Вы находитесь на странице: 1из 6



P Thelin and H-P Nee

KTH, Royal Institute of Technology, Sweden


It has been shown earlier that the airgap flux density of flux concentration), see Figure 1. A disadvantage with
PM motors with buried magnets can be calculated with buried magnets is that some of the magnet flux is lost
a satisfactory result, if saturated iron bridges, axial due to leakage through the thin iron bridges. The iron
leakage flux (from 2D-FEM) and teeth saturation are bridges are required to keep the iron lamination togeth-
taken into account. In this paper, instead of using 2D- er mechanically.
FEM, an analytical expression for the axial reluctance
is derived. By including the saturation of rotor and sta- The intention of this paper is to improve the analytical
tor yokes, the model is even further improved. The re- calculation of the airgap flux density of such motors.
sult is a complete analytical model, which takes into

account all,phenomena mentioned above, and improves
the agreement with experimental values.
The axial leakage flux is often neglected in radial flux
INTRODUCTION machines. This is due to the fact that the axial leakage
flux is normally very small. The analytical calculations
are therefore based on a cross-section of the motor. The
same assumption is normally made in FEM calcula-
tions, since 2D-FEM is more common, faster and easier
to use than 3D-FEM. For PM machines with relatively
short rotor lengths, the axial leakage flux has a larger
influence on the radial torque-producing flux.

Axial Reluctance From 2D-FEM

In the earlier paper, Thelin and Nee [7], the axial reluc-
tance "seen" by the magnet, i.e. 9 i a , when the radial re-
luctance is assumed to be infinite, was obtained from
2D-FEM calculations. This was a drawback, since the
remaining expression could be evaluated without the
aid of FEM calculations. To overcome this drawback,
Figure 1: Example of an 8-pole PM-motor design with a five different -but typical- magnet configurations will
thin squirrel cage and buried magnets in V- be studied and some analytical expressions will be pre-
shape. sented in this paper.

Permanent magnet synchronous motors (PMSM) with A typical magnetic field line plot of the axial leakage
buried magnets have been considered in a wide range of flux for a motor with buried magnets is shown in Figure
drives including both variable-speed, Morimoto et al 2. Figure 2 shows an axial cross-section view of the up-
[I]. and mains-connected (i.e. line start), Herslof [2], per half of the stator and the rotor. The two magnets,
Chalmers [3], drives. A buried magnet design can have which originally were placed in V-shape, have been re-
some advantages compared to surface mounted, Slemon placed by one single magnet at the average height of the
and Xian [4], and inset, Nipp [ 5 ] , magnet designs. V [7]. The influence of the saturation (from the radial
flux) on the axial leakage flux has been neglected. The
Buried magnet designs are likely to be less complicated iron material of the rotor and the stator is set to have a
to manufacture than surface mounted designs. For ex- very high permeability. The shaft is non-magnetic. The
ample no bandaging is required. The magnets are just axial reluctance of one pole for the two sides of the rotor
inserted into punched slots in the laminated rotor iron. is then found to be [7]
Other advantages are that the magnets are protected
from physical damage and demagnetizing currents. 4224
R, = -431,5. lo3 = ~ J ~ M H -(1)
Burying the magnets admits different magnet configu- (0,M+ 0,04)0,01257
rations, Hippner and Harley [ 6 ] .The magnets can e.g.
be placed close to the circumference, or in V-shape (i.e. by using a FEM calculation. (H"=A/Wb=ANs)

Power Electronics and Variable Speed Drives, 18-19 September 2000, Conference Publication NO. 475 @ IEE 2000

Stator \ 1 ,Rotor body
qa= 0,005507 Vs/m



qa= 0,01257 Vs/m \ -//

Figure 2: Typical field lines of the axial leakage flux.

As can be seen from Figure 2, the main part of the axial Figure 4: Axial field lines from the left half of the rotor
leakage flux is concentrated to the vicinity of the mag- when the magnet is displaced vertically.
net, though some penetrates the stator iron. The part of
the flux that penetrates the stator iron will partly link The axial reluctance is again calculated with FEM:
with the stator winding, thereby slightly contributing to
the torque production. Due to these two reasons it is
%, = ... = 4,36MH-' (3)
enough to deal with the leakage flux close to the mag-
net. Therefore the stator and the rotor shaft are omitted
in the five following cases. This measure increases the axial reluctance only slight-
Case 1. Centred magnet. Figure 3 shows a simplified
geometry of the machine in Figure 2. Due to symmetry, Case 3. Magnet close to edge. Figure 5 shows the field
lines when the magnet is moved to a new position, even

only the left half of the rotor in Figure 2 is depicted in
Figure 3.
0,005914 Vs/m
closer to the edge.
\ qa= 0,00431 Vs/m

Figure 3: Axial field lines from the left half of the rotor. Figure 5: Axial field lines from the left half of the rotor
when the magnet is close to the edge.
Again the total axial reluctance per pole for both sides
of the rotor is calculated by the use of FEM,and the re- The axial reluctance is again calculated with E M :
sult is
31, = ... = 5,69MH-' (4)
So = ... = 4,03MH-' (2)
The axial reluctance has now increased further.
As can be seen from Equation (2) the axial reluctance is
now slightly higher but still in the same range, com-
pared to the value of Equation (1).
Case 4. Thicker magnet. It is also interesting to see
Case 2. Displaced magnet. It is also interesting to see how a thicker magnet effects the axial reluctance. Fig-
how a vertical displacement of the magnet effects the ure 6 shows the field lines when the thickness of the
axial reluctance, see Figure 4. magnet slot is 10 mm instead of 5 mm.

which has a simple mathematical law.
n \ q a = 0,009338 Vslm [S) suggests the heuristic method of estimating the per-
/ /d\\ \
meances of probable flux paths. With this method ap-
plied on the problems above the field lines are divided
into two regions, one at a distance from the magnet
22 where the field lines follow a circular path and one
10 close to the magnet where the field lines follow a path
with a mean length. This results in the following ap-
22 proximate expression for the total axial reluctance per
pole of the two rotor sides:

Figure 6: Axial field lines from the left half of the rotor
with a thicker magnet slot.

The axial reluctance is again calculated with FEM: Parameter definitions, see Figure 8. [SI mentions that
functions of complex variables (i.e. the method of con-
3, = ... = 4,91 MH- (5) formal mapping, see e.g. Fisher [lo]) may also be used.
The result of this is that the factor 0,26 in Equation (7)
reduces to 0,24 and 0,22 for a thick and a thin sample,
This measure increased the axial reluctance.
respectively [8]. From this one can conclude that the ap-
Case 5. Thinner magnet. Figure 7 shows the field lines proximate method is accurate enough in this case.
when the thickness of the magnet slot is reduced to 2,5
Another approach, which shows quite good agreement
with the FEM calculated values, is to assume that the
,,,,----A\, q: = 0,003494 Vs/m flux goes the shortest way from the north pole to the
south pole, see Figure 8.

North pole

I South pole I i;
Figure 8: Derivation of axial reluctance for the left rotor
Figure 7: Axial field lines from the left half of the rotor side with a vertically displaced magnet slot.
with a thinner magnet slot.
This is -of course- a less physical model, but anyhow it
The axial reluctance is again calculated with FEM: seems to be a good estimate of the total axial reluctance.
This reluctance can be derived in the following manner:
R, = ... = 3,40MH- (6) The differential permeance of the air path from the cen-
tre of the magnet slot up to the height y in Figure 8 is
This measure reduces the axial reluctance. given by

Analytical Expressions For the Axial Reluctance

The analytical calculation of the axial reluctance is not

easy, or as Roters [8] expresses it: where w,, is the magnet width and lslo, is the thickness
of the magnet slot. The total permeance of the upper air
The precise mathematical calculation of the permean- path is then given by integrating over the height h , :
ce of flux paths through air, except in a few special cas-
es, is a practical impossibility. This is because the flux
does not usually confine itself to any particular path

The permeance of the lower air path is found in the COMPENSATION FOR IRON SATURATION
same manner: ')
The increased reluctance of the rotor and stator teeth,
due to iron saturation, can theoretically be represented
by an extra airgap in the magnetic circuit of the machine
[7]. The length of the extra airgap is found as [7]
The total axial reluctance per pole of the two rotor sides
is then given as

where L is the axial length of the stator or the rotor, and

wR is the true circumferential pole-width on the rotor
surface, found in Figure 9. In [7] Rfo, was the resulting
reluctance of the teeth in the stator 9ifSand in the rotor
R,, . By adding reluctances representing the most satu-
rated part of the stator yoke Rys and of the rotor yoke
By calculating the axial reluctance with Equation (7)
9tyr the model is further improved, that is
and (1 I), respectively, a comparison with the FEM cal-
culated values of Equations (2)-(6) can be made. In this
comparison, the FEhI calculated values are regarded as
the correct values. The results of the calculations are
summarized in Table 1. Reluctances of Teeth and Yokes
Table 1: FEM and analytically calculated values of the The resulting reluctances of the stator and rotor teeth
axial reluctance for Case 1-5. under one pole are given in [7]. In the same manner, the
reluctances representing the most saturated part of the
stator and rotor yokes under one pole are found as

1 1 4,03 4,86 I 4,16

2 1 4,36 6,19 I 5,87
I 3 I 5,69 I 20,70 I 8,84 I
4 4,9 1 6,22 587
5 3,40 4,O1 323
where H is the magnetic field intensity, B is the flux
density, L is the rotor or stator length, k f is the stacking
factor for the iron lamination, I,, is equal to the slot
width at the bottom of a stator slot, wys is the thickness
of the stator yoke (i.e. back), w y r is the width of the
most narrow part of the rotor yoke (i.e. the part that will
carry most flux per width), and 1,, is the length -inside
the pole- of the same narrow part. Further:

1 if most narrow part carries total pole-flux

2 if most nanow part carries half pole-flux
The tangential leakage flux, i.e. the inter-pole leakage
through the airgap or through airgap-statortoothtip-air-
gap, has not been taken into consideration in this paper. Magnetic Flux Density in Teeth and Yokes
There are several reasons for this:
Equations for the flux densities of the active stator and
The tangential leakage flux rotor teeth are given in [7]. The flux densities of stator
and rotor yokes are found in a similar way as
influences the shape of the flux in the airgap only
where it is less weighted in the Fourier analysis
depends on the rotor position
is contributing to the airgap flux density even in static
FEM calculations

where w,,, is the yoke width of stator and w y , is ex- where k, is the Carter factor (see e.g. [lo]), g is the air-
plained in the subsection above. gap length and gc is the fictitious extra airgap which
compensates for iron saturation given by Eq. (12).
The flux density values from the equations for the teeth
and the yokes give a good indication of how hard the Inserting Equation (I 1) into the old equation for w/i
iron is used. Values above e.g. 1,7 T -of course depend- in [7], and rearranging, gives the new equation
ing on the iron quality used- indicate that the MMF-
drops of the teeth and yokes must be taken into consid- !f -+-+...+%+-+...+x
= WFeI wFc2 w, k, (23) W62
eration, at least if the teeth and the narrow parts of the Fe1 Fe2 61 6,
yokes are relatively long.
where L is the axial length of the rotor and
Magnetic Field Intensity in Teeth and Yokes

To find the magnetic field intensity H for a given flux

density B , a simplified Langevin-expression -together
with Newtons method- was employed [ 7 ] .


where h , and h, are the two heights defined in Figure
The airgap flux density can now be calculated analyti- 8. The other geometrical parameters are defined in Fig-
cally with the following equation (compare to [7]): ure 9. The peak value of the fundamental airgap flux
density can, using Fourier analysis, be found as:

where a is the electrical angle of half the true pole

width on the rotor surface, see Figure 9.
where B, is the remanent flux density of the magnet, t
BSar is the assumed saturated flux density level of the
iron, k f is the stacking factor for the iron lamination,
I, is the thickness of the magnet, and p, is the relative
permeability of the magnet. wFe is the sum of the (dif-
ferent) iron bridge widths under one pole:

WFe = WFel + W F e , + ... (18)

w , is the sum of the (different) magnet widths under

one pole:

w , = w m 1 + w m 2 +... (19)
and the true circumferential pole-width on the rotor sur-
face is given by

n 2- Figure 9: Definition of parameters for one rotor pole.

wg = r 2 a . -d.
180 p
where r is the rotor radius, a is the electrical angle of EXPERIMENTS
half the true pole width and p is the pole-number.
Motors A-E
ki = To check the expression of Equation (17), five PM mo-
tor designs have been examined. The 2D-FEM calcula-
where Ii is the thickness of the internal airgap surround- tions were performed with the software ACE (from
ing the magnet and I, is the thickness of the magnet. ABB Corporate Research). All five PM motors have
been manufactured so experimental, back-calculated
values of the airgap flux densities are also available.

Motor A. Motor; A, which has ,8 poles and is inverter- of the iron are also shown in Table 2.
fed, has the geometry shown in Figure 1. The geometri-
cal parameters of Motor A were identified by using Fig- Motors B-E. Motors B-E all have geometries which are
ure 9. The axial leakage flux of Motor A is shown in somewhat similar to the one in Figure 1, but the rotor
Figure 2. The results of the calculations and the meas- cage has deeper bars and a higher number of bars since
urements are shown in Table 2. these motors are line-start motors. Motors B-E have 4,
6, 16 and 4 poles, respectively. The results of perform-
Table 2: Flux densities of the five examined motors A-E. ing the sa-me exercises for Motors B-E as for Motor A,
are presented in Table 2. Here the product of the Carter
Motor A B C D E factor for stator and for rotor was used in the analytical
0,88 0,64 0,82 I , 1 1 0,79 T calculations.
B(I ) & 2D-FEM

It has been shown earlier that the airgap flux density of

PM synchronous motors with buried magnets can be
calculated with satisfactory result, if saturated iron
bridges, teeth saturation and axial leakage flux are tak-
en into consideration. In this paper an analytical ex-
pression -instead of a 2D-FEM calculation- for the axial
reluctance is derived. Also the magnetic model is fur-
ther improved by including the effect of iron saturation
of rotor and stator yokes. The result is an analytical ex-
pression for the airgap flux density, that includes satu-
ration and axial leakage phenomena.


ABB Corporate Research, ABB Motors, Atlas Copco

Controls, ITT Flygt and the Swedish National Energy
The different subscripts of the peak value of the funda- Administration are gratefully acknowledged for the fi-
mental airgap flux density from the magnets of Table 2 nancial support of the work.
are explained below:
B(I)g.2D-EM : 2D-FEM calculated value. [I] Morimoto S., Sanada M., Takeda Y. and Taiguchi K.. 1994. Optimum
Machine Parameters and Design of Inverter-Driven Synchronous Mo-
B( )R : Analytical value from Eq. (17). Without axial tors for Wide Constant Power Operation, Conferencecord of thc
E E E IAS Annual Mte, Denver, Colorado, pp. 177-182.
leakage (k, =O) and without iron saturation (g, =O).
121 HerslOf U,, 1996. Design. Analysis and Verification of a Line Start
Permanent Magnet Synchronous Motor, Licentiates thesis, Royal In-
B( 1 )n.
age flux into account with 2D-FEM (%, << -
: Value from Eq.(17), taking axial leak-
). With- 131
stitute of Technology, Stockholm,ISSN-I 100-1631.
Chalmers B.J.. 1996, Exploitation of Old Phenomena and New Mate-
rials in Electrical Machines Vigo. Spain,
out iron saturation (g, a). 10-12 September, pp. 1-6.

b(l)g,,x,:Analytical value from Eq. (17), taking axial 141 Slemon G.R. and Xian L.. 1992, Modelling and design
permanent magnet motors.. WC
. optimization
Mach= and Power .Sy&DS,

leakage flux into account with Eq. (24) (k, 9).

With- U,pp. 71-92.
out iron saturation (gea). 151 Nipp E., 1995, Extending the Field-Weakening Range of Surface-
Mounted Permanent-Magnet Motors, &&@I Stock-
holm, 18-22 June. pp. 408-413.
romp :Analytical value from Eq. (17), compen-
161 Hippner M. and Harley R.G., 1992, Looking for an optimal rotor for
sated for tooth and yoke saturations (g, > 0 ). Without high speed permanent magnet synchronous machine, G&WKL&L
p .
Houston. 4-9 Oct., ynL ,
axial leakage (k, a). L pp. 265-270.
camp :Analytical value from Eq. (17), taking 171 Thelin P.and Nee H.-P., 1999, Calculation of the Airgap Flux Density
B, I ) ~ , of PM Synchronous Motors with Buried Magnets including Axial
axial leakage flux into account with Eq.(24) (k, 9). Leakage and Teeth Saturation, EMD99. Canterbury,
United Kingdom. September, pp. 339-345.
Compensated for tooth and yoke saturations g, > 0 . [8] Roters H.C., 1941, Electromagnetic Devices, Chapter V. John Wiley
& Sons Inc.
expe : Value calculated from measurements.
191 Fisher S.D. 1990. Complex Variables, 2nd Ed., Wadswonh &
BrookslCole Mathematics Series. ISBN 0-534-13260-X.
The highest flux density level in the rotor and stator 1101 Thelin P; and Nee H.-P., 1998. Analytical Calculation of the Airgap
teeth, and in the rotor and stator yokes (from FEM cal- Flux Density of PM-Motors with Buried Magnets,
culations), and the assumed saturated flux density level 1CEM98.Istanbul. Turkey, September, y~L-2pp. 1166-1 17 1.