European Journal of Scientific Research ISSN 1450-216X Vol.26 No.4(2009), pp.499-509 EuroJournals Publishing, Inc. 2009 http://www.eurojournals.com/ejsr.

htm

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

S.A. Saied Department of electrical Eng, KNTU University of technology, Iran E-mail: [email protected] K. Abbaszadeh Department of electrical Eng, KNTU University of technology, Iran S. Hemmati Department of electrical Eng, KNTU University of technology, Iran M. Fadaie Department of electrical Eng, KNTU University of technology, Iran Abstract In this paper, a new method for designing of the motor magnets is introduced. In this method the machine magnets consist of several identical segments which are shifted to a definite angle from each other. In the beginning, the dependence of cogging torque on the shifting angle and the number of the segments is mathematically proved. Since cogging torque strongly depends on those parameters, the main goal of this paper is finding the parameter values which minimize the cogging torque. In following for a six pole surfacemounted PM Motor, optimization for finding the numbers of segments and the shifting angles is done. In this optimization, first, the possible arrangements for the magnet segments are simulated by Finite Element Method (FEM) and the cogging torques are analyzed, the result of this process define the best segment arrangement for each magnet pole span. At the end, the arrangements are compared in case of average torque and the optimum one is determined. The results show a noticeable decrease in cogging torque. Keywords: Cogging torque, Permanent magnet motor, Surface mounted, Finite Element Method

1. Introduction
HIGH torque density of surface mounted permanent magnet (PM) motors makes them attractive for industrial applications. Cogging torque is one of the important draw backs of these motors, which results in shaft vibration and noise. Moreover cogging torque causes in pulsating in motor torque and prevents having a smooth rotation, especially at low speed and light loads. In the electrical machine, where the structure is symmetric, cogging torque has a periodic shape versus the mechanical position thus, from frequency analysis point of view, to cancel the vibration related to cogging torque; the magnitude of its harmonics should be reduced properly. There are different approaches to cogging torque reduction; some of these techniques recommend the

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

500

modification of the current wave form, injected into the stator windings (J.P. Martin, et al. October 2000). In structural solution category where the cogging torque originate from the interaction of the rotor magnets and stator slots, there are various techniques for magnet and slot design. A classical one is choosing the magnet pole span, an integer multiple of slot pitch (T. Li, G.R. Slemon, 1988 ; T. Ishikawa, G.R. Slemon, 1993; K.C. Lim, et al. 2002), this solution reduces the magnitude of the first order harmonic of cogging torque. Another recommended technique is Teeth pairing with different tooth width (S.M. Hwang, et al., 2001). Some researchers proposed optimizing of the magnet pole span to pole pitch ratio (T. Li, G.R. Slemon, 1988; Z. Q. Zhu and D. Howe, Dec. 2000; T. Ishikawa, G.R. selmon, 1993; A. Keyhani, et al. Jul. 1999; B. Ackermann, et al. Jul. 1992), and using a fractional number of slots per pole (T. Li, G.R. Slemon, 1988; B. Ackermann, et al. Jul. 1992; J. R. Hendershot and T. J. E. Miller,1994; . C. Hwang, et al. Mar. 1998). In theory, the cogging torque can be vanished by skewing the stator slots or the rotor magnets, the skew angle should be an integer multiple of
Figure 1: Cross Section View of a PM Motor with Magnet Segmentation

Cross section view of one pole

the slot pitch (D.C. Hanselman, 1997). Displacement of the magnets and using auxiliary slots are also recommended (T. Li, G.R. Slemon, 1988; N. Bianchi and S. Bolognani, Sep. /Oct. 2002). Magnet segmentation in which a pole magnet consists of several elementary magnet segments with the same polarities is another technique (R.Lateb, et al., 2006). In this technique the segments are placed, side by side in a pole span (as Fig. 1 shows). This arrangement, from now on is called type-A, in this paper. The new approach proposed in this paper is based on the magnet segmentation .The novelty of this paper lays in defining a shifting angle between the segments, for cogging torque minimization (as is shown in Fig.2). This proposed method makes the segment arrangement more flexible than the suggested method in (R.Lateb, et al., 2006). One of the advantages of this flexibility is the ability to consider higher order harmonics in cogging torque reduction; another is, to have more alternatives for magnet pole span, especially when the slot pitch is large. Obviously the method described in (R.Lateb, et al., 2006), is one of the possible arrangements in the new method, when there is no space left between the magnet segments.

501

S.A. Saied, K. Abbaszadeh, S. Hemmati and M. Fadaie

2. Dependence of Cogging Torque on the Number of Segments and their Shifting Angles
In a permanent magnet motor with Ns stator slots per pole pair, the cogging torque of each magnet segment can be expressed by (T. Ishikawa, G.R. Slemon, 1993):
Ts ( ) =
h =1

Th .Sin(hN s)

(1)

Where Th is the Fourier coefficient of the hth harmonic and is electrical angular position. Where magnet segments and stator slots are identical, using (1) for a reference segment (for example the fifth segment in Fig.2 which is in light color) the torque produced by the other segments can be stated by including an appropriate phase shift in (1). The shift is equal to angular displacement of the segment from the reference one and is shown by in Fig.2 .The formula for the total cogging torque is the sum of the torques of all the segments. Thus:
i Tt (, i ) = Th .Sin(hN s ( + i )) i =1 h =1 N

(2) (3)

Where:
i = K Sw +
j=1

In (2), N is the number of segments in a magnet pole, and K in (3) is the number of segment for which is calculating, from the reference one. Sw is the segment width (in this paper the width of the magnet segment is 15 electrical deg1 for all the cases). j is the gap angular width2 between those segments which are placed among ith of segment and the reference one. Generally it is not possible to find i's which make (2) equal to zero for any given value of theta (0<theta<2), therefore the best alternatives for i's are, those that minimize the following equation:
F = Tt (, i ) 2 =
Tt 2

(, i )d

(4)

In other words, we try to reduce the RMS value of cogging torque by minimizing (4).
Figure 2: Magnet Segment Arrangement in Cross section view of one pole

Where i and h are independent, (2) can be rewritten as:

N i Tt (, i ) = Th .Sin(hN s ( + i )) h =1i =1

(5)

Expanding the Sine function:

Tt (, i ) =
N i Th . Sin ( hN s ).Cos h =1 i =1

(hN s i ) + Cos(hN s ).Sin (hN s i ))

(6)

Factoring by Sin(hNs) and Cos(hNs):

1 2

In this paper all of the angles are in electrical degree As Fig. 2 shows, two neighbor segments can be placed far from each other, therefore a gap creates among them. The width of this gap called gap angular width in this paper.

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

N i N i Tt (, i ) = (Sin (hN s ). Th Cos(hN s i ) + Cos(hN s ). Th .Sin (hN s i ) h =1 i =1 i =1

502 (7) (8)

Finally, using Parseval's theorem, the torque can be expressed as:

Tt (, i ) = h .Sin (hN s ( + h ))
h =1

Where Ah and are functions of i and the number of segments:

N i N i A h = Th .Sin (hN s i ) + Th .Cos(hN s i ) i =1 i =1
2 2

N i N i = Tan 1 Th .Sin (hN s i ) . Th .Cos(hN s i ) i =1 i =1

(9)

Substituting Tt from (8) in (4):

F = Tt (, i ) 2 = (A h .Sin (hN s ( + h ))) 2 d
h =1

(10)

After integrating and simplification:

F = Tt (, i ) 2 =
h =1

(A h )

(11)

Substituting Ah from (9) in (11):

F=
N N i 2 i 2 ( Th .Sin (hN s i )) + ( Th .Cos(hN s i )) h =1 i =1 i =1

(12)

i i In (12), both terms of ThSin (hN si ) and T h Cos ( hN s i ) are functions of i and the number of segments thus, it is possible to minimize cogging torque, using appropriate angles and number of segments. To optimize F some limits like symmetry can be added to the problem to decrease the result domain.

3. Using Symmetric Arrangements

To have a symmetric magnetic field, it is necessary to use symmetric magnet segment arrangements. In the following, the application of the symmetry condition for a specific number of magnet segments is described. The machine which is considered, has the magnets consisting of eight identical segments (the number of segments is arbitrary, but practical consideration should be noticed). The 5th segment (which is the reference one) is always fixed. In order to have a symmetric shape for the machine magnet and to make the process simpler, the 4th segment is attached to the reference one and is fixed as well, therefore the gap angular width between the 4th and 5th segments always remains zero ( as Fig.2 shows). Thus, from (3): 1 = S w + 1 , 2 = 2 * S w + (1 + 2 ) 3 = 3 * S w + (1 + 2 + 3 ) , 0 = 1 * S w (0)
= 3 * S w (1 + 2 + ) 3 3
2 = 2 * S w (1 + 2 )

1 = S w 1

(13) j are the gap angular width between two neighbor segments and from now on, is called gap in this paper. As Fig. 2 shows, a vector of gap for each arrangement can be defined in the form of ['3 '2 '1 0 1 2 3]. It is obvious that each arrangement is uniquely defined by its gap vector. As mentioned before for sake of symmetry: 1 = 1 , 2 = 2 , 3 = (14) 3 Obviously by considering the symmetry, the number of variables in this case is reduced from six to three, therefore any possible result for segment arrangement can be expressed by [3 2 1 0 1 2 3] or in short form of [1 2 3]. It is clear that the magnets of type-A have a gap vector with

503

S.A. Saied, K. Abbaszadeh, S. Hemmati and M. Fadaie

arrays equal to zero thus, these types of magnets, are some of possible arrangements in the new method. One can modify the previous equation for a machine with different number of magnet segment in its poles.

4. Simulation and Finding the Appropriate Shifting Angles as well as the Number of Segments
In this part for a six-pole, three-phase, 18-slot PM motor, a four stage simulation is conducted. In each stage, cogging torque of the machine, related to different designs of its magnet, is calculated by FEM. The stages are defined as below (in all the stages, the characteristics of the machine 3 are the same, except those of the magnets): 1. Stage 1, simulation of type-A magnets of various magnet pole spans. It is done to create a base for further comparisons. 2. In the second stage, when the magnet consists of eight identical segments, the simulation is done for any possible arrangement of these segments. 3. In the third stage, the machine magnets are consisting of seven identical magnet segments in each arrangement. The simulation is done for any of them. 4. In the last stage simulation is done for the machine with six magnet segments is each pole. As mentioned earlier the width of magnet segments are 15 electrical degrees in all cases. In the first stage, where the type-A magnets are considered, it is obvious that there is no gap between two neighbor segments of this type, therefore the magnet pole spans is changed by altering the number of segments in each pole. In other words in this stage the number of magnet segments is variable. In rest of the stages, in contrary with the first one, the number of magnet segments in each pole is kept fixed (eight for the second stage and seven and six for the 3rd and 4th stages respectively), therefore in the last three stages, the magnet pole spans can be changed only, by altering the gap angular width between the segments. These changes in the gap angular width create different arrangements for magnet segments (Fig.4 shows some of different arrangements in the second stage). The simulation result of cogging torque4 in the first stage when the magnet pole span differ from 120 to 180 degree (in this stage it means, the number of magnet segments is changed from eight to twelve) is shown in Fig.3, as it shows the minimum cogging torque is occurred in magnet pole span of 135 degree.
Figure 3: Cogging Torque of the magnets of Type-A versus their Magnet Pole Spans
C ogging Torque [N /M]
3
4 3.5 3 2.5 2 1.5 1 0.5 0

120 129 135 144 150 159 165 174 180

The main machine parameters are: stator outside diameter = 68 mm, air gap length = 0.5 mm, rotor diameter = 36 mm, magnet thickness = 3.5 mm and width of stator slot openings = 2 mm. 4 Peak to peak value of cogging torque corresponding to each magnet pole span is depict in Figure 3, 5, 7 and 9

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

504

In the second stage in which the machine magnets are consists of eight segments only, all the possible arrangements for these eight segments are simulated by FEM and cogging torque related to each case is calculated; it is obvious that the arrangement for a specific magnet pole span in not necessarily unique, therefore, the arrangement with the lowest cogging torque is chosen for each magnet pole span. Table I reveals the best arrangement related to each magnet pole spans in the 2nd stage.
Table I:
Case

Different Arrangements of 8 Segments in Stage 2

[1 2 3] Elec. Degree [0 0 0] [0 3 0] [3 0 3] [3 0 6] [3 9 0] Magnet Pole Span (Elec. Deg.) 120 126 132 138 144 Case [1 2 3] Elec. Degree [3 12 0] [18 0 0] [21 0 0] [24 0 0] [21 0 6] Magnet Pole Span (Elec. Deg.) 150 156 162 168 174

Type-A* Case1 Case2 Case3 Case4

Case5 Case6 Case7 Case8 Case9

*. Type-A with eight segments

As mentioned earlier, the magnet of type-A is one of the possible arrangements of the new method, emerging in each of the last three stages. The data related to this type is expressed, in the first cell of table I to III. Comparison of Fig.3 and Fig.5 shows the effectiveness of the new method in cogging torque reduction in various magnet pole spans. The arrangements expressed in table I, are depicted in Fig.4. For example the case3 in Fig 4, shows the arrangement in which, the gap vector is [6 0 3 0 3 0 6]. As mentioned earlier where the arrangements are symmetric, for sake of simplicity only the last three digits are used in table I to III. The large pieces in Fig. 4, are consisting of two or more segments with no gap between them (in all of the shapes, the magnet is consist of eight segments. These segments are depict clearly in Fig. 2)
Figure 4: Magnet Segment Arrangements in a cross section view of one Pole for cases described in table I

Fig. 5 shows the result of cogging torque for the cases, presented in table I. According to Fig.5, the best arrangement for the magnet segments in the second stage is that of the case3, with a magnet pole span of 138 degree.

505

S.A. Saied, K. Abbaszadeh, S. Hemmati and M. Fadaie

Figure 5: Cogging Torque of Different Arrangements of 8 Segments versus their Magnet Pole Spans
C o g g in g T o rq u e [N /M ]

2 1.5 1 0.5 0 120 126 132 138 144 150 156 162 168 174

Fig. 6 shows the cogging torque of the machine when the magnet arrangements are in form of type-A and case3. As it shows the RMS value of the cogging torque is reduced considerably in case3, in comparison with that of type-A.
Figure 6: Cogging Torque Magnitude of Type-A and Case 3 versus Theta (magnet pole with eight segments)
Cogging Torque (Normalized)

In the following, the third and fourth stages are analyzed in the same way as done for the previous stage. Table II and III show the best arrangements for various magnet pole spans in the third and forth stages. As Fig. 7 and 9 show the arrangements with the lowest cogging torque, have magnet pole span of 147 and 120 degree which are related to stages three and four respectively.
Table II:
Case

Different Arrangements of 7 Segments in Stage 3

[1 2 3] Elec. Degree [0 0 0] [3 0 0] [0 0 6] [0 6 3] [0 0 12] [15 0 0] Magnet Pole Span (Elec. Deg.) 105 111 117 123 129 135 Case [1 2 3] Elec. Degree [0 3 15] [21 0 0] [6 18 0] [0 24 3] [0 24 6] [0 21 12] Magnet Pole Span (Elec. Deg.) 141 147 153 159 165 171

Type-A* Case1 Case2 Case3 Case4 Case5

Case6 Case7 Case8 Case9 Case10 Case11

*. Type-A with seven segments

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

Figure 7: Cogging Torque of Different Arrangements of 7 Segments versus their Magnet Pole Spans
C o g g in g T o r q u e [N /M ]
1 0.8 0.6 0.4 0.2 0

506

105 111 117 123 129 135 141 147 153 159 165 171

Fig. 8 depicts the cogging torque, when the seven magnet segments are in arrangements of type-A and case7 (which are expressed in table II). The reduction of cogging torque RMS value is clearly depicted in this figure.
Figure 8: Cogging Torque Magnitude of Type-A and Case 7 versus Theta (magnet pole with seven segments)
Cogging Torque (Normalized)

Table III: Different arrangements of 6 segments in stage 4

Case [1 2 3] Elec. Degree [0 0 0] [0 3 0] [0 3 3] [0 0 9] [9 3 0] [9 3 3] [6 6 6] [6 6 9] Magnet Pole Span (Elec. Deg.) 90 96 102 108 114 120 126 132 Case [1 2 3] Elec. Degree [6 6 12] [0 0 27] [0 15 15] [30 3 0] [12 21 3] [3 33 3] [3 30 9] Magnet Pole Span (Elec. Deg.) 138 144 150 156 162 168 174

Type-A* Case1 Case2 Case3 Case4 Case5 Case6 Case7

Case8 Case9 Case10 Case11 Case12 Case13 Case14

*. Type-A with six segments

507

S.A. Saied, K. Abbaszadeh, S. Hemmati and M. Fadaie

Figure 9: Cogging Torque of Different Arrangements of 6 Segments versus their Magnet Pole Spans
C o g g in g T o rq u e [N /M ] 0.8 0.6 0.4 0.2 0 90 96 102 108 114 120 126 132 138 144 150 156 162 168 174

Figure 10: Cogging Torque Magnitude of Type-A and Case 5 versus Theta (magnet pole with six segments)

The new magnet arrangements presented in table I, II and III may affect on some other aspect of the machine operation, such as the average torque. Therefore to study this effect and to find the best case among them, all the cases are analyzed by FEM and the average torque of the machine in each case is calculated. Fig.11 reveals the normalized average torque for all the cases presented in table I to III as well as type-A. The points which are named A, B, C and D in Fig.11, show the average torque related to the arrangements having the least cogging torques in stages 1 to 4 respectively. Table IV shows the main characteristics of those points. It reveals that the optimum design is related to that of the point B, if the optimizing object is cogging torque reduction without loosing average torque.

Cogging Torque (Normalized)

A New Approach to Cogging Torque Reduction in Surface-Mounted Permanent-Magnet Motors

Figure 11: Average Torque versus Magnet Pole Span (All Stages)

508

Stage 1 (Type- A) Stage 2 Stage 3 Stage 4

The arrangement related to the point B, consists of eight segments and 138 electrical degree magnet pole span (case 3 in Fig.4).
Table IV: Charactristics of the optimum points of fig 8
The Points in Fig.8 A B C D Cogging Torque [N /M] Average Torque [Normalized] 0.96 0.92 0.82 0.74 Magnet pole Span [Degree] 135 138 147 120 Number Magnet of Segments 9 8 7 6

0.74 0.28 0.17 0.14

5. Conclusion
In this paper for cogging torque reduction, the machine magnets are divided into several identical segments, then a shifting angle is defined for each segments. The dependence of the cogging torque on the number of segments as well as the shifting angles is mathematically proved. The results of FEM simulation justify the mentioned dependence as well. Moreover the simulation results reveal the fact that, the reduction in the number of segments causes in decrease of cogging torque as well as the average torque (as Fig. 5, 7, 9 and 11 show). Also, it is shown that for each magnet pole span, the type-A is not necessarily the best arrangement and for each specific number of segments there is an optimum arrangement which minimize cogging torque (as Fig. 5, 7 and 9 show), this arrangement is defined using the proposed method. In the other hand, for a specific number of segments with a definite width, increase in magnet pole span (by increasing the gap angular width, between the segments) causes in the reduction of the average torque (as Fig. 11 shows). To conclude, the new approach described in this paper which defines a gap angular width between the segments, reduces the cogging torque properly and thanks to its design, the magnetic material necessary for a specific magnet pole span can be reduced in comparison to that of the type-A.

509

S.A. Saied, K. Abbaszadeh, S. Hemmati and M. Fadaie Keyhani, C. Studer, T. Sebastian, and S. K. Murthy, Jul. 1999. Study of cogging torque in permanent magnet machines Electr. Mach. Power Syst.,vol. 27, no. 7, pp. 665678, Ackermann, J. H. H. Janssen, R. Sottek, and R. I. Van Steen, Jul. 1992. New technique for reducing cogging torque in class of brushless motors , Proc. Inst. Elect. Eng.Electr. Power Appl., vol. 139, no. 4, pp. 315320, C. Hwang, S. B. John, and S. S. Wu, Mar. 1998. Reduction of cogging torque in spindle motors for CD-ROM drive, IEEE Trans. Magn., vol. 34, no. 2, pp. 468470, C. Hanselman, 1997. Effect of skew, pole count and slot count on brushless motor radial force, cogging torque and back EMF. IEE Proc. Electron. Power Appl., Vol. 144, No. 5, pp. 325330, J. R. Hendershot and T. J. E. Miller,1994. Design of Brushless Permanent Magnet Motors. Oxford, U.K.: Clarendon, J. P. Martin, F. Meibody-Tabar, B. Davat, October 2000. Multiple-phase Permanent Magnet Synchronous Machine Supplied by VSIs Working under Fault Conditions, IEEE Industry Applications Conference, 2000, 35th IAS Annual Meeting, Roma, Italy, K. C. Lim, J.K.Woo, G.H. Kang, J. P. Hong, G.T. Kim, 2002. Detent force minimization techniques in permanent magnet linear synchronous motors, IEEE Trans. Magn., Vol. 38, No. 2, pp.1157-1160, N. Bianchi and S. Bolognani, Sep./Oct. 2002 .Design techniques for reducing the cogging torque in surface-mounted PM motors, IEEE Trans. Ind. Appl., vol. 38,no. 2, pp. 12591265,. R. Lateb, N. takorabet, F meibody-Tabar, et al, 2006. Design technique for reducing the cogging torque in large surface-mounted magnet motors, Springer, Recent Development of Electrical Drives, pp 59-72, S. M. Hwang, J.B. Eom, Y.H. Jung, D.W. Lee, B.S. Kang, 2001. Various design techniques to reduce cogging torque by controlling energy variation in permanent magnet motors, IEEE Trans. Magn., Vol. 37, No. 4, pp. 28062809, T. Ishikawa, G.R. Selmon, 1993. A method of reducing ripple torque in permanent magnet motors without skewing, IEEE Trans. Magn., Vol. 29,No. 2,pp. 2028-2031, T. Ishikawa, G.R. Slemon, 1993. A method of reducing ripple torque in permanent magnet motors without skewing, IEEE Trans. Magn., Vol. 29, No. 2, pp. 20282031, T. Li, G.R. Slemon, 1988. Reduction of cogging torque in permanent magnet motors, IEEE Trans. Magn., Vol. 24, No. 6, pp. 29012903,. Z. Q. Zhu and D. Howe, Dec. 2000. Influence of design parameters on cogging torque in permanent magnet machines, IEEE Trans. Energy Convers.,vol. 15, no. 4, pp. 407412,

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]