Available online at www.sciencedirect.

com

International Communications in Heat and Mass Transfer 35 (2008) 466 – 475

www.elsevier.com/locate/ichmt

Constrained and unconstrained melting inside a sphere ☆

F.L. Tan
Nanyang Technological University, School of Mechanical and Aerospace Engineering, 50 Nanyang Avenue, 639798, Singapore
Available online 20 November 2007

Abstract

The melting of the phase change material (PCM) inside a sphere using n-Octadecane for both constrained and unconstrained
melting is investigated. In constrained melting, the solid PCM is restrained from sinking to the bottom of the sphere. For
unconstrained melting, the solid PCM would sink to the bottom of sphere due to gravity. The experiments are carried out at three
different wall temperatures of 35 °C, 40 °C and 45 °C with a sub-cooling of 1 °C for the unconstrained melting and three different
initial sub-cooling of 1 °C, 10 °C and 20 °C at a constant wall temperature of 40 °C for the constrained melting.
© 2007 Elsevier Ltd. All rights reserved.

Keywords: Constrained melting; Unconstrained melting; Melting inside sphere; Heat transfer; Solid–liquid phase front

1. Introduction

Solidification and melting has great influences on nature and modern technologies for example, in the thermal
energy storage (TES) system. The system utilizes the phase change material (PCM) to maintain a constant temperature
over a period of time. The TES system is to provide heating or cooling from a thermal energy storage reservoir in the
system. The heating or cooling process add heat to or extract heat from the reservoir respectively. The heating effect can
be found in the use of solar collector to store thermal energy from the sun to be used later at night whereas the cooling
effect is to store thermal energy storage at a low temperature at night to be used in air conditioning during the day. The
TES system can enhance the efficient usage of energy with a reduction in the cost of electricity.
In TES system, a spherical container is most commonly used for storing PCM. This is mainly due to its low volume
to heat transfer surface area ratio [1]. There are also alternative storage devices such as plates, cans or cylindrical shapes
with or without fins. Spherical shape is preferred because it can be easily packed into the storage system. The freezing
process has been investigated in various geometrical configurations [2–4] such as slab, cylinder and sphere. Ismail et
al. [5,6] presented a numerical solution based on finite difference approximations that are either subjected to convection
boundary conditions or a constant surface temperature. Investigation on the thickness of the wall, diameter of sphere,
surface temperature and initial PCM temperature were performed. They concluded that different spherical size, thermal
conductivity of material and surface temperature had effect on the time taken for complete solidification.
The first to study the unconstrained melting of the PCM within spherical container both experimentally and
numerically was Moore and Bayazitoglu [7] where n-octadecane was used as the PCM. The mathematical simulation

☆
Communicated by W.J. Minkowycz.
E-mail address: mfltan@ntu.edu.sg.

0735-1933/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.icheatmasstransfer.2007.09.008
 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475 467

was developed by assuming that the liquid remains at the fusion temperature at all times and solved using perturbation
method. The solid–liquid interface positions and the temperature profiles were determined for various Stefan and
Fourier numbers. The experimental data agreed well with the simulation results. Roy and Sengupta [8] performed a
similar study and found an analytical solution for the melting rate at the lower surface of the solid PCM in contact with
the heated surface. Following that, Roy and Sengupta [9] performed another study on the unconstrained melting
process inside a sphere. They noted two zones of melting: a thin melting layer at the bottom and a thicker layer at the
top of the sphere. It was observed that natural convection effect played an important role in the results as large amount
of melting was found to be at the upper portion of the sphere. Bahrami [10] observed that the movement of the melting
PCM caused by the presence of gravity has a great impact on the rates of melting and heat transfer.
Experimental and numerical analysis of melting of PCM inside a plexiglas spherical container was conducted by
Felix et al. [11], where paraffin wax was used as the PCM. Investigation on the effect of the sphere radius, Stefan
number, molten fraction and time for complete melting was carried out. The results for both the experimental and
numerical were quite identical. It was found that Stefan number and the sphere radius had significant effects on the
complete melting time of phase change material.
The constrained melting of PCM within a spherical bulb was presented by Khodadadi and Zhang [12]. They
numerically studied the effects of buoyancy-driven convection. They had the same prediction as Roy and Sengupta [9]
where the conduction mode of heat transfer was dominant in the earlier stage of the melting process. At a later stage, the
melting rate increased as the buoyancy-driven convection inside the liquid PCM became more significant. There is no
other work found on the constrained melting inside the sphere to the best knowledge of the author.
This paper is to investigate the differences in the constrained and unconstrained melting of phase change material
inside the sphere under several constant surface temperature boundary conditions at several initial sub-cooling
conditions. The motivation of the work here is to understand better the melting rate of phase change material inside the
sphere with or without restraining of the solid PCM inside the sphere. This would have useful applications to the
thermal design of thermal energy storage that uses phase change material inside spherical container.

2. Experimental setup and testing

The schematic of the experimental setup is shown in Fig. 1. The experimental setup consists of a water tank, a spherical glass,
several Type K thermocouples, refrigerated bath and the HP data acquisition unit. The hot water from the refrigerated bath flows
through the outlet valve into the water tank. The water in the tank is maintained at a uniform temperature through constant stirring
(electric stirrer not shown in the schematic). The water then flows out of the tank and back into the refrigerated bath through an inlet

Fig. 1. Schematic diagram of the experimental setup.

468 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475

Fig. 2. Thermocouples inside the sphere.

valve. The flow of water in and out of the tank is regulated by the two valves so that the water level in the tank is kept constant
throughout the process. The temperature controller unit attached to the refrigerated bath sets the required water temperature to flow
into the tank.
A water bucket shown in Fig. 1 is used to achieve the initial temperature of the PCM inside the sphere. This is to provide the initial
sub-cooling of the PCM required for the experiment. Using another refrigerated bath (not shown in the schematic), the water in the
bucket is maintained at a certain initial temperature for the experiment. The sphere is submerged into the bucket while waiting for the
water in the tank to rise to the required temperature. The sphere is filled with liquid PCM and the amount of PCM required is
indicated on the over-flow tube attached to the sphere. The inner volume of the spherical container is 0.00055 m3 and the inner radius
of the sphere is 50.83 mm. When the water in the tank has reached the required temperature for the experiment, the sphere is then
taken out of the water bucket, submerged and suspended inside the water tank. As the water temperature is higher than the melting
temperature of the PCM inside the sphere, the PCM would begin to melt.
Fig. 2 shows the position of the thermocouples assembled in the acrylic tube. The distance between each thermocouple is
12.5 mm, except for K11 which is only 10.3 mm away form K07. The acrylic tube with the thermocouples is inserted into the over-
flow tube and attached to the bottom of sphere. The thermocouples are connected to the HP data acquisition unit where the
temperatures are recorded during the experiment. The data can then be read from the HP Bench-Link Data Logger into the computer.

Fig. 3. Melting inside sphere: (a) unconstrained melting, (b) constrained melting.
 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475 469

The PCM in the sphere is subjected to unconstrained and constrained melting with different initial sub-cooling (ΔT) and at
different constant surface temperatures (TS) respectively. The unconstrained melting involves no temperature measurement of PCM
inside the sphere, which means that no thermocouple is used. In unconstrained melting experiment shown in Fig. 3(a), the solid PCM
has sunk to the bottom of the sphere and the lower portion of the solid PCM is in contact with the spherical glass. Three experiments
were carried out at three different constant temperatures of 35 °C, 40 °C and 45 °C at the surface of sphere. An initial sub-cooling
temperature of 1 °C for the solidified PCM is maintained for all the three experiments.
In contrast, the constrained melting experiment in Fig. 3(b) is carried out by having the acrylic tube inside the sphere. The hollow
tube holds the thermocouples in position and also holds the solid PCM onto the tube thus restraining the solid from sinking down to
the bottom of sphere despite the effect of gravitational force and buoyancy force due to the difference in density. Three experiments
were performed at three different initial sub-cooling of 1 °C, 10 °C and 20 °C for a constant temperature of 40 °C at the surface of
sphere.

3. Results and discussion

Fig. 4 shows the superimposed phase front versus time for unconstrained and constrained melting. The same boundary condition
is applied for both experiments at constant surface temperature of 40 °C and sub-cooling temperature of 1 °C for the solidified PCM.
The sub-cooling temperature is defined as ΔT = Tm − Ti, where Ti is the initial temperature and Tm is the melting temperature of the
PCM.
For the unconstrained melting shown in Fig. 4(a), the melting pattern could be seen clearly that the solid PCM has sunk to the
bottom of the sphere. This is due to the heavier density of the solid PCM than the liquid PCM. The lower portion of the solid PCM is
always in contact with the glass sphere. Melting at the lower half of the solid PCM occurs through heat conduction to the glass. At the
upper half of the PCM, melting occurs mainly through natural convection in the liquid PCM. In contrast, the constrained melting of
Fig. 4(b) shows that the melting occurs all around the PCM inwards towards the centre of the sphere. The solid PCM is restrained
from sinking by the tube inside the sphere. There is no contact of the solid PCM with the spherical glass. Melting is mainly through
the natural convection in the liquid at the top and bottom halves of the solid PCM. The solid–liquid phase front is rather smooth at
the top half. However, some waviness seen at the bottom of the solid PCM could be observed. This is caused by the several
convection cells formed at the bottom half in the liquid region.
Under the same experimental condition, unconstrained melting inside the sphere seems to occur at a faster rate than the
constrained melting. This is due to the larger rate of heat transfer by conduction from the solid PCM to the spherical glass.

3.1. Unconstrained melting under constant sub-cooling, ΔT

For the unconstrained melting experiments, the sphere is maintained at three different constant surface temperatures of 35 °C,
40 °C and 45 °C, respectively. An initial sub-cooling of 1 °C is maintained for the solid PCM before the melting experiment begins.

Fig. 4. Series of melting phase front contours at different times.

470 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475

The melting phase front within the sphere is shown in Fig. 5. The solid PCM is opaque white while the liquid PCM is transparent. In
Fig. 5(b), melting of the solid PCM can be seen and this is due to heat conduction at the inner wall of the sphere. In Fig. 5(c), as
melting occur, the solid PCM sinks to the bottom, thus the bottom of the PCM is always in contact with the inner wall. The liquid
form at the bottom is pushed up through the side to the top half as the solid PCM sinks to the bottom. Hence, a larger liquid layer can
be seen at the top half as time progresses. Heat conduction occurs between the solid PCM and the glass sphere and causes melting at
the bottom of sphere. In Fig. 5(d), as warm liquid PCM flows up along the inner wall, the cooler liquid PCM replaces it, thus causing
natural convection in the liquid region at the top half. The top half of the PCM is melted by two natural convection cells as seen from
Fig. 5(d) to (e). The convection cell in the liquid seems to be symmetrical with respect to the vertical axis through the centre of
sphere. The top half melts below the centre of the sphere as shown from Fig. 5(f) to (h), but the phase front remains relatively

Fig. 5. Unconstrained melting phase front at 40 °C with an initial sub-cooling of 1 °C.

 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475 471

Fig. 6. Heat conduction and natural convection in unconstrained melting.

unchanged. This could be due to the higher melting rate at the bottom region. This also indicates that the melting at the top is slower
than the bottom region. Towards the end of melting as in Fig. 5(i), the solid PCM is so small and light that it may sway to and fro the
side of sphere due to some external disturbance to the sphere.
From the images in Fig. 5, it could be inferred that natural convection in the liquid at the top half and heat conduction at the
bottom half have significant effect on the melting rate and shape profile of the melting PCM inside the sphere. A schematic sketch of
the heat conduction and natural convection flow for unconstrained melting inside the sphere is shown in Fig. 6. Two natural
convection cells in the liquid region at the top half can be deduced from experimental images in Fig. 5.
Fig. 7 shows the variation of the liquid fraction versus time for various Stefan number. The Stefan number (Ste) describes the
operating condition of the sphere undergoing melting, given that the surface temperature TS directly affects the value of the Stefan
number and is defined as:

Cpl ðTS  Tm Þ
Ste ¼
L
where Cpl is the specific heat of the liquid PCM, L is the latent heat of PCM and Tm is the melting temperature of PCM. Using the
same PCM inside the sphere, a higher surface temperature gives rise to a higher Stefan number.
The melting liquid fraction is obtained from the digital images captured at different time intervals through the use of solid
modeling. The coordinates of the solid–liquid interface is captured from the digital image using the software, WinDig. The
coordinate points of the phase front are then imported into the CAD software, Solid Works. The solid model of the solid PCM is then

Fig. 7. Liquid fraction versus Stefan number in unconstrained melting.

472 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475

generated by Solid Works and the volume of the solid PCM is then determined from the software. The liquid fraction of the melting
PCM can then be determined from the following:
VS
f ¼1
VC
where f is the liquid fraction, VS is the volume of solid PCM, and VC is the volume of the spherical container.
From Fig. 7, a higher surface temperature at 45 °C melts completely in 80 min compared to 90 and 160 min for surface
temperatures at 40 °C and 35 °C, respectively. Melting occurs almost instantaneously for surface temperatures at 40 °C and 45 °C.
But for surface temperature at 35 °C, there seems to be a delay of 30 min before melting begins. a lower surface temperature results in
an opposite effect on the melting rate. The lower surface temperature causes lower heat conduction at the sphere wall and also affects
the heat transfer rate through natural convection, thus causing a slow down in the melting significantly.

3.2. Constrained melting under constant surface temperature, Ts

In constrained melting experiment, a constant surface temperature of 40 °C is maintained throughout the experiments. Three
experiments were carried out for three different initial sub-cooling of 1 °C, 10 °C and 20 °C, respectively. Fig. 8 shows the
temperatures of PCM inside the sphere for the experiment at constant surface temperature of 40 °C and an initial sub-cooling of 10 °C.
The curves in Fig. 8 represent the temperature recorded from the thermocouples during the experiment. K01 to K07 and K11 is
located on the vertical axis through the centre of the sphere. The distance between each thermocouple for K01 to K07 is about
12.5 mm apart and K04 is located at the centre of the sphere. However K11 is only 10.3 mm away from K07. The purpose of this
additional thermocouple is to obtain a more accurate temperature gradient at the bottom region. The position of K08 is located at the
inner wall of the sphere and K09 and K10 is located at the outer wall of the sphere. The melting inside the sphere is divided into three
different zones. They are the solid region, phase change region and liquid region. The temperature ranges of these regions are from
16 °C to 27 °C for solid region, from 27 °C to 32 °C for phase change region, and from 32 °C to 40 °C for liquid region.
The centre temperature of the solid PCM before it is submerged into the tank is maintained at an initial sub-cooling of 10 °C.
Once the sphere is submerged inside the tank, the solid PCM is subjected to a water temperature of 40 °C around the sphere. The
PCM is in the solid heating zone. The temperatures of K01 to K07 and K11 will increase to the liquid heating zone over a period of
140 min depending on the location that is measured. When the temperature at the inner wall reaches the phase change zone, the
melting process will begin. The whole experiment will only end after the temperature of K05 arrives at the liquid heating zone. It is
noted that before K05 enters the last zone, the rest of the thermocouples are already into the liquid heating zone.
It can be seen that the temperature of the solid PCM initially stays below the melting temperature of n-octadecane which is about
27 °C. Once the solid–liquid interface passes the thermocouple position, it can be seen that there will be a sharp rise in liquid
temperature at that position (see for example thermocouples K01 to K06 in Fig. 8). The outer and inner surface temperatures of the
glass sphere are rather constant at 40 °C (see K08 to K10). The temperatures, K07 and K11, measure the temperatures at the bottom
of the sphere. The K11 temperature gives temperature above the melting point throughout the experiment. This shows that a layer of

Fig. 8. Temperatures versus time at different positions inside the sphere for constrained melting.
 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475 473

Fig. 9. Constrained melting phase fronts at 40 °C with an initial sub-cooling of 10 °C.

liquid exists at the bottom of sphere about 3 mm away from the sphere bottom right from the start of the experiment. At 13.3 mm
above the bottom of sphere, thermocouple K07 gives a temperature above melting point after about 65 min into the experiment.
The constrained melting phase front within the sphere is shown in Fig. 9. An almost concentric melting is observed from the
images shown from Fig. 9(a) and (b) where heat conduction at the inner wall has dominated at this stage. The forming of the liquid
PCM is relatively constant at the beginning of the melting process. This is due to the direct contact between the inner wall and the
solid PCM. As melting progresses, Fig. 9(c) to (e) show that the layer of liquid PCM increases and thus heat conduction is
significantly reduced. An oval shape melting pattern is formed at the top half of the solid PCM but the profile at the bottom remains
relatively unchanged. The oval shape phase front is caused by the natural convection when the liquid layer increases. Natural
convection occurs as a result of the warm liquid PCM rises along the hot wall while the cooler liquid in the centre flows down to
replace the warmer fluid. This creates an unstable fluid circulation inside the sphere known as buoyancy-driven convection.
474 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475

Fig. 10. A schematic sketch showing natural convection in constrained melting.

Therefore, the top half melts faster than the bottom half. This occurs throughout the whole melting process inside the sphere. The
melting rate at the bottom increases gradually as the top half continues to melt at a steady rate as shown from Fig. 9(f) to (j). This
gradual increase in melting rate at the bottom is primarily due to the natural convection cells formed at the bottom due to the increase
in liquid layer at the bottom of the sphere. A schematic sketch of the natural convection flow for the constrained melting in the sphere
is shown in Fig. 10. The figure shows natural convection cells in the liquid. The three convection cell at the bottom causes the
waviness of the solid PCM at the bottom.
Fig. 11 shows the variation of the liquid fraction for different sub-cooling under constrained melting. It is observed that different
sub-cooling conditions applied on PCM do not have significant effect on the melting rate at a constant Stefan number. Initial sub-
cooling of the PCM at 10 °C and 20 °C have an almost complete melting time compared to a sub-cooling of 1 °C which has a faster
complete melting time. However, from the slope of the curves in Fig. 11, the melting rate seems to be almost the same.

4. Conclusion

In unconstrained melting, the melting of PCM at the beginning is dominated by the heat conduction across the wall.
As the PCM melts, it sinks to the bottom of the sphere. This downward movement creates additional pressure to the
liquid PCM, forcing the warm liquid between the solid PCM and the inner wall of sphere to rise to the top. The warm

Fig. 11. Liquid fraction versus time in constrained melting.

 F.L. Tan / International Communications in Heat and Mass Transfer 35 (2008) 466–475 475

and cool liquid generates circulation of the liquid PCM and causes natural convection at the top half of the solid PCM.
The melting at the bottom half is much faster due to heat conduction at the inner wall. In constrained melting, heat
conduction only exists at the beginning of the melting process. This effect causes the melting PCM to melt almost
concentrically inwards. An oval shape of the solid PCM is formed at the top half due to natural convection heat transfer
in the liquid PCM. The top half melts at a faster rate than the bottom half. Natural convection cells are also formed at the
bottom half, thus causing the waviness profile at the bottom.

Acknowledgement

The authors would like to acknowledge the work of Nanyang Technological University undergraduate student, S.S.
Lim, for his contribution on the design, fabrication and experimental testing in this work.

References

[1] L. Bilir, Z. Ilken, Total solidification time of a liquid phase change material enclosed in cylindrical/spherical container, Applied Thermal
Engineering 25 (2005) 1488–1502.
[2] A. Barba, M. Spriga, Discharge mode for encapsulated PCMs in storage tanks, Solar Energy 74 (2003) 141–148.
[3] K.A.R. Ismail, M.G.E. da Silva, Melting of PCM around a horizontal cylinder with constant surface temperature, International Journal of
Thermal Sciences 42 (2003) 1145–1152.
[4] E. Assis, L. Katsman, G. Ziskind, R. Letan, Numerical and experimental study of melting in a spherical shell, International Journal of Heat and
Mass Transfer 50 (2007) 1790–1804.
[5] K.A.R. Ismail, J.R. Henriquez, Solidification of PCM inside a spherical capsule, Energy Conversion and Management 41 (2000) 173–187.
[6] K.A.R. Ismail, J.R. Henriquez, A parametric study on ice formation inside a spherical capsule, International Journal of Thermal Sciences 42 (2003)
881–887.
[7] F.E. Moore, Y. Bayazitoglu, Melting within a spherical enclosure, American Society of Mechanical Engineers Journal of Heat Transfer 104 (1982)
19–23.
[8] S.K. Roy, S. Sengupta, The melting process within spherical enclosures, Transactions of the American Society of Mechanical Engineers 109 (1987)
460–462.
[9] S.K. Roy, S. Sengupta, Gravity-assisted melting in a spherical enclosure: effects of natural convection, International Journal of Heat and Mass
Transfer 33 (1990) 1135–1147.
[10] P.A. Bahrami, Natural Melting Within a Spherical Shell, Ames Research Center, Moffett Field, California, NASA Technical Memorandum
102822, 1990.
[11] A. Felix Regin, S.C. Solanki, J.S. Saini, Experimental and Numerical analysis of melting of PCM inside a spherical capsule, 9th AIAA/ASME
Joint Thermophysics and Heat Transfer Conference, 5–8 June, San Francisco, California, AIAA 2006-3618, 2006.
[12] J.M. Khodadadi, Y. Zhang, Effects of buoyancy-driven convection on melting within spherical containers, International Journal of Heat and
Mass Transfer 44 (2001) 1605–1618.