Journal of Energy Storage 65 (2023) 107338

Contents lists available at ScienceDirect

Journal of Energy Storage

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

Research papers

Measurement of thermophysical parameters and thermal modeling of

21,700 cylindrical battery
Qichao Wu a, Rui Huang a, b, Xiaoli Yu a, b, *
a
College of Energy Engineering, Zhejiang University, Hangzhou 310027, China
b
Key Laboratory of Automotive Intelligent Thermal Management Science and Technology, Taizhou 317200, China

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

Keywords: There is temperature unevenness inside the operating battery, and the internal temperature distribution of the
Lithium-ion battery battery has gradually attracted attention. To establish a thermal model of the 21,700 cylindrical battery that can
21,700 cylindrical battery reflect the internal temperature distribution, thermophysical parameters including anisotropic thermal con­
Thermophysical parameters
ductivity and specific heat capacity are tested through experiments. The thermal model includes the heat gen­
Temperature measurement
eration part and heat transfer part. Thermophysical parameters are essential for the heat transfer part. Hybrid
Thermal model
pulse power characterization (HPPC) tests are carried out to obtain battery internal resistance, and the entropy
coefficient of the battery is also obtained by tests. Internal resistance and entropy coefficient data are used as
inputs for the battery heat generation part. The core temperature and surface temperature of the operating
battery are measured by the thermocouple embedded inside the battery and the thermocouple pasted on the
battery surface. The calculated core and surface temperatures from the battery thermal model are both in good
agreement with the measured temperature results. The maximum core temperature of 2C discharging battery is
about 60 ◦ C and the maximum temperature difference inside the 21,700 battery reaches 3 ◦ C. The established
simulation model can be used for extended research work. When heat dissipation is enhanced by increasing the
convection heat transfer coefficient, the overall temperature of the battery decreases but the temperature non-
uniformity inside the battery is aggravated.

1. Introduction battery and heat the center with nichrome wire to measure battery
radial thermal conductivity [6]. Yu et al. tested the specific heat capacity
Cylindrical lithium-ion batteries are widely used due to the advan­ through the battery oil mixed heat transfer method [7]. Panchal et al.
tages of high performance and stable uniformity [1]. When the battery is established a 18,650 battery simulation model and calculated the tem­
operating, self-generated heat accumulates [2]. Because of the multi- perature results of different discharge rates [8]. The size of the 26,650
layer winding structure inside the cylindrical battery, the radial ther­ battery becomes larger, the capacity is also improved, and normally it
mal conductivity of the battery is much smaller than the axial thermal generates more heat [9]. There is also some related research. Drake et al.
conductivity [3]. There is temperature unevenness inside the operating measured the anisotropic thermal conductivity and heat capacity of a
battery, and the measured surface temperature can be far from the 26,650 cylindrical battery using the adiabatic heating method [10]. A
maximum temperature inside the battery [4]. The internal temperature simplified two-dimensional battery thermal model is also established.
distribution of the battery under different operating conditions is also Compared to the 18,650 and 26,650 batteries, the 21,700 cylindrical
affected by heat transfer and has gradually attracted attention [5]. battery is the relatively new standard and less studied [11]. The 21,700
Currently, the mainly used cylindrical lithium-ion batteries are battery has high capacity and energy density [12], so the temperature
18,650 (18 mm diameter, 65 mm height), 26,650 (26 mm diameter, 65 distribution inside the operating battery has research value.
mm height), and 21,700-type cells (21 mm diameter, 70 mm height). The methods to study the internal temperature distribution of cy­
There are many studies including measurement of thermophysical pa­ lindrical batteries can be divided into experimental tests and simulations
rameters and thermal modeling for the 18,650 cylindrical battery due to [13]. Experimental tests of temperature distribution require some
its long history. For example, Bhundiya et al. disassembled a 18,650 modifications to batteries, for example, Du et al. utilized infrared

* Corresponding author at: College of Energy Engineering, Zhejiang University, Hangzhou 310027, China.
E-mail address: [email protected] (X. Yu).

https://doi.org/10.1016/j.est.2023.107338
Received 13 November 2022; Received in revised form 28 March 2023; Accepted 2 April 2023
Available online 14 April 2023
2352-152X/© 2023 Elsevier Ltd. All rights reserved.
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

thermal imaging technology to visualize the temperature distribution of 2.1. Measurement of thermophysical parameters
the internal cross-section in a 18,650 cylindrical battery [14]. When
fabricating, the battery used infrared optical glass instead of metal on Thermophysical parameters including thermal conductivity and
one end of the shell, allowing the infrared light to pass through This specific heat capacity of the battery need to be measured for thermal
method can only see the temperature distribution of a surface and obtain modeling. Due to the thermal contact resistance between layers, The
the two-dimensional temperature distribution on one end of the battery multi-layer winding structure inside the cylindrical battery leads to the
without the axial temperature data. Compared with experimental tests, anisotropic thermal conductivity of the battery body. The radial and
the implementation costs of simulations are lower [15]. However, due to axial thermal conductivity of the cylindrical battery were measured
the differences in the parameters and internal structure of each battery, respectively. According to the experimental measurement method pro­
it is necessary to obtain battery thermophysical parameters before posed by Drake et al. [10], the 21,700 battery was hung in the vacuum
establishing the battery thermal model, and the accuracy of the simu­ tank, and experimental devices are shown in Fig. 1 (a). A vacuum pump
lation model needs to be verified [16]. Jeon et al. built a thermal model was used to vacuum the tank to eliminate heat conduction and heat
of a 18,650 battery and calculated the internal temperature distribution convection between the battery and the environment. It minimized the
of the battery [17]. They verified the thermal model by comparing the heat dissipation of the battery. The heating film wire and thermocouple
calculated battery surface temperature with the experimentally tested wire were drawn from the hole at the bottom of the vacuum tank. All
surface temperature. Inspired by some literature on measuring the core gaps were sealed with glue to ensure air tightness. The heating film was
temperature inside the cylindrical battery [18,19], in our view, the connected to the controllable power supply and the heating power was
verification of the battery thermal model can be further improved. The set to 1 W. The battery was heated on the side surface when measuring
surface temperature and core temperature from simulated battery the radial thermal conductivity, and the battery was heated on the
temperature distribution can be compared with temperature data bottom surface to measure the axial thermal conductivity. Calibrated
measured by temperature sensors on the battery surface and embedded OMEGA T-type thermocouples were used to measure temperature, and
inside the battery, respectively. Such verification has high requirements the thermocouple output was sampled at 1 Hz using a National In­
for the accuracy of the battery heat generation model and thermo­ struments (NI) cRIO-9214 module. The positions of the heating film and
physical parameters. To our knowledge, we are unaware of any studies thermocouple on the two heating methods are shown in Fig. 1 (b). The
that have done this. This study shows the whole process from the plastic wrapping film on the battery surface was stripped off to avoid
measurement of thermophysical parameters to the establishment of a interference in temperature measurement, and only a very small part of
21,700 battery thermal model including the heat generation part and the plastic film near the positive electrode remained for insulation
heat transfer part. Moreover, the battery core temperature and surface protection. Thermophysical parameters of the battery including aniso­
temperature measured by experiments are used to verify the simulation tropic thermal conductivity and specific heat capacity can be calculated
results. Some extended studies about the internal temperature distri­ by fitting the temperature rising curve recorded in the process of heating
bution of the battery are carried out. the battery.
The rest of the paper is arranged as follows: Section 2 states the
experimental section, including battery thermophysical parameters,
internal resistance, and entropy coefficient tests. Core and surface 2.2. Internal resistance and entropy coefficient tests
temperature measurements of the operating battery are also shown.
Section 3 presents the thermal model of the 21,700 cylindrical battery. Based on the heat generation model proposed by Bernardi et al. [21],
The thermal model mainly includes the battery heat generation part and the internal resistance and entropy coefficient of the battery are needed
the heat transfer part. Section 4 includes the results and discussion of as input for the heat generation part of the battery thermal model. Ac­
both experiments and simulations. Section 5 concludes the paper. cording to the US FreedomCAR Battery Test Manual [22], hybrid pulse
power characterization (HPPC) tests were used to obtain direct current
2. Experimental section (DC) internal resistance of the battery. Internal resistance can be
calculated by monitoring the voltage change when a fixed charging or
The research object selected in this study is a commercial 21,700 discharging current pulse is applied. Battery internal resistance under
cylindrical battery from Samsung company. The cell model is INR21700- different state of charge (SOC) was tested in the environmental chamber
50E and basic specifications [20] of the battery are given in Table 1. and the current pulse was determined by combining the specification
NEWARE CT4008-5V12A charging and discharging equipment is used manual of the battery [20]. Test steps of HPPC at a specific temperature
as the battery operating control device. ESPEC GPU-3 environmental and SOC are shown in Table 2. The fully charged battery was discharged
chamber is used to create a constant temperature environment for bat­ to 90 % SOC with 1C rate and shelved for 1 h to achieve electrochemical
tery tests. The standard battery charging process consists of charging the and thermal equilibrium. Then the battery was tested according to
battery at a constant current rate of 1/2C (1C = 4.9A) after 1 h thermal Table 2. After the current SOC test, the battery was discharged to the
equilibrium in 25 ◦ C environmental chamber, and when the charging next SOC with 1C rate and tested after 1 h of shelving. HPPC tests range
cut-off voltage is reached, the charging mode is switched to constant from 90 % SOC to 10 % SOC with an interval of 10 % SOC. Because the
voltage charging until the current dropped to 0.02C (98 mA). The bat­ battery electrochemical system is greatly affected by temperature, in­
tery is fully charged before each discharge test according to the standard ternal resistance of the battery varies at different temperatures. The
charging process. battery was fully charged according to the standard charging process
and then applied a complete suite of HPPC tests in the environmental
chamber of different temperatures. The test temperature was set from
15 ◦ C to 65 ◦ C with an interval of 10 ◦ C. Before each test, the battery was

Table 1
Basic specifications of the battery.
Cell model Nominal capacity Charging cut-off voltage Discharging cut-off voltage Weight Height Diameter Density (kg
(Ah) (V) (V) (g) (mm) (mm) m− 3)

SAMSUNG INR21700-
4.9 4.2 2.5 69 70 21 2846
50E

2
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 1. Measurement of battery thermophysical parameters (a) Experimental devices; (b) Two heating methods for anisotropic thermal conductivity tests.

content: <1 PPM) to avoid the impact of water and oxygen on the
Table 2
electrolyte, and a calibrated OMEGA T-type thermocouple was
Test steps of HPPC at a specific temperature and SOC.
embedded into the battery, as shown in Fig. 2 (a), which is described in
Step name Time (s) Rate (C) Current (A) more detail in our previous research work [27]. Fig. 2 (b) shows that the
Constant-current discharge 10 1 4.9 battery with an embedded thermocouple was discharging in the envi­
Shelve 40 0 0 ronmental chamber and another calibrated thermocouple was also
Constant-current charge 10 0.75 3.675
pasted on the center of the battery surface. The height of the core and
surface thermocouples was the same, both at half of the battery height.
shelved for 1 h to ensure thermal equilibrium. The voltage and current Discharging tests and electrochemical impedance spectroscopy (EIS)
data obtained in HPPC tests were used to calculate internal resistance of tests were both carried out to analyze the impact of the embedded
the battery. thermocouple on the battery performance. As shown in Fig. 3 (a), there
When temperature changes, the positive and negative potential of is little difference in discharging curves of both 1C and 2C before and
the battery will change, and as a whole, the open-circuit voltage will be after embedding the thermocouple. In addition, the core and surface
different [23]. The entropy coefficient related to the electrochemical temperature of the battery was recorded during 1C and 2C discharging
reactions can be used to describe this phenomenon [24]. The entropy process. Temperature data was used to compare with the calculated
coefficient is also one of the inputs for battery heat generation modeling. result of the battery thermal model in the later section. EIS tests before
It can be considered that there is a linear relationship between the and after embedding the thermocouple were carried out on 50 % SOC
battery open-circuit voltage and battery temperature, and the entropy battery with Zahner Zennium E4 electrochemical workstation to eval­
coefficient is the slope of open-circuit voltage to temperature [25]. The uate the impact on the battery electrochemical system. Tested imped­
test steps of the entropy coefficient are as follows [26]: ance curves were also quite close, as shown in Fig. 3 (b). Therefore, it
could be considered that embedding the thermocouple had little effect
(1) Because of the tiny change in the battery open-circuit voltage, it on the battery.
needed quite a long time to achieve sufficient electrochemical
and thermal equilibrium. The fully charged battery of 100 % SOC 3. Thermal model of cylindrical battery
was shelved for 4 h in 25 ◦ C environmental chamber. Then the
open-circuit voltage was measured and recorded with ZLG The thermal model of the battery was built with COMSOL Multi­
DMM6000 high-precision 6.5-digit digital multimeter. physics 5.5 software. The battery was regarded as a cylinder with a
(2) The temperature of the environmental chamber was changed to diameter of 21 mm and a height of 70 mm. The battery shell was
45 ◦ C. The open-circuit voltage of the battery was tested and ignored, and the cylindrical battery with anisotropic thermal conduc­
recorded after 4 h shelving. The difference between the obtained tivity was further simplified based on two-dimensional axisymmetry. In
value and the open-circuit voltage recorded in step (1) was COMSOL Multiphysics software, 2D Axisymmetric was selected as the
divided by the temperature difference of 20 ◦ C to obtain the en­ space dimension, and heat transfer in solids was selected as physics with
tropy coefficient of the battery at 100 % SOC. time dependent study. Therefore, the computational domain is a rect­
(3) The temperature of the environmental chamber was changed angular axisymmetric section. The grid of the axisymmetric simplified
back to 25 ◦ C. After 1 h shelving, the battery was discharged to battery model and grid independence verification is shown in Fig. 4. The
90 % SOC with 1C rate. So far, the complete test process of single dashed red line marked with r = 0 in Fig. 4 (a) represents the axis of
SOC was completed, and the process was repeated until 0 % SOC symmetry, and the temperature monitoring positions to obtain the
was tested. There were 11 groups of data in total. battery surface temperature and core temperature data in simulations
are indicated by red dots. In Fig. 4 (b), the ordinate indicates the
simulated core temperature at the end of the 1C discharging process, and
2.3. Core and surface temperature measurements of the battery the four points represent different results of Coarser, Normal, Finer and
Extra fine grid size in the simulation software. Therefore, Finer grid size
It is necessary to measure the battery temperature during operation is chosen and is sufficient for simulation calculations. The calculating
to verify the thermal model. Both the core and surface temperature of time is the total discharge duration of the battery, which varies ac­
the 21,700 cylindrical battery can be measured by transforming the cording to the discharge rate, and the time step is set to 0.1 s. The heat
battery. The battery was drilled in an argon atmosphere in the MIK­ generation model proposed by Bernardi et al. [21] is adopted to
ROUNA Super Purified glove box (Water content: <1 PPM; Oxygen

3
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 2. Temperature measurements of the transformed battery: (a) Embedding the thermocouple in the glove box [27]; (b) Measuring core and surface temperature
of the battery in the environmental chamber.

Fig. 3. Experimental tests of the battery before and after embedding the thermocouple [27]: (a) Discharging curves of different rates; (b) Impedance curves by
EIS tests.

calculate the heat generation of the battery, and it can be assumed that represents that the irreversible heat, i.e. joule heat can be calculated by
the battery is self-heated uniformly. Considering that the discharging the working current I and the voltage drop. Uocv is the open-circuit
experiment corresponding to the thermal model is carried out in the voltage, and U is the working voltage of the battery. Furthermore, the
environmental chamber, all exposed outer surfaces of the battery model irreversible heat can also be calculated by I and battery internal resis­
are set as convection air cooling boundary conditions. Temperature tance R. In Eq. (3), dUdTocv indicates entropy coefficient of the battery. The
probes are set at the center point of the outer side and the center point of reversible heat, i.e. reaction heat can be calculated based on the working
the axis respectively, corresponding to temperature measurement points current I, battery temperature T, and entropy coefficient of the battery.
with thermocouples in discharging experiments, which is convenient for Since the battery is regarded as a uniform heat source in the simulation
subsequent battery thermal model verification. process, the parameter T In Eq. (3) adopts the average temperature of
According to the heat generation theory proposed by Bernardi et al. the battery computational domain when calculating the reversible heat
[21], battery heat generation is divided into two main parts: irreversible part. The heat generation expression is written into the simulation
heat and reversible heat. As shown in Eq. (1), Q denotes the total heat software through the user-defined function.
generation power of the battery. Qirr and Qre denote the heat generation
power of irreversible heat and reversible heat, respectively. Eq. (2) Q = Qirr + Qre (1)

4
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 4. (a) Grid of axisymmetric simplified battery model; (b) Grid independence verification.

Qirr = I(Uocv − U) = I 2 R (2) q = h(Tamb − T) (5)

Qre = − IT
dUocv
(3) 4. Results and discussion
dT
The heat transfer process inside the battery is described by the 4.1. Results of thermophysical parameters measurement
transient heat conduction equation, as shown in Eq. (4). It can be used to
calculate the internal temperature of the battery in COMSOL software. Thermophysical parameters of the battery were tested under the two
Density ρ, specific heat capacity Cp and thermal conductivity k are battery heating methods, and the temperature measurement results are
physical parameters of the battery. Q̇ denotes heat generation power per shown in Fig. 5. Fig. 5 (a) shows the result of heating from the side
unit volume inside the battery. The simulation solution of T can get the surface of the battery and Fig. 5 (b) shows the result of heating from the
internal temperature distribution of the battery. The convection heat bottom surface of the battery. According to the method proposed by
transfer boundary condition of the battery is described in Eq. (5). q Drake et al. [10], the temperature curve will tend to be a linear function
denotes heat emitted from the battery surface to the environment. h is over time. The slope and intercept of the linear function can be used to
the convection heat transfer coefficient. Tamb represents ambient tem­ calculate the specific heat capacity and thermal conductivity of the
perature. Within the relatively low-temperature range in this work, the battery. Calculation methods are shown in Eqs. (6)–(8). In Eq. (6), P
thermal radiation is very small compared with the convective heat indicates the heating power; ρ is the density of the battery; R and H
transfer and can be ignored [10], so only the convective heat transfer of denote battery radius and height, respectively; Cp is the specific heat
the battery is considered. capacity of the battery; k denotes the slope of the linear function fitted.
Eq. (7) corresponds to the temperature curve fitting result of the side
ρC p
∂T
+ ∇ • ( − k∇T) = Q̇ (4) surface heating method, and b denotes the intercept of the linear func­
∂t tion. kr indicates the radial thermal conductivity of the battery. Eq. (8)

Fig. 5. Temperature results under two heating methods: (a) Heating from side surface; (b) Heating from bottom surface

5
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Table 3 Table 4
Calculated results of battery thermophysical parameters. Comparison of the thermal conductivity of different cylindrical batteries.
Heating Slope of Intercept of Specific heat Thermal Source Battery Radial thermal Axial thermal
1 1
methods fitting fitting capacity (J conductivity (W type conductivity (W m− conductivity (W m−
function function kg− 1 K− 1) m− 1 K− 1) K− 1) K− 1)

Heating from Our results 21,700 0.9 24.2

side surface 0.0166 0.635 873 0.90 Bhundiya
18,650 0.43 –
(Test 1) et al. [6]
Heating from Bhundiya
26,650 0.2 –
side surface 0.0164 0.640 884 0.89 et al. [6]
(Test 2) Drake et al.
18,650 0.2 30.4
Heating from [10]
bottom Drake et al.
0.0163 − 1.402 889 24.03 26,650 0.15 32
surface [10]
(Test 1) Ahmed et al.
26,650 0.39 –
Heating from [28]
bottom Todys et al.
0.0162 − 1.380 894 24.41 21,700 0.83 11.6
surface [29]
(Test 2) Marzook et al.
21,700 – 28.1
[30]

corresponds to the temperature curve fitting result of the bottom surface the axial thermal conductivity has been relatively large, the difference in
heating method, and kz indicates the axial thermal conductivity of the axial thermal conductivity has a small impact on the temperature dis­
battery. The latter part of the temperature curve in Fig. 5 (a) and (b) is tribution inside the battery. And the comparison of the thermal con­
linearly fitted, and the fitting result is shown with the red line. ductivity of different cylindrical batteries is shown in Table 4. Some
thermal conductivity values of 21,700 cylindrical batteries measured in
P
=k (6) other papers [29,30] are also listed. Due to the different companies and
ρπR2 HCp
models of the 21,700 battery, it is normal to have some differences in
P thermal conductivity values. And compared with these values, it can
=b (7) also indicate that the thermal conductivity measured in our study is
8πHkr
reasonable.
PH
− =b (8)
6 π R2 k z
4.2. Results of internal resistance and entropy coefficient tests
A repeated test is carried out to ensure reliable data, and the results
are very close to temperature curves of Test 1 shown above. Fitting re­ Groups of battery internal resistance data at different temperatures
sults of temperature curves from Test 1 and Test 2 are both shown in are obtained through HPPC tests, and each group includes discharging
Table 3. The specific heat capacity and thermal conductivity of the resistance and charging resistance. Then average resistance of different
battery are further calculated according to Eqs. (6)–(8), and the results SOC is calculated, as shown in Fig. 6 (a). The curve fitting function in
are also shown in Table 3. The specific heat capacity calculated by the MATLAB software is further used for polynomial fitting of battery
two battery heating methods is close, but there is a huge difference resistance data. Fig. 6 (b) shows the surface fitting result of internal
between the radial thermal conductivity calculated by the side surface resistance to SOC and temperature, with an R-square of 0.9832 and
heating method and the axial thermal conductivity calculated by the RMSE of 0.00097. The corresponding expression of resistance is also
bottom surface heating method, with a gap of more than an order of obtained, as shown in Eq. (8). This expression is used for input to the
magnitude. The main reason mentioned above is that inside the cylin­ battery heat generation model according to Eqs. (1) and (2).
drical battery, the thermal contact resistance between layers of the

R = 109.2 − 338.5 • SOC − 2.765 • T + 687 • SOC2 + 6.061 • SOC • T + 0.03797 • T 2 −

564 • SOC3 − 7.916 • SOC2 • T − 0.04098 • SOC • T 2 − 0.0001961 • T 3 + (8)
158.7 • SOC4 + 3.234 • SOC3 • T + 0.02625 • SOC2 • T 2 + 0.00007106 • SOC • T 3

multi-layer winding structure is considerable. The results of obtained

battery thermophysical parameters are averaged, and the specific heat Results of entropy coefficient tests under different SOC of the battery
capacity is about 885 J kg− 1 K− 1. The radial thermal conductivity is are obtained. Since it is difficult to fit the expression of entropy coeffi­
about 0.9 W m− 1 K− 1, and the axial thermal conductivity is about 24.2 cient changing with SOC in a simple form, the entropy coefficient data
W m− 1 K− 1. These three values are used as input parameters of the are interpolated as shown in Fig. 7. The obtained piecewise function is
simulation thermal model. Besides, it is also worth noting that the input into the battery heat generation model according to Eqs. (1) and
measured radial thermal conductivity of the 21,700 battery is already (3).
larger than that of the 18,650 and 26,650 batteries measured in some
literature [6,10,28], which may be caused by the more compact layered
4.3. Battery temperature measurements and simulation results
winding structure inside the 21,700 battery with high energy density.
The measured axial thermal conductivity of the 21,700 battery is close
1C discharging test of the battery was carried out in the environ­
to other types of batteries in the literature. Although there seem to be
mental chamber of 25 ◦ C. Both core and surface temperatures of the
some differences in the numerical value. Due to the fact that the value of
battery were monitored and recorded during the test. The experimental

6
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 6. DC internal resistance of the battery: (a) Variation of internal resistance with SOC at different temperatures; (b) Surface fitting of internal resistance to SOC
and temperature

25 ◦ C environmental chamber. At the end of discharge, the core tem­

perature of the battery is about 60 ◦ C, and the difference between the
surface temperature and the core temperature is about 3 ◦ C. Solid curves
in Fig. 9 (a) represent surface temperature and core temperature
calculated by the simulation of 2C discharging battery. Compared with
the 1C discharging simulation, heat source term changes based on Eq.
(1), and other settings remain the same. It can be seen that simulation
temperature results of 2C discharging battery are also in good agreement
with experimentally measured data as a whole. It verifies that the
established battery heat generation model is relatively accurate to some
extent. Fig. 9 (b) shows the internal temperature profile of the battery at
the end of 2C discharge. The shape of temperature distribution is similar
to the result of 1C discharge. However, due to larger heat generation and
heat dissipation during 2C discharge, the overall temperature is higher
and the maximum temperature difference inside the battery is larger,
which is 3.1 ◦ C.
When the battery is discharged at 2C rate, the maximum core tem­
perature of the battery is about 60 ◦ C. In practical use, if the battery is
operated at a high rate, it usually needs to be strengthened for cooling.
Therefore, the simulation of 2C discharging battery under different
convection heat transfer coefficients is additionally carried out to
Fig. 7. Battery entropy coefficient under different SOC.
observe the variation of battery temperature, and the initial battery
temperature and the ambient temperature are still 25 ◦ C. Calculated
measured temperatures are plotted as dotted curves in Fig. 8 (a). At the temperature results are shown in Fig. 10. Fig. 10 (a) shows the battery
end of the 1C discharging process, the measured core temperature of the core and surface temperatures when the convection heat transfer coef­
battery is about 40 ◦ C, and the measured surface temperature is about ficient is set to 50 W/(m2⋅K). Due to the enhanced heat dissipation, the
1.2 ◦ C lower than the core temperature. The established simulation maximum core temperature of the battery decreases to <45 ◦ C, but at
model of the battery was calculated correspondingly, and the battery the same time the maximum temperature difference between the surface
initial temperature and ambient temperature were set to 25 ◦ C. When temperature and the core temperature increases to 3.9 ◦ C. When the
the convection heat transfer coefficient was set to 14 W/(m2⋅K), the convection heat transfer coefficient is set to 100 W/(m2⋅K), simulated
battery core and surface temperatures calculated by the simulation are battery temperature results are shown in Fig. 10 (b). The maximum core
in good agreement with experiment results, which is shown in Fig. 8 (a). temperature drops further to <40 ◦ C. However, the maximum temper­
It also verifies that thermophysical parameters measured by the exper­ ature difference between the surface temperature and the core temper­
iment are relatively accurate to some extent. From simulation results, ature increases to 4.6 ◦ C. It can be seen that enhanced heat dissipation of
the internal temperature distribution of the battery during discharge can the battery can reduce the battery overall temperature on the one hand,
also be obtained. Fig. 8 (b) shows the battery internal temperature but aggravate the temperature non-uniformity inside the battery on the
profile at the end of 1C discharging process. Since the axial thermal other hand. This temperature unevenness inside the battery makes the
conductivity of the battery is much greater than the radial thermal measured surface temperature far from the battery maximum temper­
conductivity, the temperature gradient inside the battery is mainly in ature, which is annoying for the thermal management system. The in­
the radial direction, and the overall temperature profile presents a ternal temperature non-uniformity may have an impact on the aging
spindle shape. At the moment, the simulated maximum temperature performance of the battery, and the temperature distribution inside the
difference is 1.1 ◦ C, which is also very close to the experimental value. battery can cause local premature aging [31], although there are few
Dotted curves in Fig. 9 (a) indicate the measured surface temperature relevant studies at present.
and core temperature of the battery during 2C discharging process in

7
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 8. Temperature simulation results of 1C discharging battery: (a) Compared with the measured temperature results; (b) Temperature profile at the end
of discharge.

Fig. 9. Temperature simulation results of 2C discharging battery: (a) Compared with the measured temperature results; (b) Temperature profile at the end
of discharge.

5. Conclusion SOC of the battery. Besides, overall battery internal resistance decreases
with the rise of temperature. The calculated core temperature and sur­
This paper presents the thermal modeling process of a 21,700 cy­ face temperature from the established thermal model of the operating
lindrical battery. The thermophysical parameters of the battery were battery are in good agreement with the values measured in the experi­
measured by experiments. The specific heat capacity is about 885 J kg− 1 ment. The maximum core temperature of 1C discharging battery is about
K− 1. The radial thermal conductivity is about 0.9 W m− 1 K− 1, and the 40 ◦ C and the maximum temperature difference inside the battery is
axial thermal conductivity is about 24.2 W m− 1 K− 1. The measured >1 ◦ C. When the discharge rate is 2C, the maximum core temperature of
radial thermal conductivity of the 21,700 battery is large compared with the battery is about 60 ◦ C and the maximum temperature difference
some 18,650 and 26,650 batteries in literature. In order to establish the inside the battery reaches 3 ◦ C. In the further simulation, heat dissipa­
heat generation model, the internal resistance and the entropy coeffi­ tion of 2C discharging battery is enhanced by setting the convection heat
cient of the battery are measured, and both values change with different transfer coefficient to 50 W/(m2⋅K), and the maximum core temperature

8
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

Fig. 10. Temperature simulation results of 2C discharging battery under different convection heat transfer coefficients: (a) 50 W/(m2⋅K); (b) 100 W/(m2⋅K).

of the battery decreases to <45 ◦ C, but the maximum temperature dif­ [6] H. Bhundiya, M. Hunt, B. Drolen, Measurement of the effective radial thermal
conductivities of 18650 and 26650 lithium-ion battery cells, in: Thermal & Fluids
ference inside the battery increases to 3.9 ◦ C. When the convection heat
Analysis Workshop (TFAWS) 2018 Proceedings, 2018.
transfer coefficient increases to 100 W/(m2⋅K), the maximum core [7] G. Yu, X. Zhang, C. Wang, W. Zhang, C. Yang, Experimental study on specific heat
temperature drops below 40 ◦ C, but the maximum temperature differ­ capacity of lithium thionyl chloride batteries by a precise measurement method,
ence inside the battery increases to 4.6 ◦ C. The effect of this battery J. Electrochem. Soc. 160 (2013) A985–A989.
[8] S. Panchal, M. Mathew, R. Fraser, M. Fowler, Electrochemical thermal modeling
internal temperature non-uniformity on battery performance especially and experimental measurements of 18650 cylindrical lithium-ion battery during
the aging characteristic can be studied in our further research. discharge cycle for an EV, Appl. Therm. Eng. 135 (2018) 123–132.
[9] Y. Ye, L.H. Saw, Y. Shi, K. Somasundaram, A.A.O. Tay, Effect of thermal contact
resistances on fast charging of large format lithium ion batteries, Electrochim. Acta
CRediT authorship contribution statement 134 (2014) 327–337.
[10] S.J. Drake, D.A. Wetz, J.K. Ostanek, S.P. Miller, J.M. Heinzel, A. Jain, Measurement
Qichao Wu:Conceptualization, Methodology, Experiment and of anisotropic thermophysical properties of cylindrical Li-ion cells, J. Power
Sources 252 (2014) 298–304.
Simulation, Writing. [11] J.B. Quinn, T. Waldmann, K. Richter, M. Kasper, M. Wohlfahrt-Mehrens, Energy
Rui Huang:Funding acquisition, Project administration. density of cylindrical Li-ion cells: a comparison of commercial 18650 to the 21700
Xiaoli Yu: Supervision, Funding acquisition. cells, J. Electrochem. Soc. 165 (2018) A3284–A3291.
[12] H. Popp, N. Zhang, M. Jahn, M. Arrinda, S. Ritz, M. Faber, D.U. Sauer, P. Azais,
I. Cendoya, Ante-mortem analysis, electrical, thermal, and ageing testing of state-
Declaration of competing interest of-the-art cylindrical lithium-ion cells, e & i Elektrotechnik und
Informationstechnik 137 (2020) 169–176.
[13] H. Wang, Y. Wang, F. Hu, W. Shi, X. Hu, H. Li, S. Chen, H. Lin, C. Jiang, Heat
The authors declare that they have no known competing financial generation measurement and thermal management with phase change material
interests or personal relationships that could have appeared to influence based on heat flux for high specific energy power battery, Appl. Therm. Eng. 194
the work reported in this paper. (2021), 117053.
[14] X. Du, Q. Wu, Y. Wang, T. Pan, Y. Wei, H. Chen, W. Song, D. Fang, Visualizing two-
dimensional internal temperature distribution in cylindrical Li-ion cells, J. Power
Data availability Sources 446 (2020), 227343.
[15] V. Ramadesigan, P.W.C. Northrop, S. De, S. Santhanagopalan, R.D. Braatz, V.
R. Subramanian, G.C.U.S. National Renewable Energy Lab. NREL, Modeling and
Data will be made available on request.
simulation of lithium-ion batteries from a systems engineering perspective,
J. Electrochem. Soc. 159 (2012) R31–R45.
Acknowledgments [16] C. Ji, B. Wang, S. Wang, S. Pan, D. Wang, P. Qi, K. Zhang, Optimization on
uniformity of lithium-ion cylindrical battery module by different arrangement
strategy, Appl. Therm. Eng. 157 (2019), 113683.
This study was supported by the Natural Science Foundation of [17] D.H. Jeon, S.M. Baek, Thermal modeling of cylindrical lithium ion battery during
Zhejiang Province (Grant number: LQ20E060008) in China and the discharge cycle, Energy Convers. Manag. 52 (2011) 2973–2981.
Open Foundation of the State Key Laboratory of Clean Energy Utilization [18] D. Anthony, D. Wong, D. Wetz, A. Jain, Non-invasive measurement of internal
temperature of a cylindrical Li-ion cell during high-rate discharge, Int. J. Heat Mass
of China (No. ZJUCEU2022016). Transf. 111 (2017) 223–231.
[19] C. Forgez, D.Vinh Do, G. Friedrich, M. Morcrette, C. Delacourt, Thermal modeling
References of a cylindrical LiFePO4/graphite lithium-ion battery, J. Power Sources 195 (2010)
2961–2968.
[20] W. Kim, Specification of Product INR21700-50E, SAMSUNG SDI, 2019.
[1] M. Lu, X. Zhang, J. Ji, X. Xu, Y. Zhang, Research progress on power battery cooling [21] D. Bernardi, E. Pawlikowski, J. Newman, A general energy balance for battery
technology for electric vehicles, J.Energy Storage 27 (2020), 101155. systems, J. Electrochem. Soc. 132 (1985) 5–12.
[2] H. Fathabadi, High thermal performance lithium-ion battery pack including hybrid [22] INEEL, FreedomCAR Battery Test Manual for Power-assist Hybrid Electric Vehicles,
active–passive thermal management system for using in hybrid/electric vehicles, U.S. Department of Energy, 2003.
Energy 70 (2014) 529–538. [23] Z. Lin, D. Wu, C. Du, Z. Ren, An improved potentiometric method for the
[3] M. Fleckenstein, O. Bohlen, M.A. Roscher, B. Bäker, Current density and state of measurement of entropy coefficient of lithium-ion battery based on positive
charge inhomogeneities in Li-ion battery cells with LiFePO4 as cathode material adjustment, Energy Rep. 8 (2022) 54–63.
due to temperature gradients, J. Power Sources 196 (2011) 4769–4778. [24] G.K. Mertin, D. Wycisk, M. Oldenburger, G. Stoye, A. Fill, K.P. Birke, A.D. Wieck,
[4] J. Cao, Z. Ling, X. Fang, Z. Zhang, Delayed liquid cooling strategy with phase Dynamic measurement of the entropy coefficient for battery cells, J.Energy Storage
change material to achieve high temperature uniformity of Li-ion battery under 51 (2022), 104361.
high-rate discharge, J. Power Sources 450 (2020), 227673.
[5] G. Xia, L. Cao, G. Bi, A review on battery thermal management in electric vehicle
application, J. Power Sources 367 (2017) 90–105.

9
Q. Wu et al. Journal of Energy Storage 65 (2023) 107338

[25] K.A. Murashko, A.V. Mityakov, V.Y. Mityakov, S.Z. Sapozhnikov, J. Jokiniemi, [29] G.M. Todys, S. Mahmud, Thermal Characterization of a Cylindrical Li-ion Battery
J. Pyrhönen, Determination of the entropy change profile of a cylindrical lithium- Cell, 2020. Master's thesis in Electric Power Engineering.
ion battery by heat flux measurements, J. Power Sources 330 (2016) 61–69. [30] M.W. Marzook, A. Hales, Y. Patel, G. Offer, M. Marinescu, Thermal evaluation of
[26] S. Skoog, in: Electro-thermal Modeling of High-performance Lithium-ion Energy lithium-ion batteries: defining the cylindrical cell cooling coefficient, J.Energy
Storage Systems Including Reversible Entropy Heat, IEEE, 2017, pp. 2369–2373. Storage 54 (2022), 105217.
[27] X. Yu, Q. Wu, R. Huang, X. Chen, A novel heat generation acquisition method of [31] M. Klett, R. Eriksson, J. Groot, P. Svens, K.Ciosek Högström, R.W. Lindström,
cylindrical battery based on core and surface temperature measurements, J. H. Berg, T. Gustafson, G. Lindbergh, K. Edström, Non-uniform aging of cycled
Electrochem.Energy Convers.Storage 19 (2022). commercial LiFePO4//graphite cylindrical cells revealed by post-mortem analysis,
[28] M.B. Ahmed, S. Shaik, A. Jain, Measurement of radial thermal conductivity of a J. Power Sources 257 (2014) 126–137.
cylinder using a time-varying heat flux method, Int. J. Therm. Sci. 129 (2018)
301–308.

10