Applied Thermal Engineering 94 (2016) 846–854

Contents lists available at ScienceDirect

Applied Thermal Engineering

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / a p t h e r m e n g

Comparison of different cooling methods for lithium ion battery cells

Dafen Chen a,b, Jiuchun Jiang a, Gi-Heon Kim b, Chuanbo Yang b, Ahmad Pesaran b,*
a National Active Distribution Network Technology Research Center, Beijing Jiaotong University, Beijing 100044, China
b
National Renewable Energy Laboratory, 1617 Cole Blvd, Mail Stop 1633, Golden, CO 80401, USA

H I G H L I G H T S

• Performed 3D electrochemical-thermal modeling of four battery cooling methods.

• Thermal performance of direct air cooling, direct liquid cooling, indirect (jacket) liquid and fin cooling are compared.
• Merits and limitations of each cooling method for occupying a fixed volume are summarized.
• Temperature rise for a fixed load is lower with direct or indirect liquid cooling lower than air and fin cooling.

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

Article history: Choosing a proper cooling method for a lithium-ion (Li-ion) battery pack for electric drive vehicles (EDVs)
Received 15 April 2015 and making an optimal cooling control strategy to keep the temperature at a optimal range of 15 °C to
Accepted 5 October 2015 35 °C is essential to increasing safety, extending the pack service life, and reducing costs. When choos-
Available online 22 October 2015
ing a cooling method and developing strategies, trade-offs need to be made among many facets such as
costs, complexity, weight, cooling effects, temperature uniformity, and parasitic power. This paper con-
Keywords:
siders four cell-cooling methods: air cooling, direct liquid cooling, indirect liquid cooling, and fin cooling.
Li-ion battery
To evaluate their effectiveness, these methods are assessed using a typical large capacity Li-ion pouch
Cooling method
Cooling model cell designed for EDVs from the perspective of coolant parasitic power consumption, maximum tem-
Battery thermal management perature rise, temperature difference in a cell, and additional weight used for the cooling system. We
use a state-of-the-art Li-ion battery electro-chemical thermal model. The results show that under our
assumption an air-cooling system needs 2 to 3 more energy than other methods to keep the same average
temperature; an indirect liquid cooling system has the lowest maximum temperature rise; and a fin cooling
system adds about 40% extra weight of cell, which weighs most, when the four kinds cooling methods
have the same volume. Indirect liquid cooling is a more practical form than direct liquid cooling though
it has slightly lower cooling performance.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction behaviors of battery packs in the Honda Insight and Toyota Prius
using air cooling. An air-cooling system worked very well in HEVs
Energy-saving and environmentally friendly electric drive vehicle during standard drive cycles that could control the maximum tem-
(EDV) adoption in the market is increasing and has more potential perature below the limit of 55 °C and the temperature difference
if batteries have more energy, travel longer, and are less expen- was no more than 5 °C, but the maximum temperature was higher
sive. The battery thermal management system to keep the than the desired limit on an aggressive cycle. Choi and Kang [6] de-
temperature at an optimal range of 15 °C to 35 °C [1,2] is essential veloped a thermal model to structure the flow system and determine
for lithium-ion (Li-ion) battery packs in electrical vehicles (EVs) and the appropriate cooling capacity for an air-cooled HEV. Wang et al.
hybrid electrical vehicles (HEVs) to extend lifetime and ensure op- [7,8], Yang et al. [9] and Xu et al. [10] worked on the optimization
erating safety. During vehicle operation, considerable heat is of battery arrangement and airflow duct, respectively, to achieve
generated in the battery pack that needs to be rejected. How to better performance from air cooling. Zhao et al. [11] investigated
remove the generated heat, and keep the temperature uniform has parametric influence on a cylindrical battery module using air
become a challenge because of the high requirement of gravimet- cooling. Metal foam was added to improve the performace of air
ric and volume energy in EDVs. Several cooling methods have been cooling by Mohammadian et al. [12]. Thus the temperature unifor-
proposed and researched. Zolot et al. [3–5] evaluated the thermal mity was improved. Pesaran and Kim et al. [13,14] analyzed the
merits and shortcomings of liquid cooling and air cooling. Chacko
et al. [15] evaluated the performance of an indirect liquid cooling
* Corresponding author. battery pack and concluded that active indirect liquid cooling/
E-mail address: Ahmad.Pesaran@nrel.gov (A. Pesaran). heating would be one of the most promising means to achieve

http://dx.doi.org/10.1016/j.applthermaleng.2015.10.015
1359-4311/© 2015 Elsevier Ltd. All rights reserved.
 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854 847

battery thermal management. Yeow et al. [16] studied the fin cooling

system and discussed the advantages of using air to remove the heat
from fin ends compared to using liquid. Wu et al. [17] designed a
solution using a heat pipe to mitigate the temperature rise. Wang
et al. [18] used the heat pipes to cool and heat the EV battery and
found that cooling and heating via heat pipes is viable for EVs. A
flat heat pipe was used by Tran et al. [19] to reduce the thermal re-
sistance of a common heat sink and found it could handle more
efficiently instant increases of the heat flux than the conventional
heat sink. Kim et al. [20] researched the method of using phase
change materials for cooling and found that they can provide some
benefit to limiting peak temperature. Porous foam or carbon fiber
was used by some researchers [21–24] to enhance the perfor-
mance of PCM cooling. Liu et al. [25] suggested thermoelectric
cooling in battery thermal management systems and found it can
keep a more uniform temperature distribution.
Different cooling methods have different limitations and merits.
Air cooling is the simplest approach. Forced-air cooling can miti-
gate temperature rise, but during aggressive driving circles and at high
operating temperatures it will inevitably cause a large nonuniform
distribution of temperature in the battery [26,27]. Nevertheless, in
some cases, such as parallel HEVs, air cooling is adequate. A liquid
cooling system is more effective but its complexity, cost, and poten-
tial leakage make manufacturers hesitate to use it. Adding phase
change material to a battery pack could add unwanted mass and
volume. Phase change materials also need to exhaust the heat they
absorb during continuous cycling, but they could address maximum
temperature situations. Heat pipe cooling for Li-ion battery pack is
limited by gravity, weight and passive control [28].
Currently, air cooling, liquid cooling, and fin cooling are the most
popular methods in EDV applications. Some HEV battery packs, such
as those in the Toyota Prius and Honda Insight, still use air cooling.
Indirect liquid cooling has been adopted by the Chevrolet Volt, and
Tesla Model S. A123 used fins for heat removal and achieved tem-
perature uniformity. A fierce debate is ongoing about which kind
4.2
of cooling method should be applied to EDV battery packs. Because
1C discharge
each cooling method has unique advantages and disadvantages, the 2C discharge
purpose of this paper is to carry out cell-level thermal analysis in 4 3C discharge
the battery pack design process to assess which cooling method
should be adopted for different applications. That can help battery 3.8
designers design packs according to properties such as cell chem-
istry, geometry, and working conditions. Developing a cooling
Voltage(V)

strategy needs to be based on cell-level analysis results. In order 3.6

to compare the advantages and disadvantages of different cooling
methods and provide usable flow rate range under a specific control 3.4
target, this paper analyzes the effects of air cooling, direct liquid
cooling, indirect liquid cooling, and fin cooling. The results are then
3.2
applied to a large-capacity Li-ion pouch cell designed for EDVs. Com-
pactness and high energy density (i.e. volume efficiency) of the
battery pack draw great attention for integrating it in a vehicle with 3
limited space. In this paper, we compared the thermal perfor-
mance of each cooling method based on the same added volume 2.8
for cooling. The one that had the minimum temperature rise and 0 500 1000 1500 2000 2500 3000 3500 4000
minimum temperature variation was considered better if the volume Time(s)
were the same. Cases with different coolant flow rates or heat trans-
fer coefficients are simulated to compare the coolant flow power
consumption, maximum temperature rise, and temperature differ- Fig. 1. Battery characters and model. (For interpretation of the references to color
in this figure, the reader is referred to the web version of this article.)
ence in a cell. Additional cooling system weight is also compared,
assuming the same volume among cooling methods. Optimum flow
rate range is discussed in each case when all previously men-
tioned impacts are taken into account. tions. The picture of the battery selected for this analysis is shown
in Fig. 1(b). Fig. 1(c) contains discharge curves of the cell at 25 °C.
2. Battery modeling The cell is designed for an EDV battery pack with 128.5 Wh kg
energy density. Its anode material is graphite, and its cathode ma-
A 35 Ah prismatic pouch Li-ion cell with dimensions of 169 mm terial is a combination of NMC and MnO. According to the geometry
width, 179 mm long, and 14 mm thick is modeled for all simula- and structure of the battery, a three-dimensional model is built in
848 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854

4.2 5 45 4
Experiment Data Experiment Data
Simulation Data Simulation Data
4
4
40 3

Temperature Error(%)
3.8

Voltage Error(%)

Temperature(°C)
35 2
Voltage(V)

2
3.6
1
30 1
3.4
0

25 0
3.2
−1

3 −2 20 −1
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Time (s) Time (s)

Fig. 2. Voltage and temperature comparison of the simulation and experimental results. (For interpretation of the references to color in this figure, the reader is referred to
the web version of this article.)

ANSYS/Fluent, as shown in Fig. 1. Four parts comprise the model: used to acquire the temperature and other data. The experimental
the active volume, positive current tabs, negative current tabs, and and simulated results are shown in Fig. 2. The experimental tem-
skin. The active volume represents the stacked structure, includ- perature rise data used here are the average value of nine
ing positive and negative active materials, separator layers, and thermocouples, and the simulation temperature rise is the average
aluminum and copper foils. The positive tab and negative tab are temperature of the whole battery. Both the voltage drop and tem-
used to collect the current flow through the cell. The thin skin wrap- perature rise of the simulation compare well to the experimental
ping the active volume and part of the tabs representing the pouch data, indicating that the model can simulate the real battery.
is built to contain the internal component heat conduction through
skin surfaces in the model.
The model is solved in Fluent using the Multi-Scale Multi- 3. Cooling methods configuration and simulations
Dimensional (MSMD) battery module [29]. The electrical and thermal
fields are solved using equations 1 and 2, specifically [30]. Fig. 3 shows the schematic of each cooling method. For better
visualization, the cooling part is shown with increased thickness.
⎧∇ ⋅ (σ + ∇φ+ ) = − j All four methods use the two largest side surfaces of the cell to
⎪
⎨∇ ⋅ (σ −∇φ− ) = − j (1) reject heat. Parallel coolant passing by each cell in a battery pack
⎪ j = I Vol is assumed, so the nonuniformity in the pack is not considered.
⎩
Fig. 3(a) shows the configuration of air cooling or direct liquid
where σ is the effective electric conductivities for the electrode, ϕ cooling. In air cooling and direct liquid cooling, the coolant flows
is the phase potential for the electrode, + and − present the positive through the gap between two cells and contacts cell side surfaces
and negative respectively, j is the volumetric transfer current density directly. Although the two methods have the same configuration
computed by the current I divided by volume Vol of the sub model. at the cell level, direct liquid cooling is much more complicated
than air cooling from the perspective of the pack level and flow
⎧ ∂ρC P T  distribution. Dielectric mineral oil is used in direct liquid cooling,
⎪⎪ ∂t − ∇ ⋅ ( k∇T ) = q which adds extra weight to the battery pack and needs a more
⎨ (2)
complex circulation system, whereas air won’t gain significant
j ⎡⎢U − (φ+ − φ− ) − T
⎪q = σ + ∇ 2φ+ + σ −∇ 2φ− + dU ⎤
⎪⎩ ⎣ dT ⎥⎦ weight and is easy to circulate using fans in direct air cooling.
Further, the liquid leakage problem is worth careful consideration
where T is temperature, q is the heat generation rate during the in direct liquid cooling. Fig. 3(b) shows the configuration of fin
battery operation, k is the thermal conductive coefficient, Cp is the cooling. Heat generated during discharge/charge is conducted
heat capacity, and U is the open circuit voltage of the battery. through the largest side surfaces to the fin, which is sandwiched
A 1RC equivalent circuit model is used in this paper. To acquire between two cells. Then the heat is dissipated from the edge of
the parameters of the equivalent circuit model and validate model the fin, which is usually sitting on a cold plate or cooled by air
performance, two sets of experiments were conducted. The equiv- directly. The cold plate can be cooled by liquid or other methods.
alent circuit model parameters are calculated from multi-pulse Only one side of the fin edge is cooled in this paper. The fins,
discharge and charge data using the simulation software Matlab [31]. usually aluminum, add extra weight to the battery pack. The di-
A 95A discharge process is used to validate the voltage and tem- mension of each fin end is 169 mm × 197 mm × 1 mm . Fig. 3(c) shows
perature predicted by the model. Nine T-type thermocouples are the indirect liquid cooling configuration. A “jacket” is used to
pasted on the surface of the battery to monitor the temperature at contain the cooling liquid, conduct the heat from the cell to the
each area. During the discharge process, the battery is made to be coolant, and restrict the coolant in specific cooling channels. The
adiabatic by placing it in an adiabatic battery testing calorimeter, coolant is usually water/glycol. The jacket and coolant add some
Phi-TEC, produced by HEL [32,33]. An Arbin BT-2000 is used as the weight to the battery pack, and the leakage problem should be
discharge equipment and to monitor voltage. An Agilent 34901A is considered during operation. Fig. 3(d) shows the direct liquid cooling
 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854 849

Fig. 3. Cooling configuration with fix and the same gap between cells.

configuration; the coolant is a dielectric mineral oil. The proper- The extra weight of air cooling is negligible compared to the
ties of each material used in each cooling system are listed in battery weight (1.01 kg). Fin cooling adds maximum extra weight,
Table 1. approximately 39%, to the battery when all cooling methods have
Generally, as shown in Fig. 4, the following comprise a battery the same volume. Direct liquid cooling and indirect liquid cooling
pack cooling loop: a battery pack, a fan/pump, a heat exchanger, and add approximately 2.95% and 7.16% weight to the battery, respec-
coolant pipes [36]. In this paper, the volume for different cooling tively, which is acceptable in EDV applications. In conclusion,
methods is assumed to be the same – that is, the gap between two considering the structure and extra weight added to a battery, air
cells used for cooling in different cooling methods is the same. De- cooling is the simplest and lightest method, fin cooling adds the most
creasing the hydraulic diameter has positive effects on battery extra weight, the weight added in indirect liquid cooling and direct
cooling, whereas the power consumption of the cooling system will cooling is moderate, and direct liquid cooling add less extra weight
increase [14]. We use a 1 mm gap between two cells for cooling, than indirect liquid cooling because the density of the aluminum
as adopted by some manufactures and researchers for prismatic “jacket” is almost three times that of mineral oil.
pouch cells in air cooling [37] and fin cooling [16]. The gap used
for cooling between batteries in all methods discussed in this paper 4. Simulation results and discussion
is also set to 1 mm. The extra weight of different cooling methods
is calculated and shown in Table 2. The mass percentage is the ratio A series of simulations were conducted to estimate the effects
of cooling system weight to cell weight. of cooling by changing the flow velocity of coolant in air cooling
and liquid cooling. We let the average temperature rise at the end
of discharge reach a minimum of 4 °C so that all cooling methods
Table 1 are comparable. The velocity range is 0–20 m/s for air, 0–0.01 m/s
Material properties. for mineral oil and 0–0.05 m/s for water/glycol. Because the method
Item Air [34] Mineral Water/glycol Aluminum of conducting heat from the cooling surface of the fin cooling can
oil [34] [34] (50/50) [35] be variable, we assume that the thermal resistance of the cold plate
Density (kg/m3) 1.225 924.1 1069 2719 is negligible and specify only the heat transfer coefficient directly
Specific heat capacity 1006 1900 3323 871 on the cooling surface. The heat transfer coefficient can be calcu-
(J/kg/k) lated from the flow in the cold plate. The typical value is 5–25 W/
Thermal conductivity 0.0242 0.13 0.3892 202.4
(W/m/k)
m2/k using air cooling. A value of 390 W/m2/k was used for indirect
Kinematic viscosity 1.46e-5 5.6e-5 2.58e-6 — liquid cooling [36]. All the simulations have the same initial and inlet
(m2/s) temperature, the battery discharge current is set to 2.71C and the
operating state of charge range is from 1 to 0.2. The average heat
generation rate during the process calculated from experimental data
is approximately 15.7 W. The simulation results – the
averagetemperature of the battery in each of the different cooling
methods – are shown in Fig. 5. The cases in which air flow speed
is greater than 4 m/s in air cooling and when the heat transfer co-

Table 2
Extra weight added by cooling system for each cell.

Item Air Mineral oil Water/glycol Fin

Extra mass (kg) Negligible 0.0298 0.0723 0.394

Mass percentage (%) ≐0 2.95 7.16 39.0
Fig. 4. General schematic of battery pack cooling loop.
850 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854

Fig. 5. Average battery temperature at different flow speeds or h during 2.71C discharge. (For interpretation of the references to color in this figure, the reader is referred
to the web version of this article.)

efficient is greater than 400 W/m2/k are higher than what is currently increases dramatically. The ideal power consumption used for the
practical in an EV cooling system, so the temperature rise curves coolant driving force can be calculated using equation 3:
are plotted in dashed lines.
n
Fin cooling with a reasonable heat transfer coefficient can hardly
power = ∑ ΔPV
i i (3)
control the average temperature rise below 8 °C, even when the i =1
maximum heat transfer coefficient is 500 W/m2/k. The minimum
average cell temperature rise at the end of discharge using air cooling where ΔPi is the pressure drop from the coolant inlet to the coolant
is approximately 8 °C when the flow speed is during the realizable outlet of one coolant channel, Vi is the volume flow rate in the
range. The flow speeds or heat transfer coefficients used in those cases coolant channel, and i represents the coolant channel number. The
are in equal intervals, whereas the average temperature rise is not same inlet velocity is applied to the boundary conditions, so all Vi ,
decreased proportionally as the flow speed or heat transfer coeffi- in certain cases, are equal. Thus, the power consumption can be ex-
cient increases. The average temperature decrease rate at the end of pressed as equation 4:
discharge is reduced with the increasing flow speed or heat trans-
1 n n
Vi
fer coefficient. The average temperature of the battery reflects the
portion of heat generated during discharge that was not conducted
power = ∑ ΔPi ∑
n i=1 i=1
(4)

by the cooling system. This means that at high flow speeds, the added
cooling effect becomes less significant with the increase in coolant The ∑ni=1 ΔPi and ∑ni=1 Vi are obtained from the simulation results
flow rate, whereas the power used to blow or pump the coolant of each cooling method, as shown in Tables 3–5.
 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854 851

Table 3 are plotted in Fig. 7(a). Because various cooling methods can be used
Pressure drop and coolant flow rate of air cooling. to cool the cold plate in fin cooling, only a heat transfer coefficient
Flow speed (m/s) 2 4 6 8 10 is specified on the cooling surface. The power consumption of fin
∑i2=1 ΔPi ( Pa ) 144 296 454 618 788 cooling is not discussed here.
∑i2=1 Vi ( mL s ) 390 780 1171 1561 19524 Fig. 7(a) shows the power consumption of different cooling
Flow speed (m/s) 12 14 16 18 20
methods when we need to control the maximum temperature at
the end of a 2.71C discharge to a certain value. Much more power
∑i2=1 ΔPi ( Pa ) 966 1152 1342 1540 1745
is needed to control the maximum temperature within a certain
∑i2=1 Vi ( mL s ) 2342 2732 3123 3513 3904
range when using air cooling compared to liquid cooling. Jacket
cooling consumes less power than oil cooling.
Table 4 Fig. 7(b) shows that the average temperature at the end of a 2.71C
Pressure drop and coolant flow rate of oil cooling. discharge decreases as the mass flow rate increases, whereas the
Flow speed (m/s) 0.001 0.002 0.003 0.004 0.005 decreasing rate varies considerably. Jacket cooling has the highest
∑i2=1 ΔPi ( Pa ) 204 406 610 814 1016
rate of decline, because of the relatively large heat capacity. Air
∑i2=1 Vi ( mL s ) 0.20 0.39 0.59 0.78 0.98 cooling has the smallest decline rate because of the small heat ca-
pacity and poor thermal conductivity of air.
Flow speed (m/s) 0.006 0.007 0.008 0.009 0.01
As previously mentioned, the average temperature of the battery
∑i2=1 ΔPi ( Pa ) 1220 1422 1626 1830 2040
reflects the portion of heat generated during discharge that is not
∑i2=1 Vi ( mL s ) 1.17 1.37 1.56 1.76 1.95
rejected by the cooling system. Four cases in which average tem-
perature rise at the end of discharge is closest to 8 °C in each cooling
Fig. 6(a) shows the simulation results of discharge voltage from method are selected to analyze the temperature distribution of dif-
the cell state of charge 1 to 0.2. The average temperature when flow ferent cooling methods when all cooling methods rejected the same
rate of air cooing, direct liquid cooling, and jacket cooling is around heat from the cell. The mass flow rate, pressure drop, and power
5 g/s and when the heat transfer coefficient is 50 W/m2/k are shown consumption of each case using air cooling, oil cooling, and jacket
in Fig. 6(b). The average temperature of jacket cooling is 5 °C lower cooling are shown in Table 6. Note that the general trend of jacket
than air cooling and 2 °C lower than direct liquid cooling when the cooling being better than direct liquid cooling is for the assump-
mass flow rate of the coolant are nearly 5 g/s. tions of this study and should not be extrapolated as a general case.
The average temperature rise at the end of a 2.71C discharge com- The temperature distribution at the end of discharge of each se-
pared to the power consumption of air cooling and liquid cooling lected case is shown in Fig. 8. Fig. 8(a–d) shows the results of air
cooling, direct liquid cooling, indirect liquid cooling, and fin cooling,
respectively. The heat transfer coefficient at the cold plate of fin
Table 5 cooling in the selected case is 300 W/(m2K).
Pressure drop and coolant flow rate of jacket cooling. In air and direct liquid cooling, temperature increases gradual-
Flow speed (m/s) 0.005 0.01 0.015 0.02 0.025 ly from inlet to outlet. The positive tab (top) is hotter than the
∑i6=1 ΔPi ( Pa ) 378 756 1135 1515 1895
negative tab because more heat is generated in the positive tab and
∑i6=1 Vi ( mL s ) 0.07 0.14 0.22 0.29 0.36 heat conduction is slower than the negative tab. This is because the
positive tab, which is made of aluminum, has relatively smaller
Flow speed (m/s) 0.03 0.035 0.04 0.045 0.05
thermal and electrical conductivity than the negative tab, which is
∑i6=1 ΔPi ( Pa ) 2275 2656 3038 3421 3804
made of copper, as shown in Table 7. Temperature difference between
∑i6=1 Vi ( mL s ) 0.43 0.50 0.58 0.65 0.72
the cell surface and the coolant in air cooling is larger than that of

4.1 39
air @v=2m/s
2
fin @h=50w/(m k)
4 37 oil @v=0.00275m/s
water/glycol @v=0.025m/s
average temperature(°C)

3.9 35
Voltage(V)

3.8 33

3.7 31

3.6 29

3.5 27
0 200 400 600 800 1000 0 200 400 600 800 1000
time(s) time(s)

Fig. 6. Voltage and temperature simulation results. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)
852 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854

2 42
10
Air cooling Air cooling
Direct liquid cooling Direct liquid cooling
Jacket cooling 40 Jacket cooling

10 0
Ideal power consumption(W)

@v=20m/s

Average temperature(°C)
38
@v=8m/s
36
@v=0.009m/s
-2
10 @v=2m/s
@v=0.003m/s 34

@v=0.001m/s 32
-4 @v=0.05m/s
10

@v=0.02m/s 30

@v=0.005m/s
10 -6 28
2 4 6 8 10 12 0 0.5 1 1.5 2
Average temperature rise at the end of 2.71C discharge (°C) Mass flow rate(g/s)

Fig. 7. Simulation results of different cooling methods. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

liquid cooling as a result the relatively small thermal conductive co- poor thermal conductivity. Therefore, the operating mass flow rate
efficient of air, which is indicated by the temperature at the edge range of each cooling method should be carefully considered to
of inlet side. The temperature distribution of direct liquid cooling control both the temperature rise and temperature difference under
is similar to air cooling, but it has a larger temperature effect on set values and at the same time using minimum fan or pump power.
the battery. So the temperature gradient is larger than the air cooling For example, the control target of the temperature rise and tem-
in this case. The temperature difference of the jacket cooling is largest perature difference are both specified to be less than 4 °C (below
when the average temperature at the end of the discharge is con- the blue dashed line in Fig. 9). For air cooling, the temperature dif-
trolled to 8 °C. The temperature distribution of jacket cooling is more ference will never exceed the target value, so the temperature rise
complicated than other cooling methods because of the round- takes the dominant design point. For jacket cooling, to constrain the
about flow routing. The lowest temperature is around the inlet, temperature rise to 4 °C and the temperature difference to 4 °C, the
whereas the highest temperature is located at the top of the battery mass flow rate is 0.65 g/s and 0.70 g/s, respectively. Thus, the tem-
on the outlet side. Because the temperature of coolant becomes perature difference becomes the dominant design point, which
higher with the flow direction, the coolant at the outlet is the hottest. means that the mass flow rate should be larger than 0.70 g/s.
Heat generated by the cell around outlets is conducted to the inlet
part of the cell because of the relatively bigger temperature gradi- 5. Conclusions
ent at that direction and low temperature around inlets. So the cell
temperature around outlets is lower than the right-top. In fin cooling, In this paper, an electrochemical-thermal battery model for a pris-
heat at the opposite side of the cold plate needs to be conducted matic cell was built using ANSYS/Fluent, and its performance was
through the fins to the cooler side, therefore resulting in a rela- validated. Four cooling structures were analyzed based on the model:
tively high temperature difference in the battery. air cooling, direct liquid cooling, indirect liquid cooling, and fin
Fig. 9 shows the temperature difference ( Tdiff ) and average tem- cooling. The extra weight of the cooling systems is calculated and
perature rise ( Trise ) of each cooling method at the end of discharge compared. The cooling effect of each cooling method is accessed
as a function of mass flow rate. Although the average temperature using a series of simulations that change flow speed or heat trans-
rise decreases with the increase in mass flow rate, the tempera- fer coefficient from the perspective of coolant flow power
ture difference becomes larger with the increase in mass flow rate consumption and average temperature rise. The simulation results
at the low mass flow rate. After a certain mass flow rate, the tem- are useful to battery pack designers of electrical vehicle to assess
perature difference begins to decline with the increase in flow speed. and choose a proper cooling method under the volumetric con-
Indirect liquid cooling has the biggest maximum temperature dif- strain. They can also get a usable flow rate of different cooling
ference, because it has the longest flow channel, and it has the biggest methods for a specific control target.
decrease rate after the maximum temperature difference. Air cooling From this study, the following conclusions can be drawn:
has the smallest maximum temperature difference because of its
1. Fin cooling adds the most extra weight when all cooling methods
have the same volume.
Table 6
Design parameters of different cooling methods with same temperature rise. 2. Air cooling consumes the most parasitic power.
3. Both fin cooling and air cooling are constrained by minimum
Item Air Mineral oil Water/glycol (50/50)
average temperature rise considering the realistic heat transfer
Average temperature rise (°C) 8.4 8.2 8.3 coefficient range and fan power range in an electric car.
Flow velocity V (m/s) 4 0.00275 0.015
1 4. A maximum temperature difference exists at the low mass flow
Pressure drop ∑ni =1 ΔPi ( Pa ) 148 279 189
n rate area when the mass flow rate increases from zero.
Mass flow rate (g/s) 0.96 0.49 0.29 5. Indirect liquid cooling has the highest maximum temperature
Ideal power consumption (mW) 116 0.15 0.051
difference point because of the longest coolant channel, but the
 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854 853

Fig. 8. Temperature distribution of different cooling methods. (For interpretation of the references to color in this figure, the reader is referred to the web version of this
article.)
854 D. Chen et al./Applied Thermal Engineering 94 (2016) 846–854

Table 7 [9] N. Yang, X. Zhang, G. Li, D. Hua, Assessment of the forced air-cooling
Thermal and electrical conductivity of positive and negative tab [38]. performance for cylindrical lithium-ion battery packs: A comparative analysis
between aligned and staggered cell arrangements, Appl. Therm. Eng. 80 (2015)
Tap Material Thermal conductivity Electrical conductivity 55–65.
[10] X.M. Xu, R. He, Research on the heat dissipation performance of battery pack
( W ⋅ m −1 ⋅ k −1 ) ( S ⋅ m −1)
based on forced air cooling, J. Power Sources 240 (2013) 33–41.
Positive Aluminium 249 3.50 × 107
[11] J. Zhao, Z. Rao, Y. Huo, X. Liu, Y. Li, Thermal management of cylindrical power
Negative Copper 353 5.96 × 107
battery module for extending the life of new energy electric vehicles, Appl.
Therm. Eng. 85 (2015) 33–43.
[12] S.K. Mohammadian, S.M. Rassoulinejad-Mousavi, Y. Zhang, Thermal
management improvement of an air-cooled high-power lithium-ion battery
by embedding metal foam, J. Power Sources 296 (2015) 305–313.
[13] A.A. Pesaran, S. Burch, M. Keyser, An approach for designing thermal
management systems for electric and hybrid vehicle battery packs, Proceedings
of the 4th Vehicle Thermal Management Systems pp. 24–27 (1999).
[14] G. Kim, A. Pesaran, Battery thermal management design modeling, World Electr.
Veh. J. 1 (2007) 126–133.
[15] S. Chacko, S. Charmer, Lithium-ion pack thermal modeling and evaluation of
indirect liquid cooling for electric vehicle battery thermal management, in:
Institution of Mechanical Engineers (Ed.), Innovations in Fuel Economy and
Sustainable Road Transport, Woodhead Publishing, 2011, pp. 13–21.
[16] K. Yeow, H. Teng, M. Thelliez, E. Tan, Comparative study on thermal behavior
of lithium-ion battery systems with indirect air cooling and indirect liquid
cooling, ASME/ISCIE 2012 International Symposium on Flexible Automation,
American Society of Mechanical Engineers, 2012, pp. 585–591.
[17] M.-S. Wu, K. Liu, Y.-Y. Wang, C.-C. Wan, Heat dissipation design for lithium-ion
batteries, J. Power Sources 109 (2002) 160–166.
[18] Q. Wang, B. Jiang, Q. Xue, H. Sun, B. Li, H. Zou, et al., Experimental investigation
on EV battery cooling and heating by heat pipes, Appl. Therm. Eng. 88 (2015)
54–60.
[19] T.-H. Tran, S. Harmand, B. Desmet, S. Filangi, Experimental investigation on the
feasibility of heat pipe cooling for HEV/EV lithium-ion battery, Appl. Therm.
Eng. 63 (2) (2014) 551–558.
[20] G.-H. Kim, J. Gonder, J. Lustbader, A. Pesaran, Thermal management of batteries
in advanced vehicles using phase-change materials, World Electr. Veh. J. 2 (2008)
134–147.
[21] Z. Wang, Z. Zhang, L. Jia, L. Yang, Paraffin and paraffin/aluminum foam composite
Fig. 9. Temperature rise and difference at different mass flow rates. (For interpre- phase change material heat storage experimental study based on thermal
tation of the references to color in this figure, the reader is referred to the web version management of Li-ion battery, Appl. Therm. Eng. 78 (2015) 428–436.
of this article.) [22] A. Babapoor, M. Azizi, G. Karimi, Thermal management of a Li-ion battery using
carbon fiber-PCM composites, Appl. Therm. Eng. 82 (2015) 281–290.
[23] W.Q. Li, Z.G. Qu, Y.L. He, Y.B. Tao, Experimental study of a passive thermal
maximum temperature difference decreases more quickly after management system for high-powered lithium ion batteries using porous metal
the highest value than in other methods because of the higher foam saturated with phase change materials, J. Power Sources 255 (2014)
9–15.
heat capacity of water/glycol coolant. [24] N. Javani, I. Dincer, G.F. Naterer, G.L. Rohrauer, Modeling of passive thermal
6. Using the low mass flow rates of indirect liquid cooling to control management for electric vehicle battery packs with PCM between cells, Appl.
the temperature rise and temperature difference within a battery Therm. Eng. 73 (1) (2014) 307–316.
[25] Y. Liu, S. Yang, B. Guo, C. Deng, Numerical analysis and design of thermal
should be avoided. management system for lithium ion battery pack using thermoelectric coolers,
Adv, Mech. Eng. 2014 (2014) 1–8.
Acknowledgements [26] Z. Rao, S. Wang, A review of power battery thermal energy management, Renew.
Sustain. Energy Rev. 15 (2011) 4554–4571.
[27] T.M. Bandhauer, S. Garimella, T.F. Fuller, A critical review of thermal issues in
The authors gratefully acknowledge financial support from the lithium-ion batteries, J. Electrochem. Soc. 158 (2011) R1–R25.
U.S. Department of Energy, Vehicle Technologies Office, and China [28] T. Zhang, C. Gao, Q. Gao, G. Wang, M. Liu, Y. Guo, et al., Status and development
of electric vehicle integrated thermal management from BTM to HVAC, Appl.
Scholarship Council and Doctoral Innovation Fund of BJTU (No.
Therm. Eng. 88 (2015) 398–409.
E13JB00150). [29] G.-H. Kim, K. Smith, K.-J. Lee, S. Santhanagopalan, A. Pesaran, Multi-domain
modeling of lithium-ion batteries encompassing multi-physics in varied length
References scales, J. Electrochem. Soc. 158 (8) (2011) A955–A969.
[30] ANSYS, Inc. ANSYS FLUENT Battery Module Manual (2013).
[31] T. Huria, M. Ceraolo, J. Gazzarri, R. Jackey, High fidelity electrical model with
[1] Y. Ye, L.H. Saw, Y. Shi, A.A.O. Tay, Numerical analyses on optimizing a heat pipe thermal dependence for characterization and simulation of high power lithium
thermal management system for lithium-ion batteries during fast charging, battery cells, Electric Vehicle Conference (IEVC), 2012 IEEE International, IEEE,
Appl. Therm. Eng. 86 (2015) 281–291. pp. 1–8, (2012).
[2] A. Pesaran, S. Santhanagopalan, G. Kim, Addressing the impact of temperature [32] K. Fang, S. Chen, D. Mu, J. Liu, W. Zhang, The heat generation rate of nickel-metal
extremes on large format Li-ion batteries for vehicle applications (presentation), hydride battery during charging/discharging, J. Therm. Anal. Calorim. 112 (2)
Tech. rep., National Renewable Energy Laboratory (NREL), Golden, CO, (2013). (2013) 977–981.
[3] M. Zolot, A.A. Pesaran, M. Mihalic, Thermal evaluation of Toyota Prius battery [33] F. Bianchi, B. Tevene, Measurements in abusive tests on lithium ion polymer
pack, Tech. rep., SAE Technical Paper (2002). cells, Instrumentation and Measurement Technology Conference (I2MTC), 2015
[4] M.D. Zolot, K. Kelly, M. Keyser, M. Mihalic, A. Pesaran, A. Hieronymus, Thermal IEEE International, IEEE, pp. 1068–1071, (2015).
evaluation of the Honda Insight battery pack. Presented at the 36th Intersociety [34] G.-H. Kim, A.A. Pesaran, Battery thermal management system design modeling,
Energy Conversion Engineering Conference, Vol. 2, SAE, pp. 923–928, (2001). Presented at the 22nd International Battery, Hybrid and Fuel Cell Electric Vehicle
[5] K. Kelly, M. Mihalic, M. Zolot, Battery usage and thermal performance of the Conference and Exhibition (EVS-22), (2006).
Toyota Prius and Honda Insight during chassis dynamometer testing. The [35] J.E. Hatch, A. Association, et al., Aluminum: Properties and Physical Metallurgy,
Seventeenth Annual IEEE Battery Conference on Applications and Advances, ASM International, Geauga, OH, (1984).
2002, pp. 247–252. [36] A.A. Pesaran, Battery thermal management in EV and HEVs: issues and solutions,
[6] Y.S. Choi, D.M. Kang, Prediction of thermal behaviors of an air-cooled lithium-ion Battery Man. 43 (2001) 34–49.
battery system for hybrid electric vehicles, J. Power Sources 270 (2014) 273–280. [37] D. Chen, J. Jiang, Y. Duan, Z. Wang, F. Wen, Fluid and thermal analysis of power
[7] T. Wang, K.J. Tseng, J. Zhao, Z. Wei, Thermal investigation of lithium-ion battery Li-ion battery pack and experimental verification, in: L. Jia, Z. Liu, Y. Qin, M.
module with different cell arrangement structures and forced air-cooling Zhao, L. Diao (Eds.), Proceedings of the 2013 International Conference on
strategies, Appl. Energy 134 (2014) 229–238. Electrical and Information Technologies for Rail Transportation (EITRT2013),
[8] T. Wang, K. Tseng, J. Zhao, Development of efficient air-cooling strategies for vol. II, Springer, Berlin Heidelberg, pp. 161–170, (2014).
lithium-ion battery module based on empirical heat source model, Appl. Therm. [38] R.A. Serway, J.W. Jewett, Principles of Physics, vol. 1, Saunders College Pub., Fort
Eng. 90 (2015) 521–529. Worth, TX, (1998).