i An update to this article is included at the end

Applied Mathematical Modelling 35 (2011) 2265–2278

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Computational fluid dynamics and experimental study

of the hydrodynamics of a gas–solid tapered fluidized bed
D.C. Sau a,⇑, K.C. Biswal b
a
National Metallurgical Laboratory, Jamshedpur 831 007, Jharkhand, India
b
National Institute of Technology, Rourkela, Orissa, India

a r t i c l e i n f o a b s t r a c t

Article history: In the present work, experimental and numerical studies for the hydrodynamics in a gas–
Received 26 May 2010 solid tapered fluidized bed have been carried out. The experimental results obtained by
Received in revised form 31 October 2010 carrying out experiments in a tapered fluidized bed for glass bead (spherical) of 2.0 mm
Accepted 15 November 2010
and dolomite (non-spherical particles) of 2.215 mm in diameter, were compared with
Available online 24 November 2010
the computational fluid dynamics (CFD) simulation results, using a commercial CFD soft-
ware package, Fluent. The gas–solid flow was simulated using the Eulerian–Eulerian model
Keywords:
and applying the kinetic theory of granular flow for solid particles. The Gidaspow drag
Gas–solids fluidization
Tapered fluidized bed
model was used to calculate the gas–solid momentum exchange coefficients. Pressure
Two fluid modeling drops predicted by the CFD simulations agreed reasonably well with experimental mea-
CFD surements for both types (spherical and non-spherical) of particles. Good agreement was
Two-dimensional also obtained between experimental and CFD predicted bed expansion ratios for both types
of particles. Present study provides a useful basis for further works on the CFD of tapered
fluidized bed.
Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction

Most of the gas–solid fluidization behavior studies that have been reported are for straight cylindrical or columnar flu-
idized beds, although a considerable portion of the fluidized beds have inclined walls or have a tapered bottom section. A
velocity gradient exists in the axial direction in tapered bed leading to unique dynamic characteristics of the bed. Due to this
characteristic, tapered fluidized beds have found wide application in many industrial processes such as, waste water treat-
ment [1] immobilized biofilm reaction, incineration of waste materials, coating of nuclear fuel particles, crystallization, coal
gasification, roasting sulfide ores [2] and food processing [3], etc.
Tapered fluidized beds can be operated smoothly without any instability, i.e. with less pressure fluctuations [1]. Tapered
fluidized beds are very useful for fluidization of materials with a wide particle size distribution, as well as for exothermic
reactions [4] and also for extensive particle mixing [5]. Despite their widespread application, much of the development
and design of tapered fluidized bed reactors have been empirical in nature as the complex behavior of gas–solid flow in these
systems makes flow modeling a challenging task. In addition, numerical solutions of complex non-linear equations, with
moving phase boundaries, are difficult to obtain. However, with increasing computational capabilities and computational
fluid dynamics (CFD) tools in recent years, several researchers are involved in studying the hydrodynamics of gas–solid sys-
tems, which would be useful in the design, optimization and scale-up process. The two common approaches for modeling

⇑ Corresponding author. Tel.: +91 657 2345180 (O); fax: +91 657 2270527.
E-mail address: [email protected] (D.C. Sau).

0307-904X/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2010.11.037
2266 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

Notations

CD drag coefficient, dimensionless

di particle diameter, m
ess restitution coefficient, dimensionless
g acceleration due to gravity, m/s2
g0,ss radial distribution coefficient, dimensionless
H expanded bed height, m
Hs static bed height, m
I stress tensor, dimensionless
I2D second invariant of the deviatoric stress tensor, dimensionless
kHs diffusion coefficient for granular energy, kg/s m
Kgs gas/solid momentum exchange coefficient, dimensionless
Ka coefficient, dimensionless
Kc coefficient, dimensionless
P pressure, Pa
rPtap axial pressure gradient of tapered bed
Re Reynolds number, dimensionless
Ss,a solid phase source term at annulus, kg/m2 s2
Ss,c solid phase source term at core and hump, kg/m2 s2
t time, s
ti velocity, m/s
v 0s particle velocity fluctuation, m/s
z height coordinate measured from distributor, m

Greek symbols
ei voidage, dimensionless
Hi granular temperature, m2/s2
ki bulk viscosity, kg/m s
li shear viscosity kg/m s
qi density, kg/m3
si stress tensor, Pa
cHs collision dissipation of energy, kg/s3 m
ugs transfer rate of kinetic energy, kg/s3 m

Subscripts
fb fully fluidized
g gas
i general index
mf minimum fluidization
max maximum
o fixed bed condition
s solids
T stress tensor
tap tapered bed
con conventional bed

gas–solid fluidized beds are Eulerian–Lagrangian model, where the gas is treated as the continuous phase and the solid as the
discrete phase, and the Eulerian–Eulerian model, where the two phases are treated as interpenetrating continua. The trans-
port coefficients of the solid phases also include gas–particle interactions and particle–particle collisions terms. For the
resultant continuum approximation, certain averaging techniques and assumptions are required to obtain a momentum bal-
ance for the solids phase. The mechanisms of particle-to-particle collisions are described by the hard sphere [6] and soft
sphere approach [7]. Kawaguchi et al. [8] simulated particle motion in a spouted conventional bed using the discrete element
method (DEM).
The computational demands for Eulerian–Lagrangian approach are substantial or even impossible for systems with large
number of particles. The Eulerian–Eulerian model (so called two fluid models) is the most common approach in simulating
fluidized bed systems. Johansson et al. [9] simulated conventional fluidized beds using two different closure models for the
particle phase stresses. The first approach uses a constant particle viscosity (CPV) and predicts particle pressure. The second
approach uses the particle turbulence model (PT), which is also known as granular temperature model [10]. This model is
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2267

based on kinetic theory of granular flow (KTGF) and derived in analogy with the kinetic theory of gases [11]. The solid par-
ticles are generally considered to be identical. Johansson et al. [9] also reported that the KTGF model results are in better
agreement with the experiments.
CFD techniques is now a promising tool to model fluidized bed dynamics due to it’s many advantages such as design, opti-
mization and scale-up of processes. Some of the problems with respect to CFD model validation for gas–solid systems have
been reviewed by Grace and Taghipour [12]. Taghipour [13] presented CFD simulation of two-dimensional (2D) cylindrical
columns and observed that the motion of bubbles and particles are greatly influenced by the wall effect in a 2D column.
Recently detail CFD simulation of conical spouted bed has been carried out by Wang [14].
Some investigations have been carried out in tapered fluidized bed dynamics during last two decades [1–4]. Schaafsma
et al. [5] studied particle flow pattern and segregation phenomena in tapered fluidized bed granulator using segregation
experiment and positron emission particle tracking experiments (PEPT) and concluded that the flow and segregation phe-
nomena is totally different from conventional (cylindrical) beds. Recently, Sau et al. [15,16] also investigated the dynam-
ics of tapered fluidized beds and presented models for predicting minimum fluidization velocity and maximum bed
pressure drop. But so far, the CFD simulations studies of tapered fluidized bed have not been reported in literature. In
this present study, therefore, both the experimental and CFD simulation investigations are carried out in a tapered fluid-
ized bed column to enable meaningful comparison of the pressure drop and bed expansion ratios. Here, both spherical
(glass bead, minimum fluidization velocity = 2.01 m/s) and non-spherical (dolomite, minimum fluidization veloc-
ity = 2.41 m/s) particles have been used to carry out the investigations, although most of the investigations reported in
literature are for spherical particles and in conventional bed. Computational fluid dynamic simulation was carried out
using commercial CFD software, Fluent [17], with Gidaspow drag function to calculate the momentum exchange coeffi-
cient and a solid phase source term for the momentum balance equation for the solid phase. Experimental data were
compared with simulation results obtained with a two fluid model with closure equation based on kinetic theory of
granular flow.

2. Experimental setup

Experiments were conducted in a tapered column of 520 mm height and tapered angle of 4.61°. The tapered column was
made of Perspex sheet to allow visual observation. The diameter of the column at the bottom was 50 mm where as at the top
the diameter was 135 mm. A 60-mesh screen at the bottom served as the support as well as the distributor. The calming
section of the bed was filled with 3 mm glass beads, which offers sufficient resistance for uniform distribution of air. Two
pressure taps, one just above the distributor and the other at the top of the bed were provided to record the pressure drops.
Pressure drop was measured by manometer, which was 1 m long. Carbon tetrachloride (density = 1594 kg/m3) was used as
the manometric fluid. Air at a temperature of around 301 K (qf = 1.17 kg/m3 and lf = 1.82  105 kg/m s), used as the fluid-
izing medium was passed through a receiver and a silica gel tower to dry and control the air flow before being sent through
the tapered column. Two rotameters, one for the lower range (0–10 m3 h1) and the other for the higher range (10–120 m3
h1) were used to measure the air flow rates. Glass beads (spherical particles) of 2.0 mm diameter and density of 2600 kg/m3
and dolomite (non-spherical particles) of 2.215 mm diameter and density of 2800 kg/m3 were fluidized with air at around
301 K and atmospheric pressure. The bed pressure drops and bed expansion were measured at different gas velocities from 0
to 3.54 m/s. The detail experimental set up is shown in Fig. 1. Snapshots were also taken by using high speed digital video
camera.

Fig. 1. Experimental set-up. 1. Compressor; 2. Receiver; 3. Silica gel tower; 4. Bypass valve; 5. Line valve; 6. Rotameter; 7. Bed materials; 8. Fluidizer;
9. Pressure tapping to manometer.
2268 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

3. CFD modeling

Currently there are two approaches for the numerical calculation of multiphase flows: the Eulerian–Lagrangian approach
and the Eulerian–Eulerian approach (TFM). In the Eulerian–Lagrangian 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 par-
ticles (or bubbles, droplets) through the calculated flow field. The dispersed phase can exchange momentum, mass, and en-
ergy with the fluid phase. A fundamental assumption made in this approach is that the dispersed second phase occupies a
low volume fraction. In the Eulerian–Eulerian 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 phase volume fraction is intro-
duced. These volume fractions are assumed to be continuous functions of space and time and their sum is equal to one. For
granular flows, such as flows in risers, fluidized beds and other suspension systems, the Eulerian –Eulerian multiphase model
is generally used, and also for simulations in this study.

3.1. Computational geometry

For simulation study, a gas–solid tapered fluidized bed having bottom width 50 mm, top width 135 mm and slant height
520 mm has been considered. The grid was created using commercial software, GAMBIT 2.1.6 and exported into Fluent
6.1.22 for simulation.

3.2. Model formulation

In the present work, an Eulerian–Eulerian multi-fluid model, which considers the conservation of mass and momentum
for the solid and gas phases, has been adopted. The kinetic theory of granular flow, which considers the conservation of solid
fluctuation energy, has been used for closure.
The unsteady state equations of continuity for phase i (gas or solid) with temporal and spatial gradients is given as:
@
v i Þ ¼ 0:
ðei qi Þ þ r:ðei qi~ ð3:1Þ
@t
The phase volume fractions satisfy the following condition:
eg þ es ¼ 1: ð3:2Þ
The momentum balance for each phase given by the Navier–Stokes equation is modified to include an interphase momen-
tum transfer term and a solid phase source term:
@
v g Þ þ r  ðeg qg ~
ðeg qg ~ v g~
v g Þ ¼ eg rP þ eg r  sg þ eg qg g  K gs ð~
vg  ~
v s Þ; ð3:3Þ
@t

@ ! ! ! ! ! !
ðes qs v s Þ þ r  ðes qs v s v s Þ ¼ es rP  rPs þ r  ss þ es qs g  K gs ðv g  v s Þ þ Ss ; ð3:4Þ
@t
where the left hand side (LHS) represents the temporal and spatial transport terms and the right hand side (RHS) represents
the various interacting forces. Interactions between the phases involve various momentum exchange mechanisms such as
the drag, the lift and added mass force, etc. However, the contribution of drag force has been considered while the effect
of the other forces has been ignored.
In order to solve Eqs. (3.3) and (3.4), it is necessary to have closure equations to relate the unknowns in these equations. In
the present study the kinetic theory of granular flow developed by Lun et al. [11] was used to define the solid phase
properties.
Analogous to the thermodynamic temperature for gases, the granular temperature Hs can be introduced as a measure of
the particle velocity fluctuation, i.e.,
1
Hs ¼ ðv 02
s Þ: ð3:5Þ
3
The granular temperature conservation equation is
 
3 @ ! !
ðqs es Hs Þ þ rðqs es v s Hs Þ ¼ ðps I þ ss Þ : r v s þrðkHs rHs Þ  cHs þ /is ; ð3:6Þ
2 @t
!
where ðps I þ ss Þ : r v s is the generation of energy by the solid stress tensor because of the interaction between the normal
and shear stress matrix with the mean velocity field, kHsrHs accounts for the transport of energy due to diffusion (kHs is the
diffusion coefficient), yHs is the dissipation of energy due to collision and is given as,
pffiffiffiffi 3
cHs ¼ ðð12ð1  e2ss Þg 00 ss Þ=ds pÞqs e2s H2s :
The last term on the RHS is the exchange of kinetic energy between the solid and the gas phases, given as,
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2269

/gs ¼ 3kgs Hs :

The stress–strain tensor for phase i is given as,

! !T 2
si ¼ ei li ðr v i þr v i Þ þ ei ðki  li Þrvi I: ð3:7Þ
3
For the solid phase, s, the solids shear viscosity is the sum of the collisional viscosity, kinetic viscosity and optional frictional
viscosity and is given as
ls ¼ ls;col þ ls;kin þ ls;fr : ð3:8Þ

The collision viscosity is given as,

 1=2
4 Hs
ls;col ¼ es qs ds g 0;ss ð1 þ ess Þ ; ð3:9Þ
5 p
where g0,ss is defined as,
"  1=3 #1
es
g 0;ss ¼ 1  : ð3:10Þ
es;max
The following expression from Gidaspow [18] was used to estimate the kinetic viscosity
pffiffiffiffiffiffiffiffiffiffi  2
10qs ds Hs p 4
ls;kin ¼ 1 þ g 0;ss es ð1 þ ess Þ : ð3:11Þ
96es ð1 þ ess Þg 0;ss 5

The solid bulk viscosity as given by Lun et al. [11] is

 1=2
4 H
ks ¼ es qs ds g 0;ss ð1 þ ess Þ s : ð3:12Þ
5 p
The frictional viscosity as defined by Schaeffer [19] is

P sinðUÞ
ls;fr ¼ s p2 ffiffiffiffiffiffi : ð3:13Þ
I2D
Several models for the fluid–solid exchange coefficient are available in the literature. In the present study, however, the
widely used Gidaspow [18] drag model has been used. The Gidaspow model is a combination of the Wen and Yu [20] model
and the Ergun [21] equation.
When eg > 0.8, the fluid–solid exchange coefficient Ksg is of the following form:
! !
3 es eg qg j v s  v g j 2:65
K sg ¼ CD eg ; ð3:14Þ
4 ds
where
24
CD ¼ ½1 þ 0:15ðeg Res Þ0:687 ; ð3:15Þ
eg Res
! !
qg ds j v s  v g j
Res ¼ : ð3:16Þ
lg
When eg 6 0.8
! !
es ð1  eg Þlg qg es j v s  v g j
K sg ¼ 150 2
þ 1:75 : ð3:17Þ
eg ds ds

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, rPs, in the solids phase
momentum equation. The solids pressure is composed of a kinetic term and a second term due to particle collision. The sol-
ids pressure as given by Lun et al. [11] is

Ps ¼ es qs Hs þ 2qs ðð1 þ ess Þe2s g 0;ss Hs : ð3:18Þ

For tapered beds, there exist three distinct regions: a dilute core, a dense annular region between the core and the wall
named the annulus, and a dilute hump region above the bed surface [2]. For simulation of the tapered bed, the bed should be
divided mainly two regions: a dilute fluidized region (including both the core and the hump) and a dense defluidized region
2270 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

(annulus). At fully fluidization, the outer noncore region (dense annular) is in a state of complete fluidization [22] and cir-
culation of particles [23]. To include the additional stress exerted by the tapered side wall on the gas–solids flow, two solid
phase source terms [14] are introduced into the annulus and core regions, respectively, with the following forms:

DPtap
Ka ¼ ; ð3:19Þ
DPcon
K c ¼ f ðeg;0 ; qs ; ds ; qg ; lg ; v g;s Þ; ð3:20Þ

where Ka is the ratio of the pressure drop in a tapered fluidized bed to that in a conventional fluidized bed.
Axial pressure gradient of tapered bed is calculated by

Table 1
Boundary conditions used for the simulation of tapered fluidized bed

Description Type Comments

Inlet Velocity-inlet Uniform distribution for gas phase
No particles enter for solid phase
Outlet Pressure-outlet Atmospheric
Wall Stationary wall No slip for gas phase
Zero shear stress for solid phase

Table 2
Experimental conditions and simulation settings for tapered bed

Bottom diameter of the bed (m) 0.05

Top diameter of the bed (m) 0.135
Total height of the bed (m) 0.52
Initial static bed height (m) 0.065, 0.08, 0.055, 0.075
Inlet gas velocity(m/s) 0–3.54
Gas density (kg/m3) 1.17
Gas viscosity (kg/m s) 1.82  105
Particle density (kg/m3) 2600, 2800
Particle diameter (mm) 2.0, 2.215
Initial solids packing 0.48, 0.64
Restitution coefficient 0.9
Time steps (s) 0.0001
Convergence criterion 0.001

1200

1000
Bed pressure drop (Pa)

800

600

400

200 Experimental
CFD Simualtion

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Superficial gas velocity (m/s)
Fig. 2a. Comparison plot of experimental bed pressure drop with CFD simulation using Gidaspow drag model (particle = glass bead, diameter = 2 mm,
initial static bed height = 6.5 cm).
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2271

 2 " 4 #
1 vg D0
rP tap ¼ ð1  eg;0 Þqs g þ 1 :
2H eg;0 Di

Axial pressure gradient of conventional bed is calculated by

rP con ¼ ð1  eg;0 Þqs g

Though Kc is a function of different parameters, for simplification of the problem, Kc was assumed to be one. Based on Wen
and Yu [20] and Ergun [21] drag equations the following simple expressions were used to describe the solid phase source
term.

1200

1000
Bed pressure drop (Pa)

800

600

400

200
Experimental
CFD Simulation

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Superficial gas velocity (m/s)
Fig. 2b. Comparison of experimental bed pressure drop with CFD simulation using Gidaspow drag model (particle = glass bead, diameter = 2 mm, initial
static bed height = 8.0 cm).

1200

1000
Bed pressure drop (Pa)

800

600

400

200 Experimental
CFD Simulation

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Superficial gas velocity (m/s)

Fig. 3a. Comparison of experimental bed pressure drop with CFD simulation using Gidaspow drag model (particle = dolomite, diameter = 2.215 mm, initial
static bed height = 5.5 cm).
2272 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

When eg 6 0.8 and Z 6 H0 (in the annulus),

Ss;a ¼ es qs g þ K a ðes qs gÞ ¼ ðK a  1Þes qs g: ð3:21Þ
When eg > 0.8 (in the core and the hump),
Ss;c ¼ es qs g þ K c ðes qs gÞ ¼ ðK c  1Þes qs g: ð3:22Þ
The actual pressure gradient in a tapered bed is the combination of the gravity term and the solid phase source term in
the annulus. Different values of Ka (or different solid phase source terms) represent different values of the pressure gradient
in a tapered bed.

3.3. Solution procedures

The set of governing equations were solved using the finite volume approach. The 2D computational domain was discret-
ized by 6102 triangular cells. A time step of 0.0001 with 20 iterations per time step was chosen. This iteration was adequate

1400

1200
Bed pressure drop (Pa)

1000

800

600

400

200 Experimental
CFD Simulation
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Superficial gas velocity (m/s)

Fig. 3b. Comparison of experimental bed pressure drop with CFD simulation using Gidaspow drag model (particle = dolomite, diameter = 2.215 mm, initial
static bed height = 7.5 cm).

2.4

2.2
Bed expansion ratio (H/Hs)

2.0

1.8

1.6

1.4

Experimental
1.2
CFD Simulation

1.0
1.5 2.0 2.5 3.0 3.5 4.0
Superficial gas velocity (m/s)

Fig. 4a. Comparison of experimental and CFD simulated bed expansion ratios (particle = glass bead, diameter = 2 mm, initial static bed height = 6.5 cm).
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2273

to achieve convergence for the majority of time steps. The relative error between two successive iterations was specified by
using a convergence criterion of 0.001 for each scaled residual component. The pressure–velocity coupling is obtained by
phase-coupled SIMPLE (PC-SIMPLE) algorithm. In this study, the effect of restitution coefficient was not examined. Tables
1 and 2 summarize the boundary conditions and parameters used for the CFD simulation study respectively.

4. Results and discussion

In this work, the bed expansion and the pressure drop which are two important parameters in the investigation of gas–
solid fluidization behavior have been studied. The sensitivity of the numerical flow prediction to grid resolution is also inves-
tigated in the present study. A grid independency was checked with two different mesh sizes (2000 rectangular cells and
6102 triangular cells). The set of governing equations were solved using the finite volume approach and second order
discretization scheme. Second order discretization scheme is widely used in CFD simulation of gas–solid fluidized bed to
get proper size and shape of bubbles. In this study, grid refinement hardly changed the instantaneous flow structure, time

2.2

2.0
Bed expansion ratio (H/Hs)

1.8

1.6

1.4

Experimental
1.2 CFD Simulation

1.5 2.0 2.5 3.0 3.5 4.0

Superficial gas velocity (m/s)

Fig. 4b. Comparison of experimental and CFD simulated bed expansion ratios (particle = glass bead, diameter = 2 mm, initial static bed height = 8.0 cm).

3.0

2.8

2.6
Bed expansion ratio (H/Hs)

2.4

2.2

2.0

1.8

1.6
Experimental
1.4
CFD Simulation
1.2

1.5 2.0 2.5 3.0 3.5 4.0

Superficial gas velocity (m/s)

Fig. 5a. Comparison of experimental and CFD simulated bed expansion ratios (particle = dolomite, diameter = 2.215 mm, initial static bed height = 5.5 cm).
2274 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

3.0

2.8

2.6

Bed expansion ratio (H/Hs)

2.4

2.2

2.0

1.8

1.6

1.4

1.2 Experimental
CFD Simulation
1.0

1.5 2.0 2.5 3.0 3.5

Superficial gas velocity (m/s)

Fig. 5b. Comparison of experimental and simulated bed expansion ratios (particle = dolomite, diameter = 2.215 mm, initial static bed height = 7.5 cm).

Fig. 6a. Contours of solids volume fractions of 2D bed (velocity of gas = 2.83 m/s, drag function: Gidaspow, particle: glass bead, diameter: 2.0 mm, initial
static bed height: 6.5 cm).

Fig. 6b. Contours of solids volume fractions of 2D bed (velocity of gas = 2.83 m/s, drag function: Gidaspow, particle: glass bead, diameter: 2.0 mm, initial
static bed height: 8.0 cm).
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2275

averaged flow pattern and other predicted parameters. Since the experimental fluidized bed is three-dimensional (3D) and
CFD simulations have been carried for two-dimensional (2D); some differences between the simulations and experimental
results are expected.
Time averaged pressure drop, bed expansion ratios at different superficial gas velocities and solid volume fractions profile
with different time instant, obtained from CFD are shown in Figs. 2–9. Experimental analyses were done to obtain the steady
state pressure drop, and bed expansion ratios at different superficial gas velocities. Snapshots were taken using digital video

Fig. 7a. Contours of solids volume fractions of 2D bed (velocity of gas = 3.54 m/s, drag function: Gidaspow, particle: glass bead, diameter: 2.0 mm, initial
static bed height: 6.5 cm).

Fig. 7b. Contours of solids volume fractions of 2D bed (velocity of gas = 3.54 m/s, drag function: Gidaspow, particle: glass bead, diameter: 2.0 mm, initial
static bed height: 8.0 cm).

Fig. 8a. Contours of solid volume fractions of 2D bed (velocity of gas = 2.83 m/s, drag function: Gidaspow, particle: dolomite, diameter: 2.215 mm, initial
static bed height: 5.5 cm).
2276 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

Fig. 8b. Contours of solid volume fractions of 2D bed (velocity of gas = 2.83 m/s, drag function: Gidaspow, particle: dolomite, diameter: 2.215 mm, initial
static bed height: 7.5 cm).

Fig. 9a. Contour of solid volume fractions of 2D bed (velocity of gas = 3.54 m/s, drag function: Gidaspow, particle: dolomite, diameter: 2.215 mm, initial
static bed height: 5.5 cm).

Fig. 9b. Contour of solid volume fractions of 2D bed (velocity of gas = 3.54 m/s, drag function: Gidaspow, particle: dolomite, diameter: 2.215 mm, initial
static bed height: 7.5 cm).

camera to analyze and compare the experimental results with the CFD simulation results at different bed conditions. All the
comparisons between experimental and CFD simulated results are in reasonably good agreement with each other. One can
compare the experimental bed height snapshots but not the volume fraction of 2D simulation, because the experimental
 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278 2277

snapshot gives the picture of the volume fraction at the wall, where as in 2D simulation, the contours shown is along the
central axial plane and the wall is represented by two lines only. Hence, a 3D simulation is necessary to compare the sim-
ulation results with experimental data, which will be the scope of our future study.
Based on the analysis, the time-averaged pressures drops and bed expansion ratios were taken for comparing modeling
and experimental results. Pressure drop of the bed has been studied by considering different superficial velocities, both
above and under the minimum fluidization velocity. Fig. 2(a) and (b) shows the comparisons between the experimental
and simulated pressure drops plotted against superficial velocities for different initial static bed heights (viz. 6.5 and
8 cm) for a particular size (2 mm) of spherical particles (glass bead). Fig. 3(a) and (b) shows the comparisons between the
experimental and simulated pressure drops plotted against superficial velocities for different initial static bed heights
(viz. 5.5 and 7.5 cm) for a particular size (2.215 mm) of non-spherical (dolomite) particles. Qualitatively, good agreements
can be found from the graphs. It has been observed that in the both cases (CFD simulation and experimental), the initial var-
iation of pressure drop with the inlet gas velocity is linear. When the gas velocity increases, the gas pressure drop still in-
creases up to a maximum value. When the gas velocity increases further, the gas velocity decreases rapidly at first and
then remains constant with gas velocity.
The CFD simulated time-average bed expansion ratios using Gidaspow drag functions for various superficial gas velocities
are calculated, analyzed and compared with experimental values in Figs. 4 and 5. Good agreement can be found from the
graphs. It is also observed that the difference between the experimental and CFD simulation of bed expansion ratios are
clearly less incase of non-spherical (dolomite) particles whereas the difference are more in case of spherical particles (glass
bead) at higher gas velocities. The proposed CFD models predicted the right trends of bed expansion for both types of par-
ticles, though the predicted bed expansion ratios differ from experimental results. At higher superficial gas velocities the dif-
ferences of bed expansion ratios between the simulation and experimental results are more.
The CFD simulations were performed using the transient Eulerian-Granular model in Fluent software at fluidization con-
dition for both types of particles (spherical and non-spherical) at different time instant (1–6 s) with two different initial sta-
tic bed height and two different gas velocities. Figs. 6–9 show a contour plot of solid fractions using Gidaspow drag model.
After the initial startup period (around 3 s), almost for all the cases (Figs. 6–9) particle concentration is less at the central
region than near the wall region. This is due to the formation and rising of bubbles in the central region. Contour plots of
solids fraction by CFD simulation of spherical (glass bead) particles at different time instant (1–6 s) can be seen in Figs. 6
and 7 at different fluidization condition, whereas contour plots of solids fraction by CFD simulation of non-spherical
(dolomite) particles at different time instant can be seen in Figs. 8 and 9 at different fluidization condition. The instantaneous
break-up of the particle locking leads to a significant upward flow of particles. Bubbles are formed in the distributor, which
forces particles to move upward. This bubble grows as gas flows upward and eventually converts to a slug at the upper part
of the bed. Once the lifted particles fall back to the bed, stable fluidization is established rapidly. The influence of initial static
bed height and diameter is also observed on the solid volume fraction profile of the particles beside superficial gas velocity,
as can be seen in Figs. 6–9.These factors change the distribution of particles velocities inside the bed.

5. Conclusions

A multifluid Eulerian–Eulerian model integrating the kinetic theory for solid particles using Fluent CFD software was
capable of predicting bed dynamics and time average bed behavior of a tapered fluidized bed at fixed bed condition and flu-
idization conditions. Comparison of the model predictions, using the Gidaspow drag functions, and experimental measure-
ments on the time-average bed pressure drop and bed expansion indicated reasonable agreement for most operating
conditions. We stress, however that the current results are for two-dimensional only, and can therefore only serve to get
a qualitative insight into the physical underlying of the fluidization behavior in tapered bed. For a true, quantitative compar-
ison with experiments, clearly, full 3D simulations with different drag model and different restitution coefficient are re-
quired. This work is currently underway with particle image velocimetry (PIV) measurements and will be the subject of
future publications.

References

[1] Y.F. Shi, Y.S. Yu, L.T. Fan, Incipient fluidization condition for a tapered fluidized bed, Ind. Eng. Chem. Fundam. 23 (1984) 484–489.
[2] Y. Peng, L.T. Fan, Hydrodynamic characteristics of fluidization in liquid–solid tapered beds, Chem. Eng. Sci. 52 (1997) 2277–2290.
[3] F. Depypere, J.G. Pieters, K. Dewettinck, Expanded bed height determination in a tapered fluidized bed reactor, J. Food Eng. 67 (2005) 353–359.
[4] H.G. Kim, I.O. Lee, U.C. Chung, Y.H. Kim, Fluidization characteristics of iron ore fines of wide size distribution in a cold tapered gas–solid fluidized bed,
ISIJ Int. 40 (2000) 16–22.
[5] S.H. Schaafsma, T. Marx, A.C. Hoffmann, Investigation of the particle flow pattern and segregation in tapered fluidized bed granulators, Chem. Eng. Sci.
61 (2006) 4467–4475.
[6] D. Gera, M. Gautam, Y. Tsuji, T. Kawaguchi, T. Tanaka, Computer simulation of bubbles in large-particle fluidized beds, Powder Technol. 98 (1998) 38–
47.
[7] N. Kobayashi, R. Yamazaki, S. Mori, A study on the behavior of bubbles and solids in bubbling fluidized beds, Powder Technol. 113 (2000) 327–344.
[8] T. Kawaguchi, T. Tanaka, Y. Tsuji, Numerical simulation of two-dimensional fluidized beds using the discrete element method (comparison between
the two- and three-dimensional models), Powder Technol. 96 (1998) 129–138.
[9] K. Johansson, B.G.M. vanWachem, A.E. Almstedt, Experimental validation of CFD models for fluidized beds: Influence of particle stress models, gas
phase compressibility and air inflow models, Chem. Eng. Sci. 61 (2006) 1705–1717.
2278 D.C. Sau, K.C. Biswal / Applied Mathematical Modelling 35 (2011) 2265–2278

[10] H. Enwald, A.E. Almstedt, Fluid dynamics of a pressurized fluidized bed: comparison between numerical solutions from two-fluid models and
experimental results, Chem. Eng. Sci. 54 (1999) 329–342.
[11] C.K.K. Lun, S.B. Savage, D.J. Jeffrey, N. Chepurniy, Kinetic theories for granular flow: inelastic particles in Couette flow and slightly inelastic particles in a
general flow field, J. Fluid Mech. 140 (1984) 223–256.
[12] J.R. Grace, F. Taghipur, Verification and validation of CFD models and dynamic similarity for fluidized beds, Powder Technol. 139 (2004) 99–110.
[13] F. Taghipur, N. Ellis, C. Wong, Experimental and computational study of gas–solid fluidized bed hydrodynamics, Chem. Eng. Sci. 60 (2005) 6857–6867.
[14] Z. Wang, Experimental Studies and CFD Simulations of Conical Spouted Bed Hydrodynamics, Ph.D. Thesis, The University of British Columbia, 2006.
[15] D.C. Sau, S. Mohanty, K.C. Biswal, Minimum fluidization velocities and maximum bed pressure drops for gas–solid tapered fluidized beds, Chem. Eng. J.
132 (2007) 151–157.
[16] D.C. Sau, S. Mohanty, K.C. Biswal, Critical fluidization velocities and maximum bed pressure drops of homogeneous binary mixture of irregular
particles in gas–solid tapered fluidized beds, Powder Technol. 186 (3) (2008) 241–246.
[17] Fluent 6.1.22, User’s Guide, 20.4, Eulerian Model, Fluent Inc., 2005.
[18] D. Gidaspow, Multiphase Flow and Fluidization: Continuum and Kinetic Theory Description, Academic Press, San Diego, 1994.
[19] D.G. Schaeffer, Instability in the evolution equations describing incompressible granular flow, J. Differ. Equat. 66 (1987) 19–50.
[20] C.Y. Wen, Y.H. Yu, Mechanics of fluidization, Chem. Eng. Prog. Symp. Ser. 62 (1966) 100–111.
[21] S. Ergun, Fluid flow through packed columns, Chem. Eng. Prog. 48 (1952) 89–94.
[22] H. Toyahara, M. Kawamura, Fluidization of a tapered fluidized-bed of a binary particle-mixture, Int. Chem. Eng. 32 (1992) 164–171.
[23] H. Toyahara, M. Kawamura, Circulation of particles in the side regions of a tapered fluidized bed, Int. Chem. Eng. 33 (1993) 525–530.
 Update
Applied Mathematical Modelling
Volume 35, Issue 7, July 2011, Page 3651

DOI: https://doi.org/10.1016/j.apm.2011.03.026
 Applied Mathematical Modelling 35 (2011) 3651

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Corrigendum

Corrigendum to ‘‘Computational fluid dynamics and experimental

study of the hydrodynamics of a gas–solid tapered fluidized bed’’
[Appl. Math. Modell. 35 (2011) 2053–2584]
D.C. Sau a,⇑, S. Mohanty b, K.C. Biswal c
a
National Metallurgical Laboratory, Jamshedpur 831 007, Jharkhand, India
b
Institute of Minerals and Materials Technology, Bhubaneswar 751 013, Orissa, India
c
National Institute of Technology, Rourkela, Orissa, India

The paper has been published by mistake with Dr. Swati Mohanty’s name omitted in the author list. The author regrets
the error and apologizes for any inconvenience this may cause to the readers.

DOI of original article: 10.1016/j.apm.2010.11.037

⇑ Corresponding author. Tel.: +91 657 2345180 (O); fax: +91 657 2270527.
E-mail address: [email protected] (D.C. Sau).

0307-904X/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2011.03.026