Applied Thermal Engineering 37 (2012) 1e9

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

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

Experimental and numerical studies on melting phase change heat transfer

in open-cell metallic foams filled with paraffin
W.Q. Li, Z.G. Qu*, Y.L. He, W.Q. Tao
MOE Key Laboratory of Thermo-Fluid Science and Engineering, Xi’an Jiaotong University, 28# XianNing Road, 710049 Xi’an, China

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

Article history: In the current study, the melting phase change heat transfer in paraffin-saturated in open-celled metallic
Received 28 July 2011 foams was experimentally and numerically studied. The experiments were conducted with seven high-
Accepted 2 November 2011 porosity copper metal foam samples (3
90%), and paraffin was applied as the phase-change material
Available online 17 November 2011
(PCM). The wall and inner temperature distribution inside the foam were measured during the melting
process. The effects of foam morphology parameters, including porosity and pore density, on the wall
Keywords:
temperature and the temperature uniformity inside the foam were investigated. The melting heat
Phase-change material
transfer is enhanced by the high thermal conductivity foam matrix, although its existence suppresses the
Metallic foam
Melting
local natural convection. A numerical model considering the non-Darcy effect, local natural convection,
Non-equilibrium model and thermal non-equilibrium was proposed. The velocity, temperature field, and evolution of the solid
eliquid interface location at various times were predicted. The numerically predicted results are in good
agreement with the experimental findings. The model as well as the feasibility and necessity of the
applied two-equation model were further validated.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction electronic devices. For example, in some electronic cooling systems,

Phase-change materials (PCMs) have been widely used in many
applications, such as passive cooling for electronic devices, protec-
tion systems in aircrafts, food processing, and energy conservation in
buildings, because of their high latent heat, chemical stability, suit-
able phase-change temperature, and reasonable price. Experimental
and analytical/numerical studies in published literatures have
focused on moving boundary problems [1e4]. Agyenim et al. [5]
summarized the various applications of PCMs with different
melting temperatures in suitable thermal energy storage systems.
The size and the shape of the PCM container were also taken into
consideration to ensure the long-term stable thermal performance of
the system. Dutil et al. [6] presented main four types of numerical
solutions dealing with the thermal behaviors of PCM in solid/liquid
systems and showed the predicted results of different configurations.
Although some organic PCMs, such as paraffin, are very popular
in energy storage applications and electronic cooling systems
because of the aforementioned advantages and their low density
compared with other kinds of heat storage materials (e.g., metal
PCMs and hydrated salts). However, organic PCMs suffer from low
conductivity (z0.1 W m1 K1), which is likely to cause failure in
* Corresponding author. Tel./fax: þ86 029 82668036.
E-mail address: zgqu@mail.xjtu.edu.cn (Z.G. Qu).
the chip experiences a transient or periodic heat generation, which
requires a highly efficient coolant system to dissipate the heat in
case of chip exposure to extreme-temperature environments.
However, the poor conductivity of organic PCMs reduces the rate of
heat storage, thereby increasing the junction temperature of the
devices beyond the allowable range.
In order to address this unacceptable problem, thermal
conductivity enhancement techniques that increase heat transfer
rates have been developed. These enhancement techniques are
summarized in the following three methods: (1) Dispersing high-
conductivity particles in PCMs; Wang et al. [7] experimentally
proved that the thermal conductivity of composite PCMs is
enhanced by incorporating a b-aluminum nitride additive. (2)
Utilizing high-conductivity matrices, such as a metal or a graphite
compound, as heat delivery promoters; Kim and Drazal [8]
improved the effective thermal conductivity of paraffin by stirring
exfoliated graphite nanoplatelets (xGnP) in liquid paraffin. The
authors found that the thermal conductivity of paraffin/xGnP
composite PCMs increases as the xGnP loading content increases
without reducing the latent heat of the paraffin wax. (3) Filling
extensive surfaces such as fins into the body of PCM; Shatikian et al.
[9] numerically investigated the effect of internal fins on melting
rate, melting front profiles, and heat transfer. (4) Varying the shape
of the heat storage-heat transfer system; Banaszek et al. [10].

1359-4311/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2011.11.001
2 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9
Nomenclature
A additional term in the momentum equation
asf interfacial surface area (m1)
C constant parameter
CI inertial coefficient
cpf specific heat capacity of PCM (J kg1 K1)
cps specific heat capacity of metal foam (J kg1 K1)
ctd thermal dispersion coefficient
df fiber diameter (m)
dp pore size (m)
dk characteristic length (m)
fl liquid fraction in the pore
g gravitational acceleration (m s2)
hsf interfacial heat transfer coefficient (W m2 K1)
K permeability (m2)
k thermal conductivity (W m1 K1)
ke effective thermal conductivity (W m1 K1)
kf thermal conductivity of fluid (W m1 K1)
kfe effective thermal conductivity of fluid (W m1 K1)
ks thermal conductivity of solid (W m1 K1)
kse effective thermal conductivity of metal foam
(W m1 K1)
L latent heat of paraffin (J kg1 K1)
Nu Nusselt number
P pressure (Pa)
PPI pore number per inch
Pr Prandtl number
q heat flux (W m2)
Re Reynolds number
designed an energy conservation system using helix-shaped
channels where PCM and heat transfer fluids were separately
enclosed in every other channel. Such a design makes it possible for
the fluid to stick to the channel wall more strictly when flowing
through the channel because of centrifugal force and more effi-
ciently take away heat from the PCM. However, a disadvantage of
such a structure for heat exchange is the increased consumption
and complexity of the cooling system.
In addition to the aforementioned methods, inserting PCMs into
a porous media appears to be an attractive choice to enhance heat
transfer. A number of experimental and theoretical investigations
on this problem have been conducted in the past two decades.
Beckermann and Viskanta [11] conducted both experimental and
numerical studies on the melting of gallium in glass beads. In their
numerical study, the one-equation energy model was adopted to
predict the temperature, taking into account the natural convective
effect. To the authors’ best knowledge, the one-equation model is
suitable only when the thermal properties of the solid matrix and
the saturated PCM are in the same order of magnitude; that is, the
macroscopic temperatures of the two materials should be close
enough so that a single temperature model can describe the whole
melting process. This equilibrium model fails when the thermal
properties of the two materials are significantly different or when
the convective effect becomes vitally important [12]. Extensive
studies are conducted to improve the accuracy of the model.
Harris et al. [13] theoretically studied the phase-change process in
porous media using a two-temperature model. However, the main
drawback of this approach is its difficulty in handling complex
cases, such as buoyancy-driven natural convection in a molten
region in porous media. Krishnan et al. [14] numerically investi-
gated the thermal transport phenomena associated with phase
change in a rectangular cavity filled with metal foam. The
T temperature ( C)
Tm1 the lower limit of melting point
Tm2 the upper limit of melting point
t time (s)
u,v velocity in x and y directions (m s)
x,y cartesian coordinates
Greek symbols
r density (kg m3)
d liquid fraction (¼3 fl)
b thermal expansion coefficient (K1)
m dynamic viscosity (N s m2)
3 porosity
c tortuosity coefficient
u pore density (pore per inch, PPI)
Subscripts
amb ambient
f fluid (both solid and liquid paraffin)
fl liquid paraffin
fs solid paraffin
m melting
p pore
s solid matrix
sf surface
td thermal dispersion
Superscripts
n iteration number at the present time level
influences of the Rayleigh, Stefan, and Nusselt numbers on the
evolution of the solid/liquid interface were reported and discussed.
Although this technique can solve more complicated cases, the
thermal dispersion effects were not considered, thereby making
the technique unsuitable for cases with a high Rayleigh number.
Moreover, the effective thermal conductivity is only determined by
the volume fraction of each phase without considering the
geometric configuration effect of the metal structure, which is not
suitable for metal foam. Lafdi et al. experimentally investigated the
phase-change heat transfer within a PCM composite permeated
with aluminum foam [15], and then numerically studied the cool-
ing process of the electronic device using heat sinks of solid matrix
impregnated with PCM [16]. Zhao et al. [17] experimentally and
numerically studied the paraffin melting in metallic foam.
However, in their model, the flow motion of liquid paraffin, which
has been proved to be very important, was not taken into account.
Jegadheeswaran and Plhekar [18] have summarized various tech-
niques for enhancing the thermal performance in different heat
thermal storage systems and enumerated the pros and cons of
these enhancement techniques.
The aim of the present work was to experimentally and
numerically investigate the melting process of paraffin in copper
foams. Metal foam, a metallic porous matrix material, is a porous
media that exhibits the excellent combination of compactness, low
weight, and high thermal conductivity [19]. An experimental test
rig was built to measure the wall and internal temperatures. The
effects of porosity and pore density on heat transfer were examined
and discussed. A two-equation model was subsequently applied to
numerically study the solideliquid phase-change process in high-
porosity metallic foam. In the model, the natural convection of
liquid paraffin, the thermal dispersion effect, the irregular geom-
etry configuration, and the non-Darcy effects were fully considered.
 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9
The numerical predictions were validated by comparing the
experimental results. The velocity field of the molten paraffin, the
temperature distributions for PCM and the metal matrix, as well as
the time-related interface variations were presented and discussed.
2. Experimental apparatus and procedure
2.1. PCM melting point and latent heat measurement
The latent heat and melting point of the paraffin were measured
using a differential scanning calorimeter (DSC; TA-Q20). The results
shown in Fig. 1 indicate that the latent heat of paraffin is 102.1 J/g,
and the melting process starts at 46.48  C and ends at 60.39  C. The
related thermophysical properties were listed in Table 1.
2.2. Experimental apparatus and procedure
As shown in Fig. 2, a plexiglass cavity was designed to cage the
paraffin-saturated metal foam based on the size of the metal foam.
However, considering the volume expansion of PCM during phase
change, a 5 mm gap between the top surface and the inner glass
surface was left for the expansion of the melted paraffin. In case of
leakage during melting, two other 30 mm thick plexiglass plates
were fabricated and fixed to the left and right sides of the cavity and
tightened at the corners with four studs. All copper foam samples
having a uniform dimension of 100 mm  100 mm  45 mm were
sintered on a 2 mm thick copper substrate at the bottom to improve
the thermal conduction to the internal space and minimize the
contact thermal resistance. A 100 mm  100 mm  0.15 mm
electric insulated film heater was glued to the left side of the metal
foam, i.e., a brazed copper substrate, to provide constant heat flux.
The other four sides (top, bottom, front, and back) were kept
thermally insulated with 30 mm thick urethane foam plates. A
constant heat flux of 4000 W/m2 was supplied by a DC voltage
stabilizer to the film heater during the experiment. For each
sample, a temperature-tracking task was completed by a Keithley
2700-multimeter/data acquisition system and data were obtained
every 5 min. A small cylindrical hole (3 mm in diameter, 45 mm in
height) erected on the sintered copper plate was drilled in the
center of the foam to better understand the time-related variation
in the inner temperatures. Five T-type thermocouples were
selected, as seen in Fig. 2(b). One was attached to the copper plate
to measure the wall temperature, and the other four were fixed
Fig. 1. DSC results for paraffin.
3
Table 1
Thermal physical properties of paraffin.
Parameters Value
rf (at 20  C) 785.02
L 1.021  105
Tm1 46.48
Tm2 60.39
kfs 0.30
kfl 0.10
cpf (at 20  C) 2850
m (at 65  C) 3.65  103
b (at 65  C) 3.085  104
inside the hole at 3, 13, 28, and 43 mm from the left side of the
metal foam. The uncertainties were estimated based on the random
errors during temperature measurements. The uncertainty for the
T-type thermal couple temperature is 0.1  C. The uncertainties for
the D.C. power output and the film heater power output are 1.22%
and 5.0%, respectively.
The morphology of the metal foam used in the experiment is
shown in Fig. 3. Seven copper foam samples with different poros-
ities and pore densities were used, the geometric characteristics of
which are listed in Table 2.
3. Experimental results and discussion
3.1. Effect of the metal matrix and foam porosity
Fig. 4 displays the wall temperature variation history with time
of three metal foams with different porosities (3 ¼ 0.90, 0.95, and
0.98), with the pore density fixed at 40 PPI. The wall temperature
variation of pure PCM was used as the reference. The whole melting
process can be divided into three regions: solid, mush, and pure
liquid. In the solid region, the temperatures for pure PCM and foam-
PCM composite both increased by absorbing explicit heat from the
wall. The foam-PCM composite wall temperature was lower than
that of pure PCM and showed better thermal management ability
because the effective PCM thermal conductivity was improved by
the copper metal matrix. As time progressed, the metal ligaments
reached a temperature higher than the PCM melting point and
started triggering a local phase change. Two factors dominate the
phase change, namely, the heat conduction and the natural
convection of the molten PCM. The final performance depends on
whichever of these two factors prevails in the melting process. For
the pure PCM, the paraffin absorbed the latent heat at the solid/
liquid melting interface dominated by natural convection, thereby
increasing the temperature of pure paraffin in a flattened manner.
However, in the foam-PCM composite, the heat conduction thermal
resistance was evidently reduced by the foam matrix, although the
liquid paraffin was constrained in the metal foam pore to some
extent and suppressed natural convection, which plays a positive
role in heat transfer by accelerating the melting rate. Consequently,
heat conduction dominated the melting process, and the wall
temperature of the foam-PCM composite was lowered and
increased almost linearly, governed by the combined effects of heat
diffusion, natural convection, and the latent heat absorption at the
moving boundary. When the melting process was fully complete,
the explicit heat exerted its role again and the temperatures for
both pure PCM and the composite continued to increase linearly.
Points A, B, C, and D represented the end of the melting process in
the mush region of the composites with 0.90, 0.95, and 0.98
porosities and pure PCM (3 ¼ 1). The total time consumption
decreased in the order of decreasing porosity because a higher
porosity indicates a higher PCM volume fraction. Higher porosity
means a lower effective thermal conductivity at a fixed pore
4 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9

Fig. 2. Experimental apparatus and test section. (a) Experimental apparatus, (b) Test section.

density, and implies a smaller fiber diameter or smaller heat
transfer surface area for the prevailing heat conduction process.
Thus, higher porosity allows the sample to achieve higher wall
temperature. The effective thermal conductivity can be calculated
according to Ref. [20]. The interfacial surface area can be obtained
from Eq. (1) [21]:
3pdf
asf ¼
d2p
3.2. Effect of pore density
increasing pore density at the same time; however, the influence
was less significant for the pore density compared with the
porosity. As pore density increased, the interfacial surface area and
the effective thermal conductivity increased, thereby improving
the melting heat transfer; nevertheless, the permeability
decreased, resulting in a strong suppression of the natural
convection of molten wax. The two opposite factors competed and
strong suppression of the natural convection is comparatively
(1)
considerable, resulting in higher wall temperature for high pore
density foam and a less evident temperature difference among the
three composites.

Fig. 5 depicts the relationship between the wall temperature

and the duration time for the copper foam filled with PCM of three Table 2
pore densities (10, 20, and 40 PPI) and pure PCM with a fixed 0.9 Geometrical parameters of the sample foams.
porosity. A similar trend was observed when the wall temperature Sample Porosity Pore density (PPI) dp (mm) df (mm)
for the copper foam filled with PCM was lower than that of pure 1 0.90 10 2.540 0.309
PCM. The total melting duration time at various pore densities were 2 0.90 20 1.270 0.154
almost the same because the PCM volume fraction were identical at 3 0.90 40 0.635 0.077
a fixed porosity. The wall junction temperature became higher with 4 0.95 10 2.540 0.218
5 0.95 20 1.270 0.109
6 0.95 40 0.635 0.055
7 0.98 40 0.635 0.035

Fig. 4. Variations of Tw with time of PCM composites with different porosities at a pore
Fig. 3. Typical structures of tested metal foam (3 ¼ 0.90, u ¼ 40 PPI). density of 40 PPI.
 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9 5

PPI foam than for the 40 PPI foam. The motion of molten paraffin wax
in the 10 PPI foam was less restricted in the matrix because of the
larger pore size, leading to a stronger natural convective effect.
Consequently, the temperature distribution was more uniform in the
horizontal direction. The effect of porosity on temperature distribu-
tion uniformity is related to pore density because the total thermal
resistance is determined by both the effective thermal conductivity
and natural convection. The natural convection was severely sup-
pressed at the 40 PPI pore density, whereas it was considerable at the
10 PPI pore density. When the porosity varied from 0.95 to 0.9, the
effective thermal conductivity increased or the heat conduction
thermal resistance decreased. The influence of the effective thermal
conductivity increment by decreasing the porosity from 0.95 to 0.9 on
the total resistance was evident for the 40 PPI foam compared with
the 10 PPI foam. Hence, the improvement in temperature distribution
uniformity is more significant for the 40 PPI foam than for the 10 PPI
foam, as shown in Fig. 6(a)e(d).

Fig. 5. Variations of Tw with time for the PCM composites with different pore densities 4. Numerical simulation
(3 ¼ 0.90).

4.1. Physical and mathematical model

3.3. Internal temperature distributions
The two-dimensional schematic diagram of the present tran-
Fig. 6 quantitatively presents the internal temperature distribution
sient melting heat transfer is shown in Fig. 7. The copper foam
at four different foam composite test spots at two porosities (3 ¼ 0.90
saturated with solid paraffin measured 45 mm in length and
and 0.95) and pore densities (u ¼ 10 and 40 PPI). A comparison of
100 mm in height. The top, bottom, and right sides can be
Fig. 6(a) and (b) or Fig. 6(c) and (d) shows that the temperature
considered adiabatic. The initial temperature was set as the
difference between the two adjacent tested spots is smaller for the 10
ambient temperature at 25  C. The left side was heated at

a b

c d

Fig. 6. Internal temperature-time variation.

6 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9

Continuous equation:

vrf
!
þ V$ rf h U i ¼ 0 (2)
vt
Momentum equations:
 
rf vhui rf ! vP m m rf CI !
þ 2 ðh U i$VÞhui ¼  þ fl V2 hui  fl þ pffiffiffiffi jh U ij hui
d vt d vx d K K
þ Ahui ð3Þ

 
rf vhvi rf ! vP mfl 2 mfl rf CI !
þ 2 ðh U i$VÞhvi ¼  þ V hvi  þ pffiffiffiffi jh U ij hvi
d vt d vy d K K


þ rf g b Tf  Tm1 þ Ahvi ð4Þ

Energy equation for PCM:

! D E
Fig. 7. Schematic diagram of the physical domain.
dfl v Tf ! D E
3 rf cpf þ L D E þ rf cpf ðh U i$VÞ Tf
a constant heat flux of 4000 W m2, consistent with the input d Tf vt
power of the heating film. The following assumptions are made in
 D E
D E
the mathematical model for the present problem: (1) the metal ¼ kfe þ ktd V2 Tf þ hsf asf hTs i  Tf (5)
foam is assumed homogeneous and isotropic; (2) the liquid PCM is
Energy equation for metal matrix:
considered impressible and Newtonian, and subjected to the
Boussinesq approximation; (3) the changes of density and heat vhTs i
D E
capacity of paraffin during the whole melting process is neglected ð1  3 Þrs cps ¼ kse V2 hTs i  hsf asf hTs i  Tf (6)
vt
while the other thermal physical properties of PCM are constant at
each phase but treated separately for the solid and liquid phases. Where thermal properties ofrf, L,cpf,mfl,b are presented in Table 1.
Based on the above-mentioned considerations, the volume- The properties ofrs (8920 kg m-3), cps (380 J kg-1 K-1) and ks
averaging continuous, momentum, and two-energy equations for (401 W m-1 K-1)are density,heat capacity and the thermal
PCM and the solid matrix can be written as follows: conductivity of copper, respectively. Other parameters in the

Table 3
The employed semi-empirical correlations of metallic foam parameters.

Parameter Correlation Reference

Permeability, K 3
2 d2 [23]
k
K ¼
36ðc  1Þc
 
4p 1
c ¼ 2 þ 2cos þ cos1 ð23  1Þ
3 3
c 22:4  103
dk ¼ dp dp ¼
3c u
pffiffiffiffi
Inertial coefficient, CI FI ¼ CI ð1  3 Þ0:132 ðdf =dp Þ1:63 = K , CI ¼ 2.12  103 [26]
rffiffiffiffiffiffiffiffiffiffiffi
13
df ¼ 1:18 dp
3p
( 0:76Re0:4 Pr0:37 k =d ; 1  Re  40
d f f d
hsf ¼ 0:52Re0:5 d Pr
0:37
kf =df ; 40  Red  103
0:26Re0:6 d Pr
0:37
kf =df ; 103  Red  2  105
Local heat transfer coefficient, hsf [24]
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Red ¼ rf u2 þ v2 df =ð3 mfl Þ
Thermal dispersion conductivity, ktd Ctd pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [25]
ktd ¼ r C d u2 þ v2 , Ctd ¼ 0.36
1  3 f pf f
1
ke ¼ pffiffiffi
2ðRA þ RB þ RC þ RD Þ
4l
RA ¼
ð2e2 þ plð1  eÞÞks þ ð4  2e2  plð1  eÞÞkf
ðe  2lÞ2
RB ¼
ðe  2lÞe2 ks þ ð2e  4l  ðe  2lÞe2 Þkf
pffiffiffi
ð 2  2eÞ2
RC ¼ pffiffiffi pffiffiffi pffiffiffi
2pl ð1  2 2eÞks þ 2ð 2  2e  pl ð1  2 2eÞÞkf
2 2

Effective thermal conductivity, ke 2e [20]

RD ¼ 2
e ks þ ð4  e2 Þkf
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi pffiffiffi
2ð2  ð5=8Þe3 2  23 Þ
l ¼ pffiffiffi , e ¼ 0.339
pð3  4 2e  eÞ
( kfs;e solid region
kfe ¼ fl kfl;e þ ð1  fl Þkfs;e Mush region
kfl;e liquid region
kse ¼ ke jkf ¼0 , kfe ¼ ke jks ¼0
 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9 7

governing equations, including K (permeability), CI (inertial coef- Table 4

ficient), kse (the effective thermal conductivity of the solid matrix), Initial and boundary conditions for the governing equations.

kfe (the effective thermal conductivity of paraffin), ktd (thermal Definite conditions Positions Velocity Temperature
dispersion conductivity), and hsf (local heat transfer coefficient), are Initial condition 0  x  L, 0  y  H u¼v¼0 Tf ¼ Ts ¼ Tamb
from the semi-empirical correlations in the literature and can be x ¼ 0, 0  y  H u¼v¼0 q ¼ qfþ qs
seen in Table 3, where the values of CI and Ctd shown the best fit to ¼3 q þ (1  3 )q
x ¼ L, 0  y  H u¼v¼0 vTf vTs
the experimental data in the corresponding literatures. For inter- ¼ ¼ 0
vx vx
facial heat transfer coefficienthsf, since no exact correlation is Boundary conditions 0  x  L, y ¼ 0 u¼v¼0 vTf vTs
¼ ¼ 0
suitable for natural convection in porous media, an empirical vy vy
correlation provided by Zukauska [24] has been introduced. The 0  x  L, y ¼ H u¼v¼0 vTf vTs
¼ ¼ 0
Reynolds number is based on the cylinder diameter and Pr is the vy vy

fluid Prandtl number. In this study, the Re range was verified to be

less than 5.0 which satisfied the first term of correlation propased
by Zukauska [24] and then was applied. The bracket “hi” represents
the volume-averaging model based on the DupuiteForchheimer
relationship [22]. The second and fourth terms on the right side of
Eqs. (3) and (4) account for the extension of Darcy’s law to explain
the non-Darcy effects, and the boundary and inertial effects
proposed by Brinkman and Forchheimer, respectively. The last
terms in the momentum equations are additional source terms,
where parameter A is related to the liquid fraction in the pore
volume based on Kozeny’s equation [22], which is introduced as the
following equation:
5. Numerical results and discussion
5.1. Code validation
The numerical codes were validated with the present experi-
mental results by comparing the interface positions at three
different moments. Fig. 8(a) and (b) are pictures of interface loca-
tions captured by a digital camera at 3600 s and 3780 s for the 0.90
porosity and 10 PPI foam. The predicted solid/liquid interfaces
[Fig. 8(c) and (d)] obtained from the numerical predictions agree


C 1  fl2
A ¼ (7)
S þ fl3

8 0 Tf  Tm1
<
.
fl ¼ Tf  Tm1 ðTm2  Tm1 Þ Tm1  Tf  Tm2 (8)
:
1 Tm2  Tf

d ¼ 3 $fl (9)

where d is the liquid fraction considered as a function of PCM

temperature and porosity, and C and S are set as 1015 and 1010,
respectively, to fix the PCM velocity to zero before melting for
convenient numerical implementation. Therefore, the conduction
dominated equations for the solid phase and convection-diffusion
controlled equations for melted paraffin are unified by Eqs.
(3)e(5). The initial and boundary conditions for the governing
equations are given in Table 4. The initial temperatures for PCM and
metal foam are equal to the ambient temperature. The initial values
of velocity for PCM at both directions are zero and the same value is
applied to the velocity boundary condition based on the velocity
non-slip principle at the wall. At the left boundary, the cover plate
contacts with metal fiber or the fluid (solid or liquid), due to this
consideration, if ignoring the heat dissipation through the sintered
2 mm copper substrate, the heat is assumed distributed respec-
tively to the metal foam and PCM by each representative volume
ratio at the wall surface as applied in [26]. The rest boundaries are
set adiabatic.

4.2. Numerical procedure

The governing equations were discretized by a finite volume

method in a staggered grid system of 90  80. The combined
equations are solved numerically using the IDEAL algorithm [27].
The regions of solid paraffin and melted region are determined
automatically during the iteration by the whole domain solution
strategy. The iteration at the present time layer is considered
convergent if the maximum relative residuals of Tf are less than Fig. 8. Comparison of the solid/liquid interface between the photos and numerical
105, which can be expressed as ½ðTfn  Tfn1 Þ=Tfn1  < 105 . results. (3 ¼ 0.90，u ¼ 10 PPI) at different times.
8 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9
Fig. 9. Velocity field and temperature distributions (3 ¼ 0.95, u ¼ 20 PPI) at 2700 s.
well with the experimental results. The feasibility of the present
model is verified.
5.2. Temperature distribution
Fig. 9 shows the velocity field and temperature contours of PCM
and the solid matrix (3 ¼ 0.95, u ¼ 20 PPI) at 2700 s. The hotter
liquid moved upward because of the natural convection of molten
paraffin stimulated by buoyancy, pushing the upper portion of the
interface faster than the bottom. For the temperature distribution,
it could be clearly observed that there exist great differences
between metal foam and the PCM. Also, at the same local position,
metal foam temperature was higher than that of PCM. In the solid
PCM region, conduction-induced heat transfer produced similar
temperature profiles for the PCM and the metallic matrix; however,
the metallic foam temperature was higher than that of the solid
PCM due to the local thermal non-equilibrium effect. In the liquid
wax region, the liquid temperature pattern differs from that of the
metallic matrix. The temperature profile of the liquid paraffin was
more uniform in the horizontal direction than that of the metallic
matrix because natural convection prevailed in the liquid region,
a b
0.10
0.08
0.06
Y
Y
0.04 600s
900s
1200s
0.02 1500s
1800s
2100s
0.00
0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045
X X
ε=0.90
Fig. 10. Solid/liquid interface locations at various times (u ¼ 20 PPI).
whereas only diffusion existed in the solid paraffin region. A
difference in the temperature distributions between PCM and the
metal matrix was observed, especially at the liquid paraffin region.
This quantitative result validates the feasibility and necessity of the
two-temperature model.
5.3. Interface variations
Fig. 10(a) and (b) compare the time-related interface position for
the 0.9 and 0.95 porosities at a fixed 20 PPI pore density. At the
beginning of the melting process, the interface tended to be vertical
because the liquid fraction was relatively small, indicating that heat
conduction dominated in the process. As melting continued, the
upper part moved faster than the lower part because of the eddy
flow, promoting the fusion rate at the top and the phase change was
accelerated because of the natural convection induced by buoyancy
in the molten PCM. Moreover, the interface movement rate for the
0.90 porosity foam was higher than that of the 0.95 porosity foam.
At 1500 s, the top phase interface for the 0.90 porosity foam had
already attached to the right boundary, whereas the interface of the
0.95 porosity foam had just crossed over the middle (X/H ¼ 0.57).
0.10
0.08
0.06
600s
900s
0.04
1200s
1500s
1800s
0.02 2100s
2400s
3000s
0.00
0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045
ε=0.95
 W.Q. Li et al. / Applied Thermal Engineering 37 (2012) 1e9
The above-mentioned results are consistent with the experimental
results indicated in Fig. 4.
6. Conclusion
The melting phase-change heat transfer in paraffin PCM was
enhanced by the porous metallic foam. The mechanisms of heat
diffusion and natural convection dominated the melting phase-
change heat transfer. The enhancement of heat conduction is
more prominent than the suppression of natural convection of
molten liquid by the copper matrix in the foam-PCM composite,
whereas natural convection prevailed in the pure PCM. The final
total thermal resistance was lower for the foam-PCM composite
than the pure PCM to result in the lower corresponding wall
temperature. The influence of pore density on wall temperature
was less sensitive than that of porosity. The uniformity of the
temperature distribution inside the foam-PCM composite was
augmented either by decreasing pore density to accelerate natural
convection or by decreasing porosity to improve the effective
thermal conductivity. The feasibility of the numerical model, in
which the non-Darcy effects, the natural convection in the liquid
paraffin, and the local non-equilibrium effects were considered,
was validated by the experimental visualization. The temperature
distribution profile for the local solid matrix was quite different
from that of the filled PCM and validated the feasibility and
necessity of the applied two-equation model. The solid/liquid
movement rate was higher for the low porosity foam, which is also
consistent with the experimental results.
Acknowledgements
The current study was supported by the National Key Projects of
Fundamental R/D of China (973 Project: 2011CB610306), the
National Natural Science Foundation of China (Nos. 51176149 and
50736005), the National Excellent Doctoral Dissertation Founda-
tion of China (201041), the Doctoral Fund of the Ministry of
Education of China (200806981013).
References
[1] V.R. Voller, C. Prakash, A fixed grid numerical modelling methodology for
convectionediffusion mushy region phase-change problems, Int. J. Heat Mass
Transfer 30 (1987) 1709e1719.
[2] V.R. Voller, J.B. Swenson, C. Paola, An analytical solution for a Stefan problem
with variable latent heat, Int. J. Heat Mass Transfer 47 (2004) 5387e5390.
[3] T.J. Scanlon, M.T. Stickland, A numerical analysis of buoyancy-driven melting
and freezing, Int. J. Heat Mass Transfer 47 (2004) 429e436.
9
[4] P. Lamberg, R. Lehtiniemi, A.M. Henell, Numerical and experimental investi-
gation of melting and freezing processes in phase change material storage, Int.
J. Therm. Sci. 43 (2004) 277e287.
[5] F. Agyenima, N. Hewitt, P. Eames, M. Smyth, A review of materials, heat transfer
and phase change problem formulation for latent heat thermal energy storage
systems (LHTESS), Renew. Sust. Energ. Rev. 14 (2010) 615e628.
[6] Y. Dutil, D.R. Rousse, N.B. Salah, S. Lassue, L. Zalewski, A review on phase-
change materials: mathematical modeling and simulations, Renew. Sust.
Energ. Rev. 15 (2011) 112e130.
[7] W. Wang, X. Yang, Y. Fang, J. Ding, J. Yan, Enhanced thermal conductivity and
thermal performance of form-stable composite phase change materials by
using b-Aluminium nitride, Appl. Energy 86 (2009) 1196e1200.
[8] S. Kim, L.T. Drazal, High latent heat storage and high thermal conductive
phase change materials using exfoliated graphite nanoplatelets, Sol. Energy
Mater. Sol. Cells 93 (2009) 136e142.
[9] V. Shatikian, G. Ziskind, R. Letan, Numerical investigation of a PCM-based heat
sink with internal fins, Int. J. Heat Mass Transfer 48 (2005) 3689e3706.
[10] J. Banaszek, R. Domañski, M. Rebow, F. El-Sagier, Numerical analysis of the
paraffin waxeair spiral thermal energy storage unit, Appl. Therm. Eng. 19
(1999) 1253e1277.
[11] C. Beckermann, R. Viskanta, Natural convection solid/liquid phase change in
porous media, Int. J. Heat Mass Transfer 31 (1988) 35e46.
[12] M. Kaviany, Principles of Heat Transfer in Porous Media (Mechanical Engi-
neering Series). Springer, Berlin, 1995.
[13] K.T. Harris, A. Haji-Sheikh, A.G. Agwu Nnanna, Phase-change phenomena in
porous media-a non-local thermal equilibrium model, Int. J. Heat Mass
Transfer 44 (2001) 1619e1625.
[14] S. Krishnan, J.Y. Murthy, S.V. Garimella, A two-temperature model for solid-
liquid phase change in metal foams, J. Heat Transfer 127 (2005) 995e1004.
[15] K. Lafdi, O. Mesalhy, S. Shaikh, Experimental study on the influence of foam
porosity and pore size on the melting of phase change materials, J. Appl. Phy.
102 (2007) 083549.
[16] K. Lafdi, O. Mesalhy, A. Elgafy, Merits of employing foam encapsulated phase
change materials for pulsed power electronics cooling applications, J. Elec.
Packing 130 (2008) 021004.
[17] C.Y. Zhao, W. Lu, Y. Tian, Heat transfer enhancement for thermal energy
storage using metal foams embedded within phase change materials (PCMs),
Sol. Energy 84 (2010) 1402e1412.
[18] S. Jegadheeswaran, S.D. Pohekar, Performance enhancement in latent heat
thermal storage system: a review, Renew. Sust. Energ. Rev. 13 (2009)
2225e2244.
[19] M.F. Ashby, T.J. Lu, Metal foams: a survey, Sci. China (Series B.) 46 (2003) 521e532.
[20] K. Boomsma, D. Poulikakos, On the effective thermal conductivity of a three-
dimensionally structured fluid-saturated metal foam, Int. J. Heat Mass
Transfer 44 (2001) 827e836.
[21] V.V. Calmidi, R.L. Mahajan, Forced convection in high porosity metal foams,
J. Heat Transfer 122 (2000) 557e565.
[22] A. Donald Nield, Convection in Porous Media. Springer, New York, 2006.
[23] J.G. Fouriea, J.P. Du Plessis, Pressure drop modeling in cellular metallic foams,
Chem. Eng. Sci. 57 (2002) 2781e2789.
[24] A. Zhukauskas, Heat Transfer from Tubes in Cross Flow, Advances in Heat
Transfer. Academic, New York, 1972.
[25] J.G. Georgiadis, I. Catton, Dispersion in cellular thermal convection in porous
layers, Int. J. Heat Mass Transfer 31 (1988) 1081e1091.
[26] V.V. Calmidi, Transport phenomena in high porosity fibrous metal foams, Ph.
D thesis, University of Colorado, 1998.
[27] D.L. Sun, Z.G. Qu, Y.L. He, W.Q. Tao, An efficient segregated algorithm for
incompressible fluid flow and heat transfer problems e IDEAL (inner doubly
iterative efficient algorithm for linked equations) part I: mathematical
formulation and solution procedure, Numer. Heat Transfer, Part B. 53 (2008)
1e17.