You are on page 1of 16

Journal of Energy Storage 58 (2023) 106299

Contents lists available at ScienceDirect

Journal of Energy Storage


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

Research papers

Optimization of battery cooling system used in electric vehicles


Shrikant R. Patil a, Bhaskara Rao Lokavarapu a, *, Hareesh Karonnin Thaliyanveedu b
a
School of Mechanical Engineering, Vellore Institute of Technology, Chennai Campus, Vandalur-Kelambakkam Road, Chennai 600127, Tamil Nadu, India
b
Industry Expert, Bangalore, Karnataka, India

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

Keywords: In order to achieve superior performance and battery life, battery thermal management (BTM) has a significant
Battery Electric Vehicles (BEVs) role to play in maintaining the temperature of the battery within a defined range. In different environmental and
Battery Thermal Management Systems (BTMs) working conditions, the temperature of the battery increases, affecting its capacity. The battery cooling plate can
Heat flux
be examined at the cell or module level. Until now, all optimizations have been conducted at the cell level alone.
Cooling plate
The current study examines the optimization of battery cooling plates at a module level. Two different modules
Battery pack
are analyzed, namely Z-type and original cooling plates. As compared with the original cooling plate, the Z-type
plate provides better performance. Thermal simulations are validated based on published results. The bottom
surface of the cooling plate has been modified in order to improve its performance and to create different fluid
domains within the cooling plate. To optimize the cooling plate, the Z-type model is utilized in order to minimize
pressure drop, maximize temperature uniformity, and minimize average temperature in the contact area between
the cooling plate and batteries. This optimization provides the best design, resulting in a 23.68 % improvement in
pressure drop from 456 mbar to 348 mbar. The variation in temperature at the contact area (Delta T (solid)) has
been reduced by 23.28 %, from 8.59 ◦ C to 6.59 ◦ C. The Delta T (fluid) has decreased from 4.81 ◦ C to 2.62 ◦ C,
which represents a 45.53 % improvement.

expensive to manufacture. They are also highly susceptible to temper­


ature changes [1,7,8]. In addition, the difference between the maximum
1. Introduction and minimum temperature within a battery pack should be less than or
equal to 5 ◦ C [1,9]. Therefore it is necessary to have a well-designed
Sustainable environments are made possible by electric vehicles. The thermal management system.
demand for electrical vehicles in the automobile industry is growing. Battery life and energy capacity are highly influenced by the tem­
Electric vehicles are seen as futuristic vehicles which will have more perature of the battery [4,9,10]. During a high-temperature condition,
safety and a long-range compared to traditional vehicles [1,2]. Over the there may be thermal runaway [1,10,11]. When a low temperature oc­
past decade, there has been a consistent increase in the transportation curs, internal resistance increases, and electric current decreases. As a
industry due to a growing population, leading to an increase in fuel result, the battery capacity is reduced over time. Considering the range,
consumption throughout the world. In order to reduce pollution, it is performance, durability, and value of the battery system, the capacity of
very imperative to use renewable energy in addition to electric vehicles. a vehicle affects its value and performance in the long run. Currently,
Energy has been created in most developed countries through the use electric vehicles (EV), plug-in hybrid electric vehicles (PHEVs), and
of renewable resources, which has shown to have a positive impact [3]. hybrid electric vehicles (PHEVs) have batteries that are backed by a
During the last two decades, considerable research has been undertaken warranty of eight years and powertrains backed by a five-year warranty
on the storage of renewable energy and the availability of materials like [12]. In order to maintain the performance of the battery, the thermal
solar panels and wind energy [4,5]. One of the most popularly used management system of the battery must effectively disperse heat.
batteries is Li-ion because of its high specific energy and power [1,4]. As shown in Fig. 1, a battery pack has different levels. A cell consists
Despite being one of the most effective energy storage devices (ESS), of positive and negative electrodes and an electrolyte. The heat gener­
ineffective packaging is a common reason for battery failure [6]. In most ated is due to the conversion of chemical energy into electrical energy at
cases, faulty packaging leads to increased battery temperature as a result the cell level. Stacks of these cell levels are called modules, and they are
of inefficiency in thermal management systems. Li-ion batteries are

* Corresponding author.
E-mail address: bhaskarbabu_20@yahoo.com (B.R. Lokavarapu).

https://doi.org/10.1016/j.est.2022.106299
Received 31 May 2022; Received in revised form 12 November 2022; Accepted 28 November 2022
Available online 12 December 2022
2352-152X/© 2022 Elsevier Ltd. All rights reserved.
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Acronyms Delta T (Maximum temperature – Minimum temperature) in


contact area of the cooling plate and battery pack, K
BTMs Battery Thermal Management system Tuni Temperature uniformity
ρ Density, kg/m3 ΔPfluid P inlet – P outlet, bar
Mcell Mass of cell Tavg Average temperature defined over bottom surface of the
qcell Heat generated in cell, W plate where heat flux is applied, K
qr Heat rejected in liquid coolant, W Tσ Standard deviation of the temperature defined over bottom
Cp,c Cell specific heat, J/(K-kg) surface of the plate, K
Cfluid Fluid specific heat, J/(K-kg) Tf Face value of Average Temperature, K
ṁ Mass flow rate, kg/s ∅ Surface Temperature – Inlet Temperature, K
ΔTfluid Fluid Outlet Temperature – Fluid Inlet Temperature, K ∅ Surface average of ∅
Umean Mean velocity (Flow rate/ area), m/s Face value of the ∅
_
∅f
Dh Hydraulic diameter, m Af Area of face, m2
μ Dynamic Viscosity, Pa-s

Fig. 1. Battery cell, module and pack level [15].

Fig. 2. Battery cooling method.

arranged in different positions depending on the space available. In When used as a high-power application, the battery cell generates more
order to produce the voltage and current required, the modules are heat compared to a regular cycle. The second factor is the geographical
connected in parallel and series. A pack is a collection of modules region where it is used, such as a hot or cold climate. It is essential to
stacked together. maintain the inlet temperature when operating in hot climates. A cell’s
Heat exchangers are used in battery thermal management to remove chemistry is another factor that affects the cell temperature. Different
heat from the battery cell, pack, or module [13–15]. Batteries with BTMs chemistry generates different amounts of heat as well. Essentially, the
are capable of charging and discharging very quickly. The complexity of main function of a BTM is to cool the battery pack at high temperatures
the battery pack depends on the duty cycle under which it is operated. and to heat the battery pack at temperatures between 0 ◦ C and 20 ◦ C.

2
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 3. Heat pipe [16].

Fig. 4. Methodology.

When the temperature difference in the battery pack is large, each reactions.
cell is degrading at a different rate. In batteries, all three types of heat Studies have also demonstrated that low temperatures greatly reduce
transfer should be considered, namely, conduction, convection and ra­ the battery capacity by up to 80 % [1,4,17]. At low temperatures, the
diation. Conduction and convection play a dominant role in the heat electrical impedance increases while the motion of lithium ions within
transfer of the battery. In general, Li-ion batteries should operate be­ the electrolyte is restricted. This phenomenon is usually removed from
tween 15 ◦ C and 35 ◦ C [1,4,9,12]. An exothermic reaction is more likely batteries by heating them. The issue with any battery system is not the
if the temperature is raised over 40 degrees Celsius. As a result, the speed of charging or discharging but rather the amount of heat gener­
battery pack temperature increases and fumes are produced. Once the ated during operation [1,4].
fires are started in the battery pack, they cannot be stopped. This is
called a thermal runaway. The phenomenon of thermal runaway occurs 2. Battery cooling methods
at different temperatures based on the chemistry of the cell. Several
studies have shown that electrode modification can be reduced within Fig. 2 shows the different battery cooling systems, including air,
the cell temperature [16]. Taking into account that the temperature liquid, refrigeration, PCM and heat pipe cooling systems [1,17,18].
within the cell pack has an influence on its performance, BTMs help A passive cooling system removes heat from the battery using cabin
maintain the temperature within the cell pack [1,2,17]. It is quite air without the need for external power and is usually open circuit in
common for heat to be generated in the electrode region during reaction most cases. Passive cooling relies on cabin air as a cooling agent. Active

3
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 5. Heat transfer in battery cooling plate.

Fig. 6. Original cooling plate.

Fig. 7. Top surface of original cooling plate.

cooling is achieved by using two loops, the first cooling/heating the air lightweight when compared with liquid cooling [13,18]. It is possible to
flowing into the battery pack. The second loop of this cooling system is heat the battery with air. Air cooling systems are widely utilized for
connected through a chiller unit to the heat, ventilation, and air con­ small electric vehicles [3,9,12]. Consequently, it is difficult to target a
ditioning (HVAC) in the vehicle, which maximizes its efficiency by uti­ sufficient flow rate and air inlet temperature since they depend on the
lizing the vehicle’s climate control system to heat, ventilate, and cool the passenger’s environment as well as the surrounding air. Air cooling re­
vehicle. Air cooling has the advantage of being simple to install and quires proper ducts and space for moving air between battery packs.

4
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

laminar flow results in poor heat rejection and reduced heat transfer
efficiency, it is desirable to design liquid coolant channels with turbulent
flow [9]. Turbulent flow has a higher heat carrying capacity than
laminar flow. Turbulent flow presents the disadvantage of a high pres­
sure drop as compared to laminar flow. As a result, most liquid cooling
systems utilize laminar flow. Liquid cooling permits increasing the
cooling contact area by adding a cooling plate. Temperature uniformity
can be achieved easily with liquid cooling [2]. In addition, liquid cooling
increases the complexity and weight of other components as well. Liquid
cooling has a high thermal conductivity and efficiently transfers heat
from cells. Due to this, it has been commercialized for use in high per­
formance battery-electric vehicles [8,12]. The new system layout is
designed to reduce the complexity of the liquid cooling system. It has
also been demonstrated there is a significant effect of inlet temperature
Fig. 8. Inlet and outlet.
on the performance of the battery [20].
Compared to air or liquid cooling, phase change materials (PCM)
Increased airflow in the battery pack will lead to increased noise have the advantage of transitioning through multiple physical phases,
pollution, while increased vibration and harshness (NVH) of the vehicle generally solid to liquid when heated [3,21]. It is mainly made up of wax
will reduce passenger comfort. Cell temperatures will vary widely across and graphite based materials. The use of paraffin and graphite com­
the battery pack as a result of the air’s low specific heat capacity [19]. posite materials has been studied in some cases as a way to reduce the
A liquid cooling system does not depend on the temperature of the maximum temperatures of batteries [17,21]. This material does not
cabin. It is quieter, but there is a possibility that it could leak, but transition into a liquid phase, but rather softens the solid state. For the
thermal uniformity can be maintained by using this system [20]. As purpose of removing heat from the battery pack, air and liquid systems

Fig. 9. Z-Type cooling plate.

Fig. 10. Mesh convergence plot of pressure drop objective function.

5
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Table 1 it is not commercialized.


Cooling plate analysis parameters. A wide variety of electronic applications are cooled by heat pipes,
Material properties Values which remove heat through evaporation. The Heat pipe system must be
in proper contact with the batteries for satisfactory operation [1,21].
Coolant fluid Water-ethylene glycol
Coolant viscosity (Pas) 0.003331667 Heat pipe cooling systems are more compact than the three systems
Coolant conductivity (W/m-K) 0.419 described above. It is also critical to note that heat pipes require sec­
Coolant specific heat (J/kg-K) 3322.5 ondary cooling systems such as air or water cooling. As shown in Fig. 3, a
Coolant density (kg/m3) 1067.25 heat pipe is composed of three different regions, which are evaporative,
Solid Aluminum
Plate conductivity (W/m-K) 180.0
adiabatic, and condensation regions.
Plate specific heat (J/kg-K) 903.0 Besides its long lifespan, heat pipes have a high thermal response and
Plate density (kg/m3) 2702.0 efficiency. Heat pipes have the disadvantage of evaporating the entire
liquid upon increasing operating temperatures, which eliminates the
Dimensions possibility of heat transfer between the heat pipe and the battery alto­
Plate thickness (mm) 2.5 gether. Another disadvantage of heat pipes is their relatively high cost.
Plate length (mm) 1919 Heat pipes involve a two phase flow that is difficult to model. Therefore,
Plate width (mm) 575
they are more suitable for prismatic cells. Experimental results alone are
Fluid domain thickness (mm) 2.5
Initial fluid domain width (mm) 150 available [1]. Due to these problems, commercial availability of heat
pipes has not been achieved.
Boundary condition
As a result of the battery life and safety, LTO/LiFePO4 is the most
Coolant inlet mass flow (kg/s) 0.125–0.5 appropriate Li-ion battery chemistry due to its low specific energy,
Coolant inlet temperature (K) 298.15 performance, durability, energy management, and safety [3]. Compared
Coolant outlet pressure (Pa) 0 to today’s batteries, batteries are required to have 2–3 times more en­
Heat flux (W/m2) 2272.72
ergy density to meet the current demands of high performance cars.
According to current research, Li-ion batteries are expected to reach 500
use sensible heat while PCM uses latent heat. It takes a large amount of miles by 2022. This expectation can be met by the new research in super
latent heat to convert a solid to a liquid state. The pure PCM, therefore, is capacitors in conjunction with batteries [22].
an excellent choice for BTMs; however, the main disadvantage of the In the past, the majority of optimization methods have been evalu­
pure PCM is that its thermal conductivity is very low, 0.2–0.77 W/m-K, ated at the cell levels [1,23–26]. However, the battery cooling plate can
which is very low compared to the sensible heat of liquid [1]. Even when be installed at two levels, namely the cell level and the module level. The
composite structure or nanoparticles are added, the thermal conduc­ present study examines the optimization of battery cooling plates at the
tivity remains within the range of 2–11 W/m-K. However, this is module level. Using this method, one of the benefits is that we can
insufficient to remove heat from the battery. As another critical simulate flow and thermal stress on the entire system. In thermal anal­
parameter, when the PCM is converted to liquid form, the thermal ysis, changes in a substance’s physical properties are determined by the
conductivity drastically decreases. This causes the battery to heat up time and temperature at which they occur. By using thermal stresses it is
even more, increasing the possibility of leakage. The PCM is combined possible to optimize structural design for better performance.
with another liquid in order to reduce the effects of leakage. There is a Using a module level cooling system allows multiple batteries to be
need to take proper care when expanding the PCM, which explains why cooled simultaneously by a single cooling plate, thus reducing the cost of

Fig. 11. Geometry model [5].

6
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 12. Heat simulation results.

It is the purpose of this paper to evaluate the thermal performance of the


Table 2 existing design with constant heat flux and to improve it by using
Final results compare using Jarrett and Kim [5]. Optimate+ (HEEDS) as a design modification tool.

Geometry ΔP fluid (Pa) Tavg (K) Tσ (K) 3. Methodology

This study illustrates the implementation of a pack level cooling


Jarrett and Kim [5] 2948 306.09 2.64 method for a battery electric vehicle. Batteries are grouped together to
generate the required current and voltage. For this study, Star CCM+
software has been utilized for conjugate heat transfer simulation,
Simulation 3241.6 306.13 2.47 whereas Optimate+ (HEEDS) has been employed for optimization. The
steps involved are as follows and shown in Fig. 4:

• Study and select a battery cooling system based on the requirements.


the battery cooling system. As each cell requires a separate cooling plate, • Take the Original cooling plate model and clean up the geometry in
the number of such plates required for cell cooling will be higher. This Catia and CCM+ according to simulation requirements.
will increase both the cost and the weight of the battery cooling module.

Fig. 13. Delta T in cooling plate.

7
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 14. Z-Type model fluid domain.

Fig. 15. Z-Type cooling plate top surface.

Fig. 16. Z- Type model with different types of variables.

• It is critical to validate the flow and thermal simulation result against • Investigate the effect of geometry changes on the mass flow in all
the published result [5]. If the results do not match, change the mesh ducts or channels.
and solver settings. • Find out the temperature uniformity and temperature distribution on
the cooling plate.

8
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 17. Z- Type cooling plate: Delta T = 8.51 ◦ C, Tuni = 87.37 %.

Fig. 18. Velocity plot of Original cooling plate model.

• Perform the multi-objective optimization with respect to pressure are employed in this study in order to simplify the problem, and these
drop, temperature uniformity and average temperature. assumptions have been widely implemented in various research papers
[1,4,9,12] and in the automotive industry as well.
4. Heat transfer in battery cooling plate
1. The simulations were conducted on cooling plates in this study.
4.1. Assumptions Batteries have not been modeled in the simulation. Heating of bat­
teries has not been addressed in this study. In this study, the objective
A heat transfer diagram illustrating how heat is transferred from a is to cool battery packs uniformly. The simulation has not taken into
battery to a coolant is shown in Fig. 5. The batteries generate heat when account other electronic and mechanical components of the battery’s
charging and discharging. This heat is carried from the batteries to the thermal management system.
bottom surface of the pack and dissipated by the coolant. Thermal pads 2. Batteries dissipate heat in a uniform manner. Conduction is the only
are used to remove air gaps from the contacting zone, thereby reducing method of removing heat from batteries at the present time.
thermal resistance. Batteries have not been modeled. In place of that, a 3. Throughout this study, steady state is assumed because the temper­
constant heat flux was applied to the contact area of the cooling plate. Li- ature change in the cooling liquid is <5 ◦ C. The roughness of the
ion batteries are widely used in the automotive industry. Li-ion batteries cooling channel has not been taken into consideration.
alone are used in determining the cooling, plate simulation boundary 4. In this study, materials were considered to have only isotropic
conditions. Adding batteries, electronic components, etc., will require properties.
large amounts of computer resources. This study only examines the 5. It is assumed that radiation losses are negligible.
cooling plate, and does not examine other components, such as batteries 6. The convection from the cooling plate’s side surface has been taken
and electronic components. Similar principles have been applied to the into account.
battery cooling plate in references [5,9,11]. The following assumptions

9
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 19. Pressure plot of Original cooling plate model.

Fig. 20. Multi-objective optimization design variables.

Heat transfer in battery cooling plate: ( )


∂ui ∂ui 1 ∂P ∂ u ∂ui
CONDUCTION (heat transfer between batteries and thermal pad). + uj = − + (1)
CONDUCTION (heat transfer between thermal pad and cooling ∂t ∂xj ρ ∂xi ∂xj ρ ∂xj
plate). A battery pack generates heat when chemical energy is converted
CONVECTION (heat transfer into the coolant). into electrical energy. Eq. (2) represents the energy balance equation.
On the left side of the equation is the stored energy in the battery. On the
4.2. Governing equation and boundary condition right side of the equation is the heat generated by the battery and the
energy rejected in the liquid coolant [6]. Heat rejected by the battery in
Segregated flow solver settings have been used in Star CCM+ sim­ the liquid coolant is calculated by using Eq. (3).
ulations. It solves the problems of momentum and energy equations. The
basic governing equation of fluid flow in Star CCM+ is the Navier-Stokes dT cell
Mcell Cp,c = qcell − qr (2)
equation. This study considers only the steady state condition. Eq. (1) dt
shows the Navier- strokes equation. qr = mCp,fluid ΔTfluid (3)

10
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 21. Contact area between battery pack and top surface of cooling plate.

Battery module total Area = 0.88 m2, Total heat load = 2000 W, Heat
Table 3 flux = 2000 W/0.88 m2, Heat Flux on contact area = 2272.72 W/m2.
Original and Optimate+ (HEEDS) simulation results: Mass flow rate, Delta P and
Delta T.
5. CFD model and validation
Sr. Name Mass Flow Delta P Delta T (Solid)
No. Rate (kg/S) (mbar) (°C) There are two cooling plates examined in this study: the original and
the Z-type cooling plates. The model shown in Fig. 7 is referred to as the
1 Original cooling plate 0.125 457 8.59
model “original cooling plate”. As shown in Fig. 7, this model consists of three
distinct regions: solid surfaces on top and bottom, and fluid regions
between the solid surfaces. There is a magenta color (dark color) indi­
2 Z-Type Design Model 0.125 248 8.51

cating the location of the battery’s contact area. Fig. 8 shows the main
3 Optimate+ (HEEDS) Best 0.227 348 6.59 inlet and outlet of the cooling plate and Fig. 9 illustrates the Z-type
Design 1
cooling plate fluid domains formed by modifying the bottom surfaces.
Fig. 7 illustrates the similar geometry and contact area of the top sur­
faces of the two cooling plates. The Z-type cooling plate is modeled in
Star CCM+.
Table 4
Original and Optimate+ (HEEDS) simulation results: Delta T (Fluid), Temper­
ature Uniformity, Standard deviation.
5.1. Mesh independence study

Sr. Delta T Standard Temperature The top solid thickness is 2.5 mm, the bottom solid thickness is 1 mm,
Name
No. (Fluid) (°C) Deviation Uniformity (%) and the fluid domain thickness is 2.5 mm. In the mesh setting, different
Original cooling plate
parameters are used to create meshes in the fluid domain, including
1
model
4.81 - - mesh elements and mesh sizes. Mesh settings include elements like
polyhedral, thin mesher, and prism mesh. A conformal mesh has been
2 Z-Type Design Model 4.82 1.64 87.37 generated for the entire solid domain using thin type meshers. This
ensures that three cells appear in the vertical plane of the solid domain.
3 Optimate Best Design 1 2.62 1.24 88.97 As shown in Fig. 10, the convergence plot for the coolant pressure drop
is based on 5 million elements, which is sufficient to achieve a conver­
gence rate of 1 %. CFD analysis is performed using STAR CCM + Solver.
A prism mesh is constructed in order to simulate boundary layers and
heat transfer between fluids and solids. Table 1 presents information on
where, Mcell = Mass of cell
the material properties and boundary conditions used in the simulation.
qcell = Heat generated in cell
As a general rule, when doing volume meshing, the base size refers to the
qr = Heat rejected in liquid coolant
size of the largest elements in the mesh. If the base size is large, the mesh
Cp,c = Cell specific heat
will be coarser, whereas if the base size is small, the mesh will be finer.
Cfluid = Fluid specific heat
ṁ = Mass flow rate
ΔTfluid = Fluid Outlet Temperature – Fluid Inlet Temperature 5.2. Validation of heat simulation
The model shown in Fig. 6 is referred to as the “original cooling
plate”. According to Fig. 6, the cooling plate’s top has seven equal Fig. 11 shows the geometry and schematic of the CFD model using
contact areas. Each surface has an area of 0.13 m2, except for the last the geometry for validation of the heat simulation [5]. Simulation ma­
surface which has an area of 0.12 m2. Therefore, the total contact area of terial properties and boundary conditions were derived from the ge­
the cooling plate is 0.88 m2. The maximum heat dissipated by the bat­ ometry given in the reference [5].
tery is 2 kW. The boundary conditions for the simulation are as follows. The purpose of this study was to compare the final results of a CFD
Heat flux can be calculated by dividing maximum heat by total area. simulation with those from a reference study. The results from the CFD
Inlet mass flow ṁ = 0.125 kg/s, Outlet = 0 Pa, Bottom surface of simulation were in good agreement with the reference results, with a
small difference in pressure drop. The mesh elements used in the two

11
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 22. Z-Type cooling plate (base design) velocity vector plot.

Fig. 23. Optimate+ (HEEDS) best design1velocity vector plot.

simulations were different, which could account for the difference in mbar. The cooling plate surface must be enhanced so as to ensure that
pressure drop. A comparison of the final simulation result with Jarrett velocity is increased across the entire fluid domain in any new design.
and Kim [5] can be found in Fig. 12 and Table 2. This will guarantee temperature homogeneity. Batteries degrade rapidly
ΔP fluid (Pa) = Difference in pressure between the outlet and inlet of when temperatures rise above 35 ◦ C. Considering that temperatures do
a fluid domain. not behave uniformly, very few electric vehicles will be able to achieve
T avg. (K) = Average temperature of the bottom surface of the plate the 10 year expected battery life. The fluid domain of the Z-type cooling
under which heat flux is applied. plate is illustrated in Fig. 14. Fig. 15 illustrates the top surface of the
T σ (K) = The standard deviation of the temperature over the surface cooling plate. Solid ribs are outlined in white in the fluid domain. This
of the bottom of the plate. geometry was generated using StarCCM+. Z-type cooling plates have a
uniform velocity and low pressure drop within their fluid domain.
5.2.1. Original cooling plate heat simulation result
A similar case setup was used for the original cooling plate based on 6. Optimization
the above study. The model shown in Fig. 7 is referred to as the “original
cooling plate”. The same method and heat simulation were employed as Multi-objective optimization is the process of finding an optimal
for the original cooling plate model (Ref. Fig. 7). The cooling plate was design which satisfies all the objective functions and meets constraints.
subjected to 2 kW of heat loads. Fig. 13 illustrates the effect of tem­ This is a design which exists within the optimization domain but does
perature deviations in the contact area of the cooling plate. not cause any constraints. The performance of the best design should be
According to Fig. 13, Delta T (Solid) represents the difference be­ greater than that of any other feasible design. Engineering problems in
tween the maximum and minimum temperatures at the contact points the real world cannot be solved by improving only one function of the
between the cooling plate and battery module. According to Fig. 12, the system as that would affect the performance of the other functions.
maximum temperature is 35.864 ◦ C and the minimum temperature is Multiple objective studies are more close to real world performance.
27.273 ◦ C. Hence, the Delta T (solid) value for the battery module is HEEDS MDO decides how to change input parameters using intelligent
8.59 ◦ C. mathematical search strategies. It may choose a 3–4 different optimi­
According to Fig. 13, Delta T is 8.59 ◦ C, which can be reduced by zation algorithm which depends on simulation and response (results).
altering the design of the cooling plate, which has a pressure drop of 457 The steps in solving an optimization problem using HEEDS are as

12
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 24. Pareto front.

follows. 8. Result and discussion

1. Define the problem by specifying the inputs, outputs, and objectives. 8.1. Multi-objective optimization
2. Discretize the design space into a set of points, called design points.
3. Evaluate the objective functions at each design point. The thickness of the ribs is also taken into account in multi-objective
4. Select the best design point based on the objective functions. optimization. Fig. 20 illustrates the design variable that was used in the
5. Repeat steps 2–4 until the best design point is found. multi-objective study. Fig. 21 illustrates the contact area between the
cooling plate and the battery pack in this multi-objective study. The
7. Z- Type cooling plate simulation target pressure drop for this cooling plate is 500 mbar. In the Original
cooling plate model, the pressure drop reached 457 mbar, however, the
A geometric model and parameters are shown in Fig. 16 of the fluid temperature variation (Delta T solid) was 8.59 ◦ C, indicating that mass
domain. The geometric model has two types of ribs: (1) Fixed ribs, and flow was necessary for reducing temperature variations. In this opti­
(2) Moving ribs. The outlet position also changes in the fluid domain, but mization, the mass flow rate was kept variable. Multi-objective opti­
the inlet position does not change. Each rib is subject to change in width, mization has three objectives.
length and depth. The width of the rib changes from 6 to 20 mm and the In Multi-objective optimization the objective function are:
height of the rib changes from 6 to 500 mm and the depth from 0.001 to
2.5 mm. 1. Minimize: P fluid(d)
The mass flow rate is 0.125 kg/s and the heat flux is 2272.72 W/m2. 2. Minimize: T avg (d)
These simulations involve 57 variables in the rib, including 12 variables 3. Maximize: T uni (d)
related to height, 12 variables related to depth, 12 variables related to
width, 7 variables related to distance, 12 variables related to rib vertical Constrain: 0.5 mm ≤ Depth of the rib ≥2.5 mm
position, one variable related to outlet position, and one variable related
to mass flow rate. The top ribs except the first are fixed. The bottom ribs a. 6 mm ≤ Height of the rib ≥500 mm
are moving ribs as shown in Fig. 16. The distance range for moving ribs b. 6 mm ≤ Width of the rib ≥20 mm
can be found in Fig. 16. As shown in Fig. 17, Delta T is the difference
between the maximum and minimum temperatures in the contacting where, P fluid(d) = P inlet –P outlet
area. This model indicates a Delta T of 8.51 ◦ C and a temperature uni­ ∑
f Tf Af
formity of 87.37 % in the contact area of this model. The Original Tavg = ∑ (4)
fA
cooling plate model should be optimized in order to achieve a reduced
Delta T and temperature uniformity (Tuni). Fig. 18 illustrates the velocity ∑
f |∅ − ∅|Af
plot for the original cooling plate model. Fig. 19 depicts the pressure plot Tuni = 1 − ∑ (5)
2|∅| f Af
for the original cooling plate model. The plots were taken at 1 mm height
in the fluid domain.

13
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 25. Z-type cooling plate base design pressure, velocity and Delta T.

Tavg = Average temperature in the contact area of the cooling plate (HEEDS) Best Design 1 is the most efficient model among the three, with
and the battery pack the highest mass flow rate and the lowest delta P.
Tuni = Temperature uniformity in the contact area of the cooling A study comparing Delta T (Fluid), Temperature Uniformity, and
plate and the battery pack Standard Deviation of three different models is presented in Table 4. The
P fluid (d) = Pressure drop Original cooling plate model had a delta T of 4.81, while the Z-Type
Tf = Face value of Average Temperature design model had a delta T of 4.82. The Optimate Best Design 1 had a
∅ = Surface Temperature – Inlet Temperature delta T of 2.62. In the study, it was determined that the Optimate design
∅ = Surface Average of ∅ had the highest degree of temperature uniformity, 88.97 %, with a
standard deviation of 1.24. As shown in Fig. 21, the delta T, standard
_
∅f = Face value of the ∅
deviation, and uniformity of temperature were measured at the contact
Af = Area of a face
area between the cooling plate and battery pack. The following equation
The purpose of this study was to optimize an original cooling plate
is used to calculate the Deta T (fluid).
model for pressure drop, velocity, and Delta T. Optimate+ (HEEDS) was
used to improve the design, resulting in a minimum of 35 design itera­ Delta T (fluid) = Outlet Temperature–Inlet Temperature.
tions. The Original cooling plate model is shown in Fig. 25 along with
An example of a Pareto front derived from pressure drop and Delta T
pressure drop, velocity, and Delta T. When benchmarking the optimi­
(Fluid) average temperature can be seen in Fig. 24 of the multi-objective
zation algorithm, the lack of improvement in the last two designs is
study. HEED removed the higher Pareto-front. The Delta T (Fluid)
considered to be converged. Fig. 26 shows the best design in the thirty
temperature range of 1.25 to 4.82 ◦ C was used to color each design
five design iterations. In the best design, Delta T was reduced from
point. X-axis indicates the average temperature in the contact area be­
8.51 ◦ C to 6.59 ◦ C, but the pressure drop was increased from 252 mbar to
tween the cooling plate and battery pack. Y-axis represents pressure
348 mbar. The pressure drop is still less than the target of 500 mbar.
drop. There are three design regions represented by the Pareto-front: A,
Fig. 27 illustrates the delta T temperature plot for all designs. In the base
B, and C. A design in region A represents a design where the pressure
design, temperature uniformity is 87.37 %, while 88.97 % is achieved in
drop is low, but the average temperature is high. A design in region C
the second design.
represents a design where the average temperature is low, but the
An original, Z-type, and Best design 1 simulation result is presented
pressure drop is high. Design in region B represents a design with a low
in Table 3. This table compares the mass flow rate, delta P, and delta T of
average temperature as well as a low pressure drop. A velocity vector
three different models. The Original cooling plate model is used to
plot of the best design 1 is shown in Fig. 23. Fig. 22 shows a velocity
compare the two other models. In the Z-Type Design Model, efficiency is
vector plot of the Z-type cooling plate. The plots are taken at a height of
increased with respect to the Original cooling plate model. This model is
1 mm in the fluid domain. It has been demonstrated in the study that by
more efficient than the Original cooling plate model. The Optimate+

14
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 26. Best Design1.

using HEEDs, it has been possible to eliminate designs that have higher temperature throughout the batteries of 5 ◦ C. Aside from this, Delta T
pressure drops and higher average temperatures. Furthermore, this represents the maximum difference between a local hot spot and a local
study has demonstrated how HEEDs may be used to optimize the design cold spot. Additionally, the hot spot locations are small. A design that
of battery cooling plates. meets all constraints is the best design 1. Using the best design 1, Delta T
(solid) reduced from 8.59 ◦ C to 6.59 ◦ C, an improvement of 22.28 %, and
9. Conclusion Delta T (fluid) decreased from 4.81 ◦ C to 2.62 ◦ C, an improvement of
45.53 %. Also, the temperature uniformity increased from 87.37 % to
Lithium-ion batteries are widely used in the automotive industry 88.97 %. As a result of the increased temperature uniformity, the best
because they are light and have a high energy density. However, these design 1 provides more even surface temperatures, which leads to
batteries are highly sensitive to temperature. The working temperature improved battery performance and lifespan. Reduction of pressure drop
range for these batteries is 15–35 ◦ C. When the battery reaches a tem­ from 456 mbar to 348 mbar represents a 23.68 % improvement.
perature of >35 ◦ C, this is considered an extreme case. In this scenario,
the battery’s storage capacity and life will decrease. Liquid cooling has CRediT authorship contribution statement
been found to be the most efficient cooling method for batteries due to
its high efficiency, ability to heat the battery, and ability to achieve Shrikant R. Patil: Visualization, Investigation, Data curation,
temperature uniformity easily with liquid cooling. In this study, the Original draft preparation,
focus is on optimizing the battery cooling plate at the module level. With Bhaskara RaoLokavarapu: Conceptualization, Validation, Super­
regard to objective functions such as average temperature, temperature vision, Reviewing and Editing.
homogeneity, and pressure drop, the optimal design of battery cooling Hareesh Karonnin Thaliyanveedu: Methodology, Writing,
plate was determined using Optimate+ (HEEDS). The cooling system is Reviewing.
capable of providing the required cooling for the battery pack.
It is important that the temperature difference between the top and Declaration of competing interest
bottom surfaces of the batteries within the module is less than or equal
to 5 ◦ C. The cooling plate must be maintained at a temperature less than All authors here declare that there are no financial and personal
or equal to 30 ◦ C. This will ensure that the batteries are kept between 15 relationships with other people or organizations that could inappropri­
and 35 ◦ C.Fig. 27 illustrates that the average temperature of the cooling ately influence (bias) their work or state if there are no interests to
plate is between 28 and 29 ◦ C. Fig. 27 shows the contact area between declare
the cooling plate and batteries, which corresponds to the bottom surface
of the batteries. By using this logic, we are able to ensure uniform

15
S.R. Patil et al. Journal of Energy Storage 58 (2023) 106299

Fig. 27. Original cooling plate model and best design temperature plot.

Data availability [13] Zhonghao Rao, Shuangfeng Wang, A review of power battery thermal energy
management, Renew. Sust. Energ. Rev. 15 (9) (2011) 4554–4571.
[14] Frank Richter, Signe Kjelstrup, Preben J.S. Vie, Odne S. Burheim, Thermal
It is available in the manuscript conductivity and internal temperature profiles of Li-ion secondary batteries,
J. Power Sources 359 (2017) 592–600.
References [15] Battery University, Company webpage, http://batteryuniversity.com/learn/artic
le/types_of_battery_cells, 2017.
[16] Bumjun Park, Chung ho Lee, Chunyang Xia, Cheolsoo Jung, Characterization of gel
[1] Huaqiang Liu, Zhongbao Wei, Weidong He, Jiyun Zhao, Thermal issues about Li- polymer electrolyte for suppressing deterioration of cathode electrodes of Li ion
ion batteries and recent progress in battery thermal management systems: a batteries on high-rate cycling at elevated temperature, Electrochim. Acta 188
review, Energy Convers. Manag. 150 (2017) 304–330. (2016) 78–84.
[2] L.W. Jin, P.S. Lee, X.X. Kong, Y. Fan, S.K. Chou, Ultra-thin minichannel LCP for EV [17] Ziyuan Wang, Xinxi Li, Guoqing Zhang, Youfu Lv, Cong Wang, Fengqi He,
battery thermal management, Appl. Energy 113 (2014) 1786–1794. Chengzhao Yang, Chuxiong Yang, Thermal management investigation for lithium-
[3] Dongchang Pan, Sichuan Xu, Chunjing Lin, Guofeng Chang, Thermal Management ion battery module with different phase change materials, RSC Adv. 7 (68) (2017)
of Power Batteries for Electric Vehicles Using Phase Change Materials: A Review. 42909–42918.
No. (2016-01-1204). SAE Technical Paper, 2016. [18] John T. Warner, The Handbook of Lithium-Ion Battery Pack Design: Chemistry,
[4] A. Opitz, P. Badami, L. Shen, K. Vignarooban, A.M. Kannan, Can Li-ion batteries be Components, Types and Terminology, Elsevier, 2015.
the panacea for automotive applications? Renew. Sust. Energ. Rev. 68 (2017) [19] Lip Huat Saw, Yonghuang Ye, Andrew A.O. Tay, Wen Tong Chong, Seng
685–692. How Kuan, Ming Chian Yew, Computational fluid dynamic and thermal analyses of
[5] Anthony Jarrett, Il Yong Kim, Design optimization of electric vehicle battery Lithium-ion battery pack with air cooling, Appl. Energy 177 (2016) 783–792.
cooling plates for thermal performance, J. Power Sources 196 (23) (2011) [20] Xinhai Xu, Wenzheng Li, Xu. Ben, Jiang Qin, Numerical study on a water cooling
10359–10368. system for prismatic LiFePO4 batteries at abused operating conditions, Appl.
[6] Jaephil Cho, Sookyung Jeong, Youngsik Kim, Commercial and research battery Energy 250 (2019) 404–412.
technologies for electrical energy storage applications, Prog. Energy Combust. Sci. [21] Z.Y. Jiang, Z.G. Qu, Lithium–ion battery thermal management using heat pipe and
48 (2015) 84–101. phase change material during discharge–charge cycle: a comprehensive numerical
[7] Mark E. Steinke, Satish G. Kandlikar, Single-phase heat transfer enhancement study, Appl. Energy 242 (2019) 378–392.
techniques in microchannel and minichannel flows, in: Second International [22] Waseem Raza, Faizan Ali, Nadeem Raza, Yiwei Luo, Eilhann E. Kwon,
Conference on Microchannels and Minichannels, Rochester, New York, USA, June, Jianhua Yang, Sandeep Kumar, Andleeb Mehmood, Ki-Hyun Kim, Recent
17-19, 2004. advancements in supercapacitor technology, Nano Energy 52 (2018) 441–473.
[8] Keith Breene, Is solar set to take over the world? World economic forum. https:// [23] Yuan Yang, Xiaopeng Huang, Zeyuan Cao, Gang Chen, Thermally conductive
www.weforum.org/agenda/2016/03/is-solar-set-to-take-over-the-world/, 11 Mar separator with hierarchical nano/microstructures for improving thermal
2016. management of batteries, Nano Energy 22 (2016) 301–309.
[9] Sury Janarthanam, Neil Burrows, Bhaskara Rao Boddakayala, Factors Influencing [24] Lip Huat Saw, Hiew Mun Poon, Hui San Thiam, Zuansi Cai, Wen Tong Chong,
Liquid over Air Cooling of High Voltage Battery Packs in an Electrified Vehicle. Nugroho Agung Pambudi, Yeong Jin King, Novel thermal management system
(No. 2017-01-1171). SAE Technical Paper, 2017. using mist cooling for lithium-ion battery packs, Appl. Energy 223 (2018)
[10] Anthony Jarrett, Il Yong Kim, Influence of operating conditions on the optimum 146–158.
design of electric vehicle battery cooling plates, J. Power Sources 245 (2014) [25] Lip Huat Saw, Yonghuang Ye, Ming Chian Yew, Wen Tong Chong, Ming Kun Yew,
644–655. Tan Ching Ng, Computational fluid dynamics simulation on open cell aluminium
[11] Zhen Qian, Yimin Li, Zhonghao Rao, Thermal performance of lithium-ion battery foams for Li-ion battery cooling system, Appl. Energy 204 (2017) 1489–1499.
thermal management system by using mini-channel cooling, Energy Convers. [26] A. De Vita, Arpit Maheshwari Armando, Matteo Destro, Massimo Santarelli,
Manag. 126 (2016) 622–631. Massimiliana Carello, Transient thermal analysis of a lithium-ion battery packs
[12] Emma Arfa Grunditz, Torbjörn Thiringer, Performance analysis of current BEVs comparing different cooling solutions for automotive applications, Appl. Energy
based on a comprehensive review of specifications, IEEE Trans.Transp.Electrif. 2 206 (2017) 101–112.
(3) (2016) 270–289.

16

You might also like