Applied Thermal Engineering 225 (2023) 120226

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Comprehensive efficiency analysis of air-cooled vs water-cooled electric

motor for unmanned aerial vehicle
Min Chang a, Bingzhu Lai b, Hui Wang b, *, Junqiang Bai a, b, Zhaoyong Mao a
a
Unmanned System Research Institute, Northwestern Polytechnical University, Xi’an, Shaanxi 710072, China
b
School of Aeronautics, Northwestern Polytechnical University, Xi’ an, Shaanxi 710072, China

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

Keywords: The efficiency of an electric motor for an unmanned aerial vehicle can reduce when its temperature exceeds the
Electric unmanned aerial vehicle working temperature. In this paper, air- and water-cooling methods for electric motors in unmanned aerial
Cooling method vehicles are proposed and comprehensively compared. The effects of fan speed, water flow speed, and convection
Thermal performance
heat transfer coefficient on the outer surface of the electric motor on the cooling capacity of the electric motor
Efficiency improvement
are discussed. A comprehensive performance evaluation factor for the cooling method of the motor is proposed.
Results show that when the fan speed increased from 800 to 2000 rpm, the cooling efficiency increased by
27.78% for the air-cooled motor, whereas when the inlet flow rate of cooling water increased from 0.1 to 0.9 m/
s, the cooling efficiency increased by 67.47% for the water-cooled motor. The comprehensive performance of the
water-cooled motor reached the highest level with a velocity of 0.7 m/s. The convection heat transfer coefficient
increased from 5 to 25 W/(m2⋅K), and the cooling efficiency increased by 22.93% and only 2.77% for the air- and
water-cooled motors, respectively. These findings can provide guidance for designing highly efficient electric
motor cooling equipment for unmanned aerial vehicles.

study considered the air speed at the inlet boundary as a constant value,
1. Introduction and the effect of the fan speed on the air inlet speed was ignored. Kang
et al. [10] proposed a self-ventilating cooling structure for an electric
Electric propulsion technology has a wide range of applications in motor. They observed that the new cooling structure could reduce the
electric vehicles [1,2], high-speed electric trains [3], solar-powered temperature of permanent magnets by 7%. The fan structure was not
unmanned aerial vehicles (UAVs) [4,5], and electric commercial considered in their model, and the inlet velocity was given a constant
aircraft [6]. An electric motor is an indispensable component of electric value. Hyeon et al. [11] designed a cooling system with a flow guide to
propulsion systems. In an environment with good heat dissipation, the enhance the cooling efficiency. They observed that the flow guide
power density of the electric motor can reach a high level. When the increased the cooling effect by 28%. The fan structure was not described
temperature of the electric motor exceeds the working temperature for a in their model, and the magnitude of the inlet velocity was directly set.
long time, it can cause adverse phenomena such as aging of materials in Galloni et al. [12] evaluated the performances of different fan shapes in
the electric motor, demagnetization of permanent magnets, and geo­ an air-cooled motor. They observed that the ratio of the impeller’s
metric deformation [7,8], which significantly reduce the reliability, height to its diameter affected the airflow through the motor. Although
service life, and efficiency of the electric motor. Therefore, controlling they considered the structure of the fan, the specific structure of other
the temperature is important to improving the efficiency of the electric parts of the motor was ignored, and the influence of the overall structure
motor for UAVs under different working conditions. of the motor on the flow characteristics was not considered. Most pre­
The air cooling method removes heat by introducing air flowing into vious studies simplified fans without considering different fan speeds.
the electric motor. For example, Kim et al. [9] eliminated the influence Although a few studies have considered the fan, the other parts of the
of stagnant flow by installing air-gap fans on both sides of a motor. They motor have been ignored. The motor fan and other structures have an
observed that using an air-gap fan could increase the cooling perfor­ interactive relationship; therefore, the entire motor should be modeled
mance by 55%, and the air-gap fan on the rear side of the motor had during the research process to obtain an accurate simulation result.
higher efficiency than that of the air-gap fan on the front side. Their The liquid cooling of a motor uses liquid to flow in a fixed area of the

* Corresponding author.
E-mail address: [email protected] (H. Wang).

https://doi.org/10.1016/j.applthermaleng.2023.120226
Received 12 August 2022; Received in revised form 4 February 2023; Accepted 13 February 2023
Available online 16 February 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Nomenclature y y direction
z z direction
cp Constant pressure specific heat capacity [J/(kg⋅K)]
G Turbulent kinetic energy caused by the average velocity Greek symbols
gradient (W/m3) ε Turbulent kinetic energy dissipation rate
h Convection heat transfer coefficient (CHTC) [W/(m2⋅K)] ρ Density (kg/m3)
k Turbulent kinetic energy (J) μ Dynamic viscosity (Pa⋅s)
K Thermal Conductivity [W/(m⋅K)] Subscript
M Motor weight (kg) am Ambient
n Gradient direction cool Cooling structure
P Pressure of fluid (Pa) eff Effective value
Pr Prandtl number f Fluid
q Calorific value per unit volume (W/m3) ori Original structure
T Temperature (K) out Outer surface
u Velocity in x direction (m/s) s Solid
v Velocity in y direction (m/s) shell Motor shell
w Velocity in z direction (m/s) t Turbulence
x x direction

Fig. 1. Schematic diagram of UAV structure.

motor for heat exchange. For example, Pei et al. [13] designed a water- the heat dissipation performance; therefore, they neglected the coupling
cooling structure with a primary water-cooling plate for motor cooling effect of the water-cooling device and motor.
and studied the effect of heat dissipation on the structure. They observed As reviewed above, the electric motor model has been simplified to
that the designed water-cooling structure significantly reduced the some extent in previous studies, and the interaction between the cooling
temperature increase and prevented the motor from burning out. They device and motor has been ignored. In this paper, the motor model is
only considered the structure of the water-cooled plate and did not built in detail, the interaction between the cooling device and the motor
consider the structure of the motor, including the stator core, windings, is considered in the simulation. The heating capacity of the windings in
and permanent magnets. Chiu et al. [14] evaluated the cooling perfor­ the motor was determined through experiments, and the temperature
mance of a reluctance motor. They observed that the temperature in­ distribution of the UAV motor under different cooling modes (including
crease of the windings of the motor could be reduced to 42% after air- and water-cooling methods) was simulated. In the air-cooled motor,
cooling using liquid water. Although they modeled the motor winding the motor mainly cools the heat source by drawing in the cold air of the
and stator core, they did not consider the influence of the rotation external low temperature through the fan. In the water-cooled motor,
function on the performance of the motor. Han et al. [15] conducted an the motor cools the heat source by introducing cooling water. In this
integrated design of a 30 kW motor water-cooled radiator and investi­ paper, the influence of fan speed and convection heat transfer coefficient
gated the cooling performance of three radiator configurations. They (CHTC) on the cooling capacity of air-cooled motor, and the influence of
observed that the serpentine channel exhibited good cooling perfor­ cooling water speed and CHTC on the cooling capacity of water-cooled
mance and temperature uniformity. They did not consider the entire motor are considered. Finally, the cooling capacity of the motor under
motor structure and only modeled the designed heat sink. Ge et al. [16] different cooling modes is comprehensively evaluated. The research in
calculated the heat dissipation effect on a motor. They observed that the this paper can directly provide some guidance for the motor design on
cooling performance of the circumferential helical coolant channel was UAV in different occasions.
better than that of an axial S-shaped flow channel. They directly
simplified the components of the motor into temperature nodes without
considering the actual structure of the motor. Most scholars have
simplified the model and ignored the influence of the motor structure on

2
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 3. Speed distribution of the section of the fan (800 rpm).

2. Problem Description

2.1. Geometrical Model

The structure of UAV mainly includes chassis, battery, motor, pro­

peller, fuselage main board, landing frame and connecting arm. Fig. 1
shows a typical structure diagram of UAV, and the motor structure is
partially enlarged. During the flight of the UAV, the motor on the UAV is
powered by the power supply, and the motor converts the electric en­
ergy into kinetic energy to drive the rotor on the UAV to rotate. Elec­
tronic energy is converted into heat energy, which results in an increase
in the temperature of the motor. When the temperature of the motor
exceeds the working temperature, the working efficiency of the electric
motor and the rotational speed of the rotatory wing decrease [17].
Permanent magnet synchronous motors are widely used in UAVs
because of their high efficiency and power density [18]. To dissipate
heat for the electric motor in a UAV, air- and water-cooled motors are
designed based on a permanent magnet synchronous motor. Fig. 2 (a)
and (b) shows the physical models of the air- and water-cooled motors,
respectively. Air-cooled motors are primarily composed of the fan, filter,
central shaft, shell, stator core, winding, permanent magnets, and
cooling fins. The shell and permanent magnet parts constitute the rotor
of the motor. The airflow path in the air-cooled motor is the gap between
the stator core and winding. In the air-cooled motor, the rotating parts
Fig. 2. Schematic of the outer rotor motor model. include the fan, filter, central shaft, shell and permanent magnets, which
are driven by the power supply to rotate clockwise. The suction gener­
ated by the rotation of the fan sucks the surrounding cold air into the
Table 1 motor. After entering the motor, the cold air flows through the gap
Structural parameters for the air-cooled motor and the water-cooled motors.
between the stator core and winding and flows in the axial direction to
Parameter Air-cooled Water-cooled cool the stator core and winding. Subsequently, the air flows out from
motor motor the bottom of the motor.
Model height (m) 1.18 × 10− 1 1.25 × 10− 1 The structure of the water-cooled motor is composed of a top cover,
Fan height (m) 1.3 × 10− 2 – central shaft, baffle, shell, stator core, winding, cooling-water inlet,
Fan outer diameter (m) 1.68 × 10− 1
cooling-water outlet, and permanent magnets. Similar to the air-cooled
–
Fan inner diameter (m) 7.8 × 10− 2 –
Top cover height (m) – 1.79 × 10− 2 motor, the rotating parts of the water-cooled motor include the top
Cooling water inlet and outlet diameters – 6.2 × 10− 3 cover, the central shaft, shell and permanent magnets. In the water-
(m) cooled motor, the circulation path of the cooling water is located be­
Stator outer diameter (m) 1.56 × 10− 1 1.4 × 10− 1 tween the stator core and central shaft, and the cooling water inlet and
Stator inner diameter (m) 1.02 × 10− 1 1.02 × 10− 1
Rotor outer diameter (m) 1.684 × 10− 1 1.524 × 10− 1
outlet boundaries are located at the lower part of the motor. The shape
Rotor inner diameter (m) 1.571 × 10− 1 1.411 × 10− 1 and flow direction of the cooling-water channel are shown in Fig. 2 (c).
Central axis diameter (m) 4.2 × 10− 2 4.2 × 10− 2 During the operation of the water-cooled motor, the cooling water flows
Permanent magnet thickness (m) 3.2 × 10− 3 3.2 × 10− 3 into the cooling channel of the motor at a certain speed and temperature.
After cooling the motor through convection heat exchange, it is dis­
charged from the outlet boundaries.
Whether it is air- or water-cooled, the lower part of the motor is

3
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Table 2 ⎧
Calorific value per unit volume under different loads. ⎪
⎪
⎪ mass :
⎪
⎪
⎪ ∂ u ∂v ∂w
Working load (kg) 0 20 40 60 ⎪
⎪
⎪ + + =0
⎪
⎪ ∂x ∂y ∂z
q (W/m )3
624,312 697,317 1,334,216 1,838,125
⎪
⎪
⎪
⎪
⎪
⎪ momentum :
⎪
⎪ ( ) ( 2 )
⎪
⎪ ∂u ∂u ∂ u ∂P ∂ u ∂2 u ∂2 u
⎪
⎪
⎪
⎪ ρf u + v + w =− + μeff + +
Table 3
⎪
⎪
⎪ ∂x ∂y ∂z ∂x ∂x2 ∂y2 ∂z2
⎨ ( ) ( 2 )
Boundary conditions of the air-cooled motor. ∂v ∂v ∂v ∂P ∂ v ∂2 v ∂2 v (1)
⎪
⎪ ρ f u + v + w = − + μ eff + +
Boundary condition Value ⎪
⎪
⎪ ∂x ∂y ∂z ∂y ∂x2 ∂y2 ∂z2
⎪
⎪ ( ) ( 2 )
Ambient temperature (K) 295
⎪
⎪
⎪
⎪ ∂w ∂w ∂w ∂P ∂ w ∂2 w ∂2 w
Ambient pressure (Pa) 101,325 ⎪ ρf u
⎪ +v +w =− + μeff + 2 + 2
⎪
⎪ ∂x ∂y ∂z ∂z ∂x 2
∂y ∂z
Fan inlet temperature (K) 295 ⎪
⎪
⎪
Speed of the fan (rpm) 800/1200/1600/2000
⎪
⎪ energy :
⎪
⎪
Shell convection heat transfer coefficient (W/(m2⋅K)) 5/10/15/20/25 ⎪
⎪ ( ) ( 2 )
⎪
⎪
⎪ ∂Tf ∂Tf ∂Tf ∂ Tf ∂2 Tf ∂2 Tf
Outlet pressure (Pa) 101,325 ⎪
⎩ ρ f c p,f u + v + w = K eff + +
∂x ∂y ∂z ∂x2 ∂y2 ∂z2

connected to the static parts. Permanent magnets are attached to the where u, v, and w are the velocities of the fluid in the x, y, and z di­
surface of the shell to generate a magnetic field. The winding is wound rections, respectively, ρf is the density of the fluid, P is the pressure of the
around the stator teeth, and the motor rotates if a direct current is fluid, μeff is the effective viscosity of the fluid, cp,f represents the
adopted. During motor operation, the main heating components are the constant-pressure specific heat capacity of the fluid, Tf is the tempera­
stator core and winding. The structural parameters of the air- and water- ture of the fluid, and Keff is the thermal conductivity of the fluid. The
cooled motors are listed in Table 1. viscosity and thermal conductivity of the fluid are expressed as
The parameters in Table 1 indicate that the diameter of the air-
μeff = μf + μt (2)
cooled motor is slightly larger than that of the water-cooled motor.
This is because the cooling water channel in the water-cooled motor is cp,f μt
the area between the stator core and central shaft. However, air passage Keff = Kf + (3)
Prt
in air-cooled motor is gap between the winding and stator core. There­
fore, in order to ensure that there is enough space in the air-cooled motor where μf is the dynamic viscosity of the fluid, μt is the turbulent dynamic
as the gas flow channel, the diameter of the stator core of the air-cooled viscosity of the fluid, Kf represents the thermal conductivity of the fluid,
motor must be appropriately expanded. In this way, it can ensure that and Prt represents the Prandtl number of the fluid. The turbulent dy­
the volume of air cooled motor winding and water-cooled motor namic viscosity can be calculated using the density of the fluid (ρf), and
winding is the same, so as to ensure that the heat source power is the relationship between the turbulent kinetic energy dissipation rate (ε)
consistent. and turbulent kinetic energy (k) is expressed as follows:

2.2. Mathematical Modeling 0.09ρf k2

μt = (4)
ε
The problem has been partially simplified: (1) air is considered an The standard k-ε model with turbulent kinetic energy and dissipation
ideal gas and (2) the radiant heat effect is ignored because of low tem­ rate is adopted as follows:
perature. At the same time, because the motor speed is low, the change [( ) ]
of air density is insignificant. As shown in Fig. 3, the air flow rate in the ∂ ( ) ∂ μ ∂k
ρf kui = μf + t + G − ρf ε (5)
motor is preliminarily calculated, and the speed section at the fan is ∂xi ∂xj Prk ∂xj
shown when the motor speed is 800 rpm. It can be seen that the [( ) ]
maximum air velocity is about 7.15 m/s, and the corresponding Mach ∂ ( ) ∂ μt ∂ε ε ε2
ρ εu = μf + + C1 G − C2 ρf (6)
number is 0.02 (the local sound velocity is 344.28 m/s). The value of ∂xi f i ∂xj Prε ∂xj k k
Mach number is far less than 0.2, so the air compressibility can be
where G is the turbulent kinetic energy caused by the average velocity
ignored [19,20]. The incompressible solution method can improve the
gradient, Prk and Prε are the turbulent Prandtl numbers of the k and ε
simulation efficiency and reduce the simulation cost while ensuring the
equations, respectively. Prk takes a value of 1, and Prε takes 1.3. C1 and
accuracy of the calculation results.
C2 are constants, the value of C2 is 1.92, and the value of C1 is 1.44. In
The three conservation laws for mass, momentum, and energy con­
the above formulas, G can be calculated as follows:
servation should be satisfied. In a steady-state process, the mathematical
( )
forms for these three conservation laws are expressed as [21]. ∂ui ∂uj ∂ui
G = μt + (7)
∂xi ∂xj ∂xj
In addition, a heat transfer model of the solid region of the electric
motor must be considered. The energy conservation equation of the solid
part of the electric motor is expressed as follows:
( 2 )
∂ Ts ∂2 Ts ∂2 Ts
ks + 2 + 2 +q = 0 (8)
∂x2 ∂y ∂z

where ks is the thermal conductivity of the solid material, Ts is the

temperature of the solid region, and q is the calorific value per unit
volume of the solid part. The calorific value originates from the heat loss
in the motor, which is caused by various losses in the operation of the
motor as displayed in Table 2.
The heat source includes the stator core and winding part. In the

4
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Table 4 The initial temperature of the motor is 295 K, and the pressure is
Boundary conditions of the water-cooled motor. 101,325 Pa. The motor shell is set as the convective heat transfer
Boundary condition Value boundary condition. Assuming that the motor shell is heat exchange
with the environment, the Robin boundary condition is adopted in
Ambient temperature (K) 295
Ambient pressure (Pa) 101,325 thermodynamics [22]. The heat transfer between the motor housing and
Cooling water inlet temperature (K) 295 the environment is expressed as:
Cooling water inlet velocity (m/s) 0.1/0.3/0.5/0.7/0.9
Shell convection heat transfer coefficient (W/(m2⋅K)) 5/10/15/20/25 ∂Tshell ( )
− kshell = hshell− am Tshell,out − Tam (9)
Cooling water outlet pressure (Pa) 101,325 ∂n

where Tshell is the temperature distribution of the motor shell, Tshell,out is

simulation process of the air-and water-cooled motors, the volume of the the temperature at the outer surface of the shell, kshell is the thermal
stator core and the winding are guaranteed to be the same. The total heat conductivity of the motor shell, and hshell-am is the CHTC between the
generation of the two cooling modes is the same, which is convenient for shell and environment. The n-direction indicates outward along the
comparing the cooling effects of the different cooling methods. radial direction of the motor. The CHTC can be set to 5–25 W/(m2⋅K)
according to the flight altitude and velocity of the UAV [23], and the
2.3. Boundary Conditions specific values are listed in Table 3.
At the interface between the fluid and solid, the velocity is a no-slip
To solve the governing equations in the fluid and solid regions, the boundary condition, and the heat flux conditions as follows:
model must be set with the corresponding boundary conditions. For the ⎧
uinterface = vinterface = winterface = 0
air-cooled motor, the speed of the fan and the CHTC of the shell are ⎪
⎪
⎪
⎨
considered the research objects. The influence of different speeds and − kf
∂Tf ⃒⃒ ∂Ts ⃒⃒ (10)
interface = − ks
the CHTCs on the cooling effect of the air-cooled motor is analyzed.
interface
⎪
⎪ ∂ n ∂n
⎪
⎩
In an air-cooled motor, cold air is sucked from the upper end of the Tf ,interface = Ts,interface
motor by a fan. The cold air passes through the gas channel between the
stator core and winding, exchanges heat with the heat source and flows where uinterface, vinterface and winterface are the velocity component at the
out from the lower part of the motor. In addition, to consider the in­ interface. Since it is a no-slip wall boundary, uinterface, vinterface and win­
fluence of the rotating parts of the motor on the internal flow field and terface are equal to zero. Secondly, the heat flux on both sides of the
heat transfer performance, we must set the shell, permanent magnets, interface should be consistent, that is, the heat flux calculated by Fou­
filter, and fan to the rotating mode. The rotational speed is the same as rier’s law is equal. At the same time, the heat transfer temperature dif­
the fan speed, and the other parts remain stationary. The boundary ference between the fluid and the solid is ignored in the calculation of
conditions of the air-cooled motor are listed in Table 3. heat transfer, so that the temperature of the fluid side and the solid side

Fig. 4. Schematic of the unstructured grid for the air- and water-cooled motors (From left to right: the motor grid, the mid section grid and the grid near the wall).

5
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Table 5
Grid independence test of air-cooled motor model.
Number of grids Grid size (m3) Winding temperature (K) Deviation from average temperature (K)

Maximum Minimum Maximum Minimum Maximum (Average value is 351.57 K) Minimum (Average value is 348.43 K)
7 13
21,399,417 3.13 × 10− 4.66 × 10− 352.33 348.58 0.76 0.15
7 13
27,653,015 1.67 × 10− 4.35 × 10− 352.63 349.91 1.06 1.48
7 13
29,501,059 1.23 × 10− 3.17 × 10− 350.45 347.76 − 1.12 − 0.67
8 13
31,609,171 8.82 × 10− 1.87 × 10− 351.70 348.88 0.13 0.45
8 13
33,047,467 8.69 × 10− 1.22 × 10− 351.62 347.54 0.05 − 0.89
8 13
34,447,747 7.18 × 10− 1.09 × 10− 350.28 347.52 − 1.29 − 0.91
8 14
40,167,489 6.17 × 10− 7.35 × 10− 351.95 348.79 0.38 0.36

Table 6
Grid independence test of water-cooled motor model.
Number of grids Grid size (m3) Winding temperature (K) Deviation from average temperature (K)

Maximum Minimum Maximum Minimum Maximum (Average value is 335.92 K) Minimum (Average value is 329.01 K)
7 13
13,920,635 1.78 × 10− 5.26 × 10− 337.28 329.31 1.36 0.30
7 13
15,019,930 1.32 × 10− 3.95 × 10− 337.02 330.42 1.10 1.41
7 13
17,441,695 1.20 × 10− 3.77 × 10− 336.72 329.58 0.80 0.57
7 13
19,027,829 1.00 × 10− 2.65 × 10− 335.80 329.62 − 0.12 0.61
8 13
21,278,887 8.26 × 10− 2.58 × 10− 335.04 328.47 − 0.88 − 0.54
8 13
23,996,142 8.17 × 10− 1.95 × 10− 334.00 328.12 − 1.92 − 0.89
8 13
25,966,045 6.21 × 10− 1.65 × 10− 335.58 327.58 1.66 − 1.43

at the interface is equal. where Δy is the distance from the nearest mesh point to the wall, ut
The flow direction of the cooling water in the water-cooled motor is stands for the shear stress velocity and ν is kinematic viscosity. The wall
shown in Fig. 2 (c). The cooling water enters the cooling water channel treatment method used in this work is the enhanced wall method, which
from the inlet boundary at the lower end of the motor and flows around is not sensitive to the value of y + [25]. Previous relevant literature has
the circumference of the motor. After the water cools the motor, the proved that in the enhanced wall method, even when the y + value is
temperature of the water increases and flows out from the outlet between 3 and 10, the velocity and temperature profiles near the wall
boundary. As with the air-cooled motor, the shell and permanent mag­ can still be reasonably displayed [26]. In the simulation of this work,
nets of the motor are set as a rotating model, the speed is fixed at 800 because the model is speccific, the size of the wall grid structure pro­
rpm, and the other components remain stationary. The boundary con­ cessed at different parts is different, and the y + value in most areas is
ditions of the water-cooled motor are presented in Table 4. less than 1. However, there are still a small part’s y + value is large. For
The initial calculation conditions of the water-cooled motor are the example, the maximum y + value in air-cooled motor is 3.81, and the
same as those of the air-cooled motor, with an initial temperature of 295 maximum y + value in water-cooled motor is 4.21, which has little
K and pressure of 101,325 Pa. The inlet cooling-water temperature is impact on the energy exchange in this work [26].
maintained at 295 K, the cooling-water outlet is the pressure outlet, and In addition, the simulation used a steady state algorithm based on
the outlet pressure is the same as the atmospheric pressure. Eqs. (9) and pressure, and a single model uses 56 cores for parallel calculation, which
(10) are also applicable to the boundary calculation of a water-cooled takes about one to two days. In the spatial discrete model, the standard
motor. model is used for pressure, and the second-order upwind scheme is used
for momentum and energy discretization.
2.4. Simulation Settings In addition, because some components of the motor rotate during
operation, it needs to be set separately. In the water-cooled motor, the
In this paper, according to the designed motor model, the unstruc­ rotating parts include the central shaft, shell, permanent magnets and
tured grid is divided by ICEM software, and then the divided grid is filter. In the air-cooled motor, the rotating parts also include a fan. In
imported into FLUENT for calculation condition setting and simulation order to simulate the effect of rotating structure on the surrounding fluid
calculation. Finally, the calculation results are imported into TECPLOT in the steady-state solver, the Multiple Reference Frame (MRF) method
for post-processing. is adopted [27]. In the MRF method, the direction and speed of rotation
This section mainly introduces the simulation settings in FLUENT. need to be set for the rotating region. In this work, the rotation speed is
The semi-implicit method for pressure-linked equations (SIMPLE) al­ the rotational speed of the motor, which is 800, 1200, 1600 and 2000
gorithm and finite volume method were adopted. The temperature rpm respectively, and the rotation direction is the clockwise direction of
distribution profile of the motor was obtained by solving the three the x-axis. In addition, it also needs to set the wall surfaces of the
control equations. Among them, the viscosity model used in the simu­ rotation area as rotating walls, and the rotation direction and speed are
lation is the standard k-ε model. The standard turbulence model is the the same as the rotation area.
main tool for engineering calculation, which has wide application scope, To ensure the accuracy of the calculation results, we set the residuals
economy and reasonable accuracy. Eqs. (5)–(6) are the turbulent kinetic of the continuity variance and momentum equations to 10− 4 and the
energy and dissipation rate equations to be solved by the standard k-ε residuals of the energy equations to 10− 6. The results were output after
model in this problem. In order to simulate the flow near the solid wall, the calculated residuals converged.
it is necessary to study the consistency of the flow field and the algo­
rithm. For this reason, the dimensionless coefficient y + is proposed, and 3. Mesh Independence Test
its calculation method is expression as [24].
Δyut A schematic of the grid structure of the air- and water-cooled motors
y+ = (11) is shown in Fig. 4. To stabilize the flow state of the cooling water at the
ν

6
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 5. Workflow diagram.

Fig. 6. Experimental test device.

Table 7 Table 8
Physical properties of copper and aluminum. Temperature of the motor winding under different loads.
Parameter Copper Aluminum Working load (kg) Point 1 (K) Point 2 (K) Average temperature (K)
3
Density (kg/m ) 8978 2719 0 318.0 318.0 318.0
Thermal conductivity (W/(m⋅K)) 387.6 202.4 20 321.2 321.1 321.15
Specific heat capacity (J/(kg⋅K)) 381 871 40 337.9 343.5 340. 7
60 380.7 372.2 376.45

inlet and outlet boundaries and eliminate the influence of the boundary
effects on the water-cooled motor, we slightly extended the grid area at area is adopted to ensure the calculation accuracy of fluid area.
the inlet and outlet boundaries of the cooling water. All grids adopt After dividing the mesh, we verified the mesh independence of the
unstructured grids, and the specific size information is listed in Table 5 model to avoid large errors caused by the number of meshes. The model
and Table 6. In addition, for the mesh of fluid area, due to the complex of air- and water-cooled motor is divided by using unstructured grids of
shape of the model, the method of densifying the overall mesh of fluid different sizes [28,29]. Each type of motor is divided into seven models

7
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Table 9 simulation results. After verifying that the error between the model re­
Boundary conditions of the the verified model. sults and the test results is within the acceptable range, the simulation
Boundary condition Value method can be used to study the influence of different fan speed, cooling
water speed and CHTC on the heat dissipation capacity of the motor.
Ambient temperature (K) 295
Ambient pressure (Pa) 101,325 Fig. 6 shows the experimental test device used in this study. The
Shell convection heat transfer coefficient (W/(m2⋅K)) 15 ambient temperature during the experiments was 295 K. In the test
Working load (kg) 0/20/40/60 method, a 300 V direct current was connected to the input port of the
motor, and then the motor was applied with loads of 0, 20, 40, and 60
kg. Subsequently, the motor was started to run for a period, and when
with different grid numbers. The temperature distribution in the motor
the motor operation state was stable, a temperature measuring instru­
is calculated using the boundary conditions and settings described in
ment was used to measure the temperature of the two symmetrical
sections 2.3 and 2.4. For the air-cooled motor, the working load was 60
points on the motor winding. When the operating temperature of the
kg, the fan speed was 800 rpm, and the CHTC between the shell and
temperature measuring instrument is within the range of 273.15 to
environment was 15 W/(m2⋅K). The results for the different mesh
773.15 K, the measurement uncertainty is ± (0.75% + 1 K). The phys­
numbers are listed in Table 5.
ical parameters of copper and aluminum are depicted in Table 7, and the
For the water-cooled motor, the working load was 60 kg, the inlet
experimental results are listed in Table 8.
velocity was 0.1 m/s, and the CHTC was 15 W/(m2⋅K). The simulation
results for the different mesh numbers are listed in Table 6.
It can be seen from the Table 5 and Table 6 that the temperature of
motor winding is basically the same under different grids. In the air-
cooled motor, the maximum difference between the highest tempera­
ture and the lowest temperature is 2.05 K and 2.39 K respectively. In
water-cooled motors, the maximum difference of the highest tempera­
ture is 3.28 K, while the maximum difference of the lowest temperature
is 2.84 K. The error between calculation results do not exceed 1% of the
temperature, this shows that the number of grids has little influence on
the calculation results, and proves the independence between grids. At
the same time, we compared the difference between the calculation
results and the average value of different grids. Considering the calcu­
lation accuracy and efficiency, we selected the model with 31,609,171
grids in the air-cooled motor and 19,027,829 grids in the water-cooled
motor for calculation.

4. Experiment Test

4.1. Experimental Apparatus

Fig. 5 is the workflow diagram of this article. In this paper, the heat
dissipation of the motor is studied by experiment and simulation. Among
them, the experiment part is mainly to determine whether the conditions
set in the simulation work are appropriate to ensure the rationality of the Fig. 8. Air-cooled motor winding temperature at different fan speeds.

Fig. 7. Winding temperature obtained by experiment and simulation and their relative error.

8
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 9. Temperature and speed distribution of the air-cooled motor at different fan speeds.

4.2. Validation of the Simulation Model them, the speed of the motor can be regulated by an electronic speed
control.
In the validation model, we adopted the same settings as the test The temperature distribution of the motor was calculated using the
conditions, including ambient temperature, ambient pressure, motor simulation model and was compared to the experimental results with
speed, etc. See Table 9 for specific values. At the same time, the con­ loads of 0, 20, 40, and 60 kg as shown in Fig. 7.
vection heat transfer coefficient of the outer wall is determined ac­ From the comparison between the experimental results and the
cording to the engineering experience, assuming a constant value of 15 simulation results, it can be found that the winding temperatures ob­
W/(m2⋅K). No cooling devices are included in the validation model. tained by the two methods are basically the same. At the same time,
When the load of the motor is 0, 20, 40 and 60 kg respectively, the speed Fig. 7 also shows the relative error between the experimental and
of the motor is 1180, 950, 1308 and 1600 rpm, respectively. Among simulation results, with the maximum error not exceeding 1%. This

9
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 10. Air-cooled motor winding temperature with different convection heat
transfer coefficients.

shows that the simulation result of motor temperature is reliable, and

the model can be used to simulate the heat dissipation of motor. At the
same time, the reasons for the difference between the simulation results
and the experimental results are analyzed. First, different from the
constant environmental conditions that can be maintained in the
simulation, the environmental conditions will fluctuate during the
experiment. At the same time, due to unstable power supply voltage and
other reasons, the heating capacity of the motor may change, resulting
in differences from the simulation. In addition, there will also be some
random errors and systematic errors in the temperature measurement
process. These will lead to differences between the results.

5. Results and Discussion

5.1. Fan Speed for the Air-Cooled Motor

The initial temperature of the motor was 295 K, the CHTC between
the motor shell and environment was 15 W/(m2⋅K), and the motor load
was 60 kg. The temperature field distributions of the motor at four fan
speeds (800 rpm, 1200 rpm, 1600 rpm, and 2000 rpm) were obtained.
The maximum and minimum temperature changes in the motor winding
at different fan speeds are shown in Fig. 8. The maximum and minimum
temperatures of the air-cooled motor decreased with increasing fan
speed. Compared with the temperature of the motor shown in Fig. 7
without any cooling device, the air-cooled motor reduced the temper­
ature of the motor to a certain extent. For example, when the load was
60 kg, the motor winding temperature without any cooling device was
373.3 K. When the fan speed was 800 rpm, the maximum temperature of
the motor winding was 351.7 K, which was approximately 21.6 K lower
than that of the motor without cooling device, which meant that the
temperature increase decreased by 27.59%. As the fan speed increased,
the temperature of the winding in the motor gradually decreased. When
the fan speed increased from 800 to 2000 rpm, the maximum temper­
Fig. 11. Air-cooled motor temperature distribution for different convection
ature of the motor decreased by 6 K, and the cooling efficiency increased heat transfer coefficients.
by 27.78%. This was because, as the fan speed increased, the airflow
velocity in the cooling channel of the air-cooled motor gradually
temperature area in the motor, decreased with an increase in the rota­
increased. When the airflow velocity increased, the flow rate of cold air
tional speed of the fan. This was due to the increased pressure differ­
passing through the motor increased, thereby increasing the CHTC be­
ential across the fan as the fan speed increased, resulting in a greater
tween the cold air and motor. This ultimately reduced the temperature
inlet airflow velocity. The speed profiles in Fig. 9 (b), (d), and (f) show
of the motor.
that the speed of the air flowing through the motor increased with the
The temperature distribution of the motor on the z = 0 plane at 800,
speed of the fan. The CHTC increased when the airflow velocity
1600, and 2000 rpm was intercepted (Fig. 9 (a), (c) and (e)). The
increased. This meant that the cooling efficiency of the motor can be
maximum temperature in the motor winding, as well as the high-
improved by increasing the fan speed to increase the flow of cold air

10
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

distribution of the motor when the CHTCs were 5, 15, and 25 W/(m2⋅K),
respectively. It can be seen that when CHTC is 5 W/(m2⋅K), the tem­
perature in the winding and stator area of the motor in the section ex­
ceeds 349 K. However, with the increase of CHTC, the area of the area
where the temperature exceeds 349 K is shrinking. When the CHTC in­
creases to 25 W/(m2⋅K), only the temperature of some areas in the right
winding area exceeds 349 K. This shows that although the CHTC has no
significant impact on the maximum temperature of the motor,
increasing the CHTC can still improve the cooling efficiency of the motor
to a certain extent and reduce the area of the high-temperature area.

5.3. Inlet Flow Rate for Water-Cooled Motors

The inlet flow velocities were set as 0.1, 0.3, 0.5, 0.7, and 0.9 m/s.
The load was set at 60 kg, the initial temperature of the motor and
cooling water was set, and the ambient temperature was 295 K. The
CHTC between the motor shell and environment was set as 15 W/
(m2⋅K). The maximum and minimum temperatures of the winding at
different inlet flow velocities were determined as shown in Fig. 12.
Fig. 12. Water-cooled motor winding temperature with inlet velocities.
As the inlet flow rate increased, both the maximum and minimum
temperatures of the motor winding decreased. When the motor does not
have any cooling device, the maximum temperature of the winding in
the motor is 373.3 K (78.3 K higher than ambient temperature). In
water-cooled motors, when the cooling water inlet velocity is 0.1 m/s,
the maximum temperature of the motor winding is 335.8 K (40.8 K
higher than ambient temperature), which is 37.5 K lower than the motor
without cooling device. That is to say, compared with the motor without
cooling device, the temperature rise of water-cooled motor with inlet
velocity of 0.1 m/s decreases by 47.89%. In addition, with an increase in
the inlet flow velocity, the magnitude of the temperature decrease
gradually decreased. When the inlet velocity reached 0.7 m/s, as the
inlet flow velocity continued to increase, the maximum and minimum
temperatures did not change, indicating that the cooling limit of the
water-cooled motor was reached, and the cooling efficiency of the motor
could not be strengthened by further increasing the inlet velocity of the
cooling water. When the inlet velocity of the cooling water increased
from 0.1 to 0.9 m/s, the maximum temperature of the motor winding
decreased from 335.8 to 310.5 K. That is to say, when the inlet velocity
of cooling water is 0.1 m/s, the maximum temperature of motor winding
decreases by 37.5 K compared with that without cooling device, while
Fig. 13. Differential pressure of cooling water with inlet velocities. when the inlet velocity is 0.9 m/s, the maximum temperature of motor
winding can be reduced by another 25.3 K on this basis, and the cooling
through the motor. efficiency increased by 67.47%.
In addition, it should be noted that with the increase of the cooling
5.2. Convection Heat Transfer Coefficient of an Air-Cooled Motor water flow rate, the pressure difference between the inlet and outlet of
the motor cooling water channel will also increase, as shown in Fig. 13.
The CHTCs between the motor shell and environment were set to 5, When the inlet speed of cooling water increases from 0.1 to 0.9 m/s, the
10, 15, 20, and 25 W/(m2⋅K), and the ambient temperature was 295 K. pressure difference between the inlet and outlet of cooling water in­
The speed of the motor was 800 rpm and the load was fixed at 60 kg. The creases from 93.35 Pa to 4120.21 Pa. As the import and export pressure
influence of the different CHTCs on the motor winding temperature is difference increases, the high pump power is required. Therefore, in
shown in Fig. 10. Both the maximum and minimum temperatures in the order to obtain a better cooling effect, it is necessary to provide more
winding decreased with an increase in the external CHTC. This was power in the water-cooled motor. In practical applications, it is neces­
because, as the CHTC of the ambient convection increased, the motor sary to balance the relationship between heat dissipation capacity and
could dissipate more heat through the shell, thereby reducing the tem­ pump power, and make choices according to the actual situation. For
perature of the heat source. When the CHTC increased from 5 to 25 W/ example, compared with the inlet speed of 0.7 m/s, the inlet speed of
(m2⋅K), the maximum temperature of the winding decreased by 4.43 K, 0.9 m/s can only reduce the maximum temperature of the motor by 0.72
and the cooling efficiency increased by 22.93%. Different convection K, but it will consume about 39% higher pump work, which does not
heat transfer coefficients have influence on the temperature distribution conform to the principle of economy.
of air-cooled motor. It can be seen from the temperature cloud diagram Fig. 14 (a)–(f) show the temperature distributions of the motor and
in Fig. 10 that with the increase of the convection heat transfer coeffi­ cooling water for different cooling water inlet flow velocities. The
cient, the area where the temperature is higher than 349 K on the section temperature distribution of the motor is represented by the cross-
is decreasing. At the same time, because the convective heat transfer sectional temperature of the z = 0 plane (Fig. 14 (a), (c), and (e)).
coefficient selected in the study is small, the heat dissipation power Area A represents the cooling water and area B represents the heat
through convective heat transfer is small, so the influence of the source. The temperature on the left side of the motor was slightly lower
convective heat transfer coefficient on temperature is limited. than that on the right side, which was related to the flow direction of the
Fig. 11 (a), (b), and (c) show the cross-sectional temperature cooling water. The cooling water first entered from the inlet on the left

11
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 14. Water-cooled motor temperature distribution for different cooling water inlet velocities.

side of the motor, flowed clockwise, and then flowed out from the outlet. a poor cooling effect. The temperature on the right side of the motor was
Therefore, the temperature of the cooling water cooling the left side of higher than that on the left side of the motor. The change in the cooling
the motor was lower and the cooling effect was greater. As the cooling water temperature is shown in Fig. 14 (b), (d), and (f), and the tem­
water flowed clockwise, the temperature of the cooling water gradually perature of the cooling water increased continuously during the entire
increased because of the heat absorbed from the heat source, resulting in flow process. With an increase in the inlet velocity, the temperature

12
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

5.5. Comprehensive Performance Comparison of the Air- and Water-

Cooled Motors

On the UAV, due to the limited power that can be provided by the
power supply, there are certain requirements for the weight of the UAV
[30]. With the increase of UAV weight, the energy required to maintain
flight is also increasing. In order to improve the endurance of the UAV,
weight of UAV is usually taken as the optimization objective during
structural design [31]. In addition, additional components are added to
the motor in the air- and water-cooled method, which increases the
weight of the UAV and additional heat dissipation power consumption.
Compared with the increased mass, the power consumed by the
increased heat sink has almost negligible impact on the UAV. Therefore,
the heat dissipation efficiency and structure weight of the motor are
taken as the influencing factors for evaluating the motor performance
[32]:
Mcool Tcool
F= + (12)
Mori Tori
Fig. 15. Water-cooled motor winding temperature with convection heat
where Mori is the weight of the motor without any cooling device, Mcool is
transfer coefficients.
the weight of the air-cooled or water-cooled motor, Tori is the temper­
ature of the motor winding without any cooling device, and Tcool is the
increase of the cooling water decreased, and the high-temperature area
temperature of the motor winding after air or water cooling. The value
also decreased. This implied that an increase in the inlet flow velocity
of F is small, indicating that the motor with this cooling method has good
enhanced the heat dissipation effect of the water-cooled motor.
heat dissipation performance and is lightweight; that is, the heat dissi­
pation benefit introduced by increasing the weight of the motor is high.
5.4. Convection Heat Transfer Coefficient of Water-Cooled Motor
The motor weight (Mori), without any cooling device, was 3.682 kg.
The air-cooled motor replaced the upper top cover with a fan, and its
The cooling effect of the motor was simulated under five conditions
total weight was 3.692 kg. A water-cooled motor also must consider the
of the CHTCs with 5, 10, 15, 20, and 25 W/(m2⋅K). The simulated
weight of the cooling water. The total weight of the water-cooled motor
ambient temperature is 295 K, the inlet speed of cooling water is 0.1 m/
was 4.367 kg. When the motor load was 60 kg, the temperature (Tori) of
s, and the working load of the motor is 60 kg. The maximum and min­
the motor winding without a cooling device was 373.3 K. The values of
imum temperature changes in the motor winding are shown in Fig. 15.
the motor objective function for the different cooling methods are shown
Increasing the CHTC also slightly reduced the highest temperature in the
in Fig. 17.
water-cooled motor; however, the reduction was not significant. When
The comprehensive performance evaluation factor of the air-cooled
the CHTC increased from 5 to 25 W/(m2⋅K), the maximum temperature
motor was significantly smaller than that of the water-cooled motor.
of the motor winding decreased by only 1.04 K, and the minimum
In the water-cooled motor, the minimum value of F was 2.02, whereas
temperature decreased by only 0.8 K. The cooling efficiency of the motor
the maximum F value of the air-cooled motor was only 1.94. This
increased by only 2.77%. This was because the thermal conductivity and
demonstrates that, although the cooling capacity of the air-cooled motor
heat capacity of water was higher than those of air, and the convection
is smaller than that of the water-cooled motor, the weight requirement
heat transfer intensity between the motor and cooling water was
of the air-cooled motor is smaller than that of the water-cooled motor. If
significantly greater than the convection heat transfer intensity of the
the cooling power requirement of the motor is not large but the weight
casing, resulting in most of the heat of the motor being absorbed and
requirement is very low, an air-cooled motor is the best option. If a
removed by the cooling water. The CHTC of the shell had minimal effect
higher cooling capacity can be obtained by sacrificing weight, a water-
on the maximum and minimum temperatures in the water-cooled motor.
cooled motor can be selected.
Fig. 16 (a), (c), and (e) show the temperature distribution of the
This paper studies the difference of cooling capacity between air- and
motor section at z = 0 under different CHTCs. Area A is the cooling
water-cooled motors. The influences of different fan speed, cooling
water, and area B is the heat source. Fig. 16 (b), (d), and (f) show the
water speed and convection heat transfer coefficient on the cooling
temperature distribution of cooling water under different CHTCs.
performance are compared. For light UAV, the motor power is not large,
Although Fig. 15 shows that the maximum and minimum temperatures
but there are strict restrictions on the weight of UAV [33,34]. At this
in the motor winding were not significantly affected by the CHTC, we
time, air-cooled motor can be used, and the fan speed can be adjusted to
can observe from the cross-sectional temperature distribution that with
match the heating power of the motor. For some UAVs with large motor
the increase in the CHTC, the area of the high-temperature region in the
heating power and high temperature control requirements [35,36],
motor gradually decreased. In addition, the temperature distribution of
water-cooled motor can be used to improve the cooling capacity by
the cooling water in the motor varied with the increase in the CHTC. As
sacrificing weight. At the same time, increasing the speed of cooling
the CHTC increased, the volume of the low temperature region in the
water can also improve the cooling efficiency of water-cooled motors.
cooling water also increased slightly. Although the change of the CHTC
had a slight effect on the maximum and minimum temperature in the
6. Conclusions
winding of the water-cooled motor, the increase in the CHTC reduced
the proportion of the high temperature area in the water-cooled motor.
Two schemes, air cooling and water cooling, were designed for the
Therefore, increasing the CHTC increased the cooling efficiency of the
heat dissipation of the motor in a unmanned aerial vehicle. The
water-cooled motor to a certain extent.
maximum temperature of the air-cooled motor decreased with
increasing fan speed. When the fan speed was increased from 800 to
2000 rpm, the cooling capacity increased by 27.78%. The maximum
temperature of the air-cooled motor decreased with an increase in

13
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 16. Water-cooled motor temperature distribution for different convection heat transfer coefficients.

convection heat transfer coefficient. When the convection heat transfer When the inlet flow rate was 0.1 m/s, the maximum temperature of the
coefficient increased from 5 to 25 W/(m2⋅K), the maximum temperature motor winding was 335.8 K, which was significantly lower than that of
of the motor winding decreased by 4.43 K, and the cooling capacity the air-cooled motor. Compared with the motor without cooling, the
increased by 22.93%. The maximum temperature of the water-cooled temperature increase decreased by 47.89%. When the inlet flow rate
motor decreased as the inlet speed of the cooling water increased. increased from 0.1 to 0.9 m/s, the maximum temperature of the winding

14
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

Fig. 17. Comprehensive performance evaluation factor for different cooling methods.

decreased by 25.3 K, and the cooling capacity increased by 67.47%. [3] H. Zhou, Y. Wan, H. Ye, M. Jiang, J. Wang, Real-time energy-efficient optimal
control of high-speed electric train, Control. Eng. Pract. 112 (2021), 104825.
When the velocity of cooling water was 0.7 m/s, the comprehensive
[4] X. Li, K. Sun, F. Li, General optimal design of solar-powered unmanned aerial
performance of the water-cooled motor reached the highest level. vehicle for priority considering propulsion system, Chin. J. Aeronaut. 33 (8) (2020)
Increasing the convection heat transfer coefficient slightly reduced the 2176–2188.
highest temperature in the water-cooled motor; however, the reduction [5] Y. Xu, W. Zhu, J. Li, L. Zhang, Improvement of endurance performance for high-
altitude solar-powered airships: a review, Acta Astronaut. 167 (2020) 245–259.
was not significant. When the convection heat transfer coefficient [6] N. Thapa, S. Ram, S. Kumar, J. Mehta, All electric aircraft: a reality on its way,
increased from 5 to 25 W/(m2⋅K), the maximum temperature of the Mater. Today Proc. 43 (2021) 175–182.
winding decreased by 1.04 K, and the cooling capacity only increased by [7] Y. Gai, M. Kimiabeigi, Y. Chong, J. Widmer, M. Popescu, X. Deng, Cooling of
automotive traction motors: schemes, examples, and computation methods, IEEE
2.77%. A comprehensive performance evaluation factor is proposed to Trans. Ind. Electron. 66 (3) (2019) 1681–1692.
determine the condition that uses an air-cooled motor or a water-cooled [8] B. Melka, J. Smolka, J. Hetmanczyk, P. Lasek, Numerical and experimental analysis
motor. The research in this paper can provide some guidance for the of heat dissipation intensification from electric motor, Energy 182 (2019) 269–279.
[9] C. Kim, K.S. Lee, S.J. Yook, Effect of air-gap fans on cooling of windings in a large-
motor design on UAV in the future. capacity, high-speed induction motor, Appl. Therm. Eng. 100 (2016) 658–667.
[10] M. Kang, H. Wang, L. Guo, T. Shi, C. Xia, Self-circulation cooling structure design
CRediT authorship contribution statement of permanent magnet machines for electric vehicle, Appl. Therm. Eng. 165 (2020),
114593.
[11] S. Hyeon, C. Kim, K.S. Lee, Thermal enhancement of an air-cooled motor with a
Min Chang: Methodology, Validation, Formal analysis, Investiga­ flow guide, Int. J. Heat Mass Transf. 183 (2022) 22228.
tion, Writing – original draft. Bingzhu Lai: Methodology, Validation, [12] E. Galloni, P. Parisi, F. Marignetti, G. Volpe, CFD analyses of a radial fan for electric
motor cooling, Therm. Sci. Eng. Prog. 8 (2018) 470–476.
Formal analysis, Investigation, Writing – original draft. Hui Wang: [13] Z. Pei, J. Zhao, J. Song, K. Zong, Z. He, Y. Zhou, Temperature field calculation and
Writing – review & editing, Formal analysis, Supervision. Junqiang Bai: water-cooling structure design of coreless permanent magnet synchronous linear
Writing – review & editing. Zhaoyong Mao: Writing – review & editing. motor, IEEE Trans. Ind. Electron. 68 (2) (2021) 1065–1076.
[14] H. Chiu, J. Jang, W. Yan, R.B. Shiao, Thermal performance analysis of a 30 kW
switched reluctance motor, Int. J. Heat Mass Transf. 114 (2017) 145–154.
[15] F. Han, H. Guo, X. Ding, Design and optimization of a liquid cooled heat sink for a
Declaration of Competing Interest motor inverter in electric vehicles, Appl. Energy 291 (2021), 116819.
[16] B. Ge, J. Zhang, D. Tao, Temperature prediction and cooling structure optimization
None. of explosion-proof high pressure water-cooled double speed motor, Energy Rep. 8
(2022) 3891–3901.
[17] A. Nikbakhsh, H.R. Izadfar, M. Jazaeri, Classification and comparison of rotor
Data availability temperature estimation methods of squirrel cage induction motors, Measurement
145 (2019) 779–802.
[18] I. López, E. Ibarra, A. Matallana, J. Andreu, I. Kortabarria, Next generation electric
Data will be made available on request. drives for HEV/EV propulsion systems: technology, trends and challenges, Renew.
Sust. Energ. Rev. 114 (2019), 109336.
Acknowledgments [19] J. Jang, H. Chiu, W. Yan, M.C. Tsai, P. Wang, Numerical study on electromagnetics
and thermal cooling of a switched reluctance motor, case stud, Therm. Eng. 6
(2015) 16–27.
This work was supported by the China Postdoctoral Science Foun­ [20] A.E.L. Hocine, S. Poncet, H. Fellouah, CFD modeling and optimization by
dation (No. 2021M692632 and No. 2022T15532). metamodels of a squirrel cage fan using OpenFoam and Dakota: ventilation
applications, Build. Environ. 205 (2021) 08145.
[21] W. Yan, H. Teng, C. Li, M. Ghalambaz, Electromagnetic field analysis and cooling
References system design for high power switched reluctance motor, Int. J. Numer. Methods
Heat Fluid Flow 29 (5) (2019) 1756–1785.
[22] W. Chen, Y. Ju, D. Yan, L. Guo, Q. Geng, T. Shi, Design and optimization of dual-
[1] Z. Wang, J. Zhou, G. Rizzoni, A review of architectures and control strategies of
cycled cooling structure for fully-enclosed permanent magnet motor, Appl. Therm.
dual-motor coupling powertrain systems for battery electric vehicles, Renew. Sust.
Eng. 152 (2019) 338–349.
Energ. Rev. 162 (2022), 112455.
[23] A.F. Mills, Heat Transfer, 2nd ed., Prentice-Hall, New York, 1999.
[2] M.L.D. Klerk, A.K. Saha, Performance analysis of DTC-SVM in a complete traction
motor control mechanism for a battery electric vehicle, Heliyon 8 (4) (2022),
e09265.

15
M. Chang et al. Applied Thermal Engineering 225 (2023) 120226

[24] Y. Zhang, X. Zhang, Y. Li, M. Chang, J. Xu, Aerodynamic performance of a low- [30] X. Wang, B. Li, G. Gerad, K. Huang, I. Stone, S. Worrall, Y. Yan, A critical review on
Reynolds UAV with leading-edge protuberances inspired by humpback whale thermal management technologies for motors in electric cars, Appl. Therm. Eng.
flippers, Chin. J. Aeronaut. 34 (5) (2021) 415–424. 201 (2022), 117758.
[25] R. Cabello, A.E.P. Popescu, J.B. Ruiz, D.C. Cantarell, J. Llorens, Heat transfer in [31] X. Li, K. Sun, F. Li, General optimal design of solar-powered unmanned aerial
pipes with twisted tapes: CFD simulations and validation, Comput. Chem. Eng. 166 vehicle for priority considering propulsion system, Chin. J. Aeronaut. 33 (8) (2020)
(2022), 107971. 2176–2188.
[26] A. Muhammad, Numerical analysis of friction factor for a fully developed turbulent [32] Y. Zhang, J. Peng, R. Yang, L. Yuan, S. Li, Weight and performance optimization of
flow using k-ε turbulence model with enhanced wall treatment, Beni-Suef U J. rectangular staggered fins heat exchangers for miniaturized hydraulic power units
Basic 3 (4) (2014) 269–277. using genetic algorithm, Case Stud. Therm. Eng. 28 (2021), 101605.
[27] H. Ma, N. Cai, L. Cai, F. Si, Effects of the forced convection induced by assistant [33] M. Gadalla, S. Zafar, Analysis of a hydrogen fuel cell-PV power system for small
fans on the thermal performance of an indirect dry cooling system, Case Stud. UAV, Int. J. Hydrogen Energ. 41 (15) (2016) 6422–6432.
Therm. Eng. 35 (2022), 102141. [34] D. Sziroczak, D. Rohacs, J. Rohacs, Review of using small UAV based
[28] I.B. Celik, U. Ghia, P.J. Roache, C.J. Freitas, Procedure for estimation and reporting meteorological measurements for road weather management, Prog. Aerosp. Sci.
of uncertainty due to discretization in CFD applications, J. Fluid Eng. Trans. ASME 134 (2022), 100859.
130 (7) (2008), 078001. [35] H. Dong, L. Ruan, Y. Wang, J. Yang, F. Liu, S. Guo, Performance of air/spray
[29] P. Dubey, U. Ghia, L.A. Turkevich, Numerical investigation of sheath and aerosol cooling system for large-capacity and high-power-density motors, Appl. Therm.
flows in the flow combination section of a baron Fiber classifier, Aerosol Sci. Eng. 192 (2021), 116925.
Technol. 48 (8) (2014) 896–905. [36] P. Fan, S. Huang, H. Wang, D. Luo, H. Li, M. Sun, Fundamental frequency region-
based thermal control of power electronics modules in high power motor drive,
Microelectron. Reliab. 88-90 (2018) 1242–1246.

16