Applied Energy 112 (2013) 1222–1232

Contents lists available at SciVerse ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Numerical modeling for solid–liquid phase change phenomena in porous

media: Shell-and-tube type latent heat thermal energy storage
Zhenyu Liu, Yuanpeng Yao, Huiying Wu ⇑
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

h i g h l i g h t s

" The non-equilibrium thermal model is adopted in the numerical simulation.

" The thermal conduction is dominant as PCM embedded within metal foam.
" The performance of latent heat storage using metal foam is significantly enhanced.
" The effects of HTF should be considered in real industrial application.

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

Article history: In this paper, a numerical model is established to predict the phase change material (PCM) melting pro-
Received 21 September 2012 cess in porous media. The heat transfer enhancement technique using metal foam in a shell-and-tube
Received in revised form 29 January 2013 type latent heat thermal energy storage (LHTES) unit is investigated. The solid–liquid phase change phe-
Accepted 5 February 2013
nomenon is solved with the enthalpy porosity theory. The computational fluid dynamic technology is
Available online 13 March 2013
adopted to predict the flow and heat transfer for the liquid PCM and heat transfer fluid (HTF). The appro-
priate source terms accounting for the pressure drop caused by the presence of solid PCM and porous
Keywords:
media are added to the momentum equations. The solid–liquid interface position and the temperature
Melting process
Latent heat thermal energy storage
distribution are predicted to describe the melting process. The effects of the structural parameters of por-
Phase change material ous media and the inlet conditions of HTF on the thermal characteristics of LHTES unit are analyzed. The
Porous media numerical model can be further extended to investigate the thermal performance of different types of
storage units in LHTES system.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction change to liquid PCMs. The PCMs absorb heat in a very small tem-
perature range, which store 5–14 times energy than those of sen-
The energy crisis becomes serious all over the world during the sible storage materials with the same volume. LHTES unit using
past decades and new techniques should be implemented to save PCM, which is a shell-tube or rectangular type, is a highly attractive
and reuse the waste energy in the industrial applications [1–3]. option in the energy storage applications. For the shell-and-tube
The thermal storage technology is significant to the efficient en- type, the PCM occupies the space between the tube and the shell.
ergy saving and the new energy developing, which bridges the The heat is transferred from the heat transfer fluid (HTF), which
gap between supply and demand and improves the energy system flows through a single tube.
performance [4,5]. In Lacroix’s work [9], a theoretical analysis was performed to
The latent heat thermal energy storage (LHTES) becomes one study the transient behavior of a LHTES unit. The thermal and geo-
attractive technique in recent years due to its high energy storage metric parameters were specified considerably to improve its ther-
capacity in a small operating temperature range [6–8], in which mal performance. In Ref. [10], a shell-and-tube type LHTES system
the heat is absorbed as the phase change material (PCM) turns was numerically analyzed, in which the solidification process was
from solid to liquid. Initially, these PCMs temperature rises when dominated by the heat conduction in PCM. Zivkovic and Fujii [11]
they absorb heat, which perform like sensible storage materials. numerically investigated the melting times of PCM in the rectangu-
Then, the thermal energy is transferred when the solid PCMs lar and cylindrical storage units, in which the two storage units had
same heat transfer area and same volume. Rosler and Bruggemann
[12] numerically and experimentally analyzed the paraffin wax
⇑ Corresponding author. Tel./fax: +86 21 34205299. melting process in a shell-and-tube type storage unit, the
E-mail address: [email protected] (H.Y. Wu). natural-convection dominated phase change of PCM showed a

0306-2619/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2013.02.022
 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232 1223

Nomenclature

Symbols x,y,z Cartesian coordinates

a permeability
Am consecutive number Greek
Asf interfacial surface area (m1) b liquid volume fraction
c specific heat (J kg1 K1) c expansion factor (K1)
Ci inertia coefficient e porosity or dissipation energy (m2 s3)
Co Courant number l dynamic viscosity (kg m1 s1)
d diameter (m) q density (kg m3)
df cell ligament diameter (m) x constant number
dp average diameter (m)
g gravity (m s2) Subscript
h heat transfer coefficient (W m2 K1) 0 initial
k thermal conductivity (W m1 K1) eff effective
K turbulence energy (J kg1) fe effective thermal conductivity of fluid
L latent heat (J kg1) in inlet
p pressure (Pa) m porous matrix
PPI pore number per inch p phase change material
Pr Prandtl number sf between solid and fluid
Ra Rayleigh number se effective thermal conductivity of solid
Re Reynolds number t turbulent
S source term td thermal dispersion
Sh source term in energy equation x X direction
t time (s) y Y direction
T temperature (K) z Z direction
u X-velocity (m s1)
v Y-velocity (m s1) Superscript
V velocity (m s1) ‘ fluctuation
~
V velocity vector (m s1)
w Z-velocity (m s1)

slow process, in which the isothermal VOF method was extended ied in [6,14,15], which include different configurations for finned
to solve the temperature distribution in PCM. The PCM melting/ tubes, bubble agitation, PCM with metal foam/high conductivity
solidification time was found to be one of the essential parameters particles, microencapsulated PCM, etc. These techniques are aimed
to determine the geometrical parameters of the storage unit. to improve the PCM thermal conductivities.
The previous studies show that the major disadvantage of PCMs The metal foam embedded within PCM was found to be prom-
is the low thermal conductivities, which limits their industrial ising due to its high surface density (heat transfer surface area per
application. The thermal conductivities of most paraffin waxes unit volume) and light weight [6]. The metal foam with cellular
and inorganic salts are around 0.2 W/(m K) and 0.5 W/(m K), structure (copper or aluminum) has a very high porosity, as shown
respectively, which leads to a low charging/discharging rate [13]. in Fig. 1, which can reach 95% of the volume consist of void spaces.
The related heat transfer enhancement techniques have been stud- It also has a high ratio of strength to density, a high thermal con-
ductivity and porous structure with light weight. The open cells
in the metal foam are randomly oriented and can be considered
to be homogeneous in size and shape. The experimental work in
[16,17] shows that the PCMs embedded in porous media can en-
hance the heat conduction and increase the charging/discharging
rate. The shell-and-tube type LHTES unit using metal foam can
be an innovative implement in the real industrial application,
and few reports can be found in the open literature.
In this work, a numerical model for the PCM embedded in metal
foam is developed, which is applied to a shell-and-tube type LHTES
unit. A parametric study is performed to analyze the effects of
influential factors on the storage unit thermal performance based
on the numerical predictions.

2. Numerical model

2.1. Model equations for melting process in porous media

The early studies treated the solid–liquid phase change problem

as conduction-controlled one. In fact, the effect of the natural con-
vection was found to be of importance in the real application [18].
In this work, the enthalpy porosity theory is employed to predict
Fig. 1. SEM image of the metal foam structure. the phase change process in the numerical model. The two-phase
1224 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232

mixed region can be described with the porosity, which can be Am = 108 kg/(m3 s). In this paper, the paraffin wax RT 58 is used in
compared to the liquid volume fraction. The different source terms the numerical simulation. The consecutive number is specified with
in the momentum equations can be used to evaluate the pressure the same value as the PCMs are from the same company. The sec-
losses caused by the presence of the solid PCM and the porous ond and third items on the right side of the equal sign of Eqs.
media. (6)–(8) contain the source terms accounting for the viscous resis-
To achieve a better convergence rate, the Boussinesq model was tance and the inertia resistance due to the porous media. The buoy-
adopted in many natural convection problems [18]. The changes of ancy effect is also included in Eq. (7), in which the gravity is along
fluid density with temperature meet the Boussinesq assumption, the negative Y axis.
which are included in the volume-related items in the momentum There are two thermal models available to simulate the heat
equations. transfer process in porous media. The equilibrium thermal model
treats that the liquid PCM and the porous media have the same
q ¼ q0 ½1  cðT  T 0 Þ ð1Þ
temperature and the conservation equation for energy can be de-
where q and q0 are fluid density and reference density, T and T0 are scribed as:
 
fluid temperature and reference temperature, and c is fluid expan- @T @T @T @T
sion factor.
qc þ ðqcÞp u þv þw
@t @x @y @z
The continuity equation can be described as: !
2 2 2
@ T @ T @ T @b
@ q @ðquÞ @ðqv Þ @ðqwÞ ¼ keff þ þ  eqp L ð9Þ
þ þ þ ¼0 ð2Þ @x2 @y2 @z2 @t
@t @x @y @z
where qm and qp are densities of porous matrix material and PCM,
where t is time; u, v and w are dimensional velocities in X, Y and Z respectively; cm and cp are specific heats of porous matrix material
direction, respectively. and PCM, respectively. qc can be calculated by the following equation:
When PCM and porous media are mixed as composite materi-
als, the heat transfer process will be strengthened. It also solves qc ¼ ð1  eÞqm cm þ eqp cp ð10Þ
the problems of liquid PCM leakage and corrosion. The flow pres-
The effective thermal conductivity keff introduced in the Eq. (9)
sure loss due to the solid PCM and the porous media can be deter-
can be calculated as the volume average of the conductivities of
mined by the source terms in the momentum equations, which are
porous matrix material and PCM:
expressed as:
  ! keff ¼ ð1  eÞkm þ ekp ð11Þ
@u @u @u @u @2u @2u @2u @p
q þu þv þw ¼l þ þ  þ Sx where km and kp are thermal conductivities of porous matrix mate-
@t @x @y @z @x2 @y2 @z2 @x
rial and PCM, respectively.
ð3Þ In the non-equilibrium thermal model, the porous media and
! fluid are not assumed to be in thermal equilibrium. The conserva-
 
@v @v @v @v @2v @2v @2v @p tion equations for energy are solved separately for the fluid and so-
q þu þv þw ¼l þ þ  þ Sy lid zones. The conservation equation for the fluid zone (PCM) is
@t @x @y @z @x2 @y2 @z2 @y  
@T f @T f @T f @T f
ð4Þ eðqcÞf þ eðqcÞf u þv þw
@t @x @y @z
! !
  @2T f @2T f @2T f @b
@w @w @w @w @2w @2w @2w @p ¼ ðkfe þ ktd Þ þ þ 2 þ hsf Asf ðT f  T s Þ  eqp L
q þu þv þw ¼l þ 2 þ 2  þ Sz @x2 @y2 @z @t
@t @x @y @z @x2 @y @z @z
ð5Þ ð12Þ
And the conservation equation for the solid zone (porous media) is
ð1  bÞ2 1 l !
Sx ¼ 3 Am u þ u þ C i qujuj ð6Þ @T s @2T s @2T s @2T s
ðb þ xÞ a 2 ð1  eÞðqcÞs ¼ kse þ 2 þ 2 þ hsf Asf ðT s  T f Þ ð13Þ
@t @x2 @y @z
ð1  bÞ2 l 1 where Asf is surface area density, which can be defined as the ratio
Sy ¼ 3
Am v þ vþ C i qv jv j  qg cðT  T 0 Þ ð7Þ
ðb þ xÞ a 2 of the heat transfer surface area to the volume; hsf is heat transfer
coefficient between the fluid and the porous media.
ð1  bÞ2 l 1
Sz ¼ Am w þ w þ C i qwjwj ð8Þ 2.2. Modeling on metal foams
ðb3 þ xÞ a 2

where the first item on the right side of the equal sign of Eqs. (6)–(8) 2.2.1. Permeability and inertia coefficient
is the source term due to the presence of solid, which was intro- For the laminar flow in porous media driven by the pressure gra-
duced by Voller and Prakash [19]. q is dynamic viscosity; b is vol- dient, the permeability is widely used to establish a relationship be-
ume fraction of liquid; Am is consecutive number for the mixture tween the pressure gradient and velocity [8]. The permeability is a
region. Am can be considered as porous media permeability. In the complicated parameter and not available to be theoretically deter-
2
ð1bÞ
full liquid zone (b = 1), ðb A ¼ 0. As it undergoes from the liquid mined for the natural convection in the metal foam because the re-
3
þxÞ m
ð1bÞ2
lated experimental data are rare in the open literature. As the
phase to the solid phase, b trends to be zero and A
ðb3 þxÞ m
tends to porous media porosity is very high, the permeability a and the iner-
infinity, in which a small constant x (less than 0.0001) is intro- tia coefficient Ci can be determined by the following equations [21]:
duced. The constants mentioned above have to be specified care- a ¼ 0:00073ð1  eÞ0:224 df
1:11 0:89
dp ð14Þ
fully. In the work of Shmueli et al. [20], a parametric analysis was
performed using the PCM of Rubitherm RT 27. In their simulation,  1:63
Am varied from 105 to 1010 kg/(m3 s). The best agreement between df
C i ¼ 0:00212ð1  eÞ0:132 ð15Þ
the numerical and experimental data was achieved when dp
 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232 1225

Table 1
Correlations of metal foam parameters.

Parameter Correlation No. Reference

 
kf 0:518Rad
1=4
3 (17) Churchill and Chu [22]
df
0:36 þ 0:599 9=16
4=9 Rad ¼ gbDdf =ðaf v f Þ
½1þð Pr Þ 
8
Local heat transfer coefficient hsf = >
> 0:76Red0:4 Pr0:73 kf (18) Zhukauskas [23]
>
> df
; 1 6 Red 6 40
< !
0:52Re0:5
d Pr
0:73
kf
df
; 40 6 Red 6 103 Red ¼ qj V jdf =ðlfl Þ
>
>
>
> 0:26Re 0:5
Pr 0:6
k 3 5
: d f
; 10 6 Red 6 2  10
df
! (19) Wakao and Kaguei [24]
1 j V jd
kf ½2 þ 1:1Pr 3 ð v f p Þ
Interfacial surface area density Asf = 3pdf (20) Calmidi and Mahajan [21]
2
dp
6ð1Þ (21) Dullien [25]
dp
3pdf ð1eð1Þ=0:04Þ (22) Calmidi and Mahajan [21]
ð0:59dp Þ2

Table 2
Physical parameters of Paraffin RT 58.

PCMs Paraffin RT 58
Melting temperature (K) 321–335
Thermal conductivity (W m1 K1) 0.2
Heat of fusion (kJ kg1) 181
Thermal expansion coeff. (K1) l.l  l04
Specific heat (kJ kg1) 2.1
Density of solid (kg l1) 0.84
Viscosity (Pa s) 0.0269

Fig. 2. Physical model of experiment performed in [16].

Fig. 4. The physical model for two-dimensional simulation (d1 = 40 mm and

d2 = 82 mm).

mally, the ligaments of the structure are treated as cylinders and

its roughness can be neglected. Different formulations for these
two parameters can be found in the previous studies and are sum-
Fig. 3. Results of different numerical models compared with experimental data. marized in Table 1. The formulation (17) and (22) are adopted in
the present numerical model based on the structure characteristic
of metal foam.
2.2.2. Independent parameters for metal foam
Normally, the porosity, the pore size and the ligament diameter
2.2.4. Effective thermal conductivity and thermal dispersion
are the parameters used to describe the structure characteristics of
conductivity
porous media. These are not independent parameters according to
For the non-equilibrium thermal model in this numerical simu-
the work done by Calmidi and Mahajan [21]. For the metal foam
lation, the effective thermal conductivity is defined separately for
used in this paper, two independent parameters can be selected
each phase. It is hard to predict the accurate thermal conductivity
and the cross-relationship among these parameters can be written
due to the complicated structure of metal foam. A three structured
as:
model (tetrakaidecahedron) can be adopted to describe the effec-
rffiffiffiffiffiffiffiffiffiffiffi  tive thermal conductivity with Eqs. (23)–(29):
df 1e 1
¼ 1:18 ð16Þ pffiffiffi
dp 3p 1  eð1eÞ=0:04 2
keff ¼ ð23Þ
2ðM A þ MB þ MC þ M D Þ

2.2.3. Surface area density and heat transfer coefficient 4r

It is hard to theoretically or numerically predict the heat trans- MA ¼ ð24Þ
ð2e2 þ prð1  eÞÞkm þ ð4  2e2  prð1  eÞÞkp
fer within the metal foam due to its complicated structure. Nor-
1226 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232

Fig. 5. Liquid fraction distribution for pure PCM (–): (a) t = 1500 s; (b) t = 3000 s; (c) t = 4500 s; (d) t = 6000 s.

ðe  2rÞ2 0:36 !
MB ¼ ð25Þ ktd ¼ q cp df j V j ð32Þ
ðe  2rÞe2 km þ ð2e  4r  ðe  2rÞe2 Þkp 1e p

pffiffiffi
ð 2  2eÞ2 2.3. Governing equations for heat transfer fluid
MC ¼ pffiffiffi pffiffiffi pffiffiffi ð26Þ
2pr2 ð1  2e 2Þkm þ 2ð 2  2e  pr2 ð1  2e 2ÞÞkp
The waste heat temperature is a key factor determining the
2e waste heat recovery feasibility, which can vary significantly. In or-
MD ¼ 2 ð27Þ
e km þ ð4  e2 Þkp der to enable the heat transfer and recovery, it is necessary that the
waste heat source temperature is higher than the heat sink tem-
sp
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffiffi pffiffiffi ffi
perature. The flue gas is considered as HTF in this numerical work.
2ð2  5=8e3 2  2eÞ
r¼ pffiffiffi ð28Þ For the HTF domain, Eqs. (2), (3), (4), (5), and (9) can be employed
pð3  4e 2  eÞ and the source terms in the governing equations are all equal to
zero. To consider the turbulent effects on the HTF, the standard
e ¼ 0:339 ð29Þ K–e model is implemented in the numerical model. The turbulent
Then the effective thermal conductivities of the PCM and the kinetic energy K and dissipation rate e can be computed from the
metal foam can be described as: following transport equations:

kfe ¼ keff jkm ¼0 ð30Þ   

@ðqKÞ ! lt
þ div ðqK V Þ ¼ div lþ grad K þ GK  qe ð33Þ
@t de
kse ¼ keff jkp ¼0 ð31Þ
  
The heat transfer enhancement due to the fluid mixing in por- @ðqeÞ ! lt e e2
þ div ðqe V Þ ¼ div lþ grad e þ C 1e ðGK Þ  C 2e q
ous media at the pore scale can be represented by the thermal dis- @t de K K
persion. The thermal dispersion model based on the stochastic ð34Þ
phenomena developed by Georgiadis and Catton [26] is adopted
@u
in this paper, which can be expressed by the following equation: where GK ¼ qu0i u0j @xj , C1e ¼ 1:44, C2e ¼ 1:92, de ¼ 1:3:
i
 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232 1227

Fig. 6. Temperature distribution for pure PCM (K): (a) t = 1500 s; (b) t = 3000 s; (c) t = 4500 s; (d) t = 6000 s.

3. Numerical model validation PCM starts to melt around t = 1200 s and finishes the melting pro-
cess around t = 4000 s. In the numerical work of Ref. [16], the PCM
One test case, which is the melting process of the PCM encapsu- was treated with a constant melting point, which led to a relatively
lated in rectangular metal foams, is used to evaluate the numerical flat melting curve in their predictions. In this study, the PCM melt-
model. In the experimental work of [16], the paraffin wax was used ing process occurs in a temperature range of 321–335 K, which ob-
as PCM and the heat transfer was found to be improved by metal tains a better agreement between the numerical predictions and
foam. The heat transfer analysis was performed with a 2D numer- the experimental data. The non-equilibrium thermal model can
ical model including the effect of natural convection. Same simula- be found to be more accurate to predict the heat transfer phenom-
tion domain, initial conditions and thermal physical properties are ena in porous media compared with the equilibrium thermal mod-
used to compare the results from the numerical model developed el due to its better agreement with the experimental data.
in this paper with the data in [16].
As shown in Fig. 2, the PCMs are encapsulated in a rectangular 4. Results and discussion
metal foam sample (10 PPI, 95% porosity and copper). The sample
dimensions are 200 mm and 25 mm in X-direction and Y-direction, The physical domain includes the volume filled with a metal
respectively. The simulation domain is divided into 200  50 matrix (copper). This matrix is saturated with the PCM RT 58,
uniform mesh grids. The initial temperature is 20 °C. The bottom and the physical parameters of the PCM are shown in Table 2.
of the sample is set with a fixed heat flux. The three other sides The HTF flows through the inner tube at a temperature higher than
are in the room temperature environment and the heat loss into the PCM melting temperature, while the outer shell surface is kept
the surrounding is considered. The simulation is performed using adiabatic.
the CFD software package, Ansys Fluent. The UDF (User Define
Function) is adopted to improve present porous media model in 4.1. Two-dimensional simulation
Fluent to account for the effect of variable coefficients in the ther-
mal equations. In the Rosler and Bruggemann’s work [12], there was no differ-
Fig. 3 shows a comparison between the numerical predictions ence in the melting process along the Z axis, which was based on a
and the data from Ref. [16]. The local position of the measuring high HTF mass flow rate and the relatively small heat transfer sur-
point is 8 mm above the bottom side. The results show that the face of tube. A two-dimensional computational domain is adopted
1228 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232

Fig. 7. Velocity field of the numerical simulation for pure PCM: (a) t = 1500 s; (b) t = 6000 s.

to predict the melting process in this work. The computational do- is implemented in the numerical simulation, which assures a Cou-
main employs an unstructured mesh and the calculations were pri- rant number of Co = 0.5.
marily carried out with three different grid densities: 20,000, Figs. 5 and 6 show the liquid fraction and temperature distribu-
30,000, and 50,000 elements to perform a grid independency test. tions of pure PCM at different time step, respectively. In Fig. 5, the
The total melting time for the two fine grids are almost the same region of liquid fraction between 0 (blue1) and 1 (red) indicates the
and 3.9% larger than that for the coarse grid. For the finest grids, positions of the solid–liquid interface region. The solid–liquid
the solution is time consuming to be used practically. The final cal- interface moves upwards due to the effect of natural convection
culations are performed using 30,000 elements. The same proce- in the pure liquid PCM. The melting process is furthest advanced
dure was carried out for the three-dimensional simulation and in the upper part of the simulation domain. After the area near
the final mesh number is 0.2 million. The temperature of the inner the top of shell is occupied by liquid PCM, the rest of solid PCM
tube is set at 350 K and the outer shell is set as an adiabatic bound- in the upper part starts to melt rapidly. The thermal energy can
ary condition. The simulation of melting process starts with a uni- be hardly transferred by heat conduction because the temperature
form temperature distribution of 320 K for the PCM and metal gradient is large in the solid–liquid interface region, as shown in
foam. The physical model adopted in the numerical work is shown
in Fig. 4. A single precision solver is adopted in the numerical mod-
el to solve the second order implicit scheme. PISO algorithm is 1
For interpretation of color in Figs. 5 and 6, the reader is referred to the web
adopted for the pressure–velocity coupling. A variable time step version of this article.
 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232 1229

Fig. 12. Cross-section view of liquid fraction distribution in PCM domain (X = 0),
t = 1800 s.

Fig. 6. After 3000 s, more than half of the solid PCM in the domain
becomes liquid PCM and the temperature keeps increasing in the
Fig. 8. Liquid fraction distribution for PCM with metal foam (–), t = 600 s. upper part of simulation domain. Then, the thermal energy is
transferred by conduction downwards to the solid PCM in the low-
er part of the storage unit. The mushy region broadens and the so-
lid–liquid interface reaches the lower end of the simulation
domain and only little PCM is left at the bottom of the storage unit
after 6000 s. The similar melting process in the LHTES unit without
metal foam was experimentally observed in [12]. Fig. 7a and b
shows the velocity field and vectors induced by the natural convec-
tion in liquid PCM at t = 1500 s and t = 6000 s. The smaller vortexes
near the top of tube at higher velocities due to the buoyancy effect
can be observed at the beginning of melting process. The bigger
vortexes at two sides of tube at lower velocities can be found as
the melting process reaches the lower part of the storage unit.
Figs. 8 and 9 show the liquid fraction and temperature distribu-
tions of the PCM embedded in metal foam at 600 s, respectively.
Compared with the pure PCM results, the melting process is signif-
icantly improved due to the application of metal foam. The effect of
buoyancy is not as obvious as that in pure PCM due to the presence
of metal foam. According to the numerical prediction with the non-
equilibrium thermal model, the whole domain is almost occupied
Fig. 9. Temperature distribution for PCM with metal foam (K), t = 600 s. by the liquid PCM after 800 s, which is enhanced by more than se-
ven times compared to that of the pure PCM. The melting process
of the PCM embedded in metal foam is conduction dominant and
no large temperature gradient exists across the phase front, which
leads to a broader mushy zone. The implementation of the metal
foam can be proved to enhance the thermal characteristics of the
shell-and-tube type LHTES unit.

4.2. Three-dimensional simulation

Fig. 10. Physical model of LHTES unit. For the energy saving in real industrial applications, the influ-
ence by the HTF must be considered in the storage unit’s design
and analysis. The HTF flows into the tube at inlet temperature Tin

Fig. 11. Cross-section view of temperature distribution (X = 0), t = 1800 s.

1230 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232

Fig. 13. Temperature distribution: (a) t = 1600 s; (b) t = 1800 s; (c) t = 2000 s; (d) t = 2200 s.

Fig. 14. Liquid fraction distribution: (a) t = 1600 s; (b) t = 1800 s; (c) t = 2000 s; (d) t = 2200 s.

and inlet velocity Vin. And the average temperature of the HTF de- PCM domain, the temperature near the inner tube wall is higher
clines along the flow direction as the heat is transferred to the PCM than that near the outer shell, and the average temperature of
filled in the container. Thus, the three dimensional simulation be- PCM decreases along the Z axis. It is shown that the high temper-
comes necessary to analyze the real heat transfer phenomena in ature region boundary moves from inlet region (right) to the outlet
the LHTES unit. The outer shell is adiabatic and the gravity is in region (left), which means the PCM near the HTF inlet melts first.
the negative Y direction, as shown in Fig. 10. As the time goes on, the melting process moves towards to the
Fig. 11 shows the temperature distribution on the cross section HTF outlet. Fig. 12 shows the liquid fraction distribution on the
of X = 0 at t = 1800 s. The HTF temperature decreases along the Z cross section of X = 0 at t = 1800 s. The pure liquid region boundary
axis because the heat energy is transferred to the PCM. For the move from the inlet region to the outlet region as time goes. At
 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232 1231

Fig. 13 shows the temperature distribution vs. time. An iso-sur-

face (T = 335 K) moves from the inlet to the outlet as the melting
time goes from 1600 s to 2200 s. The right side of the iso-surface
is pure liquid and the left side is where melting process is proceed-
ing. After 2200 s melting time, the temperature change of the PCM
trends to be very slow. Fig. 14 shows the contour of the liquid frac-
tion from t = 1600 s to t = 2200 s. The PCM near the tube wall at
high temperature melts first. In the liquid PCM, the slight natural
convection can be observed due to the buoyancy effect, which
can accelerate the melting process.
Figs. 15 and 16 show a comparison of different metal foam
parameters and HTF inlet conditions. The melting rate increases
with the decrease of the metal foam porosity due to a higher effec-
tive thermal conductivity. A low porosity of metal foam means a
fast melting process, which reduces the temperature gradient in
the simulation domain. The melting rate decreases with a decrease
in the HTF inlet temperature or velocity. This is because the low
Fig. 15. Total energy of PCM domain (sensible and latent heat) in the LHTES unit vs.
values of the inlet temperature and velocity decrease the heat
time. transferred from the HTF to the PCM. Accordingly, the total energy
of PCM deceases for the low HTF inlet temperature and velocity.
Fig. 17 shows a comparison of different pore sizes (PPI, pore per
inch) to analyze its effect on the PCM phase change process. A
higher value of PPI means a smaller pore size of metal foam, which
leads to the higher values of surface area per volume and flow
resistance. As the pore size decreases from 10 PPI to 30 PPI, the
heat transfer area between the liquid PCM and the metal foam in-
creases, this speeds the melting process. As the pore size keeps
decreasing (from 30 PPI to 60 PPI), the natural convection in PCM
can be neglected due to a relatively high flow resistance and the
melting process trends to slow down.

5. Conclusions

In this paper, a numerical model is established to analyze the

phase change process in the LHTES, which includes the convective
heat transfer of the HTF and the melting process of the PCM in por-
ous media. The heat transfer enhancement using metal foam is
Fig. 16. Liquid fraction of PCM domain vs. time. investigated in the shell-and-tube type LHTES unit. Compared with
the results of the pure PCM, the phase change heat transfer can be
enhanced by more than seven times. The parametric study shows
that the structure characteristics of the metal foam and the HTF in-
let conditions have significant influences on the storage unit per-
formance. The numerical model can be extended to investigate
different type of PCM containers for LHTES in the future work.

Acknowledgements

This work was supported by the National Natural Science Foun-

dation of China through Grant no. 50925624, the National Basic Re-
search Program of China (973 Program) through Grant no.
2012CB720404, and the Science and Technology Commission of
Shanghai Municipality through Grant nos. 11XD1403100 and
12JC1405100.

References
Fig. 17. Comparison of different pore sizes (Tin = 473 K and Vin = 10 m/s).
[1] Arce P, Medrano M, Gil A, Oro E, Cabeza LF. Overview of thermal energy storage
(TES) potential energy savings and climate change mitigation in Spain and
Europe. Appl Energy 2011;88:2764–74.
[2] Reichl J, Kollmann A. The baseline in bottom-up energy efficiency and saving
t = 2200 s, the whole PCM region almost finishes the melting pro- calculations – a concept for its formalisation and a discussion of relevant
cess and the liquid PCM occupies the whole volume. It also can options. Appl Energy 2011;88:422–31.
be found that a wide mush zone exists in the domain due to the [3] Perez G, Rincon L, Vila A, Gonzalez JM, Cabeza LF. Green vertical systems for
buildings as passive systems for energy savings. Appl Energy 2011;88:4854–9.
high effective thermal conductivity of metal foam and the small [4] Yau YH, Rismanchi B. A review on cool thermal storage technologies and
temperature gradient throughout the simulation domain. operating strategies. Renew Sust Energy Rev 2012;16:787–97.
1232 Z. Liu et al. / Applied Energy 112 (2013) 1222–1232

[5] Rady M. Thermal performance of packed bed thermal energy storage units [16] Tian Y, Zhao CY. A numerical investigation of heat transfer in phase change
using multiple granular phase change composites. Appl Energy materials (PCMs) embedded in porous metals. Energy 2011;36:5539–46.
2009;86:2704–20. [17] Li M, Wu Z, Kao H. Study on preparation, structure and thermal energy storage
[6] Agyenim F, Hewitt N, Eames P, Smyth M. A review of materials, heat transfer property of capric-palmitic acid/attapulgite composite phase change
and phase change problem formulation for latent heat thermal energy storage materials. Appl Energy 2011;88:3125–32.
systems (LHTESS). Renew Sust Energy Rev 2010;14:615–28. [18] Faghri A, Zhang Y. Transport phenomena in multiphase system. San
[7] Zhou D, Zhao CY, Tian Y. Review on thermal energy storage with phase change Diego: Academic Press; 2006.
materials (PCMs) in building applications. Appl Energy 2012;92:593–605. [19] Voller VR, Prakash C. A fixed grid numerical modelling methodology for
[8] Zhang Z, Zhang N, Peng J, Fang X, Gao X, Fang Y. Preparation and thermal convection-diffusion mushy region phase-change problems. Int J Heat Mass
energy storage properties of paraffin/expanded graphite composite phase Transfer 1987;30:1709–19.
change material. Appl Energy 2012;91:426–31. [20] Shmueli H, Ziskind G, Letan R. Melting in a vertical cylindrical tube: numerical
[9] Lacroix M. Numerical simulation of a shell-and-tube latent heat thermal investigation and comparison with experiments. Int J Heat Mass Transfer
energy storage unit. Sol Energy 1993;50:357–67. 2010;53:4082–91.
[10] Ismail KAR, Alves CLF, Modesto MS. Numerical and experimental study on the [21] Calmidi VV, Mahajan RL. Forced convection in high porosity metal foams. J
solidification of PCM around a vertical axially finned isothermal cylinder. Appl Heat Transfer 2000;122:557–65.
Therm Eng 2001;21:53–77. [22] Churchill SW, Chu HHS. Correlating equations for laminar and turbulent free
[11] Zivkovic B, Fujii I. Analysis of isothermal phase change of phase change convection from a horizontal cylinder. Int J Heat Mass Transfer
material within rectangular and cylindrical containers. Sol Energy 1975;18:1049–53.
2001;70:51–61. [23] Zhukauskas AA. Heat transfer from tubes in cross flow. New York: Academic
[12] Rosler F, Bruggemann D. Shell-and-tube type latent heat thermal energy Press; 1972.
storage: numerical analysis and comparison with experiments. Heat Mass [24] Wakao N, Kaguei S. Heat and mass transfer in packed beds. New York: Gordon
Transfer 2011;47:1027–33. & Breach; 1982.
[13] Tao YB, He YL. Numerical study on thermal energy storage performance of [25] Dullien FAL. Porous media: fluid transport and pore structure. New
phase change material under non-steady-state inlet boundary. Appl Energy York: Academic Press Inc.; 1979.
2011;88:4172–9. [26] Georgiadis JG, Catton I. Dispersion in cellular thermal convection in porous
[14] Chiu JNW, Martin V. Submerged finned heat exchanger latent heat storage layers. Int J Heat Mass Transfer 1988;31:1081–91.
design and its experimental verification. Appl Energy 2012;93:507–16.
[15] Tay NHS, Belusko M, Bruno F. Experimental investigation of tubes in a phase
change thermal energy storage system. Appl Energy 2012;90:288–97.