Professional Documents
Culture Documents
IEEE TRANSACTIONS ON
POWER
APPARATUS AND
SYSTEMS
500
550
Maximumdryoperating
318
1-minutevoltage
withstand
740
60-cycle
60-cycle wet 10-second withstand
665
Impulse, 1'/,x40
Full wave withstand
1550
1785
chopped wave 2000
~~~~~~~~~~~~~~~~~~~~~~3-p.s
2-ps chopped wave
Switching surge
1050
1250
500
550
318
860
775
1800
2070
2320
1250
1250
BIlL
These interpretations will be verified soon in another HighVoltage Laboratory test program.
Table I presents the various existing, proposed, or anticipated dielectric test requirements for 500-kV circuit breakers,
VOL.
PAS-84,
NO.
10
OCTOBER 1965
TEMIA
SHEDRN
~~~~~AS
SHIELD
GOMPRESSIONSPRINGS/
WEAHER rOVER
\CL.RENT-CARRYING
| >
TORI-SHAPED
MOUNTING FLANGE
&
r
Fig. 3.
3
' ':
dH
Closeup of an EHV circuit-breaker bushing showing
of
an
bushing
and fall to the root groove without creeping along the protected undersurface. Actually, 46 percent of the total creepage is protected by the overhanging skirts, a most significant
statistic on an insulator whose creepage/striking distance
ratio of 2.8 approaches the practical maximum of 3.0.
Ceordination Theory
There are three fundamentals of insulation coordination
which must be observed in the design of a live-tank circuit
breaker.
OCTOBER
interrupter.
oGAP
\
CURRENT
INTERRUPTER
MODULES
TRANSFORMER
LIGHTNING
ARRESTER
>1.3pu
3- '-l^
J
1965
COORDINATING
-- .
4 1F1
A
excessive
withstand
if clean
and dry.
Such
variation voltage
makes any
insulator
or bushing
unreliable
porcelainstrenlgth
as a coordinating gap.
2. A coordinating gap also may carry a power-follow arc which
should not be drawn near the surface of a porcelain column.
SIXTY-CYCLE TESTS
A dry 1800-kY BIL circuit breaker is required to withstand, for one minute, a 60-cycle voltage of 860 kV. The
voltage distribution and coordination calculations indicated
that a single bushing must withstand 860 kY by 0.29 or 250
kV, and it must not break down internally during this test.
The minimum one-minute, dry withstand of any bushing
tested was 255 kY, and there were no internal breakdowns
853
VOLTAGE DISTRIBUTION
Verification of voltage distribution calculations
were of
prime importance to the remainder of the High Voltage Laboratory investigation. The technique employed to determine that proportion of the total applied voltage which appeared across each of the six series-connected interrupters
utilized electrostatic voltmeters connected to the energized
line terminal and other intermediate terminals. The electrostatic voltmeter leads were directed away from the breaker
along equipotential surfaces. Ground current from the remote, grounded terminal was held constant by varying the
supply voltage to adjust for the measuring meter's influence.
Percent of applied voltage appearing across each interrupting break as measured is:
bushing60-cycle wettest, right-hand horizontal Dog-Bone bushing is shown withstanding 180 kV while upper tilted bushing
flashes over
while each was flashed over ten times at an average of 275 kV.
Under standard rain conditions, the 10-second, wet withstand of an 1800-kV BIL circuit breaker should be 775 kV.
A single bushing must, in order to effectively coordinate,
withstand 775 kV by 0.29 or 225 kV. Some statistical scatter
is to be expected with wet tests; consequently, 10-second withstands were recorded as low as 218 kV and as high as 231 kV,
with a 226 kV average.
The ceiitral bushing in Fig. 5 is the Dog-Bone shown withstanding 180 kV wet while the adjacent bushing breaks down.
Note the distinctive water shedding pattern of this porcelain.
IMPULSE TESTS
Using similar logic, a single bushing of an 1800-kV BIL
circuit breaker must have an external, corrected impulse
withstand voltage of 520 kV. Internally, it must withstand
uncorrected crest voltages in excess of 520 kV.
With this test set up, the corrected positive full wave withstand of a single bushing was 492 kV; the negative was 532
kV. However, when the interrupter module was elevated to
its normal operating height above ground, the electrostatic
field changed sufficiently to raise the positive full wave withstand to 520 kV. The negative withstand remained acceptable after elevatioin.
Additional impulse shots taken at crest voltages above 520
kV uncorrected, with the bushing chopping the wave, verified
that a higher impulse withstand existed within the interrupter.
Both the oscillographic record of the applied voltage and the
aforementioned camera verified the absence of internal
breakdowns during tbese tests.
Tests on Complete Pole
Therefore, the coordination calculations which had assumed a 22 percent maximum distribution were somewhat
pessimistic but will not be changed. The taller, 1800-kV
BIL breaker should have slightly more favorable distribution
as the stray capacity to ground will be less.
operating voltage.
utilized.
SIXTY-CYCLE TESTS
Figure 1 shows the dry 1550-kV BIL circuit breaker withstanding 850 kV in the closed position, while Fig. 7 shows a
flashover at about 900 kV under the same conditions. In the
open position, the circuit breaker was tested with the full
1000 kV (Fig. 8) available in the Westinghouse High-Voltage
OCTOBER
INCEPTION OF AUDIBLE
600
400
CORONA
VISIBLE
/ __*
BREAKER CLOSED
AMBIENT,/
(0 -'
e0
BREAiK
R OPEN
100
200
APPLIED
300
VOLTAGE (KV)
500
400
-_____
dy-or 1
Fig. 8. 1550-kV BIL circuit breaker withstanding 1000 kV, 60cycle dry, for 1 minute in the open position
1___
Fig.9. 1550-kVBILcircuitbreakerwithstanding750kV,60-cycle
wet, in the open position
Fig. 10. 1550 kV BIL circuit breaker flashing over at 800 kV,
60-cycle wet, in the open position
855
SWITCHING SURGES
System design engineers are convinced, and have proved,
that 500 kV systems will receive their greatest insulation stress
from switching surges. These are overvoltages produced on
a transmission line whenever switching occurs and can, under
certain conditions, attain a crest voltage level higher than
2.5 times the maximum line-to-neutral crest voltage.
Much has been written in the last few years [6]-[14] about
switching surges and their effect on line and apparatus insulation. However, no definite switching surge test procedure
or voltage requirements have been written for power circuit
breakers. The technique adopted for this investigation represents a composite of those features deemed significant by
other investigators [9], [12]-[14]:
1. A wave which crests in 125 us and decays to half-voltage
by 4500 ,us was chosen to be in a range found most critical in
air. The wave shape definition is that used by Atwood, et al.
[14].
w-et
856(
IEEE
TrRAN5ACTION5
ON POWER
APPARATUS
AND
SYSTEMS5
OCTlOBER
NWITHSAND L!VEL
>
1050
AR CL
0.2
5
20
50
80
95 9999E 99.99
PROBABILITY OF FLASHOVER IN PERCENT
Fig. 11. Wet switching surge probability curve for 1550-kV BIL
live-tank circuit breaker in the closed position
negative polarity
I400C
a ~~~~~~~~~~~~
a
1300
250
~~~~~~~~~~terminal
z
50
80
95 9 9i9iT 1s
PROBABILITY OF FLASHOVER IN PERCENT
02I1
Fig. 13. Wet switching surge probability curve for 1550-kV BIL
live-tank circuit breaker in the open position
1965
500-Kv SF6
BREAKERS
857
BREAKER CLOSED
The closed circuit breaker containing one atmosphere (0 psig)
of SF6 will withstand, without internal breakdown, any of the
closed position tests discussed in this paper. In other words,
the breaker will, at 0 psig:
1. Withstand for one minute the required 60-cycle test
2. Flashover externally on 60-cycle overvoltages
3. Withstand its BIL and 3- and 2-ps impulse chopped waves
4. Withstand switching surges as graphically shown in Fig. 11.
BREAKER OPEN
In the open position, the dielectric strength across the
separated contacts becomes critical whenever all positive SF
pressure is lost. Nevertheless, the following insulation levels
exist at 0 psig where the open breaker will:
1. Withstand a 60-cycle voltage in excess of twice the maximum
line-to-neutral voltage (636 kV)
2. Withstand impulse full waves less than or equal to 1/2 the
rated BIL (900 kV)
3. Withstand switching surges less thanor equal to 2 pu (900kV).
Contamination Tests
Additional testing on the specially-mounted module (Fig.
16) compared the Dog-Bone bushing to conventional sawtooth or sine-wave shed designs normally used on horizontally
oriented porcelains. The concern given to atmospheric contamination stems from our previous studies and results agreeing with Nasser [2]:
The higher the system voltage, the greater is the percentage
of the failures caused by polluted surface flashovers and the lower
is the probability of a flashover due to lightning surges.
Hand cleaning, live washing, and the application of silicone
grease are all expensive, and the use of insulators with a satisfactory performance in polluted atmospheres is certainly the most
economic solution in the long run.6
the
insulators occur
in service when
FlashoversFlasoves
on polluted
opolute inulaors
ccu inservce
henthe
surface conductivity of the insulators exceeds a definite value
which varies with the shape of the insulators. To make it
conductive, the pollution deposit requires a sufficient amount of
moisture which dissolves the solid ion-forming parts of the
pollution layer. Dew, fog, and drizzling rain are the most
frequent causes of the thorough wetting of insulator surfaces
in service.
This was the philosophy underlying our early investigations
which ultimately resulted in the Dog-Bone shed design. To
prove the merits of this configuration, a contamination testing
procedure was developed, based primarily on the following
assumptions:
1. Applying multiple layers of contaminant added not only
considerable time, but also increased the variation between
858
:
4
3C
l|
Fig. 16. Right-hand horizontal Dog-Bone bushing contaminated with Kieselguh r and exposed to light fog (110 kV applied)
one
suddenly.
TO+5
RESULTS
An interrupter module was outfitted with one Dog-Bone
bushing and one bushing whose weather casing sheds were of
the saw-tooth form. Both bushings were of identical length,
but the Dog-Bone design offered 25 percent more creepage.
The module was mounted similarly to the left-hand module in
Fig. 16, and several comparative tests were run to evaluate the
relative dry, wet, and contamination flashover properties.
The following significant points were learned:
1. The 60-cycle dry withstand of the two bushings was identical.
OCTOBER
bushing.
Note in Fig. 16, where the contaminated Dog-Bone
bushing iS shown withstanding 110 kV while fogged, the
effectiveness of its shed design in directing the contaminant
stream around to the underside of the proce]ain.
[11] Russian).
S. C. Killian and J. H. Moran, "Characteristics of EHV station
post insulators," IEEE Trans. on Power Apparatus and Systems,
vol. 83, pp. 280-285, March 1964.
Conclusion
[12] J. B. Owens, "The determination of switching surge withstand
1. It has been shown through analytical logic, through High
voltages for EHV insulation systems," IEEE Trans. on Power
'
and
vol. 83, pp. 263-266, March 1964.
Voltage Laborator
tests
individualinterApparatus
Voltage Laboratory tests on individual
iterrupter
modules,
[13] D. E. AlexanderSystems,
and E. W. Boehne,
"Switching surge insulation
and through complete High Voltage Laboratory verification
level of porcelain insulator strings," IEEE Trans. on Power
to
bemorecapableofwithstanding
re
bumosphing i contamination,
bentaminen
shownntodbe
fof conditions,
nditong
atmospheric
evenemo
under
light fog
than a conventionally designed bushing while also having a
higher wet withstand.
5. Although effective coordination was demonstrated for
these live-tank circuit breakers, the need for parallel protective gaps has been stressed on the basis that there are many
capa
or
Acknowledgments
The authors wish to thank the Westinghouse High Voltage
Laboratory personnel for their cooperation in the accumulation and analysis of the data presented in this paper.
and Systems,
vol. 83, pp. 1145-1157, November and
Apparatus
W. Atwood,
J. F. Wittibschlager, A. R. Hileman,1964.
[14] A.
J. W. Skooglund, Jr.,
"Switchingsurge
tests on simulated EHV towerinsulator systems," presented at the 1964 Western Electronic
Show and Convention, Los Angeles, Calif., August 1964.
[15] American Standard Test Code for High Voltage Air Switches,
ASA C37.34, Proposed additions to cover switching
~~~~~testing of EHV switches, Fourth draft, January 1965. surge
Discussion
A. W. Atwood, Jr. (Southern California Edison Company,
1965
ickle
nd
breaker
859
externally.
Sound engineering practice demands not only that the three
basic tenets of insulation coordination as stated in this paper
are met, but that reasonable margins exist between external
flashover to ground, open gap flashover, and internal flashover.
In testing airbreak switches, the margin between phase to ground
and open gap flashover may be verified by establishing a flashover
level for phase to ground insulation, then deliberately increasing
the insulation to ground and establishing a flashover level across
the open gap. The ratio of minimum open gap flashover to
maximum phase to ground flashover is the margin of safety.
A convenient means sometimes used to increase the phase to
damage?
only in polluted atmospheres, but also in areas where air contamination was not considered of any danger to the insulation
ability under all weather conditions [1].
860
characteristic of
two different insulator types as a
function of the
amount or conductivity of contaminant
g
wVDv,
I-
0>
:
>
v2
vA
B
AMOUNT OF CONTAMINATION OR
SURFACE LAYER CONDUCTIVITY
The tests described in this paper which were carried out under
simulated atmospheric contamination require, however, much
more care if the results and their interpretations are to yield
correct reproducible information.
Important factors that govern results obtained from laboratory
contamination tests include surface layer distribution [2],
method of moistening, the way of applying the test voltage and
the condition of the surface and also the method of interpreting
the results.
The method of applying the contaminant as used by the
authors, although it seems very similar to the actual contamination process, is not capable of producing surface layers comparable with natural ones. Examining naturally contaminated
insulators placed at different locations of different pollution and
atmospheric conditions had shown that the periodic distribution
of surface layer along the insulator length varies between uniformity and two different extremes. After heavy rains the
contaminant is thicker underneath the sheds than on exposed
surfaces and is thicker on these surfaces after long dry periods.
In the average and after mixed weather conditions, the deposited
surface layer is usually very near to being uniform. In laboratory
tests, changing the periodic distribution of a surface layer could
change the performance of an insulator to a great extent [3].
A second point which must be given great care when performing contamination tests is the moistening of the surface.
In practice, this can take place either by dew, which is sometimes
accompanied by fog, or by light rain or drizzle. While dew
produces a more or less uniform moisture layer, rain or drizzle
causes a completely nonuniform unpredictable moisture layer,
which depends on intensity, conductivity, and direction of rain.
In order to keep the amount of variables at a minimum, and
since dew and fog are considered to have the strongest insulation
deteriorating effect, a simulated moistening method was agreed
upon, which produced results correlating with practical experience. Therefore, the overhead fog or spray nozzles used by
the authors do not represent the most severe and common conditions and could yield misleading results.
The third point, which is often overlooked by many researchers
in this area, is the dependability of insulator performance on the
height of test voltage applied. Figure 17 illustrates this fact
for two insulator types, A and B. It is seen that at a voltage
VI, insulator B is superior to A. If the voltage is lowered to
V2, the performance reverses. The test voltage should, therefore,
be equal to the actual service voltage for most realistic results.
I hope that the 150 kV used corresponds to the actual service
voltage. Experience has also shown that measuring flashover
voltages by means of applying a test voltage and increasing it to
flashover is an intorrect proredure.
k oordinate the authors' results with others, I would like
To
to know whether the surface aurrent and the resistivity of the
contaminant were measured and, if so, whieh values were
obtained.
REFERENCES
ETZ-A,ivorl. 84,spp
OCTOBER
REFERENCE
beam).
On the 50-c/s voltage wave, it is evident that there is superpositioii of a surge having the same waveform and the same
polarity of the switchiiig surge applied to the other terminal.
Such an effect is considered a result of the capacitive coupling of
the two terminals.
In the case of our switch, the crest value of the induced surge
voltage was approximately 10 percent of the crest value of the
applied switching surge; the results were accordingly corrected.
In the case described in the paper, rny feeling is that such a
phenomenion may be more pronounced than in our case, since:
1) the indoor laboratory where the tests were performed seems
to allow less clearance (and, therefore, higher coupling) than
those of our tests which were approximately 40 feet from both
the generator and the transformer cascade (see Fig. 19), and 2)
the voltage dividing capacitors, coiinected between the two
terriiinals wlhen the breaker is open, represent an additional
strong coupling.
A second point that I wish to submit to the authors is that
my experience on the behavior of rod-rod gaps, especially with
SONNENBEiIG
AND BRADO:
R. H. Bower (Pennsylvania Power and Light Company, Allentown, Pa.): In this paper, it is indicated that parallel protective
gaps should definitely be installed ahead of the current transformers on any system utilizing these breakers. The authors
also indicate that such gaps should be set to coordinate with the
Iine design factors, as well as within the spectrum of lightning
arrester characteristics, but beneath the circuit breaker impulse
or switching surge requirements. The information regarding
rNSULATION COORDINATION
OF
500-KV SF6
BREAKERS
861
existing standards.
Concern has been expressed about the contamination testing
technique, and we most heartily support the pleas for standards
in this yet undefined area. Dr. Nasser's introductory comments
summarize very well the present state of the art. Some, however, have questioned certain elements in the reported contamination testing procedure. It must be emphasized that the
contamination tests were only intended to compare two porcelains identical in all physical and voltage characteristics except
for shed configuration. They cannot be used to predict the time
to flashover under any field-encountered conditions, nor the
interval between porcelain cleanings. Most importantly, no
guarantee was made or implied that the Dog-Bone design would
withstand all forms and/or degrees of atmospheric contamination.
It was only shown to have a higher probability of withstand than
the conventional saw-tooth design.
Multiple layers of contaminant were tried, but discarded
because the deviation or scatter of the results only increased
without altering the conclusion, making the additional application and drying time unwarranted. Wind direction was not
considered because: 1) what contaminant is added to the windwar fac of a pocli she is sb race fro th lewr
bushing.
was the
Mt
O)CTOBER
effective coordination." The statement had been made following mention of the extremely random nature of the wet switching
surge flashover and pertained only to wet switching surges on
an ungapped breaker.
The High Voltage Laboratory testing on the 1300-kV and
1550-kV BIL breakers demonstrated complete coordination
under impulse, switching surge, and 60-cycle voltages; the latter
two were both wet and dry.
Dr. Carrara's observation of coupling between the switching
surge generator and the 60-cycle transformers is most interesting and appreciated. Regretfully, conformation is impossible
as we were not observing the 60-cycle wave form during the
biased switching surge tests. The effect as shown in Fig. 18
seems to be an almost 1/4 reduction in 60-cycle crest voltage.
Our tests with a line-to-line crest voltage bias (1.73 pu), if
affected, were still adequate to show that a bias at least equal to
1 pu could be tolerated.
Dry switching surge tests were conducted, the breaker was
flashed over, the voltages were in excess of the reported wet
withstand levels, and coordination was demonstrated.
We have never tried a method of demonstrating coordination
of a live-tank breaker similar to that described by Mr. Milton
Cange
tray capacitance
apactanc or lngthning
M. Smth.
and Mr.and
Smith.
Changes in
in stray
lengthening
of column members might be sufficient to upset the normal
voltage division network, if the breaker were elevated.
or
REFERENCES
[11 I. B. Johnson, V. E. Phillips, and H. 0. Simmons, Jr., "Switching
of extra-high-voltage circuits, I-System requirements for circuit
breakers," IEEE Trans. on Power Apparatus and Systems, vol.
pp.Shores,
1187-1196,
1964.V. E. Phillips, "Switching of
R. B.
J. W.December
[2] 83,
Beatty, and
extra-high-voltage circuits, II-Design of air blast circuit
breakers," IEEE Trans. on Power Apparatus and Systems, vol.
83, pp. 1206-1212, December 1964.
Interruption Research
D. R. Kurtz, Member IEEE E. H. Bogert
to circuit breaker design. This has advantages over the approach used by some of the other arc interruption investigations which are based on experimental data using step functions of current or voltage, or which use current amplitudes
significantly less than that seen by the breaker [1], [2]. These
other approaches have provided valuable information for the
particular fundamental
process being studied.
However,
untilac fundamental
theory std evel
d
are comparablealtobreak
actual breaker
example,
therm
ep conditions.
onditiandFordx/dt,
PAS-84, NO. 10
OCTOBER 1965