Sei sulla pagina 1di 7

Hydrodynamics of Gas-Liquid-Solid Three Phase Fluidised Bed -

A CFD Study
Panneerselvam R#, Jikku M Thomas*, Savithri S#, Surender G.D#
#
Process Engineering & Environmental Technology Division,
Regional Research Laboratory (CSIR), Thiruvananthapuram
* Department of Chemical Engineering, Indian Institute of Technology Madras.

1. INTRODUCTION

Gas-Liquid–solid fluidised bed operation is of considerable importance in chemical, petrochemical,


biochemical and mineral processing industries. The successful scale up, design and operation of
such multiphase reactors depend mainly on the accurate prediction of phase holdups and their
distributions, flow regimes and patterns and mixing levels of individual phases. At present, the
understanding of hydrodynamics of three phase fluidized systems is far from complete because of
the complex interactions and interphase coupling of momentum, mass and energy between the
different phases. For this reason, computational fluid dynamics (CFD) has been promoted as a
useful tool for understanding multiphase reactors for reliable design and scale up. The most
important macroscopic quantities for characterizing such multiphase flows reactors are the phase
velocities, the holdups of the disperse phases, the pressure and density of each phase. The
knowledge of these hydrodynamic characteristics are essential in determination of rates of
heat/mass transfer and desired chemical reactions, and consequently of vital importance for the
reliable design and scale-up of such chemical reactors.

The present work aims to predict the phase velocities and holdup of dispersed phases in three
phase fluidised beds for various design and operating conditions using CFD. The data of S.D.Kim
et al. [3] is chosen for the purpose of validating the numerical results obtained through CFD. The
fluidised bed used in the experimental study of S.D.Kim [1, 2] is a plexiglas column of 0.254 m i.d.
with 2.5 m height. The liquid phase is chosen as water and gas phase is chosen as air. The solid
3
phase is chosen as glass beads of size 2.3 mm with a density of 2500 kg /m

2. LITERATURE REVIEW
Chen et al. [3] proposed three flow regimes viz, dispersed bubble regime, coalesced bubbly
regime and slug flow regime of three phase fluidised bed based on the superficial gas velocities.
Some authors have concentrated on the flow regime identification and flow transition from
dispersed bubble regime to coalesced bubbly regime. Recently Larachi et al. [4] used solid phase
tracer experiments for flow regime identification in three phase fluidised bed. Briens et al. [5] used
spectral analysis of the pressure fluctuation for identifying flow regime transition from dispersed to
coalesced bubbling flow regime.

The sequential development in the modeling of the hydrodynamics of three phase fluidised bed
can be classified into development of phenomenological model or the bubble wake model, fluid
dynamic modeling of continuum approach and development of advanced modeling techniques like
discrete phase computation. Bhatia & Epstein [6] presented a generalised wake model that
predicts the phase holdup of individual phases. Gidaspow et. al., [7] developed multifluid model for
gas, liquid and solid phase, in which gas phase is treated as a particulate phase (0.4 cm diameter
dispersed spherical air bubbles) and a kinetic theory granular flow model for solid phase. Mitra-
Majumdar et. al., [8] studied three phase modeling in vertical column and used modified drag
correlation between the liquid and gas to account for the effect of solid particles and between the
solid particles and liquid phase to account for the effect of gas bubbles. Recently Gidaspow et al,
[9] developed a predictive experimentally verified computational fluid dynamic (CFD) model for
gas-liquid-solid flow and used transient 3-D simulation of coupled Navier stokes equation for each
phase. Recently Zhang et al., [10] have developed two dimensional discrete phase computation for
gas-liquid-solid fluidisation and investigated thoroughly the dynamics of a single bubble rising in a
liquid-solid fluidised bed using Eulerian fluid dynamic (CFD) method, dispersed particle method
(DPM) and the volume-of-fluid (VOF) method for accounting for the flow of liquid, solid, and gas

1
phases, respectively. Due to the prohibitively large computational time required to track the large
number of particles or bubbles in discrete phase computation, the Eulerian–Eulerian method is
used in the present study.

Thus the present work aims to predict the phase velocities and holdups of dispersed phase in gas–
liquid-solid fluidised bed for different superficial liquid and gas velocities using Multi-fluid model
based on the Eulerian-Eulerian approach and validate the reported experimental data in the
literature with the results obtained in this study.

3. GOVERNING EQUATIONS OF MULTIFLUID MODEL

Hydrodynamic Model equations used for gas-liquid-solid fluidised bed are given below in Table 1.

Table 1 Governing Equation used in present study


Continuity Equations for k= (g ,s ,l )

∂ r
.(ε k .ρ k ) + ∇.(ρ k .ε k .u k ) = 0
∂t

Momentum Equations
Liquid phase (continuous phase)
∂ r r 2 r r
∂t
( )
.(ρ l .ε l .u l ) + ∇. ρ.ε l .u l = −ε l .∇P + ∇.τ l + ρ l .ε l .g + ∑m = g , s β.(u m − u l )

Gas/Solid phase k(=g, s) (Dispersed phase)

∂ r r 2 r r
∂t
( )
.(ρ k .ε k .u k ) + ∇. ρ.ε k .u k = −ε k .∇P − ∇p k + ∇.τ k + ρ k .ε k .g + ∑m−l , g ,s β fm .(u m − u k )
m≠ k

Constitutive Equations
The momentum transfer due to drag is
3 ε r r r r
Fd = C D ρ l s u s − u k (u s − u k )
4 dp
Drag models for liquid –solid system

Based on the Gidaspow Drag model


Ergun (1952)

2
150ε s µ l 1.75ε s ρ l (u l − u s )
β= 2
+ (ε l < 0.8)
ε ld p ε ld p
Wen and Yu (1966)
β=
3 (u − u s ) f (ε )
cd ε sρ l l (ε l > 0.8)
l
4 dp
Where
24
Cd =
Re
(
1 + 0.15Re P
0.687
) Re ≤ 1000 and
C d = 0.44 Re ≥ 1000
ρ ld p u l − u s
Re p =
µl

2
Drag models for liquid –gas system
Tomiyama’s correlation

24 74 8 E o
C d = Max[min[
Re p
(
1 + 0.15Re p
0.687
, ) ],
Re p 3 E o + 4
]ε l
−0.5

Where
g(ρ l − ρ g )d b
2

Eo =
σ
Viscous stress tensor for gas and liquid phases
 2 
( )
τ k = ε k µ k ∇ u k + ∇ u k + ε k  λ k − µ k ∇.u k I
 3 
Solid pressure model
Ps = ρ s ε s Θ s (1 + 2(1 + e )g 0 ε s )
Radial distribution function
−1
 1

  εs 3 

g 0 (ε ) = 0.61 −  
 
  ε sm  
 
granular temperature (Syamlal–O’Brien)

Θs =
( )
− K ε s tr D s +
2
( ) 2
( ( ) ( ))
K 1 tr 2 D s ε s + 4K 4 ε s K 2 tr 2 D s + 2K 3 tr D s
2ε s K 4
Where
K 1 = 2 (1 + e )ρ s g
ε sg o 2
K 2 = 4d s ρ s (1 + e ) − K3
3 π 0.5 3
 π 12 
 [1 + 0.4 (1 + e )(3e − 1)ε s g o ]

d ρ 3 (3 − e )
K3 = s s  
2 
+ 8 ε g
(1 + e ) 
 s 0 1 
 5π 2 
Solid shear viscosity
µ s = µ s, kin + µ s,F
4 2 Θ
µs,F = ε s ρ s d s g 0 (1 + e )
5 π
2
5 π ρ sd p  4 
µ s, kin =  1 + (1 + e )g o ε s  Θ
48 (1 + e )g 0  5 

Solid bulk viscosity


4 Θ
ξ = ε s 2 ρ s d s g 0 (1 + e )
3 π

3. NUMERICAL SIMULATION
ANSYS CFX-10 software code is used for simulating the hydrodynamics of three phase fluidised
bed. A 2-D simulation is carried out in the present study and Table 2 summarizes the model
parameters/conditions used for present simulation. Inlet boundary conditions are employed at the
bottom of the bed to specify a uniform inlet liquid and gas velocity. Pressure boundary conditions
5
are specified for outlet boundary conditions, which are set at a reference value of 1.013×10 Pa.
The lateral walls are modeled using the no-slip velocity boundary conditions for the liquid phase
and the free slip boundary conditions for the solid phase and gas phase.

The numerical simulations of the discrete governing equations were achieved by the finite volume
method. Pressure Velocity coupling was achieved by the SIMPLE algorithm. The governing
equations were solved using the advanced coupled multi grid solver.

3
Table 2
Simulation model parameters
3
Density of solid (kg/m ) 2500
Mean particle Size (mm) 2.3
3
Density of gas. Kg/m 1.185
13 ,16 ( From
Mean bubble size, mm
experimental data)
Initial bed height, m 0.39
Solid holdup 0.607
Bed voidage 0.317
Restitution coefficient, es
0.95
Superficial gas velocity Ug ,m/s
0.01, 0.04
Superficial liquid velocity Ul,m/s
0.06
Time steps
0.001 s
No of Nodes 39900

For time dependent solution the second order implicit time discretization was used. The
simulations were carried out till the system reached the pseudo steady state. The convergence
criteria for all the numerical simulation is based on monitoring the mass flow residual and the value
of 1.0e-04 is set as converged value.

4. RESULT AND DISCUSSION


4.1 Solid Motion

Figure 1(a) depicts a typical time averaged axial solid velocity profile under a superficial gas
velocity of 0.04 m/s and liquid superficial liquid velocity of 0.06 m/s It can be observed that there
are more number of circulation cells with upward movement of solid at the center region and
downward movement of solid at the wall region, such an observation is also reported by Limtrakul
et al. [11] in her experimental work. Figure 1(b) shows the bed expansion of solid with respect to
time.

t=0.0s t=2.0s t=3.0s t=5.0s t=10.0s t=15.0s


(a) (b)
Fig 1 (a) solid circulation profiles (b) bed expansion behaviors of solid at different time step at gas
superficial velocity of 0.04 m/s and liquid superficial liquid velocity of 0.06 m/s

4
4.2 Radial Distribution of Gas Holdup

Figure 2 shows the comparison of radial variation of gas holdup. The gas phase holdup profile
predicted by CFD matches closely with experimental data reported by S.D.Kim et al. [1, 2] and it
shows that gas phase holdup decreases with an increase in radial distance.

0.5

0.4
CFD Simulation
Gas holdup

0.3
Experimental
0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
Dim ensionless Radial Position

Fig 2 Radial distribution of gas-phase holdup at liquid superficial liquid velocity Ul=0.06 m/s and
gas superficial liquid velocity Ug = 0.04 m/s

4.3 Radial distribution of bubble velocity

Figure 3 shows bubble velocity profiles in the radial direction at liquid superficial velocity of 0.06
m/s and gas superficial velocity of 0.04 m/s. The bubble velocity profile predicted by CFD matches
closely with experimental data at the wall region of the column and varies at the center region of
the column. This deviation is due to fact that the interaction between the solid and gas phases is
not considered in the present study and this interaction term dominates at the center of the column
due to the large size and fast rise velocities of bubbles that occur at the center region.

2
Bubble Rise velocity, m/s

CFD Simulation
1.5
Experimental
1

0.5

0
0 0.2 0.4 0.6 0.8 1
Dimensionless Radial Position

Fig 3 Radial distribution of bubble velocity at liquid superficial liquid velocity Ul=0.06 m/s and gas
superficial liquid velocity Ug = 0.04 m/s

4.4 Radial distribution of liquid velocity

Liquid velocity profiles in the radial direction at liquid superficial velocity of 0.06 m/s and gas
superficial velocity of 0.04 m/s are shown in figure 3. The liquid velocity profile predicted by CFD
matches closely with experimental data at the wall region of the column and varies at the center
region of the column as in the case of bubble velocity prediction. The results further indicate the
presence of the maximum liquid velocity at the center of the column and reverse flow at the wall
region of the column. This recirculating flow is induced by the radial nonuniformity of gas phase
holdup and bubble rising velocity. As can be seen fig.3, the flow direction is changed from upward
to downward i.e., inversion point at the dimensionless radial position of 0.85 coinciding with the
values reported in the experiment work of S.D.Kim

5
0.8

0.6

Axial Liquid Velocity, m/s


CFD simulation
0.4
Experimental

0.2

-0.2
0 0.2 0.4 0.6 0.8 1

Dim ensionless Radial Position

Fig 4 Radial distribution of liquid velocity at liquid superficial liquid velocity Ul=0.06 m/s and gas
superficial liquid velocity Ug = 0.04 m/s

4.5 Effect of gas velocity on gas holdup

Figure 5 shows the effect of superficial gas velocity on the gas phase holdup. Since gas holdup in
three phase fluidised bed mainly depends the superficial gas velocity, gas holdup is found to
increase with increasing superficial gas velocity.
0.2
CFD Simulation Ug=0.01 m/s

CFD Simulation Ug=0.04 m/s


0.15
Gas phase holdup

Experimental Ug=0.01 m/s

Experimental Ug=0.04m/s
0.1

0.05

0
0 0.2 0.4 0.6 0.8 1

Dimensionless radial position

Fig 5 Radial distribution of gas phase holdup at different superficial gas velocity Ul=0.01 m/s,
0.04 m/s

5. CONCLUSION
CFD simulation of hydrodynamics of gas-liquid-solid fluidised bed is carried out by employing the
multifluid Eulerian-Eulerian approach. Adequate agreement is demonstrated for phase velocities
and dispersed phase holdup between CFD simulation results and experimental findings reported
by S.D.Kim et al. [1, 2]. In the time averaged sense, more circulation cells for the solids exist in a
three phase fluidized bed where solids ascend at the center of the column and descend near the
wall. The liquid velocity profiles predicted from CFD simulation exhibits maximum velocity at the
center of the column and downward velocity at the wall and the reverse flow of liquid is observed
at the dimensionless radial position of 0.85. The gas phase holdup is predicted to have a high
value at the wall and lower value in the center.

6. REFERENCE
[1] Y.H.Yu, S.D.Kim, Bubble Characteristics in the Radial direction of Three Phase Fluidised Beds,
AICHEJ. 34 (1988) 2069-2072.
[2] Y.H.Yu, S.D.Kim, Bubble Wake Model for radial Velocity profiles of liquid and Solid Phases in
Three Phase Fluidised beds, Ind.Eng.Chem.Res. 40 (2001) 4463-4469.
[3] R.C. Chen, L.S. Fan, Flow structure in a three-dimensional bubble column and three-phase
fluidized bed, AICHE J. 40 (1994) 1093.

6
[4]. M.S Fragu, M.C. Cassanello F. Larachi, J. Chaouki, Flow Regime Transition Pointers In Three-
Phase Fluidized Beds Inferred From a Solid Tracer Trajectory, Chem. Eng. Proc. 45 (2006)
350–358
[5] L. A. Briens, N.Ellis, Hydrodynamics Of Three-Phase Fluidized Bed Systems Examined By
Statistical, Fractal, Chaos And Wavelet Analysis Methods, Chem.Eng.Sci.60 (2005) 6094 –
6106
[6] V.K.Bhatia, N.Epstein, Three-phase Fluidization: A Generalised Wake Model. In Fluidization
and its Application; Keairns, Ed., Hemisphere, Bristo PA 1974 Vol.1
[7] M.Bahary, Experimental and Computational Studies of Hydrodynamics in Three-Phase and
Two-Phase Fluidised Beds, Ph.D. Thesis, Illinois Institute of Technology,
Chicago,Illiniois,1994.
[8] D. Mitra-Majundar, B. Farouk, Y.T. Shah, Hydrodynamic Modeling of Three-Phase Flows
through a Vertical Column. Chem. Eng. Sci. 52 (1997) 4485–4497.
[9] D. Matonis, D. Gidaspow, M. Bahary, CFD Simulation of Flow and Turbulence in Slurry Bubble
Column, AICHE J. 48 (2002) 1413–1429.
[10] J.P. Zhang, Y. Li, L.S. Fan, Discrete Phase Simulation of Gas–Liquid–Solid Fluidization
Systems: Single Bubble Rising Behavior. Powder Tech. 113 (2000b) 310–326.
[11] S.Limtrakul, Hydrodynamics of liquid fluidized beds and gas–liquid fluidized beds. D.Sc.Thesis
1996, Washington University, USA.

Potrebbero piacerti anche