INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. 2013; 37:13–24

Published online 23 January 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/er.1956

ACCELERATED PUBLICATION

Thermal management of lithium-ion batteries for

electric vehicles
G. Karimi{ and X. Li*,†
Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada

SUMMARY
Thermal issues associated with electric vehicle battery packs can significantly affect performance and life cycle. Fundamental
heat transfer principles and performance characteristics of commercial lithium-ion battery are used to predict the temperature
distributions in a typical battery pack under a range of discharge conditions. Various cooling strategies are implemented to
examine the relationship between battery thermal behavior and design parameters. By studying the effect of cooling conditions
and pack configuration on battery temperature, information is obtained as to how to maintain operating temperature by
designing proper battery configuration and choosing proper cooling systems. It was found that a cooling strategy based on
distributed forced convection is an efficient, cost-effective method which can provide uniform temperature and voltage
distributions within the battery pack at various discharge rates. Copyright © 2012 John Wiley & Sons, Ltd.

KEY WORDS
lithium-ion battery; thermal management; mathematical modeling; convection cooling; discharge

Correspondence
*X. Li, Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, Ontario, N2L 3G1, Canada.
†
E-mail: [email protected]
{
On leave from Shiraz University, Shiraz, I.R. Iran

Received 26 July 2011; Revised 10 November 2011; Accepted 12 November 2011

1. INTRODUCTION system is necessary to: (i) regulate the batteries to operate

in the desired temperature range; and (ii) to reduce uneven
Development of advanced rechargeable batteries for electric temperature distribution. In order to quantify the impact of
vehicles has been the subject of intensive study worldwide temperature on the performance of a battery and thus
due to the beneficial impact of battery-powered electric on the electric vehicle, temperature-dependent battery
vehicles on energy consumption and atmospheric air quality performance models are needed. These data are then used
[1]. The lithium-ion (Li-ion) battery has been considered to to devise battery thermal management system.
be one of the most promising long-term advanced battery Mathematical simulation of heat transport within large
technologies by the U.S. Advanced Battery Consortium [2]. batteries is an effective tool to obtain knowledge about
Compared to other rechargeable batteries, Li-ion batteries whether excessive heat generated during the battery
exhibit better characteristics in terms of capacity, power and discharge process can be removed, and how operating
energy densities, charge retention, life cycles and competitive temperature can be controlled. There has been few research
cost. Presently, however, Li-ion batteries are only commer- on thermal management of Li-ion battery packs in recent
cially available in small sizes. Accordingly, large numbers years. Chen and Evans [3] were first to develop a two-
of cells have to be assembled in series and/or parallel config- dimensional model to study the effect of various cell
urations forming battery packs to achieve the desired battery components, stack size and cooling conditions on the
powers for vehicular propulsion systems. performance of Li-polymer electrolyte batteries under
The performance of an electric vehicle depends strongly different discharge rates. Based on the temperature profiles
on its battery pack performance. The Li-ion batteries obtained, they provided useful insights such as how to
generate significant amount of heat during high power maintain operating temperature by designing proper cell
discharge in some cases shortening the battery life or stacks and choosing proper cooling/insulating systems
posing a safety hazard. This presents the challenge of from heat transfer point of view. Al-Hallaj et al. [4] studied
making highly efficient, highly reliable battery packs for the thermal behavior of commercial cylindrical and
use in electric vehicles. In order to achieve the optimum prismatic Li-ion cells using electrochemical-calorimetric
performance from a battery pack, a thermal management methods. They measured the cell heat generation during

Copyright © 2012 John Wiley & Sons, Ltd. 13

 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

the cell discharge and heat consumption during the cell developed by Forgeza et al. [13]. Current pulses of different
charge with significant dependence on the state of charge magnitudes are applied to the battery to determine heat trans-
(SOC) at the end of discharge due to concentration polari- fer coefficients and heat capacity experimentally. Also, the
zation. Maleki and Shamsuri [5] evaluated numerically battery thermal resistance was determined from steady-state
the thermal performance of a notebook computer Li-ion temperature measurements. Their model is capable of simu-
battery pack under various operating conditions. They lating the internal temperature directly from the measured
found that the battery temperature rise during charge is current and voltage of the battery. Smith et al. [14] developed
dominated by the power dissipation from the control a coupled thermal/electrical model of a commercial 18-650-
electronics and during discharge by the heat dissipation size Li-ion cell and a module with 16 cells in parallel. Cell
from Li-ion cells. Mills and Al-Hallaj [6] designed and electrical and thermal responses are modeled using equivalent
simulated a passive thermal management system with a electrical and thermal circuits which are then integrated into a
graphite impregnated phase change material (PCM) for a module-level model. The module-level model is validated
Li-ion laptop battery pack. The heat generation rate for with experiment and used in a parametric study to assess the
a commercial 18650, 2.2 Ah Li-ion battery was experimen- battery thermal safety margin.
tally measured under various SOC conditions. Kizilel The goal of a battery thermal management system is to
et al. [7] demonstrated the advantage of using PCM ensure that the battery operates at an optimum average
thermal management systems over conventional active temperature with only small temperature and voltage
cooling systems for Li-ion battery packs. They showed that variations among the cells, modules and within the pack
high energy packs with PCM are discharged safely at high as identified by the battery manufacturer. In the present
currents, and degrading rate of capacity of the Li-ion packs work, further studies of the effect of thermal management
is lowered by half. They also reported that energy density system design parameters on battery thermal behavior
of the battery pack would increase with the application and performance have been conducted for planar battery
of PCM. Later, Sabbah et al. [8] and Kizilel et al. [9] packs used in electric vehicles. To evaluate battery pack
compared the effectiveness of passive cooling by PCMs design and provide solutions for battery thermal issues,
with that of active (forced air) cooling for thermal manage- we have used heat transfer principles to obtain fundamen-
ment of high power Li-ion battery packs. Numerical tal ideas concerning the optimization of cell structure from
simulations were performed at different discharge rates, the heat transfer point of view. Topics such as active
operating temperatures and ambient temperatures and cooling versus passive cooling; liquid cooling versus air
compared with the experimental results. The results cooling and application of PCMs are examined.
showed that at high discharge rates and/or at high operat-
ing or ambient temperatures, air cooling is not a proper
thermal management system to keep the temperature of 2. MODEL DEVELOPMENT
the cell in the desirable operating range without expending
significant fan power. Li-ion batteries are fabricated in many different shapes and
Inui et al. [10] developed two- and three-dimensional configurations, for example, cylindrical, coin, prismatic
simulation codes of the transient response of the tempera- (rectangular), and thin and flat. Although all four battery
ture distribution in the cylindrical and prismatic Li-ion configurations are suited for portable and low power
battery during a discharge cycle. The numerical results demand electronic applications, the prismatic and flat type
for the cylindrical battery were in good agreement with batteries seem to be the preferred power source for HE,
the experimental data. Their results indicated that battery plug in HE (PHE) or plug in electric vehicular applications.
with the laminated cross section has a remarkable effect In this work, combinations of the thin-film flat type battery
on the suppression of the temperature rise in comparison are considered for the analysis.
with the battery with square cross section. Fang et al. [11] As shown in Figure 1, an electric vehicle battery is com-
used an electrochemical-thermal-coupled model to predict posed of one or more multi-cell packs, and a battery pack is
performance of a Li-ion cell and its individual electrodes composed of many battery units connected in series and/or
at various operating temperatures. They validated the in parallel. The number of unit cells comprising a battery
model against the experimental data for constant current unit and the number of battery units within a pack is
and pulsing conditions, characteristic of hybrid electric determined by the desired battery capacity and voltage.
(HE) vehicle applications. The prediction of individual Such design is considered to be suitable for traction battery
electrode potential compared with three-electrode cell applications and exhibits high specific energy.
experimental data with good agreement. Duan and As illustrated in Figure 1, a typical Li-ion cell consists
Naterer [12] studied experimentally thermal management of a negative electrode formed from a thin layer of
of battery modules with PCMs. They examined the powdered graphite, or certain other carbons, mounted on
effectiveness of PCM thermal management under vari- a copper foil and a positive electrode composed of a thin
able simulated battery discharge conditions and variable layer of powdered metal oxide (e.g. LiCoO2) mounted on
ambient temperatures, as well as the effects of buoyancy aluminum foil. The two electrodes are separated by a
during PCM melting. A lumped-parameter thermal model porous plastic film soaked typically in LiPF6 dissolved in
of a cylindrical LiFePO4/graphite Li-ion battery is a mixture of organic solvents such as ethylene carbonate,

14 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Thermal management of Li-ion battery packs G. Karimi and X. Li

Figure 1. Schematic of a Li-ion battery and details of its components.

ethyl methyl carbonate, or diethyl carbonate [15]. The dEoc

stacked cells are generally held together by pressure from q ¼ I ðEoc  V Þ  IT (4)
dT
the battery container.
During the charge/discharge process, lithium ions are
inserted or extracted from the interstitial space between where q is the heat generation rate, I, is current (I > 0 for
atomic layers within the active materials. The chemical discharge and I < 0 for charge), Eoc is the equilibrium cell
reactions are expressed as voltage or open-circuit potential, V is the cell voltage, T is
temperature and dEoc/dT is the temperature coefficient. The
Positive reaction CoO2 þ Liþ þ e ⇌LiCoO2 (1) first term, I(Eoc  V), is heat generation resulting from
ohmic and other internal resistances present in the cell
Negative reaction LiC6 ⇌C6 þ Liþ þ e (2) and the second term,  IT[dEoc/dT)], is heat generated
or consumed because of the reversible entropy change
resulting from cell electrochemical reactions.
Total reaction CoO2 þ LiC6 ⇌LiCoO2 þ C6 (3) Different kinds of Li-ion batteries have different values
of the internal resistances and entropy changes. A thermal
In these chemical formulas, the reactions proceed from management strategy requires that these data be deter-
the left side to the right side during the discharge and in mined accurately. Since the internal resistances depend
the reverse direction during the charge cycle. on both the temperature and SOC of the battery, it should
be measured under various battery operating conditions.
2.1. Heat generation To the knowledge of the authors, the equivalent internal
resistance for the flat type batteries used in the electric
The magnitude of the overall heat generation rate from a vehicle is proprietary and not available in the open litera-
battery pack under load dictates the size and design of ture. However, this parameter has been measured for the
the cooling system. In a battery cell, heat is generated by commercially available small size Li-ion batteries [10,13]
(i) entropy change from electrochemical reactions and (ii) which is structurally very similar to those of flat batteries.
Joule’s effect (or ohmic heating) caused by current transfer In the present work, the experimentally measured internal
across internal resistances. The heat generation rate in a resistance data for the cylindrical SONY-US18650 [10]
cell can be calculated from [3]: as a function of SOC and temperature were used for the

Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd. 15
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

analysis. These experimental data are curve fitted for q_ ¼ f ðSOC; i; T Þ (10)
numerical implementation and are as follows:

Ri ¼ 2:258  106 SOC0:3952 T ¼ 20 C which has been determined experimentally [10]. Equation 9
Ri ¼ 1:857  106 SOC0:2787 T ¼ 30 C (5) is subject to the initial and boundary conditions. It is
Ri ¼ 1:659  106 SOC0:1692 T ¼ 40 C assumed that all of the battery components are at an initial
temperature, Ti
where Ri is the battery internal equivalent resistance (Ω  m3).
In addition, the entropy change, ΔS, obtained by measuring T ¼ Ti at t ¼ 0; all x; y (11)
the dependence of the battery open circuit voltage on the
temperature at different SOCs according to the theory of It is further assumed that active/passive cooling media
thermodynamics [10]: are applied at strategic locations in the battery pack to
manage the temperature distribution. For a typical battery
pack illustrated in Figure 1, the following boundary
dEoc
ΔS ¼ F (6) conditions apply
dT


@T
The experimental results indicated that over a temperature at x ¼ L1 and y k ¼ h1 ½T  Tb1 ðyÞ
range from 20 to 40 ∘C, ΔS is almost independent of the tem-
@x i (12)
@T
perature and its variations with SOC follows approximately at x ¼ L2 and y k ¼ h2 ½T  Tb2 ðyÞ
the following set of equations: @x i

Similar equations can be written to account for exter-

ΔS ¼ 99:88 SOC  76:67 0⩽SOC⩽0:77
nal heating and/or cooling effects due to the exposure of
ΔS ¼ 30 0:77 < SOC⩽87 (7)
the battery pack to the ambient via the battery pack con-
ΔS ¼ 20 0:87 < SOC⩽1
tainer. Overall heat transfer coefficients, Un, Us, Ue and
Uw are defined to represent equivalent thermal resistances
Considering these, Equation 4 can be modified, and the (combined conductive and convective) in the north,
rate of internal heat generation in power generation region south, east and west sides, respectively, as illustrated in
of the Li-ion battery during discharge can be expressed as: Figure 1. In Equation 12, L1 and L2 denote the spatial
locations of the thermal management media which in gen-
i eral could be flowing air or a dielectric liquid, or a sta-
q_ ¼ Ri i2  TΔS (8)  
nF tionary PCM. The terms k @T @x i in Equation 12
represent the conductive heat flux on the cell surface ex-
where q_ is the rate of internal heat generation per unit volume, posed to the cooling media. h1 and h2 denote the local
Ri is the internal equivalent resistance, i is the discharge cur- heat transfer coefficients, and Tb1 and Tb2 are the
rent per unit volume and n is the number of electrons partic- corresponding local bulk temperatures for medium 1
ipating in the reaction (n = 1). and 2, respectively. To account for the temperature var-
iations along the cooling media, the set of equations in
2.2. Temperature distribution Equation 12 needs to be coupled with the energy equa-
tion for the cooling media. Figure 2 and Equation 13 show
Because of the thin-film cell design, the greatly differing how the energy equations are coupled:
thermal properties of different layers may have significant
influence on heat transfer behavior within a battery. In   ΔTb
the present work, a transient two-dimensional thermal _
hWΔy Tw;l þ Tw;r  2Tb  mcΔTb ¼ rAΔyc (13)
Δt
model for a battery pack with heterogeneous thermal-
physical properties of different layers is developed. The where Tw, l and Tw, r are the wall surface temperatures on the
model can be used to assess the effectiveness of various left and right sides, respectively, W is the channel depth and
thermal management methods for optimum operation of A is the channel cross-sectional area. The energy equations
the battery pack. for cooling channels are subject to the following boundary
The temperature distribution in each layer of the cell conditions:
unit in the battery pack is governed by
at y ¼ 0 T ¼ Tb;in
R  
@ 2 T @ 2 T q_ 1 @T at y ¼ H _ b;in þ
mcT
H

hW Tw;l þ Tw;r  2Tb dy ¼ mcT
_ b;out
þ þ ¼ (9) 0
@x2 @y2 k a @t (14)

where x and y represent the spatial directions, k is the The average heat transfer coefficient, h , is calculated
thermal conductivity, a is the thermal diffusivity, and based on fully developed laminar flow in channels with

16 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Thermal management of Li-ion battery packs G. Karimi and X. Li

Figure 3. Schematic of the battery pack with 20 battery units

and two surrounding cooling ducts.

battery pack configuration, it varies between 72 V (fully

charged) and a lower value depending on the SOC. Tables I
Figure 2. Various modes of heat transfer between battery units and II list the system parameters, thicknesses of various
and a flowing cooling medium. layers forming a cell unit and the associated thermo-
physical properties. According to these data, a cell unit is
constant heat flux boundary conditions and is a strong func- 730 mm thick. Thermal management (cooling) is provided
tion of the channel hydraulic diameter [16]. by two cooling ducts mounted at both ends of the battery
pack.
All simulations start with a fully charged battery
2.3. Numerical solution (SOC0 = 1 and C0 = 20 Ah). Electric currents at different
The whole computational domain (including the battery C rates are drawn from the battery and the transient
container and thermal management ducts) are divided into temperature and voltage distributions in the battery pack
segments of appropriate sizes, and the governing equations are examined. The instantaneous SOC of the battery is
are discretized for various regions. Different mesh sizes are
it
chosen for various cell layers depending on their thermal SOC ¼ 1  (15)
diffusivity values and the expected temperature gradients. c0
The implicit alternating-direction technique is applied to
cast Equations 9 and 13 along with the boundary condi- and the individual cell voltages can be calculated from
tions, into finite difference forms which are then solved
Z E t1þΔt
by the Thomas algorithm.
dE ¼ Ri i (16)
E t1

3. RESULTS AND DISCUSSION Depending on the location of the cooling media, local
temperatures may be different from one battery unit (or cell
The schematic of a typical battery pack used in an electric unit) to another battery unit during discharge; hence, it
vehicle traction system is illustrated in Figure 3. The bat- is expected that the internal equivalent resistances be
tery pack considered for the present thermal management different as well. Equation 16 shows how cell voltage
analysis consists of a total of 20 battery units connected may vary across the battery units/pack.
in series. Each battery unit considered to be 16 cm wide Lithium-ion battery cells have demonstrated excellent
and 23 cm tall and comprised of 10 cell units connected performance when operated at ambient temperature condi-
in parallel. Each cell unit consisted of seven separate layers tions. The adverse effect of temperature excursions on
as illustrated in Figure 1. For parallel connection, each cell Li-ion cell performance is reported when the cell is
unit needs to be insulated from its adjacent cell units with a operated at temperatures below 0  C or beyond 65  C
separator. For a battery capacity of 20 Ah, the current [4,17]. Despite the relatively wide operating temperature
density is estimated to be 5.4 mA/cm2. The overall magni- range (0–65  C) for battery cells, a non-uniform tempera-
tude of the battery voltage depends on the arrangement of ture distribution in a battery pack will lead to non-uniform
all battery packs in the system; however, for the present voltage distribution in the pack which adversely affect the

Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd. 17
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

Table I. System parameters.

Battery:
Number of battery units in the pack 20
Number of cell units per battery unit 10
Cell surface area [cm2] 368
Electric capacity of each battery unit [Ah] 20
Open-circuit voltage of each cell [V] 3.6
Cut-off state of charge [%] 20
Initial pack temperature [ C] 20
Thickness of cell layers [mm] [19]:
- Graphite 120
- Cu 20
- Graphite 120
- Electrolyte 40
- LiCoO2 180
- Al 20
- LiCoO2 180
- Separator 50
Total thickness of a unit cell [mm] 730
Cooling system based on air and silicon oil:
Ambient temperature [ C ] 20
Inlet temperature of cooling media [ C ] 20
Inlet fluid velocity [m/s] 0.01-1
Heat transfer coefficient for the cooling media [W/(m2   C)] 7-250
Overall heat transfer coefficients for the battery container [W/(m2   C)] 2, 5

Table II. Thermal-physical properties for various cell components used in the Li-ion cell [19].

Property Cu Graphite Electrolyte LiCoO2 Al Separator


k [W/(m  C)] 398 1.04 0.59 4 237 0.35
r [kg/m3] 8930 1347 1223 2700 2710 1400
c [J/(kg   C)] 386 1437 1375 715 902 1551

performance durability. Therefore, the main objective of However, the concluding remarks are still valid for battery
the present thermal management system is (i) to keep the charging. In addition, to make it short, henceforth, the term
battery cell temperatures within an acceptable range for battery will be used for Li-ion battery.
which heat generation data are available (e.g. 20–40 C),
and (ii) to ensure that a relatively uniform voltage distribu- 3.1. Natural convection cooling
tion prevails within the pack.
The thermal management of the battery pack can be Figure 4 shows how temperature distributions develop in a
achieved through air or liquid cooling, or using a thermal battery pack during discharge. The discharge rate was set
storage or PCM. The thermal management system may at 2 C meaning that it would take half an hour for the
be passive (i.e. only the ambient environment is used) or battery to be fully discharged. Natural convection was
active (i.e. a built-in sink/source) provides cooling/heating) considered in the cooling ducts. The temperature data are
or a combination approach. The ultimate goal is to find a plotted for the middle of the battery pack at y = 11.5 cm.
cooling/heating system which results in small temperature The simulation starts for a fully charged battery (SOC = 1)
variations within the individual cells and battery units, and ends after 24 min when the battery SOC reaches 20%.
and small voltage variations among various battery units. As seen from this figure, the battery temperature increases
At the same time, the cooling/heating system must be reli- with time, and a maximum temperature of about 39∘C
able, compact and cost effective. If necessary, the cooling prevails in the middle battery units located at the farthest
ducts can be divided into a number of smaller channels distance from the cooling media. On the other hand, a very
to enhance heat transfer coefficients at the boundaries. small temperature drop can be observed in the corner
In the following sections, numerical results for various battery units due to heat exchange with the adjacent
thermal management strategies are presented. Since, the cooling air.
objective is to examine the thermal behavior of the Li-ion An overall picture of temperature distribution in the
battery pack during discharge, when heat is generated battery pack at the end of discharge (24 min) is shown
within the battery, only cooling strategies are considered. in Figure 5. As seen from this picture, maximum

18 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Thermal management of Li-ion battery packs G. Karimi and X. Li

40
t = 24 min temperature region (about 41 C) in the upper region
is due to the smaller overall heat transfer coefficient
considered there (Un = 2 W/(m2   C)) and higher air
35 temperature at the outlet of the cooling ducts.
Figure 6 shows how various parameters (averaged) vary

Cooling duct
Cooling duct

from the battery-to-battery units within the pack at the end

of the discharge process. Figures 6a and b display the
T [ °C]

30 variations of the internal equivalent resistance and the

corresponding internal heat generation rate for all the bat-
tery units, respectively. As seen from these figures, the
t = 10 min lower temperature at both ends of the battery pack causes
25
both the internal resistance and the resulting heat genera-
t = 4 min tion rate to increase at the end of the discharge. However,
the rise in the heat generation rate in the corner batteries
t = 0 min is not high enough to even compensate the small heat loss
20
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
at these locations; as a result, lower temperatures are
x [m] observed in the battery units in the neighborhood of the
cooling ducts as shown in Figure 6c. Figure 6d shows that
Figure 4. Transient temperature distributions at the midplane of a uniform voltage distribution prevails in the battery pack
the battery pack (y = 11.5 cm) while discharging at a rate of 2 C at the end of discharge. The associated voltage drop due
from SOC = 1 to SOC = 0.2 [Cooling media: air, natural convec- to larger internal resistances is less than 0.16%, which is
tion, Vair=0.01 m/s, h = 7 W/(m2   C)]. negligible.
Figure 7 shows how air temperature varies along the
cooling ducts surrounding the battery pack at the end
temperatures prevail in the upper region of the pack where of discharge. The air temperature rises quickly and
the maximum thermal resistance exists. Numerical results approaches the contacting solid temperatures in the
indicate that the presence of air with low velocity and entrance region. In fact, the cooling air at such a low
low heat transfer coefficient is not efficient in removing flow rate (natural convection at 0.01 m/s) behaves like an
heat from the pack even in the battery units very close to insulator in almost two third of the cooling ducts. As
the cooling ducts. The maximum temperature drop in the expected, variations of air temperature in both end ducts
corner cells is less than 1  C. The existence of the high are identical.
The presented results have shown that the amount of
heat removal by naturally convecting air is very small.
As a result, higher temperatures are uniformly developed
across the battery pack at all SOCs (or times). Although
higher temperatures generally tend to improve the battery
performance because of increased electrochemical reaction
rates (lower internal resistances), the battery lifetime
decreases because elevated temperatures increase corro-
sion. In fact there exists an optimum operating condition
for a battery pack which is normally defined based on (i)
a predefined temperature range, (ii) temperature uniformity
within and among the battery units, and (iii) the electrical
balance among battery units in the pack. The statements
(ii) and (iii) are interrelated meaning that if individual cells
and/or battery units in a pack are at different temperatures,
each battery unit will develop different voltage during the
discharge and/or charge cycle. Therefore, after several
cycles, battery units in the pack will become unbalanced,
degrading the pack’s performance [18]. In the following
sections, the application of forced convection cooling is
examined for thermal management of the battery pack.

3.2. Forced convection cooling

Figure 5. An overall picture of temperature distribution in the Figure 8 compares the heat removal ability of active air and
battery pack at the end of discharge for a rate of 2 C [Cooling liquid cooling with that of natural convection air cooling.
media: air, natural convection, Vair=0.01 m/s, h = 7 W/(m2   C)]. To compare the results, the cooling duct on the left-hand

Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd. 19
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

Figure 6. Distributions of the averaged (a) internal equivalent resistance (b) heat generation rate (c) temperature, and (d) voltage from
the battery-to-battery in the pack at the end of discharge for a rate of 2 C [Cooling media: air, natural convection, Tin = 20  C, Vair=0.01
m/s, h = 7 W/(m2   C)].

36 side of the battery pack is fed with naturally convecting

air, while forced convection air and liquid cooling are
considered in the right-hand side cooling duct, and the
temperature distributions at the midplane of the battery
32
pack (y = 11.5 cm) are plotted at the end of the discharge
process. A velocity of 1 m/s is considered for both forced
air and liquid flows at different heat transfer coefficients,
h = 15, 150 and 250 W/(m2  ∘C), respectively. The high
T [°C]

28
heat transfer coefficient for air (and liquid) can be achieved
by dividing the cooling duct to smaller mini channels as
mentioned earlier. Dielectric silicone oil (r = 920 kg/m3,
24
c = 1370 J/(kg  ∘C)) is considered as the liquid coolant.
As seen from this figure, larger amounts of heat can be
removed as the heat transfer coefficient increases. In
Left duct
20 Right duct addition to the higher heat transfer coefficients, the large
heat capacity (rc) associated with the silicone oil enhances
0.05 0.1 0.15 0.2 0.25
its heat removal capability. As a result, significant temper-
y [m]
ature drop can be observed in the battery units near the
Figure 7. Variation of air temperature along the cooling duct right-hand side cooling medium. Included in this figure is
at the end of discharge for a rate of 2 C [Cooling media: air, the resulting temperature profile due to the presence of a
natural convection, Tin = 20  C, Vair=0.01 m/s, h = 7 W/ PCM. The PCM is simulated with a flowing fluid with very
(m2   C)]. large heat capacity (rc ! 1). Numerical results in Figure 8

20 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Thermal management of Li-ion battery packs G. Karimi and X. Li

40
clearly show that even the most efficient cooling medium
38 (e.g. PCM) is unable to remove heat from the middle
battery units. This is due to the relatively small Fourier
36 and Biot numbers associated with the constructing
Maximum  
34
penetration depth materials of the individual batteries (Fo ¼ 0:82 and Bi ¼
4
7:3  10 h).
32 Figures 9a to d show how battery voltages vary in the
T [°C]

battery pack for various cooling media. The voltage

30
distribution for the natural convection boundary condition
28 is also included in the figure for comparison. It is clear
Free conv. (air) from these figures that as the cooling capacity increases,
26 Forced conv. (air, low h) the internal equivalent resistance increases, and the battery
Forced conv. (air, high h)
Forced conv. (liquid, high h) units near the cooling media provide lower voltages. A
24
Phase change material 10% deviation in voltage can be observed when a PCM
22 is used for cooling. The lack of balance in the individual
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
battery voltages and the SOC will have adverse effect on
Battery number
the battery durability as pointed out earlier.
Figure 8. Effect of active air cooling, active liquid cooling,
and application of phase change material on the temperature 3.3. Performance curves
distributions in the battery pack at the end of discharge for a
rate of 2 C [Cooling media: air, silicon oil, phase change Figure 10 displays time variations of battery voltage at
material]. various rates of discharge, C. In this figure, the time

3.1

3
V [V]

2.9

2.8
(a) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.1

3
V [V]

2.9

2.8
(b) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.1

3
V [V]

2.9

2.8
(c) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.1

3
V [V]

2.9

2.8
(d) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Battery number

Figure 9. Battery-to-battery voltage distributions for various cooling characteristics at the end of the discharge for a rate of 2 C. (a) Free
convection (air); (b) forced convection (air); (c) forced convection (silicone oil); and (d) phase change material.

Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd. 21
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

passage is expressed by the variation of SOC because

it decreases linearly at constant current discharge as indi-
cated in Equation 15. From this figure, a sudden drop in
the voltage can be observed at the start of the battery
discharge for all C rates. However, the magnitude of the
voltage drop is linearly dependent on the rate of discharge
or current according to the Ohm’s law. As time passes by,
the battery voltage decreases, and the higher the rate of
discharge, the larger the rate of voltage drop. The cell is
considered to have reached the end of discharge when
its SOC drops to 20%. The total power output can be
calculated by integrating the surface area underneath the
curves with the C rate taken into account.

3.4. Proposed cooling system

There are several approaches to designing a battery thermal

management system. Each approach depends on the avail-
ability of information (e.g. power demand, drive cycle, cell
shape, materials of construction, battery pack configura-
tion, etc.), desire level of sophistication and space and
weight limitations. Based on the numerical results just Figure 11. An overall picture of temperature distribution in the
presented, it is found that although application of naturally battery pack with distributed cooling ducts at the end of
convecting air as a cooling medium is very simple, it is not discharge for a rate of 2 C [Cooling media: air, forced convection,
as effective as air or liquid forced convection. The rate of 4 mm  4 mm channels, Vair=1 m/s, h = 50 W/(m2   C)].
heat transfer between the battery surfaces and the cooling
fluid depends on the thermal conductivity, viscosity, den- generally have lower viscosity and higher thermal conduc-
sity and velocity of the fluid. For the same flow rate, the tivity than most oils, resulting in higher heat transfer
heat transfer rate for most practical direct-contact liquids coefficients. However, because the heat must be conducted
such as silicone oil is much higher than that of air because through walls of the jacket/container or fins, the effective-
of the higher fluid heat capacity and thermal conductivity. ness for indirect-contact cooling decreases [18].
However, because of silicone oil’s higher viscosity and A distributed air cooling system is proposed in this
associated higher pumping power, a lower flow rate is study. In such a system, the original end cooling ducts
usually used, making the oil heat transfer coefficient only are substituted with several cooling ducts to thermally
1.5 to 3 times higher than that with air. Indirect-contact
heat transfer liquids such as water or water/glycol solutions 40

t = 24 min
70 35
0.2C
0.5C

1C
65
T [°C]

30

2C
V [V]

t = 10 min
60
25

t = 4 min
55
4C t = 0 min
20
0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
50
x [m]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
1-SOC Figure 12. Transient temperature distributions at the midplane
of the battery pack (y = 11.5 cm) while discharging at a rate of
Figure 10. Time variation of the battery voltage at various C 2 C from SOC = 1 to SOC = 0.2 [Cooling media: air, forced
rates [Cooling media: air, forced convection, Vair=1 m/s and convection, 4 mm  4 mm channels, Vair=0.01 m/s, h = 50 W/
h = 15 W/(m2   C) in both cooling ducts]. (m2   C)].

22 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Thermal management of Li-ion battery packs G. Karimi and X. Li

3.1

3.0

V [V]
2.9

2.8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Battery number

Figure 13. Averaged battery-to-battery voltage distribution in the battery pack with distributed cooling ducts [Cooling media: air,
forced convection, 4 mm  4 mm channels, Vair=0.01 m/s, h = 50 W/(m2  ∘C)].

manage (cool) individual batteries. The cooling ducts need performance characteristics of commercial Li-ion battery
to be smaller in size (smaller widths) to ensure the are used to investigate the thermal and electrical perfor-
compactness of the battery pack. The smaller size would mance of a battery pack at various rates of discharge.
also enhance heat transfer coefficient. The air flow through Different cooling strategies are considered to examine the
these mini channels can be provided by the moving battery pack performance. The cooling strategies include
vehicle. To this end, the battery pack is divided into ten natural convection, forced convection and application of
sections, each section containing two battery units. The a PCM. Numerical results indicated that a battery pack
cooling is provided by forced convection air through with naturally convecting air can reach unacceptably high
square channels (cross-section area: 4 mm  4 mm) temperature levels. Forced convection cooling (air and
mounted between every two battery units. The small liquid) and application of PCMs at the battery pack
channel sizes ensure high heat transfer coefficients boundaries would lead to unbalanced temperature/voltage
(50 W/(m2   C) considered here). The extra cooling distributions in the pack which deteriorate battery durabil-
channels will increase the volume of the battery pack ity. Numerical analysis indicated that a distributed cooling
by 14%. (ventilation) can reduce the temperature non-uniformity
An overall picture of the temperature distribution across between battery units and improve substantially the ther-
the battery pack with the proposed cooling system is illus- mal performance of the battery pack.
trated in Figure 11 at the end of discharge period for a rate
of 2 C. As seen from this figure, this configuration provides
relatively good temperature uniformity, and the maximum NOMENCLATURE
temperature is about 4  C lower than that of the battery
pack with two cooling ducts. A = cross-sectional area [m2]

Figure 12 shows temperature distributions at the midplane Bi = averaged Biot number [hL/k]
(y = 11.5 cm) of the fully charged battery pack at a discharge c = charge capacity per unit volume [Ah/m3]
rate of 2 C for different times. As seen from this figure, c = heat capacity [J/(kg   C)]
the generated heat within the battery units are removed C = charge capacity [Ah]
uniformly by the distributed cooling ducts. The intermediate E = cell voltage [V]
plunges in temperatures represent the local air temperature f = arbitrary function
in the cooling ducts. This figure clearly shows that the F = Faraday number; 96485 [C/mol]

temperature distribution in the individual battery units are Fo = averaged Fourier number [at/L2]
uniform, with a maximum dispersion of about 0.2 ∘C, and h = heat transfer coefficient [W/(m2   C)]
the maximum observed temperature is 36 ∘C. The uniformity
h = average heat transfer coefficient [W/(m2   C)]
in temperature ensures uniformity in the internal equivalent H = height of the battery pack [m]
resistances and the voltage distribution as depicted in i = discharge current per unit volume [A/m3]
Figure 13 which satisfies the objectives. Numerical results in- I = discharge current [A]
dicated that the maximum voltage deviation from battery-to- k = thermal conductivity [W/(m   C)]
battery units is about 1.5%. L = distance/characteristic length [m]
m_ = mass flow rate [kg/s]
n = number of electrons participating in the reaction
4. CONCLUSIONS q = heat transfer/generation rate [W]
q_ = heat generation rate per unit volume [W/m3]
The performance of an electric vehicle is affected by R = internal equivalent resistance [Ω  m3]
the operating temperature and the degree of temperature S = entropy [J/molK]
gradient in its battery pack. In this work, thermal analysis SOC = state of charge
of a Li-ion battery pack is carried out to examine the t = time [s, min]
relationship between battery thermal behavior and design T = temperature [ C ]
parameters. Fundamental heat transfer principles and U = overall heat transfer coefficient [W/(m2   C)]

Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd. 23
DOI: 10.1002/er
 1099114x, 2013, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/er.1956 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [06/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. Karimi and X. Li Thermal management of Li-ion battery packs

W = depth of cooling ducts [m] and uniformity in high-energy Li-ion battery packs at
x = spatial coordinate [m] high current and ambient temperature. Journal of
y = spatial coordinate [m] Power Sources 2008; 183:370–375.
8. Sabbah R, Kizilel R, Selman JR, Al-Hallaj S. Active
(air-cooled) vs. passive (phase change material)
Greek letters thermal management of high power lithium-ion packs:
Limitation of temperature rise and uniformity of
a = thermal diffusivity [m2/s]
temperature distribution. Journal of Power Sources
Δ = difference
r = density [kg/m3] 2008; 182:630–638.
9. Kizilel R, Sabbah R, Selman JR, Al-Hallaj S. An
alternative cooling system to enhance the safety of
Subscripts/superscripts Li-ion battery packs. Journal of Power Sources 2009;
194:1105–1112.
0 = initial condition 10. Inui Y, Kobayashi Y, Watanabe Y, Watase Y, Kitamura
b = bulk Y. Simulation of temperature distribution in cylindrical
e = east side and prismatic lithium ion secondary batteries. Energy
in = inlet Conversion and Management 2007; 48:2103–2109.
i = initial/internal/layer number 11. Fang W, Kwon OJ, Wang CY. Electrochemical
l = left side
thermal modeling of automotive Li-ion batteries and
n = north side
experimental validation using a three-electrode cell.
oc = open circuit
out = outlet International Journal of Energy Research 2010;
r = right side 34:107–115.
s = south side 12. Duan X, Naterer GF. Heat transfer in phase change
w = wall/west side materials for thermal management of electric vehicle
battery modules. International Journal of Heat and
Mass Transfer 2010; 53:5176–5182.
REFERENCES 13. Forgez C, Do DV, Friedrich G, Morcrette M,
Delacourt C. Thermal modeling of a cylindrical
1. Campanari S, Manzolini G, de la Iglesia FG. Energy LiFePO4/graphite lithium-ion battery. Journal of
analysis of electric vehicles using batteries or fuel Power Sources 2010; 195:2961–2968.
cells through well-to-wheel driving cycle simulations. 14. Smith K, Kim GH, Darcy E, Pesaran A. Thermal/
Journal of Power Sources 2009; 186:464–477. electrical modeling for abuse-tolerant design of lithium
2. Pesaran AA, Market T, Tataria H, Howell D. Battery ion modules. International Journal of Energy Research
Requirements for Plug-in Hybrid Electric Vehicles: Special Issue: Electrical Energy Storage for Future
Analysis and Rationale. Presented at the 23rd Transportation and Renewable Energy. 2010; 34(2):
Int. Electric Vehicle, Symposium and Exposition 204–215.
(EVS-23), Anaheim, California, December 2007. 15. Wakihara M. Recent developments in lithium ion
3. Chen Y, Evans JW. Thermal analysis of lithium batteries. Material Science and Engineering 2001;
polymer electrolyte batteries by a two dimensional 33:109–134.
model -Thermal behaviour and design optimization. 16. Incropera FP, Dewitt DP, Bergman TL, Lavine AS.
Electrochimica Acta 1994; 39:517–526. Fundamentals of heat and mass transfer (6th edn). John
4. Al Hallaj S, Prakash J, Selman JR. Characterization of Wiley & Sons: New York, NY, 2007.
commercial Li-ion batteries using electrochemical 17. Sony Type US18650 Lithium-Ion Battery Manual,
calorimetric measurements. Journal of Power Sources Sony Co., 1993.
2000; 87:186–194. 18. Pesaran AA, Vlahinos A, Burch SD. Thermal
5. Maleki A, Shamsuri AK. Thermal analysis and performance of EV and HEV battery modules and packs.
modeling of a notebook computer battery. Journal of Proceedings of the 14th Int. Electric Vehicle Symposium,
Power Sources 2003; 115:131–136. Orlando, Florida, 15–17 December, 1997; 15–17.
6. Mills A, Al-Hallaj S. Simulation of passive thermal 19. Cheng L, Ke C, Fengchun S, Peng T. Research on
management system for lithium-ion battery packs. thermo-physical properties identification and thermal
Journal of Power Sources 2005; 141:307–315. analysis of EV Li-ion battery. Vehicle power and
7. Kizilel R, Lateef A, Sabbaha R, Farid MM, Selman JR, propulsion conference, 2009, VPPC’ 09, IEEE,
Al-Hallaj S. Passive control of temperature excursion Dearborn, MI, 7–10 Sep. 2009; 1643–1648.

24 Int. J. Energy Res. 2013; 37:13–24 © 2012 John Wiley & Sons, Ltd.
DOI: 10.1002/er