Technical Report

STUDY OF GAS-SOLID FLUIDIZED

BED - TUNING SYAMLAL O’BRIEN
DRAG MODEL

GAURAV KATHURIA+ , ADITYA KARNIK* AND MOHIT TANDON*

+
Indian Institute of Technology, Hauz Khas, New Delhi - 110016. India.
*
CD-adapco, Pune - 411001. India.
Introduction
The inter-phase momentum transfer between gas and solid phase is accounted for by the drag force.
Several researchers [6, 4, 2, 3] have proposed correlations for modeling drag for gas - particle flows.
In this report, we will be focusing on the steps needed to tune the Syamlal O’Brien drag law.

Governing equations
At minimum fluidization condition, force balance yields the following equation -

AD
.(vmf ) = (1 − αg ).(ρs − ρg ).g. (1)
αg

where,

3 αs .αg .ρg
AD = . .CD .|vmf |. (2)
4 ds
The idea of the tuning process is to get correct estimate of CD for given minimum fluidization
velocity, vmf , such that equation 1 is satisfied.
Equation 1 can be re-cast in the following form,

3 3
CDs Re2ts = CD Re2t = Ar (3)
4 4
The above relation holds for a single particle as well as a cluster of particles. CDs is the drag
coefficient for a single particle. Rets is the Reynolds number based on terminal settling velocity of
single particle. The quantities without the s subscript are those for a group of particles. Ar, the
non-dimensional Archimedes number, is a function of the known quantities and given by,

ρg 3
Ar = d (ρp − ρg ) g. (4)
µ2g p
It is useful at this point to define Vr as the ratio of single particle terminal velocity to that of a
cluster of particles.

Ret
Vr = . (5)
Rets
Substituting for Vr in equation 3, yields following relation between single and multi-particle drag
coefficients,

CDs
CD = . (6)
Vr2

1
Syamlal-O’Brien Drag Law
The single particle drag law employed by Syamlal-O’Brien is that given by Dalla Valle:
· ¸2
4.8
CDs = 0.63 + √ . (7)
Rets
To complete the modeling, Syamlal-O’Brien choose a analytical formulation for Vr as proposed by
Garside and Al-Dibouni:
· q ¸
2 2
Vr = 0.5 A − 0.06Ret + 0.0036Ret + 0.12Ret (2B − A) + A (8)

where
A = ǫ4.14 (9)

and 
C2 ǫ1.28 ǫ ≤ 0.85
B= (10)
ǫ C 1 ǫ > 0.85

The default values of C1 and C2 are 2.65 and 0.8 respectively.

To ensure continuity of B, C1 and C2 are related by the following relationship:

log C2
C1 = 1.28 + (11)
log 0.85

Adjustment of Syamlal-O’Brien Drag law

From the known quantities, the Archimedes number can be estimated using equation 4. Substituting
eqn. 7 in eqn. 3 gives an explicit formula for Rets .

r q 2
4Ar
 4.82 + 2.52 3
− 4.8 
Rets =   (12)
 1.26 

For fluidized beds where information about the minimum fluidization velocity, Umf is available,
the multi-particle Reynolds number can be calculated as follows,

ρg Umf dp
Ret = . (13)
ǫmf µf
Note that since the minimum fluidization velocity is specified as the superficial velocity, we need
to divide it by the gas volume fraction at minimum fluidization condition, ǫmf . Typically we assume
ǫmf = 0.4.

2
 For small particles or for small Reynolds numbers (Ret < 20), the following expression obtained
from first term of Ergun equation can be used to estimate ǫmf ,

d2p (ρp − ρg ) g ǫ3mf

Umf = . (14)
150µg 1 − ǫmf
Knowing Rets and Ret , Vr can be estimated using eqn. 5. Note that during the tuning process,
the void fraction, ǫ, used in equations 9 and 10 should be ǫmf .
Finally, C2 , can be estimated from eqn. 8.
The parameter C1 is obtained from C2 using equation 11.

Demo case
To demonstrate the effectiveness of the described tuning process, the case reported by Taghipour
et.al.[5] has been chosen. Several simulations were computed using the Eulerian multiphase model in
STAR-CCM+. The parameters used in the simulations have been listed in Table 1.

Results and Discussions

According to the experimental results of Taghipour et.al. [5] the minimum fluidization velocity(determined
experimentally) and the pressure drop at that condition were 0.065 m/s and 4400 Pa, respectively.
The static bed height used in the present study is 0.47 m while that in the experiments is 0.4 m.
However according to the study of Karnik et. al.[1], the minimum fluidization velocity is independent
of the initial packing height of the column. The fluidization velocity at the inlet was varied from
0.9 - 1.5Umf , to check if the Syamlal model predicts correct fluidization conditions with original pa-
rameters. These parameters were then tuned using the procedure described before and the modified
parameters demonstrate better agreement with the experimental results. Simulations were run for 5s
of physical time and quantities have been averaged during the period of 3-5s. The screen-shots of the
column at 3s have been shown in the Figure 1, for different velocities with original value of C1 and
C2 (0.8 and 2.65 respectively).

3
 Table 1: Simulation parameters
Description Value Comment
3
Particle density, 2500 kg/m Glass beads
ρP
Gas density, ρg 1.225 kg/m3 Air
Mean Particle 275 µ m Uniform distribution
Diameter, dp
Restitution coef- 0.9 From literature
ficient, ess
Initial solids 0.60 Fixed value
packing, αs0
Superficial gas 0.14625-0.325 ms 0.9- 2Umf
velocity, U
Bed width 0.28 m Fixed value
Bed height 1m Fixed value
Static bed 0.47 m Fixed value
height
Grid interval 0.005 m Specified
spacing
Inlet boundary Velocity True velocity
condition
Outlet boundary Pressure
condition
Time-step size 0.001 s Specified
Maximum num- 20 Specified
ber of iterations

(a) (b) 0.9Umf (c) 1.0Umf (d) 1.1Umf (e) 1.5Umf

Volume
Frac-
tion of
Particle

Figure 1: Solid Volume Fraction snapshots at different fluidization velocities (C1 =0.8 and C2 = 2.65)

4
 The figure 1 clearly shows that no fluidization has occurred in the bed even with fluidization
velocity of 1.5Umf . This can be verified numerically by looking at the values of drag as predicted by
the Syamlal-O’Brien model at Umf , which is as follows:
The effective weight of particles in the bed taking into account the buoyancy force is = 1933.57 N,
while the drag value from Syamlal-O’Brien model is( AαDg .(vg - vs ).Volume of bed) = 914.537 N, which
is much less than the effective weight and hence we do not see any fluidization in the bed.
Thus it is very clear that the Syamlal-O’Brien drag model needs to be tuned to get better agreement
with the experimental results. In their numerical study, Taghipour et.al. [5] suggested C1 and C2
values of 9 and 0.28 respectively. These values were then employed in the present work and the
particle volume fractions in the bed at different velocities have been shown in Figure 2.

(a) (b) (c) (d) (e) 1.5Umf

Volume 0.9Umf 1.0Umf 1.1Umf
Frac-
tion of
Particle

Figure 2: Solid Volume Fraction snapshots at different fluidization velocities (modified C1 =9 and C2
= 0.28)

In the Figure 2, it can be seen that there are disturbances in the bed even at fluidization velocity
of 0.9Umf and an apparent bed expansion is also visible in all four cases. This suggests that with
the parameters employed by Taghipour et.al.[5], the fluidization occurs much before 0.9Umf . A quick
calculation of the value of drag with these parameters give the drag as = 2675.68, which is higher than
the effective bed weight suggesting that we should obtain an earlier fluidization with these parameters.

As has already been mentioned that the parameters can also be modified to satisfy equation 1. So,
C1 and C2 were calculated to satisfy the equation 1 (assuming αg = ǫmf = 0.4). This gives the value
of C1 and C2 as 7.23 and 0.38 respectively. The simulations were then carried out using these pa-

5
rameters and snapshots of the volume fraction of particles in the column have been shown in Figure 3.

(a) (b) (c) 1.0Umf (d) 1.1Umf (e)

Volume 0.9Umf 1.5Umf
Frac-
tion of
Particle

Figure 3: Solid Volume Fraction snapshots at different fluidization velocities (modified C1 =7.23 and
C2 = 0.38)

It is very apparent from the Figure 3 that bed does not fluidize till fluidization velocity of Umf
and disturbances in the bed along with expansion starts to occur at 1.1Umf . Also the value of drag
with these parameters at the minimum fluidization condition is = 1915.71 N, which is very close to
the effective weight of the bed.

Table 2: Height of the bed for different cases (Initial height of the bed being 47 cm)
U/Umf C2 = 0.8 C2 = 0.28 C2 = 0.38
0.9 39 cm 48.5 cm 45.5 cm
1.0 40 cm 50.25 cm 47 cm
1.1 40.5 cm 51.75 cm 48.25 cm
1.5 44 cm 87.75 cm 51.75 cm

The height of the bed for various cases was also calculated (having at least 10% of particles by
volume) and has been mentioned in Table 2 . This data clearly further endorses the idea that with
corrected parameters (i.e. tuning Syamlal-O’Brien drag model with the given method), fluidization
can be predicted rather accurately.

6
References
[1] Karnik A., Rangarajan A.M., and M.P. Tandon. Investigation of the Effect of Bed Height on
the Minimum Fluidization Velocity of a Fluidized Bed. 8th Intl. Conf. on Multiphase Flow, Jeju,
Korea, May 2013.

[2] H. Arastoopour, P. Pakdel, and M. Adewumi. Hydrodynamic analysis of dilute gas-solids flow in
a vertical pipe. Powder Technology, 62:163 – 170, 1990.

[3] R. Di Felice. The voidage functions for fluid-particle interaction system. Intl. J. Multiphase Flow,
20(1):153 – 159, 1994.

[4] M. Syamlal and T.J. O’Brien. The derivation of a drag coefficient formula from velocity-voidage
correlations. Technical Note, U.S. Department of energy, Office of Fossil Energy, NETL, Morgan-
town, WV, April 1987.

[5] F. Taghipour, N. Ellis, and C. Wong. Experimental and computational study of gassolid fluidized
bed hydrodynamics. Chemical Engineering Science, 60(24):6857 – 6867, 2005.

[6] C.Y. Wen and Y.H. Yu. Mechanics of fluidization. Chemical Engg. Prog. Symposium Series,
62(2):100 – 111, 1966.