Airflow Cooling Mechanism for High Power-Density Permanent Magnet Motor

High power-density electric machines present the benefits of high torque and speed. However, this generally comes with heating problems characterized by high temperatures that affect performance. Conventional approaches to address overheating are to include cooling fans or jackets within the stator core of the machine. This approach is challenging to implement in small-size high power-density machines. In this article, a cooling mechanism integrated in the rotor of a high power-density permanent magnet motor is proposed. It comprises a set of six holes, shrouded within a hollow shaft. The mechanism is based on conditioning air due to a centrifugal force that is produced by the rotational speed of the rotor from the inlet. A theoretical model based on flow resistance network is proposed to analyze the airflow rate. An analytical thermal model based on lumped parameter thermal network is developed to analyze the effect of the flow rate on the temperature distribution in the motor. Also, a simulation analysis model was conducted using computational fluid dynamics to analyze the effect of air flowing in the motor. An experimental prototype is developed to verify, validate, and evaluate the proposed cooling model. The cooling system is effective in reducing temperatures from speeds above 6000 min−1.

Heat flux.Air-gap width.r r [m] Radius of rotor.L r [m] Length of rotor.K f [-] Coefficient of friction.b r [-] Roughness friction.N u , R e P r [-] Nusselt, Reynold, and Prandtl numbers, respectively.R ea , R er [-] Respective axial and radial Reynold's numbers.Qcu , Qmech [W] Copper and friction loss, respectively.

I. INTRODUCTION
There are many applications of mechatronics systems in which weight and size are the design criteria.Wearable robots are designed to be portable, mechatronics systems for agricultural purposes are designed to operate in different terrains, and industrial actuators are designed to be flexible [1], [2], [3], [4], [5], [6].These mechanical systems employ either pneumatic system, hydraulic system, or electric motor for actuation [2], [7].Electric motors are desirable for construction of compact mechatronics systems because they can be optimized to meet design criteria [8].Conversely, electric motors have relatively low output power compared to pneumatic and hydraulic actuators [9].High output power applications that utilize electric motors employ permanent magnet (PM) motors because of their high power-density and speed [10].Despite these merits, PM motors are faced with many challenges, the most critical of which are the losses generating heat in the motor [11].Heating directly affects the performance and life span of the electric machine.For instance, heating facilitate wearing of winding coil insulation and causes demagnetization of PMs.Power loss analysis performed in [12] reveals that the total heat generated in interior PM motors is higher than in surface mounted PM (SPM) motors, but comparing rotor heating for both machines reveal rotors in SPM motors have higher temperatures.This can be blamed on poor heat dissipation facility in the design of high power-density SPM motors.
Depending on the application, sophisticated cooling solutions for electric motors in general, and SPM machines, have been proposed in many literature.In [13], cooling of outer-rotor SPM machine based on heat pipe is proposed.Heat is conducted through the pipes to aluminum frame heat sink.The heat absorbed by the sink is radiated to the surrounding by forced convection using external fan system.In direct fan-cooling technique (FCT) of electric machines, an impeller-type fan is mounted directly at the rear end of the rotor [14], [15].Fan-cooling system is very efficient and is powered by the rotation of the motor and is highly influenced by the speed [16], [17].Also, liquid cooling techniques (LCT), such as cold plates and hollow conductors, have been investigated for direct cooling in electric motors [18], [19].A major disadvantage of these techniques is the increase in size and weight of the motor to accommodate the cooling mechanism.The motor structure design must take into consideration the use of pipes, fan impellers, and liquid coolants in the case of cold plates and hollow conductors.
Much of the weight in electric machines is concentrated in the stator and rotor core.There are cooling techniques designed by structural adjustment of stator or rotor iron.In [20], a hollow rotor shaft design is investigated for cooling in SPM motors.During operation of the motor, air flows through the hollow shaft from one end to the outlet at the other end.This technique improves the rate of convection in the rotor but is disadvantageous because the thermal distance between the high loss regions of the stator assembly and the cooling air is large.

TABLE 1. Specifications of SPM Motor
To solve the aforementioned problems, this article proposes a fan-cooling mechanism based on rotor structural design.The solution employs a direct air-conditioning system which is powered by the rotation of the motor.The proposed airflow cooling SPM motor is shown in Fig. 1.The cooling system has air inlet and outlet sections designed on the rotor and front housing, respectively.The merit of the system is that no external impeller is required.The size of the machine is maintained and the weight is considerably reduced accordingly.In this research, a 2 kW, 15 000 min −1 , 3-phase SPM motor is used as experimental case study.The detail specification of the experimental SPM motor is shown in Table 1.The airflow cooling mechanism is evaluated in this article.
In forced cooling of electric machines, the rate of cooling increases as the mass flux of coolant increases [21], [22].The mass flow rate generated in the SPM motor by the airflow cooling system is analyzed.First, the air flow field of the SPM motor without the cooling system is investigated.A comparative study is made with the air flow field analysis of the motor in presence of the cooling mechanism.The result of the comparison shows that mass flow rate of air is generated in the motor by the cooling mechanism.The term flow rate in this article refers to the volumetric flow rate except otherwise stated.Thermal analyses of the SPM motor reveal the cooling system maintains the temperature of the winding and PMs within operating range during operation of the motor.The main contributions of this article are as follows.
1) The design of a cooling system with simple costeffective structure which is easy to implement and can achieve high cooling power at high rotational speed.2) Also, this article proposes an analytical model to describe the centrifugal force produced by the rotation of the rotor.
3) The implementation of an experimental prototype to verify and validate the proposed cooling system model.The rest of this article is organized as follows.The geometric design considerations of the proposed cooling mechanism are discussed in Section II.A theoretical model of the air flow and temperature field of the SPM motor with the airflow cooling mechanism is developed in Sections III and IV, respectively.In Section V, numerical computational fluid dynamics (CFD) simulation is performed on the CAD model of the SPM motor with and without the proposed cooling mechanism.Finally, the experimental verification of the air flow field is discussed in Section VI.In Section VII, the merits of the thermal design are clarified in comparison with selected state-of-the-art cooling systems.Finally, Section VIII concludes this article.

II. AIRFLOW COOLING SYSTEM DESCRIPTION
The airflow cooling mechanism consists of air inlet section, which is designed at the rear end of the rotor shaft, as shown in Fig. 2, and air outlet section which is on the front cover.The air inlet section employs a ducted fan-like system, which comprises a set of six holes shrouded within a hollow cylindrical ventilation air duct, as shown in Fig. 2(a).The design considerations of the inlet holes include easiness of manufacturing by using standard drilling machines.The internal structure of the duct design is shown in Fig. 2(b) and in this article, During operation of the motor, a centrifugal force is produced by the rotation of the rotor.This force exerts pressure on the mass of air which accumulates in the ventilation duct.The compressed air enters the motor through the inlet holes and circulates, as depicted in Fig. 3(a).The air outlet section is a set of six holes, each of 10.0 mm diameter designed on the front cover of the motor as shown in Fig. 1.The holes are placed toward the drive end with each hole positioned above the end-winding slot.Cooling is achieved when hot air from inside the motor flows through the outlet to the surrounding.The merit of the design is summarized as follows.
1) The mechanism is cost effective as no additional component is required.There is no need for external fan impellers as in the case of direct fan-cool systems.
2) It enable the design of compact high power-density motors.With no external component required to implement the air conditioning system, small size electric motors can be provided with ventilation support.
3) The thermal distance between hotspots in the motor and the coolant (air in this case) is highly reduced.During motor operation, the pressure generation unit produces high turbulent flow within the motor, which pushes air from the inlet duct into the motor and through all flow passages.Initial attempts to evaluate the airflow cooling mechanism are reported in [23], [24], [25].The lumped parameter thermal network (LPTN) model employed in these analyses did not include the node at fluid level.Also, only copper loss in the coils was considered and the flow rate was not evaluated.In this paper, the flow rate and temperature are estimated.The theoretical flow model is based on the flow network (FN) technique presented in [25], [26].

III. FLOW ANALYSIS
In forced cooling of electric machines, heat exchange occurs between heat generating components and the coolant.In principle, heat transfer rate in thermal systems is characterized by mass flow rate of heating and cooling mediums.The mass flow rate of coolant, having mass density ρ and flowing with velocity v through a surface of cross section A is related to the flow rate by where Q and ṁ are the volumetric and mass flow rates of the coolant, respectively.In the proposed SPM motor, the flow rate defined by ( 1) is due to the mechanical power of the rotor.In this article, the computation of the flow rate is based on the descritization of (1) by using FN technique.

A. FN MODEL
As shown in Fig. 3(b), a centrifugal force F is generated in the ventilation duct perpendicular to the axis of rotation.The force exerts a pressure P F on the air in the duct causing it to flow along the axis of rotation into the motor, as depicted in Fig. 3(a).The pressure is computed as where A D is the area of cross section of the ventilation air duct.The force generated is proportional to the square of the motor speed, ω, and is estimated as in [25] where ρ air is the density of air and the integral is computed over the volume of the ventilation duct, V , defined by the boundary z [z 0 , z 2 ] indicated in Fig. 3(b).The variation of the centrifugal force in relation to the depth of the ventilation duct and motor speed is shown in Fig. 4. It is deduced from the graph that the force: 1) increases with increase in motor speed and 2) reduces with increase in depth of the ventilation duct, i.e., the centrifugal force is stronger at z 2 than at z 0 .This characteristic of the force accounts for the change in velocity of the airflow in the duct.
The circulation of the air inside the SPM motor is constrained to hydrodynamic resistances, (or flow resistances), of the flow passages.The flow characteristic can be represented with an equivalent FN circuit [26], [27].In this research, the airflow is modeled with an equivalent two channel FN shown in Fig. 5.The circuit constitutes the flow loops in the inlet duct, air-gap, winding slot, and the outlet.Their respective hydrodynamic resistances are denoted as R in , R ag , R slt , and R out .As reported in [26], the flow resistance unit is described by where P is the pressure difference between the inlet and outlet of the flow and R H represents the flow resistance of the flow passage.The quadratic form of Q in (4) accounts for the turbulent nature of the flow.In purely laminar flow, Q is linear.The flow resistance is expressed as in [27] where μ is the dynamic viscosity of air, L and r are the length and radius of the flow passage, respectively.The pressure at the inlet and outlet are denoted as P in and P out , respectively.The boundary conditions of the flow at the inlet and outlet is subject to constant atmospheric pressure P atm .Therefore, they can be expressed as and P out = P atm .The flow rate is computed by applying Ohm's law-equivalent definition expressed in (4) to the flow circuit.This gives were z is the height of the ventilation duct and R is the effective flow resistance of the FN circuit.Each flow resistance is computed by using (5).The parameters used in the computation of ( 7) are listed in Table 1.

IV. THERMAL ANALYSIS
To model the heat transfer process in the proposed SPM motor, thermal analysis was conducted.The objective is to determine the temperature distribution in the motor.The determination of the temperature distribution is important to properly evaluate the effect of the proposed cooling mechanism as the machine operates at different speeds.
To achieve this, LPTN technique [28], [29], [30] was employed.The step-by-step procedure applied in the thermal analysis is as follows.
1) Identification of thermal nodes and construction of the LPTN network to determine the temperature distribution in the motor.2) Determination of the parameters of the LPTN network.
This include the identification and estimation of losses in the proposed motor, determination of heat conduction thermal resistances, and modeling of heat convection.3) Solving the network to obtain the temperature at each node.The analysis combines the advantages of analytical and numerical thermal modeling techniques, which are listed in [31].Summarized in Fig. 6, the losses are determined by finite element analysis (FEA) method [32] and the FN techniques are employed in the modeling of heat convection process in the motor.Finally, CFD simulation [33] is conducted with an equivalent model of the proposed motor to determine the rate of airflow in the machine and track the temperature changes with this flow.

A. ANALYTICAL MODELING
Suppose the temperature distribution in the SPM motor is defined by T .At steady-state, the heat flux generated in the motor is absorbed by the cooling air.The heat transfer process is described by where k is the effective thermal conductivity of the motor in (W/mK), q is the total loss in the motor in (W), h is the overall convective heat transfer coefficient in (W/m 2 K), A is the area of cross section of the motor, and T f and T a are the temperatures of motor frame and ambient air, respectively.Generally, the left-hand side of ( 8) is the heat conduction in presence of heat generation sources and the right-hand side is the convective heat transfer.
The analytical model to the heat conduction and convection problem in the SPM motor is obtained by discretizing (8) using LPTN method.To simplify the solution, the following assumptions were considered in the heat transfer process.
1) The heat transfer in the motor is due to conduction in the solid components and convection due to the airflow.Heat transfer by radiation is not considered.This is because the proposed cooling system implements forced air cooling method.As a result, the thermal resistance due to radiation is much larger than convection [34].
2) The temperature of a solid component is uniformly distributed.As a result, the radial and axial heat flow is considered in evaluating the mean temperature in the component.The proposed LPTN is shown in Fig. 7.The network consists of 14 temperature nodes, with each node connected to successive nodes through thermal resistances.The description of the nodes and thermal resistances are given in Tables 2  and 3, respectively.The thermal resistances are distinguished as either conduction or convection in figure.Power loss sources in the motor are indicated as heat nodes in the LPTN.The losses in the motor are categorized and estimated in Section IV-B.

B. FEA AND LOSS ANALYSIS
To conduct the loss analysis of the proposed motor, the main power loss sources were considered.As indicated in Fig. 7, The dominant power loss sources in the motor are due to copper loss, Qcu , in the winding, iron losses in the stator core, Qs , tooth, Qt , rotor, Qr , and the air friction loss, Qmech , generated by airflow dynamics in the air gap between the stator tooth and the surface of the magnet padding.

1) IRON LOSS ANALYSIS
The computation of the iron losses in the stator and rotor core are achieved through FEA analysis using JMAG-Designer-ver19.The procedure for conducting the analysis is illustrated in Fig. 8.For accurate analysis, the magnetic flux distribution in the machine is determined through a magnetic field  transient analysis.Upon determination of the magnetic flux distribution, loss analysis is conducted to compute the joule loss and hysteresis losses.The iron loss computation in the motor is then obtained as a summation of the joule and hysteresis losses.The iron loss density distribution is shown in Fig. 9(a).The iron loss is maximum at the stator tooth and is above 67.5 kW/m 3 .This is indicated with the red arrow hotspot in the figure.The iron loss in the stator and rotor is shown in Fig. 9(b).The iron loss in each component is the sum of the joule and hysteresis losses.

2) WINDING COPPER LOSS ANALYSIS
As already indicated above, the dominant loss source in the SPM motor is due to copper losses in the phase windings.This is as a result of joule heating from the high ac current (40 Arms) flowing in the coils.As reported in [35], the total joule loss Qcu in the proposed motor due to current i ac flowing in the winding is expressed as where R el is the electrical resistance of the phase winding and n is the number of phases.

3) AIR FRICTION LOSS
When in operation, the air circulating in the air-gap of the motor is in direct contact with the magnet padding and the teeth of the stator.In highly turbulent flows, the air moves with high velocity and because of increase in flow rate due to increase in speed, impact momentum increases in the air-gap.Energy is released in the form of heat in the process.A general equation for the air friction loss in the air-gap is expressed as in [35] Qmech = π r 4 r L r b r K f ρ air ω 3 (10) where b r is roughness coefficient of the component of rotor assembly in contact with the air-gap air [i.e., the magnet padding, as shown in Fig. 2(a)], K f is the friction coefficient on the surface of the component, L r is its axial length, and r r is the outer radius of the rotor assembly structure.The friction coefficient is calculated as 4R ea R er 2 0.38 (11) where R er and R ea are the Reynold's number in the radial and axial direction to the flow, respectively.The nature of the airflow is determined by the dimensionless Reynold's numbers.They can be expressed as follows: R er = ρ air μ ωr r H ag (12) and where H ag and L c is the radial and characteristic length of the air-gap, respectively, and v is the local velocity of air in the air-gap which is obtained from (1).The variation of the air-friction loss in relation to speed of rotation and flow rate is shown in Fig. 10.In the proposed SPM motor, the loss due to friction in the air-gap increases with increase in speed of rotation.

C. THERMAL RESISTANCE FORMULATION
The thermal resistances of the SPM motor are computed from the geometric and material properties of the motor components.The computation was performed using the correlations for determining thermal resistance of standard geometric shapes that are peculiar to electric machines [28], [29], [30].For heat transfer in a metal of cross section A and length L, the thermal resistance due to conduction and convection are defined, respectively, as and

D. HEAT CONVECTION MODELING
Heat transfer rate by convection in the motor is determined by the heat convection coefficient h in (15).Heat convection is caused by the circulation of air in the motor.In forced convection, the air movement is due to an external force.This is achieved in this article by the centrifugal force generated by the airflow cooling mechanism.Various empirical correlations for modeling heat transfer coefficient for both forced and natural convection are studied in [36].
In general, the empirical correlation for estimating heat transfer coefficient by forced air cooling on a surface of characteristic length L c is given as where N u is the Nusselt number and k air is the thermal conductivity of air.The Nusselt number is defined as where R e and P r are the Reynold's and Prandtl numbers, respectively.As discussed in Section IV-B3, the nature of the flow is determined by the local velocity of the air movement on the surface.The Reynold's number is determined from the velocity as in (13).The Prandtl number accounts for the effect of natural convection component of the flow.The dimensionless constant is calculated as where c air is the specific heat capacity of the cooling air.

E. TEMPERATURE COMPUTATION
Theoretical analysis of the steady-state temperature distribution in the proposed SPM motor is reported in this section.The temperature distribution is computed by solving the LPTN circuit shown in Fig. 7.A program was written in C language to compute the thermal resistance values, and the thermalequivalent nodal analysis technique was applied to the thermal circuit and solved using MATLAB software.The temperature distribution was obtained as The temperature rise at each node was calculated for different speeds of rotation.The result of the delicate parts of the stator assembly including the cover is shown in Fig. 11.It is observed that the temperature distribution in the selected nodes reduces as speed increases, which is a direct consequence of the increase in flow rate as speed increases.

V. NUMERICAL EVALUATION
In this section, the flow rate and thermal analysis models are verified by using CFD simulation.The simulation employed SOLIDWORKS flow simulation tool.To evaluate the cooling mechanism, two case studies of the SPM motor were simulated and their results were compared.The different case studies are defined as follows.
r CASE 1: SPM motor with airflow cooling system.r CASE 2: SPM motor without airflow cooling system.
To compute the flow rate, an internal analysis type was performed on 3-D CAD model of the SPM motor in CASE 1.The configuration settings during the flow simulation studies are shown in Table 4.The external flow analysis study was performed to analyze the steady-state temperature distribution in the SPM motor of both CASE 1 and CASE 2.

A. AIR FLOW RATE COMPUTATION
The flow rates at the inlets and outlets were computed during simulation of the air flow in the SPM motor.The fluid (i.e., air in this case) volume was created in the internal analysis studies of the SPM motor in CASE 1.This was done by setting the pressure conditions of the flow at the inlet and outlet to standard atmospheric pressure.Also, the k − turbulence model was selected for the condition of the turbulence flow.To verify the analytical flow characteristic define by (7), the simulation study was configured as a parametric study in which the speed of rotation was varied from 1000 to 15 000 min −1 .The result of the airflow analysis reveals two important objectives as follows.
1) The cooling system design generates air flow in the SPM motor due to the rotation of the rotor.As shown in Fig. 12, during operation, the air ventilation duct acts like a vortex tube.This generates high turbulence airflow in the motor.2) The flow rate generated in the motor increases as the speed of rotation increases.Comparison of the flow rate computed at one outlet during simulation and in theoretical calculation given by ( 7) is shown in Fig. 13.At the maximum rated speed of 15 000 min −1 , the flow rate in a single outlet is 4.2 and 5.0 L/min in simulation and as predicted in theory, respectively.

B. THERMAL COMPUTATION
The thermal design objective of the SPM motor is sustainability.The motor is required to operate at normal and peak driving torque conditions while maintaining hotspot temperatures within operating range.The operating temperature range for the proposed SPM motor is 120 °C.Steady-state temperature analysis of the SPM motor was performed at different load values during the CFD simulation.The contour map of the steady-state temperature distribution in CASE 1 is shown in Fig. 14.The contour map shows simulation result at maximum rotation of 15 000 min −1 under full load condition.It illustrates that the minimum temperature in the SPM motor is exhibited in the rotor region which is 58.32 °C during full load operation.This is safe for the PMs as the highest loss in the SPM motor is generated during full load operation.The map also illustrates the highest temperature in the motor is exhibited in the stator assembly region.The high temperature in this region is due to the high loss generated in the winding coils due to the current.
By comparing the temperature of the end winding computed in the simulation of the motor in CASE 1 and CASE 2, the influence of the cooling system on the temperature of the motor is further clarified.This is shown in Fig. 15.According to the contour map in CASE 1 shown in Fig. 15(a), at speeds above 6000 min −1 , the motor can be operated above 50% load while maintaining the end winding temperature within operating range.For instance, at full load operation, the temperature of the end winding is 72.2 °C at 15 000 min −1 rotation speed.This is safe for the motor as it is within the operating range.The contour map in CASE 2 shown in Fig. 15(b) clearly demonstrates the effect of the absence of the cooling system.Without the cooling system, the SPM motor cannot be operated above 20% load irrespective of the speed.At full load, the temperature of the end winding is 129.4 °C at 15 000 min −1 rotation speed.This is above the operating temperature.This comparison demonstrates that the cooling system is efficient.The temperature distribution at selected nodes in CASE 1 is shown in Fig. 16.The result shows that the cooling system is highly influenced by the speed of rotation of the motor.This is in accordance with the prediction of the LPTN analysis shown in Fig. 11.As the speed increases, the flow rate increases.

VI. EXPERIMENTAL VERIFICATION
To further investigate the significance of the airflow cooling system and to validate the theoretical and simulation results, a prototype SPM motor with the proposed cooling mechanism is manufactured.This is shown in Fig. 17.The main parameters of the prototype machine are listed in Table 1.The internal structure of the prototype SPM motor is compact and is shown in Fig. 17  the motor are copper-clad aluminium wire with AIW insulation type having maximum operating temperature of 220 °C.The maximum operating temperature of the PMs used in the prototype motor is 150 °C.The high power-density prototype motor is equipped with two K-type thermocouple wires with each connected to the end winding of two of the phase coils.During the experiment, the temperature of the end winding of the motor is monitored through the thermocouples.The complete coupling of the prototype motor with the air outlets on the cover is shown in Fig. 17(b).

A. EXPERIMENTAL SETUP
Multiple experiments were conducted by varying the speed of rotation.An enhanced filed-oriented control (FOC) method including disturbance observer (DOB) was implemented in the experiment to control and achieve the different required speeds.This is shown in Fig. 18.The DOB enhances the speed controller by estimating and compensating load disturbances in the experiment to satisfy speed requirements.By controlling the speed, the torque was indirectly adjusted to corresponding experimental conditions.

VOLUME 5, 2024
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.The MTL MEH-30-300PE rotary encoder was used as the speed sensor.The torque was constantly monitored by using the UNIPULSE UTM II 2 Nm torque sensor.The experimental setup is shown in Fig. 17(c).To measure the flow rate, an SMC PFM525 digital flow switch (±5% FS) was connected to each air outlet.The experiment was conducted under load and no load conditions.The MAGTROL HB-250 hysteresis brake was used as the load.The FOC control and data processing program were implemented in a DSP control board of MY-WAY PE-EXPERT4 digital control systems.The complete experimental laboratory is shown in Fig. 17(d).The setup comprised of data acquisition board for the flow switches and thermocouples.All data processing was performed in PE-ViewX software installed in the lab PC.The experimental conditions are shown in Table 5.

B. RESULT DISCUSSION 1) FLOW RATE MEASUREMENT
The flow rate was measured at the outlets under no load condition.The experimental setup is shown in Fig. 17(c).Measurements were taken from 1000 to 12 000 min −1 .Six flow sensors were connected to the six outlets.Flow rate measurement at 12 000 min −1 is shown in Fig. 19.The corresponding flow rates for each outlet indicated as f1, f2, f3, f4, f5, and f6 in figure is 3.98 L/min.This result shows that the airflow design is symmetric.The total flow rate denoted by "f" in the figure is 23.88 L/min.
Comparison of the flow rates in theory, simulation, and experiment is shown in Fig. 20.As the speed changes from 1000 to 12 000 min −1 , the flow rate increases from 0.98 to 3.98 L/min, respectively.This illustrates that the flow rate is influenced by the speed of rotation.This tendency is similar in both the simulation and theoretical model of the proposed airflow system.
There is correlation between experimental and theoretical flow rate measurement at higher speeds.From 9000 to 12 000 min −1 the flow rate measurement in both theory and experiment changes from 2.99 to 3.98 L/min.However, there is discrepancy with simulation result which are lower for all speeds.This is because the measurement points for the simulation and inside the flow sensor is slightly different.In the simulation, a lid is placed at a distance which is equivalent to the length of the connection tube to the flow sensor.

2) TEMPERATURE MEASUREMENT
The flow switches were removed when conducting the temperature experiment.Based on the theoretical result shown in Fig. 11 and evident in the simulation contour map shown in Fig. 14, the highest temperature in the SPM motor is exhibited in the end winding.The effectiveness of the cooling system is verified by observing the end winding temperature under load condition.The thermal test was conducted at 50% load.The load was monitored by using the torque sensor and the measurement is shown in Fig. 21(a).The peak-to-peak phase currents of the motor at 50% load recorded during the thermal test is shown in Fig. 21(b).
Multiple experiments were conducted by changing the speed from 7000 to 10 000 min −1 .Each experiment was conducted for an extended time of 600 s.The time is long enough to get the experiment to steady state and properly evaluate the temperature of the machine.The experimental result is shown in Fig. 22.The temperature of the machine decreases as the speed of rotation increases.This is in accordance with the CFD simulation results as is seen in Fig. 23.Both results further confirms the prediction of the analytical LPTN model introduced in Section IV.The SPM motor benefits from the high flow rate at higher speeds due to the airflow cooling mechanism.The airflow in the motor improves the rate of convection.A Ti100 infrared thermal imager was used to monitor the temperature of the air in and out of the motor.The temperature recorded by the thermal imager at 10 000 min −1  is shown in Fig. 24.The room temperature is 22.9 °C as shown in Fig. 24(a) and the temperature of the air coming out of the motor through the outlet is 66.2 °C, as shown in Fig. 24(b).This illustrates that the cooling mechanism is effective.

VII. GENERAL EVALUATION OF THE COOLING SYSTEM
A qualitative and quantitative comparative study of the proposed airflow cooling mechanism with other state-of-the-art cooling systems is shown in Table 6.In general, cooling systems are categorized as follows.1) Self-cooling techniques (SCT): Heat transfer in totally enclosed nonventilating (TENV) systems is due to conduction in the solid components of the system.Most TENV electric motors are provided with heat sink for self-cooling of the machines.2) Liquid cooling techniques: These employ passive vents in the motor or jackets on the motor housing, which contain coolant liquids.Various fluids are used as coolants in electric machines most common of which include air, water, and oil.3) Fan-cooling techniques: In this category, an impellertype fan is used to blow air into the motor to increase the rate of convection.The choice of cooling system depends on the application of the electric machine.As shown in Table 6, the airflow cooling system proposed in this article is advantageous in weight-critical applications.Its implementation requires no extra components unlike the system in [13], [15], [18], [19], [20], and both rotor and stator cooling are achieved unlike the system in [13], [18], [19], [20].Moreover, the power-density per volume of the proposed system is relatively high compared to the other systems.Also, with the exception of the system in [13], the proposed mechanism has high power-density per mass compared to other cooling systems.Also, it is observed

r
[m] Radius of flow passage.R cond , R conv [K/W] Thermal resistance.G T [W/K] Thermal conductance.R el [ ] Phase winding resistance.i ac [A] Rated current.h [W/m 2 K] Heat transfer coefficient.μ [Pa.s]Dynamic viscosity of air.H ag[m]

FIGURE 1 .
FIGURE 1. SPM motor structure with proposed airflow cooling design.

FIGURE 2 .
FIGURE 2. Rotor design of proposed airflow cooling mechanism.(a) Rotor assembly structure.(b) Section view of rotor core design.(c) Schematic design of ventilation duct.

FIGURE 3 .
FIGURE 3. Pressure generation unit.(a) Depiction of air circulation in the motor.(b) Coordinate system of centrifugal acceleration.

FIGURE 4 .
FIGURE 4. Variation of centrifugal force with depth of ventilation air duct and motor speed (assuming z 0 = 0.0 m).

FIGURE 5 .
FIGURE 5. Flow resistance network of the air circulation in the proposed SPM motor.

FIGURE 7 .
FIGURE 7. LPTN thermal network model for discretizing the heat equation of the SPM motor.

FIGURE 10 .
FIGURE 10.Variation of air friction loss in relation to the speed of rotation and flow rate.

FIGURE 11 .TABLE 4 .
FIGURE 11.LPTN computation of the temperature rise in stator assembly and front cover of the SPM motor at different rotation speeds.

FIGURE 12 .FIGURE
FIGURE 12. CFD analysis of the airflow vorticity in the SPM motor at 15 000 min −1 rotation speed.

FIGURE 14 .
FIGURE 14. Steady-state CFD analysis of the temperature distribution in the SPM motor at 15 000 min −1 rotation speed.

FIGURE 15 .FIGURE 16 .
FIGURE 15.CFD analysis of temperature distribution in end winding at different speeds of rotation and load.(a) Steady-state analysis in CASE 1.(b) Steady-state analysis in CASE 2.
(a).The coils of the phase windings of

FIGURE 17 .
FIGURE 17. Prototype of proposed SPM motor with airflow cooling system.(a) Internal structure of prototype machine.(b) Coupled prototype motor showing air outlets.(c) Testbench setup.(d) Experimental lab setup.

FIGURE 18 .
FIGURE 18. Block diagram of implementation of FOC with DOB.

FIGURE 20 .
FIGURE 20.Comparison of flow rate computation method.

FIGURE 22 .
FIGURE 22. Experimental result of temperature rise in end winding at 50% load.

FIGURE 23 .
FIGURE 23.CFD analysis at 50% load of end winding of the SPM motor with the airflow mechanism.