machine stators
S.C.Chang
R. Yaca m in i
Introduction
In many drive applications involving nonlinear converters there is a high harmonic content in the machine
input current. The magnetic force wave therefore has a
rich frequency composition which enhances the possibility of coincidences between the frequencies of force
components and the natural frequencies. This could
eventually produce significant levels of machine noise
and vibration. The accurate determination of the
machine stator natural frequencies and the associated
mode shapes are thus of importance in the solution of
machine noise problems and in the determination of
machine vibration characteristics.
Over the years a number of researchers have
attempted to calculate the mechanical response of
motor stators. Classical methods, experimental meth0IEE, 1996
ZEE Proceedings online no. 19960184
Paper received 14th March 1995
The authors are with the Department of Engineering, Aberdeen University, Aberdeen AB9 2UE, Scotland, UK
242
ods, finite element analysis, or more usually a combination of these methods, have been developed.
As summarised in [I], Carter first developed a
method using a single thin ring model to calculate the
natural frequencies of a stator core which is directly
subject to the electromagnetic force and many later
researchers used the ring model. As described in [2],
Erdelyi developed the double-ring method applied to a
medium size machine whose frame is linked to the core
by key bars. The model included two thin rings joined
by key bars. As an extension of Erdelyis work [2],Ellision and Yang investigated the calculation of the natural frequency of a stator having a thin frame coupled
solidly through key bars to a thick core loaded with
teeth and windings, taking into account bending, shear
extension and rotary inertia.
Work has also been done on the encased machine
stator. Based on the three-dimensional elasticity theory
Verma et al. [3-5] developed an analytical method in
which the energy method along with the principle of
Rayleigh-Ritz were used to derive the general frequency equation of the stator. Cylindrical shells were
used for the stator yoke and the frame, whilst the teeth,
windings and cooling fins were treated as beam type
elements. They used a double series to describe the
modeshape functions.
Garvey [6] carried out research work to analyse
vibration characteristics of large D C motor stator. In
his work, the properties of a laminated core structure
was addressed.
In recent years, the finite element method has been
used to calculate the natural frequencies of machine
stators including the work of Yang, Belmans et al. and
Watanabe et al. [7-91. This method has become increasingly popular, and today it is in more or less routine
use for structural analysis. In comparison with the classical method, the finite element method is more efficient in coping with complex stator constructions and
therefore is able to give more accurate results. However, it has been found from finite element analysis and
experimental studies carried out on a number of small
and medium size machine stators, a few important
issues, in connection with using the finite element
method, such as modelling the laminated core, modelling winding and winding overhang and damping need
to be addressed, these will be described elsewhere. Also
the finite element method requires a large capacity
computer.
Experimental methods can give direct and accurate
results so long as they are properly carried out. Experimental studies have always been used along with analytical or finite methods as a verification [2, 9, 101.
IEE Proc.-Electr. Power Appl., Vol. 143, No 3, May I996
Fig. 1 FEmodel
Regular ring
ring
Fig. 2 FE model
Ring with small notch
60
20
-20
I1
- 4 O k '
'
"I
1000
'
'
Hz
"
'
2000
"
'
"
3000
- 4 O t '
"
"
1000
Hz '
"
"
2000
"
"
'
3000
tor
60
40
20
-20
-40
-40
600
1200
HZ
1800
2400
3000
3. I Experimental set-up
Three dimensional measurements were now adopted to
include the machine stator. Fig. 7 shows a sketch of
the tested stator. The actual stator general parameters
and structure details are given in the Appendix.
Around the periphery of the internal surface of the stator core, 48 points were chosen as measuring points as
shown in Fig. 7. This number of points is sufficient to
capture the relatively high circumferential vibration
modes. The measuring points are also uniformly spaced
along the core internal periphery, which is helpful for
data processing and the identification of circumferential mode shapes. Axial measurements around the circumference were also carried out, each including nine
measuring points along the stator length.
-50
m
U
-60
-70
300
900
2100
1500
2700
3300
HZ
Fi
. 9 Measured
15
21
33
39
45
48?Hz
Circumferential (n = 2 )
27
-------c--C.---c----t-
Fi
1.0
Longitudinal (m = 0)
0.5
0
-0.5
------3
15
21
27
33
39
45
-&----o-+----c-
Fi
48?Hz
Longitudinal (m = 0)
Fi
27%4Hz
Circumferential (n = 4)
1 .o
0.5
-0.5
Fi
18f6H.z
Circumferential (n = 3)
15
21
27
33
39
45
1
9
&&--e--+-&&-
Fi
27%4Hz
Longitudinal (m = 0)
245
ten equal intervals and the y axis represents the normalised amplitude of the stator FRF along the core
length. The eight circumferential locations were identified using symbols as shown at the bottom of each diagram.
I
0.5
17
25
33
41
1
measuring point
Fig. 17 Measured modeshape waves associated with stator natural frequencies
Mode order n = 2. 488Hz
1
-1
.o
15
4
21
27
33
39
45
::o
LL
--e---------
-1
17
25
33
41
1
measuring point
Fig. 18 Measured modeshape waves associated with stator natural fiequencies
1
kLL o
-1
1
25
33
41
measuring point
Fig. 19 Measured modeshape waves associated with stator natural fiequencies
1
17
Table 1: Measured values of resonance pairs on the stator (frequencies underlined were previously identified)
Frequency values
Values, Hz
Difference
Hz
Values, Hz
488
504
16
1896
1932
36
2704
2704
3 160
2664
40
2744
40
3192
32
An examination of the circumferential and longitudinal modeshapes of these resonance pairs was then carried out. For the purpose of demonstration,
modeshapes associated with the pair 488 and 504Hz
and the pair 2704 and 2744Hz are shown below.
Fig. 20 Measured stator circumferential mode shapes associated with resonance pairs
1 .o
0.5
10
0.5
0
-0.5
-0.5
-1 .o
- 1 . O L
1
2
2
3
3
15
4
21
27
33
39
45
-----%------A-
Fig. 21 Measured stator longitudinal mode shape associated with resonuncej?equemy pair 488Hz und 504Hz
Longitudinal m
15
21
27
33
39
45
0, 488Hz
-A-+----
Fig. 24 Measured stator longitudinal mode shape associated with resonancefrequency pair 2704Hz and 2744Hz
Longitudinal m = 0, 2704Hz
10
0.5
-0 5
-1.0 t
1
3
15
21
27
33
39
45
--a------o----e-----.r----c--c-
15
21
27
33
39
45
e
-
Fig. 22 Measured stator longitudinal mode shape associated with resonance fvequency pair 488Hz und 504Hz
Fig. 25 Measured stutor longitudinal mode shape associated with resonancefrequency pair 2704Hz und 27441%
Longitudinal m = 0, 504Hz
Longitudinal m = 0, 2744Hz
247
Fig. 23 shows the circumferential modeshapes associated with the pair 2704Hz and 2744Hz in thick and
thin lines, respectively. This Figure shows that this pair
have the same circumferential mode order (n = 4) and
again take a similar circumferential modeshape. The
measured longitudinal radial vibration of the pair is
shown in Figs. 24 and 25. As may be seen from these
Figures, longitudinal vibration associated with the pair
is approximately uniform along the stator length. This
again suggests that this pair take a similar mode shape.
Further examination shows that the same is also true
for the circumferential and longitudinal modeshapes of
the other resonance pairs listed in Table 1.
Fi
62!HZ
Circumferential (n = 2)
0.5
Londitudinal
mode order
Frequency
values, Hz
n=2
m=O
672
n=3
1812
n=3
m=O
m=O
n=4
m=O
2596
n=O
m=O
3026
1596
Again, the circumferential and longitudinal modeshapes of these frequency components were examined.
It is found that these frequency components are associated with similar radial vibration modeshape patterns
to those previously revealed in this Section. For the
purpose of display, the modeshapes associated with
672Hz, 1812 and 2596Hz are shown here in Figs. 2631.
Fig. 26 shows that the circumferential modeshape of
the component 672Hz is very regular and close to sinusoidal but of low amplitude. This is a similar modeshape to the pair 488Hz and 504Hz in Fig. 20. Fig. 27
shows that the measured longitudinal radial vibration
of this component is uniform along the stator length.
This also shows that its longitudinal modeshape is a
straight line which is a similar shape to 488Hz and
504Hz in Figs. 21 and 22.
-1.01
2
3
3
9
15
21
27
33
39
45
---&.----+a-
Fi
679Hz
Longitudinal (m = 0)
Again, as may be seen from Figs. 28-31, the frequency components 1812Hz and 2596Hz take similar
circumferential and longitudinal modeshapes to that of
1896Hz and 2704Hz shown in Figs. 11-14.
Based on this analysis, some further interesting conclusions can be drawn. The asymmetries in the structure of the stator causes it to have more than one
natural frequency for each specific circumferential
mode order. Their circumferential and longitudinal
radial vibration mode shapes are similar. For the measured stator, this can be seen from frequency components 488Hz, 504Hz and 672Hz for the mode n = 2
IEE Proc.-Electr. Power Appl., Vol. 143, No. 3, May 1996
0.5
-0.5
21
27
-1.01
1
3
15
33
39
45
----A-----S--+---e--
I
T
.o t
2
3
3
9
----e--+----
Fi
15
21
27
33
39
45
2596tfz
Longitudinal (m
Conclusions
-0.5
-1
Acknowledgments
The authors wish to thank Dr R. Neilson and D. Martin for their advice on many occasions and also
acknowledge the financial support of the SERC and
the Ministry of Defence.
References
= 0)
3.4 Discussion
The analysis described so far has shown, in general, the
effects of the stator irregularity on stator vibration
behaviour and does not differentiate between frame
and core irregularities. It has been argued [20] that the
teeth are identical and uniformly spaced around the
periphery and constitute a small proportion of the core
mass and can be assumed to be spaced closely enough
for their effects to be averaged (or smeared) over the
TEE Proc Electr Power Appl , Vol 147, N o 3, May 1996
250
7. I
Appendix
Stator details
Horse power, kW
37
Number of phases
Rated frequency,
Hz
Rated voltage, V
50
415
64
Number of poles
Connection
Encased
Number of slots
48
Airgap length, m m
0.6366
Airgap radius, m m
124.459
196.46
188.46
154.46
124.78
Frame length, m m
48 1
Core length, m m
222
Height of tooth, m m
29.68
12
10
13