7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010



Two-phase Plug Flow Distribution at a Mleso-scale T-junction
Effect of Plug Bubble Length


Ju Hyuk Hong*, Sang Yong Lee* and Jin Seok Lim'

Department of Mechanical Engineering, KAIST, Science Town, Daejeon, Korea
t Halla Climate Control Corp, Sinil-dong, Daedeok-gu, Daejeon, Korea

juhrukhong akaist.ac.kr, singsr an I ii c I,.limic IsrI and jslim3 @mail.hcc.co.kr


Keywords: plug flow, plug bubble length, flow distribution, T-junction, meso-scale




Abstract

In the present work, as the preliminary study, distribution of two-phase flow at a meso-scale T-junction with the square
cross section of 600 x600 pnr2 was investigated experimentally. The experiments were conducted by using air and pure water
as the test fluids. The superficial velocity ranges of air and water were 0.24 1.22 m/s and 0.31 1.0 m/s, respectively, which
were confirmed to be in the plug flow regime. Two different types of mixers were used to observe the effect of the plug bubble
length on the flow distribution behavior. Experimental results showed that, in the meso-scale T-junction, the fractions of the
gas and liquid phases separated through the branch varied largely with the bubble length at the inlet of the main, while the
effect of the superficial velocities of each phase at the main appeared minor. More specifically, with the long plug bubbles, the
flow distribution was not much affected by the superficial velocities of the gas and the liquid as were reported by others for
larger size (mm-order) tubes. However, with the short bubbles, significant mal-distribution of phases was observed and even
the phase separation through the branch or run was available.


Introduction

Two-phase flow distribution at T-junction of small
diameter (compact heat exchanger size) tubes has been a
subject of interest, such as in Stacey et al. (2000) or Wren et
al. (2005). Most of the heat exchangers consist of parallel
channels connected to headers, and the flow distribution at
the header-channel junctions determines the heat transfer
performance of the heat exchangers. The header-channel
assembly is considered to be an accumulation of single T-
junctions and it is essential to look into the flow distribution
behavior at the single T-junctions as the basic approach (Lee
and Lee 2001, 2005).
For application to large-scale pipe systems, distribution of
two-phase mixtures at macro-scale T-junctions has been
studied extensively (Reimann et al. 1988) and it is known
that the distribution phenomena depend on the flow pattern,
quality and fluid properties as well as the branching
orientation (Saba and Lahey 1984). However, recently, the
size of the heat exchangers becomes smaller and smaller,
down to the compact heat exchanger and even to the meso-
scale heat exchanger, having channel diameter of the
compact passage size (Dh = 1 6 mml1) and the meso-channel
size (DI, = 100 pLm 1 mm) (Mehendale et al. 2000),
respectively.
For T-junctions of the compact passage size (or slightly
larger), Stacey et al. (2000), Wren et al. (2005) and Lee and
Lee (2001) have reported that the flow distribution of the
two-phase mixtures is seriously affected by the channel size
as well as the flow pattern at the junction inlet. In the slug
flow regime, smaller fraction of the gas-phase is separated


out to the branch as the channel size decrease (Wren et al.
2005). In the annular flow regime, larger fraction of the
liquid-phase flows out through the branch as the channel
size becomes smaller (Stacey et al. 2000), which is similar
to the slug-flow case.
For meso-scale T-junctions, Zun et al. (2007) have
examined the mal-distribution of two-phase flow including
the bubbly, slug and the annular flows. However, more
detailed study is still to be performed for each flow pattern.
In this size range, according to Lee and Lee (2008),
Coleman and Garimella (1999) and Chung and Kawaji
(I**-'""} the slug/plug flow regime appears predominantly
and the flow distribution of the two-phase mixture
corresponding to that regime should be studied first.
In the slug flow regime, the two-phase mixture flows as a
train of long bubbles separated by liquid slugs, and basically,
the flow distribution behaviour may be predicted from the
size and velocity of the bubbles and the liquid slugs. Similar
concept has been proposed by Arirachakaran (1990) for
macro-scale horizontal T-junctions: there, the flow was
considered to be a combination of stratified-flow part and
the liquid-slug part, where the gravity effect plays an
important role. However, Lee and Lee (2001) have showed
that the gravity effect is no longer important as the junction
size becomes smaller as the compact passage size by
comparing his experimental results (with the vertical inlet
and horizontal branch) to those of Stacey et al. (2000) (with
the horizontal inlet and horizontal branch). Obviously the
model by Arirachakaran (1990) is not applicable to the
channel of the compact passage size or smaller. Recently,
Wren et al. (2005) have attempted to find the effect of the






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

gear pump (HNP Mikrosystems, mzr-4622) with a feedback
loop to reduce pulsations and control flow rates of water.
Another inlet was connected to the pressure-regulated
building air line through an air flow meter (UIP, M2030).
Two types of mixers (mixers 1 and 2, as shown in Fig.
1(b) and (c), respectively) were adopted to generate gas
plugs with different lengths for same flow rates of air and
water at the test section inlet. Mixer 1 is a commercial Y-
shaped fitting that connects 2 mm ID tubes at the inlet and
the exit parts and generates long plug bubbles at the
downstream. On the other hand, mixer 2 has T-shaped
configuration with approximately the same cross-section
size as the test section (600x600 pr)~. Mixer 2 generates
short plug bubbles at the downstream because the cross
section sizes of the channels are much smaller than those of
mixer 1. (In other words, the length of the plug bubbles is
approximately in the same order of magnitude with the
cross-section size (diameter) of the channels as examined by
Steijn et al. (2007) and Yue et al. (2008). Thus the thinner
gas stream discharging from the gas port is more easily
disintegrated by the liquid stream.)
Figure 1(d) shows the details of the test section, which
consists of a main, run and a branch with all of them having
the square shapes. The test section was made of transparent
acryl plates for flow visualization, and placed horizontally.
The cross-sectional area of the main, run and the branch
were 590 x600 pLm 590 x600 pm2 and 600 x600 pLm ,
respectively. Fractions of the flow distribution to the run
and the branch were controlled by adjusting the valves at
each downstream part, but with the pressure at 30 mm
upstream of the junction kept constant, 41+2 kPa. The
pressure was measured using a pressure transducer
(Validyne, DPl5).


slug length on the flow distribution at T-junctions but
accurate prediction of the slug length was not successful and
concrete conclusion could not be provided.
Therefore, in the present work, flow distribution of two-
phase plug flow at meso-scale T-junction was examined
through a series of experiments especially by controlling the
length of the plug bubbles at the junction inlet.

Nomenclature


hydraulic diameter (m)
flow rate (kgs ')
quality (-)
superficial velocity (ms')


Subscripts


main
run
branch
gas
liquid


Experimental Setup

Figure 1(a) shows the experimental setup for the two-
phase flow distribution test. The system basically consists of
the liquid and gas supply lines, two-phase mixer, test section
(T-junction), the flow control/measurement part at the
downstream of the run and the branch, and the flow
visualization device.
Air and water were used as the test fluids for gas and
liquid phases. One of the mixer inlets was connected to a


Flow visualization device


(a) Experimental setup


250 mm 30 m 150 mm


Main,l 1 iI Run, 2
(590 x600 pm2)1 1 (590 x600 pm2)


Branch, 3
(600 x 600 pm2)

(d) Test section


Liquid-


(b) Mixer 1


Figure 1: Schematic diagram of (a) experimental setup, (b) two-phase mixer 1, (c) two-phase mixer 2 (d) test section


70 m

4 = 600pm

..a=6600pm



Gas
(c) Mixer 2


Liquid-
Gas*
4 mm OD 4 mm OD
2 mm ID 2 mm ID









Table 1. Experimental conditions
Two- Plug bubble
Case jG1 [m/S] jL1 [m/S] phase length/Dh
mixer (main channel)
A 1.0 0.31 105
B (A 0.01) 0.86 100
C 0.26 1 74
D 0.75 77
E 1.22 1093
F 0.24 (A00)1.6
G 1.20 22.5

The two-phase mixtures through the run and the branch
were collected by the flow separators at each end; and the
water flow rates were estimated by weighing the collected
amount (using the electronic balances, A&D, CB-2000) at
the bottom for 5 minutes, while the air flow rates by using
the air flow meters (UIP, M2030) installed at the top exits.
Also a CCD Camera (ProgRes@ MF Cool, Jenoptik) and a
stroboscope (Seorim-electronics, DX-525) were used to
record the images of the two-phase flow behavior in the test
section and to measure the plug-bubble lengths.
Table 1 summarizes the experimental conditions tested in
the present work. The ranges of the volumetric fluxes of
water and air were 0.31 1.0 m/s and 0.24 1.22 m/s,
respectively. Mixer 1 was used for cases A E (with long
bubbles) and mixer 2 for cases F and G (with short bubbles).
Cases A and B compare the effect of the water flow rate on
the distribution behavior for a fixed flow rate of air. On the
other hand, cases C E were tested to examine the effect of
the air flow rate for a fixed flow rate of water. Finally, cases
F and G were tested to see the effect of the plug bubble
length on the flow split to the branch. (Note that the air and
the water flow rates of cases F and G are almost the same
with those of cases C and E, respectively; however, the
average lengths of the plug-bubbles are drastically different
from each other as shown in the last column of Table 1.)

Results and Discussion

Flow pattern
Prior to the main test, the flow patterns of the air-water
mixture at the main were examined. For all test conditions,
the flow pattern belongs to the plug flow regime as shown in
Fig. 2 (photographs and illustrations) and also confirmed
from the flow pattern map of Chung and Kawaji (21***4), as
shown in Fig. 3, where the similar channel size (530 pLm ID)
was tested. More accurately, according to the recent work by





(a) An example of long plug bubbles by mixer 1
(Case E, jG1 = 1.22 m/s, jL1 = 0.99 m/s)



b bbe L. ~ sud

(b) An example of short plug bubbles by mixer 2
(Case G j'G1 = 1.20 m/1s, jL1 = 1.02 mI/s)

Figure 2: Flow pattern in main channel


i I ... ..... .
: Present work
_ Chung and Kawaji (2004)


0.1 1 10 100


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

Lee and Lee (2008), the present cases belong to the wet-plug
regime because of existence of thin liquid films surrounding
the plug bubbles, differentiated from the dry-plug flow in
which no liquid film exists between the bubbles and the wall.
As shown in Table 1, the plug bubbles become longer with
the decrease of the water superficial velocity and/or with the
increase of the air superficial velocity. (Compare cases A and
B for the water superficial velocity effect and cases C, D and
E or cases F and G for the air superficial velocity effect.)
Similar trend was reported by Steijn et al. (2007) and Lee
and Lee (2009), where the meso-scale T-shaped mixer and
compact passage size Y-shaped mixer were tested,
respectively. Here, it should be noted that the velocity effect
on the bubble length appeared minor compared to the
channel (mixer) size effect, which should be studied
systematically as a future work.
Figures 4 and 5 visualize the configurations of two-phase
plug flows with long and short plug bubbles, respectively. In
general, the plug bubbles were split into smaller sizes and
flows into the downstream channels. With the long plug
bubbles in the main, as shown in Fig. 4, the flow pattern
inside both the run and the branch remain to be the plug flow.
On the other hand, with the short plug bubbles in the main,
either the plug flow (Fig. 5(a)) or the bubbly-plug flow (Fig.
5(b)) is observed in the run and in the branch depending on
the fractions of the liquid and gas take-offs through the
branch.

Distribution of two-phase flow with long plug bubbles
Figures 6 and 7 show the effects of the water and air
superficial velocities, respectively, on the flow distribution
behavior of two-phase mixture with long bubbles in the main
generated by mixer 1. The results were plotted in the x-y
plane of WFG3 G1 and W,,3/17, that imply the fractions of the
gas and liquid take-offs through the branch, respectively. In
these plots, the diagonal line indicates the same mass quality
at the run and the branch (i.e., x3/x1 = x2/x1). Thus the points
above this line indicate the flow with the lower quality than
that in the main while the points under the diagonal line
indicate the higher quality flow compared to the main flow.
Therefore, larger deviation from the diagonal line means


rBubbly



Slug(Plug)


Churn


01


01



0.1


jG (m/S)


Figure 3: Flow pattern map of Chung and Kawaji (21 1 14) for
horizontal circular channel of 530 pLm inner diameter
























~T~--~uR


S--O--jL1 =0.31 m/s, case A
-V--9--j, = 0.86 m/s, case B
L19




1 .V3




Main, 1~ = 00 un, 2
,

#6 D= 600 pm Bah,3'





7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


Flow direction Liquid slug

Main Run
Plug bubble I






1 mm ;~Branch


Flow direction Liquid slug


Run


SPlug flow


1 mm Branch
(a) Case G, WG3 WG1 = 0.60, WL?3/171 = 0.25


Min


1 mm A Branch
(b) Case G, WFG3 WG1 = 0.84, 1713/1711


Branch


=0.35


Figure 5: Images of short plug flow distribution

and by Wren et al. (2005) with the compact passage size
junctions.
One important thing to note is on the effect of the channel
size on the flow distribution. As shown in Fig. 8, for similar
conditions of gas and liquid superficial velocities, the results
of flow distribution of two-phase mixtures with long plug
bubbles (about 100 times the channel diameter) at the
compact passage size T-junction by Wren et al. (2005)
appears almost the same with the present case (with the
meso-scale T-junction), although the pressure at the junction
was different from their experimental condition (148 kPa).
This disagrees with the comparison between the macro-scale


MainRu




Plug flow


1 mm I Branch


Figure 4: Sequential images of long plug flow distribution
(Case C, WG3 WG1 = 0.58, WL?3/171 = 0.52)

larger mal-distribution of the two-phase mixture. Also, the
points near the upper right-hand corner indicate larger flow
rates through the branch which can be attained by opening
the branch-side valve largely (with the run-side valve
opening smaller to maintain the junction pressure the same).
The points near the lower left-hand corner are for the
opposite valve opening conditions at the run and at the
branch.
As shown in Figs. 6 and 7, for two-phase flows with long
bubbles, the distribution appears rather even, although the
fraction of gas take-off is somewhat larger than that of the
liquid take-off. Marginally larger fraction of the gas take-off
is due to be the larger momentum of the liquid phase in the
run direction (compared to the gas phase) as explained by
Azzopardi and Whalley (1982).
As the liquid superficial velocity is increased from 0.31
m/s to 0.86 m/s (about 2.8 times larger) with the gas flow
rate fixed O'G1 = 1.0 f 0.01 m/s), the fraction of gas take-off
also increases but only slightly, as exhibited in Fig. 6. Effect
of the gas superficial velocity on the flow distribution is not
discernible although the gas superficial velocity was
increased about 4.7 times (from 0.26 m/s to 1.22 m/s), as can
be seen in Fig. 7. The mnsensitiveness of the hiquid and gas
superficial velocities was reported also by Arirachakaran
(1990) and Buell et al. (1994) with the macro-scale junctions


0.8



0.6



0.4



0.2



0.
0.(


0


0.2 0.4 0.6 0.8 1.0


Fraction of gas separated (WG3/WG1)

Figure 6: Effect of liquid superficial velocity on flow
distribution O'G1 = 1.0 f 0.01 mI/s)


Main


SRun

Bubbly-plug
Flow
























































-- jG = 1.14 m/S, Dh = 5 mmWren et al. (2005)
-- jG1 = 1.01 m/s, Dh = 0.6 mm, present work, case A







I~







Aain, 1( Run, 2
..

Branch, 3


Mixer
SI -- o Case C --- Case E


0 1 I I I


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


. .bubbles) were plotted in the same figure. Obviously, the
trend of mal-distribution is more pronounced with the shorter
----- jG1 = 0.26 m/s, case C ,bubbles even though the superficial velocities of gas and
G1~-- = 0.75 m/s, case D Ad liquid are almost the same: compare cases C and F or E and
G1~-- c = 1.22 m/s, case E : G. More specifically, larger amount of gas is separated
~~ I through the branch with the shorter plug bubbles (except for
---6GWG El Another thing to note is on points a, b and c in Fig. 9
'9;: which imply separation of the liquid phase through the
branch (point a) or the run (points b and c), evidenced by the
/ ;, photographs of the liquid-phase separation phenomenon in
.: Fig. 10: no air bubbles are found in the branch (Fig. 10(a)) or
Alin 1h= 600 C u, in the run (Fig. 10(b) and (c)). According to Menetrier-
.':~Deremble and Tabeling (2006), the distribution behaviour of
,I ~ ...t~immiscible liquid-liquid two-phase flow at micro-scale

600 pmjunctions mostly depends on the channel geometry including
.4. = 60 pm Branh, 3 the channel size and the angle between the main and the
"branches, and the effect of the disperse-phase length was not
1.0 0.2 0.4 0.6 0.8 1.0
counted seriously. However, the present observation shows
Fraction of gas separated (WG3/WG1) that the plug bubble length (disperse phase) plays an
important role in flow distribution and phase separation of
': Effect of gas superficial velocity on flow gas-liquid mixtures at meso-scale T junction. Insignificance
,n (jL1 = 1.0 f 0.03 m/s) of the disperse phase length in the study of Menetrier-
Deremble and Tabeling (2006) is probably due to the
impactt passage size systems, where the effect of the differences in the size scale of the channels and in the density
ize was substantial (with the slug flow by Wren et of the disperse phase (from ours), which is left as a different
)), as mentioned in the introduction part. This is subject for the future study.
d to be due to the predominance of the capillary Differences in fractions of gas and liquid take-offs
:r the gravitational force once the channel size between cases F and G in Fig. 9 are due not only to the
smaller than about 5 mm in hydraulic diameter. superficial gas velocity but also to the bubble length, where
the bubbles of case G is longer than those of case F by about
ion of two-phase flow with short plug bubbles 50 %.


1.0







S0.6



.C 0.4


S0 2
.


0.0
0


For two-phase flows with short plug bubbles, most of the
data points (i.e., all the points except for several in the range
of WFE3/WF1 < 0.2 for case F) fall on the region under the
diagonal line of Fig. 9, similar to the long-bubble cases. For
the comparison purpose, cases C and E (with long plug


Figure 7
distribution

and the co
channel sj
al. (2005)
considered
force ove
becomes

Distribute


Conclusions

In the present study, distribution of two-phase plug flow at
meso-scale T-junction was examined through a series of
experiments, especially by changing the plug bubble length


1.0



S0.8



S0.6






S0.2



0 0


1.0 I


- Aain, 1~h Run, 2 C


Dhz=600~u pm
Branch, 3 ,'


,d

..e- A


S0.8




S0.6




S0.2


.
0.0


0I


.
0.0


0.2 0.4 0.6 0.8 1.0


0.2 0.4 0.6 0.8 1.0


FFROMOH Of gas separated (WG3/WG1)

Figure 9: Effect of plug length on flow distribution (iL1
1.0 0.03)


Fraction of gas separated (WG/WG1)

Figure 8: Comparison of present data with that of Wren et
al. (2005) (jL1 = 0.31 mI/s)









Flow direction Liqudslg



Plug bubble


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

This work was supported by the Korea Research
Foundation Grant funded by the Korean Govemnment
(MOEHRD, Basic Research Promotion Fund) (KRF-2008-
314-D00045), and Energy Resource R&D program
(2009T100200142) under the Ministry of Knowledge
Economy, Republic of Korea.

References

Arirachakaran, S., Two-phase Flow Splitting Phenomenon at
a Regular Horizontal Side-arm Tee, Ph.D. thesis, University
of Tulsa (1990)

Azzopardi, B. J. and Whalley, P. B., The Effect of Flow
Patterns on Two-phase Flow in a T Junction, Int. J.
Multiphase Flow, Vol. 8, pp. 491-507 (1982)

Buell, J. R., Soliman, H. M. and Sims, G E., Two-phase
Pressure Drop and Phase Distribution at a Horizontal Tee
Junction, Int. J. Multiphase Flow, Vol. 20, pp. 819-836
(1994)

Chung, P. M. and Kawaji, M., The Effect of Channel
Diameter on Adiabatic Two-phase Flow Characteristics mn
Microchannels, Int. J. Multiphase Flow, Vol. 30, pp. 735-761


Coleman, J. W. and Garimella, S., Characterization of Two-
phase Flow Pattemns in Small Diameter Round and
Rectangular Tubes, Int. J. Heat and Mass Transfer, Vol. 42,
pp. 2869-2881 (1999)

Lee, C. Y. and Lee, S. Y., Influence of Surface Wettability on
Transition of Two-phase Flow Pattemn in Round Mini-
channel, Int. J. Multiphase Flow, Vol. 34, pp. 706-711 (2008)

Lee, C. Y. and Lee, S. Y., Formation of Gas and Liquid Plugs
at a Merging Y-junction, Proc. KSME Spring Conference,
Busan, Korea, pp. 163-166 (2009)

Lee, J. K. and Lee, S. Y., Dividing Two-phase Annular Flow
within a Small Vertical Rectangular Channel with a
Horizontal Branch, Proc. of the 3r Int. Conf. on Compact
Heat Exchangers and Enhancement Technology for the
Process Industries, Davos, Switzerland, pp. 677-713 (2001)

Lee, J. K. and Lee, S. Y., Assessment of Prediction Models
for Dividing Two-phase Flow at Small T-junction, Int. J. of
Heat Exchangers, Vol. 6, pp. 217-234 (2005)

Mehendale, S. S., Jacobi, A. M. and Saha, R. K., Fluid Flow
and Heat Transfer at Micro- and Meso-scales with
Application to Heat Exchanger Design, Appl. Mechanics
review, Vol. 53, pp. 175-193 (2000)

Menetrier-Deremble, L. and Tabeling, P., Droplet Breakup in
Microfluidic Junctions of Arbitrary Angles, Physics Review
E, Vol. 74, 035303 (2006)

Reimann, J., Brinnkmann, H. J. and Domanski, R., Gas-
liquid Flow in Dividing Tee-junction with Horizontal Inlet
and Different Branch Orientations and Diameters,
Kernforschungszentrun Karlsruhe, Institute fur


Liquid slug Liquid-only flow


Main Run
Plug bubble




1 mm l Branch
(c) Point c (WG3W, G1 3, W,, 1 = 0.42, case G)

Figure 10: Phase separation phenomena in flow distribution
at points a c indicated in Fig. 9

in addition to the gas and liquid flow rates. With long plug
bubbles, two-phase mixture was split to the branch and the
run rather evenly, and the distribution behaviour appeared
insensitive to the liquid and gas superficial velocities in the
main. However, as the plug bubbles become shorter
substantial mal-distribution was observed. Moreover, even
the phase separation phenomenon was observed depending
on the flow rates of gas and liquid through the run and the
branch. This implies the usefulness of T-junction for the
phase-separation purpose in meso-scale systems because of
its simplicity in geometrical shape.
So far, the two-phase flow distribution at T-junctions has
been described mostly in terms of the gas and liquid flow
rates, properties of fluids and the channel sizes. However, for
the meso-scale systems where the plug flow regime
predominates, the plug bubble length (determined by the far
upstream conditions) should be counted as another important
parameter in predicting the flow distribution behaviour. The
quantitative relationship between the flow distribution and
the plug bubble length is left as a future work.

Acknowledgements


SLiquid-only flow

1 mm

(a) Point a (WG3 TF G1 = 0, 6FF 3/FFL1 = 0.16: case F)

Liquid-only flow

S_i n
Main Rn





1 mm V Branch

(b) Point b (WG3 WG1 i L3 WL1 = 0.77, case F)






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

Wren, E., Baker, G, Azzopardi, B. J. and Jones, R., Slug
Flow in Small Diameter Pipes and T-junctions, Exp. Thermal
and Fluid Sci., Vol. 29, pp. 893-899 (2005)

Yue, J., Luo, L., Gonthier, Y., Chen, G And Yuan, Q., An
Experimental Investigation of Gas-liquid Two-phase Flow in
Single Microchannel Contactors, Chemical Engineering
Science, Vol. 63, pp. 4189-4202 (2008)

Zun, I., Gregorc, J. and Perpar, M. Two-phase Flow Mal-
distribution in a Manifold System, 6 h Int. Conf. on
Multiphase Flow, Leipzig, Germany (2007)


Reaktorbauelemente Report KfK 4399, pp. 105-119 (1988)

Saba, N. and Lahey, R. T., The Analysis of Phase Separation
Phenomena in Branched Conduits, Int. J. Multiphase Flow,
Vol. 10, pp. 1-20 (1984)

Stacey, T, Azzopardi, B. J. and Conte, G, The Split of
Annular Two-phase Flow at a Small Diameter T-junction, Int.
J. Multiphase Flow, Vol. 26, pp. 845-856 (2000)

Steijn, V., Kreutzer, M. T. and Kleijn, C. R., pL-PIV Study of
the Formation of Segmented Flow in Microfluidic T-junction,
Chemical Engineering Science, Vol. 62, pp. 7505-7514
(2007)