318 Di Gerlando et al. Figure 15. Simulated and test results of themotor of Table 2 and Fig. 4, with N tuc = 46, in sinusoidal feeding (380 Vrms, 50 Hz); x axis: torque (from no-load to T max ); y axis: input current; ∇: analytical simulation, by (15), (16), δ 0 ≥ δ ≥ δ max = ϕ z : ∇: experimental result; x: FEM simulation result [14]. Another important effect connected to the PM winding arrangements adopted in this kind of machine is the very low level of cogging: by manually handling the rotor of the unfed motor, we have verified no appreciable cogging torque, as confirmed also by FEM simulations [14]. Fig. 15 shows simulated and test results of the input current in loaded operation with sinusoidalfeeding (V=380 Vrms,f =50Hz), with thetorquerangingfrom zerotoT max :the analytical result (see also Fig. 7) is confirmed both by measurements and FEM simulation, for no-load, rated torque and pull-out torque conditions. The rated operation has been verified also by a thermal test (Fig. 16), that indicated acceptable temperature levels. Figs. 17 and 18 report some simulations, performed by integrating equation (28), aimed to show the motor dynamic behavior, evidencing its self-starting capabilities, under mains sinusoidal supply, in loaded conditions. Considering that the rotor inertia equals J r = 0.023 kgm 2 , we have considered to drive a load withthe sameinertia (→ J tot = 0.046kgm 2 ); several simulationshavebeen performed, with different mains voltage phase conditions. Figure 16. Experimental thermal test of the motor of Table 2, Fig. 4, N tuc = 46, running with sinu- soidal feeding (380 Vrms, 50 Hz), with rated torque (T n = 53 Nm); the points are the temperatures measuredby athermocouple put in contact withtheendwindings(ambienttemperature:T a = 25.5 ◦ C). III-1.2. High Pole Number, PM Synchronous Motor 319 Figure 17. Simulated transient of the motor of Table 2, Fig. 4, N tuc = 46, with sinusoidal feeding (V = 380 Vrms, f = 50 Hz): synchronization from zero speed with rated torque (T n = 53 Nm); total inertia: 0.046 kgm 2 ; response to torque steps of DT = 40 Nm. Fig. 17 refers to a synchronization from zero speed with rated torque (T n = 53 Nm), followed by two opposite torque steps of T = 40 Nm: the response appears stable and acceptable, both at starting and after load variations. Fig. 18 shows another starting transient under the same conditions of Fig. 17, except for the initial values of the supply voltages (in opposition to the previous one): the starting transient hasthe sameduration asbefore (roughly0.4 s),but torque andspeed showdifferent instantaneous values, even significantly negative. At t = 0.5 s, a torque ramp is applied, up to the pull-out torque, that occurs exactly at the analytically estimated torque value (T max = 101 Nm), with the consequent loss of synchronization. Corresponding results have been obtained also by FEM transient simulations: these simulations gave the additional information of the absence of torque ripple: this result, confirming the absence of cogging of the unfed machine, appears particularly interesting, also considering that no skewing have been applied between teeth and PMs. Experimental starting tests in loaded conditions demonstrated the correctness of the simulations, with a satisfying behavior, both at starting and during steady state operation: Figure 18. Simulated electromechanical transient of the motor of Table 2, Fig. 4, N tuc = 46, with sinusoidal feeding (V = 380 Vrms, f = 50 Hz): synchronization from zero speed with rated torque, with initial voltages in opposition to those in Fig. 17; application of a torque ramp, up to the pull-out torque (T max = 101 Nm). 320 Di Gerlando et al. the sinusoidal nature of the machine is confirmed by the practical absence of noise in any operating condition. Conclusion A PMsynchronous motorequipped with special, two-layer, concentrated coil windings have been described, capable of self-starting in loaded conditions with mains supply: the winding structure have been illustrated, together with some design criteria, developing useful figures of merits for the best choice of the main constructional parameters. Several simulations by analytical and FEM models have demonstrated the interesting performances of the machine, confirmed also by corresponding experimental tests. The activity will be intensively continued, both as regards the optimization of the mo- tor, and concerning the application of the developed winding theory to different machine configurations. References [1] E. Spooner, A.C. Williamson: “Direct coupled, PM generators for wind turbine applications”, IEE Proc. on Electr. Power Appl., Vol. 143, No. 1, pp. 1–8, January 1996. [2] E. Spooner, A.C. Williamson, G. Catto: “Modular design of PM generators for wind turbines”, ibidem, Vol. 143, No. 5, pp. 388–395, September 1996. [3] E. Spooner, A.C. Williamson: UK Patent 2278738: “Modular Electromagnetic Machine”. [4] P. Lampola: “Electromagnetic Design of an Unconventional Directly Driven PM Wind Gen- erator”, Proceedings ICEM’98, Istanbul, Turkey, pp. 1705–1710, 1998. [5] M. Lukaniszyn, M. Jagiela, R. Wrobel: “Influence of Magnetic Circuit Modifications on the Torque of a Disc Motor with Co-axial Flux in the Stator”, Proceedings ICEM’02, Brugge, Belgium, paper No. 069, 2002. [6] A. Muetze, A. Jack, B. Mecrow: “Alternate Designs of Low Cost Brushless DC Motors using Soft Magnetic Composites”, ibidem, paper No. 237. [7] Th. Koch, A. Binder: “PM Machines with Fractional Slot Winding for Electric Traction”, ibidem, paper No. 369. [8] S. Tounsi, F. Gillon, S. Brisset, P. Brochet, R. Neji: “Design of an axial flux brushless DC motor for electric vehicle”, ibidem, paper No. 581. [9] W.R. Canders, F. Laube, H. Mosebach: “PM Excited Poly-phase Synchronous Machines with Single-Phase Segments. Featuring Simple Tooth Coils”, ibidem, paper No. 610. [10] F. Magnussen, C. Sadurangani: “Winding Factors and Joule Losses of PM Machines with Concentrated Windings”, IEEE-IEMDC ’03 Conference Proceedings, Madison, Wisconsin, USA, pp. 333–339, June 1–4, 2003. [11] N. Bianchi, S. Bolognani, F. Luise: “Analysis and Design of a Brushless Motor for High Speed Operation”, ibidem, pp. 44–51. [12] N. Bianchi, S. Bolognani, P. Frare: “Design Criteria of High Efficiency SPM Synchronous Motors”, ibidem, pp. 1042–1048. [13] A. Di Gerlando, M. Ubaldini: Italian Patent Application MI2002A 001186, “Synchronous Electrical Machine with Concentrated Coils”, May 31st, 2002; International PCT Patent pend- ing. [14] Maxwell 2D and 3D FEM codes, Vol.10, Ansoft Corporation, Pittsburgh, PA, USA, November 2003. III-1.3. AXIAL FLUX SURFACE MOUNTED PM MACHINE WITH FIELD WEAKENING CAPABILITY J.A. Tapia, D. Gonzalez, R.R. Wallace and M.A. Valenzuela Electrical Engineering Department, University of Concepcion, Casilla 160-C, Correo 3 Concepcion, Chile juantapia@udec.cl, degonzalez@udec.cl, rwallace@udec.cl, anivalenz@udec.cl Abstract. In this paper an axial flux PM machine with field control capability for variable speed application is presented. To achieve such as control, surface mounted PM rotor-pole configuration is shaped so that, a lowreluctancepathisincluded.Inthisway,controllingthearmature reaction basedon vector control allows us to command the airgap flux in a wide range. Magnetizing and demagnetizing effect can be reached with a low stator current requirement. In order to handle the rotor reluctance, an iron and PM sections are include. 3D-FEA is carried out to confirm the viability of the proposed topology. Also a procedure to estimate the d-q parameters for the topology is presented Introduction Permanent magnets (PM)machineshavegainedgreatpopularity due tohigherpowerdensity and efficiency compared to conventional electromagnet excitation [1]. In fact, Modern PM based on NdFeB allows us to mounted directly on the rotor surface provide high airgap flux density [2], with no field losses, reduced volume, and lower requirement for the machine manufactured. However, for variable speed applications, PM machines offer difficulties because of the fix excitation provided for the magnets. Induce voltage increases linearly with frequency reducing the speed range over rated speed [3]. Controlling the airgap flux is the main issue for PM machines. Axial flux machines offer several advantages when high torque and power density [4,5] compared with radial flux topologies. Sandwich configuration allows us to stack several rotors and stators in a single shaft with direct control over the airgap length. These features are required especially in traction and power generation. In this paper,an axialflux surfacemounted PM(AFPM) machine configurationwith field weakening capability is proposed. The topology allows us to control the airgap flux with a reduced d-axis current requirement for operation over the rated speed. To perform such as control, machine design considers a modification of the rotor-pole magnetic configuration including an iron-pole piece. In this manner, a negative d-axis flux component can easily be generated by the armature reaction, so that the total airgap flux is reduced (or increased) accordingly [6]. As a result, magnets are not submitted to any significant demagnetizing field and control action is made over iron mostly. S. Wiak, M. Dems, K. Kom ˛ eza (eds.), Recent Developments of Electrical Drives, 321–334. C 2006 Springer. 322 Tapia et al. 3D finite element analysis (3D-FEA) has been carried out with commercial software FLUX3D provided by MAGSOFT Co. These analyses demonstrate that airgap flux can be commanded with an appropriate armature reaction control. As a result speed range can be increased without significant requirement of the stator d-axis current. In addition a procedure to estimate the d-q parameters based on armature reaction waveform analysis is used. Geometry and iron to magnet ratio define the machine reactances. In the following sections description of the proposed machine, FEA for no-load and load conditions, and parameters procedure calculation are shown. Axial flux machine topology Description The AFPM machine topology proposed is shown in Fig. 1. This machine is composed of two rotors and one central stator. Rotors are north-north (NN) PM surface mounted type containing the excitation poles. Each of these poles is assembled by two parts: PM and iron piece. PM section is a magnet part axially magnetized which provides excitation to the machine. On the other hand an Iron section which offers an easy path for the stator current armature reaction. Due to the short airgap length, the total flux per pole can be considered as two components: one associated to the magnets (high reluctance), the other associated to iron (low reluctance). The stator contains two set of three-thase AC windings (one in front of each rotor) allocated in radial slots. Stator iron yoke completes the rotor-stator magnetic circuit, so that each side can be considered an independent magnetic circuit. Control can be performed separately in each side of the machine. From the construction point of view, stator and rotor yoke, and stator teeth are made using iron lamination, which is compacted by epoxy glue and enrolled as a spiral. In that Permanent Magnets Iron pole Rotor RotorStator Stator AC winding Figure 1. Axial flux surface mounted PM machine for field weakening application. III-1.3. Axial Flux Surface Mounted PM Machine 323 way, radial flux is minimized, avoiding zigzag leakage flux and axial paths are allowed to conduct flux. Operation Below rated speed, the machine is controlled using maximum torque per ampere (MTA) trajectory [3]. dq-Axis currents are calculated according to the machine parameters and operating condition to obtain maximum torque with rated armature current. Over the rated speed, the voltage and current inverters constrain obligate an appropriated current control. In fact, linear variation of the back-emf with the speed makes this internal voltage increase above the rated value. The operating condition became critical due to fix and uncontrollable PM magnetization. Therefore, in order to reduce the total airgap flux, armature reaction is utilizedto demagnetize themachine by controlling ofthe stator current. Phasor diagram and flux distribution for the synchronous PM motor is depicted in Fig. 2. Armature reaction flux, ad , isdivided itin twocomponents. Asis observed, negative d-axis current introduces a flux component, d , which neutralizes to the PM flux. In this manner adequate stator current can be used to control the total flux on the machine. For the circuit point of view, d-axis voltage drop (jX d I d ) compensates the increment in the back-emf (E f ). Observe that both voltages have linear dependence of the frequency (speed). However, in regular PM machine, the amount of d-axis current required to perform such a control, is extremely high due to the large PM reluctance. Using this approach elevated copper losses are generated and PM demagnetization risk reduce its application. −I a E f q-axis −I q I d −I d d-axis V t jI q X q jI d X d I q I a N S ar F aq F ad F δ f g f F res F d-axis demagnetization effect of the armature Permanent Magnet and flux generated Figure 2. Phasor diagram and flux relationship for the salient pole synchronous machine. 324 Tapia et al. The proposed machine topology, total rotor-pole reluctance is modified using a small iron-pole section. In this manner, d-axis flux has two components: one for the magnets and the other for the iron. Because of the low iron reluctance, low stator current is required to perform the airgap flux control. Adequate selection of the PM to iron ratio allows us to adjust machine parameter so that the speed range can be extended for the machine over the rated speed [6]. The optimum operation condition requires at high speed both back-emf and d-axis reactance voltage drop increase in the same proportion, so that their variation are balanced [7]. As a result, stable operation is achieved. Features In addition to the natural axial flux machines advantages, the machine topology presented shows several others, in comparison with the regular PM machine such as: r Wide range of airgap flux control to reduce or increase its value, this is made with low requirement of d-axis current. r This particular configuration allows us to control the level of excitation of the machine without any demagnetization risk for the permanent magnet. r Airgap flux control allows to increase and to improve the power capability at high-speed range of the drive-motor configuration. Drawbacks However, this configuration has some problems that can be summarized as follow: r Lower power density respect to regular PM machine due to the reduction on the amount of magnet. r gsymmetrical flux density distribution over the stator teeth introduces additional satura- tion over the stator and rotor yoke. This is because of their trapezoidal shape. Finite element analysis In order to determine the effectiveness of the proposed configuration a 3D-FEA is carried out. Rotor and stator domain and 3D-mesh used to evaluate the topology is shown in Fig. 3. Stator winding representation is depicted. One detail has to be incorporated in the model. Laminationcoreforthemagneticcircuithasthepropertytocarry fluxmainlyintangential and axial directions. However, due to the interlamination airgap radial flux is reduced considerably. To take into account this effect in the model, additional radial airgaps are introduced in the 3D-FE model. In this manner, flux is forced to flow in the ordinary directions given by the iron permeability and lamination. No-load operation Forno-loadoperation,theonlyexcitationpresentonthemachineisprovidedforthemagnets. Flux density distribution for this operation is depicted in Fig. 4. As expected, there is III-1.3. Axial Flux Surface Mounted PM Machine 325 AC Winding Magnet Iron pole Interlaminatio n airgaps Figure 3. 3D-mesh for one pole of the AFPM machine. (a) Stator with armature winding (b) Rotor. magnetic activity mostly over the magnet area. Due to the no stator current, ar mature reaction does not apply flux over the iron section of the airgap. As a result, flux density is negligible in this area. Total flux crossing airgap correspond that impressed by the PMs. On-load operation Under vector controlstrategy, statorcurrent canbe positioned in any location over theairgap respect to the PM flux. According to the required demagnetization effect, current angle (γ in Fig. 2) is calculated so that necessary d-axis flux is generated to counteract magnet flux. Combined flux density distributionfor maximum d-axis demagnetization effect (I d = 1 pu) and magnet excitation is presented in Fig. 5. For the operating condition indicated, 3D-FEA establishes that armature reaction acts mostly over the iron section and flux density over the magnet has almost no variation respect to the no-load condition. Iron rotor pole offers a very low axial reluctance for the armature reaction respect to the magnet. As a result, flux imposes by the stator current increase consequently and direction over iron section of the airgap is opposite respect to the magnet flux. Total airgap flux is the difference between these two fluxes. Magnetic effort over the magnets is reduced substantially, with minimum demagnetization risk and low current. To evaluate the airgap flux as a function of the d-axis current, the airgap region is divided in two sections: one considering the area in front of the magnet and the other in front of the iron section. Each component of the airgap flux is plotted in Fig. 6. Flux associated to the magnet has minimum variation as the stator current increases. At the same time, iron section flux increases with the current. As a result, total airgap flux is varied according to the d-axis current. Iron-magnet rotor pole Iron to magnet ratio over the rotor pole has an important impact in the flux control capacity in this topology. While the iron section is reduced, saturation diminishes the armature reaction effect over the total airgap flux, as depicted in Fig. 7. As the iron section increases, higher variation of the flux is encountered. In addition a linear dependence respect to the demagnetizing current is presented. However magnets size makes that power density 326 Tapia et al. Z Figure 4. Flux distribution for no-load condition. (a) Stator teeth b) Rotor pole. III-1.3. Axial Flux Surface Mounted PM Machine 327 Z Figure 5. Flux distribution for maximum demagnetization condition. (a) Stator teeth (b) Rotor pole. . Kom ˛ eza (eds.), Recent Developments of Electrical Drives, 321 334 . C 2006 Springer. 322 Tapia et al. 3D finite element analysis (3D-FEA) has been carried out with commercial software FLUX3D provided. range of airgap flux control to reduce or increase its value, this is made with low requirement of d-axis current. r This particular configuration allows us to control the level of excitation of the. these simulations gave the additional information of the absence of torque ripple: this result, confirming the absence of cogging of the unfed machine, appears particularly interesting, also considering