Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Hydrodynamic modeling of ionic liquids and

conventional amine solvents in bubble column

Muhammad Furqan Ali a, Jieqing Gan b, Xiaochun Chen a,∗, Guangren Yu a,∗,
Yuan Zhang a , Mujtaba Ellahi c , Ahmed A. Abdeltawab d
a Beijing Key Laboratory of Membrane Science and Technology, College of Chemical Engineering, Beijing University of
Chemical Technology, Beijing 100029, PR China
b Laboratory of Simulation and Modeling of Particulate System, Department of Chemical Engineering, Monash

University, Clayton, VIC 3800, Australia

c Department of Chemistry, Faculty of Arts & Basic Science, Balochistan University of Information Technology

Engineering & Management Sciences, Quetta 87100, Pakistan

d Department of Chemistry, College of Science, King Saud University, Riyadh 11451, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: The green solvent ionic liquids (ILs) are getting attention in variety of industries, because of
Received 29 May 2017 their unique properties. Few studies explain the hydrodynamics of ILs, but the comparison
Received in revised form 27 October with conventional amine solvents in CO2 capture process has not so far been reported. The
2017 primary target of this research is to develop a state of the art computational fluid dynamics
Accepted 20 November 2017 (CFD) model that would be useful to understate hydrodynamic parameters of both ILs and
Available online 28 November 2017 conventional amine solvents in a glance. In present study, three ILs are investigated inside
a 3D flat bubble column and their higher viscosities are examined vigilantly by compar-
Keywords: ison with three amine solvents. The flow pattern, liquid velocity magnitude, CO2 holdup
Bubble column and bubble size distribution are explained. It is reported that current CFD model may be
Conventional amine solvents used for higher viscous ILs and amine solvents. Compared with amine solvents, CO2 plume
Hydrodynamics meandering behavior was not observed and very low gas holdup was obtain in ILs. The
Ionic liquids velocity magnitude of ILs increased with the height in bubble column, but a decreasing
Numerical modeling trend with column height was followed in the case of amine solvents. CO2 exhibited non-
coalescence behavior in ILs, but coalescence and breakup phenomena were seen in amine
solvents. Finally, baffled type industrial tray column is also investigated and some useful
hydrodynamic aspects in the comparison of pure ionic liquid, 90%-solution of ionic liquid
and 30%-solution of MEA are drawn. The experimentally validated CFD model would be
useful for switching to ILs from corrosive amine solvents and further investigation of new
economical ILs.
© 2017 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction class of ILs treated as green solvent, is used to purify acid gases to
reduce global warming from different power plants and oil refineries
Ionic liquids (ILs) are basically organic salts composed of organic (Keskin et al., 2007). These ILs have unique properties with the com-
cations and inorganic anions. At ambient temperature ILs are molten parisons of conventional amine solvents, such as low vapor pressure,
below decomposition point. There are more than million types of ILs less-corrosive, thermal stability, non-inflammable, easy regeneration,
presenting with their uncountable dissolving properties. Recently, a less contamination as biological aspects. Because of these properties,
ILs provide a wide range of applications and become an attractive agent
in variety of chemical industries (Plechkova and Seddon, 2008; Prado
and Weber, 2016). In a glance, ILs have the ability to become a designer’s
∗
Corresponding authors. solvent and replace conventional amine solvents efficiently.
E-mail addresses: chenxc@mail.buct.edu.cn (X. Chen), gryu@mail.buct.edu.cn (G. Yu).
https://doi.org/10.1016/j.cherd.2017.11.034
0263-8762/© 2017 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 357

capture process in CO2 /N2 -[bmim]BF4 system and predict the Sauter
mean diameter in gas-ILs system was proposed by an experimental
Nomenclature
correlation. Recently, Götz et al. (2016) measured the gas holdup in ILs
CD Drag coefficient experimentally and compared it with other organic liquids in a slurry
CL Lift coefficient bubble reactor. Although, reliable experiments conducted to explain
the variety of hydrodynamic parameters in these previous studies but
CWL Wall lubrication coefficient
the reactor size was small, and the dynamic parameters to differen-
CVM Virtual mass coefficient
tiate the hydrodynamics between ILs and conventionally used amine
d Bubble diameter (cm)
solvents in acid gas removal process were not mentioned. Industries
d32 Sauter mean diameter (cm)
are looking forward to overcome the bottleneck of conventional amine
F Interphase interaction force (N) solvents. A hydrodynamic comparison study between ILs and conven-
f Volume fraction of bubbles tional amine solvents may give rise to a significant replacement of
g Gravity force (m/s2 ) conventional corrosive amine solvents in CO2 capture system.
P Pressure (Pa) Besides experimental techniques, CFD gets more attention which
Si Source term (1/m3 s) requires simple calculation and provides broad spectrum view of veloc-
u Velocity,(m/s) ity field rather than particle image velocimetry (PIV) experiments for
Ug Superficial gas velocity (m/s) ionic liquids. A summarization of some CFD simulation studies of ILs in
Table 1 demonstrates that most researchers focused on single bubble
|Vg | Total volume of gas (m3 )
rising in ILs, and few revealed the hydrodynamics of ILs (mainly consist
M0 Morton number +
of [bmim] cation and BF4 − , PF6 − anion).
Re Reynolds number
The present study aims to provide a CFD model to understand the
Eo Eotvos Number hydrodynamics of new ILs with conventional amine solvents in a large
size laboratory scale bubble column. This objective is obtained by inves-
IUPAC name tigating three ionic liquids, involved with two different cation [bmim]
+

[bmim]SCN 1-Butyl-3-methylimidazolium thiocyanate + −

& [hmim] and three anion BF4 , PF6 − −
and SCN and three industrial
[hmim]BF4 1-Hexyl-3-methylimidazolium tetrafluorob- amine solvents AMP, MDEA and MEA. The hydrodynamic behavior of
orate ILs-CO2 and amine solvent-CO2 are compared in terms of flow pat-
[hmim]PF6 1-Hexyl-3-methylimidazolium hexafluo- tern, liquid flow field, mean gas holdup and bubble size distribution
rophosphate (BSD), which may provide useful guidelines for scaling up prototype to
AMP 2-Amino-2-methyl-1-propanol commercial flue gases absorption column.

MDEA Methyl-di-ethanolamine
MEA Mono-ethanolamine 2. Simulation methodology

Greek letters
A flat rectangular bubble column has quiescent liquid height
˛ Void fraction [–]
of 45 cm, width of 20 cm and depth of 5 cm as shown in Fig. 1(a).
! Surface tension (N/m)
To cover the whole column analysis three portions are virtu-
" Collision rate (m3 /s)
ally divided at the height of 13, 25 and 37 cm. CO2 gas jetted
# Viscosity (Pa s)
with low superficial gas velocity of 0.0014 m/s is introduced
$ Density (Kg/m3 )
for partial aeration in the bubble column to maintain homo-
% Volume of bubble
geneous steady state flow. The superficial gas velocity is kept
constant at 0.0014 m/s for all simulation cases. The dimension
Subscripts
of rectangular bubble column with superficial gas velocities
eff Effective
from 0.0014 m/s to 0.0083 m/s is used previously by Buwa et al.
i, j Bubble classes
(2006), Buwa and Ranade (2002, 2003), Díaz et al. (2008), Díaz
l,g Liquid, gas phases
et al. (2006), Ma et al. (2016), Pfleger et al. (1999). CO2 gas
lam Laminar flow
acted as a disperse phase and six different types of continuous
t turbulent flow
phases are used in rectangular bubble column as described
Abbreviations in Table 2. Initial bubble diameter (0.83 cm) of CO2 gas at the
CFD Computational fluid dynamics sparger is calculated here by the proposed correlation of Zhang
ILs Ionic liquids et al. (2014), which is more specific for ILs (relative experimen-
LDA Laser doppler anemometry tal deviation is ±7%).
PIV Particle image velocimetry The bubble column is systematically transformed into
PBM Population balance model modeling setup. In the preliminary step, grid generation plays
a vital role in computational modeling. To describe more
real phenomena, a 3D model is generated by GAMBIT 2.4 as
shown in Fig. 1(a). Grids have non-uniform structure cells
In spite of the wide use of ILs in commercial plant, few queries (more refine near inlet and center line of the column) con-
still need more research. For example, are they really different from sist of hexahedral cells and quality of mesh is fairly under
conventional amine solvents in the absorbing column? In which man-
control (skewness is almost zero and orthogonal quality is
ner the hydrodynamics of ILs behave differently from conventional
unity). In course grid, the total number of cells is 2975 with
amine solvents? In past few years, Kaji et al. (2009) investigated the
&x × &y × &z = 17 × 25 × 7. The medium grid has 16,544 cells
hydrodynamics of ILs experimentally by using conductivity method to
calculate the gas holdup. One ionic liquid [emim]EtSO4 was compared
with &x × &y × &z = 32 × 47 × 11 and the fine grid has 106628
with 85% of glycerol mixture. After that, Dong et al. (2010) conducted cells with &x × &y × &z = 61 × 92 × 19. Time step size is esti-
a single N2 bubble in ILs. An experimental drag coefficient was pro- mated through CFL condition (&t ≤ &y/|'|), where &t is the
posed for ILs, presenting good agreement with experimental data. time step size, &y is the minimum characteristic size of grid
Zhang et al. (2012), adopted cylindrical bubble column to set a CO2 cell and |'| is the magnitude of bubble rising velocity. The abso-
 358
Table 1 – Contribution of CFD modeling for the hydrodynamics of ILs.
Reactor ILs used CFD model Remarks

Hollow fiber [N1111 ][Gly] 2D finite element method CO2 removal on membrane base is studied and IL Razavi et al. (2016)
contactors acted as absorbent
Gas–liquids flow rates affected on mass transfer
Bubble column [bmim]BF4 , [bmim]PF6, 2D level set method Hydrodynamic behavior of single bubble formation, Carvajal et al. (2015)
glycerol–water mixture detachment and rising is studied
Different drag coefficient is compared
Stirred tank [bmim]BF4 3D Eulerian–Eulerian Hydrodynamics study of CO2 dispersion in IL is Ouyang et al. (2015)
transient model investigated

Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

CFD-PBM coupling showed good agreement with
experimental measurements
Cylindrical bubble [bmim]BF4 , [omim]BF4 , 2D VOF transient model New correlation of liquid side mass transfer is Bao et al. (2015)
column [bmim]NO3 , and two applied by CFD model
aqueous ILs & water CO2 bubble coalescence phenomena between two
mixture bubbles is deeply investigated
Mass flux is affected by water content
Electrostatic spray [bmim]MeSO4 with ethanol 3D electrostatic method IL used as an extractor in a ternary azeotropic Krawczyk et al. (2012)
column & n-heptane mixture of ethanol and n-heptane
Electrostatic field generate to enhance the mass
transfer
Rotating disc [4-mebupy]BF4 with 2D Eulerian–Eulerian Extraction of aromatic hydrocarbons and IL acted as Onink et al. (2010)
contractor toluene & n-heptane transient Model a solvent
CFD model explained the unexpected behavior of
gas hold up at low flux
CFD–PBM coupling is recommended for the
improvement of hydrodynamic behavior
Cylindrical bubble column [bmim]BF4 , [bmim]PF6 , [omim]BF4 2D VOF transient model Single N2 -bubble deformation, bubble aspect ratio, Wang et al. (2010b)
velocity & pressure field is investigated at different
temperatures
Cylindrical bubble [bmim]PF6 with mixture of Eulerian–Eulerian transient Dong’s drag correlation (Dong et al., 2010) is best Wang et al. (2010a)
column CO2 & N2 . Model fitted with experimental for ILs and Schiller
Naumann drag coefficient (Schiller and Nauman,
1935) over predicted the gas holdup
CFD-PBM coupling is more rigorous than CFD alone
while prediction of gas hold showed good
agreements the experimental
Rectangular bubble [bmim]SCN, [hmim]BF4 , 3D Eulerian–Eulerian Hydrodynamics of new ILs are compared with Present work
column [hmim]PF6 , and amine transient Model conventional amine solvents with constant CO2 flow
solvents AMP, MDEA, MEA rate
Important flow parameters (flow pattern, liquid flow
field, CO2 gas holdup and BSD) are investigated
A CFD model is furnished that is more
computationally economical with high viscous ILs
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 359

Fig. 1 – Flat rectangular bubble column (sizes: 20 cm × 45 cm × 5 cm) (a) 3D-isometric domain with medium grid, gas sparger
in red color 2.4 × 1.2 cm and (b) grid independency, time-averaged vertical liquid velocity in bubble column at height 25 cm.
CO2 superficial velocity is 0.0014 m/s. (For interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)

lute medium grid size is 0.0063 m and the bubble rise velocity,
as proposed by Pfleger et al. (1999) and Buwa and Ranade
(2003), is set to 0.2 m/s. With this, better computational stabil-
ity was achieved when using time step size &t ≤ 0.03 s. The grid
independency is examined for ILs-CO2 as shown in Fig. 1(b).
Results did not show significant change in these three grid res-
olutions. Experience from the previous literature (Buwa and
Ranade, 2002; Díaz et al., 2008) demonstrated that medium
grid is preferred to achieve more precise results with less com-
putational efforts, and thus it is used for the rest of studies.
CFD codes are used to solve 3D domain and governing equa-
tions are described for numerical modeling in the preceding
Section 2.1.
loss during the whole process and temperature keep uniform
throughout the column. Thirdly, liquid is completely filled and
keep at rest initially i.e., no gas traces was present inside the
column. Fourth, the sparger thickness is almost zero hence
pressure drop at inlet is negligible. Finally, the working fluid is
assumed to be isothermal and incompressible. The governing
transport equations for each phase are assembled in Eule-
rian approach after substituting averaged Reynolds number
as follows:
2.1.1. Mass conservation equations
For gas dispersed phase:

2.1. Computational model and equations

∂
($g ˛g ) + ∇. ($g ˛g ug ) = 0 (1)
∂t
Only hydrodynamic parameters are under investigation,
hence some assumptions are taken in this numerical simu-
lation as follows: firstly, there is no absorption between liquid ∂
($g ˛g fi ) + ∇. ($g ˛g ug fi ) = Si (2)
and gasses state. Secondly, there is also no evaporation or heat ∂t

Table 2 – Physical properties of ILs and amine solvents at 313 K.

$ (kg/m3 ) # (Pa s) ! (N/m)

[bmim]SCN 1060.6 0.027 0.043 Vakili-Nezhaad et al.

[hmim]BF4 1134.7 0.066 0.037 (2012)
[hmim]PF6 1279.5 0.162 0.037
30% AMP 986.4 0.0016 0.038 Khan et al. (2016)
30% MDEA 1014.04 0.0020 0.051
30% MEA 999.6 0.0015 0.056
360 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

where Si is the source term. It is normally taken as zero; when correlation gives good agreement with experimental data
single bubble group size is assumed. fi is volume fraction of within the range of ±10%.
bubbles group size-i. More detail is discussed in Section 2.2.
For liquid continuous phase: ⎧
⎨ 22.73Re−0.849 Mo0.020 , 0.5 ≤ Re ≤ 5
∂ CD = ,
($ ˛ ) + ∇. ($l ˛l ul ) = 0 (3) ⎩
∂t l l 20.08Re−0.636 Mo0.046 , 5 ≤ Re ≤ 50
$l |ug − ul |d32 (10)
Re =
2.1.2. Momentum exchange equations #l
For gas dispersed phase: g#4 ($l − $g )
Mo =
! 3 $l 2
∂
($g ˛g ug ) + ∇. ($g ˛g ug ug ) = −˛g ∇P′
∂t Second, the Grace et al. (1976) drag correlation, as given in
! " #$
+∇. ˛g #eff,g ∇ug + ∇uTg + Fg,l + $g ˛g g (4) Eq. (11), is used here for amine solvents because these solvents
exhibit lighter densities and lower viscosities relative to ILs.

For liquid continuous phase: 4

( $ − $ ) gd
l g 32
CD = ,
3 $l VT2
∂ ! " #$ (11)
($ ˛ u ) + ∇. ($l ˛l ul ul ) = −˛l ∇P′ + ∇. ˛l #eff,l ∇ul + ∇uTl #l
∂t l l l VT = Mo−0.149 (J − 0.875)
$l d32
−Fg,l + $l ˛l g (5)

where P′ is the modified pressure defined as Lift force (FLF

l,g ) or transverse force could influence the bub-
ble movement radially outward from the center to small
2 2 bubbles and inward to large size bubbles. Lift force exerted on
P′ = P + # ∇.ul + $l kl (6)
3 eff,l 3 single rising bubble is generated by three main mechanisms:
bubble rotation, bubble share in the liquid and bubble wake
where, #eff,l is the liquid phase effective viscosity, described lift. Here, bubble share induced lift force is applied and given
as #eff,l = #lam,l + #t,l + #tb . #eff,g is the effective viscosity of as
gas phase, calculated as #eff,g = #lam,g + #t,g . Hence, #tb =
C#b $l ˛g dbs |ug − ul | where, #tb is the bubble induced viscosity.
Turbulence induced by the movements of bubbles is taken into FLF
l,g = ˛g $l CL (ug − ul ) (∇ × ul ) (12)
account by an addition term proposed by Sato and Sekoguchi
(1975) with model constant C#b set as 0.6.
where, CL is the non-dimensional lift force coefficient, which
has a significance with respect to its positive, negative and
2.1.3. Interphase interaction force model constant value (Colombo and Fairweather, 2015; Elena Díaz
The interphase momentum transfer term (Fg,l ) is consist of
et al., 2009). Here, Legendre and Magnaudet (1998) lift force
added or virtual mass force (FVM D
l,g ), drag force (Fl,g ), lift force model is applied which has more robust mathematical expres-
(FLF WL
l,g ) and wall lubrication force (Fl,g ). Each type of force has its sion and good agreement with flow rules (Pang and Wei, 2011).
own importance in two phase coupling. When bubbles approach to wall region a wall lubrication
force (FWL
l,g ) is generated against bubbles due to the surface
−Fg,l = Fl,g = FVM D LF WL
l,g + Fl,g + Fl,g + Fl,g (7) tension to restrict them from adhesion on the solid surface
of wall. The general form of wall lubrication force is given by
Additional work due to the acceleration (Antal et al., 1991) and described as
" of two
# phase veloc-
ities is proportional to virtual mass force FVM
l,g . A constant 0.5
is set in current work as the virtual mass value for inviscid flow, →
as used by Deen et al. (2001), Mohd. Akbar et al. (2013). FWL 2
l,g = ˛g $l CWL |ul − ug | nw (13)

D ( 7 )
FVM
l,g = 0.5˛g $l (ug − ul ) (8)
Dt CWL = max , 0.0217Eo (14)
Re1.9

It was found that the impact of drag force (FDl,g ) on flow pat-
tern is more significant than other interfacial forces by Li and
Zhong (2015). The drag force per unit volume in an infinite
liquid is given as (Clift et al., 1978).
3 ˛g $l
FD
l,g = CD |ug − ul | (ug − ul )
4 d32
where, CD is the drag coefficient. In present study two drag
coefficients are incorporated. First, the correlation proposed
by Dong et al. (2010), as given in Eq. (10), was considered
and specifically used for ILs-CO2 system in bubbly flow which
depends on Reynolds number and Morton number. Predicted
→
where nw is the unit normal vector pointing direction away
from the wall. The wall lubrication coefficient CWL is proposed
by Hosokawa et al. (2002). As described in Eq. (14), it depends
on the relative Reynolds Number Re and Eotvos Number Eo.
(9)
2.2. CFD-PBM coupling
General concept of population balance model (PBM) in reac-
tors can be found in the literature (Ramkrishna, 1985). The
basic population balance equation (PBE) based on Ramkris-
han’s (Ramkrishna, 2000a) work is described as:
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 361

*v *∞ *∞
∂ 1
($g ni ) + ∇.($g ug ni ) = $g ( ′ ′ ′ ′
a(v − v , v )n(v − v )n(v )dv − n(v)′ ′ ′
a(v, v )n(v )dv +′
m(v′ )b(v′ )p(v, v′ )n(v′ )dv′ − b(v)n(v) )(15)
∂t 2 + ,- .
0 o v
+ ,- . + ,- . + ,- . deathrateduetobreakupintosmallerbubbles

birthrateduetocoalesenceofsmallerbubbles deathduetoCoalecenseofothers birthrateduetobreakupoflargebubbles

where ni is the number of bubbles per unit volume of group i,

momentum while one disperse phase divide into M-phases
a(v, v′ ) shows the rate of bubble coalescence of size v and v′ ;
(M > 1) and N-classes in discrete inhomogeneous method
and b (v) shows the breakup rate of bubble size v. The number
(Tomiyama and Shimada, 2001), where every class follows
of fragments or children obtained by breakup of bubbles size
its own unique velocity. We restricted our discussion here
v′ is define as m(v′ ). The probability of density function gener-
because our main focus is to get reasonably accurate BSD
ated by bubble breakup of size v′ is represented by p(v, v′ ). More
rather than testing equivalency of these methods. A valuable
details of CFD-PBM coupling and applications of the coales-
comparison between QMOM/DQMOM (Barrett and Webb, 1998;
cence and breakup effect in a two phase flow can be found in
Marchisio and Fox, 2013, 2005) and QMOM/DQMOM/CM can
studies by Chen et al. (2004), Liu and Hinrichsen (2014), Xing
be found in the literature elsewhere (Gupta and Roy, 2013;
et al. (2013). PBE is an Integral–Differential equation so that
Marchisio et al., 2003b,c; Nguyen et al., 2016; Sanyal et al., 2005;
there are unique methods to solve these equations. Generally,
Selma et al., 2010b).
these methods fall in two categories. First category, in which
In this work Homogeneous Discrete Model is adopted and
the internal coordinate (weights and abscissas) are integrate
explained. Under consideration model is also known as Class,
out for moments. The equations for moments are developed
Sectional or Bins Model. CM is relatively expensive and classes
in the context of achieving less computation expenditure. In
must be defined in preliminary. Population balance equation
standard moments of method (SMM) (Randolph and Larson,
is discretized as volume fraction of each bubble class. Bubble
1971; Marchisio et al., 2002), moments are used to transform
density keeps constant and division of classes is done on the
PBE into 4–6 sets of transport equations and then solved with
basis of bubble volume. Fixed pivot approach is developed here
respect to internal coordinates. This approach is useful when
by the method of Kumar and Ramkrishna (1996) for the distri-
entire particle size distribution is not desirable. However, it
bution of bubbles. As described in Section 2.1.1, in " continuity
#
works on the assumptions for constant particle’s aggregation
Eq. (2), fi is the volume of fraction in bubble size-i fi = ˛gi /˛g ,
and size independent growth which poses serious limitations.
and Si represents the source term for coalescence and breakup
These limitations overcome by the proposed model of McGraw
of bubbles, describes as Si = BiC − DiC + BiB − DiB . Disperse phase
(1997) as the exact closure in SMM replaced by quadrature
volume fraction is correlated with bubble number density ni
point approximate closure (Marchisio et al., 2003a,b,c). Hence,
as ni vi = ˛i , where the total volume fraction of disperse phase
quadrature method of moments (QMOM) has the ability to /
is ˛i = ˛g , and the volume fraction of individual group is
track the moments of population with very small error and
can be applied to a border range of applications. Although, i
/
this method losses performance (Fox, 2006), when applied for fi = ˛gi /˛g with fi = 1. Hence, the above PBE is rewritten in
multi-variate system. Further, an efficient product-difference i
the form of fi as:
(PD) algorithm is used to find internal coordinates that will
become more complex for multi-variate case which may be
∂
time consuming. All quadrature points in QMOM assume to (˛g $g fi ) + ∇. (˛g $g ug fi ) = $g vi (BiC − DiC + BiB − DiB ) (16)
∂t
move with one velocity due to this aggregation of particle is
not possible. The assumption of same velocity is relaxed in and the bubble number density is ni defined as:
DQMOM method (Marchisio, 2007; Marchisio and Fox, 2005) as
every point moves with different and independent velocity as
* vi +1

internal coordinate are tracked directly rather than tracking ni (t) = n(v, t)dv (17)
vi
the moments and it is possible to describe multi-variate cases
in this method. Moreover, the difference in velocity increases
The bubble birth and death rates are formulated as follows
the particles interaction and segregation. With this, secondary
phase should be divided in three subclasses to move with j≥k
different velocities (Fan et al., 2004), so that, three extra equa-
/ 1
BiC = (1 − ı )"(vj + vk )a(vj , vk )nj nk ,
tions and source terms are involved in DQMOM method. 2 jk
vi−1 ≤vj +vk ≤vi
In all moments method, transport equations are solved for (18)
/
M
moments only not for entire distributions, so that number dis- DiC = a(vi , vj )ni nj
tribution and Sauter mean diameter is obtained from lower j=1
order moments later on.
Beside this, the discretization of population balance equa-
/
M
tion or stochastic approach is classified as second category BiB = '"ij g(vi )ni
(Bertin et al., 2016; Vanni, 2000), which is famous for entire (19)
j=1
distributions and number density function. This method is DiB = g(vi )ni
applied to get Sauter mean diameter directly rather than
moments of particles. This robust numeric gives PSD as
where the breakage rate " ij is
reasonably accurate as moments method but more scalars
(classes) are needed to solve simultaneously. The discrete * vi+1 * vi+1
vi+1 − v v − vi−1
homogenous method (Ramkrishna, 2000b) assumes one dis- "ij = ˇ(v, vj )dv+ ˇ(v, vj )dv (20)
vi+1 − vi vi − vi−1
perse phase (M = 1), in which all classes travel with same vi vi
362 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

Fig. 2 – Numerical analogy framework.

⎧ " #
⎪
⎪ vi+1 − vj + vk " # 3. Numerical model
⎪
⎨ , vi ≤ vj + vk ≤ vi+1
vi+1 − vi
"= " # (21) Different momentum equations are solved for amine solvents
⎪
⎪ v + v − vi−1 " #
⎪
⎩ j k , vi−1 ≤ vj + vk ≤ vi
and ILs as described in the previous Section 2.1 and shown in
vi − vi−1 Fig. 2. The model description is as follows:
For all amine solvents, a transient formulation to solve
" #
where ˇ v, vj is fragments size distribution; and Navier–Stoke equation, second order implicit discretization is
Finally, the local mean bubble diameter (also known as applied. QUICK interpolation scheme is used to solve momen-
Sauter mean diameter d32 ) which shows the mean bubble size tum and continuity equations. Pressure–velocity equations
for all the bubbles in the domain is obtained. It is an essen- are solved by SIMPLE scheme. Eulerian frame work is adopted
tial parameter to couple the interaction in bubbly flow. d32 is for transient pressure based multiphase flow. Added mass
evaluated every time step according to Eq. (22) and return to force (Eq. (8)), drag coefficient (Eq. (11)), lift force (Eq. (12)),
the momentum equation as shown in the modeling analogy and wall lubrication force (Eqs. (13) and (14)) are incorporated
of Fig. 2. in momentum equation. The continuous phase is static at
inlet and no-slip boundary condition on the wall is imposed
1N for both dispersed and continuous phases. The outlet of the
ni d3i
d32 = 1i=1
N 2
(22) column was at atmospheric pressure. Turbulence multiphase
nd
i=1 i i mixture model is governed by standard k–epsilon equations
(Selma et al., 2010a). CFD–PBM coupling is applied to obtain
where ni is the number of bubbles against the bubble diameter
the bubble size distribution as describe in Fig. 2. CO2 gas is
di .
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 363

Fig. 3 – Validation of the model with IL ([bmim]PF6 ). Predicted and measured gas holdup at (a) 10–20 cm and (b) 50–60 cm of
bubble column at 310 K.

treated as disperse phase so that eleven classes are classified
in homogeneous discrete class models as shown in Table 3.
Around 9–15 classes were sufficient for bubble size distribu-
tion with 0.0014 m/s superficial gas velocity and same size of
bubble column (Bannari et al., 2008; Selma et al., 2010b). Break-
age kernel (Lehr et al., 2002) and aggregation kernel (Luo, 1993;
Luo and Svendsen, 1996) models are adopted for breakup and
coalescence of bubbles phenomena respectively.
The value of under relaxation factors was set as 0.3 for
pressure and 0.7 for momentum. Convergence criteria for all
momentum, turbulence and population conservation equa-
tions were set as every time step will be converged after
third order (10−3 ) magnitude of mass residue drop down.
Approximately less than 100 internal iterations are required
for convergence at medium size grid. Simulation was going on
until a periodic behavior achieved for all variable that is up to
200 s physical time. Time-averaged values were obtained after
skipping first five seconds of initial simulation. In the present
simulation for amine solvents, fixed time step size of 0.01 s was
found quite comparable with medium mesh size. Medium grid
exported on commercial CFD codes of ANSYS Fluent Release
16 for simulation. Economical computation trade is achieved
on medium resolution, i.e., less than 24 h for 200 s of dynamics.

Table 3 – Bubble class size.

Amine solvent-CO2 system
Classes index 1 2 3 4 5 6 7 8 9 10 11
Size (cm) 0.15 0.20 0.28 0.39 0.54 0.75 1.04 1.44 1.99 2.75 3.80

Ionic liquid-CO2 system

Classes index 1 2 3 4 5 6 7 8 9
Size (cm) 0.24 0.33 0.45 0.61 0.83 1.12 1.51 2.04 2.76
364 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

Fig. 4 – Instantaneous snapshots of flow pattern and white stream lines shows vortices in bubble column; air volume
fraction 4% or above is represented by white color, black color is 100% water volume fraction. (a) Right tilt plume
experimental (Upadhyay et al., 2013), Ug = 0.0013 m/s; (b) Left tilt plume experimental (Upadhyay et al., 2013),
Ug = 0.0013 m/s; (c) Right tilt plume experimental (Upadhyay et al., 2013), Ug = 0.0083 m/s; and (d) Left tilt plume
experimental (Buwa et al., 2006), Ug = 0.0083 m/s.

In ILs-CO2 system, the fifth class (0.83 cm) of CO2 bub-
ble size was set as initial bubble diameter at the gas sparger
(described in Section 2). No back flow of gas is imposed at the
upper layer of the ionic liquid because of the high viscosity. It
means only CO2 can pass through the liquid surface. Nine bub-
ble classes stated in Table 3 are spread for the distribution of
CO2 bubbles. These nine classes are found sufficient because
in most of the cases, gas is stratified in ionic liquids (Kaji
et al., 2009). Increasing number of classes generates more con-
servation equations which slow down the overall numerical
model without significant improvement of results. Experi-
enced gained from previous literature, drag coefficient (Eq.
(10)) is adequate to resolve the interphase between gas and ILs
(Bao et al., 2015; Carvajal et al., 2015; Onink et al., 2010; Wang
et al., 2010a). Inclusion of more forces poison the momen-
tum equation, like lift and turbulent dispersion terms create
divergence. Second order implicit formulation failed to get the
convergence. To reduce the truncation error, 0.0005 s time step
size was used with first order implicit scheme. Momentum
and phase equations were solved by QUICK method. Up to 120
internal iterations were required to get convergence for every
time step. On parallel processing, 3–4 h are required for 1 s sim-
ulation. Special correlation (Eq. (10)) of drag coefficient was
written C++ language using visual studio professional 2012.
User Define Functions (UDF) was directly compiled by Graph-
ical User Interface (GUI) after define the exchange property
macro.
3.1. Model validation
To verify the current CFD model, experimental data is taken
from the previous study of Wang et al. (2010a). Ionic liquid
([bmim]PF6 ) is used in cylindrical bubble column and inlet gas
flow consist of 10% CO2 & inert N2 gas. Experimental procedure
and operating conditions can be seen from the experimen-
tal work of Wang et al. (2010a). Here, 60 cm long simulation
domain of cylindrical bubble column is created to verify the
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 365

Fig. 5 – Time-averaged vertical liquid velocity validation with experimental data of (Pfleger et al., 1999) at air
Ug = 0.0013 m/s, (a) Time-averaged vertical liquid velocity at height 37 cm, 25 cm, 13 cm, and (b) Time-averaged vertical liquid
velocity and mean stream tracer in 3D-bubble column. An isometric view shows different heights.

gas holdup in the range of 10–20 cm and 50–60 cm of column
at 310 K. Experimental gas holdup in the absorbing column is
compared with present simulation as shown in Fig. 3(a) and
(b). It can be seen that the predicted gas holdup is satisfac-
tory agreement with experimental measurements and given
model is valid for ILs-CO2 .
The results obtained from proposed CFD model is bench-
marked in Figs. 4 and 5 for given medium grid resolution in
rectangular bubble column. Experimental data from the litera-
ture (Buwa et al., 2006; Pfleger et al., 1999; Upadhyay et al., 2013)
on the water–air system with same physics and operating con-
dition is validated. Gas plum meandering behavior which is
also observed in conventional amine solvents is compared
with experimental snap shots in Fig. 4(a)–(d). At a relatively
higher superficial air velocity (0.0083 m/s), plume oscillated
after every 5–6 s and air is seen to detach from the main
plume unlike the lower superficial air velocity. The predicted
time-averaged vertical liquid velocity at different height of
the rectangular column (13 cm, 25 cm, and 37 cm) is compared
with LDA measurements in Fig. 5(a). The results show that the
proposed CFD model gives good agreements with experiment
data and is also applicable for lighter viscosity liquids.
4. Results and discussions
4.1. Flow pattern
Flow pattern is characterized in disperse phase with super-
ficial velocity of 0.0014 m/s along pseudo steady state flow.
The given CFD model in Section 3 is capable to capture both
instantaneous and time-averaged flow regime.
Flow circulation mainly depends on bubble size and gas
holdup. Instantaneous snapshots of fully developed plume
in three ILs are shown in Fig. 6(a)–(c) at CO2 flow rate is
1.33 × 10−5 m3 /s. In the range of viscosity from 0.02 Pa s to
0.16 Pa s, all ionic liquids showed the same non oscillatory
flow pattern and CO2 plume did not oscillate in ILs. Two small
stationary vortices formed just above the sparger and then
CO2 moved upward continuously in viscous ILs in straight
direction without swirling mode. Fully developed plume was
consist of continues cluster of CO2 gas. Liquid moved down-
ward near wall and upward along with CO2 plume at the center
of rectangular column. Morton number in Eq. (10) played a sig-
nificant effect for interaction of CO2 with viscous media, as it
is directly proportional to the fourth power of liquid viscos-
ity.
Flow regime of amine solvents was totally different from
ILs-CO2 system. In all three amine solvents, relatively similar
dynamics is visualized and liquid circulation is observed. Slow
speed vortices are seen that attached at the bend of plume
oscillation. Direction of circulation of vortex depends on the
plume bends direction. Clockwise and anticlockwise vortices
occurred beneath the plume bends. These meandering behav-
iors are shown in Fig. 6(d)–(f). From the animation videos (not
depicted here because it takes large memory), it is noticed that
fully developed plume started oscillation after 20 s and com-
pleted its one cycle in average of 12–13 s. No detachment of CO2
gas phase or separate recirculation was obtained. The time-
averaged flow pattern of [hmim]PF6 and MEA is compared in
Fig. 7. It revealed that CO2 gas exhibited steady state flow in all
ILs, while unsteady state flow was found in all amine solvents.
366 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Fig. 6 – Instantaneous snapshots of flow pattern. ILs (a–c) in the comparison of amine solvents (d–f) with superficial CO2
velocity of 0.0014 m/s.
Fig. 7 – Comparison of time-averaged flow pattern between
[hmim]PF6 and MEA with superficial CO2 velocity of
0.0014 m/s. Volume fraction of CO2 in (a) 0–0.005 & (b)
0–0.08.
4.2. Averaged velocity profile
The time-averaged velocity magnitudes of ILs and amine
solvents are explained by CFD model at the flow rate of
1.33 × 10−5 m3 /s. As shown in Fig. 8(a)–(c), ILs circulated
throughout the column as it moved upward at the central
region and downward along the stationary wall. The stream
lines indicated that ILs moved in the vortices when CO2 started
to eject from the sparger. Small eddies formed at two sides
just above the gas sparger and frequently disappear with col-
umn height. At this lower part, ILs and CO2 velocities got
higher and more mixing took place. Mean velocity profile
of ILs is described as it is changed with respect to ions. As
shown in Fig. 8(b) and (c), two ionic liquids with same cation
and different anion showed gradual change in velocity profile
(i.e. [hmim]+ with BF4 − & PF6 − ). The low to high viscosity of
ILs ([bmim]SCN to [hmim]PF6 ) explained that higher viscosity
dampens all the movements of fluid in bubble column. The
gradual change in mean velocity magnitude with respect to
viscosity is described as [bmim]SCN > [hmim]BF4 > [hmim]PF6 .
By contrast, as shown in Fig. 8(d)–(f), amine solvents moved in
a smooth angular shape with central CO2 stream right above
the gas sparger. Amine solvents circulated in relatively bigger
mixing vortices comparing to ILs.
Investigation of mean velocity along with column height
(13 cm, 25 cm, and 37 cm) brought statistical judgment about
ILs as well as amine solvents. It can be seen from Fig. 9(a), (c)
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Fig. 8 – Time-averaged velocity magnitude. ILs (a–c) in the comparison of amine solvents (d–f) with superficial CO2 velocity
of 0.0014 m/s.
and (e), ILs velocity magnitude increased gradually along with
height. In bubble column, the increase of mean liquid velocity
with height is explained as that the ionic liquids are already
in straight upward circulation and the continuous flow rate
of coming gas exhibits less efforts to create a path in viscous
media. It is noticed that viscosity effects is more pronounced
than other physical properties of ionic liquid. Viscosity gov-
erned the flow behavior of ionic liquids rather than surface
tension and density. In amine solvents, viscosity values were
very low with respect to ILs, so that CO2 oscillated freely
and the liquid velocity kept higher throughout. As shown in
Fig. 9(b), (d) and (f), the velocity of liquid adjacent to wall was
high because of vortex formed along the wall of column. A
gradual decreasing pattern in mean velocity profile along with
height of the column is observed, which is more significant
far from gas sparger (25 cm–37 cm). The time-averaged veloc-
ity magnitude of ILs was almost 10 times less than the other
three amine solvents.
4.3. Gas hold-up
Another key parameter to explain the flow behavior inside the
bubble column is the gas holdup. Fig. 10 compared the time-
averaged CO2 gas holdup in ILs and amine solvents. In ILs
systems (Fig. 10(a), (c) and (e)), higher viscosity of ionic liq-
uids shrank down the gas holdup, consequently a decreasing
367
trend of CO2 total gas holdup is observed along the height
of bubble column. At the height of 13 cm (Fig. 10(a)), CO2
gas distribution was more concentrated at central region and
negligible volume fraction traces were found at the lateral
part of rectangular column. A damping gas hold profile near
sparger, as previously observed in Fig. 7(a), is explained as
that the much higher viscous media of ([hmim]PF6 ) tried to
bifurcate the CO2 plume. Although, viscosity has found to be
the governing factor in flow regime and higher viscosity has
a pronounced decrease effect on the overall gas holdup at a
given flow rate in bubble column (Lange et al., 2013; Urseanu
et al., 2003). At the height of 25 cm (Fig. 10(c)) and 37 cm
(Fig. 10(e)), negligible decreasing trend was found along with
column height. One possibility could be that the gas holdup
kept higher viscosity (>0.01 Pa s) within the narrow range of
investigated ILs (0.02 Pa s–0.1 Pa s), almost in the same way as
liquids. In other words, viscosity loses its significance at above
0.01 Pa s. In this condition the surface tension plays a govern-
ing role on gas holdup rather than viscosity effect. Xing et al.
(2013) also concluded that total gas holdup was almost inde-
pendent in a narrow range of liquid viscosity. Moreover, ionic
liquids with cation [bmim]+ keep relatively higher value of
surface tension than ionic liquids with cation [hmim]+ (see
Table 2). Minute difference in surface tension (0.006 N/m) gives
different peaks of time-averaged gas holdup at three sec-
tions of the bubble column. Therefore, it can be concluded
368 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

Fig. 9 – Height wise analysis of time-averaged velocity magnitude in ILs and amine solvents. Column height of 13 cm (a,b)
25 cm (c,d) and 37 cm (e,f).

that in homogeneous flow, ionic liquids with higher value
of surface tensions exhibits relatively lower gas hold up, as
conformed by the experimental investigation of Götz et al.
(2016).
In amine solvents, as shown in Fig. 10(b), (d), and (f), much
higher gas holdup is obtained than the ILs system. CO2 gas
continuously injected at a fix flow rate of 1.33 × 10−5 m3 /s. In
less viscous media, CO2 plume can easily move in the lat-
eral direction. The time-averaged gas holdup was affected
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 369

Fig. 10 – Height wise analysis of time-averaged CO2 gas holdup in ILs and amine solvents. Column height of 13 cm (a,b),
25 cm (c,d) and 37 cm (e,f).

by plume oscillation. Gas plume was less windy at above
the column height of 25 cm. The combined effect of liq-
uid properties (density, viscosity and surface tension) was
observed in amine solvents. As the comparison of the gas
holdup between ILs and amine solvents, CO2 plume tended
to confide at the center of the column in ILs, while it spread
throughout in amine solvents. The mean gas holdup was 15–30
times higher in amine solvent because of the lower viscos-
ity.
370 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Fig. 11 – Bubble size distribution at the center plane, 12 different sizes of CO2 is shown by 12 bands. ILs (a–c) in the
comparison of amine solvents (d–f).
4.4. Bubble size distribution (BSD)
In ionic liquids system, CO2 bubble sizes are pivoted in 9
classes with diameter from 0.24 cm to 2.76 cm in a geo-
metric series as vi+1 = 2r vi . The smallest bubble volume is
v1 = 7.23 × 10−3 cm3 and size ratio is set as r = 1.3. Similar
scheme was adopted for amine solvents with 11 classes. In
Fig. 11(a)–(c), contours of Sauter mean diameter predicted
by current CFD model demonstrated that the small bubble
size localized in lower section (below 13 cm) of bubble col-
umn. Small stationary eddies near the gas sparger had enough
energy to break up the big bubbles and then those bubbles
became a part of large circulation loop along the wall. The large
vortex loops carried out those small bubbles in a recirculation
where small bubbles interacted with each other, and the bub-
ble size changed from 0.31 cm to 0.82 cm due to coalescence.
Those phenomena could be seen along the column wall more
in [bmim]SCN & [bmim]BF4 , and less in [hmim]PF6 . Hence, the
lower section of the bubble column had highest fractions of
small bubbles due to high breakage rate. Minute coalescence
was seen apart from wall region (away 0.04 m from both sides
of wall, Fig. 12(a)). One reason could be that coalescence and
brakeage of bubbles were in dynamic equilibrium or because
that high viscosity bubbles did not break at the center part
in axial direction. Another reason could be that the low coa-
lescence behavior of ILs was predominated. The same kind
of non-coalescence behavior of ILs was found experimentally
by Kaji et al. (2009). It was explained that the intermedi-
ate size bubbles traveled together in ionic liquids but did
not coalescence. ILs [hmim]PF6 & [hmim]BF4 exhibited non-
coalescence bubble behavior more obviously than [bmim]SCN.
It is observed that the variation of viscosity played important
role in BSD. With the viscosity of ionic liquid increases, the
turbulence decreases consequently, and the bubble breakup
rate decreases. Thus, the predicted sequence was observed
in the order of [hmim]PF6 > [hmim]BF4 > [bmim]SCN. Further-
more, large bubble size was observed to accumulate in the
upper section (above 37 cm) of bubble column. In that sec-
tion, the bubble size distribution was more stable than the
middle and lower sections, because of the negligible breakage
rate near the wall region. Coalescence dominated on breakage
as bubble moved upward. Upper section consisted of uniform
bubble size distribution ranging from 0.89 cm to 0.95 cm.
As shown in Fig. 11(d)–(f), amine solvents exhibited very
different bubble breakage and coalescence phenomena. Large
bubbles were seen at the central line of the column and
uniformly decreased towards the wall. Bubble coalescence
occurred in axial direction and large bubble broke down into
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375 371

Fig. 12 – CO2 bubble size at the mid length (22.5 cm) of bubble column. (a) ILs and (b) amine solvents.

small bubble size in later direction. In all three amine solvents,

the middle section (25–37 cm of column height) presented
maximum size range of CO2 bubbles. With the comparisons of
ILs, it was observed that bubbles of CO2 partially occupied the
column domain and bubble density near wall was negligible
in the cases of amine solvents. Bubble coalescence and break-
age happened side by side, hence, wider bubble size range
(0.13–1.5 cm) was observed in amine solvents. Fig. 12(a) and (b)
described the difference of bubble population phenomenon.
Clearly, the maximum bubble breakage happen near wall in
ILs (see Fig. 12(a)); while negligible bubble size was found near
wall that grew gradually toward center of column in amine
solvents (see Fig. 12(b)).

4.5. Hydrodynamic aspects in industrial tray column

In the view of industrial aspects, above hydrodynamic char-

acteristics (average liquid velocity profile, mean gas holdup
and BSD) are further investigated in an example of industrial
tray column in this study. Bubble column in Fig. 1(a) is modi-
fied into an industrial tray column as two trays are located at
the vertical distance of ±10 cm from the mid length (22.5 cm)
of bubble column. These two trays acted as baffles instead of
sieve tray or bubble caps and liquid are assumed at rest ini-
tially, i.e. no liquid inlet. Since, gas–liquid interaction is focal
Fig. 13 – An industrial tray column in which two trays are
point hence; simplifications are adopted rather than solving
located at the vertical distance of ±10 cm and 2.4 × 1.2 cm
more complicated geometry. CO2 gas is injected through three
size of each inlet; ±5 cm apart horizontally from central
inlets (2.4 × 1.2 cm size of each inlet; ±5 cm apart horizontally
inlet.
from central inlet). The gas superficial velocity at each inlet
set as 0.00288 m/s as show in Fig. 13.
Ionic liquid; [bmim]BF4 and 90%-[bmim]BF4 solution is viscosity that facilities the movement of liquid and gas. The
used to compare the hydrodynamics with 30%-MEA in this conventional amine solvent (30%-MEA) as compared to ILs is
industrial tray column. As shown in Fig. 14(a)–(c), baffles play observed to move with almost 10-time faster velocity magni-
a key role in liquid circulation and recirculation. The circula- tude and fast moving liquid circulation is depicted in Fig. 14(c).
tion phenomenon promotes mixing between liquids and CO2 Time-average gas holdup is explained as CO2 moves straight
gas which was not observed in previous Fig. 8(a)–(c). Liquid direction without swirling mode in both type of ILs (pure &
velocity is higher in 90%-[bmim]BF4 (Fig. 14(b)) as compared 90% solution) and baffles interrupt this flow pattern as shown
to pure [bmim]BF4 (Fig. 14(a)) because of minute quantity of in Fig. 14(d) and (e). Besides ILs, CO2 gas holdup is higher in
water in ionic liquid (90%-[bmim]BF4 ) dramatically reduce the 30%-MEA as CO2 recirculation can be seen in Fig. 14(f) near the
372 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375

Fig. 14 – Time-averaged velocity magnitude (a–c) and CO2 gas holdup (d–f) with superficial CO2 velocity of 0.00288 m/s.

edges of baffles area. Here, the design of bubble column may
be helpful to improve the mixing flow between ILs-CO2 . Real
industrial column of bubble cap or sieve trays (Shenastaghi
et al., 2018; Tang et al., 2017) can be further investigated in
future for the CFD hydrodynamics of ILs-CO2 system.
As discussed above in Section 4.4, the BSD under goes sev-
eral types of phenomenon (segregation, breakage, effect of
small eddies and large vortex loop etc.). In tray column, BDS is
frequently changed because of additional effect of baffles. As
shown in Fig. 15(a) and (b), near the edges of baffles more coa-
lescence is achieved which is not observed in column without
baffles in Fig. 11(b) and (c) where constant BDS is depicted. In
30%-MEA solution, breakage and coalescence phenomenon in
the middle region of two baffles is more rampant because of
lighter viscosity than other ILs as shown in Fig. 15(c). Hence,
for Ionic liquids, tray column may be a suitable option with
respect to achieve strong mixing.
5. Conclusions
The hydrodynamics of ILs and amine solvents is success-
fully distinguished and explained by the developed numerical
model in this work. Experimentally validated numerical model
used to explain the hydrodynamic characteristics (flow pat-
tern, average liquid velocity profile, mean gas holdup and
BSD) inside the bubble column are successfully revealed to
differentiate the behavior of ILs-CO2 and conventional amine
solvents-CO2 . The following conclusions can be drawn:
• All the three investigated ILs have shown identical flow pat-
tern that leads to the opportunity of utilizing reported flow
pattern in industrial column for new kind of ILs. On the
other hand, instable winding plume is observed to affect
the dynamics in amine solvents.
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Fig. 15 – Bubble size distribution at the center plane, 10 different sizes of CO2 is shown by 10 bands. (a) Pure [BMIM]BF4 (b)
90%-[BMIM]BF4 and (c) 30%-MEA.
• The ILs velocity magnitude was highly affected by viscosity
which slowed down the movement of ILs more significantly
than amine solvents.
• In ILs, the CO2 gas holdup was extremely lower compared
with amine solvents, due to the higher value of viscosity.
Surface tension showed its significance when gas holdup
apparently independent of liquid viscosity.
• In ILs, the non-coalescence phenomena dominated, while
both coalescence and breakup of CO2 bubbles exhibited in
amine solvents.
• It is observed that baffle type industrial tray column
may overcome the difficulty of mixing between ILs and
CO2 .
These characteristics may be helpful to scale up indus-
trial absorption column for gas purification system, and the
purpose of distinguishing the hydrodynamics of ILs and con-
ventional amine solvent is successfully achieved. The current
hydrodynamic data may assist to replace corrosive amine sol-
vents with non-volatile and easy to regenerate ILs in CO2
capture system. The experimentally validated CFD model also
shows the ability to resolve the complex interphase between
ILs & CO2 and can be applied for more real industrial bubble
cap and sieve tray with new ILs in future.
Acknowledgments
This work is financially supported by State Key Laboratory
of Organic-Inorganic Composites, China (oic-201601012). The
authors extend their appreciation to the International Sci-
entific Partnership Program ISPP at King Saud University for
funding this research work through ISPP#0080. We also give
many thanks to Professor Zhipeng Li and Professor Ziqi Cai
for their very helpful discussion.
373
References
Antal, S.P., Lahey, R.T., Flaherty, J.E., 1991. Analysis of phase
distribution in fully developed laminar bubbly two-phase flow.
Int. J. Multiph. Flow 17, 635–652.
Bannari, R., Kerdouss, F., Selma, B., Bannari, A., Proulx, P., 2008.
Three-dimensional mathematical modeling of dispersed
two-phase flow using class method of population balance in
bubble columns. Comput. Chem. Eng. 32, 3224–3237.
Bao, D., Zhang, X., Dong, H., Ouyang, Z., Zhang, X., Zhang, S.,
2015. Numerical simulations of bubble behavior and mass
transfer in CO2 capture system with ionic liquids. Chem. Eng.
Sci. 135, 76–88.
Barrett, J.C., Webb, N.A., 1998. A comparison of some
approximate methods for solving the aerosol general dynamic
equation. J. Aerosol Sci. 29, 31–39.
Bertin, D., Cotabarren, I., Piña, J., Bucalá, V., 2016. Population
balance discretization for growth, attrition, aggregation,
breakage and nucleation. Comput. Chem. Eng. 84, 132–150.
Buwa, V.V., Ranade, V.V., 2002. Dynamics of gas–liquid flow in a
rectangular bubble column: experiments and
single/multi-group CFD simulations. Chem. Eng. Sci. 57,
4715–4736.
Buwa, V.V., Ranade, V.V., 2003. Mixing in bubble column reactors:
role of unsteady flow structures. Can. J. Chem. Eng. 81,
402–411.
Buwa, V.V., Deo, D.S., Ranade, V.V., 2006. Eulerian–Lagrangian
simulations of unsteady gas–liquid flows in bubble columns.
Int. J. Multiph. Flow 32, 864–885.
Carvajal, D., Carlesi, C., Meléndez-Vejar, V., Vásquez-Sandoval, D.,
Monteverde-Videla, A.H.A., Bensaid, S., 2015. Numerical
simulation of single-bubble dynamics in high-viscosity ionic
liquids using the level-set method. Chem. Eng. Technol. 38,
473–481.
Chen, P., Sanyal, J., Dudukovic, M.P., 2004. CFD modeling of bubble
columns flows: implementation of population balance. Chem.
Eng. Sci. 59, 5201–5207.
Clift, R., Grace, J.R., Weber, M.E., 1978. Bubbles, Drops, and
Particles. Academic Press, Inc., New York.
374 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Colombo, M., Fairweather, M., 2015. Multiphase turbulence in
bubbly flows: RANS simulations. Int. J. Multiph. Flow 77,
222–243.
Díaz, M.E., Montes, F.J., Galán, M.A., 2006. Influence of aspect ratio
and superficial gas velocity on the evolution of unsteady flow
structures and flow transitions in a rectangular
two-dimensional bubble column. Ind. Eng. Chem. Res. 45,
7301–7312.
Díaz, M.E., Iranzo, A., Cuadra, D., Barbero, R., Montes, F.J., Galán,
M.A., 2008. Numerical simulation of the gas–liquid flow in a
laboratory scale bubble column: influence of bubble size
distribution and non-drag forces. Chem. Eng. J. 139, 363–379.
Deen, N.G., Solberg, T., Hjertager, B.H., 2001. Large eddy
simulation of the gas–liquid flow in a square cross-sectioned
bubble column. Chem. Eng. Sci. 56, 6341–6349.
Dong, H., Wang, X., Liu, L., Zhang, X., Zhang, S., 2010. The rise
and deformation of a single bubble in ionic liquids. Chem.
Eng. Sci. 65, 3240–3248.
Elena Díaz, M., Montes, F.J., Galán, M.A., 2009. Influence of the lift
force closures on the numerical simulation of bubble plumes
in a rectangular bubble column. Chem. Eng. Sci. 64, 930–944.
Fan, R., Marchisio, D.L., Fox, R.O., 2004. Application of the direct
quadrature method of moments to polydisperse gas–solid
fluidized beds. Powder Technol. 139, 7–20.
Fox, R.O., 2006. Bivariate direct quadrature method of moments
for coagulation and sintering of particle populations. J.
Aerosol Sci. 37, 1562–1580.
Götz, M., Lefebvre, J., Mörs, F., Reimert, R., Graf, F., Kolb, T., 2016.
Hydrodynamics of organic and ionic liquids in a slurry bubble
column reactor operated at elevated temperatures. Chem.
Eng. J. 286, 348–360.
Grace, J.R., Wairegi, T., Nguyen, T.H., 1976. Shapes and velocities
of single drops and bubbles moving freely through immiscible
liquids. Trans. Inst. Chem. Eng. 54, 167–173.
Gupta, A., Roy, S., 2013. Euler–Euler simulation of bubbly flow in a
rectangular bubble column: experimental validation with
radioactive particle tracking. Chem. Eng. J. 225, 818–836.
Hosokawa, S., Tomiyama, A., Misaki, S., Hamada, T., 2002. Lateral
migration of single bubbles due to the presence of wall. In:
Proc. ASME Joint U.S.–European Fluids Engineering Division
Conference, FEDSM2002, Montreal, Canada, p. 855.
Kaji, R., Zhao, D., Licence, P., Azzopardi, B.J., 2009. Studies of the
interaction of ionic liquid and gas in a small-diameter bubble
column. Ind. Eng. Chem. Res. 48, 7938–7944.
Keskin, S., Kayrak-Talay, D., Akman, U., Hortaçsu, Ö., 2007. A
review of ionic liquids towards supercritical fluid applications.
J. Supercrit. Fluids 43, 150–180.
Khan, A.A., Halder, G.N., Saha, A.K., 2016. Comparing CO2
removal characteristics of aqueous solutions of
monoethanolamine, 2-amino-2-methyl-1-propanol,
methyldiethanolamine and piperazine through absorption
process. Int. J. Greenh. Gas Control 50, 179–189.
Krawczyk, M., Kamiński, K., Petera, J., 2012. Experimental and
numerical investigation of electrostatic spray liquid–liquid
extraction with ionic liquids. Chem. Process Eng. 33.
Kumar, S., Ramkrishna, D., 1996. On the solution of population
balance equations by discretization—I. A fixed pivot
technique. Chem. Eng. Sci. 51, 1311–1332.
Lange, V., Azzopardi, B.J., Licence, P., 2013. Hydrodynamics of
ionic liquids in bubble columns. In: Ionic Liquids— New
Aspects for the Future. InTech.
Legendre, D., Magnaudet, J., 1998. The lift force on a spherical
bubble in a viscous linear shear flow. J. Fluid Mech. 368,
81–126.
Lehr, F., Millies, M., Mewes, D., 2002. Bubble-size distributions and
flow fields in bubble columns. AlChE J. 48, 2426–2443.
Li, W., Zhong, W., 2015. CFD simulation of hydrodynamics of
gas–liquid–solid three-phase bubble column. Powder Technol.
286, 766–788.
Liu, Y., Hinrichsen, O., 2014. Study on CFD–PBM turbulence
closures based on k–! and Reynolds stress models for
heterogeneous bubble column flows. Comput. Fluids 105,
91–100.
Luo, H., Svendsen, H.F., 1996. Theoretical model for drop and
bubble breakup in turbulent dispersions. AlChE J. 42,
1225–1233.
Luo, H., 1993. Coalescence, Breakup and Liquid Circulation in
Bubble Column Reactors, D.Sc. Norwegian Institute of
Technology.
Ma, T., Ziegenhein, T., Lucas, D., Fröhlich, J., 2016. Large eddy
simulations of the gas–liquid flow in a rectangular bubble
column. Nucl. Eng. Des. 299, 146–153.
Marchisio, D.L., Fox, R.O., 2005. Solution of population balance
equations using the direct quadrature method of moments. J.
Aerosol Sci. 36, 43–73.
Marchisio, Daniele L., Fox, R.O., 2013. Computational Models for
Polydisperse Particulate And Multiphase Systems. Cambridge
University Press, United Kingdom.
Marchisio, D.L., Barresi, A.A., Garbero, M., 2002. Nucleation,
growth, and agglomeration in barium sulfate turbulent
precipitation. AlChE J. 48, 2039–2050.
Marchisio, D.L., Dennis Vigil, R., Fox, R.O., 2003a. Implementation
of the quadrature method of moments in CFD codes for
aggregation–breakage problems. Chem. Eng. Sci. 58,
3337–3351.
Marchisio, D.L., Pikturna, J.T., Fox, R.O., Vigil, R.D., Barresi, A.A.,
2003b. Quadrature method of moments for
population-balance equations. AIChE J. 49, 1266–1276.
Marchisio, D.L., Vigil, R.D., Fox, R.O., 2003c. Quadrature method of
moments for aggregation–breakage processes. J. Colloid
Interface Sci. 258, 322–334.
Marchisio, D.L., 2007. Quadrature method of moments for
poly-disperse flows. In: Marchisio, D.L., Fox, R.O. (Eds.),
Multiphase Reacting Flows: Modelling and Simulation.
Springer Vienna, Vienna, pp. 41–77.
McGraw, R., 1997. Description of aerosol dynamics by the
quadrature method of moments. Aerosol Sci. Technol. 27,
255–265.
Mohd. Akbar, M.H.b., Hayashi, K., Lucas, D., Tomiyama, A., 2013.
Effects of inlet condition on flow structure of bubbly flow in a
rectangular column. Chem. Eng. Sci. 104, 166–176.
Nguyen, T.T., Laurent, F., Fox, R.O., Massot, M., 2016. Solution of
population balance equations in applications with fine
particles: mathematical modeling and numerical schemes. J.
Comput. Phys. 325, 129–156.
Onink, F., Drumm, C., Meindersma, G.W., Bart, H.-J., de Haan, A.B.,
2010. Hydrodynamic behavior analysis of a rotating disc
contactor for aromatics extraction with
4-methyl-butyl-pyridinium·BF4 by CFD. Chem. Eng. J. 160,
511–521.
Ouyang, Z., Bao, D., Zhang, X., Dong, H., Yan, R., Zhang, X., Zhang,
S., 2015. Numerical simulation of CO2 -ionic liquid flow in a
stirred tank. Sci. China Chem. 58, 1918–1928.
Pang, M.J., Wei, J.J., 2011. Analysis of drag and lift coefficient
expressions of bubbly flow system for low to medium
Reynolds number. Nucl. Eng. Des. 241, 2204–2213.
Pfleger, D., Gomes, S., Gilbert, N., Wagner, H.G., 1999.
Hydrodynamic simulations of laboratory scale bubble
columns fundamental studies of the Eulerian–Eulerian
modelling approach. Chem. Eng. Sci. 54, 5091–5099.
Plechkova, N.V., Seddon, K.R., 2008. Applications of ionic liquids
in the chemical industry. Chem. Soc. Rev. 37, 123–150.
Prado, R., Weber, C.C., 2016. Chapter 1 – applications of ionic
liquids A2 – Kuzmina, Olga. In: Hallett, J.P. (Ed.), Application,
Purification, and Recovery of Ionic Liquids. Elsevier,
Amsterdam, pp. 1–58.
Ramkrishna, D., 1985. The status of population balances. Rev.
Chem. Eng., 49.
Ramkrishna, D., 2000a. Chapter 3—birth and death functions. In:
Population Balances. Academic Press, San Diego, pp. 47–116.
Ramkrishna, D., 2000b. Chapter 4—the solution of population
balance equations. In: Population Balances. Academic Press,
San Diego, pp. 117–195.
Randolph, A.D., Larson, M.A., 1971. In: Larson, M.A. (Ed.), Theory
of Particulate Processes. Academic Press, p. iv.
 Chemical Engineering Research and Design 1 2 9 ( 2 0 1 8 ) 356–375
Razavi, S.M.R., Rezakazemi, M., Albadarin, A.B., Shirazian, S.,
2016. Simulation of CO2 absorption by solution of ammonium
ionic liquid in hollow-fiber contactors. Chem. Eng. Process.
108, 27–34.
Sanyal, J., Marchisio, D.L., Fox, R.O., Dhanasekharan, K., 2005. On
the comparison between population balance models for CFD
simulation of bubble columns. Ind. Eng. Chem. Res. 44,
5063–5072.
Sato, Y., Sekoguchi, K., 1975. Liquid velocity distribution in
two-phase bubble flow. Int. J. Multiph. Flow 2, 79–95.
Schiller, Nauman, 1935. A drag coefficient correlation. Z. Ver.
Dtsch. Ing. 77, 318–320.
Selma, B., Bannari, R., Proulx, P., 2010a. A full integration of a
dispersion and interface closures in the standard model of
turbulence. Chem. Eng. Sci. 65, 5417–5428.
Selma, B., Bannari, R., Proulx, P., 2010b. Simulation of bubbly
flows: comparison between direct quadrature method of
moments (DQMOM) and method of classes (CM). Chem. Eng.
Sci. 65, 1925–1941.
Shenastaghi, F.K., Roshdi, S., Kasiri, N., Hasan Khanof, M., 2018.
CFD simulation and experimental validation of bubble cap
tray hydrodynamics. Sep. Purif. Technol. 192, 110–122.
Tang, M., Zhang, S., Wang, D., Liu, Y., Zhang, Y., 2017. CFD
simulation and experimental study of dry pressure drop and
gas flow distribution of the tridimensional rotational flow
sieve tray. Chem. Eng. Res. Des. 126, 241–254.
Tomiyama, A., Shimada, N., 2001. (N+2)-field modeling for bubbly
flow simulation. Comput. Fluid Dyn. J., 418–426.
Upadhyay, R.K., Pant, H.J., Roy, S., 2013. Liquid flow patterns in
rectangular air-water bubble column investigated with
radioactive particle tracking. Chem. Eng. Sci. 96, 152–164.
375
Urseanu, M.I., Guit, R.P.M., Stankiewicz, A., van Kranenburg, G.,
Lommen, J.H.G.M., 2003. Influence of operating pressure on
the gas hold-up in bubble columns for high viscous media.
Chem. Eng. Sci. 58, 697–704.
Vakili-Nezhaad, G., Vatani, M., Asghari, M., Ashour, I., 2012. Effect
of temperature on the physical properties of
1-butyl-3-methylimidazolium based ionic liquids with
thiocyanate and tetrafluoroborate anions, and
1-hexyl-3-methylimidazolium with tetrafluoroborate and
hexafluorophosphate anions. J. Chem. Thermodyn. 54,
148–154.
Vanni, M., 2000. Approximate population balance equations for
aggregation–breakage processes. J. Colloid Interface Sci. 221,
143–160.
Wang, X., Dong, H., Zhang, X., Xu, Y., Zhang, S., 2010a. Numerical
simulation of absorbing CO2 with ionic liquids. Chem. Eng.
Technol. 33, 1615–1624.
Wang, X., Dong, H., Zhang, X., Yu, L., Zhang, S., Xu, Y., 2010b.
Numerical simulation of single bubble motion in ionic liquids.
Chem. Eng. Sci. 65, 6036–6047.
Xing, C., Wang, T., Wang, J., 2013. Experimental study and
numerical simulation with a coupled CFD–PBM model of the
effect of liquid viscosity in a bubble column. Chem. Eng. Sci.
95, 313–322.
Zhang, X., Dong, H., Huang, Y., Li, C., Zhang, X., 2012.
Experimental study on gas holdup and bubble behavior in
carbon capture systems with ionic liquid. Chemi. Eng. J. 209,
607–615.
Zhang, X., Dong, H., Bao, D., Huang, Y., Zhang, X., Zhang, S., 2014.
Effect of small amount of water on CO2 bubble behavior in
ionic liquid systems. Ind. Eng. Chem. Res. 53, 428–439.