Journal of Materials Processing Technology 161 (2005) 288293

Design of skewed mounted permanent magnet synchronous generators based on 2D and 3D nite element techniques
Themistoklis Kefalas, George Kalokiris , Antonios Kladas, John Tegopoulos
Laboratory of Electrical Machines, Electric Power Division, Faculty of Electrical and Computer Engineering, National Technical University of Athens, 9 Heroon Polytechneiou Street, Athens 15780, Greece

Abstract This paper investigates the effects of rotor skew on the performance of permanent magnet synchronous generators. The main effects are concerning reduction of the electromotive force induced, stator current harmonics and torque ripple. The analysis performed is based on various 2D and 3D nite element techniques. Simulation results are compared to measurements. 2004 Elsevier B.V. All rights reserved.
Keywords: Finite element techniques; Harmonic frequencies; Permanent magnets; Skewed rotors

1. Introduction The distortion of voltage waveforms due to electromotive force harmonics of permanent magnet machines can affect the network power quality [1,2]. In this paper, the effects of rotor skew on electromotive force harmonics in synchronous permanent magnet generators are investigated [35]. The analysis is based on models involving 2D and 3D nite element techniques. The computed results are compared to measurements on a prototype.

into a system of algebraic equations for the discrete problem. The practical problems necessitate usually several tenths of thousands of unknowns. However, appropriate numerical techniques have been developed, enabling to obtain the solution of such systems within reasonable time, even when personal computers are used. It should be mentioned that the 3D problems require considerably higher computational resources than the 2D ones. In the present paper, the 2D nite element model adopted involves vector potential formulation, while the magnetic ux m per pole can be calculated as follows: m =
S1

2. Finite element modeling The method of nite elements, is based on a discretisation of the solution domain into small regions. In magnetostatic problems, the unknown quantity is usually the magnetic vector potential A, and is approximated by means of polynomial shape functions. In two-dimensional cases triangular elements can easily be adapted to complex congurations and rst order elements exhibit advantages in iron saturation representation. The size of elements must be small enough to provide sufcient accuracy [6]. In this way, the differential equations of the continuous problem can be transformed

B dS =
C1

A dl (2Agap )l0 =

(1)

where l0 is the length of the magnetic circuit (m), A the magnetic vector potential, Agap the vector potential value in the middle of the air-gap, B the ux density (Tesla), S1 the cross-sectional area normal to the direction of ux ow (m2 ) and C1 the contour surrounding the surface S1 (m). The electromotive force at no load can be calculated as follows: E= dm dt

Corresponding author. E-mail address: gkalok@mail.ntua.gr (G. Kalokiris).

(2)

0924-0136/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2004.07.038

T. Kefalas et al. / Journal of Materials Processing Technology 161 (2005) 288293

289

Fig. 1. One pole part of the permanent magnet machine constructed (unskewed rotor): (a) employed triangular mesh; (b) eld distribution at no load; (c) magnetic eld distribution at full load.

Fig. 2. No load voltage of permanent magnet generator without rotor skew (2D fem simulation).

Fig. 3. No load voltage of permanent magnet generator without rotor skew (experiment).

290

T. Kefalas et al. / Journal of Materials Processing Technology 161 (2005) 288293

Fig. 4. No load voltage of permanent magnet generator with rotor skew (2D fem simulation).

Fig. 5. Full load voltage of permanent magnet generator without rotor skew (2D fem simulation).

Fig. 6. Full load voltage of permanent magnet generator without rotor skew (experiment).

Fig. 7. Full load voltage of permanent magnet generator with rotor skew (2D fem simulation).

T. Kefalas et al. / Journal of Materials Processing Technology 161 (2005) 288293

291

Fig. 8. Magnetic eld distribution at no load (3D fem simulation): (a) without rotor skew; (b) with rotor skew.

The value of voltage at full load can be calculated by relation (3): V = E RI jL I (3)

where V is the voltage on stator windings (V), E the electromotive force at no load (V), R the stator resistance ( ), L is the stator leakage inductance (H), the rotor angular velocity (rad/s) and I the stator current (A). The magnetic eld distribution at full load is given in Fig. 1c. Then the magnetic ux and electromotive forces can be derived by using Eqs. (1) and (2). The authors propose the study of a permanent magnet machine prototype through a design procedure involving three steps. In a rst step, the typical design procedure is conveniently to this type of machines. In a second step, the machine characteristics are checked by means of a detailed eld calculation through nite element techniques. In a third step, a prototype is constructed in order to validate the machine performance. The effects of rotor skew on magnetic eld distribution in the air-gap as well as on the produced electromotive force waveforms are investigated by simulation.

3. Results and discussion The permanent magnet machine analysis procedure described previously has been applied to construct a 2.5 kW,

26 rad/s generator to be connected to a six pulse diode rectier bridge. The air-gap width has been chosen 1 mm while a multipole peripheral machine structure has been adopted Fig. 1. The geometry of the permanent magnet machine is shown in Fig. 1a giving also the mesh employed for the twodimensional nite element program of the machine involving approximately 2100 nodes and 4000 triangular elements. The corresponding magnetic eld distribution at no load is shown in Fig. 1b. The eld distribution at full load is given in Fig. 1c. In these gures, the dissymmetry caused in the eld distribution by the loading current can be observed. The simulation results concerning no load voltage of synchronous generator without rotor skew are presented in Fig. 2. Fig. 3 gives the measured results under the same operating conditions. A good agreement can be observed in these gures between the simulated and measured results. The effects of rotor skew obtained by using a 2D FEM model and by combining several rotor positions are shown in Fig. 4. It may also be observed that the magnitude of the voltage fundamental frequency is less important in the case with rotor skew [7,8]. The simulation results concerning full load voltage of synchronous generator without rotor skew are presented in Fig. 5. Fig. 6 gives the measured results under the same operating conditions. A good agreement can be observed in these gures between the simulated and measured results also in the case of full load. The effects of rotor skew obtained by successive rotor position by using 2D FEM

292

T. Kefalas et al. / Journal of Materials Processing Technology 161 (2005) 288293

Fig. 9. Flux density distribution in the air gap in the middle of the rotor (3D and 2D fem simulation).

Fig. 10. No load distribution of air-gap radial magnetic ux density (3D fem simulation): (a) without rotor skew; (b) with rotor skew.

simulations are shown in Fig. 7. Once again the smaller magnitude of stator voltage fundamental frequency in the case with rotor skew can be observed. In Fig. 8a and b are presented the distributions of magnetic eld at no load in one pole of the machine by using 3D simulation in the cases of unskewed and skewed rotors, respectively. Fig. 9 gives a comparison of the simulated ux density distribution in the middle of the air-gap with and without rotor skew by 3D FEM simulation and the same distribution obtained by 2D simulation. From the results, it comes up that the magnitude of magnetic ux density is less with rotor skew. In Fig. 10a is presented the magnitude of magnetic ux density in the air gap of the machine without rotor skew by using 3D analysis, while in Fig. 10b is presented the same analysis with rotor skew. The same remark that the

magnitude of magnetic ux density is less with rotor skew can equally be observed.

4. Conclusions The results obtained by using 2D and 3D nite element models provided sufcient accuracy with the measured ones in a permanent magnet synchronous generator prototype. The simulations performed have shown that the skew of permanent magnets in the rotor results in a reduction of the fundamental component of the electromotive force induced because of the reduction in the ux density in the air-gap. So, the stator current harmonics and torque ripple are estimated to be subsequently reduced in the case with rotor skew.

T. Kefalas et al. / Journal of Materials Processing Technology 161 (2005) 288293

293

References
[1] C. Marchand, Z. Ren, A. Razek, Torque optimization of a buried permanent magnet synchronous machine by geometric modication using FEM, Proceedings EMF94, Leuven, 1994, pp. 5356. [2] T. Higuchi, J. Oyama, E. Yamada, E. Chiricozzi, F. Parasiliti, M. Villani, Optimization procedure of surface permanent magnet synchronous motors, IEEE Trans. Magnet. 33 (2) (1997) 19431946. [3] A. Toba, T. Lipo, Generic torque maximizing design methodology of surface permanent magnet Vernier machine, IEEE Trans. Ind. Appl. 36 (6) (2000) 15391546. [4] G. Tsekouras, S. Kiartzis, A. Kladas, J. Tegopoulos, Neural network approach compared to sensitivity analysis based on nite element technique for optimization of permanent magnet generators, IEEE Trans. Magnet. 37 (5/1) (2001) 36183621.

[5] D.C. Aliprantis, S.A. Papathanassiou, M.P. Papadopoulos, A.G. Kladas, Modeling and control of a variable-speed wind turbine equipped with permanent magnet synchronous generator, International Conference on Electrical Machines, Helsinki, Finland, 2000, pp. 558562. [6] T. Nakata, N. Takahashi, K. Yoneda, Numerical analysis of ux distribution in permanent-magnet stepping motors, IEEE Trans. Magnet. 14 (5) (1978) 548550. [7] M.R. Dubois, H. Polinder, J.A. Ferreira, Contribution of permanentmagnet volume elements to no-load voltage in machines, IEEE Trans. Magnet. 39 (3) (2003) 17841792. [8] M.A. Alhamadi, N. Demerdash, Modeling and experimental verication of the performance of a skew mounted permanent magnet brushless dc motor drive with parameters computed from 3D FE magnetic eld solutions, IEEE Trans. Energy Conversion 9 (1) (1994) 135.