Chemical Engineering Science 57 (2002) 1849–1858
www.elsevier.com/locate/ces
Transverse migration of single bubbles in simple
shear %ows
Akio Tomiyamaa ; ∗, Hidesada Tamaia, Iztok Zunb, Shigeo Hosokawaa
aGraduate School of Science and Technology, Kobe University, Rokkodai, Nada, Kobe 657-8501, Japan
bFaculty of Mechanical Engineering, University of Ljubljana, Askerceva 6, 1000, Ljubljana, Slovenia
Received 24 October 2001; received in revised form 18 February 2002; accepted 20 February 2002
Abstract
Trajectories of single air bubbles in simple shear %ows of glycerol–water solution were measured to evaluate transverse lift force
acting on single bubbles. Experiments were conducted under the conditions of −5:56 log10 M6 − 2:8, 1:396Eo6 5:74 and
06 |dVL=dy|6 8:3 s−1, where M is the Morton number, Eo the E7otv7os number and dVL=dy the velocity gradient of the shear %ow. A
net transverse lift coe8cient CT was evaluated by making use of all the measured trajectories and an equation of bubble motion. It was
con:rmed that CT for small bubbles is a function of the bubble Reynolds number Re, whereas CT for larger bubbles is well correlated
with a modi:ed E7otv7os number Eod which employs the maximum horizontal dimension of a deformed bubble as a characteristic length.
An empirical correlation of CT was therefore summarized as a function of Re and Eod. The critical bubble diameter causing the radial
void pro:le transition from wall peaking to core peaking in an air–water bubbly %ow evaluated by the proposed CT correlation coincided
with available experimental data. ? 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Multiphase %ow; Bubble; Lateral migration; Lift force; Simple shear %ow
1. Introduction
Accurate prediction of developing bubble %ows in verti-
cal pipes cannot be carried out without su8cient knowledge
of a transverse lift force acting on a bubble, the force which
governs the direction of transverse migration of a bubble in
a shear :eld. It has been clari:ed through a number of exper-
iments that the lateral migration strongly depends on bub-
ble size, i.e., small bubbles tend to migrate toward the pipe
wall which causes a wall-peak bubble distribution, whereas
large bubbles tend to migrate toward the pipe center which
results in a core-peak bubble distribution. For an air–water
system at atmospheric pressure and room temperature, bub-
bles ranging from about 1 to 5 mm in sphere-volume equiv-
alent diameter d might correspond to small bubbles, and
bubbles larger than about 5 mm to large bubbles (Liu, 1993;
Grossetete, 1995; Sakaguchi, Ijiri, Tabasaki, & Shakutsui,
1996).
The migration of small bubbles toward the pipe wall
can be explained with the so-called shear-induced lift force
∗ Corresponding author. Fax: +81-78-803-1131.
E-mail address: tomiyama@mech.kobe-u.ac.jp (A. Tomiyama).
model (Zun, 1980; Auton, 1987; Drew and Lahey, 1987)
given by
FLF =−CLF�L �d
3
6
(VG − VL)× rotVL; (1)
where the subscripts G and L denote the gas and liquid
phases, respectively, FLF is the shear-induced lift force, CLF
the lift coe8cient, � the density and V the velocity. Zun
(1980) and Lance and Lopez de Berodano (1994) reported
that CLF for small bubbles in an air–water system takes a
value ranging from 0.25 to 0.3. As for the migration to-
ward the pipe center, Serizawa and Kataoka (1994) surveyed
available experimental data and presumed that the direction
of lateral migration would be governed by complex inter-
action between a bubble wake and a shear :eld about the
bubble. The validity of their presumption was partly con-
:rmed by Tomiyama, Zun, Sou, and Skaguchi (1993) and
Tomiyama, Sou, Zun, Kanami, and Sakaguchi (1995). They
carried out interface tracking simulation of single bubbles in
a Poiseuille %ow and pointed out that the migration toward
the pipe center relates to the presence of a slanted wake be-
hind a deformed bubble, which is apparently caused by the
interaction between the wake and shear :eld. In addition,
0009-2509/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0009 -2509(02)00085 -4
1850 A. Tomiyama et al. / Chemical Engineering Science 57 (2002) 1849–1858
their simulation indicated that another kind of transverse lift
force, FTL, is induced by the complex interaction.
In spite of these previous studies, our knowledge on the
lateral migration is still insu8cient due to the lack of rele-
vant experimental data. Fundamental experiments on lateral
migration such as the measurement of bubble trajectories
in a simple shear %ow conducted by Kariyasaki (1987) are
de:nitely indispensable for improving transverse lift force
models. Trajectories of single air bubbles in simple shear
%ows of viscous liquids were therefore measured in this
study to provide an experimental database and to present
an empirical correlation of a net transverse lift force. The
applicability of the proposed correlation to bubbles in low
liquid viscosity systems was also discussed.
2. Experiments
The Morton number M , which is a property group of the
two phases, the E7otv7os number Eo, which is the ratio of
buoyancy to surface tension forces, and the intensity of a
velocity gradient ! of a simple shear %ow were selected as
the parameters of experiments. They are de:ned by
M =
g(�L − �G)�4L
�2L�3
; (2)
Eo=
g(�L − �G)d2
�
(3)
and
!= |rotVL|=
∣∣∣∣dVLdy
∣∣∣∣ ; (4)
respectively. Here g denotes the acceleration of gravity, �
the viscosity, � the surface tension, VL the liquid velocity
in the vertical (z) direction and y the horizontal coordinate.
Note that in the case of a two-dimensional simple shear
%ow heading in the z direction, the x, y and z components
of the liquid velocity vector VL are given by (0; 0; VL(y)),
and thereby the liquid velocity gradient ! = |dVL=dy| is
equivalent to |rotVL|. These quantities were varied within
the ranges of −5:56 log10M6 − 2:8, 1:396Eo6 5:74
and 06!6 8:3 s−1.
Fig. 1 shows a schematic of the experimental setup.
Glycerol–water solution at atmospheric pressure and room
temperature was :lled in the acrylic tank, the height, width
and depth of which were 0.9, 0.45 and 0:152 m, respec-
tively. The seamless belt, 0:15 m in width, was rotated by
the servomotor at a constant speed, the value of which was
regulated by a controller within the range of 0–0:27 m=s.
The position of the rotating belt was stabilized by the two
guide plates and two pulleys so as to prevent its bending
and %uttering. A simple shear %ow with a constant velocity
gradient ! was thus realized in the 0:03 m gap between the
belt and sidewall of the tank. Distilled water was used for
making the glycerol–water solution to avoid the eJect of
Fig. 1. Schematic of experimental apparatus.
10 20 30
20
40
60
80
100
0
y [mm]
V L
[m
m/
s]
z =100 mm
z = 0 mm
Fig. 2. An example of measured liquid velocity pro:les.
surfactants on lateral migration. A single air bubble was re-
leased from a nozzle made of a brass tube. The nozzle tip
was positioned at the elevation where the liquid %ow estab-
lished a simple shear %ow. Fig. 2 is an example of liquid ve-
locity pro:les measured with a hot :lm probe at the nozzle
location (z = 0 m) and at 0:1 m downstream of the nozzle
location (z=0:1 m). Five diJerent tubes with 0.14, 0.5, 2.0,
3.0 and 4:0 mm in inside diameter were used for the nozzle
to release various bubbles. The liquid density was evaluated
as the ratio of a measured mass of the solution to its volume.
The liquid viscosity and surface tension were measured with
a rotational viscometer and a capillary tube tensiometer, re-
spectively. Bubble shapes and trajectories were recorded us-
ing a high-speed video camera (shutter speed = 1=1000 s,
frame rate=400 frame=s). Enlarged video images were used
A. Tomiyama et al. / Chemical Engineering Science 57 (2002) 1849–1858 1851
Fig. 3. Two examples of consecutive images of single bubbles in a simple
shear %ow: log10M =−5:3; ! = 3:8 s−1.
to evaluate d and its aspect ratio. The uncertainties esti-
mated at 95% con:dence for measured �L; �L; �; d and !
were 0.6%, 2.0%, 3.3%, 0.7% and 2.3%, respectively. The
measurement error of bubble position was±0:3 mm. All the
measured %uid properties are summarized in Table 1 in the
appendix.
Fig. 3 shows two examples of consecutive images of sin-
gle bubbles in a shear %ow. The small bubble in Fig. 3(a)
migrated toward the stationary sidewall, whereas the large
bubble in Fig. 3(b) migrated toward the moving belt. Sim-
ilar images were taken for various combinations of d, !
and M . As a result, a database consisting of 116 trajectories
was obtained. As an example, a dataset for log10M =−5:3
is shown in Fig. 4. The data for the other Morton number
systems are summarized in the appendix. The y∗ and z∗
in the :gure are the dimensionless horizontal and vertical
coordinates, normalized by the gap width D = 30 mm, i.e.,
0
2
4
6
8
y*=0: Belt location
y*=1: Wall location
=0 [s-1]
=3.8
=5.7
ω
ω
ω
ω
=6.2
Eqs. (7)-(9)
d=2.84 mm d=3.52 mm d=4.16 mm
d=4.85 mm d=5.54 mm
0.2 0.4 0.6 0.8 1
0.2 0.4 0.6 0.8
2
4
6
0
z*
0 0.2 0.4 0.6 0.8 1
y*
0
y*
Fig. 4. Measured bubble trajectories for log10M =−5:3.
y∗ = y=D and z∗ = z=D. This :gure clearly shows that (1)
the direction of lateral migration under a constant Morton
number is not aJected by !, but by the bubble diameter d,
in other words, by the E7otv7os number Eo, and (2) the lateral
migration length increases with !. As shown in Figs. 11
and 12 in the appendix, bubbles in the other Morton number
systems exhibited the same tendencies.
3. Evaluation of lift coe�cient under a simple shear
�ow
Bubble trajectories were calculated using a one-way bub-
ble tracking method to evaluate the transverse lift force act-
ing on a bubble. Since the measured lateral migration length
was proportional to ! both for the bubbles migrating toward
the moving belt and for the bubbles migrating toward the
stationary wall, it was assumed in the calculations that FTL
caused by the slanted wake has the same functional form as
that of the shear-induced lift force FLF , that is,
FTL =−CTL�L �d
3
6
(VG − VL)× rotVL; (5)
where CTL is the transverse lift coe8cient due to the slanted
wake. Hence the net transverse lift force FT was assumed
to be given by
FT = FLF+FTL=−(CLF+CTL)�L �d
3
6
(VG−VL)× rotVL
=−CT�L �d
3
6
(VG − VL)× rotVL; (6)
where CT is the net transverse lift coe8cient, i.e. the sum
of shear- and wake-induced lift coe8cient.
The equation of bubble motion was therefore given by
(�G + 0:5�L)
dVG
dt
=− 3CD�L
4d
|VR|VR − CT�LVR × rotVL +
(�L − �G)g; (7)
1852 A. Tomiyama et al. / Chemical Engineering Science 57 (2002) 1849–1858
0 2 4 6 8 10
-0.4
-0.2
0
0.2
0.4
ω [s-1]
C T
d=2.80
d=3.56
d=4.23
d=4.85
d=5.68[mm]
(a) log10M=-5.5
0 2 4 6 8 10
-0.4
-0.2
0
0.2
0.4
ω [s-1]
C T d=3.19
d=3.40
d=4.19
d=4.93
d=5.64[mm]
(b) log10M=-3.6
Fig. 5. EJects of liquid velocity gradient ! on CT : (a) log10M =−5:5,
(b) log10M =−3:6.
where VR is the relative velocity (=VG −VL); g the accel-
eration due to gravity (=(0; 0;−g)) and CD the drag coef-
:cient, which was evaluated by (Tomiyama, Kataoka, Zun,
& Sakaguchi, 1998b)
CD = �max
[
min
{
16
Re
(1 + 0:15Re0:687);
48
Re
}
;
8
3
Eo
Eo+ 4
]
; (8)
where � is a tuning factor to make calculated bubble ve-
locities just equal to measured values, and Re the bubble
Reynolds number de:ned by
Re =
�L|VR|d
�L
: (9)
The value of CT was adjusted so as to yield the best :ts to the
measured bubble trajectories. As shown in the solid curves
in Fig. 4, we could con:rm that all the measured trajecto-
ries were well reproduced with Eq. (7), which implies the
validity of the employed assumption for the functional form
of FT . It should be also noted that the acceleration of bubble
velocity, dVL=dt, was negligibly small for all the measured
bubbles, and thereby, even if we neglected the left-hand side
of Eq. (7), there was no diJerence in the evaluated CT . In
other words, the value of the virtual mass coe8cient, 0.5, in
Eq. (7) had no substantial eJects on the evaluation of CT .
Fig. 5 shows thus evaluated net transverse lift coe8cients
CT for two Morton number systems, (a) log10M = −5:5
0 10 20 30 40 50 60
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
log10M=-2.8
log10M=-3.6
log10M=-4.2
log10M=-5.0
log10M=-5.3
log10M=-5.5
Eq. (11)
Re
C T
Fig. 6. CT for small bubbles.
0 5 10
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
log10M=-2.8
log10M=-3.6
log10M=-4.2
log10M=-5.0
log10M=-5.3
log10M=-5.5
Eq. (11)
Eod
C T
Fig. 7. CT for large bubbles.
and (b) log10M=−3:6. As shown in these :gures, CT takes
positive values for small bubbles, whereas it takes nega-
tive values for large bubbles. In addition, CT for a constant
bubble diameter is not aJected by the liquid velocity gradi-
ent ! and is more or less constant.
As shown in Figs. 6 and 7, we also found that CT for
small bubbles is well correlated with Re, whereas CT for
intermediate and large bubbles is a function of a modi:ed
E7otv7os number Eod, which is de:ned by using the maximum
horizontal dimension of a bubble as a characteristic length
as follows:
Eod =
g(�L − �G)d2H
�
: (10)
The resulting empirical correlation of CT is given by
CT ={
min[0:288tanh(0:121Re); f(Eod)] for Eod¡4;
f(Eod) for 46Eod610:7;
(11)
A. Tomiyama et al. / Chemical Engineering Science 57 (2002) 1849–1858 1853
where
f(Eod) = 0:00105Eo3d − 0:0159Eo2d
−0:0204Eod + 0:474: (12)
Eq. (11) yields 0¡CT 6 0:288 for small bubbles migrating
toward the wall, which coincides with available experimen-
tal data of CT for small bubbles in an air–water system, e.g.,
CT =0:3 by Zun (1980) and CT =0:25 by Lance and Lopez
de Berodano (1994). On the other hand, Eq. (11) gives neg-
ative values for large bubbles, and thereby large bubbles in
a pipe %ow would migrate toward the pipe center due to the
net transverse lift force.
4. Application of CT correlation to an air–water system
The proposed CT correlation is based on the experimen-
tal data obtained in a high-viscosity system, so that in prin-
ciple it is not applicable to a low-viscosity system such
as air–water and vapor–water systems. However, as noted
above, the correlation coincidentally yields the same value
of CT with experimental data for a small bubble in an
air–water system. In view of this coincidence and the lack
of CT correlations for a low-viscosity system, it would be
worth applying the proposed correlation to a bubble in an
air–water system to examine whether or not it can explain
the tendency of bubble lateral migration in a low-viscosity
system.
4.1. Bubble lateral migration in a bubbly up8ow
In an air–water system under atmospheric pressure and
room temperature, the proposed CT correlation yields the
bubble diameter dependency shown in Fig. 8. To evaluate
dH in the de:nition of Eod, we made use of the following
empirical correlation of the aspect ratio E for spheroidal
Wall Regime: 0.4 < d < 5mm
migration toward near wall region
Neutral Regime: 0< d <0.4mm
5< d <6mm
affected by turbulence,
bubble residence time, etc.
Core Regime: 6 mm < d
migration toward pipe center
µL-controlled σ and g -controlled Possible Three Regimes
0 1 2 3 4 5 6 7 8 9 10-0.3
-0.2
-0.1
0
0.1
0.2
0.3
d [mm]
C T
Fig. 8. CT in an air–water system and postulated regimes of lateral
migration.
bubbles in a fully contaminated system (Wellek, Agrawal,
& Skelland, 1966):
E =
dV
dH
=
1
1 + 0:163Eo0:757
; (13)
where dV is the maximum vertical dimension of a
bubble.
For d¡ 4:4 mm, CT is controlled by the bubble Reynolds
number, i.e. by a viscous force. On the other hand, it is gov-
erned by the E7otv7os number for d¿ 4:4 mm. CT changes
its sign at d = 5:8 mm from positive to negative. We can,
therefore, postulate three regimes for the bubble lateral
migration in a bubbly up%ow in a vertical pipe. The :rst
( 16d¡ 5:8 − 2 where k (mm) is a small value) is a
wall regime in which CT takes a large positive value, and as
a result, bubbles would migrate toward the pipe wall. The
second (d¿ 5:8 + 3) is a core regime in which CT takes
a large negative value, and thereby bubbles would migrate
toward the pipe center. The third (5:8 − 26d¡ 5:8 + 3
and 0¡d¡ 1) is a neutral or intermediate regime in
which the bubble lateral migration might be strongly af-
fected by many other factors such as the magnitude of
bulk liquid turbulence and a bubble residence time in a
%ow domain because the net transverse lift force in this
regime keeps a low value due to the low magnitude of
CT .
Although the above-mentioned three regimes are based
on the CT correlation for single bubbles in viscous shear
%ows, they again coincide with available experimental data
on radial void pro:les in air–water turbulent bubbly up-
%ows in vertical pipes. As an example, radial void pro:les
measured by Sakaguchi et al. (1996) are replotted in Fig.
9, in which R (=15:4 mm) is the pipe radius, r the radial
coordinate, 〈J 〉 the area-averaged volumetric %ux, and 〈$G〉
the area-averaged void fraction. They measured the position
and size of each bubble using an image processing method
and classi:ed bubbles into several groups in terms of their
sizes. Then they obtained radial void pro:les $G(r; d) for
each bubble group (A–F or a–f in the :gure) and $G(r) for
all the bubbles (G or g). As shown in the :gure, bubbles less
than 5 mm (groups A–C or a–c) constitute the wall peak-
ing, bubbles of 5–6 mm (group D or d ) correspond to the
intermediate pro:le, and bubbles larger than 6 mm (groups
E and F or e and f) form the core peaking. Liu (1993) also
measured bubble sizes and void pro:les in air–water tur-
bulent bubbly %ows using a vertical pipe of R = 28:6 mm,
and concluded that the critical bubble diameter causing the
void pro:le transition from wall peaking to core peaking
is about 5–6 mm. Though not to mention all, many ex-
perimental data have indicated that the pro:le transition in
air–water bubbly %ows occurs when bubbles are larger than
about 5 mm (Grossetete, 1995; Zun, 1988). These facts
imply that the net transverse lift coe8cient in an air–water
system is not so much diJerent from the proposed CT
correlation.
1854 A. Tomiyama et al. / Chemical Engineering Science 57 (2002) 1849–1858
0
1
0
1
0
1
0
1
2
r/R
G
α
(r)
/<α
G
>
d=2-3mm
0
2
4
0
1
2
0.2 0.4 0.6 0.8 1
1
2
0
A(8)
B(122)
C(810)
D(445)
E(76)
F(5)
G(1466)
=0.504m/s
=0.017m/s
<αG>=2.18%
d=3-4mm
d=4-5mm
d=5-6mm
d=6-7mm
d=7-8mm
d=2-8mm
0
1
0
1
0
1
0
1
2
r/R
0
2
4
6
0
1
2
3
0.2 0.4 0.6 0.8 1
1
2
0
a(4)
b(231)
c(895)
d(342)
e(44)
f(2)
g(1518)
=1.01m/s
=0.017m/s
<αG>=1.34%
Fig. 9. Measured radial void pro:les as a function of d; data are quoted
from Sakaguchi et al. (1996).
4.2. Most probable radial position of a bubble in a
turbulent bubbly up8ow
Once we establish a reliable equation for the balance of
lateral forces acting on a bubble, it would be possible to
predict the most probable radial position rmp of a bubble
in an air–water bubbly %ow in a vertical pipe. The lateral
motion of a bubble in a turbulent bubbly %ow would be,
of course, aJected by many factors such as the transverse
lift force, bulk liquid turbulence, wall eJect, non-rectilinear
bubble path intrinsic to deformed bubbles, bubble collision
and so on. However, for the purpose of evaluating the most
probable radial position, we might be able to neglect the fol-
lowing eJects: bubble collision, %uctuating bubble motion
and bulk liquid turbulence, since these phenomena are more
or less stochastic. In any case, a sort of restraining condi-
tions to bubble lateral migration is indispensable to account
for wall eJects. At this stage of examining whether or not
the CT correlation can explain the tendency of rmp in a tur-
bulent bubbly %ow, we might be able to employ a wall force
model proposed by Tomiyama et al. (1995) as one repre-
sentation of wall eJects. Other possible way to
本文档为【Transverse migration of single bubble这个有升力公式来源】,请使用软件OFFICE或WPS软件打开。作品中的文字与图均可以修改和编辑,
图片更改请在作品中右键图片并更换,文字修改请直接点击文字进行修改,也可以新增和删除文档中的内容。
该文档来自用户分享,如有侵权行为请发邮件ishare@vip.sina.com联系网站客服,我们会及时删除。
[版权声明] 本站所有资料为用户分享产生,若发现您的权利被侵害,请联系客服邮件isharekefu@iask.cn,我们尽快处理。
本作品所展示的图片、画像、字体、音乐的版权可能需版权方额外授权,请谨慎使用。
网站提供的党政主题相关内容(国旗、国徽、党徽..)目的在于配合国家政策宣传,仅限个人学习分享使用,禁止用于任何广告和商用目的。