i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 2

www. i ifi i r .org

ava i lab le at www.sc iencedi rec t . com

journa l homepage : www.e lsev i er . com/ loca te / i j r e f r ig

Study on two-phase flow pattern of supercritical carbondioxide with entrained PAG-type lubricating oilin a gas cooler

Chaobin Dang*, Koji Iino, Eiji Hihara

Department of Human and Engineered Environmental Studies, Graduate School of Frontier Sciences, The University of Tokyo, 5-1-5

Kashiwanoha, Kashiwa-shi, Chiba 277-8563, Japan

a r t i c l e i n f o

Article history:

Received 25 August 2007

Received in revised form

17 January 2008

Accepted 27 January 2008

Published online 10 February 2008

Keywords:

Carbon dioxide

Transcritical cycle

Experiment

Two-phase flow

Gas cooler

Mixture

Oil

Polyalkylene glycol

* Corresponding author. Tel./fax: þ81 4 7136E-mail address: dangcb@k.u-tokyo.ac.jp (

0140-7007/$ – see front matter ª 2008 Elsevidoi:10.1016/j.ijrefrig.2008.01.014

a b s t r a c t

In order to clarify the heat transfer mechanism of supercritical carbon dioxide flowing with

a small amount of lubricating oil, visualization experiments were conducted using two sight

glass tubes with inner diameters of 2 and 6 mm. The visualization images were recorded us-

ing a high-speed CCD camera with shutter speed changing from 500 to 10,000 pps (pictures

per second). PAG-type oil, which is partially miscible with supercritical carbon dioxide, was

used. The experiments were conducted with lubricating oil concentrations of 1 and 5 wt%,

pressures between 8 and 10 MPa, and mass fluxes between 200 and 1200 kg m�2 s�1.

The visualization images revealed that the two-phase flow pattern inside the gas cooler

was determined by many factors, including the tube diameter, oil concentration, tempera-

ture, pressure, and mass flux. For a small size tube of 2 mm ID, the formation of both oil drop-

lets in the bulk region and an oil film along the inner wall of the tube was confirmed. At low

temperatures, a large number of oil droplets were observed flowing with CO2 with a slip ratio

of approximately 0.7. With an increase in the bulk temperature, both the dimension and

number of oil droplets entrained with CO2 decreased, and the flow of the oil film became

clearly visible. For a large tube of 6 mm ID, the flow pattern at a low mass flow rate was a sep-

arated wavy flow; with an increase in mass flow, the flow pattern changed to annular flow,

which corresponds to a distinct decrease in the heat transfer coefficient due to the heat re-

sistance of oil layer. The transition of the flow pattern is considered due to the shear stress

between the oil layer and bulk CO2. In addition, the dissolution of CO2 into PAG oil as well as

the change in the solubility and thermodynamic properties of CO2 with the temperature and

pressure makes the prediction of the flow patterns a challenging task. The relationship be-

tween the flow pattern and heat transfer characteristics was also discussed.

ª 2008 Elsevier Ltd and IIR. All rights reserved.

4647.C. Dang).er Ltd and IIR. All rights reserved.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 21266

Etude sur la configuration de l

’ecoulement diphasique dudioxyde de carbone supercritique avec de l’huile lubrifiante detype PAG entraınee dans un refroidisseur a gaz

Mots cles : Dioxyde de carbone ; Cycle transcritique ; Experimentation ; Ecoulement diphasique ; Refroidisseur de gaz ; Melange ; Huile ;

Polyalkylene glycol

Nomenclature

A heat transfer area, m2

cp specific heat, J kg�1 K�1

d diameter, m

DT temperature difference, �C

G mass flux, kg m�2 s�1

P pressure, MPa

q heat flux, W m�2

T temperature, �C

W weight, kg

x oil concentration, wt%

Greek symbols

a heat transfer coefficient, W m�2 K�1

Subscripts

1 inlet

2 outlet

CO2 CO2

bulk bulk

oil oil

wall wall

water water

1. Introduction

In an actual heat pump cycle, lubricating oil is commonly

used in the compressor for lubrication, cooling, and sealing

purposes. However, small amounts of lubricating oil may

be discharged by the compressor along with the refrigerant;

this oil would flow through the heat exchangers and expan-

sion devices and finally return to the compressor. Therefore,

a mixture of the refrigerant and a small amount of oil,

instead of pure refrigerant, flows inside the heat exchanger

and exchanges heat with the environment. Due to the excel-

lent lubricity and stability in the transcritical CO2 environ-

ment, PAG (polyalkylene glycol) type oils are widely

applied in the CO2 heat pump cycle (Kaneko et al., 2006). It

is, therefore, important to understand the heat transfer

performance of supercritical CO2 with entrained PAG-type

lubricating oils.

Dang et al. (2007) have investigated the heat transfer char-

acteristics of supercritical CO2 with small amounts of

entrained PAG oil. The experimental results showed a de-

crease in the heat transfer coefficient and an increase in the

pressure drop with the entrainment of lubricating oil. The

maximum reduction in the heat transfer coefficient was about

75%, which occurred in the vicinity of the pseudocritical tem-

perature. In addition, the transition of the flow pattern was

found to drastically affect the heat transfer performance.

The changes in the thermophysical properties of supercritical

CO2 with the temperature and pressure are the main parame-

ters used to predict the flow pattern and heat transfer perfor-

mance of supercritical CO2 with lubricating oil; moreover,

a change in the solubility of CO2 into PAG oil would only

make the issue more challenging. To understand the experi-

mental results of the heat transfer performance with oil en-

trainment, information concerning the flow pattern inside

the gas cooler is of significant importance. However, mainly

due to the difficulty in dealing with the high operation pres-

sures, a systematic investigation of the flow pattern under

wide experimental conditions is absent in the literature. The

current available experimental results from open literature

concerning the flow and heat transfer of supercritical CO2 con-

taminated with PAG-type oil include the following: Yun et al.

(2006) measured the heat transfer of supercritical CO2 with

PAG-type oil in microchannel with a diameter of 1.0 mm.

The mass flux was varied from 200 to 400 kg m�2 s�1 at heat

flux of 20 kW m�2, and the test section inlet pressures of CO2

were changed from 8.4 to 10.4 MPa with inlet temperatures

controlled from 40 to 80 �C. By comparing the measured value

at different oil concentration, they concluded that the average

heat transfer coefficient decreased 9% for 2 wt%, and 20.4% for

4 wt%. Gao and Honda (Gao and Honda, 2002) investigated the

heat transfer characteristics of supercritical CO2 with an oil

concentration of 1% inside a 4.5 mm ID tube at a pressure of

7.7 MPa, and reported a maximum decrease of 40% in its

heat transfer performance at pseudocritical temperature

when compared to pure CO2. Mori et al. (2002) observed the

flow pattern of a CO2–oil mixture inside a gas cooler with

a 6 mm ID tube at a pressure of 9.5 MPa. Although the oil con-

centration was unknown, it was found that the oil separated

from CO2 and formed an oil layer near the inner wall. Subse-

quent measurements of the heat transfer coefficient revealed

a decrease in heat transfer, which was explained by the heat

resistance of the oil layer.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 2 1267

In order to clarify the heat transfer mechanism of super-

critical CO2 flowing with a small amount of entrained lubricat-

ing oil, visualization experiments have been conducted using

two sapphire sight glass tubes with ID of 2 and 6 mm. In this

report, the flow pattern inside the gas coolers under different

pressure, mass flux, and temperature conditions is analyzed,

and the relationship between the flow pattern and the heat

transfer performance is discussed.

2. Experimental apparatus and experimentalconditions

2.1. Experimental loop

Fig. 1 shows a schematic drawing of the test loop, which con-

sists of three main components – a main loop, test section,

and oil sampling section. The main loop was a conventional

CO2 heat pump cycle for controlling the temperature and pres-

sure at the inlet of the test section; an oil separator was in-

stalled at the exit of the compressor to control the oil

concentration circulated in the loop. The mass flow rate of re-

frigerant was controlled by adjusting the rotation speed of the

compressor and the opening of the expansion valves. The test

section included a sapphire sight glass tube for visual observa-

tion of the refrigerant flow pattern and a tube-in-tube heat ex-

changer for the measurement of the heat transfer coefficients

under various experimental conditions. The oil concentration

was measured at the sampling section using the shut-down

method.

The test section for heat transfer coefficient measure-

ment was a 0.5 m long horizontal tube-in-tube counterflow

heat exchanger with CO2 flowing inside the inner tube and

cooling water inside the annular passage. The wall temper-

atures and CO2 side temperatures at the inlet and outlet of

the test section, Twall1, Twall2, T1 and T2, respectively, were

measured to define the temperature difference using LMTD

method:

DT ¼ ðT1 � Twall1Þ � ðT2 � Twall2Þln�

T1�Twall1T2�Twall2

� (1)

TP

Compressor

Gascooler

Subcooler

Sightglass

TP

Sampling section for oilconcentration

Expansionvalve 2

Evaporator

Mass flowmeterConstant

temperature bath

Expansionvalve 1

TP

P TP

Oil separator

Test section

Cooling water

Fig. 1 – Schematic of the test loop.

The heat flux q was calculated from heat exchange rate of

waterside as

q ¼ Gwatercpwater ðTwater out � Twater inÞ=A (2)

where A is the heat transfer area, cpwater is the specific heat of

water, and Twater in, Twater out are the temperatures of water at

the inlet and outlet of the test section.

The heat transfer coefficient was determined from the heat

flux q and temperature difference DT as

a ¼ qDT

(3)

A detailed description of the experimental loop and heat

transfer coefficient measurements can be found in the previ-

ous report (Dang et al., 2007).

2.2. Slight glass for flow visualization

The sight glass was made of sapphire in order to endure the

high operation pressure of supercritical CO2. The outside di-

ameter of the two sight glasses was both 30 mm with inner

diameters of 6 and 2 mm, therefore the wall thickness was

12 and 14 mm, respectively. The sight glasses were carefully

assembled, and test results demonstrated that they could

withstand high pressures of up to 20 MPa at an environmen-

tal temperature of 60 �C. A sight window was assembled to

enable the observation of the flow pattern inside the sap-

phire glass. The dimensions of the sight window were

10� 58 mm for the 6 mm ID sight glass and 10� 20 mm for

the 2 mm ID sight glass. Fig. 2 shows the details of the man-

ufactured sight glasses and an enlarged view of the 2 mm ID

sight glass.

The flow pattern inside the sight glass was recorded us-

ing a high-speed CCD camera (MEMRECAM fx-6000, NAC

Corp.) with a maximum shutter speed of 70,000 pps (pictures

per second). Because the CO2–oil droplet entrained flow was

much faster than the oil film, the record was repeated twice

for the same experimental conditions using different shut-

ter speeds. High-speed recording, 10,000 or 6000 pps, was

applied for the observation of the flow pattern over the

bulk regions for the 2 and 6 mm tubes, respectively, and

low-speed recording (500 pps) was applied for the observa-

tion of the flow pattern in near-wall areas. The recorded im-

ages were viewed on a monitor and transferred to

a computer simultaneously. From recorded continuous im-

ages, the velocity of oil droplets and oil film can be calcu-

lated therefore the slip ratio of the oil droplets to bulk CO2

flow, as well as the slip ratio of oil film to bulk CO2 flow,

can be determined.

2.3. Oil sampling method

The oil concentration was measured behind the heat transfer

test section using a 4.35 mm ID tube with a length of 1.5 m.

The inlet temperature of the sampling section was main-

tained at 25 �C. As shown in Fig. 3, when sampling the oil, first

the two three-way valves at the end of the sampling section

were switched to bypass the flow, then the two on/off valves

were switched to shut the mixture of CO2 and oil into the sam-

pling tube. Therefore, the sampling tube can be dismantled

High-speedCCD camera

PC

Monitor

Sight glass

Light

Sight glasses (6 mm ID and 2 mm ID)

Enlarged view of sight glass (2 mm ID) Experimental setup for visual observation

Fig. 2 – Visual observation instruments.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 21268

from the experimental cycle and the oil concentration inside it

was measured.

The weight of the whole sampling tube, with two on/off

valves attached, was measured by a balance with a capacity

of 1.2 kg and a minimum scale of 0.001 g. After that, a needle

valve with known weight was attached to the sampling tube,

and the CO2 inside the sampling tube was carefully released

through the needle valve. This releasing process usually lasts

about 1 h. Then the tube was heated up to about 50 �C by

a dryer in order to ensure that CO2 dissolved inside the oil is

expelled out of the sampling tube. The weight of the needle

valve and the sampling tube, with oil confined inside, was

measured to calculate the weight of CO2 ðWCO2 Þ released

from the sampling tube. Finally, the sampling tube and the

needle valve were cleaned with ASAHIKLIN AK-225 to remove

the attached oil completely. The weight of oil (Woil) was calcu-

lated from the difference of the weight of sampling tube and

needle valve before and after the cleaning. The oil concentra-

tion was defined as the weight of the oil to the CO2–oil mixture

confined inside the sampling tube as follows:

x ¼ Woil

Woil þWCO2

� 100% (4)

With oil concentration of 0.1 wt%, the oil weight inside the

sampling tube was about 0.02 g. Considering the accuracy of

0.001 g of the balance used, the absolute uncertainty of the

oil concentration is less than 0.1 wt%. However, it is very dif-

ficult to ensure a complete separation of CO2 from oil,

Testing

Sampling

Fig. 3 – Oil sampling method.

therefore, the measurement of oil concentration was repeated

three times at each experimental condition, and the differ-

ence of measured oil concentration was confirmed less than

0.1 wt%, which is considered the uncertainty of the oil concen-

tration measurement.

2.4. Experimental conditions

Since PAG oil is partially miscible with CO2 at the supercritical

pressure, the flow pattern of the CO2–oil mixture may vary

with the oil concentration, mass flux, temperature, and pres-

sure conditions. Table 1 denotes the experimental conditions.

Flow visualizations were conducted using 2 and 6 mm ID

tubes at pressures of 8 and 10 MPa, mass fluxes from 200 to

1200 kg m�2 s�1, and temperatures from 20 to 60 �C. Observa-

tions of the change in the flow pattern with oil concentrations

of 1 and 5 wt% were carried out under the above-mentioned

experimental conditions.

3. Visual observation results

3.1. Classification of flow pattern

Fig. 4 illustrates the observed flow pattern of supercritical CO2

flowing with PAG oil under the above-mentioned experimen-

tal conditions. Here, ‘‘V’’ denotes supercritical CO2, ‘‘D’’

Table 1 – Experimental conditions

Tube material Copper

Tube ID 2 and 6 mm

Oil type PAG100

Oil concentration 1 and 5 wt%

Mass flux

(kg m�2 s�1)

200–1200

Pressure (MPa) 8 and 10

V: CO2

D: Oil droplet

F: Oil film

Fig. 4 – Classification of flow pattern. M: mist flow; AD:

annular-dispersed flow; A: annular flow; W: wavy flow;

WD: wavy-dispersed flow.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 2 1269

denotes oil droplet, and ‘‘F’’ denotes oil film. The observed

flow pattern includes the following: (a) mist flow (M), where

a small amount of oil droplets flow with CO2 and no oil film

is observed; (b) annular-dispersed flow (AD), where both oil

droplets and an oil film are observed; (c) annular flow (A),

where no or few oil droplets are observed; (d) wavy flow (W),

where the oil film only exists at the bottom of the cross-sec-

tion; (e) wavy-dispersed flow (WD), where oil droplets are ob-

served flowing along with an oil film at the bottom of the

cross-section.

Some examples of the high-speed camera images are

shown in Figs. 5–7. The white stripe at the centerline is due

to the refraction of incoming light. Since supercritical CO2 is

transparent, the white particles moving along with CO2 are

oil droplets. The flow pattern was found changing with the

T = 25°C

T = 30°C

T = 35°C

T = 40°C

T = 50°C

P = 8 MPa, x = 1

Oil droplet

P =

a b

Fig. 5 – Flow visualization of CO2 with entraine

temperature, pressure, and oil concentration, as well as with

the tube diameter.

3.2. Flow pattern for 2 mm ID tube

Fig. 5 shows a comparison of the flow pattern for the 2 mm ID

tube under different temperature, oil concentration, and pres-

sure conditions. At a low temperature of 25 �C and an oil con-

centration of 1 wt%, it is clear that the flow pattern is mist

flow, with oil droplets flowing along with the bulk CO2 at

a slip ratio of about 0.7 in the flow direction; simultaneously,

the droplets move up and down along the radial direction

due to the disturbing velocity of the bulk CO2 flow. The aver-

age diameter of the oil droplets ranges from 50 to 100 mm. Un-

der this condition, no distinct oil film flowing along the inner

wall is observed.

Since the solubility of CO2 in the oil decreases with an in-

crease in temperature, the viscosity and surface tension of

the oil droplets increase with the temperature. As a result,

the separated oil appears to adhere to the inner wall and

forms an oil-rich layer at a high temperature with 20–60 wt%

CO2 dissolved inside it; this layer is visible as stripes at a tem-

perature of 30 �C, as shown in Fig. 5(a). With a further increase

in temperature, the oil-rich layer becomes much thicker, and

the possible movement of the oil droplets in the bulk region

cannot be observed. The layer moves at a very low speed,

which may lead to a decrease in the heat transfer coefficient

due to its heat resistance.

When comparing the flows at oil concentrations of 1 and

5 wt% at the same temperature, the oil-rich layer can be

seen clearly even at a low temperature of 25 �C under the

high-oil-concentration condition and, at the same time, the

P = 10 MPa, x = 5

Oil film

8 MPa, x = 5

c

d PAG oil; d [ 2 mm, G [ 800 kg mL2 sL1.

T = 25°C T = 30°C

T = 35°C T = 40°C

T = 45°C T = 50°C

a b

c d

e f

Fig. 6 – Comparisonofflow patternatdifferent temperature for6 mmIDtube; d [ 6 mm, P [ 8 MPa,G [ 200 kg mL2 sL1, x [ 1%.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 21270

number of oil droplets flowing along with CO2 in the bulk re-

gion also increases proportionally.

Comparing the flows under pressures of 8 and 10 MPa, no

distinct difference is observed in the flow pattern under low-

and high-temperature conditions. However, in the near-

pseudocritical-temperature region, a sharp contrast in the

flow patterns at different pressures is observed. When com-

paring the flow patterns at pressures of 8 and 10 MPa at

40 �C and an oil concentration of 5%, it is seen that the oil-

rich layer at 8 MPa is much thicker than that at 10 MPa and

the movement of oil droplets is less obvious. The difference

observed is due to the change in the pseudocritical tempera-

ture with pressure, which corresponds to the changes in the

thermophysical properties of CO2 and solubility of CO2 into

PAG oil; the pseudocritical temperature is 34.5 and 44.8 �C

at 8 and 10 MPa, respectively. At 40 �C, the CO2 flow is on

the high-temperature side of the pseudocritical temperature

when the pressure is 8 MPa, the density difference between

CO2 and the oil is large, and the solubility of CO2 into the

oil is low; consequently, the oil tends to separate from the

bulk area and adhere to the inner wall. On the contrary, at

pressure of 10 MPa, the CO2 flow is on the low-temperature

side of the pseudocritical temperature, the densities of CO2

and the oil are close to each other, and the oil layer is easily

transformed into oil droplets.

3.3. Flow pattern for 6 mm ID tube

Fig. 6 shows the transition of the flow pattern with tempera-

ture for the 6 mm ID tube at a low mass flux of 200 kg m�2 s�1.

The flow pattern forms a wavy-dispersed or wavy flow at this

T = 30°C

a

Fig. 7 – Comparisonofflow patternatdifferent temperature for6 m

tube diameter and mass flux. At the lowest temperature of

25 �C, the oil-rich layer flows at the bottom of the cross-

section with large amount of entrained oil. Because the densi-

ties of CO2 and the oil at low temperature are close to each

other and the viscosity and surface tension are low due to

the dissolution of CO2 into the oil, the oil-rich layer is easily

broken up to produce oil droplets flowing near the oil-rich

layer. With an increase in temperature, the solubility of CO2

into oil decreases, and both the viscosity and surface tension

of the oil-rich layer increase; as a result, the oil-rich layer be-

comes stable and the number of oil droplets decreases. At

a temperature greater than 40 �C, only the wavy flow with

the oil film at the bottom is observed.

In Fig. 7, the flow pattern for the 6 mm ID tube at a high

mass flux of 800 kg m�2 s�1 is shown. Unlike the flow pattern

at a low mass flux of 200 kg m�2 s�1, the oil film can be seen

formed at both the upper and bottom inner walls. This annu-

lar flow pattern is confirmed at different temperatures of 30

and 40 �C. Obviously, the wavy flow is unstable at high mass

flux due to the increased turbulent disturbance.

3.4. Comparison of flow pattern at different tubediameters

Previous researches on flow pattern change are focus on tube

diameter larger than 20 mm, the flow pattern at small sized

tube less than 2 mm is not well investigated, and no much

knowledge available for a two-phase flow with a large viscos-

ity and surface tension like lubricating oil. The flow pattern

map proposed for water–air or refrigerant was tested and

the results showed that the previous flow map could not

T = 40°C

b

mIDtube; d [ 6 mm, P [ 8 MPa,G [ 800 kg mL2 sL1, x [ 1%.

d = 2 mm d = 6 mm

a b

Fig. 8 – Comparison of flow pattern at different tube diameter, P [ 8 MPa, G [ 800 kg mL2 sL1, T [ 308C, x [ 1%.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 2 1271

explain the flow pattern change demonstrated in this re-

search. Therefore, an attempt for the flow map prediction

has been scheduled. However, the results are unavailable at

present.

A comparison of the flow pattern at tube diameters of 2 and

6 mm tubes is presented in Fig. 8. The visual observations were

conducted at pressure of 8 MPa, mass flux of 800 kg m�2 s�1,

the temperature and oil concentration is 30 �C and 1 wt%, re-

spectively. It is obvious from comparison that the oil film for

2 mm tube is thicker than that of 6 mm tube, and the oil drop-

lets are much larger with smaller number density. This result

is due to the decreased turbulent disturbance with the de-

crease in tube diameter when compared under the same

mass flux, temperature and pressure condition. The thicker

oil film for small sized tube corresponds to a much more signif-

icant effect of heat resistance. Due to the small viscosity of CO2

and high operating pressure, a small size tube is usually pro-

posed to be used for CO2 heat pump. However, the visual ob-

servation results imply that the formation of oil film and the

corresponding heat transfer deterioration due to the introduc-

tion of lubricating oil should be taken in consideration to

decide the optimum tube dimension.

4. Discussions

The relationship between the flow pattern of CO2–PAG oil and

the heat transfer performance (Dang et al., 2007) is shown in

Figs. 9 and 10 for the 2 and 6 mm ID tubes, respectively.

In Fig. 9, a comparison of the heat transfer coefficients and

the flow pattern for the 2 mm ID tube at different tempera-

tures is shown. The solid line denotes the heat transfer

d = 2 mmP = 10 MPa G = 800 kgm–2s–1

q = 12 kWm–2

9000

8000

7000

6000

5000

4000

3000

2000

1000

020 30 40

Tb

Hea

t tr

ansf

er c

oeff

icie

nt [

W/m

2 K

]

Fig. 9 – Flow pattern of CO2–oil and heat

coefficient of pure CO2 (Dang et al., 2004). It is seen that the

heat transfer coefficient decreases due to the introduction of

PAG oil, and this decrease is most significant at the pseudoc-

ritical temperature. Comparing the curves at 30 and 50 �C,

the drop in the heat transfer coefficient at 50 �C is much larger

than that at 30 �C for an oil concentration of 5%. From the vi-

sual observation, it is found that the flow pattern at low tem-

perature is mist flow with a considerable amount of oil flowing

over the bulk area and a thin oil film. At high temperature,

a thick oil film is observed, which corresponds to a sharp de-

crease in the heat transfer coefficient. This visual observation

implies that the heat resistance due to the formation of the oil

film is the main reason for the heat transfer deterioration. And

the oil droplets flowing in the bulk region do not contribute

significantly to the heat transfer deterioration.

Fig. 10 shows a comparison of the heat transfer coefficient

and flow pattern for the 6 mm ID tube at two mass fluxes, 200

and 800 kg m�2 s�1. The heat transfer coefficient decreases sig-

nificantly at 800 kg m�2 s�1 due to the heat transfer resistance

of the oil film flowing along inner wall. However, the heat

transfer coefficient does not drop significantly at a mass flux

of 200 kg m�2 s�1, even if the oil concentration is increased to

5%. From visual observation, it is found that the flow at the

low mass flux is wavy flow, and the oil film is only observed

at the bottom of the tube. However, at the high mass flux,

the oil film is formed over the entire inner wall, which leads

to a significant drop in heat transfer performance.

It is evident again by comparing the heat transfer coeffi-

cient shown in Figs. 9 and 10 that although the heat transfer

coefficient for a small sized tube is much higher than that of

large sized tube for pure CO2, by introducing the partial misci-

ble lubricating oil, much more significant deterioration of the

50 60 70

ulk [ºC]

0 1 3 5

transfer coefficient for 2 mm ID tube.

G = 200 kgm–2s–1

G = 800 kgm–2s–1

d = 6mmP = 8 MPa G = 200 kgm–2s–1

q = 12 kWm–2

d = 6mmP = 8 MPa G = 800 kgm–2s–1

q = 12 kWm–2

20 30 40 50 60 70

Tbulk [ºC]

20 30 40 50 60 70

Tbulk [ºC]

6000

5000

4000

3000

2000

1000

0

heat

tra

nsfe

r co

effi

cien

t [W

/m2

K]

18000

16000

14000

12000

10000

8000

6000

4000

2000

0

heat

tra

nsfe

r co

effi

cien

t [W

/m2

K]

0 1 3 5

0 1 3 5

a

b

Fig. 10 – Flow pattern of CO2–oil (40 8C) and heat transfer

coefficient for 6 mm ID tube.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 1 ( 2 0 0 8 ) 1 2 6 5 – 1 2 7 21272

heat transfer coefficient was observed for small sized tube

than that of large sized tube, which is the result of the changes

in flow pattern since the oil film for small sized tube may

much thicker than large sized tube. It is therefore revealed

that the effect of lubricating oil should be considered in detail

when the tube diameter is determined.

5. Conclusions

Visual observations of the flow pattern inside a gas cooler were

conducted using two sight glass tubes with ID of 2 and 6 mm.

The results show that the flow pattern is determined by

many factors, including the tube diameter, oil concentration,

temperature, pressure, and mass flux. The main conclusions

of this study are summarized as follows:

(1) For a 2 mm ID tube, mist flow, annular-dispersed flow, and

annular flow are observed. The flow pattern changes with

the temperature. The mist-flow pattern occurs at low tem-

perature. In the pseudocritical-temperature area, with an

increase in temperature, the flow pattern seems change

from an annular-dispersed flow to an annular flow.

(2) For a 6 mm ID tube at a low mass flux of 200 kg m�2 s�1,

wavy-dispersed flow and wavy flow are observed. At

a high mass flux of 800 kg m�2 s�1, usually, an annular-dis-

persed flow pattern is observed.

(3) The change in the flow pattern at different experimental

conditions is due to the change in the thermophysical prop-

erties of supercritical CO2 with the temperature and pres-

sureaswellas thechange inthe solubilityofCO2 intoPAG oil.

(4) From the comparison of the flow pattern observations and

the heat transfer measurements, it seems that the mecha-

nism behind the heat transfer performance drop is mainly

due to the formation of an oil film along the inner wall, and

the oil droplets do not contribute significantly to the heat

transfer deterioration.

Acknowledgements

The authors express their gratitude to Matsushita Electric In-

dustrial Co., Ltd. for the financial support.

r e f e r e n c e s

Dang, C., Hihara, E., 2004. In-tube cooling heat transfer ofsupercritical carbon dioxide-part 1: experimentalmeasurement. Int. J. Refrigeration 27, 736–747, doi:10.1016/j.ijrefrig.2004.04.018.

Dang, C., Iino, K., Fukuoka, K., Hihara, E., 2007. Effect oflubricating oil on heat transfer of supercritical carbon dioxide.Int. J. Refrigeration 30, 724–731, doi:10.1016/j.ijrefrig.2006.09.006.

Gao, L., Honda, T., 2002. Experiments on heat transfercharacteristics of heat exchanger for CO2 heat pump system.In: Proceedings of the Asian Conference on Refrigeration andAir Conditioning, 4 December 2002. Japan Society ofRefrigeration and Air Conditioning Engineers, Kobe, Japan, pp.A2–A4.

Kaneko, M., Ikeda, H., Tokiai, T., Yoshii, A., Suto, H., 2006. Thedevelopment of PAG refrigeration lubricants for automotive A/Cwith CO2. In: JSAE Automotive Air-Conditioning Conference,Tokyo, Japan, 13–15 March 2006, pp. 142–159.

Mori, K., Onishi, J., Shimaoka, H., Nakanishi, S., Kimoto, H., 2002.Cooling heat transfer characteristics of CO2 and CO2–oilmixture at supercritical pressure conditions. In: Proceedingsof the Asian Conference on Refrigeration and AirConditioning, 4 December 2002. Japan Society of Refrigerationand Air Conditioning Engineers, Kobe, Japan, pp. A2–A5.

Yun, R., Hwang, Y., Radermacher, R., 2006. Gas cooler heattransfer and pressure drop characteristics of CO2/oil mixturein a microchannel. In: Proceedings of the 7th IIR-GustavLorentzen Conference on Natural Working Fluids,Trondherim, Norway, pp. 503–505.