You are on page 1of 17

SINGLE BUBBLES IN STAGNANT LIQUIDS AND IN LINEAR SHEAR FLOWS

A. Tomiyama

Graduate School of Science & Tech., Kobe University, Rokkodai, Nada, Kobe, 657-8501, Japan
e-mail: tomiyama@mech.kobe-u.ac.jp

1. Introduction

The bubble drag and lift coefficients, CD and CL, are of great importance in the
numerical prediction of bubbly flows based on multi-fluid models [1], [2] and bubble
tracking methods [3]-[7]. Though a number of studies on CD and CL have been carried
out so far [8], [9], our knowledge on them is still insufficient even for single bubbles in
stagnant liquids and in simple shear flows for lack of relevant experimental data such
as the terminal velocities of single bubbles in various stagnant liquids and the lateral
migration of single bubbles in viscous linear shear flows. In my laboratory, several
fundamental experiments on single bubbles have therefore been conducted, such as
measurements of (1) shapes and motions of single bubbles in various infinite stagnant
liquids [10]-[12], (2) shapes and motions of single bubbles in various conduits [13]-[15],
(3) lateral migration of single bubbles in linear shear flows of various liquids [12], [16],
and (4) lateral migration of single bubbles due to the presence of wall [17]. Theoretical
and numerical works to develop reliable closure relations have been also carried out
along with the experimental works [10], [12], [15]. This report reviews recent studies
on single bubbles carried out in my laboratory. In section 2, the necessary and
sufficient condition of dynamical similarity for a single bubble will be deduced from the
local instantaneous field equations to make clear the forces acting on a bubble and
relevant dimensionless groups for modeling the terminal velocity and drag coefficient.
Then a theoretical terminal velocity model for oblate and prolate spheroidal bubbles
will be deduced from the local instantaneous equations and jump conditions in section
3. The lack of uniqueness in terminal velocity caused by a certain kind of nonlinear
bifurcation of terminal condition will be discussed in section 4. Then the effects of fluid
properties, bubble sizes and shear rates on the lift coefficients of single bubbles in
viscous linear shear flows will be discussed in section 5.

2. Dynamical Similarity for Single Bubbles in Stagnant Liquids

All the experimental works and numerical simulations on single-phase flows are
firmly supported by the well-known fact that the dynamical similarity of two
geometrically similar systems does exist if and only if the values of Reynolds numbers
of the two systems are the same. To the contrary, no reliable conditions of dynamical
similarity are established for gas-liquid two-phase bubbly flows. One of the causes for
this shortcoming could be attributed to the presence of various length and time scales
in bubbly flows. At any rate, the lack of similarity conditions definitely forces us to carry
out a mock-up experiment before we start an operation of a plant or a two-phase
system. To overcome this deficit, it would be better for us to review the way in which
the single-phase similarity law has been established. Though it was originally pointed
out by the well-known experiment by O. Reynolds, all of us now know that we can
deduce the condition of dynamical similarity from the continuity and Navier-Stokes
equations. In other words, correct field equations must be the source of similarity law.

3
Let us try to deduce a necessary and sufficient condition of dynamical similarity for
a single isolated bubble with a terminal velocity VT and a sphere-volume equivalent
diameter d from the above-mentioned point of view. Correct field equations must not
involve any empiricism, and thereby only local instantaneous field equations can be
made use of. If we assume that the two phases are incompressible isothermal viscous
fluids, the local instantaneous equations and jump conditions are given by

∂Vk 1 1
∇ ⋅Vk = 0 , +Vk ⋅ ∇VK = − ∇Pk + g + ∇ ⋅ τk (1), (2)
∂t ρk ρk

(VG −V L )⋅ n = 0 , − (PG − PL − κσ)n + (τ G − τ L )⋅ n = 0 (3), (4)

where the subscript k denotes the phase (k = G for the gas phase and k = L for the
liquid phase), V the velocity, t the time, ρ the density, τ the viscous stress tensor, n the
unit normal to the gas-liquid interface, σ the surface tension, and κ is 2H in which H is
the mean curvature of the interface. Since we are concerned about a single bubble
rising through an infinite stagnant liquid, the mass equation Eq. (1) and the mass jump
condition Eq. (3) do not provide any relevant dimensionless groups. Characteristic
velocity, length and time scales for the problem at hand are given by VT, d and T = d/VT,
respectively. Then the magnitude of the viscous stress tensor can be estimated as

[ ]
τ k = µ k ∇V k + (∇Vk ) ≈ µ k
T VT
d
(5)

where µ is the viscosity and the superscript T is the transpose. Hence Eq. (2) yields

PK ≅ Ck 1ρkVT2 + Ck 2ρk gd + Ck 3µ kVT / d (6)

where Cki (i =1, 2, 3) is a constant. Substituting Eqs. (5) and (6) into the momentum
jump condition Eq. (4) yields

(C1ρ L + C 2 ρ G )VT2 d 2 + (C3µ L + C 4 µ G )VT d = C5 σd + C6 ∆ρgd 3 (7)

where ∆ρ = ρL - ρG. For a gas bubble in liquid, we can postulate ρL >> ρG and µL >> µG.
Then Eq. (7) reduces to the following simple force balance:

c1ρ LVT2d 2 + c2µ LVT d = c3σd + c4 ∆ρgd 3 (8)

where ci is a constant. The above equation clearly indicates that the four different
forces, i.e., inertial (Fi), viscous (Fµ), surface tension (Fs) and buoyancy (Fb) forces,
govern the dynamics of single bubble. Several important dimensionless groups can
be easily deduced from the four forces, e.g.,

Fi ρ LVT d
Re = = (Bubble Reynolds number) (9)
Fµ µL

Fi ρ LVT2 d
We = = (Weber number) (10)
FS σ

4
Fb ∆ρgd 2
Eo = = (Eötvös number) (11)
Fs σ

Fi VT
Fr = = (Froude number) (12)
Fb ∆ρgd ρ L

Fµ4 Fb gµ 4L ∆ρ
M= = (Morton number) (13)
Fs3 Fi 2 ρ2L σ3

However we should aware that (i) only three independent dimensionless groups can
be deduced from the four forces and (ii) the constraint given by Eq. (8) decreases the
number of independent dimensionless groups to two. The dynamical similarity for a
single bubble problem will therefore be established if two dimensionless groups, in
which all the four forces are involved, are the same. For example, if we select Re, Eo
and M to develop an empirical correlation of VT, Re should be a function of Eo and M,
i.e. Re = f (Eo, M). The well-known graphical correlation proposed by Grace [18], in
which Re and bubble shape are expressed as a function of Eo and M, is now firmly
supported by the condition of dynamical similarity. It also supports (a) the work of
White & Beardmore [19], in which they correlated VT of a Taylor bubble in the form of
Fr = f (Eo, M), and (b) several attempts [16],[17],[20] to correlate drag, lift and wall
force coefficients as a function of the two dimensionless groups, Re and Eo.
The above example indicates that the application of local instantaneous field
equations and jump conditions to a particular problem of concern is of great use in
setting up a similarity condition, clarifying relevant forces and dimensionless groups,
and establishing a fundamental functional form of correlations as well. For example,
we can readily deduce correct functional forms of VT from Eq. (8) as follows:
(i) In a viscous force dominant regime ( Fi ≅ 0, Fs ≅ 0 ),

∆ρgd 2
VT = C (Stokes or Hadamard-Rybczynski equation [21]) (14)
µL

(ii) In a surface tension force dominant regime ( Fµ ≅ 0 ),

σ ∆ρgd
VT = C1 +C2 (Wave analogy equation [20],[22]) (15)
ρLd ρL

(iii) In an inertia dominant regime ( Fµ ≅ 0, Fs ≅ 0 ),

∆ρgd
VT = C (Davies & Taylor equation [23]) (16)
ρL

We should therefore make the best use of the local instantaneous field equations
and jump conditions not only for interface tracking simulations but also for finding out
the conditions of dynamical similarity and a basis of models or correlations.

5
3. VT Models for Oblate and Prolate Spheroidal Bubbles

Tomiyama et al. [10] proposed a theoretical VT O


A
model for a distorted oblate spheroidal bubble by uL,A
making use of an inviscid momentum jump condition
and a potential flow theory for a flow about an oblate b φΑ
spheroid. As will be shown later, the deduced model a
agrees well with measured data. Good agreement is
not surprising because the model is based on the local
instantaneous equations and jump conditions. Since E=b/a
the VT model given in Ref. [10] is applicable only to
distorted oblate spheroidal bubbles, let us deduce VT
models applicable to various spheroidal bubbles by Fig.1 Spheroidal bubble
following the derivation of VT model in Ref. [10].
When the viscous force is negligible, i.e. when Re >>1, the momentum jump
condition Eq. (3) reduces to the Laplace equation given by

PG = PL + σκ (17)

In the frame of reference moving with a bubble, the bubble nose is a stagnant point.
Hence, by applying the Bernoulli’s theorem to points O and A in Fig.1, we obtain

PL ,O = PL , A + ρ L u L , A − ρ L gdE (1 − cos φ A )
1 2 1
(18)
2 2

where E (= b/a) is the aspect ratio and uL,A the tangential component of the liquid
velocity at point A. On the other hand, the pressure inside the bubble satisfies,

PG ,O = PG , A − ρ G gdE (1 − cos φ A )
1
(19)
2

Note that the above equation assumes that the kinetic energy within the gas phase is
negligible. If the gas density is large, the above equation should be modified so as to
account for the kinetic energy of the gas phase.
The above three equations yield the following energy balance:

ρ L u L , A = σ(κ A − κ O ) + ∆ρgdE (1 − cos φ A )


1 2 1
(20)
2 2

The curvature κ in the above equation is given by

2E 4/3  1 1 
κ=  + 3/ 2 
(21)
d  (cos φ + E sin φ)
2 2 2 1/ 2
(cos φ + E sin φ) 
2 2 2

where φ = 0 at point O and φ = φA at point A as shown in Fig. 1. The liquid velocity uL, A
is readily obtained by making use of a potential flow theory for a flow about an oblate
spheroid (E < 1), a sphere (E = 1) or a prolate spheroid (E > 1) [24] as follows.

6
 (1 − E 2 ) sin 2 φ A 1− E 2
 T
V for E <1
 cos 2 φ A + E 2 sin 2 φ A sin −1 1 − E 2 − E 1 − E 2
3
u L , A =  VT sin φ A for E =1 (22)
2
 ( E 2 − 1) sin 2 φ A E 2 −1
VT cos 2 φ + E 2 sin 2 φ
 A A E E − 1 − tanh
2 −1
1− E 2 / E ( ) for E >1

Substituting Eqs. (21) and (22) into Eq. (20) and taking the limit, φ A → 0 , yield the
following VT models for spheroidal bubbles in surface tension force dominant regime:

 sin −1 1 − E 2 − E 1 − E 2 ∆ρgd E 2/ 3 8σ 4 / 3
 + E for E < 1
 1− E 2
2ρ L 1 − E 2
ρLd
 2 ∆ρgd
VT =  for E = 1 (23)
 3 2ρ L

( )
 E E 2 − 1 − tanh −1 E 2 −1 / E ∆ρgd E 2 / 3

8σ 4 / 3
E for E > 1
 E −1
2
2ρ L E − 1 ρ L d
2

which clearly shows that (1) VT depends not only on d and fluid properties but also on
the aspect ratio E, and (2) the surface tension σ does affect VT only when the bubble is
non-spherical. As shown in Fig. 2 (a), the terminal velocity of an oblate bubble (E < 1)
increases as d approaches to 0, whereas that of a spherical and prolate bubble (E > 1)
decreases with d. This difference is due to the negative sign in the surface tension
force term in Eq. (23). Figure 2 (b) shows the effect of E on VT. Under the condition of
constant bubble diameter, VT decreases with increasing E, in other words, VT
increases as the projected area of a bubble increases. This tendency completely
contradicts with Moore’s theory [25] in which it was concluded that VT increases with E.
As will be proved in the next section, Moore’s drag model fails in predicting the effect
of E on VT, whereas the proposed VT model gives good predictions of VT and the effect
of E on VT as well.

40 40
Air-Water System d[mm]
1
30 30
E
VT [cm/s]

VT [cm/s]

2
0.7
3
20 20
0.8
0.9 4 5
10 10
1.0
1.1 1.2 Air-Water System
00 2 4 6 8 10 0 0.6 0.8 1.0 1.2
d [mm] E

(a) Effect of d on VT (b) Effect of E on VT

Fig. 2 Application of the VT model, Eq. (23), to single air bubbles in stagnant water

7
4. Bifurcation of Terminal Velocity

4.1 VT in a low viscosity system


As shown in Fig. 3, VT of
single air bubbles in water is 50

Terminal velocity VT [cm/s]


apt to widely scatter within a
certain bubble diameter range, 30
about 0.5 < d < 10 mm. It has 20
been taken for granted that the
upper bound of the VT data 10
corresponds to bubbles in a
pure system and the presence
5
of a subtle amount of surface-
active contaminants causes a
drastic increase in drag, which
results in the scatter of VT 0.5 1 2 3 5 10 20 30
Bubble diameter d [mm]
within the hatched region of the
figure. On the other hand, Fig. 3 VT of single air bubbles in water
Tomiyama et al. pointed out
that the primal cause of this scatter is not the presence of surfactants but a slight
difference in the way of bubble release, in other words, the difference in the initial
condition [13],[16]. When a bubble is released from a nozzle with small initial shape
deformation, VT takes a low value (close to the lower bound of the hatched region), its
motion is apt to be either rectilinear or zigzag and its aspect ratio E keeps a high value.
To the contrary, when a bubble is released with large initial shape deformation, VT
becomes higher and widely scattered, the motion is likely to be either helical or
rectilinear and E takes a lower value (more deformed). As a result, even in distilled
water, measured VT widely scatters as shown in Fig. 4. Multiple terminal conditions
observed in the range of 0.5 < d < 5 mm might result from a certain kind of bifurcation
caused by the nonlinearity in the Navier-Stokes equation and a nonlinear link between
0.4
High initial shape
deformation CD =
8 Eo
0.3
Low E, High VT 3 Eo + 4
VT [m/s]

48
0.2 CD =
Re
High E, Low VT
Low initial deformation
0.1
Rectilinear
CD =
16
Re
1+ 0.15Re 0.687 ( ) Helical
Zigzag
0
0.5 1 5
d [mm]
Fig. 4 Measured VT of single air bubbles in distilled stagnant water

8
the gas-liquid interface and the flow field induced by a bubble itself. The types of
motion denoted by three symbols in the figure were determined at about 170 mm
above the nozzle tip. Motion types were also checked at 900 mm above the tip, which
was close to a free surface locating at 1000 mm above the tip. As a result, there were
no changes in the motion type at these two elevations, though a few bubbles exhibited
a transition from zigzag to helical motion as has been observed in many experiments.

0.4
d=0.8mm d=1.1mm d=1.5mm d=2.5mm d=3.5mm d=4.5mm
0.3

0.2

0.1
d=0.75-0.85mm d=1.05-1.15mm
0
d=0.9mm d=1.2mm d=1.8mm d=2.8mm d=3.8mm d=4.8mm
0.3
VT [m/s]

0.2

0.1
d=0.85-0.95mm d=1.15-1.25mm
0
d=1.0mm d=1.3mm d=2.0mm d=3.0mm d=4.0mm d=5.0mm
0.3

0.2

0.1
d=0.95-1.05mm d=1.25-1.35mm
0
0.6 0.8 0.6 0.8 0.4 0.6 0.8 0.4 0.6 0.8 0.4 0.6 0.8 0.4 0.6 0.8 1
Aspect Ratio E

Fig. 5 Measured and calculated VT of single air bubbles in a distilled water:


Equation (23) is used to calculate VT.

As shown in Fig. 5, the proposed VT model for oblate spheroidal bubbles (E < 1)
agrees well with the measured VT. Hence the following drag coefficient, which is
based on Eq. (23), can be recommended at least within the range of 0.7 < d < 5 mm.
2
2 Eo  1− E 2 
C D = 3/ 2   (24)
E (1− E ) Eo +16 E  sin 1− E − E 1− E 
2 4/ 3 −1 2 2

In the case of an air-water system, it has been taken for granted that CD for small
bubbles (d < 1.3 mm) is a function of the bubble Reynolds number Re. For example,
Levich [26] and Moore [27] proposed the following theoretical CD models for spherical
bubbles with Re > 50:

C D = 48 / Re (Levich) (25)

48  2.21 
CD = 1 −  (Moore) (26)
Re  Re 

As can be understood from Fig. 4, these CD models accurately trace the upper bound
of the measured VT for 0.5 < d < 1.3 mm (40 < Re < 480). However the aspect ratios of
bubbles in the upper bound are much smaller than unity, and thereby the assumption

9
of spherical bubble shape adopted in the derivation of Eqs. (25) and (26) does not
hold. Hence the agreement between the measured data and Eq. (25) shown in Fig. 4
is nothing but a coincidence. On the other hand, the fact that the proposed VT model
gives good predictions for 0.7 < d < 5 mm implies that the bubble diameter range, 0.7 <
d < 1.3 mm, is not the viscous force dominant regime but the surface tension force
dominant regime (This might be reasonable because Re in this region is high).
As for the effect of the aspect ratio E, both the proposed VT model and measured
data show the decrease of VT as E increases, whereas the theoretical CD model for
spheroidal bubbles proposed by Moore [25] shows the opposite tendency.

4.2 VT in high viscosity systems


To examine whether or not the 200 mm
bifurcation of VT also takes places under
low bubble Reynolds number conditions, Type A
measurements of shapes and terminal
velocities of single bubbles in viscous
Air
liquids are now being conducted. Figure
6 is a schematic of the experimental 690mm
setup. The dimensions of the acrylic tank Type B
are 690, 200 and 200 mm in height, depth
Bubble
and width, respectively. Glycerol-water
solution and air under room temperature Air
are used for the gas and liquid phases,
respectively. The range of the Morton

Liquid
number M examined is – 4.8 < logM < –
2.8 (Cf., logM = –10.6 for the air-water
system). Single bubbles are released Nozzle Syringe
from a nozzle. Two types of bubble Fig. 6 Schematic of experimental setup
injection methods are adopted to change
the magnitude of initial shape deformation. One is an injection method using an
air-filled syringe which is directly connected to the nozzle (type A), by which large
initial shape deformation is realized. The other is a bubble transport method (type B),
i.e. a measured amount of air is stored in the bubble injection line and is slowly
transported to the nozzle tip by using a liquid-filled syringe. This method enables us to
lower the magnitude of initial shape deformation. Note that low and high initial shape
deformations are also realized by using large and small inside diameter nozzles,
respectively. Hence several nozzles (1.2, 2.4, 3.2, 4.8, 6.0 and 8.5 mm in i. d.) are used
in the experiments. Measured terminal velocities and bubble Reynolds numbers for
logM = –2.8 and – 4.8 are shown in Fig. 7. Terminal velocities calculated by using the
drag coefficient for spherical bubbles in a pure liquid are also plotted. When d is small,
all the measured data scatter above the drag curve, which implies that the fluid
system of the present experiment is not contaminated with surfactants, and the cause
of the VT scatter is not attributed to the presence of surfactants, i.e. if surfactants are
the cause of the scatter, measured terminal velocities should be scattered below the
drag curve, but they are not. Surprisingly enough, the bifurcation of VT takes place
even at low bubble Reynolds number conditions, Re=O(1). As in the case of the
air-water system, bubbles with high initial shape deformation result in low E and high
VT, whereas low initial shape deformation results in high E and low VT. One may still
argue that a subtle amount of surfactants resolved in the glycerol-water solution can
be the cause of the VT scatter in these low bubble Reynolds number conditions.

10
However, as shown in Fig. 8, the bifurcation of VT occurs even with silicon oil, in which
surfactants are to be dissolved, and thereby they cannot be accumulated on bubble
surface (no Marangoni effect). Hence surfactants are not the cause of bifurcation
shown in Figs. 4 and 7. The multiple terminal conditions might result from a certain
kind of bifurcation caused by the nonlinearity in the Navier-Stokes equation and a
nonlinear link between the momentum jump condition and Navier-Stokes equation, i.e.
the link between the bubble interface and bulk fluids. The trigger causing the
bifurcation might relate with the surface free energy σS, where S is the bubble surface
area. When initial shape deformation is high, a bubble is largely stretched in the
vertical direction. As a result, the bubble starts its motion with a surface free energy
larger than that of a bubble with low initial shape deformation. Since this is merely a
speculation, further study on the cause of bifurcation is definitely indispensable.
0.10 0.2
logM= -2.8 logM= -4.8
0.08
VT [m/s]

VT [m/s]
0.06
0.1
0.04

0.02
CD =
16
Re
(
1+ 0.15Re 0.687 )
0 0
0 1 2 3 0 1 2 3
d [mm] d [mm]
(a) Measured terminal velocities

CD =
16
Re
(
1+ 0.15 Re0.687 ) CD =
24
Re
(
1 + 0.15 Re0.687 )
10 20
logM= -2.8 logM= -4.8
8
15
Re Re
6
10
4

5
2

0 0
0 1 2 3 4 0 1 2
Eo Eo
(b) Relation between measured Re and E: The onset of bifurcation is Re = O(1).
Fig. 7 Measured VT and Re of single bubbles in viscous liquids

11
0.15
Nozzle inside diameter
8.5 mm (low initial shape deformation)
1.2 mm (high initial deformation)
0.10
VT [m/s]

Silicon oil: logM = - 0.8

0.05

CD=16/Re (for 0 < Re < 1)

0
0 1 23 4 5 6
d [mm]
Fig. 8 Measured VT of single bubbles in silicon oil (logM = -0.8)

5. Lateral Migration of Single Bubbles in Viscous Linear Shear Flows

The lift force FL acting on a small bubble in a linear shear flow is given by [28]-[30]

πd 3
FLF = −C L ρ L (VG −V L )× rotV L (27)
6

When a bubble is spherical and the viscous force is negligible (Re >> 1), the value of
the lift coefficient CL is equal to 0.5 [29]. Zun [28] and Marie [31] reported that CL for
small bubbles in an air-water system takes a value ranging from 0.25 to 0.3. These
positive values of CL can explain the reason why small bubbles are apt to migrate
toward the pipe wall in a bubbly upflow in a vertical pipe. However the positive lift
coefficients cannot explain the lateral migration of large bubbles toward the pipe
center. On the other hand, Tomiyama et al. [32], [33] carried out interface tracking
simulation of single bubbles in linear shear flows and pointed out that the lift
coefficient can be negative when the Eötvös number Eo is large. They also measured
lateral migrations of single bubbles in viscous linear shear flows to evaluate CL [16],
[34]. As shown in Fig. 9 (a) and (b), measured CL for small bubbles was well
correlated with Re, whereas CL for intermediate and large bubbles was confirmed to
be a function of a modified Eötvös number Eod, which is defined by using the
maximum horizontal dimension of a bubble, dH, as a characteristic length as follows:

g∆ρd H2
Eod = (28)
σ

An empirical correlation of CL based on the measured CL is given by

 min[0.288 tanh (0.121Re ), f (Eod )] for Eod < 4


C L ( Re, Eo) =  (29)
 f (Eod ) for 4 ≤ Eod ≤ 10.7

12
0.4 0.4
logM=-2.8
0.3 0.3 logM=-3.6
logM=-4.2
0.2 0.2 logM=-5.0
CL CL logM=-5.3
0.1 Eq.(29) 0.1 logM=-5.5
logM=-2.8 Eq.(29)
0 0
logM=-3.6
-0.1 logM=-4.2 -0.1
logM=-5.0
-0.2
logM=-5.3 -0.2
logM=-5.5
-0.3 -0.3
0 10 20 30 40 50 60 0 2 4 6 8 10 12
Re Eod
(a) CL of small bubbles (b) CL of large bubbles
Fig. 9 Measured lift coefficients of single bubbles in viscous linear shear flows

where

f (Eod ) = 0.00105 Eod3 − 0.0159 Eod2 − 0.0204 Eod + 0.474 (30)

As can be understood from Fig. 9 (a), the measured CL decreases with Re. On the
other hand, Legendre & Magnaudet [35] carried out a theoretical analysis of forces
acting on a spherical bubble in a low Reynolds number linear shear flow, and clarified
that (a) CL increases with decreasing Re and (b) CL in low Re conditions depends on
the dimensionless shear rate Sr defined by

ωd
Sr = (31)
VR

where ω is the velocity gradient dVL/dy of the linear shear flow (VL: the liquid velocity
component in the vertical direction, y the horizontal coordinate) and VR the relative
velocity between a bubble and liquid. They also performed detailed numerical
simulations of a spherical bubble in linear shear flows [36] and deduced the following
semi-theoretical correlation based on numerical results and theoretical solutions:

2
 6   1 1 + 16 / Re 
2
2.255
C L ( Re, Sr ) =  2  +
3/ 2   (32)
π +   2 1 + 29 / Re 
1/ 2
 ( ReSr ) (1 0.2 Re / Sr )

which indicates that CL for 0 < Re < 5 depends on Sr and steeply increases as Re
approaches to 0, whereas CL for Re > 5 does not depend on Sr and increases with Re.
Since the experimental data shown in Fig. 9 do not cover the low Reynolds
number conditions (Re < 5) of Eq. (32), we recently restarted the measurement of CL
using a new experimental setup shown in Fig. 10. Glycerol-water solution at
atmospheric pressure and room temperature is filled in the acrylic tank, the height,
width and depth of which are 2.5, 0.2 and 0.11 m, respectively. The seamless belt of

13
100 mm in width is rotated with the Controller 110mm
servomotor at a constant speed. A Motor 100mm
viscous linear shear flow with a Belt
constant velocity gradient is thus Syringe
realized in the 30 mm gap between Nozzle
the belt and the sidewall of the Nozzle
tank. Distilled water is used to top view
make the glycerol-water solution to
avoid the effect of surfactants on Test Section

2000mm
lateral migration. Single bubbles
are released from a nozzle. The Guide
nozzle tip is positioned at the Belt
30mm Belt
elevation where the liquid flow 200mm
establishes a linear shear flow. Water Tank
The nozzle tube is orientated as
shown in Fig. 10 (b) so as not to
deteriorate fully-developed linear Bird’s-eye view
shear flows. Trajectories of single (a) (b)
bubbles are measured with a high-
speed video camera (frame rate = Fig. 10 Schematic of experimental apparatus
250 frame/sec, shutter speed =
1/10000 sec,). Then bubble trajectories are calculated by using a one-way bubble
tracking method to evaluate CL. The equation of bubble motion adopted is given by

(ρ G + 0.5ρ L ) dV B = − 3C D ρ L V R V R − C L ρ LV R × ∇ ×V L + ∆ρg (33)


dt 4d

where VB is the bubble velocity and VR is the relative velocity ( = VB – VL ). The value of
CD is adjusted so as to make calculated bubble rise velocities just equal to measured
values. Then a measured value of CL is determined as the value that gives the best fit
to the measured bubble trajectory. The ranges of the Morton number M and liquid
velocity gradient ω are –5.5 < logM < –2.8 and 3.3 < ω < 9.1 [1/sec]. Figure 11 shows
the distributions of VL at 30 mm and 150 mm above the nozzle tip. Even at the lowest
viscosity condition (logM = -5.5), the linear shear flow is well established.

1.0

0.8 z = 30 mm z = 150 mm
VL/Vw

0.6
ω [1/sec] ω [1/sec]
0.4 3.3 3.3
4.7 4.7
0.2 6.2 6.2
9.1 9.1
0.0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1
Dimensionless horizontal position y* (y*=0: belt location, y*=1: wall

Fig. 11 Measured liquid velocity distributions (VW: the speed of belt)

14
1.0 1.0
ω[1/sec] ω[1/sec]
3.3 3.3
0.5 6.2 0.5 6.2
9.1 9.1

CL
CL

0 0

-0.5 -0.5
logM= -2.8 logM= -2.8
1.0 1.0

0.5 0.5
CL

CL
0 0

-0.5 -0.5
logM= -3.2 logM= -3.2
1.0 1.0

0.5 0.5
CL

CL

0 0

-0.5 -0.5
logM= -3.6 logM= -3.6
-1.0 -1.0
0 1 2 3 4 5 6 7 0 5 10 15 20 25 30
Eo Re
(a) Dependence on Eo (b) Dependence on Re
Fig. 12 Measured lift coefficients of single bubbles in viscous linear shear flows

Measured lift coefficients are plotted against Eo and Re in Fig. 12 (a) and (b). This
figure clearly shows the following characteristics:
(1) In the case of the highest viscosity system (logM = –2.8), CL monotonously
decreases with increasing Eo, whereas a slight decrease in the liquid viscosity
results in a complicated dependency on Eo. That is, as Eo increases, CL steeply
decreases from a positive to a negative value and may reach a minimum value,
then CL steeply increases up to about 0.3, then monotonously decreases to
negative values.
(2) When bubbles are large (Eo is greater than about 2), CL is better correlated with
Eo than with Re, in other words, CL for Eo > 2 does not depend on ω if it is
expressed as a function of Eo. This tendency was already confirmed by
Tomiyama et al. [16], [34] and is implemented in the CL correlation, Eq. (29).
(3) When bubbles are small enough to be spherical, CL steeply increases to large
positive values as Re or Eo approaches to 0, and it depends on the shear rate

15
Sr (the scatter of CL in low Re and Eo regions are caused by the difference in Sr
as will be shown below). This indicates that the analytical solution of Legendre
& Magnaudet [35] for spherical bubbles in high viscosity liquids is correct.
Figure 13 is a comparison between the theoretical CL model, Eq. (32), and measured
CL for the low Re number conditions (Re < 6 and logM = –2.8). As shown in Fig. 14, the
bubble aspect ratio E is also measured for several bubbles. When Re is less than
about 2, E is about 0.96 (almost spherical), whereas E gradually decreases with
increasing Re. The measured CL for Re < 2 therefore agrees well with the CL model for
spherical bubbles. In addition, both Eq. (32) and the measured CL show the same
tendency with respect to the effect of Sr on CL, i.e. CL increases with Sr. On the other
hand, the difference between Eq.(32) and the measured data becomes larger as
bubble shape deformation becomes large.

10 1.00
Sr=0.45
Sr=0.18 Eq.(32)
Sr=0.11 0.95
0.4 < Sr<0.5
CL 0.15<Sr<0.2 E
0.09<Sr<0.13
1 0.90

0.85

logM = -2.8 logM = -2.8


0.1 0.80
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Re Re
Fig. 13 CL for low Re number bubbles Fig.14 Aspect ratios of low Re bubbles

1.0 1.0
logM= -2.8 logM= -2.8
logM= -3.2 logM= -3.2
0.5 logM= -3.6 0.5 logM= -3.6
logM= -4.2 logM= -4.2
CL CL
0.0 0.0

-0.5 -0.5
CL=0.08(5.4-Eod)
-1.0 -1.0
0 5 10 0 5 10 15
Eo Eod
(a) CL as a function of Eo (b) CL as a function of Eod
Fig. 15 Effect of Morton number on CL

16
According to the empirical correlation Eq.(29) proposed by Tomiyama et al. [16],[34],
CL for deformed bubbles must be well correlated with the modified Eötvös number Eod,
provided that 4 < Eod < 11. Hence all the measured data are plotted against Eo and Eod
in Fig. 15 (a) and (b). When CL is plotted against Eo (Fig.15 (a)), the scatter of
measured CL is large. However this plot indicates that CL changes its sign from
positive to negative at Eo ≅ 4 , irrespective of the value of Morton number. In the case
of the air-water system, Eo = 4 corresponds to the bubble diameter d = 5.5 mm. This
value coincides with the critical bubble diameter causing the void profile transition
from wall peaking to core peaking in air-water bubbly flows in vertical pipes [37],[38]. It
should be also noted that Eo = 4 corresponds to the value at which the bubble drag
coefficient based on a wave theory [20]

8 Eo
CD = (34)
3 Eo + 4

takes the maximum value, in other words, at which the buoyancy force is just equal to
the surface tension force. As can be understood from Fig.15 (b), CL is better
correlated with Eod within the range of 4 < Eod < 10, and is expressed with a rather
simple correlation given by

C L = 0.08(5.4 − Eod ) (35)

However, even in this Eod range, CL still slightly depends on M, so that further data
analysis is indispensable to develop a reliable correlation.
Since the experiment on CL is still on going in my laboratory, it may take several
months to reach the final conclusions on the lift coefficients of single bubbles in
viscous linear shear flows.

6. Conclusions

Terminal velocities of single bubbles in infinite stagnant liquids and lift coefficients
of single bubbles in viscous linear shear flows have been measured for various fluid
systems. As a result, the following preliminary conclusions are obtained:
(1) In an air-water system, the terminal velocity VT of a bubble larger than 0.7 mm is
well predicted by the theoretical model, Eq. (23), proposed by Tomiyama et al.
(2) The drag coefficient model for spheroidal bubbles proposed by Moore fails in
predicting the effect of bubble aspect ratio E on VT, whereas Eq. (23) gives good
predictions.
(3) The terminal velocity would increase as the aspect ratio decreases, both in
viscous force dominant regime and in surface tension force dominant regime.
(4) The bifurcation of VT occurs even at low bubble Reynolds number conditions, Re =
O(1), the cause of which is still open to question. However the bifurcation is
confirmed to occur whenever the shape depends on initial condition.
(5) Surfactants are not the cause of VT bifurcation.
(6) The lift coefficient CL for spherical bubbles is well predicted by the semi-theoretical
model proposed by Legendre & Magnaudet.
(7) When a bubble is non-spherical, CL is apt to decrease as Re approaches to 0.
(8) The lift coefficient changes its sign at Eo ~ 4.
(9) When Eo is high (Eo > 2), CL is well correlated with a modified Eötvös number.

17
Acknowledgement
The author would like to express his gratitude to Dr. Shigeo Hosokawa, Mr. Satoru
Yoshida, Mr. Kimio Tsunemi, Mr. Yoshihiro Adachi and Mr. Tomotada Maruo for their
assistance in experiments.

References

[1] Tomiyama, A. & Shimada, N., “A Numerical Method for Bubbly Flow Simulation
based on a Multi-Fluid Model”, Trans. ASME, J. of Pressure Vessel Technology,
23, 4, pp.510-516 (2001).
[2] Tomiyama, A. & Shimada, N., “(N+2)-Field Modeling for Bubbly Flow Simulation”,
Computational Fluid Dynamics J., 9, 4, pp.418-426 (2001).
[3] Tomiyama, A., Zun, I., Higaki, H., Makino, Y. & Sakaguchi, T., “A Three-
Dimensional Particle Tracking Method for Bubbly Flow Simulation”, Nuclear Eng.
Des., 175, pp.77-86 (1997).
[4] Tomiyama, A., “Struggle with Computational Bubble Dynamics”, on CD-ROM of
3rd Int. Conf. on Multiphase Flow, Lyon (1998), also in Multiphase Science and
Technology, 10, 4 (1998).
[5] Tomiyama, A., Miyoshi, K., Tamai, H. Zun, I. & Sakaguchi, T., “A Bubble Tracking
Method for the Prediction of Spatial-Evolution of Bubble Flow in a Vertical Pipe”,
on CD-ROM of 3rd Int. Conf. Multiphase Flow, ICMF’98-Lyon, pp.1-8 (1998).
[6] Tomiyama, A., Tamai, H., & Hosokawa, S., “Velocity and Pressure Distributions
around Large Bubbles rising through a Vertical Pipe”, on CD-ROM of 4th Int. Conf.
Multiphase Flow, New Orleans, USA, pp.1-12 (2001).
[7] Zun, I., Kljenak, I. & Moze, S., “Space-Time Evolution of the Nonhomogeneous
Bubble Distribution in Upward Flow”, Int. J. Multiphase Flow, 19, 1, pp.151-172
(1993).
[8] Clift, R., Grace, J. R. & Weber, M. E., “Bubbles, Drops, and Particles”, Academic
Press (1978).
[9] Magnaudet, J. & Eames, I., “The Motion of High-Reynolds Number Bubbles in
Inhomogeneous Flows”, Annual Rev. Fluid Mech., 32, pp.659-708 (2000).
[10] Tomiyama, A., Celata, G. P., Hosokawa, S. & Yoshida, S., “Terminal Velocity of
Single Bubbles in Surface Tension Force Dominant Regime”, Int. J. Multiphase
Flow, 28, 9, pp.1497-1519, (2002), also on CD-ROM of 39th European
Two-Phase Flow Group Meeting, Aveiro, Portugal, F-2, pp.1-8 (2001).
[11] Tomiyama, A., Yoshida, S. & Hosokawa, S., “Surface Tension Force Dominant
Regime of Single Bubbles rising through Stagnant Liquids”, on CD-ROM of 4th
UK-Japan Seminar on Multiphase Flow, Bury St. Edmunds, UK pp.1-6 (2001).
[12] Tomiyama, A., “Reconsideration of Three Fundamental Problems in Modeling
Bubbly Flows”, Proc. JSME-KSME Fluid Eng. Conf. Pre-Symposium, Nagoya,
pp.47-53 (2002).
[13] Tomiyama, A., Nakahara, Y & Morita, G., “Rising Velocities and Shapes of Single
Bubbles in Vertical Pipes”, on CD-ROM of 4th Int. Conf. Multiphase Flow, New
Orleans, Paper No. 492, pp.1-12 (2001).
[14] Tomiyama, A., Nakahara, Y., Adachi, Y. & Hosokawa, S., “Interface Tracking
Simulation of Large Bubbles in Vertical Conduits”, on CD-ROM of 5th
JSME-KSME Fluids Eng. Conf., Nagoya, Japan, 1-6 (2002) to be published.
[15] Tomiyama, A., Adachi, Y., Nakahara, K. & Hosokawa, S., “Shapes and Rising
Velocities of Single Bubbles rising through an Inner Subchannel”, Proc. 3rd
Korea-Japan Symposium on Nuclear Hydraulics and Safety, Kyeongju, Korea,
pp.1-6 (2002), to be published.
[16] Tomiyama, A., Tamai, H., Zun, I. & Hosokawa, S., “Transverse Migration of
Single Bubbles in Simple Shear Flows”, Chem. Eng. Sci. 57, 11, pp.1849-1858

18
(2002).
[17] Hosokawa, S., Tomiyama, A., Misaki, S. & Hamada, T., “Lateral Migration of
Single Bubbles due to the Presence of Wall”, on CD-ROM of ASME Fluid Eng.
Division Summer Meeting, FEDSM2002-31148, pp.1-6 (2002).
[18] Grace, J. R., “Shapes and Velocities of Bubbles Rising in Infinite Liquids”, Trans.
Inst. Chem. Eng., 51, 116 (1973).
[19] White, E. T. & Beardmore, R. H., “The Velocity of Rise of Single Cylindrical Air
Bubbles through Liquids contained in Vertical Tubes”, Chem. Eng. Sci., 17,
pp.351-361 (1962).
[20] Tomiyama, A., Kataoka, I., Zun, I. & Sakaguchi, T., “Drag Coefficients of Single
Bubbles under Normal and Micro Gravity Conditions”, JSME Int. J., Ser. B, 41, 2,
pp.473-479 (1998).
[21] Batchelor, G. K., “An Introduction to Fluid Dynamics”, Cambridge Univ. Press,
(1967).
[22] Mendelson, H. D., “The Prediction of Bubble Terminal Velocities from Wave
Theory”, AIChE J., 13, 2, pp.250-253 (1970).
[23] Daivies, R. M. & Taylor, G. I., “The Mechanics of Large Bubbles rising through
Extended Liquid in Tubes”, Proc. Roy. Soc., Ser. A, 200, pp.375-390 (1950).
[24] Lamb, H., “Hydrodynamics”, 6th Edition, Cambridge Univ. Press (1932).
[25] Moore, D. W., “The velocity of rise of distorted gas bubbles in a liquid of small
viscosity”, J. Fluid Mech., 23, 4, 749-766 (1965).
[26] Levich, V. G., “Physicochemical Hydrodynamics”, New York: Prentice Hall (1962).
[27] Moore, D. W., “The boundary layer on a spherical gas bubble”, J. Fluid Mech., 16,
pp.161-176 (1963).
[28] Zun, I., “Transverse Migration of Bubbles Influenced by Walls in Vertical Bubbly
Flow”, Int. J. Multiphase Flow, 6, pp.583-588 (1980).
[29] Auton, T. R., “The Lift Force on a Spherical Body in a Rotational Flow”, J. Fluid
Mech., 183, pp.199-218 (1987).
[30] Drew, D. & Lahey, R. T., Jr., “The Virtual Mass and Lift Force on a Sphere in a
Rotating and Straining Flow”, Int. J. Multiphase Flow, 13, pp.113-121 (1987).
[31] Lance, M. & Lopez de Bertodano, M., “Phase Distribution Phenomena and Wall
Effects in Bubbly Two-Phase Flows”, Multiphase Science and Technology, 8,
Begell House Inc., pp.69-123 (1994).
[32] Tomiyama, A., Zun, I., Sou, A. & Sakaguchi, T., “Numerical Analysis of Bubble
Motion with the VOF Method”, Nuclear Eng. Des., 141, pp.79-82 (1993).
[33] Tomiyama, A., Sou, A., Zun, I., Kanami, N. & Sakaguchi, T., “Effects of Eötvös
Number and Dimensionless Liquid Volumetric Flux on Lateral Motion of a Bubble
in a Laminar Duct Flow”, Advances in Multiphase Flow, Edited by Serizawa, A.,
Fukano, T. & Bataille, J., Elsevier, pp.3-16 (1995).
[34] Tomiyama, A., Zun, I., Tamai, H., Hosokawa, S. & Okuda, T., “Measurement of
Transverse Migration of Single Bubbles in a Couette Flow”, Proc. 2nd Int.
Symposium on Two-Phase Flow Modelling and Experimentation, 2, pp.941-948
(1999).
[35] Legendre, D. & Magnaudet, J., “A Note on the Lift Force on a Spherical Bubble or
Drop in a Low-Reynolds-Number Shear Flow”, Phys. Fluid, 9, 11, pp.3572-3574
(1997).
[36] Legendre, D. & Magnaudet, J., “The Lift Force on a Spherical bubble in a Viscous
Linear Shear Flow”, J. Fluid Mech., 368, pp.81-126 (1998).
[37] Liu, T. J., “Bubble Size and Entrance Length Effects on Void Development in a
Vertical Channel”, Int. J. Multiphase Flow, 19, pp.99-113 (1993).
[38] Grossetete, C., “Experimental Investigation of Void Profile Development in a
Vertical Cylindrical Pipe”, Advances in Multiphase Flow, Edited by Serizawa, A.,
Fukano, T. & Bataille, J., Elsevier, pp.333-346 (1995).

19

You might also like