Thermal Analysis of a Gramme‐ring‐winding High‐speed Permanent‐magnet Motor for Pulsed Alternator Using CFD

Yuan Wan,Qiang Li,Jian Guo,Shumei Cui

DOI: https://doi.org/10.1049/iet-epa.2020.0086

2020-01-01

IET Electric Power Applications

Abstract:IET Electric Power ApplicationsVolume 14, Issue 11 p. 2202-2211 Research ArticleFree Access Thermal analysis of a Gramme-ring-winding high-speed permanent-magnet motor for pulsed alternator using CFD Yuan Wan, Corresponding Author Yuan Wan wy19870328@163.com orcid.org/0000-0001-5150-9454 School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorQiang Li, Qiang Li School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorJian Guo, Jian Guo School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorShumei Cui, Shumei Cui Department of Electrical Engineering, Harbin Institute of Technology, Harbin, 150006 People's Republic of ChinaSearch for more papers by this author Yuan Wan, Corresponding Author Yuan Wan wy19870328@163.com orcid.org/0000-0001-5150-9454 School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorQiang Li, Qiang Li School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorJian Guo, Jian Guo School of Automation, Nanjing University of Science and Technology, Nanjing, 210094 People's Republic of ChinaSearch for more papers by this authorShumei Cui, Shumei Cui Department of Electrical Engineering, Harbin Institute of Technology, Harbin, 150006 People's Republic of ChinaSearch for more papers by this author First published: 12 August 2020 https://doi.org/10.1049/iet-epa.2020.0086Citations: 3AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Due to the merit of very-short end turns, Gramme-ring windings are widely used in high-speed permanent-magnet motors (HSPMMs). This study investigated the thermal issues of a 300 kW, 12,000 rpm Gramme-ring-winding HPSMM with housing water jacket and forced air cooling, based on the computational fluid dynamics (CFD) and numerical heat-transfer calculations. The reasons that cause the overheat of the prototype in the experimental test are thoroughly analysed. The high wind resistance of air gap as the ventilation duct and the large interface gap between the stator lamination and housing water jacket is thought to be the main reasons. Then improvement of the ventilation structures is studied to decrease the wind resistance. Besides, the modification of water channels is also investigated to avoid the influence of a large interface gap on the water cooling performance. Finally, a new cooling structure, combining the slot ventilation ducts and inside water channels, is proposed for the HSPMM with Gramme-ring windings. With the new cooling structure, the temperature rise of the motor is decreased, and the power density can be increased. 1 Introduction Pulsed alternator, a kinetic energy pulsed power supply, is widely thought to promise for driving electromagnetic guns [1, 2]. An high-speed permanent-magnet motor (HSPMM) is necessary to accelerate the alternator rotor to store the mechanical energy. As a mobile application, the weight and volume of the power are of great importance. Take the considerable outer diameter of an alternator into consideration, it is favourable to select a flat-configuration HSPMM, to make the system axially more compact. Therefore, an HSPMM with Gramme-ring windings was designed for this application, mounted on a common shaft of the alternator [3]. HSPMM is typically characterised by a small footprint and high power density. Due to the small heat dissipation area and high loss density, heating becomes one of the biggest problems. Besides, with Gramme-ring windings, the difficulty in the motor cooling will be further enhanced, because of extra copper loss of windings in the outer slots of the stator core. Moreover, the mechanical strength of the surface-mounted HSPMM rotor is not only threatened by large centrifugal force but also massive impact imposed by alternator discharge for this application [4-7]. Herein a considerable thick carbon-fibre retaining sleeve is mounted outside of the rotor to ensure the mechanical strength. However, regarding the poor thermal conductivity of the sleeve, it is not favourable for the heat emission of the rotor [8, 9]. Consequently, overheat of the rotor may cause irreversible demagnetisation of permanent magnets [10]. And the retaining sleeve also has temperature limits. Accordingly, it is of great significance to study thermal behaviour. Numerical techniques such as lumped parameters thermal network (LPTN) and computational fluid dynamics (CFD) are widely employed to predict and improve the thermal performance of HSPMMs [11-13]. While in LPTN, because the distribution of airflow is very complex inside of the HSPMM, it poses a great challenge to calculate the convection heat transfer coefficients for different surfaces inside of the motor based on empirical formulas. Dong et al. took the CFD approach to obtain the flow velocities in a proposed LPTN model for a 75 kW, 36,000 rpm HSPMM with forced convection cooling. Nevertheless, the advantage of time-saving will be weakened for LPTN method, since extra time is necessary for preparation and simulation of the CFD model. In recent years, despite more time-consuming, CFD combined with numerical heat-transfer calculation is increasingly utilised to investigate the thermal issues of HSPMM, with the improvement of computational resources. Because the airflow can be predicted precisely, and there is no need to especially calculate the convection boundaries. Using CFD, Dong et al. [14] studied the optimisation of the cooling structure for the aforementioned HSPMM, to allow more cooling fluid to flow through the air gap. Li et al. established a CFD analysis model for an oil-cooled 117 kW, 60,000 rpm HSPMM to predict the distribution of the cooling medium and its cooling performance. By comparison with the measured data, the thermal analysis results were thought with reasonable accuracy [8, 15]. Huang and Fang [16] analysed the distribution of fluid in the cooling air channels for a 100 kW, 32,000 rpm HSPMM with axial pressurised air flowing in the air gap, implemented in the commercial software ANSYS CFX. Ding et al. have been dedicated to analysing the three-dimensional (3D) CFD model of a PMSM with cooling ribs and junction boxes with the use of ANSYS Fluent software package [17, 18]. The above studies reveal the effectiveness of CFD in solving the fluid and thermal issues of HSPMM. However, the modelling and simulation usually have a close relationship with the cooling structures of the motor. The housing water jacket and forced air flowing through the air gap are adopted for the cooling of an HSPMM with Gramme-ring windings in this paper. Owning to the imperfections in the contact surfaces, the interface gap between the stator lamination and the housing water jacket is inevitable during the manufacturing process. Although the value of the interface gap is very low, it is still of great importance on water cooling performance, because most losses cross this surface [19, 20]. Taking the slots cutting on the outer surface of the laminated stator into consideration, the problem of interface gap becomes extremely serious. However, few papers investigate the influences of the interface gap for the Gramme-ring windings HSPMM. Moreover, the test of the prototype indicates the velocity of forced air flowing through the air gap is very low and the temperature rise is much higher than expected. Hence, it is necessary to clarify the reasons that result in the small velocity of forced air and further evaluate the influence of velocity on the temperature field. Furthermore, the analysis results will be instructive for the design of the ventilating system for HSPMMs, which is rarely discussed in the previous studies. In this paper, the specifications of the Gramme-ring-winding HPSMM are first provided. Then the CFD model is established for thermal analysis. Based on the model, the reasons causing the overheat of the prototype is thoroughly analysed. To decrease the temperature, the improvements in ventilation structures and inside the water-cooled structure are studied. Finally, a new cooling structure is proposed fit for the HSPMM with Gramme-ring windings. 2 Specifications of the Gramme-ring-winding HSPMM A Gramme-ring-winding HSPMM has been designed for driving a 50-MJ pulsed alternator, shown in Fig. 1. The rotor is constructed by magnets, pole spacers, rotor core, shaft and sleeve. On account of the centrifugal force of high-speed rotation of the rotor and the braking torque caused by pulsed alternator discharging, a considerable thick carbon-fibre retaining sleeve is assembled on the magnets to protect them from breaking [7]. The specifications of the Gramme-ring-winding HSPMM are provided in Table 1. Fig. 1Open in figure viewerPowerPoint Cross-section of the Gramme-ring-winding HPSMM Table 1. Specifications of Gramme-ring-winding HPSMM Parameters Value power, kW 300 speed, rpm 12,000 poles/slots 4/36 outside diameter of stator, mm 570 inside diameter of stator, mm 294 air gap length, mm 2 thickness of sleeve, mm 10 thickness of magnet, mm 18 effective length, mm 80 material of inner stator 35WW250 material of magnets N48UH material of retaining sleeve Carbon fibre Figs. 2 a and b show the configuration of the Gramme-ring-winding HSPMM prototype. The stator losses are intended to be mainly dissipated by the cooling water circulating in the spiral channels of the housing water jacket. The majority of rotor losses will be removed away by axially forced coolant flowing through the air gap (via a separately-installed axial flow blower). In addition, Fig. 2 c shows the contact surfaces between the stator lamination and housing water jacket. As can be seen, there are some imperfections on the surfaces. The deformation of the outer surface of stator lamination is extremely serious, as a result of punching slots on the out edge. Therefore, the interface gap of Gramme-ring-winding HSPMM is much larger than that of the conventional industrial motors. Fig. 2Open in figure viewerPowerPoint Photos of Gramme-ring-winding HSPMM prototype (a) Outside view, (b) End view, (c) Imperfections in the contact surfaces between the stator lamination and housing water jacket There are three PT100 resistance temperature detectors embedded in the middle of inner slots, outer slots and stator core to measure the temperature of them, as shown in Fig. 1. The rotor temperature is estimated by the value of the back EMF, for the reason that the remanence of permanent magnets is related to temperature. 3 Modelling for thermal analysis This paper uses CFD and numerical heat-transfer calculation for thermal analysis of the Gramme-ring-winding HSPMM, to predict the distributions of flowing air and temperature precisely. There are two kinds of fluids existing in the motor, namely cooling water and air. The changes in the velocity and temperature of the flowing water are usually small in the water channels. To simplify the analysis, the cooling effect of water is considered to be modelled by the heat transfer coefficient. The inside air of the motor is in motion influenced by the forced air and rotor rotation. The motion and heat transfer of the air satisfy the laws of conservation of mass, the conversation of the moment, and conversation of energy. The Mach number of the air-fluid is small, so the air is regarded as incompressible. 3.1 Establishment of Gramme-ring-winding model The Gramme-ring windings consist of a large number of round wires in this paper. The diameter of the wires is very small to decrease the high-frequency copper loss. The number of wires in one slot is 255. According to the author of [20, 21], it is not possible or desirable to model the position of each individual wires for the random wound windings to predict the temperature distribution. So several assumptions are introduced to establish the winding model for the following steady-state thermal analysis: (i) The Gramme-ring winding used in this paper is random wound windings. A large number of wires are fulfilled in the slot. So the wires are assumed as uniformly distributed in the slots as the literature. (ii) With use of vacuum pressure impregnation technology, the space in the slot except wires are well filled with the insulating varnish. Hence, the insulating varnish is assumed fully impregnated in the slots. (iii) With good quality of wires, the thickness of the insulation coat is uniform for the wires. According to the mechanism of heat transmission, a layered winding model is introduced to accurately describe the temperature gradient in the slot [22]. The conductors that having similar temperature are lumped together to constitute a copper layer. The thickness of the copper layer is approximately the same with the bare diameter of wires. In addition, the area of all the copper lays, with layers of insulation between them, should be the same with the actual motor. The layered model of the Gramme-ring winding is provided in Fig. 3. The end-wingdings take straight alignment instead of arc to reduce mesh elements. Fig. 3Open in figure viewerPowerPoint Layered model of Gramme-ring winding 3.2 Fluid governing equation and 3D heat conduction equation The air-fluid inside of the motor satisfies the mass conservation, momentum conservation, and energy conservation. On the basis of the fluid mechanic, the calculation control equations for the fluid domain are shown as follows [23]: (1) (2) (3) where μ is the velocity vector (m/s), μ, ν, w are the velocity in the x, y, z directions, respectively, t is the time (s), μ is dynamic viscosity (Pa·s), ρ is the density (kg/m3), P is the pressure (N/m2), and Su, Sv, Sw are the general source (N/m3), T is the temperature (K), λ is the thermal conductivity (W/K·m), c is the heat capacity (J/kg·K), and ST is the volumetric heat sources. Considering the geometrical symmetry of the motor, one-fourth physical model is established as a 3D fluid-thermal coupling analysis model, shown in Fig. 4. Both the fluid and solid domains are included in the model. The entrance of the air-fluid is applied with velocity flow inlet boundary condition, while the outlet of the air-fluid was applied with a pressure boundary condition. Since the focus of this paper is on the steady-state temperature field of the motor, the temporal aspect in the heat conduction governing equation is not considered. For the solid domain, the detailed descriptions of the heat conduction equation for the 3D solved model of the motor are expressed as [17]: (4) where λx, λy, λz are the heat conduction coefficients in the x, y, z direction (W/K·m), qv is the heat source density (W/m3), S 1 is the thermal insulation surface, S 2 is the heat transfer surface, α is the heat transfer coefficient (W/m2), Tf is the fluid temperature around heat transfer surface (K). Fig. 4Open in figure viewerPowerPoint Fluid-thermal analysis model of HSPMM Equations (1)–(3) are solved depending on the fluid flow. To determine if the flow is laminar or turbulent, the Reynolds number should be evaluated. The expression is [21]: (5) where ω r is the angular velocity of the rotor (rad/s), r is the rotor radius (m), δ is the length of the air gap (m). According to [21], the fluid gets turbulent when the Reynolds number is larger than 2300. 3.3 Boundary conditions According to the configuration of ventilation and characteristic of heat transfer for the Gramme-ring-winding HSPMM, a brief description of the boundary conditions for solving domain is given as follows: (i) The inlet velocity and outlet pressure boundary conditions for the forced air are imposed, respectively, on the two ends of the fluid domain. (ii) The rotation of the rotor is considered by setting moving-wall boundary condition on the shadow faces of the rotor. (iii) The symmetric boundary of the solving domain is considered as thermal insulation. (iv) The interfaces of fluid and solid are considered as no-slip boundary in the solving domain. (v) The walls of water channels are considered as convection boundary, and the heat transfer coefficient is calculated hereafter. The velocity of water flow is thought roughly uniform along the water channels. According to the theory of fluid mechanics, the flowing water satisfies the following equations [7]: (6) where N u is the Nusselt number, V, ν water, λ water are the velocity (m/s), kinematic viscosity (m2 /s) and thermal conductivity of water (W/(m·K), respectively. α is the heat transfer coefficient of water channel wall (W/(m2 ·K). D is the equivalent hydraulic diameter (m), and it can be further described as: (7) where A is the area of the cross-section (m2) and U is the wetted perimeter of water flow (m). Considering the turbulence of the water flow, the Nusselt number can also be expressed as: (8) where P r(CP) and P r(CT) are the Prandtl number, respectively, at the average temperature of water and the temperature of water channel walls. ɛ is the correction factor. Due to the little change of water temperature along the water channels, the Prandtl number is regarded as constant, namely P r(CP) /P r(CT) ≃1. Thereby, (8) can be further written as: (9) 3.4 Heat sources The heat sources of the Gramme-ring-winding HSPMM consist of copper loss, iron loss, friction loss and rotor eddy-current losses in the sleeve, magnets and metal pole spacers, respectively, which has been given in the authors' previous work [3]. The iron loss density is much higher in the teeth than that of the yoke, because of the higher flux density in teeth. To take the non-uniform distribution into consideration, the result of iron loss is directly coupled from the electromagnetic model to the fluid-thermal analysis model. The heat generation rate of the other parts is the quotient of the loss divided by the volume of the corresponding part. The friction loss can be obtained through CFD analysis of the fluid. With the carrier frequency of 16 kHz for the PWM voltage source, the heat generation rates applied in the CFD model are provided in Table 2. Table 2. Heat generation rates Parts Heat generation, W/m3 winding (copper layer) 385,526 permanent magnet 551,669 spacer (aluminium alloy) 266,241 sleeve 1,135,679 3.5 Material properties According to the previous studies [10, 21, 24], the thermal properties of materials are applied to the fluid-thermal analysis model, shown in Table 3. Note that the coefficients of thermal conductivity and specific heat capacity of the iron core and winding are actually both non-linear functions of temperature. However, the temperature rises are usually small for the windings and stator. Typically, the average value of temperature rise is <100 K. According to the handbook of chemistry and physics, the changes in thermal conductivity and specific heat capacity are small as well [25]. Consequently, the influence on the steady-state temperature field of the motor can be ignored. So the thermal conductivity and specific heat capacity are considered as constants at room temperature herein. Besides, due to the laminated structure, the thermal conductivity of the lamination stack is anisotropic. In addition, the dynamic viscosity of air is 1.81 × 10−5 Pa·s. According to (6)–(9), the heat transfer coefficient is 1800 W/(m2 ·K) for the water channel walls, with the flow rate of cooling water of 35 L/min. Table 3. Heat conduction coefficients Parts Thermal conductivity, W/K·m Density, kg/m3 Specific heat capacity, J/kg·K iron stack (radial) 42.5 7650 438 iron stack (axial) 2.8 copper 387.6 8933 385 slot insulation 0.15 2150 1172 slot wedge 0.29 2150 1172 permanent magnet 8 7600 502 spacer (aluminium alloy) 177 2770 875 sleeve (tangential) 5.5 1650 712 sleeve (radial) 0.76 housing (steel) 15.1 7750 502 air 0.0259 1.205 1013 4 Numerical analysis and ventilation structures 4.1 Meshing of the model The solution of the fluid-thermal coupling model is achieved in the commercial solver ANSYS Fluent, which is based on the finite-volume technique. According to (5), the Reynolds number of fluid in the air gap is 17,263 (>2300), so the standard k-ɛ turbulence model is used to solve the problem [21]. To solve the fluid governing equation and 3D heat conduction equation, the physical model has to be discretised into small elements. The quality of mesh usually has a great impact on the precision of numerical simulation and convergence of numerical computing. Fig. 5 is the schematic diagram of the meshing details of the CFD model. The mesh type herein is a hexahedron grid. The solid domain primarily finely meshed with experience, then the mesh of the fluid domain was further determined by the following grid independence test, in Fig. 6. The studies on the mesh indicate that the numerical calculation can achieve a good convergent result with the element number of 169,587 for the fluid domain. To remove the influence of mesh on the results, the mesh refinement of the whole fluid domain was studied with the element number ranging from 169,587 to 74,895,534, shown in Fig. 6 a. The temperatures herein are the highest values of corresponding parts of the motor. It can be seen that the mesh refinement to the whole fluid domain has little effect on the results. Moreover, the influence of mesh refinement to the fluid domain locally in the air gap on the results was also studied shown in Fig. 6 b. It can be seen the temperatures also change slightly with the element number. With elements of the fluid domain larger than 4,161,637, the temperature results are basically unchanged. Fig. 5Open in figure viewerPowerPoint Meshing detail of the CFD model Fig. 6Open in figure viewerPowerPoint Grid independence test of CFD analysis (a) Influence of mesh refinement to whole fluid domain on temperatures, (b) Influence of the mesh refinement to fluid in the air gap on temperature Considering the convergence of compute and computer resources, the number of the final elements is 8,572,631 for the developed CFD model (including 4,161,637 elements of the fluid domain). The convergence of computing can be achieved after about 200 times of iterations, as shown in Fig. 7. It can be seen the difference of the highest temperature of the model is 0.0002 K between the last two iterations. Fig. 7Open in figure viewerPowerPoint Convergence history of the maximum value of temperature with iterations 4.2 Numerical calculation of temperature field With forced air velocity of 15 m/s in the air gap, the distribution of temperature field is analysed for the Gramme-ring-winding HSPMM at rated load based on the above CFD model, shown in Fig. 8 a. As can be seen, the hot spot is in the retaining sleeve. The reasons can be attributed to the poor property of thermal conductivity and high heat generation of the carbon-fibre sleeve. The temperature of the magnets is slightly lower than the sleeve. Fig. 8Open in figure viewerPowerPoint Thermal analysis results of the Gramme-ring winding HSPMM based on the CFD model (a) Distributions of temperature field, (b) Streamlines of inside air, (c) Pressure of the inside air Figs. 8 b and c show the distributions of the velocity streamlines and pressure of the air inside of the motor, respectively. The velocity of the air in the air gap is much higher than that of the end region of the motor, due to the motion of the rotor. Besides, it can be seen that the air pressure is very high at the inlet, and then decreases rapidly along the axial direction. To achieve the expected velocity of the forced air, the inlet pressure should be higher than 520 MPa. Therefore, it is essential to choose a commercially high-pressure centrifugal blower. Nevertheless, such blowers usually have large volume, high weight and loud noise, which is unfavourable for the motor system. The high air pressure at the inlet is the result of the large wind resistance of the air gap as a ventilation duct. Accordingly, it is essential to improve the ventilation structure to decrease the wind resistance of the ventilation duct, to increase the velocity of forced air through it. 4.3 Influences of forced air velocity in the air gap The heat convection is generally thought as a function of the coolant velocity, which will further influence the temperature distribution of the motor. Fig. 9 a shows the variation of the highest temperature of key parts of the motor with the forced air velocity. Meanwhile, the inlet pressure corresponding to forced air velocity is also calculated, shown in Fig. 9 b. It can be seen that the highest temperatures of key parts of the motor decline dramatically with forced air velocity, especially the rotor, while the inlet pressure is increased greatly. Besides, the characteristic curves of the three chosen air blowers are also included in Fig. 9 b. The intersections are the operation point of the air blowers. It can be concluded that with the increase of the forced air velocity, the pressure of air blowers required is increased. Fig. 9Open in figure viewerPowerPoint Influences of forced-air velocity in the air gap on the temperatures and inlet pressure (a) Curves of the highest temperatures of key parts with the air velocity, (b) Curves of the inlet pressure with air velocity According to the above analysis, the pressure of the selected axial flow blower is too low to make the forced air velocity to reach the expected value, thus leading to overheating of the rotor for the actual prototype. 4.4 Influences of interface gap on temperature field As previously stated, the interface gap between the housing water jacket and stator lamination is considerable large for Gramme-ring-winding HSPMM. The interface gap is filled with air, the thermal conductivity of which is very poor. Besides, most losses will cross this surface. Therefore, it is essential to clarify the influence of the interface gap on the temperature field. The value of the interface gap is typically between 0.01 to 0.08 mm found in some industrial motors [19]. Considering that the interface gap is much larger for the Gramme-ring-winding HSPMM, the temperature field is analysed with the interface gap ranging from 0 to 0.2 mm. The variations of the temperatures of key parts of the motor with interface gap are shown in Fig. 10. It can be seen that the temperatures of all the rotor parts increase slightly with the interface gap, whereas the temperatures of stator core and windings are significantly dependent on the value of the interface gap. The reason is that the majority of the stator losses are dissipated through the water jacket, but the rotor losses are mainly removed away by the forced air. Thus, it can be concluded that the temperature rises of stator core and windings will be significantly underestimated, with no regard of the interface gap. Fig. 10Open in figure viewerPowerPoint Influence of interface gap on the temperatures of the key parts of the HSPMM 4.5 Verification of the analysis results To verify the analysis result, the comparison between the calculation and test temperature rise was performed. Fig. 11 shows the test bench designed to assess the temperature rise of the prototype. In such a test arrangement, the prototype is coupled to a dynamometer via a universal coupling. Restricted by the speed range of dynamometer, the temperature rise of the prototype was only tested up to 6000 rpm. As previously described, the temperature of the windings and stator are tested by the pt100 sensors embedded at different positions, shown in Fig. 1. And the temperature of the magnets is the average value estimated by the remanence which is related to temperature. The average forced air velocity in the air gap measured is about 2.5 m/s. Fig. 11Open in figure viewerPowerPoint Overview of test bench for the prototype By calibration of the calculation results with the measured data, they are in good agreement with the interface gap about 0.15 mm for the prototype. Table 4 shows the comparison between calculated temperature rise result and test data at rated torque, 6000 rpm. The worst error is −4.74% on the stator core. In addition, Fig. 12 shows the comparison between calculated temperature rise result and test data at different load rates, 6000 rpm. As can be seen, the worst discrepancy between the calculated and tested temperature rise is lower than ±4°C. So the fluid-thermal model can be verified for the studies on thermal issues. Table 4. Comparison between calculated temperature rise result and test data at rated torque, 6000 rpm Position Calculation, °C Test, °C Error, % inner windings 71.28 74.8 −4.71 outer windings 70.45 73.5 −4.15 stator core 65.92 69.2 −4.74 magnets 83.84 81.4 3.05 Fig. 12Open in figure viewerPowerPoint Comparisons between calculated temperature rise results and test dat