Anda di halaman 1dari 10

Proceedings of the ASME 2016 Pressure Vessels and Piping Conference

July 17-21, 2016, Vancouver, British Columbia, Canada




Yi-Gang Dinga, Xia Lua, Fu-Li Denga

School of Chemical and Engineering and Pharmacy,
Wuhan Institute of Technology,
WuHan City,
HuBei Province,
PR China

ABSTRACT chemical reaction engineering[1]. The most important factor

influencing mixing in such reactors is the local prevailing
A coupled CFD-PBM (population balance mode) model hydrodynamics, and the gas-liquid interfacial area, which
is adopted to investigate complex behavior in a rectangle depends on the bubble size distribution [2-4]. To optimize
bubble column. In this work The Euler–Euler (E–E) model was design or scale up of bubble column processes, it is essential
adopted for the liquid phase and gas phase, while accounting to know the local hydrodynamics and bubble size distribution
for bubble coalescence and breakup a PBM discrete model at different operating conditions [5] .Bubbles in two phase flow
was employed . The total gas holdup for a range of superficial can interact with each other. They can aggregate to produce
gas velocities were studied and compared with the literature larger bubbles and they can also break to generate smaller ones.
and modest agreement was found. The simulation result shows Accompanying with the gas superficial velocity increase, the
that the superficial gas velocity has great effect on bubble size flow pattern change from the homogenous regime to
distribution, and a wider bubble size distribution is found at heterogeneous regime. Therefore it’s important to study the
higher superficial gas velocity. This indicates an increasing of bubble coalescence and breakup in flow pattern and bubble
the superficial gas velocity increases the bubble coalescence size distribution prediction[6]. Considerable experiments have
and break-up rate. been carried out to investigate the bubble size distribution in
different types of laboratory-scale bubble columns. Mena[7]
INTRODUCTION measured the bubble size distribution in a mass transfer system
in a lab-scale 3D column at a low superficial gas velocity of
Gas–liquid bubble column reactors are widely used in 2.8mm/s. Daeseong[8] discussed bubble size distributions in a
phosphorus chemical industry due to its great advantage, e.g. two-dimensional packed bed by using image processing
lager mass transfer area, good material mixing property, fast techniques with a large number of bubble samples,
reaction rate and so on. In practice, the reaction in phosphorus considering the effect of coalescence and breakup.
chemical industry is fast and mixing is very important in In recent years, computational fluid dynamics (CFD)

1 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

has emerged as a powerful tool for bubble column design and centrifugal force, the volume-averaged mass and momentum
scale up. CFD modeling of dispersed gas–liquid two-phase conservation equations are expressed as:
flows has shown remarkable progress over the last decade. ∂
( ρ gα g ) +∇ i( ρ gα g v g ) = 0 (1)
The Euler–Euler (E–E) model was employed to simulation gas-
liquid two phases flows, and the population balance model was ∂
( ρ lα l ) +∇ i( ρlα l vl ) = 0 (2)
incorporated to take into account break-up and coalescence of ∂t

bubbles as well as growth or shrinkage of bubbles as a
( ρ gα g vg ) + ∇ ⋅ (α g ρ g vg vg ) = −α g ∇p
consequence of mass transfer[9]. Many attempts have been
made by coupling computational fluid dynamics with population +α g ρ g g + ∇ ⋅τ g + K gl (vl − v g )
balance model (CFD-PBM) to simulate the gas-liquid flows[10, +α g ρ g ( Flift , g + Fwl , g + FVm , g + Ftd , g )

. Montante et al[12] investigated gas-liquid flow and bubble
11] ( ρlαl vl ) + ∇ ⋅ (αl ρl vl vl ) = −αl ∇p
size distribute on a stirred tank, presenting a good prediction of +α l ρl g + ∇ ⋅τ l + K gl (v g − vl )

the number mean bubble size but a significant under-prediction +α l ρl ( Flift ,l + Fwl ,l + FVm ,l + Ftd ,l )

of the Sauter mean diameter. Miriam Petitti[13] et al discussed Where v represents the phase velocity, α denotes the
local gas hold-up and bubbles size distributions using an
volume fraction of each phase, Flift ,l , Flift , g is lift force ,
Eulerian multiplied approach coupled with a PBM model. The
PBM has been solved by resorting to the quadrature method of
Fwl ,l , Fwl , g is wall lubrication force, FVm,l , FVm, g is virtual
moments (QMOM), implemented through user defined
functions. A detailed modeling of two-phase flows via a four-
mass force, and Ftd ,l , Ftd , g is turbulent dispersion
way coupling approach, combining the standard CFD analysis
with a Population Balance Modeling (PBM), for taking into force, p denotes the pressure shared by all phases, τ is the
account the effect of bubble breakage and coalescence on the stress-strain tensor, being defined as
bubbles size, has been performed, to estimate the local bubble 2
τ g = α g µ g (∇vg + ∇v gT ) + α g (λg − µ g )∇ivg I (5)
size distribution, key information for the correct evaluation of 3
the specific surface area and the mass transfer rate. However, 2 (6)
τ l = α l µl (∇vl + ∇v lT) + α l (λl − µ l )∇ ivl I
there is no universal model for different situation due to 3
complexity of the problem, the effect of different interface
µq λq
interaction closure models and bubble coalescence and breakup Here and are the shear and bulk viscosity of phase,
models should be evaluated. The objective of this paper is to
predicting the hydrodynamics and bubble size distribution using I is a unit tensor.
a 2d model, and a full population balance model with detailed In this study the drag force, lift force, turbulent dispersion
bubble breakup and coalescence models is coupled in the CFD force and interaction force are considered, and the wall
Framework. lubrication force and turbulent dispersion force are neglected.
Water is modeled as the primary fluid, since gas appears in the
Inter-phase momentum closure relations
Governing equations of gas-liquid flow The closure terms such as drag force, lift force, virtual
Numerical simulations are performed with the Euler– mass force, turbulent dispersion force and interaction force has
Euler two fluid model. In the Euler-Euler two fluid model, the very strong influence on the simulation results. This section
different phases are treated mathematically as interpenetrating discusses the closure terms be used by the study.
continua. For unsteady-state incompressible flow in the Modeling drag force
absence of mass transfer, external body forces such as the The drag force resists the bubble motion in the surrounding

2 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

liquid. In the framework of two-fluid model, the drag force is min 0.228 tanh ( 0.121Re ) , f ( Eod )  ( Eod < 4) 
 
characterized as the mean momentum transfer between the phases. CL =  f ( Eod ) (4 ≤ Eod ≤ 10) 
The momentum transfer between the phases is considered as the −0.27( Eo > 10) 
 d
product of inter-phase exchange coefficient K gl and the relative f ( Eo ) = 0.00105Eo − 0.0159 Eo − 0.0204 Eo + 0.474
′ ′3
′ 2


velocity between the phases .The exchange coefficient for these d h = d b (1 + 0.163Eo′0.757 ) 3

types of bubbly is defined as: Modeling virtual mass force

3 C
K gl = ρlα lα g D vg − vl (7) The virtual mass force, which is particularly important
4 db in gas-liquid flows, accounts the force necessary to accelerate
the fluid around a bubble. The virtual mass force is given as,
Where CD represents a drag coefficient, differing among
d v d v  (13)
Fvm, g = − Fvm ,l = Cvmα p ρ q  q q − p p 
the exchange-coefficient models. In this work, Tomiyama  dt dt 
model is adopted, because the model is well suited to gas-liquid
Cvm is the virtual mass coefficient which
flows in which the bubbles can have a range of shapes. The value
is given by
  24 72  8 Eo  (8)
Cd = Max  min 
 Re
(1 + 0.15 Re0.687 ) , Re , 
 3 Eo + 4 
typically has a value of 0.5.The term
denotes the phase

Where Eötvös number (Eo) are defined as material time derivative of the form
g ( ρl − ρ g ) d b2 d q (φ ) ∂ (φ )
Eo = (9) = + ( vq i∇ ) φ (14)
σ dt ∂t
Re denotes the relative Reynolds number and can be written Turbulence Model
as: In order to solve the Reynolds stress in the turbulent
ρ l d b u g − ul
momentum equations, the k -ε mixture turbulence model is
Re = (10)
chosen for simulation, and expressed as follows:
Modeling lift force ∂ρ m k  µ  
+ ∇ρ m vm k = ∇   µm + t ,m  ∇k 
In addition to the drag force, the bubble experiences a ∂t  σ k  
lift force perpendicular to its relative motion. Lift forces act +Gk , m − ρ mε (15)
on a particle mainly due to velocity gradients in the primary- ∂ρ mε  µ  
+ ∇ρ m vmε = ∇   µm + t ,m  ∇ε 
phase flow field, as well as the shear resulting from the slip ∂t  σε  
between the phases. The wall peak distribution of radial gas ε
holdup profiles usually observed in upward bubbly flows, a
(C ε G
1 k ,m − C2 ε ρ m ε )

phenomenon which is predominant at small bubble ranging Where,

from0.4 to5mm due to action of the lift forces[15-17]. As mean
ρ m =α g ρ g + α l ρ l (16)
bubble diameter is increased, the peak shifts towards the
center of the pipe eventually resulting in core peaking[18]. In
µ m =α g µ g + α l µl (17)
this work the lift force is calculated as

Flift , g = − Flift ,l = −Cl ρlα g (vl − vg ) × ( ∇× vl ) (11) α g ρ g ug + α l ρl ul

vm = (18)
Cl , the lift coefficient, is given by the Tomiyama
In bubble columns, both experimental and numerical
model. studies reported the turbulence characteristics of the
continuous phase as being affected by the presence of

3 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

dispersed entities[19]. In this work, viscosity of the slurry phase of volume V. ν is the number of child particles produced per
is modeled using the Sato enhanced turbulence model[20], as
parent. a (V − V ′, V ′) defines the rate coefficient of
the following equation:
aggregation of two particles with volumes V − V ′ and V ′ .
µ t , m = µ t , s + µ t ,b (19)
Breakage kernel
µt , s The breakage kernel[21], is expressed as
represents the conventional shear-induced turbulent
g (V ′ ) β (V V ′ ) (23)
viscosity and is obtained by the standard k -ε model as:
Where g (V ′ ) is the rate coefficient for the breakage of
k2 (20)
µ t , s = ρ m Cµ
a particle size V ′ , and β (V V ′ ) defines the fragment

is a constant, which is 0.09 . distribution function.
Luo and Lehr[21, 22] developed a theoretical model for the
µt ,b
is a bubble-induced component of turbulent breakup of drops and bubbles in turbulent suspensions. It is
viscosity given by: based on the concept that the bubble breaks up when it collides
with the turbulent eddy with sufficient energy. One advantage
µt ,b = Cµ ,b ρlα g db ug − ul (21)
is that this model does not include empirical parameters.
Another advantage is that the daughter size distribution is
C1ε , C2ε , σ k and σ ε are parameters in the standard
derived directly. This model has been widely used in the
previous work. The breakup rate coefficient is modeled as:
k -ε model and the following values are selected: C1ε = 1.44 ,
(1 + ξ )
1 2

g (V ′ ) = K ∫ exp ( −bξ m )d ξ
C2ε = 1.92 , σ k = 1.0 and σ ε = 1.3 . ξmin
(1 + ξ )
ε  1
Population balance model =0.9238 FB (1 − rd )  c2 
 di 
∫ξ min ξ 11/3
The population balance model was used to obtain a
 12  f BV
+ (1 − f BV ) 2/3 − 1 σ 
× exp  −  
reasonable result about the local void fraction. Let n(V , t )  βρcε c2/3 di5/3ξ 11/3 
 
represent the number density of particles of size V at time t.
Where f BV represents the volume fraction of the parent
The population balance equation can be written as
∂ bubble that constitutes the volume of one daughter bubble per
[ n(V , t )] + ∇ ⋅ [un(V , t )] =
∂t unit volume of the continuous phase. FB is a calibration
1 V
a (V − V ′, V ′)n (V − V ′, t )n (V ′, t ) dV ′
2 ∫0
coefficient, β is equalto2, ε c is the continuous phase eddy
birth − due − to − aggregation
− ∫ a (V , V ′)n (V , t )n(V ′, t )dV ′
0 dissipation rate, σ is the interfacial tension, and ξ is the
de ath − due − to − aggregation

+ ∫ ν g (V ′ ) β (V V ′ ) n (V ′, t ) dV ′ − g (V )n (V , t ) dimensionless size of eddies in the inertial subrange of

death − due − to − breakage
birth − due − to − breakage
isotropic turbulence.

Where g (V ′ ) is the breakage frequency of particle with

The breakage PDF function contains information on the
probability of fragments formed by a breakage event. It

( )
volume V ′ , and β V V ′ represents probability density
provides the number of particles and the possible size
distribution from the breakage, and the equation is given as,
function of particles breaking from volume V ′ to a particle

4 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

C 1 − C 2   V 
 V    (25) modeled as
β (V / V ′ ) = 0.5  + 24   − 24   + 6  
′ ′  ′   V ′   
 V V  V
θijT = FCT Sij ( u 2ti + u 2tj ) (d + d )
1/2 2
= FCT i j
4 (29)
Where V and V ′ are the daughter and parent particle
( ) +( )
2 1/ 2
× 
2ε d
1/3 1/3
2ε d 1/3 1/3

 
c i c j
volumes, respectively. C represents the shape factor of the
particle breakage distribution function. The buoyancy contribution to collision frequency is
Aggregation kernel modeled as:
θ ijB = FCB Sij u rj − u ri = FCB ( di + d j )
The aggregation kernel is expressed as 2

a(V ,V ′) 4 (30)
The model of Prince and Blanch[23] assumes that the 2.14σ 2.14σ
× + 0.505gd j − + 0.505gd i
coalescence of two bubbles occurs in three steps. First, the ρc d j ρc d i
bubbles collide trapping a small amount of liquid between
Where FCT and FCB are calibration factors.
them. This liquid film then drains until the liquid film
separating the bubbles reaches a critical thickness. The film NUMERICAL STRATEGY AND SCHEME
then ruptures and the bubbles join together. The coalescence
kernel is therefore modeled by a collision rate of two bubbles A schematic representation of the column is illustrated in
and a collision efficiency relating to the time required for Fig. 1(a).just as show in fig.1 (a), the whole column
coalescence, it can be calculated as follows, dimensions is 0.2 × 0.03 × 0.7 (width, depth and height). In
the rectangle column, two phases come in contact: a gas phase
a(V , V ′) = Q ( mi ; m j ) = (θijT + θijB + θijS )ηij (26)
(air) and water as a continuous phase.

Where ηij is the collision efficiency. θijT , θijB and

θijS are the collisions contributions of turbulence, buoyancy,

and shear respectively. In this study, the θ ij is neglected.


The collision efficiency is modeled by comparing the

time required for coalescence with the actual contact time
during the collision,
− tij τ ij
ηij = e
 ρc rij3  h
 Fig.1 Sketch of rectangle bubble column
tij =  ln  0 
 16σ  h Two dimensional (2D) simulation is adopted to reduce
   f 
rij2/3 calculation cost. Fig.1 (b) shows the mesh topology which is
τ ij = 1/3
εc constructed in ICEM CFD16.0. The CFD-PBM coupled
model is performed in Fluent 16.0(Ansys Inc., USA), the
Where h0 is the initial film thickness, h f is the
aggregation kernel is compiled through user defined functions
(UDFs). The Euler–Euler two fluid model is adopted to
critical film thickness when rupture occurs, and rij is the
simulate the flow and a PBM discrete model is employed to
equivalent radius: account for bubble coalescence and break-up. The coupling
−1 scheme of CFD model and PBM is discussed as following.
 1  1 1  (28)
rij =   +   The bubble volume fraction, particle velocity are obtained by
 2  ri rj  
  
the Eulerian–Eulerian two-fluid model, which are applied to
The turbulent contributions to collision frequency are solve the moment transport equations in PBM considering

5 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

break up and aggregation terms. Once the population balance obtain more reasonable and accurate results.
equations are solved by discrete method, the PSD are obtained
and sauter mean diameter is updated for further calculation of RESULTS
the drag force and then the information of flow hydrodynamics
by CFD model is updated. Therefore, the coupling of Validation of results
Eulerian–Eulerian two-fluid model with PBM is achieved. In this work, CFD-PBM model was evaluated, comparing
By means of boundary and gradient adaption, Mesh the simulated data with the work of Y.M.Lau[24]. From Fig.3,
independence tests are carefully conducted to ensure that the it can be seen that similar trend is observed for results of
solutions do not obviously change with the increase number of experiment and simulation, and the comparison shows a
computational cells. The result showed that increasing the cell modest agreement within a deviation of13%. The gas holdup
numbers from 278901 to 326000 resulted in less obvious increases with the increase of superficial gas velocity
variation of gas hold up and axial velocity. Herein, the mesh accordingly. The simulation results are slightly larger than the
number of 278901 is adopted to perform the simulation. With experimental results, and the difference increases with the
the mesh refining technique, most additional mesh points are increase of superficial gas velocity. This is because the model
put in regions of high gradients around the gas inlet. uses a single bubble drag force model, which is larger than
According to the operating conditions, boundary bubble group drag force in fact. In order to get more accurate
conditions and initial conditions are set for the numerical results, a more accurate model of bubble group drag force
simulation. At the gas inlets, gas velocities are set ranging based on experiment is needed. In general, the model is
from 0.005m/s to 0.03cm/s. At the gas outlet, a degassing capable to predict the gas holdup distribution of the rectangle
boundary condition is employed at the free surface, from bubble column, and the calculation accuracy of the couple
where dispersed bubbles are permitted to escape but the liquid CFD-PBM model is acceptable in this work.
phase is not allowed. Along the wall, gas and liquid are treated
as no-slip. Initial bubble size distribution in the CFD-PBM
model is obtained from the experimental measurements by
Y.M.Lau[24]. In this work, fifteen bubble classes are employed,
and the specific PSDs is shown in Fig. 2.

Fig.3 Gas holdup versus superficial gas velocity

Bubble and liquid velocities
Fig.4-5 shows the snapshots of the instantaneous air and
liquid velocity. By analysis of the time series, the following is
observed. At low superficial gas velocity of 0.01m/s, the
Fig.2 The particle size applied for simulation bubbles rise upward in a homogeneous fashion at first. With
The calculation are all implemented with ANSYS 16.0 the increase of time, there are some regions where the bubbles
with a double precision mode. The phase-coupled SIMPLE rise faster near the wall and small vortical structures are
algorithm is used for the pressure-velocity coupling and formed. It is because the initial bubble size is approximate
correction. Regarding better convergence and less numerical 4mm, and the flow appears wall peak at first. With time
effort, the first-order upwind discretization is adopted to increasing, coalescence and break-up of bubbles start to play
discretize the equations. Temporal discretization scheme is set a role and mean bubble size increases accordingly. Hence, the
as first-order implicit. The convergence criterion is set to 1e- flow pattern transforms into core peak.
3, and time step is set as 1e-3 to ensure the stability and to

6 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

(a) (a)

(b) (b)
Fig.4 Snapshots of the instantaneous air velocity fields for Fig.5 Snapshots of the instantaneous liquid velocity fields for
different superficial gas velocities. (a)ug=0.01m/s different superficial gas velocities.(a)ug=0.01m/s
(b)ug=0.02m/s (b)ug=0.02m/s
In fig.5 (a), it can be seen that the liquid (water) rises Fig. 6 shows time-averaged profiles of the mean axial air
upward near the wall and downward in the central region due and liquid velocity at 0.30 m column heights. For a low
to wall peak at first, and the liquid flow patter become chaotic superficial gas velocity of 0.005m/s, a uniform up-flow of the
near the wall due to bubble coalescence and break-up after a rising bubbles. By increasing the superficial gas velocity,
few second of flowtimes. At a superficial gas velocity of bubble down-flow regions near the side walls and a bubble up-
0.02m/s, the trend is similar to that at 0.01m/s, but the flow flow region in the centre across the width of the bubble column
pattern become more chaotic. Small vortices are forming in are developed. A parabolic velocity profile character was
the whole area due to intense bubble coalescence and break- obtained and at high gas superficial velocities a radial
up. The results are consistent with the literature report[24]. symmetry is observed in the fully developed region. Same
trends are observed in the liquid phase.

7 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:




Fig. 6 Time averaged plot of the mean axial velocity
component of the gas phase (a) and liquid phase (b) at 0.30 m
column heights for different superficial gas velocities.
Bubble size distribution
The length number density of axial bubble size (c)
distribution for different superficial gas velocities is
illustrated in Fig.7. It can be seen that the change of length
number density become more intense with the increase of
superficial gas velocity. For a low superficial gas velocity of
0.005m/s, the length number density of bubble drops a little
along the axial direction due to small amount of bubble
coalescence and break-up. For a superficial gas velocity of
0.01m/s, the length number density drops quickly below
0.3m column height, and little change is observed above (d)
0.3m column height. At the same time a wide bubble size Fig.7 The axial bubble size distribution for different
distribution can be seen. This indicates that bubble superficial gas velocities (a)ug=0.005m/s (b)ug=0.01m/s
coalescence and break up almost take place almost below (c)ug=0.02 m/s (d)ug=0.03 m/s
0.3m column height, and an equilibrium between coalescence At a superficial gas velocity of 0.02m/s(see Fig.7(c) a
and break up is almost reached in the bottom section. wider bubble size distribution is observed and the fraction of
small and large bubbles increase. This indicate the flow
pattern begins to change from homogeneous regime to

8 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

heterogeneous regime. The length number density declines
along the axial direction of the column. It seems that the
equilibrium between coalescence and break-up is not This work is financially supported by the Natural
obtained in the bottom section, but still develops throughout Sciences Foundation of China (No. ZRY2014000037).
the column. Increasing the superficial gas velocity to 0.03m/s
leads to similar effects, but also to more coalescence and
break-up. Compare to literature[24], large bubble fraction
seems higher. This is probably due to over estimation of [1] Lifeng, H., L. Dadong, T. Xiaojin, et al., Study on the
coalescence frequency. hydrodynamics characteristics of external-loop airlift
slurry reactors. Petroleum Processing and
Petrochemicals, 2012. 43(6): p. 1-5.
CONCLUSIONS [2] Yang, L., J. Cheng, P. Fan, et al., Micromixing of
Solid-liquid Systems in a Stirred Tank with Double
The gas–liquid flow in a square cross sectioned bubble Impellers. Chemical Engineering & Technology,
column has been successfully simulated using a CFD-PBM 2013. 36(3): p. 443-449.
model. The flow was described with the Euler–Euler approach, [3] Weipeng, Z., Y. Yumei, Z. Guangji, et al., Mixing
where the drag, lift, virtual mass forces, bubble coalescence haracteristics and Bubble Behavior in an Airlift
and break-up were taken into account. The turbulence Internal Loop Reactor with Low Aspect Ratio.
viscosity in the liquid phase was modeled with k − ε model. Chinese Journal of Chemical Engineering, 2014(06):
An extra contribution in the effective viscosity for the p. 611-621.
turbulence induced by bubbles was taken into account. A [4] Cheng, D., X. Feng, J.C. Cheng, et al., Numerical
discrete model was adopted to model bubble coalescence and simulation of macro-mixing in liquid-liquid stirred
break-up. It can be seen that the comparison of gas holdup tanks. Chemical Engineering Science, 2013. 101: p.
shows a modest agreement with the experimental data. The 272-282.
gas holdup increases with the increase of superficial gas [5] Sujatha, K.T., B.G.J. Meeusen, J.A.M. Kuipers, et al.,
velocity accordingly. At a low superficial gas velocity of Experimental studies of bubbly flow in a pseudo-2D
0.005m/s, a uniform up-flow of the rising bubbles. For high micro-structured bubble column reactor using digital
gas superficial velocities, a parabolic velocity profile character image analysis. Chemical Engineering Science, 2015.
was obtained and a radial symmetry is observed in the fully 130: p. 18-30.
developed region. The coalescence and break up become more [6] Ghasemi;, H. and S.H. Hosseini, Investigation of
intense with the increase of superficial gas velocity. An hydrodynamics and transition regime in an internal
equilibrium between coalescence and break up is almost loop airlife reactor using CFD. Brazilian Journal of
reached in the bottom section at a superficial gas velocity of Chemical Engineering, 2012. 29(4): p. 821-833.
0.01m/s. a wider bubble size distribution is observed and the [7] Mena, P.C., Pons,M.N.,Teixeira,J.A.,Rocha,F.A.,
fraction of small and large bubble increase at a superficial gas Using image analysis in the study of multiphase gas
velocity of 0.02m/s. the equilibrium between coalescence and absorption. Chem.Eng.Sci, 2005. 60: p. 5144-5150.
break-up is not obtained in the bottom section, but still [8] Jo, D., Revankar, and S. T, Effect of coalescence and
develops throughout the column. The similar trend is observed breakup on bubble size distributions in a two-
for a superficial gas velocity of 0.03m/s, but to more dimensional packed bed. Chemical Engineering
coalescence and break seems the coalescence frequency Science, 2010. 65(14): p. 4231-4238.
is over estimated, and the size distribution is wider than the [9] Darmana, D., Deen, N.G., Kuipers, J.A.M., Detailed
literature. In order to obtain a more accurate bubble size modelling of hydrodynamics, mass transfer and
distribution, the modification of coalescence model is needed. chemical reactions in a bubble column using a

9 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use:

discrete bubble model. Chemical Engineering Science Research, 2005. 44(14): p. 5107-5151.
2005. 60(12): p. 3383-3404. [20] Sato, Y., M. Sadatomi, and K. Sekoguchi, Momentum
[10] Bannari, R., F. Kerdouss, B. Selma, et al., Three- and heat transfer in two-phase bubble flow—II. A
dimensional mathematical modeling of dispersed two- comparison between experimental data and theoretical
phase flow using class method of population balance calculations. International Journal of Multiphase
in bubble columns. Computers & Chemical Flow, 1981. 7(2): p. 179-190.
Engineering, 2008. 32(12): p. 3224-3237. [21] Luo, H., Svendsen, H.F., Theoretical model for drop
[11] Chen, P., J. Sanyal, and M.P. Dudukovic, CFD and bubble breakup in turbulent dispersions. A.I.Ch.E.
modeling of bubble columns flows: implementation of Journal, 1996. 42: p. 1225-1233.
population balance. Chemical Engineering Science, [22] Lehr, F. and D. Mewes, A transport equation for the
2004. 59(22-23): p. 5201-5207. interfacial area density applied to bubble columns.
[12] Montante, G., Horn,D.,Paglianti,A., Gas-liquid flow Chemical Engineering Science, 2001. 56(3): p. 1159-
and bubblesize distribution instirred tanks. 1166.
Chem.Eng.Sci, 2008. 63(April(8)): p. 2107-2118. [23] Prince, M.J., Blanch, H.W., Bubble coalescence and
[13] Petitti, M., A. Nasuti, D.L. Marchisio, et al., Bubble break-up inair-sparged bubble columns. A.I.Ch.E.
size distribution modeling in stirred gas–liquid Journal 1990. 36: p. 1485.
reactors with QMOM augmented by a new correction [24] Lau, Y.M., K.T. Sujatha, M. Gaeini, et al.,
algorithm. AIChE Journal, 2010. 56(1): p. 36-53. Experimental study of the bubble size distribution in a
[14] Sakaguchi, T., A. Tomiyama, H. Minagawa, et al., pseudo-2D bubble column. Chemical Engineering
Three-dimensional distributions of particles and Science, 2013. 98: p. 203-211.
bubbles in gas-liquid-solid three-phase bubbly flow in
a vertical pipe. Experimental Thermal and Fluid
Science, 1993. 7(2): p. 165.
[15] Lucas, D. and A. Tomiyama, On the role of the lateral
lift force in poly-dispersed bubbly flows. International
Journal of Multiphase Flow, 2011. 37(9): p. 1178-
[16] Tomiyama, A., Zun, I., Higaki, H., Makino, Y.,
Sakaguchi, T., A three-dimensional particle tracking
method for bubble flow simulation. Nuclear
Engineering andDesign 1997. 175: p. 77-86.
[17] Wang, T., J. Wang, and Y. Jin, Population Balance
Model for Gas−Liquid Flows:  Influence of Bubble
Coalescence and Breakup Models. Industrial &
Engineering Chemistry Research, 2005. 44(19): p.
[18] Shen, X. and T. Hibiki, Interfacial area concentration
in gas–liquid bubbly to churn flow regimes in large
diameter pipes. International Journal of Heat and
Fluid Flow, 2015. 54: p. 107-118.
[19] Jakobsen, H., H. Lindborg, and C. Dorao, Modeling
of Bubble Column Reactors: Progress and
Limitations. Industrial & Engineering Chemistry

10 Copyright © 2016 by ASME

Downloaded From: on 03/16/2017 Terms of Use: