Applied Thermal Engineering 254 (2024) 123864

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Experimental and numerical investigations on the intermittent heat transfer

performance of phase change material (PCM)-based heat sink with triply
periodic minimal surfaces (TPMS)
Jiabin Wang a , Wenhao Pu a, * , Haisheng Zhao b , Long Qiao a , Nanxin Song a , Chen Yue a
a
College of Energy and Power Engineering, Nanjing University of Aeronautics and Astronautics, Key Laboratory of Thermal Management and Energy Utilization of
Aircraft, Ministry of Industry and Information Technology, Nanjing, Jiangsu 210016, China
b
HFYC(Zhenjiang) Additive Manufacturing Co. Ltd., Zhenjiang, Jiangsu, 212132,China

A R T I C L E I N F O A B S T R A C T

Keywords: The excessive heat generated during the operation of electronic components leads to significant temperature
Phase change material increases, posing a significant risk to their service life. For the problem of efficient thermal management of
Triply periodic minimal surfaces electronic devices in confined spaces experiencing high heat flow, a scheme using triply periodic minimal sur­
Intermittent
faces and a phase change material-based active–passive cooling heat sink is proposed to control the temperature
Temperature control
Heat sink
of electronic equipment. The apparent heat capacity method is employed to simulate the intermittent process of
the phase change material-based heat sink. The optimization based on Box-Behnken Design and Nondominated
Sorting Genetic Algorithm II is analyzed to obtain an improved triply periodic minimal surface structure. The
effect of thermal performance (including base temperature, liquid fraction, and Grashof number) of an inter­
mittent heat sink based on an improved structure is discussed. Visualization and temperature testing platforms
are established. Through visualization experiments, the internal temperature and liquid fraction distribution of
the heat sink are obtained, and the simulation results are in good agreement with the test results. The results
show that the base temperature, liquid fraction, and Grashof number achieve a steady periodic variation during
the intermittent process. Specifically, the base temperature remains stable in the range of 314 K to 340 K, the
liquid fraction stabilizes between 0.8 and 1.0, and the Grashof number stabilizes between 100 and 1000. At the
heating power of 30 W or below, the base temperature of the heat sink exhibits a steady periodic variation. At the
heating power of 40 W, the maximum base temperature of the heat sink exceeds 370 K.

Li-Wu Fan et al. [8] utilized a liquid metal PCM-based heat sink for
1. Introduction thermal energy storage. Their research revealed that liquid metal
significantly extends the effective protection time during the heating
With the rapid development of electronic technology, the amount of process. Rao et al. [9] found that the adoption of a PCM-based heat sink
heat generated by electronic devices has increased significantly [1]. could make the temperature distribution inside the battery more uni­
Electronic devices in confined spaces, such as high-speed railway cam­ form. Additionally, the adoption of PCMs with low melting points and
eras, semiconductor cooling chips, lithium-ion batteries, etc., often high thermal conductivity [10–13] could improve the heat transfer
exhibit intermittent operational characteristics and frequently performance of PCM-based heat sinks.
encounter transient high heat flow during operation. Therefore, From the above, it is evident that PCMs are highly suitable for
resolving the contradiction between high heat flow and the maximum thermal control of electronic components. However, the slow melting
allowable operating temperature has become a focal point of current rate of many solid–liquid PCMs, attributed to their inherently low
research. By leveraging the substantial latent heat capacity of PCMs, thermal conductivity, limits the effectiveness of phase change heat
effective temperature control of electronic components can be achieved storage. In order to improve the thermal conductivity of PCMs, Lin et al.
[2–7]. [14] provided a review of how to enhance the thermal performance of
PCMs are often used for latent heat thermal energy storage (LHTES). PCMs and which fields the PCMs could be applied. The paraffin wax is

* Corresponding author at: College of Energy and Power Engineering, Nanjing University of Aeronautics and Astronautics, Key Laboratory of Thermal Management
and Energy Utilization of Aircraft, Ministry of Industry and Information Technology, Nanjing, Jiangsu 210016, China.
E-mail address: [email protected] (W. Pu).

https://doi.org/10.1016/j.applthermaleng.2024.123864
Received 26 February 2024; Received in revised form 3 June 2024; Accepted 2 July 2024
Available online 4 July 2024
1359-4311/© 2024 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Nomenclature Tpc Melting point temperature(K)

Tf Temperature of fluid(K)
Amush Constant of mushy zone Ta Temperature of base(K)
Cp Specific heat capacity (J/(kg⋅K)) tmelt Melting time(s)
D(T) Gaussian function [36] u Velocity (m/s)
Eh Entransy(J⋅K) [37] β Thermal expansion coefficient (K− 1)
E1 Entransy of base(J⋅K) θ Liquid fraction
F Buoyancy force(N) v Kinematic viscosity(m2/s)
Gr Grashof number ρ Density (kg/m3)
H Heat transfer coefficient W/(m2⋅K) μ Dynamic viscosity (N⋅s/m2)
I Unit vector ε A small constant
I1 Solid state sensible heat absorption stage φ Heat storage rate (J/s)
J1 Latent heat absorption stage Δ Differential operator
K1 Liquid state sensible heat absorption stage
k Thermal conductivity (W /(m⋅K)) Subscripts
L Latent heat (J/kg) avg Average
mf Fluid mass(kg) eff Effective
l1 characteristic length (m) Abbreviations
p Pressure (Pa) BTMs Battery thermal management systems
Qvh Heat capacity(J) CAD Computer-aided design
QA Input heat capacity at the base (J) LHTES Latent heat thermal energy storage
Qs Heat storage capacity(J) MFPCM Metal foam–phase change material
Qi Thermal power at the base(W) PCM Phase change materials
S Momentum sink term TPMS Triply periodic minimal surfaces
T Temperature (K)

typically combined with high thermal conductivity materials, including display patterns in three directions in a Cartesian coordinate system. In
metal meshes [15] and nanoparticles [16]. The improved PCMs are this study, we use the modeling tool called Wolfram Mathematica to
called composite PCMs (CPCMs). CPCMs are widely used for enhancing mathematically model TPMS structures.
the thermal performance of PCMs. Santosh Chavan et al. [17] studied After years of research and application, TPMS structures have
the effects of different orientations on the CPCM composed of 98 % demonstrated their excellent performance in various ways. Al-Ketan
paraffin wax and 2 % copper nanoparticles under various wall heating et al. [24] investigated the topological-mechanical properties relation­
conditions. The results indicated that the bounded domain with different ship of different structures. And they found that compared with other
orientations had a significant impact on the formation of natural con­ structures tested, TPMS structures showed more excellent mechanical
vection. Santosh Chavan et al. [18] also prepared CPCMs based on linear performance. Furthermore, in scaffold and tissue engineering [25–28],
low-density polyethylene and analyzed their performance. Thermal we can find the applications of TPMS structures. The feed spacers based
characterization revealed that the latent heat capacity was enhanced on TPMS [29] are used for reverse osmosis and ultrafiltration technol­
with the increase in the concentration of nanoparticles. The CPCMs ogies. They effectively increase the flux and reduce biological
composed of metal foams and PCMs are referred to as metal foam-PCMs contamination.
(MFPCMs) [19]. Dinesh et al. [20] considered that the pore size and More and more attention has been paid to detailed thermal charac­
porosity of the foams could affect the heat performance of metal foam- terizations and heat transfer applications of TPMS structures. Smith
PCMs (MFPCMs). Additionally, Abishek et al. [21] studied how the et al. [30] studied the thermal conductivity of composite materials based
microstructure affects the melting time of PCM using the method called on TPMS and air through an experiment. The results indicated that the
pore-scale computational analysis.” Wei Li et al. [22] studied tempera­ thermal conductivity of TPMS structures had a mathematical relation­
ture control for electronic components using a PCM-based heat sink with ship with the relative density and architecture. Deng et al. [31] estab­
honeycomb metal. Results showed effective temperature regulation and lished a visualization experimental platform for phase change thermal
extended heat sink working time, especially under high heat and low energy storage. They studied the thermal storage performance of TPMS
PCM conductivity. porous aluminum/paraffin wax composite materials. The results
In summary, to enhance the thermal characteristics of PCMs, lots of revealed that the porous structure of TPMS, while partially suppressing
studies of CPCMs have been conducted. Although nanomaterials possess the natural convection capability of paraffin, significantly enhanced the
numerous unique properties, they also exhibit several drawbacks, thermal conductivity of phase change heat sinks. Al-Ketan et al. [32]
including high production costs, poor stability, and insufficient nano­ studied the thermal characteristics of TPMS-based heat sinks and the
particle loading robustness. Traditional metal foams also have some results showed that both the porosity and architecture could affect the
drawbacks. For instance, the pore shape distribution of traditional metal thermal performance in the heat transfer process. Zahid Ahmed Qureshi
foams is random, and the structure is prone to discontinuity, leading to et al. [33] studied the thermal performance of PCM-based heat sinks
additional thermal resistance. In recent years, the triply periodic mini­ with three TPMS structures, including Gyroid, IWP, and Primitive, and
mal surfaces (TPMS) structures with controllable structures and regular the conventional Kelvin structure. They found that compared to the
pore distribution patterns have attracted attention. The development of Kelvin structure the three TPMS structures had superior thermal con­
additive manufacturing technology has led to an improvement in the ductivity. Zahid Ahmed Qureshi et al. [34] also found that the superior
production efficiency of TPMS structures [23]. Despite being named thermal performance of the heat sinks during a heat transfer process
‘minimal surfaces’, TPMS structures actually exhibit larger surface areas could be attributed to their combination with TPMS structures.
compared to Kelvin cells when considering the same cell size and From the review of the literature mentioned above, it can be found
porosity. As the name ‘triply periodic’ indicates, TPMS can periodically that previous studies have mainly focused on a melting or solidification

2
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

process. Research on the entire intermittent process remains limited. while there is no heat flux at the bottom during this phase. In the tem­
Due to the superior thermal performance of TPMS, a scheme of TPMS perature testing experiment, T-type thermocouples are arranged on the
combined with an intermittent PCM-based heat sink is proposed. TPMS specimen, and the temperature data collected by the temperature
Experimental and numerical research on the intermittent process of probe is transmitted to the computer. In the visualization study, one side
TPMS combined with a PCM-based heat sink is carried out. This study of the heat sink’s AlSi10Mg wall is replaced with quartz glass to accu­
employs the apparent heat capacity method to simulate three- rately capture the temperature change and melting front of the paraffin
dimensional intermittent processes. Guided by the BBD and NSGA-II using the infrared thermal imager. The thickness of the external wall of
methods, the PCM-based heat sink with TPMS optimization, which ex­ the heat sink mentioned above is 2 mm.
hibits optimal thermal performance, is obtained with the goal of mini­ Fig. 2(a) and Fig. 2(b) illustrate the experimental setups for tem­
mizing entransy and maximizing the heat storage rate. The numerical perature testing and visualization, respectively. The heat sink is made of
simulation investigates the thermal performance (base temperature, AlSi10Mg material and has overall dimensions of 44.5 × 44.5 × 42 mm3.
liquid fraction, and Grashof number) of the optimized PCM-based heat The AlSi10Mg heat sink is produced using selective laser melting tech­
sink during the charging and discharging cycles. Visualization and nology. The selective laser melting technology employs high-power and
temperature testing platforms have been established to obtain the in­ high-energy–density lasers to fuse metal powder particles within a
ternal temperature distribution and liquid fraction distribution of the specific region of each layer with the solid material of the previous layer,
heat sink, and the experimental results are utilized to validate the allowing for the consolidation and formation of parts. The raw material
simulation results. Experimental research on the effects of heating for 3D printing is spherical AlSi10Mg alloy powder, with a particle size
power during the intermittent process of the optimized PCM-based heat range of 15 to 53 μm. Its chemical composition (weight, %) is primarily
sink is carried out. composed of Si 9.5 ~ 10.5, Mg 0.25 ~ 0.50, Fe = 0.044, Mn < 0.01, Cu
< 0.01, and the balance being Al. PCM is cast into the heat sink, forming
2. Experimental study a PCM-based heat sink. In this study, paraffin wax is used for experi­
mental and simulation research. The thermal properties of the paraffin
2.1. Experimental setup wax are derived from the reference [35] and the thermal properties of
AlSi10Mg are measured in-house in this study. The material properties
The experimental setup for the PCM-based heat sink test system is of PCM and AlSi10Mg are shown in Table 1.
shown in Fig. 1. The components of the test platform include an In the temperature testing experiment, six T-type thermocouples
AlSi10Mg heat sink, paraffin-based PCM, temperature probe, T-type with a diameter of 1 mm are mounted on the TPMS specimen, and the
thermocouples, electric heating plate, DC power supply, computer, positions of the measurement points are shown in Fig. 3. The coordinate
MAG12 online infrared thermal imager, insulation cotton, quartz glass, information of all measurement points is shown in Table 2.
and thermal conductive silver paste. The infrared thermal imager has The TPMS structure used in this study is generated based on a
dimensions of 84 × 43 × 60 mm3, a temperature measurement range of mathematical model implemented in Wolfram Mathematica 12. The
− 293 K to 423 K, an infrared pixel count of 80 × 60, a visible light pixel improved model B1 derived from section 4.2 is used in the experiments.
resolution of 640 × 480, and a frame rate of 50HZ. The type of infrared The TPMS structure is shown in Fig. 4.The size information of the TPMS
detector of MAG12 is an uncooled focal plane array. The electric heating structure used for the experiment is shown in Table 3. The volume of
plate is placed near the bottom surface of the PCM-based heat sink to PCM is 52.81 cm3.
simulate the heat source, and thermal conductive paste is evenly applied
between them to reduce the contact thermal resistance. The DC power 2.2. Uncertainty analysis
supply (0–60 V, 0–30 A) provides power to the heating sheet, and the
heating power can be adjusted directly. To enhance the thermal storage The uncertainties in this study are mainly caused by systematic er­
efficiency of the paraffin, TPMS cells are embedded in the paraffin. rors, including the heating power of the T-type thermocouples, tem­
During the heating process, heat is transferred to the paraffin through perature collectors, and DC power. The uncertainty analysis results for
conduction and natural convection. The paraffin continuously absorbs each instrument are shown in Table 4. The temperature uncertainty
heat and undergoes a phase change from solid to liquid when reaching during the experiment is expressed by the following equation:
the melting point. During the intermittent solidification cycle, forced √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑n
convection cooling is applied to the top of the PCM-based heat sink, Δ= Δ2 (1)
i=1 i

Fig. 1. Phase change heat sink system.

3
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Fig. 2. The experimental setup. (a) Experimental setup of temperature test, (b) Visualization experimental setup.

Table 1
The thermal properties of materials.
Properties PCM [35] AlSi10Mg

Thermal conductivity k (W/(m⋅K)) 0.2 180.4

Specific heat cp (J/(kg⋅K)) 2000 970
Density ρ, solid (kg/m3) 880 2600
Density ρ, liquid(kg/m3) 760 −
Melting temperature Tpc (K) 315 −
Phase change temperature range ΔTpc (K) 4 −
Latent heat L (kJ/kg) 165 −
Dynamic viscosity μ（kg/(m⋅s)） 0.0235 −
Thermal expansion coefficient β (1/K) 0.0001 −

Fig. 4. The TPMS structure used in the experiment.

Table 3
Size of the TPMS structure used for the experiment.
surface area porosity wall thickness Surface area to cell volume
(mm2) (%) (mm) ratio(mm− 1)

41,562 80 0.8 3.25

Fig. 3. Thermocouple monitor point locations.

Table 2 Table 4
Location of thermocouple measuring point. Uncertainty for instruments.

Measuring point x(mm) y(mm) z(mm) Symbol Instruments Uncertainty

P1 11.25 11.25 0.5 1 T-type thermocouple ±1%

P2 22.25 22.25 0.5 2 Temperature collector ±0.2 %
P3 33.25 33.25 0.5 3 DC power ±1%
P4 2 27 22
P5 22.25 22.25 22
P6 39.20 18.90 22

4
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Based on Equation (1), the total uncertainty of temperature can be

obtained as 0.38 %. (1 − θ)2
S(T) = Amush u (5)
θ3 + ∊
3. Numerical simulations The term S(T) represents the momentum sink due to the decrease in
liquid fraction in the mushy region, where ε = 0.001 is a small constant
3.1. CAD model to prevent division by zero. Amush is a constant in the mushy region that
varies between 104 and 107. In this study, Amush is set to 105.
In this study, Wolfram Mathematica 12 is used to construct a TPMS The buoyancy force F causes natural convection in the liquid phase
model. The CAD modeling of TPMS can be performed by using and is calculated using the Boussinesq approximation based on Equation
approximate level-set equations for TPMS. The level-set equations for (6), where the magnitude of the gravitational acceleration g is 9.81 m/
the Primitive-type TPMS are mentioned below: s2.
− C < Cos[w1 πx] + Cos[w2 πy] + Cos[w3 πz] < C (2) (
F = ρβg T − Tpc
)
(6)
Where w1, w2, and w3 are control parameters of the TPMS structure Where Tpc is the melting point temperature and β is the coefficient of
in the x, y, and z directions, and C is a constant. By adjusting the pa­ thermal expansion.
rameters w1, w2, w3 and C in the TPMS expression, the crystal cell Energy equation:
structure can be changed. A primitive structure model is generated using ( )
mathematical models in Wolfram Mathematica 12, with C = 0.3, w1 = 1, ∂T ∂p
ρCp + ρCp u⋅∇T + ∇⋅q = βT + u⋅∇p + τ : ∇u + Q (7)
w2 = 2, w3 = 2, within the bounds of − 2 < x < 2, − 2 < y < 2, − 2 < z < 2, ∂t ∂t
as the initial model, as shown in Fig. 5.
Where q is the heat flux(w/m2), τ is the viscous stress tensor (Pa), Q
contains additional heat sources(W/m3).
3.2. Mathematical models
The phase change is characterized by the melting fraction θ ac­
cording to Equation (8). The parameter phase change temperature range
The solid and fluid heat transfer model in COMSOL Multiphysics 6.0
ΔTpc defines the range of the transition area.
is based on the apparent heat capacity method. The model is used in this
⎧
study to solve the heat transfer and flow problems of solid–liquid phase ⎪
⎪ 0, T < Tpc − ΔTpc
⎪
change involving PCM-based heat sinks. The apparent heat capacity ⎪
⎪
⎪ 2
⎪
method uses highly nonlinear specific heat to describe the latent heat of ⎪
⎪
⎪
⎪
( )
⎨ T − T − ΔTpc
phase change. In the solving process, the laminar flow model is com­ pc
θ= 2 ΔTpc ΔTpc (8)
bined with the volume force to simulate natural convection in the phase ⎪ , Tpc − ≤ T ≤ Tpc +
⎪
⎪
⎪ ΔT pc 2 2
change process. The multiphysics nonisothermal flow model is coupled ⎪
⎪
⎪
⎪
with the above two physical models for heat transfer phase change ⎪ ΔTpc
⎩ 1, T < Tpc +
⎪
⎪
simulation. 2
In order to solve the heat transfer and flow problems of PCM-based
The apparent heat capacity method uses a modified specific heat to
heat sink, the following equations [36] are presented.
define the phase change, and the latent heat is a characteristic quantity
Continuity equation:
of phase change. A Gaussian function based on Equation (9) is used with
∂ρ the phase change temperature Tpc as the center.
+ ∇⋅(ρu) = 0 (3)
∂t 2
(T− Tpc )
Where ρ is the material density (kg/m3), u is the velocity (m/s). − ( )2
ΔTpc
Momentum equation: 4
e (9)
∂u [ ( ] D(T) = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )̅ 2
ρ + ρ(u⋅∇)u = ∇⋅ − ρI + μ ∇u + (∇u)T + F − S(T)u (4)
∂t
ΔTpc
π 4

The definition of momentum source term can be described as:

The modified specific heat capacity cp is calculated according to
Equation (10), maintaining the energy balance during phase change.
( )
cp = cp1 + θ⋅ cp2 − cp1 + L⋅D(T) (10)

User-defined functions are used to describe the material properties of

solid and liquid PCM as well as the flow in the transition region. The
liquid fraction θ is used to describe the temperature-dependent material
properties. The density ρ is obtained based on Equation (11), and the
thermal conductivity k is similarly obtained based on Equation (12).
ρ = θ⋅ρ1 + (1 − θ)⋅ρ2 (11)

k = θ⋅k1 + (1 − θ)⋅k2 (12)

In the equations, Tpc is the melting point temperature, ΔTpc is the
temperature difference in the phase change region, θ is the liquid frac­
tion, k1 is the thermal conductivity of solid paraffin, k2 is the thermal
conductivity of liquid paraffin, ρ1 is the density of solid paraffin, ρ2 is the
density of liquid paraffin, cp1 is the specific heat capacity of solid
paraffin, cp2 is the specific heat capacity of liquid paraffin, and L is the
latent heat of phase change.
In the numerical simulation process, several assumptions are made as
Fig. 5. The initial model diagram.

5
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

follows:

(1) The effect of fluid compressibility is ignored, and fluid viscous

dissipation is neglected.
(2) The influence of natural convection is considered, such as buoy­
ancy force due to gravity and flow-induced buoyancy forces in
three directions.
(3) The phase change material is assumed to be isotropic and has
constant properties.

3.3. Boundary conditions

The geometric model of numerical simulation and the setting of

boundary conditions in this paper are shown in Fig. 6. The detailed
Settings are as follows:
For the heating process, a constant heat flow is used at the bottom of
the heat sink. The constant heating power Qi at the bottom is set as 30 W,
and the constant heat flux is q = Qi/A = 15150(W/m2), where A is the
heating area(44.5 × 44.5 mm2).
∂T
− kAlSi10Mg = q’’ (13)
∂yy=0
Fig. 7. Grid independence test.
Where kAlSi10Mg is the thermal conductivity of AlSi10Mg.
The top of PCM-based heat sink contacts with ambient to form nat­ 3.4. Numerical methods
ural convection boundary condition.
∂T To ensure the computational accuracy of the numerical simulation, it
− kambient = H(Tambient − T) (14) is essential to investigate the independence of the validation grid and
∂y
time step size. The TPMS cell in Fig. 4 is used for verification, and the
Where kambient is the thermal conductivity of the ambient air on the bottom surface of heat sink is heated by a constant power of 30 W.
top of the heat sink, H represents the heat transfer coefficient, which is As shown in Fig. 7, the variation of liquid fraction over time is
taken as 10 W/(m2⋅K) in this simulation. Tambient is the ambient compared for three different grid settings. It is found that the maximum
temperature. deviation of the liquid fraction between the 249,427 and 505,488 grid
The outer wall of the heat sink is used as an adiabatic boundary numbers is 2.48 %, demonstrating that the finer grid with 249,427 cells
condition. is sufficient to meet the computational accuracy.
In order to simplify the computational complexity, symmetric As shown in Fig. 8, the variation of liquid fraction over time is
boundary conditions are used. compared for three different time step size settings. It is found that the
∂T ∂T ∂ui ∂ui maximum deviation of the liquid fraction between a step size of 0.05 s
= = 0, = =0 (15) and 0.5 s is 2.82 %, demonstrating that a time step size of 0.5 s is suf­
∂x ∂y ∂x ∂y
ficient to meet the computational accuracy.
For the cooling process, the bottom of the heat sink stops providing Based on the apparent heat capacity method, numerical simulations
heat, and the top is cooled by active air cooling of 3.5 m/s. are carried out using the solid and fluid heat transfer model in COMSOL
The initial condition of the computational domain: Multiphysics 6.0. The laminar flow model is coupled with the volume
The initial liquid fraction is θPCM = 0. For most simulations, the
initial temperature (Tinitial) is set to 300 K. For intermittent numerical
and experimental research, Tinitial is set to 302 K.

Fig. 6. The geometric model of numerical simulation. Fig. 8. Time step independence test.

6
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

force to simulate natural convection. A transient study model is selected, Table 5

and the calculation tolerance is set as the physics-controlled option. The Level and range of design variables.
time stepping method of the transient solver is set to backward differ­ Parameter Level
ence formula. To ensure the convergence of time step size, the solver
− 1 0 1
step size is set to be precise, the time step is set to 0.5 s, and the
maximum step size constraint is set to auto. For non-linear controller w1 1 2 3
w3 1 2 3
consistent initialization, backward Euler method with an initial step size C 0.3 0.35 0.4
fraction of 0.01 is chosen. A segregated solver approach within the
transient solver is used for computation, with a maximum iteration
count set to 30, a tolerance factor set to 0.5, and the stopping criteria set working endurance as a phase change heat sink structure. When opti­
to solution. The stabilization and acceleration models are Anderson mizing the design of a heat sink structure, one or more criteria should be
acceleration, with an iteration space dimension of 5, a mixing parameter satisfied optimally under certain conditions. Therefore, appropriate
of 0.9, and a delay of 1. The velocity and pressure fields adopt the direct performance evaluation indicators should be selected before optimizing
fluid flow variable model, while the temperature field adopts the direct the design of the heat sink structure.
heat transfer variable model. The direct solvers in the above algorithms The objective functions are as follows:
use the PARDISO solver for solving linear equations, with a constant
(Newton) nonlinear method and a damping coefficient of 0.8. The Ja­ (1) Entransy:
cobian matrix is updated once per time step, and the termination tech­
nique is set to iteration with a count of 1. Based on the analogies between thermal and electrical conductions,
Guo et al. [37,38] proposed the physical quantity entransy, which re­
flects the ability of heat transfer. It is defined as half of the product of
3.5. Numerical model validation
heat capacity and temperature. The mathematical formula for its defi­
nition is as follows:
To validate the existing numerical model, it is crucial to compare the
simulation results with reliable data. In this study, the simulation results 1
Eh = Qvh Tf (16)
of Zahid Ahmed Qureshi et al. [33] are used for model validation. The 2
physical parameters, geometric model, and boundary conditions in this Where Qvh is the heat capacity, Tf is the temperature of fluid.
validation are consistent with the reference simulation. Fig. 9 presents During the heating process of the PCM-based heat sink, the base of
the results of the average PCM temperature for both cases: with a the heat sink is in the highest temperature region. To ensure that the
Primitive-type TPMS structure and with PCM only. The results indicate heat sink operates within a specified safe temperature range, if the base
that the numerical model can accurately predict the heat transfer pro­ temperature is relatively low during the heat transfer process, it in­
cess of PCM-based heat sinks. dicates a higher heat transfer efficiency of the heat sink. To evaluate the
endurance capacity of the heat sink during the heating process, the
3.6. Multi-objective optimization concept of “entransy” is introduced. In the entire thermal storage pro­
cess, the entransy of the base of the heat sink is defined as follows:
To investigate the influence of changing TPMS structure on heat 1 1
transfer performance, the parameters w1, w3, and C are selected as E1 = QA ⋅Ta = Qi ⋅tmelt • Ta (17)
2 2
optimization variables. The levels and ranges of the structural design
parameters are shown in Table 5. During this process, the bottom of the Where QA(J) is the input heat capacity of the bottom, Ta(K) is the
heat sink is heated at a constant power of 30 W, and the effect of natural base temperature of the heat sink, Qi(W) is the heating power of the
convection is taken into account at the top of the heat sink. bottom, and tmelt(s) is the melting time.
When evaluating the performance of heat sink structures, factors
such as safety, reliability, and economy should be considered. The most (2) Heat Storage Rate:
important considerations are the excellent heat storage capacity and
During the thermal storage process of the PCM-based heat sink, the
melting rate of the PCM can express the heat transfer capability within
the heat sink. A faster melting rate indicates a stronger heat transfer
capability of the heat sink. The conventional definition of liquid fraction
neglects the actual mass of the PCM and cannot accurately reflect the
melting rate. By introducing the concept of heat storage rate, the entire
process of completely melting all PCMs is considered as the thermal
storage process. The heat storage rate is defined as the ratio of the total
heat storage capacity to the thermal storage time.
Qs mf • L
φ= = (18)
tmelt tmelt
Where Qs is the heat storage capacity, tmelt is the melting time, mf is
the mass of fluid, and L is the latent heat of the phase change.
Table 6 presents the design schemes given by BBD and the corre­
sponding numerical simulation results of the heat sink. The BBD
experimental design scheme consists of 17 design points with three
variables and three levels, including five repeated center points. The
entransy E1 and the heat storage rate φ are performance indicators of the
PCM-based heat sink. From Table 6, it can be observed that the nu­
merical simulation results of the BBD experimental design scheme yield
a range of entransy values from 20.4939 to 24.2767, and a range of heat
Fig. 9. The model verification results.

7
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Table 6
The scheme and results of numerical simulation by BBD.
Design parameters CFD results
NO.
w1 w3 C E1(105J⋅K) φ(J/s)

1 3 2 0.4 20.4939 17.7

2 2 2 0.35 21.4268 17.3
3 2 2 0.35 21.4268 17.3
4 2 1 0.3 22.2939 17.2
5 1 2 0.3 24.2767 16
6 2 2 0.35 21.4268 17.3
7 1 3 0.35 23.2058 16.6
8 2 2 0.35 21.4268 17.3
9 3 2 0.3 21.6150 18.3
10 3 1 0.35 21.5754 18
11 2 1 0.4 21.6150 17.2
12 2 3 0.3 22.5225 17.5
13 1 2 0.4 22.8390 16.2
14 2 3 0.4 20.8399 17.5
15 3 3 0.35 20.84625 18.1
16 2 2 0.35 21.4268 17.3
17 1 1 0.35 23.8680 16.1

Fig. 10. Comparison of numerical and experimental results of base tempera­

storage rate values from 16 to 18.3. The simulation results indicate that ture at P2.
the entransy does not necessarily decrease with an increase in the heat
storage rate, demonstrating that the two do not exhibit a linear heat sink is heated for 150 s with a constant power of 30 W, and the top
relationship. is cooled for 900 s by active air cooling after heating stops. It can be
By performing multiple regression analysis on the variable values observed from Fig. 10 that the experimental base temperature data are
and numerical simulation results from Table 6, a regression model be­ in general agreement with the numerical base temperature data, with
tween the design variables and the objective function can be obtained. the maximum temperature difference occurring around 1050 s during
The multiple quadratic regression equations of objective functions the cooling stage, reaching 4.22 K with an error of 1.34 %. This period
are as follows: corresponds to the solidification process, and ideal adiabatic conditions
E1 = 37.86961 − 4.59878 • w1 + 0.630187•w3 − 50.69737 • C are applied in the simulation. However, in the actual experiment, some
heat dissipation from the outer wall of the heat sink accelerates the
− 0.016744 • w1• w3 + 1.58288•w1 • C − 5.01862 • w3 • C (19) cooling of the base, leading to errors.
Fig. 11 displays the visualized experimental (melting front) and
+ 0.7177221w1 2 + 0.229397w3 2 + 64.66875C2 numerical simulation results, demonstrating good consistency between
them. As shown in Fig. 11, the white color in the experimental results
φ = 9.5 + 3.3 • w1 − 0.05 • w3 + 21 • C − 0.1 • w1 • w3 − 4 • w1 • C represents the solid state of the PCM, while transparency indicates the
melting of the PCM. With the progress of melting, the upper solid PCM is
− 1.1447E − 08 • w3 • C − 0.2 • w1 2 + 0.1 • w3 2 − 20 • C2 (20) gradually heated and melted through natural convection and heat con­
duction. Under the combined effect of natural convection and buoyancy,
To optimize the objective function using the NSGA-II algorithm, it is
the melting expands toward the upper-middle region.
found by comparison that when the population size exceeds 100 and the
Fig. 12 displays the visualized experimental (temperature) and nu­
number of generations exceeds 400, the obtained Pareto optimal solu­
merical simulation results, demonstrating good consistency between the
tion set tends to stabilize. Therefore, for this optimization in this study, a
two. Fig. 12 shows that the relatively low-temperature region in the
population size of 200, 500 generations, a crossover and mutation
upper part gradually diminishes, indicating the attenuation of thermal
probability of 0.9, and an individual mutation probability of 0.5 are set.
When optimizing the design variables, the objective is to minimize the
entransy E1 and maximize the heat storage rate φ. Objective optimiza­
tion constraints are as follows:
⎧
⎪
⎪ Maximizeφ = Y1 (w1 , w3 , C)
⎪
⎪ Minimize E1 = Y2 (w1 , w3 , C)
⎪
⎪
⎪
⎪
⎨ 1 ≤ w1 ≤ 3
1 ≤ w3 ≤ 3 (21)
⎪
⎪
⎪
⎪ 0.3 ≤ C ≤ 0.4
⎪
⎪
⎪
⎪ tc ≥ 350s
⎩
Tmax ≤ 338K
Where tc is the time of temperature control at the maximum allow­
able temperature. Tmax is the maximum allowable temperature.

4. Results and discussion

4.1. Validation of the numerical model by experimental study

Fig. 10 shows the comparison of numerical simulation and temper­

ature experiment results of base temperature at P2. The experiment is
Fig. 11. Comparison of numerical and experimental data of melting front
conducted at an ambient temperature of 29 ◦ C, and the bottom of the during the heating process.

8
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

. .

Fig. 12. Comparison of numerical and experimental data of temperature during the heating process.

Table 7
Comparison of heat transfer performance based on heat sinks of different
models.
results
Model w1 w3 C
E1 φ

105J⋅K J/s

MInitial 1 2 0.3 24.28 16

MA1 3 3 0.3 21.42 18.51
↓(11.8 %) ↑(13.6 %)
MB1 3 3 0.3562 20.53 18.1
↓(15.4 %) ↑(11.6 %)
MC1 2.7976 3 0.4 20.01 17.7
↓(17.6 %) ↑(9.6 %)

= 3, C:0.3562), and C1(w1:2.7976, w2:2, w3:3, C:0.4) in Fig. 13. The

performance parameters of these models are compared with those of the
initial design model, as shown in Table 7. Based on the analysis provided
in Table 7, it is observed that compared to the initial model, model A1
exhibits the largest increase in heat storage rate, with an improvement of
13.6 %, while the decrease in entransy is minimal, with a reduction of
Fig. 13. Pareto optimal solution set.
11.8 %. On the other hand, model C1 achieves the smallest decrease in
entransy, with a reduction of 17.6 %, but the smallest increase in heat
storage capacity of the PCM. Eventually, the internal temperature of the storage rate, with an improvement of 9.6 %. Model B1 demonstrates a
entire PCM-based heat sink becomes more uniform. significant decrease in entransy by 15.4 % and a notable increase in heat
storage rate by 11.6 %. Compared to models A1 and C1, model B1 ex­
hibits better comprehensive performance. The CFD calculation for
4.2. Studies on multi-objective optimization on the charging cycle
model B1 results in a temperature control time of 365 s under a tem­
perature of 338 K. Therefore, based on the comprehensive analysis,
Fig. 13 illustrates the Pareto optimal solution set obtained from
model B1 is determined as the improved model.
multi-objective optimization. It can be observed that there is a positive
correlation between the entransy E1 and the heat storage rate φ. The
point A1 has the minimum entransy and the minimum heat storage rate, 4.3. Studies on the improved-model based heat sink on the charging cycle
while the point C1 has the maximum heat storage rate but the maximum
entransy. The objective of this optimization is to minimize the entransy Fig. 14 presents the temperature evolution of the base in the heating
E1 and maximize the heat storage rate φ. By observing Fig. 13, it can be process and the corresponding liquid fraction contours for different
seen that the slope in the A1B1 segment is relatively large, approximately stages based on the improved model. The overall phase change heat
0.54. In this segment, as the entransy increases, the heat storage rate transfer process can be divided into three stages: solid sensible heat
experiences a significant increase. On the other hand, in the B1C1 absorption stage I1, latent heat absorption stage J1, and liquid sensible
segment, the slope is relatively small, around 0.21. In this segment, as heat absorption stage K1. In stage I1 (0 s-50 s), PCM remains solid, and
the entransy increases, the increase in heat storage rate is relatively the solid PCM and the heat sink metal absorb heat. In stage J1 (50 s-350
smaller. s), as indicated by the liquid fraction contour, most of the PCM gradually
The numerical simulation calculations are performed for the models melts. During this process, PCM undergoes a phase transition from solid
corresponding to points A1(w1:3, w2:2, w3:3, C:0.3), B1(w1:3, w2 = 2, w3 to liquid, absorbing a significant amount of heat through latent heat,

9
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Fig. 14. Evolution curve of base temperature with time on the charging cycle.
Fig. 16. The evolution of liquid fraction.

Fig. 15. The evolution of base temperature.

Fig. 17. Comparison of effective thermal conductivity results.
causing a slower increase in the base temperature. In stage K1 (350 s-
650 s), as evident from the liquid fraction contour, PCM is completely the heat sink, and an initial temperature of 300 K for the heat sink
melted, and the PCM without latent heat capacity leads to a rapid rise in structure, the effective thermal conductivity of the heat sink is obtained
the base temperature. using Fourier’s law. Taking the structure named PCM Only (heat sink
Fig. 15 compares the base temperature evolution of the improved with only PCM) as an example, under these calculation conditions, the
model with the initial model. It can be observed that the improved calculated effective thermal conductivity is 0.21 W/(m⋅K), with an error
model achieves better temperature control during the heating process, of less than 5 % compared to the PCM thermal conductivity of 0.2 W/
with the base temperature consistently lower than that of the initial (m⋅K), indicating the accuracy of the calculations. Observing Fig. 17, the
model. Using 338 K as the temperature control limit, the improved effective thermal conductivity of the initial model is 5.7 W/(m⋅K), while
model has an additional 45 s of temperature control compared to the the improved model achieves an effective thermal conductivity of 8 W/
initial model. Fig. 16 presents the comparison of the time evolution of (m⋅K), representing a 40 % improvement in the effective thermal con­
the liquid fraction of PCM based on the improved model and the initial ductivity after optimization.
model. It can be observed that during the melting process, PCM based on Fig. 18 presents the temperature and liquid fraction contours of the
the improved model melts faster with a higher liquid fraction, saving 90 initial and improved models at 300 s. Observation of Fig. 18(a) reveals
s compared to the initial model for achieving complete melting. that the initial model exhibits higher base temperature with a lower
Fig. 17 displays a bar chart comparing the effective thermal con­ temperature region in the lower-middle portion, indicating an uneven
ductivity of the improved and initial models. Under steady-state con­ temperature distribution. In contrast, the improved model shows
ditions, with adiabatic outer walls, a uniform heat flux of 10 W/m2 at the significantly reduced base temperature and a more uniform temperature
bottom surface of the heat sink, natural convection at the top surface of distribution. Fig. 18(b) indicates complete PCM melting near the heat

10
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Fig. 18. Internal cross-sectional contours.

source and the cells in the initial model. However, a considerable

portion of PCM in the low-temperature region still has a relatively lower
liquid fraction. The improved model demonstrates improved liquid
fraction in the low-temperature region, with almost complete melting of
PCM in the lower-middle portion and an overall higher level of liquid
fraction.

4.4. Research on heat transfer performance of PCM based heat sink on

the intermittent cycle

Fig. 19 illustrates the intermittent power cycle, where the heater is

operated at 30 W for 150 s and the top surface of the heat sink is exposed
to natural convection during the cooling phase. The cooling process

Fig. 20. Evolution curve of base temperature for the intermittent process.

involves air-cooling the top surface of the heat sink, and there is no
heating applied for 900 s.
Fig. 20 depicts the time evolution of the base temperature for the
improved model during the intermittent process. It can be observed that
the base temperature increases with each cycle, and the rate of increase
in peak temperature slows down in the fourth cycle compared to the
preceding cycles. By the end of the fourth cycle, the base temperature
reaches a relatively stable state, with peak temperature below 340 K and
valley temperature below 320 K. This indicates that the intermittent
process can meet the required temperature control of 338 K.
Fig. 21 presents the time evolution of the liquid fraction of PCM
during the intermittent process. It is observed that PCM remains
Fig. 19. The intermittent power cycle.

11
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Fig. 23. The internal fluid Grashof number changes during the intermit­
Fig. 21. Evolution curve of liquid fraction for the intermittent process. tent cycle.

partially melted during the first two cycles and transitions to a solid­ stage. With the progress of the intermittent cycle, the liquid fraction
–liquid mixing zone after the cooling phase. After the heating in the increases continuously, leading to a reduction in the gradient difference
third cycle, PCM continues to melt internally through natural convection of the liquid fraction, indicating a stabilizing growth of the liquid frac­
and conductive heat transfer until complete melting. After the cooling tion in the heat sink.
phase, PCM transitions from a liquid state to a solid–liquid mixing zone. Fig. 23 presents the time evolution of the wall Grashof number (Gr)
By the end of the fourth cycle, the liquid fraction of the heat sink reaches and internal fluid velocity contours of the heat sink based on the
a relatively stable state, with peak liquid fraction at 1.0 and valley liquid improved model. Grashof number is a dimensionless quantity used to
fraction around 0.8. measure the ratio of buoyancy force to viscous force in natural con­
Fig. 22 showcases the internal cross-sectional temperature and liquid vection phenomena. Physically, a higher Grashof number indicates a
fraction contours of the heat sink at different time instances during the relatively stronger buoyancy force. Previous studies have shown that
intermittent cycle. It can be observed that heating results in higher using the Grashof number as a criterion for determining the transition of
temperatures in the lower-middle portion of the heat sink, and as the heat transfer behavior overcomes the limitations associated with the
cooling process progresses, the temperature gradient within the heat usage of the Reynolds number for forced convection scenarios. In this
sink diminishes, leading to a more uniform temperature distribution. study, the Grashof criterion is employed as the determinant for heat
However, due to the initial concentration of temperature in the lower- transfer behavior transition. The correlation equation is as follows:
middle portion and the lower overall liquid fraction, there are still sig­
nificant differences in the liquid fraction distribution during the cooling gβl1 3 ΔT (22)
Gr =
v2

Fig. 22. Internal cross-sectional temperature contours (left) and liquid fraction contours (right) at different time instances during the intermittent cycle.

12
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

Where g is the acceleration of gravity, m/s2; β is the thermal during the heating phase translates to an increase in input heat. With
expansion coefficient, 1/K; l1 is the characteristic length, m; ΔT is the higher power and larger input heat, and unchanged heat dissipation
temperature difference between the upper and lower surfaces of the capacity, the heat sink temperature rises accordingly. In the case of 40 W
fluid in the vertical direction along the heating surface, K; v is the ki­ operating condition, the input heat exceeds the dissipated heat, resulting
nematic viscosity m2/s. in difficulties in controlling the heat sink temperature during the
Observing Fig. 23, it can be seen that as the intermittent cycle pro­ intermittent process. Conversely, under lower power conditions, where
gresses, the Grashof number (Gr) reaches its maximum value in the third the input heat is less than or equal to the dissipated heat, temperature
cycle. The fluid velocity cloud map shows that the fluid velocity is higher controllability is more effective. Therefore, reducing the heating power
in the upper part of the heat sink. At this point, the PCM in the lower part maintains the base and internal temperatures of the heat sink at lower
of the heat sink is melting, and under the action of buoyancy, natural levels, while excessively high heating power makes it difficult to control
convection develops towards the upper part. After the third cycle, the the temperature rise.
temperature difference between the upper and lower surfaces of the
fluid decreases, resulting in a decrease in the trough Gr compared to the 5. Conclusion
previous cycles. In the fourth cycle, the PCM is almost completely
melted at the peak moment, and the velocity cloud map shows that the In this work, the scheme of using TPMS and PCM-based active­
internal fluid velocity is evenly distributed, and natural convection –passive cooling heat sink is proposed to control the temperature of
weakens, resulting in a decrease in the Gr. After the fourth cycle, the electronic equipment. Using structural parameters of Primitive-type
temperature difference between the upper and lower surfaces of the TPMS as design variables, the structural optimization design of the
fluid tends to balance, and the Gr evolves to a stable state, with the peak TPMS is conducted with the aim of minimizing entransy E1 and maxi­
around 1000 and the trough below 200. mizing heat storage rate φ. Based on the improved model, the evolution
trend of the heat performance of the heat sink and the effects of heating
4.5. Experimental study of the effect of heating power on the intermittent power on the temperature of the heat sink during intermittent cycle are
cycle studied. Finally, the patterns of evolution of the heat performance (base
temperature, liquid fraction, and Grashof number) and the effects of
Fig. 24 shows the temperature evolution curve at point P2 and P6 for heating power are obtained during intermittent cycle. The main con­
different heating power. The improved model is used for the experiment. clusions are given as follows:
And experiments are conducted under four heating power conditions:
10 W, 20 W, 30 W, and 40 W.The intermittent cycle under 30 W is shown 1. The improved model achieves significant reductions in entransy and
in Fig. 19. In the intermittent cycles under other power levels, only the improvements in heat storage rate compared to the initial model,
power during the heating process is adjusted, while the remaining with a 15.4 % reduction in entransy E1 and an 11.6 % increase in heat
conditions remain the same. storage rate φ.
From Fig. 24, it can be seen that increasing the power results in an 2. The thermal performance of the heat sink is improved by applying
increase in the temperature during the intermittent cycle. For example, the PCM and optimized TPMS structure. Compared to the heat sink
in the last cycle, the maximum temperature at point P2 exceeds 370 K based on the initial model, the improved heat sink exhibits an
and the minimum temperature exceeds 340 K for the power of 40 W, extended temperature control time of 45 s, a shortened melting time
with a rising trend. Decreasing the power reduces the peak and trough of 90 s, and an increased effective thermal conductivity of 40 %.
temperatures. Taking the temperature at point P2 under 20 W and 30 W 3. During the intermittent cycle, the thermal performance of the
as examples, the temperature stabilizes between 310 K and 330 K for 20 active–passive cooling heat sink gradually achieves a steady periodic
W, and between 314 K and 340 K for 30 W. P6 is located within the heat variation during the intermittent process. Specifically, the base
sink, and its temperature rise rate is relatively slower compared to that temperature remains stable in the range of 314 K to 340 K, the liquid
at the base. When the power is less than or equal to 30 W, the temper­ fraction stabilizes between 0.8 and 1.0, and the Gr stabilizes between
ature fluctuations remain stable. At 30 W, the temperature of P6 remains 100 and 1000.
around 315 K. However, when the power is increased to 40 W, the 4. When the heating power results in excessive input heat, the base
subsequent temperature exhibits a rapid upward trend, with the temperature at P2 of the heat sink becomes uncontrollable. At the
maximum temperature exceeding 350 K. power of 30 W, the temperature fluctuation of P2 remains stable,
Under the same intermittent cycle, an increase in heating power ranging from 314 K to 340 K. However, at the power of 40 W, the

Fig. 24. The temperature evolution of different measuring points at different heating power during the intermittent cycle.

13
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

maximum temperature of the heat sink exceeds 370 K, and there is Transfer 91 (2018) 274–281, https://doi.org/10.1016/j.
icheatmasstransfer.2017.12.014.
still a trend of temperature increase.
[13] S.A.B. Al-Omari, Z.A. Qureshi, E. Elnajjar, F. Mahmoud, A heat sink integrating fins
5. P6 is located within the heat sink, and its temperature rise rate is within high thermal conductivity phase change material (PCM) to cool high heat-
relatively slower compared to that at the base. At the power of 30 W, flux heat sources, Int. J. Therm. Sci. 172 (2022) 107190, https://doi.org/10.1016/
the temperature fluctuations remain stable, ranging from 310 K to j.ijthermalsci.2021.107190.
[14] Y. Lin, Y. Jia, G. Alva, G. Fang, Review on thermal conductivity enhancement,
315 K. However, at the power of 40 W, the subsequent temperature thermal properties and applications of phase change materials in thermal energy
exhibits a rapid upward trend, with the maximum temperature storage, Renew. Sust. Energ. Rev. 82 (2018) 2730–2742, https://doi.org/10.1016/
exceeding 350 K. j.rser.2017.10.002.
[15] A. Mustaffar, A. Harvey, D. Reay, Melting of phase change material assisted by
expanded metal mesh, Appl. Therm. Eng. 90 (2015) 1052–1060, https://doi.org/
CRediT authorship contribution statement 10.1016/j. applthermaleng.2015.04.057.
[16] K.Y. Leong, M.R.A. Rahman, B.A. Gurunathan, Nano-enhanced phase change
materials: a review of thermo-physical properties, applications and challenges,
Jiabin Wang: Conceptualization, Software, Validation, Formal J. Energy Storage 21 (2019) 18–31, https://doi.org/10.1016/j.est.2018.11.008.
analysis, Investigation, Writing – original draft, Project administration. [17] S. Chavan, Veershetty Gumtapure, D. Arumuga Perumal, Computational
Wenhao Pu: Methodology, Writing – review & editing, Project admin­ investigation of bounded domain with different orientations using CPCM,
J. Storage Mater. 22 (2019) 355–372, https://doi.org/10.1016/j.est.2019.02.018.
istration. Haisheng Zhao: Methodology, Supervision, Project adminis­ [18] S. Chavan, Veershetty Gumtapure, D. Arumuga Perumal, Performance assessment
tration. Long Qiao: Formal analysis, Data curation. Nanxin Song: of composite phase change materials for thermal energy storage-characterization
Investigation, Data curation. Chen Yue: Methodology, Supervision, and simulation studies, Recent Patents on Mechanical Engineering 14 (1) (2021)
75–85.
Project administration. [19] T.-U. Rehman, H.M. Ali, M.M. Janjua, U. Sajjad, W.-M. Yan, A critical review on
heat transfer augmentation of phase change materials embedded with porous
materials/foams, Int. J. Heat Mass Transf. 135 (2019) 649–673, https://doi.org/
Declaration of competing interest 10.1016/j.ijheatmasstransfer. 2019.02.001.
[20] B.V.S. Dinesh, A. Bhattacharya, Effect of foam geometry on heat absorption
characteristics of PCM-metal foam composite thermal energy storage systems, Int.
The authors declare that they have no known competing financial J. Heat Mass Transf. 134 (2019) 866–883, https://doi.org/10.1016/j.
interests or personal relationships that could have appeared to influence ijheatmasstransfer. 2019. 01.095.
the work reported in this paper. [21] S. Abishek, A. King, N. Nadim, B. Mullins, Effect of microstructure on melting in
metal-foam/paraffin composite phase change materials, Int. J. Heat Mass Transf.
127 (2018) 135–144, https://doi.org/10.1016/j.ijheatmasstransfer.2018.07.054.
Data availability [22] W. Li, S. Liu, et al., Experimental and numerical study on temperature control
performance of phase change material heat sink, Appl. Therm. Eng. 238 (2024)
122081. ISSN1359-4311.
Data will be made available on request.
[23] Z.A. Qureshi, S.A.B. Al-Omari, E. Elnajjar, O. Al-Ketan, R.A. Al-Rub, On the effect
of porosity and functional grading of 3D printable triply periodic minimal surface
Acknowledgement (TPMS) based architected lattices embedded with a phase change material, Int. J.
Heat Mass Transf. 183 (2022) 122111, https://doi.org/10.1016/j.
ijheatmasstransfer.2021.122111. ISSN 0017-9310.
The authors acknowledge financial support from the project funded [24] O. Al-Ketan, R. Rowshan, R.K.A. Al-Rub, Topology-mechanical property
by the National Natural Science Foundation of China (51606105) and relationship of 3D printed strut, skeletal, and sheet based periodic metallic cellular
materials, Add. Manuf. 19 (2018) 167–183, https://doi.org/10.1016/j.
the Science Center for Gas Turbine Project (P2021-A-I-001-002).
addma.2017.12.006.
[25] D.J. Yoo, Porous scaffold design using the distance field and triply periodic
References minimal surface models, Biomaterials 32 (31) (2011) 7741–7754, https://doi.org/
10.1016/j. biomaterials.2011.07.019.
[26] S.C. Kapfer, S.T. Hyde, K. Mecke, C.H. Arns, G.E. Schroder-Turk, Minimal surface
[1] Y. Tomizawa, K. Sasaki, A. Kuroda, et al., Experimental and numerical study on
scaffold designs for tissue engineering, Biomaterials 32 (29) (2011) 6875–6882,
phase change material (PCM) for thermal management of mobile devices[J], Appl.
https://doi.org/10.1016/j.biomaterials.2011.06.012.
Therm. Eng. 98 (2016) 320–329.
[27] M. Afshar, A.P. Anaraki, H. Montazerian, J. Kadkhodapour, Additive
[2] M.M.A. Khan, R. Saidur, F.A. Al-Sulaiman, A review for phase change materials
manufacturing and mechanical characterization of graded porosity scaffolds
(PCMs) in solar absorption refrigeration systems[J], Renew. Sust. Energ. Rev. 76
designed based on triply periodic minimal surface architectures, J. Mech. Behav.
(2017) 105–137.
Biomed. Mater. 62 (2016) 481–494, https://doi.org/10.1016/j.
[3] C. Amaral, R. Vicente, P. Marques, et al., Phase change materials and carbon
jmbbm.2016.05.027.
nanostructures for thermal energy storage: a literature review[J], Renew. Sust.
[28] A. Castro, T. Pires, J. Santos, B. Gouveia, P. Fernandes, Permeability versus design
Energ. Rev. 79 (2017) 1212–1228.
in TPMS scaffolds, Materials 12 (8) (2019) 1313, https://doi.org/10.3390/
[4] N.I. Ibrahim, F.A. Al-Sulaiman, S. Rahman, et al., Heat transfer enhancement of
ma12081313.
phase change materials for thermal energy storage applications: a critical review
[29] N. Sreedhar, N. Thomas, O. Al-Ketan, R. Rowshan, H. Hernandez, R.K.A. Al-Rub, H.
[J], Renew. Sust. Energ. Rev. 74 (2017) 26–50.
A. Arafat, 3D printed feed spacers based on triply periodic minimal surfaces for flux
[5] C. Zeng, S. Liu, A. Shukla, Adaptability research on phase change materials-based
enhancement and biofouling mitigation in RO and UF, Desalination 425 (2018)
technologies in China[J], Renew. Sust. Energ. Rev. 73 (2017) 145–158.
12–21, https://doi.org/10.1016/j.desal.2017.10.010.
[6] J.P.D. Cunha, P. Eames, Thermal energy storage for low and medium temperature
[30] S. Catchpole-Smith, R. Śelo, A. Davis, I. Ashcroft, C. Tuck, A. Clare, Thermal
applications using phase change materials-a review[J], Appl. Energy 177 (2016)
conductivity of TPMS lattice structures manufactured via laser powder bed fusion,
227–238.
Add. Manuf. 30 (2019) 100846, https://doi.org/10.1016/j.addma.2019.100846.
[7] K. Pielichowska, K. Pielichowski, Phase change materials for thermal energy
[31] D.E.N.G. Xing, L.I.U. Fei, Z.H.A.N.G. Zhengwen, et al., The effect of periodic porous
storage[J], Prog. Mater. Sci. 65 (2014) 67–123.
Al on the heat transfer performance of paraffin[J], IOP Conference Series: Earth
[8] L.-W. Fan, Wu. Yu-Yue, Y.-Q. Xiao, Y.i. Zeng, Y.-L. Zhang, Zi-Tao Yu, Transient
and Environmental Science 680 (1) (2021) 012069.
performance of a thermal energy storage-based heat sink using a liquid metal as the
[32] O. Al-Ketan, M. Ali, M. Khalil, R. Rowshan, K. Khan, A.-R. Abu, Forced convection
phase change material, Appl. Therm. Eng. Volume 109 (Part A) (2016) 746–750.
CFD analysis of architected and 3D printable heat sinks based on triply periodic
ISSN 1359–4311.
minimal surfaces, J. Thermal Sci. Eng. Appl. (2020) 1–33, https://doi.org/
[9] Z. Rao, S. Wang, G. Zhang, Simulation and experiment of thermal energy
10.1115/1.4047385.
management with phase change material for ageing LiFePO4 power battery[J],
[33] Zahid Ahmed Qureshi, Salah Addin Burhan Al-Omari, Emad Elnajjar, Oraib Al-
Energ. Conver. Manage. 52 (12) (2011) 3408–3414.
Ketan, Rashid Abu Al-Rub, Using triply periodic minimal surfaces (TPMS)-based
[10] S. Al Omari, A. Ghazal, E. Elnajjar, Z.A. Qureshi, Vibration-enhanced direct contact
metal foams structures as skeleton for metal-foam-PCM composites for thermal
heat exchange using gallium as a solid phase change material, Int. Commun. Heat
energy storage and energy management applications, International
Mass Transfer 120 (2021) 104990 p, https://doi.org/10.1016/j.
Communications in Heat and Mass Transfer, Volume 124,2021,105265, ISSN
icheatmasstransfer.2020.104990.
0735-1933, https://doi.org/10.1016/j. icheatmasstransfer.2021.105265.
[11] S.A. Al Omari, A. Ghazal, E. Elnajjar, A new approach using un-encapsulated
[34] Zahid Ahmed Qureshi, Emad Elnajjar, Oraib Al-Ketan, Rashid Abu Al-Rub, Salah
discrete PCM chunks to augment the applicability of solid gallium as phase change
Burhan Al-Omari, Heat transfer performance of a finned metal foam-phase change
material in thermal management applications, Energy Convers. Manage. 158
material (FMF-PCM) system incorporating triply periodic minimal surfaces
(2018) 133–146, https://doi.org/10.1016/j.enconman.2017.12.002.
(TPMS), International Journal of Heat and Mass Transfer, Volume 170, 121001,
[12] S.A. Al Omari, A. Ghazal, E. Elnajjar, A novel concept to enhance the applicability
ISSN 0017–9310 (2021), https://doi.org/10.1016/j. ijheatmasstransfer.
of solid gallium as phase change material for heat sinks by integrating within it
2021.121001.
discretely distributed chunks of un-encapsulated PCM, Int. Commun. Heat Mass

14
J. Wang et al. Applied Thermal Engineering 254 (2024) 123864

[35] C. Ji, Z. Qin, Z. Low, et al., Non-uniform heat transfer suppression to enhance PCM [37] Z.Y. Guo, H.Y. Zhu, X.G. Liang, Entransy—A physical quantity describing heat
melting by angled fins[J], Appl. Therm. Eng. 129 (2018) 269–279. transfer ability, Int. J. Heat Mass Transfer 50 (13–14) (2007) 2545–2556, https://
[36] S. Moench, R. Dittrich, Influence of natural convection and volume change on doi.org/10.1016/j.ijheatmasstransfer.2006.11. 034.
numerical simulation of phase change materials for latent heat storage, Energies [38] Z.Y. Guo, X.G. Cheng, Z.Z. Xia, Least dissipation principle of heat transport
(19961073). 15 (8) (2022) 2746, https://doi.org/10.3390/en15082746. potential capacity and its application in heat conduction optimization, Chin. Sci.
Bull. 48 (4) (2003) 406–410, https://doi.org/10.1360/03tb9085.

15