IEEE Transactions on Power Apparatus and Systems, Vol. PAS-99,No.2 March/April 1980
USA-USSR INVESTIGATION OF 1200-kV TOWER INSULATION
F. S. YoungFellowElectric PowerResearch InstitutePalo Alto, CA
H. M. SchneiderMemberGeneral Electric Co.Pittsfield, MA
Y. M. GutmanDirect CurrentResearch InstituteLeningrad, U.S.S.R.
N. N. TikhodeyevDirect CurrentResearch InstituteLeningrad, U.S.S.R.
ABSTRACT
Results of the joint USA/USSR investigations on
1200-kV tower air Faps and insulator strings are re-ported for a comprenensive range of times to crest ofapplied switching impulses extending from about 100microseconds to 6 milliseconds. Tests were performedin the USA with a conventional impulse generator andin the USSR with cascade transformers. As a result ofthese investigations, data are available for the firsttime on similar test objects stressed by two differenttypes of voltage waveforms. A technique for comparingthe results based on the concept of the "active part"of the waveshape is described and applied to the data.
INTRODUCTION
In the last decade, a number of major high voltagelaboratories in the world have engaged in research onclearances and insulation subjected to impulses whichsimulate switching overvoltages on EHV and UHV trans-mission lines. Tests have been made in the USA at thehighest voltages in outdoor areas with impulse genera-tors producinR aveshapes with times to crest from 1.5to 1000 ps(1, *3T and in the USSR with cascaded testtransformers generating waveshapes with times to crestfrom 3 to 5 milliseconds(4v596). However, only a fewof these tests were performed with full-scale 1200-kVmockups and with times to crest exceeding 500 ps. Thepreviously published results show a substantial dis-agreement among laboratories, although test conditionsappear to be comparable(7).
In view of the contemplated construction of1200-kV ac overhead transmission lines in the USA andthe USSR, consideration must be given to the very widerange of switching overvoltages likely to occur duringoperation of these UHV lines. Consequently, a new setof volt-time characteristics for 1200-kV tower air gapsand insulator strings extending from 100 us to severalmilliseconds was required for insulation coordination.A unique opportunity to obtain such data was providedby a joint research project in the framework of the1974 USA-USSR Agreement on Scientific and TechnicalCooperation in the Field of Energy(8). At the begin-ning of this cooperative effort, neither country couldimplement a program to generate data for a comprehen-sive range of times to crest because of the unavaila-bility of a full set of test equipment.
F 79 283-3 A paper recommended and approved by theIEEE Transmission and Distribution Commmittee of the IEEEPower Engineering Society for presentation at the IEEE PESWinter Meeting, New York, NY February 4-9, 1979.Manuscript submitted August K9, 1978; made available forprinting December 13, 1978.
In the USA, the work on this joint project wassponsored by the Electric Power Research Institute(EPRI) and carried out by the General Electric Company(GE) in Pittsfield, MA. The USA part of the effortcovered the range of times to crest from 100 to 1200ps,obtained with an outdoor impulse generator. The USSRpart of the effort involving longer times to crest wasperformed at the High Voltage Technology Laboratory ofthe Direct Current Research Institute (NIIPT) in Lenin-grad, USSR, using an outdoor test transformer cascade.Both high voltage laboratories employed similar testobjects. An important part of the program describedhere was the exchange of investigators between the USAand USSR during the period when tests were performed.
TEST OBJECTS AND TEST PROCEDURES
Test objects for the cooperative program were se-
lected taking into account the proposed designs of thefirst 1200-kV lines in the USSR(499) and the USA(10).During establishment of the program for the joint in-vestigations, attention was focused on assuring identi-cal test conditions. The main considerations in thisrespect were the height of the tower and the conductorabove ground, the geometry of the tower, the cross-
sections of the tower elements, and the design of thebundle conductor mockup. It was decided that the jointprogram tests would be made in both countries withtower mockups having the same dimensional characteris-tics as that shown in Fig. 1. This common model repre-sents actual 1200-kV towers which might be used in
Fig. 1. 1200-kV Tower window model used as the basisfor tests in both USA and USSR.
either country. In the USSR, the specially designedtower illustrated in Fig.2 was built for test purposes.In the USA, the tower window mockup shown in Fig. 3 wasdesigned and suspended between the columns of an exist-ing test portal described previously(l).
Outer phase vertical strings were suspended fromthe lateral truss extensions of the test portal (USA)as also shown in Fig. 3, and from the boom of a mobilecrane (USSR) of the same dimensions acting as a towermockup as illustrated in Fig. 4. The height to thetruss at the initial stage of the investigations wasselected to be 30.5 m in accordance with the height ofthe existing tower truss in the USA.
The bundle conductor mockups simulated a 1200-kVline configuration with 8 subconductors arranged overa 0.75 m radius. The hardware securing the bundle con-ductors to the insulator strings had no grading rings.The insulators closest to the bundle were approximatelyin line with the two adjacent subconductors.
Insulator strings were tested at a number ofconductor-to-truss distances ranging from 5 to 8 m. Inthe USA, outer phase vertical strings 5.5 m long weresuspended at a distance of 8 m from the tower leg. Inthe USSR, outer phase vertical strings 4.6 m long weresuspended at a distance of 10 m from the tower leg. Vstrings arranged in the center of the tower window were6.4 m (USA) and 8 m (USSR) long. The spacing betweenthe conductor and the truss in both countries was ad-justed by varying the length of a metal support cableto the truss, the number of units in the strings beingkept constant. To guarantee similar test conditions
in both laboratories, insulators with practicallyidentical values of the H/D ratio were selected.
Fig. 2. 1200-kV Tower mockup for V-string tests atNIIPT laboratory in Leningrad, U.S.S.R.
463Minor differences in the test objects as shown in
Table 1 did not, in the authors' opinion, affect thetest results.
In the USSR, positive polarity switching impulsesoscillatory (1-cosine) in shape and with times to crestof 2200, 4600 and 6200 Ps were obtained by dischargingcapacitor banks into the low voll ge windings of a3 x 600 /2-kV transformer cascade(3 1. The USA testswith insulator strings were made with aperiodic(double exponential)waveshapes of about 100 x 3500 ps,225 x 4000 ps, 500 x 3800 ps, and 1200 x 4800 ps pro-duced by a 6000-kV, 300-kJ impulse generator(l).
Fig. 3. 1200-kV Tower mockup for V-string tests incenter and I-strings in foreground at GE laboratory inPittsfield, MA.
Fig. 4. 1200-kV Tower mockup with crane for I-stringtests at NIIPT laboratory in Leningrad, U.S.S.R.
464
TABLE 1
CHARACTERISTICS OF TEST OBJECTS
Test ObJect Component USSR USA
Lattice sections of1200-kV tower mockup 1.4 x 1.4 m 1.2 x 1.2 m
Diameter of subconductors 2.4 cm 3.1 cm
Length of bundle mockup 60 m 30 m
Insulator type II C-12A, glass Porcelain
H = Spacing H = 138 mm H = 146 mmD = Diameter D = 260 mm D - 254 mmL Leakage distance L = 325 mm L = 280 mm
String type Single Twin, with 45 cmspacing
The test and voltage measurement procedures weresimilar in both countries. Both laboratories followeda procedure which allowed the relationship betweenflashover probability and impulse crest voltage to bedetermined. The method consisted of applying 20 vol-tage impulses with an identical crest value at each of4 to 8 equally spaced voltage levels. The resultingvalues of the relative frequency of flashovers wereused to calculate the parameters of an assumec Gaussiandistribution. The two reported parameters are U50, theimpulse crest voltage corresponding to a 50% flashoverprobability, and O/U50 the standard deviation in percent of the 50% flashover voltgge.
CALIBRATION TESTS WITH ROD-PLANE AIR GAPS
Calibration tests were performed in order to ob-tain a uniformity of the test and measurement proce-dures for flashover experiments in both laboratoriesby using rod-plane air gaps 6 and 10 m long.
Figure 5 gives volt-time characteristics for the6 and 10 m rod-plane air gaps as a function of thefull time to crest (TC) of the applied switching im-pulse. For the 6 m rod-plane air gap there is a dis-tinct minimum of the volt-time curve occurring for aswitching impulse time to crest of about 200 Us. Forthe 10 m gap, the minimum of the volt-time curve isless distinct. The volt-time characteristic for the10 m air gap is almost independent of the steepness ofthe wavefront for the range of times to crest employed.This explains why the difference between the shapes ofaperiodic and oscillatory (1-cosine) impulses for thelOm gap do not affect the values of flashover voltages.Results obtained in the two laboratories agree quitewell and permit a smooth curve to be drawn. A 10%difference in the results of two USA test series forthe same waveshape appears to be due to the effect ofabsolute humidity, similar to that reported recent-
The discrepancy between flashover voltages ob-tained in the two laboratories for the 6 m gaps withimpulses having Tc of 1200 to 2000 ps may be accounted.for by the steeper front of oscillatory impulses com-pared to that of double exponential waveshapes withthe same full time to crest. This difference inflashover voltages obtained with aperiodic and oscilla-tory impulses has already been noted with a 2m rod-plane air gap for a tc of 1000 ps(14).
TECHNIQUE FOR COMPARING DATA OBTAINED WITH APERIODICAND OSCILLATORY WAVESHAPES
Better agreement of the data obtained in the twolaboratories for different waveshapes is achieved byusing a criterion which takes into account differencesin the shapes of the impulses. This criterion is basedon the analysis of the time interval defined as the
"active part" of the impulse front, i.e., the time dur-ing which a flashover is likely to occur. Such an
approach is valid when a flashover takes place beforethe crest of a switching impulse. This is known tooccur in 6 to 10 m rod-plane air gaps with positivepolarity impulses whose times to crest exceed 250 vs.Tests made in several laboratories show that the mini-mum voltage at which a flashover is still possible isapproximately equal to U50 minus 3a. For instance dur-ing the USSR tests, the minimum actual flashover vol-tages were 0.83 U50 and 0.78 U50 for impulses withtimes to crest of 2000 and 5000 us,respectively.For the
Ps
Fig. 5. Fifty per cent flashover voltage (U5 ) as a
function of actual time to crest (Tc) for rod-plane airgap calibration tests. Data are uncorrected forweather. Solid points - USA (aperiodic impulses),Hollow points - USSR (oscillatory impulses) curve (1) -
10 m gap. Curve (2) - 6 m gap.
465
majority of waveshapes of practical importance, allpossible values of 0/U50 lie between 5 and 10%. Thusthe "active part" of the front of the waveshape cannotextend below the 70% point of the impulse crest voltagecorresponding to a 50% flashover probability. This de-fines the "active part" of the waveshape as the portionof the wavefront between 0.7 U50 and U50, with a cor-responding time interval T 7. This approach for asses-sing the shape of surges on transmission lines andtest impulses for laboratory experiments has beensuggested in the past(104). The advantages of thisapproach are (1) it permits the determination of appro-ximate equivalents of different waveshapes; (2) itagrees better with concepts of the flashover mecha-nisms; and (3) it ignores legitimately that part ofthe waveshape (below 0.7 U50) which is of little impor-tance for development of flashover.
A comparison of oscillatory and aperiodic wave-shapes is shown in Fig. 6, where switching impulses ofthese two types have been plotted with their crest vol-tages coincident. Although the shapes of these wave-forms differ considerably at low voltage levels, the"active parts" of the impulses are similar. It wouldthus be expected that if aperiodic and oscillatory im-pulses had the same value of TO 7, their flashover vol-tages would be almost identical.
It 0° 77-! (oscillatory)I (
- lro .7 -i (aperiodic)
Fig. 6. Example showing definition offor aperiodic and oscillatory impulses.
kVr
2200 F
1800
14 sAA
10002
10 inGap11 1|11 118_1----- If r X X-_ E
-I Gap
103
"active time"
Assessment of the flashover voltages for differentwaveshapes on the basis of the "active part"of the wavecan considerably improve agreement of the volt-timecharacteristics. For example, if the data for the 6 mrod-plane gap are plotted as a function of TO.7 insteadof Tc, the results would appear as shown in Fig. 7.The difference in the data for similar waveshapesin both laboratories plotted on this basis doesnot exceed 3% on the average. It should be noted thatthe 6 m gap tests carried out during winter under simi-lar weather conditions (low absolute humidity and lowtemperature) show an even better agreement. With theaid of the concept of the "active part"of the wave, thecalibration tests with rod-plane air gaps demonstrate auniformity of the test and measurement procedures re-quired for further tests.
FLASHOVER VOLTAGES OF DRY IAND V-STRINGS
Primary consideration in the joint US/USSR inves-tigation involved producing the electrical strength of1200-kV tower gaps for the entire range of waveshapescharacteristic of switching overvoltages on 1200-kVoverhead lines. The concept of the "active part" of thewaveshape was used to compare results obtained on simi-lar tower mockups with aperiodic and oscillatory impul-ses.
All data obtained on I and V-strings are presentedin Table 2, where uncorrected flashover voltages andweather conditions are reported. For V-strings, therelationship between flashover voltage and full time tocrest (Tc)is given in Fig. 8, whereas the same data are
plotted as a function of the "active" time (lo07) inFig. 9. Note that as in the case of rod-plane gaps,
2200 -
1800 - L L
1400-
10 102 103 104 PSFig. 8. Fifty per cent flashover voltage (U50) as afunction of time to crest (Tc) for V-strings. Conduc-tor to truss air gaps as follows: curve (1) - 5m,(2) - 6 m, (3) - 7 m. Solid points - USA, Hollowpoints - USSR.
kV
104 PS
Fig. 7. Data of Fig. 5 giving fifty per cent flash-over voltage (U50) as a function of "active time"(TO 7) - Solid points - USA, Hollow points - USSR.
2200 -
1800
1400-
vvv_
10 102 103 104 PS/Fig. 9. Data of Fig.8 plotted as a function of "activetime" (To 7).
t_
I
466
TABLE 2
COMPLETE DATA FOR DRY FLASHOVER TESTS ON I AND V-STRINGS TESTEDWITH POSITIVE POLARITY APERIODIC AND OSCILLATORY SWITCHING IMPULSES
Fifty per cent StandardTime to Conductor* Flashover Deviation Barometric Absolute
String Crest Truss Air Voltage, U50 a/U50 Pressure Temperature HumidityType TCG(PS) Gap (m) (kV) (%) (mm Hg) (OC) (g/m3)
* For I strings, conductor-to-tower by clearances were 8 m (USA) and 10 m (USSR)the use of the concept of the "active part" of the wavefor tower V-strings in Fig. 9 yields a useful volt-timecharacteristic which extends over a wide range of im-pulse shapes. Minima of the volt-time curves werefound at full times to crest of about 200 us or less,decreasing with gap spacing. Such a variation of theminimum as a function of time to crest has been ob-served previously for simple electrode configurationssuch as rod-plane gaps(1) and tower windows(3). As thetime to crest increases from 200 Us at the minimum ofthe, volt-time curve to 6000 Vs, the electrical strengthof the air gaps increases by about 20%.
The volt-time characteristics of vertical stringsin terms of Tc and To.7 are given in Figs. 10 and 11respectively. These curves are similar in shape tothose given in Figs. 8 and 9 for V-strings. Minima of
the curves lie between 100 and 200 vs. The flashovervoltage of a 6 m vertical string increased by about 25%as the time to crest increased from 250 to 4600 Vs.Long impulses were not used to test strings of morethan 6 m, since the voltages required for such testswere beyond the capability of the test transformer cas-cade.
In comparing flashover voltages for I and V-strings in dry conditions, it may be seen from Table 2that in the case of aperiodic waveshapes with Tc be-tween 100 and 500 us, there is no statistically signi-ficant difference in flashover voltages for the sameair clearance between conductor and upper truss. How-ever, for oscillatory waveshapes with Tc between 2200and 6200 Us, the I-strings have higher strength thanV-strings by an average of 5% when comparing equalconductor-truss clearances.
467
kV
10 102 103
Fig. 10. Fifty per cent flashoverfunction of time to crest (xc)I-strings. Conductor to truss -
follows: curve (1) - 6 m, (2) - 7 Ipoints - USA, Hollow points - USSR.
kV
2600
2200
1800
1400
nl
voltage (U50)for outer
air gaps a
m, (3) - 8 m.
32.
TO.7
I I I11 1 1 vI II II I II
102 103Fig.-ll. Data of Fig. 10 plotted"active time" (T ).
0.7
104as a functi
PRECIPITATION EFFECT TESTS
It is well known that the negative polarityover voltage of dry air gaps with insulator sis higher than that of the same gaps tested withtive polarity(l). In order to evaluate the effprecipitation, a few tests were made in the USboth polarities on I and V-strings with an artirain spray system and in natural rain and snow.artificial precipitation tests were conducted wisame test objects and test procedures used fortests. However, when natural precipitation occan up-and-down test method was used. The rainsystem was located on the tower truss and adjustdownward spraying of the gaps. The rain intensitadjusted to a rate of approximately 4 mm/minute iconditions. Care was taken to report results official rain tests only when the water spray envthe insulators, since during windy periods the wwas much less uniform. In addition, some testmade with natural precipitation. Experimental dareported in Table 3, where they are also comparedry flashover test results.
strings in a similar way as in the dry flashover tests.
For V-strings, the positive polarity impulses inwet conditions gave rise to the lowest flashover vol-tages. Tests with negative polarity in wet conditionsalways resulted in a higher flashover voltage than withpositive polarity in dry conditions. The maximum de-crease of the electrical strength of the V-string towerwindow gap in artificial rain was 5% of the dry flash-over voltage.
For I-strings tested in artificial rain, theflashover voltages for positive polarity were alwaysless than those obtained with positive polarity in dryconditions. The maximum reduction of the flashover
104 PS voltage in positive polarity artificial rain tests was5% of the dry flashover voltage.Negative polarity testsin artificial rain sometimes gave lower flashover vol-
as a tages than either positive polarity dry or wet. As mayphase be seen in Table 3, the maximum reduction in electrical
are as strength for I-strings in negative polarity artificialSolid rain tests was 13% of the positive polarity dry value.
It is possible that for the same conductor-truss clear-ance but with more insulators and a shorter metalcable, the wet flashover voltages would be closer tothe dry flashover values.
There was no significant difference between wetand dry flashover voltages for either I or V-stringsunder the natural precipitation conditions of rain andsnow. It should be noted that the above observationson precipitation effects are based on a very limitednumber of tests.
CONCLUSIONS
1. Volt-time characteristics were obtained for thefirst time over a wide range of times to crest (100 Vsto 6 ms) for 6 m and 10 m rod-plane air gaps and for1200-kV tower mockups with I and V insulator strings.These new data have substantially increased the availa-ble information which is required for the insulationcoordination of 1200-kV tower designs using techniquessuch as those proposed in the USSRR15) and in the
PS USA(2).Lon of
2. The notion of an "active part" of the waveshape wasdeveloped and justified experimentally. This permits amore accurate comparison of flashover voltages obtainedfrom impulse generators and transformer cascades. Sucha comparison is particularly important for studying the
flash- range of times to crest necessary to simulate switchingtrings overvoltages on 1200-kV power transmission lines.posi- 3. Based on a very limited number of tests dur-
ect of ing precipitation, a maximum reduction equal to 13% of
ficial positive polarity dry flashover voltage occurred for I-
The strings tested with negative polarity in artificialththe rain.dththedry ACKNOWLEDGENENT
urred,,sprayed fory wasn calmarti-
eloped,ettings wereta ared with
For positive polarity, the time to crest of theimpulses used in the precipitation tests affected theelectrical strength of air clearances and insulator
The work covered on the USA portion of these testsis part of the Project UHV research program sponsoredby the Electric Power Research Institute.
REFERENCES
1. EPRI Transmission Line Reference Book/345 kV andAbove, Electric Power Research Institute, 1975.
2. IEEE Working Group Report,"Guide for Application ofInsulators to Withstand Switching Surges", IEEE PAS-94,pp. 58-67, 1975.
3. W. C. Pokorny, R. W. Flugum, "UHV Tower InsulationParameters Determined by Full-Scale Testing", IEEEPAS-94, pp. 518-529, 1975.
.2200 11 III -.&- 11ou-
r
3 ol-.2 -o
A1800 d -
1Tc
1400 1 I I I I I I III I I I I I III
468
TABLE 3
COMPARISON OF DRY AND PRECIPITATION TESTS ON I AND V-STRINGS
4. K. P. Kryukov, S. C. Merkhalev, N. N. Tikhodeyev,"The Electrical Strength of Large Air Gaps and Se-lection of Insulation Clearances for Overhead Lines".Paper presented at USA-USSR AC Power Transmission Sym-posium, Washington, 1975.
5. G. N. Aleksandrov, Y. M. Gutman, V. L. Ivanov,V. E. Kiesewetter, A. S. Maikopar, S. D. Merkhalev,A. A. Philippov, V. S. Rashkes, N. N. Tikhodeyev,"Di-electric Strength of Line Insulation", CIGRE Report No.417, 1966.
6. G. N. Aleksandrov, V. P. Redkov, Y. I. Lyskov,"Electrical Strength of Air Gaps between Conductor andTower at Switching Surges", Elektrichestvo, No. 5.1972.7. I. V. Fateyeva, A. N. Sherentsis, "Selection of In-sulation Clearances for UHV Power Transmission Lines",Elektrichestvo, No. 1, 1978.
8. F. S. Young and R. S. Gens, "UHV Transmission Re-search in the USSR", EPRI Journal, June 1976, pp.24-27.
9. V. V. Bourgsdorf et al., "Design of the UHV 1150-kVAC Transmission Line", CIGRE Report No. 31-03, 1976.
10. S. A. Annestrand, G. A. Parks, "Bonneville PowerAdministration's Prototype 1100/1200 kV TransmissionLine Project", IEEE PAS-96, pp. 357-366, 1977.
11. G. Elstner, S. Franke, W. Zimmermann, H. Bachmann,G. E. Krastin, A. G. Levit, A. A. Malygin, N. N. Tikho-deyev, "Trends for the Development in the Field of In-sulation Tests of UHV Transmission Lines and Equipmentsas well as Testing Plants for this Purpose, WELC ReportNo. 2014, Moscow, 1977.
12. A. A. Philippov, "Methods of Obtaining VoltageImpulses of up to 2 MV Crest Value with Waveshapes Cor-responding to Internal Overvoltages in AC and DC
Lines", Izvestiya NIIPT, No. 8, 1961.
13. W. Busch, "Air Humidity: An Important Factor for
UHV Design", Paper F78039-9, presented at the IEEE PAS
Winter Meeting, New York, N. Y. January 29- February 3,1978.
14. A. Colombo, G. Sartorio, A. Taschini, "Phase-to-Phase Clearances in EHV Substations as Required bySwitching Surges", CIGRE Report No. 33-11, 1972.
15. D. E. Artemyev, N. N. Tikhodeyev, S. S. Shur, "Co-ordination of Insulation of Power Transmission Lines",Energiya Publishers, Moscow, 1966.
Frank S. Young was born in Salt Lake City,Utah. He received a BSEE degree from StanfordUniversity in 1955 and an MSEE degree fromthe University of Pittsburgh in 1962.Mr. Young joined the Electric Power
Research Institute in 1975 as Managerof the AC& DC Overhead Lines Research Program. In
this capacity he was responsible for planning andimplementing research and development pro-jects to improve the electrical and mechanicalcapabilities of overhead transmission lines. In-
cluded in his duties were supervision of Project UHV where research ontransmission up to the 1500 kV level is being conducted. In 1977, hebecome Manager of Program Planning. He is responsible for technicalplanning and budgeting for EPRI as a whole and participates in the
development of the EPRI Overview and Strategy. Before coming toEPRI, Mr. Young was with the Westinghouse Electric Corporation inEast Pittsburgh, Pennsylvania. He joined Westinghouse in 1955 on theGraduate Student Course. In 1963 he became Sponsor Engineer for the
Northeastern Zone of the company, working on a wide variety of utilityproblems involving planning, design and operation of electric utility
469systems. In 1966, he was appointed Manager of the Waltz MillUnderground Transmission Test Facility and was responsible fordesign, construction, and operating of this 1100 kV research station.The Waltz Mill Station is the focal point of industry research onunderground transmission systems.In 1972, he become Manager ofUltra High Voltage research for Westinghouse with assignment todesign other special purpose test facilities , and market UHV researchefforts for Westinghouse.Mr. Young serves on the U.S. - U.S.S.R. Joint Commission on Scien-
tific and Technical Cooperation as a member of the project groups onUHV Transmission Technology and HVDC Transmission System Ex-periment Design. He has made four trips to the Soviet Union in thisassignment. He is also active in IEEE work, and is a member of theIEEE Ethics Task Force and Transmission and Distribution Commit-tee. He is a fellow of the Institute. In addition, he is a member ofCIGRE, Tau Beta Pi and a Registered Professional Engineer.
Yully M. Gutman was born in Leningrad, USSR on July 10, 1934. Hereceived the Electrical Engineer and the Candidate of TechnicalSciences degrees in 1957 and 1967, respectively, from the LeningradPolytechnical Institute.
After graduation in 1957 he worked in the power apparatus industry.Since 1961 he has been with High Voltage Technology Laboratory ofDirect Current Research Institute, Leningrad. His primary respon-sibilities are research on dielectric strength of air gaps and line insula-tion and its applications to line and substation insulation design. Heparticapated in a number of projects related to EHV and UHV AC andDC lines and substations.Mr. Gutman has published some 30 technical papers.
Nikolay N. Tlkhodeyev was born in the USSR on December 7, 1927. Hereceived the Electrical Engineer, the Candidate of Technical Sciences,and the Doctor of Technical Sciences degrees in 1952, 1955, and 1966,respectively, from the Leningrad Polytecnical Institute; at present helectures at the LPI as Professor. In 1979 he was elected CorrespondingMember of the USSR Academy of Sciences. Within CIGRE he is theUSSR representative in Study Committee 41 "Future of Power Systemsand Power Transmission Lines."
Since 1955 Dr. Tikhodeyev has been with High Voltage TechnologyLaboratory of Direct Current Research Institute, Leningrad. He hasdirected the Laboratory since 1958. He does research in many fields ofhigh voltage engineering, but the area of his major interest is selectionand coordination of insulation for EHV and UHV AC power transmis-sion lines.
Dr. Tikhodeyev is the author of 5 books and some 130 technicalpapers.
Discussion
C. L. Nellis and R. L. Brown (Brownville Power Administration, Van-couver, Washington): The authors have presented interesting new dataon full-scale 1200 kV mockup towers for a very wide range of switchingimpulse waveforms. The data with wavefronts between 1000 and 6000 ,uis of particular interest. BPA has performed full-scale tests on 1100 kVmockup towers with times to crest ranging from 100 to 900As. For theBPA tests, standard deviation levels with wavefront times less than 400ps average 4-50%o, but for wavefronts from 400-900 As 5-6% levels wereobtained. The authors data has average standard deviations of 4.53%for wavefronts of 255 ps or less and 7.070o for longer wavefronts. Dothe authors have an explanation for this increase in standard deviationlevels with time to crest?The authors state that "as the time to crest increase from 200 Ms at the
minimum volt-time curve to 6000 lAs, the electrical strength of the airgap increased by about 20%". We feel that this statement is somewhatmisleading. We agree that the CFO level is increased by about 20%, butthe voltage levels two or three standard deviations below CFO are not.Take for instance the 5 m air gap data with CFO's of 1630 and 1930 kVfor the time to crests of 110 and 6200 ps respectively. If the average stan-dard deviations of 4.53 and 7.04%are used, then the U-3s withstandlevels are 1409 and 1522 kV respectively. The 6200 ;As wavefront withs-tand level for these standard deviations is only about 8% greater thanthe 110 ps wavefront, not 18% as the CFO levels indicate. And if the ac-tual standard deviations obtained for these tests are used (4.3 and10.3%), then these levels are 1420 and 1334 kV respectively. For thiscase the longer wavefront had a U-3s level that is actually 6% lower.
Since transmission lines are designed to operate with insulation levels2 or 3 standard deviations below CFO, it is important to know what
470
standard deviation value to use. For EHV line designs BPA uses 5% forthe standard deviation level. What do the author feel this level shouldbe for UHV lines?Do the authors have an explanation for the test results that had stan-
dard deviation greater than 10%? Was the weather changing rapidlyand how much time was involved in performing the test? Do the authorsfeel the 10%o levels are valid or is it their opinion that for someunknown reason the dielectric strength of air during these tests did notfit a Gaussian distribution and therefore the calculated standard devia-tion should not be used?
During BPA's tests it was discovered that grading has a large in-fluence on whether the switching impulse flashover arc follows the in-sulator string or whether it takes the air gap. During the authors testswere any attempts made to study effects of grading? Was the flashoverarc path pattern of the single "relatively flat" Vee string different fromthe double "relatively deep" Vee string?
Manuscript received January 2, 1979.
F. S. Young, H. M. Schneider, Y. Gutman, and N. N. Tikhodeyev: Theauthors wish to thank the discussors for their interest in the subject andfor stressing the importance of these data for future transmission linedesign.With respect to the first question on the apparent increase in standard
deviation with time-to-crest of the applied voltage, it is helpful to ex-amine the curves shown in Fig. 1 of this closure, where standards devia-tions obtained for each test in Table 2 of the paper are plotted for I- andV-strings. Also given in Fig. 1 are mean values for these two configura-tions. For the I-strings, the average standard deviation increases from4.5% at 90 lAs to 6.5% for a 500 gAs time-to-crest and then decreases to4.5-5% for the longer wavefronts. For V-strings, the mean standarddeviation increases from 4.5% at 110 jAs to almost 9% at 2200 As andthen decreases to 6.5% for the 6200 lAs time to crest. The general trendof increasing standard deviation for longer wavefronts has been observ-ed in the past, according to references (5) and (6) of the paper.However, it may be seen in Fig. 1 that the range or spread of standarddeviation values about the mean also increases dramatically forV-strings at the longer times to crest. Since only three or four points areavailable at each time to crest, the mean value is of more significancethan the individual data points.The authors do not agree with the discussors' point concerning a
"misleading statement" about increasing electrical strength with timeto crest, since the authors' comments refer to the mean breakdownstrength U. The fact that the "withstand" (U-3o) levels follow a dif-ferent variation with time-to-crest if the standard deviation varies with
time-to-crest as noted by the discussors has been noted previously [11].For the design ofUHV lines, it appears that if overvoltages with long
times to crest are anticipated, standard deviations which are ap-
propriate for these longer times-to-crest should be taken into account.However, considering the large amount of supportive information inestablishing the 5% level for shorter times-to-crest, it would be wor-
thwhile to expand the data base on standard deviations for the longertimes-to-crest before utilizing data from such a small statistical samplefor design.
In regard to changes in weather influencing the standard deviation bychanging the probability distribution shape during the course of a test,the authors do not believe that this was the case. The duration of thetest was at most two hours, and those cases in which weather did changeradically were not included in the report. Consequently it is felt that theresults are valid as given in the paper.
Grading effects were not investigated since the original protocol forthe test program stipulated that grading rings were not to be used.The flashover patterns observed in the U.S.A. and the U.S.S.R. were
essentially the same.
O'/u'01%)
4
glo 10 10TIMETOCREST
Fig. I - Standard Deviation of I- and V-string Flashover Voltages as aFunction of Time to Crest of the Applied Voltage.
Reference
[1]. C . Memenlis and G. Harbec, "Coefficient of Variation of thePositive-Impulse of Long-Air Gaps", IEEE PES Tran. Power andSystems, Vol. PAS-93, May/June 1974, pp. 916-927.
