Performance evaluation of multi-phase permanent magnet synchronous motor based on different winding configurations and magnetization patterns

Received Jul 17, 2018 Revised Feb 14, 2019 Accepted Mar 25, 2019 Permanent magnet synchronous motor (PMSM) is the most reliable and efficient machine that widely used in robotics and automation, industrial applications, electric vehicles, home appliances, aircraft and aerospace technology due to its high efficiency, good dynamic performance and high torque density. In this paper, the influence of various types of winding configuration and different magnetization patterns in the performance of a five-phase PMSM is investigated. Three types of magnetization patterns such as radial magnetization (RM), parallel magnetization (PaM), and multisegmented Halbach magnetization (SH) are applied to the five-phase 10slot/4-pole PMSM during open-circuit and on-load conditions. A 2D finite element method (FEM) is intensively used in this investigation to model and predict the electromagnetic characteristics and performance of the PMSM. The detailed results from the finite-element analysis (FEA) on the cogging torque, induced back-emf, airgap flux density and electromagnetic torque are analysed. The induced back-emf of the machine is computed further into its harmonic distortions. Additionally, the skewing method for minimization of cogging torque of PMSM is proposed. From the results, it is observed that the five-phase, 10-slot/4-pole PMSM with double layer distributed winding and parallel magnetization gives the best machine performance.


INTRODUCTION
For the past decades, transportation sector's fuel consumption has grown at a higher rate than other industrial sectors. This increase was largely derived from the new demand for personal use of vehicles powered by conventional internal combustion engines. With increased awareness of air pollutions and rising fuel prices, electric vehicles have emerged as an alternative solution to replace the conventional vehicles [1,2]. Permanent magnet synchronous motor (PMSM) is a very good candidate for the electric vehicle applications due to its advantages such as high power density and high efficiency with an efficient heat lost to the surrounding [3,4]. However, electric vehicle applications require extremely high reliability that made the traditional three-phase PMSM facing a severe challenge. Multi-phase drives have additional degrees of freedom that can be used for different purposes, such as additional torque generation or fault tolerance when a part of the system fails [5][6][7]. With the increase in the number of phases, the multi-phase PMSM can obtain the highest possible reliability. Therefore, multi-phase PMSM drive system offers another solution for the electric vehicle drive system. In order to achieve a high performance PMSM, both rotor and stator structures of the motor must be considered. PMSM has two types of windings which can be applied to the stator core such as concentrated windings and distributed windings [8,9]. The concentrated windings can reduce the copper loss, while the distributed windings may contribute to higher torque output of the motor. But the copper loss from distributed windings is usually higher than that of concentrated windings. In this paper, a five-phase, 10-slot/4-pole PMSM is designed based on an analytical calculation to determine the machine dimensions using sizing equations. The machine performance is modelled using analytical sub-domain model, and further analysed using the finite element method that involves with three different winding types i.e. double-layer distributed winding (DLDW), single-layer concentrated winding (SLCW), and double-layer concentrated winding (DLCW). Three types of magnetization patterns such as radial magnetization (RM), parallel magnetization (PaM) and segmented Halbach (SH) magnetization are also considered for this five=phase, 10-slot/4-pole PMSM. The influences of winding configurations and magnetization patterns on the phase back-emf and the electromagnetic torque are analyzed and compared. Additionally, the cogging torque is analysed and further minimized by applying the stator skewing method. The machine model with the best machine performance will be selected. The research starts by determining the machine design parameters using an analytical model and sizing equations. Then, six models of PMSM that have different magnetization patterns and winding configurations are developed. The PMSM motor performances i.e. the magnetic vector potential, flux density distribution, phase back-emf, cogging torque, and electromagnetic torque for each design are analyzed. Finally, a skewing method is applied in order to reduce the cogging torque.

RESEARCH METHOD 2.1. Machine design parameters
The machine dimensions for the five-phase, 10-slot/4-pole PMSM are determined based on analytical calculation. In order to calculate the dimensions, some parameters are fixed to ensure this machine model is applicable in hybrid electric vehicles (HEVs) and electric vehicles (EVs). The fixed parameters for the five-phase PMSM are referred from [10] as shown in Table 1. In [10], the five-phase PMSM applied surface-mounted segmented Halbach (SH) magnetization with double-layer concentrated winding (DLCW) configuration. The others machine parameters such as stator tooth body, stator yoke, rotor yoke, and tooth tip are identified using sizing equations [11]. The average air gap flux density Bg is determined using (1), where the symbols are defined in Table 1. Magnetic flux from the rotor magnet will cross the air gap and enter the stator teeth. Therefore, the total flux per stator tooth ∅ is obtained using (2), where Ns is the number of slots. Assuming that saturation level of flux density Bsat in the stator and rotor core is not more than 1.6T, then the stator tooth body wtb is calculated using (3). For this 10-slot/4-pole PMSM, the total flux passing through the stator tooth body will not split into two halves when reaching the stator yoke. Therefore, the stator yoke wsy will have the same height as the stator tooth body wtb as represented in (4). The rotor yoke wry is assumed higher than the theoretical value. The minimum value of rotor yoke is calculated using (5). Total flux per magnet pole ∅ is determined using (6), where p is the pole-pair number of the PMSM. Figure 1 shows the 10-slot/4-pole PMSM model with different magnetization patterns i.e. RM, PM, and SH.The performances of machine i.e. phase back-emf and output torque are affected by the allocation of slots for the phase coils. Therefore, it is important to determine the winding arrangements in the five-phase, 10-slot/4-pole PMSM which maximizes the machine performance. The allocations of phase windings in the 10-slot/4-pole PMSM are shown in Figure 2 ISSN: 2088-8694  Performance evaluation of multi-phase permanent magnet synchronous motor based on (Dahaman Ishak) The magnetic field distributions, phase back-emf, electromagnetic torque, and winding inductance can then be calculated. Few assumptions are used in formulating the analytical sub-domain model i.e. infinite permeability in the rotor and stator cores; no conductivity in the rotor and stator cores; the eddy current reaction field is neglected in this model; the teeth are spoke shaped (radial slot boundaries); end-effect is neglected; and linear magnet properties [19]. In order to simplify the analytical sub-domain modelling, in this paper, only two regions are considered, i.e. magnet region and airgap region. The magnetic flux density in the airgap is obtained analytically using (7) and (8) for its radial and tangential components respectively [20] where Mn is the magnetization component for the permanent magnet, is the relative permeability, Rr is the rotor radius, Rm is the magnet radius, Rs is the stator inner radius, Qs is the slot number, p is the pole-pair number, and is the permeability of the air. While , , and are the Fourier series coefficients of the complex relative permeance function (CRPF).

RESULTS AND ANALYSIS 3.1. Magnetic vector potential
It is important to ascertain that the flux patterns are correct as expected by field distributions in the permanent magnet. Applying the magneto-static analysis [21] which is available in the 2D finite element analysis (FEA), the open-circuit magnetic field distributions across the machine sectional area can be predicted as illustrated in Figure 4. Each machine is modelled with three different magnetization patterns for the surface-mounted rotor magnets i.e. radial magnetization (RM), parallel magnetization (PaM), and segmented Halbach magnetization (SH). The magnetic flux lines flow from one magnet to adjacent magnets with different polarities. Since this motor has four magnetic poles, there are four magnetic flux semi-circles in the rotor core.

Flux density distribution
The open-circuit flux density distributions for RM, PaM and SH magnetized magnets are computed from 2D FEA as shown in Figure 5. The highest values of magnetic flux density are 1.52T, 1.51T, and 1.55T for machines with RM, PaM and SH respectively. These highest values are observed mainly in the stator teeth and stator yoke. The magnetic flux density from the SH magnetization in the 10-slot/4-pole machine during open-circuit condition is the highest compared with the RM and PaM magnetization patterns. During on-load condition, the magnetic flux density in the machine is further influenced by the types of winding in the machine. Referring to Figure 6, it can be observed that the flux density is varied throughout the machine cross-sectional area. The PM machines are excited with sinusoidal phase current of 5A peak. Higher flux density inside tooth tip area is observed during the on-load operation. However, among these three winding configurations, DLDW produces the highest magnetic flux density as shown in Figure 6

Phase back-emf and harmonics
The phase back-emf waveforms of 10-slot/4-pole PMSMs for RM, PaM and SH are shown in Figure  8. It is noted that the phase back-emf for the machine with RM and PaM are trapezoidal in shape, but the machine with PaM exhibits slightly higher magnitude of phase back-emf than the machine with RM. Additionally, the phase back-emf for the machine with SH is more sinusoidal and higher in magnitude with DLDW compared to RM and PaM. Meanwhile, the machine with SLCW and DLCW configurations generate exactly the same magnitude and shape of phase back-emf.

Cogging torque
Generally, cogging torque exists from the interaction of permanent magnet MMF harmonics and the air gap permeance harmonics due to slotting. A low cogging torque may be achieved when the ratio of the slot number to pole number is fractional, and the smallest common integer between the slot number and pole number is high. Since the PM machines under consideration have 10-slot/4-pole numbers, therefore, the smallest common integer is 20. As a result, the cogging torque waveform is repeating every 18 0 mech. or 36 0 elect. The system variable RMTORQUE in OPERA2D software computes the cogging torque when the machine is simulated during open-circuit condition as shown in Figure 9. As can be seen, both PM machines with PaM and RM have an almost similar shape of the cogging torque waveform. However, machine with PaM has slightly smaller cogging torque compared to machine with RM. As expected, a PM machine with SH exhibits almost zero cogging torque.

Electromagnetic torque
During the on-load condition, electromagnetic (EM) torque is simulated in 2D FEA rotating machine analysis by exciting phase currents of 5A peak into the stator windings. The EM torque waveforms of the five-phase, 10-slot/4-pole machine for three magnetization patterns are shown in Figure 10. The EM torque for the motor with PaM is the highest than RM and SH. The machine with SH has the lowest EM torque. This is quite expected since the fundamental back-emf is highest for the machine with PAM and lowest for SH, as shown earlier in Table 2. In addition, the machines with DLDW produces the highest EM torque, while the SLCW has a lowest EM torque value compared to other types of windings. Particularly, the torque ripple in the machines which have DLDW are the highest compared to others. The average EM torque and peak-to-peak torque ripple are listed in Table 3.

Skewing method
The cogging torque in 10-slot/4-pole PMSM with RM and PaM can be further minimized by applying the skewing method in 2D FEA [22][23][24][25]. There is no more cogging torque minimization required for SH since it already exhibits almost zero cogging torque i.e. 0.04 Nm peak. Using OPERA2D software, the stator core ISSN: 2088-8694  Performance evaluation of multi-phase permanent magnet synchronous motor based on (Dahaman Ishak) 1205 of 10-slot/4-pole PM machine has been sliced into three axial segments. Each axial segment of the stator is skewed with an inclination angle equal to 6 o mechanical, making total skew angle αi of 18º mechanical. Figure 11 shows the cogging torque reduction for the 10-slot/4-pole PM machine for RM and PaM respectively. The resultant cogging torque is nearly nine times smaller than the initial value. However, PM machine with skewed stator has slightly lower phase back-emf due to the skewing effect. Figure 11. Cogging torque minimization using skewing method for five-phase, 10-slot/4-pole PMSM

CONCLUSION
This paper has investigated the influence of different winding configurations and magnetization patterns on the performance of a five-phase, 10-slot/4-pole PMSM. Generally, 10-slot/4-pole PMSM can produce different machine performance due to the differences in magnetization patterns and winding configurations. Further, it is also shown that better machine performance can be achieved by applying the skewing method in 2D FEA to minimize the cogging torque. In terms of phase back-emf, the machine with segmented Halbach exhibits the most sinusoidal waveform. Whereas, the PMSM with radial and parallel magnetizations induce more trapezoidal shape of phase back-emf. The five-phase, 10-slot/4-pole PMSM with double-layer distributed winding (DLDW) for parallel magnetization (PaM) gives the best machine performance compared to other types of machines, in terms of higher magnitude of fundamental phase back-emf, lower cogging torque, and higher electromagnetic torque. Future research will be focused on the torque versus speed of the PMSM in order to determine its suitability in the electric vehicles and hybrid electric vehicles.