Chemical Engineering Science 59 (2004) 5201 – 5207

www.elsevier.com/locate/ces

CFD modeling of bubble columns flows: implementation of

population balance
P. Chena , J. Sanyalb , M.P. Dudukovica,∗
a Chemical Reaction Engineering Laboratory, Department of Chemical Engineering, Washington University, One Brookings Drive,
Campus box 1198, St. Louis, MO 63130, USA
b Fluent Inc., Lebanon, NH 03766-1442, USA

Received 3 March 2004

Available online 3 September 2004

Abstract
The Bubble Population Balance Equation is implemented both in the two-fluid model and in the Algebraic Slip Mixture Model. Two
and three-dimensional numerical simulations of two-phase (air–water) and pseudo-two-phase (air–Therminol-Glass beads) transient flows
are performed for laboratory scale bubble columns of two different diameters operated over a range of superficial gas velocities (8 to
20 cm/s) in the churn turbulent regime. The ability of the simulations to predict liquid recirculation and gas radial holdup profile, which
drives the recirculation, is illustrated. The simulations also capture the effects of increased superficial gas velocity and of surface tension
on bubble size distribution in agreement with observations. At all conditions a uni-modal bubble size distribution is predicted.
䉷 2004 Elsevier Ltd. All rights reserved.

Keywords: Bubble columns; Computational fluid dynamics; Dynamic simulation; Population balance; Modeling; Multiphase flow

1. Introduction dispersion, driven by bubble wakes and meandering path-

Bubble column reactors are widely used. In their design,
scale-up and scale-down of such reactors the understanding
of the fluid dynamics is important. From the industrial point
of view, churn-turbulent flow is of most interest as it ensures
high volumetric productivity. Yet this flow is poorly under-
stood. An important task is to describe the large scale mix-
ing of the liquid and gas phase in columns operated in the
churn-turbulent flow regime. It has been demonstrated that
gas holdup radial profile is almost parabolic (Kumar et al.,
1997) with the highest value at the center of the column. This
profile becomes column height invariant after one to 1.5 di-
ameters from the sparger (George et al., 2000; Shollenberger
et al., 2000; Ong, 2003) and it drives, in a time-averaged
sense, a single cell liquid recirculation (Degaleesan, 1997).
Superimposed on this large scale convective motion, is eddy
∗ Corresponding author. Tel.: +1-314-935-6021; fax: +1-314-935-4832.
E-mail address: dudu@poly1.che.wustl.edu (M.P. Dudukovic).
ways, with axial diffusivity exceeding radial by an order of
magnitude (Degaleesan, 1997). This experimental evidence,
imbedded in phenomenological physically based models
has been highly successful in predicting liquid (Degaleesan
et al., 1996) and gas phase mixing (Gupta et al., 2001).
Since such phenomenological models can predict liquid and
gas tracer dynamic responses faithfully (Degaleesan et al.,
1996; Gupta et al., 2001), it is important to examine whether
the parameters needed by such models can be computed
by CFD. This requires computation of gas radial holdup
distribution, liquid recirculation velocity, eddy diffusivities,
mean bubble size and interfacial area per unit volume.
For the churn turbulent flow regime only the Euler–Euler
k-fluid model or ASMM seem practical. However, most pre-
vious attempts (e.g. Sanyal et al., 1999) utilized an assumed
mean bubble size for evaluation of the drag. This assumed
size was often adjusted based on attempts to match model
predictions to some measured result. Hence, the model was
not fully predictive. Moreover, even with 3D simulations

0009-2509/$ - see front matter 䉷 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2004.07.037
5202 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207
holdup profiles were not always predicted well and inter-
facial areas were pre-judged by assumption of the bubble
size. In addition, in churn-turbulent flow, bubble–bubble in-
teractions result in widely distributed bubble sizes that may
be substantially different from the “mean” bubble size as-
sumption. A remedy to this situation can be developed by
implementation of the bubble population balance equation
(BPBE) in the Eulerian–Eulerian model or in ASMM. This
eliminates the assumption of single constant bubble size and
should improve the gas holdup profile prediction. Most im-
portantly, the implementation of BPBE allows one to predict
the bubble size distribution locally, and eliminates the trial-
and-error procedure regarding the unknown mean bubble di-
ameter mentioned above, while also providing the estimate
of the interfacial area concentration locally throughout the
column.
2. Experimental techniques used
At the Chemical Reaction Engineering Laboratory, Wash-
ington University in St. Louis, the unique Computer Auto-
mated Radioactive Particle Tracking (CARPT)–Computed
Tomography (CT) facilities allow non-invasive monitoring
of the velocity and holdup profiles of two phases in opaque
multiphase reactors on a single platform. Details of the
experimental setup and the procedures for particle track-
ing and the tomographic techniques are available elsewhere
(Devanathan et al., 1990; Yang et al., 1992; Kumar et al.,
1995, 1997; Degaleesan, 1997; Chen et al., 1999).
3. Computational models
3.1. The Eulerian multiphase model
The governing equations in this approach can be derived
by ensemble averaging the fundamental conservation equa-
tions for each phase to describe the motion of liquid and gas
in a bubble column. Both the continuous and the dispersed
phase are modeled in the Eulerian frame of reference as in-
terpenetrating continua. The mass and momentum balance
equations are written for each phase separately. The momen-
tum equations of the phases interact with each other through
inter-phase momentum exchange terms. The model details
can be found elsewhere (e.g., Drew, 1983). The equations
used have been reported earlier (Sanyal et al., 1999; Pan
et al., 2000; van Baten and Krishna, 2001). In the present
computation, we include the drag force only, while the added
mass force and lift force are neglected in the model for inter-
phase momentum exchange. The drag force coefficient as
suggested by Schiller and Naumann (1935) is used

 1+0.15Re0.687 
24 Re Re
1000
CD = . (1)
0.44 Re > 1000
The local bubble diameter is used as input parameter for
the simulation. In the past it was either estimated for the op-
erating conditions used from former experience or by a trial-
and-error method by comparing model prediction with data.
Sometimes it was calculated from available mean bubble
diameter correlations (e.g., Wilkinson, 1991). In this study
we obtain the local bubble diameter via the solution of the
bubble population balance equation.
Turbulence in the liquid phase is modeled through a set of
modified k-ε equations with extra terms that include inter-
phase turbulent momentum transfer (Launder and Spalding,
1974; Elghobashi and Abou-Arab, 1983). For the dispersed
gas phase, turbulence closure is effected through correlations
from the theory of dispersion of discrete particles by homo-
geneous turbulence (Tchen, 1947). The full set of equations
is described by Sanyal et al. (1999).
3.2. The algebraic slip mixture model
The algebraic slip mixture model (ASMM) is essentially
a simplified Eulerian–Eulerian model. Like in the Eulerian–
Eulerian approach, the ASMM assumes the phases to be
completely interpenetrating and allows the phases to move at
different velocities. However, unlike the Eulerian–Eulerian
model, the mixture model does not require the solution of
mass and momentum balance equations for each phase,
but only for the mixture phase. This simplification is made
based on the assumption that the phases are in local equi-
librium over short spatial length scales. This implies that
the accelerating dispersed phase particles/bubbles attain the
terminal velocity after traveling a distance which is much
smaller than the system’s length scale. In other words, the
dispersed phase entity (bubble) always slips with respect
to the continuous phase at its terminal Stokes’ velocity in
the local acceleration field. In ASMM, the standard k–ε
model is used to model the mixture turbulence. The details
of the ASMM can be found elsewhere (Manninen et al.,
1996).
3.3. Bubble population balance equation
In bubble columns, the initial bubble size is determined
by the formation of bubbles at the sparger. However, this
initial bubble size may not be stable due to turbulence, in-
terfacial instability, wake entrainment, size dependent rise
velocity difference and shear layer induced velocity dif-
ference. In all these cases, the bubble size is further de-
termined by a break-up and/or coalescence mechanism. In
reaction systems, in addition to the break-up and coales-
cence, mass transfer should also been considered. Phase
change and pressure change may also need to be taken into
account.
In the present implementation of BPBE in the two-fluid
model, all the bubbles are assumed to move at identical ve-
locity which equals the local ensemble averaged gas phase
 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207
velocity, ug , obtained from the solution of the two-fluid
model. One can theoretically solve the continuity and mo-
mentum equations for N gas phases (N equals to the number
of distinct bubble sizes to be tracked) and the liquid phase.
However, it is very difficult, if not impossible, to get a large
number of highly non-linear momentum equations to con-
verge in a realistic period of time (e.g., 2(N + 1) equations
in 2D and 3(N + 1) in 3D). To circumvent the problem of
solving momentum equations for N + 1 separate phases,
only the continuity and momentum equation for two phases
(gas and liquid) are solved. The gas phase is discretized into
n subclasses according to bubble size. The number density
based population balance equation for the ith bubble class
can be written as
* Here we focus on describing the effect of the number
ni + ∇ · (ug ni ) = Si , (2)
*t
where ni is ith bubble class local number density. Reaction
and phase change are neglected in the present study and the
source term due to pressure change is neglected as well.
Therefore, the source term Si is only due to breakup and co-
alescence of bubbles. The appropriate closures are discussed
briefly below.
Eq. (2) is solved along with the two-fluid model in a
sequential manner to obtain for each bubble class its local
number density ni . The local mean bubble size, which is
needed to calculate the drag force, is given as
N
i=1 ni di3
db = N . (3)
i=1 ni di
2
We have also implemented BPBE in the ASMM in two
ways. In one we have again allowed all bubbles to move at
the same velocity, and in the other the bubbles move at the
velocity characteristic of that bubble class with respect to
the liquid phase depending on the diameter of a given bub-
ble class. For the results presented here, different bubbles
classes have different velocity in ASMM implementation.
Chen (2004) showed that the difference in the implementa-
tion method does not have significant effect on the simulated
results.
4. Results and discussion
In our studies (Chen et al., 2004a,b) we have examined
the effect of various closures for coalescence and breakup.
The type of closure used did not have a great effect on the
computed holdup and velocity field. However, in all cases,
to match the experimental observations of fully developed
flow (e.g., gas holdup radial profile invariant with column
height) it was necessary to increase the breakup rate pre-
dicted by the model by a factor of ten. Hence, the Luo and
Svendsen’s (1996) breakup model, adjusted as described
5203
above, and Luo’s (1993) coalescence model, are used for
the results presented here. The most likely reason for the
mismatch in magnitude between the calculated breakup and
coalescence rate is the inability of the k–ε model to provide
physically realistic values of the turbulent energy dissipa-
tion (Bertola et al., 2003). A better turbulence model, which
properly models the energy containing eddies, is needed in
the future.
We have shown (Chen et al., 2004a,b) that the simulations
from the BPBE produce much better comparison with data
for holdup profiles and liquid velocity than “unoptimized”
predictions based on single bubble diameter calculated from
correlations. We have also shown that the Euler–Euler model
and ASMM with BPBE implemented in each yield compa-
rable results.
of bubble classes tracked in the BPBE on the predicted re-
sults. We also consider the ability of the model to predict
an increase in the number of large bubbles with the increase
in superficial gas velocity and its ability to account for the
surface tension effect. In each simulation the initial bubble
size at the sparger was calculated from Miyahara et al.’s
(1983) correlation and used as the boundary condition. With
the previously discussed coalescence and adjusted breakup
model a stable holdup profile and bubble size distribution is
always obtained after 1–1.5 column diameters in agreement
with data.
To investigate the effect of the number of bubble classes,
9 and 16 classes (Table 1) were used to solve the popu-
lation balance equation. Two-dimensional axis-symmetric
simulations, via ASMM implementation, were performed
for an 8 in diameter (19.0 cm i.d.) air–water cylindrical
bubble column, The gas superficial velocity simulated was
12 cm/s, characteristic of churn-turbulent flow. In Fig. 1a
the computed time-averaged liquid axial velocity profiles
are compared with the experimental results obtained by
CARPT. The centerline axial velocity is somewhat over-
predicted by simulations. However, the general overall
shape of the velocity profile is captured and the agreement
with data improves as one moves radially outwards in the
column. Moreover, the cross-over point (i.e., the radial lo-
cation where the axial velocity component becomes zero)
is well predicted. Fig. 1b shows the comparison for the
time averaged radial gas holdup profile obtained from the
simulations and from Computed Tomography (CT). The
assumption of cylindrical axi-symmetry prevents lateral
motion of the dispersed gas phase and leads to an unreal-
istic gas holdup distribution wherein a maximum holdup
is away from the centerline. This indicates the necessity
of three dimensional simulation. The difference in the pre-
dicted results using 9 and 16 bubble classes are not signifi-
cant. Hence 9 bubble classes are used in most of the other
simulations.
Three dimensional simulations were performed with
BPBE implemented in the two-fluid model framework and
9 bubble classes tracked. Quantitative comparisons for
5204 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207

Table 1
Bubble classes tracked in simulation

9 classes
Class index 1 2 3 4 5 6 7 8 9
Bubble diameter (mm) 1.00 1.60 2.50 4.00 6.35 10.08 16.00 25.40 40.00
. . . .
16 classes
Class index 1 2 3 4 5 6 7 8
Bubble diameter (mm) 1.00 1.26 1.60 2.00 2.50 3.20 4.00 5.04
Class index 9 10 11 12 13 14 15 16
Bubble diameter (mm) 6.35 8.00 10.08 12.70 16.00 20.16 25.40 32.00

60 0.4

50 0.35
16 Class
Axial Liquid Velocity, cm/s

40 0.3
9 Class

Gas Holdup
30 CARPT 0.25

0.2 16 class
20
9 class
0.15
10 CT
0.1
0
0 0.2 0.4 0.6 0.8 1 0.05
-10
Dimensionless Radius 0
-20
0 0.2 0.4 0.6 0.8 1
-30 (b) Dimensionless Radius
(a)

Fig. 1. Comparison of time-averaged (a) axial liquid velocity and (b) gas holdup distribution via 2D simulation (DC = 19 cm, Ug = 12.0 cm/s, P = 1 bar).

predicted mean axial liquid velocity and gas holdup profile
with data are obtained for columns operated over a wide
range of superficial velocity, operating pressure, physical
properties, and column diameter (Chen et al., 2004b). Such
three dimensional simulations produce a better agreement
with holdup data. Moreover they are superior to the sim-
ulations based on single predicted bubble size. The lack
of suitable bubble size distribution data in churn turbulent
flow makes it difficult to completely validate the BPBE
model. Nevertheless, qualitative examination of the BPBE
predicted results is possible and reveals some of the trends
predicted by the model.
Fig. 2 shows the simulated time and azimuthally-
averaged volume fraction profiles for different diameter
bubbles in a 44-cm diameter air–water bubble column
operated at 10 cm/s superficial gas velocity. The volume
fraction of small bubbles (d
4 mm) is uniformly dis-
tributed, especially for tiny bubbles (d < 2.5 mm). For
larger bubbles, however, the volume fraction distribution
is parabolic as shown in Fig. 2b. In Fig. 3, the simu-
lated cumulative bubble volume fraction is given for a
16.2-cm diameter air–Therminol-Glass beads slurry bub-
ble column operated at 8, 14 and 20 cm/s superficial gas
velocity. As found in the air–water system (Chen et al.,
2004b), the cumulative volume fraction of smaller bubbles
(up to 4 mm) does not change with increase in superficial
gas velocity (14 and 20 cm/s), for operation deep into the
churn-turbulent flow regime. These findings are in line with
the hypothesis and reported experimental observations of
Krishna and Ellenberger (1996) who asserted that in churn-
turbulent flow the holdup attributed to small bubbles stays
constant. However, single modal bubble size distribution is
found in the simulation as shown in Fig. 4, contrary to the
assertion of Krishna and Ellenberger (1996) who claim a
bimodal size distribution based on the dynamic gas disen-
gagement technique (DGD). The superficial gas velocity,
when the flow is deep into the churn-turbulent regime, has
little effect on volume fraction of small bubbles (1 and
1.6 mm). However, the superficial gas velocity effect be-
comes pronounced for larger bubbles. The bubble size distri-
bution becomes wider and the Sauter mean bubble diameter
becomes larger as the superficial gas velocity increase. This
is in line with observations reported by Fan et al. (1999) in
high pressure small diameter column. Bimodal bubble size
distribution postulated by Krishna and Ellenberger (1996)
is not seen in any of our simulations or in data by Fan et al.
(1999).
The bubble class holdup-based probability function
is illustrated in Fig. 5 for air–water bubble column
(Ug = 10 cm/s) and air–Therminol–Glass beads slurry
 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207 5205

0.05 0.08
class 5
0.07 class 6
0.04 class 7
class 8
0.06
class 1 class 9
class 2 0.05
0.03
Holdup

Holdup
class 3
class 4 0.04
0.02 0.03

0.02
0.01
0.01
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(a) Dimensionless Radius (b) Dimensionless Radius

Fig. 2. Time-averaged bubble class holdup profile (DC = 44 cm, Ug = 10.0 cm/s, Air–Water).

0.3 0.07 Ug = 14 cm/s

Ug = 20 cm/s
0.06
0.25
0.05
Cumulative Holdup

0.2
Ho ldup

0.04
0.15 0.03
0.02
0.1 Ug = 8 cm/s
Ug = 14 cm/s 0.01
0.05 Ug = 20 cm/s
0
1

16

40
6

35

25
8
2.
1.

.0

0
6.
10

0 10 20 30 40
Bubble Diameter, mm
Bubble Diameter, mm
Fig. 4. Bubble classes holdup distribution for Air–Therminol-Glass beads
Fig. 3. Overall bubble classes cumulative holdup (Air–Therminol-Glass slurry bubble column.
beads).

bubble column (Ug = 8 cm). The physical properties of
liquid (slurry) phase for these two columns are listed in
Table 2, and the operating conditions are listed in Table 3.
The major difference in these two systems is the surface
tension of the liquid (slurry). The simulated overall gas
holdup for these two columns are about the same. How-
ever, the bubble class holdup-based probability density
function for these two systems, as expected, is signif-
icantly different because of the difference in the liquid
(slurry) phase surface tension. There are many more smaller
bubbles in the air–Therminol–Glass beads slurry bubble
column than in the air–water bubble column operated at
similar superficial gas velocity, and this is captured by the
simulation.
5. Summary and conclusions
Three-dimensional simulation and incorporation of the
BPBE is necessary to capture the gas holdup radial pro-
file in bubble columns faithfully. For this purposes it seems
sufficient to track 9 bubble classes. The lack of measured
bubble size distribution data in churn-turbulent flow makes
it difficult to validate the BPBE model completely. Nev-
ertheless, model predictions agree qualitatively well with
the experimental observations regarding the effect of super-
ficial gas velocity on small and large bubbles and in ac-
counting for the surface tension. A single modal bubble size
distribution is found in all simulations of churn-turbulent
flow.
5206 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207

0.35 Ug = 10 cm/s, Subscripts

Air-Water
0.3 Ug = 8 cm/s, Air- b Bubble
Therminol g Gas phase
Bubble class p.d.f

0.25 i, j bubble class index

0.2
Superscript
0.15
i Bubble class index
0.1

0.05 Acknowledgements

0 Two of the authors (P. Chen and M.P. Duduković) ac-

knowledge the support of the industrial sponsors of the
16

40
1

4
35

.4
8

Chemical Reaction Engineering Laboratory (CREL).

1.

2.

.0

25
6.
10

Bubble Diamter, mm
References
Fig. 5. Bubble classes holdup-based probability distribution.
Bertola, F., Grundseth, J., Hagesaether, L., Dorao, C., Luo, H., Hjarbo,
K.W., Svendsen, H.F., Vanni, M.B., Jakobsen, H.A., 2003. Numerical
analysis and experimental validation of Bubble size distributions in
two phase bubble column reactors. 3rd European–Japanese Two-Phase
Table 2 Flow Group Meeting, Certosa di Pontignano.
Physical properties Chen, J., Li, F., Degaleesan, S., Gupta, P., Al-Dahhan, M.H., Dudukovic,
M.P., Toseland, B.A., 1999. Fluid dynamic parameters in bubble
Liquid Density Viscosity Surface tension
columns with internals. Chemical Engineering Science 54 (13–14),
(Slurry) phase (kg m−3 ) (cp) (N/m)
2187–2197.
Water 1000 1.0 0.072 Chen, P., 2004. Fluid dynamic modeling of bubble column flows. D.Sc.
Therminol-Glass beads 1015 1.08 0.017 Thesis, Washington University, Saint Louis, Missouri.
(9.1% solids loading) Chen, P., Sanyal, J., Dudukovic, M.P., 2004a. Numerical simulation of
bubble columns flows: effect of different breakup and coalescence
closures. Chemical Engineering Science, accepted.
Chen, P., Sanyal, J., Dudukovic, M.P., 2004b. Three dimensional
simulation of bubble columns flows with bubble coalescence and
Table 3 breakup. A.I.Ch.E Journal, accepted.
Operation condition Degaleesan, S., 1997. Fluid dynamic measurements and modeling of liquid
mixing in bubble columns. D.Sc. Thesis, Washington University, St
DC (cm) Ug (cm/s) System Overall gas holdup (sim.) Louis.
Degaleesan, S., Roy, S., Kumar, S.B., Dudukovic, M.P., 1996. Liquid
16.2 8 Air–Therminol 0.165
mixing based on convection and turbulent dispersion in bubble columns.
44 10 Air–Water 0.162
Chemical Engineering Science 51 (10), 1967–1976.
Devanathan, N., Moslemian, D., Dudukovic, M.P., 1990. Flow mapping
in bubble columns using CARPT. Chemical Engineering Science 45,
2285–2291.
Drew, D.A., 1983. Mathematical modeling of two-phase flow. Annual
Notation Review of Fluid Mechanics 15, 261–291.
Elghobashi, S.E., Abou-Arab, T.W., 1983. A two-equation turbulence
CD Drag coefficient, dimensionless model for two-phase flows. Physics of Fluids 26 (4), 931–938.
Fan, L.-S., Yang, G.Q., Lee, D.J., Tsuchiya, K., Luo, X., 1999.
d, dB Bubble diameter, m Some aspects of high-pressure phenomena of bubbles in liquids and
n, N Number density of bubble class, m−3 liquid–solid suspensions. Chemical Engineering Science 54, 4681–
Re Bubble Reynolds number, dimensionless, Re = 4709.
db |ul − ug |
l /l George, D.L., Shollenberger, K.A., Torczynski, J.R., 2000. Sparger effect
Si Source term, m−3 s−1 on gas volume fraction distributions in vertical bubble-column flows
as measured by gamma-densitometry tomography. ASME 2000 Fluids
u Velocity, m/s Engineering Division Summer Meeting, Boston, MA.
t Time, s Gupta, P., Al-Dahhan, M.H., Dudukovic, M.P., Toseland, B.A., 2001.
Comparison of single- and two-bubble class gas-liquid recirculation
Greek letters models—application to pilot-plant radioactive tracer studies during
methanol synthesis. Chemical Engineering Science 56 (3), 1117–1125.
Krishna, R., Ellenberger, J., 1996. Gas holdup in bubble column reactors
 Dynamic viscosity, kg m−1 s−1 operating in the churn-turbulent flow regime. A.I.Ch.E Journal 42,

Density, kg/m3 2627–2634.
 P. Chen et al. / Chemical Engineering Science 59 (2004) 5201 – 5207
Kumar, S.B., Moslemian, D., Dudukovic, M.P., 1995. A gamma-ray
tomographic scanner for imaging voidage distribution in two-phase
flow systems. Flow Measurement and Instrumentation 6 (1), 61–73.
Kumar, S.B., Moslemian, D., Dudukovic, M.P., 1997. Gas-holdup
measurements in bubble columns using computed tomography.
A.I.Ch.E Journal 43 (6), 1414–1425.
Launder, B.E., Spalding, D.B., 1974. Mathematical Models of Turbulence,
Academic Press, London.
Luo, H., 1993. Coalescence, breakup and liquid circulation in bubble
column reactors. D.Sc. Thesis, Norwegian Institute of Technology.
Luo, H., Svendsen, H.F., 1996. Theoretical model for drop and bubble
breakup in turbulent dispersions. A.I.Ch.E Journal 42, 1225–1233.
Manninen, M., Taivassalo, V., Kallio, S., 1996. On the mixture model
for multiphase flow. Technical Research Center of Finland, VIT
Publications.
Miyahara, T., Matsuba, Y., Takahashi, T., 1983. The size of bubbles
generated from perforated plates. International Chemical Engineering
23 (3), 517.
Ong, B., 2003. Experimental investigation of bubble column
hydrodynamics—Effect of elevated pressure and superficial gas
velocity. D.Sc. Thesis, Washington University, Saint Louis, USA.
Pan, Y., Dudukovic, M.P., Chang, M., 2000. Numerical investigation of
gas-driven flow in 2-D bubble columns. A.I.Ch.E Journal 46 (3),
434–449.
5207
Sanyal, J., Vasques, S., Roy, S., Dudukovic, M.P., 1999. Numerical
simulations of gas–liquid dynamics in cylindrical bubble column
reactors. Chemical Engineering Science 54, 5071–5084.
Schiller, L., Naumann, Z., 1935. Über die grundlegenden Berechungen
bei der Schwerkraftbereitung. Z. Ver. Deutsch. Ing. 77, 318.
Shollenberger, K.A., George, D.L., Torczynski, J.R., 2000. Effect of
Sparger geometry on gas-volume-fraction in bubble-column flows
measured by gamma-densitometry tomography (GDT). Sandia National
Laboratories.
Tchen, C.M., 1947. Mean value and correlation problems connected with
the motion of small particles suspended in a turbulent fluid. D.Sc.
Thesis, TU, Delft, The Netherlands.
van Baten, J.M., Krishna, R., 2001. Eulerian simulations for determination
of the axial dispersion of liquid and gas phases in bubble columns
operating in the churn-turbulent regime. Chemical Engineering Science
56 (2), 503–512.
Wilkinson, P.M., 1991. Physical aspects and scale-up of high pressure
bubble columns. D.Sc. Thesis, University of Groningen, The
Netherlands.
Yang, Y.B., Devanathan, N., Dudukovic, M.P., 1992. Liquid backmixing
in bubble columns. Chemical Engineering Science 47, 2859–2864.