Professional Documents
Culture Documents
J.W. Wood
H.G. Sedding
W.K. Hogg
I.J. Kemp
H. Zhu
1 Introduction
Abstract: Partial discharge (PD) specifications
exist for all types of electrical equipment except Partial discharge (PD) acceptance tests are now applic-
motors and generators. Whereas specifications able to most HV power equipments. For example:
essentially require that the equipment is discharge-
free the stator conductor bar insulation of the
gas insulated switchgear (GIS) [l] < 2-5 pC
bushings [a] < 1opc
motors and generators under consideration is < 5 pC at 1.5 U N
cables [3]
micaceous and has excellent PD resistance. Conse- < 300 pC at 1.3 U N
quently, the intention in a specification would be transformers [4]
to permit a certain level of PD. It is therefore par-
< 500 pC at 1.6 UN
ticularly important that the problem of over- where U, = nominal voltage. Some years ago at the sug-
coming the attenuation between the PD site and gestion of a user, Ontario Hydro, collaborative work
the measurement equipment at the winding ter- began with Parsons Turbine Generators, a manufacturer,
minals be solved. Within the general context of to investigate the possibility of developing a PD test spe-
examining the feasibility of a PD specification, the cification to apply to HV generator and motor stators,
problem of PD signal attenuation is addressed. which is the last area of power equipment to require spe-
The approach adopted was to seek a region of low cifications to be defined.
attenuation by examining the whole of the pos- The probability of a ‘true’ insulation failure initiating
sible range of PD measurement frequencies. The from a PD within a high-voltage machine is small.
investigations described consist of experiments on However, machine users would welcome the introduction
a full-size 500 MW endwinding model and on two of an acceptance test for economic reasons [SI. When an
large high-voltage motor windings. Careful screen- outage is required in service, the costs can easily exceed
ing to reduce radiative pick-up from the cali- E100OOO per day for a base-load nuclear unit. Added to
bration generator simulating a PD source and a this are maintenance and perhaps extensive refur-
direct earthing scheme enabled the essential ele- bishment costs. Consequently, there is a strong desire to
ments of PD pulse transfer through a winding to reduce maintenance costs by ensuring that the machine
be determined. The frequencies investigated has been correctly manufactured and serviced. An over-
ranged from kHz to GHz. Pulse transmission and voltage test does not necessarily guarantee this. An arbi-
frequency spectrum measurements were made. It trary PD specification would not be acceptable to a
is concluded that the PD pulse transmits essen- supplier. However, suppliers recognise that failures affect
tially as a travelling wave, the high-frequency their reputation and this is of prime concern. An accept-
components being highly attenuated. No signifi- ance test might additionally enable important design
cant crosscoupling was apparent through the end- improvements to be achieved, e.g. operation at higher
winding region. The windings acted as a low-pass electrical stresses.
filter, the cut-off frequency being dependent on the The setting of a PD specification for stators is a
winding length, and being below 1 MHz. Conse- complex undertaking and involves theoretical and experi-
quently, it is concluded that it should be possible mental studies. At the start of the project there were
to measure PD, using equipment with a passband several problems to be solved before this could be
below the low-frequency cut-off, with low atten- attempted; some of these exist now in the instances above
tuation. where PD acceptance tests have been established. In
0IEE, 1993
Paper 9651A (S2), first received 9th February and in revised form 11th
This contribution has been assembled from experi-
May 1993 mental work at Ontario Hydro, Parsons Turbine
J.W. Wood is with the Generator Design Department, Parsons Turbine Generators and Glasgow Caledonian University,
Generators Ltd., Heaton Works, Newcastle on Tyne, NE6 2YL, UK and from discussions with B.K. Gupta and G.L.
H.G. Sedding is with the Ontario Hydro Research Division, 800 Ford at Ontario Hydro and G.C. Stone at Iris
Kipling Avenue, Toronto, Ontario, Canada M8Z 5S4 Power Engineering. Some components of the cali-
W.K. Hogg is at 32A Offington Lane, Worthing, Sussex, UK bration work were supported by the Electric
I.J. Kemp and H. Zhu are with the School of Engineering, Glasgow Power Research Institute under RP2577-1.
Caledonian University, Cowcaddens Road, Glasgow, G4 OBA, UK
winding as a transmission line with respect to the trav- b Cross-section of model core
elling wave component of the discharge signal.
detection frequencies from approximately 35 to 250 kHz
A typical arrangement was as shown in Fig. 2. The was used in certain instances to obtain a relevant quant-
cable details are given in Table 1. Matched junctions with iative measure of discharge signal attenuation. The six
unequally spaced cable lengths between them enabled conductor bars (three top and three bottom bars) have
waveforms to be closely studied for evidence of reflections epoxy mica paper insulation and a low conductivity
from the injection point. No significant reflections were coating on their outer surface covering the model core
observed with the IEC method of injection. With a length (approximately 1 m); total length of each half bar
Trichel pulse generator (point-hemisphere gap) attached from one end (stator centre line) to endwinding nozzle is
to the energised cable no photographic records could be approximately 5 m.
obtained, but a similar conclusion was reached from
visual observation of waveforms on the CRO in so much When experimentation began radiative pickup was
found to be signficant. Fig. 4a shows the signal level on a
a
35m 40m 25m
short (1 m) length of conductor bar approximately 3 m
i i away from the injection point, lying on the earthed floor
I I plane. An unexpected amount of screening, six layers of
aluminium foil, was necessary to reduce the signal picked
up to an acceptable level. As appropriate to the experi-
mental condition the bars in the model were screened to
this degree to ensure accurate measurements were being
_L made. Fig. 4b shows the coupling from individual bars
designated A and B in the model with and without
Fig. 2 Measuring arrangements for cable pulse tests screening of bar B (which affects A transmitted pulse
412 IEE PROCEEDINGS-A, Vol. 140, NO. 5 , S E P T E M B E R 1993
waveshape). An example of the measurement logic is that connected via a coupling capacitor to the other, the
by joining at their nozzle positions two fully screened top detected signal varied with spacing as shown in Fig. 5.
bars situated in the same slots as two similarly screened
bottom bars it was possible to examine any pickup due
to common magnetic flux. No evidence of any significant
pulse coupling was observed. Fig. 4c shows a typical
example of propagation by travelling wave in which the
peak pulse amplitude is reduced as the wave progresses
-1 0
1
along the bar.
1 - 50 1
0 0.25 0.5 0.75 1.00 1.25 1.5 1 '5
endwinding separation, mm
0
which assures signals in that frequency band are trans-
mitted without distortion. timebase 500nsldiv
The form of the detected signal as a simulated dis- C
charge pulse progresses from an injection point around Fig. 8 Voltage induced in adjacent endwindingsfrom discharge simu-
one coil of motor A is shown in Fig. 7. In this instance lated in endwinding ofA2-I, motor A
a Applied voltage
IA 6 b Induced voltage
c As above, with slower sweeps
> (i) A2-1; (ii) A2-2; (iii) A2-3; (iv) A2-4
e
> 20
E
R 10
al
m 0
-?
c
0
U
U
a J -.1 0
I
= -2 0
I I I ul
I I I I E - 30
4 0 ~
.50 I I iiiiiii I I IIIIIII i iiiiiiii -T MII I
I IIIIIII
lo4 lo5 lo6 10’ lo8 lo9
frequency,Hz
a
20
10
m
u 0
5
u -10
=c - 20
E - 30
LO
50
1o4 lo5 1o6 td lo9
frequency,Hz
b
Fig. 9 Motor A stator winding frequency characteristics
a Injection points: 6 of A2-1, A2-2, A2-3
Detection point: 0 of A2-1
-A2-1 _______ A2-2
-A2-3
-A2-4 - _ _ _ _ _A2-5
_ A2-6~
4 Discussion