Date post: | 17-Dec-2015 |
Category: |
Documents |
Upload: | ashley-dean |
View: | 228 times |
Download: | 0 times |
Torque in Permanent Magnet Couplings : Comparison
of Uniform and Radial Magnetization
Romain Ravaud, Guy Lemarquand, Valerie Lemarquand, Claude Depollier
To cite this version:
Romain Ravaud, Guy Lemarquand, Valerie Lemarquand, Claude Depollier. Torque in Perma-nent Magnet Couplings : Comparison of Uniform and Radial Magnetization. Journal of AppliedPhysics, American Institute of Physics (AIP), 2009, 105, pp.053904. .
HAL Id: hal-00366546
https://hal.archives-ouvertes.fr/hal-00366546
Submitted on 9 Mar 2009
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
Larchive ouverte pluridisciplinaire HAL, estdestinee au depot et a` la diffusion de documentsscientifiques de niveau recherche, publies ou non,emanant des etablissements denseignement et derecherche francais ou etrangers, des laboratoirespublics ou prives.
1Torque in PM Couplings: Comparison of
Uniform and Radial Magnetization
R. Ravaud, G. Lemarquand, V. Lemarquand and C. Depollier
Abstract1
We present a three-dimensional study of the torque transmitted between tile permanent magnets2
uniformly magnetized. All this study is based on the Coulombian model. The torque is calculated semi-3
analytically by considering all the surface densities that appear on the tiles. In addition, no simplifying4
assumptions are done in the expressions given in this paper. Consequently, the evaluation of the torque is5
very accurate and allows us to show the drawbacks of using tile permanent magnets uniformly magnetized6
instead of using tile permanent magnets radially magnetized. Such an approach is useful because it allows7
us to realize easily parametric studies.8
Index Terms9
Magnetic coupling, Tile permanent magnet, Torque, Uniform magnetization, Radial magnetization,10
Three-dimensional calculation.11
I. INTRODUCTION12
MAGNETIC couplings are often realized with tile permanent magnets radially or uniformly13
magnetized. Tile permanent magnets radially magnetized allow us to obtain great couplings14
but they are difficult to fabricate. Consequently, an alternative experimental method consists in using15
tiles uniformly magnetized. Indeed, they are simpler to fabricate than tiles radially magnetized and thus16
cheaper. Unfortunately, they are also less efficient and can lower the quality of transmission between tiles17
located on the stator and tiles located on the rotor. Therefore, it is interesting to predict theoretically the18
Manuscript Received October 25,revised December 8, 2008
The authors are with the Laboratoire dAcoustique de lUniversite du Maine UMR CNRS 6613, Avenue Olivier Messiaen, 72085Le Mans Cedex 9, France
2way the torque is transmitted between tile permanent magnets uniformly magnetized. Such a model allows19
us to realize easily parametric studies and thus to know when the use of tile permanent magnets radially20
magnetized is necessary or not. Indeed, the angular width has a great influence on the radial field created by21
a tile permanent magnet uniformly magnetized. As this radial field is not perfectly symmetrical, the torque22
transmitted is modified. Therefore, the aim of this paper is to determine precisely with a three-dimensional23
model how the torque transmitted between tiles changes according to the way the tiles are magnetized.24
Many authors have studied magnetic couplings [1]-[4] and structures using permanent magnets [5]-[10].25
Historically, we can say that the first models used to study magnetic couplings were the two-dimensional26
models [11]-[19]. The main reason lies in the fact that a two-dimensional approach is fully analytical and27
allows us to make an easy parametric optimization of the permanent magnet dimensions. However, this28
2D-approach is not valid when we determine the magnetic field far from one magnet [20].29
Tree-dimensional approaches seem to be more difficult to realize parametric studies because they are30
not fully analytical. In fact, the calculation of the radial and axial magnetic field components created by31
tile permanent magnets is not strictly analytical but is necessary based on special functions [21]- [40].32
More generally, all the semi-analytical or analytical approaches used by several authors allow manu-33
facturers to optimize devices using permanent magnets [41]-[48].34
This paper presents a semi-analytical approach based on the Coulombian model for calculating pre-35
cisely the torque transmitted between tile permanent magnets uniformly magnetized. This semi-analytical36
approach is three-dimensional. We explain in the second section how this problem can be solved. Then,37
we present a semi-analytical expression of the torque transmitted between two tile permanent magnets38
uniformly magnetized. Eventually, we present the main drawbacks of tile permanent magnets uniformly39
magnetized.40
II. MODELING TILE PERMANENT MAGNETS WITH THE COULOMBIAN MODEL41
This section presents the geometry studied and the model used for the modeling of the torque transmitted42
between two tile permanent magnets uniformly magnetized.43
3b
1
2
3
4
u
uu
r
z
r rr
r in1 out1out2in2
0
z
z
c
d
za
z
Fig. 1. Representation of the geometry considered in three-dimensions : two tile permanent magnets uniformly magnetized
A. Notation and geometry44
The geometry considered and the related parameters are shown in Fig 1. A two-dimensional repre-45
sentation of the geometry is shown in Fig 2. We consider here two tile permanent magnets uniformly46
magnetized. The inner radius of the outer tile is rin1 and its outer one is rout1. Its height is zb za and47its angular width is 2 1. The inner radius of the inner tile is rin2 and its outer one is rout2. Its height48is zdzc and its angular width is 43. The two magnetic polarization vectors ~J1 and ~J2 are expressed49as follows:50
~J1 = cos(2 + 12
)~ur sin(2 + 12
)~u (1)
51
~J2 = cos(4 + 32
)~ur sin(4 + 32
)~u (2)
The magnetic field created by tile permanent magnets can be obtained by using the Coulombian Model.52
Indeed, a permanent magnet can be represented by a magnetic pole surface density that is located on53
the faces of the magnet and a magnetic pole volume density that is located inside the magnet. In the54
configuration presented in Fig 1, the magnetic pole volume density is 0 because the magnetic polarizations55
of the magnets are uniform. Consequently, each tile permanent magnet is modelled by its magnetic pole56
surface density, which is determined as follows: by denoting ~nij , the unit normal vector of the face i of57
42
u
u
r0
1
2
3
4
n
n
n
n
n
nn
n
1
2
3
4
5
67
8
J J1
Fig. 2. Representation of the geometry considered in two dimensions : two tile permanent magnets uniformly magnetized
Surface densities Scalar Product Expression
11~J1.~n1 J1 cos( 1+22 )
12~J1.~n2 J1 sin(
212 )
13~J1.~n3 J1 cos( 1+22 )
14~J1.~n4 J1 sin(
212 )
25~J2.~n5 J2 cos( 3+42 )
26~J2.~n6 J2 sin(
432 )
27~J2.~n7 J2 cos( 3+42 )
28~J2.~n8 J2 sin(
432 )
TABLE I
DEFINITION OF THE MAGNETIC POLE SURFACE DENSITIES LOCATED ON THE MAGNETS
the magnet j, the corresponding surface density ij is determined with (3).58
ij = ~Jj . ~nij (3)
Thus, four faces of the two magnets are charged with a magnetic pole surface density. All the surface59
density calculations are represented in Table I. The torque transmitted between two tile permanent magnets60
can be determined in two steps. The first step consists in calculating the magnetic field created by one tile61
permanent magnet. The second step consists in integrating the magnetic field created by the first tile on62
5the second one. Let us first consider the tile permanent magnet located on the right of Fig 1. By using the63
Coulombian approach, we can write that the azimuthal field created by this tile is expressed as follows:64
H(r, , z) =J1
40
21
zbza
cos( 1 + 22
)~u(rin1)
|~u(rin1)|3 rin1ddz
J140
21
zbza
cos( 1 + 22
)~u(rout1)
|~u(rout1)|3 rout1ddz
+J1
40
rout1rin1
zbza
sin(2 1
2)~v(2)
|~v(2)|3 drdz
+J1
40
rout1rin1
zbza
sin(2 1
2)~v(1)
|~v(1)|3 drdz
(4)
where65
~u(x) = (r x cos( ))~ur x sin( )~u + (z z)~uz (5)
and66
~v(y) = (r r cos( y)~ur r sin( y)~u + (z z)~uz (6)
The next step is thus to express the torque transmitted to the second tile permanent magnet uniformly67
magnetized (as shown in Fig 1). By using (4), the torque T can be determined as follows:68
T =J1J2
40
43
zdzc
rin2H(rin2,, z)rin2 cos(
3 + 4
2)d
dz
J1J240
43
zdzc
rout2H(rout2,, z)rout2 cos(
3 + 4
2)d
dz
+J1J2
40
rout2rin2
zdzc
rH(r, 4, z) sin(4 3
2)drdz
+J1J2
40
rout2rin2
zdzc
rH(r, 3, z) sin(4 3
2)drdz
(7)
B. Semi-analytical Expression of the Torque69
The torque transmitted between two tile permanent magnets can be written as follows:70
T =J1J2
40
( 21
43
dT(1) + T
(2) +
zbza
dT(3) +
21
dT(4)
)(8)
60 0.25 0.5 0.75 1 1.25 1.5Angle @radD
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
Torq
ue@N
.mD
Fig. 3. Representation of the torque transmitted between two tiles ; thick line: the magnetization is uniform, dashed lines:the magnetization is radial; 4 3=2 1=
pi
12, rin1 = 0.025m, rout1 = 0.028m, rin2 = 0.021m, rout2 = 0.024m,
zd zc = 0.003m, zb za = 0.003m, za = 0.001m, r = 0.024m, J1 = J2 = 1T
where dT(1) , T
(2) , dT
(3) and dT
(4) are given in the appendix. Strictly speaking, this expression is three-71
dimensional and we take into account all the contributions between the two tiles. However, it is noted that72
all the contributions have not the same weight and the interaction between the surface densities located73
on the inner and outer faces of each tile permanent magnet is in fact the most important. Furthermore, it74
is noted that this expression could probably still simplified and led to incomplete elliptical integrals of the75
first, second and third kind. In addition, we use the Cauchy principal value for determining the singular76
cases that appear for example when one tile permanent magnet is exactly in front of the other one.77
III. COMPARISON OF THE TORQUE TRANSMITTED BETWEEN TWO TILES RADIALLY MAGNETIZED78
AND TWO TILES UNIFORMLY MAGNETIZED79
Tiles uniformly magnetized are generally less efficient than tiles radially magnetized for magnetic80
couplings. However, this loss of torque depends greatly on the tile angular width. Therefore, it is interesting81
to determine this loss of torque for different tile angular widths. For this purpose, we represent in the same82
figure the torque transmitted between two tiles radially magnetized and the torque transmitted between83
two tiles uniformly magnetized. On Figure 3, the angular width of each tile is the same ( pi12 ). Then,84
we represent in Figs 4, 5 and 6 the torque transmitted between tiles whose angular widths are pi6 ,pi485
and pi3 . Figs 3, 4, 5 and 6 show clearly that the more the tile angular width is important, the less86
the magnetic torque between tiles uniformly magnetized is important. Consequently, we deduct that a87
manufacturer should use tiles uniformly magnetized only if their widths are small. Furthermore, Figs 3,88
70 0.25 0.5 0.75 1 1.25 1.5Angle @radD
0
0.0025
0.005
0.0075
0.01
0.0125
0.015
Torq
ue@N
.mD
Fig. 4. Representation of the torque transmitted between two tiles ; thick line: the magnetization is uniform, dashed lines:the magnetization is radial; 4 3=2 1=
pi
6, rin1 = 0.025m, rout1 = 0.028m, rin2 = 0.021m, rout2 = 0.024m,
zd zc = 0.003m, zb za = 0.003m, za = 0.001m, r = 0.024m, J1 = J2 = 1T
0 0.25 0.5 0.75 1 1.25 1.5Angle @radD
0
0.0025
0.005
0.0075
0.01
0.0125
0.015
Torq
ue@N
.mD
Fig. 5. Representation of the torque transmitted between two tiles ; thick line: the magnetization is uniform, dashed lines:the magnetization is radial; 4 3=2 1=
pi
4, rin1 = 0.025m, rout1 = 0.028m, rin2 = 0.021m, rout2 = 0.024m,
zd zc = 0.003m, zb za = 0.003m, za = 0.001m, r = 0.024m, J1 = J2 = 1T
0 0.25 0.5 0.75 1 1.25 1.5Angle @radD
-0.0025
0
0.0025
0.005
0.0075
0.01
0.0125
0.015
Torq
ue@N
.mD
Fig. 6. Representation of the torque transmitted between two tiles ; thick line: the magnetization is uniform, dashed lines:the magnetization is radial; 4 3=2 1=
pi
3, rin1 = 0.025m, rout1 = 0.028m, rin2 = 0.021m, rout2 = 0.024m,
zd zc = 0.003m, zb za = 0.003m, za = 0.001m, r = 0.024m, J1 = J2 = 1T
8
12
6
4
3
Angular Width @radD102030405060
Loss of Torque @%D
p = 6
p = 8
p = 12
p = 24
Fig. 7. Representation of the mean loss of efficiency of the torque transmitted between two tiles uniformly magnetized versus theangular width of the tiles. This loss of torque is calculated in comparison with the torque transmitted between two tiles radiallymagnetized. The points correspond to the figures 3, 4, 5 and 6. We have : rin1 = 0.025m, rout1 = 0.028m, rin2 = 0.021m,rout2 = 0.024m, zb za = zd zc = 0.003m, J1 = J2 = 1T.
4, 5 and 6 show that the torque transmitted between two tile permanent magnets radially magnetized89
does not depend very much on their angular widths whereas it decreases greatly with the increase in the90
angular width when the tiles are uniformly magnetized.91
We can also represent this decrease by calculating its rate of loss versus the angular width of tiles (Fig92
7).93
Figure 7 is consistent with Figs 3, 4, 5 and 6. Indeed, when the angular width of the tile permanent94
magnets tends to zero, we can use either tiles radially magnetized or tiles uniformly magnetized. This is95
the only case in which the choice has not a great importance. Consequently, as tiles uniformly magnetized96
are cheaper to fabricate, their use is more judicious. However, when the angular width of the tiles used97
becomes greater, the choice is certainly more difficult and other considerations must be taken into account.98
For example, if the first property required in a coupling is really the value of the torque transmitted, the99
magnetization of the tiles should be radial and not uniform.100
IV. APPLICATION : ALTERNATE MAGNET STRUCTURES USING TILES RADIALLY OR UNIFORMLY101
MAGNETIZED102
We can illustrate the expression established in the previous section by studying the torque transmitted103
in an alternate magnet structure with 8 tile permanent magnets on each rotor. For this purpose, we use104
the principle of superposition with (7) for the calculation of the torque transmitted between the leading105
9-3 -2 -1 0 1 2 3Angle @radD
-0.4
-0.2
0
0.2
0.4
Torq
ue@N
.mD
Fig. 8. Representation of the total torque transmitted between an eight tile rotor and an eight tile stator versus the angle .The tiles are either radially magnetized (dashed line) or uniformly magnetized (thick line) rin1 = 0.025m, rout1 = 0.028m,rin2 = 0.021m, rout2 = 0.024m, zb za = zd zc = 0.003m, J1 = J2 = 1T
rotor and the lead rotor. We represent this torque versus the shifting angle in Fig 8.106
Figure 8 is consistent with the previous representations of the torque transmitted between two tiles107
radially or uniformly magnetized. First, we note that the torque transmitted in the alternate magnet structure108
is sixteen times greater than the one transmitted between only two tile permanent magnets. Then, we note109
that the torque transmitted between tiles uniformly magnetized is smaller than the one transmitted between110
tiles radially magnetized, which is still consistent with Fig 4.111
V. CONCLUSION112
This paper has presented a three-dimensional expression of the torque transmitted between two tile113
permanent magnets uniformly magnetized. Then, we have compared the torque transmitted between two114
tiles uniformly magnetized with two tiles radially magnetized. An alternate magnet structure has been115
studied to illustrate the three-dimensional expression of the torque transmitted between tiles radially or116
uniformly magnetized. Some theoretical results have been analytically determined. First, tiles uniformly117
magnetized are less interesting than tiles radially magnetized for realizing couplings. Indeed, the more the118
tile angular widths are important, the less the torque transmitted between two tiles uniformly magnetized119
is important in comparison with the torque transmitted between two tiles radially magnetized. However,120
the cost of the magnets must be taken into account. Tiles uniformly magnetized are easier to fabricate121
10
than tiles radially magnetized (and thus cheaper...).122
APPENDIX123
We give the expressions of the parameters used for calculating the torque transmitted between two124
tile permanent magnets uniformly magnetized. The torque can be expressed in terms of semi-analytical125
expressions using one or two numerical integrations or not.126
T =J1J2
40
( 21
43
dT(1) + T
(2) +
zbza
dT(3) +
21
dT(4)
)(9)
where127
dT(1) = + cos(
1 + 22
) cos( 3 + 4
2) sin(
)A[rin1, rin2]dd
cos( 1 + 22
) cos( 3 + 4
2) sin(
)A[rout1, rin2]dd
cos( 1 + 22
) cos( 3 + 4
2) sin(
)A[rin1, rout2]dd
+cos( 1 + 22
) cos( 3 + 4
2) sin(
)A[rin2, rout1]dd
(10)
128
T(2) = +sin(
2 12
) sin(4 3
2)B[2, 4]
+ sin(2 1
2) sin(
4 32
)B[1, 4]
+ sin(2 1
2) sin(
4 32
)B[2, 3]
+ sin(2 1
2) sin(
4 32
)B[1, 3]
(11)
129
dT(3) = sin(
2 12
) cos(2 1
2 4 3
2)C[2, rin2]dz
+ sin(2 1
2) cos(
2 12
3 + 42
)C[1, rin2]dz
sin(2 12
) cos(2 1
2 3 + 4
2)C[2, rout2]dz
sin(2 12
) cos(2 1
2 3 + 4
2)C[1, rout2]dz
11
(12)
130
dT(4) = +cos(
1 + 22
) sin(4 3
2)D[4, rin1]d
cos( 1 + 22
) sin(4 3
2)D[4, rout1]d
+cos( 1 + 22
) sin(2 1
2)D[3, rin1]d
cos( 1 + 22
) sin(2 1
2)D[3, rout1]d
(13)
The calculation of the torque expression can be done as follows. Eqs (4) and (7) show four fundamental131
expressions, written A[r1, r2], B[i, j ], C[i, ri] and D[j , rj ] that are necessary for calculating the torque132
T. These expressions are expressed as follows:133
A[r1, r2] =
zbza
zdzc
r21r22(r21 + r
22 2r1r2 cos( ) + (z z)2)
3
2
dzdz (14)
134
B[i, j ] =
rout2rin2
zdzc
rout1rin1
zbza
rr sin(j i)(r2 + r
2 2rr cos(j i) + (z z)2) 32drdzdrdz (15)
135
C[i, ri] =
rout1rin1
zbza
zdzc
r2i r sin( i)(r2 + r2i 2rri cos( i) + (z z)2)
3
2
drdzdz (16)
136
D[j , rj ] =
rout2rin2
zdzc
zbza
r2j r sin(j i)(r
2+ r2j 2rj r cos(j ) + (z z)2)
3
2
dzdrdz (17)
The torque transmitted between two tile permanent magnets could be calculated directly by numerical137
means with (8). However, the computational cost would be too long. Therefore, we give here four reduced138
semi-analytical expressions of A[r1, r2], B[i, j ], C[i, ri] and D[j , rj ]. We obtain:139
A[r1, r2] = A(1)[za, zc]A(1)[zb, zc]A(1)[za, zd] + A(1)[zb, zd] (18)
with140
A(1)[zi, zj ] =r21r
22
r21 + r
22 + (zi zj)2 2r1r2 cos( )
r21 + r22 2r1r2 cos( )
(19)
12
141
B[i, j ] =
rout2rin2
(B(1)[za zc, rout1] + B(1)[zb zc, rout1]
)dr
+
rout2rin2
(+B(1)[za zd, rout1]B(1)[zb zd, rout1]
)dr
+
rout2rin2
(B(1)[za zc, rin1]B(1)[zb zc, rin1]
)dr
+
rout2rin2
(B(1)[za zd, rin1] + B(1)[zb zd, rin1]
)dr
(20)
with142
B[y, ri] = r
r2 2rrx + y + r2x log
[r rx +
r2 2rrx + y
]
+ri(x +1 + x2
r2 y)
21 + x2 log[A]
+ri(x +
1 + x2)
r2 y
21 + x2 log[B]
(21)
143
A =2i(rr(1 + x2) + r2(x x3 + x21 + x2))
(x +1 + x2)(r + r(x +1 + x2))(r2 y) 32
+2i1 + x2(y + i
r2 y
r2 2rrx + y)
(x +1 + x2)(r + r(x +1 + x2))(r2 y) 32(22)
144
B =2irr(1 + x2) 2ir2x(1 + x2 + x1 + x2)
(x +1 + x2)(r + r)(x +1 + x2)(r2 y) 32
+21 + x2(iy +
r2 y
r2 2rrx + y)
(x +1 + x2)(r + r(x +1 + x2))(r2 y) 32
(23)
145
x = cos(i j) (24)
13
C[i, ri] =
rout1rin1
C(1)[i, ri]dr (25)
C(1)[i, ri] = ri
r2i + r
2 + (zc z)2 2rir cos(3 i)
rir2i + r
2 + (zd z)2 2rir cos(3 i)
rir2i + r
2 + (zc z)2 2rir cos(4 i)
+ri
r2i + r
2 + (zd z)2 2rir cos(4 i)
+ri(zc z) log[zc + z +
r2i + r
2 + (zc z)2 2rir cos(3 i)]
ri(zd z) log[zd + z +
r2i + r
2 + (zd z)2 2rir cos(4 i)]
+ri(zd z) log[zd + z +
r2i + r
2 + (zd z)2 2rir cos(3 i)]
ri(zc z) log[zc + z +
r2i + r
2 + (zc z)2 2rir cos(3 i)]
(26)
D[j , rj ] =
rout2rin2
(D(2)(zb, zc)D(2)(za, zc) + D(2)(za, zd)D(2)(zb, zd)
)dr (27)
with146
D(2)(x, y) =sin(i j)r2j r
r2j +
r2
+ (x y)2 2rj r cos(j i)r2j +
r2 2rj r cos(j i)
(28)
REFERENCES147
[1] J. P. Yonnet, Permanent magnet bearings and couplings, IEEE Trans. Magn., vol. 17, pp. 11691173, Jan 1981.148
[2] J. P. Yonnet, Rare-earth Iron Permanent Magnets, ch. Magnetomechanical devices. Oxford science publications, 1996.149
[3] J.P.Yonnet, S. Hemmerlin, E. Rulliere, and G. Lemarquand, Analytical calculation of permanent magnet couplings, IEEE150
Trans. Magn., vol. 29, pp. 29322934, Nov. 1993.151
[4] P. Elies and G. Lemarquand, Analytical study of radial stability of permanent magnet synchronous couplings, IEEE Trans.152
Magn., vol. 35, pp. 21332136, July 1999.153
[5] J. P. Yonnet, Passive magnetic bearings with permanent magnets, IEEE Trans. Magn., vol. 14, pp. 803805, Sept. 1978.154
14
[6] J. Delamare, E. Rulliere, and J. Yonnet, Classification and synthesis of permanent magnet bearing configurations, IEEE.155
Trans. Magn., vol. 31, pp. 41904192, Nov. 1995.156
[7] G. Lemarquand, Ironless loudspeakers, IEEE Trans. Magn., vol. 43, pp. 33713374, August 2007.157
[8] P. Samanta and H. Hirani, Magnetic bearing configurations: Theoretical and experimental studies, IEEE Trans. Magn., vol. 44,158
pp. 292300, Feb. 2008.159
[9] S. Mukhopadhyay, J. Donaldson, G. Sengupta, S. Yamada, C. Chakraborty, and D. Kacprzak, Fabrication of a repulsive-type160
magnetic bearing using a novel arrangement of permanent magnets for vertical-rotor suspension, IEEE Trans. Magn., vol. 39,161
pp. 32203222, Sept. 2003.162
[10] R. Moser, J. Sandtner, and H. Bleuler, Optimization of repulsive passive magnetic bearings, IEEE. Trans. Magn., vol. 42,163
pp. 20382042, Aug. 2006.164
[11] F. Bancel and G. Lemarquand, Three-dimensional analytical optimization of permanent magnets alternated structure, IEEE165
Trans. Magn., vol. 34, pp. 242247, Jan. 1998.166
[12] V. Lemarquand, J. F. Charpentier, and G. Lemarquand, Nonsinusoidal torque of permanent-magnet couplings, IEEE Trans.167
Magn., vol. 35, no. 5, pp. 42004205, 1999.168
[13] J. Charpentier and G. Lemarquand, Optimization of unconventional p.m. couplings, IEEE Trans. Magn., vol. 38, pp. 1093169
1096, March 2002.170
[14] G. Akoun and J. P. Yonnet, 3d analytical calculation of the forces exerted between two cuboidal magnets, IEEE Trans.171
Magn., vol. 20, no. 5, pp. 19621964, 1984.172
[15] J. Hilton and S. McMurry, Halbach cylinders with improved field homogeneity and tailored gradient fields, IEEE Trans.173
Magn., vol. 43, pp. 18981902, may 2007.174
[16] K. Halbach, Design of permanent multiple magnets with oriented rec material, Nucl. Inst. Meth., vol. 169, pp. 110, 1980.175
[17] M. Abele and H. A. Leupold, A general method for flux confinement in permanent magnet structure, J. Appl. Phys., vol. 64,176
no. 10, pp. 59885990, 1988.177
[18] O. M. Kwon, C. Surussavadee, M. Chari, S. Salon, and K. S.vasubramaniam, Analysis of the far field of permanent magnet178
motors and effects of geometric asymmetries and unbalance in magnet design, IEEE Trans. Magn., vol. 40, pp. 435442,179
march 2004.180
[19] M. Abele, J. Jensen, and H. Rusinek, Generation of uniform high fields with magnetized wedges, IEEE Trans. Magn., vol. 33,181
no. 5, pp. 38743876, 1997.182
[20] R. Ravaud, G. Lemarquand, V. Lemarquand, and C. Depollier, Analytical calculation of the magnetic field created by183
permanent-magnet rings, IEEE Trans. Magn., vol. 44, pp. 19821989, Aug 2008.184
[21] J. Conway, Inductance calculations for noncoaxial coils using bessel functions, IEEE Trans. Magn., vol. 43, pp. 10231034,185
march 2007.186
[22] J. Conway, Noncoaxial inductance calculations without the vector potential for axisymmetric coils and planar coils, IEEE187
Trans. Magn., vol. 44, pp. 453462, oct. 2008.188
[23] K. Kim, E. Levi, Z. Zabar, and L. Birenbaum, Mutual inductance of noncoaxial circular coils with constant current density,189
IEEE Trans. Magn., vol. 33, pp. 43034309, sept. 1997.190
15
[24] S. Babic and C. Akyel, Magnetic force calculation between thin coaxial circular coils in air, IEEE Trans. Magn., vol. 44,191
pp. 445452, April 2008.192
[25] S. Babic, C. Akyel, S. Salon, and S. Kincic, New expressions for calculating the magnetic field created by radial current in193
massive disks, IEEE Trans. Magn., vol. 38, pp. 497500, march 2002.194
[26] S. Babic and M. Gavrilovic, New expression for calculating magnetic fields due to current-carrying solid conductors, IEEE195
Trans. Magn., vol. 33, pp. 41344136, may 1997.196
[27] S. Babic and C. Akyel, An improvement in the calculation of the magnetic field for an arbitrary geometry coil with rectangular197
cross section, International Journal of Numerical Modelling: Electronic Networks, Devices and Fields, vol. 18, pp. 493504,198
November 2005.199
[28] S. Babic and C. Akyel, Calculating mutual inductance between circular coils with inclined axes in air, IEEE Trans. Magn.,200
vol. 44, pp. 17431750, July 2008.201
[29] J. P. Selvaggi, S. Salon, O. -M. Kwon, and M. V. K. Chari, Calculating the external magnetic field from permanent magnets202
in permanent-magnet motors - an alternative method, IEEE Trans. Magn., vol. 40, pp. 32783285, Sept. 2004.203
[30] J. P. Selvaggi, S. Salon, O. -M. Kwon, M. V. K. Chari, and M. DeBortoli, Computation of the external magnetic field, near-field204
or far-field from a circular cylindrical magnetic source using toroidal functions, IEEE Trans. Magn., vol. 43, pp. 11531156,205
Avril 2007.206
[31] E. P. Furlani, Formulas for the force and torque of axial couplings, IEEE Trans. Magn., vol. 29, pp. 22952301, Sept. 1993.207
[32] E. Furlani, S. Reznik, and A. Kroll, A three-dimensonal field solution for radially polarized cylinders, IEEE Trans. Magn.,208
vol. 31, pp. 844851, Jan. 1995.209
[33] E. Furlani, Field analysis and optimization of ndfeb axial field permanent magnet motors, IEEE Trans. Magn., vol. 33,210
pp. 38833885, Sept. 1997.211
[34] B. Azzerboni, E. Cardelli, and A. Tellini, Computation of the magnetic field in massive conductor systems, IEEE Trans.212
Magn., vol. 25, pp. 44624473, november 1989.213
[35] B. Azzerboni and E. Cardelli, Magnetic field evaluation for disk conductors, IEEE Trans. Magn., vol. 29, pp. 24192421,214
november 1993.215
[36] B. Azzerboni and G. Saraceno, Three-dimensional calculation of the magnetic field created by current-carrying massive disks,216
IEEE Trans. Magn., vol. 34, pp. 26012604, september 1998.217
[37] E. Perigo, R. Faria, and C. Motta, General expressions for the magnetic flux density produced by axially magnetized toroidal218
permanent magnets, IEEE Trans. Magn., vol. 43, pp. 38263832, oct. 2008.219
[38] H. Rakotoarison, J. Yonnet, and B. Delinchant, Using coulombian approach for modeling scalar potential and magnetic field220
of a permanent magnet with radial polarization, IEEE Trans. Magn., vol. 43, no. 4, pp. 12611264, 2007.221
[39] Y. Zhilichev, Calculation of magnetic field of tubular permanent magnet assemblies in cylindrical bipolar coordinates, IEEE222
Trans. Magn., vol. 43, pp. 31893195, July 2007.223
[40] E. Durand, Electrostatique, Masson Editeur, Paris, France, vol. 1, pp. 248251, 1964.224
[41] T. Ohji, C. Mukhopadhyay, M. Iwahara, and S. Yamada, Performance of repulsive type magnetic bearing system under225
nonuniform magnetization of permanent magnet, IEEE Trans. Magn., vol. 36, pp. 36963698, Sept. 2000.226
16
[42] T. Ohji, S. Ichiyama, K. Amei, M. Sakui, and S. Yamada, Conveyance test by oscillation and rotation to a permanent magnet227
repulsive-type conveyoer, IEEE Trans. Magn., vol. 40, pp. 30573059, April 2004.228
[43] M. Berkouk, V. Lemarquand, and G. Lemarquand, Analytical calculation of ironless loudspeaker motors, IEEE Trans. Magn.,229
vol. 37, pp. 10111014, March 2001.230
[44] L. Yong, Z. Jibin, and L. Yongping, Optimum design of magnet shape in permanent-magnet synchronous motors, IEEE231
Trans. Magn., vol. 39, pp. 35234205, nov. 2003.232
[45] O. Cuguat, J. Delamare, and G. Reyne, Magnetic micro-actuators and systems (magmas), IEEE Trans. Magn., vol. 39,233
pp. 36073612, nov 2003.234
[46] J. Wang, G. W. Jewell, and D. Howe, Design optimisation and comparison of permanent magnet machines topologies,235
vol. 148, pp. 456464, IEE.Proc.Elect.Power Appl., Sept 2001.236
[47] A. A. Hussien, S. Yamada, M. Iwahara, T. Okada, and T. Ohji, Application of the repulsive-type magnetic bearing for237
manufacturing micromass measurement balance equipment, IEEE Trans. Magn., vol. 41, pp. 38023804, Oct. 2005.238
[48] M. Aydin, Z. Zhu, T. Lipo, and D. Howe, Minimization of cogging torque in axial-flux permanent-magnet machines: design239
concepts, IEEE Trans. Magn., vol. 43, pp. 36143622, sept. 2007.240