Water-Heated Pool Boiling of

Different Refrigerants on the

Tailian Chen
Department of Mechanical Engineering,
Gonzaga University,
Spokane, WA 99258-0026
e-mail: chent@gonzaga.edu
Outside Surface of a Horizontal
Smooth Tube
Pool boiling heat transfer has been extensively studied over decades, but the effect of
boundary heating conditions on boiling received little attention. In this work, heat trans-
fer coefficients during pool boiling of five different refrigerants (R123, R245fa, R236fa,
R134a, and R22) on the outside surface of a smooth copper tube were measured at the
saturation temperature of 6.7  C; water flows inside the tube and provides heat to the
refrigerants to boil (thus, water-heated boiling). Measurements showed that the refriger-
ant of a higher vapor pressure has a higher heat transfer coefficient, with the exception
that R22 performs nearly the same as R134a. A correlation previously developed for
electrically-heated pool boiling on cylindrical tubes underpredicts by 30%–46% the heat
transfer coefficients during water-heated boiling of the five refrigerants. Among the pool
boiling correlations reviewed in this work, the Cooper correlation (for pool boiling on
cylindrical tubes) predicts the boiling heat transfer coefficients of R22 and R245fa rea-
sonably well (within 68.5%), but not as well those of the other three refrigerants (under-
predicts by nearly 30% for R134a and R236fa and overpredicts by nearly 40% for R123).
It is found that the predicted boiling heat transfer coefficients of the five refrigerants by
the modified Gorenflo correlation (simply adding a constant multiplier of 1.47 to the Gor-
enflo correlation) are in excellent agreement with their respective measurements.
[DOI: 10.1115/1.4004902]

Keywords: pool boiling, cylindrical tubes, refrigerants, boiling correlations, chiller

efficiency, energy conservation

Introduction heat transfer correlations for pure halogenated and flammable

refrigerants boiling on electrically-heated boiling on smooth
Ever-increasing energy demand accentuates the importance of
tubes, i.e., heated by the electrical heating elements attached
efficient use of energy. Chillers produce chilled water for climate
inside the tube. Recently, Sarma et al. [13] proposed a correlation
control in the commercial and industry buildings and constitute a
for boiling of water and ethyl alcohol on cylindrical tubes based
major portion of total energy use in these buildings. Enhancing
on the data of Borishansky et al. [14]. A summary of these corre-
heat transfer in evaporators and condensers is one of the major
lations is given in Table 1. The Gorenflo correlation contains
approaches to increasing chiller efficiency. Moreover, phase-out
fluid-specific reference heat transfer coefficients for a number of
of some refrigerants due to environmental issues raises concerns
boiling fluids; those for halogenated refrigerants included in the
about heat transfer performances of their replacements. Thome [1]
correlation are reproduced in Table 2. It is noted that among the
and Gorenflo [2] provided reviews on boiling of new refrigerants
five refrigerants tested in this work, R245fa and R236fa are not
and refrigerant mixtures.
included in the Gorenflo correlation.
As an important heat transfer process in energy systems, pool
Though research of boiling heat transfer has been performed
boiling has been extensively studied and correlations have been
over decades, it is only recently that the effect of boundary heat-
proposed to predict pool boiling heat transfer. Commonly-used
ing conditions (water-heated versus electrically-heated) on the
nucleate boiling correlations fall into two categories: reduced
heat transfer during the boiling process started being noted. The
pressure-based and thermophysical properties-based correlations.
modeling work by Unal and Pasamehmetoglu [15] showed that
The reduced pressure-based correlations that predict the boiling
the different heating methods led to different boiling curves. The
heat transfer from the macroscopic system perspective include
fluid-heated smooth tube boiling heat transfer by McManus [16]
Cooper correlation [3], Borishanski correlation [4], Mostinski cor-
differs by as much as 50% from the electrically-heated boiling
relation [5], and Gorenflo correlation [6]. The thermophysical
data by Memory [17]. In an experimental study of pool boiling of
properties-based correlations generated on the basis of the micro-
R123 on four commercial enhanced tubes, Kedzierski [18] found
scopic heat transfer mechanisms associated with the bubble dy-
that boiling curves generated in the fluid-heated boiling experi-
namics include Rohsenow correlation [7], Stephan and
ments were located to the left of those from the electrically-heated
Abdelsalam correlation [8], and Forster and Zuber correlation [9].
boiling experiments, indicative of better boiling heat transfer per-
Gorenflo et al. [10] studied the effects of fluid thermophysical
formance for fluid-heated boiling; Darabi et al. [19] also observed
properties and they found that the effects of thermophysical prop-
similar behavior. In spite of these counter-intuitive observations
erties are consistent with those of reduced pressure on boiling heat
on the effect of the boundary conditions, understanding of the
transfer coefficient. Jung et al. [11,12] proposed two pool boiling
underlying mechanisms responsible for this effect is not yet avail-
able due to scarce water- or fluid-heated boiling data in relation to
Contributed by the Heat Transfer Division of ASME for publication in the
JOURNAL OF HEAT TRANSFER. Manuscript received December 24, 2010; final
electrically-heated boiling data.
manuscript received August 12, 2011; published online December 13, 2011. Assoc. This paper reports the heat transfer measured during pool boil-
Editor: Louis C. Chow. ing of five different refrigerants on the outside surface of a

Journal of Heat Transfer Copyright V

C 2012 by ASME FEBRUARY 2012, Vol. 134 / 021502-1

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms

 Table 1 A list of reviewed pool boiling heat transfer correlations; “General” denotes the correlations are not specified for a
particular fluid or geometry

Correlations Fluids Geometries

 0:67
Based on Cooper h ¼ C  55P0:120:4343lnR
r
P
ð0:4343lnPr Þ0:55 M0:5 q== (1) General Flat
reduced correlation [3] C: constant (C ¼ 1 for flat surfaces, C ¼ 1.7 for cylindrical surfaces); surfaces and
pressure Pr: reduced pressure; M: molecular weight of the boiling fluid; cylindrical
q//: heat flux (W/m2); h: boiling heat transfer coefficient (W/m2 K); surfaces
and Rp: surface roughness in mm (Rp ¼ 1 mm for unknown surfaces).
Borishanski q== ¼ ðA Þ3:33 ðTW  Tsat Þ3:33 ½FðPr Þ3:33 ; A* and F(Pr) are given by (2) General General
correlation [4] Mostinski correlation; Tw: wall temperature ( C); Tsat:
saturation temperature ( C); q//: heat flux (W/m2).
Mostinski Proposed relations in the Borishanski correlation: (3) General General
correlation [5] A ¼ 0:1011P0:69 1:7 1:2
c , and FðPr Þ ¼ 1:8Pr þ 4Pr þ 10Pr ; Pc,
10

the critical pressure of the boiling fluids (bar);

Pr is the reduced pressure.
 nf
Gorenflo == (4) Halogenated Cylindrical
h ¼ h0 FPF q== =q0 ðRP =RP0 Þ0:133 ; h0: reference values
correlation [6] refrigerants surfaces
given in Table 2; FPF ¼ 1:2P0:27 r þ 2:5Pr þ Pr =ð1  Pr Þ; (Table 2)
==
q// has a unit of W/m2; q0 ¼ 20,000 W/m2; nf ¼ 0:9  0:3P0:3 r ;
and Rp0 ¼ 0.4 lm; Rp is surface roughness and taken
as 0.4 lm for unknown surfaces.
 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi0:33  
Based on Rohsenow 1 Cp;l 1 r 0:67 (5) Water, carbon General
thermo correlation [7] h¼  0:67 q== ðPrl Þ1:7 tetrachloride,
Csf hfg ll gðql  qv Þ
physical Isopropyl
properties As a first approximation, Csf ¼ 0.074 for refrigerants on alcohol,
copper surfaces; Cpl, heat capacity (J/kg-K); hfg, latent heat (J/kg); n-butyl alcohol
ml, liquid viscosity (Ns/m2); r, surface tension (N/m); ql, qv,
liquid and vapor densities (kg/m3); g, specific gravity (9.8 m/s2);
and; q00 , heat flux W/m2.
Stephan ==
0:745
0:581 (6) Refrigerants; General
hDb q Db qv
and ¼ 207 ðPrl Þ0:533 for other fluid,
kl kl Tsat ql
Abdelsalam this correlation
correlation [8] has different forms
Db ¼ 0.146b[2r/g (ql - qv)]0.5;
q//, heat flux (W/m2); kl, thermal conductivity (W/m-K); ql, qv,
liquid and vapor densities (kg/m3); Tsat, saturation temperature;
r, surface tension (N/m); Prl, Prandtl number; and the
contact angle b is 35 .
" 0:79 #
Forster and kl C0:45pl ql
0:49 (7) General General
Zuber q== ¼ 0:00122 0:5 0:29 0:24 0:24 ðTw  Tsat Þ1:24 DP0:75 sat
correlation [9] r ll hfg qv
q//, heat flux (kW/m2); kl, thermal conductivity (kW/m-K);
Cpl, heat capacity (kJ/kg-K); ql, qv, liquid and vapor
densities (kg/m3); Tw and Tsat are wall and fluid saturation
temperatures ( C); r, surface tension (N/m); hfg, latent heat
(kJ/kg); ml, liquid viscosity (Ns/m2); and DPsat,
saturation pressure (Pa).
Empirical Jung et al. [11]
C1 (8) Halogenated Cylindrical
kl q== Db 1:4
(curve-fit h ¼ 10 P0:1
r ð1  Tr Þ ðPrl Þ0:25 refrigerants electrically-
Db kl Tsat
of test data) heated
Db ¼ 0:0146b½2r=gðql  qv Þ0:5 , where the contact surfaces
angle b is 35 and it is noted that the constant 0.0146
was used in Jung et al. [11,12] while 0.146 was used in the
Stephan and Abdelsalam correlation [8] for calculation of
bubble diameter Db; c1 ¼ 0:855ðqv =ql Þ0:309 ðPr Þ0:437 ; kl,
thermal conductivity (W/m-K); Pr, reduced pressure; Tr is
reduced temperature (Tr ¼ Tsat/Tc, and Tc is critical temperature);
ql, qv, liquid and vapor densities (kg/m3); and Prl is Prandtl number.
Jung et al. [12]
C 2
(9) Flammable Cylindrical
kl q== Db q 0:53
h ¼ 41:4 ð log 10 Pr Þ1:52 1  v refrigerants electrically-
Db kl Tsat ql
heated
C2 ¼ 0:835ð1  Pr Þ1:33 ; surfaces
the parameters and their units are the same as in (8).
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:22 !0:55
Based on Sarma et al. [13] (10) Water and Cylindrical
q== r D PD
the data of ¼ 3:8  106 P0:72
r
ethyl glycol surfaces
Borishansky ll hfg gðql  qv Þ dt ll h0:5
fg
 0:825
et al. [14] r
ðql  qv ÞgD2
in which dt ¼ kl * DT/q// is the thermal boundary
layer thickness; and D is tube diameter (m).

021502-2 / Vol. 134, FEBRUARY 2012 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms

 Table 2 Fluid-specific reference heat transfer coefficients h0 ber. The closed water loop maintains water through the test sec-
for halogenated fluids in the Gorenflo correlation [6]; R245fa tion in the boiling chamber at desired conditions (flow rate and
and R236fa are not included in the correlation temperature). The inlet temperature of water at a given flow rate
is controlled by adjusting the amount of heat transferred to water
Fluids Pc M h0 (W/m2 K)
in the water/steam heat exchanger included in the loop; each inlet
R11 44.0 137.4 2800
temperature corresponds to a heat transfer rate from the water to
R12 41.6 120.9 4000 the refrigerant. The chilled water loop provides chilled water to a
R13 38.6 104.5 3900 tube bundle in the condensation chamber. The condensation
R13B1 39.8 148.9 3500 chamber and the boiling chamber are open to each other both at
R22 49.9 86.5 3900 the top and at the bottom. By adjusting the flow rate of the chilled
R23 48.7 70.0 4400 water through the tube bundle in the condensation chamber, a
R113 34.1 187.4 2650 desired pressure in the boiling chamber and thus the saturation
R114 32.6 170.9 2800 temperature of the refrigerant is obtained.
R115 31.3 154.5 4200 In this work, experiments were performed at the pressures cor-
R123 36.7 152.9 2600
R134a 40.2 102.0 4500
responding to the saturation temperature of 6.7  C (44  F) for all
R152a 45.2 66.1 4000 the five refrigerants. Table 3 lists the refrigerant properties at the
R226 30.6 186.5 3700 saturation temperature of 6.7  C in the order of increasing vapor
R227 29.3 170.0 3800 pressure: R-123, R-245fa, R-236fa, R-134a, and R-22. For each
RC318 28.0 200.0 4200 refrigerant, experiments were performed at three flow rates of
R502 40.8 111.6 3300 water inside the tube in the range of 1.70 to 2.50 m/s (typical in
R245fa 36.4 134.0 (3641) chiller applications) that corresponds to Reynolds number in the
R236fa 32.0 152.0 (3641) range of 40,000 to 65,000 depending on the water inlet tempera-
ture. For each refrigerant at a given flow rate, the test section aver-
age heat flux depends on the water inlet temperature; a fixed
horizontal smooth tube heated by water flowing inside the tube, maximum water inlet temperature about 26.7  C (80  F) at each
i.e., it is water-heated boiling in comparison to electrically-heated flow rate results in the heat flux ranging from 10,000 to 100,000
boiling. Most empirical boiling correlations were developed on W/m2 covered in this work.
the basis of electrically-heated boiling data; however, fluid-heated
boiling is closer to the operating conditions in actual boiling appli-
cations where the heat is usually provided from another heat trans- Test Section
fer medium, such as water in chillers, to the boiling refrigerant. The test section consists of two 10-ft-long smooth copper tubes
Providing a boiling correlation applicable to the actual boiling installed horizontally side-by-side in the boiling chamber. In addi-
operating conditions for accurate prediction of boiling heat trans- tion to the 8-ft-long boiling segment, each test tube also includes
fer is of paramount significance to industry engineers. two 1-ft-long segments for sealing, plumbing, and instrumentation
at the inlet and the outlet of each tube. Figure 2 is a schematic of
the test section; the two tubes were positioned on a horizontal
Test Facility plane (with a center-to-center separation distance of 3.5 in. and
Figure 1 is a schematic of the test rig that includes a chilled thus a tube-to-tube clearance of 2.5 in.) and boiling on one tube is
water main, a water steam main, a closed water loop, an open not influenced by the other. A thermally insulated flexible hose
chilled water loop, a boiling chamber, and a condensation cham- connects the two tubes at one end of the boiling chamber while

Fig. 1 A schematic of the test rig and instrumentation; the boiling and the
condensation chambers are open to each other both at the top and at the bottom

Journal of Heat Transfer FEBRUARY 2012, Vol. 134 / 021502-3

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms

 Table 3 Properties of the refrigerants tested in this work at 6.7 C [27]

k(mW/m-K) Density(kg/m3)

Prandtl
Refrigerants M Pc(kPa) P(kPa) hfg(kJ/kg) r(mN/m) Liquid Vapor Cp(J/kg-K) Liquid Vapor l  106(Pa s) number

R123 152.9 3670 43.91 178.75 17.349 81.636 8.133 988.1 1509.4 2.98 518.37 6.3
R245fa 134.0 3640 71.96 200.71 16.372 93.768 11.716 1285.9 1386.5 4.30 519.33 7.1
R236fa 152.0 3200 140.61 156.01 12.199 78.783 11.317 1222.3 1418.8 9.81 360.96 5.6
R134a 102.0 4060 370.28 193.28 10.581 88.995 12.118 1360.6 1271.8 18.23 244.34 3.7
R22 86.5 4940 615.26 199.39 10.681 91.663 9.883 1189.3 1257.9 26.23 200.46 2.6

water enters the first tube and exits the second tube on the other
end, which forms a two-pass water flow. The outside diameter of
the test tubes is 1 in. and the wall thickness is 0.044 in. The inside
diameter of the boiling chamber is 6 in. and its length is 8 feet.
The test tubes run through the boiling chamber and are sealed at
the two end-flanges. The boiling chamber is nearly 2/3 filled with
a pool of refrigerant and the two 8-ft boiling segments are
immersed in the refrigerant pool.
A computer-based data acquisition system was used to record
data including water temperatures, water flow rate, and refrigerant
temperature and pressure in the boil chamber; data were recorded
refrigerant. Uo is the overall heat transfer coefficient with respect
to the outside surface area, ho is the average outside heat transfer
coefficient, hi is the inside heat transfer coefficient, Rwall is the
wall thermal resistance, and kwall is the wall thermal conductivity.
In Eq. (14), the fouling factor is neglected since the test tubes
were new tubes from raw stock and each tube was well cleaned
and dried before being put into the test chamber. In Eq. (15), Do,
Di, and L are respectively the outside diameter, the inside diame-
ter, and the length of each tube. For turbulent flow inside the tubes
[20]

only when the thermal and the hydrodynamic steady states were hi Di l 0:14
reached. Calibration was performed on all temperature sensors, ¼ 0:027Re0:8 Pr1=3 : (16)
kwall lwall
pressure transducers, and flow meters prior to experiments. The
refrigerant temperature measured in the boiling chamber matches
In Eq. (16), the viscosity, l, and the Prandtl number, Pr, are at the
well (60.5%) the measured pressure in the chamber.
average water temperature, Tave, and the Reynolds number is
Heat transfer calculations can be readily performed for each of
given by Re ¼ 4m _ w =ðplDi Þ. lwall is at the average inside wall
the two tubes
temperature Twall,i over the tube length obtained by
_ w Cp ðTin  Tout Þ
m
q== ¼ (11) Twall;o ¼ Tsat þ q== =ho (17)
Ao  
q== Twall;i ¼ Twall;o þ q== t =kwall : (18)
Uo ¼ (12)
LMTD
Tin  Tout Tsat in Eq. (17) is the saturation temperature obtained by averag-
LMTD ¼ (13) ing the measurements from four temperature sensors in the refrig-
ln½ðTin  Tsat Þ=ðTout  Tsat Þ
erant pool, and Twall,o is the average outside wall temperature. In
1 1 1 Ao Eq. (18), t is the wall thickness of the tube. The outside heat trans-
¼  Ao Rwall  (14)
ho Uo hi Ai fer coefficient ho is obtained by an iteration of Eqs. (14) through
lnðDo =Di Þ (18). Following the method of Kline and McClintock [21], the
Rwall ¼ (15) uncertainties in boiling heat transfer coefficients are estimated to
2pkwall L be in the range of 3.8%–6.1%; a larger uncertainty occurs at a
lower heat flux.
in which q00 is the heat flux with respect to the outside surface
area Ao of each 8-ft-long boiling segment, m _ w is the water flow
rate, Tin and Tout are the water inlet and outlet temperatures of
each tube, and Cp is the specific heat of water at its average tem- Results and Discussion
perature over the boiling segment, Tave ¼ ðTin þ Tout Þ=2. LMTD Figure 3 gives the average boiling heat transfer coefficients of
is the log mean temperature difference between the water and the R134a on the first and the second tubes (Fig. 2) to compare the

Fig. 2 A schematic of the test section; the two tubes are positioned on a
horizontal plane

021502-4 / Vol. 134, FEBRUARY 2012 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms

 Fig. 3 Boiling heat transfer coefficients measured on the two
individual tubes
boiling performances measured on the two individual tubes. For
each tube, there are a total of 27 data points corresponding to three
water flow rates (1.70, 2.16, and 2.50 m/s) and nine water inlet
temperatures at each flow rate. It is apparent that, also as
expected, the boiling heat transfer coefficients measured on the
two tubes are essentially the same, but the first tube covers a range
of larger heat fluxes than the second tube. This holds true for all
the five refrigerants tested in this work. At a given condition
(water flow rate and inlet temperature), the average water temper-
ature in the first tube is higher than in the second tube as the water
temperature decreases downstream, which results in a larger range
of heat flux for the first tube. The same heat transfer coefficients
measured on the two tubes indirectly confirmed that boiling on
one tube was not influenced by the other.
Figure 4 presents the measured boiling heat transfer coefficients
of the five refrigerants; the data points for each refrigerant include
those from both the first and the second tubes. For instance, there
Fig. 4 Boiling heat transfer measurements of the five refrigerants;
data points for each refrigerant contain those measured on
both of the two tubes
Journal of Heat Transfer
Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms
Fig. 5 Predicted boiling heat transfer coefficients of the five
refrigerants by the correlations developed by Jung et al. [11]
are a total of 54 data points for R134a (27 data points for each
tube). Throughout the tests, the maximum water inlet temperature
at each flow rate for all the five refrigerants was fixed at nearly the
same temperature (26.7  C); the maximum heat flux for a refriger-
ant depends on its boiling performance. The maximum heat flux
for R123 is the lowest as it performs lowest among the five refrig-
erants. For each refrigerant, the boiling heat transfer coefficient
increases with increasing heat flux within the heat flux range for
each refrigerant. Figure 4 also shows that the refrigerant of a
higher vapor pressure has a higher heat transfer coefficient, with
the exception that R22 performs nearly the same as R134a
although the vapor pressure of R22 is almost 60% higher than that
of R134a. This seemingly “abnormal” phenomenon for R22 and
R134a was also observed in a recent work by Park et al. [22] who
performed nucleate boiling of different refrigerants (R32, R22,
R134a, and R123, in the order of decreasing vapor pressure at a
given temperature) on an electrically-heated horizontal flat
smooth copper surface. Their results showed that, although the
heat transfer coefficients of different refrigerants generally
Fig. 6 Comparison of the predictions by the Cooper correlation
with the measurements
FEBRUARY 2012, Vol. 134 / 021502-5
 increase with increasing vapor pressure (i.e., R32 performs high-
est and R123 performs lowest), the performances of R22 and
R134a differ very little. Similar results have been seen in the work
by Webb and Pais [23]. This “abnormal” behavior reiterates that
boiling heat transfer is a complicated process involving compound
effects of thermophysical properties of the boiling fluids. Table 3
lists the thermal and transport properties of the five refrigerants.
An examination of these properties indicates that the heat transfer
coefficients measured in this work inversely follow the trend of
their surface tension values. In particular, R134a and R22 have
the same surface tension values, as with nearly the same heat
transfer coefficients of the two refrigerants. The importance of
surface tension to boiling heat transfer is widely recognized; how-
ever, the degree of its importance relative to those of other proper-
ties (including reduced pressure) awaits further investigation.
The correlation proposed by Jung et al. [11] was based on
electrically-heated pool boiling of pure halogenated refrigerants on
the smooth tubes; comparison of its predicted results with the meas-
urements in this work would provide insights into the effect of
boundary heating conditions on boiling heat transfer. Toward this
Fig. 7 Predictions by different nucleate boiling heat transfer correlations for the five refrigerants tested in this work
021502-6 / Vol. 134, FEBRUARY 2012
Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms
end, the boiling heat transfer coefficients of the five refrigerants are
predicted at the saturation temperature (6.7  C) by the correlation
(Eq. (8) in Table 1) and the predictions are plotted in Fig. 5 with
measurements. Figure 5 shows that the electrically-heated boiling
correlation underpredicts the water-heated boiling performances by
30 to 45% depending on the refrigerant and the heat flux. As noted
above, better heat transfer performance during the fluid-heated boil-
ing relative to the electrically-heated boiling has also been observed
by Kedzierski [18] and Darabi et al. [19]. Kedzierski postulated that
the interactions between the fluctuating wall temperature and the
fixed electrical heat flux induced a higher degree of superheated liq-
uid on the electrically-heated surface than on the fluid-heated sur-
face. For the same time-averaged heat flux, a larger fraction of it is
used to superheat liquid for electrically-heated boiling than for
fluid-heated boiling, which leads to a higher heat transfer coefficient
for fluid-heated boiling. Darabi et al. [19] extended an explanation
based on the fact that the temperature profiles along the tube (axi-
ally) for the water-heated and the electrically-heated cases are quite
different. In spite of these attempted explanations, it is reasoned
here that the difference in the performances is attributed, at least
Transactions of the ASME
 Table 4 The disparity between the predictions given in Fig. 7 and the measurements obtained in this work; “-” indicates that the
correlations underpredict the boiling performances
R123 R245fa
Rohsenow [7] 20 to 50% 47 to 65%
Stephan and Abdelsalam [8] 51 to 66% 59 to 66%
Gorenflo [2] 33 to 37% 30%
Mostinski [5] 13 to 47% 43% to 59%
partially, to the circumferential nonuniformity of heating during
water-heated boiling in comparison to the uniform heat flux during
electrically-heated boiling. For boiling on cylindrical surfaces, bub-
bles (except those on the very top portion) slide along the tube cir-
cumference to the top portion before they depart into the bulk
liquid. As a consequence, the boiling characteristics (forms of bub-
bles and their activities) vary along the tube circumference due to
accumulation and possible agglomeration of bubbles on the top por-
tion. Compared with the uniform heat flux during the electrically-
heated boiling, the heat flux over the tube circumference for the
water-heated boiling is most likely nonuniform due to heat redis-
tribution and strong turbulence within the fluid, and it is suspected
that the circumferentially-uniform heat flux during the electrically-
heated boiling results in boiling characteristics on the tube surface
not as effective a heat transfer process.
It is desirable to examine the applicability of commonly-used
nucleate pool boiling correlations for predicting the boiling heat
transfer during water-heated boiling on smooth tubes. Compared
with boiling on flat surfaces, the gravity effect during boiling on
cylindrical tubes is quite different. Nishikawa et al. [24] showed
that at heat fluxes below a threshold value, the boiling heat trans-
fer increases as the inclination angle of the boiling surface
increases from 0 deg (facing upward) to 180 deg (facing down-
ward). Carey [25] attributed the increase in boiling heat transfer
on an inclined surface to two effects: (1) the thicker natural con-
vection boundary layer on an inclined surface favorable for bubble
nucleation and (2) the promoted heat convection and film evapo-
ration due to sweeping of bubbles over an inclined surface since
bubbles on an inclined surface must travel along the surface to its
lateral edge before escaping to the ambient. The Cooper correla-
tion [3] has been recommended as one of the most accurate corre-
lations in predicting nucleate boiling heat transfer [26]; for boiling
on cylindrical tubes, Cooper recommended adding a multiplier of
1.7 to the Cooper correlation (for boiling on upward-facing flat
surfaces) to account for the gravity effect [3]. Figure 6 compares
the measurements obtained in this work with those predicted
by the Cooper correlation for the five refrigerants, which shows
that the Cooper correlation predicts very well the performances of
R22 and R245fa (within 68.5%), but underpredicts those of
R134a and R236fa by nearly 30% and overpredicts those of R123
by nearly 40%.
To further examine the applicability of commonly-used pool
boiling correlations for predicting water-heated boiling heat trans-
fer on cylindrical tubes, four correlations (Rohsenow correlation,
Stephan and Abdelsalam correlation, Gorenflo correlation, and
Mostinski correlation; details of each are given in Table 1) are
selected and their predictions are presented in Fig. 7 for the five
refrigerants corresponding to the test conditions in this work. In
order to use the Gorenflo correlation to predict the boiling heat
transfer coefficients of R245fa and R236fa that are not included in
the correlation, the averaged value of the given reference heat
transference coefficients for all the halogenated fluids is assigned
as the reference value (h0 ¼ 3641 W/m2 K) for the two refrigerants
(Table 2). A comparison with the measurements indicates that all
these four correlations underpredict the boiling heat transfer coef-
ficients during water-heated boiling on smooth tubes; the dispar-
ities are given in Table 4. A close examination of Table 4 reveals
that there is a wide scatter of the disparities for the Rohsenow,
Stephan and Abdelsalam, and Mostinski correlations. However,
Journal of Heat Transfer
Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms
R236fa R134a R22
29 to 61% 47 to 57% 24 to 39%
51 to 66% 73% 72%
25 to 30% 28 to 30% 27 to 28%
47 to 69% 66 to 71% 53 to 63%
the disparities by the Gorenflo correlation are almost the same,
i.e., the predictions for all the five refrigerants by the Gorenflo
correlation are all nearly 30% lower than their respective meas-
urements over the entire heat flux range covered by each refriger-
ant. In particular, the Gorenflo correlation captures the ‘abnormal’
behavior of the same heat transfer coefficients of R22 and R134a.
It is noted in Table 4 that Stephan and Abdelsalam correlation pre-
dicts the same performances of R22 and R134a, but their dispar-
ities (73%) and the disparities for the other refrigerants (65%)
differ considerably.
In spite of the disparities by the Gorenflo correlation in Table 4,
its capability of excellently capturing the trend of water-heated
boiling heat transfer coefficients of the five refrigerants sets it
apart from the other correlations, which may be attributed to the
fact that the Gorenflo correlation utilizes an experimentally-
determined heat transfer coefficient h0 (at reduced pressure of 0.1,
heat flux of 20,000 W/m2, and surface roughness of 0.4 lm) for
each fluid as the reference value. With the effect of thermophysi-
cal properties of the boiling liquid being contained in the fluid-
specific reference heat transfer coefficient h0, the Gorenflo
correlation predicts the heat transfer coefficients at other condi-
tions by accounting for the effects of heat flux, reduced pressure,
and surface roughness. Motivated by this observation (the pre-
dicted heat transfer coefficients of the five refrigerants are all
nearly 30% lower), the Gorenflo correlation is modified by adding
a multiplier of 1.47. The predicted boiling heat transfer coeffi-
cients by the modified Gorenflo correlation are given in Fig. 8,
which shows good agreement with the boiling heat transfer coeffi-
cients of the five refrigerants (within the range of 3.3–9.5%) dur-
ing water-heated boiling on smooth tubes. In this work, the
refrigerants were tested at a fixed temperature of 6.7  C (44  F),
the standard temperature of chiller operation. However, the varia-
tion in the cooling load during the actual chiller operation may
Fig. 8 The predicted boiling heat transfer coefficients by the
modified Gorenflo correlation are in good agreement (3.3%–
9.5%) with those measured in this work
FEBRUARY 2012, Vol. 134 / 021502-7
 lead to a maximum variation of 10  C in the refrigerant saturation q ¼ density, kg/m3
temperature. The follow-up of this work is to determine whether D ¼ difference
or not the modified Gorenflo correlation predicts the boiling heat
transfer coefficients at various refrigerant saturation temperatures.
Subscripts
Summary l¼ liquid
v¼ vapor
Heat transfer coefficients during boiling of five different refrig- w¼ water
erants on an 8-ft-long smooth tube were measured at the satura- wall ¼ tube wall
tion temperature of 6.7  C (44  F). Water flows inside the test
section and provides heat to the refrigerants to boil. The refriger- References
ant of a higher vapor pressure was measured to have a higher boil- [1] Thome, J., 1996, “Boiling of New Refrigerants: A State of the Art Review,”
ing heat transfer coefficient, with the exception that R22 has Int. J. Refrig., 19, pp. 435–457.
essentially the same performance as R134a although the vapor [2] Gorenflo, D., 2001, “State of the Art in Pool Boiling Heat Transfer of New
pressure of the former is almost 66% higher. A correlation previ- Refrigerants,” Int. J. Refrig., 24, pp. 6–14.
[3] Cooper, M. G., 1984, “Heat Flow Rates in Saturated Nucleate Pool Boiling—A
ously developed for electrically-heated pool boiling on cylindrical Wide Range of Examination Using Reduced Properties,” Adv. Heat Transfer,
tubes underpredicts by 30%–46% the boiling heat transfer coeffi- 16, pp. 157–239.
cients of the five refrigerants heated by flowing water inside the [4] Borishanski, V. M., 1969, “Correlation of the Effect of Pressure on the Critical
tubes. Compared with the uniform heat flux along the tube cir- Heat Flux and Heat Transfer Rates Using the Theory of Thermodynamic
Similarity,” Problems of Heat Transfer and Hydraulics of Two Phase Media,
cumference during electrically-heated boiling, nonuniform heat Pergamon Press, pp. 16–37.
flux along the tube circumference during the water-heated boiling [5] Mostinski, I. L., 1963, “Application of the Rule of Corresponding States for
is suspected to be, at least partially, responsible for the more Calculation of Heat Transfer and Critical Heat Flux,” Teploenergetika (Mos-
effective boiling heat transfer. The Cooper correlation (for pool cow, Russ. Fed.), 4, pp. 66–71.
[6] Gorenflo, D., 1993, “Pool Boiling” in VDI Heat Atlas (English Version), VDI-
boiling on cylindrical tubes) predicts the boiling heat transfer Verlag Dusseldorf, Germany.
coefficients of R22 and R245fa reasonably well (within 68.5%), [7] Rohsenow, W. M., 1952, “A Method of Correlating Heat Transfer Data for Sur-
but not as well for the other three refrigerants (underpredicts by face Boiling of Liquids,” Trans. ASME, 74, pp. 969–975.
nearly 30% for R134a and R236fa and overpredicts by nearly [8] Stephen K., and Abdelsalam, M., 1980, “Heat Transfer Correlations for Natural
Convection Boiling,” Int. J. Heat Mass Transfer, 23, pp. 73–87.
40% for R123). It is found that the modified Gorenflo correlation [9] Forster, H. K., and Zuber, N., 1955, “Dynamics of Vapor Bubbles and Boiling
(simply adding a constant multiplier of 1.47 to the Gorenflo corre- Heat Transfer,” AIChE J., 1(4), pp. 531–539.
lation) predicts very well the boiling heat transfer coefficients dur- [10] Gorenflo, D., Chandra, U., Kotthoff, S., and Luke, A., 2004, “Influence of
ing water-heated pool boiling of the five refrigerants on the Thermo-Physical Properties on Pool Boiling Heat Transfer of Refrigerants,”
Int. J. Refrig., 27, pp. 492–502.
outside surface of a smooth tube (within the range of 3.3–9.5%). [11] Jung, D, Kim, Y., Ko, Y. and Song, K., 2003, “Nucleate Boiling Heat Transfer
Coefficients of Pure Halogenated Refrigerants,” Int. J. Refrig., 26, pp. 240–248.
[12] Jung, D., Lee, H., Bae, D., and Oho, S., 2004, “Nucleate Boiling Heat Transfer
Nomenclature Coefficients of Flammable Refrigerants,” Int. J. Refrig., 27, pp. 409–414.
Ai ¼ tube inside area, m2 [13] Sarma, P. K., Srinivas, V., Sharma, K. V., Subrahmanyam, T., and Kakac, S.,
Ao ¼ tube outside area, m2 2008, “A Correlation to Predict Heat Transfer Coefficient in Nucleate Boiling
Cp ¼ specific heat, J/kg-K on Cylindrical Heating Elements,” Int. J. Therm. Sci., 47, pp. 347–354.
[14] Borishansky, V. M., Bodrovich, B. I., and Minchenko, F. P., 1961, “Heat
Di ¼ tube inside diameter, m Transfer During Nucleate Boiling of Water and Ethyl Alcohol,” Aspects of
Do ¼ tube outside diameter, m Heat Transfer and Hydraulics of Two Phase Mixtures, Government Energy
g¼ gravity acceleration, m/s2 Publishing House, Moscow, pp. 75–93.
h¼ heat transfer coefficient, W/m2 K [15] Unal, C., and Pasamehmetoglu, K. O., 1993, “A Numerical Investigation of the
Effect of Heating Methods on Saturated Nucleate Pool Boiling,” Int. Commun.
hi ¼ heat transfer coefficient on the tube side, W/m2 K Heat Mass Transfer, 21(2), pp. 167–177.
hfg ¼ latent heat, J/kg [16] McManus, S. M., Marto, P. J., and Wanniarachchi, X, 1986, “An Evaluation of
ho ¼ boiling heat transfer coefficient, W/m2 K Enhanced Heat Transfer Tubing for Use in R114 Chillers,” ASME Heat Trans-
kwall ¼ wall thermal conductivity, W/m-K fer Air Cond. Refrig. Equip., HTD 65, pp. 222–230.
[17] Memory, S. B., Sugiyama, D. C., and Marto, P. J., 1944, “Nucleate Pool Boiling
L¼ tube length, m of R114 and R114/Oil Mixtures from Smooth and Enhanced Tubes—Part I:
M¼ molecular weight, g/mol Single Tubes,” Int. J. Heat Mass Transfer, 38(8), pp. 1347–1361.
m_w¼ water flow rate, kg/s [18] Kedzierski, M. A., 1955, Calorimetric and Visual Measurements of R-123 Pool
P¼ pressure, kPa Boiling on Four Enhanced Surfaces, National Institute of Standards and Tech-
nology, Gaithersburg, MD, Report No. NISTIR 5732.
Pc ¼ critical pressure, kPa [19] Darabi, J., Ohadi, M. M., Fanni, M. A., Dessiatoun, S. V., and Kedzierski, M.
Pr ¼ reduced pressure, P/Pc A., 1999, “Effect of Heating Boundary Conditions on Pool Boiling
Pr ¼ Prandtl number Experiments”, HVAC&R Res., 5(4), pp. 1–14.
q00 ¼ heat flux, W/m2 [20] Sieder, E. N., and Tate, G. E., 1936, “Heat Transfer and Pressure Drop of
Liquids in Tubes,” Ind. Eng. Chem., 28, pp. 1429–1435.
Re ¼ Reynolds number [21] Kline, S. J., and McClintock, F. A., 1953, “Describing Uncertainties in Single
Rwall ¼ wall thermal resistance, K/W Sample Experiments,” Mech. Eng., 75, pp. 3–9.
t¼ tube wall thickness, m [22] Park, K. J., Jung, D., and Shim, S. E., 2009, “Nucleate Boiling Heat Transfer
Tave ¼ water average temperature,  C Coefficients of Halogenated Refrigerants up to Critical Heat Fluxes,” J. Mech.
Eng. Sci., 223, pp. 1415–1424.
Tin ¼ water temperature at the inlet,  C [23] Webb R. L., and Pais, C., 1992, “Nucleate Pool Boiling Data for Five Refriger-
Tout ¼ water temperature at the outlet,  C ants on Plain, Integral-Fin and Enhanced Tube Geometries,” Int. J. Heat Mass
Tsat ¼ refrigerant saturation temperature,  C Transfer, 35(8) pp. 1893–1904.
Twall,i ¼ average tube inside wall temperature,  C [24] Nishikawa, K., Fujita, Y., Uchida, S., and Ohta, H., 1983, ‘‘Effect of Heating
Surface Orientation on Nucleate Boiling Heat Transfer,’’ Proceedings of
Twall,o ¼ average tube outside wall temperature,  C ASME-JSME Thermal Engineering Joint Conference, Y. Mori and W. J. Yang,
Uo ¼ overall heat transfer coefficient, W/m2 K eds., Honolulu, HI, Vol. 1, pp. 129–136.
[25] Carey V. P., 1992, Liquid-Vapor Phase Change Phenomena: An Introduction to
the Thermophysics of Vaporization and Condensation Processes in Heat Trans-
Greek Letters fer Equipment, Taylor and Francis, London.
l ¼ viscosity, kg/m s [26] Collier, J. G., and Thome, J. R., 1994, Convective Boiling and Condensation,
3rd ed., McGraw-Hill, New York.
a ¼ thermal diffusivity, m2/s [27] NIST Thermodynamic and Transport Properties of Refrigerants and Refrigerant
r ¼ surface tension, N/m Mixtures—refprop Version 8.0.

021502-8 / Vol. 134, FEBRUARY 2012 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 05/12/2013 Terms of Use: http://asme.org/terms