q
q
+ + = )
3
2
( ) (

(5)
where
q
and
q
are the shear and bulk viscosity of phase q.
For the solid phase s, the solids shear viscosity is the sum of the collisional viscosity,
kinetic viscosity and the optional frictional viscosity, as shown in equation (6).
+
+ =
fr s kin s col s s , , ,
(6)
The collision viscosity is modeled as:
) (
) 1 (
5
4
2 1
, 0 ,

+ =
s
ss
ss
s
s
s
col s
e
g
d
(7)
where d
s
is the diameter of the solid particles, g
0,ss
is the radial distribution function, and
FLUENT employs the following expression as equation (8), e
ss
is the coefficient of restitution,

s
is the granular temperature.
(
(

=
|
|
|
.
|

\
|

max ,
3 1
1
, 0
1
s
s
ss
g
(8)
The following expression from Gidaspow (1994) is used to estimate the kinetic
viscosity.
(

+ +
+

= ) 1 (
5
4
1
) 1 ( 96
10
, 0
2
, 0
,
e
g
g
e
d
ss s
ss
ss
ss s
s s s
kin s

(9)
In our simulation, the frictional viscosity was not considered, and the solid bulk
viscosity used either the following form from Lun et al. (1984) or a constant value of zero.
) (
) 1 (
3
4
2 1
, 0

+ =
s
ss
ss
s
s
s s e
g
d
(10)
Fluid-solid exchange coefficients:
The fluid-solid exchange coefficient K
sl
can be written in the following general form:

s
s
s
sl
f
K
=
(11)
where f is defined differently for the different exchange coefficient models, and
s
, the
particulate relaxation time, is defined as

18
2
l
s s
s
d
=
(12)
In FLUENT, there are three models for fluid-solid exchange coefficient, and the
Gidaspow drag model is applied in this work.
Gidaspow drag model (1994):
The Gidaspow model is a combination of the Wen and Yu model (1966) and the Ergun
(1952) equation.
When
8 . 0 >
l
, the fluid-solid exchange coefficient K
sl
is of the following form:

65 . 2
4
3

=
l
s
l s
l
l
s
D sl
d
v v
C K
(13)
where,
(

+ = )
Re
( 15 . 0 1
Re
24
687 . 0
s
l
s
l
D

C
(14)

l
l s s l
s
v v d

=
Re
(15)
When
8 . 0
l

d
v v
d
K
s
l s
s
l
s
l
l
l s
sl

75 . 1
) 1 (
150
2
(16)
Solids Pressure:
For granular flows in the compressible regime (i.e., where the solids volume fraction is
less than its maximum allowed value), a solids pressure is calculated independently and used
for the pressure gradient term,
P
s

, in the solid phase momentum equation. The solids

pressure is composed of a kinetic term and a second term due to particle collisions, as shown
in equation (17) (Fluent Inc., 2003b).

+ +

=
s
ss
s ss
s
s
s
s s
g
e P
, 0
2
) 1 ( 2

(17)
The solid phase source term:
It was found that the ratio of the pressure drop at stable spouting to the pressure drop
at stable fluidization is usually smaller than one for both the cylindrical spouted beds and the
conical spouted beds (Mathur and Epstein, 1974; Mukhlenov and Groshtein, 1964, 1965). At
partial spouting state, however, the above ratio is usually bigger than one in the ascending
process. Thus, the gravity term in the axial solid phase momentum equation, equation (4),
needs to be modified. To do so, a solid phase source term, S
s,a
, is introduced into the axial
solid phase momentum equation for both the spout and annulus regions:
When
8 . 0
l
and
H
Z
0

(in the annulus)

g
k
g
k
g
S
s
s a
s
s a
s
s a s

) 1 ( ) (
,
= + =
(18)
When
8 . 0 >
l
(in the spout and the fountain)
g
k
g
k
g
S
s
s s
s
s s
s
s a s

) 1 ( ) (
,
= + =
(19)
where Z is the axial height, H
0
is the static bed height, k
a
and k
s
are the ratio of the pressure
drop of spouted beds to the pressure drop at stable fluidization, which are functions of
operating conditions and the geometrical structure of the bed. In the gas velocity ascending
process, when the spouted bed is operated at partial spouting state, k
a
is usually greater than
one. To simplify the problem, k
s
was usually set to one.
By applying the above solid phase source term, it is possible to use FLUENT to
simulate a spouted bed operated at partial spouting or stable spouting state for both the
ascending and the descending processes.
3. Simulating Conditions
Table 1. List of simulation conditions
Description Value Comment
Operating gas velocity, U
i
24 m/s Based on D
i

Gas density,
l
1.225 kg/m
3
Air
Gas viscosity,
l
1.789410
-5
kg/(ms) Air
Particle density,
s
2500 kg/m
3
Spherical glass beads
Particle diameter, d
s
0.00116 m Uniform distribution
Initial solid packing,
s,0
0.61 Fixed value
Packing limit,
s,max
0.61 Fixed value
Solid viscosity,
s
Gidaspow Eq. (7) + Eq. (9)
Solid bulk viscosity (Base case),
s
0 Fixed value
Cone angle, 45 Fixed value
Diameter of the upper section, D
c
0.45 m Fixed value
Total height of the column 1.6 m Fixed value
Gas inlet diameter, D
0
0.019 m Fixed value
Diameter of the bed bottom, D
i
0.038 m Fixed value
Static bed height, H
0
0.396 m Fixed value
Solver
2 dimensional, double precision, segregated,
unsteady, 1
st
order implicit, axisymmetric
Multiphase Model Eulerian Model, 2 phases
Viscous Model Laminar model
Phase Interaction (Base case)
Fluid-solid exchange coefficient: Gidaspow Model
Restitution coefficient: 0.9
Time steps (Final value) 0.00001s Fixed value
Convergence criterion 10
-3
Default in FLUENT

Table 2. Boundary conditions
Description Comment
Three types of radial distributions are tested for fluid phase
Inlet
No particles enter for solid phase
Uniform velocity distribution for fluid phase
Outlet
No particle exits for solid phase
Axis Axisymmetric
Non-slip for fluid phase
Wall
Zero shear stress for solid phase
In the simulation, the bed geometrical structure and dimensions (as shown in Fig. 1
(b)), the spouting gas, the bed material as well as operating conditions used are kept almost
the same as in the actual experiment. The operating gas velocity used in simulations is about
2% higher than the experiment, and the total column height is much longer than the actual
experimental setup. Because of the influence of the outlet structure on flow field, comparisons
between the experiment and simulation will not be considered above the bed surface. Details
on simulation conditions are listed in Table 1, with boundary conditions given in Table 2.
In order to investigate all possible factors that may affect simulation results, parameters
such as the fluid inlet velocity profile, solid bulk viscosity, restitution coefficient, exchange
coefficient and the source term are selected for the sensitivity analysis. At the same time, three
kinds of mesh/grid partitions of the bed are also studied. All conditions investigated are
summarized in Table 3.
Table 3. Summary of conditions used for sensitivity analysis
Grid
Partition
Fluid Inlet
Radial Profile
Bulk
Viscosity
Restitution
Coefficient
Exchange
Coefficient
Source
Term
Uniform
Parabolic
0
Lun et al.
0.9
0.81
0.99
K
sl
(Gidaspow)
0.8* K
sl
(Gidaspow)
1.2* K
sl
(Gidaspow)
k
a
=1.0
k
a
=0.7
k
a
=0.5
k
a
= k
s
=0.5
Partition 1
(10497 cells)
k
a
=0.41
k
a
=1.0
k
a
=0.7
Partition 2
(4102 cells)
k
a
=0.5
k
a
=1.0
k
a
=0.7
Partition 3
(2598 cells)
1/7
th
power law
0
0.9
K
sl
(Gidaspow)
k
a
=0.5
Notes: a. In simulations, k
s
equals to1.0 unless having further indications;
b. Conditions for the base case are as follows: partition 1; parabolic fluid inlet profile; zero value of the
solid bulk viscosity; restitution coefficient equals to 0.9; fluid-solid exchange coefficient estimated by the
Gidaspow model, k
a
equals to1.0.
4. Experiments
Experiment was conducted in two conical spouted to generate data for the validation
of the simulation results. A schematic diagram of the experimental unit is shown in Fig. 1 (a).
Conical spouted beds (both a semi-circular half column and a full column) are made of
Plexiglas with an included angle of 45
o
. The diameter at the conical base D
i
is 0.038 m, the
diameter of the nozzle D
0
is 0.019 m, and the diameter of the upper cylindrical section D
c
is
0.45 m. Glass beads of 0.00116 m in diameter were used as the bed material, compressed air
at the ambient temperature was used as the spouting gas. Other particle properties and
detailed operating conditions are shown in Table 4.

(a) (b)
Fig. 1. Schematic drawing of the Plexiglas conical spouted bed unit.
Table 4. Particle properties and operating conditions
Particle diameter
d
s
, (m)
Particle density

s
, (kg/m
3
)
Lose-packed
voidage,
l,0
Geldarts
classification
Static bed height
H
0
, (m)
Velocity
U
i
, (m/s)
0.00116 2500 0.39 D 0.396 23.5
The distribution of the static pressure was measured by static pressure probes
connected to differential pressure transducers; the distribution of the vertical solid velocity and
solid hold up were measured by optical fiber probes which were calibrated separately in
advance. During experiments, all probes can be radially traversed to measure the profile at
each vertical position.
5. Results and Discussion
Notes for Fig. 2 to Fig. 7:
Figure (a) Figure (b)
Red lines: Z=38mm; Blue lines: Z=89mm;
Magenta lines: Z=191mm; Cyan lines: Z=292mm
Red lines: Z=140mm; Blue lines: Z=241mm
Magenta lines: Z=343mm
The grid partition
The effect of grid size or grid partition on the simulation results is first examined by
comparing the simulation results from three grid sizes. As shown in Figure 2, the grid size
within the range investigated in the current simulation has little effect on the radial distribution
of the static pressure, although some influence on the distribution of the axial solid velocity is
observed, especially in the spout region. Thus, the more accurate grid partition with the
smallest grid size, partition 1, was used in the following sensitivity analysis. It is also seen from
Fig. 2 that simulated results on the axial solid velocity agree very well with experimental data,
but not for the static pressure profile under the base operating conditions without the
consideration of the solid phase source term.
0.00 0.04 0.08 0.12 0.16
r (m)
0
1000
2000
3000
4000
5000
6000
P

(
P
a
)
292
191
89
38
Z(mm)
0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 2. Comparison between experimental results and simulated results with different grid
partitions at k
a
=1.0 (k
s
=1.0, 1/7
th
power law). Symbols are experimental results, and lines are
simulated results. (Solid lines correspond to partition 1, dotted dash lines correspond to
partition 2, dash lines correspond to partition 3.)
The fluid inlet profile
The influence of fluid inlet profiles on the simulation result is shown in Fig. 3. Although
fluid inlet profiles have little effect on the distribution of the static pressure, the influence on the
distribution of the axial solid velocity is shown clearly, especially in the spout region. Simulated
static pressures overestimated experimental data significantly when k
a
was chosen to be equal
to 1.0, although the simulated particle velocity profile is quite close to the experimental data
except for the case when a parabolic inlet gas velocity profile was used. Therefore, 1/7
th
power
law or turbulent distribution inlet velocity profile was used in subsequent simulations.
0.00 0.04 0.08 0.12 0.16
r (m)
0
1000
2000
3000
4000
5000
6000
P

(
P
a
)
292
191
89
38
Z(mm)

0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
12.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 3. Comparison between experimental results and simulated results with different fluid inlet
profiles at k
a
=1.0 (k
s
=1.0). Symbols are experimental results, and lines are simulated results.
(Solid lines correspond to the 1/7
th
power law or turbulent flow, dashed lines correspond to the
parabolic profile or laminar flow, dotted dash lines correspond to the uniform profile.)
The solid bulk viscosity
0.00 0.04 0.08 0.12 0.16
r (m)
0
1000
2000
3000
4000
5000
6000
P

(
P
a
)
292
191
89
38
Z(mm)
0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 4. Comparison between experimental results and simulated results with different solid bulk
viscosity at k
a
=1.0 (k
s
=1.0, 1/7
th
power law). Symbols are experimental results, and lines are
simulated results. (Solid lines correspond to zero value for the solid bulk viscosity, dashed
lines correspond to the expression from Lun et al. for the solid bulk viscosity.)
Figure 4 shows the influence of different models for estimating the solid bulk viscosity.
It is seen that, within the range of our investigations, the solid bulk viscosity almost has no
effect on simulated results. Therefore, a zero value is assigned to the solid bulk viscosity in
most of our subsequent simulations.
Restitution coefficient
The restitution coefficient is varied from 0.81 to 0.99 to study its effect on the
simulation result. Comparing with the base case with e
ss
=0.9, a 10% increase of the restitution
coefficient affects significantly on simulated results. On the other hand, a 10% decrease of the
restitution coefficient has almost no effect on the distribution of the static pressure and the
axial solid velocity. A value of 0.9, which is the typical value used in most simulations in the
literature for glass bead particles, is thus chosen and used in the simulations throughout this
work.
0.00 0.04 0.08 0.12 0.16
r (m)
0
1000
2000
3000
4000
5000
6000
P

(
P
a
)
292
191
89
38
Z(mm)

0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 5. Comparison between experimental results and simulated results with different
restitution coefficient at k
a
=1.0 (k
s
=1.0, 1/7
th
power law). Symbols are experimental results, and
lines are simulated results. (Solid lines correspond to e
ss
=0.9, dashed lines correspond to
e
ss
=0.81, dotted dash lines correspond to e
ss
=0.99.)
The fluid-solid exchange coefficient
0.00 0.04 0.08 0.12 0.16
r (m)
0
1000
2000
3000
4000
5000
6000
P

(
P
a
)
292
191
89
38
Z(mm)

0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 6. Comparison between experimental results and simulated results with different fluid-solid
exchange coefficient at k
a
=1.0 (k
s
=1.0, 1/7
th
power law). Symbols are experimental results, and
lines are simulated results. (Solid lines correspond to the fluid-solid exchange coefficient K
sl

from Gidaspow drag model, dashed lines correspond to 80% of K
sl
, dotted dash lines
correspond to 120% of K
sl
.)
Figure 6 shows the effect of the fluid-solid exchange coefficient. Within the range of
variation, there is little influence of the drag coefficient on profiles of the static pressure,
although there is a significant effect on the axial solids velocity distribution. The Gidaspow drag
model appears to be a good choice for estimating the fluid-solid exchange coefficient, and was
used throughout this study.
The axial solid phase source term
It is seen from figures 2 to 6 that the base case setting of the CFD code with proper
inlet velocity profiles, grid size, and parameters on solids bulk viscosity, restitution coefficient
and interphase exchange coefficient can properly capture the radial particle velocity
distribution profiles in the conical spouted bed. However, variations of these key parameters
failed to bring the simulation results close to the static pressure profiles. As pointed at the
beginning of the paper, the annulus region in the spouted bed cannot be treated as a fluidized
bed, and a source term needs to be introduced to correct the gravitational term in the vertical
momentum balance equation for the particle phase. The effect of the solids source term on
static pressure and axial particle velocity profiles are simulated based on equations (18) and
(19), with the simulation results shown in Fig. 7. It is seen that the axial solid phase source
term has a significant impact on the static pressure profile, but has very little effect on the
distribution of the axial solid velocity. Compared to experimental data, a selection of k
a
=0.7
seems to give the best agreement with the experimental data on the axial solid velocity, while
a slightly smaller value of k
a
gives better agreement with the static pressure data (see Fig. 8).
Therefore, a single constant value of k
a
may not be sufficient for simulating conical spouted
beds.
0.00 0.04 0.08 0.12 0.16
r (m)
0
500
1000
1500
2000
2500
3000
P

(
P
a
)
292
191
89
38
Z(mm)
0.00 0.02 0.04 0.06 0.08
r (m)
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
V
s

(
m
/
s
)
140
241
343
Z(mm)

(a) (b)
Fig. 7. Comparison between experimental results and simulated results with different axial
solid phase source term (k
s
=1.0, 1/7
th
power law). Symbols are experimental results, and lines
are simulated results. (Solid lines correspond to k
a
=0.5, dashed lines correspond to k
a
=0.41,
dotted dash lines correspond to k
a
=0.7.)
0 1000 2000 3000 4000 5000
P
exp
(Pa)
0
1000
2000
3000
4000
5000
P
c
a
l

(
P
a
)
K
a
= 1
K
a
= 0.7
K
a
= 0.5
K
a
= 0.41

Fig. 8. Comparison between experimental results and simulated results with different axial
solid phase source term.
6. Conclusions
Among all factors investigated, the axial solid phase source term has the most
significant influence on static pressure profiles, followed by the restitution coefficient, with other
factors almost having no effect. Apart from the solid bulk viscosity, almost all other factors
affect the distribution of the axial solid velocity, although the effect of the axial solid phase
source term is minimal. For a complex system like the conical spouted bed, the introduction of
a source term to modify the gravitational force in the annulus region is proven to be essential,
although not perfect, in order to give reasonable agreement with the static pressure profile and
the particle velocity data.
In the spout region and fountain region, the movement of particles are quite complex,
particles were first accelerated from the bed bottom, and then decelerated in the fountain
region, thus, the drag force varies a lot in these two dilute regions because of the acceleration
and deceleration, while, these two dilute regions are simulated by fluidized beds at dynamic
balancing, and the drag force were described by Wen and Yu model (1966). Thus, to simulate
these two regions properly, more accurate models for the drag force are needed.
Notations
C
D
drag coefficient
D
0
gas inlet diameter, m
D
c
diameter of the upper cylindrical section, m
D
i
diameter of the bed bottom, m
d
s
particle diameter, m
e
ss
restitution coefficient
f drag function
g gravitational acceleration, in axial direction, the value is g=-9.81 m/s
2
; in radial
direction, the value is zero

g
0,ss
radial distribution function
H
0
static bed height, m
k
a
the ratio of the pressure drop of spouted beds to the pressure drop at stable
fluidization in the annulus region
K
ls
=K
sl
the momentum exchange coefficient between fluid phase l and solid phase s
k
s
the ratio of the pressure drop of spouted beds to the pressure drop at stable
fluidization in the spout and fountain region
n total number of phases
P static pressure, Pa
P
cal
simulation results on the static pressure, Pa
P
exp
experimental data on the static pressure, Pa
P
s
solid pressure, Pa
r radial position, m
Re
s
relative Reynolds number
Ss
solid phase source term
S
s,a
axial solid phase source term
t time, s
U
i
superficial gas velocity based on D
i
, m/s
vq
velocity of phase q, q can be fluid phase l or solid phase s, m/s
V
s
axial particle velocity, m/s
Z axial height of the bed, mm
Greek letters

q
volume fraction of phase q, q can be fluid phase l or solid phase s

l,0
loose-packed voidage

s,0
initial solid packing

s,max
packing limit

s
the granular temperature
included cone angle, degree

q
bulk viscosity of phase q, q can be fluid phase l or solid phase s, kg/(ms)

q
viscosity of phase q, q can be fluid phase l or solid phase s, kg/(ms)

s,col
solid collisional viscosity, kg/(ms)

s,fr
solid friction viscosity, kg/(ms)

s,kin
solid kinetic viscosity, kg/(ms)

l
gas density, kg/m
3

s
particle density, kg/m
3

l
fluid phase stress-strain tensor

s
solid phase stress-strain tensor

s
particulate relaxation time
Acknowledgements
This work is supported by the Natural Sciences and Engineering Research Council of
Canada (NSERC) under the Discovery Grant program.
References
Aguado, R.; Olazar, M.; San Jose, M. J.; Aguirre, G.; Bilbao, J., Pyrolysis of sawdust in a
conical spouted-bed reactor. Yields and product composition, Ind. Eng. Chem. Res., 39(6),
1925-1933 (2000a).
Aguado, R.; Olazar, M.; Barona, A.; Bilbao, J., Char-formation kinetics in the pyrolysis of
sawdust in a conical spouted bed reactor, J. Chem. Technol. Biotechnol., 75(7), 583-588
(2000b).
Duarte, C. R.; Murata, V. V.; Barrozo, M. A. S., A study of the fluid dynamics of the spouted
bed using CFD, Brazilian Journal of Chemical Engineering, 22(2), 263-270 (2005).
Ergun, S., Fluid Flow through Packed Columns, Chem. Eng. Prog., 48(2), 89-94 (1952).
Fluent Inc., FLUENT 6.1 UDF Manual (2003a)
Fluent Inc., FLUENT 6.1 User's Guide (2003b)
Gidaspow, D., Multiphase flow and fluidization: Continuum and kinetic theory descriptions,
Academic Press, London (1994).
Goldschmidt, M. J. V.; Kuipers, J. A. M.; van Swaaij, W. P. M., Hydrodynamic modelling of
dense gas-fluidized beds using the kinetic theory of granular flow: effect of coefficient of
restitution on bed dynamics, Chem. Eng. Sci., 56(2), 571-578 (2001).
He, Yurong; Zhao, Guangbo; Bouillard, Jacques; Lu, Huilin, Numerical simulations of the
effect of conical dimension on the hydrodynamic behaviour in spouted beds, Can. J. Chem.
Eng., 82(1), 20-29 (2004).
He, Y.-L.; Qin, S.-Z.; Lim, C. J.; Grace, J. R., Particle velocity profiles and solid flow patterns
in spouted beds, Can. J. Chem. Eng., 72(4), 561-568 (1994a).
He, Y. L.; Lim, C. J.; Grace, J. R.; Zhu, J. X., Measurements of voidage profiles in spouted
beds, Journal of Chemical Engineering, 72(2), 229-234 (1994b).
Huilin, L.; Yongli, S.; Yang, L.; Yurong, H.; Bouillard, J. Numerical simulations of
hydrodynamic behaviour in spouted beds, Chemical Engineering Research and Design,
79(A5), 593-599 (2001).
Kawaguchi, T.; Sakamoto, M.; Tanaka, T.; Tsuji, Y., Quasi-three-dimensional numerical
simulation of spouted beds in cylinder, Powder Technology, 109(1-3), 3-12 (2000).
Kucharski, J. & Kmiec, A., Hydrodynamics, heat and mass transfer during coating of tablets in
a spouted bed, Can. J. Chem. Eng., 61(3), 435-439 (1983).
Lu, Huilin; He, Yurong; Liu, Wentie; Ding, Jianmin; Gidaspow, Dimitri; Bouillard, Jacques,
Computer simulations of gas-solid flow in spouted beds using kinetic-frictional stress model of
granular flow, Chem. Eng. Sci., 59(4), 865-878 (2004).
Lun, C. K. K., Savage, S. B., Jeffrey, D. J., and Chepurniy, N., Kinetic theories for granular
flow: Inelastic particles in Couette flow and slightly inelastic particles in a general flowfield", J.
Fluid Mech., 140, 223-256 (1984).
Markowski, A. S., Drying characteristics in a jet-spouted bed dryer, Can. J. Chem. Eng.,
70(5), 938-944 (1992).
Marnasidou, K. G.; Voutetakis, S. S.; Tjatjopoulos, G. J.; Vasalos, I. A., Catalytic partial
oxidation of methane to synthesis gas in a pilot-plant-scale spouted-bed reactor, Chem. Eng.
Sci., 54(15-16), 3691-3699 (1999).
Mathur, K. B. & Epstein N., Spouted beds, Academic Press, New York (1974).
McKeen, Tim and Pugsley, Todd, Simulation and experimental validation of a freely bubbling
bed of FCC catalyst, Powder Technology, 129(1-3), 139-152 (2003).
Mukhlenov, I. P. & Gorshtein, A. E., Hydrodynamics of reactors with a spouting bed of
granular material, Vses. Konf. Khim. Reactrom Novosibirsk, (3) 553-562 (1965).
Mukhlenov, I. P. & Gorshtein, A. E., Hydraulic resistance of a fluidized layer in a cyclone
without a grate, Zh. Prikl. Khim., 37(3), 609-615 (1964).
Olazar, M.; Aguado, R.; San Jose, M. J.; Bilbao, J., Kinetic study of fast pyrolysis of sawdust
in a conical spouted bed reactor in the range 400-500 C
o
, Journal of Chemical Technology &
Biotechnology, 76(5), 469-476 (2001).
Olazar, M.; Aguado, R.; San Jose, M. J.; Bilbao, J., Performance of a conical spouted bed in
biomass catalytic pyrolysis, Recents Progres en Genie des Procedes, 14(76), 499-506
(2000a).
Olazar, M.; Aguado, R.; Bilbao, J.; Barona, A., Pyrolysis of sawdust in a conical spouted-bed
reactor with a HZSM-5 catalyst, AIChE J., 46(5), 1025-1033 (2000b).
Passos, M. L.; Oliveira, L. S.; Franca, A. S.; Massarani, G., Bixin powder production in conical
spouted bed units, Drying Technol., 16(9&10), 1855-1879 (1998).
Passos, M. L.; Massarani, G.; Freire, J. T.; Mujumdar, A. S., Drying of pastes in spouted beds
of inert particles: design criteria and modeling, Drying Technol., 15(2), 605-624 (1997).
Pham, Q. T., Behavior of a conical spouted-bed dryer for animal blood, Can. J. Chem. Eng.,
61(3), 426-434 (1983).
Reyes, A.; Diaz, G.; Blasco, R., Slurry drying in gas-particle contactors: fluid-dynamics and
capacity analysis, Drying Technol., 16(1 & 2), 217-233 (1998).
San Jose, M. J.; Olazar, Martin; Alvarez, Sonia; Izquierdo, Miguel A.; Bilbao, Javier, Solid
cross-flow into the spout and particle trajectories in conical spouted beds, Chem. Eng. Sci.,
53(20), 3561-3570 (1998).
Uemaki, Osamu; Tsuji, Toshiro, Gasification of a sub-bituminous coal in a two-stage, jet-
spouted bed reactor, in Fluidization V: Proceedings of the Fifth Engineering Foundation
Conference on Fluidization, Engineering Foundation, New York, 497-504 (1986).
Wen, C. Y.; Yu, Y. H., Mechanics of fluidization, Chemical Engineering Progress, Symposium
Series, 62(62), 100-111 (1966).
Yang, Ning; Wang, Wei; Ge, Wei; Li, Jinghai, CFD simulation of concurrent-up gas-solid flow
in circulating fluidized beds with structure-dependent drag coefficient, Chemical Engineering
Journal (Amsterdam, Netherlands), 96(1-3), 71-80 (2003).