7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Bubble behavior in glass rectangular microchannels
Chiwoong Choi, Donin Yu and Moohwan Kim
,Mechanical Engineering Department, Pohang University of Science and Technology
San 31, HyojaDong, NamGu, Pohang, 156093, Republic of Korea
cwchoi@ucla.edu,
Keywords: bubble, microchannel, pressure drop, void fraction
Abstract
Bubbles in a rectangular microchannel are surrounded with nonuniform liquid film. Therefore, bubble behavior in a rectangular
microchannel is different with that in a circular microchannel. Experiments of adiabatic twophase flow in rectangular microchannel were
conducted with five rectangular microchannels. Their geometrical parameters are a hydraulic diameter and an aspect ratio. This study was
extended works of previous study by Choi et al. (2009). Visualization of flow regime was carried out using a highspeed camera and a long
distance microscope. From visualized images, bubble velocity and void fraction were evaluated. The bubble velocity has a linear relation
with total superficial velocity similar to conventional theory of Wallis (1969). The void fraction has linear relation with volumetric quality
similar to Armand (1964). The critical parameter is the hydraulic diameter rather than the aspect ratio. Therefore, the new correlation was
proposed as a function of the hydraulic diameter. The pressure drop in a single bubble was evaluated using a unit cell model. The pressure
drop has also dependency of the hydraulic diameter. Finally, new correlation of the pressure drop in the elongated bubble in a rectangular
microchannel was proposed. In addition, it is applicable in various aspect ratio rectangular microchannels.
Introduction
In phase change process in a microchannel, an incepted
single bubble grows, and makes a longer bubble due to
confinement effect of microchannel. It is called elongated
bubble. This flow pattern is one of major flow pattern in a
microchannel (flow pattern). There are many experiments
related with twophase flow in a microchannel for different
kinds of topics. For example, heat transfer, pressure drop,
flow pattern, void fraction, etc. However, there are less
studies for bubble motion in a microchannel. Bubbles in
circular microchannel can be assumed 1 dimensional shape,
if gravitational force is negligible. It means the thickness of
liquid film surrounding a bubble is same for radial direction.
On the other hand, Bubble in rectangular microchannel has
a radial profile of the liquid film. Wong et al.(1995) reported
bubble motion in a rectangular capillary has less hydraulic
resistance than that in a circular capillary due to the corer
effect in rectangular crosssection. Choi et al.(2009)
conducted experiments for bubble motion in rectangular
microchannels with nitrogen gas and water liquid. And they
visualized bubble motion in the rectangular microchannels.
They reported bubble velocity, void fraction and pressure
drop. In addition, using image processing, the pressure drop
in single bubble in rectangular microchannel was evaluated.
In this study, we extended same experiments with various
diameters and aspect ratios.
Nomenclature
area (m2)
aspect ratio factor
CaB Capillary number base on bubble velocity
D diameter (m)
f friction factor
G mass flux (kg/m2s)
H height (m)
j superficial velocity (m/s)
L length (m)
P pressure (kPa)
Re Reynolds number
t Time [s]
u velocity (m/s)
w width (m)
Greek letters
A difference
a Void fraction
p Volumetric quality
p Density (m3/kg)
6 Thickness (m)
Subsripts
L liquid phase
nose bubble nose
G gas phase
B bubble
S liquid slug
frame image frame
h hydraulic
UC unit cell
ch microchannel
measured measured value
j Based on total superficial velocity
EB elongated bubble
Glass Microchannels
To visualize bubble motion, glass microchannel was
fabricated using photosensitive glass.
MEMS(microelectromechanical system) fabrication
techniques and procedure of the microchannels was
described in Choi et al.(2008). Figure.1 shows test section
image of microchannel. Two inlets for each phase, one
outlet and three pressure ports were patterned. To mix water
liquid and nitrogen gas, Tjunction was used. Total length of
microchannel and distance between pressure ports are 60
mm and 15 mm, respectively. In this study, five
microhcnanels were fabricated using control of height and
width of microchannels. The crosssectional dimensions of
the rectangular microchannels were measured by the 3D
profiler (VeecoWyko DMEMS NT1100). Different
hydraulic diameters and aspect ratios of the rectangular
microchannels are shown in Tablel. MC 1, 4 and 5 are
similar aspect ratios. And MC 2, 4 and MC 3, 5 are similar
hydraulic diameters, respectively.
nitrogen gas inlet
Tjunction mixing region
Figure 1: Test section (top view): microchannel, inlets,
outlet and pressure ports.
Table 1 Dimensions of test sections.
Microchannel Number Dh (Wch x
Hch) [ftm], AR
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Three kinds of pressure transducers (Druck LPM 9000 and
Setra 209 25PSID) were used for appropriate range.
Pressures and temperatures at upstream of both of liquid and
gas flows were measured to confirm a steady condition. The
visualization was achieved using a highspeed camera
(MotionXtra HG 100K) and a longdistance microscope
(Infinity PhotoOptical KC/STM) with cold light. Visualized
region is between last two pressure ports as shown in Figure
1. All measured data were gathered by data acquisition
system (Agilent 34970A) and a personal computer with
sampling rate of 0.5Hz.
Highspeed camera water liquid flow
S nitrogen gas flow
p) flowmeter A
longdistance      
microscope (T ,
filter
TI I* / i Pre KP e vessel
S Testsection T fl owmeter v
(Figure 4)
U filter
Cold light
regulate
Water tank t
Data acquisition system Personal computer I Hot plate
Figure 2: Experimental apparatus (Choi et al.2009).
Table 2 shows experimental ranges of this study. All flows
were included laminar flow. Range of gas superficial
velocity was relatively narrow because we focused only
bubble flow regime. 8 and 10 cases for liquid and gas
superficial velocities were selected for five microchannels.
And several data of lower liquid superficial velocity for
smaller microchannel could not be obtained. Therefore, total
number of data was 355 for five different crosssection
microchannels. Uncertainty analysis was conducted and
results are shown in Table 3.
e vessel
MC 1 479 (510 x 450), 0.88
MC 2 322 (501 x 237), 0.47
MC 3 143 (503 x 85), 0.17
MC 4 304 (332 x 280), 0.84
MC 5 165 (181 x 151), 0.83
Experimental Procedure and Uncertainties
Figure 2 shows an experimental loop. A flow was
controlled by a pneumatic pump, which was controlled by
the electric regulators (SMC ITV1000 series) with a
pressure vessel. Helium and nitrogen gases were used as
pressurized gas for liquid flow and gas flow, respectively.
Before conducting experiments, water in a water tank was
degassed using a USP method, which was validated by
Curley et al.(2004). Flow rates were measured by
flowmeters (OMEGA FMA1602A, 1603A, FVL1604 and
1619A) for both liquid and gas flows. And inline filters were
used for both liquid and gas flow paths. Pressures were
measured directly through the embedded pressure ports.
Table 2 Experimental ranges.
Variables
Liquid mass flux, GL [kg/m s]
Gas mass flux, GG [kg/m2s]
Liquid Reynolds number, ReL
Gas Reynolds number, ReG
Liquid superficial velocity, jL [m/s]
Gas superficial velocity, in [m/sl
Table 3 Uncertainties.
Variables
Diameter [tm]
Area [tm 2]
Pressure [kPa]
Temperature [C]
Mass flux, G [kg/m2s]
Friction factor
Superficial velocities
Ranges
133 600
0.075 0.75
19 286
0.6 20
0.13 0.6
0.06 0.66
Uncertainty
17 (4.5 %)
6311 (3.8 %)
0.005, 0.034, 0.085
S0.5
S2%
9%
10 %
Single Phase Pressure Drop
To validate experimental apparatus and measurement
techniques and obtain friction factor for each microchannel,
the experiment of single phase water pressure drop was
carried out. The value of the friction factor multiplied by
Reynolds number is constant for laminar flow. The fRe =16
for circular tube. And Hartnett and Kostic(1989) proposed
fRe correlation for rectangular duct with different aspect
ratios as Equation (1). And Equation (2) is friction factor for
single phase laminar flow.
f Re= 24(11.3553AR+1.9467AR2
(1)
1.7012AR3 +0.9564AR4 0.2537AR5)
f=Dh 2( (2)
f 2ph A(
Where, f is a friction factor, Re is Reynolds number and AR
is an aspect ratio factor, which is defined shorter length over
longer length, the smaller AR means higher aspect ratio. The
aspect ratios of the rectangular microchannels were
described in Tablel. Figure 3 indicated that Hartnett and
Kostic's correlation was well agreement with experimental
pressure drop in a rectangular microchannel. Moreover, this
result validated that pressure drop measurement and
experimental apparatus in this study were well developed.
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Unit Cell Model
Visualization results shows bubble flow was fine periodic
flow pattern. It means that we can define the repetitive
region and whole phenomenon in the bubble flow pattern
was magnified by the number of the periodic region. This
region could be defined as unit cell. Therefore, to analyze
bubble behavior, unit cell model concept was applied to the
result of visualization of bubble flow in the rectangular
microchannels. Figure 4 describes conceptual image of the
unit cell model. Based on the unit cell model, length of
bubble and liquid slug, bubble velocity and the number of
unit cell was obtained.
Circular(fRe=16)
a MC2 Experiment
A MC4 Experiment
MC1 (Hamett and Kostic)
MC3 (Hamett and Kostic)
MC5 (Hamett and Kostic)
* MC1 Experiment (a)
o MC3 Experiment
* MC5 Experiment
 MC2 (Hamett and Kostic)
MC4 (Hamett and Kostic)
o 1 'o
0.01
1000
Figure 4: Unit cell model: (a) visualized bubble flow
pattern, (b) conceptual image of unit cell model.
Bubble Velocity
From the image processing, the bubble velocity could be
evaluated using following equation.
S nose
frame
100
/
E
100
, MC1 Experiment (b)
A MC2 Expriment
o MC3 Expriment
A MC4 Expriment
* MC5 Expriment
MC1 Hartnett and Kostic
MC2 Hartnett and Kostic
MC3 Hartnett and Kostic
MC4 Hartnett and Kostic
MC5 Hartnett and Kostic
1000 10
Re
Figure 3: Results of single phase pressure drop in
rectangular microchannels: (a) friction factors, (b) pressure
drop.
Where, ALnose is the moving distance of a nose of bubble
in two images and Atframe is differential time of captured
two images. The fifty velocity data were evaluated and
averaged. The bubble velocity of a slug bubble in horizontal
tube can be defined as Equation (4) (Wallis 1969).
UB = Cj
Where, coefficient Cl is distribution factor, which can be
evaluated as the ratio of the bubble velocity and total
superficial velocity. In other words, Cl is area ratio of
bubble and microchannel (Choi et al. 2009). Therefore, the
smaller crosssectional area of the bubble made the higher
bubble velocity. Cubaud and Ho(2006) report Cl to be
approximately 1, and Fukano and Kariyasaki (1993)
measured bubble velocities in 1, 2.4 and 4.9 mm diameter
capillaries. The evaluated Cl value was close to 1.2. And
new correlation for the bubble velocity was proposed. Their
correlation is shown in Equation (5),
10 r
~3~t~ 3
uB = Cj105 (5)
Where C1 values were 1.09, 1.17 and 1.21 for diameters of
4.9 mm, 2.4 mm and 1.0 mm, respectively. As diameter
decreased, the coefficient Cl increased. Thulasidas et al.
(1995) reported liquid film flow rate in a corer of a square
duct is higher than that in a circular tube. For rectangular
cross section, there are two kind geometric parameters; the
aspect ratio and the hydraulic diameter.
Figure 5 shows measured bubble velocities for the five
different rectangular microchannels.
3
o MC1 (D= 479 iml) A MC2 (D= 322 /m)
2.5 MC3(Dh143uni) A MC4(Dh 304 an)
0 MC5 (Dh= 165 iml) Correlation MC I
Correlation MC2 Correlation MC3
2 Correlation MC4 Correlation MC5
[ A
0.5
O aU1
A _l______________________
0
Figure 5: Bubble
microchannels.
velocities in different
rectangular
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
C1 values were evaluated using Equation (4). Figure 6
shows dependency of the bubble velocity in the rectangular
microchannel for the aspect ratio and hydraulic diameter.
There is no correlation of the aspect ratio in the bubble
velocity, however, the C1 is decreased as the hydraulic
diameter decreased. This result is opposite to result of
Fukano and Kariyasaki (1993). It means the smaller
diameter makes larger diameter ratio of the bubble and the
rectangular microchannel. I proposed new C1,distribution
factor in Equation (4) as function of the hydraulic diameter
for the rectangular microchannel as follows:
C = 965Dh +0.82
Where, the unit of Dh is meter. Applicable range is 143 
179 umr for Dh and 0.17 0.88 for AR. And the relative
error of this correlation is 4.44 %.
Void Fraction
In open literature, there are two kinds of correlations for
void fraction. The one is Armand (1946)'s correlation,
which indicates linear relation with volumetric quality, P as
shown in Equation (7).
a B = 0.833 G
G + jL
The other is proposed by Chung and Kawaji (2004), which
indicates nonlinear relation with volumetric quality. And
Xiong and Chung (2007) developed for different diameters
as shown Equation (89).
mr/05
a=
1 (1 m)/05
0.266
1+13.8
1+13.8.e688Dh
0.2 0.4 0.6 0.8
AR
1.298 *
In this study, void fraction was evaluated using following
basic relation.
=JG
a=
1.132 1.11.4
Figure 7 indicates that as same to Armand's correlation, the
void fraction for the five rectangular microchannels has
0.993
linear relation with volumetric quality. Based on Equation
"".0.969 (7), coefficient CA was evaluated using regression method.
Figure 8 shows dependency of CA coefficient for the aspect
ratio and the hydraulic diameter. The hydraulic diameter is
more influential parameter in void fraction relation as
0 100 200 300 400 500 600 shown in Figure (8). In our experimental range, new Ca
Dh [pm] coefficient was proposed as a function of the hydraulic
S. .. s diameter as following equation.
giF ure 6: The C1 coefficients of five rectangu
diameters.
C,= 1.11675Dh
1.298 ,
1.132
*
0.993
1
S0.969
1.114
Ucji
0.9
0.8
1;'i
Where, a unit of Dh is meter. And relative error of this
correlation is 3.03%.
MCI(Dh=47911m)
SMC2 (Dh 322 anm)
1 0 MC3 (Dh, 143 piu)
A MC4 (Dh 304 pim)
0.8 MC5(Dh= 165 pl)
Armand
Xiong and Chung (D = 100 11ll)
0.6 :
0.4
0.2
0
Figure
microc
0.8
0.6
0.4
0.2
0
1.2
1
0.8
S0.6
0.4
0.2
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
modified C as a function of hydraulic diameter, mass flux,
and heat flux etc. In this study, we focused on only bubble
regime in a rectangular microchannel. Especially, we
studied the pressure drop in single bubble in the rectangular
microchannel. Based on unit cell model, the pressure drop
in unit cell is consisted of liquid slug, bubble body and
interface region. The liquid slug bubble is same to single
phase flow, therefore, the pressure drop in liquid slug region
can be evaluated using Equation (11)
s 2 fp 2 f Ls
Dh
Where, Ls is liquid slug length, which was satisfied region
of Equation (11). He and Kasagi (2008) calculated bubble
0 0.2 0.4 0.6 0.8 1 motion in capillary tubes. They reported the length satisfied
P single phase motion (Equation (11)) was shorter than real
S7: Void fraction in different rectangular length of the liquid slug with order of the diameter. We
channelss with correlations. applied their relation to calculate the pressure drop in the
liquid slug. The pressure drop in bubble body was negligible
because of no gradient of pressure drop. In addition, density
1.00 1.01 (a) of gas phase is much lower than liquid phase. In this study,
S0.89 the pressure drop in a single bubble was defined as the
S0.92 pressure drop in bubble body and interface regions. Finally,
The pressure drop in a single bubble was calculated using
0.79 following Equation (11) and (12).
AP
MB measured Ms
NUC
Where, Nuc is the number of the unit cells. Figure 9 shows
typical result of the pressure drop in single bubble in
0 0.2 0.4 0.6 0.8 1 rectangular microchannel. Bretherton (1961) reported the
AR pressure drop in long slug bubble was correlated with
Capillary number based on the bubble velocity. Therefore,
the pressure drop was ploted with CaB. Choi et al.(2009)
1.01 92 (b) reported that the pressure drop in the bubbly flow was lower
10 0 . than that in the elongated bubble and the data of the
1.00 _ 0.79 pressure drop in the elongated bubble pattern follows an
0.89 ~ asymptotic line. Figure 9 confirms same results. Therefore,
the pressure drop for the only elongated bubble pattern was
extracted with criterion of LB/Wch > 1.5. Figure 10 shows
the pressure drop for the elongated bubble pattern in five
different rectangular microchannels. The pressure drop in
the single elongated bubble was correlated following
equation.
0 1
0 100 200 300 400 500 600
Dh [um]
Figure 8: The dependency of coefficient CA in five
rectangular microchannels: (a) for the aspect ratio, (b) for
the hydraulic diameter.
Pressure Drop In Single Bubble
Many researchers have intensively studied pressure drop in
twophase flow in microchannels during last decade.
However, most approaches are based on modification of C
value in Lockhart Martinelli model (Awad 2006). They
APJ = MCa'
ERB
The correlations of coefficient m and M were compared for
the aspect ratio and the hydraulic diameter. The coefficient
m was approximately same to 1.0. So, we assumed that
coefficient m = 1.0. the coefficient M was inverse
proportional to the hydraulic diameter as shown in Figure
11(b). Figure 12 shows pressure drop and CaB/Dh relation.
Finally, the new correlation for the pressure drop in the
single elongated bubble in a rectangular microchannel was
proposed as Equation (14).
d
a.i
1
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
60 51.42
S60.02
Where, the unit of Dh is meter. The relative error of this
correlation was 8.9 %.
He and Kasagi (2008) proposed pressure drop in transition
between liquid slug and bubble as following Equation.
APEB 42.4
APEB 2 0.07++ (15)
(pu2/2) Rej
Where, Re, is defined for total superficial velocity and
APEB is nondimensionlized pressure. Figure 13 shows He
and Kasagi's correlation underestimated our experimental
results. So, we modified He and Kasagi's correlation as
Equation (16)(shown in Fig. 13). The relative error of
modified He and Kasagi's correlation is 12.4 %.
94.16
AP' = 0.07 +
Rej
26.47
*
S25.82
17.32 ,
0 0.2 0.4 0.6 0.8 1
m 60.02
50 51.42
40
30
25.82
Asymptotic for Enlongated bubble
(8)
(7)
r6 (6)
T (5)
ILzz (4)
bmw7 (3)
o a a (2)
ra a (1)
0 100 200 300
Dh [IPm]
400 500 600
Figure 11: The correlation of coefficient M for five
different rectangular microchannels: (a) for the aspect ratio,
(b) for the hydraulic diameter.
(1) (2) (3)(4)(5)(6)(7)(8)
0.001 0.010 0.100
CaB
Figure 9: Trend of the pressure drop in single bubble and
visualized flow pattern in MC 1 for jL = 0.6 m/s.
SMC1 (Dh= 479 i/n)
o MC3 (Dh= 143 i1n)
* MC5 (Dh= 165 pgm)
SMC2 (Dh= 322 Imi)
* MC4 (Dh= 304 fml)
< 0.1 A qo
0.01
0.001 0.01 0.1
CaB
Figure 10: The pressure drop in single elongated bubble for
five different rectangular microchannels.
10
0.1
, 0.1
I MC I (Dh= 479 im) A MC2 (D= 322 'iln)
o MC3 (Dh= 143 mn) A MC4 (Dh= 304 nm)
MC5 (Dh 165 ff1) New Correlation
10 0.008 ,
10 100 =0Ca 0
1 10 100 1000
CaB/Dh
Figure 12: New correlation for pressure drop in the
elongated bubble in rectangular microchannel.
0.008
APEB = CaB
AB CaD B
Dh
0.10
ff
3
n MCI (D =479 yu)
2.5 A MC2 (Dh 322 Lm)
c MC3 (DI, 146 yi)
MC4 (Dh 304 gIn)
2 MC5 (1%= 165 Al)
SHe and Kasagi
1.5 Modified He and Kasagi
0.5
0.5 
0
0 100 200 300 400 500 600 700 800 900 1000
Rej
Figure 13: Comparison experimental result with He and
K.,..,I, 2i" 11 's correlation and modified correlation.
Conclusions
This study is extension of our previous work (Choi et al.
2009). To investigate bubble behavior including bubble
velocity, void fraction and pressure drop in single bubble,
experiments for bubble flow pattern in five rectangular
microchannels were conducted. Five different aspect ratio
and hydraulic diameters are used to study their influence on
bubble behavior. From visualization with the unit cell model,
Bubble velocity, void fraction and pressure drop were
analyzed. From the results, following conclusions are
derived.
1. The bubble velocity has linear relation with total
superficial velocity and the coefficient C is function of
the hydraulic diameter not the aspect ratio. We
proposed new C as function of the hydraulic diameter.
2. The void fraction has linear relation with a volumetric
quality as similar to Armand's correlation. The
coefficient CA was newly proposed as a function of the
hydraulic diameter.
3. We reconfirmed that the pressure drop in a bubbly
pattern is lower than that in an elongated bubble pattern.
The pressure drop in the single elongated bubble was
correlated with Capillary number has dependency of the
hydraulic diameter. Finally, new correlation for the
pressure drop in the single elongated bubble was
proposed.
Acknowledgements
References
Wong, H., Radke, C.J. and Morris, S. The motion of long
bubbles in polygonal capillaries: part. Tin films, J. Fluid
Mech., Vol. 292, 7194 (1995)
CHOI, C.W., Yu, D.I. and Kim, M.H. Bubble dynamics and
pressure drop of an elongated single bubble in a rectangular
microchannel, 7th International Conference on Nanochannel,
Microchannel and Minichannel, Paper82104 (2009).
CHOI, C.W. and Kim, M.H. The fabrication of a single
glass microchannel to study the hydrophobicity effect on
twophase flow boiling of water, J. Micromech. Microeng.,
Vol. 18, 105016 (2008).
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Curley, T., Forsythl, R., Sun, S., Fliszar, K., Colletto, M.
and Martin, G. P. Measurement of Dissolved Oxygen as a
Determination of Media Equilibrium During Dissolution
Testing, Dissolution Technologies, Vol. 11, 611 (2004).
Hartnett, J. P. and Kostic, M. Heat transfer to newtonian and
nonnewtonian fluids in rectangular ducts, Advances in Heat
transfer, Vol. 19, 247356 (1989).
Wallis, GB. Onedimensional Twophase
McGrawHill, New York, U.S., 262302 (1969).
Flow,
Cubaud, T., Ulmanella, U. and Ho, C.M. Twophase flow in
microchannel with surface modifications, Fluid Dynamics
Research, Vol. 38, 772786 (2006).
Fukano, T. and Kariyasaki, A. Characterisitics of gasliquid
twophase flow in a capillary tube, Nuclear Engineering and
Design, Vol. 141, 5968 (1993).
Thulasidas, T.C., Abraham, M.A. and Cerro, R.L.,
Bubbletrain flow in capillaries of circular and square cross
section, Chemical Engineering Science, Vol. 50, 183199
(1995).
Armand, A. A. The resistance during the movement of the
twophase system in horizontal pipes, Izv. Vses. Teplotekh.
Inst., Vol. 1, 1623 (1946).
Chung, P. M.Y and Kawaji, M. The effect of channel
diameter on adiabatic twophase flow characteristics in
microchannels, Int. J. Multiphase Flow, Vol. 30, 735761
(2004).
Xiong, R. and Chung, J.N. An experimental study of the
size effect on adiabatic gasliquid twophase flow patterns
and void fraction in microchannels, Phys fluids, Vol. 19,
033301 (2007).
Awad, M. M. and Muzychka, Y S. Bounds on twophase
frictional pressure drop gradient in minichannels and
microchannels, 4th International Conference on
Nanochannels, Microchannels and Minichannels, Limerick,
Ireland (2006).
He, Q. and Kasagi, N. Numerical investigation on flow
pattern and pressure drop characteristics of slug flow in a
micro tube, 6th International Conference on Nanochannels,
Microchannels and Minichannels, Darmstadt, Germany
(2008).
Bretherton, F. P. The motion of long bubbles in tubes, J.
Fluid Mech., Vol. 10, 166188 (1961).
