N°434

1

Soft Magnetic Composite Axial Flux Seven-Phase Machine F. Locment, E. Semail, and F. Piriou

Abstract—A fault-tolerant seven-phase axial-flux permanent magnet (AFPM) machine with a soft magnetic composite (SMC) stator has been designed thanks to analytical and 3D finite element methods. In this paper, we present its characteristics and show that a high number of phases allows to get a good quality of the torque with a design simpler than this one of the three-phase AFPM machines. Experimental back-electromotive forces (backEMF) and the cogging torque are then compared with predeterminations. Finally, experimental torque of a drive composed of a seven-leg Voltage Source Inverter (VSI) supplying the machine in a vector control is presented. Index Terms—multiphase machine, axial flux, soft composite material.

R

I. INTRODUCTION

ARE-earth magnets with high flux density reduce the importance of a high permeability for electromagnetic devices. When the use of laminated steel is not easy, other materials such as SMC can become interesting [1]-[3]. The AFPM machines are an example since one of their critical issues is the manufacturing of the stator with laminated steel. If we suppose that the core is made of one sheet, the axial direction of the flux imposes to wound this steel sheet in a spiral fashion to form a ring. Moreover if the stator must be slotted, it is still more difficult because for example it is not easy to preserve the insulation between the steel sheets during the manufacturing. Compared with solid steel which is also easy to work, the SMC presents a higher resistivity and prevents therefore eddy currents. Consequently, various slotted AFPM machines with SMC have been developed [4][6]. For the AFPM machine, another issue is the winding arrangement. Compared with the radial flux machines, there is an asymmetry for the windings since there is less space at inner radius than at the outer radius. So, simple windings are used with usually one conductor per pole and per phase. The consequence is that the back-EMF are not sinusoidal [6], [7] unless using special shapes of magnets [3], [8], [9]. If the Manuscript received June 30, 2006. This work is part of the project ’Futurelec2’ within the ’Centre National de Recherche Technologique (CNRT) de Lille’ with EDF and AREVA/Jeumont. F. Locment and E.Semail are with the L2EP, ENSAM, 8 Bd Louis XIV 59046 FRANCE, (e-mail: [email protected]). F. Piriou is with the L2EP, USTL, Bât P2 59655 FRANCE, (e-mail: [email protected]).

machine has only three phases this implies a difficulty to get a torque without ripples since there are interactions between the first harmonic of current with the 5th and 7th harmonics of back-EMF [10]. That is a problem when the AFPM machine must be used in low speed drives without reduction gear. For multiphase machines it has been shown that in steady state the first harmonic of the pulsating torque due to interaction between back-EMF and currents increases with the number of phases. Moreover the torque ripple can be theoretically equal to zero even with non-sinusoidal backEMF due to the fact that interactions between harmonics are different for multi-phase machines [11], [12]. For multiphase machines it is not necessary to achieve complex shapes of magnets in order to get low torque ripple. The active surface of the rotor magnets is then more important. The last issue widely studied for AFPM machine is the cogging torque. We will discuss only of the case of AFPM machine with two external rotors and one stator since the machine presented in the paper is of this type. In order to reduce the cogging torque, various techniques are proposed for 3-phase machines such as shifting between the two rotor discs [7], magnet skewing [7], [9], [10], various pole pitches [12] or stator side displacement [15]. For our multiphase machine shifting between the two rotor discs has been chosen. This simple technique is more interesting with a 7-phase machine than with a 3-phase one. Besides, multi-phase machines offer also another way to reduce the cogging torque by using fractional winding. A 9-phase AFPM machine is studied in [16] and a 7-phase radial-flux machine in [17]. Moreover, multiphase machines are intrinsically more reliable than 3-phase ones [18]. This property that could be interesting for off-shore wind generator and more generally for embedded systems is studied in [19]. In the paper, the experimental machine is presented and main features are justified. Then experimental results such as back-EMF and cogging torque are compared with results obtained by 3D finite element method (3D-FEM) software1. Finally, we examine in generator mode the drive composed of the AFPM machine and a seven-leg VSI: efficiency and torque in steady state are presented.

1

CARMEL developed in our laboratory

N°434

2 II. PRESENTATION OF THE PROTOTYPE

Among the various existing structures of AFPM machine, the studied prototype (Fig. 1) is a slotted TORUS AFPM machine (called TORUS-S in [9] and [10]). It possesses one torus-shape stator (Fig. 2) sandwiched between two external rotors (Fig. 1 and Fig. 3). Even if slotless TORUS AFPM machine [19] have not cogging torque and are easier to manufacture than AFPM machine with slots, the weak values of inductances and consequently of the time constants imply that the carrier frequency of the Pulse Width Modulation (PWM) VSI must be high. It is the reason why the TORUS-S configuration has been chosen. This kind of AFPM machine has still two different topologies depending on the flux paths (see Fig. 4). They are described in [9] and called NN type TORUS-S and NS type TORUS-S. The studied prototype is an NN type TORUS-S AFPM machine. This kind of topology allows a Gramme-Ring winding which presents fewer constraints for end-winding at inner radius than the NS type TORUS-S AFPM machine. On the contrary the thickness of the stator is more important.

Fig. 3 One of the two rotors with shape of one magnet.

N

S

Φ

N

S

Φ

S

N

(a) NN type

N

S

(b) NS type

Fig. 4 Two different topologies for TORUS-S AFPM machine. TABLE I PARAMETERS AND MACHINE DIMENSIONS Number of poles (2p) Number of phases Number of slots Pole-arc-ratio Core outer diameter Do (mm) Core inner diameter Di (mm) Axial length of stator core (mm) Axial length of one rotor core (mm) Axial length of one magnet (mm) Length of one airgap (mm) Slot dimensions (mm)

Fig. 1 One sixth of the machine.

Fig. 2 Torus stator of the AFPM machine.

The characteristics of the prototype are resumed in table I. Designed in our laboratory, the prototype has been manufactured by Selem Electrotechnologies ([email protected]).

Number of wires in a slot Slot fill factor (%) Wire type Core soft composite material type (QMP) Magnet type Mean flux density in airgap (T) Cogging torque peak (Nm) Cogging torque p.u. Electrical loading at (Do+Di)/4 (kA/m) Nominal Speed (rpm) Nominal Torque with 50°C in winding (Nm) Nominal power at 750 rpm (kW) Nominal current at 65 Nm (A)

6 7 42 0.8 310 154 78.3 18.4 2.8 1 11.5 × 6.3 40 44 AWG18 ATOMET EM1 NdFeB N48 0.7 4 0.06 11,5 750 65 5.1 5

A 4/5 arc pole for the magnet repartition (Fig. 3) allows the cancellation of the fifth harmonic. The reason of this choice is linked with considerations of vector control of the machine. A wye-coupled 7-phase machine can be considered effectively as the association of three 2-phase fictitious machines M1, M2 and M3 magnetically independent [12]. Each machine is characterized by a harmonic family given in

N°434

3

Table II. In order to implement algorithms of vector control already developed for 3-phase machines it is sufficient that for each fictitious machine only one harmonic exists [12]. Considering that high harmonics are naturally weak and will be negligible, it has been chosen to cancel the fifth harmonic. The M2 (resp. M1 and M3) machine will be then characterized by the ninth harmonic (resp. the first and the third harmonics). Each 2-phase fictitious machine has then almost sinusoidal back-EMF. For each one of this machine, it has been implemented algorithms of vector control developed for 3-phase machines with sinusoidal back-EMF. TABLE II HARMONIC CHARACTERIZATION OF THE THREE FICTITIOUS MACHINES FOR WYE-COUPLED 7-PHASE MACHINE Fictitious 2-phase machines

Families of odd harmonics

M1 M2 M3

1, 13, 15, 27, …, 7h ± 1 5, 9, 19, 23, …, 7h ± 2 3, 11, 17, 25, …, 7h ± 3

In addition, the choice of the number of phases is linked with three other simple constraints: Cr1. the ability to work in fault mode without too much torque ripple, Cr2. the ability to use the third harmonic to increase the torque density of the machine, Cr3. a minimum number of legs for the VSI which is supposed to supply the machine. These three constraints eliminate the 5-phase machine for which it is difficult to verify simultaneously the constraints Cr1 and Cr2. The double-star 6-phase machine can not verify the Cr2 constraint with only six legs. III. 3D-MODELLING OF THE MACHINE AND EXPERIMENTAL RESULTS

A. Magnetic flux density distribution The magnetic flux density distribution with no-load is given in Fig. 5 for the machine. In the stator core the maximum of flux magnetic density is about 1.2 T and 1.5 T in rotor core. In the airgap, the mean value of density is 0.7 T (Fig. 6) with peak values at 1.1 T (Fig.7).

Fig. 5 Magnetic flux density distribution for one sixth of the machine.

Fig. 6 Magnetic flux density distribution under a pole in the airgap.

Fig. 7 Magnetic flux density distribution in the airgap under a pole at mean radius (Do+Di)/4.

B. Axial forces and cogging torque Using electromagnetic pressure B2 ( 2µ 0 ) it is easy to estimate the axial force existing between one rotor and the stator. An approximate value of 7150 N is confirmed by 3D determination (Fig. 8). The cogging torque has been predetermined by 3D-FEM. If there is not a shifting between the two rotors, the peak value of the cogging torque (16 Nm) represents 25% of the nominal torque.

Fig. 8 Axial force between one rotor and the stator under one pole.

N°434 A shifting of one slot (mechanical angle 4.3°) has been introduced (mechanical angle). The cogging torque drops to 4 Nm which is acceptable (0.06 p.u.). This method gives good result because the mechanical angle is not too weak and consequently not too much sensitive to positioning of the rotors. Moreover as a 4.3° mechanical angle corresponds with a 12.9° electrical angle for the studied seven-phase machine, the first harmonic of back-EMF is reduced by only a factor 0.994. Experimental cogging torque measured thanks to a TM 211 MAGTROL torque transducer gives result in agreement with 3D predeterminations (Fig. 10). It must be noticed that a three-phase machine with the same number of slots (42) and also one slot per pole and per phase has 14 poles. A shifting of one slot (4.3° mechanical angle) represent an electrical shifting of 30° for a three phase machine. The first harmonic of back-EMF of the three-phase machine is reduced by a factor 0.966.

4 harmonics with the fundamental is of the same order (Fig. 12). The cancellation of the fifth harmonic is confirmed. The shifting between the two rotors has also a positive effect on the spectrum of the back-EMF since the 13th harmonic is cancelled. It is interesting because this harmonic belongs to the family of the M1 machine and can consequently interact with the first harmonic of current to create torque ripples.

Fig. 11 Experimental and 3D-FEM back-EMF of one phase at 1 rd/s.

Fig. 9 Predicted 3D-FEM cogging torques with and without shifting between the two rotors.

Fig. 12 Comparison of spectrum of back-EMF.

IV. EXPERIMENTAL INVESTIGATION IN GENERATOR MODE

Fig. 10 Experimental and 3D-FEM cogging torques in case of a mechanical shifting of 4.3°.

C. Back-EMF Experimental measurements of back-EMF at no-load confirm 3D-FEM predetermination (Fig. 11). There is a relative error of 6 % for the first harmonic (experimental first harmonic: 44 V for 250 rpm). Proportionality of the

A. Presentation of the drive The seven-phase machine is driven by a DC motor (Fig. 13). The carrier frequency of the PWM seven-leg VSI has been set to 20 kHz. In generator mode, the drive of the sevenphase machine and the seven-leg VSI supplies a 5 kW electronic charge connected in parallel with a DC supply which imposes a 300 V DC voltage. A vector control, depicted in Fig.17, allows us to impose the value of the torque. The digital control is implemented on a dSPACE DS1005 Controller board. Currents and back-EMF are sampled at 6.6 kHz after a second-order antialiasing filter with a cut-off frequency at 19 kHz. Consequently, the parasitic effects of the PWM of the VSI are not observed on the currents.

N°434

5

Fig. 13 Experimental set-up of the prototype.

Fig. 15 Experimental torques for the two considered tests.

B. Torque in steady state Two different tests have been achieved at a 250 rpm speed (12.5 Hz for the first harmonic of current). In the first test, only the M1 machine has been used to obtain a mechanical torque of 65 Nm on the shaft. The currents are sinusoidal (Fig. 14). As the back-EMF of M1 machine is almost sinusoidal (cancellation of the 13th harmonic) torque ripple should be negligible. Experimental torque delivered by the torque transducer with a bandwidth of 5 kHz confirms this prediction. Even if the signal contains perturbations we can not observed significant pulsations at the 14th harmonic (175 Hz). In the second test, we have used also the M2 machine to produce a little more torque. As the 3rd harmonic of back-EMF represent 20% of the 1st harmonic, we have imposed i M3q = 0.2 × i M1q in order to work at maximum torque for a

C. Efficiency at nominal torque First tests of efficiency of the drive (7-phase machine and VSI) in generator mode have been achieved. As thermal modeling is not complete, temperature has been measured in the windings thanks to a type K thermocouple in order to verify that the values of temperatures are acceptable. The mechanical torque has been imposed to 65 Nm and measured with the torque sensor. The speed has been tuned by the DC drive and active power of the electronic DC load is measured. Since the torque is constant the copper losses also. Consequently the efficiency is increasing with the speed. It will be interesting to deduce experimentally in real conditions losses in the SMC for further works.

given modulus of the vector current. We can observe in Fig. 15 the same form of torque with an average value of 67.5 Nm which corresponds with the predicted increase of + 4 %.

Fig. 16 Efficiency of the drive for a torque of 65 Nm.

V. CONCLUSION

Fig. 14 Experimental current in one phase at 250 rpm speed for different repartitions of the torque between the M1 and M3 fictitious machines.

Due to manufacturing constraints, the structures of the AFPM machines are usually simple. The filtering of harmonics by winding is not easy. Consequently back-EMFs are not sinusoidal unless complex shape forms of the magnets. For applications where smooth torque is required the threephase machines are not well adapted. A seven-phase machine has been proposed in this paper. With a simple manufacturing it is possible to reach very good performances in terms of

N°434

6

pulsating torque. Experimental results confirm the predeterminations obtained by 3D-FEM for the sensitive quantities which are the harmonics of back-EMF and the cogging torque. Finally, a vector control confirms that the mechanical torque has not significant pulsations. Moreover, this machine, simple to manufacture, has also an aptitude to fault mode operation. This feature is developed in [19]. REFERENCES [1]

[2] [3] [4]

[5] [6] [7]

[8] [9]

A.G. Jack, B.C. Mecrow, P. G. Dickinson, D. Stephenson, J. S. Burdess, N. Fawcett, and J. T. Evans, “Permanent-Magnet Machines with Powdered Iron Cores and Prepressed Windings”, IEEE Trans. on Industry Appl., Vol. 36, issue 4, pp. 1077-1084, July-Aug 2000. J. Cros, P. Viarouge, and A. Halila, “Brush DC motors with concentrated windings and soft magnetic composites armatures”, Proc. of IEEEIAS’01, Vol.4, pp 2549-2556, Chicago (USA), Oct. 2001. S.M. Madani, T.A. Lipo, C.E. Nino, and D. Lugo, “A New Trapezoidal Shaped-Pole Permanent Magnet Machine”, Proc. of IEEE-IEMDC’05, pp. 1715-1719, San Antonio (USA), May 2005. A.G. Jack, B.C. Mecrow, G. Nord, and P.G. Dickinson, “Axial Flux Motors Using Compacted Insulated Iron Powder and Laminations – Design and Test Results”, Proc. of IEEE-IEMDC'05, pp. 378-385, San Antonio (USA), May 2005. P. Anpalahan, A. Walker, S. Meister, S. Tsakok, and M. Lampérth, “Axial Flux Machine Stator Construction with Concentrated Windings”, Proc. of ICEM’04, Sept 2004, Cracow (Poland), CD-ROM. Y. Chen, and P. Pillay, “Axial-flux PM wind generator with a soft magnetic composite core”, Proc. of IEEE-IAS’05, Hong-Kong, Vol. 1, pp. 231-237, Oct. 2005. F. Caricchi, F. G. Capponi, F. Crescimbini, and L. Solero, “Experimental study on reducing cogging torque and no-load power loss in axial-flux permanent-magnet machines with slotted winding”, IEEE Trans. on Industry Appl., Vol. 40, issue 4, pp. 1066-1075, July/Aug. 2004. A. Parvianen, M. Niemela and J. Pyrhonen, “Modeling of Axial Flux Permanent-Magnet Machines”, IEEE Trans. on Industry Appl., Vol. 40, issue 5, pp. 1333-1340, Sept/Oct 2004. S. Huang, M. Aydin and T.A. Lipo, “TORUS Concept Machines: PrePrototyping Assessment for Two Major Topologies”, Proc. of IEEEIAS’01, pp. 1619-1625, Chicago (USA), Oct. 2001.

0

Repartition

Tref

iM1qref

vM1αref

PI

++-

R(θ) PI

+-

of reference

0

torques

iM2qref

R(9θ) PI

+-

iM3qref

PI +-

vM2βref

M1α M1β M2α M2β M3α M3β

R(3θ) PI iM1d

iM1α R-1(θ)

iM2d iM2q

iM1β

vBref

[ C7 ]

vDref

PWM

7-Phase

vEref

VSI

AFPM

vFref ABCDEFG

vGref

M1α M1β M2α M2β M3α M3β

iM2α R-1(9θ)

iM3d iM3q

vM3βref

vAref

vCref

vM3αref

iM1q ⎡ cos pθ − sin pθ ⎤ R(θ) = ⎢ ⎥ ⎣ sin pθ cos pθ ⎦

vM1βref vM2αref

PI

+-

0

[10] S. Huang, M. Aydin, and T.A. Lipo, “Torque Quality Assessment and Sizing Optimization for Surface Mounted Permanent Magnet Machines”, Proc. of IEEE-IAS’01, pp. 1603-1610, Chicago (USA), Oct. 2001. [11] H.A. Toliyat, T.A. Lipo, and J. C. White, “Analysis of a Concentrated Winding Induction Machine for Adjustable Speed Drive Application Part 2 (Motor design and Performance)”, IEEE Trans. on Energy Conversion, Vol. 6 no. 4, pp. 684-692, Dec. 1991. [12] E. Semail, X. Kestelyn, and A. Bouscayrol, “Right Harmonic Spectrum for the back-electromotive force of a n-phase synchronous motor”, Proc. of IEEE-IAS’04, Vol. 1, pp.71-78, Seattle (USA), Oct. 2004. [13] H-M Ryu, J-W Kim and S-K Sul, “Synchronous Frame Current Control of Multi-Phase Synchronous Motor, Part I. Modeling and Current Control Based on Multiple d-q Spaces Concept Under Balanced Condition”, Proc. of IEEE-IAS’04, Vol. 1, pp. 56-63, Seattle (USA), Oct. 2004. [14] M. Aydin, R. Qu, and T. A. Lipo, “Cogging torque minimization technique for multiple-rotor, axial-flux, surface-mounted-PM motors: alternating magnet pole-arcs in facing rotors”, Proc. of IEEE-IAS’03, Vol. 1, pp. 555-561, Salt Lake City (USA), Oct. 2003. [15] A. Letelier, J. A. Tapia, R. Wallace, and A. Valenzuela, “Cogging Torque Reduction in an Axial Flux PM Machine with Extended Speed Range”, Proc. of IEEE-IEMDC'05, pp. 1261-1267, San Antonio (USA), May 2005. [16] D. Vizireanu, X. Kestelyn, S. Brisset, P. Brochet, and E. Semail, “Experimental tests on a 9-phase Direct Drive PM Axial-Flux Synchronous Generator”, Proc. of ICEM’06, Chania, (Greece), Sept. 2006, CD-ROM. [17] F. Scuiller, J.F. Charpentier, E. Semail, and S. Clenet, “A global design strategy for multiphase machine applied to the design of a 7-phase fractional slot concentrated winding PM machine”, Proc. of ICEM’06, Chania, (Greece), Sept. 2006, CD-ROM. [18] T.M. Jahns, “Improved reliability in solid state ac drives by means of multiple independent phase-drive units”, IEEE Trans. on Industry Appl., vol. IA-16, pp. 321-331, May-June 1980. [19] F. Locment, E. Semail, and X. Kestelyn, “A vector controlled axial flux seven-phase machine in fault operation”, Proc. of ICEM’06, Chania, (Greece), Sept. 2006, CD-ROM. [20] B.J. Chalmers, W. Wu, and E. Spooner, “An axial-flux permanentmagnet generator for a gearless wind energy system”, IEEE Trans. on Energy Conversion, Vol. 14, n°2, pp. 251-257, June 1999.

[ C7 ]

−1

iM2β

iA,iB,iC,iD,iE,iF,iG

iM3α R-1(3θ)

iM3β

ABCDEFG

θ Fig. 17. Synoptic of the vector control of the seven-phase machine.