Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Chemical Engineering Science 230 (2021) 116218

Contents lists available at ScienceDirect

Chemical Engineering Science


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

CFD modeling of reactive absorption of CO2 in aqueous NaOH in a


rectangular bubble column: Comparison of mass transfer and
enhancement factor model
Junda Liu, Ping Zhou, Liu Liu ⇑, Si Chen, Yanpo Song, Hongjie Yan
Central South University, School of Energy Science and Engineering, Lushannan Road 932, 410083 Changsha, China

h i g h l i g h t s

 The Henket1 model was found to be the most accurate among the assessed four mass transfer models.
 The enhancement factor model of Hlawitaschka performs better than that with a constant of 1.
 The mass transfer model has very little effect on pH variation at the end of the reaction.

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

Article history: Presently various mass transfer rate models have been developed to predict the mass transfer rate in the
Received 22 May 2020 column. Also, different enhancement factor models were proposed to include the influence of chemical
Received in revised form 24 September reaction on mass transfer process. However, a comprehensive assessment of these models has rarely been
2020
performed, leaving the model accuracy arguable. In this study, four typical mass transfer rate models and
Accepted 16 October 2020
Available online 20 October 2020
two enhancement factor models were employed to simulate the gas-fluid mass transfer behaviors of a
bubble plume rising in a rectangular bubble column, and their accuracies were evaluated by comparing
the simulation results with the published experimental data. The results show that the Henket1 model is
Keywords:
Bubble column
the most accurate among the assessed four mass transfer models. The model of Hlawitschka performs
Two-phase flow better than that with a constant enhancement factor. The mass transfer model has very little effect on
Chemical reaction pH variation at the end of the reaction.
Mass transfer rate model Ó 2020 Elsevier Ltd. All rights reserved.
CFD

1. Introduction Hlawitschka et al., 2017; Kováts et al., 2017, 2018). Bubble size dis-
tribution, liquid velocity and gas hold-up can be obtained by Parti-
Bubble columns are widely used in chemical, biotechnological cle image velocimetry (PIV) and high-speed shadowgraphy.
and environmental engineering industries due to the advantages Darmana et al., (2007) measured the time-dependent pH-value at
of simple construction, low operating costs and good mass transfer a single location in the column using a glass electrode.
capacity (Liu et al., 2019a,b). Various flow patterns exist in bubble Hlawitschka et al. (2017) and Kováts et al. (2017) employed a 2-
columns as a result of different superficial gas velocities and oper- Tracer Laser Induced Fluorescence(2 T-LIF) for the sake of the mea-
ating conditions used. Moreover, mass transfer together with the surements of the local pH-value. However, it is quite difficult to
chemical reaction prevail in the column, which makes the transfer acquire comprehensive data by present experimental techniques
phenomenon very complicated. Therefore, comprehensive studies due to the complexity of multiphase reaction flow. Fortunately,
are required in order to reveal the flow behavior and mass transfer computational fluid dynamics (CFD) is getting more and more effi-
mechanism in the column. cient and powerful for understanding these complex reaction flow
Both experiments and simulations are two approaches to study behaviors. Jia and Zhang (2017) studied the chemisorption process
the phenomena happened in the column. Experiments of the reac- of a spherical CO2 bubble in NaOH solution, and the results show
tive absorption of CO2 in aqueous NaOH in bubble column have that the chemical reaction is the decisive factor affecting the local
been performed by many researches (Darmana et al., 2007; mass transfer. Zhang et al. (2009) coupled a bubble number density
equation in the simulation of the chemisorption process of CO2
⇑ Corresponding author. bubble in NaOH solution, and the results show that the bubble size
E-mail address: l.liu@csu.edu.cn (L. Liu). in the core of the bubble plume is larger than those close to the

https://doi.org/10.1016/j.ces.2020.116218
0009-2509/Ó 2020 Elsevier Ltd. All rights reserved.
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

Nomenclature

Symbols Means k In liquid phase or gas phase


a Volume fraction L In liquid phase
q Density,kg=m3 t Shear-induced turbulence
l Dynamic viscosity,kg=ðm  sÞ
m Kinematic viscosity,m2 =s Subscript
r Surface tension,N=m B Bubble
a Specific surface area of bubble when with subscript,1=m BIT Bubble-induced turbulence
c Molar concentration,mol=m3 G In gas phase
D Diffusion coefficient of species,m2 =s k In liquid phase or gas phase
d Diameter of bubble, m L In liquid phase
E Enhancement factor t Shear-induced turbulence
!
g Gravity constant,m=s2
k Mass transfer coefficient,m=s Superscript
k1 Chemical reaction constant eff Effective
M Interfacial forces,N=m3
_ j jth species
m Mass source due to transfer,kg=ðm3  sÞ T Turbulence
p_ Pressure, Pa  Interfacial equilibrium value
S Mass source of species due to chemical,kg=ðm3  sÞ
CO2 Carbon dioxide
T Temperature, K
!
u Velocity vector,m=s
Dimensionless number
Y Mass fraction
Symbol Name (Expression)
ðq -q Þgd 2
Eo Eötvös number (Eo¼ L rG B )
Subscript pffiffiffiffiffiffiffiffiffiffiffiffi
k Dc
B Bubble Ha Hatta number (Ha¼ 1kl OH )
! !
BIT Bubble-induced turbulence Re Reynolds number (Re¼ ju G -u L jdB
mL )
G In gas phase Sc m
Schmidt number (Sc¼ D)

@  
walls. Gruber et al. (2015) investigated the impact of bubble break- !
ðak qk Þ þ r  ak qk u k ¼ m
_k ð1Þ
age and coalescence (B&C) on the mass transfer, and the results @t
indicated that B&C provide a critical effect for low to moderately
@    
large pH while shrinkage of bubbles dominate for high pH values. ak qk ! !!
u k þr  ak qk u k u k ¼ - ak rpk þr
Darmana et al. (2007) carried out a detailed numerical modeling of @t
h  i
! !
hydrodynamics, mass transfer and chemical reactions in a bubble  ak leff
k r! ! T
u k þðr u k Þ þak qk g þMk ð2Þ
column using a discrete bubble model and compared the numeri-
cal results with their experimental data. The modeling results indi- In two-fluid model, both gas and liquid phases are considered to
cate that the overall mass transfer rate is lower compared to its be continuous, fully interpenetrating continua, coupled by an
experimental counterpart. Solsvik, (2018) evaluated the mass interaction term. The volume fraction of phases is represented by
transfer correlations by means of Euler-Lagrangian method, and ak . The subscript k¼ G; L denotes the gas phase and the liquid
the results indicate that the accuracy of the assessed models are phase respectively. The volume fractions are conservative:
not good. Up to now, several mass transfer models have been pro-
aG þaL ¼ 1 ð3Þ
posed to reproduce the experimental data(Darmana et al., 2007;
Hlawitschka et al., 2017, 2016; Jain et al., 2015; Kraub and Since the gas phase (CO2 ) was absorbed by the liquid phase in
Rzehak, 2017; Buffo et al., 2017). However, the comprehensive bubble column, the balance between the gas phase m _ G and liquid
comparisons and assessments of the typical mass transfer models _ L can be presented as follows:
phase m
have not been reported in the literature.
_ L¼ - m
m _ G ¼-m
_ CO2 ð4Þ
In this paper, a detailed CFD modeling of reactive absorption of
CO2 in aqueous NaOH in a rectangular bubble column has been where the mass source of CO2 m_ CO2
is calculated by the model of
performed with the help of Euler-Euler two-fluid approach. Four !
mass transfer described in Section 2.2. The interfacial force M k is
typical mass transfer rate models, two enhancement factors were
formulated as a sum of the drag, lift, wall, turbulence dispersion
respectively coupled in the two-fluid model to assess their applica-
and virtual mass forces:
bility and accuracy. The chemical reaction rate and the flow field
were obtained from the CFD modeling and then analyzed. Further-
! ! ! ! ! !
MG ¼ Fdrag þFlift þFwall þFdisp þFVM ð5Þ
more, four mass transfer models were assessed and the most suit-
able one was recommended by comparing with the experimental ! !
ML ¼ - MG ð6Þ
data of Darmana et al. (2007).
Expressions of each force and the corresponding closures in Eq.
(5) can be found in Table 1. It should be noted that the surface ten-
2. Numerical modeling
sion coefficient is calculated by the correlation provided by Rzehak
and Krepper (2016) with considering the effects of temperature:
2.1. Two-fluid model

In this paper, the Euler-Euler two-fluid model was employed. r¼ ð118:31-0:158TÞ  10-3 ð7Þ
The continuity and momentum equations are given as follows:
2
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

Table 1
Interfacial forces considered in the present work.

Force Closure referee


! ! ! ! !
Fdrag ¼ - 4d3B CD qL aG j uG - uL jð uG - uL Þ Re   Ishii and Zuber (1979)
CD ¼ maxð 1 þ 0:1Re0:75 ;
24  pffiffiffiffiffiffi 
2 8
min Eo; Þ
3 3
! 8
! ! ! > ½ ð Þ; f ðEo? Þ; Eo? < 4 Tomiyama
Flift ¼ -CL qL aG ð uG - uL Þ  ðr  uL Þ < min 0:288tanh 0:121Re
CL ¼ f ðEo? Þ; 4 < Eo? < 10 et al. (2002)
>
:
-0:27; 10< Eo?
gðqL -qG Þd?
2
Eo? ¼ ;
f ðEo? Þ¼ 0:00105Eo? 3 -0:0159Eo? 2 -0:0204Eo? þ0:474 r
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 0:757
d? ¼dB 1 þ 0:163Eo
! ! ! 2 b CWL ¼CW ðEoÞ Frank et al. (2008)
Fwall ¼ -CWL qL aG j uG  uL j n W
1
max½0;
CWD
1-ðyw =CWC dB Þ
P -1

yw  ðyw =CWC dB Þ
8
>
> 0:47; Eo < 1
>
>
< e-0:933Eoþ0:179 ; 1  Eo  5 CWC ¼ 10; CWD ¼ 6:8;
CW ðEoÞ¼ 0:00599Eo-0:0187;
>
>
> 5 < Eo  33 P  ¼ 1:7
>
:
0:179; 33 < Eo
! ! ! l rT ¼ 0:9
Fdisp ¼ - 34 CD adGB j uG - uL j rL;TT ð a1L þ a1G ÞraG

(Burns et al., 2004)


! ! ! CVM ¼ 0:5
uG - D uL Þ
FVM ¼ -CVM qL aG ð D Dt Dt

Mougin and
Magnaudet (2002)

In Eq. (2), the leff


k is the effective viscosity of gas or liquid phase.
In Eq. (10), Dj is the turbulent diffusion coefficient. Versteeg and
The gas phase is assumed to be laminar due to the low density and van Swaalj (1988) proposed the following correlation to predict Dj
small spatial scales, thus and this formula was also used in Rzehak and Krepper (2016) and
the present work.
leff
G ¼lG ð8Þ
-2119
where lG is the molecular viscosity of CO2 with a value of ¼ 2:35  10-6 expð
CO2
D Þ ð11Þ
T
1.5105 kg=ðm  sÞ. For liquid phase, turbulence is unavoidable.
The liquid turbulence consists of two parts, i.e., shear-induced tur- S_ j is the source term and a detailed definition can be found in
bulence and bubble induced turbulence. The shear-induced turbu- the work of Hlawitschka et al. (2016).
lence is described by the SST k-x model (Menter, 2009). For For each phase and cell, the mass fractions of the species should
bubble-induced turbulence modeling, presently two approaches obey the basic rule as following:
are popular. One is the Sato BIT model (Sato et al. 1981) in which X
Y j¼ 1 ð12Þ
an additional viscosity is used. Another approach is using addi-
tional source terms in the k and e/x equations (Rzehak and Species j leaves the gas phase and is absorbed by the liquid
Krepper, 2013; Krauß and Rzehak, 2018; Fleck and Rzehak, 2019; phase through the interface. The mass of the absorbed gas phase
Liu et al., 2019ab). In the present work, the Sato BIT model is used. is expressed in terms of m_ j . The transfer of the species through
Therefore, the effective viscosity of liquid phase in Eq. (2) can be the interface can be explained by two-film theory. For the trans-
expressed as: port across the phase interface, we consider that the gas-side resis-
leff _ j is given by:
L ¼lL þlL; t þlL;BIT ð9Þ tance to mass transfer is negligible. Thus, m
8  
where lL is the molecular viscosity with a value of 8.899104 > Ek aG;L qL YL -YLj
j
< ; if j¼CO2
kg=ðm  sÞ, lL;t is the shear-induced turbulence viscosity which is _ j¼
m ð13Þ
>
:
solved in the SST k-x model (Menter, 2009) and lL;BIT is the 0 ; if j¼others
bubble-induced turbulence viscosity which equals to
! ! where YL represents the mass fraction of species CO2 at the inter-
0:6qL aG dB j uG - uL j (Sato et al. 1981).
face, which is calculated by Henry constant
2.2. Mass transfer models cLj YLj qL
H j¼ ¼ ð14Þ
cGj
YGj qG
The species transport equations(Hlawitschka et al., 2016) were
employed to describe the mass transportation of each species: where qL is the density of liquid with value of 1000 kg/m3, qG is the
density of gas with value of 1.977 kg/m3, Henry constant of CO2 is
@    
ak qk Y j þr  ak qk Y j !
uk ¼ r given by Versteeg and van Swaalj (1988):
@t
 
_ j þaS_ j 2044
 ak qk Dj rY j þm ð10Þ H CO2
¼ 3:59  10-7 RT expð Þ ð15Þ
T

3
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

Table 2
Coefficients in different mass transfer models.

Mass transfer model a b c Reference Mark

Bird 0.6415 0.5 0.5 (Hlawitschka et al., 2017) B1


Brauer 0.015 0.89 0.7 (Darmana et al., 2007) B2
Henket1 1.25 0.5 0.33 (Jain et al., 2015) H1
Henket2 0.43 0.58 0.33 (Jain et al., 2015) H2

Based on the assumption of spherical bubbles, the specific sur- L=200 mm


face area of a single bubble can be expressed as aG;L ¼ 6adG . The mass
j
transfer rate k can be related to Sherwood number Sh as follows:
j j
j Sh D
k ¼ ð16Þ
d
The Sherwood correlation (Jain et al., 2015) for a single bubble
with forced convection is presented in a general form as below:
j jb jc
Sh ¼ 2þa  Re  Sc ð17Þ
probe
Different values were proposed for the coefficients a; b and c in
different mass transfer models, which can be found in Table 2.
After transferring to liquid phase, the gas reacts with ions. The
chemical reaction process consists of two steps. The reaction step
I is:
H=1000 mm
CO2 þOH- HCO-3 ð18Þ
The reaction step II is:

HCO-3 þOH- CO2-


3 þH2 O ð19Þ
The species which are consumed or generated due to the chem-
ical reaction are considered by S_ j in Eq. (10). A detailed information
of the reaction rate constants can be found in the previous work
z
(Hlawitschka et al., 2016). It should be noted that the reaction of y
CO2 and H2O was not considered in the work of Hlawitschka W =30 mm
x sparger
et al. (2016). The complete model considering all reactions has
been given in the work of Kraub and Rzehak. (2017). In the future
Fig. 1. Sketch of geometry for the experiment from Darmana et al. (2007). The
work, this complete model will be used and the corresponding origin of coordinates is taken in the left front of the column bottom.
results will be compared to the present work to clarify the effects
of the reaction of CO2 and H2O.
Hlawitschka et al. (2016) have validated this chemical reaction Table 3
and proposed a correlation for the enhancement factor. The corre- Boundary conditions.

lation can reproduce the results of Fleischer et al. (1996) which Boundary Boundary condition for gas Boundary condition for liquid
carried out the experiment of the chemisorption of CO2 in NaOH name
solution with same initial pH level as Darmana et al. (2007). The Gas sparger uG;z ¼ 0:14 m=s !
uL ¼ 0
correlation can be formulated as follows: Wall uG;V ¼ 0 !
uL ¼ 0
( Liquid surface p¼ 1:01  105 Pa
!
uL ¼ 0
1; if YOH- < 1:8  10-6
E¼ ð20Þ uG;V mean the velocity along the normal of boundary.
1241:3YOH- þ1:0069; if YOH- 1:8  10-6

where the enhancement factor is a function of the mass fraction of in the first experiment (without considering chemical reaction),
hydroxyl ions, which decreases as the chemical reaction proceeding. and then the gas phase was switched from N2 to CO2 in the second
experiment (considering chemical reaction). In both experiments
3. Simulation and results the gas superficial velocity was set to a constant value of 0.007
ms1 . The measured quantities are the time-dependent pH value,
3.1. Experimental data and numerical setup gas velocity and plume oscilaiton frequency.
All simulations were performed by means of the open-source
As an experimental reference the results of Darmana et al. flow solver OpenFOAM v6 released by OpenFOAM Foundation.
(2007) have been used. In their study a lab scale bubble column The flow domain corresponds to the bubble column used for the
was used. The geometry of the column is sketched in Fig. 1. experiments. The gas sparger was simplified as a 30mm10 mm
The column is filled with liquid up to a level of 1000 mm during (length  width) rectangle. The corresponding computational grid
the experiment. An aqueous NaOH solution with an initial pH of consists of 29975 cells and the time step for each run is set to
12.5 was used as the liquid phase. The gas was introduced into 5104 s. The detailed boundary condition can be found in Table 3.
the column by 21 needles located at the center of the bottom plate The dispersed phase is represented by a mono-dispersal neglecting
with a square pitch of 5 mm. The diameter of the generated bub- the effects of coalescence/break-up and shrinking due to the mass
bles near the needle is 5.5 mm. Nitrogen was used as the gas phase lost by the absorption of CO2 into the liquid. The monodisperse
4
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

size of 5.5 mm was employed in this work. In addition, four mass


transfer rate models and two enhancement factors were taken into
consideration in the simulations. The detailed information of the
selected cases is shown in Table 4.
The detail of mass transfer rate models was shown in Table 2.

3.2. Results and discussions

3.2.1. Flow structures


The instantaneous bubble position observed in the experiment
of Darmana et al. (2007), the gas hold-up, gas velocity, liquid veloc-
ity, and the time-averaged gas hold-up of case H1-EF are presented
in Fig. 3.
The gas hold-up in the simulation indicates that the trace of
bubbles shows a ‘‘S” shape, which is similar to the experimental
results. However, the gradient of gas hold-up near the wall in the
simulation is larger than that in the experiment. As the bubbles
Fig. 2. The pH curve with constant values of 3.5 mm (measured at the location of are absorbed by the liquid, the gas hold-up decreases gradually
the pH measurement point), 4.6 mm (averaged according to the experimental data)
with the height increasing. Vortexes are formed near the bubble,
and 5.5 mm (initial bubble size).
and the vortexes at the top of the column is small due to the
bubble size model and the model considering the bubble shrinking decline of the drag force. In Fig. 3(e), the averaged gas hold-up is
due to chemisorption of CO2 are popular bubble size distribution very low in the vicinity of the wall due to the effect of wall lubri-
models. The former ignores the shrinking of bubble over the height cation force. The gas hold-up at the bottom and top of the column
of the column, which might lead to a wrong interfacial area. How- is more concentrated.
ever, the error could be minimized by employing a suitable con-
stant bubble size model in which a more accurate averaged 3.2.2. Chemical reaction process
interfacial area could be provided. Three simulations were run Fig. 4 shows the history of time-dependent species concentra-
with constant values of 3.5 mm (measured at the location of the tion of case H1-EF. The reaction in the solution changes as the
pH measurement point), 4.6 mm (averaged according to the exper- hydroxyl iron constantly consumed. In first 75 s of the reaction,
imental data) and 5.5 mm (initial bubble size). The results show the concentration of bicarbonate ions increases, and then it turns
that the pH curve obtained from a bubble size of 5.5 mm is closest to decrease, which indicates that the consumption rate of hydroxyl
to the experiments (see Fig. 2). Therefore, the monodisperse bubble ions reduced significantly, and reaction II begins to be reversed.

Table 4
Cases of overall simulation.

Enhancement factor Mass transfer rate models


Bird Brauer Henket1 Henket2
E = 1 (E1) B1-E1 B2-E1 H1-E1 H2-E1
Eq. (20) (EF) B1-EF B2-EF H1-EF H2-EF

G
uG

[m s-1 ] [m s-1 ]
( (

Fig. 3. Instantaneous bubble position at y=W ¼ 0:5, the corresponding instantaneous gas hold-up, velocity and time-averaged gas hold-up.(a) flow structures observed in the
experiment of Darmana, (b) Gas hold-up at t¼ 44 s, (c) Velocity vector of bubble at t¼ 44 s, (d) Velocity vector of liquid at t¼ 44 s, (e) Time-averaged gas hold-up.

5
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

cates that the mass transfer rate predicted by this model is the
cOH- most accurate. The B1 model overestimates while the B2 and H2
0.03
models underestimate the mass transfer rate. Both Darmana
Concentration [mol/m3]

cHCO-3
et al. (2007) and Krauß and Rzehak (2018) employed B2 model
cCO2-3 in their simulations, and also got a overestimated pH curve.
0.02 cCO2 A closer look at H1 model reveals that the numerical results of
cases H1- E1 predict a slightly higher pH value compared to the
experimental data especially during the time interval from 150 s
to 200 s. Similar phenomena can be observed in previous work
0.01 (Hlawitschka et al., 2016, 2017; Krauß and Rzehak, 2018). While
H1-EF cases predict a slightly lower value during the whole period.
Another interesting finding is that the pH values are consistent
with the experimental results at the end of the reaction even
0.00
0 50 100 150 200 250 though the reaction rates are different in the beginning of the reac-
t [s] tion. This is due to that the pH of the solution at the end of the
Fig. 4. Concentration history of each species involved in the chemisorption process
reaction depends on the equilibrium constant of the chemical reac-
of case H1-EF at x=W ¼ 0:5, y=D ¼ 0:5; z=H ¼ 0:98. tion while independents on the mass transfer rate model. There-
fore, the equilibrium constant of the chemical reaction I was
validated indirectly by the consistency of the simulation results
In addition, there is another noticeable change of the pH curve and experimental results at the end of the reaction.
slope when the absorption of carbon dioxide in the solution
reaches a saturation state. This change in the simulation results 3.2.3. Mass transfer rate
occurs at around 200 s, which is agreement with the experiment. In order to quantitatively assess the various mass transfer rates,
The chemical reaction process can be visually reflected by pH the rates of hydroxyl consumption at the initial stage were ana-
curve in the solution. Fig. 5 shows the history of the time- lyzed, which are proportional to the rate of mass transfer of carbon
dependent pH value at point x/W = 0.5, y/D = 0.5, z/H = 0.98. The dioxide at the beginning of the reaction (Darmana et al., 2007). The
carbon dioxide enters the solution through the phase interface, fol- consumption rate of hydroxyl in the beginning of reaction was
lowed by the reaction with the hydroxyl. The reaction rate varies at obtained by fitting the time-dependent hydroxyl curve at the ini-
different transfer rates. In Fig. 5, it is obvious that there are signif- tial stage (0–50 s for B1 model, 0–70 s for others). In Table 5, the
icant differences in the simulation results using different mass hydroxyl consumption rates predicted by different models were
transfer rate models. This indicates that the mass transfer rate summarized, in which the predicted results of both cases H1-EF
model is the main factor which can dramatically influence the and H1-E1 agree with the measurements quite well.
accurate prediction of the chemical reaction process.
The simulation results predicted by mass transfer rate model 3.2.4. Time-averaged velocity profiles
H1 always agree well with the experimental results, which indi- Fig. 6 shows the simulation and experimental results of the
time-averaged gas vertical velocity at z/H = 0.75, y/W = 0.5. The

13
0.35
12
0.30
Velocity of gas [m/s]

11
0.25
pH [-]

10
0.20
9
0.15 Exp.Darmana
Exp.Darmana
8 B1-EF B1-E1 0.10 B1-EF B1-E1
B2-EF B2-E1 B2-EF B2-E1
7 H1-EF H1-E1
H2-EF H2-E1
0.05 H1-EF H1-E1
H2-EF H2-E1
6 0.00
0 50 100 150 200 250 0.0 0.2 0.4 0.6 0.8 1.0
t [s] x\L [-]
Fig. 5. The comparison of time-dependent pH value of simulation results and Fig. 6. The comparison of time-averaged axial component of the gas velocity of
experiment result of Darmana at x/W = 0.5, y/D = 0.5, z/H = 0.98. simulation results and experiment result of Darmana at z/H = 0.75, y/W = 0.5.

Table 5
Hydroxyl consumption rate.

Enhancement factor Mass transfer rateð10-4 mol=ðm3  sÞÞ


Bird Brauer Henket1 Henket2
E = 1(E1) 5.53 2.55 4.18 2.58
Eq. (20) (EF) 5.70 3.09 4.36 3.16

The hydroxyl consumption rate of experiment was 4.2310-4 mol=ðm3  sÞ.

6
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

0.8 0.25
(a) (b) f=0.1 Hz
0.6 0.20

amplitude [m/s]
0.4
uG,y [m/s]

0.15

0.2
0.10
0.0
0.05
-0.2
0.00
0 10 20 30 40 50 0.0 0.1 0.2 0.3 0.4 0.5
t [s] frequency [Hz]

Fig. 7. (a) Time history of the vertical gas velocity at x/L = 0.25, y/W = 0.5, z/H = 0.5 for case H1-EF (b) the corresponding frequency domain for the case H1-EF showing the
meandering frequency.

Table 6
The plume oscillation frequencies of different cases.

Enhancement factor Frequency (Hz)


Bird Brauer Henket1 Henket2
E = 1 (E1) 0.12 0.12 0.12 0.12
Eq. (20) (EF) 0.10 0.12 0.10 0.10

The frequency in the experiment is 0.10 Hz.

gas velocity achieves a maximum value in the middle of the col- fer rate predicted by Henekt1 model was found to be much more
umn and gradually decreases toward both side walls in both exper- accurate than that predicted by other models. The varied enhance-
iments and simulations. Although the numerical simulations can ment factor of Eq. (20) can give more accurate mass transfer
reproduce the overall trend of the experiment, the maximum gas behavior when the Henket1 mass transfer rate model was
vertical velocity predicted by all models is higher than the experi- employed. In addition, an overestimation of pH value within
mental data. 150 s to 200 s may be caused by the underestimation of the reverse
reaction rate of reaction II. The mass transfer model has very little
effect on pH variation at the end of the reaction.
3.2.5. Bubble plume oscillations
The periodic oscillations of the bubble plume were observed in
the experiments due to the centrally distributed sparger. The CRediT authorship contribution statement
plume oscillation frequency was obtained via taking the fast Four-
ier transformation (FFT) of the instantaneous velocity in the work Liu Junda: Formal analysis, Visualization, Investigation, Writing
of Darmana et al. (2007). Similarly, the instantaneous gas velocity - original draft. Zhou Ping: Methodology, Writing - original draft,
and the corresponding FFT result are show in Fig. 7. The values of Supervision. Liu Liu: Conceptualization, Methodology, Writing -
the oscillation frequencies for both experiments and numerical review & editing, Project administration. Chen Si: Software, Valida-
simulations can be found in Table 6. It can be seen that the fre- tion, Resources. Song Yanpo: Data curation. Yan Hongjie:
quencies of cases B1-EF, H1-EF, H2-EF are the same with the exper- Validation.
imental value of 0.1 Hz, and frequencies of other cases are also in
quite good agreement with the experimental results considering
Declaration of Competing Interest
all the simplifications and approximations that have been made
in the modeling. All plume oscillation frequencies predicted by
The authors declare that they have no known competing finan-
the present models are closer to the experimental value than that
cial interests or personal relationships that could have appeared
predicted by the models used in the work of Darmana et al. (2007).
to influence the work reported in this paper.
It can be concluded from the above analysis that, the H1 model
provides a more accurate mass transfer rate than other models. In
addition, the simulation results are more accurate when using the Acknowledgments
EF enhancement model. Therefore, the results of case H1- EF show
a better agreement over those of other cases. This work was supported by the National Natural Science Foun-
dation of China (Grants Nos. 51676211, 51906262) and the Natural
Science Foundation of Hunan Province, China (Grant No.
4. Conclusions 2020JJ5735).

In this paper, four mass transfer rate models and two enhance-
ment factors were considered in order to investigate the mass References
transfer process of a bubble plume rising in the bubble column
Buffo, A., Vanni, M., Marchisio, D.L., 2017. Simulation of a reacting gas–liquid bubbly
with the help of the Euler-Euler two-fluid model. Comparing with flow with CFD and PBM: Validation with experiments. Appl. Math. Model. 44,
the experimental results of Darmana et al. (2007), the mass trans- 43–60.

7
J. Liu, P. Zhou, L. Liu et al. Chemical Engineering Science 230 (2021) 116218

Burns, A.D., Frank, T., Hamill, I., Shi, J.M., 2004. The Favre averaged drag model for Kováts, P., Thévenin, D., Zähringer, K., 2018. Characterizing fluid dynamics in a
turbulence dispersion in Eulerian multi-phase flows. In: Proceedings of the 5th bubble column aimed for the determination of reactive mass transfer. Heat
International Conference on Multiphase Flow, ICMF2004, Yokohama, Japan. Mass Transf. 54 (2), 453–461.
Darmana, D., Henket, R.L.B., Deen, N.G., Kuipers, J.A.M., 2007. Detailed modelling of Krauß, M., Rzehak, R., 2018. Reactive absorption of CO2 in NaOH: An Euler-Euler
hydrodynamics, mass transfer and chemical reactions in a bubble column using simulation study. Chem. Eng. Sci. 181, 199–214.
a discrete bubble model: Chemisorption of CO2 into NaOH solution, numerical Kraub, M., Rzehak, R., 2017. Reactive absorption of CO2 in NaOH: Detailed study of
and experimental study. Chem. Eng. Sci. 62 (9), 2556–2575. enhancement factor models. Chem. Eng. Sci. 166, 193–209.
Fleck, S., Rzehak, R., 2019. Investigation of bubble plume oscillations by Euler-Euler Liu, L. et al., 2019a. A systematic experimental study and dimensionless analysis of
simulation. Chem. Eng. Sci. 207, 853–861. bubble plume oscillations in rectangular bubble columns. Chem. Eng. J. 372,
Fleischer, C., Becker, S., Eigenberger, G., 1996. Detailed modeling of the 352–362.
chemisorption of CO2 into NaOH in a bubble column. Chem. Eng. Sci. 51 (10), Liu, L. et al., 2019b. Euler-Euler modeling and X-ray measurement of oscillating
1715–1724. bubble chain in liquid metals. Int. J. Multiph. Flow 110, 218–237.
Frank, T., Zwart, P.J., Krepper, E., Prasser, H.M., Lucas, D., 2008. Validation of CFD Menter, F.R., 2009. Review of the shear-stress transport turbulence model
models for mono- and polydisperse air–water two-phase flows in pipes. Nucl. experience from an industrial perspective. Int. J. Comput. Fluid Dynam. 23 (4),
Eng. Des. 238, 647–659. 305–316.
Gruber, M.C., Radl, S., Khinast, J.G., 2015. Rigorous modeling of CO2 absorption and Mougin, G., Magnaudet, J., 2002. The generalized Kirchhoff equations and their
chemisorption: The influence of bubble coalescence and breakage. Chem. Eng. application to the interaction between a rigid body and an arbitrary time-
Sci. 137, 188–204. dependent viscous flow. Int. J. Multiph. Flow 28 (11), 1837–1851.
Hlawitschka, M.W., Drefenstedt, S., Bart, H.J., 2016. Local Analysis of CO2 Rzehak, R., Krepper, E., 2013. CFD modeling of bubble-induced turbulence. Int. J.
Chemisorption in a Rectangular Bubble Column Using a Multiphase Euler- Multiph. Flow 55, 138–155.
Euler CFD Code. J. Chem. Eng. Process Technol. 7, 3. Rzehak, R., Krepper, E., 2016. Euler-Euler simulation of mass-transfer in bubbly
Hlawitschka, M.W., Kováts, P., Zähringer, K., Bart, H.J., 2017. Simulation and flows. Chem. Eng. Sci. 155, 459–468.
experimental validation of reactive bubble column reactors. Chem. Eng. Sci. Sato, Y., Sadatomi, M., Sekoguchi, K., 1981. Momentum and heat transfer in
170, 306–319. twophase bubble flow-I. Int. J. Multiph. Flow 7, 167–177.
Ishii, M., Zuber, N., 1979. Drag coefficient and relative velocity in bubbly, droplet or Solsvik, J., 2018. Lagrangian modeling of mass transfer from a single bubble rising in
particulate flows. AIChE J. 25, 843–855. stagnant liquid. Chem. Eng. Sci. 190, 370–383.
Jain, D., Kuipers, J.A.M., Deen, N.G., 2015. Numerical modeling of carbon dioxide Tomiyama, A., Tamai, H., Zun, I., Hosokawa, S., 2002. Transverse migration of single
chemisorption in sodium hydroxide solution in a micro-structured bubble bubbles in simple shear flows. Chem. Eng. Sci. 57, 1849–1858.
column. Chem. Eng. Sci. 137, 685–696. Versteeg, G.F., van Swaalj, W.P.M., 1988. Solubility and Diffusivity of Acid Gases
Jia, H.W., Zhang, P., 2017. Mass transfer of a rising spherical bubble in the (CO2, N2O) in Aqueous Alkanolamine Solutions. Chem. Eng. Data 33, 29–34.
contaminated solution with chemical reaction and volume change. Int. J. Heat Zhang, D., Deen, N.G., Kuipers, J.A.M., 2009. EulerEuler Modeling of Flow, Mass
Mass Transf. 110, 43–57. Transfer, and Chemical Reaction in a Bubble Column. Ind. Eng. Chem. Res. 48
Kováts, P., Thévenin, D., Zähringer, K., 2017. Investigation of Mass Transfer and (1), 47–57.
Hydrodynamics in a Model Bubble Column. Chem. Eng. Technol. 40 (8), 1434–
1444.

You might also like