Applied Thermal Engineering 231 (2023) 120959

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Thermal parameters of cylindrical power batteries: Quasi-steady state heat

guarding measurement and thermal management strategies
Hong Yu a, Hengyun Zhang a, *, Jinghe Shi a, Shunbo Liu a, Zhaozang Yi a, Shen Xu a,
Xinwei Wang b
a
School of Mechanical and Automotive Engineering, Shanghai University of Engineering Science, 333 Longteng Road, Songjiang, Shanghai 201620, China
b
Department of Mechanical Engineering, 2025 Black Engineering Building, Iowa State University, Ames, IA 50011, USA

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

Keywords: In this work, a new quasi-steady state heat guarding measurement method for the thermophysical parameters of
Cylindrical lithium-ion battery cylindrical batteries is proposed. The effectiveness of the heat guarding method is evaluated by the finite element
Axial thermal conductivity analysis and the measurement of a standard ANSI 304 stainless steel sample. The heat loss is minimized to 2.4%
Specific heat
maximum for cylindrical batteries under test based on numerical simulation. With the optimization of the test
Heat guarding method
Thermal management strategies
configuration, the experimental characterization of batteries was conducted to determine the thermal parameters
accurately. The experimental results show that the axial thermal conductivity of 18650 and 21700 batteries are
within 11.8–15.4 W⋅m− 1 ◦ C− 1 and 12.6–16.7 W⋅m− 1 ◦ C− 1, respectively, which are much lower than the material
test values by the laser flash method. Axial thermal conductivity increases with temperature and SOC, and the
specific heat increases linearly with temperature but varies little with SOC. This study demonstrates a fast, cost-
effective and nondestructive way to obtain the axial thermal conductivity and specific heat of cylindrical bat­
teries simultaneously and accurately. The subsequent thermal analysis on the thermal design of cylindrical
batteries is also conducted based on the measured thermal parameters, which facilitate appropriate thermal
management strategies for different types of cylindrical batteries.

and safety of lithium-ion power batteries are becoming predominate

issues in the development of electric vehicles [12–15]. To come up with
1. Introduction optimal thermal management solutions, first and foremost, the ther­
mophysical parameters of lithium-ion batteries in different forms are to
In recent years, electric vehicles have played an important role in be determined accurately, including thermal conductivity and specific
mitigating energy crisis and environmental pollution, and are being heat [16]. For a typical cylindrical lithium-ion battery, the radial ther­
developed rapidly all over the world. The lithium-ion battery has mal conductivity is significantly different from the axial thermal con­
become the preferred power source for electric vehicles due to its ad­ ductivity due to the spiral roll-like structure of the anode-separator-
vantages of high-power density, long lifetime and low self-discharge rate negative material layers, together with other accessory parts and com­
[1–3]. However, the characteristics of lithium-ion battery are closely ponents [17].
related to the operating temperature [4–5]. When the battery temper­ Unfortunately, the thermophysical parameters are usually unavai­
ature reaches the high threshold, the decomposition of the solid elec­ lable from the suppliers’ specifications. There still exist discrepancies in
trolyte interface layer would occur, or even worse, resulting in the the determination of the thermal parameters with high fidelity in liter­
battery thermal runaway accompanied with smoke, fire or explosion ature. In many cases, the thermal conductivity is simply taken as the
[6–8]. On the other hand, a low temperature leads to an increase in weighted average of the battery core layers without considering the
internal resistance and a sharp decline in the discharge capacity of necessary internal structure such as top and bottom insulation layers.
battery and driving performance of electric vehicles. It was reported that The weighted average is mostly inaccurate due to the unconsidered
capacity fade of lithium-ion battery after 600 cycles was only about factors such as the tab connection between the core materials and the
25.8% at − 10 ◦ C [9] and in extreme cases, it even prevents electric terminals as well as the non-trivial contact resistances among layers
vehicles from starting directly [10–11]. Clearly, thermal management

* Corresponding author.
E-mail address: [email protected] (H. Zhang).

https://doi.org/10.1016/j.applthermaleng.2023.120959
Received 11 October 2022; Received in revised form 26 April 2023; Accepted 11 June 2023
Available online 16 June 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Nomenclature Tn , Tp average temperatures of negative terminal and positive

terminals, respectively
A the cross-sectional area channel of the cooling pipeline ΔT temperature difference between the positive and negative
(m2) terminal surfaces of the battery (◦ C)
c specific heat of battery (J⋅kg− 1 ◦ C− 1) ΔTr radial temperature rise (◦ C)
cw specific heat of water (J⋅kg− 1 ◦ C− 1) ΔTw temperature rise of water (◦ C)
FO thermal Fourier number ΔTz axial temperature rise (◦ C)
h convective heat transfer coefficient (W⋅m− 2 ◦ C− 1) ΔTca temperature difference between the test configuration and
H length of cylindrical battery (m) the environment (◦ C)
kr radial thermal conductivity of battery (W⋅m− 1 ◦ C− 1) v velocity of flow (m⋅s− 1)
kz axial thermal conductivity of battery (W⋅m− 1 ◦ C− 1) Z axial direction
q volumetric heat generation rate of the cylinder model
(W⋅m− 3) Greek symbols
ρ density (kg⋅m− 3)
q’’H heat fluxes normal to the planes Z = 0 (W⋅m− 2)
ρw density of water (kg⋅m− 3)
Qt heat generated by the cylinder (W)
r radial direction Acronym
R radius (m) ARC accelerated rate calorimeter
Rr,e radial equivalent thermal resistance (◦ C⋅W− 1) DOE design of experiment
Rz,e axial equivalent thermal resistance (◦ C⋅W− 1) DSC differential scanning calorimetry
S terminal surface area of the cylindrical battery (m2) FEA finite element analysis
t time (s) NCM-523 Li[Ni0.5Co0.2Mn0.3]O2
T temperature (◦ C) NCM-811 Li[Ni0.8Co0.1Mn0.1]O2
T1 T4 measured positive and negative terminal temperatures SOC state of charge
shown in Fig. 3 (◦ C) TCSG thermally conductive silicone grease
Tamb environment temperature (◦ C) HCGS highly conductive graphite sheet
Tmax , Tout maximum and surface temperatures of cylindrical battery
respectively in Fig. 7 (◦ C)

[18–19]. Maleki et al. [20] utilized the laser flash technology to measure Sheng et al. examined the thermophysical parameters for prismatic
the thermal conductivity and specific heat of the electrode material lithium-ion batteries [27–28], in which the operating temperature had a
layers of the batteries, the in-plane conductivity values were measured greater effect on the specific heat of the battery than on the thermal
within 29.99–39.90 W⋅m− 1 ◦ C− 1 at 45 ◦ C. Nonetheless, the material test conductivity, whereas SOC had less effect on the two parameters. Drake
at the electrode level may not be directly applicable to the battery at the et al. [17] measured the thermal conductivity and specific heat of the
cell level due to the lack of consideration of the cell structure such as tab cylindrical batteries by heating the battery from the periphery, and the
connections with positive and negative electrode terminals. 18650-type and 26650-type cylindrical lithium-ion batteries were used
A number of research works were devoted to develop the measure­ as experimental specimens. Anisotropic thermal conductivities were
ment techniques on the thermophysical parameters of lithium-ion bat­ obtained with 0.20 W⋅m− 1 ◦ C− 1 in radial direction and 30.4 W⋅m− 1 ◦ C− 1
teries. Chen et al. [21] estimated the overall specific heat of the battery in axial direction for 18650 lithium-ion battery. The axial and radial
by consulting a large amount of data on the specific heat of each material specific heat results of 26650 lithium battery were 1605 J⋅kg− 1 ◦ C− 1 and
that made up the battery. Villano et al. [22] tested the specific heat of 1895 J⋅kg− 1 ◦ C− 1, respectively, with an inconsistency of 18.1%. In
each component of the battery by the method of differential scanning another study from the same group, the radial thermal conductivity was
calorimetry (DSC) to calculate the overall specific heat by weighted determined to be 0.39 W⋅m− 1 ◦ C− 1 for 26650 lithium-ion battery [29].
average. It is noted that the above measurement methods are limited to The heat loss along the radial heating direction was also characterized,
the layered materials without considering the heterogenous structure which amounted to 50% or more, varying with the test process. As such,
inside the battery cells. In [23–24], the thermophysical parameters can more efforts should be made to minimize the heat loss and improve the
be obtained by matching the numerical simulation results with the measurement accuracy.
experimental measurements without dissection of the battery cell. The literature survey shows that some progress has been made in the
Zhang et al. [23] simultaneously estimated the specific heat and thermal measurement of thermophysical parameters but scattering in the data is
conductivity of large-format laminated batteries. Feng et al. [24] pro­ still found among the different experimental studies. For cylindrical
posed a graphical method to determine the normal and spanwise ther­ batteries, the radial thermal conductivities varied from 0.15 W⋅m− 1 ◦ C− 1
mal conductivity of a prismatic ternary lithium-ion battery. to 3.4 W⋅m− 1 ◦ C− 1, and the axial thermal conductivities varied from
Nevertheless, the above methods require not only the experimental test 20.06 W⋅m− 1 ◦ C− 1 to 32.0 W⋅m− 1 ◦ C− 1, whereas the specific heat varied
but also the three-dimensional numerical simulation to determine from 760 J⋅kg− 1 ◦ C− 1 to 1720 J⋅kg− 1 ◦ C− 1 [15–16]. It is not likely that
thermophysical parameters, which requires long computational time the variations in battery types and formulations in different studies
and more efforts. A twin-type calorimeter was used to measure the could solely cause such large deviations [30]. Instead, the differences
specific heat of a cylindrical 18650 battery [25], which showed that the could be more attributed to the insufficient investigations on the
specific heat increased linearly with operating temperature from 20 ◦ C experimental measurement methods of the battery thermophysical
to 90 ◦ C. Bazinski et al [26] measured the specific heat of a pouch properties, which may lead to different levels of measurement uncer­
battery at different states of charge (SOC) by using an isothermal calo­ tainty. Though DSC is often used to measure the specific heat of the
rimeter and the specific heat was found to vary little with SOC but individual materials by dissecting the battery cell [22,31], it cannot
largely with the temperature. Efforts have also been made to obtain the determine the average specific heat of the battery cell as a whole. The
thermal conductivity and specific heat simultaneously. However, the use accelerated rate calorimeter (ARC) can be used to test the specific heat of
of dedicated instruments increases the cost as well as the testing time. a battery cell by adapting the test chamber size to the test cell, which

2
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

however is difficult to cope with the tests of realistic batteries of varying dimensional heat flow in the axial direction and thus the axial thermal
sizes. This equipment is also expensive accompanied with high opera­ conductivity and specific heat of the cylindrical battery can be measured
tional cost and long test time [32]. Less work was conducted on the axial accurately, even if the axial length is much greater than its diameter for
thermal conductivity of cylindrical battery. In the few measurements cylindrical batteries. The influences of the operating temperature and
such as Drake et al. [17], single battery cells of 18650 and 26650 were SOC on the thermophysical parameters of the battery were also studied.
used for the experimental test under quasi-steady state condition. It is Different from previous studies [20], the heat guarding measurement
important to note that the quasi-steady state measurement method is can determine the axial thermal conductivity and specific heat of cy­
fast in obtaining the thermal parameters including thermal conductivity lindrical batteries simultaneously in situ in a straightforward yet fast
and specific heat. However, heat loss would accumulate to a significant manner without dissecting the battery. In addition, the verification test
portion due to the essentially low thermal conductivity and moderate of the present measurement method was also conducted based on the
heat absorption rate of a battery cell. If the heat loss is not adequately standard stainless-steel samples.
accounted for, it would lead to incorrect estimation of specific heat and
thus inaccurate prediction in battery temperature and thus thermal 2. Measurement method
safety state. An overestimation in the specific heat, which is typical in
lack of heat loss analysis, would delay the warning time and increase the The schematic of a cylindrical lithium-ion battery subject to axial
risk of premature thermal runaway [16]. Some researchers used guarded heating is shown in Fig. 1. For a well-designed heat guarding case as
plate to measure thermal conductivity [33] to minimize the heat loss. shown in Fig. 1(c), a constant heat flux q˝H is input at the surface of Z = H
However, development of heat guarding method for quasi-steady state referred to the negative terminal surface, whereas the other boundaries
measurement for battery is seldom reported. Although there were are set to be adiabatic.
currently various methods available for measuring the thermophysical Subjected to the heat flux input q˝H , the temperature of the battery
parameters of batteries, it is challenging in the validation with a stan­ rises with time. According to the transient heat conduction theory [34],
dard sample of the same size and similar properties. In addition, there is the quasi-steady state can be reached after a short period of heating time
insufficient reporting on the impact of operating temperature on the if all the other surfaces are adiabatic. Namely, the solid structures at the
thermal physical performance of batteries. Therefore, the development quasi-steady state have the same temperature rise rate everywhere and
of accurate measurement method and verification against standard the temperature gradient or heat flux does not change with time.
samples is necessary to guarantee the measurement accuracy of battery Assuming the thermal conductivity and specific heat are constant
thermal parameters. quantities given a small temperature rise range, the following equation
In this work, a new quasi-steady state heat guarding measurement is obtained according to energy conservation.
method for the thermophysical parameters of cylindrical batteries is
proposed. In this method, the axial thermal conductivity and specific dT
q˝H S = mc (1)
heat of the batteries can be measured in situ within a short testing time dt
without destroying the battery. To shield the heat loss and improve the
where q˝H is heat flux normal to the negative terminal surface of Z = H, S
measurement accuracy, a pair of batteries was arranged at the center of
is the terminal surface area of the cylindrical battery, m and c are the
the fixture socket with the negative terminal surfaces facing to each
mass and average specific heat of the battery, and dT/dt represents the
other. Multiple pairs of batteries of the same type were symmetrically
average temperature rise rate across the battery.
arranged around the central battery pair under test to guard the unfa­
Thus, the average specific heat of the battery can be expressed as
vorable heat leaking from the lateral surfaces of the central battery pair.
In this way, the lateral heat leakage was minimized to ensure one-

Fig. 1. Schematic of axial heating for cylindrical lithium-ion battery (a) outer dimensions, (b) actual heat flow, (c) one-dimensional axial heat conduction under
adiabatic conditions.

3
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

q˝H with the heating time from the beginning. The curves of the temperature
c= (2)
ρH dT rises and their time derivatives (temperature rise rate) of positive and
dt
negative terminal surfaces of the battery in the heating process are
where ρ and H are density and length of the battery, respectively. shown in Fig. 2. Here the average temperature of the positive terminal is
For the case of axial heating of the battery, the heat conduction denoted as Tp = (T1 + T4)/2 whereas the average of the negative ter­
differential equation and boundary conditions are given by minal is Tn = (T2 + T3)/2. t1 is denoted as the start of the initial quasi-
steady state stage, at which the difference between the temperature
∂2 θ ρc ∂θ rise rates of the positive and negative terminal is reduced within 5%. As
= (3)
∂z2 kz ∂t the heating process goes on, the temperature of the battery keeps rising,
and the temperature difference between the negative and positive ter­
∂θ 1 ˝
= q at Z = H (4) minal remains at the same level, indicating the quasi-state heat transfer
∂z kz H is reached. However, the temperature difference ΔTca between the test
configuration and the environment keeps increasing, which leads to
∂θ
= 0at Z = 0 (5) more heat leakage from the test configuration to the environment. To
∂z
avoid large measurement error, the initial phase of the quasi-steady state
where θ(z, t) is the temperature rise, kz is the axial thermal conductivity (t1 ~ t2) is set with ΔTca not more than 15 ◦ C. In this stage, the mea­
of the battery. surement can be implemented with higher accuracy due to negligible
Considering that the initial temperature of the battery is same as the heat leakage from the lateral surfaces to the environment, which is to be
ambient temperature, i.e., θ(r, 0) = 0, the thermal differential equation discussed in the later section.
(3) is solved by Laplace transform method or variable separation
method, and the analytical solution of temperature field in time and 3. Experimental
space is obtained.
( ) ∑ The commercial cylindrical NCM-523 18650 lithium-ion batteries
n+1 (nπz)
q˝ q˝
θ(z, t) = H t + H Z 2 −
H2
+
∞ (− 1) 2q˝H H
cos
2
e(− FO (nπ) ) supplied from LG Chemical Co., Ltd. and NCM-523 21700 lithium-ion
ρcH 2kz H 3 n=1
kz (nπ)2 H batteries supplied from Panasonic Co., Ltd. were taken as the test bat­
(6) teries. The basic parameters of the battery are shown in Table 1.

where FO is the thermal Fourier number. The thermal Fourier number is 3.1. Test section
calculated during the axial conduction of the cell as follows
kz t The test device before and after assembly is shown in Fig. 3(a-b). The
FO = (7) test section consisted of seven pairs of 18650 cylindrical lithium-ion
ρcH 2
batteries in a fixture socket made from 3D printing, with the pair of
According to the formula (6), an expression relating to the axial the batteries under test located in the center and the other six pairs
thermal conductivity kz of the battery can be obtained by the difference
between θ(0, t) and θ(H, t)

q˝H H ∑∞ (− 1)n+1 2q˝ H

ΔT = − + 2 n=1 H
e(− FO (nπ )2 )
(8)
2kz kz (nπ)2

where ΔT is the temperature difference between the positive and

negative terminal surfaces of the battery.
The series term in Eq. (8) rapidly decays with the increase of heating
time due to the exponential dependence of time. A quantitative analysis
shows that the numerical value of the series terms can be ignored for FO
greater than 0.5. Therefore, the axial thermal conductivity kz of the
battery can be simplified to.

q˝H H
kz = − (9)
2ΔT
It is seen that the specific heat and axial thermal conductivity can be
attained based on the quasi-steady state heat transfer under the adia­
batic boundary conditions with the one-dimensional axial heat con­
duction as shown in Fig. 1(c). Nonetheless, heat leakage becomes larger
along with the heating time in the actual experimental test, especially
from the large lateral surface of the cylindrical battery such as 18650
batteries, which must be minimized to ensure the measurement accu­
racy. In addition, the initial phase of quasi-steady state should be uti­
lized to determine the axial thermal conductivity and the specific heat to
avoid excessively high battery temperature and thus enlarged heat
leakage.
In this work, the determination of the initial phase of quasi-steady
state is as follows. Upon heating, the battery temperature rises quickly
at the maximum rate before heat propagation to the inside. Before soon,
the initial temperature rise rate is reduced to a constant level due to heat
propagation to the battery core along the thermal path. In contrast, the
temperature rise rate of the positive terminal surface increases steadily Fig. 2. (a) Temperature variations and (b) temperature rise rates for the pos­
itive and negative terminals.

4
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Table 1 is to be discussed in the following section. In addition, the test equip­

Basic parameters of the test batteries. ment and materials are not difficult to implement without resorting to
Parameters 18650 21700 expensive equipment. In comparison, the ARC equipment would take
several hours to obtain the specific heat due to its gradual temperature
Chemical material composition NCM-523 NCM-523
Nominal capacity (Ah) 2.5 4.5 rise rate essentially requiring thermal equilibrium between the bulky
Nominal voltage (V) 3.6 3.7 chamber and the test sample [30].
Charge cut-off voltage (V) 4.2 4.2
Discharge cut-off voltage (V) 2.5 2.5
3.3. Finite element analysis and method validation
Max. charge current (A) 4 2
Max. discharge current (A) 30 10
Standard charge current (A) 1.25 1 The effectiveness of the heat guarding test device is visualized with
Standard charge end condition (A) 0.2 0.1 the finite element analysis (FEA). Taking 18650 and 21700 lithium-ion
Weight (g) 46 69 batteries as an example, the FEA on three different test configurations as
Length (mm) 65 70
shown in Fig. 4(a-c) was conducted in ANSYS FLUENT to quantify the
Diameter (mm) 18.2 21.2
heat leakage and thus visualize whether the heat guarding test device is
effective or not. The material parameters and boundary conditions in the
located evenly at the peripheral as the heat guarding batteries. Each pair numerical analysis are shown in Table 2.
of batteries of the same type was pressed end to end in between inserted Fig. 4 (d-e) shows the temperature contour plot of the battery
with an electric heating film encapsulated with polyimide sheets. The simulation model for the test configuration with heat guarding batteries
symmetric arrangement of the two central batteries also guarantees the in the quasi-steady state at 600 s for 18650 and 21700 lithium-ion
terminal heating without heat loss along the axial direction. Four T-type batteries. Although the temperature distribution in the heat guarding
thermocouples (wire diameter of 0.1 mm, National first-class accuracy) batteries is non-uniform due to heat leakage, good temperature unifor­
were allocated at the two ends of the two central batteries. For each mity is found for the central batteries. It ensures that the heat applied to
central battery, one thermocouple was fixed on the positive terminal the negative terminal surface of the central battery can be transferred to
surface (Z = 0), and the other was soldered at the lateral surface 2 mm the positive terminal surface mainly along the axial direction. The
away from the negative terminal surface (Z = 63 mm). The location of maximum temperature difference between the central battery pair and
the thermocouple away from the negative electrode avoids the direct the heat guarding batteries at the same axial position is only 0.9 ◦ C and
heating of the thermocouple by the heating film. 1.3 ◦ C for 18650 and 21700 lithium-ion batteries, indicating that the
Special care must be taken to ensure the heating power to go through heat leakage from the central battery pair can be successfully prevented
the battery along the axial direction. The polyimide electric heating by the heat guarding test device.
films with a thickness of 0.14 mm, the power rating of 10 W, the The heat leakage rate can be numerically determined by heat flows
applicable temperature range of − 150 ◦ C to 220 ◦ C were selected to from the lateral side of the central battery and the results are listed in
provide constant power heat input for the battery negative terminals, Table 3 at the different temperature difference ΔTca . It is seen that the
with 18 mm for 18650 battery and 21 mm for 21700 battery in diameter. maximum proportion of the heat leakage from the lateral surface in the
A highly conductive graphite sheet with a thickness of 46 µm was initial phase of the quasi-steady state is only 0.95% and 2.40% of the
attached to both ends of the heating film to homogenize the temperature total heat for the test configuration for 18650 and 21700 lithium-ion
of the battery. In addition, the HM-712N thermal conductive silicon batteries as given in Table 3 based on the simulation models in Fig. 4
grease with the thermal conductivity of 2.38 W⋅m− 1 ◦ C− 1 was thinly (a). Maximum deviations occur for the single-battery configuration, in
coated on the heating face of the battery, which was to reduce the which the relative deviation is 12.98% and 18.34% for 18650 and 21700
contact thermal resistance with the heating film. lithium-ion batteries, respectively, one order of magnitude higher than
Two layers of 10 mm thick aerogel were wrapped around the seven that of the present heat guarding device. In comparison, the present heat
pairs of batteries to reduce heat leakage and mimic an approximately guarding test device creates nearly adiabatic condition for the central
adiabatic environment. The silica aerogel blanket has a thermal con­ battery pair, which avoids excessive heat leakage from the lateral
ductivity as low as 0.02 W⋅m− 1 ◦ C− 1 at 25 ◦ C, which acts as excellent surface.
thermal insulation material to minimize the heat leakage from the large The measurement method was validated for the dummy battery
lateral side of the battery cell. samples made of standard ANSI 304 stainless steel with the three test
configurations, as shown in Fig. 4(a-c). The measurement results are
3.2. Test system shown in Table 4 as against the literature values in [36].
It is obvious from Table 4 that the measurement accuracy of thermal
Fig. 3(d) shows the schematic diagram of the test system. It consisted conductivity and specific heat of the test configuration with thermal
of the test section, a thermostatic chamber (GDJ-250, with the accuracy insulation device is much higher than that of the other two test con­
of 1 ◦ C), DC power supply (GWINSTEK GPD-2303S, with the voltage figurations without heat guarding batteries. In comparison, the thermal
accuracy of 0.01 V, and the current accuracy of 0.001 A), temperature conductivity and specific heat of the single battery test configuration
data acquisition equipment (AGILENT 34972A, with the uncertainty of were overestimated by 9.45% and 30.11%, respectively. As such, it is
0.3%) and computer. not surprising that the thermal parameter such as the specific heat for
The thermocouples were fixed on the positive terminal surface and the single battery test configuration could be obviously higher than the
the lateral surface 2 mm away from the negative terminal surface of the material test values reported in literature [20].
two batteries under test, recorded as T1 ~ T4 in turn, which are shown in It is noted that the deviation of specific heat is still high up to
Fig. 3(d). The thermocouple to measure ambient temperature was 10.70%. After scrutinizing the test setup, the thermal interfaces between
placed at the lower end of the acrylic box to reduce the influence of air the heating film and the battery were optimized to improve the mea­
convection. The enclosed acrylic box provided a constant ambient surement accuracy by incorporating a thin layer of graphite film (46 µm)
temperature for the test section without being disturbed by the fan flow between the heating film and the batteries. The adding of the graphite
in the thermostatic chamber. The advantages of the testing system in this film greatly enhanced the temperature uniformity of the negative ter­
experiment lie in the several aspects. The test can be completed in a time minals and further improves the measurement accuracy of specific heat.
period as short as several hundred seconds due to the essentially tran­ For this purpose, the test with TCSG and HCGS was conducted for the
sient measurement method. The measurement accuracy is improved by dummy battery with a heating power of 1.19 W as against the test with
one order of magnitude with the present heat guiding technique, which TCSG only. The test was repeated three times to minimize the random

5
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Fig. 3. Test configuration: (a) Before assembly; (b) After assembly; (c) Disassembly of battery; (d) Schematic diagram of the test system.

error due to personnel operation. The average thermal conductivity and negative electrodes of the battery at the initial stage of quasi-steady state
specific heat under 25 ◦ C of operating temperature are also tabulated in is maintained within a temperature rise of 15 ◦ C. The average temper­
Table 4. The initial results show that the measurement errors of axial ature of positive and negative terminals is taken as the average battery
thermal conductivity and specific heat of ANSI 304 stainless steel operating temperature. Since both the operating temperature and SOC
dummy battery reached 1.01% and 10.7% with the silicone thermal would affect the thermophysical parameters of the battery [35], the
grease (TCSG) only as thermal interface. With the addition of an ultra- experimental tests of 18650 and 21700 lithium-ion batteries were car­
thin layer of graphite film between the heating film and the battery, ried out at four operating temperature levels (-15 ◦ C, 5 ◦ C, 25 ◦ C, 45 ◦ C)
the associated measurement errors could be reduced to 0.13% and and five SOC levels (0, 0.25, 0.5, 0.75, 1) were controlled with the
4.99%, respectively. Clearly, the use of the ultrathin but highly charge station supplied by Neware Shenzhen. The SOC level is defined at
conductive graphite sheet helps to homogenize the temperature at the 25 ◦ C. When testing the thermophysical parameters of batteries at
battery bottom and thus improve the measurement accuracy. different SOC levels, the 14 batteries need to be disassembled and placed
This improvement can be explained that the TCSG may reduce the separately in the tester for charging and discharging to the specified
contact thermal resistance of the contact surface between the battery SOC. The changes of the axial thermal conductivity and specific heat of
and the heating film, but cannot eliminate the uneven heat generation the battery in the temperature range [-4.5 ◦ C, 55.5 ◦ C] were investi­
inherent in the heating film with in-built heating wire structure. When gated. It is worth noting that batteries under test were maintained at the
the TCSG combining with the highly conductive graphite sheet was used same SOC.
as the thermal interface material, the graphite sheet with a high in-plane
thermal conductivity can equalize the battery temperature, reduce the
4.1 Axial thermal conductivity measurement
fluctuation of battery temperature difference and temperature rise rate
and thus improve the measurement accuracy of thermal conductivity
The axial thermal conductivity of 18650 and 21700 lithium-ion
and specific heat.
batteries was measured three times using the quasi-steady state with
heat guarding device and the average of the three measurements was
4. Results and discussion
reported. The variations of the axial thermal conductivity of the battery
at different SOCs and temperatures are illustrated in Fig. 5. The results
In this section, the test results for the thermal conductivity and
show that the axial thermal conductivity of the battery increases almost
specific heat of cylindrical lithium-ion battery are presented and dis­
quadratically with temperature and SOC. The axial thermal conductivity
cussed. For consistency, the average thermal conductivity and specific
of the battery mainly depends on not only the thermal conductivity of
heat capacity at the initial stage of quasi-steady state are taken as the
the spiral roll of the battery core with metal current collectors, active
measured values. Setting the temperature of the incubator at a fixed
materials layers and the separator, but also the tab connection condi­
temperature such as 25 ◦ C, the temperature range by the positive and
tions between the core materials with the two terminals and interlayer

6
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Fig. 4. Three test configurations (a) with heat guarding batteries; (b)&(c) without heat guarding batteries; (d-e) the temperature nephogram of the thermal
simulation in the test configuration with heat guarding batteries for 18650 and 21700 lithium-ion batteries under heating power of 1 W and 2 W.

Table 2 Table 3
The parameters in simulation model. Comparison of heat leakage of three different test structures in Fig. 4.
Parameter Value for 18650 Value for 21700 Parameter ΔTca (◦ C) 3 6 9 12 15
batteries at 25 ◦ C batteries at 25 ◦ C
18650 Qloss for 0.09% 0.17% 0.36% 0.63% 0.95%
Diameter of battery (mm) 18.2 21.2 batteries Fig. 4(a)
Length of battery (mm) 65 70 Qloss for 2.08% 4.50% 6.47% 8.65% 10.64%
Density of battery (kg⋅m− 3) 2708 2792 Fig. 4(b)
Specific heat of battery 953 1028 Qloss for 2.85% 5.78% 8.15% 10.60% 12.98%
(J⋅kg− 1 ◦ C− 1) Fig. 4(c)
Axial thermal conductivity of 14.2 (this study) 15.2 (this study) 21700 Qloss for 1.81% 1.93% 2.05% 2.22% 2.40%
battery (W⋅m− 1 ◦ C− 1) batteries Fig. 4(a)
Radial thermal conductivity of 1.045 [35] 1.167 [35] Qloss for 2.76% 5.81% 8.18% 10.15% 11.92%
battery (W⋅m− 1 ◦ C− 1) Fig. 4(b)
Initial temperature (◦ C) 25 25 Qloss for 5.93% 9.92% 12.94% 15.70% 18.34%
hupper (W⋅m− 2 ◦ C− 1) 5 5 Fig. 4(c)
hlower (W⋅m− 2 ◦ C− 1) 5 5
hlateral (W⋅m− 2 ◦ C− 1) 10 10
q˝H (W⋅m− 2) 3844 5666 along the axial direction, leading to smaller contact resistance among
the battery core, tab connection and the two battery terminals. On the
other hand, the thermal conductivity also increases with increasing SOC,
contact conditions. One of the reasons why the axial thermal conduc­
for which the negative materials intercalated with more lithium ions
tivity increases with increasing temperature is that the thermal con­
may expand, leading to higher internal stress and thus better thermal
ductivities of the electrolytes increase with temperature rising, though
contact. It can also be seen from the figure that the axial thermal con­
the thermal conductivities of the solid compositions do not change much
ductivity of 21700 lithium-ion batteries is slightly higher than that of
in such small temperature range. The thermal contact characteristics
18650 lithium-ion batteries. This minor difference could be attributed to
could also be improved due to the thermal expansion of solid material

7
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Table 4
Test results and error analysis of dummy batteries made of ANSI 304 stainless steel.
Condition ANSI 304 stainless steel Single dummy battery Two dummy batteries 7 pairs of dummy batteries 7 pairs of dummy batteries
(TCSG) (TCSG + HCGS)

Standard value Test value Error Test value Error Test value Error Test value Error
1 ◦ − 1
Thermal conductivity (W⋅m− C ) 14.9 [36] 16.30 9.45% 15.50 4.33% 14.75 1.01% 14.92 0.13%
Specific heat (J⋅kg− 1 ◦ C− 1) 477 [36] 621 30.11% 600 25.85% 528 10.70% 501 4.99%

Fig. 5. Axial thermal conductivity of the 18650 and 21700 lithium-ion batteries at different SOCs and temperatures.

the differences in battery type, electrolyte, thickness of positive and

negative materials, etc. kZ,18650 = 11.86800 + 0.032380T + 3.93578SOC − 0.014707T
In this work, the axial thermal conductivity of 18650 lithium-ion × SOC − 0.000199T 2 − 0.851429SOC2 (10)
batteries falls in between 11.8–15.4 W⋅m− 1 ◦ C− 1, whereas those for
21700 lithium-ion battery fall in between 12.6–16.7 W⋅m− 1 ◦ C− 1. Our kZ,21700 = 12.68107 + 0.032410T + 3.04030SOC − 0.009181T
battery test results are smaller than the material test values such as given × SOC + 0.000100T 2 − 0.648496SOC2 (11)
in Maleki et al’ work [20]. In their work, the laser flash method was used
2
to measure the in-plane conductivity in the range of 29.99–39.90 The determination coefficient R of 18650 and 21700 lithium-ion
W⋅m− 1 ◦ C− 1 at 45 ◦ C. It is noted that the large deviations between the battery functions are 0.9976 and 0.9965 respectively. It indicates that
material test and our cell test are mainly due to the additional thermal temperature, SOC, and their second-order terms can explain 99.76% and
resistance associated with the tab connections with terminal metals in 99.65% of the change of the axial thermal conductivity, indicating a
real battery, along with other factors such as different material com­ good fitting degree of the function.
positions and structures and contact resistances among layers. In com­
parison, the present work is based on cell level test, which is more
4.2. Specific heat measurement
suitable for the realistic battery applications.
Based on the measurement results, the axial thermal conductivity of
The specific heat of the 18650 and 21700 lithium-ion batteries varies
18650 and 21700 lithium-ion batteries can be correlated into a function
with temperature at different SOCs is illustrated in Fig. 6(a) and Fig. 6
with respect to the temperature and SOC. Factor analysis using DOE
(c), respectively. The results show that the specific heat of 18650
yields the following expression.
lithium-ion batteries falls in between 884–1016 J⋅kg− 1 ◦ C− 1, whereas
those for 21700 lithium-ion battery fall in between 892–1082

8
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Fig. 6. Specific heat of the 18650 and 21700 lithium-ion batteries at different SOCs and temperatures in comparison with Sheng et al’s data.

J⋅kg− 1 ◦ C− 1. The specific heat of both batteries increases linearly, with compositions and manufacturing processes related to the different sup­
the increments of 13.18% and 19.05%, respectively, for temperature pliers. Sheng et al. [35] used NCM-811 cylindrical 18650 lithium-ion
varying from − 4.5 to 55.5 ◦ C. In general, the specific heat of the solid battery and NCM-523 21700 lithium-ion battery supplied from Shenz­
components in the battery, including the shell, separator and current hen BAK BATTERY Co., Ltd. In comparison, the test batteries in our
collector, increases with the increase of temperature. In addition, the study are the NCM-523 18650 lithium-ion batteries from LG Chemical
molecular rotation energy, lattice vibration energy, internal energy, and Co., Ltd. and NCM-523 21700 lithium-ion batteries from Panasonic Co.,
electronic kinetic energy of the involved materials all increase with Ltd.
temperature rising, and the increase in these four energies leads to the Based on the measurement results, the function between specific
increase structural entropy of materials. Thus, the specific heat of the heat of the 18650 and 21700 lithium-ion batteries and the temperature
battery increases with increasing temperature. This is consistent with was fitted by the least square technique dependent on temperature.
the trend obtained in literature [31] and [37]. Loges et al. [31]
c18650 = 899.02786 + 1.99582 T (12)
measured the specific heat of six types of batteries with different ma­
terials. The results showed that the specific heat of the battery was be­
c21700 = 907.52396 + 2.97268 T (13)
tween 870 and 1040 J⋅kg− 1 ◦ C− 1 at 25 ◦ C. The specific heat increased
2
with the increase of temperature, was less affected by the state of charge The determination coefficient R of this function is 0.99101 and
(SOC), and varied with the type of battery material. Wu et al. [37] 0.98029, indicating a good fitting degree for the specific heat.
measured and investigated the specific heat of 18650 cylindrical
lithium-ion batteries at different temperatures using a new calibrated
4.3. Comparison with the existing test methods in literature
calorimetric method. The results showed that the specific heat of the
battery increased with increasing temperature. On the other hand, the
The present heat guarding method is also compared with existing
SOC has little effect on the specific heat, as shown in Fig. 6(b) and Fig. 6
measurement methods on thermophysical parameters of batteries. As
(d), which is consistent with the statement in Ref. [26].
listed in Table 5, both the c and axial thermal conductivity kz parameters
The measurement data for cylindrical 18650 batteries and 21700
simultaneously can be measured by the present heat guarding method in
batteries at 0.5 SOC by Sheng et al. [35] based on the principle of quasi-
a shortest measurement time. In addition, the operating temperature
steady state heat conduction are also plotted in Fig. 6(a) and Fig. 6(c).
range of the heat guarding method is wider than that of the other test
With the similar trend in temperature, the specific heat measurement
methods of the published work, which caters for the application sce­
results of 18650 lithium-ion batteries in [35] are 6.28%− 11.05% higher
narios. In a word, the present heat guarding method has the advantages
than our measurement results and those for the 21700 lithium-ion bat­
of accurate and fast measurement of both specific heat and thermal
teries are within 4% from our measurement results for the varying
conductivity without resorting to expensive equipment, which is a novel
temperature. Such deviations could be due to different chemical
work in the battery thermal characterization and thermal management

9
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Table 5
Comparison with the published work on thermophysical parameters of batteries.
Source Battery type Test method Operating temperature Test Multiple parameters at the same time?
measurement time

Drake [17] 18650 and 26650 Quasi-steady state measurement 25 ◦ C 1000 s Yes (c and thermal conductivity)
cells
Maleki [20] pouch cells DSC (The battery needs to be disassembled) 45 ◦ C 60 s No (only c)
Zhang [23] large-format Simultaneous estimation (needs to compare 25 ◦ C 3000 s Yes (c, thermal conductivity and thermal
laminated cells with simulation) conductance)
Muhammad 26650 cells Time-varying heat flux method room temperature 2100 s No (only kr)
[29]
Wang [32] 18650 cells ARC 25 C–45 C
◦ ◦
80 min No (c and heat generation measured
separately)
This work 18650 and 21700 Heat guarding measurement method − 15 ◦ C to 45 ◦ C ~600 s Yes (c and kz)
cells

design.
1 H
Rz,e = (17)
2 kz S
4.4. Analysis of thermal management strategies
1
The axial thermal conductivity would have a significant impact on Rr,e = (18)
4πkr H
the thermal management strategies of the cylindrical battery, which is
analyzed in this section. Fig. 7(a) shows a homogeneous cylinder with Dividing Eq. (16) by Eq. (15), the ratio of the maximum temperature
heat production rate q, length H, radius R and cross-sectional area S, and rises based on radial cooling and bottom cooling is obtained as follows
the heat generated by the cylinder is Qt. The thermal management can be ΔTr Rr,e k z R2
implemented either in the axial direction with bottom cooling channels = = (19)
ΔTz Rz,e 2kr H 2
such as the design for Rivian EV battery pack [38] or in radial direction
with serpentine cooling channels such as the design for Tesla EV battery When the ratio ΔTr
ΔTz
> 1, the axial thermal resistance is smaller and
pack [39]. The present heat transfer analysis is to demonstrate the thus the bottom cooling is preferred to reduce the maximum tempera­
dependence of thermal design on the axial thermal conductivity of the ture . Vice versa, a lateral side cooling is preferred for ΔT
ΔTr
< 1 to reduce
cylinder. In realistic assembly, there exist air gaps or structural glues
z

the maximum temperature of the cylindrical battery.

between battery cells, which have much lower thermal conductivity and
The curves of the maximum temperature rise ratio of radial and axial
thus the cell-to-cell heat conduction is thus not considered for both
direction for different types of batteries with the ratio of axial and radial
lateral cooling and bottom cooling. Due to the small amount of heat
thermal conductivity are drawn in Fig. 8. Note that the radial thermal
generation of the battery cell under normal operation, the temperature
conductivity by Sheng et al. [35] has been adopted to perform the
distribution along the high-performance cold plate is not considered.
thermal design analysis for the present 18650 and 21700 batteries.
For bottom cooling in Ref. [40], we assumed v = 1 m/s coolant,
It can be seen from Fig. 8 that, at the ratio of axial and radial thermal
according to the energy balance equation.
conductivity is less than 104 for 18650 lithium-ion batteries, ΔTΔTr
z
< 1 and
ΔTw =
Qt
(14) thus a lateral side cooling effect is better than the bottom cooling. For
ρw Avcw 21700 lithium-ion batteries, a lateral side cooling effect is better at the
ratio of axial and radial thermal conductivity is less than 89. For the
where ΔTw is water temperature rise, ρw is density of water, A is the present 18650 and 21700 batteries measured axial thermal conductiv­
cross-sectional area channel of the cooling pipeline, cw is the specific ity, the temperature rise ratios are both less than one and thus it can be
heat of water. A calculation shows that the fluid temperature increases concluded that the lateral side cooling is applicable for both types of
by 0.049 ◦ C for the battery cell operated at 2.5C (1.39 W) [40], which is batteries.
negligible for the present cell level analysis. The 46800 lithium-ion battery with full-tab design is identified to be
According to heat transfer analysis, the maximum temperature rises one of the next-generation high power batteries for vehicle application
of ΔTz for the bottom cooling along with axial heat transfer occurs at the [42], which is also analyzed here. However, the battery is not available
top surface as shown in Fig. 7(a), whereas the maximum temperature in the market. To facilitate the analysis, we assumed the kr of 46800
rises of ΔTr along with radial heat transfer occurs at the center line of the batteries is the same as 21700 lithium-ion batteries measured by Sheng
battery. et al. [35], whereas the axial thermal conductivity is the same as the
For the bottom cooling, the maximum temperature rise is obtained as material test values by Maleki et al. [20] since their measurement was
[41] focused on the battery material which is more conformable to the 46800
1 H lithium-ion battery with full-tab design, avoiding the high thermal
ΔTz = Tmax − Tout = Qt = Qt Rz,e (15) resistance associated with the single tab connection for conventional
2 kz S
cylindrical batteries. The temperature difference ratio of 46800 batteries
For the lateral cooling typically with serpentine tubes, the maximum with assumed thermal conductivity vs the ratio of kz and kr is shown in
temperature rise is given as Fig. 8. It is seen that the ratio of ΔTr over ΔTz is larger than one, indi­
Qt cating a bottom cooling is preferred for the new 46800 battery. It should
ΔTr = Tmax − Tout = = Qt Rr,e (16) be noted that the present thermal management is a preliminary analysis
4πkr H
with measured battery thermal parameters through the simplified
The equivalent axial and radial equivalent thermal resistances,
thermal resistance network at the cell level. More careful analysis of
defined by the ratio of the temperature difference to the heat flow, can
thermal management in full scale could be conducive in the next stage
be expressed in term of geometrical dimensions and material properties,
work.
namely.

10
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

Fig. 7. (a) bottom cooling; (b) axial cooling with curved tube; (c) the size parameter, boundary condition and equivalent thermal resistance along the axial direction;
(d) heat conduction model with equivalent thermal resistance along the radial direction.

In this work, the uncertainty of T-type thermocouple with level one

accuracy is estimated to be 1% considering the temperature accuracy of
0.1 ◦ C within 10 ◦ C temperature range. The uncertainty of Agilent data
acquisition instrument, vernier caliper and electronic balance is 0.3%,
0.1% and 1%, respectively. In addition, the accuracy of the voltage and
current of the DC supply is 0.01 V and 0.001 A, respectively, and thus the
uncertainty of the heating power is about 0.3%. The uncertainty caused
by specific experimental instruments in each experiment is shown in
Table 6.
Summing the uncertainty of experimental equipment and the
random error, the uncertainties of 18650 battery, the heat flux ( q˝H ),
Δq˝
H

density (Δρρ), length (ΔH

H ) and temperature ( T ) are obtained, which are
ΔT

1.64%, 1.13%, 0.23% and 2.60%, respectively. The uncertainties of

21700 battery, heat flux ( q˝H ), density (Δρρ), length (ΔH
H ) and temperature
Δq˝
H

(ΔT
T ) are 1.96%, 1.12%, 0.22% and 2.60%, respectively.
According to the root-sum-square technique [43], the uncertainties
of the axial thermal conductivity and specific heat of the battery can be
calculated by the following equations.
Fig. 8. Variations of the maximum temperature rise ratio with the axial and
radial thermal conductivity ratio for different types of batteries.

4.5. Experimental uncertainty analysis Table 6

Uncertainty analysis of experimental instruments.
The test uncertainty of the axial thermal conductivity and specific Experimental instrument Accuracy Uncertainty
heat of the battery can be attributed to the inaccuracy of test equipment, Thermocouple National first-class accuracy 1%
the random error associated with the dimension’s temperatures Agilent data acquisition instrument 0.3% 0.3%
measured with thermocouples, and some human factors caused by the Vernier caliper 0.1% 0.1%
Electronic balance 1% 1.0%
operation.
DC supply 0.01 V, 0.001 A 1.0%

11
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( ˝ )2 ( ) ( )2 ( )2 Declaration of Competing Interest
Δc ΔqH Δρ 2 ΔH ΔT
= ˝
+ + + (20)
c qH ρ H T
The authors declare that they have no known competing financial
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( ˝ )2 ( )2 ( )2 interests or personal relationships that could have appeared to influence
Δkz ΔqH ΔH ΔT the work reported in this paper.
= + + (21)
kz q˝H H T
Data availability
The uncertainty of the axial thermal conductivity kz and specific heat
c is approximately 3.1% and 3.3% maximum for the 18650 lithium-ion
Data will be made available on request.
batteries and 3.3% and 3.5% for the 21700 lithium-ion batteries
respectively by plus the random error due to personel assembly and
Acknowledgement
operation.
Financial supports from Shanghai Municipal Natural Science Foun­
5. Conclusions
dation (21ZR1426300) and Natural Science Foundation of China
(51876113) are acknowledged.
A novel quasi-steady state heat guarding measurement method is
proposed for measuring the axial thermal conductivity and specific heat
References
of cylindrical battery based on the principle of quasi-steady state heat
conduction. The axial thermal conductivity and specific heat of 18650 [1] X.T. Liu, Z.H. Chen, C.B. Zhang, J. Wu, A novel temperature-compensated model
and 21700 cylindrical lithium-ion batteries with large a length to for power Li-ion batteries with dual-particle-filter state of charge estimation, Appl.
diameter ratio were measured at different levels of temperature and Energy 123 (2014) 263–272, https://doi.org/10.1016/j.apenergy.2014.02.072.
[2] S.S.Y. Ng, Y.J. Xing, K.L. Tsui, A naive Bayes model for robust remaining useful life
SOC. This method has the advantages of fast, cost-effective easy oper­ prediction of lithium-ion battery, Appl. Energy 118 (2014) 114–123, https://doi.
ation and nondestructive, and can obtain the axial thermal conductivity org/10.1016/j.apenergy.2013.12.020.
and specific heat of cylindrical batteries simultaneously and accurately. [3] M. Mathew, Q.H. Kong, J. McGrory, M. Fowler, Simulation of lithium-ion battery
replacement in a battery pack for application in electric vehicles, J. Power Sources
In addition, the impact of the axial thermal conductivity on thermal 349 (2017) 94–104, https://doi.org/10.1016/j.jpowsour.2017.03.010.
management strategies for different batteries is analyzed and discussed. [4] Z. Lu, X.L. Yu, L.C. Wei, F. Cao, L.Y. Zhang, X.Z. Meng, L.W. Jin, A comprehensive
The experimental results show that the axial thermal conductivity of experimental study on temperature-dependent performance of lithium-ion battery,
Appl. Therm. Eng. 158 (2019), 113800, https://doi.org/10.1016/j.
18650 and 21700 battery are within 11.8–15.4 W⋅m− 1 ◦ C− 1 and applthermaleng.2019.113800.
12.6–16.7 W⋅m− 1 ◦ C− 1, respectively, whereas those for specific heat are [5] S. Ma, M. Jiang, P. Tao, C. Song, J. Wu, J. Wang, T. Deng, W. Shang, Temperature
within 884–1016 J⋅kg− 1 ◦ C− 1 and 892–1082 J⋅kg− 1 ◦ C− 1, respectively. effect and thermal impact in lithium-ion batteries: A review, Prog. Nat. Sci.: Mater.
Int. 28 (6) (2018) 653–666.
Axial thermal conductivity of the battery increases almost quadratically
[6] Q.S. Wang, P. Ping, X.J. Zhao, G.Q. Chu, J.H. Sun, C.H. Chen, Thermal runaway
with temperature and SOC, whereas specific heat increases linearly with caused fire and explosion of lithium-ion battery, J. Power Sources 208 (2012)
temperature but varied little with SOC. 210–224, https://doi.org/10.1016/j.jpowsour.2012.02.038.
[7] Z.C. Feng, Y.W. Zhang, Safety monitoring of exothermic reactions using time
The numerical simulation and experimental results show that the
derivatives of temperature sensors, Appl. Therm. Eng. 66 (2014) 346–354, https://
heat guarding measurement method composed of guarding batteries doi.org/10.1016/j.applthermaleng.2014.02.023.
possesses an excellent thermal shielding effect. The heat leakage can be [8] X. Feng, D. Ren, X. He, M. Ouyang, Mitigating Thermal Runaway of Lithium-Ion
minimized to 2.4% in the initial phase of the quasi-steady state based on Batteries, Joule 4 (2020) 743–770, https://doi.org/10.1016/j.joule.2020.02.010.
[9] Y.C. Zhang, C.Y. Wang, X.D. Tang, Cycling degradation of an automotive LiFePO4
the finite element analysis, which was one order of magnitude less than lithium-ion battery, J. Power Sources 196 (2011) 1513–1520, https://doi.org/
the single battery test configuration without heat guarding batteries. 10.1016/j.jpowsour.2010.08.070.
The maximum temperature difference between the central battery pair [10] C.K. Huang, J.S. Sakamoto, J. Wolfenstine, S. Surampudi, The Limits of Low-
Temperature Performance of Li-ion Cells, J. Electrochem. Soc. 147 (2000)
and the heat guarding batteries was experimentally identified to be 2893–2896, https://doi.org/10.1149/1.1393622.
around 1.3 ◦ C at the same height along the battery axis. [11] Z.Y. Ling, X.Y. Wen, Z.G. Zhang, X.M. Fang, X.N. Gao, Thermal management
Experimental results show that the axial thermal conductivity of the performance of phase change materials with different thermal conductivities for Li-
ion battery packs operated at low temperatures, Energy 144 (2018) 977–983,
lithium-ion battery increases almost quadratically with battery tem­ https://doi.org/10.1016/j.energy.2017.12.098.
perature and SOC. The specific heat of the lithium-ion battery increases [12] Y.Y. Liu, Y.Y. Zhu, Y. Cui, Challenges and opportunities towards fast-charging
linearly with the increase of battery temperature, which remains almost battery materials, Nat. Energy 4 (2019) 540–550, https://doi.org/10.1038/
s41560-019-0405-3.
unchanged at different SOCs. The presently obtained data are useful in
[13] Z.H. Rao, S.F. Wang, A review of power battery thermal energy management,
the design and optimization of battery thermal management system. Renew. Sustainable Energy Rev. 15 (2011) 4554–4571, https://doi.org/10.1016/j.
Equivalent thermal resistance analysis is conducted to assess the effec­ rser.2011.07.096.
tiveness of bottom cooling and lateral side cooling techniques for [14] H. Jouhara, N. Khordehgah, N. Serey, S. Almahmoud, S.P. Lester, D. Machen,
L. Wrobel, Applications and thermal management of rechargeable batteries for
different types of cylindrical batteries. The results show that the lateral industrial applications, Energy 170 (2019) 849–861.
side cooling is more suitable for the present slim 18650 and 21700 [15] G. Offer, Y. Patel, A. Hales, L.B. Diza, M. Marzook, Cool metric for lithium-ion
lithium-ion batteries. For the new-type 46800 lithium-ion battery with batteries could spur progress, Nature 582 (2020) 485–487, https://doi.org/
10.1038/d41586-020-01813-8.
assumed thermal conductivity, the bottom cooling is recommended [16] Y. Zeng, D. Chalise, S.D. Lubner, S. Kaur, R.S. Prasher, A review of thermal physics
instead as the optimal cooling configuration. and management inside lithium-ion batteries for high energy density and fast
charging, Energy Storage Mater. 41 (2021) 264–288.
[17] S.J. Drake, D.A. Wetz, J.K. Osanek, S.P. Miller, J.M. Heinzel, A. Jain, Measurement
CRediT authorship contribution statement of anisotropic thermophysical properties of cylindrical Li-ion cells, J. Power
Sources 252 (2014) 298–304, https://doi.org/10.1016/j.jpowsour.2013.11.107.
Hong Yu: Methodology, Investigation, Formal analysis, Writing – [18] R. Ponnappan, T.S. Ravigururajan, Contact thermal resistance of Li-ion cell
electrode stack, J. Power Sources 129 (2004) 7–13, https://doi.org/10.1016/j.
original draft, Visualization. Hengyun Zhang: Conceptualization, Re­ jpowsour.2003.11.006.
sources, Supervision, Validation, Writing – review & editing, Funding [19] Y. Yang, X.P. Huang, Z.Y. Cao, G. Chen, Thermally conductive separator with
acquisition, Project administration. Jinghe Shi: Data curation, Investi­ hierarchical nano/microstructures for improving thermal management of batteries,
Nano Energy 22 (2016) 301–309, https://doi.org/10.1016/j.nanoen.2016.01.026.
gation. Shunbo Liu: Writing – review & editing, Supervision. Zhaozang
[20] H. Maleki, H. Wang, W. Porter, J. Hallmark, Li-ion polymer cells thermal property
Yi: Writing – review & editing. Shen Xu: Data curation, Investigation. changes as a function of cycle-life, J. Power Sources 263 (2014) 223–230, https://
Xinwei Wang: Data curation, Investigation. doi.org/10.1016/j.jpowsour.2014.04.033.

12
H. Yu et al. Applied Thermal Engineering 231 (2023) 120959

[21] S.C. Chen, C.C. Wan, Y.Y. Wang, Thermal analysis of lithium-ion batteries, [33] A. Palacios, L. Cong, M.E. Navarro, Y.L. Ding, C. Barreneche, Thermal conductivity
J. Power Sources 140 (2005) 111–124, https://doi.org/10.1016/j. measurement techniques for characterizing thermal energy storage materials-A
jpowsour.2004.05.064. review, Renew. Sustainable Energy Rev. 108 (2019) 32–52, https://doi.org/
[22] P. Villano, M. Carewska, S. Passerini, Specific heatacity of lithium polymer battery 10.1016/j.rser.2019.03.020.
components, Thermochim. Acta 402 (2003) 219–224, https://doi.org/10.1016/ [34] H.P. Hu, On the Conditions Necessary and Sufficient for Establishing
S0040-6031(02)00612-3. Quasistationary Temperature Distribution, J. Appl. Sci. 22 (2004) 91–93.
[23] J.B. Zhang, B. Wu, Z. Li, J. Huang, Simultaneous estimation of thermal parameters [35] L. Sheng, Z.D. Zhang, L. Su, H.Y. Zhang, H. Zhang, Y.D. Fang, K. Li, W. Ye, Quasi
for large-format laminated lithium-ion batteries, J. Power Sources 259 (2014) steady state method to measure thermophysical parameters of cylindrical lithium
106–116, https://doi.org/10.1016/j.jpowsour.2014.02.079. ion batteries, J. Power Sources 485 (2021), 229342, https://doi.org/10.1016/j.
[24] X.N. Feng, J.J. Li, L. Wang, J. Gao, X.M. He, M.G. Ouyang, Experiments and jpowsour.2020.229342.
modeling of anisotropic thermal conductivity of lithium-ion batteries, [36] F.P. Incropera, D.P. DeWitt, T.L. Bergeman, Fundamentals of Heat and Mass
J. Automotive Safety and Energy 3 (2012) 158–164, https://doi.org/10.3969/j. Transfer, sixth ed., John Wiley & Sons Inc, 2007.
issn.1674-8484.2012.01.009. [37] Q.Y. Wu, H.Y. Zhang, J.W. Li, Calibrated Calorimetry for Measuring the Specific
[25] K. Onda, T. Ohshima, M. Nakayama, K. Fukuda, T. Araki, Thermal behavior of heat and Heat Generation Rate of Lithium-ion Battery, Automot. Eng. 42 (2020)
small lithium-ion battery during rapid charge and discharge cycles, J. Power 59-65, 10.19562/j.chinasae.qcgc.2020.01.009.
Sources 158 (2006) 535–542, https://doi.org/10.1016/j.jpowsour.2005.08.049. [38] J. King, Electric vehicle thermal management system with battery heat storage, U.
[26] S.J. Bazinski, X. Wang, Experimental study on the influence of temperature and S Patent 15, 716, 333[P]. 2018-3-29. https://patents.google.com/patent/
state-of-charge on the thermophysical properties of an LFP pouch cell, J. Power US20180086224A1/en.
Sources 293 (2015) 283–291, https://doi.org/10.1016/j.jpowsour.2015.05.084. [39] C.R. Zhao, W.J. Cao, T. Dong, F.M. Jiang, Thermal behavior study of discharging/
[27] L. Sheng, L. Su, H.Y. Zhang, Y.D. Fang, H.F. Xu, W. Ye, An improved calorimetric charging cylindrical lithium-ion battery module cooled by channeled liquid flow,
method for characterizations of the specific heat and the heat generation rate in a Int. J. Heat Mass Transfer 120 (2018) 751–762, https://doi.org/10.1016/j.ijheat
prismatic lithium ion battery cell, Energy Convers. Manage. 180 (2019) 724–732, masstransfer. 2017.12.083.
https://doi.org/10.1016/j.enconman.2018.11.030. [40] H.Y. Zhang, X.Y. Wu, Q.Y. Wu, S. Xu, Experimental investigation of thermal
[28] L. Sheng, L. Su, H.Y. Zhang, Experimental determination on thermal parameters of performance of large-sized battery module using hybrid PCM and bottom liquid
prismatic lithium ion battery cells, Int. J. Heat Mass Transfer 139 (2019) 231–239, cooling configuration, Appl. Therm. Eng. 159 (2019), 113968, https://doi.org/
https://doi.org/10.1016/j.ijheatmasstransfer.2019.04.143. 10.1016/j.applthermaleng.2019.113968.
[29] M.B. Ahmed, S. Shaik, A. Jain, Measurement of radial thermal conductivity of a [41] J.J. Zhu, H.Y. Zhang, G.P. Wu, S.L. Zhu, W. Liu, Thermal performance of
cylinder using a time-varying heat flux method, Int. J. Heat Mass Transfer 129 cylindrical battery module with both axial and radial thermal paths: Numerical
(2018) 301–308, https://doi.org/10.1016/j.ijthermalsci.2018.03.008. simulation and thermal resistance network analysis, J. Energy Storage 49 (2022),
[30] F. Richter, S. Kjelstrup, P.J.S. Vie, S.O. Burheim, Thermal conductivity and internal 104197, https://doi.org/10.1016/j.est.2022.104197.
temperature profiles of Li-ion secondary batteries, J. Power Sources 359 (2017) [42] T.G. Tranter, R. Timms, P.R. Shearing, D.J.L. Brett, Communication—Prediction of
592–600, https://doi.org/10.1016/j.jpowsour.2017.05.045. thermal issues for larger format 46800 cylindrical cells and their mitigation with
[31] A. Loges, S. Herberger, P. Seegert, T. Wetzel, A study on specific heat capacities of enhanced current collection, J. Electrochem. Soc. 167 (2020), 160544, https://doi.
Li-ion cell components and their influence on thermal management, J. Power org/10.1149/1945-7111/abd44f.
Sources 336 (2016) 341–350, https://doi.org/10.1016/j.jpowsour.2016.10.049. [43] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Therm. Fluid
[32] L. Wang, X.N. Feng, G. Xue, etc. ARC experimental and data analysis for safety Sci. 1 (1988) 3–17, https://doi.org/10.1016/0894-1777(88)90043-X.
evaluation of Li-ion batteries, Energy Storage Sci. Technol. 7(2018): 1261-1270.
10.12028/j.issn.2095-4239.2018.0161.

13