INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. (2017)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/er.3770

Investigation of heat transfer during melting of a PCM

by a U-shaped heat source
Mohammad Bashar and Kamran Siddiqui*,†
Department of Mechanical and Materials Engineering, University of Western Ontario, London, Ontario, Canada

SUMMARY
An experimental study was conducted to investigate the melting process of a phase change material (PCM) and the
associated convection heat transfer due to a U-shaped heat source embedded in the PCM. The experiments were conducted
at four input heat fluxes that varied from 3450 to 5840 W/m2. The results showed that the heat transfer behavior, interface
movement, and the heat transfer coefficients differed both axially and vertically inside the chamber. The local convective
heat transfer coefficient in the inner region, enclosed by the U-tube, was found to be about 35% higher than that in the outer
region over the input heat flux range, resulting in faster melting in the inner region than in the outer region. As melted
domain grew vertically from 15% to 100%, it was observed that the overall h in the inner region increased by 40–55% from
the lowest to highest heat flux. The melting rate was also found comparatively high up until 65–70% of the total PCM
volume melted because of the higher contribution from the inner region. It was also observed that the Rayleigh number
increased by approximately 23% in the inner region and 18% in the whole domain as the heat flux increased from 3450
to 5840 W/m2. A new Nusselt–Rayleigh number correlation is proposed for the heat transfer during the melting process
due to a U-shaped heat source. Copyright © 2017 John Wiley & Sons, Ltd.
KEY WORDS
PCM; U-shaped tube; latent heat; thermal energy storage; heat transfer coefficient; interface; melt fraction

Correspondence
*Kamran Siddiqui, Department of Mechanical and Materials Engineering, University of Western Ontario, 1151 Richmond Street,
London, ON, N6A 5B9, Canada.
†
E-mail: ksiddiqui@eng.uwo.ca

Received 9 December 2016; Revised 13 April 2017; Accepted 14 April 2017

1. INTRODUCTION the thermal energy is stored by simply increasing the

Thermal energy storage (TES) is a means to store thermal
energy for a certain period of time for its intended later
use [1]. In general, TES systems improve the energy
efficiency by reusing waste heat as well as by reducing
the mismatch between supply and demand [2]. It is a
crucial component in energy systems utilizing intermittent
renewable energy resources such as solar energy that can
also fluctuate unpredictably [3]. TES also improves the
performance and reliability of an energy system by
plummeting peak operating loads and helps the system to
work within an optimal range [4]. Hence, TES has enormous
potential to perform numerous roles in conserving energy,
which in turn helps to attain energy sustainability [5–7].
For thermal energy in a TES to be stored, a physical
media is needed that would hold the heat. Heat can be
stored in two ways, either by using a physical method or
by applying thermo-chemical energy method. Through
the use of physical techniques, heat can be stored in
sensible or latent heat form [8]. In sensible heat storage,
temperature of the storage media, although the mass and
material property of the media play important roles. On
the other hand, latent heat storage is based on the
capture/release of energy when a material undergoes a
phase change, typically between solid and liquid phases.
Latent heat thermal energy storage (LHTES) is particularly
attractive because of its two important features: Firstly, it
can provide a high-energy storage density (amount of heat
per unit mass), and secondly, it has the capacity to store
energy at nearly a constant temperature or over a small
range of temperature variations around the phase transition
temperature [9,10].
The LHTES is a potential and proven solution to
realistically and effectively offset the unreliability issues
associated with renewable energy resources [2,11–15]
and has also been used in a diverse range of industrial
applications, such as metal processing, food preservation,
solidification of castings, building systems, and electronics
cooling [8,16–22]. TES has also been used as an integrated
component in solar thermal systems [23,24]. However, for

Copyright © 2017 John Wiley & Sons, Ltd.

Bashar M. and Siddiqui K.
its effective use as a potential heat storage media, desirable
thermodynamic, kinetic, and chemical properties of these
phase change materials (PCMs) must first be carefully
assessed [25]. Some of the key parameters that need to
be considered when selecting a PCM are latent heat,
melting point, and heat source/sink temperatures. As the
latent heat is the primary mode for heat storage, higher
latent heat of a PCM is highly desirable. This further
implies that for the effective utilization of the LHTES, it
should operate within a narrow temperature band about
its melting point. Larger difference between the PCM
melting point and source/sink temperatures results in a
higher fraction of heat storage in the form of sensible heat,
which undermines the underlying benefit of LHTES. In
addition, longevity, economic, and the environment effects
of PCMs must also be taken into considerations [3,26].
The design of the heat source that carries the heat
transfer fluid (HTF) is one of the key factors that dictates
the overall heat transfer rate and performance of the
storage system. During heat addition/extraction to/from
the PCM, the heat transfer occurs in both solid and liquid
domains of the PCM. Because the fraction of each phase
continuously changes (one increasing, while other
decreasing) the local heat transfer within the PCM domain
is transient in nature. However, the correlation between the
average heat transfer and the melting rate has not been well
established [27]. Conduction is the heat transfer mode in
the solid phase, while natural convection governs the heat
transfer in the liquid phase. The rate of heat transfer is
typically higher in the liquid domain than in the solid
domain because of the presence of convective motion
induced by buoyancy, which accelerates the transportation
of heat. As the buoyancy acts vertically (aligned with
gravity), the orientation of the heat source/sink plays a
crucial role in the generation of buoyancy-induced flow
and its magnitude and extent. This, in turn, directly
influences the heat transfer within the PCM domain. Hence,
an improved understanding of the natural convection
process within the liquid domain with respect to the heat
source/sink orientation is vital to design efficient TES.
Gau and Viskanta [28] reported the presence of
convective flow patterns through the visualization of the
melting process of paraffin wax in a rectangular enclosure
heated from below. Diaz and Viskanta [29] observed that
multiple Benard-like cells are formed during initial stages
of the melting of n-octadecane through a bottom heat
source, which grew and merged into a single cell with
time. The growth of buoyancy-driven instabilities that
control the convection heat transfer, along with the upward
movement of the solid–liquid interface, increases the
complexity of the problem [30]. However, Hsu et al. [31]
argued that the timescale of interface movement is much
lower than that of the convective motion and, hence, its
influence on the convective flow can be ignored.
There are limited studies investigating the heat transfer
during phase change, and most of them considered a
sidewall as the heat source. In one of the pioneering
experimental investigations on the PCM (n-octadecane)
Heat transfer during melting of a PCM by a U-shaped heat source
melting inside a rectangular enclosure through a heated
sidewall, Hale and Viskanta [32] reported the significant
effects of convection on the melting pattern, nature of
the solid–liquid interface, and heat transfer coefficients.
In a similar study, Ho and Viskanta [33] experimentally
and numerically investigated the melting of wax inside a
similar chamber. They reported that except for the initial
stages of heat conduction period, natural convection
significantly affected the melting and was the dominant
mechanism and controlled the melting rate and the
solid–liquid interface profile. Gadgil and Gobin [34]
argued that, in such case, once the PCM started melting,
the effects of conduction may be ignored in further heat
transfer analysis. Yuksel et al. [35] developed a
theoretical model to predict the time and temperature
during the charging and discharging of a PCM TES. They
argued that the entry and average properties related to the
system can be implemented in the analysis. They also
observed the influence of internal conduction resistance
during the melting of solid and that of natural convection
in the liquid state.
Recently, Bondareva and Sheremet [36] numerically
simulated the melting process inside a square cavity heated
from the bottom wall. They found that an increase in
Rayleigh number has a profound effect on the flow
behavior as well as heat transfer. They observed the
formation of two thermal plumes, which behaves
asymmetrically as the buoyancy forces dominate over
inertial forces. The authors extended this work by
simulating the melting process in a cube subjected to
bottom wall heating [37]. They found that the velocity
and temperature fields in the middle of the cavity were
similar to the two-dimensional case. At moderate Rayleigh
numbers, they observed single thermal plume, while two
thermal plumes were observed at high Rayleigh numbers.
In a separate study, the authors also observed that an
inclined magnetic field in the cavity influences both flow
and temperature patterns [38].
In an experimental investigation during melting and
solidification processes of pure gallium inside a rectangular
chamber heated from a sidewall, Gau and Viskanta [28]
correlated the changing volume of the melt with the heat
transfer rate through some relevant dimensionless
parameters, like the Nussult number (Nu) and the Rayleigh
number (Ra). In another numerical study for melting of tin
in a 2-D rectangular enclosure heated through a sidewall,
Christiano and Luiz [17] observed that the total melting
time reduced with an increase in the Rayleigh number
(Ra) and argued that this increase (of Rayleigh number)
is the consequence of the enhancement in the natural
convection.
There is a wide variation in PCM-based TES systems
from the design point of view. These include but not
limited to various shapes like flat plate, annular cylindrical
or square tubes, micro/macroencapsulations with different
shapes of the containers, and slurries containing PCM as
well. Furthermore, these storage systems may have
different arrangements of the heat sources based on
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
practical applications. In addition, the heat source
configuration could be a combination of different
geometric arrangements that further complicates the
problem.
A major challenge in designing a PCM-based thermal
storage is the effective heat transfer to and from the PCM
medium during the phase change, which greatly influences
system efficiency [3]. This heat transfer mechanism is
directly affected by the geometric configuration of the heat
source/sink because of buoyancy. The complexity further
arises because of the simultaneous changes in liquid and
solid fractions. The geometric configuration of the heat
source/sink in thermal storage system is likely to be more
complicated than that of the simple heat source
configurations of sidewall or bottom wall reported in the
literature. Hence, there is a need to extend this research
to consider more complicated configurations of the heat
source/sink that are of practical interest and to conduct a
thorough investigation of the phase transformation patterns
and the associated heat transfer.
In real-world thermal storage systems, one practical
way to charge and discharge the heat is via an HTF that
runs through piping embedded into the storage medium
(i.e., PCM). To the best of our knowledge, there is scarcity
of studies investigating the phase change behavior and
associated heat transfer in such configurations. The present
study used a piping configuration embedded into the PCM
that is very practical for real applications, that is, a U-
shaped heat source. No study has been reported in the
literature that investigated the heat transfer and phase
change process in this or similar geometry. Another
importance of this heat source configuration is that it
combines different heat source orientations and their
influence on the melting process. The present study
comprises experimental investigation to improve
understanding of the underlying processes during solid–
liquid phase transition due to a U-shaped heat source.
The specific focus is on the investigation of the transient
solid–liquid interface movement and the associated
convection heat transfer in the melted region.
2. EXPERIMENTAL SETUP
2.1. Thermal energy storage chamber
Experiments were conducted in a thin rectangular chamber
that served as the PCM storage. A schematic and
photograph of the experimental setup are shown in
Figure 1(a) and (b), respectively. The PCM chamber has
the inside dimensions of 100 mm horizontally (L),
100 mm vertically (H), and a depth of 12 mm (W). An
optically clear cast acrylic plate (8 mm thick) cut into
a U-shaped structure formed the shape of the cavity
(Figure 2a). All three sides of this U-shaped plate were
25 mm wide. To form the front and back walls of the
chamber, one 5-mm-thick plain clear glass pane and an
acrylic sheet of the same thickness with the same cross
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
section of 150 mm × 125 mm were used, respectively.
Both the front side and backside of the U-shaped acrylic
plate were separated from the front glass pane wall and
the back acrylic wall, respectively, by two pieces of 2-
mm-thick rubber gaskets, cut into the same size and
contour of the U-shaped acrylic plate as shown in
Figure 2. The rubber gaskets, which were placed all along
the contact area between the main chamber and both the
front and back walls, were helpful in preventing the fluid
leakage and minimizing the heat loss from the chamber
to the surroundings.
A U-shaped tube made of thin-wall smooth copper tube
(Copper Alloy 122, wall thickness 0.355 mm, outer
diameter 4.76 mm) was used as the heat source for the
storage chamber. This copper tube, hereinafter, is referred
to as the HTF tube, because the heat source was a high-
temperature fluid passing through this tube. The tube was
inserted in the chamber from the open top end and was
positioned vertically (Figure 2). A clearance of
approximately 3.5 mm between the tube and the front
and back walls were kept throughout the experiments.
The distance between the two legs (left and right side)
of the U-shaped HTF tube was approximately 50 mm
over its vertical length. The outer edge of each leg was
approximately 22 mm away from the corresponding
sidewall of the chamber (Figure 2b). The maximum
immersion depth of the tube inside PCM was
approximately 95 mm. The total linear length of the tube
in contact with the PCM inside the chamber was
approximately 250 mm. All sides of the chamber except
the front glass wall and the open top end were insulated
(FC-2300-LD, Fibre Cast Inc., Burlington, Ontario,
Canada) to avoid heat losses to surroundings. The storage
chamber was vertically attached to a metal frame using bolt
and nuts. The upper part of the chamber was left open to
room atmosphere to accommodate thermal expansion of
the PCM during melting. The open space at the ceiling of
the chamber helped minimize the pressure buildup due to
the volumetric expansion of the PCM upon melting [39].
The heat transferred through the open top prevented the
heat buildup in the chamber and hence simulated the
condition of a continuous heat transfer in a real thermal
storage as the heat continued to transfer from the heat
source towards the boundaries of the storage domain.
2.2. Thermal energy storage media
A mixture of fine paraffin wax and copolymer alloys in
the form of milk-white pellets, commercially available
as ‘Polyfin’, was used as the PCM in this study. Various
thermo-physical properties of the PCM as per published
literature are listed in Table I. One of the key features
of ‘Polyfin’ for its suitability for the present research is
that it becomes transparent once melted, which was a
very useful and crucial feature for accurately imaging
the melting process and tracking the solid–liquid
interface. In addition, it is safe to handle and its melting
temperature is in the safe range of operation in the lab
Bashar M. and Siddiqui K. Heat transfer during melting of a PCM by a U-shaped heat source

environment. It is to be noted that paraffin wax has been
widely used in experimental research studies on PCMs
[29,32,40–42]. Very small amount of silver-coated
hollow glass sphere having an average diameter of 9-
13 μm was used along with the PCM as the tracer
particles to investigate the behavior of the molten liquid
flow through imaging. The volume change of the wax in
the liquid stage was estimated to be approximately 10%
from the solid state.
2.3. Heat source
As mentioned earlier, a U-shaped HTF tube acted as the
heat source in this study. The HTF (hot air) continuously
passed through the tube at a controlled flow rate and
temperature. In order for the pressure fluctuations in the
HTF to be minimized, the incoming compressed air from
the laboratory supply line was initially passed through a
settling tank to dampen pressure fluctuations. Downstream

Figure 1. (a) Schematic diagram of the experimental setup. (b) Photograph of the experimental setup. The U-tube heat source is shown
inside the phase change material (PCM) chamber. DAQ, data acquisition. [Colour figure can be viewed at wileyonlinelibrary.com]

Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source Bashar M. and Siddiqui K.

of the air settling tank, a pressure regulator was employed
to control the pressure of the air in the system. The air was
then passed through a flowmeter to control its flow rate
before passing through the heater (Figure 1). The
controlled air was then heated while passing through an
air heater (Zesta ZA0017) equipped with a PID controller
(Zesta ZEL-9100) to precisely control the temperature of
the HTF fluid. The accuracy of the PID controller was
within ±2°C. The air flow rate was maintained at
40 L/min. The accuracy of the flow meter was estimated
as ±5%. Four different controlled temperatures of the
HTF were used as inlet temperatures (150°C, 175°C,

Figure 2. Schematic of the phase change material (PCM) chamber. (a) Isometric view. (b) Schematic diagram. Locations of the
thermocouples are marked with black open circles. HTF, heat transfer fluid. [Colour figure can be viewed at wileyonlinelibrary.com]

Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
Table I. Thermo-physical properties of phase change material at 25°C (VWR CA 27900268) [43,44].
Thermal conductivity
Melting point Specific Specific heat
o (W/m/K)
( C) gravity (kJ/kg/K)
55 0.92 2.890 Solid Liquid
~0.24 ~0.18
200°C, and 225°C) by adjusting the heater controller,
which provided four different input heat fluxes of 3450,
4370, 5020, and 5840 W/m2, respectively.
2.4. Temperature data
A total of 28 T-type thermocouples (T30-2-507; Zesta
Engineering Ltd., Toronto, Ontario, Canada), hereinafter
referred to as TC, were used to measure the temperatures
at various locations in the experimental setup, shown in
Figure 2(b) as small black open circles. Out of these 28
TCs, two TCs were placed in direct contact with the air flow
to measure its (HTF) temperatures at the inlet and outlet of
the PCM chamber. The temperature on the surface of the
HTF tube inside the chamber was measured at 11 different
locations along the tube length. Out of these 11 TCs, 10
TCs were affixed on the surface of the two vertical legs (five
TCs on each leg)) with an equal spacing of about 20 mm
between them, and the other TC was fixed in the middle
of the curved part of the U-shaped tube (Figure 2b). All
these 11 TCs were affixed at the backside of the tube to
avoid any imaging obstructions. For measuring the
temperature inside the PCM domain, 15 TCs were arranged
in an array format (five rows and three columns), with one
column each on the left, right, and middle of the U-tube
(Figure 2). The left and right columns were located 14 mm
from their respective sidewalls. The center TCs’ column
was located along the vertical centerline of the chamber
and tube, at a distance of 50 mm from each sidewall. The
vertical distance between each TC row was also kept at
20 mm, and they were at the same vertical height as that
of the tube surface TCs (Figure 2b). The bottom TCs’
row was located approximately 5 mm above the curved
surface of the HTF tube near the bottom of the chamber.
All TCs in the PCM domain were inserted individually into
the chamber through the insulated acrylic back wall.
The temperature data from all 28 TCs were recorded
simultaneously at a rate of 3 Hz via data acquisition cards
(TC modules: NI-9214 and NI-9211) controlled by the
LABVIEW software. For accuracy and similarity to be
ensured, the temperature data recording for all
experimental cases was started 1 min prior to the beginning
of the heating process and continued 5 min after the PCM
was completely melted.
2.5. Calibration of thermocouples
Calibration of all TCs was accomplished by using an
insulated hot water bath. A high-accuracy thermometer
Heat transfer during melting of a PCM by a U-shaped heat source
Kinematic viscosity Latent heat of fusion Thermal diffusivity
2 2
(m /s) (kJ/kg) (m /s)

6
8
5.50 × 10 180 7.985 × 10
(Kessler 10/30 ST Jointed Thermometer) was used as the
reference. The calibration was conducted over a
temperature range from 90°C to 0°C with increments of
about 10°C between each reading. All calibration curves
showed almost linear relationship. The uncertainty of
temperature measurements from TCs was estimated as
±0.4%.
2.6. Phase change material chamber
insulation
As mentioned earlier, the front wall was made of a clear
glass pane. This was required for this study because the
clear glass pane allowed optical access to image the
melting process inside the chamber by a high-resolution
camera. This was the only uninsulated wall of the chamber,
and hence, some heat loss occurred through this glass pane.
Therefore, it was necessary to ensure that this heat loss
from the front wall did not influence the melting behavior
inside the chamber. For this purpose, another set of
experiments was conducted under the same conditions, in
which the front glass pane was also covered with the same
insulation. Note that only temperature measurements were
carried out in this set of experiments because of no optical
access. The temperature data obtained from this set of
experiments were compared with the temperature data
from the corresponding cases with the uninsulated front
glass pane. The results show that the PCM temperature
within the domain at each heat flux case was slightly
higher for fully insulated chamber than for the uninsulated
front glass pane, as expected. The difference in the mean
temperatures for the two chamber configurations was on
average 7.5%. Because the melting process was transient
in nature, to further confirm that the melting behavior
was similar, we also compared over the entire melting
period for both insulated and uninsulated cases, the
temperature profiles at different locations within the PCM
domain. The results show that the temperature profiles
for both insulated and uninsulated cases were very similar,
hence confirming that the uninsulated pane had no
considerable influence on the transient melting behavior.
2.7. Image acquisition and image
processing
A 4-megapixel high-resolution charged-coupled device
camera (Flare 4M180-CL) with a resolution of
2048 × 2048 pixels was used to capture the images of the
phase change process and to track the solid–liquid
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
interface and the flow behavior of the melted PCM. The
camera was connected to an image acquisition system
(DVR express core; IO Industries, London, Ontario, Canada)
to record images. The camera was controlled via the DVR
express CoreView software installed on a PC. The field of
view of the camera was set at 94 mm × 94 mm. Two 13-W
LED lights were used to enhance the image quality. The
lights were placed more than 50 cm away from the front
wall of the chamber. Considering the power, distance,
and nature of light emitted by an LED, the radiation effects
on the PCM chamber wall were assumed to be negligible.
For each run, images were captured from the beginning of
the melting process until the PCM was completely melted.
The camera frame rate was set at 40 frames per second
(40 Hz) for different experimental runs. The stored images
were transferred from the DVR core to an external
storage media in suitable format for further processing.
2.8. Experimental data analysis
The heat gain by the PCM was considered equal to the heat
loss by the HTF passing through the PCM domain inside
the chamber and was calculated from the following
equation [12]:
q_ total ¼ mc
_ p ðT in
T exit Þ (1)
where q_ total is the heat transfer rate, m_ is the mass flow rate
of air, cp is the specific heat of air, and Tin and Texit are the
inlet and exit temperatures of the air (HTF). Because the
TCs to measure inlet and outlet temperatures were located
approximately 1 cm off from the physical entrance and exit
of the PCM domain, the heat loss from the HTF tube
between Tin and Texit TC locations and the chamber inlet
and outlet locations, respectively, were calculated. The
net heat transferred to the PCM (q_ net ) was then calculated
by subtracting the inlet/outlet heat loss, from the total heat
00
lost by the HTF. The net heat flux (qnet ) into the PCM
domain was computed by dividing ( q_ net ) by the surface
area of the HTF tube inside the domain. The uncertainty
in estimating the heat flux was on average within
approximately ±10%.
As the liquid fraction continued to increase with the
progression of melting, the local heat transfer coefficients
were computed when the melted fraction reached up to a
particular TC. Because of the nature and pattern of melting
in a U-shaped tube configuration (heat source), the local
heat transfer coefficient (h) was measured based on the
change in the vertical extent of the melted PCM in the
vicinity of the heat source. It is the same approach used
for natural convection heat transfer from vertical tubes
[40]. Hence, the local heat transfer coefficient values were
obtained in each zone with the change in the height of the
melted PCM. The vertical locations of TCs were
considered as the reference lengths; that is, the local heat
transfer coefficient in each zone was computed when the
melted PCM domain reached subsequent TC within a
given column in the respective zone. At a given vertical
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
location, the local heat transfer coefficient was computed
using the equation
00
q ¼ h ðT s
T bulk Þ (2)
00
where q is the heat flux assumed to be equal to the average
heat flux through the HTF tube, Ts is the local surface
temperature of the HTF tube, and Tbulk is the local bulk
temperature of the melted PCM. Note that the tube surface
temperature and bulk liquid temperature were computed as
the average from all corresponding TCs that were
immersed in the liquid PCM in each zone.
2.9. Interface detection
Tracking the solid–liquid interface and detecting its form
as the melting progressed provided an in-depth insight
for the better understanding of the underlying process.
Using an in-house algorithm in a Matlab platform, the
solid–liquid interface was detected in each image over
the whole melting process. The algorithm initially
preprocessed raw images to increase their signal-to-noise
ratio by rescaling the gray-scale values between solid and
liquid phases (upper end gray-scale values correspond to
the solid PCM, and lower end gray-scale values
correspond to the liquid PCM) for accurate identification
of each phase and its transformation into a binary format
for the precise detection of the interface. A series of
morphological operations were performed to remove any
noise from the data. The coordinates of the detected
interface in the binary image were recorded at every spatial
location, which when processed generated the actual
profile and position of the solid–liquid interface at a given
time. Because of the physical presence of various TCs in
the PCM domain, the interface profile in the close
Figure 3. A snap shot showing the real image of the chamber
during phase change material melting for one experimental run
2
(heat flux of 4370 W/m ). The interface is marked as the red line.
The dark-gray area represents the melted phase change
material, and the white area represents the solid phase
(excluding the tube legs). [Colour figure can be viewed at
wileyonlinelibrary.com]
Bashar M. and Siddiqui K. Heat transfer during melting of a PCM by a U-shaped heat source

proximity of a TC was locally influenced. Hence, in the
subsequent analyses, the interface data in the locality of
each TC were excluded. The uncertainty in detecting the
interface was within ±2 pixels, which correspond to
±90 μm. Figure 3 shows an instantaneous image of the
PCM domain. The detected solid–liquid interface in the
corresponding image is also plotted with a red line.
influence of the geometry of the heat source. The
temperature data acquired throughout the storage chamber,
along with the simultaneous imaging, provided detailed
information about the thermo-fluidic behavior during the
solid–liquid phase change process.
3.1. Instantaneous temperature fields and
melting behavior

3. RESULTS It was mentioned earlier in the experimental setup section

that the temperature was measured at 26 locations inside
The objective of this study was to study the heat transfer the chamber (11 locations on the surface of the tube and
phenomena during the phase change process and the 15 locations inside the PCM domain) to characterize the

Figure 4. (Left) Colormaps of the temperature field in the phase change material (PCM) chamber at different stages of melting. Center
area marked by black lines in U-shape profile indicates the location of the heat transfer fluid tube. (Right) corresponding image of the
PCM domain. Dark-gray region corresponds to the liquid PCM, and white region corresponds to the solid PCM (excluding the U-tube).
[Colour figure can be viewed at wileyonlinelibrary.com]

Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
thermal behavior of the PCM during the melting phase.
The variation of the temperature field inside the chamber
at different stages of melting is illustrated in Figure 4 (left
side) along with the corresponding images showing the
melting trends through the front glass panel (right side).
Figure 4(a) shows the data soon after the beginning of
the heating process. At this stage, heat transfer is in the
early phase via conduction through the PCM. The heat lost
by the HTF is evident through a gradual decrease in the
tube surface temperature. The bulk PCM temperature at
this stage is not significantly influenced by the heat transfer
and the PCM domain remains in the solid phase (see the
corresponding image on the right).
Figure 4. (continued)
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
As the heat transfer to PCM continued and once the
melting temperature of the PCM was reached, the solid
phase of the PCM started to melt. The U-tube surface in
contact with the PCM naturally was at the highest
temperature near the top left corner of the chamber where
the HTF entered the PCM domain. Because the HTF was
continuously losing heat to the PCM, the temperature of
the HTF decreased along the tube, which caused an
increase in the PCM domain temperature as well as a
decrease in the tube surface temperature. Because of the
higher HTF and hence the tube surface temperatures, the
melting process initiated near the top left corner of the
chamber (Figure 4b). As the time passed, the melted zone
Bashar M. and Siddiqui K.
started to expand downward and horizontally. Thus, the
pool of melted PCM acquired the shape of an inverse cone
and continued to grow.
Following the formation of the melted zone in the
entrance region of the HTF, another smaller melted region
having a similar conical shape quickly developed around
the HTF tube where the HTF exited the PCM domain near
the top right of the chamber (Figure 4c). This is an
unexpected trend because as the HTF loses its temperature
along the tube, one expects the melting front to progress
along the same path. A plausible explanation for the
observed trend is described as follows. As mentioned
earlier, the melted zone formed in the HTF entrance zone
started to expand horizontally and vertically, indicating
the movement of the solid–liquid heated interfacial front.
The heat transfer through this interfacial front was higher
because the convection heat transfer that is induced in the
liquid PCM enhances the rate of heat transfer relative to
the conduction. Thus, the heat more rapidly transferred
horizontally from the left leg of the tube at the upper end
of the PCM towards the right leg as well as the left
sidewall. But because the heat was also being transferred
from the right leg near the upper end, the PCM temperature
in the vicinity of the right leg reached the melting point
faster than the other regions of the PCM. Another
interesting behavior observed at this stage was the change
in the trend of the tube surface temperature, which
corresponds to the similar changes in the HTF temperature
inside the tube. As Figure 4(a) shows, the tube surface
temperature decreased gradually from the inlet to the outlet
before the commencement of melting. However, once the
melting began and the heat started to propagate more
rapidly in the horizontal direction, the surface temperature
of the right leg near the outlet became higher than that of
the tube on the immediate upstream side almost up to the
tube bend at the bottom of the chamber. A plausible
explanation for this trend is described in the succeeding
texts. When the PCM is melted, because of the buoyancy
effect and convective motion the warmer liquid PCM
started to rise and accumulated near the top. Furthermore,
the liquid PCM also acted as a bridge to directly transfer
heat from the left leg to the right leg, horizontally. As a
result, the temperature difference between the tube surface
and the liquid PCM is lower near the top exit than near the
bottom. Hence, the heat transfer rate near the bottom is
higher than that near the top exit. Inside the tube, the hot
air transferred heat to the tube wall. While the air
temperature decreased because of the heat loss, but
considering the smaller geometry of the setup, this
temperature drop is not very large particularly in the
section between the bottom and the exit. Hence, the
temperature difference between the air and the tube wall
near the top exit is smaller than that near the bottom. This
resulted in a slightly higher tube wall temperature near the
exit. Such trend is not likely to occur in a large setup.
As the melting further progressed with time, the melted
fraction continued to grow in the upper region of the PCM
chamber surrounding the two tube legs and a new melted
Heat transfer during melting of a PCM by a U-shaped heat source
zone started to develop just above the U-bend of the HTF
tube in the lower mid region of the chamber and the overall
temperature of the melted PCM continued to rise. The
formation of this third zone above the tube bend was due
to the higher contribution of buoyancy. The heat source
in this region was near horizontal, where the buoyancy
forces supported the transport of heat to the PCM above
the tube; that is, the buoyancy forces aligned with the
direction of the solid–liquid interfacial front. As the
melting process further continued, all three melted zones
continued to grow (Figure 4(d)) and eventually merged
(Figure 4e). A blob of solid PCM remained in the middle
of the U-tube, which eventually melted. The results also
indicate that the temperature gradients are stronger in the
melted PCM than in the solid PCM. It is observed from
the image sequence in Figure 4 that the melting was more
rapid in the PCM enclosed by the U-tube (inner zone) that
in the region surrounding the U-tube (outer zone). Once
the PCM in the inner zone was fully melted, the melting
rate increased in the outer zone. The image sequence also
showed that the solid–liquid interface was nonlinear in all
three zones and its nature continued to change with the
growth of these zones. The interface acquired a relatively
linear form once the PCM was fully melted in the inner
zone.
It is also observed that the melted zone did not extend
much beneath the U-tube bend. This is likely due to the
reason that the heat transfer from the tube was dominated
by buoyancy-driven convective motions, which were
present above the tube bend. Hence, the heat transfer to
the PCM was primarily above the bend. This further
indicates that in a real PCM-based thermal storage, the heat
source should be designed in a way that the supply heat
should also reach closer to the bottom end of the storage,
which otherwise will not be able to effectively store heat.
One option to achieve this is by adding extended surfaces
or fins to the HTF tube that penetrate into the bottom layer
and induce convective motion.
Figure 5. Instantaneous melted fraction versus time for four
input heat fluxes. [Colour figure can be viewed at
wileyonlinelibrary.com]
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
3.2. Melting rate and melt fraction of phase
change material
Using an in-house algorithm, we analyzed the images
recorded during the melting process to compute the melted
fraction (mf) in each image, which is the ratio of the melted
PCM volume to the total PCM volume in the chamber. The
change in the total melted fraction with time that expresses
the instantaneous melting rate is plotted in Figure 5 for all
four heat flux cases. As expected, the melting rate of the
PCM was highest for the largest heat flux, which decreased
monotonically with a decrease in the heat flux. From the
graphs, it is observed that at the initial stages for a small
period of time, there was no melting at any heat flux. This
time duration corresponds to the period during which PCM
in the vicinity of the tube reached its melting temperature
after the HTF fluid started passing through the PCM
domain.
From Figure 5, it is also observed that the melting rate
was comparatively higher up until approximately 65–
70% of the total PCM volume was melted and, after that,
the melting rate decreased. A plausible explanation for this
trend is described later. As seen earlier in Figure 4, the
PCM in the inner region melted first and at a higher rate
because of the presence of two heat sources (i.e., left and
right legs of the tube). Thus, the higher melting rate
observed in Figure 5 corresponds to the melting of the
PCM in the inner region and the slower rate corresponds
to the melting of the PCM in the outer region.
3.3. Interface profile
As shown in Figure 3, the solid–liquid interface was
detected using an in-house algorithm. The algorithm was
used to detect the interface developed in the upper region
of the chamber because of the melting caused by the left
Figure 6. Time history of the solid–liquid interface position
(shown at partial melting phase). The time interval between
each interface line is 66.67 s. PCM, phase change material.
[Colour figure can be viewed at wileyonlinelibrary.com]
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
and right legs of the tube. The purpose was to investigate
the melting behavior and the associated heat transfer due
to the vertical legs of the heat source. The bottom melting
near the U-tube bend was not considered because the
interface detection and analysis would have become very
complex. The interface was tracked up until the bottom
melting zone merges with the top melting zones. Tracking
of the interface in each image allowed to analyze the
transient behavior of the interface dynamics. Figure 6
shows the time history of the detected interface at the heat
flux of 4370 W/m2 as an example.
From the profile dynamics, it is clearly evident that the
interface movement was nonlinear in both space and time.
The interface moved more rapidly near the heat sources as
expected. The results also show that the interface
movement was not symmetric about a given heat source
(left or right leg); that is, the interface movement was
relatively faster on the heat source facing the inner region.
Similarly, the interface movement was relatively faster at
the left heat source, which was the heat inlet, as expected.
Because the HTF has to enter from one leg and exit from
other, this behavior is the practical reality because of
geometry of the heat source. Hence, for a better
characterization of this behavior, the PCM chamber was
divided into four separate imaginary zones as shown in
Figure 7, which are denoted as zones A, B, C, and D. It
is to be noted that the boundary line between zone B and
zone C was commonly shared by the column of five TCs
vertically arranged along the centerline of the chamber
(Figure 2). Hence, the same set of temperature data
obtained from these five TCs was used for both zone B
and zone C to calculate each zone’s parameters. Various
thermo-physical parameters were calculated by
Figure 7. Schematic of the phase change material chamber
showing four separate imaginary zones marked as A, B, C, and
D by dashed lines. [Colour figure can be viewed at
wileyonlinelibrary.com]
Bashar M. and Siddiqui K.
determining the local parameters first, and then the average
values were taken into account to find the global
parameters.
3.4. Local convective heat transfer
coefficient (h)
The convective heat transfer coefficient (h) in the liquid
(melted) PCM was calculated using Newton’s law of
cooling [12]:
00
 keep similarity in computations, data from only three upper
q ¼ h T Surf
T Bulk (3)
00
where q is the heat flux, TSurf is the average surface
temperature of the HTF tube that is adjacent to the liquid
PCM, and TBulk is the average temperature of the liquid
PCM. As the liquid domain of the PCM expanded with
time, the local heat transfer coefficients associated with
the top melting zones, that is, formed by the left and right
vertical heat sources (zones A, B, C, and D), at different
instances were computed using the aforementioned
equation. The local values of h were computed at time
Figure 8. Values of the local convective heat transfer coefficient
(h) at different heights in the chamber. Results are presented for
2
two input heat fluxes, (a) 3450 and (b) 5840 W/m . [Colour figure
can be viewed at wileyonlinelibrary.com]
Heat transfer during melting of a PCM by a U-shaped heat source
instances when the liquid fraction reached consecutive
TCs embedded in the PCM domain. Furthermore, similar
to the interface detection, the local values of h were
computed up until the bottom melting zone merges with
the top melting zones. A final value of h was calculated
when the inner region was fully melted. It is to be noted
that in the moment when the central region bounded by
the inner area of the U-tube (zones B and C) was
completely melted, some parts at the lower regions in
zones A and D were still not melted completely. Thus, to
TCs were considered for local h calculations.
The values of the local convective heat transfer
coefficients (h) were calculated in a transient manner as
the melted PCM expanded for all heat flux cases.
Figure 8(a) and (b) shows the values of h for the lowest
and highest heat flux cases (3450 and 5840 W/m2),
respectively. In the plots, y is measured from the top of
the chamber and H is the height of the chamber. Hence,
y/H = 0 corresponds to the top of the PCM chamber. The
results show a general trend that the heat transfer
coefficient in the liquid PCM increased with the expansion
of the melted domain in each chamber zone (A, B, C, and
D) on both sides of each HTF tube leg. This increase in the
local magnitude of h is due to two main reasons. The first is
the increase in the HTF tube contact length with the liquid
that strengthens the magnitude of the buoyancy-induced
flow, and the second is the expansion of the liquid PCM
zone that allows the growth of convective cells formed in
the liquid zone. These convective cells increase the
transportation of heat from the source to the other
boundaries of the liquid zone.
At a given heat flux, the results show a clear influence
of the given PCM zone relative to the heat source
orientation; that is, the local values of h were significantly
larger in the PCM zones within the inner region (zones B
and C) than in the outer region (zones A and D). It is found
that the local values of h in the inner region are on average
about 35% higher than the local values of h in the outer
region at both the highest and lowest heat fluxes.
Relatively larger melted zone volume and higher liquid
contact length of the HTF tube in the inner region are the
likely causes for higher h values in that region.
Comparison of Figure 8(a) and (b) shows that the
magnitude of h increased with an increase in the heat flux
in the inner region. The results also show that the variation
of h with height became more linear with an increase in the
heat flux. This is likely due to the reason that an increase in
the heat flux increases the buoyancy magnitude, which
resulted in better mixing.
Figure 8 presented the transient variation in the local h
in four chamber zones at the lowest and highest flux cases.
For the results to be further generalized, the average value
of the local h in each chamber zone is presented in Figure 9
for all heat flux cases. The trends are similar to that
observed in Figure 8; that is, the highest values of the heat
transfer coefficient, h, are observed in zone B and then
zone C, which are in the inner region. The outer region
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
Figure 9. The average values of local convective heat transfer
coefficient (h) in the four chamber zones at four input heat
fluxes. [Colour figure can be viewed at wileyonlinelibrary.com]
(zones A and D) has almost similar magnitudes. Results
also show that in each zone, the value of h increased
monotonically with an increase in the heat flux. It is found
that the value of h in the inner region (zones B and C)
increased approximately by 10% as the heat flux increased
from 3450 to 5840 W/m2, while in the outer region (zones
A and D), the value of h increased approximately by 5%
over the same heat flux range.
3.5. Local Rayleigh number (Ra)
Rayleigh number (Ra) is a non-dimensional parameter that
characterizes the magnitude of buoyancy-driven force to
viscous dissipative force in the fluid domain and is
calculated using the following equation [12]:
Figure 10. Average Rayleigh number (Ra) in different chamber
zones at four input heat fluxes. [Colour figure can be viewed at
wileyonlinelibrary.com]
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
gβðT S
T B Þl 3
Ra ¼ (4)
να
where g is the acceleration due to gravity, β is the
coefficient of thermal expansion of the liquid PCM, TS is
the surface temperature of the HTF tube, TB is the bulk
temperature liquid PCM domain, l is the characteristic
length (in the present case, it is the liquid contact length
of the HTF tube), ν is the kinematic viscosity, and α is
the thermal diffusivity of the liquid PCM. As the HTF tube
bounded one side of each zone, that side was used to
compute the characteristic length for the corresponding
zone. The local variations in the Rayleigh number were
calculated in the four chamber zones in the transient
manner similar to the computation of the local heat transfer
coefficient. The average values of the local Rayleigh
number in different zones at all heat fluxes are shown in
Figure 10. The results show the highest values of Rayleigh
number in zone B, which is followed by that in zone C.
The values in the outer region (zones A and D) were in
general further lower. While overall the trends in Rayleigh
number variations are similar to that of the heat transfer
coefficient (Figure 9), one prominent difference was in
zone C. The values of heat transfer coefficient in zones B
and C were relatively comparable, indicating that the flow
patterns in both zones are quite similar. However, the
Rayleigh number in zone C was significantly lower than
that in zone B and approximately in the same range as
the Rayleigh number in the outer region. A plausible
explanation for this trend is that the Rayleigh number is
primarily influenced by the fluid temperature rather than
the fluid flow behavior. As seen earlier in Figure 5, the
HTF tube surface temperature was significantly higher in
zone B than in zone C, hence a higher liquid PCM
temperature in zone B, which caused a higher reduction
in the liquid viscosity. The results also show that the
Rayleigh number increased almost monotonically with
the heat flux in the chamber zones. It is observed that the
Rayleigh number on average increased by 23% in the inner
zone and 18% for the whole domain as the heat flux
increased from 3450 to 5840 W/m2.
3.6. Determination of the bulk parameters
The transient local behavior of the heat transfer
coefficient in different spatial zones is presented earlier.
The overall behavior can be characterized in a transient
manner by averaging the local heat transfer coefficient
values in different zones at a given height that
corresponds to the distance of the propagating melting
front (which is moving in a transient manner). However,
as seen in Figure 4, the PCM in the inner region (zones
B and C) was fully melted, while the bottom section in
the outer region (zones A and D) never melted. Thus,
the overall transient behavior of the heat transfer
coefficient was characterized by averaging the local heat
transfer coefficient values in the inner region (zones B
and C) only. The results are presented in Figure 11.
Bashar M. and Siddiqui K.
Figure 11. Average values of convective heat transfer
coefficient (h) at different heights of the chamber in the inner
region of the phase change material domain. Results are
presented for four input heat fluxes. [Colour figure can be
viewed at wileyonlinelibrary.com]
The results show that the local heat transfer coefficient
continuously increased as the liquid domain continued
to expand and the heat transfer coefficient reached a
maximum value when the inner region was fully melted.
As mentioned earlier, the plausible cause is the stronger
buoyancy-induced flow and larger domain for the
convective cells to grow. The plot shows a similar trend
of the heat transfer coefficient increase at all heat fluxes,
while the magnitude of h at a given height increased
monotonically with the heat flux. The results show that
at the lowest heat flux, the value of h increased by
40% as the vertical distance increased from y/H = 0.15
to y/H = 1. Over the same vertical distance range, the
value of h increased by 55% at the highest heat flux.
For the bulk behavior to be further characterized, the
heat transfer coefficient values averaged over space and
time are plotted in Figure 12 at all heat flux cases. The
results as expected show that the overall heat transfer
coefficient increased with an increase in the heat flux;
Figure 12. Convective heat transfer coefficient (h) as a function
of input heat flux (q00 ). [Colour figure can be viewed at
wileyonlinelibrary.com]
Heat transfer during melting of a PCM by a U-shaped heat source
however, the increasing trend is nonlinear. The results also
show that the overall heat transfer coefficient increased by
approximately 10% as the heat flux increased from 3450 to
5840 W/m2.
3.7. Nusselt number (Nu) correlation
The heat transfer behavior in a given configuration during
natural convection is typically generalized in terms of the
relation between Nusselt number, Nu (a ratio between the
convective and conductive heat transfer), and Rayleigh
number, Ra. Such relationship provides an understanding
of how the convection heat transfer is influenced by the
buoyancy effects. The local Nusselt number was
computed as [12]
Nu ¼ hl=k (5)
where h is the local convective heat transfer coefficient, l
is the corresponding local characteristic length (in the
present case, it is considered as the liquid contact length
of the HTF tube), and k is the thermal conductivity of
the liquid PCM. The local Nusselt number is plotted
versus the local Rayleigh number in Figure 13 for all heat
flux cases. The results show as expected that the local
Nusselt number increased with an increase in the
Rayleigh number. To the best of authors’ knowledge,
the results in Figure 13 provide, for the first time, the
relation between the Nusselt and Rayleigh numbers in
the transient melting process using a U-shaped heat
source. A generalized expression for this relation is
obtained using a power law fit to the experimental data,
which is presented as
Nu ¼ 0:0071 Ra0:4501 ; 3:0106 < Ra < 1:5109 (6)
The geometry considered in the present study
comprised a combination of vertical (side) heat sources
and a bottom heat source. In the literature, a well-
Figure 13. Nusselt number (Nu) versus Rayleigh number (Ra) at
different heat fluxes. Empirical correlations obtained from Lim-
Bejan [49] and LeFevre-Ede [48] are also plotted for comparison.
[Colour figure can be viewed at wileyonlinelibrary.com]
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
established Nu–Ra correlation is reported for heat transfer
from a vertical cylinder [45–47] that was originally
proposed by LeFevre and Ede in 1956 [48]. The
correlation is presented below:
 41
4 7RaH Pr 4ð272 þ 315 PrÞH
NuH ¼ þ (7)
3 5ð20 þ 21 PrÞ 35ð64 þ 63 PrÞD
where H represents the height of the chamber, D is the
diameter of the heat source, and the dimensionless
Rayleigh and Prandtl numbers are represented by Ra and
Pr, respectively. It is to be noted that, for this correlation,
the medium considered was a single-phase fluid.
Lim and Bejan [49] also investigated a two-phase
system and proposed a correlation for the melting phase
of a PCM where a vertical wall acted as the heat source.
Their correlation is expressed as
1
0:35 Ra4
Nu ¼ h i49 (8)
9 increased with time.
1 þ ð0:143=PrÞ16
These correlations are also plotted in Figure 13 for
comparison. It is observed that the present results agree
more closely with the correlation of Lim and Bejan [49]
than with LeFevre and Ede [48]. It is expected because
Lim and Bejan’s [49] correlation is for the phase change
process with a vertical heat source, while LeFevre and
Ede’s [48] correlation is for single-phase convection from
a vertical heat source. A close comparison of the present
results with Lim and Bejan’s [49] correlation show that
the Lim and Bejan [49] correlation predicts higher Nusselt
number in the lower range of Rayleigh number; however,
as the Rayleigh number increases, the Nusselt number in
the present geometry exceeds the Nusselt number
predicted by Lim and Bejan’s [49] correlation. This
difference in the trend is due to the differences in the
geometry of the heat source. The present results in the
lower range of Rayleigh number corresponds to the
duration when the PCM started to melt from the top and
propagated downward and horizontally until it merges with
the melted PCM zone emerging from the bottom. The
correlation of Lim and Bejan [49] was based on the
constant temperature condition of the vertical heat source,
while in the present case, the temperature of the HTF tube
surface was decreasing because of heat loss from the HTF
(Figure 4). Hence, the buoyancy forces in this case were
relatively weaker than those in the case with a constant
temperature heat source. This could likely be the reason
for the lower Nusselt number values in this Rayleigh
number range. The comparison also shows that beyond this
lower Rayleigh number range, the Nusselt number in the
present configuration started to become larger than that in
Lim and Bejan [49]. This is likely due to the reason that
beyond this Rayleigh number, the bottom melted zone
merged with the top melted zones (Figure 4e), which
enhanced the convective currents across the domain
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
because of the additional contribution from the bottom heat
source. The Nusselt number values in the higher range of
Rayleigh number correspond to the condition when the
PCM in the inner region was fully melted (Figure 4f). In
this condition, the buoyancy-induced convective currents
from the bottom tube surface supported the flow induced
by buoyancy from the vertical sections of the tube and
played a significant role in the overall convective motion,
hence resulted in higher Nusselt number values than that
predicted by Lim and Bejan’s [49] correlation because
such heat source was not present in the configuration of
Lim and Bejan [49]. Comparison of the present correlation
with that of LeFevre and Ede [48] shows that at a given
Rayleigh number, the Nusselt number values predicted
by LeFevre and Ede’s [48] correlation are higher than
those by the present study. A plausible explanation for this
difference is that LeFevre and Ede’s [48] correlation
considers single-phase fluid in the entire domain, which
influenced the convective flow from the initial stage of
heating, while in the present case, the liquid domain
4. CONCLUSIONS
An experimental study was conducted to investigate a
time-dependent heat transfer process during the melting
phase of a PCM inside a rectangular chamber that
embedded a U-shaped heat source. The temperature fields
and the melting behavior were recorded via TCs and
imaging system, respectively, at four input heat fluxes.
The presented results showed that the geometry of the
heat source has a significant effect on the solid–liquid
interface dynamics throughout the melting period as well
as the temperature field. The results further show that
the melting rate also changed nonlinearly over the melting
phase. The simultaneously recorded temperature data and
the images were analyzed, and that enabled to compute
various important thermo-physical parameters of the
PCM during the melting process to help better understand
the overall heat transfer mechanism. The results show that
the heat source geometry has a significant influence on the
local heat transfer coefficient. The heat transfer coefficient
in the region bounded between the U-tube legs was about
35% higher than that in the region outside the U-tube. It is
also observed that the overall convective heat transfer
coefficient increased rapidly at the early stage of melting
and the rate of increase gradually decreased at the latter
stage of melting. As the vertical distance increased
from y/H = 0.15 to y/H = 1, the value of h increased by
40–55% as the input heat flux increased from 3450 to
5840 W/m2. Over the same input heat flux range, the
Rayleigh number increased by approximately 23% in the
inner region and 18% in the whole domain. A new
correlation for the Nusselt number and Rayleigh number
is proposed to characterize the heat transfer during the
melting phase with a U-shaped heat source.
Bashar M. and Siddiqui K.
ACKNOWLEDGEMENTS
The authors are thankful to the Natural Sciences and
Engineering Research Council of Canada (NSERC) and
the University of Western Ontario for the research support.
REFERENCES
1. Murray RE, Groulx D. Experimental study of the
phase change and energy characteristics inside a
cylindrical latent heat energy storage system: part 2,
consecutive charging and discharging. Renewable
Energy 2015; 63:724–734.
2. Dinçer I, Rosen MA. Thermal Energy Storage:
Systems and Applications (2nd). Wiley: Chichester,
UK, 2011. ISBN:9780470747063.
3. Viskanta R. Solar heat storage: latent heat materials,
volume i: background and scientific principles. In
Chapter 5: Phase-change Heat Transfer, Lane GA
(ed.). CRC Press: Boca Raton, FL, 1983.
ISBN:0849365856.
4. Chiu JNW. Heat transfer aspect of using phase change
material in thermal energy storage applications,
licentiate. PhD thesis, KTH School of Industrial
Engineering and Management, Division of Heat and
Power Technology, SE-100 44 Stockholm, Sweden,
2011.
5. Akgun M, Aydin O, Kaygusuz K. Experimental study
on melting/solidification characteristics of a paraffin as
PCM. Energy Conversion and Management 2007;
48(2):669–678.
6. Sharma A, Tyagi VV, Chen CR, Buddhi D. Review on
thermal energy storage with phase change materials
and applications. Renewable and Sustainable Energy
Reviews 2009; 13:318–345.
7. Ettouney H, El-Dessouky H, Al-Ali A. Heat transfer
during phase change of paraffin wax stored in
spherical shells. Journal of Solar Energy Engineering
2005; 127(3):357–365.
8. Feng Y, Li H, Li L, Bu L, Wang T. Numerical
investigation on the melting of nanoparticle-enhanced
phase change materials (NEPCM) in a bottom-heated
rectangular cavity using lattice Boltzmann method.
International Journal of Heat and Mass Transfer
2015; 81:415–425.
9. Paris J, Falardeau M, Villeneuve C. Thermal storage
by latent heat; a viable option for energy conservation
in buildings. Energy Sources 1993; 15:85–93.
10. Cabeza LF (Ed). Advances in Thermal Energy
Storage Systems: Methods and Applications.
Woodhead Publishing: Cambridge, UK. ISBN:978-
1-78242-088-0.
Heat transfer during melting of a PCM by a U-shaped heat source
11. Farid MM, Khudhair AM, Siddique AR, Al-Hallaj S.
A review on phase change energy storage: materials
and applications. Energy Conversion and Management
2004; 45:1597–1615.
12. Incropera FP, Wewitt DP, Bergman TL, Lavine AS.
Fundamentals of Heat and Mass Transfer (7th). John
Wiley & Sons, ISBN 13 978-0470-50197-9: Hoboken,
NJ, 2011.
13. Holman JP. Heat Transfer (10th). Mc-Graw-Hill
Companies Inc: New York, NY, 2010.
14. Burmeister L. Convective Heat Transfer (2nd). John
Wiley & Sons: New York, NY, 1993.
15. Gebhart B, Jaluria Y, Mahajan RL, Sammakia B.
Buoyancy-induced Flows and Transport. Hemisphere
Publishing Company: New York, NY, 1988.
16. Liu C, Groulx D. Experimental study of the phase
change heat transfer inside a horizontal cylindrical
latent heat energy storage system. International
Journal of Thermal Sciences 2014; 82:100–110.
17. Christiano GSS, Luiz FM. Numerical study of melting
of tin within a rectangular cavity including convective
effects. Proceeding of COBEM 2011, Brazilian
Congress of Mechanical Engineering, Natal, RN,
Brazil, 2011.
18. Zhu Y, Zhang Y, Li G, Yang F. Heat transfer
processes during an unfixed solid phase change
material melting outside a horizontal tube.
International Journal of Thermal Sciences 2001;
40:550–563.
19. Tzivanidis C, Antonopoulos KA, Kravvaritis ED.
Parametric analysis of space cooling systems based
on night ceiling cooling with PCM-embedded piping.
International Journal of Energy Research 2012;
36:18–35.
20. Ismail KAR, Castro JNC. PCM thermal insulation in
buildings. International Journal of Energy Research
1997; 21:1281–1296.
21. Zhang Z, Bejan A. Melting in an enclosure heated at
constant rate. International Journal of Heat and Mass
Transfer 1989; 32(6):1063–1076.
22. Rajeev Rai KN, Das S. Numerical solution of a
moving-boundary problem with variable latent heat.
International Journal of Heat and Mass Transfer
2009; 52:1913–1917.
23. Joseph A, Kabbara M, Groulx D, Allred P, White MA.
Characterization and real-time testing of phase-change
materials for solar thermal energy storage.
International Journal of Energy Research 2016;
40:61–70.
24. Tamme R, Bauer T, Buschle J, Laing D, Müller-
Steinhagen H, Steinmann W. Latent heat storage
above 120°C for applications in the industrial process
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Heat transfer during melting of a PCM by a U-shaped heat source
heat sector and solar power generation. International
Journal of Energy Research 2008; 32:264–271.
25. Nabeel SD, Khodadadi JM. Melting and convection of
phase change materials in different, shape containers: a
review. Renewable and Sustainable Energy Reviews
2015; 43:449–477.
26. Naterer GF. Heat Transfer in Single and Multiphase
System. CRC press: Boca Raton, FL, 2003.
ISBN:0-8493-1032-6.
27. Beckermann C. A general correlation for melting in
rectangular enclosures. Journal of Heat Transfer
1989; 111:1111–1115.
28. Gau C, Viskanta R. Melting and solidification of pure
metal on vertical wall. Journal of Heat Transfer 1986;
108:174–181.
29. Diaz LA, Viskanta R. Visualization of the solid-liquid
interface morphology formed by natural convection
during melting of a solid from below. International
Communication of Heat and Mass Transfer 1984;
11:35–43.
30. Jellouli Y, Chouikh R, Guizani A, Belghith A.
Numerical study of the moving boundary problem
during melting process in a rectangular cavity heated
from below. American Journal of Applied Sciences
2007; 4(4):251–256.
31. Hsu CF, Sparrow EM, Patankar SV. Numerical
solution of moving boundary problems by boundary
immobilization and a control-volume based finite-
difference scheme. International Journal of Heat and
Mass Transfer 1981; 24:1335–1343.
32. Hale NW Jr, Viskanta R. Photographic observation
of the solid–liquid interface motion during melting
of a solid heated from an isothermal vertical wall.
Letters in Heat and Mass Transfer 1978;
5:329–337.
33. Ho CJ, Viskanta R. Heat transfer during melting from
an isothermal vertical wall. Journal of Heat Transfer
1984; 106:12–19.
34. Gadgil A, Gobin D. Analysis of two dimensional
melting in rectangular enclosures in presence of
convection. ASME Journal of Heat Transfer 1984;
106:20–26.
35. Yuksel N, Avci A, Kilic M. A model for latent heat
energy storage systems. International Journal of
Energy Research 2006; 30:1146–1157.
36. Bondareva NS, Sheremet MA. Mathematical
simulation of melting inside a square cavity with a
Int. J. Energy Res. (2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/er
Bashar M. and Siddiqui K.
local heat source. Thermophysics and Aeromechanics
2016; 23:553–565.
37. Bondareva NS, Sheremet MA. 3D natural convection
melting in a cubical cavity with a heat source.
International Journal of Thermal Sciences 2017;
115:43–53.
38. Bondareva NS, Sheremet MA. Effect of inclined
magnetic field on natural convection melting in a
square cavity with a local heat source. Journal of
Magnetism and Magnetic Materials 2016;
419:476–484.
39. Mariani VC, Coelho LS. Natural convection heat
transfer in partially open enclosures containing an
internal local heat source. Brazilian Journal of
Chemical Engineering 2007; 24(3):375–388.
40. Zalba B, Marin JM, Cabeza LF, Mehling H. Review
on thermal energy storage with phase change:
materials, heat transfer analysis and applications.
Applied Thermal Engineering 2003; 23:251–283.
41. Mathematical simulation of melting inside a square
cavity with a local heat source. Thermophysics and
Aeromechanics 2016; 23:553–565.
42. 3D natural convection melting in a cubical cavity with
a heat source. International Journal of Thermal
Sciences 2017; 115:43–53.
43. Himran S, Suwono R, Mansoori GA. Characterization
of alkanes and paraffin waxes for application as phase
change energy storage medium. Energy Sources 1994;
10:117–128.
44. Ukrainczyk N, Kurajica S, Sipusic J. Thermo-physical
comparison of five commercial paraffin waxes as latent
heat storage materials. Chemical and Biochemical
Engineering Quarterly Journal 2010; 24(2):129–137.
45. Boetcher SKS. Natural Convection from Circular
Cylinders. Springer publications, 2014. ISBN:978-3-
319-08131-1.
46. Bejan A. Convection Heat Transfer (4th). John Wiley
& Sons: Hoboken, NJ, 2013.
47. Ede AJ. Advances in Free Convection. Advances in
Heat Transfer. Academic Press: New York, NY, 1967.
48. LeFevre EJ, Ede AJ. Laminar free convection from the
outer surface of a vertical cylinder. Proceedings of the
9th international congress on applied mechanics,
Brussels, Belgium, 1956; 175–183.
49. Lim JS, Bejan A. The Prandtl number effect on
melting dominated by natural convection. Journal of
Heat Transfer 1992; 114(3):784–787.