Publication

7i-2
snu.78

c 0 i ri l Ir s s | 0 l {E t r c T R 0 T E icl H
| 0u E I 1 { E
TR}tAT|01{ALE
(affili6e
i l'0rganisation
Internationale - lS0)
deNormalisation
. ] I I | R M ED EL [ CEI

I I I T ER 1 { A TIIIl A
l LE L E C T R ( l T E C H I {C
I Cl lM
AMt ISSIl|l{
(affiliated
to theInternational
0rganization - 180)
for Standardization
I EG ST A IIIl A R I|

Publication
7l-2
. Deuxidme - Second
6dition edition
rF
1976'

deI'isolemenl
Coordination
partie:Guide
0euxi0me d'applicalion

Insulation
co-ordination
Parl2: Application
Guide

0roitsdereproduction - Copyright
rdserv6s - all rightsreserved

Bureau
Central
delaCommission
Eiectrotechnique
Internationale
1,ruedeVarembd
0endve,
S'risse
R6vision de Ia pr6sente publication Revision of this publication

Le contenu technique des publications de la CEI est cons- The technical content of IEC publrcations is kept under
tamment revu par la Commission afin d'assurer qu'il refldte constant revierr b) th.- I E C. thus ensuring that the content
bien l'6tat actuel de la technique. reflects current technolt'g1.

Les renseignements relatifs i ce travail de rdvision, d l'6ta- Information on the sork rrf rc'rtsitrfl. the issue of rerised
blissement des 6ditions r€visees et aux mises ir jour peuvent editions and amendn^*nt shc'eti !rla) bc obtained from IEC
€tre obtenus auprds des Cornit6s nationaux de la CEI et en National Committcc: anJ :rom 1l'1.'fq-rllqruingIEC sources:
consultant les documents ci-dessous:

O Bulletin de la CEI O IEC Bulletin

O Rapport d'activit6 de la CEI O Report on I E C Actiritiex

Publi6 annuellement Published 1'earl1'

O Catalogue des publications de Ia CEI O Catalogue of IEC Publicatioos

Publi6 annuellement Published 1'earll'

Terminologie utilis6e dans la pr6sentepublication Terminolog)' usd in this publication

Seuls sont d6finis ici les termes speciaux se rapportant d la Onll' special terms required ft'rr thc purposc of this publication
pr6sente publication. are defined herein.

En ce qui concerne la terminologie generale, le lecteur se For general terminologl', readers are referred to IEC Publi-
reportera d la Publication 50 de la CEI: Vocabulaire Electro- cation 50: International Electrotechnical Vocabulary (I.E.V.),
technique International (V.E.L), qui est dtablie sous forme de which is issued in the form of separate chapters each dealing
chapitres sdpar6straitant chacun d'un sujet ddfini, l'Index g6n6- with a specific field, the General Index being published as a
ral etant publi6 s6parement. Des ddtails complets sur le V.E.L separate booklet. Full details of the I.E.V. will be supplied
peuvent €tre obtenus sur demande. on request.

Symboles graphiques et litt6raux Graphical and letter symbols

Seuls les symboles graphiquc-set litteraux spdciaux sont Only special graphical and letter symbols are included in
inclus dans la prdsentepublication. this publication.

Le recueil complet des symbolesgraphiquesapprouv6spar The complete seriesof graphical symbols approved by the
la CEI fait I'objetde la PublicationI l7 de la CEI. IEC is givenin IEC Publicationl17.

Les symboleslittdraux et autres signes approuvds par la Letter symbols and other signs approved by the IEC are
CEI font I'objetde la Publication27 de la CEL containedin IEC Publication27.

Autre publication de la CEI 6tablie par le m6me Other IEC publication prepared by the same
Comit6 d'Etudes Technical Committee

Publicationno''71-l: Coordinationde l'isc-rlement.

Premiere Publication No. 7l-l: Insulation co-ordination, Part l:
partie: Termes,d€finitions,principeset rdgles. Terms, definitions,principlesand rules.

fi
 621.316.93.015.33::621.319.51.004.14(083.85)
C.D.U./U.D.C.:621.315.61.01s.008.03

c 0 i lMt I s s | 0 1E{t r c T R 0 T E icl rH0 uE I l r TE R 1 { A T | 0 1 { A t E

(affilide
d l'Organisation - 180)
deNormalisation
Internationale
I I I I R MlEl Et A CEI

I i l T E R i I A T I t l I IEALLE C T R ( l T E C H N ICCt Al M
LMISSI(|l{
(affiliated
to thelnternational
Organization - lS0)
for Standardization
I E CS T A } I I l A R l l

7l-2
Publication
- Second
6dition
Deuxidme edition
1976
9PuN - PB.

deI'isolement
Coordination
partie:Guide
l|euxiOme d'appllcalion

ctl-tlrdination
lnsulation
Parl2: Applicalion
Guide

Pi:'DIfiT1KAAN
P U S T T P E T I Y E I I OII\ i I M A S A I . A HI ( E L I S T F I K A ^ !

9245
s DF;1e?7

dereproduction
Droits rdservds - all rightsreserved
- Copyright

Aucuneparliede cettepublication
ne psut6trereproduite sous No part ol this publication
ni utilisGe may be reproducedor utilizadin any
quelqueformeque ce soit et par aucunproc0d6, ou mdca- formor by anymeans,
6lectronique electronic
or mechanical, photocopying
including
y compris
niqua, la photocopie
et lesmicrofilms, withoutpermission
€critdel'6diteur. and microfilm,
sansI'accord in writinglrom the publisher.

Bureau de la Commission
Central Electrotechnique
Internationale
1, rue de Varembd

6enive,Suise

;L,,: lm.-
 - 3 -

CONTENTS

Page
Fonrwonp 5
Pnnrlce 5

SscrroN ONs - ScopB AND REFERENcETo EXrsrrNG IEC pusucATroNs

Clause
1. Scope 7
2. Reference to existing I EC publications 7

SrcnoN Two - Vorrace srRESsEs rN sERVrcE

3. General . 9
4. Power-frequencyvoltages 1l
5. Temporary overvoltages ll
6. Switching and lightning overvoltages l5
7. Determination of expectedovervoltagelevels l9

SncnoN TnnBn - INsur.q,TroNwrrHsrAND

8. General . . 2 3
9. Insulation behaviour at power-frequencyvoltagesand temporary overvoltages . 2 5
10. Probability of disruptivedischargeof insulation under impulsevoltages . 2 5
ll. Apparatus with windings . 3 1

SncrroN Foun - PnorrcrrvE DEVrcEs

12. General . 3l
13. Non-linear resistor-typesurge arresters 31
14. Expulsion-type surge afresters 33
15. Spark gaps 35
16. Application of different protective devices 37

SecnoN FtvB - Co-onprN.q.TroN

BETwEEN
sTRESSES
AND wrrHsrAND
CoNsroBnmoNs coMMoNTo RANGesA, B nwo C

17. Insulation design to power-frequencyoperating voltage and temporary overvoltages.Problems of pollution and ageing 39

18. Pollution 39
19. Ageing 4l

SncrroN Srx - Co-onorNarroN BETwEENsrREssEsAND wrrHsrAND IN voLTAcE naNcs A

20. Selectionof the rated power-frequencywithstand voltage 4l

21. Selectionof the rated lightning-impulsewithstandvoltage 4l

SecrroN SeveN - srREssEsAND wrErsrAND IN voLTAGE,uNcr

Co-onorNATIoN BETWEEN B
22. Selectionof the rated power-frequencywithstand voltage and the rated lightning-impulse withstand voltage 5l

SBcrroN Ercnr - Co-onorNATroN BETwEENsrREssEsAND wITHsTANDrN voLTAGEruNcB C

23. Insulation design with respectto power-frequencyvoltage and temporary overvoltages 53
24. Insulation designwith respectto switching and lightning overvoltages. 53
25. Block-diagram of the insulation design and co-ordination of an electrical installation 6l

ApprNox A - Surge transferencethrough transformers . 67

AppsNDrx B - Validity of the testsin Clauses52, 53 and 54 of I EC Publication 7I-l . 77
Appnwox C - Statistical evaluation of the protective effect of spark gaps 83
AppBNox D - Examplesof application . 89
ApprNox E - Clearancesin air between live conductive parts and earthed structures to secure a specified impulse withstand
voltage for dry conditions 93
TlnrBs . 9 5
Frcunrs . 106
 - 5 -

INTERNATIONAL ELECTROTECHNICAL COMMISSION

INSULATION CO-ORDINATION
Part 2: ApplicationGuide

FOREWORD

l) The fomal decisions or agreementsof th€ IEC on technical matte$, prepared by Technical Committ!€s on which all the National
Comnittees having a spocial interest thercin are reprcsented,exprcss,as Dearly a.spossible, an internatioDal cons€trsusof opiniotr on the
subjects d€alt with.
2) They have the form ofrecomn€ndations for intemational us€ and they are accept€dby the National Committr€s itr that sense.
3) In order to promote i[temational unifcation, the I E C express€sth€ wisb that all National Cofunitte€s should adopt the t€xt ofthe I E C
recommetrdation for th€ir trational rul€s in so far as national conditions will permit. Aay divergencebetweel the I EC r€comnrendation
and ihe corr€sponding national rules should, as far as possible, be cl€arly indicat€d in the latter,

PREFACE

This publicationhasbeenpreparedby IEC TechnicalCommitteeNo. 28, InsulationCo-ordination.

It constitutesthe secondpart of IEC Publication71, InsulationCo-ordination,and supersedes
IEC Publica-
tion 71A (1962): Supplementto Publication 71, Recommendationsfor insulation co-ordination. Application
Guide.
Work relating to the Application Guide was f.rst begunduring the meetingheld in Milan in 1971;the drafts
concerningthe varioussectionsof the text werediscussed duringthe meetingheld in Athensin 1972.As a result of
this meeting, the draft, Document 28(Central Ofrce)42, was submitted to the National Committees for approval
under the Six Months' Rule in November1973.
The following countriesvoted explicitlyin favour of publication:
Argentina Poland
Australia Romania
Austria South Africa (Republic of)
Belgium Spain
Denmark Sweden
Finland Switzerland
France Turkey
Germany Union of Soviet
Israel Socialist Republics
Italy . United Kingdom
-
Netherlands United Statesof America
Norway Yueoslavia
Whenthe final text wasbeingrevisedfor publication,modificationsstemmingfrom changesin the text ofthe draft
ofIEC Publication71-1wereincluded.Moreover,modificationsin the wording suggested by the National Com-
mitteeswereintroduced,Thesedo not affectthe meaningof the original text.

Othet IEC publicationr quoted in this publicariot:

Publications Nos. 60: High-voltage T€st T€cbniqu€s.

7l-l'. InsulationCo.ordination,
Part l: Terms, Definitions, Principles and Rul€s.
99-l: LightningAr.esters,
Part l: Non-linear R€sistor Typ€ Arr€steN for A.C, SystefiE.
99-1Ar First Supplementto Publication99-1.
gg-2r Part 2: Expulsion-rypeLighhing Aresters.
28L1t High-voltageFuses,
Part l: Currentlimiting Fuses.
 - 7 -

INSULATION CO.ORDINATION
Part 2: ApplicationGuide

SECTION ONE SCOPE AND REFERENCE TO EXISTING IEC PUBLICATIONS

1. Scope

This publication provides some guidanceon the selectionof the electric strength of equipment, of surge arresters
or protective spark gaps, and on the extent to which it will be useful to control switching overvoltages.

The object of this publication is not to give strict rules for insulation co-ordination and design,but to provide
guidancetoward rational and ec6nomical solutions.

Therefore it is intended to consider only a few basic casesin this publication, it being evident that installations
constituting exceptionsto normal design,or included within systemshaving exceptionalcharacteristics,will require
special study by experiencedengineers.

This Application Guide is based on apparatustypes and ratings in use at the time of publication. As new equip-
ment and equipment characteristicsare developedand proved, this guide should not be interpreted as a limit to
their adoption.

Like Publication 7l-I, PublicationTl-2 coversonly phase-to-earthinsulation.

Like Publication 7l-1, PublicationTl-2 dealsseparatelywith the three following rangesof the highest voltage for
equipment:
- range A: above 1 kV and lessthan 52 kV;
- range B: from 52 kV to lessthan 300 kV;
- range C: 300 kV and above.

It covers installations of all kinds and in all situations involving voltageshigher than I kV, whether or not thel'
are exposedto lightning, with the exceptionof overheadlines. However, the test proceduresapply also to the latter.

2. Reference to existing IE C publications

Several IEC publications give standards which deal respectivelywith the withstand and protective levels con-
sideredseparately.Thus:

a) IEC Publication 71-I, Insulation Co-ordination, Part 1: Terms, Def,nitions, Principles and Rules, contains
the definitions of terms to be found here; it gives the seriesof standard values of the rated lightning and switch-
ing impulse withstand voltage, the rated withstand voltages of short duration at power frequency, as well as
the tables of the recommendedcombinations between the highest voltage for equipment and these rated withstand
voltages.

b) IEC Publicationgg-I, Lightning Arresters, Part 1: Non-linear Resistor Type Arresters for A.C. Systems,
gives the protective characteristicsunder lightning impulses of non-linear resistor type surge arrestersfor ratings
up to 198 kV. Standard ratings above 198 kV are not included in IEC Publication 99-1. However, values for
guidanceare given up to 396 kV. Valuesfor srvitchingimpulsesare under consideration.
 - 9 -

The characteristicsgiven in IEC Publication 99-l refer to surgearrestersin generaluse. However, surge arresters
with better protective characteristics are available and their standardization is at present under consideration.
IEC Publication 99-14, First Supplementto Publication 99-1, givesguidanceon the selectionof surge arresters.

c) IEC Publication 99-2, Lightning Arresters, Part 2: Expulsion-type Lightning Arresters, gives the impulse
characteristicsof expulsion-typesurge arresters.No referenceis made in IEC Publication 7l-1 to this type of
arrester,but it is still being used on some supply systems,particularly for voltagesin range A.

d) IEC Publication2S2-1, High-voltage Fuses,Part 1: Current-limiting Fuses,lists the maximum overvoltages

permitted during the interruption of fault current by a high-voltage fuse.

e) IEC Publication 60, High-voltage Test Techniques,describesthe methods for generationand measurementof
test voltages and currents and the appropriate test procedures,and recommendsmethods for calibration and
checking of measuring devices.

There is no IEC publication for the protective characteristicsof spark-gaps.

SECTION TWO VOLTAGE STRESSESIN SERVICE

3. General

Dielectricstresses
on insulationmay be classif.edas follows:
- power-frequency
voltages,under normal operatingconditions;
- temporaryovervoltages;
- switchingovervoltages;
- lightning overvoltages.

In IEC Publication71-1,overvoltagesare classifiedwith referenceto the shapeof the voltagewave uhich dc-
terminestheir effecton insulationand on protectivedevices,without referenceto the causeof the overvoltages.

The term "temporary overvoltages"refersto sustainedovervoltages,

or to oveNoltageshavingseveralsuccessive
peaks,with a decrementof amplitudesuchasto be corrparablewith a sustainedovervoltageat powerfrequencyor
at harmonic frequency.

The terms "switching overvoltages" or "lightning overvoltages" refer to overvoltagesfor which only the highest
peakvaluehasto be consideredandwhichcanbe represented, with regardto their effectson insulationand protective
devices,by the long-frontedstandardswitchingimpulsesor short-frontedstandardlightningimpulsesusedfor test
purposes.The foregoing nameshave beenchosenbecausesuch overvoltagesoften-but not always---originatefrom
switching operations or lightning discharges.

For example,the energizationof a transformer-terminatedline givesrise to an overvoltagethat may be regarded

as a "switching" or temporaryovervoltagedependingon the decrementofthe successive peaks(i.e. dependingon
the circuit parameters).As another example: a phase-to-earthfault, although actually a switching operation (the
samephenomenonwould ariseif a phaseconductorwasconnectedto earthby a circuit-breakeroperation),may give
rise to steep-frontovervoltages,
similar to thosedue to lightning; on the other hand, a lightning surgetransferred
through a transformerby inductivecoupling betweenwindingsmay produceon the secondaryside of the trans-
former long-frontedwaves,similarto thosedue to switchingoperations.
 - 1 1-

4. Power-frequencyYoltrg€s

In insulationco-ordinationprocesses, sinceovervoltagesand impulsevoltagesare definedin termsof their peak

valuesto earth, it is also convenientto make use of the phase-to-earthpeak value of the systemvoltage, which is
\/21 \/T :0.816 timesthe usualr.m.s.phase-to-phase voltage.
Under normal operating conditions, power-frequencyvoltage can be expectedto vary somewhatin magnitude
and may be describedby meansof a probability distribution about the averageoFrating value. The parametersof
this distribution will differ from one point of the systemto another. For purposes of insulation design and co-
ordination, power-frequencyvoltage should howeverbe consideredas constant and equal to the highest voltage for
equipment, which in voltage range C does not materially differ from the highest systemvoltage, with a phase-to-
earth peak value ot V^,'r/21 1/T
In range A and in range B up to 72.5 kY, the highest voltage for equipment may be susbtantially higher than the
highestsystemvoltage,asindicatedin the note underClause5 ofIEC Publication71-1.For the sakeof standardiz-
ation, it is however assumedthat equipment insulation will always be able to operate satisfactorily at the highest
voltage for equiprnentimmediately above,if not equal to, the highestsystemvoltage.

'
5. Temporary oYerYoltages

The severity of temporary overvoltages is mainly characterizedboth by their amplitude and duration.

The importance of temporary overvoltages in insulation co-ordination is twofold:

- on the one hand, the characteristicsof temporary overvoltages at the surge arrester location are of great import-
ance in surge arrester selection;
- on the other hand, the repetition of successive
overvoltagepeaks of opposite polarity, even if of lower amplitude
than some other overvoltages,may determinethe design of both the internal insulation of equipment as well as
the external insulation (surfacesexposed to contamination).

Temporary overvoltages generally arise from:

a) earth faults;

D/ sudden changesof load;

c) resonanceand ferro-resonance.

5.1 Earth faults

The overvoltage at power frequency on the sound phaseswhen another phase is accidentally earthed depends,at a
given point of the system, on the treatment of the system neutral with respect to earth, as characterized by its
earth fault factor at that point.

Note. - In theevaluationof earthfault factors,the followingremarksshouldbeconsidered:

- In general,in order to evaluatethis factor at a given location it is assumed,for simplicity, that the fault is located at the point
for which the factor is desired; but, in some specialcases,it may be desirableto investigatethe effect of other fault locations
on this factor.
- In principle, there are as many particular valuesof the earth fault factor at a given location as different possibleconfigurations
of the system.The factor which characterizesthe location consideredis the highest of the valuesthat correspondto the differ-
ent systemconfigurations which may occur in practice.
- The systemconfigurationswhich have to be consideredare thosewhich exist during a fault; thus one should take into consider-
ation thosechangesin the systemwhich may be produced by the fault itself, for example,on account of the operation of circuit-
breakers.
- For many systems,it will be sufficient to consider only one value of the earth fault factor which covers all possible fault lo-
cations on the system.
- Attention is drawn to the fact that the highest voltage at power frequencywhich may appear on a sound phaseduring a par-
ticular earth fault does not depend only on the value of the earth fault factor, but also on the value of the phase-to-phase
voltage at the time of the fault. This phase-to-phasevoltage will generally be taken as the highest systemvoltage, as given in
 -13-

Defuitioa 4 of I EC Publication 71-1; but, in sorne cas€s,i[ order to predict the operation of Fot€ctive deeic€sand sperify
their charapteri$tiG, it is n€cessaryto take into account the incrqs€d value of the phas€-to-phas€voltage that may appe{ at
the sel€ctedlocatioD utrder abnormal operating codditions not covered by the defnition referred to above.

Within range A and in some caseswithin range B, many systemsor installations are operated with their neutral
earthedthrough a high impedance,an arc-suppressioncoil or with their rieutral isolated. For the purpose of insul-
ation co-ordination, particular attention must therefore be paid in thesecasesto the earth fault factor.

Independently of the earth fault factor, particularly high overvoltagesmay arise in range A and range B systems
in the caseof:
aJ earth faults in a systemthe neutral of which is earthed tbrough an arc-suppressioncoil when the circuit is
under-compensated ;
b) arcing earth faults in a systemthe neutral of which is isolated and in some casesin a system the neutral of
which is earthed through an arc-suppressioncoil.

5,2 Swldm changesof load

In usual conditions of operation, the phase-to-phasevoltage doesnot exceedthe value of the highestvoltage of the
systemas given by Clause4 concerningdefinitions of IEC Publication 7I-I; but higher values may temporarily be
reached in the case of sudden disconnection of large active and reactive loads; these values depend on the system
layout after disconnection and on the characteristics of the sources (short-circuit power at the station, speed and
voltage regulation of the generators, etc.).

This voltage rise may be especiallyimportant in the caseof load rejection at the remote end of a long line (Ferranti
effect). It affects mainly the apparatus at the station connected on the source side of the remote open circuit-breaker.

Note. - From the point of viewof overvoltages, a distinctionshouldbemadebetween varioustypesof systemlayouts.As extremecases
wecanconsider:thosewith relativelyshortlinesandhighvaluesof theshort-circuit powerat theterminalstations;andthose
with Ionglinesandlow valuesof theshort-circuit powerat thegenerating
site.With thelatterlayouts,asareusualin an extra-
high-voltagesystemin its initial stage,muchhigherovervoltages at powerfrequencymay resultwhena largeload is suddenly
disconnected.

Due to the characteristics of the systems, overvoltages of this kind are more severein voltage range C than in
voltage range B; overvoltagesof this kind, in voltage range A, occur in generator-transformercircuits.

5.3 Resonanceand ferro-resonance

Temporary overvoltages due to these causesgenerally arise when circuits with large capacitive elements (lines,
cables, series-compensatedlines) and inductive elements (transformers, shunt reactors) having non linear-magnetiz-
ing characteristicsare energized, or as a result of sudden changesof load.

These situations are generally found for systemsin ranges B and C in the following cases:

a) A lightly loaded line, fed or terminated by a transformer, can show for example harmonic oscillations and
pronounced overvoltages if the natural frequency of the linear part of the system corresponds to one of the har-
monics of the transformer magnetizing current.

b/ Subharmonic oscillationsand overvoltagescan occur in systemscompensatedby seriescapacitorsand termin-

ated by lightly loaded power transformers or shunt reactors if the impressedvoltage, the effectivecircuit resistance,
which is strongly influenced by synchronous machines,and the circuit capacitancefall between certain limits.

c) If harmonic filters are connected to a system containing saturable elements, oscillations due to resonances
between these elementsand the fllter capacitors can develop.

,f 7-
 - 1 5 -

These ferro-resonanceeffects following energization processescan either be sustainedor last several cycles of the
power frequency depending on the time constant of transformer inrush currents.

Causesof resonanceand ferro-resonancein systemsof range A are:

d/ Resonance between inductive and capacitive components, e.g. when capacitors are used for power-factor
correction
e) Ferro-resonancewhich can occur in non-simultaneous closure or disconnection of the phasesof a transformer
the secondary of which is loaded by a small capacitanceonly.

f) Ferro-resonanceoccurring on a transformer having on the secondary side an iron-cored inductive load such
as a voltage transformer, particularly when being switched in.
g/ Ferro-resonance which can occur when one phase of a circuit supplying an unearthed primary transformer is
opened particularly if the circuit is a cable. At the highest end of range A, transformer bushing capacitance may
be sufficient to cause ferro-resonanceunder these conditions.

6. Switching and lightning overvoltages

For the purpose of this Guide, switching overvoltagesare, as stated above, of a type which can be simulated by a
standard switching impulse, i.e. an aperiodic wave with a front duration of the order of hundreds of microseconds
and a tail duration of the order of thousandsof microseconds.They stressthe various parts of an insulation in about
the same proportions as power-frequency voltages, but are not repetitive and only one peak of either polarity is
normally significant.

Lightning overvoltagesare those which can be simulated by a standard lightning impulse, i.e. an aperiodic wave
with a front duration of the order of one microsecondand a tail duration of the order of severaltens of microseconds.
Due to the front steepness, they stressmore than the former the longitudinal insulation of inductive windings, and
becauseof their shorter duration, a generally somewhat higher stressat equal amplitude can be withstood by a given
insulation. The magnitude of this effectdependson the type of insulation being considered.

These overvoltages generally arise from:

a) line energization and re-energization;

b/ faults and fault clearing;
c/ switching of capacitive currents and of small or moderate inductive currents;
d) loadrejections;
e) Iightning strokes (first or subsequentcomponents of a lightning flash).

Note. - Theovervoltages in casesa) to d) dependon thecharacteristics

of theequipment,especially
of thecircuitbreakers,transformers
andshuntreactors. Theymaygenerally bereduced by suitable
choiceof thesecharacteristics,
i.e.in casea) by usingpre-insertion
resistorsor shuntreactors,andin casec) by usingsurgearresters
connectedbetween thecircuit-breakerandthetransformer.

6.I Switching overvoltagesdue to line energizationand re-energization

Overvoltages due to closing and to single-phaseand three-phasereclosing are of great importance in the selection
of system insulation in range C.
Theseovervoltages,exceptin specialcases,are not of particular importance in the other rangesof voltages.

6.2 Switching overvoltagesdue to faults and fault clearing

In range A and in relatively few casesin range B under the conditions listed in Items a) and,b) of Sub-clause5.1,
high switching overvoltagescan arise at the initiarion oi .:,rault.

,/-
 In all the voltageranges,high overvoltagesmay arisedue to faults to earth in compositecircuitsincluding over-
head lines and cable-connectedtransformers.

At the highestvoltagesof rangeC, a high degreeof control of overvoltages causedby line energizationand re-
energizationis normally attempted.For this reason,switchingovervoltagesdue to faults and fault clearing(single
and doublephase-to-earth faults, and iheir clearance)needcarefulconsideration.

6.3 Switchingotervoltagesdue to switchingof inductfueand capacitiyecurrents

In range A, the switching of inductive or capacitive currents can give rise to overvoltages,which may require
attention,both in high-voltagedistributionsystemsand in industrialinstallationsand powerstations.In the caseof
the former, high overvoltages
may ariseifthe circuit breakerdeionizesso rapidly asto forcethe currentprematurely
to zero, so-calledcurrentchopping.
In particular,the following switchingoperationsshouldbe taken into consideration:
a, interruption of the startingcurrentsof motors;
bJ interruptionof inductivecurrents,e.g.wheninterruptingthe magnetizingcurrentofa transformeror reactor;

c) switchingand operationofarc furnacesand their transformerswhich may leadto currentchopping;

dJ switchingof unloadedcablesand of capacitorbanks;

e, interruption of cumentsby high-voltage fuses.

In range B, overvoltagesdue to the interruption of capacitivecurrents(switchingotr unloadedlines, cablesor

capacitorbanks) may be particularly dangeroussincethe use of restrike-freecircuit-breakerscannot alwaysbe
assumed.

6.4 Switchingotervoltagesdue to suddenchangesof load

Overvoltagesdue to suddenchangesof load may start with a high switchingsurgefollowed by a temporary

overvoltage.

Overvoltagesof this kind are particularlyimportant at the highestvoltagesin range C wherea high degreeof
control of reclosingsurgesis attempted.

6.5 Lightning ottervoltages

Lightning overvoltages are causedeitherby direct strokesto the phaseco ductors,back-flashovers, or as a result
of strokesto earth very closeto the line which produceinducedlighlning surges.The overvoltagesby which sub-
stationinsulationis stressed are a function ofthe line constructionand the systemconfiguration.The configuration
of the stationitself hasa greatinfluenceif the travellingtime of surgeswithin the stationis not negligiblein relation
to the front time of the surge.

Dependingon the systemconfiguration,overvoltages

with time parametersin the rangeof switchingsurgesmay
also arise as a result of lightning strokes.

Lightning discharges
which producesignificantovervoltagesin rangesB and C are confinedto direct strokesto
phaseconductorsor strokesto towersor earthwireswith subsequent back-flashovers.

In range A, induced lightning surges must also be considered. Furthermore, in this range, surges transferred
through transformersfrom a higher voltage slstem needcare:ul
"-onsideration.
 - 1 9 -

Ievels
7. Determinationof expectedoYerYoltage

7.l RangeA

For voltagesofless than 52 kV, swirchingovervoltagesdo not generallyconstitute a seriousproblem for overhead
supply systemsand insulation co-ordination is basedon lightning overvoltages'

Switching overvoltagestransferred from an overheadline into a plant through transformers or lengths of cable
can, in general,be ignored for ttre samereason.An exception is the caseof an installation connectedto the lower-
voltage terminals of a high-voltage transformer feeder, particularly if resonanceoccurs between the two systems
during one- or two-phaseenergizing'

In industrial plants and power stations,the amplitudesand waveshapesofswitching overvoltagesgenerat€dwithin

the installationvary ov€r a very wide range.In the great majority of casesthey are innocuous;in some cases,
serious overvoltage amplitudes and rates of changecan occur. Thus sudden voltage swings can be causedwhen a
switching devic€re-strikes; the resulting rate-of-changeof voltage can equal that causedby a severeclose lightning
strike.
A very large amount of practical operational experienceis available from diferent industrial plants and power
stations, and tle most severeovervoltagesor voltage swingscan usually be avoided by eliminating resonanceand
by correct choiceof the switching deviceto be employed.
Detailed representationof the systemunder consideration on a digital computer or transient network analyser
can only be economicallyjustified in particular casesat thesevoltage levelssinceaccuraterepresentationis neededto
obtain accurateresults and a complex plant frequently consistsof many piecesof equipment and electrical connec-
tions. Furthermore, the operation of sometypes of switching devicesand arcing earthsis difrcult to sirnulatewith a
sufficient degteeof accuracy.Exlrrience is often the best guide and, in exceptional cases,deliberate switching te3ts
with simultaneousrecording(both high-speedand low-speed)will producethe most valuableinformation so that
remedial measurescan be taken as the result of subsequentcalculations and confirmatory t€sts.

The amplitudes,waveshapesand frequency of occurrenceof lightning overvoltageson systemsin range A can be

estimatedwith a reasonable degreeof accuraoy.As the impulseflashovervoltageofinsulatorsusedon overheadlines
in this range is quite low as comparedwith the voltageimprdssedon sucha line by a direct lightning stroke,the
stresses to which substationequipmentis liableto be subjectedareprimarily determinedby the type ofline construc-
tion. Thus careful protection of substation equipment is required if it is connectedto a wood-pole line with un-
earthed cross-arms.Reducedprotection is permissiblewherethe lines are erectedon steelmasts,reinforced concrete
poles or where metal cross-armsare otherwiseearthed.

Amplitudes and waveshapesare also affectedby the following factors which characterizethe constitution of the
systemand arrangementof the station:
a) Surgeimpedanceofthose lines or cableswhich are connectedto the station. For example,when only one line
is connectedto a terminal transformer, the surgeis reflectedat the termination and is doubled in voltageamplitude;
whenn linesofthe samesurgeimpedanceare connectedto the busbarsof a station,and if the lightningstroketo
the line doesnot oc.curcloseto the station, the voltage at the busbarsbecomes2 ufn, where z is the amplitude of
the surgevoltage transmitted along the line on which the lightning surgeoriginated; in the caseof lightning in the
vicinity ofthe sub-station,wavereflectionmust be taken into accountand calculationmay proveto be necessary.
6) Cableswith an earthedmetallic sheathwhich are in serieswith the line or connectedbetweenthe station bus-
barsand apparatusto be protected,Any cabletendsto reducethe steepness of the wavesenteringthe station,but
the cablelengthmustbe at leastI or 2 km long beforeit hasa significanteffecton the surgeamplitude.In the case
of a directlightningstroketo the last spanin front of a station,a cablesectionbetweenoverheadline and station
afiordspracticallyno reliefto the stationequipment.More detailsarc givenin Sub-clause 21.4.

c) Protectiveearthwireson the overheadlinesextendingup to a few kilometresfrom the station;theseare efect-

ive againstcloselightning strikesto the line near the station uhich are the most dangerous.This pre-supposes
 -21 -

that shieldingby the earth wiresis sufficientlywell designedto preventdirect strokesreachingthe phaseconduc-
tors, and that the earthingresistanceof the tower is sufficientlylow to reducethe risk of back flashover,as given
in Item e) below.
d) Prctectivespark gapsor protectiveearth wiresextendingover one or two spansin front of the station; these
can materiallyreducethe amplitudesof incomingsurgeson lineswith high insulationto earth,e.g.on fully insul-
ated wood-polelines.
e,) Earthingresistancesand inductancesof the down leadsof towers,particularlyoloseto the station.In the cases
of high valuesof earthingresistanceor the inductanceof the down lead of the tower or pole, a lightning strike to
sucha tower or pole or to an earth wire may causehigh overvoltageson the phaseconductorsby back flashover
acrossthe line insulatorsto one or more phaseconductors.

In range A, lightning surgestransferred through transformers are also important. Analytical expressionsfor the
electrostaticand electromagnetic termsof the transferfedvoltagesare derivedin AppendixA.

7,2 RangeB

In rangeB, as well as in rangeA, the insulationlevelsare generallysuchthat switchingovervoltages

are seldoma
major problemand that insulationco-ordinationis still mainly basedupon lightningovervoltagesoccuningin over-
headline systems.

Furthermore,also in this voltagerange,thereis usuallyno decisiveeconomicincentivetowardsa detailedstudy

of oYervoltagestresses.

Thus the considerations

in Sub-clause
7.1 continueto apply.

7.3 RangeC

For this voltagerange,the inportance of the switchingovervoltagesin insulation co-ordinationbecomespre-

dominant,all the more so as the voltagelevelincreases. This is whatjustiflesthe substitutionof a switchingimpulse
test,consideredmore representative, for the traditional 1-minpowerfrequencytest.

The high cost of equipmentthen compelsconsiderationof more economicaldesignsof insulationco-ordination.

while, in turn, the seriousconsequences
of a failure necessitate
a more preciseestimationof the overvoltagesto be
expected.Thesehaveto be evaluatedfor eachtype of significantovervoltagein the particularsystemconsidered.

Becauseofthe extensivecomputationalrequirements, virtually all practicalovervoltagepredictionsmustbe made

using transientnetwork analisersor digital computers.

Experience with studiesof a wide varietyof systemshasshownthat developmentof generalized formulaefor over-
voltagesis difficult becauseof the large number of parametersaffecting the overvoltagevalue.

Both analogueand digital techniquesof transientsolution requirea high level of skill in problern-solving. These
skills are principallyusefulin the selectionof significantcases(it beingimpracticalto studyall possibilities),in the
reductionof the systemto a reasonablenumber of busbarsand tines (it is not practicalto representthe entire
systemon eitherTNA or digital solutions)and in the descriptionof systemconstantsand apparatuscharacteristics.

Wheneverpossible,f.eld teststo checkthe validity of the parametersusedare recommended.

In the sophisticatedapproachesto insulationco-ordinationnow becomingmore and more usual for the highest
valuesof voltage,the amplitudesof the overvoltages
to be expectedat a givenlocationdue to a giventype of event
cannotbe definedby a singlevalue(seeFigure 1, seepage106).It is only possibleto statethe probabilityf.(U) dU
 -23 -

that an overvoltage value comprised between U and U + dU may occur,f"(tJ) being the overvoltage probability
density.The probability F"(U') that the value U' may be exceeded
is then givenby:

F o ( U ' ) :I f " ( q d a (1)

U'

SECTION THREE INSULATION WITHSTAND

8. General

8.1 Self-restoringand non-sef-restoringinsulation

Clauses10and 11,concerningdefinitionsofIEC Publication71-1,subdivideinsulationinto self-restoring and non-

self-restoringinsulationaccordingto its behaviourin the caseof occurrenceof a disruptivedischargeduring a di-
electric test. On the former kind of insulation, it is possibleto carry out testsunder conditions that imply an appreci-
able risk of such discharges,e.g. by applying a large number of impulsesat the rated impulse withstand voltage, or
evenin conditions with deliberately applied dischargesas in a 50olodisruptive dischargetest carried out at voltages
above the rated impulse withstand level.

On non-self-restoring insulation,a disruptivedischargedestroysthe insulatingproperty of the insulationand a

largenumberof impulsesat ratedwithstandvoltagemay resultin a gradualdeteriorationofthe insulation.Non-self-
restoring insulation is for these reasonstested by application of a limit€d number of impulses at rated withstand
voltage.
The degreeof information on the dielectricstrengthofthe equipmentdirectlyobtainablecanthusbe much higher
for self-restoringinsulation.However,in the caseof non-self-restoring
insulation,the economicimportancefor the
manufacturerof the risk of havingthe equipmentrejectedtendsto compelhim to designthe equipmentfor a very
low probability of failure under test. Taking these two factors together,no differenceis made in IEC Publi-
cation 71-l betweenimpulsewithstandlevels,in relationto the kind of insulationor the natureofthe test.

While self-restoringinsulationdoesnot loseor rnodify its insulatingability following a disruptivedischargein a

dielectric test, it should not be inferred that damagemay not occur in serviceifthe disruptive dischargeis followed b1
an intensepower arc.
Furthermore,possibledamageto equipmentis not the only considerationto be introducedin the selectionof an
acceptable
risk of dischargein service,asthe effecton continuityof supplyalsohasto be considered.For example,a
nuch lowerprobabilityofinsulation failureis requiredin the caseofbusbarsthan on individuallines.

It must be emphasizedthat the insulating structuresof a pieceof equipment are always made up of self-restoring
and non-self-restoring parts. Generally it cannot therefore be stated that the insulation of an apparatus is self-
restoringor non-self-restoring. But tlte probability that discharges may occur acrossor through non-self-restoring
parts in the presenceof self-restoringparts can,for diferent typesofequipment, be negligibleor otherwise.Due to the
diferent voltage-timesparkovercharacteristics of solid and air insulations,this probabitity tendsto increasewith
increasingimpulsevoltageamplitudes;thus it may be negligibleat the rated withstandvoltage,but may becomeap-
preciable around the 50% disruptive dischargevoltage.

8.2 Selectionof the type of test

For some types of apparatus, within the range of overvoltagesthat tests have to simulate, the probability of a
dischargeoccurring across a non-self-restoring part is negligible. In this case,the dischargeprobability coincides
 -25-

with that of the self-restoringpats of th€ apparatusand its insulationmay be called "essentiallyself-restoring"
or, for the sakeof simplicity,"self-restoring".Disconnectingswitchesmay be consideredan exampleof this type;
in fact, €venwhenapplyingimpulseswell abovethe 50% dischargevoltageduring a 507odischargetest, sparkover
takesplaceusuallyin air without any punctureof the porcelain.For this type of equipment,the test of Clause52
of IEC Publication71-l is possibleand recomnended.

For tl,p€sof apparatuswith mainly or a large part of non-self-restoring insulation,the high cost of this part of
insulationexcludestestsofthe completeapparatusat higherthan ratedwithstandvoltage.The testvoltageis there-
fore limited to the rated withstand voltage asthis already implies an overinsulation of the non-self-restoringparts so
as to give negligibleprobability of dischargeup to ratedwithstandvoltage.

Sucht)?es of apparatusthat alsohavea self-restoringpart ofinsulation that cannotbe testedseparatelyfrom the

part, e.g.bushingsand sometypesof instrumenttransformers,would requirea large numberof
non-self-restoring
impulsesto provethe adequacyof the self-restoringinsulation.The numberof impulsesmust howeverbe restricted
becauseof the possibilityof gradualdeteriorationof the non-self-restoringparts of insulation.Insulation of this
type of equipment,that can be calledcombined,shouldbe testedaccordingto Clause53 of IEC Publication71-l,

For sometypesof equipmentwith mainly non-self-restoring insulation,e.g.powertransformers,the self-restoring

insulation(e.g.bushings)can be testedseparatelyaccordingto Clause52 of IEC Publication71-1.This kind of
equipmentis called "essentiallynon-self-restoring"or, for the sakeof simplicity "non-self-restoring".Insulation
of this kind is verifed by meansofthe testin Clause54 of IEC Publication71-1.

on the validity ofthe testsin Clauses52,53 and 54 of IEC Publication71-1are given in Appen-
Considerations
dix B.

9. Insulation behaviourrt power-frequencyyoltagesrnd temporary oyervoltag€s

In general,sparkoverunder power-frequencyvoltage in normal operating conditions and under temporary

overvoltageswill be causedby progressivedeteriorationofthe insulatingpropertiesofthe equipmentor by excep-
tional reductionsin insulationwithstanddueto severeambientconditions.
In the latter case,the conceptof probabilityis applicableto the degreeofcontamination(seeClause4).
Becauseofthe difrcultiesinvolved,no useof statisticalconceptswill be madein this Guidein respectof insulation
behaviourat power-frequency voltagesand temporaryovervoltages(seealso Clauses17 and 18).

10. Probability of di$uptive rlischrrgeof insulation under impulsevoltages

The ability of a given insulation to withstand the dielectric stressescausedby the application of an impulse of
givenwaveshape and peakvalue U is, in most cases,a randomphenomenon,evenif we considera time interval so
small (suchasthat neededto carry out a dielectric test on equipment)that the ambient and insulation conditions may
be consideredconstant, at least with respectto parameterssuch as pressure,temperature,humidity, etc., which can
be measuredand which are usedto definethe ambient and insulation conditions during tests.

The disruptive dischargeprobability of insulation for an impulse of given waveshapeand polarity, and for a peak
value U in a short time intervalas definedabove(e.g.in a dielectrictest)can be determined,if the insulationis self-
restoring,by applyingthe impulseU N timessuccessively within this time interval,and countingthe numberNl of
. N l
discharges.From the fraction , a numericalvalue can be obtainedfor this probability which will be the more
accuratethe greater is the valu; of N.
 -27 -

The dischargevoltagesof a populationof non-self-restoring insulationcan be describedstatisticallyby meansof

a distribution curve that gives the relation betweenthe amplitude of the dischargevoltage and the fraction of the
populationthat will break down at that voltageamplitude.For the determinationof this distributioncurve.
tests
must be performed vrith Yoltageamplitudes increasedup to breakdown on a sampleof piecesfrom the population.
The accuracyin the determination of the curve will be increasedby increasingthe number of piecesin the sample.
Becausea disruptive dischargenormally resultsin the destruction ofthe test piece,the number ofpieces in the sample
must be limited for reasonsof economy.
No method is at presentknown for the determination ofthe probability of disruptive dischargeofa singlepiece
of
non-self-restoringinsulation.
If we consider either switching or lightning impulsesof different peak values U, we shall be able to assoctatewith
every possiblevalue of U a dischargeprobability P,, thus establishinga relafionship pr(U) for a given insulation in a
short time interval At or, for the sakeof simplicity, at a time t (seeFigure 2a, page 107).
The valuesofP,(LQ increasefrom near 0 to near 100% probability in a more or lessnarrow band ofvoltage values.
In general,the resulting curve can be definedby a biparametric law, one paraneter being associatedwith the position
ofthe voltageband and givingan indicationof the withstandlevel,and the otherassociated with the bandwidthand
giving an indication of the scattering of the voltage values which give appreciableproportions of both
discharees
and non-discharges.
Generally, in a laboratory, the parameterthat definesthe position of the probability curve is taken as the voltage
Ur uowhich conespondsto the 50% discharge(or withstand) probability. The standard deviation of the distribution
(o), defined by the squareroot of the sum of the squaresofthe deviations with respectto the mean is
usualty taken
as the parameter which expressesthe scattering. For a Gaussiandistribution, it correspondsto half the difierence
betweenthe voltagesthat give dischargeprobabilities of 16Voand 84vo.
In service,the ambient and insulation conditionsdo not remain constant.Thereforethe discharge probability
curve of insulation, as defi.nedabove for the time t, is bound to changefrom one moment to another (.pr, pr..
. .)
(seeFigure 2a). The variations are determined,as regardsexternal insulation, mainly by atmospheric
conditions.

Taking the ambientand insulationconditionsas random,it will be nec€ssary to considerfor eachinsulation,in

addition to the dischargeprobability P,(t/), as defi.nedabove, a dischargeprobability of insulation p1({/) to
over-
vdltagesof amplitude U liable to occur at any instant of a long time interval z of operation.For the purpose
of
insulation design,it is this seconddistribution which is of interest (seeFigure zc, pageiol).

Similarly to Pt(U), P(Lf can be defined by the voltage (U1 50)which coruespondsto the 50% discharge(or $.irh-
stand) probability and by the standard deviation of the distribution oa.
The variations of Pt( t/) within the time interval AZ may beconvenientlydefinedby the probability densitypo([.'.,01
where Ut 60is consideredas a random variable (Figure 2b, page 107).This latter function in turn may be charactcr-
ized by the 50olodischargevoltage Ua 50and by its standard deviatiou oo.
On the simplifying assumption that the standard deviation ar of Pr(U) is constant within the time interval A2.
the following relation holds:
1l-=__---
o, -- ll o?+ o? (2)
v
In IEC Publication71-1,the parameterthat definesthe position of the probability curve p(Lf is taken
as the
voltagewhich comesponds to a withstandprobability of90% (seeClause26 ofIEC pultication il-l), atthoughthe
50% dischargevoltage which was referred to aboveis a convenientmeasurefor piecesof insulation that can
be sub-
mitted to a 50% disruptive dischaxgetest.
The reasonfor this choiceis that the 50% disruptive dischargetest cannot be generally applied to all kinds of
insu-
lation. Thus, in order to have the samevalues of the rated impulse withstand voltages foi all tylrs of
equipment,
whatever its insulation, and to use these values directly in the definitions of statistical distributions,
it has been
appropriate to refer to a higher value (90%) of the withstand probability, the rated impulse withstand voltage
!3em9d
Uav being identical with the lowest perrnissiblevalue of the statistical impulse withstand voltage under
specifed
t€st conditions 4 eo.
 -29 -

For risk-of-failureevaluations,it is howeverconvenientto expressthe probabilitycurvesofinsulation

dischargein
termsof their 50% dischargevoltagesand standarddeviations.
Assumingfor Pr(U) a Gaussiandistributionwith standarddeviationor, the differencebetween
the 50"/odischarge
(or withstand)voltageand the statistical(or 90%) withstandvoltageis givenby:

tr Ut so
utso:
r-l-3 o, (3)

o, dependsupon numerousparameters(waveshape, polarity, confi.gurationof the object,dielectricnature,etc.).

However,on presentinfo nation, and for equipmentof rangeC in air, ot is taken as 0,03or 0.06
dependingon the
type of impulse,lightning or switching, unlessanother value has beenspecifiedby the relevant
Apparaius Committee
(seeClause52 of IEC Publication71-l).
The probability of dischargeP,(Lf of a piece of equipment which during the test conforms to
what is specified
in IEC Publication71-1 can then be definedin terms of its 50% dischargevoltageand its
standarddeviation,as
follows:

(Jrro:*r-r r#a (4)

The probability of dischargeP,(U), defined by the above parameters,refers to the most severetest
conditions for
the equipment,sinceUis, is the rated switchingor lightning impulsewithstandvoltage.Thercfore,
if impulse tests
haveto be madein both wet and dry conditions,the probabilityP,(tr) refersgenerallyto wet
condifions.

The probability of dischargein servicep1((/) of a given piece of equipment can only be deduced
from field tests,
dependingon the site of insiallation.
However,as a broad indication,the probabilityPa(Lf of a pieceof equipmentconformingto the
specifications
of
IEC Publication7l'1 can be defined,recallingequation(2), in termsofits 50% dischargevoltage
and its standard
deviation,as follows:

Un*
Lhso > k------
l-1.3 o,
-t
/-
(s)
o,-ll ,?+ol
v
where/c is the ratio betweenthe 50% dischargevoltageof a givenequipmentin serviceduring a
time interral lI
and the 50% dischargevoltageunderthe mostsevereimpulsetestfor the equipment(wet or dry, positive
or negatrve
polarit .
For switchingimpulsesU of positivepolarity, the valuesof k and oo relevantto time intervals
AT of fine dry
weatheror variousbad weatherconditionsdo not show appreciabledifferences. The samemay be said concerning
the degreeof atmosphericpollution, at leastin the rangefrom cleanconditionsto lightly polluted
conditions.
switching impulses Lr of negativepolarity, the valuesof /<and o,, are highly dependenton the type
.For of weather
within the time intervalA? under consideration.Concomitanceofbad weather(rain, snow,fog, misi,
erc.; and not
negligible pollution leadsto a low value of k ; bad weather also increasesthe value of oo.

Valuesof k : I and oo : 5oloare suggested in this Guide for normal conditionsand a time intervalAZ equalto
the seasonal
cycleto coverthe worsepolarity impulse.This valueofoo resultsin a valueofar a little lower than g%.

The same values of ft and oo are also suggestedfor lightning impulses. This gives a value of aa equal
to 6%
approximately.
The information given above is to be consideredmerely as broadly indicative and it is recommended
that use
should be made of more detailed data derived from field and laboratory tests,if available.
 11. Apporetuswith windings

An apparatuswith high-voltage windings, such as a transformer or reactor, designedto withstand only full-wave
tests,is vulnerable,to a certainextent,to a surgeof high amplitudechoppedin its vicinity becausehigherinternal
stresses than underfull-waveconditionscan be developedacrossadjacentturns and coils.All flashoversto earthin
a substationresultin choppedwavesofvarious degreesof amplitudeand steepness. If becauseof the useof protec-
tive spark gaps, flashoversare liable to occur frequently in service,the withstand strength of the windings against
surgesmust be determinedby testingwith a suitablechoppedwave.The provision for such a test is left to the relevant
apparatus committee.

Where non'linear resistor-type surge arresters are used for the protection of transformers, chopped wave
tlpe
surgestend to be lesssevereand they are lesslikely to arise; in practice,choppedwavetestsare not usually required.

For all typesof apparatushavingwindings,suchasrotatingmachines,transformersand reactors,rapidly changing

voltagesdueto the restrikingofswitchingdevicesmay alsoproducenonJinearvoltagedistributionssimilarto
those
causedby lightning overvoltages. For this reason,it is recommended that suchequipment,irrespectiveof whether
or not it is to be usedin installationssubjectedto lightning overvoltages,
shouldbe testedwith a lightning-impulse
voltageto sheckthe winding insulationfor voltagewithstandacrossturns and coils.

SECTION FOUR PROTECTIVE DEVICES

12. General

These devicesfall under three classes:

- non-linear resistor-type surge arresters;
- expulsion-type surge arresters (range A only);
- spark gaps.

The choice betweenthesethree devices,which do not provide the same degreeof protection, dependson various
factors, e.g. the importance of the equipment to be protected, the consequencesof an interruption of service,
etc.

Their characteristics
will be consideredbelowfrom the point of view of insulationco-ordination.

13. Non-linear resistor-typesurgeanesters

Theseprotectivedevicesshould be designedand installedto limit t}re magnitudesof overvoltagesagainstwhich

theyprotectequipmentso that the total surge-arrester voltageduring operationdoesnot exceedan acceptable value.
Surgearrestersare deflnedand their characteristics are givenin IEC Publication99-1.Their rating is definedas the
specifiedmaxirnum permissibler.m.s. value of power-frequencyvoltage betweentheir terminals at which they are
designedto operate correctly; this voltage may be applied continuously to surge arresterswithout changing their
operatingcharacteristics.In addition to this definedcapability, sometypes of surgearresters1 can successfully
withstand either aJ higher than rated voltage for a specifiedshort duration, or D) a specifiednumber of successive
discharges.

In eithercase,a controllingfactor in the selectionof the surgearresteris its ability to interruptfollow current at
its ratedvoltageor in the caseof temporaryovervoltages, exceedingits rated voltage.

I Thesetypes of surge arresters

are currently available onl)' for range C.
 Aprimary point is that thevoltageproducedacrossthe terminalsofthe arresterat anymomentprior to and during
operationmust be consideredin the determinationof the switchingimpulseprotectivelevel and the lightning im-
pulse protective level.

13.1 Lightning impulseprotective level

The lightning impulse protective level of a surge arrester is characterizedby the following voltages:

a) the sparkover voltage for a standard full lightning impulse (seeTable VI of IEC Publication 99-1)1;

6,) the residual (discharge)voltage at the selectedstandard nominal current (see Table YII of IEC Publi-
cation 99-1);
c) the front-of-wavesparkovervoltage(seeTable VI of IEC Publication99-1).
The protectivelevel under lightningimpulsesis taken for insulationco-ordinationpurposesas the highestof the
three following values:
- maximumsparkovervoltagewith 1.2/50impulse;
- maxinum residual voltage at the specifledcurrent;
- maximumfront-of-wavesparkovervoltagedividedby l.l5 (seeClause33 of IEC Publication71-1).

This evaluationof protectivelevel givesa conventionalvaluerepresentinga generallyacceptableapproximation.

For a better definition of wave-frontprokction by a surge arrester,referenceshould be made to IEC Publi-
cation99-1.

13.2 Switching impulseprctective level

The swirchingimpulseprotectivelevelof a surgearresteris characterized

by the following voltages:
a) the maximum sparkovervoltagefor the standardwave-shapes
specifledin Sub-clause61.4 of IEC Publica-
tion 99-1;
,) the total surge-arrestervoltage exhibited by the surgearresterwhen dischargingswitching surges.

The protectivelevel for switchingimpulsesis the highervalueof a/ or 6). Until a standardtest for D) has been
specifledby IEC TechnicalCommitteeNo. 37, LightningArresters,referenceshould be madeto the surge-arrester
manufacturers,

14. Expulsion-typesurge arresters

Theseprotective devicesoperateto limit overvoltagesand interrupt follow currentswithin their ratings. They have
low residualvoltages.The characteristicsof thesedevicesare givenin IEC Publication99-2.

The impulse sparkover characteristicsresemblethose of protective spark gaps,but are in generallower and flatter
for the samesparkover distance.
Thesearrestersmay not appreciablylimit the amplitudeof the follow currentbeforeinterruptingit and may have
current-interruptingratingswhich mustbe comparedwith the prospectivefault currentand the prospectivetransient
recoveryvoltageat the point of installation.

I The tables mentioned here give for each surge arrester voltage rating the upper limit for each of the above voltag€s. If bettei character-
istics than those specified in I EC Publication 99-1 are available, the actual voltages for the specific sulgp arester will be obtainable from
the manufacturers. Thus, it is recommended that the actual voltages for the surge arrester prctectiv€ chaRcteristics be used foi co-ordi-
lation studies.
 -35-

15. Spark glps

The spark gap is a surge-protectivedevicewhich consistsofan open air gap betweenan energizedelectrodeand an
earth electrode.
On supply systemsoperating at voltages up to 245 kV, spark gaps have proved satisfactory in practice in some
countries with moderatelightning aotiyity. The adjustment of the gap settingsis often a compromise betweenpro-
tection and servicecontinuity, but this difficulty can be largely overcomeby the useof rapid automatic reclosing.

The sparkover voltage and the time-to-sparkover ofthe gap dependessentiallyon the distanc€ between the elec-
trodes; they are influenced by the shapeof the electrodesand also by their disposition and distsnce relafivd to the
neighbourilg live and earthed parts.
In order to improve the operation of a spark gap under steep-fronted surgesand to provide a flatter impulse
sparkover-voltag€time characteristic,the geometricalconf.guration ofthe simple rod-rod electrodearrangementcan
be modified,for instanceby appropriateshapingofthe electrodesand by provisionof a centralauxiliary electrode.
In range A, duplex-type spark gaps have also proved advantageousin regions where birds or small animals are
troublesome.

ls,L Prctective characteristicsof a spark gap

Becauseof:
a) the dispersion of the sparkover voltage of an air gap,
,, the increasein the sparkovervoltage with increasingamplitude ofthe applied wavewhen sparkovertakesplace
on the front of the wave,
the protection obtained by meansof spark gaps is lessprecise and the protective level cannot be given asprecisely
as the protective level of non-linear surge arresters.
The performance of a spark gap under impulse (switching or lightning) is characterizedby the 50olovalue and the
standard ddviation of its dischargevoltage under standard laboratory conditions. Since spark gaps constitute
typical self-restoringinsulation,the contentsof Clause10 apply also to them. Furthermore,becauseof the reasons
givenin Iten b) above,knowledgeof the times-to-sparkover of the spark gap for valuesof the appliedimpulses
well above the 50% sparkover value is often needed (seeAppendix C).

15.2 Limitations of protective spark gaps

c) When the spark gap operateson a voltagesurgeand a power-arcresults,this arc frequentlypersistsuntil dis-
connectedby a fault-protectivedeviceof the system;this constitutesa phase-to-€arth fault in the caseof a system
with directly earthed neutral, entails mechanicalstresseson the various parts of the systemequipment and may
causedisturbancesto users.The location of the spark gap should therefore be consideredin relation to its efect on
the systemprotection aud operation.
b) The spark gap is unacceptablefrom the point of view of servicecontinuity if its presencenoticeably increases
the number of circuit outages,provided theseflashoversare neither self-extinguishingnor interrupted by means
of high-speedtripping circuit-breakersfollowed by high-speedreclosing.
c) Spark-gapoperation causesohoppingof the wave, thus increasingthe probability of producing chopped
wavescloseto the terminalsof protectedapparatus,This hasto be takeninto considerationfor insulationofhigh-
voltagewindings(seealso Clause11).
d) Damage to the equipment may be causedby the power-arc acrossthe spark gap if this is not installed in a
suitableposition. For instance,if a spark gap is fitted to the bushingof a transforrreror circuit-breaker,its dis-
tance from the bushing surfacemust be sufficiently large to prevent a power-arc being blown againstthe insulator.

e) The relativearrangementofthe spark-gapsofeach ofthe phasesmust be chosenso asto minimizethe risk of

an arc spreadingto neighbouringphases,which would transforma single-phase
fault into a tfuee-phase
fault.
 -37 -

16. Application of different protective devices

l6.l Protection with non-linear resistor-typesurge arresters

The distanceseparatingthe objectto be protectedfrom the surgearresterreducesthe efficiencyof the latter. In

fact, when the surgearresteris separatedfrom the object to be protected,the latter is subjectedto an overvoltage
exceedingthe protectivelevelof the arrester.This increaseis due, in the first place,to the inductivevoltagedrop in
the arresterconnectionsand in the link betweenarresterand apparatusto be protected;secondly,ifthe duration of
propagationbetweenarresterand apparatusis not negligibleascomparedto the durationofthe incomingwavefront,
due to an increaseof short duration of the voltageat the terminalsof the apparatuswith respectto the protective
level of the arrester.Thus, it is a generalrule to locatethe arresteras closeas possibleto the apparatusto be pro-
tected.In particular,surgearrestersshouldpreferablyeitherbe installedon the transformertank or its high-voltage
and earthterminalsshouldbe connectedto the transformerby the shortestpossibleconnections.
Similarly, surgearrestersshould be fi.ttedcloseto cableterminationsif they needprotection,with the shortest
possible connectionsbetween the terminals of the surge arrester and the phase conductor and the cable
sheathrespectively.

Note. - In rhe caseofsurge afieslersclo6eto theapparaluslo be protected,the followingsafetyfadors are recommended.

a, Range A, - A safety factor of approximately 1.4 should be provided for Series I equipment between the rated lightning
impulse withstand level of the apparatus 10 be protected and the impulse protective level of the surge arrester (see Claus€s4l
and 42 of I EC Publication7lJ). For SeriesII equipment,factorsdown to 1.2are acc€ptedfor sp€cificcases.
6., RangesB and C. - Saf€tyfactors ofat least 1.2a.e normally provided for lightning overvoltag€s.
c, RangeC. - Safety factors of at least l.l5 for switching overvoltagEsand 1.25for lightning overvoltagesare normally
adopted.
The considerationsin I EC Publication99-lA shouldalso be taken into account.

The installationof surgearresterscloseto the apparatusto be protectedcan be achievedmore easilyin voltage

rangeA than in voltagerangesB and C.
The increasedue to both thesefactorsdependson a numberof conditions:distanceof the surgearresterand its
locationaheadof, or behind,the apparatusto be protected,characteristics of the line, capacitance
of the apparatus
to be protected,arrangementof the stationand steepness of the incomingwave.This increasecan be limited by all
arrangements which limit the steepnessof the surgearriving at the station (extensionof shieldingwires,localized
capacitance,cable(evenshort),largenumberof connectedlines).

16.2 Protection tuith expulsion4ypesurgearrestefs

Thesearrestersare sometimesusedon high-voltagedistributioncircuits whereshieldingagainstlightning is nor

provided(rangeA).
The impulsesparkovervoltage-timecharacteristicof suchan arresteris flatter than that of a rod gap of the same
sparkoverdistance,but not quite asflat asthat of sometypesof equipment,for examplea transformer$ inding or a
cable.
For this reason,an adequatemargin of safetyis required,not only for the lightning-impulsesparkovervoltageof
the arrssterand of the equipmentto be protected,but also for the correspondingfront-of-wavdsparkovervoltage.
Theseconditionsare assistedby the usualpracticeof installingthesetypesof arrestercloseto the equipmentto be
protected.
For further information concerningthe applicationsof thesedevices,referenceshould be made to IEC Pub-
lication 99-2.

16.3 Prctection tith spark gaps

The impulsesparkovervoltage-timecharacteristic of a sparkgap is usuallymuch morecurvedthan thoseof some

of the typesof apparatusto be protected,particularlythoseof transformsrsand cables.
Due to the curvedshapeof the voltage-timecharacteristicof a spark gap, the distanceover which protectionis
ensuredfor all surgesis very small, usuallynot more than a few metres.If a spark gap is appliedfor protection
 -39-

against surgesof a limited front steepness(considerablylower than the steepnessof the standard lightning-impulse
test voltage wave), a distanceof severaltens of metresbetwe€nthe spa.rkgap and the object to be protected does not
appreciably modify the conditions for the protection provided against such surges.

A spark gap is therefore liable to operate not infrequently when stressedby lightning surges,and occasionally
when stressedby switching surges,the amplitudes of which are below the lightning-impulse withstand voltages of
the apiaratus to be protected.In a large number of cases,the operation of the spark gap causesa circuit outage if the
gap is on the supply side of the opening switch. If the supply can be restored quickly by high-speedautomatic re-
closing, the setting of the spark gap can be adjusted so as to provide an acceptabledegree of protection to the
apparatuswithout causingan excessivenumber of supply interruptions troublesometo consumers.

Note. - Safety fa.lors of tlre order of thos€ gived for su!8p arr€steE gcoerally ensure satisfactory prot€ction provided the occurren€ of
wry st€€p fro ed surg€s is excluded (s€e AppeDdix C).

SECTION FIVE _ CO-ORDINATION BETWEEN STRESSES AND WITHSTAND

CONSIDERATIONS COMMON TO RANGES A. B AND C

17. Insulation design to F)wer-frequency operating yoltage anrl temtrorary oyeryoltrge, hoblems of pollution
and rgeing

Clauses38 and 39 of IEC Publication71-1leaveit to the relevantapparatuscommitteesto prescribethe long

duration power-frequency tests,intendedto demonstratethe behaviourof equipmentwith respectto ageingof in-
ternalinsulationor to externalpollution. Generalguidanceonly is givento the apparatuscommittees;it is indicated
that as regardsthe voltage under normal operating conditions, the insulation shall withstand permanentoperation at
the highestvoltagefor equipment.

18, Pollution

For insulation susceptibleto contamination, the problem of specifyinga suitabletest method and pollution severity
levelsis at presentunder considerationin variousapparatuscommittees.When contaminationtestsare established,
it is anticipated that the systemengineerwill specifya degreeof pollution sevGrityin relaiion to the pollution of the
environment in which the equipment is installed.
Table I, page 95, gives a provisional basis to the systemengineerfor establishing a classification of pollution
severity levels.
A scale defined in quantitative terms with referenceto a test method should be associatedwith each of these
qualificative levelsof natural pollution severityfor various tytrrs of insulator.
Besidesbeingreproducible,the testmethodshould,asfar aspossible,satisfythe requirementsof validity, i.e. ofa
saiisfactorysimulation of the natural conditionsin which the equipmentis to be installed.Therefore,the most
satisfactorytests,among those at present adopted, may vary from caseto case.

It should be emphasizedthat Table I does not cover some environmental situations, such as desert areaswhere
long dry periodsare followedby condensationor light rain.
As an example,Table II, page 97, givesthe relationship betweenthe levelsofnatural pollution and their simulation
in laboratories according to two types of tests.The table also gives an indication of the requircd creepagedistance,
although it is recognizedthat the performanceof surfaceinsulation is greatly affect€dby insulator shape.Moreover,
at the presentstageof research,theseindicationsare valid only for pin-and-captype insulators.

\."..!
 -41 -

The data in Table II are intendedto cover the behaviourof equipmentat the appropriatesingle-phase
voltage,
i.e. U^l t/l However,in the caseof a systemwhich may operatewith a phaseearthedfor long durations,these
f.gures are valid for the phase-to-phasevoltage U-.

If temporary overvoltagesare frequent and severe,it may be necessaryto take them into account in specifyingthe
pollution test.

In the case of stations with a high degreeof pollution, when it may be impossible or extremely expensiveto
requirethe necessary performanceof equipmentunder pollution conditions,the alternativesof greasingor washing
the insulating surfacesshould be considered.

19. Ageing

For insulation susceptibleto ageing,the problem of specifyingsuitable test methodsis also at presentunder con-
sideration in various apparatuscommittees.

SECTION SIX - CO.ORDINATION BETWEEN STRESSESAND WITHSTAND

IN VOLTAGE RANGE A

20. Selectionof the roted lnwer-frequencywithstandvoltage

The tablesof standardizedvaluesfor SeriesI and II in IEC Publication71-1 indicate only one value of rated
power-frequencywithstatrd voltage for each systemvoltage (/-.

21. Selectionofthe rated lightning-impulsewitlstand yoltage

TableI of standardized valuesfor SeriesI in IEC Publication71-l leavesthe choic€openbetweentwo rated light-
ning-impulsewithstand voltages,correspondingto list 1 and list 2. Table II for SeriesII indicatesrepresentatiyeyalues
of rated lightning-impulse withstand voltage dependingupon whetherthe power of the equipment is below or equal
to 500 kvA, or above.

The reducedrated lightning-impulse withstand voltagesof SeriesI, list l, wereintroduced with the avowedaim to
make thesevaluesacceptableto all apparatuscommittees.Reducedrated lightning-impulse withstand voltageshave
beenusedwith good resultsin many countriesand for a wide rangeof equipmentover long periodsof time. Com-
prehensivetests have also beenperformed on different types of equipment for this voltage range to determinetheir
inpulso withstand voltages,both for standard lightning impulsesand representativeswitching impulses.It has been
found, in particular, that the breakdown voltage of insulation under typical switching impulsesis alwayshigher than
the peak of the power-frequencytest voltage.This is onereasonwhy it wasnot found necessaryto introduce a separate
switching-impulsewithstand test in voltage range A.

The choicebetweenlist I and list 2 is to be madein accordancewith Clause42 of IEC Publication71-1and tbe
following considerationsrelevant to the equipment installations:

a, equipmentwith no connectionto an overheadline;

D,/ equipmentconnectedto an overheadline through a transformer;
c, equipmentconnectedto an overheadline eitherdirectly or through a cable, *
:
i

;'il
 _43_

21.1 Equipmentwith no corneclion to ot overheadline

A wide variety of installations is covered by this category, e.g. large underground cable networks in cities; many
industrial installations;power stationsand ship installations.Equipmentin suchlocationsis not subjectedto any
external (lightning) overvoltages,but may be subjectedto switching overvoltages(seeSub-clause 6,3).

Clause42 of IEC Publication71-1specifiesthe conditionsfor the use of equipmentaccordingto list 1 in such

installations.In all other cases,equipmentaccordingto list 2 should be usedand, with few exceptions,no surge
protection is required. One suchexceptionis an electric arc-furnaceinstallation wherehigh overvoltagesare liable to
developdue to current chopping by a circuit-breaker. Protection by specialsurgearrestersmay be required in such a
caseboth betweenphasesand betweenphasesand earth.

21.2 Equipmentconnectedto an overheadline througha transformer

A) GENBRALcoNsrDERATroNs
Equipment connectedto the lower-voltageside ofa transformer supplied on the higher-voltageside from an over-
head line is not directly subjectedto lightning or switching overvoltagesoriginating on the overheadline. However,
du€ to electrostatic and electromagnetictransferenceof such overvoltagesfrom the higher-voltage winding to the
lower-voltage winding of the transformer, such equipment can be subjectedto overvoltageswhich, in certain cir-
cumstances, can exceedits withstandvoltage.
Analytical expressionsfor the electrostatic and electromagneticterms of the transferred voltage are given in
Appendix A.
For a given transformer, the magnitudesand waveshapesof thesetransferred overvoltagesare mainly dependent
on the natureof the lower-voltagecircuit and, for this reason,it is convenientto considerthe selectionofthe rated
lightning-impulse withstand voltage of the equipment and its protection separatelyfor the two basic categoriesof
installation as follows :

Category 1. - Equipment connectedthrough transformers to higher-voltage overheadlines, and incorporating

connectionsof moderatelength,sayuP to 100m, betweenthe lower-voltagesideof the transformerand the equilF
ment, such a8the main supply switchgearof a cable distribution netlvork or an industrial installation.

Category2, - Generator-transformerinstallation.

B) BAsrc cnrDANcB

a) Category 1 equipme t
Factors which tend to increasethe magnitude of transferred overvoltagesfor such equipnent are:
i) a transformer having a high-voltage ratio and high capacitancebetweenwindings;
i, a transformer disconnectedfrom its load on the lower-voltage side;
lr) low-capacitanceconnectionsbetweena transformer and its associatedequipment;
rv, a higher-voltage winding which is not earthed (e.g. delta or unearthed star) or having a star point which is
earthedthrough a high reactance(e.g.arc-suppression coil);
v) surgeshaving steepwavefronts and surgeshaving long durations;
yt switching surgesdue to energizing a transformer from a remote point on an overheatlline
system(i.e. ener-
gizing a transformer-feeder).

Seealso note below.

Estimatesof the magnitudesof transferred overvoltagescan be made,and methodsof calculation, with examples,
are describedin Appendix A.
 -45-

Category 1 equipment can usually be protected by surge arresters, and where such protection is provided as
normal practice,it is not necessaryto makethesecalculations.For other cases,basicguidanceis givenbelowon the
nature of the transferred voltages, the general influence of circuit conditions and the criteria which can be usedto
determinewhetler precautions are necessary'
Note, - Resorunt orerrolrage: A, c..ndition ofr€sonance between two systemscoDn€ctedby a traDsfornrer cantcauseabtromally large volt-
agEsto be tlansfer€d through thc traosformer. It is reaonunended that an examin4tioo of the clruits for possible lesonance
should be ruade. atrd modifications should be made as n€cessaryto avoid resonance.

The applicationofshort duration,or steeplyrising,voltagesurgesto the higher-voltage windingofthe transformer,

e.g.a lightning stroketo the transmissionline very closeto the transformercan,throughcapacitivecoupling,give a
short duration voltage"spike" on the lower-voltagewinding. This may exceedthe impulsetest voltagesgivenin
Table I, lists 1 and 2, or Table II of IEC Publication7l-1. On the other hand, the shortestpossiblefront time,
determinedby surgeimpedanceof the line and transformerinput capacitance, is often so long that the capacitively
transferredvoltagecan be ignored.
For CategoryI equipment,the most severecondition ariseswhen the load circuitsare disconnected, i.e. on the
transformersideof the lower-voltageswitchgear,sinc€with the load connectedits capacitanceis usuallysufficient
to reduc€the amplitude of the initial voltage "spike" to a safevalue.

If the capacitanceof the connectionsbetweenthe transformerand lower-voltageswitchgearis not sumcientto

reducethe amplitudeofthe initial voltage"spike", eitheradditionalcapacitance
canbe connectedbetweenthe trans-
former terminalsand earth,or equipmentin accordancewith Table I, list 2, or Table II of IEC Publication7l-1,
may be used.
It may alsobe desirableto consideraddingsurgearresters.Attention is also drawnto the possibilityof increasing
inductivelytransferredovervoltagesby additional capacitance.This increasecan be reducedby a seriesdamping
resistor of carefully adjusted resistancevalu€.

The applicationof a longerdurationvoltagesurgeto the higher-voltage winding of the transformer,e.g. a light-

ning stroketo the transmissionline somedistancefrom the transformeror a switchingsurge,will, through inductive
coupling,give a voltagesurgeon the lower-voltagewinding of the transformerhaving a longer duration and an
amplitudesimilarto the peakvalueof the power-frequency testvoltagegivenin TableI ofIEC Publication71-1.

Dangerouslyhigh temporaryovervoltages canbe transferredto the lower-voltage

windingofa transformerthrough
capacitivecouplingfrom the higher-voltagewindingwhenan earthfault existson the higher-voltage systemand when
the neutral of the lower-voltagesystemis earthedthrough an arc-suppression coil or when it is isolated.In such
cases,the connectionof additionalcapacitancesbetweenthe low-voltageterminalsof the transformerand earthis a
widely-used protection method.

b) Category2 equipment
Recommendationson the needfor overvoltageprotection of generator-transformerinstallations and on the choice
of suitable types of protective equipment need to be basedon consideration of overvoltagesof atmosphericorigin
only sincestudieshayenot revealedmore severeconditionslikely to arisefrorn transferenceof switchingsurges,
Correspondingto the front of an incidentlightningsurge,or to the collapseof voltagedue to wavechopping,there
can be a capacitivelytransferredvoltageofshort duration(initial voltage"spike"). This is independentofthe longer-
duration voltage surgewhich is usually transferredby the combinedefect ofinductive and capacitivecouplings.

The maximum amplitude of the initial voltage "spike" is highly dependenton details of the design of the instal-
lation. Where theseare such as to assistcapacitivetransference,there may be justif.cation for making a low-voltage
surge-injectiontest on the installation or on the generator-transformerconnectedto a circuit simulating the generator
and its external connections.
Factors which tend to increasethe magnitude of transferred overvoltagesfor such equipment are:
l'J high capacitancebetweenthe transformer windings;
fi,) low-capacitanceconnectionsbetweentransformer and generator;
 - 4't -

,rr, high-voltage ratio of transformer;

iy, lower-voltage transformer winding not connectedto a generator;
vJ surgeshaving steepwavefronts and surgeshaving long duration.

Ifthere are indications that the amplitude ofthe initial voltage spike should be reduced,this can be done effectively
by connecting non-inductive capacitors betweeneach phase and earth, preferably at the transformer terminal, by
means of low-inductance connections. Attention is drawn however to the possibility of increasing inductively-
transferred overvoltagesby additional capacitors.
The longer-duration transferenc€generally takes the form of a unidirectional voltage with superimposedoscil-
lations having a frequency of severalkilohertz and if reduction of this is necessary,consideration should be given
to the addition of surge arresters.
However, voltage division between the reactancesof the generator transformer and the generator normally
ensuresthat the amplitudeof the longer-durationtransference doesnot warrantthe useof surgearresters.For large
generatorinstallations,surgearresterson the generatorare not generallyacceptableand checkcalculationsshould
be made.In so far as surgearresterscan be readilyappliedto smallinstallations,thereis no necessity in thesecases
for calculationsof transferenceto be hade. If the generatortransformercan be energizedfrom the high-voltage
systemwhen the generatoris disconnected,voltage division betweengenerator and transformer doesnot occur and
consideration should be given to the higher amplitude of the longer-duration transferenceaffecting that part of the
lower-voltagecircuit which remains connectedto the transformer.

The etrectsofthe applicationof longer-durationvoltagesurgesto the higher-voltage windingofa transformerand

their transferenceto the lower-voltagewinding when an earth fault existson the higher-voltagesystemand when the
neutralofthat systemis earthedthroughan arc-suppression coil, or ifit is isolated,are subjectto the sameconsider-
ations as describedfor Category I equipment.

C) Srr,ecnor oF TNSTJLATToN
LB\tsL

The choiceof whetherto uselist I or 2 of Table I of IEC Publication71-1and whetheradditionalovervoltage

protectionis necessaryshouldbe basedin the flrst placeon serviceexperience with similarinstallations.It may also
be useful to make neasurementson a similar existinginstallation, using a low-voltage impulse-injectionmethod.

For a large generator-transformerinstallation, for which the necessarydata concerningthe transformer and the
protective equipment are available, it will be useful to calculate the overvoltagesliable to be transfered and to
comparethe resultswith the appropriatewithstandvoltagesof the equipmentto be protected.This is normalll-
advisableonly for direct connectionsbetweengenerator and transformer and for low-voltage tertiary windings on
largesystemtransformers.If a circuit-breakeris installedbetweena generator-transformer and its associatedgen-
erator,considerationshouldbe givento the two casesof whenthe breakeris closedand whenit is open,althougha
load is usually connectedto the lower-voltagewinding of the transformer wherebytransferred overvoltagesmay bc
reducedeven in the latter case.
Severalmethodsof calculation havebeenpublished and, on the whole, theseseemto give similar results. Although
no absoluteaccuracycan be claimedfor any method of calculation,comparisonbetweencalculationand exper-
irrental results on a variety of installations has shown satisfactory agreement.It is therefore deemedappropriate to
illustrate a nethod of calculation by referenceto two numerical examples,covering Cat€goriesI and 2 respectively.
Theseexamplesare given in ClauseA'2 of Appendix A.

21,3 Equipmmt connecteddirectly to an overheadline

Equipmentinstalledin a substationconnecteddirectlyto an overheadline is subjectto direct or indirectlightning

overvoltages. Suchequipmentshould,as a generalrule, complywith the rated lightning-impulsewithstandvoltages
specifiedin list 2 (TableI) or SeriesII (TableII) in IEC Publication?l-1.
All equiprrentand, in particular,transformersin such locationsrequireprotection by surgearrestersor spark
gaps'Taking into accountthe flat impulsesparkovervoltage-timecharacteristicof a transformerwinding, trans-
 -49-

formers should preferablybe protectedby non-linearresistor-typesurge arrestersin areasof intenselightning

activity. In areasof moderatelightning activity, expulsion-typelightning arresterscan be used.Where lightning
activity is slight,protectivespark gapshaveproved adequate,particularlywherethe transformeris connectedto a
line with earthedcross-armsor wherethe transformer is designedto withstand steep-frontedchoppedwaves.

The bushings of circuit breakers, instrument transformers and substation insulators having curved impulse
sparkover voltage-time characteristics,can be effectively protect€d by existing protective devices on the trans-
formers. Considerationsrelative to the distance between protective device and protected apparatus developedin
Subclause16.1are alsovalid here.
In areasof moderate or low lightning activity, equipment having rated lightning-impulse withstand voltages in
acrordancewith list I of Table I of IEC Publication71-1may be usedbut, in that case,carefulattentionmust be
paid to adequateovervoltageprotection. In a systemthe neutral of which is earthedthrough a low resistance,surge
arrestersor spark gapscan be usedfor this purpose.In a systemthe neutral of which is earthedthrough an arc-
suppressioncoil, adequateovervoltageprotectionmust be provided.If surgearrestersare used,those which can
withstand repeatedoperations during the persistenceof ficing earths are recommended.

ln the absenceof any overvoltage-protective device,lightning surgesimpressedon an overheadline are limited

only by sparkoverson the line at the weakestpointswhich the surgesmeetduringtheir propagation.If not correctly
localized,suchspa;koverscancausedamageto equipmentasa resultofsurgereflectionsbetweenthe point of spark-
over and a vulnerable apparatus,such as a transformer winding.

In the caseof a substation with a number of lines normally connectedto the busbars,the surgevoltage arising at
apparatusin the station(seeSub-clause
the busbarsis likely to be sufficientlyreducedso asnot to overstress 7.1).

However, such a solution (no overvoltage-protective device) may be acceptable in practice on overhead supply
systemsin regions of very low lightning activity, at least if equipment according to list 2 of Table I of IEC Publi-
cation 71-1 is used.

2L.4 Equipmmt connectedto an overheadline througha cable

Insulationco-ordination,in this case,is not only concernedwith the protectionof the substationequipmentbut
also with that of the cable.
When a lightning surgepropagatedalong an overheadline impinges on a cable, it breaks up into a reflectedwave
and a transmitted wave. The amplitude U, ofthe surgeflowing along tlle cable is given by:

., : 2Zz
u' z, +z; u'
The amplitude Lrr,ofthe reflectedwave is given by:

a " : ,z"-
n i uz,,
where:
Ur : amplitude of surgevoltage on overheadline
ZL : surgeimpedanceof overheadline; in practice,400O to 500O
Zg : surgeimpedanceof cable;in practice25O to 50O, but for sometypesof cableit may be aslow as 5 C)

This initial surgeis reflected at the station end of the cable in accordancewith the effective surge impedanceat
the stationbusbar.Subsequent reflectionsat the cableterminalscontinueto be governedby the aboveequationswith
due regardto the fact that 4 and Z, invariabll refer to the wavewhich impingeson a point of reflectionwhile Ut
and Z, refer to the transmitted wave and UE refers to the reflectedwave.
 - 5l -

Provided at least one further cable of a few hundred metres in length is permanently connected to the busbar,
the surgevoltag€ to which the cable and the station equipment are subjectedis notably lower than that on the line
on which the surgeoriginated and this reduction is all the greater,the lower the surgeimpedanceofthe cable.

For a stationto which at leasttwo cablesare connected,a decisionon the adequacyof equipmentaccordingto
list 1 of Table I of I EC Publioation71-l or the needfor overvoltageprotectioncan be determinedfrom the above
equation.
However, in the caseof a terminal station, the ultimate surge-voltageamplitudes developedat the cable termin-
ations asa result of successivereflectionsare a function of the amplitude and duration of the original lightning-surge
voltage on the line, the length of the cable and, if the stroke is fairly close to the cable, also the reflections from
the point of strike.For lineswith fully insulatedcross-arms, the resultingvoltageamplitudesare so high that, even
using substation equipment and a oable with lightning-impulse withstand voltages according to list 2 of Table I
of IEC Publication71-1, surgearresters,nrrt be used at the line/cablejunction. As an example,the maximum
cablelengthsare plotted in Figure 3, page108,for which the cableand the substationequipmentcan be protected
by surgearrestersat the line/cablejunction only; the figure demonstrates the considerablebenefitof the useof a
cableoflow surgeimpedance.The protectionis fully eflectiveagainstdirectand indirectlightningimpulsesand back
flashoversprovidedtheseoriginatea few spansdistantfrom the line/cablejunction. If the cablelength exceedsthe
valu€sindicatedin Figure 3, additionalsurgearrestersare requiredat the substationend of the cable.If surgear-
resterswith sparkovervoltageslower than those specifiedin IEC Publication99-1 are used,the cable lengths
indicatedin Figure3 canbe increasedin proportionto the differences indicatedby comparing,for example,a 10.5kY
with a 12 kY surgearrester.

However,ifthe cablesuppliesonly a terminaltransformer,the incidentwaveis doubledin amplitudeat the trans-

former. As a resultof successivereflectionsat both cableterminations,this yoltagebuild-upincreases
towardstwice
th€ amplitude until no further energy is supplied by the original surge.
For lines with earthed cross-armsfeeding a cable with terminal transformer, the impulse flashover voltage to
earthof the line insulationis only slightly higher than the correspondingvalue in list 2 of Table I of IEC Publi-
cation 7l-1. ln sucha case,surgearrestersmdl be requiredat the line/cablejunction and it may also be necessary
to use theseat the station end of the cable.

The foregoing considerationsapply to direct strokes a few spansdistant from the cable termination. Full protec-
tion against very close strokesis generally not possible.
In areasof moderateor low lightningactivity,protectivesparkgapscan be usedin placeof surgearresters.How-
ever,if the sparkgapsat the line/cablejunction are earthedthrougha low resistance (the usualcase)and ifthe cable
is terminatedin a transformer,dangeroussurgevoltagescan be developedacrossthe transformerwindings.The
spark gapsat the line/cablejunction should thereforebe earthedthrough a resistanceof severaltens of ohms.
equallingideallythe surgeimlxdance of the cable.Greatly improved protectioncan be achievedby installing ad-
ditional spark gapsacrossthe line insulatorson the fust and secondpolesin front of the line/cablejunction and,
in this case,the e4rthing resistaocesof theseadditional spark gaps are immaterial.

SECTION SEYEN _ CO-ORDINATION BETWEEN STRESSESAND WITHSTAND

IN VOLTAGE RANGE B

22. Selectionof the rated power-frequencyysithstandyoltsge and the rated lighhing-impulse witlstand yoltrge

Many considerations concerningvoltagesin rangeA alsoapplyto rangeB. However,the varietyofequipmentand

locationsis not as great as in rangeA.
 -53-

In Tablc III of IEC Publication71-1,one to three valuesof the rated lightning-impulsewithstandvoltageare

associatedwith eachvalue of U-. For eachof thesevalues,there is a corresponding singlevalue ofthe rated power-
frequencywithstand voltage. Thereforethere will be one choicefor the rated power-frequencywithstand voltage and
the rated lightning-impulse withstand voltage.
The choice betweenthe possiblealternativesfor U- above 72.5 kV must take account of:
- the neutral earthing conditions;
- the existenceof protective devices,their characteristicsand their distancefrom the equipment considered.

The safety factors normally employed in the application of surge arresters in range B are to be found
in Subclause16.1.

SECTION EIGHT - CO.ORDINATION BETWEEN STRESSESAND WITHSTAND

IN VOLTAGE RANGE C

23. lrsulation designwith r€sp€ctto power-ftequencyvoltagernd temp,oraryoyervoltrges

For this range of voltage, power-frequencytests are to be specifiedby the relevant apparatuscommittees,in ac-
cordancewith the considerationsin Section Five, and taking into account that the temporary phas€-to-earthover-
voltageswill not usually exceed1.5p.u. for I s on eachoccasion.

24, Insulation d€signwith respectto switchingrnrl lightning overvoltages

IEC Publication7l-1 proposestwo methodsfor co-ordinationof insulationin respectof srvitchingand lightning

overvoltages:a conventional and a statistical method.

24.1 Conventionalmethod

The conventional method is basedupon the establishedconceptsof maximum overvoltagesstressinginsulation

and of minimum strengthofthe insulation(deflnitionsin ClausesZ and 27 ofIEC Pubtication7l-l). The notion of
rrinimum strength and that of maximum overvoltage are rather arbitrary since a rigorous rule can seldom be fol-
lowed in the evaluation of the upper and lower limits of the insulation strength and overvoltage value, which are
intrinsicaLly random variables.
Insulation is selectedin such a way as to achievea sufrcient margin betweenthe maximum overvoltageand the
minimum strength. This margin is intended to cover the uncertainties of the designer in the evaluation of
the maximum overvoltageand of the minimum strength, and no endeavouris made to assessquantitatively the risk
that insulation may break down.
The safety factors normally employed in the application of surge arresters in range C are to be found
in Sub-clause16.1.

2.4.2 Sratistical merhod

The statistical method attemptsto quantify the risk of failure for useas a safetyindex in insulation design.
 -55-

Rational insulationdesignof a transmissionsystemought to be basedon the minimum of the installationcost

plus the capitalizedyearly operationalcost and yearly cost of failure, the latter being calculatedas the estimated
cost of failure of insulation rrultiplied by the averageexpectednumber of insulation failures per year.

In order to evaluatethe aveiageexpectednumber of failures per year of a piece of insulation located at a given
point ofthe systemin consequenoe of overvoltages, all the eventsgiving rise to overvoltageswhich may afect insu-
lation design should be taken into consideration. Then for each type of event considered, the yearly frequency
of occurrenceand a statistical distribution of the relevant overvoltageamplitudes are required.l

However,a processof insulationdesignand co-ordination,asoutlinedabove,involvestoo many difrculties.The

statistical approach consideredhere is therefore restricted to checking that the risk of insulation failure due to any
foreseeable type of eventcausingovervoltages in the systemis within acceptablelimits. Theselimits dependon the
frequenoyof occurrenceof the type of event and on the consequences of the failure of the piece of insulation under
consideration.
Fortunately, the types of eventwhich are signif.cantin insulation designare generallysumcientlyfew in number as
to allow an analyticalapproach.For instance,the insulationlvithstandto srvitchingsurgesof many piecesof in-
sulation of equipment in a systemis, in general,determinedprimarily by reclosing overvoltages,

lf the frequency distribution of overvoltagescausedby a given type of event and the corresponding insulation
strength are known, the risk of failure can be expressednumerically, as will be shown below.

L€t the withstand strength of a given piece of insulation within a given time interval AZbe definedby the prob-
ability PT(LDof disruptive dischargeof the insulation when it is subjectedto an overvoltage of value U (Figure 4,
page109).Furthermore,let the distributionof the overvoltages stressingthe samepieceof insulationfor the specific
type of eventconsideredbe defined by the probability density/o((D. Then theprobabilitythatan overvoltageofvalue
comprisedbetweenU' and U' f dUmay occuris/.(U') dU. The probability densityof failure ofthe insulationdue
to an overvoltage of value U' is therefore the product of the probability density that an overvoltage of value U'
may occur and the probability that the insulation may fail under an overvoltageof value u'. Thus:

dR : f,(u,). Pr(U,)du (o
The probability of failure for a value of U taken at random, i.e. the risk of failure R for an event of the type
considered.will then be:

u - | f,(u) . Pr(u).du (7)

0

This expressionshowsthe generalprincipleofthe methodby which the probability of failure may be assessed.
It
assumesthat/o(y) and P1(Lr) are uncorrelated.

lVole, - In principle, formula (7) applies to a siele-phase piece ofinsul,atiotr only. If several piec€s of equipmedt, coDn€cteditr parallel
on the same phase, are subjected to the same overvoltage th€n it can be assu{red that the overall risk is equal to that of a single
pi€c. of equiplnett multiplied by the number of pieces io paralel. This is valid if we take hto consideration the fact that the
acc€ptablerisk offailure for substation iDsulatioD is lrually very low.

I It is evident that the amplitud€s ofall the overvoltages oc.uring in a system cainot b€ combined in ode distdbutior, but that ody ov€r-
voltages identified by the samelocatioD and causecan be considered as statisticslly homoggdeous.Actually, siDcethe overvoltage s€verity
differs for wavdhap€s which are companble resp€ctiv€lywith a lightding impulse and with a switching impuls€ (s€€Note to Clauses2l and
22 ofl EC Publicatiou 7ll), tl€ overvolrage amplitude can be said to be homogpneousonly ifidentifed by the sarF locatio!, cause and
shape. However, ov€rvoltag€s due to the same cause at a given location have broadly a similar shapeand, thercforc, those identified by
the samecauseand location may, for the sake ofsimplicity, b€ r€garded as homoggneous.Ifproblefi$ ofstandardizatioo ofthe equipmcnt
of atr efltirc network are to be dealt with, an extensioo of the con@pt of a homogeneous group of overvoltag€s 4€eds to be considered.
In this cas€,a group ofovervoltagps rnay be said to be homogen€ousif tie overvoltagps occur in similar locations of the system due to the
samecause.For exaErple,the reclosing overvoltages on the busbars (seoding.€Nrd)of say subBtation of the systEnrmay be coNid€red as a
horDogeoeousgroup of overvoltagEs.
 Sonretifires it is trec€ssaryto evaluate the risk of failure of at least on€ phas€ of a thr€e-pbas€ section of the system followitrg
a switchiqg op€ratioo (e.9. a closiDg operatioo), This risk may be obtained by multipllng by three the risk evaluat€d according to
formula (7) ifthe probability deqsity/o(U) of the overvoltages rnay b€ assuned to be equal oo all three phas€s.
An altemative nrethod is to c€tablish the overvoltage probability density /o(Lr) by considering only the highest valu€ ofthe over-
voltagEscaus€dotr the tfuee phasesby a swirchi4 opemtion. Thed the dsk of failure is evaluated by makiog use of forrnula (7).

The form€r approaph gives risk valu€s higher than the aptual ones; the latter lowcr. Obviously the two approaches give resultJ
ditreling by l€ss than 3: 1.

The mathematical model choset for deflning the severity of an overvoltagein formula (7) is based upon a few
simplifications. In fact, the following assumptionsare made:

a) Peaksother than the highest one in the waveshapeof overvoltagesare disregarded.

6/ The waveshapesof all of the lightning impulses and the switching impulses belonging to the distribution
defined byl.(Lf are assumedto be identical to the waveshapeof the highest ovenoltage.
cJ The highestovervoltagepeaksare assumedto be all of the samepolarity, although, during switching,a balanced
distributionduring the two polaritiescan be assumedin practice;this assumption,which leadsto the calculation
of greaterrisk than the actual risk, makesit possibleto take all of the piecesof insulation into account, regardless
of the differenoesof performancebetweenthe two polarities. To be on the safeside, the more severepolarity will
be used.

As regardsswitching overvoltages,which are the transient overvoltagesof predominant importance in the insu-
lation designof EHV systems,assumptiona, is. such as to give a calculated risk of failure lower than the actual
risk. Assumption c) results in a calculated risk higher than the actual one for the reasonalreadyindicated and since
the standard waveshapesare so chosenasto establishthe lowest withstand of apparatus(seeClause51 of I E C Publi-
cation 71-1).

In general,considering the opposite effectsof the assumptionsmade, the risk of failure calculated by means of
formula (7) grves risk values greater by about 0.5 to 1 decade (3 to 10 times) than the actual values. Norrrally
formula (7) is therefore conservative. As said above, fornula (7) can be applied for all the specifictypes of event
signiflcant in insulation design.

Furthermore, it is clear that the accuracyin the calculation of the risk failure greatly dependson the accuracyin
the determinationof the overvoltages and the disruptivedischargeprobability of insulation(seeAppendicesC and
D). Sinceaccuracyoftheseis seldomsatisfactory,the accuracyofthe calculatedrisk offailure canbecorrespondingly
poor.

However, the risk of failure has a precise physical meaning (contrary to the safety factor). By making use of
statistical methods,it is therefore possibleto co-ordinate the security levels of the various parts of the systemaccor-
ding to the consequences ofa fault. Furthermore,it is possibleto carry out sensitivityanalyses(e.g.the effectof a
changein the overvoltageseverityor insulationwithstandcapabilityon the probabilityoffaults). Statisticalmethods
do therefore enablethe engineerto take a decision on a rational basis.

According to the statistical method, insulation is selectedin such a way as to obtain a calculated probability of
failure lower than, or equal to, a pre determined value that characterizesthe required safety level. Referring to
Figure 4, page 109, a changein the insulation level shifts the curve representingthe dischargeprobability of the
insulation Pr(t/) along the U axis with a consequentmodification of the shadedarea,{ which representsthe pro-
bability of failure R for a random value of U.

The statistical approach may require successiveseriesof tentative designsand evaluationsof risk, until a solution
is found that correspondsto the predeternined risk.

Formula (7) can also be applied to determine the probability of failure of an insulation protected by spark gaps
or surgearresters,ifPl(Lf is takenasthe dischargeprobabilityof the insulationin presenceof the protecdivedevice.
If the time-to-dischargeof the protective devicecan be consideredalways shorter than that of the insulation to be
protected,an equallyvalid and simplermethodis to useformula (7) and to take/,(t/) asthe overvoltag€probability
densitymodifiedby the protectivedevice(seeAppendixC).
 -59-

The useof digital computersfacilitatesevaluationof the risk of failure, ard thereforethe insulationdesign,once
the statisticalovervoltagedistributionand the dischargeprobabilitycurveof insulationare known.

24.3 Simplifed stqtistical melhod

Sensitivity aialyses and ready evaluations of the risk of failure can be made on the basis of simplified statistical
methodsin which the cirlculations are performed onceand for all by making somegenerallyacceptableassumptions
concerningthe mathematicallawsby which the actualdistributionsofthe overvoltages and the dischargeprobability
of insulationare represented e.g.by assumingthem to be Gaussianwith known stardard deviations.

With theseassumptions, the cornpletedistributionof overvoltageand the dischargeprobabilitycurveofinsulation

can be defi:redby onepoint only corresponding to a givenreferenceprobabilityand called,in IEC Publication71-1,
"statisticalovervoltage"(Clause23) and "statisticalimpulsewithstandvoltage'' (Clause26), respectively. The risk
of failure can be correlated with the margin betweenthesetwo valuesso that the approach becomesrather similar to
that of the conventional method.

Figure 5, page111,givesa graphicalexplanationof the method.Figure 5a showsfrequencydistributionsof over-

voltageand insulationstrength,wherethe statisticalovervoltageis indicatedby ys and the statisticalwithstandvolt-
ageby Lr,p.In Figure 56, the overvoltagedistribution and the electricstrengthdistribution are superimposed for
threevalues1.0, 1.2and 1.4 of the statisticalsafetyfactor ), taken as equalto the runo Ualu,n.The correlationbe-
tween the statistical safetyfactor 7 and the risk of failure R is given in Figure 5c.
The referenceprobability of the overvoltagesis chosenin this guide asequal to 20lo.The reasonsfor this choiceare
givenbelowin this clause.As regardsthe referenceprobability of the withstandvoltage,the 90% valuewaschosen
in IEC Publication71-1for the reasonsouflinedin Clause10 of this guide.
As regardsswitchingsurges,Figves 6,7 and 8, pages112,113and I14, illustratethe relationshipbetweenthe risk
of fai.lure and the statistical safety factor for air insulation in different cases.
The dischargeprobabilitycurveof insulationis assumedto be Gaussianasstatedin Clause10with /c(Formula (5)
in Clause10)equalto I andor (formula(2) in Clause10)equalto 6%o,8Yo and 10%(Figures6, 7 and 8 respectively).
Ifft is takenas differingfrom 1, the statisticalsafetyfactor givenfor k : 1 shallbe multipliedby l/k,

Figure 6 is applicableto laboratory conditions,while Figure 7 is generallyapplicableto serviceconditions.

Figure8 may be usedfor particularlysevereconditions(high valuesofon in formula (2)).In all threecases,the over-
voltagedistributionsareassumedto be Gaussiantruncatedat threeand four timesthe standarddeviation06,or not
truncatedand with standarddeviationo" equalto 10o/o,l5To znd 20Vo.Figur€s6 to 8 givethe averagecorrelation
betweenthe statisticalsafetyfactor and risk of failure as well as the upper and lower envelopesof the correlations
obtainedwhenconsideringthe nine overvoltagedistributionsresultingfrorr all the possiblecombinationsof values
of standard deviation and upper truncation point.

The choiceof a Gaussiandistributionto definethe overvoltageseveritydoesnot meanthat other disFibutions

(e.g.extremevaluedistributions)may not give betterapproximations,but simplythat Gaussiandistributionsmatch
actual distributions reasonablywell over the range of interest.

The correlation between the statistical safety factor and risk of failure appearsto be only slightly atrectedby
changesin the shapeof the overvoltagedistribution. This is due to the fact that the zyo value chosenas a reference
probability ofthe overvoltagesfalls in that part of the overvoltagedistribution which givesthe major contribution to
the risk offailure in the rangeofrisk considered.If, on the contrary,a much lower or highervaluewerechosen,the
influenceofthe shap ofthe overvoltagedistribution would be very pronounced.

Figures6 to 8 give the risk of failure of a pieceof single-phase

equipment(e.g.a post insulator).If the risk of
failure of severalpiecesof equipmentis required,referenceshouldbe madeto the noie to Sub-clause 24.2above.
For example,if the numberof single-phase piecesofequipmentat a line entranceis equalto 2l (7 on eachphase)
and the risk of failure of eachfor a three-phasereclosingis 10a,then the risk of failure for the whole line entrance
 -61 -

will be closeto 2.1 ' 10-8.This simplifi.edcalculationis only valid under the condition of low risk (seenote of Sub-
clause24.2) and if the number of parallel piecesof equipment is not too great.
Extension of this method to overheadlines is possiblebut presentsproblems particularly for long lines, with which
this guide doesnot attempt to deal.

This method may be appliedlo IEhtnW szrges,keepingin mind the practical limitations indicated in Appndix D.
In faot, it is not usedat the presenttime. Analogous correlations betweenthe statistical safetyfactor and the risk of
failure are givenin Figures9, l0 and 11,pages115,116and 117,for comparisonspurposesonly, for air insulation
submitted to lightoing surges.
Assumptions similar to those of the previous sub-clauseshave been made as regards the dischargeprobability
curve of insulationto lightning surges.

The lightning overvoltagedistributionsare assumedto be Gaussianand not truncated,with standarddeviations

equalto 40% a'Jid60V,.It is thought that suchdistributionsapproximateactuallightningovervoltagedistributions
quite well around the 2% value.
Figures 9, 10 and I I give the correlation betweenthe statistical safetyfactor and the risk of failure for borh over-
voltage distributions and for standard deviationsof insulation strengthequal to 3To, 5o/oand7o/orespectively.

lvore' - The corr€lations between tie statistical saf€ty factor ard the risk of failure givea above apply to self-rettoring insulation oDly.
How€ver, they can be considered acceptablefor th€ entirc equipmetrt io most cas€sfor the following reasons:

The testsdescribedin Clauses52 and 53 of IEC Publication 7l-l areintended to ascertainthe impulse withstand strength of self-
restoring equipment and combined equipment respectively.Such testsdo not altow any dischargeon the non-self-restoringparts
of the apparatus. Consequently,when both the self-restoringand non-self-restoringparts of insulation of the same apparatus
are designedon the basis of the samerisk of failing the test, the non-self-restoringparts will have an inherently lower discharge
probability than the self-restoringparts in respectof overvoltagesof the sameamplitudes as those of the impulsesapplied during
the tests.
Consequently,it can be stated that risk of failure of the non-self-restoringparts of self-restoringequipment (testedaccording to
Clause 53) is lower than that of the self-restoringparts if the major contribution to the overall risk of failure is given by over-
voltagesin the range of U56'tr t 2 o.
For combined insulation equipment, the non-self-restoringparts will be designedfor a low risk of failure at test voltage. In the
caseswhere the major contribution to the failure risk (calculatedon the basisof the statistical withstand voltage) is given by over-
voltagesaround the test voltage, the evaluation of the risk of failure can be carried out as for self-restoringinsulation.

The foregoing considerationsnaturally presupposethat wave chopping during sparkover by a self-restoringpiece of insulation
doesnot causeseriousstressesin the non-self-restoringinsulation of equipment and doesnot accelerateageingof insulation.

25. Block-diagran of the insulation designand co-orrlinationof an elechical installation

Most predictionsor analysesof systemovervoltagelevelsassumethat a pieccofequipment(e.g.a circuit breaker,

etc.)will operateasdesigned.In othercases,an arbitrarylimit shallbe ptacedon the "credible" severityofsurges,as
is often done in the caseof lightning. It is obvious that surge levels based on such assumptionswill sometimesbe
exceeded.
Whether or not it is necessaryto take this into account dependslargely on the consequenceof failure resulting
from such abnormally high overvoltages. For example,the consequencesof transformer or reactor failure are so
seriousthat their insulation co-ordination must usually provide for evenextremecontingencies.This is achievedby
applying surgearrestersat their terminals. There are other types of equipment such as post insulators, disconnecting
switches,etc., wherethe consequen@offailure is not so seriousasto warrant the application of surgeaffesters.

In developing a sequenceof insulation design and co-ordination of an electrical systemin the form of a block-
diagram,it is convenientto differentiatebetweena caseI (surgearresterprotection)and a caseII (no surgearrester
ot remotesurgearresterprotection).A block-diagramillustratingthe methodto be adoptedis shownin Figure 12,
page119.
 -63-

The first stepin insulationco-ordination(block 5),commonto all tlpes ofequipment,hasthe purposeof ensuring
the ability of equipmentto withstandpower-frequency voltageunder normal operatingconditionsand during tem-
porary overvoltages.The systemengineershall specifyan equivalentpollution severitytest level for insulation
susceptibleto contamination;howevgr,no specialspecif.cations will be given for insulationsusceptible
to ageing
(seeClause23).

If the expectedphase-to-earth temporaryovervoltages (block 3) are more severethan the overvoltagestaken into
consideration by the relevant Apparatus Technical Committee in specifyingthe power-frequencytests discussedin
Clause39 of IEC Publication71-1,it will be necessary to specifydiferent voltagelevelsor durationsof the test or
to adopt suitablemeansor operationalproceduresto reducetemporaryovervoltages in the system(feedbackfrom
block 5 to block 2, dashedline).

Insulation designas regardsoperatingvoltagesand temporaryovervoltagesleadsto a certainwithstandof the

equipmentto both switchingand lightningimpulses.For instance,if a givenwithstandsalinityis requiredfor a post
insulator,a minimum distancein air is obtrined which variesaccordingto the post insulatortype. The equipment
will thereforeexhibit a certainwithstandto switchingsurgesdueto requirementsimposedby the operatingvoltages
and temporaryovervoltages. Theseinfluencesare indicatedin Figure 12by dottedlines(e.g.from block 5 to block 7
and block 8).

to considercaseI and caseII separately.Examplesof selectionof the rated switchingand

Then it is necessary
lightningimpulsewithstandvoltageare givenin AppendixD.

As regardsapparatusof the first type (caseI), choiceof the rated switchingand lightningimpulsewithstandvolt-
agesis usuallymadeas follows:

1) Choosethe ratedvoltageof the surgearresterson the basisof the temporaryovervoltages (block 6); thus the
protectivelevelsof the arrestersunder switchingand lightning impulseswill also be determinedat leastwithin
c€rtainlimits.

2) Choosethe rated switchingand lightning impulsewithstandvoltagesof the apparatuson the basisof safety
(seeSub-clauses
factorsdictatedby experiences 15.1and 16.1).

3) Adopt suitablemeansin systemdesignor suggestsuitableoperatingproceduresto reducetemporary over-

voltagesif an economicalincentiveexiststo reduc€insulationlevels(start againfrom block 2).

This procedureof selectingthe rated switchingand lightning impulsewithstandvoltageof the apparatusdis-

regardsthe severityof the actualswitchingand lightningovervoltages by which the apparatusconnectedin parallel
with the surgearrestersmay be stressedsinceit is basedon the protectivelevelofthe surgearrestersonly.

The insulationlevelsof the apparatusof caseII are usuallychosenaccordingto the following steps:

1) Choosethe ratedswitching-impulse withstandvoltageof the apparatuson the basisof an acceptablerisk s hich

canbe estimateddirectlyon the basisofthe expecteddistributionsof the overvoltages and ofthe dischargelohages
(statisticalmethod),or by meansof the correlationsgivenin Sub-clause 24.3 betweenthe risk of failure and the
statisticalsafetyfactor (simplifiedstatisticalmethod).Actually, thesecorrelationsapplyto self-restoring
insulation
only, but normallythey can be consideredacceptablefor the entireequipment(seenote in Sub-clause 24.3).

lvore. - ProtectioD against switching surgesafforded by surge arresters installed close to €quipnrent in case I and spark gaps installed at
the line-€ntraoce can often be discounted as regards caseII apparatus for the following r€asons:
a, Most of the types of €quipm€ot belooging to case II (especially lin€-ento4ce appantus) c4n at times be isolated ftom the
surgp a$esters installed in the station to prot€ct caseI appantus.
.
,, With the present technology ofsurge arresters,the protective level against switching impulses is often gre3ter than, or equal
to the highest switching overvoltags which cad occur with corect behaviour of system apparatus. Insulation must therefore
b€ d€signed to withstaod these ov€rvoltages.
c) Spark gaps cannot provide a substantial degree of protection against switching surgps if undesircd spa.ko!€N are to be
avoided, This point is given in Appendix C.
In Figu!€ 12, th€ block bet\reen block 4 and block 8 is therefore iodicated by dashedlioee.
 -6 5-

2) Adopt suitable means in system design to reduce switching overvoltages,if this is possible and if an economical
incentive existsin reducing the rated switching impulse withstand voltage of apparatus. No economical incentive
may exist in reducing insulation; for instance, if the withstand. to normal operating voltages and temporary over-
voltages calls for "higher insulation" than switching surges(dotted line between block 5 and block 8).

3) Verify that the rated lightning impulse withstand voltage,corresponding(in Table IV of IEC Publication 71-1)
to the rated switching impulse withstand voltage determined above (Items I and 2), guaranteesa satisfactory
performance of the apparatus under lightning overvoltages. This should be done on the basis of the expected
distribution of the lightning overvoltages and of the discharge voltages or by means of the correlation given in
Sub-clause24.3, but for the sake of simplicity it is often done on a conventional basis (seeAppendix D).

Consider that only the highest value of rated lightning impulse withstand voltage of each line should be used for
apparatusnot effectivelyprotected by a surgearrester(seeClause49 of IEC Publication 71-l).

4) Provide meansto reducethe amplitude of lightning surges1 (feed-backfrom 12 to 10 on the block-diagram of

Figure 12) or choose a rated lightning impulse withstand voltage higher than that determined on the basis of
Table IV of I E C Publication 7 L-l if too high a risk of failure to lightning overvoltagesresultsfrom Item 3. In the
latter case, the value of the rated lightning impulse voltage shall be selectedfrom the series given in Item b ) of
Clause46 of IEC Publication 7l-1.

1 Consider possiblechangesin line design such as tower footing and shielding wires, install surge arrestersother than those intended to
protect case I apparatus, make use of protective spark gaps. The actual degreeof protection provided by spark gaps is discussedin
Appendix C.
 -67 -

APPENDX A

SURGE TRANSFERENCE THROUGH TRANSFORMERS

A1. Derivation of expressions

41.1 Initial mpacitive voltagespike

During the initial period ofabout I prsunder the conditions of a lightning surge,thc transformer can be approxi-
matelyrepresented as shownin Figure 13a,page120,as a capacitancevoltagedivider of ratio s whereJ < l. ff q
is the sum of the capacitancesof the higher-voltage and lower-voltage arms of this divider, the initial transference
can be simulatedas in Figuresl3b, l3c and 13d,page120,by a seriescircuit comprisinga source[/o : rq, a ca-
pacitance C, and the capacitanceor resistanceof the external lower-voltage system. Ur is the surge voltage on the
higher-voltageterminal during the initial period. The sourcevoltage Uois the open-circuit transferred voltage.

If tho extemalsystemcan be represented

by a capacitanceC", asin Figure 13c,the equivalentcircuit is a voltage
divider having the ratio:

Ct
(8)
Ct* C"

If' duringthe initial period,the impedanceofthe externalsystemis the surgeimpedanc€ofa cableor the resistance
ofthe load,this systemcan be represented by a resistanceR asin Figure l3d. Typical valuesare from 10O to a few
hundredohms.The transferredsurgevoltageis then dependenton the steepness as well as on the amplitudeof the
surge.For high valuesof R, the initial voltage is approximately Uoand for low valuesof R, it is given approximately
by Ur" : s,9RC,,where ,Sis the maximum steepnessof the surgein volts/second.

The aboveexpressions do not take into accountthe effectof superpositionof the surgevoltageon the power-
frequency voltage. Allowance can be made for the power-frequenry voltage by substituting for 4 the actual peak
voltage y! and by introducingthe factor p. For a star/deltaor delta/starconnectedtransformerthe value ofp is
typioally about 1.15.For a star/staror delta/deltaconnectedtransformer,the value ofp is truically about 1.05.
However,slightlyhighervaluesthan thesemay be encountered. For switchingsurges,the valueofp can be taken as
unity. The amplitude Un will be limited to the front-of-wave sparkover value of the lightning arrester or spark gap
on the higher-voltagesideof the transformer(seeIEC Publication99-1).
The amplitude ofthe initial voltage "spike" on the open lower-voltageside is given by:

Uw : sPUo (9)

For a transformer without eiternal connectionsto the lower-voltageterminals, the value of factor s can rangefrom
0 to at least0.4, dependingon the winding arrangement.The valueof J can be measuredin a low-voltageimpulse
responsetest (e.9. with a recurrent-surgeoscillograph). Values of C, generally lie in the range of l0-8 to l0-e F.

.iYole. - The values ofx and ct arc difrcult to cilculate for new d€sigds of transfomer, atrd the maDufacturer caE oDly be exp€cted to
givc a rough €stimate without guarantee.

The value of Ur" should be comparedwith the appropriate impulse test voltage of Table I, list I or 2, or Table II
of IEC Publication71-1.
The amplitude of the transferred surgemay be reducedby:
l) using a surgearrester with a lower front-of-wave sparkover voltage on the higher-voltageside;
2) addingcapacitance
betweeneachphaseand earth on the lower-voltageside;
3) addinga surgearresteron the lower-voltagesidebetweeneachphaseand earth.
For numericalexamples,seeClauseA2.
 -69-

A1.2 Inductively transferredvoltage

The transferenc€of e.m.f. by inductive couptng betweenwindings in a three-phasetransformer or bank of trans-

formers can be evaluatedfor any winding connectionsby consideringthe surgevoltage as a single-phasealternating
voltage.
The efrect of delta windings on the z€ro-phase-sequence
component of the single-phasevoltage should be taken
into account.
Figure 14, page 121,.showsthe resultsfor eight differentconnectionsof the transformer,assumingthe system
voltage ratio is unity.
As in the analogouspower-frequencyconsiderations,the transferred voltagesat the terminals are determined by
the e.m.f.'sand by voltagedivision betweenth€ internal impedanceof the transformer and the extemal circuit
impedance.The e.m.f.'scan be assumedto havethe samewaveshayras the surgeon the higher-voltagesystem,if
the effects of internal oscillations in the windings are neglected.The responseof the lower-voltage systemto these
e.m.f.'sis usuallyin the form ofa voltageof similarshapeto the incidentsurgewith a superimposed oscillation.

The.arnplitude of the inductively-transferredvoltage dependsalso on the voltage ratio and three-phaseconnec-

tions of the transformer and on the relative impedancesof the lower-voltagesystemand the transformer.
The voltage on the lower-voltage side of the transformer which, as is shown in Figure 14, can be a phase-to-earth
or phase-to-phase voltage,is givenby:

Uz : pqrUnlN (10)
where:

4 : responsefactor ofthe lower-voltagecircuit to the transferred surgee.m.f,

r : factor dependingon the transformer connections(seeFigure l4).
(,Ie : fpak voltage to earth on the higher-voltageside
if : systemvoltage (phase-to-phase)ratio of the transformer

ofthe surgeand on the electricalparametersofthe lower-voltagecircuit.

The valueof4 dependson the waveshape

For lightning surgeson a transformerhaving Category1 equipment(seeSub-clause 21.2),without appreciable

load connectedto the lower-voltageside,the valueof q is generallynot greaterthan about 1.3althoughthis value
may be exceeded.For switohing surgeson a similar systemwithout appreciable load, the value of 4 is not grealer
than about 1.8.

Generally,lower valuesof 4 apply if an appreciableload is connecteddue to voltagedivision betweenthe load

impedanceand the leakageinductance of the transformer (seenote below).

For Category 2 equipment (see Sub-olause21.2), voltage division takes place between the leakage inductancc
of the transformer and the subtransientinductance of the generator, and if these are about equal, 4 has the value
of about 0.9 for lightning and switching surges.

Values of r for a surge on one phase only (e.g. a lightning surge) and for equal surgesof opposite polarity on
two phases(one type of switching surge) are shown in Figure 14 for eight different three-phaseconnectionsof the
transformer.

The calculated value of U, is an estimate of the longer-duration transferred voltage, which in practice includes
longer-term effects of capacitive transference and transferred voltages corresponding to oscillations within the
windings. Its amplitude will be linited by the protective level of the surge arrester or protective spark gap. In the
caseof the former, this will be the higher of the standard lightning impulse sparkover value and the residual voltage
value for the lightning surges(seeIEC Publication99-l). For switchingsurges,exceptwhen the transformeris
connectedto a highly-inductiveload, such as an induction motor, overvoltageson the higher-voltageside may
be assumednot to generallyprovoke overvoltageexceeding3 p.u. (seenote).
 The value of Lr, should be compared with the peak values of the appropriate power-frequencytest-voltagesof
Table I or Table II of IEC Publication 71-1.Ii may be found necessaryto reducethe value of the lightning or
switching surge on the higher-voltage side of the transformer. Adding extra capacitance to the lower-voltage
side has little effect upon the amplitude of the inductively-transferred voltage, but it may be desirableto consider
the addition of surg€arresters.

Numerical examplesare to be found in Clause42.

Nor?. - Wheo the circuit is switched off on thc higher-voltagc side of a traDsfomer which may be loaded on the lo!€r-voltage sidb by
leactors or any other inductive load, then dangerous overvoltages may be attained under the most unfavourablc conditions of
operatioo but, in gen€ral, U! does not exceedthe peak value ofthe power-fr€quency t€st voltage since 4 is l€ssthao 1.0 on account
ofvoltage division betw€en thc transformer and the load hductanoe.

A2. Numerical examples

A2.l Exatnple A: Category l installation

Transformer. 145kY stax/l2 kV delta
Lightning impulsewithstandvoltage 145kV side : 550kV
Impulsetest voltage 12 kY side 60 kV (assuminglist I of Table I of IEC
Publication71-l)
Power-frequency test voltage 12 kV side : 28 kV
Surge arrester on higher-voltageside of the transformer:
rated voltage : 120kV
front-of-wave sparkover voltage : 463 kY
1.2/50impulsesparkoverand residualvoltage : 400 kV

a) Lightn@ surgeon onephase

Initial voltage "spike" for transformer disconnectedfrom load:

Uzo: spUo: 0,4 x 1.15x 463 : 213 kV

Assuminga ratio of 1.25beiweenthe impulsetest voltageand the serviceovervoltage,i.e. for a value no1ex-
60
- -II-
ceeding : 48 kV, the cable connectionsfrom the transformer would have a capacitanceof not lessthan C.
t-zi
suchthat:
c, 48
< : o'23
c,+ c" 213
hence C, > 3,44 Ct,

Having obtained from the transformer manufacturer a value of Ct : 10{ F, the cable capacitanc€per phase
shouldbe at least3.42 x 10{ F.
If the load is connected,then this will reduce further the peak voltage od the lower-voltage side.

Due to inductive transference:

Uz : pqrUllN

_ 1.15x 1.3 x 0.577x 400: 2g kV

t2.r
145
wherelr-_:12.1
t2
 28 kV givesa ratio of 1.4 with the 39.6kY peak test yoltage(28 {2j so that the insulationto list 1 of Table I
of IEC Publication71-l is satisfactory.

b) Switchingswges of oppositepohrities on two phases

Assuminga p.u. switchingovervoltageon two phasesof 2.5, and,p : l.O, lt : l.S andr : -] :1.15, then:
\/)

1.0x 1.8x 1.15x 145x lT x2.5

Uz:
1/l x I2.I
:50kV

In this case,the voltagepeak exceedsthe peak of the test voltage(39.6kV) of Table I (IEC Publication7l-l).
To allow for this, eitherinsulationoflist 2 of TableI (IEC Publication71-1)shouldbe used(i.e.havingtest voltages
of 75 kV lightning impulsewithstandand 28 kV power frequency(r.m.s.)or 39.6 kY peak), or switchingsurges
betweenphaseson the higher-voltagetransmissionsystemshould be limited, in which casea surgearrestermay
be selectedon the higher-voltagesideofthe transformerhavinga lower protectivelevel or alternativelyconsider-
ation may be given to fitting a surge arrester on the lower-voltage side of the transformer.

422 Example B: Category2 installation

Delta generator transformer 300 kV starlA kY
Lightning impulsewithstandvoltage 300kV side : 1050 kV
Power-frequency test voltage 24 kV side : 50 kY
Impulsetest voltage 24 kV side : 125kV (assuminglist 2 of Table I of IEC
Publication7l-l
Surgearresteron higher-voltagesideof transformer:
rated voltage : 240 kV
front-of-wave sparkover voltage : 900 kV
1,2/50impulsesparkoverand residualvoltage 785 kV

a) Lightning surgeson onephase

Initial voltage"spike" for transformerdisconnected

from Z kV system:

Uu : sPUo
. : O.2x 1.15x 900 : ZO7kY (maximum)

(the value for s having been obtained from the transformer manufacturer).
This valuewould obviouslybe too greatfor insulationon the lower-voltageside.
Assuminga ratio of 1.25betweenthe impulsetest voltageand the serviceovervoltage,i.e. a valuenot exceeding
: 100kV, would requirethe addition of externalcapacitance.
12511.25 In this case:

c, 100
c"+c,(-:o'48
hence C" > 1.07Cr,
The value of C, is obtained from the transformer manufacturer for a wave steepness,S : I 200 kv/p.s (seeIEC
Publication99-l).
The other methods listed in this Appendix for reducing the capacitive peak (use of surge aresters) may also be
considered.
 _ 75 _

The overvoltagedue to inductivetransference

is:

U2 : pqrUllN

1.15x1.0x0.577x785
: ---l2J : 41 kv

41 kY is less than the 70.7 kv (50 1/f) peak testsvoltage, and this gives a ratio of 1.7, which can be
consideredadequate.

b) Switchingsurgesof oppositepola ties on two phases

Assuming a p.u. switching ovenoltage on two phasesof 2, then the inductive transferenceis :

ff 1 . 0x 1 . 0x 1 . 1 5x 3 0 0 x 1 / T x 2
L't2 :

-45kV

w h e t epi : 1 . 0 q, : 1 , 0a n dr : + : 1 . 1 5 .
\/J

45 kV is about equalto the inductivetransference

of lightningsurge(41 kV).
 -77 _

APPENDIX B
VALIDITY OF THE TESTS IN CLAUSES 52, 53 AND 54 OF IEC PUBLICATION
7I-I

81. Confidence
limits of the test

Tests can provide only estimatesof varying accuracyof the true values of the withstand strength
of equipment.
An increasein accuracycan be obtained by an increasein the extent of the test. The extent of
testsmust however
be limited for reasonsof cost, the diminishing return of gain in the accuracyand possibledestructive
eflectson the
equipment.For thesereasons,IEC Publication7l-l prescribesthree diferent test methodsin Clauses
52,53 and
54 according to the type of equipment.
Inrtem a) of Clause52 dealingwith the 50% disruptivedischargetest,the note statesthat ,.There
are a numberof
proceduresavailable,and any of thesemay be usedprovidingthat the accuracyof the
determinationis within one-
half of the standarddeviationwith a conf.dence
levelof 95% . " It canbe saidthat thereshouldbe a 95% probabilitl.
that the 50% dischargevoltageof the equipmentat the time of the test is within the boundariesgiven
by the value
estimatedfrom the test,plus or minus one-halfof the standarddeviatron.
A test procedurefulfilling this requirementis the "up and down" test with 30 impulses.
The accuracyof the 15impulseswithstandtestis considerablyless.The 95% confdencelimits
for the probability
of dischargein this test are:

for 0 discharge approximately0 to 0.213

for I discharge approximately 0 to 0.32
for 2 discharges approximately0.015to 0.40
for 3 discharges approximately0.045to 0.4g.
for 4 discharges approximately0.0g to 0.54
for 5 discharges approximately0.12 to 0.61
Basedon 957oconfidencelimits, it canbe seenthat from a testwith 15impulsesonly it is not possible
to conclude
that the dischargeprobability is lessthan 0.10,or than the probability of withstandis higher
than 0.90,evenfor 0
discharge.When the number of dischargesexceeds4, the test is significantto prove the hypothesis
that the prob-
ability of withstandis lessthan 0.90.

82. Examination of difrerent m€thodsof t€st

Acceptingthe abovebasislimitationsof testswith a smallnumberofimpulses,the following examines

the validity
of diferent methodsof dischargetestingand the balancebetweenthe risk to the manufacturerand
the customer,
bearingin mind the practicalnecessityfor a limited number of impulsesand, for the manufacturer,
to designhis
products to have an economically acceptablerisk of failure on tests.
For the sakeof simplicity,it is assumedin this appendixthat the dischargeprobabilityp,(U) of
the diferent pieces
of equipmentin a givenpopulation(sametype of equipmenton which an impulsetest is to be
carriedout) follows
a Gaussianlaw with a constantstandarddeviationo, equalfor all the piecesof equipment.In
this case,ono para-
meteronly, ( givenin p.u. of or*see Figure 15, page122)is sufrcientto determinethe deviation
of the insulation
strengthofone pieceof equipment.,i,, at the time , (4
o0)from the specifiedvalue(Up111)
Tbe 90% withstand strengthof a piece of equipmentin a population varies from one specimen
to another.
Figure 16, page 123,showshow this can be describedstatisticallyin termsofK. The value
ofon is very small for
thosetypes of equipment(e.g. disconnectors)which may be consideredessentiallyair insulatin! structures,slnce
tolerancesin dimensionsare alwaysvery small.
To ensurethe repeatabilityof tests,ambient and insulationconditionsshould be kept as
constantas possible
during tests(or correctionfactorsshouldbe used)and standardizedtestingtechniquesshould
be adopted.In prin-
ciple,therefore,the dischargeprobabilityof a givenpieceof insulationundertestconditions
shouldnot be exDected
 -79-

to change.In other words, the curve P'(Y) (seeFigure 15) should be the samefor diferent tests of pieces
of equip-
ment of the sametype.
However, the value of the 90% withstand strength of insulation may show variations from the averagevalue
de-
rived from severaltests carried out by the same methods in the samelaboratory at diferent times or in
different
laboratories,dueto differencesin the ambientand insulationconditions,or in the testcircuits.Figure l6D,page123,
shows how laboratory inaccuracy can be described statistically considering the averagevalue of the insulation
strength of one specimenas the "true value".

Assuming that the distributions of Figures 16a and 16b, page 123, are Gaussian with known standard
devi-
ations and that the design value of LIp has been chosen by the manufacturer, the probability density
of the
deviation of the rreasured 90% withstand strengtl of the population of a piece of equipment can be
calculated
(seeFigure l6c, page 123).

An ideal test should be such as to prevent equipment having, at the time of the test, either an insulation
with-
stand lower than that specifledto pass the test or a dischargevoltage higher than that specified.
Self-restoringinsulationhaving,at the rated impulsewithstandvoltageappliedduring the test, a probability
of
withstandequal to, or higher than, the referenceprobability (90%) shouldhave a probability of passing
the test
equal to 1, while insulation having a probability of withstand lower than the referenceprobability should
have
no chanceof passingthe test.

The probability of passingan ideal test, i.e. for which the selectivity would be perfect for an apparatus
whose
withstand during the test ditrers by 1(o, frorr the specifredvalue, is representedin Figure l?, page 124,as a function
of K by the solid line.
Actual tests,however, depart from the ideal test and follow in the (Q, rK) plane curvessimilar to the dashed
line.

Figure 17 showsthe curvesfor the testsproposedin Clauses52, 53 and 54 of IEC publication?l-1 and for
the
test recommended ]n Item d) of Clause27 of IEC Publication71 (fourth edition).l As far as the test of Clause52
is concerned,an "up and down" test basedon 30 impulses was taken into consideration.

The probability density of the deviation of the measuredwithstand strength (Figure l6c) of a given population
is representedin Figure 18, page 125, by curve 1. By taking into account the probability of passing a given
test
procedure as a function of K, along with the deviations betweenlaboratories, curve 2 is obtained.r
The half-plane of positive ordinates of Figure 18 is thus divided into severalsectionsfor which the areas
have
the following signif.cance:
1) Area (B + c) located to the left of the axis of the ordinates included betweencurve I and the two half-axes
of co-ordinates, is equal to the proportion s of equipment of the population presenting a withstand probability
lower than 90% (failure during tle ideal test).
2) Area (A + D) located to the right of the axis of the ordinates, included betweencurve 1 and the two half-
axesof co-ordinates, is equal to the proportion of equipment of the population presenting a withstand prob-
ability equal to or greater than 90% (successfulideal test).

Ivore.- Th€sumofthesetwoarea.s,
or I +, + C + D, is equalto udty.

3) Area C is equal to the proportion of equipment of the population which, although presenting a withstand
probability lower than 90olo,is satisfactory under actual testing; consequently,it is the customer's risk
R.

r This latter test is defiDedas follows: a sed€'t

of five impulses is applied. If Do discharg€ occurs, the €quipnFdt is coDsideled good. If t$o
discha!8ps occur, th€ equipment is considered,bad, If a single discharge occu$, ten rew impulsas are applied. Duriag
if !o discharg€ occuN, the equiprneot is considered good. If one dischargp occurs, th€ equiprneN is coDsiAerea - this new attempt,
UO.

I In the case where dr : O,thc combination would

be reduced to th€ multiplicatio' of tbe ordiaates.
s A rigorous fo.mulatiotl would
irtoduce herc the probability that an apparatus of the populatiod would have a withstand probability
lower than 9O%.
 - 81-

4) Area B is equal to the proportion of equipmentof the population which presentsa withstand probability
lower than 90% and which is not satisfactoryunder actualtesting.
5) Area D representsthe proportion of equipment of the population which has a withstand probability equal
to or greater than 90''/o and which is satisfactory under actual testing.
6) Area I representsthe proportion of equipmentof the population which has a withstandprobability equal
to or greater than 90o and which none the lessis not satisfactory under actual testing; consequently,it is the
manufacturer'srisk R-.
The sum of areasA I B is equalto the proportion of equipmentof the populationwhich doesnot prove satis-
factory under actualtesting,regardlessof its withstandprobability.As a rule, the manufacturerdesignsthe insu-
lation IZ of his product in sucha way as to minimizethe sum of ar€asI f B, whichcanalsobe interpretedasthe
probability P, of failure under actualtesting.
By repeatedcalculation of R" and Pr for different values of the intended value of the insulation strength ( I4l),
curvescan be constructedwhich show the relation betweenthe risk of the customerof acceptinga deflcient product
and the manufacturer'sprobabilityPr of his product failing the test.
Figure 19, page 126, shows such curves on the assumptionthat standard deviations ds, ol a'rrdol are thosestated
in the legendsof the figures.
Figure 20, page127, givesthe correlationbetweenX- (manufacturer'srisk) and Ro (customer'srisk) for the
casesconsideredin Figure 19.
It shouldbe noted that if on and o, are equalto zero (homogeneous populationand no laboratory inaccuracy),
it is R" (customer'srisk)
one ofthe two risks is annulled.In fact, if the equipmentconformsto the specifications,
whichis equalto zero; if the equipmentdoesnot conform, it is R- (manufacturer'srisk) which is equalto zero.
In other words,thereis only onerisk eitherfor the customeror for th€ manufacturer.This risk, as well as the
value of P,, can be obtained directly from Figure 17, page 124.
 -83-

APPENDTX C

STATISTICAL EVALUATION OF THE PROTECTIVE EFFECT OF SPARK GAPS

For a given waveshapeof the applied impulses,let us call:

- Pt(A) and Po(U) the disruptive dischargeprobabilities of the insulation and of the spark gap as a function of
the crest value U of tle impulse 1;
- ProI(D the probability that the insulation may discharge before the protective gap sparkover, as a function
of the crest value U of the impulse.

The dischargeprobability curves of insulation rit4 a"a of the spark gueP:Q) connected in parallel arc
expressedby:
r l l
- P"(u)l
P,(u): P,(u)11 + Pt(u).pe(u).piry,(u) 01)
r l l t l
po(a): p!(u)11-p,(u)l
+ pDQ).p{u)l
I - B'(u)l (.l,2)
L l t l

Notes1, - ODthe assumptionthat times-to-discharge of the iuulation and of thc sparkgapfollow a BaussiaD
law, whate\€rmay be the
crestvalu€ U of tho appliedimpulsc,thc probabilityPr"(y) is givenby thc formula:

(13)

wh€f,€:
Tr@) - T"@)
r,ng,):Tffi6
Te(a) - fi7"veJrc of the tin€-to-sparkoverofthe sparkgap,asa fudctionofthe cr€6tvalueU ofthe applicdimpu.lse
Tr\U) : So%valw ofthe tidFto-dischargeof th€insulatio4 asa functiotrofthe crestvalue{,rof the appliedimpuls€
drly) : stsndird deviationof the time-to-sparkoverof the spark gap, as a fuDctiooof the clest valu€ U of the applied
rmpuse
dtr(U) : statrda.rddeviatio! of thc tfun€-todischargE
of the insulation,as a fuoctioo of the oest valueof appli€dimpuls€
2. - To writ€ equatiods(t 1) arrd(12),it wasassurned
that th€ idsulatiodand the spark gapcaDlot strikesimultaqeously.

If, for a particular combination of the insulation to be protected and of the spark gap, there is a negligible prob-
ability that the time-to-discharge of equipment may be lower than the time-to-sparkover of the spark gap in the
entire range 0 < U < U^ *, Prn((D becomes zero and formula (11) may be written:

niru>:P{u)
ft-p"tol (14)
L I
In the case of combinations of insulation and protective spark gaps behaving thus, we shall consider the be-
haviour of the protectivespark gap as being ..ideal".
By making use of yforrrula (7), the risk of failure of a protected piece of insulation can be evaluated by means
of the following formula:
ay .l a4o
-r |
x ', : J" p , (ou ) . l l -l p , ( u- ) | . 1 "l ( u ) d" u + I o & ( u ) . p D ( u ) . p r y , ( a ) . f ' ( u ) d u( 1 5 )

Underthe conditionsof validity of formula (13),it is possibleto expressthe fault risk by the formula:

* u q o
R, : Jo P,(u) - fP"(u)l
l
. f,(u)du (16)
Ll
r The formulae given here
are generallyvalid for two piecesof insulation in parallel.
 -85-

The ratio of the 50% lightning impulse sparkover voltage to the 50% switching impulse sparkover voltage
of a spark gap can be chosenfrom a wide range (l-1.5) by changingthe electrodeconfiguration.It is therefore
possible to selectthe dischargeprobability curve qf the gap Pr(U) to switching impulse almosi irrespectiveof the
dischargeprobability curve to lightning impulses.

Design of a spark gap with a view to switching impulses will make the averageexpectednurnber of flashovers
of the spark gap per year, due to switching surges,equal to a value l{" such as not to make systemperformance
under switching surgessignificantly worse. Therefore, the probability of sparkover of the spark gap when switching
overvoltagesreach the level evaluated on the assumption that equipment operates as designed (see Clause 25)
must be made very low.

Consequently,even if the spark gap behaveslike an "ideal" protective device, the spark gap will provide no
protection against switching overvoltages and insulation must be designedto withstand this type of switching
overvoltage.This is evidentif we recallformula (14),(caseof "ideal" spark gap).

P: : P,(u) - P"(4] ! Pi(r/) (17)

[t

As regardsswitching overvoltagesexceedingthe valuesbasedon the assumptionof correct behaviourof equipment,

we can assumethat the overvoltagevalue is suchthat a gap sparkover will almost certainly be oaused.Formula (1)
then becomes:

Pt(U): P{U)'Ptn(U) (18)

and formula (15)

U^ro
*
R,: J Pr(U) . Prn(U).f,(U) . dU (le)
0

The protective effect is, in this case,due only to Prn(U),i.e. to the relationship betweenthe time-to-dischargeof
the insulation and that ofthe spark gap. It is thereforenecessaryto make Pb([/) as closeas possibleto zero over the
entire range of interest of U.

Design of a spark gap with a view to lightning impulseswill be such as to limit the averageexpectednumber of
flashoversper year ofthe spark gap to lightning surgesto an acc€ptablevalu6 Nr,
that in many casesthe sparkoverof the sparkgap doesnot leadto any
In this connection,it mustbe remembered
supplyinterruption.
Let us considerthe exampleof a spark gap installedon the line sideof a circuit-breaker.If the l.ightningstroke
causesthe line to flashover,a coincidentalsparkoverof the sparkgapis ofno significance. On the other hand,if the
overvoltage amplitude does not reach the sparkover level of the line at the point struck, it is unlikely to causea
sparkover of the spark gap, even if the spark gap withstard level is lower than that of the line. This is due to the
reduction in surge amplitude at the point of installation of the spark gaps becauseof attenuation during wave
propagation as well as the possiblepresenceof other lines and surgearrestersat the station.

Thus, in contrast to the caseof switching surges,it is possibleto accept in certain casesspark gaps with a 50%
dischargevoltage to lightning impulses lower than that obtained basing the design of the apparatus on switching
surgesand usingthe combinationsofthe ratedimpulsewithstandvoltagesgivenin TableIV ofIEC Publication71-1.

Therefore, it can be concludedfrom formulae (11) and (14) that spark gapsmay ofer a limited degreeof protec-
tion in the case of lightning overvoltages of the order of the rated switching impuls€ withstand voltage
of the apparatus.
As regardslightningovervoltages muchhigherthan the ratedlightningimpulsewithstandvoltageofthe apparatus,
gap sparkoverwill almost certainly be caused.In this case,forrrula (18) and the sameconsiderationsas regards
.Prr(t/.)previouslymadein respectof switchingsurgesapply.
 -87 -

For air insulation of equipment, the condition Pre(U) - 0 may be fulfilled both for lightning and switching
impulsesby makinguseof gapshavinga high critical sparkovervoltageto switchingsurges(kv/crn), i.e. sparkgaps
much shorter than the air distancesof the insulation of the apparatus, for instanceby making use of spark gaps
having a conductor-rod confuuration for the electrodes.

For non-self-restoringinsulation ofequipment, the checkthat factor Pb((/) is reasonablysmall should be madeby
meansof choppedwave withstand tesb carried out at a voltage level basedon the highest overvoltagewhich can be
expectedin the systemand with a cut-of time chosenon the basisof the time-to-sparkover of the spark gap.

Thesetestsare not laid down in IEC Publication71-1and, if necessary,

shouldbe agreedupon betweenuserand
supplier.
 -89-

APPENDX D

EXAMPLES OF APPLICATION

Table III, page98, illustrates the choiceof the insulation level ofa pieceof equipment for caseI (seeClause25),
protected against both switching and lightning overvoltagesby surgeanesters mounted at its terminals. Examole 1
refers to a 420 kV transformer and example2 to a 765 kY hansformer.

The insulation levels of the transformers dependon the protective levels of surgearrestersagainst both swirching
and lightning impulses(block 7 of Figure 12, page 119).The protective level of a particular surge arrester depends,
in its turn, on its characteristicsand rating. The rating of surgearrestersin both examplesI and 2 is chosenas the
available rating immediately above the temporary overvoltagesanticipated on the system (block 6). Temporary
overvoltagesinclude voltage rise during faults, overvoltagesdue to inrush currents, suddenload rejection and other
causes.The overvoltageto be expectedis influencedby the earth fault factor, systemconfiguration, the characteristics
of systemequipment and operating practices(seeClause5).

' Table lV, page 101,showsthe choiceof the insulationlevel of a pieceof equipmentfor
caseII with no surge
anesterprotection or with remote surgearresterprotection.Example 1 refersto the line-to-earthinsulationof a
420 kV disconnecting switch on the line-side of the circuit-breaker, no surge arrestersbeing installed at the line
entrance.Example 2 refers to a 765 kV disconnectingswitch at the samelocation.

The rated switchingimpulsewithstandvoltageis selectedfrorn TableIV of IEC Publication71-l on the basisof

the statistical switching overvoltagelevel at the equipment location and on the basis of an acceptablerisk of failure
(blocks4 and 8 of Figure 12).

For both examples,it is assumedthat the only critical switchingovervoltages are thosedueto line ie-energization
(i.e. it is necessaryto check the risk of failure due to reclosingop€rations only).

By making useof the sirnplifiedstatisticalapproach(seeSub-clause 24.3)and choosingan appropriatestandard

deviation,it is possibleto determinethe statisticalsafetyfactor y correspondingto the maximumadmissiblerisk of
failure; from factor Z, it is possibleto derive the rninimum statistical switching impulse withstand yoltage and then
to selectthe rated switching impulse withstand voltage immediately above,
In the example,the correlation betweenrisk of failure and the statistical safety factor given in Figurc 7, pageI 13
(or : 8%) is used.

Oncethe rated switchingimpulsewithstandvoltagehas beenobtained,a correspondingrated lightning impulse

withstandvoltageis selectedfrom the sameline of Table IV of IEC Publication71-1,consideringthat only the
highestvalue ofrated lightning impulse withstand voltage should be usedfor apparatus not effectively protected by
surgearresters(seeClause49 of IEC Publication71-1).
It is then necessaryto verify that this value guaranteessatisfactoryperformanceunder atmosphericovervoltages,
i.e. a risk of failure not higher than the permissibleone. This can be done in a similar way aspreviously for switching
impulses@locks9 to 12in Figure 12),

In the caseunderconsideration,an approximatedistributionof the lightning overvoltages, at leastfor the caseof

disconnectingswitchesin the open position, may be evaluated quite sirrply on the basis of the line characteristics
and of errpirical or semi-errpirical laws for wave attenuation.
However,lightning overvoltagestressesvary from point to point in a substation.In general,it is thereforeextremely
difficult and time-consuming to achievethe knowledge of stressesnecessaryto make use of the statistical or sim-
plified statistical approach.
Only the highest overvoltage stressesin the most common contingencies(most common position of circuit-
breakers and disconnectors)are therefore evaluated.It is then verified that the rated lightning impulse withstand
voltage selectedas stated above, starting from the rated switching impulse withstand voltage, exceedsthe maximum
 -91 -

credible atmospheric overvoltages by a suitable margin (: l0%). If a rated lightning impulse withstand voltage
higher than that determined on the basis of Table IV of IEC Publication 7l-1 is desirable,the new value shall be
selectedfrom the seriesin Item b) of Clause46 of IEC Publication 7l-1.

In Table V, page 103, there is an evaluation of the increase of the risk of failure for a rated switching impulse
withstand voltage lower than the selectedvalue given in Table IV of IEC Publication 7l-1.
 -93-

APPENDIX E
CLEARANCES IN AIR I BETWEEN LIVE CONDUCTTVE PARTS
AND EARTHED STRUCTURES TO SECURE A SPECIFIED IMPULSE WITHSTAND VOLTAGE
FOR DRY CONDITIONS

In installations which, for various reasons,cannot be impulse tested,it is advisableto take steps
to avoid flash-
over occurring below the impulse withstand level which has been specifiedfor the equipment under
test.

The condition to be fulfilled is that the statistical switching and lightning impulse withstand voltages
in air between
live parts and earth should be equal to the rated switchingand lightning impulsewithstandvjtage
as specifled
in IEC Publication71-1.This resultsin a minimum clearanceto be observedwhich dependson
the-configuration
of the live parts and the nearby structures (electrodeconfiguration).

No distanceis indicated for an equipmentwhich has an impulse test included in its specifcation
sincecomputsory
clearancesmight hamper the design of the equipment, increase its cost and impede progress.
The impulse test,
even when only a type test, is sufrcient to prove that the impulse withstand condition is fulfilled.

Tables YI A and YI B, page 105,are suitable for generalapplication, providing as a first approximation
a clear-
anc€to be specifiedin relation to the insulation level.
In Table VI A, referencein made in the first column to the rated lightning impulse withstand
voltage and in
the secondcolumn to the air clearancesfor unfavourable confi.gurationsof live and earthed parts.
In Table vI B,
referenceis made in the first two columnsto the valuesdefiningthe insulationlevelsand i;
the third andfourth
columns to the air clearancesfor electrodeconfigurations of the "conductor-structure" type and .,rod-structure,,
typc.
The "rod-structure" conf.guration is the worst eleotrode configuration normally encounteredin practice;
the
"conductor-structure" configuration covers a large range of normally used configurations.
In Tabie vI B. ref-
erenceis made to the electrodeconfiguration becauseof its notable influencefor u- > 245 kv.

The values of air clearancesgiven in Tables YI are the minimum values dictated by electrical considerations,
and do not include any addition for construction tolerances, effect of short circuits, wind effects and
safetv of
p€rsonnel,etc.

These values are valid for altitudes not exceeding I 000 m (3 300 ft).

r Thcse indications refer to a simple

distance tbrough an air spac€,without consid€f,atio! of the surfaoc sparkover voltage alo1g a le{kagc
path of th€ idsulator relat€d to the contarniEatioD Droblem.
 -95-

T.q'Bu I

Provisional scale of natural pollution levels

Pollution level Environment Performance of existing lines

No significant Areas without industries and with low density No faults are observed in high humidity con-
pollution of houses equipped with heating plants; ditions (fog, mist, etc.) on 145 kV lines even
areas with low density of industries or when equippedwith lessthan 9-10insulators
housesbut subjectedto frequent winds and/ of the normal type 1, nor on 245 kV lines
or rainfalls. All areas must be situated far even when equipped with less than 15 such
from the sea or at a high altitude, and must insulators
not in any case be exposed to winds from
the sea

Light Areas with industries not producing particu- Faults occur in fog conditions on 145 kV lini:s
I
larly polluting smokes and/or with average with lessthan 9-10 insulators of the normal
density of houses equipped with heating type t and on 245 kV lines equipped with
plants; areas with high density of houses Iessthan 15 such insulators
and/or industries but subjected to frequent
clean winds and/or rainfalls; areas exposed
to winds from the seabut not too closeto the
coast (at least about I km)

Heavy Areas with high density of industries and sub- Faults occur in fog conditions, or when the
urbs of large cities with high density of wind blows from the sea, on H.V. Iines
heating plants producing pollution; areas equippedwith normal-typeinsulatorsl,unless
close to the sea or in any case exposed to the number of units per string is exception-
relativety strong winds from the sea ally high: more than 11-12 units on 145 kV
lines and more than 18 units on 245 kV
Iines

Very heavy 3 Areas gbnerally of moderate extension, sub- Faults occur in fog conditions or during salt-
jected to industrial smokes producing par- storms on H.V. Iines, even when equipped
ticularly thick conductive deposits; areas with antipollution-type insulators 2, unless
generally of moderate extension, very close the number of units per string is exception-
to the coast and exposedto very strong and ally high: more than 11-12 antipollution
polluting winds from the sea units on 145kV lines and more than 18 anti-
pollution units on 245 kV lines '

r Refereocc is made to norm4l-type

insul4tors with the folowiDg charactedstics: spacing; 146 mm; diametgr, 255 rnm; arcepagediltaDce,
300 mtrL
r TheJefer€o@ to adtipollution-tyP
itrsulatoN is sorDewhatvague, due to the great variety of antipollution-typ€ insulators which ar€ iD
s€rvice at the pres€nt tirne on H,V. liocs.
s Areasof moderal€extersiod
very clos€to highways,wh€Iea mixture ofsalt atrd bitumeo car!causes€vcledepqsitson the insulators
may b€ subject€dto a high pollutiotr lcwl.
 -97 -

Lc.nI,r II

Provisionalrelationbetween
pollutionlevels,test levelsand creepage
distances

Type of test

Salt fog method Solid layer method Creepagedistance/phase-

Pollution level to-earth voltage
(cm/kV r.m.s.)
Withstand salinity Layer conductivity
(ke/m3) (ps)

No significantpollution <5 5-10 2.0-2.s

Light pollution to-20 r0-20 3.0-3.5

Heavy pollution 40-80 20-40 4.0-5.0

Very heavy pollution > 160 >50 > 6

Notes 1. - The valuesreportedin the table were establishedon the basisof normal cap-and-pininsulators.For other types of insulator
and particularlyfor very largeinsulatorsin substations,the correlationwith thesetest methods,betweentestlivels and creep-
age distancesand serviceexperience,is not yet sufficient to give more deflnite indications.

2. - The creepagedistancesgiven in the table are those recommendedfor the various pollution levelsand do not necessarilyagree
with the creepagedistancesderived from the third column of Table I, which refer to existing overheadlines whosebehaviour
under power frequency voltage may or may not be satisfactory.
 -99-

T,c,Br,nIII

Choice of the insulation level of case I equipment

(e.g. transformerprotected by surge qrresters)

Example I Example 2

1 ) Basic data

Highest voltage for equipment U^(r.m.s.) KV 420 765

Corresponding phase-to-earthvoltage
- r.m.s.value
kv 243 442
- peakvatue
tJ^ft lvt kV (p.u.) 343 (1.00) 625 (1.00)
Determining temporary overvoltage (derived from systemstudies)
- r.m.s.value
KV 320 60s
- peak value
kV (p.u.) 4s2 (r.32) 855(1.37).
Minimum safety factor
- for switching overvoltages
1.15 1.15
- for lightning overvoltages
1.25 t.25

2) Characteristics of surge arrester

Rated voltage (r.m.s.) KV 324 612
Maximum switching impulse sparkover voltage kv 765 230
Maximum lightning impulse sparkover voltage KV 805 400
Maximum front-of-wave sparkover voltage KV 1010 660
Maximum residual voltage at rated dischargecurrent kv 735 400

3) Protection level
to switching impulses kv (p.u.) 765 Q.23) I 230 (1.96)
to lightning impulses1 kV (p.u.) 878 (2.s6) | 440 Q.30)

4) Insulation level (phase-to-earth)

Switching impulses:
Minimum conventionalswitching impulse withstand voltage kV (p.u.) 880 Q.66) | 4rs (2.26)
Rated switching impulse voltage kV (p.u.) 950 (2.77) | 42s(2.28)
Ratio of the rated switchingimpulsewithstand voltageto the switch-
ing impulse protection level
r.24 t.16
Lightning impulses:
Minimum conventionallightning impulse withstand voltage kV (p.u.) 100(3.20) I 800 (2.76)
Rated lightning impulse withstand voltage kV (p.u.) r1s (3.42) I 800 (2.87)
Ratio of the rated lightning impulse withstand voltage to the light-
ning impulse protective level 1.33 t.25

I Maximum front'of-wave sparkover

voltage divided by l.l5 (seeSub-clausel3.l).
 -101 -

Tenlr IV

Choice of the insulation level of case II equipment

( e.g. unprotecteddisconnectingswitch)

Example I Example 2

Basic data
I
I
Highest voltage for equipment U^ (r.m.s.) | w 420 765
Correspondingphase-toearth voltage
I
- r.m.s. value I kV 245 442
- peakvarue
u^lfr l\t I kv 0.u.) 343 (1.00) 625 (1.00)
Statistical overvoltage at the equipment location (value exceededin I
2l of casesonly) due to reclosingoperations I kV (p.u.) er0 (2.6s) I 255 (2.0)
Maximum acceptedrisk of flashover to earth for reclosing operation l0-3 10-4

Statistical safety factor 7 corresponding to the maximum accepted

risk of failure (derivedfrom Figure 7, page 113) 1.13 1.24

Minimum statistical switching impulse withstand voltage I 550

Rated switching impulse withstand voltage selected I 550

Risk of flashovercorrespondingto the rated switching impulse with-

stand voltage selected 4.lo'4

Recommended rated lightning impulse withstand voltage to be

associatedwith the rated switchingimpulse withstand voltage kV (p.u.) 1 42s (4.1s) | 2 400 (3.84)
 -103-

TAsrr V

Risk of failure
of inwlation as a function of the rated switchingimpulsewithstandvoltage
of the equipmentof TableIV

Example I Example 2

Alternative a)

Rated switching impulse withstand voltage (selectedvalue) kV (p.u.) I 050 (3.06) l 550 (2.48)
Statistical safety factor 1.16 t.24
Risk of failure corresponding to the above statistical safety factor 4.10'4 10-4

Alternative b)

Rated switching impulse withstand voltage kV (p.u.) eso(2.76) | 42s (2.28)

Statistical safety factor 1.05 1.14
Risk of failure 5.10-3 9.10-{

Alternative c)

Rated switching impulse withstand voltage l 300 (2.08)

Statistical safety factor 1.04
Risk of failure 6.10'8
 -105-

T.q.BrrVI A

Correlationsbetweeninsulationlevels
and minimumphase-to-earthair clearances

Rated lightning impulse withstand I Minimum phase-to-earth air

voltage I clearance
(kv) | (-tn)

4 o l o o
6 0 l s o
7s I tzo
e5 | too
rzs I zzo
r4s I zto
t7o I EZO
2so I +eo
32s | 0ro
4s0 | q00
550 | r roo
650 | r roo
7so I r soo

T.q.Br,nVI B

Correlations between insulation levels

and minimum phase-to-earthair clearances

Impulse withstand voltages Minimum phase-to-earth air clearance

(kv) (m)

Rated switching impulse Rated lightning impulsewithstand Conductor-structure

withstand voltage voltage

(650) 750 1.4 1.5

(650) 850 1.5 1.7
750 850 1.6 1.9
750 950 1.7 1.9
850 950 1.8 2.4
850 I 050 1.9 2.4
950 I 050; I 175 2.2 2.9
050 1 7 5 ;l 3 0 0 : 1 4 2 5 2.6 3.4
t75 300; 1 425;1 550 3.1 4.1
300 425;1 550; 1 800 3.6 4.8
42s 5 5 0 ;1 8 0 0 ;21 0 0 4.2 5.6
550 800; 1 95O;2400 4.9 6.4
 -106-

1
Fo (Ul

0,75

0,50

Fo(U'l
Fo(U'\
0,25

u (p.u.)
2t9176

Ftc. la. - Surtensionsde mise sous tension de ligne.

Line energization overvoltages.

0,75
Fo(U)

0,50

0,25
Fo(U)

Fo (U'l

Flc. lb. - Surtensions apparaissant sur une chalne d'isola-

teurs suite d des coups de foudre sur Ie pyl6ne.
Overvoltages across an insulator string due to
lightning strokes to the tower.

(rt\ densit6de la probabilitd de surtension

tro\v ) -_ overvoltage probability
density
F (rA _ probabilitd (cumul6e) de surtension
' o\v'' -
overvoltage (cumulative) probability

Ftc. 1. - Surtensions.
Overvoltages.
 -107-

t / .
Py"(U) ,f Py,(lJ)./ 'Pt(J')
-f----f----/
- f f
f l
t t
l t
l l
t l
t t
- I - ' --
T--T--
1 l
t l
l l
t l ,
t t F
I t l t
- - - r - -t- - r t - !- - / - l /

dPr(U)
dtJ ri
I

--r-
2ot

Ur io

Frcuns 2c

Flc. 2. - Probabilit6de ddchargedisruptived'une isolationsoumisei destensionsde choc.

Probability of disruptivedischargeof insulationunder impulsevoltage.
 -108-

t (m)

I
I
60 I
ilI
I

\\
I
40
lr
t

\
,\ Tensi onnomi naldu
e parafoudre
\ Surgearresterrating
t
\ |'
\
\
\ 1 0 , 5k V ' 7), I I
N v=
\
\ l \
12kV
36 kV
\ \-.._
I
\ \ , \_ . _ _ {.=.
r - - _ _

*
I I I
0 10 20 30 40 50 z(0)

imp€danced'onde du cAble en ohms

-, -_
cable surge impedancein ohms
longueur du cAbleen mdtres
L, -_
cable length in metres

Ftc. 3. - Longueur de cAblemaximale admissibledans le cas d'un parafoudre i la jonction a6rosouterraine.

Maximum permissible cable length with surge arrester of line/cable junction only.
 -109-

'ro(r')
:f: Pr(r')
ffiA
F = .[ PrQ) .fou) du
0

o -_ aire hachuree
" shaded area

Ftc. 4. - Evaluation du risque de d6faut d'une isolation.

Evaluation of the risk of failure of a piece of insulation.
 - 1 1 1-

Us- Uw
foU)

fo(U)

Ua Uw
Uw = 1,2Ug

Pr(Ul [1 - Pr(U)] PrUl

140% 0% fo(U\ 100%

U8 Uw U
Uw:1,4 U8
226176 227176

Frc. 5a. - Us : statistical overvoltage placed on the Frc. 5b. - Three attempts at determining the risk of failure
probability density curve (the shaded (arca A) for statistical safety factors.
area (2%) represents the reference
probability).
Uw : U*
statistical withstand voltage placed on
i t-z: T :l.o,1.2andl.4
the discharge probability curve OO% g

representsthe referenceprobability).

Rz ---t---
I
f l
t l
F3 - - - b - - l

Frc. 5c. - Relation between the statistical safety factor y

and the risk of failure R (measuredby area A). y1 y2 Ys v
228176
Ftc. 5. - Simplified statisticalmethod.
 -ttz-

R
I
5
- Distribution t r o n q u 6 ei ^
o 6"
2 Distributiontruncatedat
Distribution t r o n q u 6 ei
10-l *o 6s
Distributiontruncatedat
5 D i s t r i b u t i o nn o n t r o n q u 6 e
Distributionnot truncated
2
10'2

5
\

s
2
\
10-3

5
$i..
2
NN\*'\.
10-4

5
N '\

\.\
./
Envelo p p es u p 6n e u r e
U p p e re n v e l o p e

2 --
Envelcp p e i n f 6 rarrr)
i
10-s - Lower e n v e l o p e N\.'i'
5
,\N\\
\\N,tl'- 6s:20%
2
10-6
\\x"
\\\
\
5
\ d\r
2
\'
\
.N...t
lo" :15%

10'7
\i\ ? \ \

5
N\ \.\- \

2
ds:10%
tN t \tt

10-8
\\'l
\ \
t
\tt \
I

\ \ \ \
5 ''
2
\ t\\ tI
\r
\r
10-e \\ \r
0,7 0,9 111 112 1,3 1'4 1,5 1'6 1r7 Y

229 I 76

Ecart type de la distribution des surtensionsd, : 10, 15 et20/"

Ecart type de I'isolation or : 6Y
Standarddeviation of overvoltagedistribution o, : 10, 15 and 2OY"
Standarddeviation of insulation oT : 6y

Flc. 6. - Corr6lationsentrele risquede d6faut(R) et le coefficientde s6curitestatistique(y) pour diversesdistri-

butionsde surtensions de manceuvre.
Correlations between risk of failure (R) and statistical safety factor (7) for various switching surge
distributions.
 - 1 1 3_ _

10-1

10-2
\
\
10-3 -En l sup6rteure
veloppr
-Iur rp8l el| uelope
\ v l
10-4
N \
.)
l l o p p ei nf6rieui
Envr
10-5 Lower enveope \\
\\ \
\' \
\
10-6 \ \
\ \ \
\ \ \
10-7 \ \ \
\

10-8
5
2
10-e
0,7 0,9 0,9 1 1,1 1,2 1,3 1,4 1,5 1,6 y
23t176

Ecart type de I'isolation or : 87o

Standard deviation of insulation 6T : 8y

FIc. 7. - Corr6lations entre le risque de ddfaut (.R) et le coefrcient de sdcuritdstatistique(y) pour diversesdistri-
butions de surtensionsde maneuvre.
Correlations between risk of failure (.R) and statistical safety factor (y) for various switching surge
distributions.
 -Ir4-

R
1

10-1

10-2
\
Ent'eloppe
sup6rieure
perenvelope
10-3 \ fuo,
\.

10'4 Lower envell)pe

N
Envelo p p e i nf6rieurr
\.
\
\
\
10-5
\\
\ \

10{ \

10-7

10-8
5
2
10-e
0,7 1'1 1,2 1,3 114 1,5 1,6 y
232176

Ecart type de I'isolation 6r : lOYo

Standarddeviation of insulation 6r : l0y

Frc. 8. - Corr6lations entre le risque de d€faut (.R)et le coefrcient de s6curit6statistique (7) pour diversesdistri-
butionsde surlensionsde mancuvre.
Correlations between risk of failure (R) and statistical safety factor (7) for various switching surge
distributions.
 -115-

R
1
5
l -

10-1 \
\
5 \
\
o":60%7
\
10'2 \

5
\

10'3

5
\ 1
\\

oa:40% A
10-4

10'5
0,5 0,6 0F 0,9 0,9 1,0 1,1 1r2 1,3 .y
2t3176

Ecart type de I'isolation oT : 3oA

Ecart type de la distribution des surtensions d, : {Q et 60%
Standarddeviation of insulation or : 3y
Standard deviation of overvoltage distribution o" : 4O and ffiY"

FIc. 9. - Correlations entre le risque de d€faut (X) et le coefrcient de sdcurit6 statistique(/) pour diversesdistri-
butions de surtensionsde foudre.
Correlations between risk of failure (R) and statistical safety factor (y) for various lightning surge
distributions.
 -116-

R
1
5

10'1 \
\
5 \
\
o":UO%7
\
10-2 \

5 \

.{
10-3
\

5 \\

6e:40% /\
10-4

10-5
0,5 0,6 0,7 0,9 0,9 1,0 1,1 1,2 1,3 y

234176

Ecart type de I'isolation 6r : 57o

Ecart type de la distribution des surtensions 6" : {Q et 60%
Standarddeviation of insulation 6T : 5yo
Standard deviation of overvoltagedistribution o" : 40 and 607"

FIc. 10. - Corrdlations entre le risque de d6faut (R) et le coefrcient de s6curit6statistique (7) pour diversesdistri-
butionsde suriensionsde foudre.
Conelations between risk of failure (R) and statistical safety factor (7) for various lightning surge
distributions.
 -tt7-

R
1
5

10-1 \
5 \
\
.d8 = 60%_7
\
10-2 \ /
\ \ /
5
\
(
10-3

5 N
os:40% t \
10-4

10-5
0,5 .0,6 0,7 0,g 0,9 1,0 1r1 7r2 1,3 y

Ecart type de I'isolation or : 7 7o

Ecart type de Ia distribution des surtensions o, - 40 et 600/"
Standard deviation of insulation oT : 7 y
Standard deviation of overvoltagedistribution o, : 4O and @7.

FIc. 11. - Corrdlations entre le risque de difaut (.R) et le coefrcient de s6curit6statistique (7) pour diversesdis-
tributions de surtensionsde foudre.
Correlations between risk of failure (R) and statistical safety factor (y) for various lightning surge
distributions.

-='{*b
 - 1 1 9-

Operating
voltage
undernormal
conditlons

Characteristlcs
of the system

Calculatlon Introduction
of means Calculation
of temporary @ to reducetemporaryor of switching
overvoltages g overvoltages
switchin overvoltages

Selection Insulationdesign
of surge as regards operating
arresterg voltagesand temporary
rated-voltage @ overvoltages

Selectlon Selectlon
of rated lightnlng of rated
and switchlngwithstand and switchingwithstand
impulsevoltage impulsevoltage
of case I apparatus of case ll apparatus
Introductionof meansto
reduce llghtning overvoltages

Lineinsulation
level, Selectlon
Lightning Calculatlon of rated
overvoltage n u mb e r
of lightning lightningwithstand
characteristics of earthwires
overvoltages impulsevoltage
sparkgaps
of case ll apparatus
at the line-entrance

Flc. 12. - Block-diagramof insulationco-ordinationand design.

 -r20-

"--ll--- Ct
Uo - sUl

237176

Frc. l3a, - Transformateur repr6sent6par un diviseur de Frc. 13b. - Circuit dquivalent pour la transmission
tension capacitif. capacitive.
Transformer as a capacitive voltage divider. Equivalent circuit of capacitancetransmission.

tl
-ll-l
c t l
U o:sul
: sUl + cg Uo=

239176

Frc. l3c. - R6seauextdrieur de capacit6 C". Frc. l3d. - Rdseauextdrieur de rdsistanceR.

External system having capacitanceC.. External system having resistanceR.

Frc. 13. - Pointeinitiale de tensioncapacitive.

Initial capacitivevoltagespike.
 -t2t-

Choc sur une phase sculement Chocs dc polarit6s oppos6cs sur 2 phascs
Cas Couplagc du ttansformcteur Surge on one phase only Surgcs ofopposite polarity on 2 phases
Case Transformer connecuorul gr": g -1, Uc: 0
Us: l, U": Us: l, Us:

Enroulement Enroulemcnt Enroulement Enroulement Enroulemont Enroulemcnt

No haute tension basse tension Tertiaire haute tension basse tension hautc tension basse tension
No. Higher.voltage Lower-voltagc Tcrtiary Higher-voltage Lower-voltagc Higher-voltage Lower-voltagc
winding winding winding winding winding winding

I Y(e) y (e) (- v)
t

,'tla ,'f-,
2 Y(e) v (i) (-, y)
x, -(
-1ls -1/3 )[- l- -l

'<?
0

tr
3 Y(e) d (-, y, O
)1,
\-tr
.v.
2

4 r(t) y (e, i) (-, Y, d)

0
.4.
I
0
.,'/rr-+1"/, A o + - 1

5 v(r) d (-' Y, d)
$r
,4.
t
'(i ,lvr<'_l;
''"uru
0 0

( lve ,lv'
5 Y(f) z (e, i) (-, y, O
)(, -)-,tvi -,lv'
7 D y (e, i) (-, Y, d)
x'u_ I lVa
s * -1lVs rl/;$-nlr'

8 D d (-, y, O
<' -''<:; .<:,
21r176

Ftc. 14. - Valeursdu facteurr.

Valuesof factor r.
 -t22-

Pt 1-Pt
(%, (%,
100 0

d" I'appareilde r6f6rence

'o,(,,,
r\v'
for the referencepiece of equipment

Probabilit6
de r6f6rence.
Referenceprobabllity
90

100
nominaleaux chocs
URw : tensionde tenue
ratedimpulsewithstandvoltage
K en p.u. de o1
K in p.u. of o1

212176

Ut n- Unw

Ftc. 15. - D6f,nition de la tenue di6lectrique d'un appareil d I'instant t a I'aide du paramdtre K.

Definition of the insulation strength of a piece of equipment at the time I as function of the parameter K.
 -r23-

Tensiontenuenominaleau choc
RatedimpulsewithEtandvoltage

0
I

dpl: [A;+4

243176

a) Distribution de la tenue didlectrique (90%) des appareils de aJ Distribution of the 90[withstand voltage of the apparatus
Ia population p. of population p.
b) Distribution des diff6ren@s entre Ia tenue di6lectrique 90% D/ Distribution of the difference between the 90Yo withstand
d'un appareil donn6 mesur6e dans diff6rents laboratoires et voltage of a given apparatus measuredin various laboratories
la tenue di6lectrique reelle. and the actual one.
c) Distributionde la tenue di€lectrique 901l d'un appareil c/ Distribution of the90% withstand voltage of any apparatus
quelconque de la population p dans divers laboratoires. of population p in various laboratories.

Frc. 16. - Densit6 de frdquence de la tenue didlectrique (90%) mesur6ed'une population d'appareils.

Frequency density of the measured 90oh withstand strength of a population of apparatus.

 -124-

0,8
,r'ift=:'=
I
I a
a

,f 0,,
,
i/ t' t
a
a

/ .m,u T a
a

i/ t' o'5
/' ilt

,,/
Essaiid6al
ldeal test

att'

/
a2

-0'8 -O'6 {'4' 4'2 0,2 0,4 0,6 0,8 1

K en p.u. de o1
K in p.u. of o1
241176

l - essai par mont6es et descentes30 chocs (article 52 de la 1 - up and down test 30 impulses(Clause52 of IEC Publica-
Publication 71-1 de la CEI) tion 7l-l)
, , _ 15/2 test (Clause53 of IEC Publication 71-1)
2 - essai15/2(article 53 de la Publication 71-1de la CEI)
3 - essai3/0 (article 54 de la Publication 7l'l de la CEI) 3 - 3/0 test (Clause54 of IEC Publication 71-1)
.
{ 5lO (article 27d) de la Publication 71 de Ia I SIO test (Clause 27d) of IEC Publication 71,
4 - essar 5/l 4 - fourth edition)
I + 10/0 cEI, quatridme6dition) I 5/1 + 10i0

Frc. 17. - Probabilitdpour un mat6rielde r6ussird diff6rentstypes d'essaien fonction de ses caractdristiques
d'isolation.
Probability of an equipmentto pass the different types of tests as a function of its insulating
characteristics.
 C o u r b eo
Curve

K en p.u. de o1
K in p.u. of o1
24s176

FIc. 18. - Probabili6 Pt d'dchouer d I'essai(aire A + B).

Risque du constructeur R* (aire ,,4).
Risque du client Ro (aire C).

Probability Pt of test failure (area A + B).

Manufacturer's risk R- (arca A).
Customer's risk R" (area C).
 -t26-

Rc ( oP: t%
F-
o t: 5 % 'z
1 - a%
\ot

( oP: +%
ot:o%
1
( ot :6%
10-1

u
t.li
f
10-2 I

/ / I )I
. , J
I
2/ / I t I
10€
tI

/
/
/
1
t
/ ,
I r
10-4

3 I f\
, I
10-5 F
t I I
2 ,
I
I I I ,
10€ I ,

,
1i I
a

/ I !a
10'7'

10{
10-5 10-4 10€ 10{ t0-1 1 P g

26176

I : essaide d€chargedisruptive 50f | : 507. dischargetest

2 : essai15/2 Z : l5l2 test i a

3 : essai3/0 3 : 3/0 test

It

Frc. 19. - Risquedu client en fonction de la probabilit6d'6checi I'essai(Pt).

Customer'srisk as a function of probabilityPt to fail the test.

F 4 * '
qf
 -127-

Fc

____
t dP:4%
dt :5ol

t ot :6%

6P:4%
--- C r :t 0 %

{ ot :6%
10-l

/
f I
t

10-2 .t
r t

I ,
21 / I
/ f
l /
I
a
10€
1 ,t
/ a I ,J
tl
/ l
l

10-4
/

3l t
/ t a

/
.L /
10-5

/ 2 I

1"" 10€
I
I
t
I

I
1t l I

10-7
!s

10-8
10-5 104 1O-3 lo-2 10.1 1 tr?m
247176

I : essaide dechargedisruptive 50f | : 507, dischargetest

2:essail5/2 2:lll2tast
3 : essai3/0 3 : 3/0 test

Ftc. 20. - Risque du client R" en fonction du risque du constructeur Rm.

Customer's risk R" as a function of manufacturer's risk Ro,.