Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

10.

Partially Aerated Bubble Column


10.1 Problem Description
Gas-liquid bubble ow is studied in a at bubble column with rectangular
cross-section and essentially two-dimensional ow pattern. The main dimen-
sions of the experimental apparatus are a length of 0.5 m and height of 1.5 m,
which give a height-to-length ratio of 3. (See Figure 10.1.1.)

pressure inlet

water:
ρ=1000 kg/m3
µ=9e-4 kg/m-s

0.5 m 1.5 m

15 cm
air:
ρ=1.293 kg/m3
µ=1.72e-5 kg/m-s
v=6.64e-4 m/s
inlet width=3 cm

Figure 10.1.1: Problem Speci cation

The gas is injected into the water column through a small inlet located 15 cm
away from the left side wall. The inlet is 3 cm wide, and the air's velocity at
the inlet is very small, 6.6e-4 m/s.

c Fluent Inc. February 25, 1999 10-1


Partially Aerated Bubble Column

The physical properties for water and air are as follows:

Physical Properties Water Air


Density (kg/m )3
1000 1.293
Viscosity (kg/m-s) 9e-4 1.72e-5

The mean diameter for the air bubbles is 1 cm.

10.2 Grid Setup


A rectangular grid with 35  49 live cells is used. Non-uniform weighting
factors are used at the bottom of the container, where the bubbles enter the
domain.

X-GRID SEGMENT INFORMATION


WEIGHTING-FACTORS
SEGMENT START-POINT LENGTH # CELLS START-POINT END-POINT
1 0.0000E+00 5.0000E-01 35 0.0000 0.0000

Y-GRID SEGMENT INFORMATION


WEIGHTING-FACTORS
SEGMENT START-POINT LENGTH # CELLS START-POINT END-POINT
1 0.0000E+00 1.5000E+00 49 6.0000 1.0000

10.3 Boundary Cells


The ow enters the domain through the small velocity inlet at the bottom of
the container and the top of the column is modeled with a pressure inlet. The
sides and bottom of the container are modeled with walls.

10-2
c Fluent Inc. February 25, 1999
10.3 Boundary Cells

Y Partially aerated bubble column Jul 07 1998


Z X Grid ( 36 X 50 ) Fluent 4.50
Fluent Inc.

Figure 10.2.1: Grid for the Partially Aerated Bubble Column

51
50

40

30

20

10

J1I1 10 20 30 37
Partially aerated bubble column Sep 18 1998
I = 26
Computational Grid Fluent 4.51
J = 49 Fluent Inc.

Figure 10.3.1: Boundary Cell De nition for the Partially Aerated Bubble Col-
umn


c Fluent Inc. February 25, 1999 10-3
Partially Aerated Bubble Column

10.4 Physical Models


Many publications (e.g., [1] and [2]) have been devoted to the description of
bubble column problems and various models for them. The main issues are
listed below:

 Formulation depends on how the dispersed phase is treated: Euler-Euler


or Euler-Lagrange
 Steady state or dynamic simulation
 Numerical schemes that avoid numerical di usion
 Turbulence modeling
In the Euler-Lagrange model, the ow eld is rst calculated from the balance
equations of the quasi-homogeneous gas-liquid dispersion, and then the local
gas holdup is determined by tracking the individual bubbles. This type of
formulation has several advantages, including straightforward consideration of
bubble-bubble and bubble-liquid interaction e ects and no numerical di usion.
There is a main drawback, however: when the gas concentration increases, it
is impossible to track all of the individual bubbles.
The Euler-Euler formulations do not require as much computer memory stor-
age as the Euler-Lagrange formulations, unless the grid is very ne. However,
very accurate solution techniques are required since the coupled balance equa-
tions for gas and liquid have to be solved simultaneously. Some numerical
di usion may appear if the discrete momentum sources are not accurately de-
scribed, but the use of very accurate numerical solution algorithms avoids this
problem.
A review of literature shows that computer power limitations have obliged
analysts to run steady-state formulations, despite experimental evidence that
the ow structures of bubble columns are dynamic. Since the dynamic models
can compute the vortices accurately and the transient simulation with FLUENT
is fast enough, an unsteady simulation has been performed here.
It is necessary to use accurate numerical schemes to avoid the numerical di u-
sion e ect that can appear in convection terms. Simulations have shown that
rst-order schemes are unable to properly simulate bubbly ows (bubbles are
carried vertically, and no vortices appear). The discretization schemes chosen
for this simulation are QUICK for both velocity components and power law
for the volume fraction.
Most of the publications on bubble columns use the standard k- turbulence
model (see reviews [1] and [2]) to account for the in uence of the rising gas

10-4 c Fluent Inc. February 25, 1999


10.4 Physical Models

bubbles in the turbulent viscosity of the gas-liquid ow. However, this model
leads to a viscosity increase of several orders of magnitude, which dampens the
transient character of the ow and eliminates the vortices that are observed
in the experiments.
Our simulations have been carried out considering laminar gas-liquid ow,
which leads to good qualitative agreement with experiments. There are, how-
ever, several ways of improving the results, including the following:
 Increase the liquid viscosity arti cially, in order to account for the in-
uence of the rising gas bubbles upon the turbulent viscosity of the
gas-liquid mixture.
 Use the large eddy simulation (LES) method to model turbulence. (Cur-
rently, this method requires a very ne grid and excessive computer time
and storage.)
For this simulation, the Eulerian-Eulerian multiphase model is used, with the
drag law of Schwarz et al.
These settings can be speci ed using the Models and Multiphase Parameters
panels
De ne ,!Models...
or the MULTIPLE-PHASES and DRAG-LAW text commands.
SETUP1 ,! DEFINE-MODELS ,!MULTIPLE-PHASES
MULTIPLE-PHASES ,! MULTIPHASE-OPTIONS ,! DRAG-LAW
The rest of the multiphase parameters are set using the Multiphase Parameters
panel or the SOLUTION-PARAMETERS text command:
MULTIPLE-PHASES ,! MULTIPHASE-OPTIONS ,! SOLUTION-PARAMETERS

(SOLUTION PARAMETERS)
YES STABILITY TERM IN P.D.E
1.0000E-01 FALSE TIME STEP FOR UNDER-RELAXATION (DIM)
1.0000E-07 LOWER LIMIT FOR ALL VOFs (DIM)
YES OUTLET PRIMARY PHASE MASS CORRECTION
NO REVERSE THE MOMENTUM EXCH. COEFFICIENT
NO COUPLED TDMA FOR MULTIPHASE (ELSE FEA)

The gravitational acceleration is set to ,9:81 m/s2 in the y direction in the


BODY-FORCES table.
EXPERT ,!BODY-FORCES


c Fluent Inc. February 25, 1999 10-5
Partially Aerated Bubble Column

10.5 Boundary Conditions


The boundary conditions for the inlet are a volume fraction of 1 and a uniform
vertical velocity of 6.6e-4 m/s for the air phase. Atmospheric pressure is
speci ed at the top of the container.

- VELOCITY/PRESSURE BOUNDARY CONDITIONS -

ZONE PHASE U-VEL. V-VEL. NORMAL PRESSURE


------- -------- -------- -------- -------- --------
W1 WATER 0.00E+00 0.00E+00 N/A N/A
AIR 0.00E+00 0.00E+00 N/A
I1 WATER 0.00E+00 0.00E+00 N/A N/A
AIR 0.00E+00 6.60E-04 N/A
I2 FIXED PRESSURE 0.00E+00

- VOLUME FRACTION BOUNDARY CONDITIONS -

ZONE AIR
---------- ----------
W1 LINK CUT
I1 1.00E+00
I2 0.00E+00

10.6 Calculation Procedure


As explained earlier, rst-order discretization schemes do not lead to an ac-
curate representation of the ow. The simulation starts with a second-order-
upwind scheme for velocities and volume fraction (for the sake of stability),
but after one second of simulation, the QUICK scheme is enabled for velocities
and the power-law scheme is enabled for the interpolation of volume fraction.

10.7 Solver Parameters


The underrelaxation factors used for the calculations are as follows:
Variable Underrelaxation Factors
water velocities 0.6
air velocities 0.6
air volume fraction 1.0
pressure 0.6
body forces 1.0
viscosity 1.0

10-6 c Fluent Inc. February 25, 1999


10.7 Solver Parameters

These values are set in the Underrelaxation panel or the UNDERRELAX-1 table.
Solve ,! Controls ,!Underrelaxation...
EXPERT ,!UNDERRELAX-1
The multigrid solver will be used only for the pressure correction equation,
and the solver sweeps are set as shown below.

Variable Number of Sweeps


pressure correction 50
u velocity 1
v velocity 1
air u velocity 1
air v velocity 1
air volume fraction 1

A time step of 0.005 second is used for the simulation.


c Fluent Inc. February 25, 1999 10-7
Partially Aerated Bubble Column

10.8 Results
Figure 10.8.1 shows the good convergence behavior for the simulation. The
vortices visible in Figure 10.8.2 have been captured because of the choice of the
discretization scheme (QUICK for velocities, power law for volume fraction).
Figure 10.8.4 shows further evolution of the vortices in time. At t = 60 the
vortices are larger than at t = 20, and there are fewer of them.

- Pressure 1.000E-01
- Water U
- Water V
- Air U
1.000E-02
- Air V
- Air Vof

1.000E-03

1.000E-04

1.000E-05

1.000E-06
0 2000 4000 6000 8000 10000 12000
Time Steps
Partially aerated bubble column Sep 21 1998
Normalized Residuals Fluent 4.51
Fluent Inc.

Figure 10.8.1: Normalized Residuals

10-8 c Fluent Inc. February 25, 1999


10.8 Results

2.16E+00
1.90E+00
1.65E+00
1.40E+00
1.15E+00
8.97E-01
6.45E-01
3.93E-01
1.42E-01
-1.10E-01
-3.62E-01
-6.14E-01
-8.65E-01
-1.12E+00
-1.37E+00
-1.62E+00
-1.87E+00
-2.12E+00
-2.38E+00
-2.63E+00
-2.88E+00
-3.13E+00
-3.38E+00
-3.63E+00
-3.89E+00
-4.14E+00
-4.39E+00
-4.64E+00
-4.89E+00
-5.14E+00
-5.40E+00

Y Partially aerated bubble column Jul 07 1998


Z X Water Stream Function (M2/S) Fluent 4.50
Max = 2.155E+00 Min = -5.396E+00 Time = 2.000E+01 Fluent Inc.

Figure 10.8.2: Water Stream Function and Velocity Vectors for t = 20 sec

1.00E-02
9.67E-03
9.34E-03
9.01E-03
8.68E-03
8.35E-03
8.02E-03
7.69E-03
7.36E-03
7.03E-03
6.70E-03
6.37E-03
6.04E-03
5.71E-03
5.38E-03
5.05E-03
4.72E-03
4.39E-03
4.06E-03
3.73E-03
3.40E-03
3.07E-03
2.74E-03
2.41E-03
2.08E-03
1.75E-03
1.42E-03
1.09E-03
7.60E-04
4.30E-04
1.00E-04

Y Partially aerated bubble column Jul 07 1998


Z X Air Volume Fraction Fluent 4.50
Max = 1.000E-02 Min = 1.000E-04 Time = 2.000E+01 Fluent Inc.

Figure 10.8.3: Air Volume Fraction and Water Velocity Vectors for t = 20 sec


c Fluent Inc. February 25, 1999 10-9
Partially Aerated Bubble Column

1.18E+01
1.10E+01
1.03E+01
9.56E+00
8.82E+00
8.09E+00
7.35E+00
6.61E+00
5.87E+00
5.13E+00
4.39E+00
3.65E+00
2.91E+00
2.17E+00
1.43E+00
6.96E-01
-4.32E-02
-7.82E-01
-1.52E+00
-2.26E+00
-3.00E+00
-3.74E+00
-4.48E+00
-5.22E+00
-5.95E+00
-6.69E+00
-7.43E+00
-8.17E+00
-8.91E+00
-9.65E+00
-1.04E+01

Y Partially aerated bubble column Jul 07 1998


Z X Water Stream Function (M2/S) Fluent 4.50
Max = 1.178E+01 Min = -1.039E+01 Time = 6.000E+01 Fluent Inc.

Figure 10.8.4: Water Stream Function and Velocity Vectors for t = 60 sec

1.00E-02
9.67E-03
9.34E-03
9.01E-03
8.68E-03
8.35E-03
8.02E-03
7.69E-03
7.36E-03
7.03E-03
6.70E-03
6.37E-03
6.04E-03
5.71E-03
5.38E-03
5.05E-03
4.72E-03
4.39E-03
4.06E-03
3.73E-03
3.40E-03
3.07E-03
2.74E-03
2.41E-03
2.08E-03
1.75E-03
1.42E-03
1.09E-03
7.60E-04
4.30E-04
1.00E-04

Y Partially aerated bubble column Jul 07 1998


Z X Air Volume Fraction Fluent 4.50
Max = 1.000E-02 Min = 1.000E-04 Time = 6.000E+01 Fluent Inc.

Figure 10.8.5: Air Volume Fraction and Water Velocity Vectors for t = 60 sec

10-10
c Fluent Inc. February 25, 1999
10.8 Results

2.500E-03

2.000E-03

1.500E-03

1.000E-03

5.000E-04

0.000E+00
0.000E+00 1.000E-01 2.000E-01 3.000E-01 4.000E-01 5.000E-01
I-DIRECTION LENGTH (M)
Partially aerated bubble column Aug 19 1998
Node Values Along J-Position = 26 Fluent 4.51
Air Volume Fraction Vs. I-Direction Length (M) Fluent Inc.

Figure 10.8.6: Air Volume Fraction along J=26

- Water Velocity Magnitude


4.000E-01
(M/S)
- Air Velocity Magnitude
(M/S)

3.000E-01

2.000E-01

1.000E-01

0.000E+00
0.000E+00 1.000E-01 2.000E-01 3.000E-01 4.000E-01 5.000E-01
I-DIRECTION LENGTH (M)
Partially aerated bubble column Aug 19 1998
Node Values Along J-Position = 26 Fluent 4.51
Fluent Inc.

Figure 10.8.7: Water and Air Velocity Magnitude along J=26


c Fluent Inc. February 25, 1999 10-11
Partially Aerated Bubble Column

10.9 References
1. Sokolichin, A. and Eigenberger, G., Gas-Liquid Flow in Bubble Columns
and Loop Reactors: Part I. Detailed Modeling and Numerical Simula-
tion, Chem. Eng. Science, 49(24B):5735{5746, 1994.
2. Jakobsen, H.A., Sannaes, B.H., Grevskott, S., and Svendsen, H.F., Mod-
eling of Vertical Bubble-Driven Flows, Ind. Eng. Chem. Res., 36(10):4052{
4074, 1997.

10.10 Test Details


Solver: FLUENT 4.5
Test Date: Mar. 5, 1998
Platform: Sun Ultra
Case File: PARTIAL.CAS
Data Files: PARTIAL 4000
PARTIAL 12000
Journal Files: RUN.COM
POST.COM
Pro le Files: PARTIAL 1.PF
PARTIAL 2.PF
Transcript File: PARTIAL.TRN
The les associated with this case are arranged as follows. The FLUENT case
le(s) are in the top-level bubbl1/fluent/v4.5 directory. FLUENT journal
les to run the case(s) are in the subdirectory bubbl1/fluent/v4.5/std. Data
and pro le les generated at Fluent Inc. are in bubbl1/fluent/v4.5/std/OUT.
If you wish to rerun this validation example automatically on your own plat-
form, please follow these steps:

1. Change directories to the bubbl1/fluent/v4.5/std subdirectory.


2. Create a subdirectory called out. The journal les will save the data
les to this subdirectory (and not overwrite the data les supplied by
Fluent Inc., which are in the OUT subdirectory). The journal le will not
function unless this out subdirectory exists.
3. Start FLUENT.
4. Read the journal le(s) ( rst for the calculation run, then for the post-
processing run).

The RUN.COM journal le will

10-12 c Fluent Inc. February 25, 1999


10.10 Test Details

1. Read the case le.


2. Adjust solution control parameters.
3. Adjust physical models, properties, and boundary conditions.
4. Calculate.
5. Write data le(s) to the out subdirectory.
6. Quit from FLUENT.

The POST.COM journal le will

1. Read the rst data le.


2. Save the XY pro les for the plots shown in this writeup.
3. Read the next data le, and save the pro les, etc.
4. Quit from FLUENT.

You can then restart FLUENT, read in the case and data les, examine the
results, and compare them with the printed results in this writeup. You can
also read the XY pro les saved by the POST.COM journal le, generate the XY
plots, and compare them with the results shown here.
When carrying out any comparison, bear in mind that minor di erences in
numerics from platform to platform are quite common. Typically, such dif-
ferences a ect the residual and variable histories. For transient multiphase
calculations involving thousands of time steps (especially bubble columns),
roundo errors can accumulate and lead to some quantitatively appreciable
di erences in the nal results. While the results on di erent platforms may
not be quantitatively equivalent, the qualitative results are generally compa-
rable across platforms.


c Fluent Inc. February 25, 1999 10-13
Partially Aerated Bubble Column

10-14
c Fluent Inc. February 25, 1999

You might also like