chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

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

An experimental investigation of phase separation

of gasliquid two-phase ow through a small break
Fachun Liang , Lianmin Song, Yuan Sun
College of Pipeline and Civil Engineering, China University of Petroleum, No. 66 The Yangtze River West Road,
Qingdao 266580, Shandong, China

a r t i c l e

i n f o

a b s t r a c t

Article history:

This paper proposes a specially designed splitting device to study the phase separation of

Received 26 March 2015

gasliquid two-phase ow through a small break. The inner pipe diameter of the main test

Received in revised form 7 June 2015

section is 40 mm. A small hole with 2.5 mm diameter was drilled at the main pipe wall to

Accepted 15 June 2015

simulate the break. Three break orientation angles were tested, including 0 (side), 45

Available online 22 June 2015

(inclined) and 90 (bottom) from horizontal orientation. Experiments were conducted in

an airwater two-phase ow loop with a horizontal test section. Stratied wavy, annular

Keywords:

and slug ows were observed. Experimental results show that phase separation is affected

Two-phase ow

by the break location, ow pattern and gas and liquid supercial velocities. The fraction of

Phase separation

liquid taken off of slug ow is observed much larger than that of stratied wavy or annular

Break

ows due to its particular ow behavior. A simplied correlation of break pressure difference

Flow pattern

is proposed in terms of break outlet mass ow rate and gas quality.

2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1.

Introduction

Gasliquid two-phase ow is widely found in a variety of

applications such as power generation, chemical process,
nuclear energy, and hydrocarbon production industries. When
gasliquid mixture is introduced into a dividing T-junction,
uneven distribution of the phases will inevitably take place,
i.e., the qualities of the two outlets are different, which are
not equal to that at the inlet (Roberts et al., 1997; Stacey et al.,
2000; Mak et al., 2006). This phenomenon is called phase separation. In the last several decades, extensive studies have been
carried out on phase separation at T-junctions (Seeger et al.,
1986; Shoham et al., 1987; Azzopardi, 1999; Mohamed et al.,
2011; Elazhary and Soliman, 2012; Chen et al., 2015).
When two-phase ow passes through a pipe with a break,
severe phase separation may also occur, depending on the
location of the gasliquid interface relative to the break (Welter
et al., 2004; Bartley et al., 2010). If the break is submerged
in liquid phase, liquid will preferentially ow into the break.
The opposite is observed when the entrance is above the liquid level and the discharge is gas predominantly. Zuber (1980)

reviewed the two-phase phenomena at a small branch on the

side of a large reservoir containing stratied layers of gas and
liquid uid phases. He pointed out that, if gas/liquid interface was below the break, liquid may be entrained into the gas
predominating ow through the break. Similarly, gas may be
entrained into the predominant liquid ow in form of vortex
or vortex-free motion when the break is below the gas/liquid
interface.
Prediction of the discharged mass ow rates from a
small break is one of the most important safety issues in
two-phase ow systems (Castiglia and Giardina, 2010). For
instance, light water nuclear reactors (LWRs) during a lossof-coolant accident (LOCA), pipeline networks transferring
hazardous uid, offshore oil-well lines, and chemical batch
or continuous reactors (Reimann and Khan, 1984). Owing
to the inherent complexity of the two-phase ow, it is still
a challenge to accurately predict the discharged mass ow
rate and quality. As mentioned above, phase separation has
a signicant inuence on the gas and liquid ow rates
through the break. Hence, knowledge of phase separation
phenomena at the break is essential for developing a model

Corresponding author. Tel.: +86 15053259392; fax: +86 053286981822.

E-mail address: Liangfch@upc.edu.cn (F. Liang).
http://dx.doi.org/10.1016/j.cherd.2015.06.027
0263-8762/ 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

162

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

Table 1 Experiments of two-phase ow through breaks.

Authors

Simulant

Flow pattern

D (mm)a

d (mm)b

Break structure

Orientation angle

Reimann and Khan

(1984)
Smoglie and
Reimann (1986, 1987)
Yonomoto and
Tasaka (1991)
Maier et al. (2001)
Lee et al. (2007)
Bartley et al. (2008)
Bowden and Hassan
(2011)

Airwater

Stratied ow

206

6, 12, 30

Branch

90

Airwater

Stratied ow

206

6, 8, 12, 20

Branch

90 , 0

Air water

Stratied ow

190(square duct)

10, 20

Branch

90 , 0

Airwater
Airwater
Airwater
Airwater

Stratied ow
Stratied ow
Stratied ow
Stratied ow

255
184
104
50.8

6.35
16, 24.8
6.35
6.35

Branch
Branch
Branch
Branch

0 , 10 , 30 , 60 , 90
0 , 30 , 45 , 60 , 90
0, 30 , 60 , 90
0 , 45 , 90

a
b

Main pipe diameter.

Branch diameter.

that can predict the discharged mass ow rate and quality.

Many experiments have been performed on two-phase ow
discharging through a small break in recent years, including
analysis of airwater or steamwater ows through branches
with different orientations and diameters, as well as various
operating conditions. Typical experimental investigations are
summarized in Table 1.
These publications mainly focused on the onset of gas or
liquid entrainment, but the phase separation mechanism of
gasliquid two-phase ow at the break had not been studied
thoroughly. However, as mentioned above, the knowledge of
the phase separation phenomena involved is extremely vital
for the break prediction model. Besides, the break was simulated by a T-junction, which consists of a small diameter
branch attached to a main pipe with larger diameter or a container. The length of the branch is usually several times of
its diameter. As well known, the branch resistance is different
from that of the break, which may affect gasliquid two-phase
ow discharge characteristics of the break. The branch diameters in previous experimental studies were all larger than
6 mm. Experimental data of break smaller than 2.5 mm is not
available yet. In addition, most of the previous experimental
studies focused on steady stratied ow, which dont reect
the real gasliquid two-phase pipe ow where the annular and
slug ows are common ow patterns. Therefore, through literature review, studies about annular or slug ow discharge
seem to be unavailable at this moment.
The objective of the present study is to experimentally and
theoretically investigate the phase separation characteristics
of two-phase ow discharge through a small break at the pipe
wall. A splitting device was specially designed for experimental study and a 2.5 mm hole was used as the break. The phase
separation inuencing factors, such as the break location, ow
pattern and gas/liquid supercial velocities were studied. In
addition, a correlation was developed to describe the relationship among break pressure difference, break outlet quality and
mass ow rate based on gasliquid two-phase orice equation.

2.

Experimental setup

2.1.

The structure of the small break splitting device

The schematic of the small hole splitting device in the present

study is shown in Fig. 1. The splitting device mainly consists of two sections: Sections 1 and 2. The inner diameter
and wall thickness of the two sections are 40.0 mm and

5.0 mm, respectively. The front of Section 1 and the end of

Section 2 are connected to a gas liquid two phase ow loop.
A circular hole with a 2.5 mm diameter is set on the wall of Section 2 to simulate the small break. The hole is surrounded by
an annular uid receiving room. When gasliquid two-phase
ow passes through the test section, the uid through the hole
will be collected in the uid receiving room and then enter the
side branch. The side branch is connected to a metering separator, where the gasliquid mixture is separated and metered.
The Rosemount pressure transducer and pressure difference
transducers were used to monitor the pressure and pressure
drop at the small hole.
Sections 1 and 2 are connected by ange 1 and ange 2.
A packing plate is placed between the two anges to prevent
leakage. Section 2 can rotate around its axis, which results in
the whole range of angles, 90  90 , could be covered. An
angle indicator is applied to indicate the current location of
the break. The orientation angle of the break is determined
by the plumb line at the dial plate. Three orientation angles,
including 0 , 45 and 90 , were experimentally investigated
in this study.

2.2.

Gasliquid two-phase ow loop

The fractions of gas and liquid taken off are often used to
describe the phase splitting behaviour of gasliquid two-phase
ow. The fractions of gas taken off, KG , and liquid taken off,
KL , are dened by the following equations:
KG =

M3G
M1G

(1)

KL =

M3L
M1L

(2)

where M is mass ow rate in the main pipe 1, kg/s; KG and KL

represent the fraction of gas or liquid taken off; subscripts 1
and 3 denote main pipe 1 and break 3, respectively; subscripts
G and L represent gas and liquid phases.
Once M1G , M1L , M3G and M3L were measured, the fraction of
gas and liquid taken off could be easily obtained according to
Eqs. (1) and (2). Experiments were carried out in an airwater
two-phase ow loop in order to obtain the fractions of gas and
liquid taken off.
Fig. 2 presents the schematic of the gasliquid two-phase
ow loop, mainly consisting of an experimental splitting
device, an outlet tank, an airwater metering separator, a
water circulation pump, an air compressor, a water storage
tank and pipelines.

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

163

Fig. 1 The schematic diagram of the small break splitting device.

The gas mass ow rate (M1G ) and liquid mass ow rate
(M1L ) of the main pipe were measured before mixing. M1G was
metered by a Yokogawa DY040-type vortex ow meter with
an accuracy of 0.75%; while M1L was measured by a Coriolis
mass ow meter made by Emersons Micro Motion Company,
which has an accuracy of 0.5% of the full range.

The splitting device was horizontally installed in the test

section. A short transparent section of the pipe, made of plexiglass, was mounted at upstream of the test section for ow
pattern observation. When airwater two-phase ow entered
into the test section, the uid was divided into two parts. The
main stream owed downstream of the test section, while the

164

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

Fig. 2 Airwater two-phase ow loop. (1) Compressor; (2) DY040-type vortex gas meter gas owmeter; (3) mixer; (4) liquid
owmeter; (5) ow pattern observation section; (6) small break splitting device; (7) branch control valve; (8) metering
separator; (9) liquid level control valve; (10) inverted U-tube; (11) electronic balance; (12) DY025-type vortex gas meter; (13)
main pipe control vale; (14) cyclone separator; (15) water tank; (16) pump.
other small portion of the uid was discharged through the
small break. The main stream then entered a cyclone separator located at the end of the loop, where the gas and liquid
were separated. The gas was vented into the atmosphere and
the water owed back into the water tank for circulation after
being measured.
The two-phase uid owing through the small hole was
collected in the uid receiving room and then went into the
metering separator. The metering separator was a vertical perspex cylinder, whose inner diameter and height were 70 mm
and 350 mm, respectively. A Yokogawa DY025-type vortex ow
meter at the top of the separator was used for gas ow rate
metering, with an accuracy of 0.75% in the full range of
0803 L/min. The liquid mass ow rate through the break M3L ,
was determined based on weightingtime method. The time
for collecting water was within 120600 s, depending upon the
velocity of the water.
All ow meters, temperature transducers, pressure and
pressure difference transducers were calibrated beforehand.
Experimental data are recorded by National Instruments data
acquisition system.

2.3.

The test matrix and experimental procedure

Experiments were performed under the conditions of room

temperature and the outlet pressure of the break close
to atmospheric pressure. Seven gas supercial velocities,
USG = 3.0 m/s, 6.0 m/s, 9.0 m/s, 11.0 m/s, 13.0 m/s, 17.0 m/s and
21.0 m/s, and three liquid supercial velocities, USL = 0.05 m/s,
0.10 m/s and 0.15 m/s were investigated. Three orientation
angles,  = 0 , 45 and 90 , were tested. Fig. 3 presents a
theoretical airwater ow pattern map in a horizontal pipe,
according to Taitel and Dukler (1977). As shown in Fig. 3, the
ow patterns of the current test matrix are predicted as stratied wavy, slug and annular ows.
The experimental procedures are as follows:

(1) The break location was adjusted to a desired orientation

angle. Orientation angles of 0 , 45 and 90 were tested
in the current experiments,

Fig. 3 Experimental data on a ow pattern map.

(2) Gas and liquid ow rates were adjusted to desired values.

The ow patterns were recorded when the ow become
steady,
(3) Different split ratio was obtained by using the main pipe
control valve and the branch control valve together,
(4) The data of M1L , M1G , M3L , M3G , P1 and P13 were measured
and recorded,
(5) The fractions of gas and liquid taken off were calculated
using Eqs. (1) and (2). Then the splitting characteristic
curves of the break were plotted and analyzed.

3.

Results and discussions

3.1.

Effect of break location on phase splitting

Fig. 4(ai) shows the effect of break location on phase splitting.

All data was plotted as the fraction of the gas taken off versus
the fraction of liquid taken off. The equal splitting line in Fig. 4
means that the gas and liquid are distributed at the same proportion. In other words, the quality of uid extracted into the
break is identical to that of the main stream and no phase
separation occurs, namely KG = KL . From Fig. 4, it is found that
the gas and liquid are not discharged evenly and severe phase
separation occurs.

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

165

Fig. 4 Effect of break location on phase splitting

When  = 45 and  = 90 , it is observed that the splitting
curves are above the equal splitting line. It means that the
liquid is prone to enter the break with  = 45 and  = 90 .
On the contrary, as to  = 0 break, the splitting curve is below
the equal splitting line, which means that the gas prefers to
ow through the break with  = 0 and the split is characterized
by gas predominantly taken off.

These phenomena can be explained by the region of inuence theory, which has been proposed for T-junction split
(Azzopardi and Whalley, 1982; Castiglia and Giardina, 2010).
As shown in Fig. 5, the uid extracted into the small break
comes from the zone in the vicinity of the break, which is
called as region of inuence. The inuenced region can be
divided into two sub zones: one is gas inuence region and

166

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

Fig. 5 Region of inuence.

the other is liquid inuence region. hb is the elevation difference between the bulk water level, where the onset of the
liquid entrainment or gas pull-through begins. hb determines
the area of inuence region and can be calculated with the
following equation (Smoglie et al., 1987):
hb =

K(A3 )0.4
g0.2

 P 
13
L G

(3)

bubbles. Alteration of liquid slug and long gas bubble causes

high uctuations of pressure in slug ow. Generally pressure
in the lm zone is much lower compared with that in the liquid
slug zone.
Considering the special ow characteristic of slug ow,
the break discharge process can be divided into two stages,
i.e. the liquid lm leak stage and liquid slug leak stage. During the liquid lm leak stage, the split characteristic is similar
to that of wavy ow. Only a small fraction of gas and liquid
can enter into the break at this stage due to the lower pressure difference of the break. However, when the liquid slug
approaches the break, the local pressure increases rapidly,
which causes the pressure difference across the break is
much larger than that of the liquid lm stage. The slug containing small bubbles will ll the whole pipe cross section
and causes large quantity of liquid ow through the break
at this stage. It was observed during the experiments that
the liquid which enters the break mainly comes from the
liquid slug leak stage. That is why there are more liquid discharge into the break than that of annular ow and wavy
ow.

3.3.
where  is the ow coefcient and is gas expansion
coefcient. A3 is the break sectional area. g is the gravity acceleration. L , G is the density of liquid and gas, respectively.
P13 is the pressure difference across the break. K is constant
and can be determined from the experiments.
Eq. (3) indicates that the area of the inuence region is
dependent on the pressure difference. hb will increase with
the increase of P13 , which leads to an increase of ow inuence area thus more uid will be discharged through the break.
Quality in the ow through the side break is determined by
the void fraction distribution in the inuence zone. When h is
larger than hb , there is no entrainment and only single continuous phase is discharged through the break. When h equals to
hb , onset of entrainment begins. As h becomes smaller than hb ,
two phase ow with gas or liquid entrainment is discharged
through the break.
When  = 0 , the inuence region of gas is much larger than
that of liquid. As a result, more gas ows through the break
and phase splitting data are below the equal splitting line. On
the contrary, for  = 90 and  = 45 , the gas inuence region
is smaller compared with that of the liquid, which results in
more liquid entering into the break and the splitting curve will
be above the equal splitting line.

3.2.

Effect of gas velocity on phase splitting

The effect of gas velocity on phase split of break is also shown

in Fig. 6. For  = 0 break, under annular ow and wavy ow, it
is found that the fraction of liquid taken off increases with
increasing gas supercial velocity. The distribution of gas
and liquid in the main test section contributes to this phenomenon.
When gas velocity is small, the ow pattern is wavy ow
and the liquid tends to ow at the bottom of the pipe, shown in
Fig. 8. With the increase of the gas velocity, the liquid lm will
creep up the side wall of pipe (Zhang et al., 2003). The ow
pattern may transform into semi-annular ow and further
into uniform annular ow at higher gas velocity. As described
above, the uid entering into the break is mainly from the zone
of inuence. The liquid inuence zone area will become larger
at a higher supercial gas velocity than that at a lower gas
supercial velocity. Therefore, more liquid will be extracted
into the break with the increasing of gas velocity.
Contrarily, for  = 45 and  = 90 break, when the gas
supercial velocity increases, the fraction of liquid taken off
decreases. The  = 45 and  = 90 break were covered by a
layer of water. When the supercial gas velocity grows up,
the local liquid lm will become thinner, reducing the area
of liquid inuence zone.

Effect of ow pattern on phase separation

3.4.

As mentioned in Section 2, stratied wavy ow, slug ow and

annular ow were observed in the experiments. An interesting phenomenon can be observed that the split characteristics
in slug ow are extraordinary different from that of stratied
wavy and annular ow, which was illustrated in Fig. 6. When
the fraction of gas taken off remains the same, the fraction of
liquid taken off in slug ow is much larger than that of wavy
ow and annular ows. This phenomenon is mainly due to
the particular ow behavior of slug ow.
Slug ow is one of the most complex ow patterns. Fig. 7
shows schematically the characteristic of a stable slug ow in
a horizontal pipe. A slug unit is classied into two sections: the
liquid slug zone and the lm zone. The lm zone consists of
a liquid lm and an elongated gas bubble, which is similar to
stratied ow. The slug is a section of liquid entrained by small

Critical fraction of liquid taken off

As shown in Fig. 9, a critical value of the fraction of liquid taken

off is observed for 90 break. It can be seen that the fraction of
gas taken off remains zero until the fraction of liquid taken off
reaches the critical value. At a given liquid supercial velocity,
the critical fraction of liquid taken off is affected by gas supercial velocity. This critical value decreases with increasing gas
supercial velocity.
This phenomenon is related to the gas entrainment.
According to Eq. (3), when the break pressure difference is low,
the area of inuence region is small. The break is below the
gas/liquid interface and it is difcult for the gas to penetrate
the liquid layer and enter the break .So the break outlet is lled
by continuous single phase liquid (see Fig. 10(a)). As the pressure difference of the break increase, the area of inuence

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

167

Fig. 6 Effect of ow pattern on phase splitting.

Fig. 7 Schematic diagram of slug ow.

region increases too. When the distance between the break

and the gasliquid interface, h, equals to hb , a thin tube of gas
forms and reaches the vicinity of the break (see Fig. 10(b)). This
critical point is called the onset of gas entrainment. Further
increase of the pressure difference will cause the gas phase
intensively discharge through the break (see Fig. 10(c)).
The local liquid lm thickness decreases with the increase
of gas supercial velocity, which makes it more easily for the
gas to penetrate the liquid lm and enter into the break. Hence,

with the increase of gas supercial velocity, the critical fraction

of liquid taken off reduces.

3.5.
The relationship among pressure difference, mass
ow rate and gas quality
The break can be regarded as an orice. Accordingly, the pressure difference across the break can be calculated based on
gasliquid two-phase ow orice equation. Here, the orice

168

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

Fig. 8 Effect of gas supercial velocity on leak.

Fig. 9 Critical fraction of liquid taken off.

equation proposed by Lin (1982) is used to calculate the pressure difference at the break.

 P 1/2
13

P0

=  + X3




L 1/2
G




(P0 )

1/2

M3 1 4

1/2

Cd A3 (2L )1/2

(5)

(4)

where P13 is pressure drop of the break;  is modication

factor determined by the gasliquid density ratio; X3 is orice
outlet quality; G and L are the densities of gas and liquid,
respectively. P0 represents the pressure drop across orice
assuming total ow to be liquid, and is a function of total twophase ow rate M3 at the same time. P0 can be calculated by
the following equation.

where M3 is the mass ow rate of gasliquid two phase ow

represent the break
through the break; A3 , , L , Cd , and
ow area, the break diameter to pipe diameter ratio, the liquid
density, the discharge coefcient and the thermal correction
factor, respectively.
Combining Eqs. (4) and (5), we could obtain the correlation
for pressure difference across the break, P13 , in terms of mass
ow rate, M3 , and break outlet quality, X3 .

169

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

160000
140000

=0
=45
=90

(p13)^(1/2)/M3

120000
100000
80000
60000
40000
20000
0

0.0

0.2

0.4

0.6

0.8

1.0

X3

(a)Effect of break position on Eq.(8)

140000
120000

=90

Ann
wavy
slug

Sqrt(p13)/M3

100000
80000
60000
40000
20000

Fig. 10 Illustration of mechanisms for gas entrainment.

(P13 )

1/2

M3 1


4 1/2


 + X3

1/2
( Cd A3 ) (2L )



L
G

1/2

0
0.0




0.2

0.4

0.6

0.8

1.0

X3

(b)Effect of flow pattern on Eq.(8)

(6)

Fig. 11 The relationship between

P13 /M3 and X3 .

Eq. (6) can be rewritten as follows:

(P13 )
M3

1/2

1 4

1/2

1/2
( Cd A3 ) (2L )

(L /G )

1/2

 

 1 4

1/2
( Cd A3 ) (2L )

1/2
(7)

X3

Then, Eq. (7) could be simplied to:

(P13 )
M3

1/2

= (A + BX3 )

(8)

4.

where


A=

and 90 break, all the experimental data fall in the same

straight line. It means that Eq. (8) is valid for all the break
orientation angles and the break location has no effect on the

relationship between X3 and P13 /M3 . Fig. 11(b) shows the

effect of ow pattern on Eq. (8). It is also found that Eq. (8) is
not affected by inlet ow patterns.
In fact, Eq. (8) constructs a relationship among pressure
difference, mass ow rate and gas quality. Once coefcient A
and B are known, Eq. (8) can be used with other correlations
to predict the quality and ow rate discharged into the break.

1 4



1/2

1/2
( Cd A3 ) (2L )

B=

(L /G )

1/2

 

 1 4

1/2
( Cd A3 ) (2L )

1/2 

Eq. (8) indicates that there exists a linear relationship

between P13 /M3 and X3 . The value of A, B is functions of

break size, gas and liquid density and can be determined from
experiments.
In the current experimental study, P13 and M3 were
measured and X3 was obtained from M3G and M3L . Fig. 11
demonstrates a good linear relationship between the break

quality X3 and P13 /M3 . A = 9516.5, B = 144283.2 are selected

based on least squares tting and the corresponding R-square
is 0.98. From Fig. 11(a) we can also observe that for  = 0 , 45 ,

Conclusion

An experimental investigation has been carried out to study

the phase separation of a gasliquid two-phase mixture owing through a small break at the pipe wall. According to the
experimental results, the following conclusions can be drawn:
(1) The gas and liquid through the break mainly comes from
the inuencing region. The break location has a great
effect on the phase separation. The liquid prefers to enter
into the  = 45 and  = 90 break. On the contrary, the
gas tends to ow through the  = 0 break and the split is
characterized by gas taken off predominantly.
(2) The splitting characteristic of slug ow is extraordinarily
different from that of annular ow and wavy ow. The fraction of liquid taken off under slug ow is much larger than

170

chemical engineering research and design 1 0 2 ( 2 0 1 5 ) 161170

that of annular ow and wavy ow due to its special ow

behavior.
(3) When the break is below the gasliquid interface, there
exists a critical fraction of liquid taken off. When critical
fraction of liquid taken off is reached, the gas above the
interface can be entrained due to the pressure drop at the
interface produced by the water acceleration in the vicinity
of the break.
(4) A simplied correlation is proposed for break pressure
difference (P13 ) prediction based on Lins Orice equation. The pressure difference (P13 ) is expressed in terms
of break outlet mass ow rate (M3 ) and gas quality (X3 ).
Experimental results conrmed there exists a good linear

relationship between X3 and P13 /M3 .

Acknowledgements
The authors express their great thanks to National Natural
Science Foundation of China (grant no. 51006123), Shandong
Province Natural Science Foundation (grant no. ZR2010EQ016)
and Fundamental Research Funds for the Central Universities
(grant no. 14CX05028A) for nancial support. The authors also
express their great thanks to Dr.Jianjun Zhu for his careful
language revision.

References
Azzopardi, B.J., 1999. Phase separation at T junctions. Multiphase
Sci. Technol. 11 (4), 223329.
Azzopardi, B.T., Whalley, P.B., 1982. The effect of ow patterns on
two-phase ow in a T junction. Int. J. Multiphase Flow 8 (5),
491507.
Bartley, J.T., Soliman, H.M., Sims, G.E., 2010. Experimental
investigation of two-phase discharge from a stratied region
through a small branch mounted on an inclined wall. Int. J.
Multiphase Flow 36 (7), 588597.
Bartley, J.T., Soliman, H.M., Sims, G.E., 2008. Experimental
investigation of the onsets of gas and liquid entrainment from
a small branch mounted on an inclined wall. Int. J. Multiphase
Flow 34 (10), 905915.
Bowden, R.C., Hassan, I.G., 2011. The onset of gas entrainment
from a owing stratied gasliquid regime in dual discharging
branches: Part I: Flow visualization and related phenomena.
Int. J. Multiphase Flow 37 (10), 13581370.
Castiglia, F., Giardina, M., 2010. A semi-empirical approach for
predicting two-phase ow discharge through branches of
various orientations connected to a horizontal main pipe.
Nucl. Eng. Des. 240 (10), 27792788.
Chen, J., Wang, S., Zhang, X., Ke, H., Li, X., 2015. Experimental
investigation of two-phase slug ow splitting at a micro
impacting T-junction. Int. J. Heat Mass Transfer 81, 939948.
Elazhary, A.M., Soliman, H.M., 2012. Two-phase ow in a
horizontal mini-size impacting T-junction with a rectangular
cross-section. Int. J. Multiphase Flow 42, 104114.

Lee, J.Y., Hwang, S.H., Kim, M., Park, G.C., 2007. Onset condition of
gas and liquid entrainment at an inclined branch pipe on a
horizontal header. Nucl. Eng. Des. 237 (10),
10461054.
Lin, Z.H., 1982. Two-phase ow measurements with sharp-edged
orices. Int. J. Multiphase Flow 8 (6), 683693.
Maier, M.R., Soliman, H.M., Sims, G.E., 2001. Onsets of
entrainment during dual discharge from a stratied
two-phase region through horizontal branches with
centrelines falling in an inclined plane: Part 2-Experiments on
gas and liquid entrainment. Int. J. Multiphase Flow 27 (6),
10291049.
Mak, C.Y., Omebere-Iyari, N.K., Azzopardi, B.J., 2006. The split of
vertical two-phase ow at a small diameter T-junction. Chem.
Eng. Sci. 61 (19), 62616272.
Mohamed, M.A., Soliman, H.M., Sims, G.E., 2011. Experimental
investigation of two-phase ow splitting in an equal-sided
impacting tee junction with inclined outlets. Exp. Therm.
Fluid Sci. 35 (6), 11931201.
Reimann, J., Khan, M., 1984. Flow through a small break at the
bottom of a large pipe with stratied ow. Nucl. Sci. Eng. 88
(3), 297310.
Roberts, P.A., Azzopardi, B.J., Hibberd, S., 1997. The split of
horizontal annular ow at a T-junction. Chem. Eng. Sci. 52
(20), 34413453.
Seeger, W., Reimann, J., Mller, U., 1986. Two-phase ow in a
T-junction with a horizontal inlet. Part I: Phase separation. Int.
J. Multiphase Flow 12 (4), 575585.
Shoham, O., Brill, J.P., Taitel, Y., 1987. Two-phase ow splitting in a
tee junctionexperiment and modelling. Chem. Eng. Sci. 42
(11), 26672676.
Smoglie, C., Reimann, J., 1986. Two-phase ow through small
branches in a horizontal pipe with stratied ow. Int. J.
Multiphase Flow 12 (4), 609625.
Smoglie, C., Reimann, J., Mller, U., 1987. Two phase ow through
small breaks in a horizontal pipe with stratied ow. Nucl.
Eng. Des. 99, 117130.
Stacey, T., Azzopardi, B.J., Conte, G., 2000. The split of annular
two-phase ow at a small diameter T-junction. Int. J.
Multiphase Flow 26 (5), 845856.
Taitel, Y., Dukler, A.E., 1977. A model for slug frequency during
gasliquid ow in horizontal and near horizontal pipes. Int.J.
Multiphase Flow 3 (6), 585596.
Welter, K.B., Wu, Q., You, Y., Abel, K., McCreary, D., Bajorek, S.M.,
Reyes, J.N., 2004. Experimental investigation and theoretical
modeling of liquid entrainment in a horizontal tee with a
vertical-up branch. Int. J. Multiphase Flow 30 (12),
14511484.
Yonomoto, T., Tasaka, K., 1991. Liquid and gas entrainment to a
small break hole from a stratied two-phase region. Int. J.
Multiphase Flow 17 (6), 745765.
Zhang, H.Q., Wang, Q., Sarica, C., Brill, J.P., 2003. Unied model for
gasliquid pipe ow via slug dynamicspart 1: model
development. J. Energy Resour. Technol. 125 (4),
266273.
Zuber, N., 1980. Problems in modeling of small break LOCA. In:
Division of Reactor Safety Research. Ofce of Nuclear
Regulatory Research, US Nuclear Regulatory Commission.