Professional Documents
Culture Documents
539
and
Zhidong Jia
ABSTRACT
The velocity of ultrasonic waves, which were launched using the pulse-echo method
and propagated in a stator insulation,, was measured to assess the insulation condition of the stator bars. A statistical minimum velocity (Vmin)and a statistical average velocity Way)were computed for a new and three sets of stator bars which
were in operation for 16, 18 and 23 years. After velocity measurements,' several
conventional electrical parameters such a s dissipation factor, capacitance, and
partial discharge were measured, and microscopic examinations of dissected insulation of tested b a r s were examined with a scanning electron microscope and a n
optical microscope. Experimental results showed that Vmi, and V,, decreased and
more microscopic defects were formed in the insulation with increasing operating
time of the stator bar. For the three sets of aged stator bars, the insulation conditions based on Vmi, and V,, were consistent with those assessed by measuring electrical diagnostic parameters. Therefore, Vmi, and V,, were used to assess the insulation condition of large generators. The measurement of ultrasonic waves may offer a broad prospect for stator insulation diagnosis.
Index Terms
Stator bars insulation, insulation diagnostics, ultrasonic, dissipation factor, partial discharges, capacitance, scanning electron microscope.
1 INTRODUCTION
T.
540
2 THEORETICAL BACKGROUND
2.1 INSULATION AGING
The stator insulation of large generators generally consists of epoxy-mica insulation. The manufacturing process
of this insulation may he described briefly as follows.
Firstly, glass-fiber reinforced mica tape is wound around
copper strands to a desired thickness, and numerous layers of mica tape make the insulation with layered structure. Then, the layered insulation as well as the copper
conductor is vacuum-impregnated with epoxy resin. After
impregnation and cure, the mica tape layers and the epoxy
resin are normally hounded, and a rigid and compact insulation is formed. At the manufacturing stage, the insulation is prone to producing gas-filled voids. The voids often
occur between layers and/or at the resin-copper interface.
More voids occur due to the deterioration of the adhesive
strength of the epoxy resin in the insulation under operating stresses. During a long-term aging process, some of
these voids are enlarged gradually. In severely aged insulation, delamination between layers, delamination of insulation from copper strand and even cracks through layers
appear. The thermal, electrical and mechanical stresses
concentrate on these microscopic defects, accelerating the
aging of the whole insulation material.
cable
111
...........................
... ... ... ...
...
..
..
.,
..
...
..
.
..
...
..
.
..
...
..
.
. . .
541
(4)
where, d is the nominal thickness of the ground wall insulation, T, and TB are the propagation time of wave F and
wave B respectively, t = TB1T, is the propagation time
of the ultrasonic waves through twice the thickness of
ground wall insulation. The thickness of the ground wall
insulation may slightly increase due to the insulation expansion in the aging process, so d in the equation ( 2 )
stands for nominal thickness.
Provided that the gas in the insulation is essentially atmospheric, Vz in equation (4) is equal to the velocity of
the ultrasonic wave in air, 331.45 m/s. Then, l/V, = 3.02
x10-3 s/m. The ultrasonic velocity
is not constant
during insulation aging, due to the changing microscopic
structure of the insulation material. A value of 2.4X 103
m/s may be adopted for y in equation (4). Thus, both
1/y and l/y2 are calculated to be 1 / ~ . = 4 . 1 7 X 1 0 - 4
s/m, l/y2= 1.74X lo- s/m. It is obvious that 1/y >>
l/y2 and l/Vz >-> 1/y2. So, equation (5) may be substituted for equation (4).
dS,
dr=-+-
dS,
V V ,
AS
A t = - + A
v,
542
3 EXPERIMENTAL
3.1 SPECIMENS
The tested stator bars were classed into four sets according to the operating time. Set I was cut from a new
stator bar, which was a standby stator bar for a watercooled turbo-generator (300 MW, 18 kV), and was 3 m in
length. The new stator bar had not been subjected to any
operating stresses and was used as a reference bar for test
comparison purposes. Sets 11, 111 and IV had been in service for 16, 18 and 23 years, respectively in three watercooled turbo-generators (300 MW, 18 kV), that were produced by the same manufacturer and were in base operation. Sets 11, I11 and IV were cut into 3 m, 3 m, and 6 m in
lengths, respectively when they were taken from the generators. Sets 11, 111 and IV have 3, 12 and 6 stator bars,
respectively. All the stator bar insulation were made of
epoxy-mica composite insulation. The nominal thickness
of the ground wall insulation was 5.5 mm. The winding
temperature in selvice was about 75 C.
\-I
Figure 4. Typical ultrasonic A-scan results from stator bare for a new and after different years in service. a, new; b, 16 years; c, 1 R years; d, 23
years.
diagnostic tests, all stator bars were dried at 100 "C for 24
h. A QS-27 Schering bridge was used for dissipatlon factor and capacitance measurements, and a TE571 digital
PD analyzer manufactured by Haefely Trench Corporation was used for partial discharge measurements. The
partial discharge behavior of each stator bar was measured over a duration of 2 nun.
The electrical diagnostic parameters for this experiment
included the dissipation factor tip-up (Atan6 = tanSE AC C,-Co
tan So), the change of capacitance (=
), the
C"
CO
maximum discharge magnitude (9") and the mean discharge magnitude (q,,,,) at operating voltage (10 kV).
tanSE and C, are the dissipation factor and the capacitance at operating voltage, respectively, tan&, and COare
the dissipation factor and the capacitance at 0.2 times the
operating voltage (2 kV) respectively.
After the electrical diagnostic tests, the stator insulation was dissected carefully from-one part of the tested
~
2.4
2 2.2j
1
.-.,
'E
2.01
1.6
2.6
(9)
.18
*
.
.23
(8)
(8)
16
K" = cI V,/N
...
..
I..
1.8
2.8
Vmi,=min(V,, V2. V3,,,Vn,,.VN]
543
z2.1
d
>g
2.2
:;
;.
A
2.0
,~
.-.
1.8
16
18
operating time (Y)
(b)
23
544
stator bars with a low-velocity diamond saw, and microscopic examinations with scanning electron microscope
(SEM) and optical microscope were performed.
71
6
c
m
41
.'
I
~
2L
1
k*
0
16
18
23
16
23
(a)
(b)
6l
1.o
.1
0'
18
16
18
16
23
18
(C)
AC
(d)
Figure 6. Plots against operating timc. a, Atan6; b, -; c, qmax;d, qmern
CL.
23
that the insulation of the two stator bars that were in service for 23 years failed at the rated voltage (18 kV) during
the measurements of Atan6, AC/C,. Dissecting the insulation of two failed bars, revealed that macroscopic ruptures existed among the insulation layers.
..
545
..
546
oid
5 CONCLUSIONS
delamination
Figure 0. Photograph of microscopic defects in stator insulation
taken with an optical microscope, Magnification, 150.
4.4 ANALYSIS
The microscopic examinations mentioned above revealed that more and more microscopic defects such as
voids and delamination were formed in the process of insulation aging. These defects could result in changes in
the ultrasonic velocity and the electrical diagnostic parameters.
According to equation (7), the ultrasonic velocity decreases as the dimension of the defects is enlarged. In the
aging process of stator bar insulation, voids are formed
and enlarged continually, therefore VmLnand V, decrease
with increasing operating time of the stator bar.
As is well known, the electrical diagnostic parameters
such as the AtanS, the AC/C,, the q,, and the q,,,,
increase with the development of microscopic defects in
insulation. So the AtanS, the AC/C,, the qmaXand the
q,,,, increase as the operating time of the stator bar extends.
Based on the experimental results presented here, it can
be seen that both the ultrasonic velocity and the electrical
diagnostic parameters depend on the operating time of
the stator bar. However, comparing the changes of ultrasonic velocity with those of the electrical diagnostic pa-
ACKNOWLEDGMENTS
We gratefully acknowledge the financial support by National Natural Science Foundation of P. R. China for part
of research (the project No. 59837260).
REFERENCES
111 K. Kimura, "Progress of Insulation Aging and Diagnostics of
HV Rotating Machine Windings in Japan", IEEE Electrical Insulation Magazine, Val. 9, pp. 13-20, 1993.
121 T. Tsukui, M. Takamura and Y. Kako, "Correlations Bchveen
Nondestructive and Destructive Tests on High-voltage Coil Insulation for Rotating Machines", IEEE Trans. EI, Vol. 15, pp.
118-127, 1980.
L31 Y. J. Kim and J . K. Nelson, "Asscssment of Deterioration in
Epoxy/mica Machine Insulation", IFEE Trans. El, Vol. 21, pp.
1026-1039, 1992.
[4] G. C. Stone, H. G. Sedding, R. A. Lloyd and B. K. Gupta, "The
Ability of Diagnostic Tests to Estimate the Remaining Life of
Stator Insulation", IEEE Trans. Enorgv Conversion. Vol. 3, pp.
833-840, 1988.
151 B. K. Gupta and 1. M. Culbert, "Assessment of Insulation Condition in Rotating Machine Stators", IEEE Trans. Energy Conversion, Vol. 7, pp. 500-508, 1992.
[61 J. K. Nelson, S. Aziri-Ghannad and H. Li, "Theory and Application of Dynamic Aging for Life Estimation in Machine Insulation",IEEE Trans. DEI, Vol. I , pp.773-781, 2000.
[71 G. C. Stone, H. G. Sedding, N. Fujimoto and J . M. Braun,
"Practical Implementation of Ultra-wideband Partial Discharge
Detectors", IEEE Trans. El, Val. 27, pp. 70-81, 1992.
[SI J. C. Song, H. K. Xie and Y. H. Cheng, "Study OD UWB Frequency Charactcrirtics ai Partial Discharge as a Criterion ai Aging Degree of Stator Winding Insulation", Proc. 6th IEEE Intern. Conf. Properties and Applications Dielectric Materials
(ICPADM), Xi", China, Vol. 1, pp. 181-184, 2000.
[9] A. K. Xie, "Ageing Extent Assessmcnt of Large Generator Insulation Using Advanced Characterization Techniques", Proc.
IEEJ Intcm. Sympos. Electrical Insulating Materials (ISEIM),
Japan, pp. 653-656, 2001.
547
~I