Applied Thermal Engineering 31 (2011) 3827e3838

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Experimental and numerical studies on performance of PCM-based heat sink

with different configurations of internal finsq
S.F. Hosseinizadeh a, F.L. Tan b, *, S.M. Moosania a
a
Department of Mechanical Engineering, Babol University of Technology, Babol, Islamic Republic of Iran
b
School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents both experimental and numerical investigations on the application of a PCM-based
Received 8 March 2011 heat sink for the purpose of thermal management. A comparison was carried out between heat sink with
Accepted 17 July 2011 and without phase change material (PCM). The effects of various parameters such as power levels,
Available online 27 July 2011
number of fins, fin height and fin thickness were studied. The results showed that increasing the number
of fins and fin height resulted in an appreciable increase in overall thermal performance. Increasing the
Keywords:
fin thickness only gave a slight improvement. There was an optimum fin thickness, above which the heat
Phase change material
sink performance showed no further improvement. Increasing the power level input, as expected,
Heat sink
Melting
increased the melting rate of the PCM. In all cases, heat conduction was the primary mode of heat
Thermal management transfer at the initial stage of melting. At the later stage, free convection played a more crucial role in
Numerical simulation enhancing the melting of the PCM.
Experimental study Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Operating the microprocessors outside the functional temper-

ature range can degrade the system performance and can cause
Thermal management of electronic components has today logic errors or component damage. Furthermore, temperatures
become a major consideration in the design of electronic products. exceeding the maximum operating limit of the component may
This is due to the fact that with increasing power and functionality result in irreversible changes in its operating characteristics. The
of components, there is an increase in heat dissipation. The rapid microprocessor temperature depends on the component power
development of electronics packaging technology means that the dissipation, thermal characteristics of the microprocessor package
sizes of electronic components are getting smaller. These two and the microprocessor thermal solution (i.e. heat sink). Today’s
factors contribute to a very high heat flux releases by the compo- computer microprocessors which operate at very high frequencies
nent. One such component is the desktop and mobile micropro- are usually attached with a heat sink. There are two types of heat
cessors. Microprocessors, as with all other electronic devices, are sinks that can be used, namely active or passive heat sink. Active
very reliable at their recommended operating temperature [1]. heat sinks utilize power for operation of a fan or other cooling
Different microprocessors have different maximum operating devices. Passive heat sinks do not require electrical power and are
temperatures, with the maximum allowable temperature between 100% reliable as they have no moving parts.
80
C and 115
C. Mithal [2] studied the effect of temperature on In recent years, passive thermal management schemes using
electronic component reliability. Their experimental results indi- PCMs in spacecrafts, portable electronic devices [3e5] and personal
cated that a 1
C decrease in a component temperature would computing have been widely researched on due to its potential. The
lower its failure rate by as much as 4%. desirable properties of PCM are its high latent heat of fusion per
unit mass and small volume change during phase change. This
means high amounts of energy can be stored in a fixed amount of
q This work is a remembrance of Seyed Farid Hosseinizadeh, who unfortunately PCM during the phase change period. The PCM absorbs the heat
died in a car accident in January 2011. His youth and talent in computational from the component and later discharges it to the ambient during
modeling and simulation work would be forever remembered by the co-authors the cooling period. However, most PCMs also possess the undesir-
and others who may be inspired by his research work.
able property of low thermal conductivity, which can result in the
* Corresponding author.
E-mail addresses: [email protected] (S.F. Hosseinizadeh), mfl[email protected]. overheating of the chip [6,7]. In order to make PCMs suitable for
sg (F.L. Tan). electronics cooling applications, it is necessary to add material with

1359-4311/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2011.07.031
3828 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

A PCM-based heat sink [8] is a form of passive cooling which

Nomenclature requires no mechanical or moving parts. As such, they are 100%
reliable. The goal is to transfer heat from the microprocessor to the
F external body forces (N) heat sink which is absorbed by the PCM. The PCM in solid form will
g gravitational acceleration (m/s2) then start to melt (charging process) until it turns into its fully
h enthalpy (W/m2 K) liquid state. Once completely melted, the heat of the PCM has to be
k thermal conductivity (W/m K) released to the surrounding by re-solidifying the molten PCM in the
Sm source term heat sink (discharging process). A PCM-based heat sink [9] is made
T temperature (
C) up of four side walls to contain the PCM. A top cover is needed to
t time (s) prevent the spillage of molten PCM due to vibrations or in the event
u velocity (m/s) of a mechanical shock.
xi Cartesian coordinate direction The objective of this paper is to study the effect of several
parameters of the PCM-based heat sink on the temperature of the
Greek symbol heat sink and melting of the PCM inside the heat sink. The
a fluid volume fraction parameters of the heat sink that are studied in this paper are the
b thermal expansion coefficient (K1) number of fins, fin thickness and fin height. Several heat sinks are
dij Kronecker delta fabricated for experimental testing. They are the heat sink without
m dynamic viscosity (kg/m s) fin, heat sinks with 3 fins, 5 fins and 9 fins, heat sinks with the fin
r density (kg/m3) thickness at 2 mm, 4 mm and 6 mm, and heat sinks with the fin
height at 10 mm, 20 mm and 40 mm. Both experimental testing and
Subscripts numerical simulations are being performed in the present paper.
f fin
i component
l liquid 2. Experimental setup and procedures
S solid
Fig. 1 shows the schematic of the experimental setup for the
experimental studies. The base of the PCM-based heat sink is
high thermal conductivity into the PCM system which will aid in heated up using an electrical heater attached to the base. The
the heat conduction. These materials are known as thermal dimension of the heater is 50 mm  50 mm. The power input to the
conductivity enhancers. The poor thermal conductivity of the PCM electrical heater is controlled by an AC power variable transformer
is one of the disadvantages limiting to the use of the PCM for connected to the 230 V AC main. The voltage across the heater and
widespread applications. The use of fins in our PCM-based heat sink the current through the heater are measured to determine the
is considered as one way to enhance the thermal conductivity of the power input to the heater. The PCM-based heat sink is insulated on
PCM. The number of fins, thickness and height of the fin would be all sides to minimize heat loss. Some experimental measurements
factors to consider in the design of a PCM-based heat sink. These of the temperature difference across the insulation are recorded to
parameters would be explored in this paper through experimental determine the heat loss. It is found that the worst case heat loss
testing and numerical simulations. under 45 W input to the heater is about 12.4%. The thermocouples

Fig. 1. Schematic of the experimental setup.

 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838 3829

attached to the heat sink are connected to PC-based data acquisi-

tion system for temperature measurements.
Heat sinks showing from Figs. 2e4 are fabricated for the
experimental studies. The overall dimension of the heat sink is kept
constant. The experimental models are fabricated using the CNC
machining. The heat sinks are made from aluminum alloy 6061,
which has a thermal conductivity of 180 W/m K. Each heat sink has
sixteen strategically located thermocouples. These thermocouples
are meant to record the temperature distribution on the heat sink
for the duration of the heating process. The thermocouples are
located on the outside vertical fin surface, inner vertical fin surface,
between the fins, and at the inner and outer base of the heat sink.
The locations of these thermocouples are shown in Fig. 5. At certain
critical points such as inner and outer base, two thermocouples (T1
and T2) are placed at the outer base of the heat sink and two
thermocouples (T3 and T4) are placed at the inner base of the heat

Fig. 3. Heat sinks with different fin thickness (a) 2 mm fin thickness, (b) 4 mm fin

thickness, (c) 6 mm fin thickness.

sink, so that an accurate temperature measurement of the base

temperature or case temperature is obtained through the averaging
of the four values. Thermocouples T8, T9 and T10 are used to
measure the temperature of the PCM. They are held in place at the
middle of the two fins by a 4 mm diameter plastic rod.
Type-K (AWG 24) thermocouples are used. It has a sensitivity of
about 41 mV/
C. This type of thermocouple comes in the form of two
wires, high and low voltage. In its original form, it is encapsulated
in a layer of insulating material. The preparation process for the
thermocouple involves stripping and spot welding using a ther-
mocouple welding unit. The thermocouples are cut to a length of
0.8 m for ease of connection to the PC-based data acquisition
system. They are then calibrated using a ThermaCalÔ calibration
Cooling/Heating source. This calibration unit has a temperature
range between 25
C and 600
C. Five temperature points are used
for the purpose of calibrating the thermocouples at 35
C, 70
C,
105
C, 140
C and 175
C. These temperature values are chosen
because they represent the expected temperature range for the
experimental studies. The calibration results, in terms of gain and
offset are input into the Hewlett PackardÔ Bench Link Data Logger
System. The thermocouples are calibrated to within 0.75
C
according to ASTM standards [10]. Random noise is estimated at
0.1
C. Following the references [11e14], the thermocouple
temperature uncertainty is estimated at  0.5
C. The thermocou-
ples are pasted at the sixteen locations shown in Table 1 using
OmegabondÔ 100 thermal paste epoxy. This epoxy has good
thermal conductivity and electrical insulating properties. They
Fig. 2. Heat sinks with different number of fins (a) without fin, (b) 3 fins, (c) 5 fins, come in the form of the catalyst and resin, which have to be pre-
(d) 9 fins. mixed before use.
3830 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

Table 1
Location of thermocouples and their functions.

Thermocouple Location Function

T1 Base outer Outer base temp
T2 Base outer Outer base temp
T3 Base inner Inner base temp
T4 Base inner Inner base temp
T5 Fin Fin 8 mm above inner base
T6 Fin Fin 24 mm above inner base
T7 Fin Fin 24 mm above inner base
T8 PCM 8 mm above inner base
T9 PCM 24 mm above inner base
T10 PCM 40 mm above inner base
T11 Side wall outer Outer side wall 8 mm above inner base
T12 Side wall outer Outer side wall 24 mm above inner base
T13 Side wall outer Outer side wall 40 mm above inner base
T14 Styrofoam inner Outer side wall temp
T15 Styrofoam outer Outer acrylic insulator temp
T16 Heater base Heater temp

2300 V). Two thermocouples are installed at inside (T14) and

outside (T15) of the Styrofoam insulation. This is to measure the
temperature drop between the insulation in order to determine the
heat loss.
The PCM, RubithermÔ RT-80 which comes in the form of pellets
are melted on a heating plate. Upon reaching its melting point, the
PCM melts. The molten PCM is then transferred to a measurement
beaker, and poured into the heat sink. The heat sink is filled up to
1 mm from the brim. The heat sink with the PCM is then left to cool
down to room temperature. This typically takes between 3 and 4 h.
Upon solidification, due to the difference in density of the solid and
liquid PCM, the PCM contracts upon solidification. The distance of
the solidified PCM from the top of the heat sink is between 4 and
5 mm.
The Hewlett PackardÔ Bench Link Data Logger software system
Fig. 4. Heat sinks with different fin height (a) 10 mm fin height, (b) 20 mm fin height,
(c) 40 mm fin height.
is then activated for temperature measurements. The experiment is
started at room temperature (T ¼ 27
C) by supplying power to the
heater via the variable transformer. The PCM-filled heat sink is
The thermocouples are connected to the Hewlett PackardÔ placed on top of the heater and the HewlettePackardÔ Bench link
Bench Link Data Logger System card which interfaces with the Data Logger data acquisition system software is initialized. The
HewlettePackardÔ Data Acquisition Unit. In order to minimize temperature readings are recorded at every 30 s interval. A total of
heat loss from the four side walls of the PCM heat sink, they are eleven different experiments with different heat sinks with fin
insulated with acrylic and heat insulation cloth. The insulating configurations and at different power levels as shown in Table 2 are
material (Styrofoam) is wrapped around the PCM heat sink using performed.
masking tape. The acrylic has a thermal conductivity of 0.2 W/m K
while the Styrofoam has a thermal conductivity of 0.054 W/m K.
3. Numerical models
The thermal conductivity of the Styrofoam was measured using the
Holometrix Micromet thermal conductivity tester (Model: Lamba
The flow of the molten PCM and air in heat sink assumed to be
laminar, incompressible and unsteady. FLUENTÔ solves the two

Table 2
Experimental runs.

Experiment Heat sink model Description Power

(W)
No. of fins Fin thickness
(mm)
Run1 3 6 Without PCM 45
Run 2 3 6 With PCM 45
Run3 3 6 With PCM 25
Run4 3 6 With PCM 35
Run5 3 6 With PCM 45
Run6 3 2 With PCM 45
Run7 5 2 With PCM 45
Run8 7 2 With PCM 45
Run9 3 2 With PCM 45
Run10 3 4 With PCM 45
Run11 3 6 With PCM 45
Fig. 5. Location of thermocouples in the heat sink.
 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838 3831

Table 3
RT-80 properties.

Melting Congealing Heat storage Density Density Volume Specific Kin. Thermal
temperature point
C capacity kJ/kg (solid) (Liquid) expansion heat viscosity conductivity


C kg/m3 kg/m3 coefficient capacity m/s w/m K
KJ/kg K
81 80 175 920 770 0.001 Liquid 1.8 Solid 2.4 9.4  106 0.2

  
vui vuj 2 vu
sij ¼ m þ  m l dij
vxi vxi 3 vxl

dij h 0 for isj
1 for i ¼ j

During the phase change process, the solid PCM will melt and
change into its molten (liquid) state. The PCM in its solid state has
a density of 920 kg/m3 while the liquid state has a density of
770 kg/m3. As the mass of the PCM remains constant throughout
this process, this change in density will inherently cause an
increase in volume. As this foreseen problem will occur in the real
life application of the PCM-based heat sink, the numerical model is
also modeled similarly. Following Shatikian et al. [8], the PCM is
filled up only until 85% of the heat sink height. The remaining 15%
of the height is modeled to accommodate the air region for the
expansion of the PCM.
A densityetemperature relation is used for air. The model PCM
is based on the properties of commercially available paraffin wax
Fig. 6. Computational domain for 2-dimensional model. (RubithermÔ RT-80) which is summarized in Table 3. The density
of PCM in solid phase is kept constant but after the melting period
conservation equations associated with this type of flow. They are
at temperatures between 80
C and 81
C, the density of liquid PCM
the mass conservation equation (continuity equation):
is given as follows:
vr v rl
þ ðrui Þ ¼ Sm r¼
vt vxi ½bðT  Tl Þ þ 1
And the momentum conservation equation:
where rl is the density of PCM at the melting temperature and b is
v v
 vr vsij the thermal expansion coefficient.
ðrui Þ þ rui uj ¼ þ þ rgi þ Fi The presence of two or more fluids is termed as a multiphase
vt vxj vxi vxj
model. The volume of fluid (VOF) model is available in FLUENTÔ to
where the stress tensor sij is given by: perform this type of modeling. In the VOF model, a single set of

300

250

200
temperature(°C)

150 with PCM (2d)

with PCM (3d)

100
with PCM (exprimental)
without PCM (exprimental)
50
without PCM (2d)

without PCM (3d)

0
0

8
10

12

14

16

18

20

22

24

26

28

30

time(min)

Fig. 7. Case temperature versus time for heat sink with and without PCM.
3832 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

Fig. 8. Plan view of melt front with time for heat sink with 3 fins e experimental study.

momentum equations is shared by the fluids, and the volume frac- fraction of one (or more) of the phases. For the qth phase, this
tion of each of the fluids in each computational cell is tracked equation has the following form:
throughout the domain. In each control volume, the volume fraction
of all phases sum up to unity. The fields for all variables and prop- vaq vaq
þ ui ¼ Saq
erties are shared by the phases and represent volume-averaged vt vxi
values. Thus, the variables and properties in any given cell are In this simulation study, the two phases are defined as follows:
either purely representative of one of the phases, or a mixture of the
phases. The qth fluid’s volume fraction in the cell is denoted as aq , Primary phase: air.
then the following three conditions are possible: Secondary phase: PCM.

aq ¼ 0 the cell is empty The volume fraction equation will not be solved for the primary
aq ¼ 1 the cell is full phase. The primary-phase volume fraction will be computed based
0 < aq < 1 the cell contains the interface between the fluids on the following constraint:

Based on the local value of aq , the appropriate properties and X

n
aq ¼ 1
variables will be assigned to each control volume within the
q¼1
domain.
The tracking of the interface(s) between the phases is accom- The enthalpy-porosity technique is used in FLUENTÔ for
plished by the solution of a continuity equation for the volume modeling the phase change process. In this technique, the melt

Fig. 9. Plan view of melt front with time for heat sink with 3 fins e numerical simulation.
 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838 3833

220

180
temperature(°C)
25W (2d)
140
35W(2d)

45W (2d)

100 25W (3d)

35W(3d)

45W(3d)
60
25W(exprimental)

35W(exprimental)

20 45W(exprimental)

48
0

30

33

36

39

42

45
3

9
12

15

18

21

24

27
time(min)
Fig. 10. Case temperature versus time for different input power.

interface is not tracked explicitly. Instead, a quantity called the is to investigate the thermal performance of the system with
liquid fraction is associated with each control volume in the respect to time, the unsteady time formulation with a first order
domain. The liquid fraction is computed at each iteration. This implicit time-stepping method is chosen.
liquid fraction allows us to track the melting of the PCM. The VOF model [15] is used to describe the PCM-air system with
The enthalpy-porosity model solves the following form of the a moving interface but without interpenetration of the two fluids.
energy equation: The solidification and melting model is enabled to define the phase
change phenomenon of the PCM. The laminar model is used to
v v v v specify the viscous model of the system. The atmospheric pressure
ðrhÞ þ ðrui hÞ þ ðrDHÞ ¼ ðkVTÞ þ S
vt vt vt vxi is set at 101325 Pa with a downward gravitational acceleration of
9.81 m/s2. All the simulations are initialized at a temperature of
300 K (27
C). The ambient air temperature is kept constant at
27
C.
4. Numerical simulation
The Pressure-Implicit with Splitting of Operators (PISO) algo-
rithm is used for the pressureevelocity coupling. At the early stage
Numerical simulations are carried out for various cases to study
of simulation, a time step of 0.0001 s is chosen. After 3 min of flow
the effects of different parameters on the PCM-based heat sink. The
simulation, the time step is increased to 0.001 s. And then during
above mentioned experimental models of the PCM-based heat
the melting process, the time step is changed to 0.01 s. The
sinks are all simulated in the numerical simulations. The compu-
convergence is checked at each time step. The convergence crite-
tational domain is discretized into finite volumes superimposed on
rion is set at 0.01 for thex-direction velocity, y-direction velocity
mesh points. The pressure-based solver with an implicit solution
and z-direction velocity in 3-d model and continuity.
formulation is used. The pressure-based solver enables the
Also, a grid dependency check is carried out to determine the
pressure-based NaviereStokes equation. As the focus of this project
effects of varying grid size on the numerical solution of the
computational model. Three different grid sizes are analyzed for
1 a PCM-based heat sink with 3 fins with 6 mm fin thickness. The
three different grids have 11264, 2816 and 704 cells for two-
dimensional model and 1441792, 180244 and 22528 cells for
0.8 three-dimensional model. The meshes with 704 cells for two-
liquid fraction

dimensional model and with 22528 cells for three-dimensional

0.6 model are chosen for subsequent computations for good accura-
25W cies and reasonable computing time. Fig. 6 schematically shows the
35W physical model for the two-dimensional simulation.
0.4
45W

0.2 Table 4
Performance of heat sink with varying power levels.

0 Power level Melting temp Duration

(W) (
C) (mins)
0

8
12

16

20

24

28

32

36

41

45

49

25 80e106 14e32
time(min) 35 86e106 12e23
45 87e110 7e16
Fig. 11. Liquid fraction versus time for different input power.
3834 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

200
temperature(°C)

150

without fin(3d)
3fin(3d)
100
5fin(3d)
7fin(3d)
3fin(exprimental)
5fin(exprimental)
50
7fin(exprimental)
7fin(2d)
5fin (2d)
3fin (2d)
0
0

10

12

20

22

24

26

28

30
14

16

18
time(min)

Fig. 12. Case temperature versus time for different number of fins.

5. Results and discussions in application intermittently where there is a need for the molten
PCM to release its heat to the surrounding for re-solidification.
5.1. Effect of PCM One interesting thing to note is that even though the melting
only starts at 7 min, the heat sink with PCM seems to be able to
Fig. 7 compares the thermal performance of the PCM-based heat maintain a lower temperature compared to the heat sink without
sink for two cases: heat sink without PCM and heat sink filled with PCM from the 1st minute. This is due to the fact that besides having
PCM. We can notice from Fig. 7 that for the case of heat sink without a large latent heat absorption capability, the PCM also has a larger
PCM there is a steady temperature rise. The case temperature rises specific heat compared to air. For the case of PCM RT-80, it has
to 100
C in 5 min. However, for the case of the heat sink with PCM, a specific heat of 2100 J/kg K. As such, the PCM continually absorbs
the temperature rise is lower throughout. The case temperature this amount of sensible heat per unit kg and 1 K of temperature rise
only reaches 100
C at 12 min. There is also a period between 10 as the base of the heat sink is heated.
and 20 min where the temperature rise seems to be low. This The numerical simulations for the 2d and 3d models compared
period coincides with the melting of the PCM in the heat sink. The reasonably well with the experimental data. For the heat sink
difference between the two cases is due to the fact that for the heat without the PCM, the 3d simulation model agrees much better to the
sink without PCM there is no effective latent energy storage to experimental data. This difference could be due to numerical accu-
maintain the temperature. For the heat sink with PCM, the PCM racies of the simulations caused by the large disparity in the thermal
stores the heat during melting as latent heat, thus keeping the conductivity between the air region and aluminum heat sink. For the
temperature at around the melting temperature of the PCM. The PCM-based heat sink, both 2d and 3d simulation models agree
PCM-based heat sink does lower the base temperature of the heat reasonably well with the experimental data. The thermal conduc-
sink through the absorption of heat by the PCM compared to the tivity of the PCM is ten times better than that of air which could
heat sink without PCM. However, once the solid PCM is fully mel- explain the better agreement for the PCM-based heat sink.
ted, the base temperature of the PCM-based heat sink can rise to The melt front evolution versus time for the heat sink with PCM
the same temperature value of about 150
C as the heat sink is shown in Fig. 8 for the experimental study. Fig. 9 shows the melt
without PCM within the same duration of 30 min. Thus, the PCM- front profile when it is viewed from the top of the heat sink
based heat sink cannot operate continuously. It can only be used obtained through the numerical simulation. The white portion in

Fig. 13. Temperature contours for heat sink without fin at t ¼ 30 s.

 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838 3835

Fig. 14. Temperature contours for heat sink with 3 fins at t ¼ 30 s.

Fig. 8 shows the solid PCM melting inside the heat sink. At time the temperature contours for the two cases. The first case is for the
zero, the heat sink is filled completely with solid PCM. After heat sink without fins (Fig. 13) and the second case is for the heat
8.5 min, some PCM melts at the four walls of the heat sink. More sink with 3 fins (Fig. 14).
and more PCM is melted with time. By 14.5 min, most of the solid Firstly, it is important to note that the PCM RT-80 has a thermal
PCM has melted. At the 15th minute, the remaining PCM volume conductivity of 0.2 W/m K while the aluminum fin has a thermal
sinks down to the base of the heat sink. The comparison of Figs. 8 conductivity of 180 W/m K. This makes the fins good heat
and 9 shows that the melt front profile obtained through numer- conductors while the PCM behaves almost like an insulator. If we
ical simulation agrees reasonably well with the melt front profile are to observe the temperature profile of the heat sink without fins,
obtained through experiments. we can see that as the base is heated, heat is distributed uniformly
throughout the 4 mm thick base and 2 mm thick side walls. This is
5.2. Effect of power input level because of high thermal conductivity of aluminum. Due to the low
thermal conductivity of the PCM, heat (before melting phase) is
Figs. 10 and 11 show the case temperature and liquid fraction only absorbed by PCM where adjacent to warmer walls. Certainly
versus time for different power level, respectively. As expected, with larger surface area of the walls, we have larger amount of heat
higher power levels results in a higher temperature for the base of absorbs by the PCM. In heat sink with internal fins, extra surface has
the heat sink. Through visual inspection it is observed that the been added and so PCM-based heat sink can absorb more heat at
melting rate increases with higher power levels. In fact for the a faster rate.
experiment at 25 W power input, the experiment extends beyond Next we examine the results of melt fraction versus time for the
30 min as the PCM has yet to melt. This is expected as a higher different number of fins. Two obvious observations from the graph
power input will result in a higher energy absorption rate by the in Fig. 15 are the melt starting time and the melting period. Firstly,
PCM. One interesting observation is that there is three different we are able to notice that the melt starting time is delayed with an
melting durations as shown in Table 4. The higher the power input increase in the number of fins. The main difference between the
the shorter the melting duration. two models is that the 7-fin heat sink has a higher aluminum mass
Another observation that we can make from Fig. 10 is that once compared to the 3-fin heat sink. Due to this the 7-fin heat sink has
the PCM is fully melted, the temperature curves for all 3 experi- a higher heat storage capacity. Heat inputs into both models are
ments diverge greatly. Take for instance at time 30 min. The similar at 45 W. The end result is that the temperature rise for the
experiment at 35 W power input has a temperature which is 36
C 3-fin heat sink is higher than that of the 7-fin heat sink. Secondly,
higher than the experiment at 25 W power input. The experiment we can notice that the melting period is reduced with the increase
at 45 W power input has a temperature which is 48
C higher than in number of fins. This can be attributed to the volume of PCM. As
that of the experiment at 35 W power input. This is expected as the mentioned earlier, the increase in number of fins also causes
PCM has reached it melting point and any further increase in the
power input simply increases the temperature of the heat sink. 1
The numerical simulation results on case temperatures for both
2d and 3d simulation models compare reasonably well with the
experimental data. There is no comparison of simulation models 0.8
with the experimental data on liquid melt fraction as it is not
liquid fraction

possible to measure the amount of PCM melted during the exper-

0.6
imental testing. The only way to measure the amount of PCM
melted is to use the pour-out method. This method entails the no fin
experiment be stopped at each time interval to pour out the liquid 0.4
3 fin

PCM for measurement. The experiment is then repeated for the 5 fin

next time interval. This is too tedious and laborious for the exper- 7 fin

iments. Thus, there are no experimental data on liquid melt fraction 0.2
to compare with the simulation results.

5.3. Effect of number of fins 0

0

9
12

15

18

21

24

27

30

Fig. 12 shows a huge difference between the PCM heat sink

time(min)
without fins and the other three models which all have different
number of fins. This phenomenon can be explained by comparing Fig. 15. Melt fraction versus time with different number of fins.
3836 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

250

200
temperature(°C)

150
2mmthick(3d)
4mmthick(3d)
6mmthick(3d)
100
2mmthick(exprimental)
4mmthick(exprimental)
6mmthick(exprimental)
50
6mmthick(2d)
4mmthick(2d)
2mmthick(2d)
0
10

12

14

16

18

20

22

24

26

28

30
0

time(min)

Fig. 16. Case temperature versus time for different fin thickness.

a decrease in the PCM volume. This smaller volume corresponds to 5.5. Effect of fin height
a lesser specific and latent heat capacity of the PCM. Thus the PCM
takes a shorter duration to melt. From Fig. 18, we can see that the performance of the heat sink
increases with increasing fin height. This is expected because the
heat transfer surface increases with increasing fin height. For the
5.4. Effect of fin thickness
10-mm fin height, heat is transferred from the base to the fins. The
fins then transfer heat only to a small portion of the PCM. As the
Fig. 16 shows the temperature evolution with time for various
thermal conductivity of the PCM is low, further heat transfer
fin thickness. It can be seen that the 1-mm and 2-mm fin thickness
toward the upper regions of the PCM in inhibited. This causes local
give almost similar results. Both the 4-mm and 6-mm fin thickness
overheating at the fins and the base which increases the case or
give slightly better results. Also as shown in Fig. 16, during the PCM
base temperature.
melting period, the heat sink with thicker fins give a slightly lower
For the 40-mm fin height, heat from the fin is transferred to
temperature. This is because of that the thermal resistance of fins
a larger portion of the PCM. Once the base and sides of the PCM are
reduces with larger fin thickness. This causes an increase in
heated and melted, internal natural convection of the melted PCM
temperature of fins and rate of heat releases to PCM and so the
takes place. This increases the rate of the heat transfer to the upper
overall temperature of heat sink decreases. We should note that the
regions of the PCM.
heat sink with thicker fins have less PCM volume. We can also see
From the melt fraction in Fig. 19, we can notice that the melt
that fins thicker than 4 mm do not further improve the perfor-
starting time is delayed with the increase in fin height. This again
mance of the heat sink. We can say that the heat sink reaches it
can be attributed to the fact that shorter fins have a lower specific
maximum performance capability at a 4-mm fin thickness. Too
heat capacity than the taller fins. We must bear in mind that all heat
thick a fin allows for only little volume of PCM, this limits its
sink models have the same fin thickness. This lower heat capacity
functioning as a latent heat storage material.
Another thing to note is that the 4-mm and 6-mm thick fins give
almost similar results in Fig. 16. This is possibly due to the fact that
1
the heat sink has reached its optimum performance with 4-mm
thick fins. Although 6-mm thick fins give similar results, it is
expected that any further increase in thickness will reduce the 0.8 2 mmthick
performance of the heat sink. 4 mmthick
liquid f ract ion

Fig. 16 also shows that the simulation results compare reason- 6 mmthick
ably well with the experimental data unlike Fig. 12. 0.6
Fig. 17 shows that the melt starting time is delayed as the fin
thickness increases. This is because when the PCM is in the solid
phase, the main mode of heat transfer in the PCM is through heat 0.4
conduction. Heat conduction requires a temperature gradient. So
the larger the temperature gradient, the greater is the heat
0.2
conduction. We can see from Fig. 17 that the thinner fins have
a higher temperature gradient throughout as compared to the
thicker fins. This will cause a higher heat conduction rate in the heat 0
sink with thinner fins. Thus, the PCM will reach its melting
12

15

18

21

24

27

30
0

temperature faster. Also we can observe that the thicker fins have tim e(m in)
a shorter melting period. This is due to the lower volume of PCM
which means lower latent heat capacity. Fig. 17. Melt fraction versus time for different fin thickness.
 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838 3837

250

200
temperature(C)

150

10mmheight(3d)

20mmheight(3d)
100
40mmheight(3d)

40mmheight(2d)
50
20mmheight(2d

10mmheight(2d)

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
time(min)
Fig. 18. Case temperature versus time for different fin height.

1 fin thickness beyond this optimum point will yield no further

improvement in thermal performance. Melt starting time is
delayed as the fin thickness is increased. The thicker fins also have
0.8
a shorter melting period. For thicker fins, the melt front moves in
parallel with the fin. For thin fins the melt front movement is not
liquid fraction

0.6 uniform. Also it is noticed that in earlier periods, the lower regions
10m m height
of the PCM for the thin fins seem to melt faster. This indicates that
20m m height the primary heating is from the base of heat sink. For both cases
0.4
40m m height however, in later periods of melting, the upper regions of the PCM
seem to melt faster, indicating internal fluid convection. Internal
0.2 fluid convection is a far more effective heat transfer mechanism
compared to heat conduction within PCM. The performance of the
0 heat sink increases with increasing fin height. The melt starting
time is delayed with an increase in fin height. Also the melting
12

15

18

21

24

27

30
0

period is reduced with increasing fin height.

time(min)

Fig. 19. Melt fraction versus time for different fin height.

References
storage means a faster rate of temperature rise for a given heat [1] R. Tummala, E. Rymaszewski, A. Klopfenstein, Microelectronic Packaging
input. This will cause the fins and PCM to reach the PCM melting Handbook. Van Nostrand-Reinhold, New York, 1989, p. 389.
temperature faster and thus begin to melt earlier. We also can [2] P. Mithal New York, ASME, Design of Experimental Based Evaluation of
Thermal Performance of a Flichip Electronic Assembly, ASME EEP Proceedings,
notice that the melting period is reduced with an increase in fin vol. 18 (1996) 109e115.
height. This is due to the lower content of PCM in the heat sink with [3] S. Krishnan, S.V. Garimella, Analysis of a phase change energy storage system
taller fins. for pulsed power dissipation, International Journal of Heat and Mass Transfer
49 (2006) 1833e1844 IEEE Transactions on components and packaging
Technologies, volume 27, No. 1, March 2004.
6. Conclusion [4] E.M. Alawadhi, C.H. Amon, Performance Analysis of An Enhanced PCM
Thermal Control Unit, Thermal and Thermomechanical Phenomena in Elec-
tronic Systems 2000, Las Vegas, NV, USA, 23e26 (May 2000).
Heat sinks with internal fins give better thermal performance [5] M.J. Vesligaj, C.H. Amon, Transient thermal management of temperature
(lower temperature) compared to one without internal fins. The Fluctuations during time varying Workloads on portable electronics, IEEE
greater the number of fins, the lower is the case temperature during Transactions on Components.
[6] Z.X. Gong, A.S. Mujumdar, Flow and heat transfer in convection-dominated
melting. Also, a heat sink with more fins starts melting later as melting in a rectangular cavity heated from below, International Journal of
compared to a heat sink with lesser fins (of the same size). Also, the Heat and Mass Transfer 41 (17) (September 1998) 2573e2580.
melt period is reduced with an increase in the number of fins. This [7] K.C. Nayak, S.K. Saha, K. Srinivasan, P. Dutta, A numerical model for heat sinks
with phase change materials and thermal conductivity enhancers, Interna-
is due mainly to the smaller PCM volume in the heat sink. It is
tional Journal of Heat and Mass Transfer 49 (11e12) (June 2006) 1833e1844.
observed that there is more convective fluid motion in the heat sink [8] V. Shatikian, G. Ziskind, R. Letan, Numerical investigation of a PCM-based
with more fins. This results in a higher melting rate. heatsink with internal fins, International Journal of Heat and Mass Transfer
An increase in fin thickness shows only a slight improvement in 48 (17) (August 2005) 3689e3706.
[9] P. Lamberg, Approximate analytical model for two-phase solidification
thermal performance. Also it is observed that there is an optimum problem in a finned phase-change material storage, Applied Energy 77 (2)
point to which the fin thickness can be increased to. Increasing the (February 2004) 131e152.
3838 S.F. Hosseinizadeh et al. / Applied Thermal Engineering 31 (2011) 3827e3838

[10] G.W. Burns, M.G. Scroger, G.F. Strouse, M.C. Croarkin, W.F. Guthrie, Temper- different phase-change materials, Energy Conversion and Management 37
ature-electromotive Force Reference Functions and Tables for Letter- (12) (1996) 1775e1785.
designated Thermocouple Types Based on the ITS-90", NIST Monograph, vol. [13] M. Esen, A. Durmus, A. Durmus, Geometric design of solar-aided latent heat
175 (1993) 250. store depending on various parameters and phase change materials, Solar
[11] B.N. Taylor and C.E. Kuyatt, Guidelines for Evaluating and Expressing the Energy 62 (1) (1998) 19e28.
Uncertainty of NIST Measurement Results, NIST Technical Note 1297 [14] M. Esen, Thermal performance of a solar-aided latent heat store used for space
(1994). heating by heat pump, Solar Energy 69 (1) (2000) 15e25.
[12] M. Esen, T. Ayhan, Development of a model compatible with solar assisted [15] C. Hirt, B. Nichols, Volume of fluid (VOF) method for the dynamics of the
cylindrical energy storage tank and variation of stored energy with time for boundaries, Journal Of Computational Physics 39 (1981) 201.