Journal of Thermal Vol. 3, No.

1, 2008

Science and
Technology
Performance Evaluation and Optimization of A
Refrigerant Distributor for Air Conditioner*
Shun YOSHIOKA**, Hyunyoung KIM** and Kazushige KASAI**
**Daikin Air-conditioning and Environmental Laboratory, Ltd.,
1304, Kanaoka-cho, Kita-ku, Sakai 591-8511, Japan
E-mail: [email protected]

Abstract
Refrigerant mal-distribution in a distributor located at the inlet of a heat exchanger
used for an air conditioning system plays an important role in the heat exchanger
performance. Distribution performance in the distributor is greatly affected by the
flow conditions as well as the geometrical parameters of the distributor. To clarify
the distribution characteristics, it is essential to know flow rates of both liquid and
vapor states at every branch tube after distribution. This paper proposes a relatively
simple test method, which enables to measure the inlet quality and the flow rate of
refrigerant at every branch of the distributor. By using the proposed evaluation
method, optimization on the geometrical parameters of the distributor was
conducted to reduce the refrigerant mal-distribution. It was confirmed that the
optimized distributor was able to reduce the mal-distribution of the refrigerant over
branches despite the flow condition changes. By the flow visualization of the
refrigerant in the distributor, it was observed that certain amount of liquid
refrigerant remained and swayed unstably at the bottom of the distributor. It is
supposed that the liquid refrigerant behavior in the distributor great affects the
distribution performance.

Key words: Distributor, Two-Phase Flow, Air-Conditioner, Visualization,

Optimization

1. Introduction

1.1. Background
In the past few decades, the decrease of carbon dioxide (CO2) emission has been of
great issue to reduce the global warming problem. Regarding this concern, it has become
essential to reduce energy consumption in residential use, which gradually increases with
time. Since the energy consumption for air conditioning systems comprises about one third
the total energy for residential use, many studies have been conducted aiming at improving
the coefficient of performance (COP) of air conditioning systems. As a result, the COP of
the separated type of residential air conditioners in Japan is reaching beyond the value of
6.0, which is almost twice as high as that of ten years ago. This has mainly resulted from the
improvement of the air conditioning system’s main components; such as compressors, fans
and heat exchangers.
The fin and tube heat exchanger, which are composed of round copper tubes for
refrigerant flow passage and aluminum fins for air passage, is most widely used for
residential air conditioning systems. To improve the performance of such a type of heat
*Received 5 Sep., 2007 (No. 07-0504) exchanger, a great deal of research has been conducted, focusing on the development of the
[DOI: 10.1299/jtst.3.68]

68
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

slit or louver pattern of the fin and groove shape of the tube. On the other hand, in recent
times, the diameter of the tube tends to be smaller and now reaches almost 5mm in outer
diameter to reduce airside pressure drop and obtain a higher heat transfer coefficient of
refrigerant flow. However, a small diameter tube also causes a large pressure drop of
refrigerant flow, which leads to decrease in heat transfer performance. Therefore, to reduce
the pressure drop of the refrigerant flow, multiple numbers of parallel refrigerant passages
are commonly employed by using a distributor. It is well known that the performance of the
heat exchanger is greatly affected by the refrigerant distribution, in which good distribution
performance of the refrigerant distributor in wide operating condition is essential in
maximizing the heat transfer performance of the heat exchanger.
This paper aims at clarifying distribution phenomena in a cylinder type refrigerant
distributor that is widely used for the middle capacity of residential air conditioning units in
Japan. To clarify the distribution phenomena, we propose a unique and simple experimental
method, in which the inlet quality and the mass flow rate of every branch tube can be
obtained. By using the orthogonal experiment matrix, parameter study on the geometry of
the distributor is conducted and the optimized shape of the distributor is proposed to obtain
good distribution performance. The refrigerant flow behavior in the distributor is observed
by the visualization technique to compare what takes place in the conventional and
optimized distributors.

1.2. Previous Test Method

To evaluate the distribution performance of the vapor-liquid two-phase flow, many
measuring data points, such as the refrigerant flow rate of each liquid and the vapor phase at
every branch tube in the distributor is required. The evaluation methods for the distribution
performance employed by previous researchers can be categorized into three types: (a)
First, instead of using refrigerant as a working fluid, either an air-water or a nitrogen-water
mixture, are used. The flow rates of the gas and liquid are measured, respectively, at every
branch tube after the distributor(1). This method is relatively simple to test due to the
simplicity of the separation of the gas-liquid phase. However, the density ratio of water to
air is almost 30 times larger than that of the liquid to vapor phase of the Freon refrigerant
used for the actual air conditioning system. This will cause large differences in flow regimes
and behaviors in the two-phase flow, which may result in different distribution
characteristics. (b) Second, the refrigerant at every branch tube after the distributor is heated
by electric heater to produce a single-phase vapor with a certain super-heated state. By
comparing the heater inputs at every branch tube, the distribution performance can be
evaluated(2). This method also allows simple set up of the test apparatus. The heater inputs
at each branch tube are only measured data and therefore do not provide information for the
refrigerant state at the inlet of each branch tube, such as flow rates of each phase, however
can be useful for the confirmation of the presence of the critical mal-distribution at the
distributor. (c) In the final evaluation method, the liquid and the vapor flow rates of the
refrigerant at every individual branch tube after distributions are measured, respectively. In
this case, the number of flow meters and liquid-vapor separators corresponding to the
number of branch tubes are required(3,4). This method is the most precise and gives enough
information to understand the distribution phenomena. However, the test set up becomes
quite complicated. Furthermore, many flow rate measuring sensors or switching of the
circuit for the flow measurement are required. Consequently, when the number of branch
tubes increase, using the third method (c) becomes expensive and/or time consuming in
terms of the test setup.
In this paper, an evaluation method with a relatively simple experimental setup based
on method (b) is proposed, which allows that the refrigerant states, such as the quality and
the flow rate, at every branch tube can be obtained.

69
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

Nomenclature
C : correction constant
Cp : specific heat, kJ/kg K
D : diameter, mm
F : statistic F
L : length, mm
M : mass flow rate, kg/h
Pr : pressure of refrigerant, kPa
Qh : heater input, kW
Qr : heat transfer rate of the refrigerant, kW
S : square summation
Tsh : super-heated temperature, K
V : variance
x : quality
∆hs : enthalpy difference between saturated liquid and vapor, kJ/kg
θ : inclined angle, deg.
ρ : sensitivity
Σ : average value of the standard deviation of all branch tube
σ : standard deviation
Subscripts
d : distributor
o : entrance of the distributor

2. Experiment

2.1. Experimental Apparatus

Figure 1 shows the schematic diagram of experimental apparatus used in this study. The
experimental apparatus mainly consists of a refrigerant supply system, a test distributor and
the heating section. The refrigerant supply system supplies the refrigerant (R410A) into the
Power Supply
A
& Power Meter V ～
Heating Section (×4）
Header
ΔP T
Test Section
Path-A

ΔP T

Path-B

P T
ΔP T

Path-C

ΔP T

P
Path-D
T

from Refrigerant Supply Source to Refrigerant Supply Source

(control pressure P, flow rate M, and quality x )

Fig. 1 Experimental apparatus

70
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

test distributor while maintaining a certain pressure, quality and flow rate. The refrigerant in
each path is heated by each electric heater up to having a certain fixed super-heated
temperature by controlling heater inputs of every path, individually. The vapor refrigerant
returns to the refrigerant supply system through a header. The heating section consists of
copper tubing with a 7mm outer diameter and 4.2m length, and a tape heater winding the
copper tubing surrounded by insulating material with a 30mm thickness. The four branch
tube paths (A, B, C and D) are prepared. The heater input, temperature of the refrigerant at
the outlet of each path and the pressure drop in each heating section are measured by power
meter, T-type thermocouples and differential pressure sensors, respectively.

2.2. Assessment Method

To obtain the refrigerant state at every branch tube inlet, which is at the point after the
distribution takes place, the map of the relation between the quality and the flow rate in
terms of the power input and the pressure drop at various flow conditions are prepared.
Before installing the distributor, the refrigerant is supplied to only one certain path among
the four while varying the refrigerant flow rate and inlet quality, when the heater input and
pressure drop are measured at each test point. In this preliminary test, the outlet pressure
and the super-heated temperature of the refrigerant are fixed at 991.8 kPa and 3 K,
respectively.
The measured data is shown in Fig. 2. The refrigerant flow rate and the inlet quality can
be correlated with the heated input and the refrigerant pressure drop as described by the
lines. For all other paths, similar tests are conducted and it is confirmed that the difference
of the data can be small enough to ignore the individual differences. Therefore, the
correlation curve of the map can be obtained with the data of all paths as follows:
Qh = C1 ⋅ Qr (1)

[
Qr = ∆hs ⋅ (1 − x) + c p ⋅ Tsh ⋅ ] 3600
M
(2)

where C1 is a correction factor to adjust the heat balance, which is ranged from 0.95 to
1.02 and expressed by a function of the flow rate and the quality based on the experimental
data. This correction factor includes heat loss from the heater to environmental and
uncertainties of the every measured data, such as flow rate, temperature and inlet quality of
refrigerant.
By the similar way, the pressure drop is correlated as;
∆Pr = C2 ⋅ x 0.148 M 1.8 (3)
where C2 is 0.0771 in present experimental apparatus and experimental condition.

2.5 x =0.1 x =0.2

x =0.3
2.0 x =0.4

x =0.5
1.5
Qh [kW]

1.0 M =40 Exp.

x≒0.1
M =30 x≒0.2
0.5 x≒0.3
M =20 x≒0.4
x≒0.5
0.0
0 20 40 60 80
ΔP r [kPa]

Fig. 2 Map of Qh and ∆Pr in terms of x and M

71
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

Table 1. Experimental conditions

Flow rate Quality, Installation

Condition Mo xo angle, θ
[kg/h] [-] [deg]
J1 0.2
J2 100 0.3
J3 0.4 0
J4 75
J5
50 0.3
J6
10
J7 100

With equations (1) ~ (3), the inlet quality and the flow rate of the branch tubes can be
obtained from the measured data of the power input and the pressure drop of the
corresponding branch tube.
The uncertainties of the obtained inlet quality and the flow rate of each branch tube are
very small, respectively, to assess the mal-distribution at the distributor.

3. Result and Discussion

3.1. Conventional Distributor

By using the map shown in Fig. 2, we evaluated the conventional cylinder type of
distributor shown in Fig. 3, which is used in package type air conditioners produced by
Daikin Co. Ltd. Japan. This distributor splits the main refrigerant flow coming from the
entrance tube into four branch tubes. Capillary tubes, each of which is 2.0mm in inner
diameter and 250 mm in length, are connected between the distributor and every path of the
heating section.
Experimental conditions are described in Table 1, in which the refrigerant total flow
rate, the inlet quality and the installation angle of the distributor are varied in consideration
of the actual operating conditions of the residential air conditioner. The outlet pressure of
the refrigerant and the outlet super-heated temperature are fixed at 991.8 kPa and 3 K,
respectively, which are identical to the conditions used for preparing the map in Fig. 2.

Path-A
Path-D Path-B

Path-C

Branch Tube
(to heating section, I.D.2.0,L200 Capillary)

Lc,0 Distributor
Ld,0 Dd,0

Entrance Tube

Lp,0
θ
Refrigerant

M0 , x0
Fig. 3 Test distributor (conventional)

72
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

Figure 4 shows the experimental results. Fig. 4(a) and 4(b) show the non-dimensional
flow rate and inlet quality deviations at the each branch tube, respectively, with respect to
the path and experimental conditions.
It can be found that the refrigerant flow rate ratio at each path varies significantly
according to the change of the refrigerant condition from J1 to J7 in Fig. 4(a). In designing
the heat exchanger path for the actual air conditioner, the refrigerant flow rate of each path
is adjusted to match to the required heat load of the path, by changing the pressure drop of
each path, that is, by changing the diameter or the length of the capillary tube attached
between the distributor and inlet of the path and/or by changing the individual path length in
the heat exchanger. However, Figure 4(a) implies that the certain heat exchanger that is
designed for having maximum performance at certain refrigerant conditions significantly
decreases its performance due to the mal-distribution of the refrigerant conditions at the
distributor, such the condition changes shown in table 1.
In Fig. 4(b), it can be found that the deviation of the inlet quality at each branch tube
also varies according to the path and refrigerant conditions. Comparing Fig. 4(a) and (b),
the mass flow rate decreases with increasing the inlet quality of the branch tube. Since each
test path at the heating section consists of the same tube length, the refrigerant in high
quality at the branch tube inlet induces the high pressure drop of the test tube due to the
high refrigerant velocity, which results in the decrease of the mass flow rate at that path.
This implies that the non-uniformity of the inlet qualities over the branch tubes induces
mal-distribution of the refrigerant flow rate in the heat exchanger.

1.4 0.2
J1 J1
J2 J2
1.2 J3 0.1 J3
Mi * = Mi /(M 0 /4)

J4
Δxi =xi - x 0

J4
1 J5 0 J5
J6 J6
J7 J7
0.8 -0.1

0.6 -0.2
path-A path-B path-C path-D path-A path-B path-C path-D
(a) Distribution of mass flow rate (b) Distribution of inlet quality

Fig. 4 Experimental results (conventional)

3.2. Optimization of Distributor Geometry

By analyzing the previous test results, it is found that the refrigerant distribution at the
conventional distributor is significantly influenced by the change of the refrigerant flow
conditions that are uncontrollable in the actual operation of the air conditioners. Therefore,
it is required for the distributor geometry that the refrigerant distribution remains stable over
the various operating range.
To find an optimized geometry of the distributor, the experiment is conducted with the
geometry parameters and operating conditions with orthogonal experimental tables. Table 2
shows the experimental parameters and levels. Table 3 describes the orthogonal
experimental table with levels of each parameter. For every geometry case shown in Table
3, the experiment is performed while changing the operating conditions described in Table
1.
The most desirable characteristic for a distributor is that the inlet qualities at every
branch tube will not be influenced despite operating condition changes. Therefore, the
standard deviation of the inlet quality at each path over all operating conditions can be an
index for the evaluation of distribution performance, which can be expressed as:

73
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

Table 2. Levels of each geometrical parameters

Parameters Level 1 Level 2 Level 3

Diameter of Distributor Dd 0.8 Dd,0 1.0Dd,0 -
Length of Distributor Ld 1.0 Ld,0 1.5 Ld,0 2.0 Ld,0
Insertion length of Branch tube Lc 0.5 Lc,0 1.0 Lc,0 1.5 Lc,0
Straight length of Inlet tube Lp 0.1 Lp,0 0.5 Lp,0 1.0 Lp,0

Table 3. Experimental matrix of L9（21×33）

Parameters
Dd Ld Lc Lp
# of exp.
1 1 1 1 1
2 1 2 2 2
3 1 3 3 3
4 1 1 2 3
5 1 2 3 1
6 1 3 1 2
7 2 1 3 2
8 2 2 1 3
9 2 3 2 1

σi = 
 1  n
(
∑ xi ,k − xi )
2
(4)
 n − 1  k =1
where i = path A, B, C, D , n = 7 (conditions)

Since there are four paths in total, the mean value of these standard deviations over the
four paths is selected to evaluate the distribution performance of the distributor, which is
expressed as follows;

1
Σ= (σ A + σ B + σ C + σ D ) (5)
4

The optimum geometry of the distributor can be found by minimizing the mean value
of the standard deviations, Σ.
Table 4 shows the test results of the value of Σ. The well-known statistical method of

Table 4. Results of Σ

Parameters
Dd Ld Lc Lp Σ
# of exp.
1 1 1 1 1 0.031767
2 1 2 2 2 0.010845
3 1 3 3 3 0.009553
4 1 1 2 3 0.017415
5 1 2 3 1 0.028687
6 1 3 1 2 0.015804
7 2 1 3 2 0.069295
8 2 2 1 3 0.031984
9 2 3 2 1 0.027353

74
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

Table 5. Result of ANOVA

Parameters f S V F0 ρ
Dd 1 0.001139 0.001139 72.0 42.84
Ld 2 0.000765 0.000382 24.2 27.97
Lc 2 0.000450 0.000225 14.2 15.97
Lp 2 0.000252 0.000126 8.0 8.40
e 1 0.000016 0.000016 4.82
Total 8 0.002622 100

0.05

0.04
Σ

0.03

0.02

0.01
Level 1 2 1 2 3 1 2 3 1 2 3
Parameter Dd Ld Lc Lp

Fig. 5 Σ with the level of the parameters

analysis of variance technique (ANOVA) is employed with the value Σ, to clarify the
influence of each parameter on the value of Σ. The result of ANOVA is shown in Table 5.
The parameters influencing the value of Σ are mostly the diameter of the distributor, Dd,
which is followed by the length of the distributor, Ld, and inserting length of the branch
tube, Lc.
Figure 5 shows the statistically calculated value of Σ with respect to the levels of the
geometrical parameters. It can be estimated that taking the level 1 of the Dd, level 3 of the
Ld and level 2 of the Lc can minimize the value of Σ from Fig. 5, which will result in most
small variations of the inlet qualities over the all operating conditions. For the confirmation
of the effect at the statistically calculated optimum level of the geometrical parameters, the
new distributor shape is fabricated as described in Fig. 6.

Branch Tube
(to heating section

Lc Distributor

Dd
Ld

Lp
θ
Refregerant

M0 , x 0
Fig. 6 Optimized distributor

75
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

0.2 1.4
J1 J1
J2 J2
0.1 J3 1.2 J3

M i * = Mi /(M 0/4)
Δxi = xi - x 0
J4 J4
0 J5 1 J5
J6 J6
J7 J7
-0.1 0.8

-0.2 0.6
path-A path-B path-C path-D path-A path-B path-C path-D

(a) Distribution of flow rate (b) Distribution of inlet quality

Fig. 7 Experimental results for the optimized distributor

Figure 7 shows the results of the inlet quality and the flow rate distribution with this
new distributor. It is obvious that the variation of these of the each branch tube over the
operating conditions is significantly reduced in comparison with that of the conventional
distributor shown in Fig. 4.

3.3. Visualization
To clarify the phenomena in liquid and
vapor flow inside the distributor, the flow Branch tube
visualization is performed. Figure 8 shows
the sectioned drawing schematic of the test
piece for the visualization. This is
fabricated with polycarbonate resin
allowing observation of the refrigerant
flow inside.
Figure 9 and 10 show the refrigerant
behavior at the conventional and optimized
distributor, and Figure (d) describe the Poly-carbonate
schematic of the observed refrigerant resin
behavior to help the understanding of the
liquid movement. Figure 9 shows the flow
Inlet tube
visualization result for the conventional
distributor. When the refrigerant flow rate,
Fig. 8 Visualization section
Mo, is 50 kg/h, it is found that a certain
amount of liquid remains at the bottom of
the distributor (highlighted area in Figure) while swaying irregularly. The mass of liquid
intermittently rises up to the branch tube dragged by the inlet refrigerant flow supplied from
the inlet tube. However, the mass of liquid is not supplied into each branch tube evenly,
since the intake flow has a certain inclined tendency of direction. As the flow rate increases
[Fig. 9(b) and (c)], the flow changes become more chaotic and the amount of liquid
remaining at the bottom becomes smaller. However, it is still observed that the mass of
liquid is supplied into a certain branch tube irregularly as described in Fig. 9(d).
Figure 10 shows the flow visualization result at the optimized distributor. In
comparison to Fig. 9, the liquid maintaining at the bottom of the distributor is apparently
located far from the branch tube. Unlike the refrigerant behavior in the conventional
distributor, the irregular direct strike of the bulk liquid remaining at the bottom of the
distributor into the branch tube was not observed and it can be observed that the liquid and
vapor are mixed while rising along the distributor, as described in Fig. 10(d). Consequently,

76
Journal of Thermal Vol. 3, No. 1, 2008
Science and Technology

.
(a) Mo = 50 kg/h (b) Mo = 100 kg/h (c) Mo = 150 kg/h (d) Schematic
Fig. 9 Conventional distributor

.
(a) Mo = 50 kg/h (b) Mo = 100 kg/h (c) Mo = 150 kg/h (d) Schematic

Fig. 10 Optimized distributor

it is supposed that the inlet quality at the branch tube is less likely to be influenced by the
liquid remaining at the bottom that is influenced significantly by the intake flow.

4. Conclusion
To clarify the distribution characteristics of the distributor, a unique test method,
allowing to yield the inlet quality and the flow rate at the each branch tube after distribution,
is proposed. By using this proposed evaluation method, it is found that the condition of the
refrigerant introduced into the distributor, such as the quality and the flow rate, significantly
influences the distribution phenomena. The optimization for the geometry of the distributor
is conducted by using the orthogonal experimental matrix. An optimized geometry is
proposed, allowing that the distribution performance is less influenced even if the
refrigerant flow conditions and inclination angle is changed. It is confirmed that the
mal-distribution of the refrigerant at the distributor can be significantly reduced with the
proposed distributor; no matter the change of the intake refrigerant flow condition.

References
(1) J.K. Lee and S.Y. Lee, "Distribution of two-phase annular flow at header-channel
junctions", Experimental thermal and fluid science, Vol. 28 (2004), pp. 217-222
(2) M. Nakayama et al., "Improvement of the distributor performance for the liquid-vapor
two-phase flow in room air conditioner"（in Japanese）, Proc. of 34th JSRAE joint
conference, Tokyo, Japan (2000).
(3) S. Vist and J. Pettersen, "Two-phase flow distribution in compact heat exchanger
manifolds", Experimental thermal and fluid science, Vol. 28 (2004), pp. 209-215.
(4) M. Watanabe, M. Katsuta, K. Nagata and K. Sakuma, "Two-phase flow distribution in
multi-pass tube"（in Japanese）, Trans. Jpn. Soc. Mech. Eng., Vol. 60, No. 580B (1994), pp.
167-172.

77