Identification of principal factors causing unbalanced

magnetic pull in cage induction motors

K. j . Binns, B.Sc, A.F.I.M.A., C.Eng., M.I.E.E., and M. Dye, B.Sc, Ph.D.

Indexing terms: Cage induction motors, Magnetic circuits, Rotors, Machine theory

ABSTRACT
The precise effect of the static eccentricity of the rotor of an induction machine on the magnetic field in the air-
gap is demonstrated and explained. The relation between the unbalanced pull and the eccentricity of a cage rotor
is determined through measurements of the electromagnetic field at the rotor surface, and the significance of the
tangential-flux component in the machine airgap is assessed.

LIST OF PRINCIPAL SYMBOLS may be considered as a 2-pole modulation of the airgap field.
This modulation gives rise to induced currents in the cage
B airgap flux density r.m.s. rotor and in parallel paths in the stator winding, when such
normal flux density at point A on rotor surface are present, and these currents reduce the amplitude of the
: flux-density amplitudes associated with ith har- 2-pole modulation and consequently decrease the u.m.p.
monic
normal component of flux density There is an abundance of published work on the subject of
n
slip-frequency component of flux density (peak) u.m.p., resulting in a large number of differing formulas de-
Bo pending on a variety of assumptions. It is difficult to draw
Bot slip-frequency component of tangential flux density
(peak) any clear conclusions, since many proposed equations have
tangential component of flux density not been tested; and the test results which have been publish-
unbalance component of flux density (peak) ed are not sufficiently comprehensive, in that they apply to
only one or two designs and mainly relate to wound-rotor
B
ut = unbalance component of tangential flux density machines. In a review of the literature on u.m.p. that was
(peak)
published in 1918, Gray and Pertsch 1 found that all the pub-
J
l = fundamental flux density (peak) lished expressions for the magnetic pull were of the form
eccentricity
normal surface stress KB2e/g, for small values of eccentricity. K was a factor
mean airgap which might depend on a number of variables, such as pole
g
number and the ratio of the airgap reluctance to that of the
H n = normal component of magnetic-field strength whole magnetic circuit. Most of the machines for which
tangential component of magnetic-field strength
i = harmonic number these formulas were to apply had salient poles, but Rosen-
berg 2 developed a rule which also applied to nonsalient-pole
K
m = rate of change of pull with displacement machines. Attempts have been made to extend the range of
stiffness of rotor shaft
L = rotor length this simple expression to cover larger values of eccentri-
P unbalanced pull city. Gray and Pertsch found that the most popular modifi-
number of pairs of poles cation was to make the pull proportional to the expression
P
unbalance pull due to tangential flux density e/g
r = radius of cylindrical surface
S surface area of surface of integration U-(e/g)2|3/2
a, \ = phase angles
phase angle of ith harmonic while von Kaehne,3 in his survey of work on u.m.p. published
in 1963, considered the most generally accepted expression to
Mo = primary magnetic constant be of the form
relative permeability
o = slip, p.u. e/g
U) = synchronous angular velocity

1 INTRODUCTION Both these expressions rise more quickly with increasing

eccentricity than the linear relationship. The increases are
The calculation of unbalanced magnetic pull (u.m.p.) in in- fairly small over the range of practical interest, but the two
duction machines owing to the rotor being mounted eccen- expressions give very different values for the increase. It
trically with respect to the stator bore, and the examination should be noted that the analysis on which they are based
of practical means of reducing this pull, are subjects which takes no account of iron saturation. Friese and Jordan 4 de-
have received considerable attention over the last 50 years. rive a multiplying factor equal to 1. 31 at 45% eccentricity,
The unbalanced force necessitates an increase in shaft dia- but Bradford reports that measured values were directly
meter and bearing size and has the effect of reducing criti- proportional to eccentricity right up to a value of 45%.
cal speed, and the practical importance of its accurate pre-
diction is beyond question. Yet, fcr cage machines, the pub- As the voltage applied to the stator winding is increased,
lished formulas are of doubtful value, and comprehensive saturation tends to limit the flux density in the iron in the
measurements of the variations in the harmonic fields in the region where the airgap is smallest. The result is that the
airgap of a cage-rotor machine with rotor eccentricity have u.m.p. does not increase indefinitely with voltage. The effect
not been published. Induction-motor rotors can be turned on of a cage on a rotor moving at an appreciable speed is to re-
a lathe to make the rotor surface true with respect to the duce the difference in flux density around the rotor, so that
bearings, thereby minimising the incidence of rotating eccen- the effects of saturation are less marked. Little attention
tricity, but static eccentricity, which occurs when the rotor has been paid to the effect of load, but leakage fluxes, which
and stator axes are not coincident, is not easily cured and is increase with load, play a significant part in tooth-tip satura-
present to a significant degree in most production machines. tion. Most of the methods of accounting for saturation ignore
This form of eccentricity, which causes a unidirectional pull, this, and so their results apply strictly only to the light-load
tending to decrease the airgap at its smallest point, is the condition. The published methods for evaluating the effect of
subject of this paper. The effect of this form of eccentricity saturation are based on an effective increase in airgap, which
is used in determining an equivalent eccentricity, which, in
turn, replaces the actual eccentricity in the simple formulas
Paper 6889 P, first received 18th September and in revised for the u.m.p. Simple methods utilising the magnetisation
form 20th December 1972 curve have been proposed by Rosenberg2 and Robinson.5
Since the machine is not uniformly saturated, if eccentricity
Mr. Binns is, and Dr. Dye was formerly, ivilh the Department is present, it has been suggested that an equivalent airgap at
of Electrical Engineering, The University, Southampton a series of points around the rotor should be calculated and
S09 5NH, England. Dr.Dye is now ivilh IBM UK Labora- a resulting equivalent eccentricity determined. In his survey
tories Ltd., Horsley Park, Winchester, Hants., England of the literature, von Kaehne suggested that the effect of satu-
PROC. IEE, Vol. 120, No. 3, MARCH 1973 349
P 11
ration could be accounted for in terms of a change in relative the factor V( 1 — K m /K s ), as has been shown by Rosenberg2
permeability in the force expression and Crawford.14
When a machine is operating at a speed near to a critical
B2 speed, vibrations will be present, and these give rise to a
(1 - I/Mr) combination of both static and rotating eccentricity.
Frohne 10 is of the opinion that only disturbances in the air-
as used by Miles-Walker.6 In view of the slotted nature of gap field caused by rotating eccentricity can affect the criti-
the boundaries, y.T would not be an actual permeability but an cal speed; hence the equalising action of a cage rotor does
'equivalent' value, which would be very difficult to determine. not affect the critical speed. He also suggests that only the
short-circuited damper-winding set in the stator slots can
The effects of slotting on u.m.p. have received rather less reliably reduce vibration. While the cage rotor cannot res-
attention than those of saturation. The general approach has pond to and reduce the effects of a rotating eccentricity, it
been to calculate the pull for a smooth stator and rotor and will respond to changes in the airgap field caused by vibra-
then to multiply this by a function of the gap-extension factor. tion. A thorough investigation of the dynamic behaviour of a
The effects of slotting are twofold. The total flux is reduced rotor operating near to a critical speed would seem to be
and the flux is concentrated over the tooth region. It has not needed before any of the methods of reducing u.m.p. can be
always been appreciated that the equivalent unslotted mach- disregarded as a means of minimising the effect on critical
ine must carry the same flux as the actual machine for the speed.
comparison to be meaningful. Thus the equivalent smooth-
airgap machine has a larger airgap and a reduced pull com- It is possible to transform an eccentric annulus representing
pared with the slotted machine, in which the flux is concen- the airgap region into a concentric annulus by a bilinear
trated in the tooth region. Rosenberg2 was in no doubt about transformation. This method has been used by Swann15 to
this point and suggested that the u.m.p. is increased in the find the component of field strength normal to the surface of
ratio of the slot pitch to the tooth width. The effect of fring- the rotor of an induction motor with a nonuniform airgap. It
ing flux at the tooth corners can be represented by an imagi- is necessary to assume the surfaces to be infinitely perme-
nary increase in tooth width; the density over the increased able, and the damping effect of the cage cannot reasonably be
tooth width is assumed uniform, with zero density over the represented. It is interesting to note that Swann1 s result for
remainder of the tooth pitch. Since the same flux as in the the flux distribution around the rotor surface is in very close
'equivalent' machine is concentrated over the effective tooth agreement with the apparently crude approximation given in
width, the local surface traction is increased. Bradford7 Miles-Walker's book.6
suggests an approach involving the actual tooth and slot
widths, using flux densities over each determined by experi-
ment. Unfortunately, a comparison was made with an unslot- 2 RELATION BETWEEN AIRGAP FIELD AND
ted machine carrying more flux than the actual one, so that a
reduction in u.m.p. was predicted. If the same experimental UNBALANCED FORCES
results had been used to make a comparison with an unslotted The force on the rotor is completely determined by the mag-
machine with the same working flux, an increase in u.m.p. netic field in the airgap; and, by integrating the surface-field
would have been found to be slightly larger than would have stress over a cylindrical surface enclosing the rotor, the
been obtained by multiplying the unslotted value by the gap- total force may be found by using Maxwell's stress method.
extension factor, a method suggested by Carr. 8 It is very unlikely that the end surfaces of the rotor contri-
bute appreciably to the u.m.p., and so this part of the surface
It is well known that a cage rotor at speed has a lower value integral will be neglected, Previous large-scale experiment-
of u.m.p. than a wound rotor, owing to induced currents in the al work has involved flux measurements on the stator bore,
cage opposing the unbalanced field component. Miles-Walker6 but, as will be seen later, the most convenient reference
and Covo9 both realised that a cage reduced the u.m.p., but frame for field measurements, and indeed analysis, is one
they thought that the reduction was dependent upon the slip fixed with respect to the rotor. The surface over which the
frequency and disappeared at synchronous speed. They field is examined is one just outside the rotor, and it is
appear to have assumed that the equalising action depended apparent that the tangential component of stress there can-
on the total current carried by the rotor bars. More recent not contribute to a radial pull, and so consideration can be
work has indicated that a cage-rotor motor at low slip can be limited to the normal component, given by
expected to experience about one-third of the pull of an
equivalent wound-rotor machine. There is little improve- (1)
ment during run-up, since the cage is moving slowly com-
pared with the unbalance in the field.
or
An equalising winding on the stator can be used to reduce
u.m.p. at all speeds, as suggested by Gray and Pertsch, 1 or (2)
an arrangement of parallel paths in the stator winding can be
employed. Short-circuited damper windings in the stator
have not often been used, because of the extra cost, but, re- since the surface lies in the airgap. For a practical iron
cently, Frohne 10 has suggested that there are significant rotor, the distribution of F n over the surface does not show
advantages over the use of parallel paths. He claims that how the applied force is distributed, but, when integrated
parallel connections often result in forced vibrations being over the surface, it gives the total force on the rotor.
set up, whereas a passive equalising winding does not. If only static eccentricity is present, any search coil mounted
Ellison and Yang11 have recently investigated the effect of on the rotor surface sweeps out the surface of integration
parallel paths on acoustic noise and found the favourable con- during one revolution of the rotor. If all the bars of the cage
nections to be the same as those suggested by Krondl12 for are of similar composition, measurement of the airgap field
reducing u.m.p. Rosenberg claims that the most effective over a rotor tooth pitch will, in time, gather information for
arrangement in all cases is to put diametrically opposite a complete description of the field surrounding the rotor.
coils in parallel and then connect these pairs in series. The use of rotor-mounted coils avoids the problem of inac-
Stringer 13 claims that reductions of as much as 30: 1 can be curacy due to the subtraction of quantities of similar magni-
obtained from such an arrangement, though this seems rather tude, one of the problems involved in measuring on the stator
unlikely. Unfortunately, space restriction does not permit surface. The effect of eccentricity appears as a flux-density
the inclusion of the measurements taken with parallel con- component, alternating at a frequency corresponding to the
nections, and it is hoped to present these results in a future angular speed of the rotor. The various field components all
paper. appear at different frequencies and are modulated by the
main flux, which is at slip frequency; so frequency-selective
It is necessary to design a rotating machine so that the filters can be used to separate harmonic field components.
critical speed differs from the running speed by a safe mar- Transient u.m.p. cannot be determined from flux measure-
gin, while, in some cases, keeping this margin as small as is ments on a single tooth, but this paper is concerned only
reasonable for economic reasons. If u.m.p. is assumed to be with the steady-state condition.
proportional to eccentricity, its effect is equivalent to a
negative component of shaft stiffness. Thus the critical To examine the influence of different field harmonics on
speed of an induction motor is reduced by u.m.p. If the rate u.m.p., consider any point A on the rotor surface, as in Fig. 1.
of change of pull with displacement is Km, and the stiffness The Maxwell stresses due to normal and tangential field
of the shaft is Ks, the reduction in critical speed is given by components may be summed separately, and consideration
350 PROC. IEE, Vol. 120, No. 3, MARCH 1973
will be given first to the normal flux density B nA . This con- This represents a force rotating with an angular velocity cou
sists of a slip-frequency component and a number of m.m.f. with respect to the rotor and stationary relative to the sta-
harmonics, together with a series of harmonic components, tor. The pull on the bearings is therefore given by
each partially or completely modulated by the slip frequency.
B,, S
Initially, consider only the slip-frequency component and P= (7)
that due to its 2-pole modulation resulting from the eccen-
tricity, which will be called the unbalanced component.
where S is the surface area of the rotor.
In the same way, it can be shown that the pull due to the tan-
gential component of flux is

(8)

So far, the only harmonic field considered has been that in-
volving the unbalance component. In a similar way to the
above, it can be shown that all harmonic fields produce a ro-
tating force relative to the rotor, each harmonic having its
characteristic angular velocity. However, only those fields
which involve the unbalance component or its harmonics are
stationary with respect to the stator. Hence the forces due
to other harmonics produce high-frequency vibrations, but
no steady force of any significance. Modulations due to
eccentricity of higher m.m.f.harmonics produce forces in-
volving the product of two components, both of which have
orders of magnitude lower than the corresponding terms in
eqn. 7 and so can be neglected.
In much of the literature, it is assumed that the flux density
stator core varies round the airgap in a sinusoidal manner, according to

= KB 1 (9)
axis of
eccentricity
This would lead to a pull, given by
Fig.l _ KB12(e/g)S
Airgap of machine with static eccentricity (10)

which is equivalent to the assumption that

These may be expressed as B u = K B 1 e/g (11)
B nA = B o sin owt + {B u ' + B u sin (au>t + au)} Since K includes the damping action of the cage as well as
sin (wut + \u) (3) saturation and the effect of parallel paths in the stator wind-
ing, it is difficult to calculate. It was therefore decided to
The contribution to the u.m.p. from the radial-flux compo- explore by experiment the dependence of B u on eccentricity.
nent is found by integrating the component of the resulting
normal surface stress round the rotor as
3 EXPERIMENTAL INVESTIGATION
2V
= Lj r cos 9 d6 (4) A 7 5 kW 6-pole motor with a diecast cage was chosen for
o 2/u0 the experimental investigation. It was large enough for a
comprehensive search-coil assembly to be mounted on the
If the rotor has turned through the angle 6 relative to the rotor, but not so large that repeated dismantling would in-
field in time t, volve unnecessary labour. A 2-pole motor was not used,
since it presents a special case, with inherent equalising of
6 = otw/p the flux between poles. The stator bore and rotor surface
were machined slightly to improve regularity, until both de-
and so the pull can be expressed as parted from true cylinders by no more than 0"025 mm; in so
doing the airgap length increased by 0075 mm to 0525 mm.
r2pn B2 /CTO)t\ To simplify eccentricity adjustment, pedestal bearings were,
P=L I cos (ou>t/p)rdl 1 (5) used; these were of the self-aligning type, with outer frames
o 2n0 \ p / that could be split for removal of the rotor. The eccentricity
Substituting from eqn. 3 in eqn. 5 gives of the machine was adjusted by movement of the stator rela-
tive to the bedplate. Shims were used under the stator feet
Lr to increase eccentricity, and adjusting screws fitted to the
p = bedplate were used to ensure that the horizontal position of
o the stator was unchanged. The axis of eccentricity was
L rr r2pn r
therefore always vertical, with the smaller airgap at the bot-
+— J B o sin acot {B ' + B., sin (owt + a.M tom. In this way the u.m.p. was made to act in the same
Un 0 direction as the weight of the rotor, thereby avoiding move-
ment of the rotor due to any free play of the bearings. We
decided not to try to measure u.m.p. with strain gauges, since
the force is determined by the field, which was to be accu-
Lr .2P7T ,
-.2P7T rately measured, and since the main aim was to investigate
T1 f | B U ' + B u sin (owt + au)}2 sin 2 (wut the force mechanism rather than to obtain force measure-
^° ° ments for a single machine.
+ XU) c o s ( The stator winding was a double layer in 48 slots, divided into
The integrands in the first and last terms are periodic in 18 coil groups, the ends of each being brought out to a termi-
the interval of integration, and so they integrate to zero, leav- nal board so that a variety of parallel connections could be
ing the middle term, which gives made. We decided to use a set of long narrow search coils
extending over the entire length of a rotor tooth, each coil
P = 2TT — B a B u sin (w u t + Au') (6) spanning one-tenth of the tooth width. Thus the average radial
flux density, taken over the length of the tooth, could be found
PROC.IEE, Vol. 120, No. 3, MARCH 1973 351
for a series of positions extending across the tooth. A single 4.1 Slip-frequency component
coil enclosing one tooth was also required, so that the aver- The component at slip frequency relative to the rotor is the
age radial flux density for the whole tooth could be corrobo- fundamental flux wave. No significant variation with eccen-
rated, so checking the calibration of the coils. For the mea- tricity was observed; the distribution of the normal and tan-
surement of radial flux density, very accurate printed-cir- gential components over a slot pitch is shown in Fig. 3. At
cuit coils spanning 1 mm were used. These were formed of light load, the normal flux density is nearly constant over a
copper on thin plastic laminate and were stuck to the rotor tooth pitch. As the load increases, the mean level drops
surface after they had been accurately positioned. slightly owing to the voltage drop in the leakage reactance.
The measurement of the tangential flux component at the A more significant effect is the reduction in flux near the
rotor surface presented a problem. The permissible projec- trailing edge of the tooth owing to the rotor leakage field.
tion from the rotor surface into the airgap was severely r e - The tangential component is, on average, less than 10% of the
stricted, and, further, the search coils were required to effect- radial component, and its value at halfload is shown.
ively discriminate against the relatively strong radial field
to produce signals representing the weak tangential field.
Prior to assembly in the machine, the coils were mounted in
a set of Helmholtz field coils on a jig, which could be rotated stator-slot
opening
so that the sensitivity to flux in directions corresponding to radial search tangential
the radial and tangential ones could be found. Any coils coil search coil
which did not show a sufficiently good discrimination were
inspected and adjusted until a discrimination of at least 25: 1
was obtained for each coil. The total thickness of the coil
assembly, including adhesive, was 0-18 mm.
The position of the rotor in the stator could be explored at
standstill by using feeler gauges inserted in the airgap at the
ends of the machine, and this method was used for setting up
each rotor position. The eccentricity increases under the
action of u.m.p.; so it was necessary to measure the airgap
at a number of points around the machine while it 16 was run-
ning. A technique has been described by Bradford whereby
the variation in capacitance between an electrode on the sta-
tor surface and the rotor surface itself may be used to find
the rotor position, and a simplified version of this apparatus
was used. Since any eccentricity present is increased by
u.m.p., it is useful to know the stiffness of the rotor shaft so
that the increase in eccentricity can be confirmed. The external radius
stiffness was found by measuring the movement at the centre of endring
point of the rotor when a known force was applied.
Fig. 2
Saturation tends to limit unbalanced pull, since the permea- Rotor-search-coil positions
bility of the iron is reduced, and the reluctance to the un-
balanced component of flux is thereby increased. As a pre-
liminary investigation, the saturation of the main flux path
was investigated for a range of voltages from 25% to 135% of
the rated voltage, and it was decided to uprate the machine 121—
from 346 V to 415 V to make the density level typical of a
modern design. ^
When flux measurements with the search coils were being 10 • \

made, high-grade active filters were used to separate all the \a

significant components of each signal from the coils. The
exact frequency of each component depended on the motor O8 —
loading, and three values were chosen:
light load : 0-2 Hz slip
medium load : 0-8 Hz slip 0-6

full load : 15 Hz slip

0-2
A series of eccentricities was used, ranging approximately
from 57O to 40% of the airgap width. For each eccentricity, a
combination of active filters was used to select each har- I I I I I I I
monic in turn; the outputs from the filters were displayed on
an accurate oscilloscope and the maximum value was mea-
sured on the screen. Since all components were modulated centre centre
by slip frequency, an oscilloscope with a storage facility was rotation
necessary to find the maximum value accurately. Two sets
of printed-circuit coils were used as a check, and all mea- Fig. 3
surements were repeated several times so that mistakes in Slip-frequency flux component
measurement could be eliminated.

4 RESULTS 4.2 Unbalance component

After the rotor bearings had been .positioned relative to the The unbalance-flux component results directly from eccen-
stator, the nominal eccentricity was computed from 16 tricity and is of prime importance. Fig. 4 shows the distri-
measurements of the airgap taken at each end of the ma- bution of this component for an eccentricity of 0"043 mm.
chine. The increased eccentricity due to the unbalanced pull The variation with eccentricity of the mean values of radial
was found with the capacitance-probe device in the manner and tangential components over a slot pitch are shown in
described earlier, and a 'true' eccentricity was determined. Fig. 5. For small eccentricities, the unbalance field increa-
For each eccentricity, the peak-to-peak value of the e.m.f. at ses more or less linearly with eccentricity, the effect of load
the maximum point was measured and recorded for each being to increase this component owing to the increased
harmonic. The positions of the rotor search coils are shown leakage flux. However, a further increase in eccentricity re-
in Fig. 2. The flux-density components corresponding to the veals a saturation effect as the permeability of the iron
measured e.m.f.s were calculated, and their variation over a starts to fall and the component remains fairly constant
rotor-slot pitch are considered in turn, starting with the over a range of eccentricity variation. However, for large
fundamental flux. values of eccentricity, there is a further increase, although
352 PROC.IEE, Vol. 120, No. 3, MARCH 1973
not at the initial high rate experienced at small eccentrici- with flux density. The higher flux density in the region of
ties. The tangential component shows a gradual increase as reduced airgap leads to some saturation and consequently to
the eccentricity increases, and its value relative to the radial an asymmetrical distortion of the 2-pole modulation, which
component varies from 28% at low eccentricity to 36% at leads to even harmonics. Fig. 6 shows the variation of this
high eccentricity. component over a slot pitch. No significant variation with
true eccentricity was found.

30

20
o \o
25
15

10
20

I I I I I I I 15
•1 2 34 56 7 89 10 11 f
Slot slot
centre centre
rotation
10
Fig. 4
Unbalance-flux component for 0-043 mm eccentricity
a No load
b Half load '
c Full load I I I I I
d Tangential component at half load 7 8 9 10 11 f
slot
centre
30 — Fig. 6
Second harmonic of unbalance-frequency-flux component
a No load
25 b Half load
c Full load

20 4.4 Other harmonic components

No significant change with eccentricity was observed, either
for the m.m.f. harmonics or the slot-ripple components.
15 4.5 Product of slip and unbalance components
It has been shown earlier that the u.m.p. is proportional to
oa the product of the unbalance- and slip-frequency components.
The slip-frequency component does not change with eccen-
o 10
0)

30 >vT
f ' \
\
\ b
25 <
0 10 20 30 40 c c \ \ \
eccentricity airgap width,°/o
<D O
Du a \ \
<
c: u 20 <
Fig. 5
ai xo \° \ 0

Variation of unbalance component with eccentricity \ \ V . c

o "—< •f 1
\
a No load
b Half load
15
) J
c Full load
d Tangential component at half load m I I I | I I 1 1 1 1
1 7 8 9 10 11 f
t
slot slot
4.3 Second harmonic of unbalance centre rotation centre
The second harmonic of the unbalance-flux component may Fig. 7
be generated in two ways. Since the stator bore was not ex-
actly round, the variation in airgap length around the peri- Typical distribution of product of slip-frequency and un-
phery contained other components, apart from the 2-pole one. balance flux components for 0-079 mm eccentricity
One of these was a 4-pole component, which appears as a a No load
second harmonic of the unbalance frequency. In addition, a b Half load
second harmonic arises because of the permeability change c Full load
PROC.IEE, Vol. 120, No. 3, MARCH 1973 353
tricity, but the distribution is not uniform over the rotor- slot that B u is proportional to B x and, since B a and B x are the
pitch; so its effect is that of a weighting curve applied to the same in magnitude, the unbalanced pull may be expressed as
distribution of the unbalance component. A typical distribu-
tion of the product is shown in Fig. 7. It will be seen that the
leading half of the tooth makes a much greater contribution Pa sf(-
to the integral on load than does the trailing half, so that, in
this region, accurate measurements are important. This does
not necessarily represent the relative proportions of the The nature of the variation with eccentricity departs from
actual pull on the tooth as discussed earlier. The variations the commonly assumed relationships to a significant degree.
in phase of the 2-pole 'unbalance' flux and of the main flux It has been shown that, for eccentricities not greater than
are negligible over a rotor-tooth pitch and so they are ig- 10%, the u.m.p. is proportional to eccentricity and the rate of
nored. The average value of the product over a slot pitch, increase is high. As the eccentricity is increased from 10%
which represents the u.m.p., is shown in Fig. 8, plotted to 30%, the increase in u.m.p. is very small. The way in
against eccentricity. This shows a fairly linear region at which it depends upon load in this region suggests that it is
small values of eccentricity and a steady increase with load. limited by saturation of the leakage-flux paths. At even larg-
A levelling off occurs at all loads, followed by a gradual in- er eccentricities, the u.m.p. starts to increase again, although
crease until maximum-eccentricity values were achieved. there is not much practical importance in eccentricities
Bradford7 constructs a straight line through his graphs of greater than 30%. The vast majority of published work1)3*4
u.m.p. against eccentricity, although it is interesting to note predicts a steady increase in slope with eccentricity. This
that the actual points have a tendency to follow a trend not is not borne out by the experimental measurements described
unlike that described here. He used a motor with wider slot in this paper, and those published by Bradford are interpreted
openings, and so the effect of saturation might well be differ- as supporting the assumption of a linear relationship.
ent. The contribution from the tangential component is not Measurements of the tangential-flux components show that
shown in Fig. 8, since it is of the order of 3% of the contribu- these can be neglected in any further work on the subject.
tion from the radial component.

6 ACKNOWLEDGMENTS
—
30 — The work described in this paper forms part of an investiga-
tion of the electromagnetic behaviour of induction motors,
made possible by the generous financial support of the UK
o „''/ Science Research Council. This work formed the basis of a
—
25— Doctor of Philosophy thesis submitted by one of the authors
0
'$'
s
$/
•
to the University of Southampton. The authors wish to thank
_9—- ^^ Prof. P.Hammond for his encouragement.
3 O ~_o/

- 20 —
I /
o/

6
/
11 —-6 7 REFERENCES
Ck
P LI
JO
o Pi
1 GRAY, A., and PERTSCH, J. G.: 'Critical review of the
JO
c
3
X)
o
15 —
~ 'Jj bibliography on unbalanced magnetic pull in dynamo
electric machines', Trans. Am. List. Elec. Eng., 1918, 37,
Pt.H, pp. 1417-1424
2 ROSENBERG, E. : 'Magnetic pull in electric machines',
f 10 - I ibid., 1918, 37, Pt. II, pp. 1425-1469
a
3 VON-KAEHNE, P. : 'Unbalanced magnetic pull in rotating
electric machines'. ERA report Z/T142, 1963
1 1 1 4 FRIESE, W. and JORDAN, H. : 'Einseitige magnetische
E "0 10 20 30 40 Zugkrafte in Drehstrommaschinen', Elektrotech. Z, 1962,
83, pp. 299-303
a , 5 ROBINSON, R. C. : 'The calculation of unbalanced mag-
eccentricity/airgap width ,°/o netic pull in synchronous and induction motors', Trans.
Fig. 8 Am. Inst. Elec. Eng., 1948, 62, pp. 620-624
Variation of product of slip-frequency and unbalance flux 6 MILES-WALKER : 'Specification and design of dynamo-
components with eccentricity electric machinery' (Longmans, Green, 1915), pp. 57-61
a No load 7 BRADFORD, M. : 'Unbalanced magnetic pull in a 6-pole,
b Halfload induction motor', Proc.IEE, 1968,115, (11), pp. 1619-1627
c Full load 8 CARR, L. H. A. : 'Recent progress in induction motor
construction, J.IEE, 1936, 78, pp. 383-403
9 COVO, A. : 'Unbalanced magnetic pull in induction motors
with eccentric rotors', Trans. Am. Inst. Elec. Eng., 1954,
5 CONCLUSIONS
73, Pt.IIIB, pp. 1421-1425
The flux density at a point on the rotor surface can be des- 10 FROHNE, H. : 'The practical significance of unbalanced
cribed by a series of harmonic components, and it has been - magnetic pull. Possibilities of its calculation and damp-
found by experiment that the effect of static eccentricity on ing', Conti Electro Ber., 1967, 13, pp. 81-92 (in German)
the field is confined to one of these. It has been shown that 11 ELLISON, A. J., and YANG, S. J. : 'Effects of rotor eccen-
the unbalanced pull can be related to the unbalance and slip- tricity on acoustic noise from induction machines', Proc.
frequency flux components by using surface-integral methods IEE, 1971, 118, (1), pp. 174-184
of force evaluation, through the simple equation 12 KRONDL, M. : 'Les vibrations radiales autoexcite'e du
rotor de machines a induction a voies d'enroulement
B, paralleles', Bull. Ass. Suisse Elec, 1956, 47, pp. 581-588
P=-
13 STRINGER, L.S. : 'Unbalanced magnetic pull and shaft
voltage in electromagnetic rotating machines', Engineer,
The common assumption of a sinusoidal variation in gap den- April 1968, p. 631
sity round an eccentric airgap results in the relation 14 CRAWFORD, W.G. : 'Unbalanced magnetic pull and the
mechanical stability of rotating electric machines',
IV Engineering, 1951, 171, pp. 504-505
P= S -
15 SWANN, S.A. : 'Effect of rotor eccentricity on the mag-
netic field in the air-gap of a non-salient-pole machine',
There is ample experimental evidence that, for cage-rotor Proc.IEE, 1963, 110, (5), pp.903-915
machines, the pull is proportional to the square of the vol- 16 BRADFORD, M. : 'Unbalanced magnetic pull in a 6-pole,
tage; for example, Bradford7 found that voltage changes of 10 kW, induction motor with a series-connected stator
25% produced force changes very close to 56%. This means winding'. ERA report 5216,1968
354 PR OC. IEE, Vol. 120, No. 3, MARCH 1973