applied

sciences
Article
Design and Optimization of Cooling Plate for Battery
Module of an Electric Vehicle
Ben Ye, Md Rashedul Haque Rubel and Hongjun Li *
Faculty of Mechanical Engineering and Automation, Zhejiang Sci-Tech University, Hangzhou 310018, China;
850644662@163.com (B.Y.); rashed019iubat@gmail.com (M.R.H.R.)
* Correspondence: lihongjun@zstu.edu.cn




Received: 17 January 2019; Accepted: 18 February 2019; Published: 21 February 2019


Abstract: With the development of electric vehicles, much attention has been paid to the thermal
management of batteries. The liquid cooling has been increasingly used instead of other cooling
methods, such as air cooling and phase change material cooling. In this article, a lithium iron
phosphate battery was used to design a standard module including two cooling plates. A single
battery numerical model was first created and verified as the basis of the module heat transfer model.
Orthogonal experimental design method was adopted in the module thermal model to optimize
the main parameters in the module: Battery gap, the cross-section size, and the number of coolant
channels of the cooling plate. The Surrogate Model was then utilized to further optimize geometry
of the cooling plate. Finally, the optimized geometry was rebuilt in the module thermal model for
analysis. The comparison showed that the maximum and minimum temperature difference in the
cooling plate was reduced by 9.5% and the pressure drop was reduced by 16.88%. It was found
that the battery temperature difference and the pressure drop decreased with the increase of the
cross-section and number of the coolant channel when the coolant flow rate was constant at the inlet.

Keywords: battery module; design of experiment; surrogate model; optimization

1. Introduction
As an important type of energy storage units, lithium batteries have been developed for many
years and their performance has been greatly improved. They have been gradually applied to artificial
satellites, robots, and electric vehicles (EV) [1–4]. Nowadays, more and more electric vehicles are being
produced, and lithium batteries are being widely used, but a lot of the heat in the process of using
has appeared [5]. The service life, capacity, and internal resistance of the batteries are sensitive to
temperature changes. In order to prolong the service life of the battery, it is particularly important to
design a battery module with good heat dissipation performance.
Air cooling can meet the requirements of the vehicles in common conditions, but the battery
temperature will be higher. When the EV changes the speed frequently or it is at high velocity,
the battery is discharging at a high rate, air cooling is unable to meet the requirements of cooling
alone [6,7]. Although the cooling effect can be optimized by changing the position, number, and angle
of the entrance, adding a guide plate and changing the battery arrangement, it is still difficult to meet
the needs of different electric vehicles [8–10]. Phase change material (PCM) cooling system controls
the temperature of the battery module by the heat absorption and heat release when its material phase
changes. Power battery cooling experiments using PCM are easier to meet the needs of the lithium
battery cooling system, but the cost is high, and it is used less in electric vehicles [11–13].
Since the introduction of a liquid cooling system with high cooling efficiency and reliability [14–16],
it has gradually occupied the electric vehicle market. The liquid cooling system in the BMW I3 and the
Tesla model S, have good sealing and reliability, and can take away the heat of each battery evenly and

Appl. Sci. 2019, 9, 754; doi:10.3390/app9040754 www.mdpi.com/journal/applsci

Appl. Sci. 2019, 9, 754 2 of 20

exhibit good performance in electric vehicles. Patil et al. [17] studied the cooling performance of 20 Ah
lithium-ion pouch cell with cold plates along both surfaces by changing the inlet coolant mass flow
rates and the inlet coolant temperatures. The enhanced cooling energy efficiency was achieved with
a low inlet coolant temperature, low inlet coolant mass flow rate, and a high number of the cooling
channels. Panchal et al. [18] experimentally investigated the temperature and velocity distributions
within the mini-channel cold plates placed on a prismatic lithium-ion battery cell using water cooling
methods. Wang et al. [19] carried out experimental and simulations to study the effect of cooling
channels, flow rates, and flow directions at different discharge C-rates. It was found that the maximum
temperature reached within the battery decreased as the amount of thermal silica plates and liquid
channels increased. Wang et al. [20] designed a new liquid cooling strategy based on thermal silica
plates combined with the cooling effect of water. The experimental results demonstrated that the
addition of thermal silica plates can greatly improve the cooling capacity.
In this paper, a lithium iron phosphate battery was used to design a standard module which can
be quickly interchanged by EV, and then the liquid cooling plate for the module was analyzed by
numerical heat transfer analysis. A surrogate model was utilized to further optimize the geometry of
the cooling plate.

2. Thermal Analysis of a Single Battery

The governing equations which were used to solve the time dependent three-dimensional flow
problems include the continuity equation, momentum equation, and the energy equation. The equation
of state was given in Equations (1)–(5) [21–24]:
Continuity equation
∂ρ
+ ∇·(ρu) = 0 (1)
∂t
X-momentum
∂(ρu) ∂P
+ ∇·(ρuu) = − + ∇·(µ∇u) + S Mx (2)
∂t ∂x
Y-momentum
∂(ρv) ∂P
+ ∇·(ρvu) = − + ∇·(µ∇v) + S My (3)
∂t ∂y
Z-momentum
∂(ρw) ∂P
+ ∇·(ρwu) = − + ∇·(µ∇w) + S Mw (4)
∂t ∂z
Energy equation
dTc
mC p = ∇·(λc ∇ Tc ) + Q g (5)
dt
The main working parameters of the lithium iron phosphate battery [25] are shown in Table 1.

Table 1. Working parameters of lithium batteries.

Parameters Values
Nominal voltage (V) 3.2
Nominal capacity (Ah) 10
Internal resistance (mΩ) ≈10
Charging current (A) ≤ 10
Continuous discharge current (A) ≤ 20
Maximum discharge current (A) 50
Upper cut-off voltage (V) 3.65 ± 0.05
Lower cut-off voltage (V) 2.5
Cycle life (/) ≥ 2000
Weight (g) 275 ± 5
Dimension (mm) 131 × 65 × 16
Appl. Sci. 2019, 9, 754 3 of 20

The interior part of the battery was simplified as an equivalent solid model and the following
assumptions were made for the model:
(1) The material properties in lithium batteries were uniformly distributed. Because of the
multi-layer structure and manufacturing process of lithium batteries, only the thermal conductivity
was anisotropic;
(2) Thermal radiation and convection can be neglected inside the lithium battery;
(3) The specific heat capacity and thermal conductivity of materials in the lithium batteries were
constant and independent of the temperature;
(4) When the battery was charged and discharged, the current and heat generation were considered
uniformly distributed.
It was difficult to accurately obtain the heat generation rate of batteries due to the complexity of
vehicle operating conditions and environment. The internal resistance of the battery was assumed
constant under ambient temperature, and the battery resistance was set to 10 mΩ. According to the
classic model proposed by Bernardi et al. [26], the heating generation rate of batteries was established
below in Equation (6). The polarization heat, chemical reaction heat and the electrode cap heat was not
considered in the model.
I ∂U0
q = [(U0 − U ) − T ] (6)
V ∂T
where, V represents the volume of the battery, in cubic meters, U0 , the open circuit voltage in volts, U,
the working voltage of the battery in volts, T, the temperature in Kelvin, ∂U 0
∂T is measured experimentally,
the value is very small at room temperature with a low discharge rate and can be neglected.
Therefore, Equation (6) can be simplified and expressed as follows:

I I2 (RP + Re ) I2 R
q= (U298.15 − U) = = (7)
V V V
U298.15 represents the open circuit voltage of the battery in the temperature of 298.15 K, in volts;
R is total internal resistance which is obtained by the internal resistance, Re , and the polarization
internal resistance, Rp , in ohms.
In addition to the complex heat generation inside the battery, the heat generation outside the
battery will also occur, such as the positive and negative electrodes, the confluent and the welding
position of the conductor. These heats can be neglected when studying the heat generation of the
battery. Equation (7) was used in this paper to estimate the heat generation of the battery. The heat
generation rate of the lithium battery at 2C discharge rate was 29, 359.953W/m3 . C-rate is the
measurement of the charge and discharge current with respect to its nominal capacity. Considering the
experimental environment and the boundary conditions of simulation, the Boussinesq hypothesis was
used for the calculation: (1) The dissipation of fluid viscosity was neglected during the process of fluid
flow; (2) except the fluid density, other thermal properties were constant with varying temperature;
(3) for density, only the terms related to volume force in momentum equation were included and the
temperature of 25 ◦ C was used as the reference temperature for calculation. The thermal properties of
the battery are shown in Table 2:

Table 2. Thermophysical parameters [25].

Density Specific Heat Capacity Thermal Conductivity Viscosity

Name
(kg·m−3 ) (J·kg−1 ·k−1 ) (W·m−1 ·k−1 ) (kg·m−1 ·s−1 )
Cell 1958.7 733 0.9/2.7/2.7 -
Air 1.185 1005 0.0263 0.0000184

The dimension of the battery is 131mm × 65mm × 16mm, the positive and negative electrode
columns were not included in the model. The simplified model of the battery was established by
CATIA as shown in Figure 1b, the fluid field was created according to the cooling method in Ge’s
 Cell 1958.7 733 0.9/2.7/2.7 -
Air 1.185 1005 0.0263 0.0000184

The dimension of the battery is 131mm × 65mm × 16mm, the positive and negative electrode
Appl. Sci. 2019,were
columns 9, 754 not included in the model. The simplified model of the battery was established 4 of 20by
CATIA as shown in Figure 1b, the fluid field was created according to the cooling method in Ge’s
experiment [16], as shown in Figure 1c. In the analysis, the natural cooling process of the lithium
experiment
battery was [16], as shown
simulated at in
2CFigure 1c. In
discharge thefor
rate analysis,
a periodtheofnatural
1800 s. cooling process ofevolution
The temperature the lithiumwas
battery was simulated at 2C discharge rate for a period of 1800 s. The temperature evolution
monitored and outputted at the end of each time step. The results of the cell surface temperature after was
monitored
1800 s at and 25 °Coutputted
are shown at in
theFigure
end of2.each
The time step. The
maximum results ofwas
temperature the 325.8
cell surface
K and temperature
located in the
after 1800 s at 25 ◦ C are shown in Figure 2. The maximum temperature was 325.8 K and located
central area of the battery surface. Surface temperature decreased gradually to the periphery. The
inlowest
the central area of the
temperature battery
of the surface.
battery Surface
was 322.2 K attemperature decreased
the four corners. gradually todifference
The temperature the periphery.
of the
The lowest
whole temperature
battery of the
surface was 3.6battery
K. was 322.2 K at the four corners. The temperature difference of
the whole battery surface was 3.6 K.

(a) (b) (c)

Appl. Sci. 2019, 9, x FOR PEER REVIEW 2 of 26
Figure1.1.(a)
Figure (a)Battery
Batterycell;
cell；(a) cell
(b) cell geometry；(b)
geometry; (c) heatheat transfer
transfer model.
model.

The velocity streamline diagram of air in a natural convection condition surrounding the lithium
battery is shown in Figure 3. It can be seen that air flew from boundaries of air domain to the surface
of the batteries, and buoyancy increased with the increase of the temperature on the surface of the
battery. The air velocity in the central region leaving the battery surface with the highest temperature
was 0.1362 m/s.

Figure 2. Surface temperature.

Figure 2. Surface temperature.
The velocity streamline diagram of air in a natural convection condition surrounding the lithium
battery is 2019,
Appl. Sci. shown9, x;in Figure
doi: 3. ItREVIEW
FOR PEER can be seen that air flew from boundaries of air domain to the surface
www.mdpi.com/journal/applsci
of the batteries, and buoyancy increased with the increase of the temperature on the surface of the
battery. The air velocity in the central region leaving the battery surface with the highest temperature
was 0.1362 m/s.
To verify the thermal model of the single battery, comparison with the experimental results in
paper [25] was plotted in Figure 4. The surface temperature was measured by a K-type thermocouple
during the experiment. During the discharging process, the temperature at the monitoring point
gradually increased with time. The highest temperature in the experiment and numerical analysis
were 53.84 ◦ C and 51.39 ◦ C, respectively, implying a difference within 5%. The reasons for this error
 Appl. Sci. 2019, 9, 754 5 of 20

Appl. Sci. 2019, 9, x FOR PEER REVIEW 3 of 26

may result from the assumptions made in the simulation and the heat generation equation did not
take into account the polarization heat, chemical reaction heat, and the electrode cap heat.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 26

and the heat generation equation did not take into account the polarization heat, chemical
reaction heat, and the electrode cap heat.
Figure 3. Streamline of air velocity.
Figure 3. Streamline of air velocity.

To verify the thermal model of the single battery, comparison with the experimental results in
paper [25] was plotted in Figure 4. The surface temperature was measured by a K-type thermocouple
during the experiment. During the discharging process, the temperature at the monitoring point
gradually increased with time. The highest temperature in the experiment and numerical analysis
were 53.84°C and 51.39°C, respectively, implying a difference within 5%. The reasons for this error
may result from the assumptions made in the simulation

Figure 4. Temperature curve of simulation and experiment.

Figure 4. Temperature curve of simulation and experiment.
The cell discharged at 25 ◦ C without cooling, the maximum temperature can reach 53.84 ◦ C,
Thewas
which cellhigher
discharged at optimum
than the 25 °C without cooling,
operating the maximum
temperature range oftemperature can reach
the battery. Because 53.84
of the °C,
various
which was higher than the optimum operating temperature range of the battery. Because
usage conditions of the EV battery, high power discharge will occur inevitably. When power batteries of the
various usage conditions
are assembled of the EV battery,
in large quantities, the heathigh power discharge
dissipation efficiencywill occurlow.
becomes inevitably. When
To solve power
this problem,
batteries are assembled
it is necessary in large
to design quantities,
a standard batterythemodule
heat dissipation efficiency
and incorporate becomessystem
a cooling low. Totosolve
ensurethis
the
problem,
workingitenvironment
is necessary temperature
to design a of
standard battery module and incorporate a cooling system to
batteries.
ensure the working environment temperature of batteries.
3. Design of Standard Battery Module
3. Design of Standard
48 volts Battery
was chosen Module
as the standard voltage for the battery module. The voltage of the lithium
iron48phosphate
volts waspower
chosenbattery
as theisstandard
3.2 V, and 15 batteries
voltage were
for the usedmodule.
battery in a module. According
The voltage to dimension
of the lithium
iron phosphate power battery is 3.2 V, and 15 batteries were used in a module. According to
dimension of traction battery for electric ve
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 26

hicles, GB/T 34013-2017 (China standard), the following arrangement scheme was preliminarily
designed, as shown in Figure 5.
Appl. Sci. 2019, 9, 754 6 of 20
The liquid cooling plate of the battery module was made of an aluminum plate with a thickness
of 2 mm. According to the scheme, the sizes of the aluminum plate and the fluid channel patterns
were determined
of traction battery for as shown
electric in Figure
vehicles, 6. The
GB/T geometry
34013-2017 of thestandard),
(China entrance the
andfollowing
exit was 1arrangement
mm (height) ×
15 mm (width), and the fluid channels were symmetrically
scheme was preliminarily designed, as shown in Figure 5. arranged in the aluminum plate. For the
initial design, all the parallel channels were the same, and the size was 1 mm (height) × 5 mm (width).

Batterymodule.
Figure5.5.Battery
Figure module.

The liquid cooling plate of the battery module was made of an aluminum plate with a thickness
of 2 mm. According to the scheme, the sizes of the aluminum plate and the fluid channel patterns were
determined as shown in Figure 6. The geometry of the entrance and exit was 1 mm (height) × 15 mm
Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 26
(width), and the fluid channels were symmetrically arranged in the aluminum plate. For the initial
design, all the parallel channels were the same, and the size was 1 mm (height) × 5 mm (width).

Appl. Sci. 2019, 9, x FOR PEER REVIEW 7 of 26

Figure 6. Initial design of cooling plate.

CATIA was employed to build Figurethe 3-dimensional
6. Initial battery
design of cooling module. The module had fifteen
plate.
lithium
CATIA was employed to build the 3-dimensional battery module.7.The
batteries arranged in the form of a 1 × 15, as shown in Figure Themodule
batteries were
had connected
fifteen lithium
in series, and
batteries the total
arranged voltage
in the form of
of the
a 1 × 15, as shown in Figure 7. The batteries were connected in series,
and the total voltage of the module was 48 V. Cooling plates were placed on the top and bottom sides
module
of the battery.was
At48 V.same
the Cooling plates
time, were gel
the silica placed
pads onwere
the top andbetween
added bottom sides of the battery.
the cooling At the
plates and the
same time,tothe
batteries silica gel
improve pads
the heatwere added
transfer betweenas
efficiency, the cooling
shown in plates
Figureand the batteries
8. The dimension to improve the
of the whole
heat transfer
module was efficiency,
288 mm ×as 131shown
mm × in71
Figure
mm. 8. The dimension of the whole module was 288 mm × 131
mm × 71 mm.

(a) (b)
Figure
Figure7.7.(a)
(a)Simplified
SimplifiedCATIA
CATIAmodel;
model;(b)
(b)battery
batteryarrangement.
arrangement.

In this study, 50% ethylene glycol and water mixed solution was used as the coolant, whose
working temperature range was 5–35 ◦ C. The flow in the channels of the cooling plates was considered
as laminar flow. The thermal properties of the coolant at 25 ◦ C were tabulated in Table 3. This work
used the ANSYS/FLUENT [27] to model the fluid zone and the solid zone was built, respectively,
the model was meshed as shown in Figure 9, and the size of the grids was chosen in such a way that
the resultant temperature and pressure drop were independent of the grid size. The inlet flow velocity
was 0.5m/s, and the heat generation rate of each battery at 2C discharge rate was 29,359.953W/m3 .
Appl. Sci. 2019, 9, 754 7 of 20

Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 26

The inlet coolant temperature was fixed at 25 ◦ C.
To model the heat convection at the contact surfaces
among cooling plate, silicone pad and battery, the couple walls were established in the interfaces.
Radiation heat transfer was not considered in this model.

Figure 8. Detailed module structure.

Table 3. ThermalFigure 8. Detailed

properties module
of the Silica gelstructure.
pad and mixed solution.

Density
In this study, 50% Specific
ethylene glycolHeat
andCapacity Thermal
water mixed Conductivity
solution was used asDynamic Viscosity
the coolant, whose
Properties
(kg/m3 ) (J/kg·K) (W/m·K) (kg/m·s)
working temperature range was 5–35°C. The flow in the channels of the cooling plates was
Silica gel pad
considered as laminar3050 0.832
flow. The thermal properties of the coolant 3at 25°C were tabulated- in Table 3.
Mixed solution 1071.11 3300 0.384 0.00339
ThisSci.
Appl. work
2019, used the
9, x FOR ANSYS/FLUENT
PEER REVIEW [27] to model the fluid zone and the solid zone was9 built,
of 26
respectively, the model was meshed as shown in Figure 9, and the size of the grids was chosen in
such a way that the resultant temperature and pressure drop were independent of the grid size. The
inlet flow velocity was 0.5m/s, and the heat generation rate of each battery at 2C discharge rate was
29,359.953W/m3. The inlet coolant temperature was fixed at 25 °C. To model the heat convection at
the contact surfaces among cooling plate, silicone pad and battery, the couple walls were established
in the interfaces. Radiation heat transfer was not considered in this model.

Table 3. Thermal properties of the Silica gel pad and mixed solution.

Specific heat Thermal Dynamic

Density
Properties capacity conductivity viscosity
（kg/m3）
（J/kg·K） （W/m·K） （kg/m·s）
Silica gel pad 3050 0.832 3 -
Mixed solution 1071.11 3300 0.384 0.00339

Figure 9. module
Module mesh.

In order to verify the grid independence, different grid sizes were selected to discretize the
model. The result of the grid independence study was plotted in Figure 10. Both the pressure drops
and the temperature difference of the cooling plate tended to be stable at the grid of 6.7 × 106. The
mesh sizes used for the following
 Appl. Sci. 2019, 9, 754 8 of 20

In order to verify the grid independence, different grid sizes were selected to discretize the
Appl. Sci. 2019, 9, x FOR PEER REVIEW
model.
10 of 26
The result of the grid independence study was plotted in Figure 10. Both the pressure drops and the
temperature
study were difference
4 mm, of0.5themm,
cooling
andplate tended
0.3 mm fortothe
be battery,
stable at silica
the grid
gelofpad, 106 . cooling
6.7 ×and The mesh sizes
plate,
used for the
respectively. following study were 4 mm, 0.5 mm, and 0.3 mm for the battery, silica gel pad, and cooling
plate, respectively.

Figure 10. Mesh independence study.

Figure 10. Mesh independence study.
The orthogonal experimental design strategy was chosen in the numerical analysis to screen the
optimal configuration
The orthogonal and shapedesign
experimental parameters
strategyof was
the battery
chosen module. Three factors
in the numerical and
analysis to three
screenlevels
the
were carried
optimal out to determine
configuration theparameters
and shape effect of eachoffactor on the module.
the battery pressure Three
difference between
factors the inlet
and three and
levels
were carried out to determine the effect of each factor on the pressure difference between the inletin
outlet of the cooling plate (pressure drop), the surface temperature difference of the cooling plate
contact
and outletwith thecooling
of the silica gel pad(pressure
plate (plate temperature
drop), thedifference), and the temperature
surface temperature difference ofdifference among
the cooling plateall
inthe batterywith
contact surfaces in thegel
the silica module (battery
pad (plate temperaturedifference),
temperature difference).and
As illustrated in Tabledifference
the temperature 4, the cross
section of each channel, battery spacing, and number of channels were taken as design
among all the battery surfaces in the module (battery temperature difference). As illustrated in Table factors with
4,three levels.
the cross section of each channel, battery spacing, and number of channels were taken as design
factors with three levels.
Table 4. Factors and levels in orthogonal experimental design.

OrthogonalFactors
test results were presented in Table 5. In Table 6, kij (i = A,B,N,C; j = 1, 2, 3) was the
average Levels A
value of the calculated results shown in Table 5 Bfor each factor at the same C level, and the
(Cross Section, mm) (Battery Spacing, mm) (Number of Channels)
equation of range analysis can be given as follow:
1 1×3 2 5
2 1×4 R = max(k ) − min3(k ) 6 (1)
3 1×5 4 7
In the Equation (8), Ri was the range value of the evaluation index.
Orthogonal test results were presented in Table 5. In Table 6, kij (i = A,B,N,C; j = 1, 2, 3) was the
The results of the orthogonal test and the range value of three evaluation indexes were tabulated
average value of the calculated results shown in Table 5 for each factor at the same level, and the
in Tables 5–7. Factor A was the main factor in the three indexes with the largest range, indicating the
equation of range analysis can be given as follow:
greatest influence on the indexes, while factor B had the least impact on the three evaluation indexes.



Ri = max kij − min kij (8)

In the Equation Table

(8), Ri4.was
Factors and levels in orthogonal experimental design.
the range value of the evaluation index.
The results of the orthogonal test and the range value of three evaluation indexes were tabulated
Factors
A the main factor in the three
in Tables 5–7. Factor A was B indexes with the largest range,
C indicating the
greatest (Cross section, mm) （Battery spacing, mm） (Number of channels)
Levels influence on the indexes, while factor B had the least impact on the three evaluation indexes.
1 1×3 2 5
2 1×4 3 6
3 1×5 4 7

Table 5. Orthogonal test results.

Appl. Sci. 2019, 9, 754 9 of 20

Table 5. Orthogonal test results.

Test Pressure Battery Temperature Plate Temperature

A B N C
Number Drop, kPa Difference, ◦ C Difference, ◦ C
1 1 1 1 1 7.59 6.4 2.4
2 1 2 2 2 6.64 6.1 2.1
3 1 3 3 3 6.39 6.2 2.2
4 2 1 2 3 4.50 5.9 1.9
5 2 2 3 1 5.81 6.1 2.0
6 2 3 1 2 5.37 6.0 1.9
7 3 1 3 2 4.28 5.9 1.9
8 3 2 1 3 3.97 5.9 1.9
9 3 3 2 1 5.28 6.1 2.1

Table 6. Range analysis of the evaluation indexes.

Evaluation Indexes A B N C
1 2.233 2.067 2.067 2.167
Plate temperature
difference 2 1.933 2.000 2.033 1.967
3 1.967 2.067 2.033 2.000
1 6.233 6.067 6.100 6.200
Battery temperature
difference 2 6.000 6.033 6.033 6.000
3 5.967 6.100 6.067 6.000
1 6.873 5.457 5.643 6.227
Pressure drop
2 5.227 5.473 5.473 5.430
3 4.510 5.680 5.493 4.953

Table 7. The range values of the evaluation indexes.

Evaluation Indexes A B N C
Pressure drop 2.363 0.223 0.170 1.274
Battery temperature difference 0.266 0.067 0.067 0.200
Plate temperature difference 0.300 0.067 0.034 0.200

The effect of factors in different levels on indexes were presented in Figure 11. For the main
factor A, the pressure drop index decreased sharply from level 1 to level 3. The battery temperature
difference decreased as well, but the slope became smaller. The plate temperature difference decreased
from level 1 to level 2 and then increased slightly from level 2 to level 3. Therefore, the level A3 is best.
The trend of factor C was the same as the main factor A, and the level C3 was chosen. Factor B has the
least influence on all indexes. Considering the compactness of the battery module, the level B1 was
preferred. The optimal scheme, A3 B1 C3 was implemented for the following study.
The accuracy of the numerical analysis was verified by comparing the theoretical cooling fluid
temperature rise with the simulation value, as shown in Equation (9):

Nφ0
∆tcoolant = = 1.13 ◦ C (9)
2min Cp

∆tcoolant is the temperature difference for coolant; N is the number of batteries; φ0 is the heat
generated per unit time of the cell, φ0 = I2 R; min is the flow rate of the coolant at the entrance.
The numerically predicted outlet temperature rise was 1.1 ◦ C, and very close to the theoretical
value. The surface temperature distribution of 15 batteries in the battery module was shown in
Figure 12. It can be seen that the maximum temperature of the battery surface was 32.3 ◦ C and the
 7 3 1 3 2 4.28 5.9 1.9
8 3 2 1 3 3.97 5.9 1.9
9 3 3 2 1 5.28 6.1 2.1

difference decreased as well, but the slope became smaller. The plate temperature difference
Appl. Sci. 2019, 9, 754 10 of 20
decreased from level 1 to level 2 and then increased slightly from level 2 to level 3. Therefore, the
level A3 is best. The trend of factor C was the same as the main factor A, and the level C3 was chosen.
minimum temperature
Factor B has was 26.4 ◦on
the least influence C. The high temperature
all indexes. zonethe
Considering was close to the of
compactness outlet
the in the middle
battery module, of
the
themodule. The preferred.
level B1 was minimum The
temperature was located
optimal scheme, A3B1on
C3 the
wassurface of the battery
implemented for thenear the inlet
following of the
study.
cooling plate and the battery temperature difference was 5.9 ◦ C.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 12

in Figure 14. The lowest temperature on plate surface was 25.2 °C at the inlet end, while the hi
temperature was at the plate corners near the outlet end, with a temperature difference of 1.9 °
in large, the plate surface temperature gradually increased from inlet end to out end. The highes
lowest temperatures of each battery were plotted in Figure 15, and the maximum temper
Effect
Figure 11.was
difference of factors
less than 5 in
°C.different levels on indexes.

Figure 11. Effect of factors in different levels on indexes.

The accuracy of the numerical analysis was verified by comparing the theoretical cooling fluid
temperature rise with the simulation value, as shown in equation (9):
Nϕ
∆t = = 1.13℃ (2)
2m C
∆t is the temperature difference for coolant; N is the number of batteries; ϕ is the heat
generated per unit time of the cell, ϕ = I R; m is the flow rate of the coolant at the entrance.
The numerically predicted outlet temperature rise was 1.1°C, and very close to the theoretical
Figure 12. Temperature
12. Temperature distribution
distribution on battery surface
Figure
value. The surface temperature distribution of 15 batteries on battery
in the surface.
battery module was shown in Figure
12. It can be seen that the maximum temperature of the battery surface was 32.3 °C and the minimum
The pressure distribution of the cooling plate was shown in Figure 13. The pressure distribution
temperature was 26.4 °C. The high temperature zone was close to the outlet in the middle of the
and pressure drop for each channel were similar, revealing good flow consistency. The pressure drop
module. The minimum temperature was located on the surface of the battery near the inlet of the
of the cooling plate was 3.85 kPa. The temperature distribution on the cooling surface was illustrated
cooling plate and the battery temperature difference was 5.9 °C.
in Figure 14. The lowest temperature on plate surface was 25.2 ◦ C at the inlet end, while the highest
The pressure distribution of the cooling plate was shown in Figure 13. The pressure distribution
temperature was at the plate corners near the outlet end, with a temperature difference of 1.9 ◦ C.
and pressure drop for each channel were similar, revealing good flow consistency. The pressure drop
By in large, the plate surface temperature gradually increased from inlet end to out end. The highest
of the cooling plate was 3.85 kPa. The temperature distribution on the cooling surface was illustrated
and lowest temperatures of each battery were plotted in Figure 15, and the maximum temperature
Appl. Sci. 2019, 9, x FOR PEER REVIEW
difference was less than 5 ◦ C.

Figure 13. Pressure distribution of the13.

Figure cooling plate.
Pressure distribution of the cooling plate
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 14 of 26
Appl. Sci. 2019, 9, 754 11 of 20

Appl. Sci. 2019, 9, x FOR PEER REVIEW 15 of 26

Figure 14. Surface temperature distribution of the cooling plate.

Figure 14. Surface temperature distribution of the cooling plate.

Figure 15. Battery temperature.

Figure 15. Battery temperature.
4. Surrogate Model-Based Optimization
4. Surrogate
Apart fromModel-based
the analysis ofOptimization
the main parameters of the lithium battery modules in the above section,
the arrangement
Apart from andthegeometric
analysisdetails
of the of channels
main also had
parameters of influence on the
the lithium battery
battery temperature
modules in the and
above
thesection,
cooling the
platearrangement
temperature.andIn this section, the channels in the cooling plate were further optimized
geometric details of channels also had influence on the battery
bytemperature
the surrogateand model of the plate
the cooling cooling plate in the
temperature. Inworkbench
this section,platform [28,29].
the channels The
in the flow plate
cooling chart were
of
optimization was illustrated
further optimized in Figuremodel
by the surrogate 16. of the cooling plate in the workbench platform [28,29]. The
Geometric
flow parameterization
chart of optimization was carried
was illustrated out on16.
in Figure a quarter of the cooling plate, as shown in
Figure 17. L1 , L2 , L3 , H1 and H2 were the geometric parameters of the cooling plate and the range of
the geometric parameters were tabulated in Table 8.
In the surrogate model, only the cooling plate was considered. To obtain the thermal load on the
plate surface, heat flux distribution in the fluent model on the plate surface was first calculated as
shown in Figure 18, and then heat flux values were extracted for seven straight lines on the surface of
the cooling plate facing the batteries, as illustrated in Figure 19. The extracted values of seven lines
were curve fitted by quadratic polynomial and the fitted curves were shown in Figure 20. All the heat
flux curves increased towards the center of the plate and reached peak values at Z = 0 mm. The peak
value at X = −126 mm was greater than that at X = 126 mm because the temperature difference between
the coolant and plate at inlet was greater than that at the outlet.
Appl. Sci. 2019, 9, 754 12 of 20
Appl. Sci. 2019, 9, x FOR PEER REVIEW 16 of 26

Parameterization of
geometric model

Determination of Determination of
optimization objectives design variables

Establishment of
analytical model

Design of
experiment

Fitting surrogate models

Optimization

Comparisons of calculated results with Fluent

Appl. Sci. 2019, 9, x FOR PEER REVIEW numerical models 17 of 26

Figure 16. Flow chart of surrogate model-based optimization.

Figure 16. Flow chart of surrogate model-based optimization.

Geometric parameterization was carried out on a quarter of the cooling plate, as shown in Figure
17. L1, L2, L3, H1 and H2 were the geometric parameters of the cooling plate and the range of the
geometric parameters were tabulated in Table 8.

Figure 17. Geometric parameters of the cooling plate (1/4).

Figure 17. Geometric parameters of the cooling plate (1/4).
Table 8. Geometric parameters of the cooling plate.
Table 8. Geometric parameters of the cooling plate.
Parameters
Parameters L1 L2 L3 H1 H2
(mm)
L1 L2 L3 H1 H2
(mm)Initial value 17.5 35.5 52.5 15 15
Initial
Uppervalue
limit value 25 17.5 43 35.5 60 52.515 15
17.5 15
UpperLower limit value
limit value 10 25 28 43 45 6012.5 15
15 17.5
Lower limit value 10 28 45 12.5 15
In the surrogate model, only the cooling plate was considered. To obtain the thermal load on the
plate surface, heat flux distribution in the fluent model on the plate surface was first calculated as
shown in Figure 19, and then heat flux values were extracted for seven straight lines on the surface
of the cooling plate facing the batteries, as illustrated in Figure 18. The extracted values of seven lines
were curve fitted by quadratic polynomial and the fitted curves were shown in Figure 20. All the heat
flux curves increased towards the center of the plate and reached peak values at Z = 0 mm. The peak
value at X = −126 mm was greater than that at X = 126 mm because the temperature difference between
the coolant and plate at inlet was greater than that at the outlet.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 18 of 26
Appl. Sci. 2019, 9, 754 13 of 20

Appl. Sci. 2019, 9, x FOR PEER REVIEW 19 of 26

Figure 18.
Figure Line location
18. Line location for
for heat
heat flux
flux extraction.
extraction.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 20 of 26

Figure 19. Heat flux on the surface of the plate.

Figure 19. Heat flux on the surface of the plate.

Figure 20. Fitting curves of heat fluxes at different location.

Figure 20. Fitting curves of heat fluxes at different location.

The heat flux obtained above was expressed in the following equations. These functions were
written as User Defined Functions (UDF) files and interpreted as thermal boundary conditions in
fluent.
q （ ） = 923.7 + 2.8Z − 21171. 1Z (3)
q （ ） = 914.4 − 1.9Z − 17800.7Z (4)
q （ ） = 914.0 + 1.9Z − 17325.3Z (5)
Appl. Sci. 2019, 9, 754 14 of 20

The heat flux obtained above was expressed in the following equations. These functions were
written as User Defined Functions (UDF) files and interpreted as thermal boundary conditions in fluent.

qflux(−126) = 923.7 + 2.8Z − 21171.1Z2 (10)

qflux(−108) = 914.4 − 1.9Z − 17800.7Z2 (11)

qflux(−54) = 914.0 + 1.9Z − 17325.3Z2 (12)

qflux(0) = 913.7 − 3.0Z − 17670.5Z2 (13)

qflux(54) = 912.3 − 1.3Z − 18340.0Z2 (14)

qflux(108) = 909.8 + 1.4Z − 19588.5Z2 (15)

qflux(126) = 914.8 + 1.5Z − 21417.0Z2 (16)

where, qflux was the heat flux on the cooling plate surface; Z was the coordinate variable along the
coordinate Z direction;
According to the battery number shown in Figure 7b, the fitting results of X = 0 mm were applied
to the corresponding regions of the cooling surface as the thermal boundary conditions of the cell
numbers 7, 8, and 9. The fitting results of X = 54 mm were applied to the corresponding regions of the
cell numbers 10, 11, and 12. The fitting results of X = −54mm were applied as the thermal boundary
conditions of the cell numbers 4, 5, and 6. The fitting results of X = −108mm were applied to the
corresponding region of batteries No.2 and No.3. The fitting results of X = 108 mm were applied to the
corresponding region of batteries No.13 and No.14. The fitting results of X = −126 mm and X = 126 mm
were applied to the corresponding regions of batteries No.1 and No.15, respectively. The heat flux
between two batteries on the cooling plate was set to a constant value of 300 W/m2 .
The simplified cooling plate was imported into workbench and the parameters were set.
The maximum temperature on the surface of the cooling plate and the pressure drop of the cooling
plate were taken as the output parameters.
In the experimental design [30–32] of surrogate models, a reasonable number of sample points
was selected, which can reflect the spatial characteristics of the design in a limited design space by
using the mathematical method. The quality of sample points determined the accuracy of the fitted
model directly. The central composite experimental design was utilized as the sampling method for all
concerned. Surrogate model [33,34] was an approximate mathematical model using an approximate
method to fit the discrete data (sample points). The response and change of the target were predicted
by design variables in this model. By comparison with the numerical model (Figure 9) in Section 2,
the surrogate model can significantly reduce the computational cost. The sample points obtained
by the central composite experimental design was fitted by Kriging model [35]. The determination
coefficient was 1 and the root mean square error was 1.7538 × 10−7 and 3.3085 × 10−10 in the fitting
quality of temperature and pressure response. The fitting accuracy was acceptable and the model
was feasible.
The sensitivity of the plate maximum temperature and pressure drop in response to the parameters
were shown in Figure 21. The influence of L1 and H1 on the pressure drop was more significant than
other parameters. The channel distance parameters affected the plate temperature difference in the
sequence of L3 > L2 > L1 . Surrogate models with parameter combinations of L1 , H1 and pressure
drop and combination of L3 , L2 , and surface temperature were established, respectively. As shown in
Figure 22, the parameters L2 and L3 in the surrogate model were positively correlated with temperature,
and the influence of L3 on temperature was greater than that of L2 . The range of temperature response
values was 26.8–27.6 ◦ C. In Figure 23, the parameters L1 and H1 were also positively correlated with
pressure drop and the range of pressure drop response values was 3.16–3.48 kPa.
 difference in the sequence of L3 > L2 > L1. Surrogate models with parameter combinations of L1, H1
and pressure drop and combination of L3, L2, and surface temperature were established, respectively.
As shown in Figure 22, the parameters L2 and L3 in the surrogate model were positively correlated
with temperature, and the influence of L3 on temperature was greater than that of L2. The range of
temperature response values was 26.8–27.6°C. In Figure 23, the parameters L1 and H1 were also
Appl. Sci. 2019, 9, 754 15 of 20
positively correlated with pressure drop and the range of pressure drop response values was 3.16–
3.48 kPa.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 22 of 26

Figure 21. Sensitivity of parameters.
Figure 21. Sensitivity of parameters.

Figure 22. L2 , L3 and temperature surrogate model.

Figure 22. L2, L3 and temperature surrogate model.
The 2 ◦ C temperature difference of the cooling plates was set as the constraint, and the minimum
pressure drop was taken as the objective to optimize this model.
Three candidate designs were obtained, and the optimal parameters were selected as shown
in Table 9. The optimized size of the cooling plate was presented in Table 10. The pressure drops
calculated was 3.2176 kPa and the maximum temperature of cooling plate was 26.99 ◦ C. Comparing
with the results, which were not optimized, the plate temperature difference and pressure drop were
reduced by 5.24% and 16.88%, respectively.
Appl. Sci. 2019, 9, 754 16 of 20

Table 9. Candidate points.

Candidate H1 H2 L1 L2 L3 Pressure Maximum

Point (mm) (mm) (mm) (mm) (mm) Drop (Pa) Temperature (◦ C)
1 12.861 15.089 10.995 41.832 55.503 3217.6 26.99
2 12.846 15.792 12.106 38.232 57.340 3226.1 26.99
3 13.356 16.041 11.201 36.216 59.964 3238.3 26.98

Table 10. Optimal parameter values.

Appl. Sci. 2019, 9, x FOR PEER REVIEW

Parameter H1 H2 L1 L2 L3 23 of 26

Value(mm) 12.50 15.00 11.00 42.00 55.50

Figure
Figure L1,LH11, and
23.23. H1 and pressure
pressure drop surrogate
drop surrogate model. model.

The
The numerical modeldifference
2 °C temperature (Figure 9)
of was then rebuilt
the cooling withsetthe
plates was optimized
as the parameters.
constraint, Battery surface
and the minimum
temperature
pressure drop distribution,
was taken asthe
thecooling
objectiveplate temperature
to optimize distribution and coolant pressure distribution
this model.
Three candidate
were shown in Figures designs
24–26,were obtained, It
respectively. and the
can beoptimal parameters
seen that were selected
the maximum as shown
temperature in cooling
of the
Tablewas
plate 9. The
26.9 ◦ C and the
optimized size of the cooling
pressure drop wasplate
3.2 was
kPa. presented
The errors in were
Table0.33%
10. The pressure
and 0.55% drops
in comparison
Appl. Sci. 2019, 9, x FOR PEER REVIEW 24 of 26
calculated
with was 3.2176
the surrogate kPa and
model the maximum
for the temperature
plate temperature of cooling
difference andplate was 26.99°C.
pressure Comparing
drop, respectively.
with the results, which were not optimized, the plate temperature difference and pressure drop were
reduced by 5.24% and 16.88%, respectively.

Table 9. Candidate points.

Candidate H1 H2 L1 L2 L3 Pressure Maximum

point (mm) (mm) (mm) (mm) (mm) drop (Pa) temperature (°C)
1 12.861 15.089 10.995 41.832 55.503 3217.6 26.99
2 12.846 15.792 12.106 38.232 57.340 3226.1 26.99
3 13.356 16.041 11.201 36.216 59.964 3238.3 26.98

Table 10. Optimal parameter values.

Parameter H1 H2 L1 L2 L3
Value(mm) 12.50 15.00 11.00 42.00 55.50
Figure 24. Battery surface temperature distribution with optimized geometry.
Figure
The numerical 24.(Figure
model Battery9)
surface temperature
was then distribution
rebuilt with with parameters.
the optimized optimized geometry.
Battery surface
temperature distribution, the cooling plate temperature distribution and coolant pressure
distribution were shown in Figures 24–26, respectively. It can be seen that the maximum temperature
of the cooling plate was 26.9°C and the pressure drop was 3.2 kPa. The errors were 0.33% and 0.55%
in comparison with the surrogate model for the plate temperature difference and pressure drop,
respectively.
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 25 of 26

Appl. Sci. 2019, 9, 754 17 of 20

Appl. Sci. 2019, 9, x FOR PEER REVIEW 26 of 26

Figure 25. Temperature distribution on cooling plate with optimized geometry.

Figure 25. Temperature distribution on cooling plate with optimized geometry.

Figure 26. Pressure distribution of the cooling plate with optimized geometry.
Figure 26. Pressure distribution of the cooling plate with optimized geometry.

5. Conclusions
5. Conclusions

The temperature resultsresults

The temperature predicted by by
predicted thethesingle
single battery thermal
battery thermal modelmodel
showedshowed good agreement
good agreement
with experiments by a difference less than 5%, implying that the heat generation and
with experiments by a difference less than 5%, implying that the heat generation model modelthe and the
assumptions were reasonable.
assumptions were reasonable.
A methodology for the design and optimization of the cooling plate for the battery module was
A methodology for the heat
proposed. A complex design andmodel
transfer optimization
for the whole ofmodule
the cooling plateincluding
was created, for the batteries,
battery twomodule was
proposed. A complex
cooling plates,heat transfer
silicone model
gel pads, for theOrthogonal
and coolant. whole module was design
experimental created,
wasincluding
implemented batteries,
by two
the numerical
cooling plates, siliconeanalysis
gel pads, to optimize the mainOrthogonal
and coolant. parameters of experimental
the module. The design
cooling plate
wasgeometry
implemented by
was further optimized by the surrogate model method. With the optimized geometry, the cooling
the numerical analysis to optimize the main parameters of the module. The cooling plate geometry
plate was rebuilt in the module thermal model for the analysis. The comparison showed that the
was furthermaximum
optimized andby the surrogate
minimum temperaturemodel method.
difference in theWith
coolingtheplate
optimized
was reducedgeometry,
by 5.24%the andcooling
the plate
was rebuiltpressure
in the module
drop was thermal
reduced bymodel
16.88%.for the analysis. The comparison showed that the maximum
and minimum Ittemperature
was concluded from the orthogonal
difference designplate
in the cooling analysis
was that the battery
reduced by temperature
5.24% and difference
the pressure drop
and the pressure drop decreased with the increase of the cross-section and number of the coolant
was reduced by 16.88%.
channel when the coolant flow rate was constant at the inlet. From the sensitivity analysis of the plate,
It wasthe
concluded from the orthogonal
maximum temperature design
and pressure drop analysis
in response thatplate
to the thegeometric
battery parameters
temperature in thedifference
and the pressure
surrogatedrop
models, decreased
it was found with
that the increase
the center of distance,
channel the cross-section and
L1, and the size of number of the coolant
the inlet plenum
exhibited
channel when the greatest
the coolant flowinfluence
rate wason the pressureat
constant drop.
the inlet. From the sensitivity analysis of the plate,
the maximumAuthor Contributions: Conceptualization,
temperature and pressure drop H.L. andinB.Y.; methodology,
response to theH.L.; software,
plate B.Y.; validation,
geometric B.Y.;
parameters in the
formal analysis, B.Y.; investigate
surrogate models, it was found that the center channel distance, L1, and the size of the inlet plenum
exhibited the greatest influence on the pressure drop.

Author Contributions: Conceptualization, H.L. and B.Y.; methodology, H.L.; software, B.Y.; validation, B.Y.;
formal analysis, B.Y.; investigation, B.Y. and M.R.H.R.; resources, H.L.; data curation, B.Y.; writing—original draft
Appl. Sci. 2019, 9, 754 18 of 20

preparation, B.Y. and M.R.H.R.; writing—review and editing, H.L. and M.R.H.R.; visualization, B.Y.; supervision,
H.L.; project administration, H.L.; funding acquisition, H.L.
Funding: This research was supported by the National Natural Science Foundation of China under Grant
No. 51605444.
Conflicts of Interest: The authors declare no conflict of interest

Nomenclature
A Factor in orthogonal experimental design, cross section
B Factor in orthogonal experimental design, battery spacing
C Factor in orthogonal experimental design, number of channels
Cp Specific heat capacity, J kg−1 K−1
∆tcoolant Temperature difference for coolant, ◦ C
H Geometric parameters at the inlet of the cooling plate
L Channel distance in the cooling plate
m Mass, kg
min Flow rate of the coolant at the entrance, kg s−1
N Number of batteries
P Pressure, Pa
Qg Generated heat
q Heat generation rate of batteries, W m−3
q f lux Heat flux on the cooling plate surface
R Total internal resistance, Ω
Re Internal resistance, Ω
Rp Polarization internal resistance, Ω
SMx , SMy , SMz Body forces
T Temperature, ◦ C
t Time, s
U Working voltage of the battery, V
U0 Open circuit voltage of the battery, V
U298.15 Open circuit voltage of the battery in the temperature of 298.15 K, V
u Flow velocity vector
u Velocity in x direction, m s−1
V Volume of the battery, m3
v Velocity in y direction, m s−1
w Velocity in z direction, m s−1
Z Coordinate Z direction, mm

Greek symbols
φ0 Heat generated per unit time of the cell, Ws−1
ρ Fluid density, kg m−3
λc Fluid viscosity, Pa s

Subscripts
c Cell
x x direction
y y direction
z z direction

Abbreviations
EV Electric vehicles
PCM Phase change material
UDF User defined functions
Appl. Sci. 2019, 9, 754 19 of 20

References
1. Tarascon, J.M.; Armand, M.B. Issues and Challenges Facing Rechargeable Lithium Batteries. Nature
2001, 414, 359–367. [CrossRef] [PubMed]
2. Kennedy, B.; Patterson, D.; Camilleri, S. Use of lithium-ion batteries in electric vehicles. J. Power Sources
2000, 90, 156–162. [CrossRef]
3. Wakihara, M. Recent developments in lithium ion batteries. Mater. Sci. Eng. R Rep. 2001, 33, 109–134.
[CrossRef]
4. Tamura, K.; Horiba, T. Large-scale development of lithium batteries for electric vehicles and electric power
storage applications. J. Power Sources 1999, 81–82, 156–161. [CrossRef]
5. Panchal, S.; Dincer, I.; Agelin-Chaab, M.; Fraser, R.; Fowler, M. Transient electrochemical heat transfer
modeling and experimental validation of a large sized LiFePO4/graphite battery. Int. J. Heat Mass Transf.
2017, 109, 1239–1251. [CrossRef]
6. Zolot, M.D.; Kelly, K.; Keyser, M.; Mihalic, M.; Pesaran, A. Thermal Evaluation of the Honda Insight
Battery Pack: Preprint. In Proceedings of the 36th Intersociety Energy Conversion Engineering Conference
(IECECí01), Savannah, GA, USA, 29 July–2 August 2001.
7. Kelly, K.J.; Mihalic, M.; Zolot, M. Battery usage and thermal performance of the Toyota Prius and Honda
Insight during chassis dynamometer testing. In Proceedings of the Seventeenth Annual Battery Conference
on Applications and Advances, Long Beach, CA, USA, 18 January 2002; pp. 247–252.
8. Pesaran, A.A.; Burch, S.D.; Keyser, M. An Approach for Designing Thermal Management Systems for Electric
and Hybrid Vehicle Battery Packs. In Proceedings of the Fourth Vehicle Thermal Management Systems
Conference and Exhibition, London, UK, 24–27 May 1999.
9. Park, H. A design of air flow configuration for cooling lithium ion battery in hybrid electric vehicles. J. Power
Sources 2013, 239, 30–36. [CrossRef]
10. Park, S.; Jung, D. Battery cell arrangement and heat transfer fluid effects on the parasitic power consumption
and the cell temperature distribution in a hybrid electric vehicle. J. Power Sources 2013, 227, 191–198.
[CrossRef]
11. Kizilel, R.; Lateef, A.; Sabbah, R.; Farid, M.M.; Selman, J.R.; Alhallaj, S. Passive control of temperature
excursion and uniformity in high-energy Li-ion battery packs at high current and ambient temperature.
J. Power Sources 2015, 183, 370–375. [CrossRef]
12. Javani, N.; Dincer, I.; Naterer, G.F.; Rohrauer, G.L. Modeling of passive thermal management for electric
vehicle battery packs with PCM between cells. Appl. Therm. Eng. 2014, 73, 307–316. [CrossRef]
13. Sabbah, R.; Kizilel, R.; Selman, J.R.; Al-Hallaj, S. Active (air-cooled) vs. passive (phase change material)
thermal management of high power lithium-ion packs: Limitation of temperature rise and uniformity of
temperature distribution. J. Power Sources 2008, 182, 630–638. [CrossRef]
14. Jin, L.W.; Lee, P.S.; Kong, X.X.; Fan, Y.; Chou, S.K. Ultra-thin minichannel LCP for EV battery thermal
management. Appl. Energy 2014, 113, 1786–1794. [CrossRef]
15. Zhao, J.; Rao, Z.; Li, Y. Thermal performance of mini-channel liquid cooled cylinder based battery thermal
management for cylindrical lithium-ion power battery. Energy Convers. Manag. 2015, 103, 157–165. [CrossRef]
16. Basu, S.; Hariharan, K.S.; Kolake, S.M.; Song, T.; Sohn, D.K.; Yeo, T. Coupled electrochemical thermal
modeling of a novel Li-ion battery pack thermal management system. Appl. Energy 2016, 181, 1–13.
[CrossRef]
17. Patil, M.; Panchal, S.; Kim, N.; Lee, M.Y. Cooling Performance Characteristics of 20 Ah Lithium-Ion Pouch
Cell with Cold Plates along Both Surfaces. Energies 2018, 11, 2550. [CrossRef]
18. Panchal, S.; Khasow, R.; Dincer, I.; Agelin-Chaab, M.; Fowler, M. Thermal design and simulation of
mini-channel cold plate for water cooled large sized prismatic lithium-ion battery. Appl. Therm. Eng.
2017, 122, 80–90. [CrossRef]
19. Wang, C.; Zhang, G.; Meng, L.; Li, X.; Situ, W.; Lv, Y.; Rao, M. Liquid cooling based on thermal silica plate for
battery thermal management system. Int. J. Energy Res. 2017, 41, 2468–2479. [CrossRef]
20. Wang, C.; Zhang, G.; Li, X.; Huang, J.; Wang, Z.; Lv, Y.; Meng, L.; Situ, W.; Rao, M. Experimental examination
of large capacity liFePO\r, 4\r, battery pack at high temperature and rapid discharge using novel liquid
cooling strategy. Int. J. Energy Res. 2018, 42, 1172–1182. [CrossRef]
Appl. Sci. 2019, 9, 754 20 of 20

21. Saw, L.H.; Ye, Y.; Tay, A.A.O.; Chong, W.T.; Kuan, S.H.; Yew, M.C. Computational fluid dynamic and thermal
analysis of Lithium-ion battery pack with air cooling. Appl. Energy 2016, 177, 783–792. [CrossRef]
22. Zhang, W.; Chen, X.; Yang, H.; Liang, H.; Wei, Y. Forced convection for flow across two tandem cylinders
with rounded corners in a channel. Int. J. Heat Mass Transf. 2019, 130, 1053–1069. [CrossRef]
23. Zhang, S.; Li, X.; Hu, B.; Liu, Y.; Zhu, Z. Numerical investigation of attached cavitating flow in
thermo-sensitive fluid with special emphasis on thermal effect and shedding dynamics. Int. J. Hydrog.
Energy 2019, 44, 3170–3184. [CrossRef]
24. Wei, Y.; Yang, H.; Dou, H.S.; Lin, Z.; Wang, Z.; Qian, Y. A novel two-dimensional coupled lattice Boltzmann
model for thermal incompressible flows. Appl. Math. Comput. 2018, 339, 556–567. [CrossRef]
25. Ge, Z.J. Research on Air-cooled Heat Dissipation System for Lithium Iron Phosphate Battery Pack of Electric
Vehicle. Master’s Thesis, South China University of Technology, Guangzhou, China, 2016.
26. Bernardi, D.; Pawlikowski, E.; Newman, J. A General Energy Balance for Battery Systems. J. Electrochem. Soc.
1985, 132, 5–12. [CrossRef]
27. ANSYS Fluent 18.0; ANSYS, Inc.: Canonsburg, PA, USA, 2017; Available online: https://www.ansys.com/
products/fluids/ansys-fluent (accessed on 17 February 2019).
28. Myers, R.H.; Montgomery, D.C.; Anderson-Cook, C.M. Response Surface Methodology: Process and Product
Optimization Using Designed Experiments, 4th ed.; Wiley: Hoboken, NJ, USA, 2016.
29. Li, W.; Xiao, M.; Peng, X.; Garg, A.; Gao, L. A Surrogate Thermal Modeling and Parametric Optimization of
Battery pack with Air Cooling for EVs. Appl. Therm. Eng. 2019, 147, 90–100. [CrossRef]
30. Montgomery, D.C. Introduction to Statistical Quality Control, 7th ed.; Wiley: Hoboken, NJ, USA, 2012.
31. Rangappa, R.; Rajoo, S. Effect of thermo-physical properties of cooling mass on hybrid cooling for lithium-ion
battery pack using design of experiments. Int. J. Energy Environ. Eng. 2018. [CrossRef]
32. Fang, K.T.; Lin, D.K.J.; Winker, P.; Zhang, Y. Uniform Design: Theory and Application. Technometrics
2000, 42, 237–248. [CrossRef]
33. Yang, D.C.; Jang, I.S.; Jang, M.H.; Park, C.N.; Park, C.J.; Choi, J. Optimization of additive compositions for
anode in Ni-MH secondary battery using the response surface method. Met. Mater. Int. 2009, 15, 421–425.
[CrossRef]
34. Buhmann, M.D. Radial basis functions. Acta Numer. 2000, 9, 1–38. [CrossRef]
35. Stein, M.L. Interpolation of Spatial Data: Some Theory for Kriging; Springer: New York, NY, USA, 1999.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).