Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Two-phase split of refrigerants at a T-junction

Sang-Jin Tae
a
, Keumnam Cho
b,1,
*
a
Graduate School, Sungkyunkwan University, 300 Chunchun-dong, Changan-ku, Suwon 440-746, Republic of Korea
b
School of Mechanical Engineering, Sungkyunkwan University, 300 Chunchun-dong, Changan-ku,
Suwon 440-746, Republic of Korea
Received 8 September 2005; received in revised form 27 February 2006; accepted 27 February 2006
Available online 5 June 2006
Abstract
The present study experimentally investigated the two-phase ow split of refrigerants at a T-junction. As geometric param-
eters, the direction of the inlet or branch tube and the tube diameter ratio of branch to inlet tube were chosen. As inlet ow pa-
rameters, the inlet mass ux and quality were varied from 100 to 700 kg m
2
s
1
and from 0.1 to 0.9, respectively, for the
condition of distribution header of a multi-pass evaporator in the general refrigeration system. All experiments were performed
for R-22, R-134a, and R-410A. The measured data were compared with the values predicted by the models developed for aire
water or steamewater mixture in the literature. We propose a modied model for application to the reduced T-junction and ver-
tical orientation of tubes. Among the geometric parameters, the branch tube direction showed the largest sensitivity to the mass
ow rate ratio for the gas phase, while the inlet quality showed the largest sensitivity to the mass ow rate ratio among the inlet
ow parameters.
2006 Elsevier Ltd and IIR. All rights reserved.
Keywords: Refrigerant; R22; R134a; R410A; Two-phase ow; Junction; Pipe; Shape; Experiment; Flow; Gas; Liquid
Comportement diphasique des frigorige`nes
dans les raccords en T
Mots cles : frigorige`ne ; R22 ; R134a ; R410A ; E

coulement diphasique ; Raccordement ; Tuyauterie ; Geometrie ; Experimentation ; Debit ;


Gaz ; Liquide
1. Introduction
Dividing T-junction is a common component of piping
systems, especially in the distribution header of a multi-pass
evaporator, or the distribution kit of a system (multi)
air-conditioner. When gas and liquid two-phase ow is mov-
ing in a pipe with a T-junction, it rarely splits with the same
ow distribution and phase separation ratio. The dividing
characteristics of the gas and liquid two-phase ow into
the branch and outlet tubes are complex because of the
many inuencing variables. During the past two decades, re-
searchers have published their experimental data and analyt-
ical models for two-phase owin a T-junction. But, due to its
* Corresponding author. Tel.: 82 31 290 7445.
E-mail address: keumnamcho@skku.edu (K. Cho).
1
IIR B1 member.
0140-7007/$35.00 2006 Elsevier Ltd and IIR. All rights reserved.
doi:10.1016/j.ijrefrig.2006.02.004
International Journal of Refrigeration 29 (2006) 1128e1137
www.elsevier.com/locate/ijrefrig
complexity, most of the results could be applied only within
their experimental range or condition.
The effect of the inlet mass ux and the inlet quality on
the two-phase split in a T-junction was studied by Saba and
Lahey [1], Seeger et al. [2], Shoham et al. [3], Ballyk et al.
[4], etc. Shoham et al. [3], Ballyk et al. [4], Azzopardi and
Whalley [5], and Reimann and Seeger [6] studied the effect
of branch diameter and branch direction on two-phase split
in a T-junction. Prediction methods for two-phase split in
the T-junction were also suggested by Shoham et al. [3],
Azzopardi and Whalley [5], Hart et al. [7], and Hwang
et al. [8], etc.
Even though researchers have reported experimental or
analytical results for the behavior of a two-phase split in
the T-junction, most of the results cannot be applied to re-
frigerant ow. One reason is that all of the prediction
methods were developed by using airewater or steame
water two-phase ow instead of refrigerant ow. The other
reason is that there is little experimental data for two-phase
owof refrigerant in the T-junction that can be used to verify
a developed prediction method.
The purpose of the present study is to perform the exper-
iment for two-phase ow of R-22, R-134a, and R-410A in
the T-junction within the wide range of geometric and inlet
ow parameters. The measured data were compared with the
values predicted by the models provided in the literature to
nd the best-t model. Also, a modied model for vertical
direction of branch or inlet tubes, and for reduced T-junction
will be suggested.
2. Experiments
The experimental apparatus is schematically shown in
Fig. 1. The systemconsists of a test section with a T-junction,
a gaseliquid separator, gas and liquid ow meters, a pre-
heater, a plate heat exchanger and a variable-speed refriger-
ant pump (0e20 kg min
1
). The sub-cooled refrigerant ow
rate was controlled by the variable-speed refrigerant pump.
The inlet mass ow rate of the refrigerant was measured
by the Coriolis-type mass owmeter (0e20 kg min
1
range,
0.01% resolution). The quality at the inlet of the test section
was xed by the pre-heater, which controlled the heat ux.
The two-phase refrigerant discharged from the test section
was separated inside the gaseliquid separator, and the ow
rate of each phase was measured by the turbine-type gas
mass ow meter (0e0.8 kg min
1
range, 0.5% resolution)
and liquid mass owmeter (0e5 kg min
1
range, 0.01%res-
olution), installed after the gaseliquid separator. The refrig-
erant in the branch tube and the refrigerant vapor separated
by the gaseliquid separator were merged, and then sub-
cooled in the plate heat exchanger.
Fig. 2 shows the details of the test sections. The lengths
of the inlet, outlet, and branch tubes were 400 mm, and the
pressure measuring positions were marked. All of the tubes
in the test section were copper tubes.
The experiments were performed under various geomet-
ric and inlet ow conditions. The experimental geometric
and inlet ow parameters and their ranges are presented in
Table 1. The baseline condition (Q 0

, U 0

) consisted
of horizontal inlet and horizontal branch ow (HeH). Based
on the baseline condition, the inlet or the branch ow direc-
tion was changed from horizontal to vertical upward or ver-
tical downward.
The tube diameters of the inlet (outlet) and branch tubes
were also varied. For the case where the inlet diameter was
the same as the branch tube diameter, the tube inner diameter
was changed to 4.95, 8.12 and 11.3 mm. The tube diameter
ratio (D
3
/D
1
) was varied to 1, 0.72, and 0.44 based on the
largest inlet tube diameter of 11.3 mm.
For each geometric condition, the inlet mass ux was
changed to 100, 300, 500, 700 kg m
2
s
1
, and the inlet
quality was changed from 0.1 to 0.9. The test refrigerants
were R-22, R-134a and R-410A. The saturation temperature
was varied from 3.5 to 11.0

C for each refrigerant.
Nomenclature
A cross-sectional ow area (m
2
)
a distance of dividing streamline (m)
C coefcient of two-phase multiplier
for T-junction
D tube diameter (m)
F mass ow rate ratio, M
3
/M
1
G mass ux (kg m
2
s
1
)
j supercial velocity (m s
1
)
M mass ow rate (kg s
1
)
R radius of curvature (m)
T temperature (

C)
u velocity (m s
1
)
X
tt
LockarteMartinelli parameter
x quality
Greek letters
a void fraction
d liquid lm thickness (m)
Q angle of branch tube (degree)
q center angle of owarea inAppendix A(degree)
r density (kg m
3
)
U angle of inlet tube (degree)
Subscripts
1 inlet tube
2 outlet tube
3 branch tube
G gas phase
L liquid phase
1129 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
The measured data for the two-phase ow distribution
and phase separation were presented in terms of the mass
ow rate ratios of gas and liquid phases. They were dened
as follows:
F
G

M
G
3
M
G
1
1
F
L

M
L
3
M
L
1
2
The uncertainty of the experimental data was analyzed
by the method suggested by Moffat [9]. The uncertainty
ranged from 2.3 to 5.1% and 1.5 to 4.8% for mass ow
rate ratios of gas (F
G
) and liquid phases (F
L
), respectively.
3. Prediction model
A prediction model for two-phase ow in a T-Junction in
the literature was developed for a specic ow regime such
as annular ow, stratied ow, etc. For this reason, the ow
5
7
6
11
4
3
1
2
T
10
9
P
T
P
T
T
8
Flow
1. Test section 2. Gasliquid separator
3. Gas mass flowmeter 4. Liquid mass flowmeter
5. Plate heat exchanger 6. Refrigerant port
7. Receiver 8. Filter
9. Refrigerant pump 10. Liquid mass flowmeter
11. Pre-heater
Fig. 1. Schematic diagram of the present experimental apparatus.
75 mm 25 mm
Inlet
(400 mm)
Branch
(400 mm)
Outlet
(400 mm)
Sight-glass
Sight-glass Sight-glass
Fig. 2. Details of the test section.
1130 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
pattern in the test section with a T-junction was examined
rst. The predicted owpatterns of the two-phase refrigerant
(R-22) are shown in Figs. 3 and 4 for horizontal and vertical
ows, respectively. As shown in Fig. 3, most of ow patterns
shown by Taitel and Duklers [10] map for horizontal ow
described annular ow. We also observed this pattern
through the sight-glass installed at the inlet of the test sec-
tion. Hewitt and Roberts [11] map can be applied if the
tube diameter is larger than that of the present study. As
shown in Fig. 4, most of predicted ow patterns were annu-
lar ow patterns, as observed through the sight-glass.
When a gas and liquid two-phase refrigerant owing in
the inlet tube passes the T-junction, many factors affect the
two-phase ow separation through the junction. They might
be the pressure gradient in the junction, gravitational force,
and diameter of the branch tube as well as the ow condition
of each phase entering the junction. The mass ow rate
ratio of gas, F
G
, may be usually different from that of liquid,
F
L
. It can be assumed that there are two dividing streamlines
for gas and liquid phases, which distinguish the ow enter-
ing the branch tube from that exiting to the outlet tube.
These streamlines were used by previous researchers
such as Shoham et al. [3], Hwang et al. [8], etc. Fig. 5 shows
the distance of the dividing streamline for gas and liquid
two-phase ows from the tube wall of the branch tube in
the inlet tube by parameters a
G
and a
L
. The major assump-
tion of the dividing streamline concept is that the gas or
the liquid ow in the area located on the right side of the di-
viding streamlines is diverted into the branch tube. The rela-
tionships between the distances of the dividing streamlines,
a
G
and a
L
, and the mass ow rate ratios, F
G
and F
L
, are
presented in Appendix A.
To nd out the relations between F
G
and F
L
, the mea-
sured data were compared with the values predicted by the
previous models in the literature. Most of the previous
models were developed by using airewater or steamewater
data instead of refrigerant data for a tube with a large
diameter.
Table 1
Experimental parameters and ranges
Parameter Unit Range Baseline case
Refrigerant e R-22, R-134a, R-410A R-22
Inlet mass ux (G
1
) kg m
2
s
1
100, 300, 500, 700 300
Inlet quality (x
1
) e 0.1, 0.3, 0.5, 0.7, 0.9 0.3
Saturated temp. (T
sat
)

C 3.5, 6.0, 8.5, 11.0 8.5
Tube inner diameter (D) mm 4.95, 8.12, 11.3 8.12
Diameter ratio (D
3
/D
1
) e 1, 0.72, 0.44 1
Branch direction (Q)

0

(HeH) 0

(HeH)
90

(HeVU)
90

(HeVD)
Inlet direction (U)

0

(HeH) 0

(HeH)
90

(VUeH)
90

(VDeH)
X
tt
Annular
Bubbly
Plug/
Slug
Wavy
x = 0.1
x = 0.9
R-22
100
300
500
700
D = 11.3 mm
D = 8.12 mm
D = 4.95 mm
G [kg/m
2
s]
{

G
j
G
2
/
(
(

L
-

G
)
D
g

c
o
s

)
}
0
.
5
10
2
10
2
10
-1
10
-1
10
-2
10
10
1
1
Fig. 3. Flow pattern for horizontal ow of R-22 based on Taitel and
Duklers map (T
sat
8.5

C).
1
10
10
5
1 10
2
10
4
10
4
10
3
10
2
x = 0.1
x = 0.9
Churn
Bubbly
-Slug
Bubbly
Wispy
Annular
Annular
R-22
300
500
700
G [kg/m
2
s]

G
j
G
2
[
k
g
/
s
2
m
]

L
j
L
2
[kg/s
2
m]
Fig. 4. Flow pattern for vertical ow of R-22 based on Hewitt and
Roberts map (T
sat
8.5

C).
1131 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
Fig. 6 shows the comparison between the measured data
for R-22 and the values predicted by several previous
models. Prediction by the other models that are not listed
in Fig. 6 was fairly much different due to the different con-
ditions. As shown in Fig. 6, the model by Hwang et al. [8]
showed the best agreement with the present data. The
models by Azzopardi and Whaley [5] and Roberts et al.
[12] under-predicted the F
G
at the xed F
L
because both
models are based on the assumption of same dividing
streamline for both gas and liquid phases even though the di-
viding streamline of the gas phase is actually located far
from the branch side wall than that of the liquid phase.
The models by Shoham et al. [3] and Hart et al. [7] over-
predicted the experimental data. The reason is as follows:
the density ratio of gas and liquid phases of a refrigerant
ow is greater than that of airewater ow. For example,
the density ratio (r
G
/r
L
) for airewater at 0.1 MPa, 25.0

C
is 0.001, while that for R-22 at the saturated pressure of
0.65 MPa is 0.022. As the density ratio was increased, F
G
was decreased for the xed F
L
.
Hwang et al. [8] developed a phenomenological model
for the horizontal T-junction with same diameters based on
different dividing streamlines for gas and liquid phases.
For the separated two-phase ow such as stratied and annu-
lar ows, it was found that the inuence of the interfacial
drag force is relatively small and may be neglected. The
model then simplies to a balance between centrifugal
forces of the two phases as follows:
r
G
u
2
G
R
G

r
L
u
2
L
R
L
3
where the radii of the curvature of gas and liquid phases divid-
ing streamlines are assumed to satisfy the following relation:
R
G
R
L

_
a
L
D
1
_
n
L
_
a
G
D
1
_
n
G
4
The exponent n
k
in Eq. (4) was determined empirically by
Hwang et al. [8] as:
n
k
5 20 exp
_
53
_
a
k
D
1
__
5
The subscript k is G for gas phase and L for liquid phase.
Thus, the relation between the dividing streamlines of gas
and liquid phases is:
_
a
L
D
1
_
n
L
_
a
G
D
1
_
n
G

r
G
u
2
G
r
L
u
2
L
6
According to the two-phase Bernoulli equation, the pres-
sure gain at the junction between the inlet and outlet tubes is
due to the decrease of the ow rate. The two-phase Bernoulli
Dividing
streamline
for liquid
Dividing
streamline
for gas
A
G3
A
L3
a
G
a
L
Branch
side

LG
Fig. 5. Dividing streamlines for gas and liquid two-phase ows.
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
F
L
F
G
Azzopardi and
Whalley (1982)
Shoham et al.
(1987)
Hwang et al.
(1988)
Hart et al.
(1991)
Roberts et al.
(1997)
Measured
R-22, D = 8.12 mm
G
1
= 300 kg/m
2
s, x
1
= 0.3
Fig. 6. Comparison of experimental data with published models.
1132 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
equation between the inlet and the outlet tubes is described
as follows:
DP
1e2

1
2
_
G
2
1
_
x
2
1
a
1
r
G

1 x
1

2
1 a
1
r
L
_
G
2
2
_
x
2
2
a
2
r
G

1 x
2

2
1 a
2
r
L
__
7
The void fraction, a, was calculated from the model by
Graham et al. [13] for the horizontal ow and by Klausner
et al. [14] for the vertical ow.
The pressure change at the junction between the inlet and
the branch tubes is caused by two terms. The one is the mo-
mentum change due to the decrease of the ow rate, which
can be described as the reversible pressure change. The other
one is the irreversible pressure change due to the change of
the owdirection (turning to the branch direction) and due to
orice effect at the entrance of the branch tubes. The pres-
sure change at the junction between the inlet and the branch
tubes is calculated as follows:
DP
1e3

J
DP
1e3

rev
DP
1e3

irr
8
DP
1e3

rev

1
2
_
G
2
1
_
x
2
1
a
1
r
G

1 x
1

2
1 a
1
r
L
_
G
2
3
_
x
2
3
a
3
r
G

1 x
3

2
1 a
3
r
L
__
9
DP
1e3

irr

K
1e3
2
G
2
1
1 x
1

2
r
L
_
1
C
1e3
X

1
X
2
_
10
K
1e3
0:951 F
L

2
0:8F
L
1 F
L
1:3F
2
L
11
The single-phase friction loss coefcient, K
1e3
, was calcu-
lated by Gardels [15] correlation. The C
1e3
was suggested
by Chisholm and Sutherland [16] for the two-phase ow at
the T-type branch as follows:
C
1e3

_
l Cl
_
r
L
r
G
r
L
_
0:5
___
r
L
r
G
_
0:5

_
r
G
r
L
_
0:5
_
12
Chisholm and Sutherland [16] proposed l 1 and C 1.75
for T-type branch ow.
Fig. 7 shows the comparison between the predicted and
measured pressures in the test section. Even though the inlet
ow conditions were same for the cases, the pressure gradient
after the junction was different for each branch direction. The
reason was that the ow distribution rates for gas and liquid
phases were different as the branch orientation was changed.
The predicted values were agreed with the measured data
within the maximum error of 25%. The maximum difference
was mainly due to the ow disturbance within approximately
100 mmof the outlet and branchtubes just after the T-junction.
4. Results and discussions
Fig. 8shows the mass owrate ratioof gas andliquidphases
through the T-junction with horizontal inlet and branch tubes
for various refrigerants. As shown in Fig. 8, the differences of
Fs for different refrigerants were negligible. The momentum
ux ratios (r
G
u
G
2
/r
L
u
L
2
) under the conditions used in Fig. 8
were 0.193, 0.196 and 0.182 for R-22, R-134a and R-410A, re-
spectively. Since the momentum ux was not changed signi-
cantly, the as were not changed greatly, so the Fs did not show
signicant difference. Fig. 9 shows the effect of the inlet mass
ux on Fs for R-22 when the inlet quality was 0.3 in the hori-
zontal inlet and branch tubes with the diameter of 8.12 mm. As
the inlet mass ux was increased, the F
G
at the xed F
L
increased. The momentum ux ratio (r
G
u
G
2
/r
L
u
L
2
) at G
1
of 100 kg m
2
s was 0.373, but decreased to 0.149 at G
1
of
700 kg m
2
s
1
because the gas velocity increased more rap-
idly than the liquid velocity as the inlet mass ux was
increased. As the momentum ux ratio was decreased, gas
owed more easily into the branch tube, and therefore, the
mass ow rate ratio, F
G
, increased at the xed F
L
.
Fig. 10 shows the effect of the inlet quality on Fs for R-
22 when the inlet mass ux was 300 kg m
2
s
1
in the hori-
zontal inlet and branch tubes with the diameter of 8.12 mm.
The mass owrate ratio of gas, F
G
, was decreased at the xed
F
L
, as the inlet quality was increased. Also, it may be ex-
plained by the momentum ux ratio of gas and liquid phases.
As the inlet quality was increased from 0.1 to 0.9, the mo-
mentum ux ratio (r
G
u
G
2
/r
L
u
L
2
) was also increased from
0.115 to 0.502 at the xed inlet mass ux of 300 kg m
2
s
1
.
Fig. 11 shows the effect of the saturation temperatures of R-
22 on Fs for horizontal inlet and branch tubes with the same
diameter of 8.12 mm at G
1
of 300 kg m
2
s
1
and x
1
of 0.3.
As shown in Fig. 11, the prediction model was not sensitive
to the difference of the saturation temperature. The measured
mass ow rate ratios of gas at the xed F
L
were slightly de-
creased as the saturated temperature was increased possibly
due to the increase of the momentum ux ratio by the increase
in the gas density and decrease in the liquid density as the
junction
inlet
(1)
run(2)
branch(3)
0.642
0.645
0.648
0.651
0 200 400 600 800 1000
L(mm)
P

(
M
P
a
)
Case H-H
Measured
Predicted
H-VD H-VU
Fig. 7. Pressure proles in the test section (G
1
300 kg m
2
s
1
,
x
1
0.3, F
L
0.5, D8.12 mm).
1133 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
saturated temperature was increased. The predicted values
were agreed with the experimental data within the maximum
error of 23%.
Fig. 12 shows the effect of the tube diameter on Fs. As
the tube diameter was increased, the values predicted by
Hwang et al. [8] for F
G
were decreased slightly at the xed
F
L
. The values of the measured data were not decreased
signicantly as the tube diameter was increased. The mo-
mentum ux ratio for the tube diameter of 4.95 mm was
0.175 and that for the diameter of 11.3 mm was 0.207. Since
the momentum ux ratio due to the increase of the diameter
was increased less than that due to the variation of the inlet
mass ux or the inlet quality, the Fs were not changed
greatly. The predicted values agreed with the measured
data within the maximum error of 21%.
The case with the different tube diameters of the inlet and
branch tubes could not be considered in the model suggested
by Hwang et al. [8]. Results by Buell et al. [17], Walters et al.
[18], Azzopardi [19], and Wren et al. [20] showed that the
mass owrate ratio of gas (F
G
) at the xed mass owrate ra-
tio of liquid (F
L
) was increased for the reduced T-junction
(D
3
/D
1
< 1). Thus, the modied model for the reduced T-
junction is suggested in Eq. (13) by modifying Eq. (6) and
considering the diameter ratio effect.
_
a
L
D
1
_
n
L
_
a
G
D
1
_
n
G

r
G
u
2
G
r
L
u
2
L
_
D
3
D
1
_
k
13
The k was determined as k 1.25 by utilizing the present ex-
perimental data for refrigerants. The values predicted by
modied model were compared with the measured data for
the reduced T-junction in Fig. 13.
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
R-22
R-134a
R-410A
Predicted Measured
D = 8.12 mm
G
1
= 300 kg/m
2
s, x
1
= 0.3
Fig. 8. Mass ow rate ratio (F) for various refrigerants.
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
Predicted Measured
100
300
500
700
D = 8.12 mm
R-22, x
1
= 0.3
G
1
[kg/m
2
s]
Fig. 9. Effect of the inlet mass ux on Fs.
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
Measured Predicted
0.1
0.3
0.5
0.7
0.9
D = 8.12 mm
R-22, G
1
= 300 kg/m
2
s
x
1
Fig. 10. Effect of the inlet quality on Fs.
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
3.5C
6.0C
8.5C
11.0C
8.5C (Predicted)
D = 8.12 mm
R-22, G
1
= 300 kg/m
2
s,
x
1
= 0.3
Fig. 11. Effect of saturation temperature on Fs.
1134 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
The model developed by Hwang et al. [8] can be applied
only for horizontal tube orientation. It cannot be applied for
a vertical upward or downward inlet and branch tubes with
T-junction. Therefore, a modied model is necessary to
compare the predicted values with the measured data for
the vertical tube orientation.
The modied model for the reduced T-junction with the
vertical tube orientation is suggested in Eq. (14) by modify-
ing Eq. (13) by considering both the angles of the branch
tube (Q) and inlet tube (U).
_
10:52sinQ0:48sinU
a
L
D
1
_
n
L
_
10:65sinQ0:28sinU
a
G
D
1
_
n
G

r
G
u
2
G
r
L
u
2
L
_
D
3
D
1
_
1:25
14
Based on the measured data, the constants on the left hand
side of Eq. (14) showed almost the same values within an ex-
perimental uncertainty for vertical upward and downward
directions for different refrigerants. Since its difference
was within the error range, Eq. (14) is suggested for the
whole experimental range.
Fig. 14 shows the effect of the orientation of the inlet and
branch tubes on Fs. The mass ow rate ratios of gas at the
xed F
L
were large for the vertical upward branch (HeVU)
case, followed by the vertical downward inlet (VDeH) case,
horizontal T-junction (HeH) case, the vertical upward inlet
(VUeH) case, and the vertical downward branch (HeVD)
case. For the vertical upward branch (HeVU) case, the grav-
itational force was acted in the opposite direction to the
branch ow direction at the junction area. Due to the differ-
ence of the densities between gas and liquid phases, the act-
ing gravitational force of liquid was also larger than that of
gas. Thus, the mass ow rate ratio of liquid was rapidly de-
creased for the HeVU case. For the HeVD case, the direc-
tion of gravitational force was parallel with the direction of
branch ow, so a large amount of liquid was extracted into
the branch tubes.
For the VUeHcase, the gravitational force acts in the op-
posite direction to the inlet ow. Around the junction area, the
gravitational force pushes the liquid with the larger density
into the branch tube, while the gas with the smaller density
ows up to the outlet tube. Thus, the mass ow rate ratio of
the liquid phase was increased more than that of the HeH
case. For the VDeH case, the gravitational force acted in
the direction of the inlet ow, so more liquid owed down-
ward into the outlet tube. Thus, the mass ow rate ratio of
the liquid phase was decreased. Similar phenomena were ob-
served for R-22, R-134a and R-410Arefrigerants. The effects
of the properties of the three refrigerants on the two-phase
ow split through the T-junction were not signicant. The
D = 4.95 mm
D = 8.12 mm
D = 11.3 mm
Predicted Measured
D = D
1
= D
3

R-22,
G
1
= 300 kg/m
2
s, x
1
= 0.3
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
Fig. 12. Effect of the tube diameter on Fs.
D
3
/D
1
= 1
D
3
/D
1
= 0.72
D
3
/D
1
= 0.44
Predicted Measured
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
R-22, D
1
= 11.3 mm
G
1
= 300 kg/m
2
s,
x
1
= 0.3
Fig. 13. Effect of the tube diameter ratio on Fs.
H-VU
H-VD
VU-H
VD-H
H-H
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
F
L
F
G
R-22
D = 8.12 mm
G
1
= 300 kg/m
2
s,
x
1
= 0.3
Fig. 14. Effect of the orientation of inlet and branch tubes on Fs.
1135 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
predicted values were agreed with the measured data within
the maximum error of 25%.
The sensitivities of the geometric and inlet ow parame-
ters were presented in Table 2. The sensitivity of parameter
was dened as the maximum difference between reference
values at the F
L
of 0.5 for the baseline cases listed in Table
1 and the value at the different conditions. As shown in Table
2, the branch tube direction was the most sensitive parameter
among the test parameters in the present study. The inlet
quality was the most sensitive parameter among the inlet
ow parameters, while the branch tube direction was the
most sensitive one among the geometric parameters. Table
2 also showed the maximum prediction error between the
measured data and predicted values by the modied model.
5. Conclusions
This present study experimentally investigated the two-
phase ow split of refrigerants at a T-junction. The modied
model that can be applied for the reduced T-junction (D
3
/
D
1
< 1) and the vertical orientation of the tubes were
suggested.
(1) Among the models for the two-phase ow split at the
T-junction in the literatures, the models developed by
Hwang et al. [8] showed the beat agreement with the
measured data at a horizontal T-junction for refrigerants.
(2) Hwang et al.s [8] model was modied for the reduced T-
junction and vertical orientation of the tubes. The values
predicted by the modied model agreed with the mea-
sured data within the maximum error of 25%.
(3) The characteristics of two-phase ow split at the T-
junction were the most sensitive to the branch tube
direction among the test parameters and to the inlet
quality among the inlet ow parameters.
Acknowledgement
This study was supported by Brain Korea 21 Project in
Korea and SFARC at Sungkyunkwan University.
Appendix A
The following equations showed the relationship be-
tween dividing streamlines and mass ow rate ratios for
each phase. The equations were obtained from the geometric
conguration.
A
G
1

p
4
D2d
2
Aa A:1
A
L
1
pdDd A1 a A:2
A
G
3

q
G
2
_
D
2
d
_
2

_
D
2
d
__
D
2
a
G
_
sin
q
G
2
A:3
A
LG
3

q
LG
2
_
D
2
d
_
2

_
D
2
d
__
D
2
a
L
_
sin
q
LG
2
A:4
A
L
3

q
L
8
D
2

D
2
_
D
2
a
L
_
sin
q
L
2
A
LG
3
A:5
q
G
2 cos
1
_
D2a
G
D2d
_
A:6
q
L
2 cos
1
_
D2a
L
D
_
A:7
q
LG
2 cos
1
_
D2a
L
D2d
_
A:8
F
G

A
G
3
A
G
1
A:9
F
L

A
L
3
A
L
1
A:10
References
[1] N. Saba, R.T. Lahey Jr., The analysis of phase separation phe-
nomena in branching conduits, Int. J. Multiph. Flow 10 (1)
(1984) 1e20.
[2] W. Seeger, J. Reimann, U. Muller, Two-phase ow in a T-
junction with a horizontal inlet, Part I: phase separation, Int.
J. Multiph. Flow 12 (4) (1986) 575e585.
[3] O. Shoham, J.P. Brill, Y. Taitel, Two-phase ow splitting in
a Tee junction experiment and modeling, Chem. Eng. Sci.
42 (11) (1987) 2667e2676.
Table 2
Sensitivity of parameters
Parameter Range Sensitivity
(%)
Maximum prediction
error (%)
Inlet ow Refrigerant R-22, R-134a, R-410A 2.5 17
Inlet mass ux (kg m
2
s
1
) 100, 300, 500, 700 23.4 17
Inlet quality 0.1e0.9 43.6 17
Saturated temp. (

C) 3.5, 6.0, 8.5, 11.0 1.8 23


Geometric Tube diameter (mm) 11.3, 8.12, 4.95 4.4 21
Diameter ratio 1, 0.72, 0.44 26.5 17
Branch direction H, VU, VD 198 25
Inlet direction H, VU, VD 181 25
1136 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137
[4] J.D. Ballyk, M. Shoukri, A.M. Chan, Steamewater annular
ow in a horizontal dividing T-junction, Int. J. Multiph.
Flow 14 (3) (1988) 265e285.
[5] B.J. Azzopardi, P.B. Whalley, The effect of ow patterns on two-
phase owinaT-junction, Int. J. Multiph. Flow8(1982) 491e507.
[6] J. Reimann, W. Seeger, Two-phase ow in a T-junction with
a horizontal inlet, Part II: pressure differences, Int. J. Multiph.
Flow 12 (4) (1986) 587e608.
[7] J. Hart, P.J. Hamersma, J.M.H. Fortuin, Phase distribution
during gaseliquid ow through horizontal dividing junctions,
Nucl. Eng. Des. 126 (1991) 293e312.
[8] S.T. Hwang, H.M. Soliman, R.T. Lahey Jr., Phase separation
in dividing two-phase ows, Int. J. Multiph. Flow 14 (4)
(1988) 439e458.
[9] R.J. Moffat, Using uncertainty analysis in the planning of an
experiment, Trans. ASME J. Fluids Eng. 107 (1985) 173e182.
[10] Y. Taitel, A.E. Dukler, A model for predicting ow regime
transitions in horizontal and near horizontal gaseliquid
ow, AIChE J. 22 (1976) 47e55.
[11] G.F. Hewitt, D.N. Roberts, Studies of Two-phase Flow Patterns
by Simultaneous X-ray and Flash Photography AERE-M 2159,
HMSO, 1969.
[12] P.A. Roberts, B.J. Azzopardi, S. Hibberd, The split of hori-
zontal annular ow at a T-junction, Chem. Eng. Sci. 52 (20)
(1997) 3441e3453.
[13] D.M. Graham, H.R. Kopke, M.J. Wilson, D.A. Yashar,
J.C. Chato, T.A. Newell, An Investigation of Void Fraction
in the Annular/Stratied Flow Regions in Smooth,
Horizontal Tubes ACRC TR-144, Air Conditioning and
Refrigeration Center, University of Illinois, Urbana, USA,
1998.
[14] J.F. Klausner, B.T. Chao, S.L. Soo, An improved method for
simultaneous determination of frictional pressure drop and va-
por volume fraction in vertical ow boiling, Exp. Therm.
Fluid Sci. 3 (1990) 404e415.
[15] A. Gardel, Pressure drops in ows through T-shaped pipe-
ttings, Bull. Tech. Suisse Romande 9 (1957) 122e130.
[16] D. Chisholm, L.A. Sutherland, Prediction of pressure gradi-
ents in pipeline systems during two-phase ow, in: Sympo-
sium on Fluid Mechanics and Measurements in Two-phase
Flow Systems, Leeds, paper 4, 1969.
[17] J.R. Buell, H.M. Soliman, G.E. Sims, Two-phase pressure
drop and phase distribution at a horizontal Tee junction, Int.
J. Multiph. Flow 20 (5) (1994) 819e836.
[18] L.C. Walters, H.M. Soliman, G.E. Sims, Two-phase pressure
drop and phase distribution at reduced tee junctions, Int. J.
Multiph. Flow 24 (1998) 775e792.
[19] B.J. Azzopardi, The effect of side arm diameter on phase
split at T-junctions, in: SPE Annual Technical Conference
and Exhibition, Houston, October, 1999, pp. 1e8.
[20] E. Wren, B.J. Azzopardi, S. Rea, Geometric effects on phase
split in a large diameter T-junction, in: Second International
Symposium on Two-Phase Flow Modelling and Experimenta-
tion, Pisa, 1999, pp. 23e26.
1137 S.-J. Tae, K. Cho / International Journal of Refrigeration 29 (2006) 1128e1137

You might also like