Applied Thermal Engineering 236 (2024) 121518

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Thermal management of 21700 Li-ion battery packs: Experimental and

numerical investigations
Tien-Fu Yang a, Wei-Mon Yan b, c, *, Pei-Yi Lin b, c, Cong-You Lin d, Chang-Chong Yang e,
Uzair Sajjad b, c, *
a
Department of Refrigeration, Air-Conditioning and Energy Engineering, National Chin-Yi Universityof Technology, Taichung 41170, Taiwan
b
Department of Energy and Refrigerating Air-Conditioning Engineering, National Taipei University of Technology, Taipei 10608, Taiwan
c
Research Center of Energy Conservation for New Generation of Residential, Commercial, and Industrial Sectors, National Taipei University of Technology, Taipei
10608, Taiwan
d
Aurora Borealis Technology Co., Ltd., Tainan 71150, Taiwan
e
Institute of Green Energy and Environment, ITRI, Hsinchu 31040, Taiwan

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

Keywords: Due to its increased cell size, LIB 21700 (Lithium-ion battery) format has surpassed the existing formats as it
Lithium-ion battery pack offers larger capacity and higher energy density. However, the battery pack’s extended life and appropriate
Thermal management performance greatly relies on the temperature. Therefore, the thermal performance assessment of LIBS is quite
Charging and discharging
essential. In line with this, the objective of this study is to investigate the heat generation inside the 21,700
Lumped model
battery or battery pack experimentally and numerically. The numerical analysis is carried out at different
Discharge rate
Energy efficiency discharge rates by the lumped model, and two distinct arrangement designs of 5 × 6 and 2 × 15 are compared in
the battery pack. The numerical results at each discharge rate in the 5 × 6 battery pack are completely matched
with the experimental data, and this thermal model is used to predict the temperature distribution at several
discharge rates in the 2 × 15 battery pack. The results indicated that 5 × 6 battery pack offers greater heat
dissipation performance at the battery’s entrance and exit. However, at the center, the temperature surpasses the
working temperature range due to the tight arrangement, resulting in a substantial temperature difference be­
tween the 5 × 6 battery pack’s batteries. It was also found that both battery packs operate under 40 ◦ C only for
1C discharge rate while for discharge rates of 2C and 5C, internal temperature of battery is greater than 50 ◦ C
and 94 ◦ C, respectively.

1. Introduction safety risks, including the overheating, combustion, explosion, etc. [5].
As a result, the efficient cooling systems must be used as a priority for
Lithium-ion batteries (LiBs) are excellent selection for the energy maintaining the maximum cell temperatures, reducing the temperature
storage in electric vehicles (EVs) because they have great energy and gradient, and ensuring that the battery operates in its optimal operating
power density, long lifetime, low self-discharging rate, faster charging temperature range during high charging and discharging rates [6–9].
capacity, higher capacity and efficiency, etc. [1]. This is because the There are two main categories of research approaches for studying
battery capacity has a significant impact on electric vehicle performance the thermal properties of lithium-ion batteries: direct and indirect
and range [2]. The battery pack’s extended life and appropriate per­ calculation methods. Through experimental measurement, the direct
formance however greatly relies on the temperature [3]. In particular, methods examine the battery’s heat generation properties, whereas the
the LIBS have high sensitivity to the storage and working temperatures. indirect calculation methods rely on the battery models [10]. Indirect
In these batteries, heat generation occurs during their operation that calculation techniques can be used to analyze the internal state of the
results in the increase of the temperature of the batteries, and finally battery, which is challenging to measure experimentally. Direct methods
provides some temperature-induced degradation phenomena [4]. This are complex, expensive, and difficult to describe the complete thermal
has negative effects on the performance of the batteries and has some state and temperature distribution of batteries comprehensively. Hence,

* Corresponding authors.
E-mail addresses: [email protected] (W.-M. Yan), [email protected] (U. Sajjad).

https://doi.org/10.1016/j.applthermaleng.2023.121518
Received 5 March 2023; Received in revised form 30 August 2023; Accepted 5 September 2023
Available online 6 September 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

it is essential to foresee the dynamic response of a LIB in real time, reveals that mini and micro channels have been used for batteries’
distinct, and dynamic operational conditions other than the experi­ temperature control, and with proper design of flow paths and channel’s
mental testing [2]. Therefore, determining exact algorithms to predict configurations, the batteries can be operated safely [19,20]. In addition,
the battery’s thermal behavior is quite essential [11]. porous materials have also shown their effectiveness to achieve tem­
In general, electrochemical heat and lumped models are used to perature control in batteries [21]. The battery surface temperature was
evaluate the BTMS (battery thermal management system). Lumped also discovered to be kept at the lowest level while being discharged at a
model however is more common owing to its simplicity and accuracy 3C current rate using a mix of phase change material (PCM), metal foam,
[12,13]. Numerous attempts have been made to predict the heat gen­ and fins [22]. Similarly, another study reveals that PCM based battery
eration, temperature profile, or thermal response of the battery. module is more effective compared to its conventional counterpart at
Recently, a cycle and calendar aging model was proposed for 144Ah various discharge rates [23].
NMC/graphite battery that could predict DoD, C-rates, and multi- The literature review indicated that the temperature has the
temperatures with a respectable accuracy of RMSE = 0.85% [14]. important role in the battery for avoiding thermal runaway, and it
Recently, a numerical and experimental study was carried out to should be maintained at the standard range. Also, the potential of 21700
investigate the heat generation in NCM-21700 at different discharge Li-ion battery to increase the energy density and capacity in an
rates [15]. It was found that the final battery surface temperature errors economical manner as well as the availability of very limited literature
were 2.34%, 1.04%, 1.88%, 1.35%, and 1.67%, between the experi­ necessitates the further investigation of this subject. In particular, there
mental measurement and numerical results for the discharge C-rates of has been little attention paid to the thermal management of 21700 Li-ion
1C, 2C, 3C, 4C, and 5C, respectively. Besides, neural networks have been battery and battery packs. Till date, no experimental and numerical
developed that can predict the battery’s heat generation rate with an investigation exists on thermal management of different arrangement
error less than 5% [16]. designs of 21700 battery packs. Due to the importance of this issue, in
In addition, both experimental and numerical investigations have this investigation, both experimental and numerical studies are con­
been performed to evaluate the BTMS along with various cooling sys­ ducted on the cooling of the 21700 battery and battery packs with two
tems. The cold plate, composite phase change materials (PCMs), hybrid distinct arrangement designs of 5 × 6 and 2 × 15. The electrochemical
BTMS with PCMs and liquid cooling, and heat pipes are some of the model is utilized to predict the battery’s irreversible and reversible heat
examples of the existing advanced BTMS [17,18].The existing literature generation. The studies are performed for different discharge rates and

Fig. 1. (a) The photo of the experimental system, (b) the schematic diagram of the experimental system, and (c) the schematic diagram of the cooling channel in the
battery test station.

2
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

two distinct arrangement designs of 5 × 6 and 2 × 15 in the battery pack. check whether the initial conditions of the experiment are Normal and
stable, the inspection during the experiment, the adjustment of experi­
2. Experimental setup and procedures mental conditions, and the data recording during the experiment are
briefly described as follows:
The photo of the experimental setup is shown in Fig. 1(a). The 21700
cylindrical lithium batteries are used in this work. The MOLICEL INR- 2.1. Ambient temperature and humidity adjustment
21700-P42A battery is a recent type of lithium-ion cylindrical batte­
ries. The lithium battery is a ternary material (NMC), the positive This experiment studies the temperature changes of lithium batte­
electrode (anode) is lithium nickel cobalt manganese (NMC), the nega­ ries. In order to measure the performance of lithium batteries, the in­
tive electrode (cathode) is graphite, and the separator is polycarbonate. fluence of environmental factors must be considered. Environmental
Porous membranes made of olefins have the advantages of high energy factors greatly affect the performance of lithium batteries. The constant
density, good cycle performance, and high battery life and life. The changes in temperature and humidity in the outdoor environment will
lithium battery used in the study has a nominal voltage of 3.6 V, a charge directly affect the performance of lithium batteries. Therefore, in order
cut-off and a discharge cut-off of 4.2 V and 2.5 V, and a nominal capacity to avoid the influence of different environmental temperature and hu­
of It is 4200mAh, and the charging working temperature is 0 ℃ to 45 ℃, midity on the performance of lithium batteries, a constant temperature
and the discharging working temperature is − 40 ℃ to 60 ℃. The and humidity box is used to maintain the required environmental con­
increased size (21 mm by 70 mm) provides higher energy density than ditions. The constant temperature and humidity box is composed of two
the 18,650 battery. The specifications of INR-21700-P42A battery are parts: regulation and humidification. Through the fan installed inside
listed in Table 1. In Fig. 1(a), the battery, direct current power supply, the box, the air is discharged into the box to realize gas circulation and
data logger, DC electronic load, and thermocouples can be observed. For balance the temperature and humidity in the box. Adjust the tempera­
preventing the effects of the environment and preserving the experi­ ture and humidity conditions required for the experiment by adjusting
mental working situation, the temperature of battery pack is kept at the settings of the constant temperature and humidity chamber, and let
25 ◦ C. The direct current power supply is used to charge the lithium-ion it stand for a period of time to ensure that the environmental conditions
battery. The direct current electronic load is in constant current mode are stabilized.
(CC) for discharging the battery. The temperature changes of the battery
are recorded during its operation and the data processing are conducted
2.2. Lithium battery pack charging
in the experiments by using the data logger.
Fig. 1(b) is a view of the experimental setup which is used to explore
Before carrying out the battery discharge test, the battery to be tested
the battery capacity, discharge voltage, and temperature of lithium
needs to be charged first. The battery to be tested is placed in the battery
batteries for various discharge rates. In the single battery discharge
test station and left to stand first. After the battery temperature stabi­
experiment, the discharge rate varies between 1C and 5C, while for the
lizes, it starts to charge. During the process, the constant current mode
battery pack, the discharge rate is between 1C and 4C. During discharge
(Constant Current, CC) for charging. After fully charged, the battery
process, the experimental measurements are conducted in the constant
must be left to stand for a period of time. After the internal chemical
current mode. To decline the instability of chemical materials and
materials of the battery are stabilized, proceed to the next step. Note that
maintain battery efficiency, first the battery is kept at a fixed tempera­
C-rate is the measurement of the charge and discharge current with
ture test section box, and the experiments are conducted after the sta­
respect to its nominal capacity.
bilization of the chemical materials inside the battery. In the discharge
experiment, the battery is continuously discharged until the battery
voltage dropped to 2.5 V. As shown in Fig. 1(b), the battery test station is 2.3. Discharge rate setting
equipped with a cooling fan and temperature sensors to regulate effec­
tively the temperature. During the discharging process with heat dissi­ During the battery discharge experiment, it is necessary to conduct
pation, the forced convection air cooling is formed in the battery test discharge experiments with different discharge rates. The difference in
station due to the use of cooling fan. Fig. 1(c) illustrates a schematic discharge rate will directly affect the performance change of the battery.
diagram of the cooling channel employed in the battery test station. In Therefore, the discharge rate is adjusted. The discharge rates used in this
this work, the quantities of the temperature, voltage, ampere and time experiment are 1C, 2C, 3C, and 4C. Four different discharge rates, the
are measured and presented. The uncertainties of the temperature, whole process is discharged in constant current mode (Constant Current,
voltage, ampere and time are, respectively, are 0.2 ◦ C, 0.005 V, 0.005A CC). After the discharge rate adjustment is completed, the battery
and 0.1 sec which are relatively small and can be ignored. discharge experiment is started.
The lithium battery pack is placed in the battery test station for
discharge experiments. Discharge is performed according to different 2.4. Discharge experiment
discharge rates, and the performance data of the lithium battery after
the discharge is recorded. In order to ensure the stability of the experi­ During the discharge experiment, the performance changes of the
ment process, before the battery discharge experiment, it is necessary to lithium battery/lithium battery pack at different discharge rates were
recorded, including discharge capacity, discharge depth, discharge cut-
off voltage, positive electrode, negative electrode, and surface temper­
Table 1 ature, to facilitate subsequent research.
Specifications of INR-21700-P42A-battery used in this work.
Name of the parameter Parameter type Range 2.5. Data analysis
Battery capacity Typical 4200mAh
Minimum 4000mAh After the discharge experiment, the data is sorted out, and the
Battery voltage Nominal voltage 3.6 V
different changes in the lithium battery pack under different discharge
Peak voltage 4.2 V
Discharge cut-off voltage 2.5 V rates are analyzed, and the battery performance is analyzed through the
Operating temperature Charge 0 ~ 45 ◦ C measurement and recording data, so as to facilitate the follow-up dis­
Discharge − 40 ~ 60 ◦ C cussion. In this work, the T-type thermocouple was used to measure the
Dimensions Diameter 21.7 mm temperature of the battery pack. The uncertainty of the T-type ther­
Height 70 mm
mocouple is 0.2 ◦ C.

3
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

3. Numerical method Table 2

Related dimensions used for the 5 × 6 and 2 × 15 battery pack simulation
In the present study, the 21700 cylindrical lithium battery is used. models.
The battery has positive and negative electrodes, the electrolyte, and the Parameters Units 5×6 2 × 15
separator. Fig. 2(a) and (b) are schematic diagrams of the 5 × 6 and 2 × L mm 425 625
15 models of the battery pack, respectively. The upper picture is a side W mm 240 180
view and the lower one is a top view. In Fig. 2, the blue section indicates H mm 210 210
the fluid inlet, red section indicates the fluid outlet, and the green sec­ L1 mm 140 140
L2 mm 145 345
tion shows the battery. For simulation of the flow inside the battery test
L3 mm 140 140
section, a specific flow field is considered, where the battery pack in­ W1 mm 70 70
cludes a fan intake and an open exhaust. Table 2 presents the related W2 mm 70 70
dimensions used for the 5 × 6 and 2 × 15 battery pack. L, W, and H are H1 mm 140 140
the total length, total width, and total height, respectively. During the
steady-state heat transfer process from the battery pack to the air, when
(3) The constant thermophysical properties are considered for the
the battery has not yet started to charge and discharge, the time t < 0,
batteries and air.
the flow field inlet velocity is uin , and the cell temperature are main­
(4) The uniform heat is created in the battery.
tained at the ambient temperature T∞ . when the time t⩾0, the battery
starts to charge or discharge, a small amount of energy is converted into
Based on these assumptions, the following governing equations are
heat, which makes the temperature of the battery core rise. To simulate
used [23,24]:
the cooling fan in the simulated battery test station, after the experi­
Continuity equation:
mental measurement, the corresponding boundary conditions are inlet
velocityuin = 2 m/s, fluid inlet temperatureTin = T∞ = 25 ℃, the upper, ∇⋅→
uf =0 (1)
lower, left, and right walls of the flow channel meet the non-slip con­
dition. Besides, the external environment and the walls are subjected to In this equation, → u f indicates the flow velocity vector.
a convection condition with convective heat transfer coefficient h• The Momentum equation:
flow and temperature fields at the outlet are assumed to be fully ⎡
( )
⎤
→
developed, and the velocity field at the interface between the fluid and ρf ⎣
∂ u f
+ →u f ⋅∇ →u f ⎦ = − ∇P + μf ∇2 → uf (2)
the battery cell is assumed to be zero. Based on the Reynolds number, ∂t
the fluid flow mode is turbulent. In air cooling thermal analysis, the
intervals between battery cells form cooling channels, and the heat In this equation, ρf and μf demonstrate the density and dynamic
generated by the battery is discharged by the air through the cooling viscosity of fluid, respectively and P indicates the pressure.
channels. The distribution of air flow determines the surface convective Energy equation:
heat transfer coefficient, which ultimately affects the battery surface ∂T
( )
temperature distribution. ρf Cpf f + ρf Cpf → u f ⋅∇Tf = kf ∇2 Tf (3)
∂t

3.1. Governing equations In this equation, Cpf and kf demonstrate the speific heat and thermal
conductivity of the fluid, and Tf is the temperature of fluid.
The FVM is used for the simulation of this problem. The following The governing equation of the battery pack is:
assumptions are also used to simplify the problem: ∂Ts
ρs Cps = ks ∇2 Ts + q̇g (4)
∂t
(1) The steady condition and turbulent regime are considered.
(2) The effects of thermal radiation are not considered. In this equation, ρs , Cps , and ks are the effective density, the effective

Fig. 2. The schematic diagram of the 3D model of the battery pack, the upper picture is a side view, and the lower picture is a top view. (a) 5 × 6 battery pack; (b) 2
× 15 battery pack.

4
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

specific heat, and the effective thermal conductivity of the battery, work is SST model, and the calculated Reynolds number Re is about
respectively, Ts indicates the temperature of battery, and qg is the 5525, which belongs to the low Reynolds number. In this work, the low
amount of heat generated per unit volume when the battery is charged Reynolds SST K-Omega model was adopted in which the y+ is required
and discharged. to be between 0 and 5. In this work, when y+ is close to 1, the results
During the operation of the lithium-ion battery, the heat is mainly have high accuracy. In order to observe the change of battery gap and
generated by both irreversible and reversible processes. The internal fluid in this work, it must have high accuracy between the walls.
heat generation of the battery is introduced by Bernardi et al. [25]. It is Therefore, the y+ value used in the analysis is about 1.
the sum of reversible and irreversible thermal effects. In this work, the
voltage difference of the battery is achieved by the experiments, and the 3.3. Grid-independence test
entropy coefficient is achieved by the data of Yun et al. [26]. The
volumetric rate of heat generated is calculated by the following formula In this work, the CFD software of ANSYS was used to simulate the
[25]: problem. The physical model to be described can also be regarded as the
( ) control volume, which is divided into network grid unit to describe, and
I dEoc
q̇g = Eoc − U − T (5) use Gauss’s Law to convert the volume integral in the partial differential
V dT
equation of the divergent term into a surface integral, the sum of the grid
In this equation, I, Eoc, and U are the current, open circuit voltage, items per unit is calculated to obtain the flux of the finite volume sur­
and voltage of the battery, respectively. T and dEdToc are the battery tem­ face, and the flux of the given volume. The flux is the same as the flux
perature and the temperature coefficient, respectively. In addition, V leaving the adjacent volume, indicating compliance with the conserva­
indicates the battery volume. tion law. Mesh establishment is a very important procedure in CFD. In
Here, the charge–discharge reaction analysis of single battery was this study, air fluid is used to cool the lithium battery. After the cooling
performed using the NTGK model developed by Kwon et al. [27]. The air enters, it flows in from the left side and flows out from the right side.
simple semi-empirical electrochemical model proposed by Kwon et al. The fluid flows in through the battery gap to achieve the cooling effect.
[27] argues that the distribution of potential and current density on Each lithium battery is presented in the form of a cylinder. In the nu­
battery electrodes at constant current is a function of discharge time, merical analysis of this study, the interface between the lithium battery
and they tested this argument by comparing discharge curves at and the fluid is crucial to the simulation results. In the analysis of this
different ambient temperatures. The heat generated by the battery is a study, the mesh on the surface of the fluid and the battery must be of
function of the discharge time, which is employed for the prediction of high quality, so the mesh densification is performed on the contact
the temperature distribution of lithium-ion batteries. Equation (6) rep­ surface of the lithium battery and the fluid, and the y+ is kept within an
resents an algebraic equation of the volumetric current transfer rate acceptable range throughout the densification process.
associated with the potential field, and the change in discharge depth The temperature of the battery is used for the validation and grid-
(DoD). Elapsed time can be obtained by solving equations (6) and (7). In independence test. The discharge rate of 3C is selected for this test.
the numerical model, the discharge curve fitting of the lithium battery The grid numbers of 80 × 80 × 150, 100 × 110 × 200, and 120 × 130 ×
must be established to calculate the Y and U coefficients to obtain the 250 are tested. Fig. 3 shows the predictions with different grid numbers
current density distribution in the electrode. and results of comparison with the experimental data. As illustrated in
this figure, the predictions for the three grid numbers agree with the
QAh experimental data. Besides, it is clearly noted in Fig. 3 that before t =
jECh = ⋅Y[U − (ϕ+ − ϕ− ) ] (6)
Qref Vol 600 s, the predictions of the three grid numbers are similar and almost
∫ t
the same. After t = 600 s, the predictions for three grids are rather
Vol different. Table 3 presents the deviations among different grid numbers
DoD = jECh dt (7)
3600QAh 0 and times. It can be observed that the relative errors of the predictions
In this equation, φ indicates the potential of the battery pack, Qref and for the grid numbers of 100 × 110 × 200 and 120 × 130 × 250 are about
QAh are the experimental charge capacity and the total charge amount, 0.004%–0.085%. For providing the balance between accuracy of the
respectively. Vol indicates the volume of the charge activity area, t predicted data and the running time, the grid number of 100 × 110 ×
demonstrates the discharge time, and Y and U indicate the function of 200 is employed in this work.
discharge depth in battery.
4. Results and discussion
3.2. SST turbulence model
4.1. Battery performance curve analysis
For the accurate simulation of the turbulent fluid flow, the SST model
[28,29] is employed. The governing equations for this model are: The time evolutions of the single battery discharge voltage at
( ) ( ) discharge rates of 1C, 2C, 3C, 4C, and 5C are presented in Fig. 4. The
[ ]
∂ ρf k ∂ ρf Vi k ∂ ∂k predicted discharge voltages agree well with those of the present mea­
+ = P̃k - β∗ ρf kω + (μ + σ k μt ) (8)
∂t ∂xi ∂xi ∂xi surement. Besides, it is found that the initial discharge voltage decreases
from 4.05 V to 3.7 V when the discharging rate is increased between 1C
( ) ( ) [
∂ ρf ω ∂ ρf Ui ω ∂ to 5C. And the discharge cut-off voltages are all about 2.6 V for various
+ = αρf S2 − βρf ω2 + (μ
∂t ∂xi ∂xi discharge rates. For the INR-21700-P42A-battery, the peak voltage is
∂ω
]
1 ∂k ∂ω about 4.2 V. Therefore, at 1C discharge rate, the overpotential drop of
+ σ ω μt ) + 2(1 − F1 )ρf σ ω2 (9) the battery is about 0.15 V. While at the high discharge rate of 5C, the
∂xi ω ∂xi ∂xi
overpotential is increased to 0.5 V, which indicates that the battery has a
In this equation, μt indicates the turbulence viscosity, P̃k demon­ larger voltage drop under a higher discharge rate. This is due to the
strates the kinetic energy term provided by the turbulence, σ k and σω lower efficiency of chemical reaction inside the battery for a higher
demonstrate the turbulent Prandtl numbers, α,β and β∗ demonstrate the discharge rate.
constants in the turbulence model, F1 indicates the mixing function, Vi Fig. 5 presents the predicted and measured time evolutions of the
indicates the fluid velocity, k demonstrates the turbulent flow energy, ω maximum surface temperature of single battery at several discharge
demonstrates the turbulence dissipation rate, S indicates the strain rate, rates. The operating ambient temperature of the battery at different
and ρf demonstrates the fluid density. The turbulence model used in this discharge rates is all maintained at 25 ◦ C. An overall inspection of Fig. 5

5
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

Fig. 3. Predictions of different grid-independence tests and compared with the measured data at discharge rate of 3C.

discharge. In addition, the measure surface temperatures are slightly

Table 3
higher than predictions.
Grid-independent tests.
Fig. 6 presents the battery temperature distribution at 1C to 5C
Nx × Ny × Nz Time(s) 80 × 80 × 150 100 × 110 × 200 120 × 130 × 250 discharge rates when the battery fully discharges. As disclosed in Fig. 6,
200 27.7735 27.7735 27.7748 at the 1C discharge rate, the temperature gradient in the battery is quite
400 42.714 42.718 42.705 small, and the maximum temperature difference within the battery is
600 47.610 47.682 47.652
about 0.1 ◦ C. But, at 5C discharge rate, there is a highest battery tem­
800 52.562 53.222 53.188
1080 60.611 62.975 62.921
perature about 56 ◦ C near the center of the battery. The temperature
Relative error 0.004%~3.67% 0.004%~0.085% 0% gradually decreases toward the battery surface. It is also noted in Fig. 6
that higher temperature non-uniformity is found for a higher discharg­
ing rate.
reveals that the predicted surface temperatures for various discharging
rates agree with those of measured results during discharging. And the
surface temperature of the battery increases with time rapidly in the 4.2. Battery pack performance curve analysis
initial charging. At the late discharging, the surface temperature in­
creases gradually with time. As shown in Fig. 5, the maximum predicted Fig. 7 reveals the measured discharge voltage curves with time at
surface temperature rises for 1C, 2C, 3C, 4C and 5C are about 4 ◦ C, 11 ◦ C, discharge rates of 1C, 2C, 3C, and 4C for a 5 × 6 battery pack. It is clear
16.5 ◦ C, 24.5 ◦ C and 31.5 ◦ C, respectively, when the battery is fully that the initial discharge voltage reduces with increasing discharging

Fig. 4. Comparison of predicted and measured time evolutions of discharge voltages at different discharge rates for a single battery.

6
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

Fig. 5. Comparison of experimental measurement and numerical analysis of surface temperature and time for single cells at different discharge rates.

Fig. 6. Temperature distributions of single battery at fully discharged state for 1C to 5C. (a) 1C, (b) 2C, (c) 3C, (d) 4C, (e) 5C.

Fig. 7. The discharge voltage curves at different discharge rates for a 5 × 6 battery pack.

7
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

rate and varies from 4.05 V to 3.7 V. Comparison of the corresponding Fig. 8 illustrates the temperature distributions of the 5 × 6 battery
curves in Fig. 4 and Fig. 7 indicates that the discharge voltage curves for pack at fully discharge state with different discharge rates. In Fig. 8, the
battery pack are similar to those for a single battery. This is because each left is the side view and the right is cross-sectional view of y = 47.5 mm.
battery of the 5 × 6 battery pack is connected in parallel, so the Besides, the blue part of the side view indicates the air inlet, the red part
discharge voltage curves are in the similar trend for both two. demonstrates the air outlet, the blue arrow in the cross-sectional view is

Fig. 8. The temperature distributions of the 5x6 battery pack with different discharge rates with the left pictures being the side view, and the right pictures being
cross-sectional view of y = 47.5 mm. (a) 1C; (b) 2C; (c) 3C; (d) 4C; (e) 5C.

8
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

the air inlet, and the red arrow is the air outlet. As can be seen from that is much higher than the working temperature, which may lead to a
Fig. 8, the 5 × 6 battery pack with different discharge rates experienced decrease in the working efficiency of the battery pack. For a discharge
with considerable temperature distributions. At the 1C discharge rate, rate of 3C, the maximum surface temperature after discharging reaches
most of the battery pack temperature shows a dark blue temperature 63.6 ◦ C. At this time, in addition to the decrease in the working effi­
distribution with maximum temperature about 36 ◦ C, and at the 2C ciency of the battery, it may also accelerate the aging of the internal
discharge rate, the temperature of the battery pack gradually produces a chemical materials of the battery pack. At a discharge rate of 4C, the
light blue temperature distribution with maximum temperature about maximum surface temperature at the end of the discharge is as high as
51 ◦ C. At the 3C discharge rate, the battery pack temperature distribu­ 79.2 ◦ C. In addition to greatly reducing the working efficiency and life of
tion gradually appears green, at the 4C discharge rate, a yellow tem­ the battery, such a high temperature may result in the danger of thermal
perature distribution is generated, and at the 5C discharge rate, the runaway of the battery pack. At the discharge rate of 5C, only the pre­
battery pack produces a red temperature distribution with maximum diction of the maximum surface temperature of the battery pack is
temperature about 95 ◦ C. Such results show that only 1C discharge rate presented. The maximum surface temperature is about is about 95 ◦ C at
at the battery pack temperature is close to the operating temperature of fully discharge. Such a high temperature is already at the critical value
40 ◦ C. At discharge rates of 2C and 3C, the temperature distributions are of thermal runaway of the battery, indicating that the battery pack may
noted with significant rise relative to those of 1C-rate due to the lower catch fire at any time.
energy transfer efficiency for a higher discharge rate. While at 4C and 5C Fig. 10 illustrates the temperature distributions of the 2 × 15 battery
rates, the battery pack produces very high temperature distributions, pack at several discharge rates. Comparison of Figs. 8 and 10 indicates
which may lead to the danger of thermal runaway of the battery pack, that the 2 × 15 battery pack and the 5x6 battery pack have similar
and the results also indicate considerable temperature gradient in the temperature distributions. As the discharge rate boosts, the temperature
battery pack at 5C discharge rate. The high temperature rise and non- distributions of the battery pack change from a dark blue temperature
uniformity may be caused by the tight arrangement design of the 5 × distribution at the 1C discharge rate to a red temperature distribution at
6 battery pack, when the air cannot effectively flow into the internal gap a 5C discharge rate. Similar to 5 × 6 battery pack, the surface temper­
of the battery pack. This means that the convection heat transfer coef­ ature of 2 × 15 battery pack is under safe operating temperature of 40 ◦ C
ficient between the internal battery pack will be much lower than that of only for 1C discharge rate. At a high discharge rate, the 2 × 15 battery
the external battery pack. Therefore, it cannot effectively dissipate heat. pack has significantly exceeded the operating temperature, which will
The accumulation of internal battery heat results in mutual heating, so lead to a decrease in the working efficiency of the battery pack. It can be
that the temperature rise is more considerable. Therefore, the temper­ observed that the 2 × 15 battery pack is experienced at a higher tem­
ature near the interior of the battery pack will increase significantly. perature distribution. This can be made plausible by noting the fact the
The comparisons of the time evolution of surface temperature of 5 × long strip arrangement of the 2 × 15 battery pack will reduce the heat
6 battery pack between the experimental and numerical studies at dissipation caused by the close arrangement in the flow direction. This
several discharge rates are disclosed in Fig. 9. To protect the battery causes a higher temperature distribution for each battery in the 2 × 15
safety, the measurement of 5 × 6 battery pack for 5C discharge rate is battery pack.
not performed. It is clearly fond that the predictions agree well with the The predicted surface temperature distributions of the 2 × 15 battery
measured data. Besides, it is noted that at the discharge rate of 1C, due to pack during discharging for several discharge rates are revealed in
the longer discharge time, the battery temperature rises relatively Fig. 11. The temperature rise trend of the 2 × 15 battery pack is roughly
smoothly. After discharging, the maximum surface temperature is about the same as that of the 5 × 6 battery pack (see the results in Fig. 9 and
36.5℃, which is close to the working temperature of 40℃, which can Fig. 11). The discharge time is long at 1C discharging rate, so the battery
achieve better working efficiency. For the discharge rate of 2C, the temperature rises relatively gently. After discharging, the surface tem­
maximum surface temperature of the battery pack rises at a faster rate perature is about 36.2 ◦ C that is close to the working temperature of
with time. After discharging, the surface temperature is about 51.6 ◦ C 40 ◦ C, which can achieve better working efficiency. At a 2C discharge

Fig. 9. Comparisons of experimental measurement and numerical analysis of battery surface temperature and time for a 5 × 6 battery pack at different
discharge rates.

9
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

Fig. 10. T The temperature distributions of the 2 × 15 battery pack with different discharge rates with the left pictures being the side view, and the right pictures
being cross-sectional view of y = 47.5 mm. (a) 1C; (b) 2C; (c) 3C; (d) 4C; (e) 5C.

rate, the surface temperature of 2 × 15 battery pack rises rapidly. After temperature rise profile to the 5 × 6 battery pack. Although the 2 × 15
the discharge, the surface temperature is about 51℃. At 5C discharge design has a large contact area between the air and the battery surface,
rate, the surface temperature of the battery pack after discharge is about but due to the long airflow path, the air is heated by the heat of the
94.6 ℃. The temperature rise of the 2 × 15 battery pack is identical to battery during the flow process. This results in a decrease in the tem­
that of the 5 × 6 battery pack. perature difference between the air and the battery which in turn
As shown in Figs. 8-11, the 2 × 15 battery pack has a similar reducing heat transfer. As for the 5 × 6 battery pack, the air cannot

10
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

Fig. 11. Numerical analysis of battery surface temperature and time at different discharge rates for a 2 × 15 battery pack.

effectively flow into the internal gap of the battery pack for heat dissi­ critical value of thermal runaway of the battery, indicating that the
pation. This causes mutual heating and increases the temperature of the battery pack may catch fire at any time.
battery. In terms of the difference between the two arrangement types, 5 - The spacing between the batteries should be carefully considered to
× 6 is a short and wide arrangement similar to a square, and 2 × 15 is a enhance the uniformity of air flowing into the battery.
long and narrow arrangement similar to a rectangle. Although the 5 × 6 - The temperature rise of the 2 × 15 battery pack is similar to that of
arrangement has a shorter airflow path, the heat accumulation problem the 5 × 6 battery pack.
of the battery pack causes the temperature to increase. While the 2 × 15
arrangement has a larger contact area with the air, due to the influence
of the airflow path, the heated air cannot effectively dissipate heat with Declaration of Competing Interest
the battery, resulting in high temperature. The convective heat transfer
coefficient of these two arrangements is reduced due to different reasons The authors declare that they have no known competing financial
such as close arrangement and airflow path, so they have similar trends interests or personal relationships that could have appeared to influence
in the process of temperature increase. the work reported in this paper.

5. Conclusions Data availability

In this work, by performing a thermal analysis of a single battery/ Data will be made available on request.
battery pack, a lumped model is employed for the simulation of the heat
generated in the battery at different discharge rates, and the analysis Acknowledgement
results are quite consistent with the experimental measurement. In the
battery pack, two different arrangements of 5 × 6 and 2 × 15 are This research was funded by both Industrial Technology Research
compared experimentally and numerically. Although the total heat of Institute and Ministry of Education (Higher Education Sprout Project
lithium batteries is mainly dominated by irreversible heat, ignoring the through Research Center of Energy Conservation for New Generation of
effect of reversible heat may cause certain errors, especially at low Residential, Commercial, and Industrial Sectors) in Taiwan.
discharge rates in which the effect of reversible heat is more obvious. In
this work, a lumped model based on irreversible heat and reversible heat References
is used to calculate the average heating value, which can better predict
the temperature distribution of lithium batteries at different discharge [1] W. Li, et al., An internal heating strategy for lithium-ion batteries without lithium
plating based on self-adaptive alternating current pulse, IEEE Trans. Veh. Technol.
rates. The following findings can be concluded as follows. (2022).
[2] S. Vashisht, et al., Thermal behaviour of Li-ion battery: An improved
- The results indicated that 5 × 6 battery pack offers greater heat electrothermal model considering the effects of depth of discharge and
temperature, J. Storage Mater. 70 (2023), 107797.
dissipation performance at the battery’s entrance and exit. However, [3] M.-K. Tran, et al., A review of lithium-ion battery thermal runaway modeling and
at the center, the temperature surpasses the working temperature diagnosis approaches, Processes 10 (6) (2022) 1192.
range due to the tight arrangement, resulting in a substantial tem­ [4] Y. Saito, M. Shikano, H. Kobayashi, Heat generation behavior during charging and
discharging of lithium-ion batteries after long-time storage, J. Power Sources 244
perature difference between the 5 × 6 battery pack’s batteries.
(2013) 294–299.
- Both battery packs (5 × 6 and 2 × 15) operate under 40 ◦ C only for [5] J.X. Weinert, A.F. Burke, X. Wei, Lead-acid and lithium-ion batteries for the
1C discharge rate. Chinese electric bike market and implications on future technology advancement,
J. Power Sources 172 (2) (2007) 938–945.
- For a discharge rate of 2C, internal temperature of the battery is
[6] Y. Deng, et al., Effects of different coolants and cooling strategies on the cooling
greater than 50 ◦ C. performance of the power lithium ion battery system: A review, Appl. Therm. Eng.
- It is found that the maximum surface temperature is about is about 142 (2018) 10–29.
95 ◦ C at fully discharge. Such a high temperature is already at the [7] C. Roe, et al., Immersion cooling for lithium-ion batteries–A review, J. Power
Sources 525 (2022), 231094.

11
T.-F. Yang et al. Applied Thermal Engineering 236 (2024) 121518

[8] A.K. Thakur, et al., A state of art review and future viewpoint on advance cooling [20] S. Sarvar-Ardeh, et al., A review on the applications of micro-/mini-channels for
techniques for Lithium–ion battery system of electric vehicles, J. Storage Mater. 32 battery thermal management, J. Therm. Anal. Calorim. (2023) 1–21.
(2020), 101771. [21] S. Rashidi, A. Ijadi, Z. Dadashi, Potentials of porous materials for temperature
[9] Q. Wang, et al., Experimental investigation on EV battery cooling and heating by control of lithium-ion batteries, J. Storage Mater. 51 (2022), 104457.
heat pipes, Appl. Therm. Eng. 88 (2015) 54–60. [22] H.N. Khaboshan, et al., Improving the cooling performance of cylindrical lithium-
[10] Z. Chen, et al., Numerical study on the heat generation and thermal control of ion battery using three passive methods in a battery thermal management system,
lithium-ion battery, Appl. Therm. Eng. 221 (2023), 119852. Appl. Therm. Eng. 227 (2023), 120320.
[11] R. Braga, et al., Transient Electrochemical Modeling and Performance Investigation [23] V. Choudhari, et al., Experimental and numerical investigation on thermal
Under Different Driving Conditions for 144Ah Li-ion Cell with Two Jelly Rolls, SAE characteristics of 2× 3 designed battery module. Available at SSRN 4367193,
Technical Paper, 2023. 2023.
[12] T. Dong, P. Peng, F. Jiang, Numerical modeling and analysis of the thermal [24] E. Jiaqiang, et al., Effects of the different air cooling strategies on cooling
behavior of NCM lithium-ion batteries subjected to very high C-rate discharge/ performance of a lithium-ion battery module with baffle, Appl. Therm. Eng. 144
charge operations, Int. J. Heat Mass Transf. 117 (2018) 261–272. (2018) 231–241.
[13] E. Gümüşsu, Ö. Ekici, M. Köksal, 3-D CFD modeling and experimental testing of [25] L. Sheng, et al., Effect analysis on thermal profile management of a cylindrical
thermal behavior of a Li-Ion battery, Appl. Therm. Eng. 120 (2017) 484–495. lithium-ion battery utilizing a cellular liquid cooling jacket, Energy 220 (2021),
[14] S. Panchal, et al., Development and Validation of Cycle and Calendar Aging Model 119725.
for 144Ah NMC/Graphite Battery at Multi Temperatures, DODs, and C-Rates. [26] F.-L. Yun, et al., Analysis of capacity fade from entropic heat coefficient of Li
2023, SAE Technical Paper. [NixCoyMnz] O2/graphite lithium ion battery, J. Electrochem. Soc. 163 (5) (2016)
[15] T.-F. Yang, et al., Numerical and experimental study on thermal management of A639.
NCM-21700 Li-ion battery, J. Power Sources 548 (2022), 232068. [27] K.H. Kwon, et al., A two-dimensional modeling of a lithium-polymer battery,
[16] C. Wu, et al., Experimental and numerical studies on lithium-ion battery heat J. Power Sources 163 (1) (2006) 151–157.
generation behaviors, Energy Rep. 9 (2023) 5064–5074. [28] F.R. Menter, Review of the shear-stress transport turbulence model experience
[17] A.K. Joshi, et al., Numerical analysis of battery thermal management system using from an industrial perspective, Int. J. Comput. Fluid Dynam. 23 (4) (2009)
passive cooling technique. 2023, SAE Technical Paper. 305–316.
[18] A.K. Thakur, et al., A state-of-the art review on advancing battery thermal [29] L.H. Saw, et al., Computational fluid dynamic and thermal analysis of Lithium-ion
management systems for fast-charging, Appl. Therm. Eng. 226 (2023), 120303. battery pack with air cooling, Appl. Energy 177 (2016) 783–792.
[19] A. Salimi, M. Khoshvaght-Aliabadi, S. Rashidi, On thermal management of pouch
type lithium-ion batteries by novel designs of wavy minichannel cold plates:
Comparison of co-flow with counter-flow, J. Storage Mater. 52 (2022), 104819.

12