Applied Thermal Engineering 248 (2024) 123202

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Pool boiling heat transfer characteristics of low-GWP refrigerants in a

horizontal tube bundle configuration
M. Muneeshwaran a, *, Cheng-Min Yang a, Ercan Cakmak b, Kashif Nawaz a, *
a
Building Technologies Research and Integration Center, Oak Ridge National Laboratory, TN, USA
b
Material Science and Technology Division, Oak Ridge National Laboratory, TN, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Heat transfer enhancement techniques have been adapted on the shell side to improve the overall performance of
Metal foam the flooded evaporators, such as finned tubes. In recent years, it has been demonstrated that the metal foam
Pool boiling structure can offer enhanced heat transfer performance under pool boiling conditions. However, there are not
Tube bundle
much research in the open literature that examine the feasibility of metal foam embedded tubes in horizontal
low-GWP refrigerant
tube bundle configurations. Therefore, this paper proposed a novel metal foam embedded tube (i.e., foam
Heat transfer coefficient
embedded outside the tube) to improve the heat transfer behavior of flooded evaporators. The experiments were
performed on a horizontal tube bundle with a staggered arrangement. Moreover, the performance of low global
warming potential (GWP) refrigerants (R-1234yf and R-1234ze(E)) is compared against R-134a for both plain
tubes and metal foam tubes with porosities of 81%, 75%, and 62%. The results showed that a metal foam tube
with a porosity of 62% showed a maximum heat transfer coefficient (HTC) enhancement of 291% compared to
the plain tube. As compared with R-134a, the HTCs of R-1234yf and R-1234ze(E) are nearly 10% higher and 5%
lower, respectively.

1. Introduction including tube spacing, tube diameter, tube arrangement (staggered or

in-line), heat flux, refrigerant, number of tube rows, and type of tubes
Flooded evaporators are commonly employed in industrial air- (smooth or enhanced) [1]. As opposed to single tube, convection effects
conditioning and refrigeration systems. The configuration of the floo­ caused by bubble agitation are significant in tube bundle. The convec­
ded evaporators is analogous to the shell and tube heat exchangers. tion effects could improve the HTC of the tubes in the top rows of the
Typically, the shell side is exposed to the refrigerant boiling (i.e., phase tube bundle. The convection-induced enhancement in the tube bundle is
change process), whereas the tube side undergoes single-phase heat known as bundle effect. For example, Gupta [2] reported a 7-fold in­
transfer. The heat transfer coefficients (HTC) on both tube and shell crease in HTC for the upper row tubes than the bottom row tubes in a 5 x
sides determine the size and capacity of the flooded evaporator. 3 in-line tube bundle. Wang et al. [3] studied the bundle effect on eight
Furthermore, the performance of the flooded evaporator, which is a part reentrant cavity tubes arranged in two columns using R134a. They
of the whole system, can significantly influence the overall system ef­ observed that the bundle effect enhanced the HTC by 1.05–1.3 times,
ficiency. The plain tubes (i.e., smooth tubes) offer a lower HTC than indicating that the enhancement ratio is more for the top row tubes than
enhanced tubes and thus increase the size of the flooded evaporator. that in the middle and bottom row tubes.
Therefore, improving the performance of the flooded evaporator has The enhanced tubes are generally deployed in flooded evaporators to
been an ongoing research development for the past few decades. achieve a higher HTC and smaller size [4,5]. Ayub and Bergles [6]
Accordingly, several enhanced tube geometries have been developed conducted experiments to measure the performance of different tubes,
over the years to increase the shell side HTC in flooded evaporators, such as smooth, GEWA-T, and GEWA-K tubes. They observed that the
which can improve the overall HTC of the flooded evaporator. enhanced tubes (i.e., GEWA-T and GEWA-K) offered 2- and 1.6-fold
The tube bundle configuration is commonly encountered in practical increments in HTC compared with the smooth tubes for R-113 and
applications, and its performance is dependent on several factors, water, respectively. They noted that the GEWA tubes performance was

* Corresponding authors.
E-mail addresses: murganm@ornl.gov (M. Muneeshwaran), nawazk@ornl.gov (K. Nawaz).

https://doi.org/10.1016/j.applthermaleng.2024.123202
Received 30 December 2023; Received in revised form 13 April 2024; Accepted 16 April 2024
Available online 18 April 2024
1359-4311/© 2024 Elsevier Ltd. All rights reserved.
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Fig. 1. (a) Experimental facility, (b) staggered arrangement of tube bundle, (c) configuration of cartridge heater, and (d) dimensional details of the plain and metal
foam tubes.

strongly dependent on the gap width, and the optimum gap width would the enhanced tube could provide a more than 5-fold improvement in
vary with working fluids. Similarly, Webb and Pais [7] compared the HTC over the smooth tubes. Ustinov et al. [10] developed the tubes with
performance of plain tubes with integral finned, GEWA-SE, GEWA- microstructures and verified their performance with R-134a under a
TX19, and Turbo-B tubes. They studied these tubes performance under tandem tube bundle arrangement. Their results showed that the
different refrigerants, including R-134a, R-11, R-22, R-12, and R-123. microstructured tubes can initiate the boiling inception at a wall-
Gorenflo et al. [8] performed the tests to examine the strong dependence superheat of 2 K that is significantly lower than the plain tubes. Hsieh
of pool boiling HTC on saturation pressure and the surface roughness of et al. [11] introduced the plasma-coated tubes to augment the HTC of
the tube. Gorgy and Eckels [9] compared the TBIILP tube’s performance the flooded evaporators. Their experimental results on R134a showed a
against smooth tube using R-123 refrigerant. They demonstrated that bundle factor of 3.82 for the plasma-coated tubes. The performance

2
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

summary of enhanced tubes used in flooded evaporators can be found focused on investigating the metal foam embedded tube’s pool boiling
elsewhere [12,13]. performance using different low-GWP refrigerants (R-1234yf and R-
Recently, the metal foam structure has been proposed to augment the 1234ze(E)). Furthermore, the HTCs of R-1234yf and R-1234ze(E) are
pool boiling HTC. Arbelaez et al. [14] used the FC-72 fluid to analyze the compared against R-134a. Additionally, the influence of metal foam’s
pool boiling behavior of flat aluminum foam. At low heat fluxes, the porosity (81%, 75%, and 62%) on pool boiling heat transfer is also
metal foams could increase the HTC due to a smaller number of bubbles studied.
generated on the metal foam surface. At high heat fluxes, the bubble
takeoff from the pores would become difficult because of the intense 2. Experimental setup
bubble generation, which could deteriorate the metal foam’s pool
boiling HTC at high heat fluxes. Similarly, Athreya et al. [15] studied the The tube bundle, with four tubes in a staggered horizontal arrange­
flat aluminum foam’s pool boiling performance using FC-72. They ment, was placed inside the pressure vessel, as shown in Fig. 1(a). The
concluded that the high pore density could degrade the HTC, whereas tube pitch of 19.05 mm was maintained. The pressure vessel was
the HTC increased with decreasing metal foam thickness. Xu et al. [16] completely insulated to prevent any heat loss or gain. In this study,
examined the influence of PPI (pores per inch) on the pool boiling HTC aluminum tubes were selected for investigation and the outer diameter
of flat copper foams. The PPI had a significant effect on metal foam HTC. of the tube was 9.52 mm. The cartridge heaters were inserted inside the
The metal foam with low PPI performed better at low wall superheats, aluminum tubes to provide the necessary heating power. Note that the
whereas the metal foam with high PPI showed good thermal perfor­ indium foil with high thermal conductivity of 86 W m− 1 K− 1 was placed
mance at moderate and high wall superheats. Yang et al. [17] demon­ at the interface of the heater and the tube to reduce the thermal contact
strated a 2–3 fold increment of pool boiling HTC in the flat copper foam resistance. The heaters’ heated length and outer diameter were 63.5 mm
than that in the plain copper surface. Analogous to Xu et al. [16], Yang and 6.35 mm, respectively. The lead end of the cartridge heater had an
et al. [17] also concluded that the PPI could greatly influence the pool unheated length of 25.4 mm, while the disc end had 6.35 mm of un­
boiling HTC of foams. Despite a greater number of nucleation sites and heated length, as shown in Fig. 1(c). The lead end of the heaters was
surface area, large PPI foams would suffer from high bubble departure applied with epoxy to prevent axial heat loss through the cartridge
resistance due to intense bubble generation. Shi et al. [18] studied the heaters. The heater power was controlled using an SCR (silicon-
influence of foam thickness using flat copper foam. They showed that controlled rectifier) controller, which helped attain the specific heat
the HTC decreases with increasing thickness. In addition to the flat metal flux. And the Watt transducer (Ohio Semitronics, PC5-020X5Y25)
foams, few studies have analyzed the performance of metal foams (i.e., measured the heaters power, which has a 0–3 kW measurement range.
metal foams embedded inside the tube) under flow boiling conditions. After placing the tube bundle inside the ASME certified pressure vessel,
Nosrati et al. [19] compared the R134a flow boiling performance of the vessel was vacuumed for over 12 h to remove any non-condensable
plain tubes against the tubes with metal foam (i.e., inside the tube). The gases from the system and then it was charged with the refrigerant. In
metal foam tubes (MFT) offered a 220 % enhancement in HTC, but with this study, R-134a, R-1234yf, and R-1234ze(E) refrigerants were used.
an adverse pressure drop increase. Hu et al. [20] analyzed the influence Upon charging the refrigerant into the vessel, the heaters were powered
of wettability on metal foam filled tubes. The foam with hydrophobic to provide the specific heating power to the aluminum tubes. Once the
coating provided a higher HTC than that of uncoated foams, whereas the tube walls were heated up, the refrigerant started to boil and produced
hydrophilic coated foams offered a lower HTC than the uncoated ones. vapor. The vapor was passed to the condenser through buoyancy, where
Based on the above review, it is clear that the heat transfer perfor­ it was condensed into liquid. Note that the plate heat exchanger was
mance of metal foams is superior to the plain surface. However, most of employed as a condenser in this study, and it was placed outside the
the studies were mainly focused on investigating the performance of flat pressure vessel. The condensed liquid refrigerant was passed to the
metal foams under pool boiling conditions and metal foam filled tubes in vessel through the liquid return line with the help of gravity. The cold-
flow boiling contexts. For flooded evaporator configurations, the metal water delivered from the chiller (Thermo Scientific, Thermoflex 7500)
foam embedded tubes (i.e., metal foam embedded on the outside of the offered cooling power to the condenser. The saturation pressure was
tube) can offer better performance because of their larger surface area- maintained by controlling the cold-water temperature and flow rate.
to-volume ratio and greater nucleation site densities. Muneeshwaran The refrigerants’ saturation temperature was monitored with two T-type
et al. [21] and Yang et al. [22] studied the metal foam embedded tubes thermocouples in the liquid pool, as presented in Fig. 1(a). There is no
pool boiling characteristics using dielectric fluid (HFE7000). Their re­ significant difference in bath temperature was observed. The variation
sults showed that the metal foam embedded tubes can offer a 2.1–4.8 in bath temperature was within 0.2◦ C, which is within thermocouple
times enhancement in HTC over the smooth aluminum tubes. The re­ uncertainty. Furthermore, the saturation temperature was verified with
frigerants, such as R-11 (GWP = 4750), R-12 (GWP = 10900), R-22 the vessel pressure measured using an absolute pressure transducer
(GWP = 1810), and R-134a (GWP = 1430) are widely employed in (Omega PX409). Four 1.1 mm holes were made in each tube at 90⁰ apart
commercial flooded evaporators, Due to the efforts aimed at achieving for measuring the wall temperature of the tubes, and the thermocouples
net-zero goals by 2050, the above high-GWP refrigerants are being were inserted into the holes, as presented in Fig. 1(b). A tight fit was
replaced with the low-GWP refrigerants in the commercial flooded maintained between the thermocouple and the hole. Additionally,
evaporators [23]. In this context, research on the pool boiling of low- similar to the heater, indium foil with a high thermal conductivity of 86
GWP refrigerants in metal foam embedded tubes is essential, which W m− 1 K− 1 was inserted into the gap between the thermocouple and the
could help understand the MFT performance under flooded evaporator hole. The tube was then heated to the melting point of indium to facil­
operating conditions. Nevertheless, the pool boiling studies of low-GWP itate melting and filling of the gap. For tubes 2–4, all four thermocouples
refrigerants related to the metal foam embedded tubes (outside of the were positioned at 25.4 mm from the rear end of the tube. Whereas for
tubes) are limited in the open literature [24]. Therefore, this study tube 1, three thermocouples were placed at 25.4 mm from the rear end
of the tube, and one thermocouple was located at 12.7 mm from the rear
end. The offset between the thermocouples in tube 1 helped estimate the
Table 1
longitudinal heat conduction loss through the tube wall. Note that the
Details of measuring instruments with their accuracy.
estimated heat loss through longitudinal heat conduction was negli­
Parameter Instrument Measurement range Accuracy
gible. And the wall temperatures were also measured using the T-type
Temperature T-type thermocouple − 250 to 350 ⁰C ±0.2 ⁰C thermocouples. The calibration of thermocouples stood within the ac­
Pressure Absolute pressure sensor 0 to 500 psi ±0.5 % F.S. curacy of ± 0.2⁰C. The temperature, pressure, and power measurements
Power Watt transducer 0 to 3000 W ±0.5 % F.S.
were acquired using the National Instruments data acquisition system.

3
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Fig. 2. (a) Picture of the plain aluminum tube and metal foam embedded tubes with different porosities and (b) X-ray tomography images of metal foam tubes for
different porosities [21].

foam’s geometric details are given in Fig. 1 (d) and Table 2. The foam
Table 2
porosity was evaluated using the X-ray computed tomography (XCT)
Dimensional details of the plain tube and metal foam.
technique. The porosity measurements were carried out on a Zeiss
Parameter Values Xradia Versa 520 instrument operated at 80 kV at 7 W. The metal foams
Outer diameter of plain tube 9.52 mm were scanned at multiple stages with overlaps to include the whole
Length of the tube 76.5 mm specimen; later, the X-ray images were joined together. For better res­
Heated length of the tube 63.5 mm
olution and field of view, a 0.4x scintillator objective was employed
Metal foam thickness 2.54 mm
Metal foam porosity 62 %, 75 %, and 81 %
before the CCD camera handled with a 2 X 2 binning, providing a 22.23
Tube pitch 19.05 mm µm voxel size. The Dragonfly PRO software v.3.5 was used to carry out
the segmentation and data analysis. The histogram thresholding method
was used to segment the solids and voids of the metal foam. For porosity
Furthermore, a high-speed camera (Photron) was utilized to visualize measurement, histogram-based segmentation was used to segment the
the boiling phenomenon, and a temporal resolution of 1000 fps (frames pores (air) vs the metal. The total volume of the segmented pores was
per second) was maintained during the recording. The details of then used to determine the total porosity by taking its ratio to the entire
measuring instruments and their accuracy are presented in Table 1. volume of the porous cylinder. The output images of the XCT techniques
Besides the plain tubes, the metal foam embedded tubes were also are grayscale images. The grayscale images are scaled from black to
tested in this study. The pictures of the MFTs are presented in Fig. 2. The white, and everything in between is scaled based on absorption in­
MFTs were manufactured by brazing the aluminum metal foam (6101) tensities. A threshold-based segmentation is used to cover the air (i.e.,
around the plain aluminum tubes, which was obtained through pores) that has darker pixels (i.e., lower absorption) compared to denser
aluminum dip brazing process. After brazing, the metal foam tubes were material (i.e., higher absorption).
T6 heat treated. To understand the influence of metal foam porosity on
pool boiling HTC, metal foams with three different porosities of 62%,
3. Data reduction
75%, and 81% were tested. Note that the thickness of 2.54 mm and 40
PPI were retained for all the foam porosities. The tube’s and metal
The power provided to the cartridge heaters (Q) was assessed with

4
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

the help of a Watt transducer, and the heat flux is estimated using the
below relationship:
Q
q″ = (1)
πdl

where d and l represent the outer diameter (m) and heated length (m) of
the tube, respectively.
The wall temperature of the tube at each measurement point can be
estimated through Fourier’s heat conduction law,
⎛ ( )⎞
d
⎜ln dth ⎟
Tw,j = Tj − Q⎜
⎝ 2πkl ⎠
⎟ (2)

where j the subscript indicates the temperature measurement point in

the tube, Tw,j is the wall temperature at each measurement point (K), Tj
represents the measured temperature at the corresponding measure­
ment point (K), dth denotes the diameter at which the thermocouples are Fig. 3. Validation of experimental facility − experimental results are compared
placed in the tube (m), and k indicates the conductivity of the aluminum against the Cooper [25] and Gorenflo [8] correlations.
( )
tube W m− 1 K− 1 .
As mentioned earlier, the wall temperature of each tube was assessed performed at a saturation temperature of 14⁰C for all the refrigerants,
at four points. Therefore, the average wall temperature of each tube can and the heat flux was ranged between 7 kW m− 2 and 150 kW m− 2.
be estimated through the following expression:

1∑ 4
4.1. Validation of the experimental facility
Twall,i = Tw,j (3)
4 j=1
The experiments on a single plain tube were performed for validating
where the subscript i indicates the tube number and Twall,i is the average the experimental setup. The study on a single tube was carried out using
wall temperature of each tube (K). R-134a refrigerant. The experimental HTCs of the single tube data were
Similarly, the bath temperature was measured at two locations in the compared against the Cooper [25] and Gorenflo [8] correlations, as
pool (i.e., bottom and top of the pool), and their average value is used in presented in Fig. 3. According to Fig. 3, the experimental data match
estimating the average HTC of the individual tube, as shown in the well with the correlations, with a maximum deviation of 20%, which
below expression: validates the test facility. The difference between the experimental data
and the correlations can be attributed to the experimental uncertainties.
q″ The Cooper and Gorenflo correlations are presented below.
havg,i = ( ) (4)
Twall,i − Tbath,avg Cooper correlation [25]:
( )
where havg,i is the average HTC of the individual tube W m− 2 K− 1 and hcooper = 90M − 0.5
( − log(Pr ) )− 0.55
(Pr )(0.12− 0.2(Rp ) )
(q″)0.67 (7)
Tbath,avg denotes the average temperature of refrigerant pool (K). ( )
The average HTC of the tube bundle can be predicted using the below where M indicates the molecular weight of the refrigerant g mol− 1 , Pr
relationship: denotes the reduced pressure, and Rp indicates the surface roughness
1 ∑N (μm). The plain tube roughness was measured with the help of Mitutoyo
havg,tb = havg,i (5) SJ210 surface roughness tester, and it was measured to be nearly 0.85 ±
N i=1
0.5 µm.
( ) Gorenflo correlation [8]:
where havg,tb is the average HTC of the tube bundle W m− 2 K− 1 and N
is the number of tubes in a bundle. Note that the tube bundle contains 4 ( )(0.9−
q″ r )
0.3P0.3 (
P
)(
Rp
)0.133
tubes in this study. hGorenflo = ho 1.2P0.27
r + 2.5Pr + (8)
q″o 1 − Pr Rp,o
The uncertainty of the experimental parameters can be estimated
using: where the subscript o indicates the reference value of the respective
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√[( )2 ( )2 ( )2 ]
̅ parameter.
√ ∂ λ ∂ λ ∂ λ
Uλ = √ Y + Y + ... + Y (6)
∂x1 1 ∂x2 2 ∂xn n
4.2. Tube bundle effect
( )
where λ is the function of independent variables x1 , x2,... xn and the The HTCs of the individual tubes in a plain aluminum tube bundle
uncertainties of those independent variables are Y1 , Y2 , ...Yn , respec­ are presented in Fig. 4. In plain tube bundle, tube 1 (i.e., row 1) showed
tively. nearly 20% enhancement in HTC compared to the bottom row tubes (i.
e., tubes 2–4). The bubbles generated at the bottom row of tubes can
4. Results and discussion impinge on the top row of tubes, creating a convective effect that results
in an enhanced heat transfer coefficient (HTC). As the bubbles rise from
In this section, the experimental results of plain and MFT tube the bottom tubes to the top of the vessel, they randomly collide with the
bundles under pool boiling conditions are discussed. Three different top tubes. Each collision causes a local disturbance in the fluid sur­
porosities of 81%, 75%, and 62% were studied for MFT tube bundle. rounding the tube, which induces a convective effect. This convective
Furthermore, the HTCs of low-GWP refrigerants (R-1234yf and R- effect can increase the HTC. Whereas the HTCs of tubes 2 and 3 (i.e., row
1234ze(E)) are compared against those of R-134a. The experiments were 2) are almost similar to those of tube 4 (i.e., bottom row). By considering

5
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Fig. 4. HTCs of individual tubes in (a) plain aluminum tube bundle and (b) Metal foam tubes with 62% porosity.

the experimental uncertainties, the bundle effect is marginal for the bundle remained identical irrespective of the tube location. Conse­
tubes 2 and 3 due to the limited number of tube rows. Qiu and Liu [26] quently, the average HTC is used to depict the performance of the metal
and Wang et al. [3] studied the bundle effects on a ~ 7–8 row tube foam tubes in the following Sections.
bundle. Their results suggested that the HTCs of the bottom row tubes Metal foam tubes are typically fabricated by brazing metal foams
are almost similar. And the HTCs increment at the bottom row tubes around plain tubes. However, variations in the number of contact points
(bottom 3 rows) are significantly lower than at the top row tubes (top 3 between the metal foam and the plain tube wall can lead to a scattered
rows). Since the number of tube rows is limited to 3 in this study, no trend in heat transfer coefficients (HTC) within a metal foam tube
significant tube bundle effects are observed for the second-row tubes (i. bundle. Additionally, the tortuous nature of the metal foam can intro­
e., tubes 2 and 3). The tube bundle effect in MFTs is shown in Fig. 4 (b). duce randomness in pool boiling behavior, such as bubble departure and
Unlike plain tubes, the HTCs of individual tubes showed a scattered generation, further contributing to the scattered behavior observed in
behavior for a given heat flux. Note that the experiments were per­ metal foam tube bundles. Moreover, the lack of tube bundle effect in
formed by interchanging the tube positions in the MFT tube bundle case. MFT can be explained with the help of the suppression of bubble
The results showed that the individual tube’s HTC remained similar sweeping phenomenon. In the plain tubes, the bubble formation is pri­
irrespective of the tube location. Hence, the average HTC of MFT tube marily occurred on the tube wall. In this scenario, the bubbles leave

6
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

to the bundle effect in plain tube bundles. More details on the bubble
sweeping phenomenon can be found in our previous paper [21].
Whereas in the metal foam tubes, most of the nucleation sites are within
the metal foam structure (thickness is 2.54 mm) unlike the plain tube,
which isolates the bubbles from the neighboring tubes and prevents the
bubble sweeping in the top row tubes. Overall, the combination of
variation in number of brazing points in each metal foam tubes and the
suppression of bubble sweeping phenomenon could have contributed to
the lack of bundle effect in the metal foam tubes.

4.3. Metal foam tubes

The metal foam tube is proposed to enhance the pool boiling HTCs of
the tube bundle. Metal foams with three distinct porosities of 81%, 75%,
and 62% were tested to analyze the effect of foam porosity on the pool
boiling HTC. The average HTCs of MFTs with different porosities are
compared against the plain aluminum tube bundle, as shown in Fig. 5. In
Fig. 5. Performance comparison of metal foam tube bundle against the plain General, for all the porosities, the MFTs showed a higher HTC as
aluminum tube bundle (Working fluid – R-134a). compared with the plain aluminum tube. Among the MFTs, 62%
porosity exhibited a higher HTC than that of a plain tube, followed by
from the bottom wall could collide with the top row tubes and eventu­ 75% and 81% porosities. The MFTs with 62% improved the HTC by
ally, they will sweep the cluster of bubbles from the top row tubes, 102–153% and 162–291% over the plain tubes at high (>39 kW m− 2)
which is known as bubble sweeping phenomenon or early bubble de­ and low (<39 kW m− 2) heat fluxes, respectively. Similarly, compared
parture. This bubble sweeping phenomenon could also have contributed with the plain tubes, the MFTs with 75% porosity offered an enhance­
ment of 95–127% and 133–202% at high (>39 kW m− 2) and low (<39

Fig. 6. Pool boiling behavior of R-134a at different heat fluxes (a) plain aluminum tube bundle, (b) metal foam tube bundle – porosity = 81 %, (c) metal foam tube
bundle – porosity = 75 %, and (d) metal foam tube bundle – porosity = 62 %.

7
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Fig. 7. Comparison of R-1234yf and R-1234ze(E) performance against R-134a (a) plain aluminum tubes, (b) metal foam tubes with 81% porosity, (c) metal foam
tubes with 75% porosity, and (d) metal foam tubes with 62% porosity.

kW m− 2) heat fluxes, respectively. Whereas in the MFTs with 81% the below expressions [27]:
porosity, the enhancement of HTC remained identical at 80–90% over
PPI 1
the tested range of heat flux. Generally, the enhancement in HTC can be = (9)
25.4 dp + df
attributed to the increased surface area of the metal foams and a greater
number of nucleation sites. Moreover, Xu et al. [16] observed that the ⎛ ⎞
√̅̅̅̅̅̅̅̅̅̅̅
enhanced HTC in metal foams could be due to the nucleation of tiny df 1 − ε⎜ 1 ⎟
bubbles at the junctions and ligaments in the open cell metal foams. = 3.39 ⎜ ( )⎟ (10)
dp 3π ⎝ ⎠
Generally, the HTC increases with decrease in porosity. The reason
1− ε
1 − exp − 0.04
for increase in HTC with decreasing porosity is mainly attributed to the
decrease in pore diameter and increase in surface area. The pore Based on the above expressions, the pore diameters of metal foams with
diameter can be predicted using two metal foams parameters: (1) 81%, 75%, and 62% porosities are 0.43 mm, 0.41 mm, and 0.38 mm.
porosity and (2) PPI. And the pore diameter can be calculated by solving Note that the PPI for all the foams is 40. The capillary force, Pc = 4σdcosθ , is
p

8
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Fig. 7. (continued).

a function of surface tension and pore diameter. Typically, for the given fluxes. This trend could be explained with the help of the pore size and
refrigerant (i.e., same surface tension), the capillary force increases as bubble generation frequency. As compared with 81% porosity metal
the porosity reduces due to the reduction in pore diameter. As a result, foam, the pore size in 62% and 75% porosity metal foams is smaller. Due
the metal foams with low porosities could accelerate the liquid replen­ to the larger bubble generation frequency at high heat fluxes, a greater
ishment to the tube wall due to the increase in capillary force, which number of bubbles can be trapped inside the small pores, which slows
helps maintain a steady liquid feed to the tube and thereby attain better down the bubble departure phenomenon due to larger bubble departure
pool boiling heat transfer [28,29]. In addition to the capillary force, the resistance and results in a relatively low enhancement percentage in
increased surface area at lower porosities would also result in greater HTC at high heat fluxes. Whereas quite a few bubbles would be gener­
number of nucleation sites, which in turn the main factor for HTC ated at low heat fluxes than at high heat flues, which makes the bubble
improvement. departure relatively easier and yields a high enhancement percentage in
As mentioned above, the 62% and 75% porosities showed distinct HTC. The pool boiling behavior of plain tube and MFTs is presented in
enhancement trends at low and high heat fluxes. The enhancement Fig. 6, and it can be visualized in the videos provided in the supple­
percentage at low heat fluxes is typically more than that at high heat mentary files. As opposed to the 62% and 75% porosities, the pore size in

9
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

Table 3 √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Selected properties of refrigerants at a saturation temperature of 14⁰C.
σ
Dd = 0.0208β (11)
g(ρl − ρv )
Properties R-134a R-1234yf R-1234ze(E)

Saturation pressure, kPa 473.2 495 354.3 where Dd indicates the bubble departure diameter (m), β = 35⁰ for re­
Critical pressure, kPa 4059 3382 3632 frigerants, σ is the surface tension (N m− 1), g represents the acceleration
Pressure ratio 0.117 0.146 0.098
due to gravity (m s− 2), and ρl and ρv denote the density of liquid and
Density at liquid state, kg m− 3 1247 1131 1198
Density at vapor state, kg m− 3 23.03 27.48 19.02 vapor refrigerant (kg m− 3), respectively. According to the above
Latent heat of vaporization, kJ kg− 1 186.6 153.7 174.7 expression, the bubble departure diameters for R-134a, R-1234yf, and R-
Surface tension, N m− 1 0.009494 0.007571 0.01039 1234ze(E) are calculated based on the properties of refrigerants ob­
Viscosity at liquid state, Pa s 2.22E-04 1.74E-04 2.18E-04
tained at a saturation temperature of 14⁰C. The departure diameter of R-
Viscosity at vapor state, Pa s 1.15E-05 1.08E-05 1.16E-05
Liquid thermal conductivity, W m− 1 K− 1 0.088 0.067 0.079 1234yf is nearly 0.608 mm, followed by 0.647 mm for R-134a and 0.690
Vapor thermal conductivity, W m− 1 K− 1 0.013 0.013 0.013 mm for R-1234ze(E). It is apparent that, against R-134a, the bubble
departure diameters of R-1234yf and R-1234ze(E) are almost 6 % lower
and 6.6 % higher, respectively. When the bubble departure diameter is
the 81% porosity metal foam is relatively larger, as shown in Fig. 2. The bigger, it may coalesce with the neighboring bubbles and form a local
81% case revealed a consistent enhancement percentage for the tested vapor blanket over the heated surface, which retards further nucleation
range of heat flux because of the smoother bubble departure phenom­ activities and adversely affects the pool boiling heat transfer. The chance
enon caused by the larger pore size. of coalescence with neighboring bubbles will be reduced when the
bubble departure size reduces, and that leaves a bare surface for forming
4.4. Low-GWP refrigerants further nucleation sites, which helps increase the pool boiling HTC. As a
result, among the three different refrigerants tested at a saturation
According to the Paris Agreement, efforts are being accelerated to temperature of 14⁰C, R-1234yf with a lower bubble departure diameter
succeed the high-GWP refrigerants with low-GWP refrigerants. In this rendered a higher HTC at high heat fluxes, followed by R-134a and R-
regard, several low-GWP refrigerants have been created, such as 1234ze(E).
hydrofluoroolefins (HFO). It is expected that the commonly used R-134a
(GWP = 1300) is likely to be replaced with other low-GWP refrigerants.
5. Conclusions
Therefore, in this study, the performance of low-GWP refrigerants,
including R-1234yf (GWP = 4) and R-1234ze(E) (GWP = 7) is tested and
This study experimentally analyzed the behavior of metal foam tubes
compared against that of R-134a (GWP = 1300). The heat transfer
in pool boiling conditions in a horizontal tube bundle. The experiments
characteristics of R-134a, R-1234yf, and R-1234ze(E) for both plain tube
were performed for metal foam tubes with different porosities of 81%,
and MFTs are presented in Fig. 7. When q is under ~ 50 kW m− 2, the
75%, and 62%, and their performance was compared against the plain
HTCs of R1234ze(E) and R-1234yf are identical to those of R-134a.
aluminum tube bundle. Furthermore, in this study, the behavior of low-
However, when q is more than ~ 50 kW m− 2, the HTCs of R-1234yf are
GWP refrigerants, such as R-1234yf and R-1234ze(E), was studied and
nearly 10 % higher than those of R-134a for both plain tube and MFTs,
compared with the results of R-134a. The main conclusions are as
while a 5 % reduction in HTC is observed for R-1234ze(E) compared
follows:
with R-134a. At low heat fluxes, the bubble generation rate is low
compared with the one at high heat fluxes, which makes the bubbles to
1. In comparison with plain aluminum tubes, metal foam tubes with
smoothly depart from the nucleation surface. The local vapor blanket
62% porosity provided a higher HTC, followed by 75% and 82%
formation is highly dependent on bubble departure size and the nucle­
porosities.
ation site densities. Since the number of nucleation sites are relatively
2. When compared with the plain tubes, metal foam tubes with 62%
fewer at low heat fluxes, the chance of local vapor blanket formation is
porosity showed nearly 102–153% and 162–291% increases in HTC
minimum. As a result, at low heat fluxes, the impact of bubble departure
at high (>39 kW m− 2) and low (<39 kW m− 2) heat fluxes, respec­
diameter and pressure ratio (related to nucleation sites activation) is
tively. Similarly, the HTC improvements of 95–127% and 133–202%
minimum on bubble coalescence and subsequent vapor blanket forma­
were noticed in the metal foam tubes with 75% for high (>39 kW
tion, which indirectly indicates the less influence over HTC. Nonethe­
m− 2) and low (<39 kW m− 2) heat fluxes, respectively.
less, at high heat fluxes, intense bubble generation could increase chance
3. In contrast to 62% and 75% porosities, metal foam tubes with 81%
of bubble coalescence and form a vapor blanket. And at this stage, the
porosity provided an 80–90% increment in HTC for the entire heat
effect of departure bubble size and nucleation site densities becomes
flux range.
dominant, which can significantly influence the HTCs. Note that for all
4. At a saturation temperature of 14⁰C, the HTC of R-1234yf and R-
the refrigerants, the experiments were conducted at a saturation tem­
1234ze(E) was 10 % higher and 5 % lower than that of R-134a for
perature of 14⁰C. At a saturation temperature of 14⁰C, the reduced
both plain and metal foam tubes, respectively.
pressure of R-134a, R-1234yf, and R-1234ze(E) are 0.117, 0.146, and
0.098, respectively, as listed in Table 3. The reduced pressure of R-
CRediT authorship contribution statement
1234yf is nearly 25 % higher than that of R-134a, whereas the reduced
pressure of R-1234ze(E) is almost 16 % lower than that of R-134a.
M. Muneeshwaran: Conceptualization; Experimental facility
Thome [30] and Wang et al. [31] showed that at a higher pressure ratio,
development, Experimental Campaign, Methodology and Data analysis,
a greater nucleation site densities can be activated on the surface, which
Writing original draft. Cheng-Min Yang: Writing – review & editing,
leads to a higher HTC for R-1234yf. Besides, the departure bubble size
Investigation, Conceptualization. Ercan Cakmak: Writing – review &
also plays a vital function in determining the pool boiling characteris­
editing, Resources, Investigation. Kashif Nawaz: Writing – review &
tics. The bubbles can early be removed from the surface when the bubble
editing, Project administration, Funding acquisition, Conceptualization.
diameter is smaller, which leaves a bare surface and activates new
nucleation spots and enhances the nucleate boiling heat transfer. Fritz
[32] proposed the below relationship to estimate the bubble departure Declaration of competing interest
diameter [33]:
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence

10
M. Muneeshwaran et al. Applied Thermal Engineering 248 (2024) 123202

the work reported in this paper. [13] S. Casciaro, J.R. Thome, Thermal performance of flooded evaporators, part 2:
review of void fraction, two-phase pressure drop, and flow pattern studies,
ASHRAE Trans. 107 (2001) 919.
Data availability [14] F. Arbelaez, S. Sett, and R. L. Mahajan, “An experimental study on pool boiling of
saturated FC-72 in highly porous aluminum metal foams,” in: 34th National Heat
Data will be made available on request. Transfer Conference, p. 759, 2000.
[15] B. Athreya, R. Mahajan, and S. Sett, “Pool boiling of FC-72 over metal foams: effect
of foam orientation and geometry,” AIAA/ASME Joint Thermophysics and Heat
Acknowledgement Transfer Conference, 2002.
[16] J. Xu, X. Ji, W. Zhang, G. Liu, Pool boiling heat transfer of ultra-light copper foam
with open cells, Int. J. Multiph. Flow 34 (11) (2008) 1008–1022.
The authors appreciate Tim Dyer, Chavez Chelo, Charles Pierce, and [17] Y. Yang, X. Ji, J. Xu, Pool boiling heat transfer on copper foam covers with water as
Michael Day for their technical support. The authors would like to thank working fluid, Int. J. Therm. Sci. 49 (7) (2010) 1227–1237.
the aid given by US Department of Energy Building Technologies Office [18] J. Shi, X. Jia, D. Feng, Z. Chen, C. Dang, Wettability effect on pool boiling heat
transfer using a multiscale copper foam surface, Int. J. Heat Mass Transf. 146
(BTO) and the technology manager, Mr. Antonio Bouza. (2020) 118726.
[19] A. Nosrati, M. Akhavan-Behabadi, B. Sajadi, P. Razi, R. Mohammadi, Experimental
Appendix A. Supplementary data study on the effects of using metal foam on R-134a flow boiling in annular tubes,
Int. J. Therm. Sci. 177 (2022) 107546.
[20] H. Hu, Z. Lai, Y. Zhao, Heat transfer and pressure drop of refrigerant flow boiling in
Supplementary data to this article can be found online at https://doi. metal foam filled tubes with different wettability, Int. J. Heat Mass Transf. 177
org/10.1016/j.applthermaleng.2024.123202. (2021) 121542.
[21] M. Muneeshwaran, C.-M. Yang, E. Cakmak, K. Nawaz, Augmentation of pool
boiling heat transfer on tube bundles using metal foam, Appl. Therm. Eng. 236
References (2024) 121812.
[22] C.-M. Yang, “Enhanced Pool Boiling Of Low-Pressure Refrigerants On Round
[1] S. Moharana, A. Bhattacharya, M.K. Das, A critical review of parameters governing Tubes-An Experimental Evaluation,” International Refrigeration and Air Conditioning
the boiling characteristics of tube bundle on shell side of two-phase shell and tube Conference, Purdue, 2022.
heat exchangers, Thermal Science and Engineering Progress (2022) 101220. [23] A.K. Dewangan, A. Kumar, R. Kumar, Experimental study of bubble behaviors
[2] A. Gupta, Enhancement of boiling heat transfer in a 5× 3 tube bundle, Int. J. Heat during boiling of a hydrocarbon refrigerant, Int. J. Therm. Sci. 166 (2021) 106989.
Mass Transf. 48 (18) (2005) 3763–3772. [24] C.-M. Yang, M. Muneeshwaran, and K. Nawaz, “Experimental investigation on
[3] Y. Wang, Z. Ma, J. Zhang, Precise determination of R134a boiling bundle effect on nucleate boiling heat transfer of low GWP refrigerants over metal-foam enhanced
a column of reentrant cavity tubes, Appl. Therm. Eng. 199 (2021) 117612. tube bundles,” presented at the 26th International Congress of Refrigeration Paris,
[4] A.K. Dewangan, A. Kumar, R. Kumar, Nucleate boiling of pure and quasi-azeotropic France, 2023.
refrigerants from copper coated surfaces, Appl. Therm. Eng. 94 (2016) 395–403. [25] M. Cooper, “Saturation Nucleate Pool Boiling: A Simple Correlation,” Department of
[5] A.K. Dewangan, A. Kumar, R. Kumar, Nucleate Pool Boiling Heat Transfer of Engineering Science, Oxford University, England, vol. 86, 1984.
Refrigerants Using Coated Surfaces, Advanced Cooling Technologies and [26] Y.-H. Qiu, Z.-H. Liu, Boiling heat transfer of water on smooth tubes in a compact
Applications (2019) 85. staggered tube bundle, Appl. Therm. Eng. 24 (10) (2004) 1431–1441.
[6] Z.H. Ayub, Pool boiling from GEWA surfaces in water and R-113, Iowa State [27] L.L. Manetti, A.S.O.H. Moita, E.M. Cardoso, A new pool boiling heat transfer
University, 1986. correlation for wetting dielectric fluids on metal foams, Int. J. Heat Mass Transf.
[7] R.L. Webb, C. Pais, Nucleate pool boiling data for five refrigerants on plain, 171 (2021) 121070.
integral-fin and enhanced tube geometries, Int. J. Heat Mass Transf. 35 (8) (1992) [28] C. Li and G. Peterson, “Parametric study of pool boiling on horizontal highly
1893–1904. conductive microporous coated surfaces,” 2007.
[8] D. Gorenflo, P. Sokol, S. Caplanis, Pool boiling heat transfer from single plain tubes [29] Y. Jiang, G. Zhou, J. Zhou, F. Zhou, X. Huai, Saturated pool boiling heat transfer of
to various hydrocarbons, Int. J. Refrig 13 (5) (1990) 286–292. HFE-7100 on sintered copper powder and wire mesh microporous surfaces: A
[9] E. Gorgy, S. Eckels, Convective boiling of R-123 on enhanced-tube bundles, Int. J. comparison study, Appl. Therm. Eng. 216 (2022) 119067.
Heat Mass Transf. 134 (2019) 752–767. [30] J.R. Thome, Enhanced boiling heat transfer, Hemisphere, Newyork, 1990.
[10] A. Ustinov, V. Ustinov, J. Mitrovic, Pool boiling heat transfer of tandem tubes [31] A. Kumar, C.-C. Wang, Nucleate pool boiling heat transfer of R-1234ze (E) and R-
provided with the novel microstructures, Int. J. Heat Fluid Flow 32 (4) (2011) 134a on GEWA-B5H and smooth tube with the influence of POE oil, Appl. Therm.
777–784. Eng. 201 (2022) 117779.
[11] S.-S. Hsieh, G.-Z. Huang, H.-H. Tsai, Nucleate pool boiling characteristics from [32] W. Fritz, Berechnung des maximalvolumes von dampfblasen, Physik. Zeitschr 36
coated tube bundles in saturated R-134a, Int. J. Heat Mass Transf. 46 (7) (2003) (1935) 379–384.
1223–1239. [33] V. Guichet, S. Almahmoud, H. Jouhara, Nucleate pool boiling heat transfer in
[12] S. Casciaro, J.R. Thome, Thermal performance of flooded evaporators, part 1: wickless heat pipes (two-phase closed thermosyphons): A critical review of
Review of boiling heat transfer studies/Discussion, ASHRAE Trans. 107 (2001) correlations, Thermal Science and Engineering Progress 13 (2019) 100384.
903.

11