International Journal of Heat and Fluid Flow 54 (2015) 211219

Contents lists available at ScienceDirect

International Journal of Heat and Fluid Flow

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

Oscillation and breakup of a bubble under forced vibration

Mohammad Movassat, Nasser Ashgriz , Markus Bussmann
Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON M5S 3G8, Canada

a r t i c l e

i n f o

Article history:
Received 14 August 2014
Received in revised form 5 March 2015
Accepted 24 May 2015

Keywords:
Drops and bubbles
Bubble oscillations
Chaotic oscillation
Microgravity ows

a b s t r a c t
Coupled shape oscillations and translational motion of an incompressible gas bubble in a vibrating liquid
container is studied numerically. The bubble oscillation characteristics are mapped based on the bubble
Bond number (Bo) and the ratio of the vibration amplitude of the container to the bubble diameter (A/D).
At small Bo and A/D, the bubble oscillation is found to be linear with small amplitudes, and at large Bo and
A/D, it is nonlinear and chaotic. This chaotic bubble oscillation is similar to those observed in two coupled
nonlinear systems, here being the gas inside the bubble and its surrounding liquid. Further increases in
the forcing, results in the bubble breakup due to large liquid inertia.
2015 Published by Elsevier Inc.

1. Introduction
Oscillations and translational motion of gas bubbles in a host
uid under forced vibrations are encountered in many two phase
ow applications. For instance, in mixers and reactors (Ni and
Gao, 1996; Krishna and Ellenberger, 2002; Knopf et al., 2006),
and in thermal management systems in microgravity (Gaul et al.,
2010; Zhang et al., 2009; Weislogel et al., 2009). In most such
devices, bubbles enhance the heat and mass transfer coefcients
and reaction rates in the mixture.
Depending on the imposed amplitude and frequency of the
vibrations, bubble oscillations range from small amplitude and linear to large amplitude and nonlinear oscillations. Further increase
of the amplitude and frequency of the vibration results in bubble
breakup.
Experimental study of the bubble oscillation under forced vibration requires a reduced gravity condition to trap the bubble and
study its motion. As a result, there are very limited studies using
forced vibration. The available studies have shown that the bubble
undergoes an oscillatory translational motion at the same frequency as the forcing (Ishikawa et al., 1994; Friesen et al., 2002;
Farris et al., 2004). Vibration amplitude and frequency were kept
small in these studies resulting in a small deviation of the bubble
from the spherical shape. Since the oscillation frequencies were
orders of magnitude lower than the Minnaert frequency (Devaud
et al., 2008), no volume oscillations were considered.
Several studies have investigated bubble oscillation by acoustic
levitation, where an acoustic force was used to trap a bubble and
Corresponding author.
E-mail address: Ashgriz@mie.utoronto.ca (N. Ashgriz).
http://dx.doi.org/10.1016/j.ijheatuidow.2015.05.014
0142-727X/ 2015 Published by Elsevier Inc.

modulate its oscillation. Eller and Crum (1970) found a threshold

for the beginning of large amplitude shape oscillations of a gas
bubble as a function of the bubble radius. Their results showed that
the pressure threshold for the beginning of shape oscillation
decreases with increasing bubble radius, since surface tension
force decreases as the bubble size increases. In the case of large
amplitude oscillations, it was shown that the interaction among
volume oscillations, shape oscillations, and translational motion
results in a chaotic bubble response (Akhatov and Konovalova,
2005; Watanabe and Kukita, 1993). Chaotic response is because
of the coupling of large amplitude and nonlinear shape and volume
oscillations, and the translational motion. Due to the coupling,
excitation of any of these motions, e.g. the translational motion,
can excite the rest (Doinikov, 2004; Mei and Zhou, 1991;
Benjamin and Ellis, 1990).
If the acoustic forcing is strong enough, and if the shape of the
bubble is distorted from the spherical shape due to the asymmetries in the ow eld, a liquid jet can form which penetrates into
the bubble and pierces it (Crum, 1979). The impact of this high
velocity liquid jet is the dominant mechanism leading to cavitation
damage. Bubble breakup has also been studied in the context of
sonoluminescence, where bubbles violently collapse. Calvisi et al.
(2007) and Blake et al. (1999) showed that in the case of surface
instabilities which can be induced by rigid boundaries, neighboring
bubbles, ow disturbances, and asymmetric ows around a bubble,
a breakup liquid jet can form within the bubble core as the bubble
collapses. In a recent study by Yoshikawa et al. (2010) large amplitude shape oscillations of a bubble in response to forced vibration
was studied experimentally in a parabolic ight. Results suggested
that as the forcing increases, bubble undergoes large amplitude
shape oscillations and breakup.

212

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

In a previous work, the authors studied regular and chaotic

response of a bubble to the forced vibrations in 2D (Movassat
et al., 2012). In the present study, the dynamics of a single bubble
in response to forced vibration is studied using a 3D numerical
scheme. As mentioned, available studies on the bubble oscillation
under forced vibration are limited to small amplitude oscillations
where the bubble shape remains spherical. The focus of the present
study is to understand the coupling between shape oscillations and
translational motion and characterize the nonlinear large amplitude and chaotic bubble motion, and bubble breakup in frequencies which are orders of magnitude lower than those in acoustic
levitation. Since the imposed frequency in this work is lower than
Minnaert frequency and also because isothermal and adiabatic
conditions are assumed, no volume oscillation is considered and
the bubble response is limited to the shape oscillations and the
translational motion.

with initial condition of, d(x, 0) = u(x), where sign(u) = 1 for u < 0,
signu 1 for u > 0, and signu 0 for u 0. In Eq. (7), d is the
distance function representing the interface between the two uids
and e is the integration variable representing time step. The major
drawback of Level Set method has been its ability to conserve mass.
In TransAT, this error is minimized using a correction factor in the
solver settings. This correction factor is applied locally, for each cell,
and globally, for the whole domain, to assure that the mass is conserved from one time step to the next. Comparing to other interface
capturing methods such as Volume of Fluid, Level Set provides a
sharper property change across the interface.
n, where r is
The surface tension force is modeled as ~
F ST rjd~
the coefcient of surface tension, j is the interface curvature, and d
is the Dirac delta function, which is dened as, du dHe =du.
Unit normal vectors and Curvature are calculated based on the
level set function as,

2. Numerical model

~
n
The equations governing the motion of an incompressible bubble in a liquid domain are the mass and momentum conservation
equations,

ru 
jruju0

jr

V 0
r~

@~
V ~
1
1
1
2
V  rp r  s ~
F A2pf cos2pft
V  r~
@t
q
q
q SF

where ~
V is the velocity vector, p is the uid pressure, q is the uid
density, s is the shear stress tensor, ~
F SF represents the surface tension force per unit volume, which is applied on the interface
between two uids. The last term on the right hand side of Eq. (2)
is the imposed oscillation force as a result of the forced vibrations,
in which A and f are the amplitude and frequency of the vibration,
respectively. Since the uids are Newtonian,
T

V r~
V
s lr~

where l is the uid dynamic viscosity.

The TransAT software was used for the simulations. The simulations were run on 32 processors on SciNet clusters at University of
Toronto. The Level Set (LS) method was used to capture the interface between the two uids. If the interface is dened by C, a function, u, is dened as u > 0 in the liquid, u < 0 in the gas, and u = 0
on the interface, C. Since the interface moves with the uids, u
must be advected by the following equation,

@u ~
V  ru 0
@t

In the LS method, the interface is assumed to have a nite thickness. Thus a smoothed density and viscosity, denoted as qe u and
le u, are dened in each computational cell as,

qe u qg ql  qg He u

ru
jruj

9
u0

The surface tension force is then discretized as a volume force

using a modied CSF method. CSF was rst introduced by
Brackbill et al. (1992). This model has been modied in TransAT
to overcome the shortcomings of the original model and minimize
smoothing of this force across the interface (Lakehal et al., 2002;
Liovic et al., 2006; Liovic and Lakehal, 2007). In TransAT the
governing equations are discretized on a collocated mesh in which
velocities are dened at cell centers. Adaptive time stepping is
controlled by specifying the following limits: CFL maxjuijk jDt=Dx;
jv ijk jDt=Dy; jwijk jDt=Dz < 0:3, DIFF maxlDt=qDx2i < 0:3, and
p
STN maxDt rjjjdni =qDxi < 0:3, which are time restrictions for the convective, viscous, and surface tension terms,
respectively.
As mentioned in the introduction, experimental studies for bubble oscillation under forced vibration is limited to small amplitude
oscillations in which shape of the bubble remains spherical. In a
previous paper (Movassat et al., 2012), authors showed that simulation results for translational motion of the bubble match well
with experimental data. As frequencies applied in forced vibration
are orders of magnitude smaller than acoustic frequencies, no comparison can be made to acoustic studies where bubble undergoes
large amplitude shape oscillations. Also, experimental studies
involve multi bubbles and the interaction force among the bubbles
is an effective force. The focus of this work was to understand the
large amplitude shape oscillations and translational motion, and
chaotic interaction of these two motions, for a bubble under forced
vibration.
To identify the non-dimensional numbers, bubble diameter, D,
vibration frequency, f, and surface tension coefcient, r, are used
to non-dimensionalized the governing equations. Resulting
2

le u lg ll  lg He u

where subscripts g and l represent gas and liquid, respectively. The

modied Heaviside function, He, is dened as, He(u) = 0 for u < e,
He(u) = 1 for u > e, and He(u) = 0.5[1 + u/e + 1/psin(pu/e)] for
|u| 6 e. At each time step the LS function is reinitialized without
changing its zero level set. This is achieved by solving the following
partial differential equation,

@d
signu1  jrdj
@s

non-dimensional numbers are Bond number, Bo qAf D2 =r,

Reynolds number, Re qAf D=l, and the ratio of the vibration
amplitude to the bubble diameter, A/D. It will be shown that at
large amplitude oscillations, the viscous forces play a smaller role,
and Bo and A/D are sufcient to predict the bubble behavior. One
can use Weber number, We, as a non-dimensional number to characterize this ow. However, the Weber number in this case is writ2

ten as We qAf D=r, which is a multiplication of Bo and A/D:

We Bo  A=D, and it is not an independent non-dimensional
number in this problem. Therefore, it is sufcient to describe the
problem in terms of Bo and A/D.

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

3. Results and discussions

A single bubble with D = 4 mm is positioned in the middle of a
liquid container of dimensions 10  20  10 mm (in x, y, and
z-directions, respectively). The container is vibrated in the vertical
y-direction. A uniform mesh with resolution of 60 cells/D is used.
Fig. 1 illustrates the xy view of the problem conguration and
the mesh. The uid properties used are ql 1000 kg=m3 , qg
1 kg=m3 ,

ll 103 Pa s, lg 2  105 Pa s, and r = 0.073 N/m.

20 mm

10 mm
Fig. 1. Problem conguration (xy view), on the left, and the mesh resolution, on
the right.

213

These are physical properties of water and air at 20 C. Due to symmetry in the x- and z-directions, half of the domain in these directions is solved using symmetry boundary conditions for symmetry
planes and no-slip boundary conditions for the rest of the
boundaries.
3.1. Regular and chaotic responses
Two vibration frequencies are considered: (i) A frequency of
f = 106.8 Hz with an amplitude of A = 0.08 mm, which result in a
case with Bo = 0.2, A/D = 0.02, and Re = 34.2, and (ii) a frequency
of f = 75.5 Hz with an amplitude of A = 0.4 mm, which result in
Bo = 0.5, A/D = 0.1, and Re = 120.8. Figs. 2 and 3 illustrate the bubble shape, velocity vectors, and contours of velocity magnitude
during the rst period at times t = T/4, T/2, 3T/4, and T for the above
two cases: Bo = 0.2 and Bo = 0.5, respectively. The plane in which
the bubble shape and contours are plotted is the middle xy plane,
which goes through the bubble center-of-mass. Due to the symmetry, zy plane yields the same results.
Figs. 2 and 3 illustrate the ow patterns during one period.
During the rst quarter, the top part of the bubble is pushed downward by the liquid on the top of the bubble. During the third quarter, the bottom part of the bubble is pushed upward by the liquid
below the bubble. These top and bottom forces result in a
non-uniform and asymmetric pressure distribution along the
y-direction, which results in an asymmetric velocity distribution
along the y-direction. For Bo = 0.5, it is observed that the imposed
oscillation force is strong enough to deform the bubble after one
period, while for Bo = 0.2, the surface tension force is such that
the deviation from the spherical shape is small.

Fig. 2. Bubble shape, velocity vectors, and velocity magnitude contours for Bo = 0.2, A/D = 0.02, and Re = 34.2 at (a) t = T/4, (b) t = T/2, (c) t = 3T/4, and (d) t = T.

214

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

Fig. 3. Bubble shape, velocity vectors, and velocity magnitude contours for Bo = 0.5, A/D = 0.1, and Re = 120.8 at (a) t = T/4, (b) t = T/2, (c) t = 3T/4, and (d) t = T.

Fig. 4 shows the location of the center-of-mass of the bubble in

the vertical direction with time for the two cases. For Bo = 0.2
(Fig. 4a), the bubble slowly moves to y = 0.05 mm and it then stably oscillates about this point. For Bo = 0.5 (Fig. 4b), the bubble has
a signicant drift along the y direction. The bubble moves downward and after 44 cycles it impacts the bottom wall. The translational oscillation of the bubble center-of-mass, as shown in

Fig. 4b, does not appear periodic, even though the forcing frequency is periodic. This is due to the coupling of the large amplitude shape oscillations with the translational motion, resulting in
a nonlinear and non-repeating pattern for Bo = 0.5.
In order to characterize the nonlinear behavior of the bubble
oscillation for the large Bo number case of Fig. 4b, the resulting
bubble shape variation is analyzed. The bubble shape in each

Fig. 4. Variation of bubble center of mass location in the vertical direction with time for (a) Bo = 0.2, A/D = 0.02, and Re = 34.2, (b) Bo = 0.5, A/D = 0.1, and Re = 120.8. For the
small amplitude case, the bubble oscillates locally with a small drift, while for the large amplitude case the bubble drift is large and the bubble eventually impacts the bottom
wall of the container.

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

period is decomposed into its Legendre polynomials, which are

mathematical functions used to decompose any arbitrary shape
to a linear sum of different modes:

r D=2

1
X
cn tPn cos h

10

n0

where r is the distance from any point to the bubble center-of-mass,

D is bubble diameter, Pn cos h is the nth mode harmonic, and cn is
the amplitude of the nth harmonic. The rst 10 harmonics were calculated. The rst two coefcients, c0 and c1, are zero because c0
reects volume oscillation, and c1 is associated with bubble center
of mass motion. c2 to c9 are calculated from the numerical model
results using a least squares algorithm (Press, 1992).
The variation of the coefcients of various harmonics are
obtained with respect to time. The variation of the second and
third mode coefcients, non-dimensionalized by bubble diameter,
c2/D and c3/D are shown in Fig. 5. Coefcients are calculated at
t = nT. Fig. 5a and b shows c2/D and c3/D for Bo = 0.2, A/D = 0.02,
and Re = 34.2. In this case, c2/D converges to 0.013 after 50 cycles
indicating that bubble is only slightly deviated from a spherical
shape. Since the bubble shape deformation is small, the bubble
translational motion remains periodic. For this case, c3/D converges

215

to zero indicating that the third mode does not contribute to the
shape of the bubble after several oscillation.
Fig. 5c and d shows the bubble shape coefcients for Bo = 0.5,
A/D = 0.1, and Re = 120.8. Neither c2 nor c3 converge to any value.
In this case, the magnitude of the coefcients is much larger than
the corresponding values for the previous small amplitude case.
The average values of c2 and c3 coefcients are about 10% and 8%
of the bubble diameter, respectively, resulting in a nonlinear and
chaotic shape oscillation. The asymmetry, associated with the third
mode, results in an imbalance between the top and the bottom
parts of the bubble in the y-direction. As a result, the imposed
oscillation force enhances the imbalance as the bubble oscillates
in time.
The variation of the surface area of the bubble for two cases
with time at t = nT is shown in Fig. 6. The area is
non-dimensionalized with the initial spherical area. For Bo = 0.2
case, the maximum deviation from the spherical shape is less than
0.06%, which occurs during the rst period. As the oscillations continue, the area converges to only 0.02% larger area than that of the
initial spherical shape. For the chaotic case of Bo = 0.5, the area of
the bubble in successive cycles oscillates, and as the oscillations
continue, the pattern of the oscillations changes. The maximum
surface area for this case, which occurs during the 36th period,

Fig. 5. Variation of Legendre polynomial coefcients with time for (i) Bo = 0.2, A/D = 0.02, and Re = 34.2, (a) c2/D, (b) c3/D and for (ii) Bo = 0.5, A/D = 0.1, and Re = 120.8, (c) c2/D,
(d) c3/D. Results suggest that the oscillations are regular for (i) since c2/D converges to 0.013 while there is almost no contribution from the third mode, but chaotic with no
convergence for (ii), where the amplitude of oscillations is comparable to the bubble diameter.

216

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

Fig. 6. Variation of the surface area of the bubble with time at t = nT, (a) Bo = 0.2, A/D = 0.02, and Re = 34.2, (b) Bo = 0.5, A/D = 0.1, and Re = 120.8.

reaches a 4% larger area compared to that of the initial spherical

shape.
Shape oscillations and translational motion show that for the
small amplitude case, both translational motion and shape oscillations are regular, i.e. the bubble behavior becomes repeating in
successive periods. For the larger amplitude case, shape of the bubble is not repeating in successive cycles as there is no convergence
in the Legendre coefcients. Also, there is not a stable position for
the bubble center-of-mass. The results suggest that the bubble
behavior at larger Bo and A/D becomes chaotic.

3.2. Bubble breakup

If the forcing is strong enough, the bubble deformation may
become so large that the bubble breaks up. Fig. 7 shows deformation of a bubble subject to a large amplitude forcing with Bo = 0.7,
A/D = 0.125, and Re = 160. At t = 0.25T, the liquid inertia forms a
small dimple at the top of the bubble (Fig. 7a). During the second
quarter, the bubble decelerates and the small dimple on the top
of the bubble grows in time. As a result, a downward moving liquid
ow forms at the core of the bubble at t = 0.35T (Fig. 7b). Since the

Fig. 7. Bubble shape, velocity vectors, and velocity magnitude contours for Bo = 0.7, A/D = 0.125, and Re = 160 at (a) t/T = 0.25, (b) t/T = 0.35, (c) t/T = 0.45, (d) t/T = 0.5, (e) t/
T = 0.55, (f) t/T = 0.66. The bubble is pierced due to the formation of a liquid jet within the core of the bubble.

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

217

bubble is decelerating, the velocity and the inertia of the liquid

cause the core liquid ow to penetrate into the bubble. Breakup
of the bubble by the core liquid ow continues during the remainder of the second quarter. After half a period, t = 0.5T, bubble attens and the core liquid ow continues to penetrate through the
bubble during the third quarter until the front of the liquid ow
reaches the bottom of the bubble. When the jet impacts the bottom
surface of the bubble, the bubble may break up forming tiny bubbles within the liquid, and tiny liquid drops may form inside the
bubble. After break up, a toroidal bubble is formed. As the forcing
continues, the toroidal bubble undergoes large amplitude shape
deformations and smaller bubbles are formed.
3.3. Viscous effect
Bo and A/D do not depend on the uid viscosities. To evaluate
the importance of viscous forces in the cases where bubble break,
the inviscid solution for the case with Bo = 0.7 and A/D = 0.125 is
compared to the viscous results presented in the previous sections,
having Re = 160. For the inviscid case, the viscosities of both liquid
and gas are set to zero. Results show that similar to the viscous
case, a liquid jet penetrates into the bubble core, causing the bubble breakup. Fig. 8 plots the distance from the front of the liquid jet
to the bottom of the container for viscous and inviscid cases. The
difference between the two cases is very small: the mesh size is
0.067 mm, and the maximum difference between the two cases
is about two mesh sizes. Although the difference is small, the gure
does show that in the inviscid case the front of the jet is closer to
the bottom of the container, i.e. that the jet has penetrated further
at the same time compared to the viscous case. Based on the above
comparison it is observed that the viscous forces do not alter the
bubble behavior. Consequently, for the range of parameters studies
here, the bubble response to the vibrations, Bo and A/D are sufcient to characterize the bubble behavior.
3.4. Bubble oscillation map
A parametric study is performed to predict the effect of Bo and
A/D on the bubble response to the forced vibrations. Fig. 9 illustrates the summary of the parametric study. Three types of
responses are characterized: regular oscillations, chaotic oscillations, and bubble breakup. Bubble breakup corresponds to the
cases in which shape oscillation is so large that breakup occurs.
Three lines are also shown in this gure. These are constant

Fig. 8. The variation of the distance between the liquid jet front and the container
bottom wall with time for viscous and inviscid solutions for Bo = 0.7 and
A/D = 0.125.

Fig. 9. Bubble response to forced vibration as a function of Bo and A/D. Three types
of responses are characterized: regular oscillations, chaotic oscillations without
piercing, and chaotic oscillations with piercing. The three straight lines are constant
frequency lines which correspond to the rst three resonant frequency modes.

frequency lines corresponding to the second, third, and fourth resonant frequencies of the shape oscillations. For a 4 mm air bubble
in water these frequencies are 52.6, 96, and 144 Hz for the second,
third, and fourth modes, respectively (Lamb, 1930).
Based on the map in Fig. 9, regular oscillations occur at small Bo
and small A/D. Cases with A/D 6 0.08 and Bo 6 0.4 (except for two
cases) respond regularly to the forced vibrations. For regular cases,
the second mode contributes to the bubble shape, while the asymmetric third mode does not contribute. Two chaotic (without
breakup) cases are observed within the regular region, Bo = 0.4,
A/D = 0.02, and Bo = 0.4, A/D = 0.05. Frequencies of these two cases
coincide with the third and fourth mode resonant frequencies.
Since the forcing frequency and the resonant frequency are very
close for these two cases, large amplitude oscillations are excited
and the bubble response deviates from the regular oscillations.
The triangular points represent cases with large amplitude
oscillations in which no breakup occurs within the rst 10 periods
of oscillations. These points mainly lie between the regular oscillation and the breakup cases. In these cases, higher harmonics than
the second mode are excited, and the amplitude of the oscillations
are much larger compared to the regular oscillation cases. As a
result of a coupling of two nonlinear systems (liquid and the gas)
both a chaotic shape oscillation and a chaotic translational motion
is observed.
As A/D and Bo increase, the inertia of the surrounding liquid
interacts with the bubble shape oscillations to result in bubble
breakup. Since the shape oscillation is chaotic, and because bubble
breakup depends on the shape of the bubble, the transition
between the chaotic with no breakup and chaotic with breakup
is not precise. For instance, at Bo = 0.6, breakup occurs for
A/D = 0.05, while increasing the amplitude to A/D = 0.06 results in
a no-breakup case. Detailed analysis of the results show that the
bubble shape for A/D = 0.05 provides the condition for breakup to
occur, while for A/D = 0.06, the bubble shape and the ow eld prevents the formation of a core ow to break up the bubble.
To summarize, there are two factors that combine to lead to
bubble breakup under forced vibrations:
i. Bubble shape: Nonlinear large amplitude oscillations of the
bubble shape allow for the onset of bubble breakup. At small
amplitude shape oscillations the bubble does not deviate

218

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219

greatly from the equilibrium spherical shape. Surface tension is strong enough to prevent the formation of a core liquid ow that pierces the bubble. As the oscillation amplitude
increases, regions of larger and smaller curvature are
formed. As well, the bubble shape changes the ow eld of
the surrounding liquid. A large amplitude shape oscillation
is a necessary condition for the occurrence of bubble
breakup.
ii. Liquid inertia: Due to the density difference between the gas
in the bubble and the surrounding liquid, any small change
in the velocity of the liquid results in a great change in the
nearby gas velocity. As a result, when the surrounding liquid
pushes the bubble, the velocity of the gas which is close to
the liquid, increases. As well, the liquid is able to deform
and penetrate into the bubble. Because of the symmetry of
the problem considered here, the maximum liquid velocity
and consequently, the maximum inertia transfer from the
liquid to the gas occurs at the bubble core. As a result,
breakup initiates and occurs within this region.

3.5. Effect of mesh size

Finally, to demonstrate that the mesh resolution which was
used in this work is good enough to capture the physics of the
problem, the case of Bo = 0.7 and A/D = 0.125 was modeled with

t/T=0.25

t/T=0.5

a ner mesh. The mesh resolution was rened to 90 cells per diameter, for a total of 5,695,312 cells in the domain. Instead of using 32
processors as for the coarse case, 64 processors were used to model
the ne case. Fig. 10 compares the shape of the bubble and the vertical velocity contours for the coarse and ne meshes. The comparison shows that while there are tiny differences in the velocity
contours, especially at t = T/2 when the bubble center of mass is
at rest and the velocities are small, the coarse mesh can still predict
the breakup and the penetration of the liquid jet within the bubble
core. The results are very similar in the two cases. It should be
noted that the formation of the smaller bubbles and the lm drainage is a micro-scale problem, and it is beyond the scope and
numerical capacity of this work. The goal of this mesh renement
study is to show that the bubble breakup is a physical phenomenon rather than a numerical error.
4. Summary and conclusions
The response of a bubble in a liquid container in response to
forced vibrations was studied. Forced vibration induces an oscillatory force on the bubble. As a result, the bubble undergoes an oscillatory translational motion. Both the bubble shape and the bubble
location inside the container oscillate. The volume oscillation is
small and it is neglected as the applied frequency is much lower
than the Minnaert frequency. Bond number (Bo), A/D and, and
Reynolds number (Re) can describe the dynamics of this type of
bubble oscillation. It is shown that the viscous forces do not play
an important role in the range of parameters studies here, and
the effect of the Re is not considered. It was shown that:
 Increasing Bo and A/D change the bubble behavior from a regular oscillation to a chaotic one. The second mode of the oscillation is the dominant mode contributing to the shape of the
bubble in regular oscillations. In the chaotic oscillations, the
higher modes are also excited resulting in large amplitude nonlinear shape oscillations. The translational motion also becomes
chaotic and non-repeating for the large amplitude cases.
 Further increase of the amplitude and frequency of oscillations
results in bubble breakup. As a result of the inertia of the oscillating liquid, a core liquid ow is formed within the bubble core.
This ow penetrates into the bubble and results in a pierced
bubble with a toroidal shape. The toroidal bubble undergoes
through large amplitude oscillations and the main bubble
breaks into smaller bubbles.
 A map indicating the outcome of the bubble oscillation, i.e. regular, chaotic, and breakup, as a function of Bo and A/D is provided by performing a parametric study. The bubble shape
oscillation is chaotic in large forcing, and there is not always a
clear transition from regular to chaotic and then to breakup as
the shape of the bubble plays an important role in the oscillation outcome. Vicinity of the applied frequency to the natural
frequency of different modes also changes the oscillation
outcome.

References

t/T=0.66
Fig. 10. Bubble shape and vertical velocity contours for Bo = 0.7 and A/D = 0.125,
coarser mesh (60/D) on the left and ner mesh (90/D) on the right.

Akhatov, I.S., Konovalova, I.S., 2005. Regular and chaotic dynamics of a spherical
bubble. J. Appl. Math. Mech. 69, 575584.
Benjamin, T.B., Ellis, A.T., 1990. Self-propulsion of asymmetrically vibrating bubbles.
J. Fluid Mech. 212, 6580.
Blake, J.R., Keen, G.S., Tong, R.P., Wilson, M., 1999. Acoustic cavitation: the uid
dynamics of non-spherical bubbles. Phil. Trans. R. Soc. Lond. A 357, 251267.
Brackbill, J.U., Kothe, D.B., Zemach, C., 1992. A continuum method for modeling
surface tension. J. Comput. Phys. 100, 335354.
Calvisi, M.L., Lindau, O., Blake, J.R., Szeri, A.J., 2007. Shape stability and violent
collapse of microbubbles in acoustic traveling waves. Phys. Fluids 19 (4),
47101-1-47101-15.

M. Movassat et al. / International Journal of Heat and Fluid Flow 54 (2015) 211219
Crum, L.A., 1979. Surface oscillations and jet development in pulsating bubbles. J.
Phys. Paris 41, 285288.
Devaud, M., Hocquet, T., Bacri, J., Leroy, V., 2008. The Minnaert bubble: an acoustic
approach. Eur. J. Phys. 29, 12631285.
Doinikov, A.A., 2004. Translational motion of a bubble undergoing shape
oscillations. J. Fluid Mech. 501, 124.
Eller, A., Crum, L., 1970. Instability of the motion of a pulsating bubble in a sound
eld. J. Acoust. Soc. Am. 47, 762767.
Farris, S.C., Bugg, D.J., Gabriel, K.S., 2004. The motion of bubbles in a sinusoidally
oscillating liquid in microgravity. Microgravity Sci. Technol. XV/3, 2835.
Friesen, T.J., Takahira, H., Allegro, L., Yasuda, Y., Kawaji, M., 2002. Numerical
simulations of bubble motion in a vibrated cell under microgravity using level
set and VOF algorithms. Ann. N.Y. Acad. Sci. 974, 288305.
Gaul, L., Papas, Z., Kurwitz, C., Best, F.R., 2010. Equilibrium interface position during
operation of a xed cylinder vortex separator. Space, Propulsion & Energy
Sciences International Forum, Washington D.C., 2010.
Ishikawa, M., Nakamura, T., Yoda, S., Samejima, H., Goshozono, T., 1994. Responsive
motion of bubbles to periodic g-jitter. Microgravity Sci. Technol. VII/2, 164168.
Knopf, F.C., Ma, J., Rice, R.G., Nikitipoulos, D., 2006. Pulsing to improve bubble
column performance: I. Low gas rates. AIChE 52 (3), 11031115.
Krishna, R., Ellenberger, J., 2002. Improving gasliquid mass transfer in bubble
columns by applying low-frequency vibrations. Chem. Eng. Technol. 25 (2),
159162.
Lakehal, D., Meier, M., Fulgosi, M., 2002. Interface racking towards the direct
simulation of heat and mass transfer in multiphase ows. Int. J. Heat Fluid Flow
23, 242257.

219

Lamb, H., 1930. Hydrodynamics. Cambridge University Press.

Liovic, P., Lakehal, D., 2007. Multi-physics treatment in the vicinity of arbitrary
deformable gasliquid interfaces. J. Comp. Phys. 222, 504535.
Liovic, P., Rudman, M., Liow, J., Lakehal, D., Kothe, D., 2006. A 3D unsplit-advection
volume tracking algorithm with planarity-preserving interface reconstruction.
Comput. Fluids 35, 10111032.
Mei, C.C., Zhou, X., 1991. Parametric resonance of a spherical bubble. J. Fluid Mech.
229, 2950.
Movassat, M., Ashgriz, N., Bussmann, M., 2012. Chaotic shape and translational
dynamics of 2D incompressible bubbles under forced vibration in microgravity.
Microgravity Sci. Technol. 24, 3951.
Ni, X., Gao, S., 1996. Mass transfer characteristics of a pilot pulsed bafed reactor. J.
Chem. Tech. Biotechnol. 65, 6571.
Press, W.H., 1992. Numerical Recipes in FORTRAN: the Art of Scientic Computing.
Cambridge University Press.
Watanabe, T., Kukita, Y., 1993. Translational and radial motions of a bubble in an
acoustic standing wave eld. Phys. Fluids A 5, 26822688.
Weislogel, M.M., Thomas, E.A., Graf, J.C., 2009. A novel device addressing design
challenges for passive uid phase separations aboard spacecraft. Microgravity
Sci. Technol. 21, 257268.
Yoshikawa, H.N., Zoueshtiagh, F., Caps, H., Kurowski, P., Petitjeans, P., 2010. Bubble
splitting on oscillatory ows on ground and in reduced gravity. Eur. Phys. J. E.
31, 191199.
Zhang, H., Mudawar, I., Hasan, M.M., 2009. Application of ow boiling for thermal
management of electronics in microgravity and reduced-gravity space systems.
IEEE Trans. Compon. Packag. Technol. 32 (2), 466477.