Journal of Energy Storage 84 (2024) 110889

Contents lists available at ScienceDirect

Journal of Energy Storage

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

Research Papers

Numerical and experimental investigations of thermal performance of

lithium-ion battery with hybrid cooling system under dry-out condition
P.M. Sutheesh , A.P. Atul, B. Rohinikumar *
CFD Laboratory, Department of Mechanical Engineering, National Institute of Technology, Calicut, Kerala, India, 673601

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

Keywords: Thermal management is essential for Lithium-ion batteries used in electrical vehicles. The present study focussed
BTMS on numerical simulation of thermal performance of six configurations of hybrid cooling system at dry out
Hybrid cooling condition (HCSDC) and compared with simple air cooling system (SACS) under different discharge and flow
Air cooling
conditions. SACS could not keep the required average temperature and temperature uniformity at 2C and 3C
Lithium-ion battery
conditions for Re values of 764–6113. BP-4 configuration of HCSDC has the best thermal performance in higher
CFD
Numerical method Re values at all discharge conditions. Maximum increase in temperature uniformity of 27.54 % over SACS and 26
% over threshold and maximum reduction in average temperature is 2.86 K lower than SACS for BP-4 at 2C
condition. Simulation predictions of SACS and BP-4 configurations are validated with experimental results ob­
tained and the maximum deviation in absolute temperature was observed as 0.48 % and 0.31 %, respectively.
Actual capacity of the cell is 54 % of the rated capacity by experiments. Experimental and simulation results
indicate that the penultimate cell experiences highest temperature than last one due to difference in the wake
regions. The non-uniformity of temperature within the battery is associated with the temperature difference
between the first and penultimate cells.

EVs and energy storage systems rely on complex arrangements of

1. Introduction cells in series and parallel to form battery packs, but their adoption is
hindered by limited battery pack temperature limits [5]. The operating
Automobiles powered by internal combustion engines that utilize temperature of battery packs plays a critical role in determining their
gasoline and diesel fuels are recognized as significant emitters of carbon lifespan [8]. Maintaining the temperature range between 20 and 40 ◦ C
dioxide, a greenhouse gas responsible for global warming [1]. The for lithium-ion batteries, is crucial for their longevity [9–11]. Without
transportation sector contributes almost 25 % of the world’s greenhouse an effective thermal management, excess thermal energy can accumu­
gas emissions [1,2]. Capping global warming below 1.5 K will demand a late, posing safety risks and potentially leading to explosions. Operating
significant reduction of emissions from the transportation sector [1]. A the battery pack beyond optimal temperature limits accelerates its
pivotal phase in reducing carbon emissions will be the electrification of degradation. Additionally, the temperature homogeneity of the battery
transportation sector. Recent advances with electric vehicles (EVs), pack affects its charging and discharging capabilities [12]. In colder
indicate a positive trend for the EV industry. With the major automakers conditions, the power of the battery pack is controlled by the cell with
joining the race, production and sales of EVs are expected to increase the lowest temperature, while in elevated temperatures, the cell with the
rapidly in the upcoming years [1,3]. In modern EVs, the rechargeable highest temperature assumes the control. This temperature disparity
battery is a critical component that requires outstanding performance. results in uneven charging and discharging rates among individual cells,
Lithium-ion batteries (LIBs) have emerged as the preferred choice for causing electrical imbalances that negatively impact the battery pack’s
EVs due to their superior energy and power density, long operational operational effectiveness.
lifespan [4–6], and environmental-friendly characteristics compared to Several investigations explored different cooling strategies for ther­
alternatives such as Ni–Cd, Ni-MH, and Lead-acid batteries [7]. mal regulation of batteries, including cooling by air, water, and phase

Abbreviations: BTMS, Battery Thermal Management System; eff , Effective; Calib, Calibration; BP, Battery Pack; EV, Electric Vehicle; TCS, Thermal Conductance
Structure; SACS, Simple Air Cooling System; HCSDC, Hybrid Cooling System under Dry out Conditions; SEI, Solid Electrolyte Interface; LCP, Liquid cooling plate;
PCMs, Phase Change Materials.
* Corresponding author.
E-mail address: rohinikumar@nitc.ac.in (B. Rohinikumar).

https://doi.org/10.1016/j.est.2024.110889
Received 30 August 2023; Received in revised form 31 January 2024; Accepted 8 February 2024
Available online 16 February 2024
2352-152X/© 2024 Elsevier Ltd. All rights reserved.
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Nomenclature Q̇total Volumetric heat generation (W m− 3)

Dh Hydraulic diameter (mm)
T Temperature (K) k Turbulence kinetic energy (J/kg)
C Charge/Discharge rate Vn Normal velocity (m/s)
ρ Density (kg m− 3) S@Tsen Sensitivity of the heat flux sensor
c Lithium concentration at electrolyte-electrode interface ηi Over potential between positive and negative electrodes
(mol m− 3) (V)
t Time (s) t+ Transport number
V Voltage (V) F1 Blending function in SST model
ṁ Mass flow rate (kg s− 1) Rcc Electrical resistance of connector contact (Ω)
u x-directional velocity (m/s) Acc Connector contact area on the terminal (m2)
D Diameter (mm) Ei Equilibrium potential (V)
Nu Mean Nusselt number e Effective
→
V Velocity vector (m/s) σk Turbulent Prandtl numbers for turbulence kinetic energy
k Thermal conductivity (W m− 1 K− 1) σω Turbulent Prandtl numbers for specific dissipation rate
R Characteristic gas constant (J/kg K− 1) y Nearest wall distance (mm)
F Faraday’s constant (9.64845604C mol− 1) CDkω Cross-diffusion term
v y-directional velocity (m/s) kf Coolant thermal conductivity (W m− 1 K− 1)
P Pressure (Pa) TS Average surface temperature (K)
f Coefficient of molar activity of electrolyte Tm,i Mean temperature at inlet (K)
w z-directional velocity (m/s) Uavg Average velocity at inlet (m/s)
U Average velocity (m/s) Tm,o Mean temperature at outlet (K)
Re Reynolds number Ainlet Area at inlet (m2)
H Width (m) ∀ Volume (m3)
L Length (m) ke Cell effective thermal conductivity (W m− 1 K− 1)
Z Height (m) cp Cell effective specific heat capacity (J/kg K− 1)
h Average convective heat transfer coefficient ji Specific reaction
Subscripts and Superscripts ai Surface area (m2)
cp Specific heat capacity (J/kg K− 1) Ca Cell capacity (A h)
σeff ,i Effective electric conductivity (S m− 1) td Cell discharge time (s)
keff ,i Rate constant of electrochemical reaction (m/s) Id Discharge current (A)
ϕs,i Electric potential (V) Greek symbols
∅ Properties (Pressure, Temperature etc.)
ΔϕSEI Voltage drop at SEI film (V)
ω Rate of specific dissipation (s− 1)
ϕe,i Electrolyte potential (V)
Cl Solution phase electrolyte concentration (mol m− 3) α, β* SST model constant
μ Dynamic viscosity (Pa s)
h Average heat transfer coefficient (W m− 2 K− 1)
Δ Small change
TS Average surface temperature (K)
V0 Open circuit voltage (V)

change material (PCM) [13]. Yang et al. [14] conducted a comparative inlets within the BTMS. Mangrulkar et al. [21] studied the staggered BPs
study on battery pack arrangements, revealing the impact of longitudi­ with and without a splitter plate, finding that the introduction of a
nal distance variations on average cell temperature in both aligned and splitter plate increased the Nusselt number while reducing pressure
staggered configurations. Notably, aligned arrangements demonstrated drop. Shahid et al. [22] developed a non-active technique using an inlet
superior thermal performance. Zhao et al. [15] investigated the influ­ plenum in the BP, optimized the geometrical dimensions and Reynolds
ence of a liquid-cooled cylinder around cylindrical Li-ion cells, number through numerical analysis and experimental validation.
observing enhanced heat transfer rates with an increased number of Sheng et al. [23] employed the numerical techniques to assess six
channels. Saw et al. [16] evaluated thermal performance of air-cooled distinct serpentine-channel liquid cooling plate (LCP) designs for pris­
battery packs through numerical simulations, establishing a correla­ matic lithium-ion cells. Parameters such as coolant flow rate, inlet/
tion between Reynolds number and Nusselt number for estimating outlet positioning, C-rate, and single versus double serpentine channels
thermal management effectiveness. were systematically studied. Key performance metrics, including
Jaguemont et al. [17] reviewed the performance loss and degrada­ maximum temperature, temperature difference and pumping power
tion of LIBs in cold climates and compared conventional cooling ap­ were analyzed. The results highlighted the significant impact of inlet/
proaches using gas, liquid, and PCM. Choudhari et al. [18] numerically outlet locations and flow directions on cell temperature distribution and
compared different battery pack (BP) configurations with and without LCP power consumption ratio. Increased fluid flow rates were found to
PCM and with fins inside the PCM, concluded that the design with PCM effectively reduce maximum temperature rise, while channel width
and fins significantly reducing the maximum temperature during high C- influenced power consumption. The study highlights the superior ther­
rates. Saw et al. [19] compared the thermal behaviour of mist-cooled mal management capabilities of serpentine-channel design, offering a
and air-cooled battery thermal management systems (BTMS), promising approach for efficient and cost-effective LIB thermal systems.
concluded that mist cooling can effectively maintain battery pack tem­ Shang et al. [24] have introduced an innovative cellular liquid cooling
perature below the optimum levels. Shahid et al. [20] investigated the jacket design for thermal management of Li-ion 21,700 cells. Through a
impact of additional inlets on thermal management in simple air-cooled comprehensive analysis encompassing both experimental and simula­
BPs, inferred that thermal uniformity was improved with additional tion methodologies, the findings elucidate that the adoption of

2
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

interlaced flow directions yields a notably more uniform thermal dis­ 2. Physical system
tribution. A meticulous comparison between numerical and experi­
mental outcomes indicated a negligible deviation of <1.5 ◦ C. Electric vehicles employ a battery pack (BP) comprised of inter­
Wei et al. [25] conducted a comprehensive experimental investiga­ connected cells arranged in series and parallel configuration, which
tion on a novel hybrid cooling method aimed at enhancing heat transfer deliver the required power. The system considered for the present work
through evaporative cooling. Hydrophilic fibers were strategically consists of a sub module of the BP with eight rows and twenty-three cells
placed around each cell in the study, which yielded to superior thermal in each row as shown in Fig. 1. To ensure optimal performance, it is
efficiency compared to configurations employing liquid cooling or no crucial to keep the temperature of the BP within a specific range of 20 ◦ C
cooling. Furthermore, in a separate set of experiments [26], Wei et al. to 40 ◦ C, with a maximum temperature difference of 5 ◦ C within the BP
explored a system that utilized a hybrid cooling approach combining [9–11]. Hence, thermal management of the BP is necessary. Hybrid
evaporative cooling and simple air cooling. The incorporation of hy­ cooled BTMS and simple air cooled BTMS are the two different thermal
drophilic fibers between the cells in the system resulted in a substantial management strategies considered in the present work.
improvement in thermal efficiency compared to configurations without The main goal of the hybrid cooling technique is to enhance the rate
cooling or relying solely on air cooling. of heat dissipation from cells, especially during rapid charging/dis­
Lu et al. [27] numerically analyzed the impact of channel dimensions charging. A hybrid cooled BTMS consisting of forced-air cooling and
and air flow pattern on the thermal performance of staggered cell ar­ passive evaporative cooling, as shown in Fig. 2. The cells are arranged in
rangements within the BP. The results showed improved thermal per­ a rectangular channel and surrounded by fibers. These fibers are
formance when the inlet was located on top and the outlet on the bottom partially submerged in a water bath and possess hydrophilic properties.
of the BTMS. Jiang et al. [28] analyzed different cooling strategies by Capillary action allows water to move up through fibers. As air flows
adjusting the inlet and outlet positions and introducing baffles within through the system, evaporation of water in the fibers takes place,
the BP. A design equipped with baffles and inlet and outlet on different absorbing latent heat from the surrounding water and air during the
sides demonstrated the improved thermal performance. Ji et al. [29] phase transition. Consequently, the passing air becomes humidified and
developed a numerical model for optimizing the placement of cells cooled. This cooled air, at a lower temperature, passes over the higher
within the BP cooling channel based on arithmetic and geometric ratios. temperature cells, resulting in a higher rate of heat dissipation from the
The BP configuration with cells arranged based on arithmetic ratio cells and subsequently reducing their temperature.
achieved the better temperature uniformity. The hybrid system functions as a simple air cooling system (SACS) in
Lai et al. [30] conducted a computational analysis to optimize the the absence of fibers and water bath, which is one of the thermal man­
dimensions of a thermal conductance structure while maintaining agement solutions studied in the present work. Hybrid cooling system
thermal performance. The study focused on parameters such as height, under dry out conditions (HCSDC) is the second thermal management
diameter, and surface angle to make the structure more compact. The strategy considered in the current work. HSCDC is the configuration of
optimized structure successfully maintained the highest temperature hybrid cooling system in the absence of water and with the presence of
and temperature difference within the optimal range. Xu et al. [31] fibers.
performed a computational analysis to optimize the design of serpentine
and U-shaped cooling channels. The results showed that the serpentine 3. Problem definition
channel exhibited better thermal performance compared to the U-sha­
ped channel. Elmekawy et al. [32] conducted a computational analysis 3.1. Computational domain
on the effect of a splitter plate in a BP with staggered cells. The presence
of a splitter plate reduced the pressure drop and enhanced the Nusselt The thermal condition and fluid-cell interactions in the battery
number. Shahid et al. [33] conducted research on a hybrid cooling module, with one inlet and one exit, are considered when defining the
system using a numerical model, including vortex generators, multiple system. A single layer of eight cells, shown in Fig. 3, is chosen for nu­
inlets, and a liquid cooling jacket surrounding the cells. The findings merical and experimental studies of the battery module [34]. The
demonstrated a significant improvement in temperature uniformity computational domain for SACS will resemble Fig. 3 and for HCSDC, the
within the BP. domain is slightly modified with the inclusion of fibers.
In the field of thermal management for lithium-ion batteries, earlier
investigations were focused on evaluating the performance of hybrid 3.2. Modelling of the system
cooling systems with the combined effect of air and evaporative cooling
under the normal operating conditions. Regrettably, the performance of Mathematical model is developed to simulate both HCSDC and SACS.
the hybrid cooling systems during dry out conditions, which result in The model development is based on the following assumptions and
suboptimal behaviour of the BTMS, has not been explored in the liter­ considerations:
ature. Additionally, the influence of number of fibers, position of fibers
and inter-fiber distance on the fluid flow and subsequently thermal • System is in steady state
performance of lithium ion battery is yet to be investigated. In the light • Fluid flow is three-dimensional, incompressible, and turbulent
of these research gaps, the present study is focussed to identify a supe­ • Effect of gravity is neglected
rior configuration of hybrid-cooled battery pack under dry out condi­ • Homogeneous heat generation with in the cell
tions, surpassing the capabilities of simple air cooling. This study • All thermo-physical properties of the working fluid, except density,
presents a novel approach, wherein the authors have developed a nu­ are assumed to remain constant
merical model of the BTMS incorporating fibers in hybrid cooling system • Effect of radiation is neglected
under dry out conditions. Subsequently, the performance of various
configurations of the hybrid-cooled BTMS at dry out conditions has been
simulated and compared against simple air cooling system. The most 3.2.1. Heat generation and conduction in the cells
promising hybrid cooling arrangement determined through simulation Heat generation in the cell stems from reversible processes,
and simple air cooling system have been fabricated and subjected to connector resistance, polarization, and internal resistance [14,35]. In­
experimental analysis. Simulation predictions of both the systems are ternal resistance encompasses electrical resistance, ion conductive
validated with the corresponding experimental results. resistance, and solid electrolyte interface (SEI) film resistance on the
negative electrode. Hence, the general thermal energy equation for the
cell is expressed as follows

3
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 1. Schematic of the battery submodule.

( )2 ( )
∂ϕs,i ∂ϕe,i 2
Q̇ohm =σ eff ,i + keff ,i
∂x ∂x
( ) (5)
2keff ,i RT(1 − t+ ) dlnf ∂lnCl ∂ϕe,i
+ 1+ + ai ji FΔϕSEI
F dlnCl ∂x ∂x
Where ji is the specific reaction, ai is the specific surface area, F is
Faraday’s constant. The product of these three parameters yields the
current Ii generated by the given area [26].
Ii = ai ji F (6)
The Bernardi Eq. [36] is a widely used simplified method for
calculating the rate of heat generation, and it is commonly employed for
this purpose, given by
dVo
Q̇total = Ii (V0 − V) − Ii T (7)
dT
The right-hand side of Eq. (7) consists of two terms. The first term
integrates Eq. (3) and represents irreversible heat generation due to over
potential across the entire circuit, encompassing polarization and in­
Fig. 2. Schematic diagram of hybrid cooling system. ternal resistance. The second term integrates Eq. (5) and represents
reversible entropic heat generation resulting from the overall chemical
∂T reaction within the cell [36].
ρc p = ∇.ke ∇T + Q̇p + Q̇rea + Q̇c + Q̇ohm (1)
∂t
3.2.2. Fluid flow and heat transfer in the system
Where ke is the overall effective thermal conductivity of the cell in The governing equations for fluid flow and heat transfer in the
the BP. Q̇p , Q̇rea , Q̇c and Q̇ohm represent the rate of heat generation due to domain are obtained using mass, momentum, and energy balances as.
polarization, reversible chemical reactions, resistance of connectors, and Continuity:
internal cell resistance, respectively, and they are calculated using [26]. ( →)
∇ • ρV = 0 (8)
Q̇p = ai ji Fηi (2)
Momentum conservation:
dEi
Q̇rea = ai ji FT (3) → (→) →
dT V .∇ V = − ∇P + μ∇2 V (9)
( )
Menter [37] introduced the SST k-ω turbulence model, which com­
Ii2 RAcccc
bines the k-ϵ and k-ω turbulence models. The k-ϵ model is used for the
Q̇c = (4)
v free stream region, while the k-ω model is effective in predicting flow
near the walls.
[ ]
∂(ρUi k) ̃ ∂k ∂k
= Pk − β * ρk ω + (μ + σk μt ) (10)
∂xi ∂xi ∂xi

4
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 3. System for the present study.

∂(ρUi ω) ∂
[
∂ω
]
1 ∂k ∂w respectively. Table 1 illustrates the boundary conditions associated with
= αρS2 − β* ρω2 + (μ + σk μt ) + 2(1 − F1 )ρσω2 the system.
∂xi ∂xi ∂xi ω ∂xi ∂xi
(11)
{{ [ ( √̅̅̅ ) ] }4 } 3.4. Computational domain, grid generation and mesh independence
k 500v 4ρσω2k
F1 = tanh min max * , 2 , study
β ωy y ω CDkω y2

(12) 3.4.1. Computational domain

( ) In HSCDC, fibers act as solid obstructions to airflow and do not
1 ∂k ∂ω − contribute significantly to heat transfer between the air and fibers. This
CDkω = max 2ρσω2 , 10 20
(13)
ω ∂xi ∂xi study considers different configurations of HCSDC alongside SAC. A
three-dimensional computational domain is created. The distance be­
Energy equation
tween adjacent cell centers in the flow direction and transverse direction
( )
→
V .∇(ρcP T) = ∇ kf ∇T (14) is 32 mm and 34 mm (2H), respectively. The stream wise length of
computational domain is 1056 mm. The fiber structure is having a
Ideal gas relation is used for finding the density, diameter of 3 mm and height of 65 mm. The total height of the
P = ρRT (15) computational domain is 65 mm, which is equal to the height of the Li-
ion cell. The detailed dimensions of the computational domain are
shown in Fig. 3.
3.3. Boundary conditions
3.4.2. Grid generation
Interaction between the cells in the system and fluid is existing in the
The computational domain is discretized using unstructured hex­
domain between upstream and downstream regions within the x-co­
ahedral and triangular prism elements. The solid-liquid interface is
ordinates x = 0 and x = L. Where, L is the total length of the channel.
refined with a growth rate of 1.2 to capture boundary layer effects. A
The point at the centre inlet of channel serves as the reference point for
biasing is applied in the direction perpendicular to the main stream flow
the coordinate system. The height and width of the domain in which the
to ensure fine mesh near the walls compared to free stream region. Fig. 4
fluid and cells are interacting are 2Z and 2H, in z and y directions,
shows the mesh for SACS, and different HCSDC configurations are also
meshed accordingly.
The SIMPLE scheme is used to interconnect the pressure and velocity
Table 1 fields. A second-order partial discretization method is employed to
Boundary conditions.
discretize the energy, momentum, pressure, and turbulence equations.
# Parameter Flow condition Thermal Under-relaxation factors of 0.3, 0.7, and 0.8 are applied to pressure,
condition momentum, and turbulent kinetic energy, respectively, while a factor of
u (0.y.z) = uaverage T (0.y.z) = 1 is used for density, turbulent viscosity, and energy. Convergence is
1 Inlet
v (0.y.z) = 0, w (0.y.z) = 0 Tambient achieved by maintaining residuals below 10− 3 for continuity and mo­
2 Outlet P (L.y.z) = Pambient ~
mentum, and 10− 8 for energy.
∂T(x, y, +Z)
Top (z = +Z) and =0
3 u = 0, v = 0, w = 0 ∂z
bottom wall (z = − Z) ∂T(x, y, − Z) 3.4.3. Mesh independence study
=0
∂z A mesh sensitivity study was conducted to ensure numerical precision
∂w(x, +H, z) ∂u(x, +H, z) ∂T(x, +H, z)
∂y
= 0,
∂y
= = and optimize the computational time. Meshes with varying numbers of
∂y
Lateral surfaces ∂w(x, − H, z) 0 elements, ranging from 2 × 104 to 150 × 104 , were considered. The dif­
0, = 0,
4 y = +H and y = − H ∂y ∂T(x, − H, z) ference in cell count between adjacent meshes was maintained above 20 %
(Symmetry face) ∂u(x, − H, z) ∂y
=
[22]. For SACS at 1 m/s air velocity with a 5C charge/discharge condition,
= 0,
∂y 0
v =0
a grid independence study was performed using the outlet temperature as
the parameter. The results of each mesh were compared with those of a

5
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 4. Discretized computational domain.

Fig. 5. Mesh independence study.

Fig. 6. Comparison of variation of outlet temperature with Reynolds number.

very fine mesh consisting of 4.4 × 106 elements. Fig. 5 illustrates the
variation of the battery module’s outlet temperature with the number of Table 2
elements. A mesh with 139.56 × 104 cells was selected as the grid- Comparison of steady state mean temperature of the cells.
independent mesh for SACS simulation. A similar approach was adopted Sl. No. Average battery module temperature (K) Percentage deviation
for simulating the different configurations of HSCDC. Current model Tete et al. study [37]
(%)

1 299.83 299.14 0.23

3.5. Validation of the model

To establish the reliability of model, a comparison was made be­ was compared with the model by P. R. Tete et al. [38], as shown in
tween numerical results and an analytical solution. The outlet temper­ Table 2. Temperature of the cells at steady state from both studies was
ature of the battery module was determined for various input velocities compared, revealing a difference of 0.23 %. Hence, the current model
with thermodynamic energy balance using can effectively simulate the performance of BTMS considering fluid flow
and heat transfer phenomena.
Q̇total
Tm,o = + Tm,i (16)
ṁcp
4. System simulation
Where, Q̇total is the total heat released from cells. Tm,o and Tm,i are the
temperature of coolant at outlet and inlet, respectively. Fig. 6 shows the 4.1. Different configurations for simulation
comparison of the outlet temperature obtained from numerical simula­
tion and analytical results. The maximum difference between the two is Six arrangements of fibers were considered for different configura­
0.001 %. Furthermore, the numerical model developed for this system tions of HSCDC, along with one SAC configuration in this study. The

6
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Table 3 cells in equidistant manner in BP-1, BP-2 and BP-3a configurations,

Specifications of different configurations. respectively. Five fibers are arranged between two adjacent cells with
Sl. Battery module Number of Geometry Specification varying distance between the fibers in BP-3b and BP-3c configurations.
No. configuration fibers The gap between the fibers are in geometric progression (GP) and
1 BP-0 0 Simple air-cooling arithmetic progression (AP), respectively in BP-3b and BP-3c. The
2 BP-1 3 3 equidistant fibers common ratio and first term in the GP in BP-3b is 0.5 mm and 8 mm,
3 BP-2 4 4 equidistant fibers respectively, whereas the common difference and first term in the AP in
4 BP-3a 5 5 equidistant fibers BP-3c is − 2 mm and 8 mm, respectively. In BP-3b and BP-3c, the gap
5 fibers with varying distance (GP
5 BP-3b 5 configuration with common ratio 0.5
between fibers increases towards the center to enhance density in the
mm and first term as 8 mm) high-velocity region for improved heat transfer rate. BP-4 was an
5 fibers with varying distance (AP extension of BP-2, with an additional fiber placed between each cell and
6 BP-3c 5
configuration with common symmetry plane on both sides, resulting in two layers of fibers for each
difference − 2 mm and first term as 8
cell (except the last cell), with the first layer having two fibers and the
mm)
4 fibers with equal distance between second layer having four fibers. BP-0 represents SACS without any
7 BP-4 4+2 them and two fibers at the minimum fibers.
area

4.2. Flow chart of simulations and inputs to the simulation

fibers, with a diameter of 3 mm and length of 65 mm, were placed
downstream of each cell in the flow direction, except for the last cell. Table 4 presents the thermophysical properties of the working fluid.
Table 3 provides the specifications of the various configurations used, The simulations include SAC and HCSDC configurations at 1C, 2C, and
while schematic representations of the same are depicted in Fig. 7. Three 3C charge/discharge conditions, with inlet velocities ranging from 0.25
fibers, four fibers and five fibers are placed between each two adjacent to 2 m/s. The battery module outlet is assumed to be open to ambient.
Symmetric faces have a zero magnitude of normal gradient for all

Fig. 7. Top view of different configurations: (a) BP-0, (b) BP-1, (c) BP-2, (d) BP-3a, (e) BP-3b, (f) BP-3c, and (g) BP-4.

7
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Table 4 Table 5
Properties of the working fluid. Specifications of the cell. [39].
Property Unit Value Property Specifications/Values

Specific heat J/kgK 1006.43 Chemical notation NMC

5
Dynamic viscosity Pa s 1.7894 × 10− Chemistry LiNiMnCoO2
Thermal conductivity W/mK 0.242 Weight (g) 45.9
Density (kg/m3) 2775
Specific heat (J/kg K) 880
solution variables (pressure, velocity, and temperature). Fig. 8 illus­ Thermal conductivity (W/m K) 3
trates the simulation flow chart. Samsung 18,650-30Q cells are used in Diameter of cell (mm) 18
both SACS and HCSDC battery modules, with their parameters listed in Height of cell (mm) 65
Rated nominal capacity (mAh) 3000
Table 5. Different boundary parameters in the computational domain
Maximum cell voltage (V) 4.2
are listed in Table 6. The entropy coefficient for the simulated lithium- Cut-off voltage (V) 2.5
ion cell is 0.01116 [38]. Volumetric heat generations inside the cell at
various C-rates are estimated using Bernardi formula and presented in
Table 7 [36,38]. Table 6
Inputs to simulation.
4.3. Performance parameters Location/Region Parameter Value

Velocity 0.25 to 2 m/s

Inlet
The steady-state heat transfer rate from the cells to the air in the Temperature 300.15 K
battery module can be calculated using [16]. Outlet Pressure 101,325 Pa
Ambient Temperature 300.15 K
Q̇total = Q̇convection + Q̇radiation + Q̇losses (17) Symmetry surface All solution variables (∅)
∂∅
=0
∂n
Taking into account the physical conditions of system and the un­
derlying assumptions of the model, the Q̇total is given by
Table 7
Q̇total = Q̇convection (18) Rate of heat generation at different discharge rates.

From thermodynamic energy balance, rate of heat transfer is eval­ Parameters Values
uated by, C-rate 0.5 1 2 3 4 5
( ) Discharge
1.5 3 6 9 12 15
Q̇convection = ṁcp Tm,o − Tm,i (19) current (A)
Volumetric heat
Using heat transfer principles, the convection heat transfer rate is generation 3733 12,907 47,579 104,017 182,220 282,189
calculated with [16] rate (W/m3)
[ ( )]
Tm,o + Tm,i
Q̇convection = hAext TS − (20)
2
Where TS is the average temperature of the cell’s surface. Reynolds
The average convective heat transfer coefficient is given by number is estimated using
Q̇ ρUavg Dh
h= [ convection
( )] (21) Re = (22)
Aext TS −
Tm,o +Tm,i μ
2
The mean Nusselt number is computed with

Fig. 8. Flow chart of simulations.

8
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

hD 5.2. Battery module of the system

Nu = (23)
k
The SACS battery module comprises eight Samsung INR 18650-30Q
Temperature uniformity in a battery module can be expressed in
cells connected in 8S1P configuration and housed within a rectangular
terms of the difference between the module’s highest and lowest tem­
channel made of transparent acrylic sheets (4 mm thick) with di­
peratures as
mensions of 65 mm × 34 mm. The cells are placed in alternative manner
ΔT = Tmax − Tmin (24) in which the positive terminal of the first cell near the inlet is on top
while the last cell, positive terminal is at the bottom side. The experi­
The actual deliverable capacity of the cell is calculated from exper­
ment is carried out by discharging the cell from the maximum cut off
imental results using
voltage of 4.2 V to the lower cut off voltage of 2 V at specified constant
td current condition. A fan with a speed controller is connected to the inlet
Ca = × Id × 1000 (25)
3600 through a converging rectangular duct. Two K-type thermocouples are
Where, td is the battery discharge time, and Id is the constant current positioned on alternate cells in the system, 28 mm from the positive
maintained during the discharge experiment. terminal [40]. One thermocouple is placed on the rear side of the cell,
The volumetric average temperature of each cell is given by where the wake region is present, anticipating higher temperatures. The
∫ second thermocouple is placed on the cell’s surface near the side wall,
1∑ where the distance between the cell and the wall is minimal, as shown in
n
Td∀ = Ti ∀i (26)
∀ i=1 Fig. 9, with lowest anticipated temperature. Additional thermocouples
measure ambient and outlet temperatures. Differential temperature
5. Experimental studies thermopile based heat flux sensor (PHFS-01) made by Fluxteq is wrap­
ped around the cell to measure the heat flux and estimate the actual heat
5.1. Determination of heat generation rate of a cell generation rate. The specifications of heat flux sensor used in the
experiment are given in Table 8. Heat flux sensor consist of thin layer of
Ampere-hour capacity is specified for a cell at a given current or thermally resistive material and a series of electrically connected ther­
time, as increased currents result in more heat dissipation across the mocouples. The thermocouples generates voltage proportional to the
cell’s internal resistance, altering chemical reactions and leading to temperature difference across the thin layer, which is measured to es­
shorter discharge times. The volumetric heat generation rate predicted timate the heat flux by using Eq. (27).
by the Bernardi Eq. [36] is suitable for ideal cells, but commercially
available cells exhibit different heat generation rates. The actual rate of
Table 8
volumetric heat generation and deliverable capacity of commercially Heat flux sensor specifications [41].
available cells were determined through experiments conducted on a
Property Specifications/Values
single cell. During heat generation experiment, the cell is subjected to
cooling by forced circulation of air in order to avoid significant rise in Sensor type Differential temperature thermopile
average temperature of the cell. Heat generated within the cell is Encapsulation material Kapton (polyimide)
Nominal sensitivity 8.0 mV/(W/cm2)
dissipated to the ambient medium and surrounding surfaces through Sensor Thickness (t) 380 μm
convection and radiation, respectively. Notably, convective heat trans­ Specific thermal resistivity 0.9 K/(kW/m2)
fer dominates due to a higher convective heat transfer coefficient. To Absolute thermal resistance 1.0 K/W
quantify heat transfer, a precision heat flux sensor mounted on the cell Heat flux range ± 150 kW/m2
Temperature range − 50 ◦ C to 120 ◦ C
surface measures heat flux, allowing for accurate estimation of heat
Response time 0.6 s
generation rate. Sensor surface thermocouple Type-T
Sensing area 6.45 cm2

Fig. 9. (a) Thermocouple locations on a cell, (b) thermocouples connected to a cell, (c) heat flux sensor on a cell, and (d) thermocouple locations in battery module.

9
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

ΔV
q́ = (27)
S@Tsen
S@T◦ C is the sensitivity of the heat flux sensor, which is estimated
using
S@Tsen = [0.00334 Tsen + 0.917] Scalib (28)
Tsen is the temperature of thermocouple in C associated with the flux
◦

sensor. Scalib is the sensitivity of the flux sensor at 25◦ C, and it is

1.34 ± 0.03 μV/(W/m2 ). The experimental setup of HSCDC is the
modified version of SACS with the inclusion of fibers. Table 9 shows the
thermophysical properties of the high density polyethylene (HDPE) fiber
structure used in the experiment.
Fig. 10. Schematic diagram of the experimental setup.
5.3. Experimental setup of the system
Uncertainty in heat generation is calculated with
A 32 V, 5 A DC supply was used to charge battery, while a 20 Ω, 4 A √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2 ( )2 ( )2̅
rheostat served as the discharger (load) for the battery. The battery, q́gen δScalib δV δT
δ = 2 + + (32)
discharger, and ammeter were connected in series, the resistance q́gen Scalib V T
adjusted manually to maintain a constant current. The air velocity was
The maximum values obtained during the experiments are used as
measured using a thermal anemometer positioned at 10 mm upstream of
references to estimate uncertainties, which are shown in Table 11.
the first cell. Temperature and voltage data were recorded using a data
acquisition (DAQ) system at a frequency of 1 Hz. The experimental
6. Results and discussions
configuration of both SAC and HCSDC systems is illustrated in Fig. 10.
The charging process follows a constant current constant voltage
6.1. Simple air-cooling system
(CCCV) protocol until the cells reach their maximum voltage. A specific
time interval is maintained between the charging and discharging pro­
Numerical simulations were performed at different inlet velocities
cesses to ensure the battery module returns to room temperature.
(0.25–2 m/s) and charge/discharge rates to evaluate thermal behaviour
Throughout the experiment, the load is adjusted to maintain a constant
of different BTMSs. The effectiveness of various BTMSs was compared
current in the circuit. One-day gap is maintained between consecutive
based on the volume averaged temperature and temperature uniformity
experiments. Fig. 11 illustrates the experimental setup for the charging
of battery. The battery module centerline temperature profile was uti­
and discharging experiments.
lized to determine the temperature variation in the wake region between
two neighbouring cells in the battery module.
5.4. Uncertainty analysis

6.1.1. Flow pattern inside the battery module

Moffatt [44] proposed a method to calculate uncertainty in experi­
Fig. 12 shows the temperature contour of the battery module at a
ments, where R represents a variable dependent on several other
discharge rate of 2C and an inlet velocity of 1 m/s. As the fluid flows
variables.
from the inlet to the outlet, the temperature of the cells increases due to
R = R(X1 , X2 X3 ….Xn ) (28) heat interaction between the cells and the progressively warmer fluid.
The seventh cell experiences the highest temperature than eighth cell
Where X1 , X2 , X3 , ….Xn represent the independent variables. Let δXi
due to changes in fluid dynamics and heat transfer after the eighth cell.
represent the uncertainty in the measurement of Xi , so that the mea­
Fig. 13 illustrates the presence of a wake region between adjacent cells,
surement of Xi can be represented as Xi ± δXi . The uncertainty in Xi
where the fluid circulates at a lower velocity. The fluid in the wake re­
affect the variable R as given below
gions has elevated temperatures compared to the fluid flowing close to
δR the symmetry walls, as the prevailing dynamic conditions and heat
δR = δX i (29)
δXi transfer mechanisms differ in these two locations. The cell surface
Final uncertainty associated is given as temperature near the wake region is higher than that near the symmetry
face, indicating improved heat interaction with the fluid near the sym­
{ )2 }1/2
n (
∑ δR metry face. In contrast, the high-velocity region near the symmetry face
δR = δXi (30) shows less heat interaction, resulting in inefficient utilization of air for
δXi
heat exchange. The wake region after the eighth cell is the largest,
i=1

The uncertainties associated with different instruments are given in allowing sufficient area for heat transfer to the surrounding high-
Table 10. velocity fluid.
Volumetric heat generation rate in the cell is given by
6.1.2. Centerline temperature
4V
q́gen = (31) Fig. 14 illustrates the simulation of the system’s centerline temper­
([0.00334 Tsen + 0.917] Scalib )0.018
ature along its length at a discharge rate of 2C and fluid velocity of 1 m/
s, highlighting the hottest region within the system. Each peak in the
trend represents temperature variation along the centerline of a cell in
Table 9 the battery module, following the flow direction. The temperature shift
Thermophysical properties of HDPE fiber structure [42,43]. between successive maxima corresponds to temperature variation along
the center of the wake region between cells. The absence of cells beyond
Property Specifications/Values
the eighth cell results in a larger wake region. This expanded wake re­
Density (Kg/m3) 954 gion increases the contact area between the wake and the high-velocity
Specific heat (J/kg K) 1600
fluid regions. The high-velocity fluid, having a lower temperature,
Thermal conductivity (W/m K) 1.21

10
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 11. (a) Battery module of HCSDC, (b) Battery module of SACS, and (c) Battery charge/discharge experimental setup.

absorbs heat from the wake region, resulting in lower temperatures in

Table 10
the wake region and the last cell.
Uncertainties of different instruments.
Parameter Uncertainty 6.1.3. Effect of Reynolds number
Thermocouple and DAQ ±1.03 + 0.0075 × reading (◦ C) Fig. 15 shows the simulation results illustrating the impact of Reynolds
Temperature number on the performance of the SACS during 1C discharge. At a lower
DAQ Voltage ±(0.00004 + 0.000006) × reading (V) + (0.000006
Reynolds number of 764, the average temperature is 308.71 K, which is
+ 0.000001) × Range
Anemometer ±0.1 m/s + 0.005 × measured value comfortably below the limiting temperature of 313.15 K. As the Reynolds
Ammeter ±0.003 × measured reading number increases, the average temperature of the module decreases which
indicates enhanced heat transfer at higher mass flow rates. The percentage
difference in average temperature between Reynolds numbers 3056 and
Table 11 4584 is 0.25 %, while it is only 0.135 % between Reynolds numbers 4584
Maximum uncertainty of different variables. and 6112. This indicates that the rate of reduction in average module
temperature decreases as Reynolds number increases.
Parameter Reference Absolute bias Uncertainty associated with
value error each value (%)
The heat transfer coefficient between the fluid and the surface of the
cells increases with Reynolds number. This is because the higher ve­
Temperature 34.99 1.29
locity of fluid leads to frequent replacement of fluid near the cell surface.
±3.68
(◦ C)
Heat flux (W/ 66,536.39 3.380 ±5.08 Additionally, increasing the velocity improves the temperature unifor­
m2) mity of the battery module. Under 1C discharge conditions, the SACS
Voltage (V) 30.22 0.00069 ±0.0023 effectively maintains temperature uniformity below the optimum level
Velocity (m/s) 1.57 0.107 ±6.87
of 5 K at higher Reynolds numbers. However, when the Reynolds
number is <1528, the temperature uniformity exceeds 5 K. Specifically,

Fig. 12. Temperature contours of SACS at 2C discharge and 1 m/s.

11
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 13. Velocity contours in the flow field of SACS at 2C discharge and 1 m/s.

of each cell at 1C discharge with an inlet velocity of 1 m/s. The average

cell temperature generally increases in the direction of the flow, except
for the final cell, which has a lower temperature compared to the
second-to-last cell. The first cell experiences the lowest temperature
among all the cells since it interacts with the air at the lowest temper­
ature, resulting in increased heat transfer rate and lowest temperature.
As the air temperature increases in the flow direction, the volumetric
average temperature of the cells rises accordingly.

6.1.4. Effect of charge and discharge rates

Simulations were conducted at discharge rates of 1C, 2C, and 3C to
evaluate the performance of the SACS. Fig. 17 illustrates the comparison
of thermal performance of the battery module at different charge/
discharge rates based on the average temperature of the module and the
temperature difference within the module. The trends of the temperature
and temperature difference with Reynolds number are similar at higher C-
rates compared to the 1C-rate. Increasing the C-rate at a specific Reynolds
number results in higher volume-averaged temperature and temperature
difference across the module. This is due to the increased rate of heat
generation in the cells during higher charge/discharge rates, interacting
with air under the same thermal conditions. The SACS successfully keeps
Fig. 14. Centerline temperature of SACS at 2C discharge and 1 m/s. the average temperature of the battery module below the limiting tem­
perature at 1C and 2C-rates for higher Reynolds numbers (Re ≥ 2292).
at the lowest Reynolds number of 764, the temperature nonuniformity is However, at lower Reynolds numbers (Re ≤ 2292), the SACS fails to
38.06 % higher than the optimum value. maintain the required optimal temperature at 2C-rate. At the 3C-rate, the
The average temperature of the battery module provides a compre­ average temperature of the module exceeds the limiting value even at
hensive overview of the temperatures of individual cells within the higher Reynolds numbers. Additionally, the SACS does not achieve tem­
module. It is crucial to maintain the temperature of each cell below the perature homogeneity across the entire range of Reynolds numbers at
optimum value. Fig. 16 illustrates the volumetric average temperature higher charge/discharge rates. Temperature homogeneity in the system is

Fig. 15. Variation of (a) average temperature of the module and temperature difference in the module (b) average heat transfer coefficient of the SACS with Reynolds
number at 1C discharge.

12
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

plane for different configurations at a 2C discharge and 1 m/s velocity.

The seventh cell consistently exhibits the highest temperature in all the
configurations. Among the different configurations of HSCDC, BP-4
demonstrates the lowest temperatures due to augmented fluid velocity
resulting from the reduced flow area caused by the incorporation of fi­
bers in the minimum area section.
Fig. 20 illustrates the variations of temperature and temperature
difference of the system with Reynolds number for different configura­
tions at a velocity of 1 m/s and a 2C charge/discharge condition. The
trends of temperature and temperature difference are consistent across
all the configurations. The inclusion of fibers in the HCSDC system does
not significantly improve the temperature of the battery module
compared to SACS, except for the BP-4 configuration. At lower Reynolds
numbers, all HCSDC configurations exhibit lower temperatures
compared to SACS. Among the different configurations, BP-4 maintains
the best temperature uniformity throughout the entire Reynolds number
range compared to SACS. The temperature uniformity of BP-4 is 27.4 %
better than that of BP-0 at lower Reynolds numbers below 764, and as
the Reynolds number increases, the deviation in temperature uniformity
between the two configurations decreases.
Fig. 16. Volumetric average temperature of the cells according to location at Fig. 21 presents a comparison of the trends of centerline temperature
1C condition. and cell-wise volumetric temperature for different configurations. Each
peak in the plots represents the diametrical temperature variation inside
achieved when the Reynolds number exceeds 1528 and 3056 at the 1C the cell, while the region between two peaks signifies the wake region
and 2C discharge rates, respectively. where air with very low velocity is present. Notably, the temperature of
the wake region is lower than the solid cell temperature, facilitating heat
6.2. Hybrid cooling system at dry out condition transfer from the cell to the coolant air in the wake region. The intro­
duction of a fiber structures in the flow field serves to divert the low-
The hybrid cooling system at dry-out condition (HCSDC) employs temperature main stream flow away from the near-symmetry region
solid fibers placed between the cells of the system, which influences the into the intercellular space. This diversion contributes to a reduction in
fluid flow and heat transfer characteristics, resulting in a different the size of the wake region in the direction perpendicular to the main
thermal management approach compared to SACS. To analyze the stream flow. Consequently, the centerline temperature trend in all
impact of position of and number of fibers on thermal performance, six HCSDC configurations differs from that of the SACS configuration. Each
different configurations of HSCDC were simulated at a 2C discharge rate battery pack configuration experiences a redistribution of centerline
with a fluid velocity of 1 m/s. Fig. 18 depicts streamlines plotted for temperature while maintaining a comparable average temperature. BP-4
HCSDC and SACS configurations at the horizontal central plane of the configuration, in particular, exhibits superior performance with the
system, illustrating the flow patterns. The inclusion of fibers helps to minimum centerline temperature and individual cell temperatures
redirect the flow towards the gaps between consecutive cells, reducing compared to all other configurations. The enhanced performance of BP-
the wake region compared to BP-0 configuration. BP-4 configuration 4 is attributed to the presence of two additional fibers in the minimum
exhibits the maximum velocity due to the minimum flow area it pos­ cross-sectional flow area, which accelerates the fluid and increases the
sesses. Fig. 19 displays temperature contours at the central horizontal convective heat transfer coefficient.

Fig. 17. Variation of (a) average temperature of battery and (b) temperature difference in the battery with Reynolds number at different discharge conditions.

13
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 18. Velocity contours of (a) BP-0, (b) BP-1, (c) BP-2, (d) BP-3a, (e) BP-3b, (f) BP-3c, and (g) BP-4 configurations at 1 m/s and 2C condition.

Out of the six configurations of HCSDC, better temperature unifor­ Fig. 25 shows the temporal variation in heat generation rate from a
mity and average module temperature were demonstrated by BP-4 single cell, which aligns with cell voltage variation. The rate of heat
configuration. Hence, thermal performance between BP-0 and BP-4 generation increases significantly at the beginning and end of the
configurations at various discharge rates is simulated and compared as experiment. Initially, the cell is in thermal equilibrium with the sur­
shown in Fig. 22. At all discharge rates, the temperature difference be­ roundings, and heat generated internally is conducted to the cell surface
tween the two configurations decreases as Reynolds number increases. and then transferred through the heat flux sensor. Heat is further con­
Even at 3C-rates, the BP-4 configuration shows significant improvement vected from the sensor to the ambient. Fig. 25 illustrates that thermal
in temperature uniformity compared to BP-0. delay and contact resistance between the cell and sensor cause a delayed
Fig. 23 illustrates the variation of pressure drop with Reynolds number heat generation rate at the start of the experiment. A larger thermal
for different configurations at a 2C discharge rate. As Reynolds number gradient between the cell and surroundings enhances the heat transfer
increases, there is a corresponding increase in pressure drop for all con­ rate to the surroundings. Towards the end of the experiment, the load
figurations, following the classic pressure drop relationship with flow resistance is reduced to maintain a constant current, resulting in higher
velocity. The change in pressure drop among the different configurations rate of heat generation due to the same current being extracted at a
is attributed to the presence of fibers, which alter the flow resistance. lower cell voltage. The heat generation within the cell is influenced by
Except for BP-0 and BP-4 configurations, the pressure drop ranges are changes in cell voltage. The average heat generation rate for a current of
similar across all configurations. BP-0 exhibits the lowest pressure drop, 1.55 A is 18,374 W/m3, which is approximately 3.3 times higher than
while BP-4 has the highest pressure drop due to the minimum and the value predicted by Bernardi equation.
maximum number of fibers, respectively. It is recommended to give more The capacity of a cell was determined by discharging it at two
priority to the temperature requirement on safety point of view, to avoid different current conditions of 1.55 A and 2 A and it is shown in
thermal runaway conditions than the increased pumping power. So, it is Table 12. The discharge time until reaching the cut-off voltage was
important to consider a minimal pressure drop for effective thermal recorded to calculate the actual deliverable capacity. It was observed
management while maintaining temperature uniformity and adhering to that the commercially available cell has a capacity lower than its rated
temperature limits across a wide range of Reynolds numbers. value, and the capacity decreases as the C-rate increases. This decrease is
attributed to the increased energy loss in the form of heat at higher C-
rates. The reduction in cell capacity as well as the higher deviation in
6.3. Experimental results
actual heat generation rate compared to analytically predicted values
indicate suboptimal performance of the domestically procured cell. To
6.3.1. Heat generation from cell
estimate the heat generation, it is recommended to use Bernardi equa­
Experiments were conducted on a single cell and an eight-cell battery
tion for standard cells while an experimental approach using heat flux
module under a constant current of 1.55 A to measure the heat generation
sensor is apt for locally available cell in market.
rate during discharge. Fig. 24 depicts the voltage variation of both the
single cell and the system of eight cells throughout the discharge process.
6.3.2. Temperature variation within the battery module
Initially, a gradual decline in voltage is observed, continuing until around
Experiments were conducted on BP-0 and BP-4 configurations of
3100 s. After this point, there is a rapid decrease in voltage. The experi­
SACS and HCSDC, respectively. Fig. 26 shows the temperature
ment continues until the cell reaches its cut-off voltage of 2.5 V.

14
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 19. Temperature contours of (a) BP-0, (b) BP-1, (c) BP-2, (d) BP-3a, (e) BP-3b, (f) BP-3c, and (g) BP-4 at 2C discharge and 1 m/s velocity.

variations of alternate cells in SACS under different discharge conditions rate of heat generation increased, causing accelerated temperature rise
and at 1.06 m/s. During the initial stage of discharge, there is a sudden in each cell. During steady-state operation at a higher discharge rate of 2
increase in temperature due to the cell’s transient start-up condition. A, the highest temperature of a cell remained below the optimum value.
However, after 747 s, the temperature variation becomes negligible, Towards the end of the discharge period at 2 A discharge and 1.06 m/s
indicating a nearly steady-state thermal condition during discharge at velocity, the system’s maximum temperature approached the threshold
1.55 A. The maximum temperature recorded by a cell during steady- value.
state operation is 303.5 K, with the seventh cell showing the highest Four experiments are carried out on HCSDC with two different inlet
temperature in the battery module. The temperature non-uniformity is velocities and discharge conditions as shown in Table 13.
1.56 K in steady-state conditions and maximum temperature non uni­ Fig. 27 shows the temperature variations of alternate cells in the BP-4
formity of the system is 3.8 K at 1.55 A discharge. Thus, the battery configuration during different discharge conditions. Rapid temperature
module successfully kept with its maximum temperature below the increases are observed at the beginning and end of the discharging
optimal level at 1.55 A discharge and 1.06 m/s velocity. The thermo­ process due to voltage decrease in these phases. In the intermediate
couples placed on the leeward side consistently measured higher tem­ stage, where the voltage remains steady, the cell temperature remains
peratures compared to those near symmetry face for each cell. relatively constant with time. Steady-state conditions are observed be­
Additionally, when the battery module reached its cut-off voltage, the tween 750 s and 3000 s for 1.55 A discharge, and between 750 s and

15
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 20. Variation of (a) average temperature and (b) temperature difference of different configurations of battery with Reynolds number at 2C and 1 m/s condition.

Fig. 21. Variation of (a) centerline temperature and (b) volumetric average temperature of cells in different configurations at 2C discharge with 1 m/s velocity.

2500 s for 2 A discharge. Temperature non-uniformity is observed in the uncertainties were considered, and the simulation predictions show
system, with cell temperatures increasing in the flow direction at various excellent agreement with a maximum difference of 0.31 % for BP-4 and
discharge conditions and velocities. The seventh cell consistently ex­ 0.48 % for BP-0 configurations with the experimental results. Due to the
hibits the highest temperature among all the cells in the HCSDC exper­ higher air velocity, the maximum rise in temperature from the ambient
iments. Initially, the cells are in thermal equilibrium with the ambient condition is observed in seventh cell and it is measured as 2.23 K from
conditions, and the temperature rise from the initial temperature de­ the experiment and the numerically predicted value is 1.39 K in HCSDC.
creases as airflow velocity increases. For a specific cell, the temperature It is evident that the lower C rate discharge (1.55 A and 2 A) will result in
rise is higher in the 2 A constant current discharge condition compared comparatively lower rate of heat generation and there by insignificant
to 1.55 A. temperature rise, which can be handled by air cooling strategy.
Experiments were conducted on SACS and the BP-4 configuration of
HCSDC at 2 A discharge with 1.57 m/s velocity. The experimental heat 7. Conclusions
generation rate data during the steady state was used as input for steady
state simulations of the BP-0 and BP-4 configurations under the same The present study focussed on the unexplored area of performance
discharge conditions. Fig. 28 compares the steady-state temperatures evaluation of HSCDC at higher C rates. The study involves the devel­
observed at two different locations of the cells in BP-0 and BP-4 con­ opment of a numerical model capable of simulating six different battery
figurations with the corresponding simulation results. All measurement module configurations of HSCDC. Performance of these configurations

16
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 22. Variation of temperature difference in BP-0 and BP-4 configurations

with Reynolds number at different discharge conditions. Fig. 25. Variation of rate of heat generation of single cell with time.

Table 12
Capacity of lithium-ion cell.
No. of Discharge Cut-off Duration of Capacity of
cells current voltage Discharge Cell
Id (A) (V) td (s) Ca (mAh)

1 1.55 2.5 3910 1625

1 2 2.5 2840 1575

Fig. 23. Variation of pressure drop with Reynolds number for different con­
figurations of the battery at 2C discharge. Fig. 26. Variation of the temperature of cells in SACS with time at (a) 1.55 A
and (b) 2 A.

Fig. 24. Variation of voltage with time during the discharge process of (a) single cell and (b) battery with eight cells.

17
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Table 13 • The temperature non-uniformity of the SACS is above 5 K for Re <

Experimental conditions of HCSDC. 1528 at 1C discharge. At 2C and 3C discharge, the non-uniformity is
Sl. No Inlet velocity (m/s) Constant discharge current (A) above the required optimum for all Re values. Among the different
configurations, the HCSDC with six fibers (BP-4) performs the best,
1 1.06 1.55
2 1.57 1.55 satisfying temperature uniformity and average temperature at higher
3 1.06 2 Re values for all discharge conditions. When Re is 764, the temper­
4 1.57 2 ature uniformity of BP-4 is 27.54 % higher than BP-0, and when Re is
6112, it is 17.84 % higher.
• Except for BP-4 configurations, the location of the fibers has minimal
are compared with a SACS under elevated C rates and various dynamic
effect on temperature uniformity when Re is above 3056. Incorpo­
coolant conditions. The simulation results identified the battery module
rating fibers into the minimum flow area of BP-4 configuration en­
with 6 fibers exhibits the best thermal performance at elevated C rates,
hances thermal performance in terms of temperature uniformity and
which is fabricated along with SACS and subjected to experimental
average module temperature.
testing to validate the numerical findings. The following conclusions
• Experimental results show that the temperature on the leeward side
shed valuable light on the superiority of HSCDC compared to SACS,
is higher than at the minimal flow area for all the cells in BP-0 and
providing insights into the thermal management of the system across
BP-4 configurations.
diverse operational conditions.
• The 7th cell consistently exhibits the highest temperature across all
the configurations in both simulations and experiments.
• At 1C discharge, the SACS maintains an average temperature within
• Under experimental conditions of 1.06 m/s velocity, the temperature
the allowable range, but, at 3C discharge, it exceeds the allowable
non-uniformity and maximum temperature of the SACS meet the
range for all Re values considered. For 2C discharge, the system
allowable range at a discharge current of 1.55 A. In experiments with
meets the requirements only at higher Re values.

Fig. 27. Variation of temperature of cells in BP-4 configuration with time at (a) 1.55 A with 1.06 m/s, (b) 1.55 A with 1.57 m/s, (c) 2 A with 1.06 m/s, and (d) 2 A
with 1.57 m/s.

18
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

Fig. 28. Comparison of simulation predictions with experimental results of (a) simple air cooling and (b) BP-4 configuration of HCSDC.

the BP-4 configuration, at discharge currents of 1.55 A and 2 A, the References

average temperature of the battery module increases by 0.78 K and
1.13 K above the ambient, respectively, at a velocity of 1.57 m/s. The [1] K. Parajuly, D. Ternald, R. Kuehr, The Future of Electric Vehicles and Material
Resources, UNU/UNITAR-SCYCLE UNEP-IETC. (2020).
temperature difference within the battery is 1.29 K and 1.87 K, [2] X. Yuan, C.W. Su, M. Umar, X. Shao, O.R. Lobonţ, The race to zero emissions: can
respectively. renewable energy be the path to carbon neutrality? J. Environ. Manage. 308
• Increasing the discharge current by 29 % at a velocity of 1.06 m/s (2022) https://doi.org/10.1016/j.jenvman.2022.114648.
[3] L.K. Singh, G. Mishra, A.K. Sharma, A.K. Gupta, A numerical study on thermal
results in a 0.5 % average temperature increase and 7 % temperature management of a lithium-ion battery module via forced-convective air cooling, Int.
difference increase for BP-4 configuration. J. Refrig. 131 (2021) 218–234, https://doi.org/10.1016/j.ijrefrig.2021.07.031.
• The steady-state numerical models used to simulate the SACS and [4] Y. Fan, Y. Bao, C. Ling, Y. Chu, X. Tan, S. Yang, Experimental study on the thermal
management performance of air cooling for high energy density cylindrical
BP-4 of HSCDC are validated with experiments. The simulation lithium-ion batteries, Appl. Therm. Eng. 155 (2019) 96–109, https://doi.org/
predictions are in excellent agreement with experimental results. 10.1016/j.applthermaleng.2019. 03.157.
The maximum rise in temperature from the ambient condition is [5] J. Jaguemont, L. Boulon, P. Venet, Y. Dube, A. Sari, Lithium-ion battery aging
experiments at subzero temperatures and model development for capacity fade
observed in seventh cell, and it is found that the difference in tem­
estimation, IEEE Trans. Veh. Technol. 65 (2016) 4328–4343, https://doi.org/
perature rise between simulation and experiment is 0.83 K in BP-4 of 10.1109/TVT. 2015.2473841.
HCSDC. [6] D.K. Sharma, A. Prabhakar, A review on air cooled and air centric hybrid thermal
management techniques for Li-ion battery packs in electric vehicles, 41 (2021).
[7] W. Chen, J. Liang, Z. Yang, G. Li, A review of lithium-ion battery for electric vehicle
CRediT authorship contribution statement applications and beyond, Energy Procedia 158 (2019) 4363–4368, https://doi.org/
10.1016/j. egypro.2019.01.783.
P.M. Sutheesh: Writing – original draft, Visualization, Validation, [8] Q.L. Yue, C.X. He, M.C. Wu, T.S. Zhao, Advances in thermal management systems
for next-generation power batteries, Int. J. Heat Mass Transf. 181 (2021), https://
Software, Investigation, Data curation. B. Rohinikumar: Writing – re­ doi.org/10.1016/j.ijheatmasstransfer.2021.121853.
view & editing, Visualization, Supervision, Project administration, [9] M. Alipanah, X. Li, Numerical studies of lithium-ion battery thermal management
Investigation, Data curation. systems using phase change materials and metal foams, Int. J. Heat Mass Transf.
102 (2016) 1159–1168, https://doi.org/10.1016/j.
ijheatmasstransfer.2016.07.010.
Declaration of generative AI and AI-assisted technologies in the [10] H. Zhou, F. Zhou, Q. Zhang, Q. Wang, Z. Song, Thermal management of cylindrical
writing process lithium-ion battery based on a liquid cooling method with half-helical duct, Appl.
Therm. Eng. 162 (2019), https://doi.org/10.1016/j.applthermaleng.2019.114257.
[11] M. Zhi, R. Fan, X. Yang, L. Zheng, S. Yue, Q. Liu, Recent research progress on phase
During the preparation of this work, the authors used ChatGPT in change materials for thermal management of lithium-ion batteries, J. Energy
order to improve readability and language. After using this tool, the Storage. 45 (2022) 103694, https://doi.org/10.1016/j.est.2021.103694.
[12] B. Wang, C. Ji, S. Wang, J. Sun, S. Pan, D. Wang, C. Liang, Study of non-uniform
authors reviewed and edited the content as needed and takes full re­
temperature and discharging distribution for lithium-ion battery modules in series
sponsibility for the content of the publication. and parallel connection, Appl. Therm. Eng. 168 (2020) 114831, https://doi.org/
10.1016/j.applther- maleng.2019.114831.
Declaration of competing interest [13] Y. Zhang, J. Huang, M. Cao, Z. Liu, Q. Chen, A novel flexible phase change material
with well thermal and mechanical properties for lithium batteries application,
J. Energy Storage. 44 (2021) 103433, https://doi.org/10.1016/j.est.2021.103433.
The authors declare that they have no known competing financial [14] N. Yang, X. Zhang, G. Li, D. Hua, Assessment of the forced air-cooling performance
interests or personal relationships that could have appeared to influence for cylindrical lithium-ion battery packs: a comparative analysis between aligned
and staggered cell arrangements, Appl. Therm. Eng. 80 (2015) 55–65, https://doi.
the work reported in this paper. org/10.1016/j. applthermaleng.2015.01.049.
[15] J. Zhao, Z. Rao, Y. Li, Thermal performance of mini-channel liquid cooled cylinder
Data availability based battery thermal management for cylindrical lithium-ion power battery,
Energ. Conver. Manage. 103 (2015) 157–165, https://doi.org/10.1016/j.
enconman.2015.06.056.
Data will be made available on request. [16] L.H. Saw, Y. Ye, A.A.O. Tay, W.T. Chong, S.H. Kuan, M.C. Yew, Computational
fluid dynamic and thermal analysis of Lithium-ion battery pack with air cooling,
Appl. Energy 177 (2016) 783–792, https://doi.org/10.1016/j.
apenergy.2016.05.122.

19
P.M. Sutheesh et al. Journal of Energy Storage 84 (2024) 110889

[17] J. Jaguemont, L. Boulon, Y. Dubé, A comprehensive review of lithium-ion batteries [30] Y. Lai, W. Wu, K. Chen, S. Wang, C. Xin, A compact and lightweight liquid-cooled
used in hybrid and electric vehicles at cold temperatures, Appl. Energy 164 (2016) thermal management solution for cylindrical lithium-ion power battery pack, Int.
99–114, https://doi.org/10.1016/j.apenergy.2015.11.034. J. Heat Mass Transf. 144 (2019), https://doi.org/10.1016/j.
[18] V.G. Choudhari, A.S. Dhoble, S. Panchal, International Journal of Heat and Mass ijheatmasstransfer.2019.118581.
Transfer Numerical analysis of different fin structures in phase change material [31] H. Xu, X. Zhang, G. Xiang, H. Li, Optimization of liquid cooling and heat
module for battery thermal management system and its optimization, 163 (2020). dissipation system of lithium-ion battery packs of automobile, Case Stud. Therm.
[19] L.H. Saw, Y.J. King, M.C. Yew, T. Ching Ng, W.T. Chong, N.A. Pambudi, Feasibility Eng. 26 (2021) 101012, https://doi.org/10.1016/j.csite.2021.101012.
study of mist cooling for lithium-ion battery, Energy Procedia 142 (2017) [32] A.M.N. Elmekawy, A.A. Ibrahim, A.M. Shahin, S. Al-Ali, G.E. Hassan, Performance
2592–2597, https://doi.org/10.1016/j.egypro.2017.12.197. enhancement for tube bank staggered configuration heat exchanger – CFD study,
[20] S. Shahid, M. Agelin-Chaab, Development and analysis of a technique to improve Chem. Eng. Process. - Process Intensif. 164 (2021) 108392, https://doi.org/
air-cooling and temperature uniformity in a battery pack for cylindrical batteries, 10.1016/j.cep.2021. 108392.
Therm. Sci. Eng. Prog. 5 (2018) 351–363, https://doi.org/10.1016/j. [33] S. Shahid, B. Chea, M. Agelin-Chaab, Development of a hybrid cooling concept for
tsep.2018.01.003. cylindrical li-ion cells, J. Energy Storage. 50 (2022) 104214, https://doi.org/
[21] C.K. Mangrulkar, A.S. Dhoble, S.G. Chakrabarty, U.S. Wankhede, Experimental and 10.1016/j.est.2022. 104214.
CFD prediction of heat transfer and friction factor characteristics in cross flow tube [34] R. Mahamud, C. Park, Reciprocating air flow for Li-ion battery thermal
bank with integral splitter plate, Int. J. Heat Mass Transf. 104 (2017) 964–978, management to improve temperature uniformity, J. Power Sources 196 (2011)
https://doi.org/10.1016/j.ijheatmasstransfer.2016.09.013. 5685–5696, https://doi.org/10.1016/j.jpowsour.2011.02.076.
[22] S. Shahid, M. Agelin-Chaab, Experimental and numerical studies on air cooling and [35] L.H. Saw, Y. Ye, A.A.O. Tay, Electrochemical – thermal analysis of 18650 Lithium
temperature uniformity in a battery pack, Int. J. Energy Res. 42 (2018) 2246–2262, Iron phosphate cell, Energ. Conver. Manage. 75 (2013) 162–174, https://doi.org/
https://doi.org/10.1002/er.4018. 10.1016/j. enconman.2013.05.040.
[23] L. Sheng, L. Su, H. Zhang, K. Li, Y. Fang, W. Ye, Y. Fang, Numerical investigation on [36] D. Bernardi, E. Pawlikowski, J. Newman, A general energy balance for battery
a lithium ion battery thermal management utilizing a serpentine-channel liquid systems, Electrochem. Soc. Ext. Abstr. 132 (1970) 164–165.
cooling plate exchanger, Int. J. Heat Mass Transf. 141 (2019) 658–668, https://doi. [37] F.R. Menter, Two-equation eddy-viscosity turbulence models for engineering
org/10.1016/j.ijheatmasstransfer.2019.07.033. applications, AIAA J. 32 (1994) 1598–1605, https://doi.org/10.2514/3.12149.
[24] L. Sheng, H. Zhang, L. Su, Z. Zhang, H. Zhang, K. Li, Y. Fang, W. Ye, Effect analysis [38] P.R. Tete, M.M. Gupta, S.S. Joshi, Numerical investigation on thermal
on thermal profile management of a cylindrical lithium-ion battery utilizing a characteristics of a liquid-cooled lithium-ion battery pack with cylindrical cell
cellular liquid cooling jacket, Energy 220 (2021) 119725, https://doi.org/ casings and a square duct, J. Energy Storage. 48 (2022) 104041, https://doi.org/
10.1016/j.energy.2020.119725. 10.1016/j.est.2022.104041.
[25] Y. Wei, M. Agelin-Chaab, Development and experimental analysis of a hybrid [39] Samsung, Specsheet INR18650-30Q, (2015) 0–19.
cooling concept for electric vehicle battery packs, J. Energy Storage. 25 (2019), [40] S. Shahid, M. Agelin-Chaab, Analysis of cooling effectiveness and temperature
https://doi.org/10.1016/j. est.2019.100906. uniformity in a battery pack for cylindrical batteries, Energies 10 (2017), https://
[26] Y. Wei, M. Agelin-Chaab, Experimental investigation of a novel hybrid cooling doi.org/10.3390/en10081157.
method for lithium-ion batteries, Appl. Therm. Eng. 136 (2018) 375–387, https:// [41] FluxTeq, PHFS-01 Heat Flux Sensor Datasheet, (2022). https://www.fluxteq.
doi.org/10.1016/j.applthermaleng.2018.03.024. com/phfs-01-heat-flux-sensor.
[27] Z. Lu, X. Yu, L. Wei, Y. Qiu, L. Zhang, X. Meng, L. Jin, Parametric study of forced air [42] L. Shi, J.H. Gong, D.M. Huang, X.Y. Liu, G.M. Zhang, Numerical study on the
cooling strategy for lithium-ion battery pack with staggered arrangement, Appl. spontaneous combustion of high-density polyethylene, Procedia Eng. 211 (2018)
Therm. Eng. 136 (2018) 28–40, https://doi.org/10.1016/j. 621–628, https://doi.org/10.1016/j.proeng.2017.12.056.
applthermaleng.2018.02.080. [43] Y. Qin, D. Liang, Y. Wang, J. Chai, H. Zhou, Experimental study on mechanical
[28] K. Jiang, G. Liao, J.E.F. Zhang, J. Chen, E. Leng, Thermal management technology properties of concrete with cold joints, Paiguan Jixie Gongcheng Xuebao/journal
of power lithium-ion batteries based on the phase transition of materials: a review, drain, Irrig. Mach. Eng. 41 (2023), https://doi.org/10.3969/j.issn.1674-
J. Energy Storage. 32 (2020) 101816, https://doi.org/10.1016/j.est.2020.101816. 8530.21.0328.
[29] C. Ji, B. Wang, S. Wang, S. Pan, D. Wang, P. Qi, K. Zhang, Optimization on [44] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Thermal
uniformity of lithium-ion cylindrical battery module by different arrangement Fluid Sci. 5 (1988) 3–17, https://doi.org/10.1016/0894-1777(88)90043-X.
strategy, Appl. Therm. Eng. 157 (2019) 113683, https://doi.org/10.1016/j.
applthermaleng.2019.04.093.

20