Experimental Thermal and Fluid Science 82 (2017) 359–366

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science

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

Effects of column diameter and liquid height on gas holdup in air-water

bubble columns
Shohei Sasaki, Kengo Uchida, Kosuke Hayashi, Akio Tomiyama ⇑
Graduate School of Engineering, Kobe University, 1-1 Rokkodai, Nada, Kobe, Japan

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

Article history: Experiments on the total gas holdup, aG, in air–water cylindrical bubble columns were carried out to
Received 29 June 2016 investigate effects of the column diameter, DH, and the initial liquid height, H0, on aG. Ranges of DH
Received in revised form 28 September and H0 were 160 6 DH 6 2000 mm and 400 6 H0 6 4000 mm, respectively. The superficial gas velocity,
2016
JG, was varied from 0.025 to 0.35 m/s. The characteristics of gas holdup showed that all the flows in
Accepted 27 November 2016
Available online 29 November 2016
the present experiments were pure heterogeneous. The following conclusions were obtained for aG in
air–water bubble columns: (1) the effects of DH and H0 on aG are negligible when scaling up from small
to large bubble columns, provided that aG in the small columns are obtained for DH P 200 mm and
Keywords:
Bubble column
H0 J 2200 mm. The height-to-diameter ratio is useless in evaluation of the critical height, above which
Gas holdup aG does not depend on H0, (2) for the above ranges of DH and H0, Akita-Yoshida’s and Koide’s correlations
Column diameter can give good evaluations of aG for a wide range of JG by tuning the model constants, (3) for DH < 200 mm,
Initial liquid height the decrease in DH increases the population of large bubbles, which results in the decrease in aG, and (4)
Froude number for H0 [ 2200 mm and DH P 200 mm, aG at a constant JG decreases with increasing H0 and approaches an
asymptotic value, and the Froude number using JG and H0 as the characteristic scales well correlates aG in
this regime.
Ó 2016 Elsevier Inc. All rights reserved.

1. Introduction Wilkinson et al. [7] investigated the effects of DH on aG at sev-

Bubble column reactors have been widely used in chemical, bio-
chemical and metallurgical industries and so on [1,2]. The total gas
holdup, aG, of a bubble column is basic information required in col-
umn design, scale-up and the optimization of operating conditions.
There are various parameters affecting aG such as the superficial
gas velocity JG, column geometries, fluid properties, the types of
gas spargers and so on. Many studies on the effects of these param-
eters have therefore been carried out [3–15]. Among the geometric
parameters, the column diameter, DH, and the liquid height in a
column are of great importance in scale-up [7]. When dealing with
the effects of the liquid height on aG, the liquid height, HC, in oper-
ation (aeration height) or the initial liquid height, H0, before start-
ing aeration has often been used. It is known that DH and H0 affect
aG when they are less than certain critical values, whereas aG is
independent of DH and H0 at larger values [3–7,11]. Hence the
knowledge on aG in lab-scale bubble columns obtained for DH
and H0 larger than the critical values would be useful in designing
pilot- and industrial-scale columns [7].
⇑ Corresponding author.
E-mail address: tomiyama@mech.kobe-u.ac.jp (A. Tomiyama).
eral DH and system pressures. Comparisons between aG at
DH = 150 and 230 mm showed that DH has little influence on aG
for the pressure ranging from 0.1 to 0.62 MPa. They also compared
their data at DH = 150 mm with those at DH = 50 mm [16] and
pointed out that the latter are much larger than the former due
to the presence of wall affecting the flow structure. Many studies
after Wilkinson et al. [7] assumed that aG is independent of DH,
provided that DH P 150 mm. However, in these comparisons
between aG at different DH, H0 were not the same. Lemoine et al.
[11] investigated effects of DH on aG in alumina–loaded slurry bub-
ble columns by using a neural-network-based aG correlation devel-
oped by Behkish et al. [17,18]. The correlation indicated that DH
affects aG even for DH > 150 mm and DH is required to be
>700 mm to make aG independent of DH. Leonard et al. [2] sup-
ported this criterion in their recent review paper for bubble col-
umn reactor technology. The effects of H0 were however not
accounted for in the neural-network-based correlation as a param-
eter in the input layer of the neural network. The uncertainty in the
predicted aG due to the neglect of the H0 effect is therefore not
clear. Although Lemoine et al. [11] pointed out that the aG data
obtained by Vandu and Krishna [19] showed that aG decreased
with increasing DH up to DH = 630 mm, H0 also increased with
increasing DH in their experimental condition.

http://dx.doi.org/10.1016/j.expthermflusci.2016.11.032
0894-1777/Ó 2016 Elsevier Inc. All rights reserved.
360 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366

Thus the DH effect on aG must be investigated while keeping H0 Column diameter DH

constant. Koide et al. [20] measured aG in a large-scale air–water
bubble column of DH = 5500 mm and compared aG with those in Cylindrical
bubble column
columns of DH from 100 to 600 mm. In spite of some scatter in
the aG data, aG in the large DH column were similar to those in
the smaller columns. Koide et al. [5] also investigated the effects C Air compressor
of DH on aG in air–water bubble columns for 100 mm 6
DH 6 300 mm and at H0 = 1500 mm. They concluded that aG in
the air–water heterogeneous bubbly flows is independent of DH. D Air dryer

Column height
The data however clearly show some DH effect for DH less than
218 mm.

Initial liquid height H0

In contrast to the studies on DH, there are only a few studies on
the H0 effect on aG [7,10,12,14,15]. As is well known, the increase
R Regulator

in H0 decreases aG at H0 smaller than a certain critical value.

Several criteria for the critical H0 have been proposed in the dimen-
sionless form as H0/DH, e.g. H0/DH = 4 [21], 5 [7] and 7 [5]. Most of Control valve
Diffuser
studies on bubble columns therefore carried out experiments for plate
DH > 150 mm and H0/DH > 5. The critical H0 is 10,000 mm for the
Wilkinson criterion when DH = 2000 mm. It has not been investi- T Thermometer z
gated using large bubble columns whether or not H0 affects aG x
up to that large value of H0. The H0/DH might be an inappropriate Flow Pressure
meter gauge
indicator for representing the effects of H0 at least for Air
DH P 150 mm since DH should be excluded in consideration of F P Chamber
the dynamic similarity if DH plays no role in aG.
Experiments on aG of air–water heterogeneous bubbly flows in
Fig. 1. Experimental setup.
a column of DH = 200 mm at various H0 up to 1000 mm were car-
ried out in our previous study [14]. The parameters in the experi-
ments were JG and H0 only, and the aG data were well correlated in
terms of the Froude number defined by using JG and H0 as the char- Circular hole ph = 25mm
acteristic scales and an aG correlation in terms of the Froude num- dh = 1.4 mm
ph
ber was proposed. However the applicability of the Froude number
in correlating aG in columns of different sizes has not been ph
examined.
Total gas holdups in air–water bubble columns were measured
in this study to obtain systematic databases of aG at various H0 and
DH, which allowed us to investigate the DH and H0 effects indepen-
dently. The ranges of DH, H0 and JG were 160 6 DH 6 2000 mm,
400 6 H0 6 4000 mm and 0.025 6 JG 6 0.35 m/s, respectively.
Plate thickness:
5.0 mm

2. Experimental Fig. 2. Diffuser plate (Nh = 37).

2.1. Experimental setup

Fig. 1 shows the experimental setup. The cylindrical bubble col-
umns of DH = 160, 200 and 300 mm were used. They were made of
transparent acrylic resin for visualization. The stainless steel dif-
fuser plate of 5 mm thickness was placed at the bottom of the col-
umns. Bubbly flows in bubble columns are often classified into
either homogeneous or heterogeneous flow regimes. The former
tends to be formed at low JG, and the increase in JG makes flows
heterogeneous. Flows in bubble columns having diffuser plates
with gas inlet holes of large diameters, dh, can however be hetero-
geneous even at low JG as reported in literature, e.g. Wilkinson
et al. [7], Zahradník et al. [9] and Ojima et al. [22]. The bubbly flow
is referred to as the pure-heterogeneous flow when the homoge-
neous regime does not appear, and the pure-heterogeneous regime
is realized for dh > 1 mm [9]. The present study focused only on aG
in the pure-heterogeneous regime. Therefore dh = 1.4 mm in the
present experiments. The ratio, rh, of the total hole area to the
cross-sectional area of the column was set at 0.18%. Consequently
the numbers, Nh, of holes were 23, 37 and 83 for DH = 160, 200 and
300 mm, respectively. The holes were located so as to be equidis-
tant from each other as shown in Fig. 2, where the hole pitch, ph,
was 25 mm.
The bubble column was initially filled with tap water at room
temperature (19 ± 1 °C) and atmospheric pressure. The H0 was var-
ied from 400 to 1800 mm. Air supplied from the compressor
(Iwata, RDG-150C) flowed into the column through the air dryer
(Iwata, SLP-1501 EB), the air chamber and the diffuser plate. The
gas volume flow rate was measured using the flowmeters (Nippon
flow cell, NVP-I, FLT-H; Tokyo Keiso, AM-1000, full-scale accuracy
±1.5%). The measured flow rate was converted into the volume
flow rate at the middle height of the liquid level by taking into
account gas expansion due to the decrease in static pressure. The
range of JG was from 0.025 ± 0.001 to 0.35 ± 0.01 m/s, where the
uncertainties were evaluated at 95% confidence.
The physical properties of the gas and liquid phases are as
follows: the liquid density qL = 998 kg/m3, the gas density
qG = 1.2 kg/m3, the liquid viscosity lL = 1.0
103 Pas, the gas vis-
cosity lG = 1.8
105 Pas, and the surface tension r = 0.072 N/m.
Two larger bubble columns of 7000 mm high were also used to
measure aG for larger H0, i.e. up to H0 = 4000 mm. Their DH were
450 and 2000 mm. The spargers for the former and latter were a
plate-type and an arm-type sparger, and (dh, Nh, rh, ph) = (5.0 mm,
152, 1.88%, 22 mm) and (5.0 mm, 372, 0.23%, 10 mm) respectively.
The JG was ranged from 0.057 ± 0.002 to 0.28 ± 0.01 m/s. The water
 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366 361

temperature was 16 ± 1 °C, and therefore, qL = 999 kg/m3, 900

lL = 1.1
103 Pas, r = 0.073 N/m.

2.2. Total gas holdup measurement 800

H(t) [mm]
The total gas holdup in the bubble column was measured by
using the image processing method [14] as shown in Fig. 3. Images 700
of the free surface (Fig. 3(a)) were taken using a high-speed video
camera (IDT, Motion Pro X-3) with the spatial and temporal reso-
lutions of 0.36 mm/pixel and 1/100 s, respectively. They were H(t)
600
Time-averaged H(t)
transformed into binary images (Fig. 3(b)). After noise reduction
(Fig. 3(c)), the instantaneous liquid height, H(x, t), at the horizontal
position x and the time t was detected using a region growing 500
method [23] (Fig. 3(d)). The line-averaged liquid height at t was 0 10 20 30
calculated as t – t1(= T) [s]
Z DH =2
1 Fig. 4. Instantaneous and time-averaged HðtÞ (JG = 0.35 m/s, H0 = 400 mm and
HðtÞ ¼ Hðx; tÞdx ð1Þ DH = 300 mm).
DH DH =2

The total gas holdup was then calculated as

Z t 1 þT
1 HðtÞ  H0
aG ¼ dt
T t1 HðtÞ
where t1 is the time, at which the recording was started, and T the
sampling time. Instantaneous and time-averaged HðtÞ at
JG = 0.35 m/s are shown in Fig. 4, in which T is varied from 0 to
30 s to examine the convergence of the averaged HðtÞ. The instanta-
neous HðtÞ fluctuates largely, whose standard deviation is 32 mm,
whereas the time-averaged HðtÞ at T = 30 s converges to within
±0.5% deviations. Therefore T was set at 30 s in all the measure-
ments to assure the accuracy of mean values. The relative standard
errors in aG were within ±2% at DH = 160, 200 and 300 mm. In the
following, several data of aG at DH = 200 mm were quoted from
Sasaki et al. [14].
ð2Þ
3. Results and discussion
3.1. Effects of column diameter on gas holdup
Fig. 5 shows typical flows in the bubble columns. The bubbly
flows under the present experimental conditions were heteroge-
neous, i.e. the flows were largely fluctuated, bubbles much larger
than those at the gas inlet were formed due to bubble coalescence
and disturbed the flows, and strong recirculation of the liquid flow
was formed, which was clearly observed through remarkable retar-

(a) (b)

Binary image
Free surface

DH

x
(c) (d)

Noise reduction H(x, t)

H(t)

Fig. 3. Image processing for aG measurement (JG = 0.35 m/s, H0 = 400 mm and DH = 300 mm).
362 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366
z[mm]
DH
1400
1200
1000
(= H0)
800
600
400
200
0 Fig. 11 shows aG at various H0, where aG in air–water bubble
(a) DH =160 mm (b) 200 mm (c) 300 mm
Fig. 5. Bubbly flows at H0 = 1000 mm and JG = 0.30 m/s.
dation of rising velocities of small bubbles close to the side walls
[1,14].
The maximum size of observable bubbles increased with
increasing JG. At DH = 160 mm, bubbles of widths, wB, comparable
to DH were observed. Synchronized-recording images taken from
the front and side of the column showed that such bubbles occu-
pied most of the column cross section as shown in Fig. 6. Bubbles
of wB >0.8DH (= 160 mm) were also observed at DH = 200 mm. The
widths of the largest bubbles in the column of DH = 300 mm were
not >0.8DH (= 240 mm), but were as large as those in the column
of DH = 200 mm. Bubbles of such sizes were referred to as the EL
(extremely large) bubbles in the following. The EL bubbles were
formed at z  400 mm, where z is the elevation from the bottom
of the column. Even though the maximum stable diameter of a
bubble in stagnant water is about 5 cm [24], the EL bubbles were
much larger than this size and they kept their sizes until they
reached the free surface. The side wall may play a role in the for-
mation of large bubbles [1], i.e. bubbles tend to accumulate in
the upward-liquid flow region (center region) due to the presence
of side wall and frequent bubble coalescence in that region main-
tains the bubble size against bubble breakup due to Rayleigh-
Taylor instability. The bubble frequency, fB, of large bubbles, whose
widths were >100 mm (Fig. 7), were evaluated from front images
for 900 6 z 6 1200 mm by using a high-speed video camera (Fast-
cam SA-X2, Photron Ltd.). The time-averaged fB measured using
front images and that using side images were confirmed to be
the same, provided that the sampling time was sufficiently long.
In the present measurements, the sampling time of 60 s gave the
correct time-averaged values. Fig. 8 shows fB at DH = 160, 200
and 300 mm. The trends of fB are the same, i.e. fB increases with
increasing JG. The fB at DH = 200 and 300 mm are almost the same
whereas fB at DH = 160 mm is larger than those at DH = 200 and
300 mm except at JG = 0.025 m/s, at which all fB are close to zero.
Hence the increase in DH mitigates the formation of the large bub-
bles for DH < 200 mm, whereas further increase in DH does not
affect fB at least for DH P 200 mm.
Total gas holdups at H0 = 400 and 1000 mm are shown in Figs. 9
and 10, respectively. The relation between aG and JG is the same as
that in the so-called pure heterogeneous regime, i.e. aG monoto-
nously increases with increasing JG and the increase rate gradually
decreases [12,25]. The comparisons between aG at the three DH
show that aG depends on DH up to a certain DH between 160 and
200 mm, whereas it is independent of DH at least for DH P 200 mm.
The cause of this trend of aG can be understood as follows. Increas-
ing DH from 160 to 200 mm increases fB. The increase in the popu-
lation of the fast large bubbles decreases aG. On the other hand, the
population of the large bubbles does not change even with increase
in DH for DH P 200 mm, and therefore, similar flow structures hav-
ing the same aG are formed at DH = 200 and 300 mm. The agree-
ment of aG of different DH at JG = 0.025 m/s can also be attributed
to the same fB.
Even though various values were proposed for the critical col-
umn diameter, e.g. DH = 150 mm [7] and 700 mm [2,11], aG of
heterogeneous flows in air–water bubble columns do not depend
on DH at least for DH P 200 mm at H0 = 400 and 1000 mm.
3.2. Effects of initial liquid height on gas holdup
columns only for DH P 200 mm are plotted to exclude the DH
effect. The data show that all the flows are classified into pure
heterogeneous flow. The increase in H0 decreases aG at small H0,
whereas aG at large H0 are independent of H0. As mentioned above,
EL bubbles were formed at z  400 mm. The region of z [ 400 mm
was therefore under flow development and bubble coalescence
was dominant. The gas holdup in this region might be larger than
that for z J 400 mm due to relatively smaller bubble sizes. The
ratio of the length of the latter developed region to that of the for-
mer developing region increased with increasing H0 since the
length of the former would not depend on H0. Thus the total gas
holdup approaches the gas holdup in the latter region as H0
increases.
The aG are re-plotted against H0 in Fig. 12, in which several aG
for DH > 200 mm (Table 1) are quoted from Refs. [19,26–28], to
make the H0 effect on aG clearer. The data of aG at H0 larger than
about 2200 mm are independent of H0. The aG at smaller H0
decrease with increasing H0 and approach constant values. The
comparisons of the aG data also reconfirm that the increase in DH
does not affect aG for DH P 200 mm. The present condition of
H0 = 4000 mm and DH = 2000 mm corresponds to H0/DH = 2, which
is much smaller than the critical H0/DH proposed in the
literature, i.e. H0/DH = 4–7. This result makes it clear that
H0/DH is inappropriate as an indicator for the critical liquid
height. It should also be noted that geometrical configuration of
spargers has no influence on aG within the range of the data
collected.
 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366 363

1200
DH = 160 mm

145 mm

150 mm
z [mm]

900
(a) front view (b) side view
Fig. 6. Extremely large bubble at JG = 0.35 m/s and H0 = 1000 mm (z is the vertical distance from the column bottom).

1200
DH = 160 mm
z [mm]

900
(a) 0.10m/s (b) 0.35 m/s (c) 0.35m/s
Fig. 7. Bubbles much larger than those at inlet (H0 = 1000 mm).

3.3. Applicability of available aG correlations
The dependencies of aG of the air–water heterogeneous bubbly
flows on the relevant parameters discussed above are summarized
as
8 air–water and an air–slurry bubble columns of DH = 200 mm in
< f D ðDH ; H0 ; J G Þ for DH < 200 mm and H0 K 2200 mm
>
aG ¼ f H ðH0 ; JG Þ for DH  200 mm and H0 K 2200 mm
>
:
f J ðJ G Þ for DH  200 mm and H0 J 2200 mm
ð3Þ
Let us first discuss an expression of fH for DH P 200 mm and
H0 [ 2200 mm. Most of the studies on the height effect focused
on the critical dimensionless liquid height, H0/DH, and available
aG correlations were developed without accounting for the H0
effect. We therefore investigated the H0 effect on aG in an
our previous studies [14,15] and pointed out that aG in these sys-
tems can be correlated in terms of the Froude number defined by
JG
FrH ¼ pffiffiffiffiffiffiffiffi
ffi ð4Þ
gH0
364 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366

5.0 0.50

DH [mm]
4.0 160 0.40
200
300

3.0 0.30
fB [Hz]

G
H0 [m] DH [mm]
2.0 0.20 Present 400 200
Present 600 300
Present 1000 300
1.0 0.10 Besagni & Inzoil [26] 3000 240
Present 3000 2000
Present 3500 2000
Present 4000 2000
0.0 0.00
0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4
JG [m/s] JG [m/s]

Fig. 8. Frequency fB of large bubbles for z = 900–1200 mm (H0 = 1000 mm). Fig. 11. aG plotted against JG for DH P 200 mm at various H0.

0.45 0.50
Present Ref.[26] [27] [19] [28]
JG [m/s] (200)(300)(450)(2000) (240) (305)(630)(800)
0.05
0.10
0.40 0.15
0.30 0.20
G

0.30
DH [mm]
160 JG [m/s]
G

0.15 200 0.20

300 0.15
0.20
0.10

0.05
0.00
0.0 0.1 0.2 0.3 0.4 0.10
JG [m/s]

Fig. 9. Total gas holdup aG at H0 = 400 mm.

0.00
0 1000 2000 3000 4000 5000
0.40 H0 [mm]

Fig. 12. Effects of H0 on aG at various DH (=200–2000 mm) and JG (Values in the

parentheses denote DH [mm]).

DH = 300 mm are well correlated in terms of FrH. The following

empirical correlation was also proposed in our previous study [14]:
G

0.20
" #
DH [mm] C R1
1 Fr H C R2
1 Fr H
160 aG ¼ max ; ð5Þ
200 1 þ C R1 R2
2 Fr H 1 þ C 2 Fr H
300
where C1 and C2 are coefficients. Since the increasing rate of aG with
respect to JG largely changed at JG  0.2 m/s, the use of single values
for C1 and C2 was not appropriate to obtain accurate evaluations.
0.00 Therefore (C1R1, C2R1) = (10.6, 19.9) and (C1R2, C2R2) = (7.7, 11.4) were
0.0 0.1 0.2 0.3 0.4
used for aG at small FrH (Regime 1) and for aG at large FrH (Regime
JG [m/s]
2), respectively. This equation agrees well with the data as shown in
Fig. 10. aG at H0 = 1000 mm. the figure.
Fig. 14 shows aG in air–water bubble columns for DH 6 200 mm.
Here the data at DH = 100 and 150 mm were quoted from Hikita
where g is the acceleration of gravity. The aG data for et al. [30] and Jhawar and Prakash [29], respectively. The H0 for
H0 [ 2200 mm and at DH = 200 and 300 mm are plotted against DH = 100 and 150 mm are 650 and 1450 mm, respectively. The
FrH in Fig. 13. Not only aG at DH = 200 mm but also those at effects of H0 and JG at each DH are well correlated in terms of FrH.
 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366 365

Table 1
Column diameters and gas diffusers in Refs. [19,26–28].

Reference DH [mm] Type of diffuser dh [mm] Nh rh [%]

Vandu and Krishna [19] 630 Spider sparger 2.5 64 0.10
Besagni and Inzoil [26] 240 Spider sparger 2–4 – –
Godbole et al. [27] 305 Diffuser plate 1.7 749 2.33
Guan et al. [28] 800 Diffuser plate 2.5 492 0.48

0.50 where CA = 0.20. Note that DH in the R.H.S vanishes though it is

kept in the three factors to make clear relevant dimensionless
groups, i.e. qLgDH2/r is the Bond number, qL2gDH3/lL2 the inverse
pffiffiffiffiffiffiffiffiffi
0.40 viscosity number and JG = gDH the Froude number. Experimental
data used in developing Eq. (6) were for 150 6 DH 6 600 mm,
DH [mm] H0 [mm] 2000 6 HC 6 3000 mm and 0.003 6 JG 6 0.40 m/s, where HC is the
0.30 400 600 1000 aeration liquid height. Koide et al. [5] proposed the following aG
200
correlation for the range of 140 6 DH 6 300 mm, H0 P 1000 mm,
G

300
1.64
104 6 JGlL/r 6 2.92
102 and 1.69
1011 6 glL4/qLr3 6
0.20 2.84
106:
Eq. (5) (C1, C 2)
 0:918  0:252
(10.6, 19.9)
aG J l g l4L
¼ CK G L ð7Þ
0.10 (7.7,11.4)
ð1  aG Þ4 r qL r3
where CK = 0.277. It should be noted that these correlations are
0.00 based on data including aG for DH < 200 mm.
0.00 0.05 0.10 0.15 0.20 0.25
Fan et al. [31] collected a large number of aG data not only of
FrH gas-liquid two-phase bubble columns but also of gas–slurry bubble
columns and proposed the following aG correlation:
Fig. 13. aG correlated in terms of FrH (0.025 6 JG 6 0.35 m/s).
!0:21M0:0079  0:096M0:011
aG 4:1 J G qG
4
qG
¼ 2:9½coshðM 0:054 Þ ð8Þ
0.50 1  aG rg qL
where M is the Morton number defined by M = lL4(qL  qG)g/qL2r3.
Eq. (5) with
0.40 The aG data cover the ranges of 100 6 DH 6 610 mm, HC/DH > 5,
Decreasing DH 0.05 6 JG 6 0.69 m/s, 668 6 qL 6 2965 kg/m3, 0.29 6 lL 6 30 mPas,
0.019 6 r 6 0.073 N/m and 0.2 6 qG 6 90 kg/m3.
0.30 These correlations are compared with the experimental data for
DH P 200 mm and H0 J 2200 mm in Fig. 15. Though the applica-
G

ble range of Fan’s correlation covers JG up to 0.69 m/s, agreement

0.20 for JG > 0.1 m/s is poor. The Akita-Yoshida and Koide correlations
DH [mm] H0 [mm]
Present 200 400-1000 qualitatively agree with the data, whereas aG are underestimated.
Present 160 400-1000 The underestimation might be because the aG data used in devel-
0.10 Jhawar & Prakash [29]150 1450 oping their correlations include aG for DH < 200 mm, under which
Hikita et al. [30] 100 650 aG decreases with decreasing DH. The correlations can however
be well fitted to the data by tuning CA and CK as shown in the fig-
0.00 ure. Most of the data are to within ±10% errors.
0.00 0.05 0.10 0.15 0.20
FrH
0.50
Fig. 14. aG vs. FrH for DH 6 200 mm. H0 [mm] DH [mm]
Present 3000 2000
Present 3500 2000 Eq. (8)
However the DH effect should be taken into account to correlate all 0.40 Present 4000 2000
the data. Multiplying the factor, / = min[2.61–1.11/(0.6 + 7
106 Godbole et al. [27] 2600 305
 Besagni & Inzoil [26] 3000 240
e11:4DH ), 1] to Eq. (5), where DH⁄ is the ratio of DH to 200 mm, yields Guan et al. [28] 4100(= H) 800
a tentative expression for fD(DH, JG, H0) which gives reasonable 0.30
evaluations of aG as shown in the figure. Although the expression
G

Eq. (7) (C K = 0.391)

of the DH effect multiplier can be improved by obtaining aG data
at various DH for DH 6 200 mm, the improvement for small DH col- 0.20
Eq. (6) (C A = 0.20)
umns, from the practical point of view, would be less important
compared with the importance of taking into account another rel- Eq. (7) (C K = 0.277)
evant parameters such as fluid properties. 0.10
Various empirical correlations for aG independent of DH and H0 Eq. (6) (C A = 0.24)
have been proposed so far. The following equation is the well
known correlation proposed by Akita & Yoshida [4]: 0.00
0.0 0.1 0.2 0.3 0.4
!1=8 !1=12
aG q 2
L gDH q 2
L gDH
3
JG JG [m/s]
¼ CA pffiffiffiffiffiffiffiffi
ffi ð6Þ
ð1  aG Þ4 r l2
L gDH
Fig. 15. Comparison between measured and predicted aG.
366 S. Sasaki et al. / Experimental Thermal and Fluid Science 82 (2017) 359–366
4. Conclusion
Experiments on the total gas holdup, aG, in air–water cylindrical
bubble columns were carried out to investigate effects of the col-
umn diameter, DH, and the initial liquid height, H0, on aG. The DH,
H0 and JG ranged from 160 to 2000 mm, from 400 to 4000 mm
and from 0.025 to 0.35 m/s, respectively. The characteristics of
gas holdup showed that all the flows in the present experiments
were pure heterogeneous. The following conclusions were
obtained for aG in air–water bubble columns:
(1) The effects of DH and H0 on aG are negligible when scaling up
from small to large bubble columns, provided that aG in the
small columns are obtained for DH P 200 mm and
H0 J 2200 mm. The height-to-diameter ratio is useless in
evaluation of the critical height, above which aG does not
depend on H0.
(2) For the above ranges of DH and H0, the Akita-Yoshida and the
Koide correlations can give good evaluations of aG for the
wide range of JG by tuning the model constants.
(3) For DH < 200 mm, the decrease in DH increases the popula-
tion of large bubbles, which results in the decrease in aG.
(4) For H0 [ 2200 mm and DH P 200 mm, aG at a constant JG
decreases with increasing H0 and approaches an asymptotic
value. The Froude number using JG and H0 as the character-
istic scales well correlates aG in this regime.
Acknowledgement
The authors would like to express their thanks to NIPPON STEEL
& SUMIKIN ENGINEERING CO., LTD. for permitting us the use of the
large columns. This work has been supported by the Japan Society
for the promotion of Science (JSPS) (Grants-in-Aid for Scientific
Research (B), No. 15H03920).
References
[1] N. Kantarci, F. Borak, K.O. Ulgen, Bubble column reactors, Process Biochem. 40
(2005) 2263–2283.
[2] C. Leonard, J.H. Ferrasse, O. Boutin, S. Lefevre, A. Viand, Bubble column reactors
for high pressures and high temperatures operation, Chem. Eng. Res. Design
100 (2015) 391–421.
[3] F. Yoshida, K. Akita, Performance of gas bubble columns: volumetric liquid–
phase mass transfer coefficient and gas holdup, AIChE J. 11 (1965) 9–13.
[4] K. Akita, F. Yoshida, Gas holdup and volumetric mass transfer coefficient in
bubble columns, Ind. Eng. Chem. Process Des. Dev. 12 (1973) 76–80.
[5] K. Koide, A. Takazawa, M. Komura, H. Matsunaga, Gas holdup and volumetric
liquid–phase mass transfer coefficient in solid–suspended bubble columns, J.
Chem. Eng. Jpn. 17 (1984) 459–466.
[6] F. Yamashita, Effect of liquid depth, column inclination and baffle plates on gas
holdup in bubble columns, J. Chem. Eng. Jpn 18 (1985) 349–353.
[7] P.M. Wilkinson, A.P. Spek, L.L. van Dierendonck, Design parameters estimation
for scale-up of high-pressure bubble columns, AIChE. J. 38 (1992) 544–554.
[8] R. Krishna, J. Ellenberger, Gas holdup in bubble column reactors operating in
the churn–turbulent flow regime, AIChE J. 42 (1996) 2627–2634.
[9] J. Zahradník, M. Fialová, M. Růžička, J. Drahoš, F. Kaštánek, N.H. Thomas,
Duality of the gas–liquid flow regimes in bubble column reactors, Chem. Eng.
Sci. 52 (1997) 3811–3826.
[10] B.N. Thorat, A.V. Shevade, K.N. Bhilegaonkar, R.H. Aglawe, U. Parasu Veera, S.S.
Thakre, A.B. Pandit, S.B. Swant, J.B. Joshi, Effect of sparger design and height to
diameter ratio on fractional gas hold–up in bubble columns, Trans. Inst. Chem.
Engrs. 76 (1998) 823–834.
[11] R. Lemoine, A. Behkish, L. Sehabiague, Y.J. Heintz, R. Oukaci, B.I. Morsi, An
algorithm for predicting the hydrodynamic and mass transfer parameters in
bubble column and slurry bubble column reactors, Fuel Process. Technol. 89
(2008) 322–343.
[12] M.C. Ruzicka, J. Drahoš, M. Fialová, N.H. Thomas, Effect of bubble column
dimensions on flow regime transition, Chem. Eng. Sci. 56 (2001) 6117–6124.
[13] R. Lau, P.H.V. Lee, T. Chen, Mass transfer studies in shallow bubble column
reactors, Chem. Eng. Process. 62 (2012) 18–25.
[14] S. Sasaki, K. Hayashi, A. Tomiyama, Effects of liquid height on gas holdup in
air–water bubble column, Exp. Therm. Fluid Sci. 72 (2016) 67–74.
[15] S. Sasaki, K. Uchida, K. Hayashi, A. Tomiyama, Effects of particle concentration
and slurry height on gas holdup in a slurry bubble column, J. Chem. Eng. Jpn.
49 (2016) 824–830.
[16] K. Idogawa, K. Ikeda, T. Fukuda, S. Morooka, Effects of gas and liquid properties
on the behavior of bubbles in a column under high pressure, Int. Chem. Eng. 27
(1987) 93–99.
[17] A. Behkish, Hydrodynamic and mass transfer parameters in large-scale slurry
bubble column reactors, PhD Thesis, University of Pittsburgh, 2004.
[18] A. Behkish, R. Lemoine, L. Sehabiague, R. Oukaci, B.I. Morsi, Prediction of the
gas holdup in industrial–scale bubble columns and slurry bubble column
reactors using back–propagation neural networks, Int. J. Chem. Reactor Eng. 3
(2005) 1–37.
[19] C.O. Vandu, R. Krishna, Volumetric mass transfer coefficients in slurry bubble
columns operating in the churn–turbulent flow regime, Chem. Eng. Sci. 59
(2004) 5417–5423.
[20] K. Koide, S. Morooka, K. Ueyama, A. Matsuura, F. Yamashita, S. Iwamoto, Y.
Kato, H. Inoue, M. Shigeta, S. Suzuki, T. Akehata, Behavior of bubbles in large
scale bubble column, J. Chem. Eng. Jpn. 12 (1979) 98–104.
[21] F. Kaštánek, J. Zahradník, J. Kratochvíl, J. Čermák, Modeling of large–scale
bubble column reacters for non–ideal gas liquid systems, Front. Chem. React.
Eng. 1 (1984) 330–344.
[22] S. Ojima, K. Hayashi, S. Hosokawa, A. Tomiyama, Distributions of void fraction
and liquid velocity in air–water bubble column, Int. J. Multiphase Flow 67
(2014) 111–121.
[23] S.A. Hojjatoleslami, J. Kittler, Region growing: a new approach, IEEE Trans.
Image Process. 7 (1998) 1079–1084.
[24] R. Clift, J.R. Grace, M.E. Weber, Bubbles, Drops, and Particles, Academic Press,
1978.
[25] S. Sharaf, M. Zednikova, M.C. Ruzicka, B.J. Azzopardi, Global and local
hydrodynamics of bubble columns–effect of gas distributor, Chem. Eng. J.
288 (2016) 489–504.
[26] G. Besagni, F. Inzoli, Comprehensive experimental investigation of counter–
current bubble column hydrodynamics: holdup, flow regime transition,
bubble size distributions and local flow properties, Chem. Eng. Sci. 146
(2016) 259–290.
[27] S.P. Godbole, S. Joseph, Y.T. Shah, Hydrodynamics and mass transfer in a
bubble column with an organic liquid, Can. J. Chem. Eng. 62 (1984) 440–445.
[28] X. Guan, Y. Gao, Z. Tian, L. Wang, Y. Cheng, X. Li, Hydrodynamics in bubble
columns with pin–fin tube internals, Chem. Eng. Res. Design 102 (2015) 196–
206.
[29] A.K. Jhawar, A. Prakash, Influence of bubble column diameter on local heat
transfer and related hydrodynamics, Chem. Eng. Res. Design 89 (2011) 1996–
2002.
[30] H. Hikita, S. Asai, K. Tanigawa, K. Segawa, M. Kitao, Gas hold-up in bubble
columns, Chem. Eng. J. 20 (1980) 59–67.
[31] L.-S. Fan, G.Q. Yang, D.J. Lee, K. Tsuchiya, X. Luo, Some aspects of high–pressure
phenomena of bubbles in liquids and liquid–solid suspensions, Chem. Eng. Sci.
54 (1999) 4681–4709.