Applied Energy 239 (2019) 142–156

Contents lists available at ScienceDirect

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

Role of porous metal foam on the heat transfer enhancement for a thermal T
energy storage tube
⁎
Xiaohu Yanga,b, Jiabang Yua, Zengxu Guoa, Liwen Jina, Ya-Ling Heb,
a
Institute of the Building Environment & Sustainability Technology, School of Human Settlements and Civil Engineering, Xi’an Jiaotong University, Xi’an 710049, China
b
Key Laboratory of Thermal Fluid Science and Engineering of MOE, School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an 710049, China

H I GH L IG H T S

• Metal foam should be arranged in both heat transfer fluid and phase change material;
• Metal foam in phase change materials improves uniformity of temperature field;
• Full melting time was maximally reduced by 88.548% compared with smooth tube;
• j-factor was increased by 5186.91% if HTF and PCM were inserted with metal foam.

A R T I C LE I N FO A B S T R A C T

Keywords: Thermal energy storage technology has attracted extensive attentions due to its remarkable energy-saving
Phase change materials benefits. However, the low thermal conductivity of phase change materials seriously limits the energy storage
Metal foam efficiency, which put forward more stringent requirements for heat transfer enhancement. In this study, a two-
Thermal energy dimensional axisymmetric simulation model with natural convection was established for the shell-and-tube
Porosity
thermal energy storage unit. Open-cell metal foam with a porosity of 0.94 and pore density of 15 pore per inch
Numerical simulation
was employed to be arranged in either heat transfer fluid or phase change materials domains. The effects of the
metal foam location and the metal foam porosity on the heat storage performance were studied. The numerical
method was verified by experimental measurement, achieving good agreement. Results demonstrated that metal
foam can significantly enhance heat transfer due mainly to the reduction of thermal resistance in heat transfer
fluid. The case that both domains for heat transfer fluid and phase change materials were embedded in porous
media can provide the best heat transfer enhancement. Compared with smooth tube without metal foam, the full
melting time for this case was reduced by 88.548%; meanwhile, temperature response rate, heat flux and j-factor
was increased by 834.27%, 774.90%, 5186.91% respectively. Besides, embedding metal foam into phase change
materials can improve the temperature uniformity of phase change materials.

1. Introduction ones. One noticeable reason is the fact that waste heat is directly re-
leased into surrounding environment through waste gas, solid reject or
These days, world energy consumption has been growing at an waste water. About 12.2% of global energy consumption is rejected by
alarming rate. According to the International Energy Agency, fossil fuel waste heat [2,3]. It is therefore necessary to take measures to improve
energy still accounts for 81% of global energy demand in 2017 [1]. the waste heat recovery rate, being beneficial to both energy con-
Except for the proportion of fossil fuel energy in Organization for servation and emission reduction. It has been proved that TES tech-
Economic Co-operation and Development (OECD) countries, which is nology is capable of collecting low-quality energy (such as industrial
lower than the world average value, the data of developing countries is waste heat), which can be directly used in industrial process, as well as
much higher than this ratio [1]. This has caused concerns about the to produce hot water and central heating for buildings. And it’s appli-
depletion of energy resources, the destruction of the ozone layer, global cation fields have expanded from the production of domestic hot water
warming and environmental degradation. At present, industrial energy to heating [4,5], refrigeration and air conditioning [6], seawater de-
efficiency in developing countries is much lower than that in developed salination [5,6], industrial heating [6,7], thermal power generation [6],

⁎
Corresponding author.
E-mail address: [email protected] (Y.-L. He).

https://doi.org/10.1016/j.apenergy.2019.01.075
Received 30 October 2018; Received in revised form 29 December 2018; Accepted 13 January 2019
Available online 01 February 2019
0306-2619/ © 2019 Elsevier Ltd. All rights reserved.
X. Yang et al. Applied Energy 239 (2019) 142–156

Nomenclature RR¯ the integral average temperature response rate (K·s−1)

r the inner radius for the copper tube (mm)
Abbreviation S source term
T temperature (K)
FVM finite volume method t time (s)
HTF heat transfer fluid
LHS latent heat storage Greek symbols
PCM phase change material
PPI pore per inch α cross-sectional area ratio of node to solid ligament
TES thermal energy storage αsf specific area (m−1)
UDF user defined functions β thermal expansion coefficient (K−1)
OCED organization for economic co-operation and development δ numerical constant
ε porosity
Symbols ρ density (kg·m−3)
μ dynamic viscosity (N·s·m−2)
Am numerical coefficient for damping velocity σ liquid fraction in the porous medium
CE inertia coefficient (m−1) χ flow tortuosity
cd drag coefficient ω pore density
cp specific heat capacity (J·kg−1·K−1) ϕ control volume
dp pore diameter (m) ε1 the porosity of metal foam inserted in the HTF domain
e thickness ratio of node to solid ligament ε2 the porosity of metal foam inserted in the PCM domain
f friction factor ϕ̄ the integral average value
fm melting fraction
G shape function for metallic ligaments Subscript
h the average heat transfer coefficient (W·m−2·K−1)
hsf interstitial heat transfer coefficient (W·m−2·K−1) e effective parameter
j heat transfer factor f interstitial medium
k thermal conductivity (W·m−1·K−1) full full solidification rate
kε effective thermal conductivity (W·m−1·K−1) HTF heat transfer fluid
ktd thermal dispersion coefficient (m2·s−1) i initial state
L latent heat of PCM (J·kg−1) lig metallic ligament
Nu Nusselt number PCM phase change material
P pressure (Pa) sf interstitial
ΔP pressure drop s skeleton
Pr Prandtl number td thermal dispersion
q̄ the integral-mean heat flux (W·m−2) tube heat transfer tube
R the inner radius for the TES tube (mm) w wall
Re Reynolds number

etc. was also increased by 3.7–5.5 times. Compared with fins, expanded
Current TES technologies can be classified into three types ac- graphite, carbon fiber and other materials, metal foam exhibits superior
cording to their methods: sensible heat storage, latent heat storage heat transfer performance. Porous metal foam can compensate for the
(LHS), and chemical energy storage. Compared with sensible heat sto- low thermal conductivity of PCM due to its high thermal conductivity,
rage and chemical energy storage, LHS has the following advantages: abundant specific surface area and low density. The use of metal foam
(1) high heat storage density; (2) during the phase change process, the as a heat transfer medium for phase change has been widely paid at-
PCM temperature is kept constant; (3) PCM has a wide range of tem- tention to in these days. Meng et al. [10] experimentally investigated
perature for options and its price is relatively low; (4) LHS technology is the performances of the tube-in-tank LHS unit filled with paraffin-
also relatively mature at present; The energy storage combined with copper foam (composite PCM) during the charging and discharging
LHS technology and PCM has significant energy saving benefits, but the processes. The results showed that the LHS unit with composite PCM
thermal conductivity of widely-used PCMs is relatively low (paraffin is demonstrated good heat transfer performance. Zhang et al. [11] ex-
about 0.2 W·m−1·K−1), which limits the TES efficiency of the industrial perimentally and numerically investigated the melting heat transfer
waste heat system. In order to strengthen the feasibility of PCM in TES characteristics of the paraffin-copper foam composite PCM. Compared
field, researchers mainly proceed from the improvement of the thermal with pure paraffin, composite PCM showed a better heat transfer per-
conductivity of PCM, and strive to maximize the heat transfer of the formance. Yang et al. [12] experimentally studied the temperature
heat exchanger. variations and the solid–liquid interface evolution for two samples,
At present, the way to increase thermal conductivity of PCM is including that of pure paraffin, paraffin–copper foam composite PCM. It
mainly to fill thermal spreaders into PCMs. Yang et al. [8] conducted can be found that under the same operating conditions, the complete
numerical investigations on the melting process of a shell-and-tube melting time of PCM in the composite was 1/3 shorter than that of pure
LHTES unit with annular fins, and found that the full melting time could paraffin. Moreover, the effect of the HTF temperature on the charging
be maximally reduced by 65% by inserting annular fins into PCM. Jiang process was more significant than that of the HTF flow rate. Son et al.
et al. [9] prepared composite PCM by using paraffin and carbon fibers. [13] added expanded aluminum foam to C5H12O2, and found that the
Depending on the density of paraffin-carbon fibers composite PCM, it thermal conductivity of composite PCM increased by 30 times and 60
was found that the in-plane thermal conductivity was 18–57 times times, respectively, when aluminum foam with a mass fraction of 7%
higher than pure paraffin, while the out-of-plane thermal conductivity and 14% was added. Siahpush et al. [14] experimentally investigated

143
X. Yang et al. Applied Energy 239 (2019) 142–156

the performances of composite PCM by adding copper foam to 99% full melting time. Yao et al. [23] captured the phase and temperature
pure eicosane. The thermal conductivity of PCM increased from fields of paraffin-foam and pure paraffin samples to visualize the de-
0.423 W·m−1·K−1 to 3.06 W·m−1·K−1 due to the presence of copper tailed melting phenomena. Xiao et al. [24] prepared paraffin/nickel
foam, and the TES time was decreased from 500 min to 250 min. Chen foam and paraffin/copper foam composite PCM by vacuum impregna-
et al. [15] explored the influence of structural parameters of metal foam tion method. The thermal conductivities of the composite PCMs were
on heat transfer and flow characteristics of staggered tube bundles measured and the thermal behaviors were analyzed. Abishek et al. [25]
through experimental research. The results demonstrated that with the performed a pore-scale computational analysis to characterize the
increase of porosity of metal foam, Nusselt number of heat transfer tube performance of a composite PCM with a porosity of 0.94. The simula-
bundles increased due to the enhanced permeability, and the heat tions were carried out using open-cell foam by employing the enthalpy-
convection dominated the heat transfer. At low Reynolds number, the porosity formulation for modeling phase change during charging pro-
higher pore density metal foam had lower Nusselt number and poorer cess. Xu et al. [26] numerically investigated the melting performance of
insulation effect. Yang et al. [16] carried out numerical and experi- PCM in a horizontal concentric–tube TES unit with consideration of
mental studies on solidification of pure distilled water in open-cell natural convection. Performances of different porous configurations
metal foam, and obtained good agreement by experimental measure- were compared to optimize the location of porous insert. Wang et al.
ment to verify the analytical model and numerical method. It can be [27] embedded paraffin in copper foam to form composite PCM. A
found that the phase interface was flat in the macroscopic, but irregular platform was constructed to test the thermophysical properties of pure
under the microscope. In addition, the temperature difference between paraffin and composite PCM, and the paraffin and composite PCM were
the saturated PCM and metallic ligament was negligible, and local processed using numerical simulations to analyze their phase change
thermal equilibrium was suitable for low Ste numbers. Natural con- process. Feng et al. [28] numerically investigated the melting process of
vection was found to be negligible at pore-scale during discharging PCM infiltrated in a finned metal foam by using two approaches: (a)
process. Xiao et al. [17] immersed the paraffin in copper foam with pore-scale and (b) volume-averaged numerical simulations. The
different porosity and pore size under vacuum conditions, and mea- thermal performance of the finned metal foam was compared with
sured the effective thermal conductivity of composite PCMs. The results conventional plate-fin and metal foam structures, demonstrating its
showed that the thermal conductivity of the composite PCM was greatly superiority as thermal conductivity enhancer of PCM. Jin et al. [29]
improved, the thermal conductivities of the paraffin-copper foam performed a visualized study at the pore-scale to explore the pore size
composite PCMs fabricated by the copper foams with the porosities of effects on melting heat transfer of a paraffin wax saturated in a copper
96.95%, 92.31%, 88.89% and pore size of 25 PPI were about thirteen, foam. Mahdi et al. [30] used a compound porous-foam/nanoparticles
thirty-one, forty-four times larger than that of pure paraffin, respec- enhancement technique to significantly improve melting of PCM in a
tively. Lafdi et al. [18] added aluminum foam into paraffin and found triplex-tube heat exchanger. A numerical model that took into account
that the addition of higher porosity aluminum foam allowed PCM to the non-Darcy effects of porous foam and Brownian motion of nano-
reach a steady-state temperature more quickly. Similarly, for the foam particles was formulated and validated with previous related experi-
with larger pore size achieved the steady state faster than the smaller mental studies. Tao et al. [31] numerically investigated the TES per-
one. This was due mainly to the strong natural convection in both the formance of metal foams/paraffin composite PCM by Lattice Boltzmann
higher porosity and the larger pore size. However, aluminum foam with method. The effects of metal foam PPI and porosity on PCM melting
lower porosity would reduce the heater temperature. Yang et al. [19] rate, heat storage capacity and heat storage density were investigated.
experimentally investigated the solidification behavior in open-cell Metal foam can enhance phase change heat transfer mainly by uti-
metal foam with gradient morphology. Open-cell foam was cut into lizing its abundant specific surface area and highly-conducting skele-
pieces and stacked, and layered metal foam with gradient morpholo- tons. Generally speaking, porous media are filled in the PCM domain.
gical characteristics was compared and analyzed under different mor- The enhancement of the overall TES performance is limited. In addi-
phological properties including porosity, pore density and materials for tion, porous media can also be inserted into the HTF domain to help
porous matrix. The results showed that the stacked foams with mor- extract more energy from HTF, which is to improve the driving force for
phological gradients enhanced heat transfer compared to the single- the phase change from the heat source. Compared with waste water, the
layer foams with fixed morphology, which would more effectively re- thermal resistance in the side of exhaust gas is higher. When the ex-
duce the total solidification time. Zhao et al. [20] experimentally and haust gas is used as HTF, in order to efficiently obtain waste heat from
numerically investigated the heat transfer enhancement by inserting the exhaust gas, the addition of metal foam into the HTF domain has
metal foam in paraffin. Compared with the pure PCM sample, the in- promising potentials with respect to practical applications. This can
fluence of metal foam on melting phase change heat transfer was more bring huge economic benefits for industrial waste heat recovery and
significant, especially in the solid region of PCM, which can enhance provide a significant contribution to energy conservation and emission
the heat transfer rate by 5–20 times. During the melting process (two- reduction. However, there are few studies on the influence of the in-
phase region) and the pure liquid region, the addition of metal foam sertion location and porosity of metal foam on TES performance. In
increased the overall heat transfer rate by 3–10 times. In addition, it can order to explore the value and impact of metal foam on waste heat
be found that smaller the porosity and the pore density, the better the recovery of exhaust gas, this paper proposed a new TES unit. To this
heat transfer performance. Zhu et al. [21] numerically investigated the end, four kinds of TES tubes filled with metal foam separately in HTF,
thermal performance of composite PCM formed by 90% porosity alu- PCM and both sides were designed and studied. Numerical models of a
minum foam and paraffin. By changing the pore density of metal foam reduced two-dimensional computational domain were established to
and modifying the shape of the cold wall, it can be observed that in- study the influence of the insertion location and porosity of metal foam
creasing pore density or changing the shape of the cold wall can im- on the melting phase change process. Open-cell metal foam with the
prove the thermal response of composite PCM. In addition, an optimi- porosity of 0.94 and pore density of 15 PPI (pore per inch) was em-
zation method was proposed to improve the thermal energy storage ployed as heat transfer spreader. The numerical results were compared
efficiency of composite PCM by 83.32% compared with pure paraffin. with the experimental results, and the results were analyzed based on
Yang et al. [22] predicted melting heat transfer in paraffin/metal foam the validated simulations. Melting fraction, temperature response, heat
composite PCM by volume-averaging method, and compared with di- transfer rate and j-factor were analyzed and evaluated regarding the
rect numerical simulation. Direct numerical simulation results showed overall thermal performance. On this basis, it can provide effective
that there was a significant temperature difference between skeleton guidance and reference for the design of new and efficient TES unit in
surface and PCM. Natural convection can significantly affect the the industrial sector.
melting solid-liquid phase interface, temperature distribution and the

144
X. Yang et al. Applied Energy 239 (2019) 142–156

2. Numerical simulation shell-and-tube TES system, PCM was generally arranged in the shell
side, and HTF flowed in the tube side, as shown in Fig. 1(b). Injecting
2.1. TES tube design HTF with high temperature at the top, heat transfer occurred between
the copper tube wall and the saturated PCM, with thermal energy
The physical model for a vertical shell-and-tube TES tank was stored in PCM. Open-cell metal foam can be inserted into the HTF or the
shown in Fig. 1(a). Due to the symmetrical distribution of the tube PCM domain. The effect of porous media on heat transfer enhancement
bundle, one of the TES tubes was usually selected as the object for was investigated by employing copper foam with the porosity of 0.94
numerical simulation (see Fig. 1(b)). In practical application of the and pore density of 15 PPI. In order to reveal the influence of metal

Fig. 1. (a) TES tank; (b) physical model; (c) four designed TES tube.

145
X. Yang et al. Applied Energy 239 (2019) 142–156

→
foam location on TES performance, four models are designed as shown
in Fig. 1(c): HTF and PCM domain were not filled with metal foam
where Forchheimer-Darcy equation ( μHTF
K
+
ρHTF CE
K )
| 〈U 〉 | 〈u〉 was em-
ployed to describe the additional flow resistance that metal foam pro-
(labeled as 1.00–1.00), namely model A; HTF domain was not filled duced.
with metal foam but PCM domain was (labeled as 1.00–0.94), namely Energy equation for HTF:
model B; HTF rather than PCM domain was filled with metal foam
(labeled as 0.94–1.00), namely model C; both domains were filled with ερHTF cp,HTF
∂〈THTF 〉 →
+ ρHTF cp,HTF 〈U 〉·∇〈THTF〉
metal foam (labeled as 0.94–0.94), namely model D. ∂t
= ∇ ·((ke,HTF + ktd ) ∇〈THTF〉 ) + Ssf (6)
2.2. Computational domain and boundary conditions
Ssf = −hsf asf ( 〈THTF〉 − 〈Ts〉 ) (7)
Given the circularly symmetric nature of the physical model, 2D Energy equation for metal foam:
computational domain was built as shown in Fig. 2. The total length for
the TES unit (L) was 400 mm, and the inner radius for the TES tube (R) ∂〈Ts 〉
(1 − ε ) ρs cs = ∇ ·(ke, s ∇〈THTF〉 ) − Ssf
and the copper tube (r) were 50 mm and 10 mm respectively. The ∂t (8)
copper tube had a thickness of 2 mm. Nine representative points were where hsf was the interstitial heat transfer coefficient between the
selected in the PCM domain for exploration of temperature response. As porous metallic surface and the saturating PCM; here we took the
can be seen from Fig. 2, the five selected points along the radial di- commonly-used empirical model [32] to account for the interstitial heat
rection were of the same height, all located at the intermediate height transfer condition, expressed as,
of the PCM domain, which was 200 mm from the bottom of the TES
tube. The intervals between these five points were 7 mm, and P1 was 0.4 0.37
⎧ 0.76Re Pr kf /d, 0 < Re⩽40
⎪
5 mm away from the outer wall of the copper tube. The five selected hsf = 0.52Re0.5Pr 0.37 kf / d, 40 < Re⩽1000
points along the axial direction were the same distance from the outer ⎨
⎪ 0.26Re0.6Pr 0.37 kf / d, 1000 < Re⩽20000 (9)
wall of the copper tube, all of which were 19 mm, located in the middle ⎩
position along the radial direction of the PCM domain. The intervals where d denoted the average ligament diameter of open-cell foam, Re
between these five points were 70 mm, and P9 was 60 mm away from and Pr separately illustrated the Reynolds and Nusselt number. The
the top of the TES tube. specific area αsf for metal foam and can be determined by [33]
Boundary conditions were as follows: the left curve was symmetry
1.18ω
axis, inlet and outlet was set as constant velocity and outflow (a asf = 3π (1 − ε )
0.0224 (10)
boundary condition in ANSYS-Fluent), in respective; all of the other
curves were adiabatic. For the above four models, all the operation where dp and ω denoted the average pore size and pore density for the
parameters were the same: the HTF inlet velocity was kept constant at open-cell foam.
0.05 m/s and the inlet temperature was 343 K, corresponding to a For both cases with/without metal foam, conjugate heat transfer is
Reynolds number lower 2300 inside the HTF tube and annulus. considered at the interface between HTF and tube wall:
Therefore, the flow can be considered as laminar flow. Except for the
∂THTF ∂T
density of liquid PCM, all thermophysical properties of PCM remain THTF = Ttube, kHTF = k tube tube
∂n ∂n (11)
constant with temperature (see Table 1). The initial temperatures both
for PCM and copper tube wall were 293 K for the charging process.

2.3. Governing equations

2.3.1. Convective heat transfer in HTF

For the HTF, two situations may exist: one was smooth tube and the
other was filled with metal foam. For convenience, the governing
equations were written in the unified format with respect to convective
heat transfer in porous media. If smooth tube was considered, let por-
osity ε in the following equations equal to unit, thereby all the source
terms equaling to zero and the extrinsic average of a quantity over a
control volume 〈ϕ〉 equaling ϕ .
The convective heat transfer is governed by the following equations:
Continuity equation:
∂pHTF →
+ ∇ (ρHTF 〈U 〉 ) = 0
∂t (1)

Momentum equations:
ρHTF ∂〈u〉 ρ → ∂〈p〉 μ
+ HTF ( 〈U 〉·∇) 〈u〉 = − + HTF ∇2 〈u〉 + Su
ε ∂t ε2 ∂x ε (2)

ρHTF ∂〈v〉 ρ → ∂〈p〉 μ

+ HTF ( 〈U 〉·∇) 〈v〉 = − + HTF ∇2 〈v〉 + Sv
ε ∂t ε2 ∂y ε (3)

μ ρ CE → ⎞
Su = −⎛ HTF + HTF
⎜ | 〈U 〉 | 〈u〉⎟

⎝ K K ⎠ (4)

μ ρ CE → ⎞
Sv = −⎛ HTF + HTF
⎜ | 〈U 〉 | 〈v〉⎟

⎝ K K ⎠ (5) Fig. 2. Computational domain and boundary conditions.

146
X. Yang et al. Applied Energy 239 (2019) 142–156

Table 1 df ∂〈TPCM 〉 →
Thermophysical properties of PCM (RT54HC) and air.
ερp,PCM ⎛cp,PCM + L m ⎞
⎜ + ρp,PCM cp,PCM 〈U 〉·∇〈TPCM〉
⎟

⎝ dt ⎠ ∂t
Material Property State Value = ∇ ·((ke,PCM + ktd ) ∇〈TPCM〉 ) + Ssf (17)
Paraffin Melting temperature range (°C) 53–54 For copper foam:
Latent heat (kJ·kg−1) 200
Density (kg·m−3) Solid/liquid 850/800 ∂〈Ts 〉
Specific heat capacity (J·kg−1·K−1) 2000
(1 − ε ) ρs cp, s = ∇ ·(ke, s ∇〈TPCM〉 ) − Ssf
∂t (18)
Thermal conductivity (W·m−1·K−1) Solid/liquid 0.2/0.1
Dynamic viscosity (kg·m−1·s−1) 2.508 × 10−3 In the above governing equations, the symbol 〈 〉 represented the ex-
Thermal Expansion coefficient (K−1) 7.5 × 10−4 trinsic average of a quantity over a control volume, σ was the liquid
Air Density (kg·m−3) 1.225 fraction in the porous medium (σ = εfm ); fm was the melting fraction; L
Specific heat capacity (J·kg−1·K−1) 1006.43 was the latent heat for PCM; ρ , cp, μ, β , ke and ktd represented the
Thermal conductivity (W·m−1·K−1) 0.0242
density, specific heat, dynamic viscosity, thermal expansion, effective
Dynamic viscosity (kg·m−1·s−1) 1.789 × 10−5
thermal conductivity and thermal dispersion coefficient respectively;
ε , αsf , hsf , K and CE were the porosity, specific area, interstitial heat
Here, ρHTF, cpHTF and kHTF were the density, specific heat and thermal transfer coefficient, permeability and inertial coefficient for open-cell
conductivity respectively; T and t illustrated temperature and time; metal foam; S depicted source term; the subscripts HTF, tube, PCM, sf, s,
subscripts HTF and tube separately stood for HTF and heat transfer e separately denoted heat transfer fluid, copper tube, phase change
tube. material, interstitial, skeleton and effective medium.

2.4. Thermophysical properties for metal foam

2.3.2. Phase change in PCM
The PCM in solid phase was going to be melt with time elapsed and The thermo-physical properties of both the PCM and injected HTF
natural convection occurred in the melt phase. Two situations existed: are assumed to be constant (see Table 1), except the density of PCM in
one was pure PCM and the other was PCM embedded in open-cell metal liquid phase, favoring the assumption of Boussinesq free convection.
foam. Volume-averaged method was employed for describing the Besides, the volume expansion due to PCM melting was also neglected.
transient phase change heat transfer with natural convection. Volume-averaged method considered porous media to be equivalent
Continuity equation: fluids with effective thermophysical properties. Homogeneous and
isotropic assumption for porous media was also made for simplification.
∂pPCM →
+ ∇ (ρPCM 〈U 〉 ) = 0 Effective thermal conductivity ke [34,35]:
∂t (12)
ke (1 − ε ) kf
Momentum equations: = + ε
ks 1−e+( 3e
2α ) ⎡⎣3(1 − e) + 3
2
αe⎤
⎦
ks
(19)
ρPCM ∂〈u〉 ρ → ∂〈p〉 μ
+ PCM ( 〈U 〉·∇) 〈u〉 = − + PCM ∇2 〈u〉 Thermal dispersion conductivity ktd [36]:
σ ∂t σ2 ∂x σ
(1 − fm )2 0.36
− Am 〈u〉 + Su ktd = ρ cpf d u2 + v 2 + w 2
fm3 + δ (13) 1−ε f (20)
Permeability K and inertial coefficient CE [37]:
ρPCM ∂〈v〉 ρ → ∂〈p〉 μ
+ PCM ( 〈U 〉·∇) 〈v〉 = − + PCM ∇2 〈v〉 ε [1 − (1 − ε )1/3]
σ ∂t σ2 ∂y σ K= dp 2
108[(1 − ε )1/3 − (1 − ε )] (21)
+ ρPCM gβ (Tm1 − 〈TPCM〉 )
−1
(1 − fm )2 cd ε ⎛1.18 1 − ε 1 ⎞
− Am 〈v〉 + Sv CE = 0.095 ⎜ ⎟

fm3 + δ (14) 12 3(χ − 1) ⎝ 3π G ⎠ (22)

In the above-mentioned models for determining thermophysical
μ ρ CE → ⎞ properties, the subscripts e, f, s separately demonstrated the (effective)
Su = −⎛ PCM + PCM
⎜ | 〈U 〉 | 〈u〉 ⎟

⎝ K K ⎠ (15) thermal conductivity of the composite PCM, interstitial medium and

skeleton; e and α denoted the dimensionless thickness and cross-sec-
μ ρ CE → ⎞ tional area of a ligament (e = 0.3, α = 1.5); cd was the drag coefficient
Sv = −⎛ PCM + PCM
⎜ | 〈U 〉 | 〈v〉 ⎟
(cd = 1.56 [38]); χ and G represented the tortuosity and shape function
⎝ K K ⎠ (16)
for metal foam, with expressed by:
(1 − fm )2
where Am 〈u〉 denoted the source term for damping the velocity ε
fm3 + δ χ=
in solidified phase, δ and Am were two numerical constants, one was 1 − (1 − ε )1/3 (23)
sufficient small ∼10−4 and the other was relatively large ∼108. fm
G = 1 − e−(1 − ε )/0.04 (24)
represented the melting fraction, where fm = 0 (solid phase) made this
item to be infinity to damp the velocity in the momentum equations and
fm = 1 (liquid phase) caused this item to be zero. Due to the tempera- 2.5. Simulation setup
ture difference in fluid phase, free convection occurred. The convection
behavior was modelled by Boussinesq free convection, denoted by Finite volume method (FVM) was employed to solve the transient
ρf gβ (Tm1 − 〈Tf 〉 ) in momentum equation. heat transfer with phase change phenomena. 2D quadrilateral elements
There existed large difference in thermophysical properties between were generated by ANSYS-ICEM 14.5 to discretize the HTF and PCM
copper ligaments and PCM (e.g. paraffin for the present study). Thermal domains. The numerical solutions were obtained using the commercial
non-equilibrium state was used to model the heat transfer between the CFD software ANSYS-Fluent 14.5 [39]. The Pressure-Linked Equations
interstitial and metal skeletons. algorithms (SIMPLE) was used for pressure-velocity coupling for forced
Energy equation for PCM: convection in HTF and natural convection in the melt phase. The

147
X. Yang et al. Applied Energy 239 (2019) 142–156

PRESTO scheme was adopted for the pressure correction equation. If gradually increased from zero to one. For model A, PCM began to melt
porous medium was inserted into either HTF or PCM domain, two en- in 3, 480 s, and the full melting time approximately required 80, 020 s.
ergy equations for the metal foam and the interstitials (air or paraffin) Setting model A as the comparison basis, the duration before melting
were solved with the assumption of thermal non-equilibrium state. was reduced by 85.92%, 97.99%, and 94.83% for model B, C and D, in
Accounting for this, user defined functions (UDF) were developed, ex- respectively; and the full melting time was correspondingly reduced by
cept for the effective thermophysical properties for porous metal foam. 25.84%, 65.90%, and 88.54%. As the charging process continued, the
To improve the computation accuracy, second-order upwind melting fraction of the four models gradually increased, but the melting
scheme was employed to solve the governing equations. Since transient speed (curve slope) was arranged in descending order of D, C, B, and A.
problem was dealt with, the sensitivities of mesh quality and time step In practical applications, if it was necessary to quickly obtain heat from
were tested. Three sets of meshes with 136,586, 394,534 and 454,158 the exhaust gas (especially high-temperature exhaust gas), it was re-
elements were tested and numerical results showed a reasonably small commended to insert metal foam into both domains. In addition to
deviation less than 1.1%, indicating that the set of mesh with 394,534 reducing heat loss in time and improving energy utilization efficiency,
elements was fine enough for simulation. Furthermore, three time steps it can also quickly supply heat source to users who take heat.
with 0.3 s, 0.1 s and 0.03 s were tested, as well. It was demonstrated by By comparing these four models, it can be clearly found that the
the numerical simulation that the time step of 0.1 s was sufficient small participation of open-cell metal foam can effectively improve the heat
for calculation accuracy since the deviation was less than 0.87% be- transfer capability and shorten the full melting time. In the case that
tween the time step of 0.1 s and 0.03 s. For save computation cost, the both HTF and PCM domains were inserted with metal foam, the thermal
set of mesh with 394,534 elements and time step of 0.1 s were chosen resistance was minimized to provide optimum heat transfer enhance-
for the following numerical simulations. The solution was considered to ment among the four models. It was worth noting that although both
be converged, only if the residuals of continuity, momentum and energy model B and model C had the addition of metal foam on one side, the
equations were less than 10−6 at each time step. full melting time of model B was almost twice that of model C, which
indicated that embedding metal foam into the HTF (air) domain can
3. Model validation bring better heat transfer enhancement effect than that into the PCM
domain. This was due mainly to the fact that the surface heat transfer
The essential for the phase change in composite PCM was to deal coefficient (13.36 W·m−2·K−1 for model A) between HTF and the
with the heat transfer between the metallic skeleton and the interstitial copper tube was much smaller than the one (40.61 W·m−2·K−1 for
PCM. Accounting for this, volume-averaged method was employed for model A) between PCM (paraffin) and the copper tube. Thermal re-
treatment of porous media. The core thermal process in the present sistance mainly existed in the HTF side. Therefore, the thermal re-
study laid on the transient phase change heat transfer with a moving sistance can be greatly reduced by embedding metal foam into the HTF
boundary in PCM region. Thermal non-equilibrium state did exist be- domain, which can significantly reduce the full melting time of PCM.
tween the metallic ligaments and saturating PCMs, involving two- However, from the perspective of economy, the addition of low-por-
temperature model in energy equations. Therefore, if the transient osity metal foam into the air domain would greatly increase the flow
phase change heat transfer was solved and validated by comparing with resistance and thus increase the pump consumption, so it was necessary
experimental data, the overall thermal process for the present heat to consider a suitable case in practical applications.
transfer configuration can be validated. Given this, we employed the Besides that, to further explore the effect of the porosity of metal
experimental measurement [11] on temperatures for ligaments and foam on the overall melting heat transfer for the TES tube, the metal
saturating PCM in a cavity as benchmark to validate our numerical foam with an average porosity of 0.94 was simultaneously inserted into
model for transient phase change heat transfer in porous media. Zhang the HTF and the PCM domain, which was recorded as ε1 − ε2 . The re-
et al. [11] measured temperatures at metallic skeleton and the inter- lationship between ε1 and ε2 was shown in the following equation:
stitial PCM at pore room using two thermocouples at the center of the
paraffin-copper foam composite layer. Fig. 3 compared the temperature ε1 + ε2
= 0.94
variation at the two points calculated by the present numerical model 2 (25)
with the experimental data [11], achieving good agreement between
where ε1 was the porosity of metal foam inserted in the HTF domain, ε2
the two approaches (maximum deviation ∼1.02%). This indicated the
was the one in the PCM domain. The following five models were se-
applicability of the developed numerical models with the treatment on
lected for simulation: 0.90–0.98, 0.92–0.96, 0.94–0.94, 0.96–0.92,
effective thermophysical properties and the thermal non-equilibrium
0.98–0.90. For convenience, 0.90–0.98, 0.92–0.96, 0.96–0.92 and
assumptions.
0.98–0.90 were recorded as model D1, model D2, model D3 and model
D4, respectively. Fig. 4(b) showed the variation of PCM melting fraction
4. Results and discussion
360
4.1. Comparisons of melting fraction

To quantitatively summarize the effect of metal foam location on 340

the melting heat transfer, Fig. 4(a) compared the variation of melting
fraction over time for the four models. In the four models, the melting
T/K

tendency of PCM was basically the same. The variation of melting 320
fraction was mainly divided into two stages. In the first stage, since the Expt, ligament
initial temperature of PCM was lower than the phase change tem- Expt, paraffin
300
perature, the melting fraction for the four models was zero at the be- NS, ligament
ginning of the charging process, which meant that PCM remained in NS, paraffin
solid state. This stage was the sensible heat storage stage. As more and 280
0 2000 4000 6000 8000 10000
more thermal energy was transferred into the PCM domain, the PCM t/s
temperature rose rapidly. When the PCM temperature reached the
phase change temperature, PCM began to melt, and the TES process Fig. 3. Validation of numerical model by comparing temperature variations of
entered the second stage, at which latent heat storage and sensible heat numerical results with literature [11] for melting process in metal foam/par-
storage were simultaneously performed. And the PCM melting fraction affin composite PCM.

148
X. Yang et al. Applied Energy 239 (2019) 142–156

(a) 1.0 linear relationship between the porosity and the melting rate. Based on
the full melting time for model D, the full melting time for model D1, D2
0.8 and D4 was increased by 16.47%, 5.56% and 0.33% respectively; while
the one for model D3 was reduced by 1.64%. In the case of 0.96–0.92
0.6 (model D3), where the metal foam inserted in HTF domain had a por-
fm

Porosity: HTF - PCM osity of 0.96 and the one inserted in PCM domain had a porosity of
0.4 Model A (1.00 - 1.00) 0.92, the thermal resistance of the heat exchanger can be minimized to
Model B (1.00 - 0.94) provide optimum heat transfer enhancement. In fact, we numerically
0.2 Model C (0.94 - 1.00) studied nine models from 0.90–0.98, 0.91–0.97 to 0.98–0.90 for por-
Model D (0.94 - 0.94) osity comparison. It can be found from Fig. 4(d) that the 0.96–0.92
0.0 model had the shortest full melting time in the nine models. Therefore,
0 15000 30000 45000 60000 75000 90000
the case of 0.96–0.92 (model D3) was the optimal case to enhance heat
t/s
transfer. It should be pointed out that in the early heat storage stage,
(b) 1.0 although the upward trend of model D was larger than the rest models,
the difference among the five models was not obvious. However, in the
0.8 later heat storage stage, model D3 and model D4 increased faster and
Partial enlargement
exceeded model D, and finally model D3 was the first to complete the
0.6 Porosity: HTF - PCM phase change. This was due to the fact that PCM kept mainly in a solid
Model D1 (0.90 - 0.98)
fm

state during the early heat storage stage, and the heat transfer was
0.4 Model D2 (0.92 - 0.96) dominated by heat conduction. The five models all significantly re-
Model D (0.94 - 0.94) duced the thermal resistance in the HTF side, therefore the heat transfer
0.2 Model D3 (0.96 - 0.92) capability of the five models was not much different at the beginning
Model D4 (0.98 - 0.90) and the melting speed was also almost the same. However, in the later
0.0 heat storage stage, as more and more PCM was melted, the thermal
0 2200 4400 6600 8800 11000
resistance in the PCM side was increased, and the heat transfer en-
t/s
hancement effect of metal foam was more prominent. Since the porosity
(c) 1.0 of metal foam was inversely proportional to its thermal conductivity,
the smaller the porosity was, the greater the thermal conductivity was.
Model D3 and D4 had smaller porosity of metal foam inserted in the
0.9
PCM domain, therefore the melting speed of model D3 and D4 exceeded
Porosity: HTF - PCM model D in the later heat storage stage and the phase change was
0.8 Model D1 (0.90 - 0.98)
fm

completed earlier.
Model D2 (0.92 - 0.96)

0.7 Model D (0.94 - 0.94) 4.2. Propagation of solidification front

Model D3 (0.96 - 0.92)
Model D4 (0.98 - 0.90) Fig. 5 showed the comparisons of transient solid-liquid phase in-
0.6 terface distribution at 2000 s, 4000 s, 6000 s, 8000 s for four models
6000 7000 8000 9000 10000 11000
(model A, B, C, D) with different metal foam locations. In four models,
t/s the solid-liquid phase interface evolved towards right and down con-
(d) tinuously with time elapsed. As can be seen from the figure, the solid-
liquid phase interface of model D developed in the fastest rate among
the four models, and its melting fraction was much higher than the
others. This was because metal foam of model D significantly weakened
the thermal resistance of both the HTF side and the PCM side, max-
imizing heat transfer enhancement. Besides, through careful observa-
tion, it can be found that model A already had a solid-liquid phase
interface at 6000 s and 8000 s, but model B remained solidified at the
same time, and no solid-liquid phase interface appeared. This may be
caused by the following reasons: model B embedded metal foam in the
PCM side, which improved the effective thermal conductivity of PCM,
but at the same time, little heat was transferred from HTF to the interior
of PCM. Therefore, in the initial stage, this inhibited the melting of
PCM. In addition, it can be observed that there was a huge difference in
Fig. 4. Comparison of PCM melting fraction for different models: (a) four
the propagation speed of the solid-liquid phase interface between
models with different metal foam locations; (b) five models with different metal model B and model C. Although the two models both had the addition
foam porosity; (c) partial enlargement of figure (b); (d) comparison of full of metal foam, the solid-liquid phase interface of model C spread much
melting time for nine models with different metal foam porosities. faster than that of model B. At 8000 s, nearly one-third of PCM was
melted for model C, while model B had no solid-liquid phase interface.
It can be explained that during the charging process, the thermal re-
over time for the five models. From the figure, it can be observed that in
sistance was mainly existed between the HTF and the copper tube. It
the first heat storage stage, the time at which the five models began to
was notable that the melting fraction gradually decreased along the
melt was 90 s, 140 s, 180 s, 220 s, 300 s, respectively. In the second heat
axial direction, and the slant solid-liquid phase interface can be ob-
storage stage, the phase change end time for the five models was
served. First of all, Nusselt number was reduced along the axial direc-
10,680 s, 9680 s, 9170 s, 9020 s, and 9200 s, respectively. By comparing
tion of the copper tube, and the heat transfer capacity of the copper
the five models, it can be clearly found that the porosity of metal foam
tube was also weakened. When considering natural convection, the
had a certain influence on the melting heat transfer, but there was no
high temperature liquid PCM flowed upwardly, which enhanced the

149
X. Yang et al. Applied Energy 239 (2019) 142–156

Fig. 5. Schematic of PCM solid-liquid phase interface for four models with different metal foam locations at different time: (a) 2000 s; (b) 4000 s; (c) 6000 s; (d)
8000 s.

melting rate of the upper PCM and weakened the one of the lower PCM. 4.3. Temperature response
Therefore, under the combined effect of the two factors, the solid-liquid
phase interface was non-uniform distribution, with inclination along In order to clarify the effect of the metal foam location on the
the axial direction. temperature response for PCM, in Fig. 7, the temperature variation of
Fig. 6 compared the numerical results of transient solid-liquid phase the five selected points along the radial direction (P1, P2, P3, P4 and
interfaces at 2000 s, 4000 s, 6000 s, and 8000 s for five models (model P5) were plotted according to the numerical results for the four models.
D, D1, D2, D3, D4) embedded open-cell copper foams with different Fig. 8 demonstrated the temperature evolution history of the five se-
porosities. In five different models, a slant solid-liquid phase interface lected points along the axial direction (P6, P7, P3, P8, and P9) for the
can be observed. It can be seen from the figure that at each moment, the four models. It can be seen from Fig. 2 that the five selected points
lower the porosity of metal foam inserted in the PCM domain was, the along the radial direction were of the same height, all located at the
smaller the slant angle of the interface was. This was due mainly to the intermediate height of the PCM domain, which was 200 mm from the
fact that lower porosity resulted in higher thermal conductivity and bottom of the TES unit. The intervals between these five points were
higher flow resistance in the liquid phase. Higher thermal conductivity 7 mm, and P1 was 5 mm away from the outer wall of the copper tube.
can transport more heat from heating source to the unmelt region. And The five selected points along the axial direction were the same dis-
higher flow resistance suppressed local natural convection, favoring tance from the outer wall of the copper tube, all of which were 19 mm,
improving the dominated role of heat conduction. It was expected that located in the middle position along the radial direction of the PCM
little slant in interface can be found when metal foam with a certain low domain. The intervals between these five points were 70 mm, and P9
porosity was inserted in PCM. was 60 mm away from the top of the TES unit. As can be seen from
Figs. 7 and 8, for the four models, the temperature variation trends of
PCM were similar. The phase change process was mainly divided into
three stages: the first stage was the heat conduction stage through the

150
X. Yang et al. Applied Energy 239 (2019) 142–156

Fig. 6. Schematic of PCM solid-liquid phase interface for five models with different metal foam porosities at different time: (a) 2000 s; (b) 4000 s; (c) 6000 s; (d)
8000 s.

solid state, and the thermal energy of HTF was absorbed by the solid and high skeleton thermal conductivity, the transported heat was easily
PCM through sensible heat. Since the sensible heat of PCM was rela- dissipated by PCM. Therefore, the combination of the two factors re-
tively low, the PCM temperature rose quickly. The second stage was the sulted in uniformity distribution of temperature along the radial di-
phase change stage, in which PCM began to melt from the solid state to rection. By comparing Fig. 7(d) with Fig. 7(b), it can be observed that
the liquid one. Due to its high latent heat, the PCM temperature rose model D further inserted metal foam into the HTF domain relative to
slighter. The third stage was the heat conduction stage, and the liquid model B. The temperature difference between the five selected points
PCM was transferred to the sensible heat storage again, thereby leading was further enlarged, and the radial PCM temperature gradient in-
to a rapidly-rising temperature trend. After the above three stages, the creased. This was due mainly to the addition of metal foam, which can
temperature of the selected points in the PCM domain gradually re- take more heat away from HTF and significantly improve the surface
mained stable, approaching 343 K, which indicated that the heat sto- heat transfer coefficient (4017.70 W·m−2·K−1 for model D) between
rage was completed. HTF and the inner wall of the copper tube. The heat transferred at the
As can be seen from Fig. 7(a) and (c), for model A and model C, the same temperature difference was also remarkably increased, which led
closer the selected point was to the heating boundary wall during the to the more apparent influence of the temperature difference for PCM
first and second stage, the faster the temperature rose for the point. It along radial direction. Therefore, inserting metal foam into HTF can
can be observed that the temperature difference between the five points increase the PCM temperature gradient along the radial direction; while
was gradually reduced from the third stage, and the temperature var- adding metal foam in PCM will improve the temperature uniformity
iation curve of the five points almost overlapped until the later stage. along the radial direction.
The temperature difference among the five points is larger for model C It can be discovered from Fig. 8(a) and (b) that in the first stage and
than that for model A, due mainly to the fact that more heat can be in the early stage of the second stage, the temperature difference be-
extracted from the HTF with metal foam inserted in HTF. When in- tween the five selected points along the axial direction was small and
serting metal foam in PCM, a distinctive phenomenon can be found in the temperature variation curves almost overlapped. This was due
Fig. 7(b) that the temperature variation curves of the five selected mainly to two aspects: first, the thermal resistance between HTF and
points along the radial direction in model B substantially overlapped the inner wall of the copper tube was high, so that the heat passing
and the radial temperature gradient was close to zero. This may be through the tube boundary wall was less. As a result, the HTF tem-
caused by the following two reasons: (1) the surface heat transfer perature difference between inlet and outlet was small (ΔT1 = 2.91 K
coefficient between HTF and the inner wall of the copper tube was for model A, ΔT2 = 4.16 K for model B), leading to an approximately
small (12.82 W·m−2·K−1 for model B), and thus limited amount of constant temperature boundary wall along the axial direction. Conse-
thermal energy was transferred from HTF to PCM; (2) due to the em- quently, the influence of the heating boundary wall on the same radial
ployment of metal foam in PCM, with its abundant specific surface area position was almost identical. Second, at this stage, the melting fraction

151
X. Yang et al. Applied Energy 239 (2019) 142–156

350 350
Third stage
340 340 Second stage
First stage
330 330
Model A (1.00 - 1.00) Model B (1.00 - 0.94)

T/K
320 320
T/K

point1 point1
point2 point2
310 310
point3 point3
300 point4 300 point4
point5 point5
290 290
0 18000 36000 54000 72000 90000 0 14000 28000 42000 56000 70000
t/s t/s
(a) (b)
350 350

340 340

330 330
Model C (0.94 - 1.00) Model D (0.94 - 0.94)
320 320
T/K

T/K
point1 point1
point2 point2
310 310
point3 point3
300 point4 300 point4
point5 point5
290 290
0 6000 12000 18000 24000 30000 0 2000 4000 6000 8000 10000
t/s t/s
(c) (d)
Fig. 7. Plot of temperatures at selected points along the radial direction in the four models with different metal foam locations: (a) model A; (b) model B; (c) model C;
(d) model D.

350 350

340 340

330 330
Model A (1.00 - 1.00) Model B (1.00 - 0.94)
320 320
T/K

T/K

point6 point6
310 point7 310 point7
point3 point3
300 point8 300 point8
point9 point9
290 290
0 18000 36000 54000 72000 90000 0 14000 28000 42000 56000 70000
t/s t/s
(a) (b)

350 350

340 340

330 330
Model C (0.94 - 1.00) Model D (0.94 - 0.94)
320 320
T/K

T/K

point6 point6
310 point7 310 point7
point3 point3
300 point8 300 point8
point9 point9
290 290
0 6000 12000 18000 24000 30000 0 2000 4000 6000 8000 10000
t/s t/s
(c) (d)
Fig. 8. Plot of temperatures at selected points along the axial direction in the four models with different metal foam locations: (a) model A; (b) model B; (c) model C;
(d) model D4.

152
X. Yang et al. Applied Energy 239 (2019) 142–156

of PCM in the annular gap was little, and the natural convection effect To quantitatively demonstrate and compare the temperature re-
was faint. Based on the two reasons, the temperature variation curves at sponse capability, an integral average value during the entire charging
this stage almost overlapped. However, in the later stage, as the melting process was introduced [16], expressed as:
fraction of PCM increased, natural convection was triggered on and its
tfull ϕ
intensity was continuously increasing, leading to the fact that tem- ϕ¯ = ∫0 t full (Ti − Tw )
dt
(26)
perature variation curves were tardily dispersed. Compared with model
A, the involvement of metal foam in PCM significantly improved the where ϕ̄ was the integral average value during the full melting time tfull,
temperature uniformity along axial direction. Fig. 8(c) and (d) depicted and Ti and Tw were the initial temperature and the tube boundary wall
that inserting metal foam into HTF (model C and model D) separately temperature, respectively. Therefore, the integral average temperature
enlarged the temperature difference among the five selected points in ¯ ) can be used to quantitatively describe the effect of
response rate (RR
comparison with model A and model B. More heat was removed from metal foam location on melting process, with its expression as:
HTF to PCM, thereby making the HTF temperature difference between
inlet and outlet further increased (ΔT3 = 5.66 K for model C, tfull 1 T (ti ) − T (ti − 1)
¯ =
RR ∫0 dt
ΔT4 = 18.44 K for model D). At this time, the tube boundary wall t full (ti − ti −1) (27)
cannot be regarded as a constant temperature wall. Compared with
where T(ti) and T(ti−1) were the temperatures at ti and ti−1, and tfull
model A and C, the involvement of metal foam in PCM can still improve
represented the full melting time for each model. Fig. 9(c) demon-
the temperature uniformity along axial direction in PCM.
strated the average response rates of P3 for the four models and made a
To compare the effect of the metal foam location on the temperature
quantitative comparison. It can be seen that model D had the highest
response, the representative point P3 in PCM was selected for the four
temperature response rate. Taking the temperature response rate of
models, and their temperature evolution histories were displayed in
model A as the benchmark, model B increased by 15.4%, model C in-
Fig. 9(a). It can be seen that during the charging process, the fastest
creased by 218.55% and model D increased by 834.27%. In some
temperature response was model D, followed by model C, B and A. The
special occasions (such as heating a room), we need to reach the set
results revealed that the thermal resistance of the TES unit on the HTF
temperature quickly, which has high requirements for temperature re-
side is much higher than that on the PCM side. Therefore, for the re-
sponse. Therefore, the addition of metal foam into both domains can
covery of waste heat in exhaust gas, the heat transfer enhancement on
meet this demand.
the HTF side should be considered with respect to the case of waste
Similarly, the same method was used to compare the effect of metal
water. In the practical TES system, we often need to consider the
foam porosity on temperature response. The representative point P3 in
temperature field uniformity of TES medium. For example, in order to
PCM for the five models was selected, as well. Its temperature evolution
obtain hot water with stable temperature, this puts forward high de-
history was plotted in Fig. 9(d). It can be seen that model D3 had the
mands on the temperature field uniformity of PCM. According to the
highest temperature response rate, followed by model D, D4, D2 and D1.
simulation results, it’s especially important to insert metal foam into
Taking the temperature response rate of model D as the benchmark,
PCM domain from the perspective of the temperature field uniformity.
model D3 increased by 0.88%, while model D4, D2 and D1 decreased by

350 350

340 340

330 330
P3, Porosity: HTF - PCM
320 320
T/K
T/K

Model D1 (0.90 - 0.98)

P3, Porosity: HTF - PCM
Model D2 (0.92 - 0.96)
310 Model A (1.00 - 1.00) 310
Model B (1.00 - 0.94) Model D (0.94 - 0.94)
300 Model C (0.94 - 1.00) 300 Model D3 (0.96 - 0.92)
Model D (0.94 - 0.94) Model D4 (0.98 - 0.90)
290 290
0 15000 30000 45000 60000 75000 90000 0 2200 4400 6600 8800 11000
t/s t/s
(a) (b)

60 60
P3, Porosity: HTF - PCM P3, Porosity: HTF - PCM
51.57 51.57 52.02
49.23 50.27
50 50
44.87

40 40
RR (×10-4)

RR (×10-4)

30 30

20 17.58 20

10 5.52 6.37 10

0 0
Model A Model B Model C Model D Model D1 Model D2 Model D Model D3 Model D4
(1.00 - 1.00) (1.00 - 0.94) (0.94 - 1.00) (0.94 - 0.94) (0.90 - 0.98) (0.92 - 0.96) (0.94 - 0.94) (0.96 - 0.92) (0.98 - 0.90)
(c) (d)
Fig. 9. Comparison of temperature response rate for different models: (a) temperature variation of Point 3 for the four models with different metal foam locations; (b)
temperature variation of Point 3 for the five models with different metal foam porosity; (c) average temperature response rate of Point 3 for the four models with
different metal foam locations; (d) average temperature response rate of Point 3 for the five models with different metal foam porosity.

153
X. Yang et al. Applied Energy 239 (2019) 142–156

2.52%, 4.54% and 13.00%, in respective. This indicated that the por- where q̄ was the integral-mean heat flux over the full melting time.
osity of metal foam in HTF should be equal or a little bit higher than Fig. 10(c) showed the average heat flux of the four models for quanti-
that in PCM to further enhance melting heat transfer, if both sides were tative comparison. It can be seen that model D had the maximum
inserted with porous media. average heat flux (maximum heat transfer enhancement), followed by
model C, B and A. Based on the average heat flux of model A, model B
increased by 41.27%, model C increased by 182.26%, and model D
4.4. Comparisons of heat flux
increased by 774.90%.
For convenience, the effect of metal foam porosity on heat transfer
To specify the effect of the metal foam location on the heat transfer
rate was compared by using the same method. The simulation results of
rate, in Fig. 10(a), the heat flux variation on the outer wall of the copper
the five models were plotted in Fig. 10(b), and it can be observed that
tube was described for the four models. It can be seen that for the four
there was no significant difference in the heat transfer rate of the five
models, although there was a large difference in numerical value of
models. Fig. 10(d) demonstrated the average heat flux of the five
heat flux, the variation trends were similar. Heat flux on the tube sur-
models for quantitative comparison. It can be found that model D3 had
face was aiming to melt solid paraffin. Therefore, the higher the heat
the maximum average heat flux (maximum heat transfer enhancement),
flux was, the higher the melting fraction was resulted in. Model D had
followed by model D4, D, D2 and D1. Based on the average heat flux of
the highest heat flux and followed by model C, B and A, which was in
model D, model D4 and D3 increased by 1.67% and 2.65%, in re-
consistence with the melting fraction illustrated in Fig. 4(a). It can be
spective; model D2 and D1 decreased by 5.28% and 15.85% respec-
clearly known from Fig. 10(a) that model D was distinguished from
tively.
other models, and the variation curves of model A and model B were
substantially coincident. This meant that the TES rate can be sig-
nificantly increased if metal foam was arranged on the HTF side. For 4.5. j-f factor evaluation
comparison, the contribution of metal foam arranged on the PCM side
to the melting process was less remarkable. However, Fig. 10(a) only Enhancement in heat transfer commonly existed with the increase in
qualitatively compared the effect of metal foam location on the heat flow resistance. It was therefore necessary to comprehensively consider
flux for the four models. In order to quantitatively summarize the effect the dual effects of heat transfer and flow resistance to obtain the ob-
of metal foam location on heat flux, we took integral-mean value over jective evaluation result. Faced with a variety of different heat transfer
the entire charging process for evaluation, adopting integral-mean heat enhancement techniques, different evaluation indicators were pro-
flux q̄ to quantitatively describe the TES rate for different metal foam posed. In order to quantitatively reveal the heat transfer enhancement
locations, with its expression as: capacity of different models, j-f factor was introduced for evaluation.
t The heat transfer factor j and the friction factor f separately represented
∫0 full q (t )dt the heat transfer performance and the resistance performance for the
q¯ =
t full (28) heat transfer surface, with their expressions as:

2500 3000
Porosity: HTF - PCM Porosity: HTF - PCM
2000 Model A (1.00 - 1.00) 2400 Model D1 (0.90 - 0.98)
Model B (1.00 - 0.94) Model D2 (0.92 - 0.96)
1500 Model C (0.94 - 1.00) 1800 Model D (0.94 - 0.94)
q / W·m-2

q / W· m-2

Model D (0.94 - 0.94)

Model D3 (0.96 - 0.92)
1000 1200 Model D4 (0.98 - 0.90)

500 600

0 0
0 15000 30000 45000 60000 75000 90000 0 2200 4400 6600 8800 11000
t/s t/s
(a) (b)

1200 1200
Porosity: HTF - PCM Porosity: HTF - PCM
1037.63 1027.77
1010.86 1010.86
1000 1000 957.54
850.63
800 800
q / W· m-2
q / W· m-2

600 600

400 326.12 400

200 163.23 200

115.54

0 0
Model A Model B Model C Model D Model D1 Model D2 Model D Model D3 Model D4
(1.00 - 1.00) (1.00 - 0.94) (0.94 - 1.00) (0.94 - 0.94) (0.90 - 0.98) (0.92 - 0.96) (0.94 - 0.94) (0.96 - 0.92) (0.98 - 0.90)

(c) (d)
Fig. 10. Comparison of heat flux for different models: (a) heat flux variation of the four models with different metal foam locations; (b) heat flux variation of the five
models with different metal foam porosity; (c) average heat flux of the four models with different metal foam locations; (d) average heat flux of the five models with
different metal foam porosity.

154
X. Yang et al. Applied Energy 239 (2019) 142–156

500 respective. In practical applications, economy and TES efficiency are

Porosity: HTF - PCM 431.52 our main focus. On the basis of enhancing heat transfer, we need to
400 further reduce the operating costs of the system. According to the j-f
factor evaluation, we can get the optimal scheme to give scientific
guidance for the actual project.
j / f 1/3(×10-4)

300
5. Conclusions
200
A two-dimensional axisymmetric model with natural convection
100 was established for a shell-and-tube thermal energy storage unit, and
38.85 the melting charging process of phase change materials was numeri-
8.16 9.91 cally simulated. The effects of metal foam location and metal foam
0 porosity on the thermal performance for the studied thermal energy
Model A Model B Model C Model D
(1.00 - 1.00) (1.00 - 0.94) (0.94 - 1.00) (0.94 - 0.94) storage unit were analyzed to obtain the optimal heat transfer en-
(a) hancement. Based on the current work, the following conclusions can
be drawn:

500 (1) The thermal resistance for thermal energy storage unit was mainly
Porosity: HTF - PCM 461.81
431.52 present on the heat transfer fluid (air) side, whose thermal re-
400 sistance was much higher than that on the phase change materials
363.07 side. In practical applications, metal foam should be arranged in
334.24
both the two domains simultaneously to achieve the best thermal
j / f 1/3(×10-4)

300 storage performance. The full melting time can be reduced about
88.54%; the temperature response rate was increased by 834.27%;
200.78
200 the average heat flux was increased by 774.9% and the compre-
hensive heat transfer performance (j-f factor) was increased by
5186.91%.
100
(2) The porosity of metal foam had a certain influence on the melting
heat transfer, but there was no linear relationship between the
0 porosity and the heat transfer enhancement. In the case of inserted
Model D1 Model D2 Model D Model D3 Model D4
porosity of 0.96–0.92, the optimum heat transfer enhancement can
(0.90 - 0.98) (0.92 - 0.96) (0.94 - 0.94) (0.96 - 0.92) (0.98 - 0.90)
be provided.
(b)
(3) Embedding metal foam into the phase change materials domain can
Fig. 11. Comparison of JF value for different models: (a) JF value of the four better transfer the thermal energy from the hot boundary wall to
models with different metal foam locations; (b) JF value of the five models with the interior of phase change materials, which not only improved the
different metal foam porosity. temperature field uniformity of phase change materials, but also
prolonged the constant temperature duration (phase change time)
h for phase change materials. In contrast, embedding metal foam into
j= Pr 2/3
ρucp (29) the heat transfer fluid domain destroyed the temperature field
uniformity of phase change materials and reduced the constant
ΔP temperature duration for phase change materials.
f=
1/2ρu2 (30) (4) The porosity of metal foam had a significant effect on the tem-
perature field of phase change materials. The interpenetrating pore
where h was the average heat transfer coefficient between HTF and structure provided space for local natural convection. The higher
PCM, ρ, cp and Pr separately denoted the average density, specific heat the porosity of metal foam inserted in the phase change materials
capacity and Prandtl number of the HTF, u represented the flow velo- domain, the stronger the natural convection. This resulted in a
city of HTF, and ΔP was the pressure drop between inlet and outlet. larger slant angle of the solid-liquid phase interface and a more
Here, we introduced a comprehensive performance coefficient (JF) for uneven temperature field.
evaluation,

JF = j/ f 1/3 Acknowledgement
(31)

The larger the JF value was, the better the overall heat transfer This work was supported by the National Natural Science
performance was when the pump power was constant. To quantita- Foundation of China (51506160), Natural Science Foundation of
tively check the effect of metal foam location on the overall heat Shaanxi Province (2017JQ5007) and the fundamental research funds
transfer performance, Fig. 11(a) demonstrated the JF values of the four for central universities (xjj2016042).
models. It can be seen that model D had the maximum JF value (op-
timum heat transfer enhancement), followed by model C, B and A. By References
comparing the JF value of model A, model B, C and D increased by
21.45%, 375.98% and 5186.91%. [1] Cabeza LF, Palacios A, Serrano S, Ürge-Vorsatz D, Barreneche C. Comparison of past
Similarly, the effect of metal foam porosity on comprehensive heat projections of global and regional primary and final energy consumption with
historical data. Renew Sustain Energy Rev 2018;82:681–8.
transfer performance was summarized by the same method. Fig. 11(b) [2] Sharma SS. Determinants of carbon dioxide emissions: empirical evidence from 69
showed the JF values of the five models for quantitative comparisons. It countries. Appl Energy 2011;88:376–82.
can be observed that model D3 had the maximum JF value (optimum [3] Chua KJ, Chou SK, Yang WM, Yan J. Achieving better energy-efficient air con-
ditioning – a review of technologies and strategies. Appl Energy 2013;104:87–104.
heat transfer enhancement), followed by model D, D4, D2 and D1. By [4] Chen J, Wu J, Ji X, Lu X, Wang C. Mechanism of waste-heat recovery from slurry by
comparing the JF value of model D, only model D3 increased by 7.02%, scraped-surface heat exchanger. Appl Energy 2017;207:146–55.
model D1, D2 and D4 decreased by 53.47%, 22.55% and 15.86%, in [5] Miró L, Gasia J, Cabeza LF. Thermal energy storage (TES) for industrial waste heat

155
X. Yang et al. Applied Energy 239 (2019) 142–156

(IWH) recovery: a review. Appl Energy 2016;179:284–301. simulation with volume-averaged method on composite phase change materials for
[6] Brückner S, Liu S, Miró L, Radspieler M, Cabeza LF, Lävemann E. Industrial waste thermal energy storage. Appl Energy 2018;229:700–14.
heat recovery technologies: an economic analysis of heat transformation technol- [23] Yao Y, Wu H, Liu Z, Gao Z. Pore-scale visualization and measurement of paraffin
ogies. Appl Energy 2015;151:157–67. melting in high porosity open-cell copper foam. Int J Therm Sci 2018;123:73–85.
[7] Fang H, Xia J, Zhu K, Su Y, Jiang Y. Industrial waste heat utilization for low tem- [24] Xiao X, Zhang P, Li M. Preparation and thermal characterization of paraffin/metal
perature district heating. Energy Policy 2013;62:236–46. foam composite phase change material. Appl Energy 2013;112:1357–66.
[8] Yang X, Lu Z, Bai Q, Zhang Q, Jin L, Yan J. Thermal performance of a shell-and-tube [25] Abishek S, King AJC, Nadim N, Mullins BJ. Effect of microstructure on melting in
latent heat thermal energy storage unit: role of annular fins. Appl Energy metal-foam/paraffin composite phase change materials. Int J Heat Mass Transf
2017;202:558–70. 2018;127:135–44.
[9] Jiang Z, Ouyang T, Yang Y, Chen L, Fan X, Chen Y, et al. Thermal conductivity [26] Xu Y, Ren Q, Zheng ZJ, He YL. Evaluation and optimization of melting performance
enhancement of phase change materials with form-stable carbon bonded carbon for a latent heat thermal energy storage unit partially filled with porous media.
fiber network. Mater Des 2018;143:177–84. Appl Energy 2017;193:84–95.
[10] Meng ZN, Zhang P. Experimental and numerical investigation of a tube-in-tank [27] Wang C, Lin T, Li N, Zheng H. Heat transfer enhancement of phase change com-
latent thermal energy storage unit using composite PCM. Appl Energy posite material: copper foam/paraffin. Renewable Energy 2016;96:960–5.
2017;190:524–39. [28] Feng S, Shi M, Li Y, Lu TJ. Pore-scale and volume-averaged numerical simulations
[11] Zhang P, Meng ZN, Zhu H, Wang YL, Peng SP. Melting heat transfer characteristics of melting phase change heat transfer in finned metal foam. Int J Heat Mass Transf
of a composite phase change material fabricated by paraffin and metal foam. Appl 2015;90:838–47.
Energy 2017;185:1971–83. [29] Jin HQ, Fan LW, Liu MJ, Zhu ZQ, Yu ZT. A pore-scale visualized study of melting
[12] Yang J, Yang L, Xu C, Du X. Experimental study on enhancement of thermal energy heat transfer of a paraffin wax saturated in a copper foam: effects of the pore size.
storage with phase-change material. Appl Energy 2016;169:164–76. Int J Heat Mass Transf 2017;112:39–44.
[13] Son CH, Morehouse JH. Thermal conductivity enhancement of solid-solid phase- [30] Mahdi JM, Nsofor EC. Melting enhancement in triplex-tube latent heat energy
change materials for thermal storage. J Thermophys Heat Transfer 2015;5:122–4. storage system using nanoparticles-metal foam combination. Appl Energy
[14] Siahpush A, O’Brien J, Crepeau J. Phase change heat transfer enhancement using 2017;191:22–34.
copper porous foam. J Heat Transfer 2008;130:082301–82311. [31] Tao YB, You Y, He YL. Lattice Boltzmann simulation on phase change heat transfer
[15] Chen K, Guo L, Xie X, Liu W. Experimental investigation on enhanced thermal in metal foams/paraffin composite phase change material. Appl Therm Eng
performance of staggered tube bundles wrapped with metallic foam. Int J Heat 2016;93:476–85.
Mass Transf 2018;122:459–68. [32] Žukauskas A. Heat transfer from tubes in crossflow. Adv Heat Transfer
[16] Yang X, Feng S, Zhang Q, Chai Y, Jin L, Lu TJ. The role of porous metal foam on the 1972;8:93–160.
unidirectional solidification of saturating fluid for cold storage. Appl Energy [33] Calmidi VV. Transport phenomena in high porosity fibrous metal foams. Boulder,
2017;194:508–21. CO: University of Colorado; 1998.
[17] Xiao X, Zhang P, Li M. Effective thermal conductivity of open-cell metal foams [34] Yang XH, Kuang JJ, Lu TJ, Han FS, Kim T. A simplistic analytical unit cell based
impregnated with pure paraffin for latent heat storage. Int J Therm Sci model for the effective thermal conductivity of high porosity open-cell metal foams.
2014;81:94–105. J Phys D Appl Phys. 2013;46:255302.
[18] Lafdi K, Mesalhy O, Shaikh S. Experimental study on the influence of foam porosity [35] Yang XH, Bai JX, Yan HB, Kuang JJ, Lu TJ, Kim T. An analytical unit cell model for
and pore size on the melting of phase change materials. J Appl Phys the effective thermal conductivity of high porosity open-cell metal foams. Transp
2007;102:083549. Porous Media 2014;102:403–26.
[19] Yang X, Wang W, Yang C, Jin L, Lu TJ. Solidification of fluid saturated in open-cell [36] Georgiadis JG, Catton I. Dispersion in Cellular thermal-convection in porous layers.
metallic foams with graded morphologies. Int J Heat Mass Transf 2016;98:60–9. Int J Heat Mass Transf 1988;31:1081–91.
[20] Zhao CY, Lu W, Tian Y. Heat transfer enhancement for thermal energy storage using [37] Yang XH, Lu TJ, Kim T. An analytical model for permeability of isotropic porous
metal foams embedded within phase change materials (PCMs). Sol Energy media. Phys Lett A 2014;378:2308–11.
2010;84:1402–12. [38] Bhattacharya A, Calmidi VV, Mahajan RL. Thermophysical properties of high por-
[21] Zhu F, Zhang C, Gong X. Numerical analysis and comparison of the thermal per- osity metal foams. Int J Heat Mass Transf 2002;45:1017–31.
formance enhancement methods for metal foam/phase change material composite. [39] Reh S, Beley JD, Mukherjee S, Khor EH. Probabilistic finite element analysis using
Appl Therm Eng 2016;109:373–83. ANSYS. Struct Saf 2006;28:17–43.
[22] Yang X, Bai Q, Guo Z, Niu Z, Yang C, Jin L, et al. Comparison of direct numerical

156