1 DESIGN STUDY OF LOW-SPEED DIRECT-DRIVEN PERMANENT-MAGNET MOTORS WITH CONCENTRATED WINDINGS F. Libert, J. Soulard Department of Electrical Machines and Power Electronics, Royal Institute of Technology 100 44 Stockholm, Sweden e-mail: florence@ets.kth.se Abstract— In this study, different designs of radial-flux direct-driven permanent magnet (PM) motors are compared. They should replace an induction motor and its gearbox for an industrial application requiring 5 kW and 50 rpm. The goal is to show if the use of concentrated windings brings some benefits for this application, the PM motors having a high number of poles. Different combinations of the number of slots and number of poles are investigated for the concentrated windings designs. Neodymium Iron Boron PM and ferrite PM are tested and the cost of the different machines is estimated. 1. INTRODUCTION Replacing induction machines with permanent magnet (PM) synchronous machines has recently gained great interest, as the environmental concern increases worldwide and the price of PM materials decreases. For low speed applications, using PM machines may further eliminate the need of a gearbox, which is traditionally coupled to a standard induction machine. Since the gearbox is costly, decreases the efficiency of the drive, and needs maintenance, a direct-driven PM machine is an attractive solution that can compete with the induction motor and the gearbox. Direct-driven PM machines are nowadays mostly used for boat propulsion, wind turbines and elevators [1]. In the present paper, the industrial application is a mixer used in wastewater treatment plants, which requires 5 kW and a rated speed of 50 rpm. Since the speed is very low, the machines are lighter when they have a high number of poles. Different radial-flux PM machines with high pole number, distributed windings, and different PM positions in the rotor were investigated in [2]. The present study shows how concentrated windings can improve the designs of high pole number direct-driven PM machines. The advantages and drawbacks of the concentrated windings are well known when the number of poles is under 30 [3], [4]. With a higher number of poles, the possibilities in the design of permanent magnet machines with concentrated windings are wider. Especially there are many more combinations of number of slots and number of poles to choose between. The choice of the number of poles and number of slots of the investigated designs is therefore justified in detailed. Designs with surface mounted and buried PM magnets with different concentrated windings are compared using results from finite element methods (FEM) simulations. The cost of the active material of different designs is also given and different PM materials are investigated. The benefits that bring the use of concentrated windings in the direct-driven PM machines are emphasized. 2. THE INVESTIGATED PM MOTOR CONFIGURATIONS A. Description of the motor configurations Surface mounted permanent-magnet (SMPM) motors are the most common configuration of radial-flux PM motors for direct-drive application. Both inner rotor and outer rotor motor designs are investigated. The other considered configuration has buried permanent magnets. It is referred to as tangentially-magnetized PM (TMPM) motor (figure 1). The salient rotor consists of different pieces of iron and permanent magnets that are fixed together on a non-ferromagnetic shaft. With a ferromagnetic shaft, a large portion of the flux generated by the permanent magnets would leak through the shaft. With this configuration, the flux generated by the PM is concentrated in the rotor iron. High airgap flux densities can thus be achieved with a low PM weight. Fig. 1. Cross-section of a TMPM motor (one pole pair) 2 The compared designs have Neodymium Iron Boron (NdFeB) permanent magnets. Designs with ferrite magnets are also investigated for the tangentiallymagnetized PM motors. B. Design procedure The designs are conducted by solving an optimization problem. The objective function to be minimized is the active weight of the motor. The optimized design should fulfill different requirements and constraints. Thus the exterior diameter is limited to 500 mm, a minimum efficiency is guaranteed by limiting the copper losses to 15% of the nominal power of the machine and a limit is set to 5 kg of PM material (for the NdFeB magnets) because of the high cost of the PM. All the designs are calculated for the same nominal torque and nominal speed of 840 Nm and 50 rpm and the same amount of copper losses (700 W). The different steps of the design procedure are depicted in figure 2. More about this procedure can also be found in [2]. The analytical design procedure is common for the different configurations and adapted for each case. For example, the reluctance torque is taken into account for the tangentially-magnetized PM motor, even though the reluctance torque is less than 5% of the rated torque. The saliency ratio is indeed less than 3. NdFeB permanent magnets or ferrite magnets can be used. The analytical calculations of the fundamental airgap flux density and flux density in the teeth are corrected using FEM simulation results. The difference between the analytical calculations and FEM simulations results of the fundamental airgap flux density is less than 5% for designs with distributed windings and less than 8% for designs with concentrated windings, the number of poles being between 20 and 80. 3. THE INVESTIGATED WINDINGS A. Distributed windings The higher the number of pole, the lower the active weight of the direct-driven motor. Since the outerdiameter of the motor is limited and the number of poles is high (over 20), the number of slots per pole per phase q is set equal to 1. Otherwise, with a higher number of slots per pole per phase, the teeth are too numerous and too thin to guarantee a rigid structure. With q = 1 and without any skewing, the torque ripple is high, due to the interaction between the airgap flux and the space harmonics in the magneto-motive force. As table I shows, the torque ripple is the highest for the tangentially-magnetized PM motors and the lowest for the inner-rotor surface-mounted PM motors, but it is still very high. The torque ripple is simulated with FEM. Fig. 2. Followed procedure to optimized the design of low-speed PM motors Table I Ratio between the torque ripple and the mean torque in % Pole number p = 30 p = 50 p = 70 Inner-rotor SMPM 19 16 10 Outer-rotor SMPM 28 27 26 Tangentiallymagnetized PM 70 46 47 B. Concentrated windings Using concentrated windings has several well-known advantages: the end-windings are much shorter, the windings are easy to mount and a low torque ripple can be achieved [3], [4]. Furthermore, the concentrated windings give the possibility of having a low number of teeth, around three times lower than for windings with q = 1. This is an important advantage when considering high pole number machines with limited dimensions, since the teeth are less saturated and more rigid. Several investigations have been made on the choice of the pole number (p) and slot number (Qs) of PM motors with concentrated windings, [4], [5], [6]. The highest winding factors are obtained for concentrated windings with the number of poles close to the number of teeth. 3 Table II Winding factors for different pole numbers p and slot numbers Qs Table II shows the winding factors for windings with different number of poles and number of slots. Windings with the same layout pattern and the same number of slots per pole per phase have the same winding factor. Machines with concentrated windings, whose layouts are not symmetrical, have problems with vibration and noise because of an unbalanced magnetic pull. The combinations of slot and pole number of such windings are for example when Qs = 9 + 6k (k=0,1,2...) and p = Q s ± 1 , [5]. The winding layout of these motors is not symmetrical since the coils of one phase are all located on one side of the stator. In this paper, the investigated slot and pole number combinations give high winding factors and windings layouts with many symmetries. They are the combinations with the number of slots per pole per phase q equal to 2/5 and 2/7 giving a fundamental winding factor kw1 equal to 0.933, q = 3/8 and 3/10 with k w1 = 0.945 and q = 5/14 and 5/16 with kw1 = 0.951. saturating the teeth. This allows the motors with concentrated windings to be shorter and therefore lighter. Fig. 3. Active motor weight as a function of the pole number for SMPM motor designs with concentrated and distributed windings. 4. RESULTS FOR SURFACE-MOUNTED PM MOTORS A. SMPM motors with inner rotor Figure 3 shows the active weight as a function of the pole number for different values of the number of slots per pole per phase q. The tendency for SMPM motors with concentrated windings is that the active weight decreases with an increasing pole number. Some points do not follow this tendency due to the fact that the active weight also varies with the number of slots. Thus, the designs with q = 2/5 or 2/7 are slightly heavier than the others, because of their lower winding factor. Figure 3 reveals also that the motors with concentrated windings are about 15 kg lighter than those with distributed windings (q = 1). Since there are fewer teeth in designs with concentrated windings, both the slots and teeth can be wider, as figure 4 shows. The permanent magnets are allowed to be thicker and the airgap flux density to be higher without Fig. 4. Geometries of two 70-pole SMPM motors, one with concentrated windings (q=3/10) and one with distributed windings. All these designs have equal copper losses (700 W). Moreover, the torque ripple is reduced from 9.3 % to less than 3 % for motors with concentrated windings. B. SMPM motors with outer rotor With a larger airgap diameter, the SMPM motors with outer-rotor are lighter than the inner-rotor SMPM motors (compare figure 3 to figure 5). With concentrated windings, the outer rotor SMPM motors are approximately 20 kg lighter than with distributed windings. The drawback of the outer-rotor SMPM motors with q = 1 is their high torque ripple (table I). 4 With concentrated windings, it is successfully reduced from 26 % to less than 5 %, as figure 6 shows. Fig. 5. Active motor weight as a function of the pole number for outer-rotor SMPM motor designs with concentrated and distributed windings (q=1). and 4.5 kg. Due to the flux concentration, the airgap flux density reaches easily 1.1 T with less than 5 kg permanent magnet. Fig. 7. Active motor weight as a function of the pole number for TMPM motors with concentrated and distributed windings (q=1). Table III Comparison of 70-pole tangentially-magnetized PM motor designs with NdFeB magnets (FEM results) q 3/10 2/5 5/14 1 Fig. 6. Torque of outer-rotor SMPM motors with concentrated and distributed windings at load conditions. 4. RESULTS FOR TANGENTIALLYMAGNETIZED PM MOTORS A. Tangentially-magnetized PM motors with NdFeB magnets Figure 7 shows the active weight of different tangentially-magnetized PM motor designs with concentrated and distributed windings. The difference in weight between the designs with different concentrated windings and same pole number is very small, although designs with q = 2/5 and q = 7/5 are slightly heavier. As for the surface-mounted PM motors, designs with concentrated windings are much lighter than with distributed windings. However, the tangentiallymagnetized PM motor designs with concentrated windings have an important advantage compared to the surface-mounted PM motors. In addition to a lower active weight, the found optimized designs have lower permanent magnet weights than the maximum allowed 5 kg. Their permanent magnet weight is only between 3.5 Active weight [kg] 55.9 57.9 56.2 76.4 PM weight [kg] 3.4 3.3 4.1 5.5 Torque ripple [%] 4.2 4.8 4.2 41.7 Active length [mm] 101 100 119 163 Outer diameter [mm] 500 498 500 500 Furthermore, the lower bound of the constraint on the machine length, which is 100 mm, is reached for some designs. The external diameter is then reduced and becomes less than 500 mm. This can be observed in table III that gives different features of 70-poles tangentiallymagnetized PM motor designs with different winding types. The torque ripple is also considerably reduced with concentrated windings, as shown in table III. Tangentially-Magnetized PM motor designs with outerrotor are also possible and have been simulated. As for the surface-mounted PM motor, the active weight is lower for the outer-rotor designs due to a larger airgap diameter. A 70-pole PM motor with outer rotor is thus 5 kg lighter than an inner-rotor design with 3 kg PM weight. However the torque ripple is slightly higher with 6 %. B. Tangentially-magnetized PM motors with ferrite permanent magnets For the tangentially-magnetized PM motor designs, the NdFeB PM can be replaced with ferrite magnets. Indeed, the flux concentration in the rotor allows reaching high airgap flux density with a low amount of PM. Furthermore, the designs with NdFeB magnets are light 5 thanks to the use of concentrated windings. With ferrite magnets, the designs, though heavier, might still be lighter than the induction motor and its gearbox that are weighting 150 kg together. The ferrite magnets have poorer magnetic properties than the NdFeB magnets but they are much cheaper. The solution with ferrite magnets can then be a better compromise between the machine’s weight and the machine’s cost. Ferrite magnets have other advantages. They are electrically non-conducting, which means that there are no eddy currents in these magnets. Their temperature of demagnetization is high (over 200° C). The ferrite magnets are also easier to handle when assembling the rotor since they are weaker than the NdFeB magnets. Finally their density is lower (5000 kg/m3 against 7500 kg/m3). With ferrite magnets, the airgap flux density is chosen to be lower than 1 T in order to use the PM efficiently. Figure 8 shows the magnetization curve of the ferrite magnet: for a 56-pole design with a PM height of 30 mm, the operating point of the PM is close to the maximum energy product. The PM thickness is also adjusted so that the iron between two PM does not saturate. The obtained airgap flux density for the 56-pole design with ferrite magnet is then 0.96 T. In order to achieve the nominal torque, the current loading is higher compared to a design with NdFeB. This can be seen in figure 9. The slots of the motor with ferrite magnets are bigger than the other motor (they both have the same amount of copper losses). Features from two designs with 56 poles, 63 slots and different magnets are compared in table IV. Airgap flux density, torque ripple and iron losses are calculated with FEM. The design with ferrite magnets is 35 kg heavier than the design with NdFeB. The ferrite magnet weight is high with 16.4 kg but these magnets are cheap. The iron losses are slightly higher for the ferrite magnet design due to the bigger quantity of iron, but the eddy current losses in the NdFeB magnets were not taken into account whereas the ferrite magnets have no eddy current losses. Fig. 8. Magnetization curve of the ferrite magnet material with operating points of designs having different magnet widths wm and same magnet thickness. Bδ is the fundamental of the airgap flux density Fig. 9. Geometries of two 56-pole tangentially-magnetized PM motors with concentrated windings, one with NDFEB magnets, one with ferrite magnets. Table IV Comparison of 56-pole, 63-slot tangentially-magnetized designs with NdFeB and ferrite PM. Results Airgap flux density Magnet width Magnet height Torque ripple PM weight Rotor active weight Total active weight Iron losses (stator + rotor) Units NdFeB Ferrite T 1.16 0.96 mm mm % kg 10.2 9.4 3.7 4.8 30 14 3.7 16.4 kg 12.1 32.2 kg 62.2 97.2 W 73.9+ 6.3 76.7+ 12.8 5. COST The different designs have been calculated to give the lower active weight with a limited PM weight for the NdFeB magnets. The total cost of the motor is also very important. An estimation of the cost of the machine is therefore made by calculating the cost of the active materials, which are the stator and rotor iron, the copper and the permanent magnets. The prices of these materials per kilogram are given in table V. 6 The costs of the active material of the different machines with 56 poles are given in table VI. These machines have the same external diameter and amount of copper losses. As can be seen, the active materials of the motors with concentrated windings are cheaper than the motor with distributed windings. Tangentially-magnetized PM motors are lighter and cheaper than the surface-mouted PM motors, due partly to the lower PM weight needed to reach the demanded torque and speed. The machine with ferrite magnets is about 100 € cheaper than the cheapest motor with NdFeB PM. It is also about 40 kg heavier. If the cost of the machine is an issue and not the weight, the ferrite magnets are a solution to decrease the price of active material. Table V Material cost NdFeB Price [€/kg ] 55 Ferrite Lamination plates Copper wire 5.5 1.65 3.1 Rotor SMPM Inner SMPM Inner SMPM Outer SMPM Outer TMPM Inner TMPM Inner TMPM Outer TMPM Inner NdFeB Active weight [kg] 98 Cost of material [€] 483 NdFeB 79 425 NdFeB 87 461 NdFeB 66 385 NdFeB 78 440 NdFeB 62 377 NdFeB 59 356 Ferrite 97 251 Winding Type PM Material DW q=1 CW q=3/8 DW q=1 CW q=3/8 DW q=1 CW q=3/8 CW q=3/8 CW q=3/8 The results show that high pole number PM machines with well-designed concentrated windings are very attractive for low-speed direct-driven applications. High pole number motors with concentrated windings are lighter than the conventional motors with distributed windings. They are also cheaper because the PM weight required to reach the torque is lower at same value of copper losses. The construction of the stator is also made easier because of the bigger and less numerous slots. Buried PM machines with tangentially-magnetized PM and concentrated windings are appropriate for the application. Because of the flux concentration in the rotor, cheaper PM can be used. Combined with concentrated windings, the cost and weight of the highpole number tangentially-magnetized PM motor is very attractive. REFERENCES Table VI Material cost for various designs with 56 poles Motor 6. CONCLUSION [1] T. Haring, K. Forsman, T. Huhtanen, M. Zawadzki, “Direct Drive – Opening a New Era in Many Applications”, Pulp and Paper Industry Technical Conference, pp. 171 –179, 16- 20 June 2003. [2] F. Libert, J. Soulard, “Design Study of Different DirectDriven Permanent-Magnet Motors for a Low Speed Application”, Nordic Workshop on Power and Industrial Electronics Trondheim, Norway, 2004. [3] D. Ishak, Z.Q. Zhu, D. Howe, "Comparative Study of Permanent Magnet Brushless Motors with All Teeth and Alternative Teeth Windings”, IEE International conference on Power Electronics and Electrical Machines (PEMD), Edinburgh, United Kingdom, 2004. [4] J. Cros, P. Viarouge, “Synthesis of High Performance PM Motors With Concentrated Windings”, IEEE Transactions on Energy Conversion, Vol. 17, Issue 2, pp. 248-253, 2002. [5] F. Libert, J. Soulard, "Investigation on Pole-Slot Combinations for Permament Magnet Machines with Concentrated Windings", International Conference on Electrical Machines, Cracow, Poland, 2004. [6] D. Ishak, Z.Q. Zhu, D. Howe, "Permanent Magnet Brushless Machines with Unequal Tooth Widths and Similar Slot and Pole Numbers", Industry Applications Conference, 39th IAS Annual Meeting. Conference Record of the 2004 IEEE, Vol. 2, pp. 1055-1061, 3-7 Oct. 2004.