IET POWER AND ENERGY SERIES 53

Condition Assessment
of High Voltage
Insulation in Power
System Equipment
Other volumes in this series:
Volume 1 Power circuit breaker theory and design C.H. Flurscheim (Editor)
Volume 4 Industrial microwave heating A.C. Metaxas and R.J. Meredith
Volume 7 Insulators for high voltages J.S.T. Looms
Volume 8 Variable frequency AC-motor drive systems D. Finney
Volume 10 SF6 switchgear H.M. Ryan and G.R. Jones
Volume 11 Conduction and induction heating E.J. Davies
Volume 13 Statistical technjiques for high voltage engineering W. Hauschild and
W. Mosch
Volume 14 Uninterruptible power supplies J. Platts and J.D. St Aubyn (Editors)
Volume 15 Digital protection for power systems A.T. Johns and S.K. Salman
Volume 16 Electricity economics and planning T.W. Berrie
Volume 18 Vacuum switchgear A. Greenwood
Volume 19 Electrical safety: a guide to causes and prevention of hazards J. Maxwell
Adams
Volume 21 Electricity distribution network design, 2nd edition E. Lakervi and
E.J. Holmes
Volume 22 Articial intelligence techniques in power systems K. Warwick,
A.O. Ekwue and R. Aggarwal (Editors)
Volume 24 Power system commissioning and maintenance practice K. Harker
Volume 25 Engineers handbook of industrial microwave heating R.J. Meredith
Volume 26 Small electric motors H. Moczala et al.
Volume 27 ACDC power system analysis J. Arrillaga and B.C. Smith
Volume 29 High voltage direct current transmission, 2nd edition J. Arrillaga
Volume 30 Flexible AC Transmission Systems (FACTS) Y-H. Song (Editor)
Volume 31 Embedded generation N. Jenkins et al.
Volume 32 High voltage engineering and testing, 2nd edition H.M. Ryan (Editor)
Volume 33 Overvoltage protection of low-voltage systems, revised edition P. Hasse
Volume 34 The lighting ash V. Cooray
Volume 35 Control techniques drives and controls handbook W. Drury (Editor)
Volume 36 Voltage quality in electrical power systems J. Schlabbach et al.
Volume 37 Electrical steels for rotating machines P. Beckley
Volume 38 The electric car: development and future of battery, hybrid and
fuel-cell cars M. Westbrook
Volume 39 Power systems of electromagnetic transients simulation J. Arrillaga and
N. Watson
Volume 40 Advances in high voltage engineering M. Haddad and D. Warne
Volume 41 Electrical operation of electrostatic precipitators K. Parker
Volume 43 Thermal power plant simulation and control D. Flynn
Volume 44 Economic evaluation of projects in the electricity supply industry
H. Khatib
Volume 45 Propulsion systems for hybrid vehicles J. Miller
Volume 46 Distribution switchgear S. Stewart
Volume 47 Protection of electricity distribution networks, 2nd edition J. Gers and
E. Holmes
Volume 48 Wood pole overhead lines B. Wareing
Volume 49 Electric fuses, 3rd edition A. Wright and G. Newbery
Volume 50 Wind power integration: connection and system operational aspects
B. Fox et al.
Volume 51 Short circuit currents J. Schlabbach
Volume 52 Nuclear power J. Wood
Volume 905 Power system protection, 4 volumes
Condition Assessment
of High Voltage
Insulation in Power
System Equipment
R.E. James and Q. Su
The Institution of Engineering and Technology
Published by The Institution of Engineering and Technology, London, United Kingdom
2008 The Institution of Engineering and Technology
First published 2008
This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act, 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publishers at the undermentioned address:
The Institution of Engineering and Technology
Michael Faraday House
Six Hills Way, Stevenage
Herts, SG1 2AY, United Kingdom
www.theiet.org
While the authors and the publishers believe that the information and guidance given
in this work are correct, all parties must rely upon their own skill and judgement when
making use of them. Neither the authors nor the publishers assume any liability to
anyone for any loss or damage caused by any error or omission in the work, whether
such error or omission is the result of negligence or any other cause. Any and all such
liability is disclaimed.
The moral rights of the authors to be identied as authors of this work have been
asserted by them in accordance with the Copyright, Designs and Patents Act 1988.
British Library Cataloguing in Publication Data
James, R. E.
Condition assessment of high voltage insulation in power
system equipment. - (Power & energy series; v. 53)
1. Electric insulators and insulation - Testing 2. High Voltages
I. Title II. Su, Q. III. Institution of Engineering and Technology
621.31937
ISBN 978-0-86341-737-5
Typeset in India by Newgen Imaging Systems (P) Ltd, Chennai
Printed in the UK by Athenaeum Press Ltd, Gateshead, Tyne & Wear
Contents
Preface xi
1 Introduction 1
1.1 Interconnectionof HV power systemcomponents 2
1.1.1 Alternatingvoltagesystems 2
1.1.2 Direct-voltagesystems 8
1.2 Insulationcoordination 9
1.3 High-voltagetest levels 10
1.3.1 Power-frequencyvoltages 13
1.3.2 Lightning-impulsevoltages 13
1.3.3 Switchingsurges 14
1.3.4 Veryfast transient tests(VFTT) 14
1.3.5 Direct-voltagetests 14
1.4 Power systemdevelopments 15
1.4.1 Reliabilityrequirements 15
1.4.2 Conditionof present assets 15
1.4.3 Extensionof power systemlife 16
1.4.4 Newsystemsandequipment 16
1.5 Futureinsulationmonitoringrequirements 17
1.6 Summary 17
1.7 References 17
1.8 Problems 18
2 Insulating materials utilized in power-system equipment 21
2.1 Reviewof insulatingmaterials 22
2.1.1 Gases 22
2.1.2 Vacuum 25
2.1.3 Liquids 25
2.1.4 Solids 27
vi Condition Assessment of High-Voltage Insulation
2.2 Characterizationof insulationcondition 33
2.2.1 Permittivity() andcapacitance(C) 33
2.2.2 Resistivity() andinsulationresistance(IR) 33
2.2.3 Timeconstants 34
2.2.4 Dielectricdissipationfactor 34
2.2.5 Partial discharges(PD) 35
2.2.6 Physical andchemical changes 35
2.3 Modesof deteriorationandfailureof practical insulating
materials 36
2.3.1 Dielectriclosses 37
2.3.2 Partial discharges sources, formsandeffects 39
2.3.3 Ageingeffects 46
2.4 Electrical breakdownandoperatingstresses 48
2.5 Development of insulationapplications 50
2.6 Summary 50
2.7 References 51
2.8 Standardsrelatedtoinsulatingmaterials 53
2.9 Problems 54
3 Introduction to electrical insulation design concepts 55
3.1 Overviewof insulationdesignrequirements 55
3.1.1 Electrical requirements 56
3.1.2 Physical limitations 56
3.1.3 Workingenvironment 56
3.1.4 Mechanical requirements 57
3.1.5 Thermal conditions 58
3.1.6 Processing 58
3.1.7 Reliability 59
3.2 Electricstressdistributionsinsimpleinsulationsystems 60
3.2.1 Multipledielectricsystems 61
3.2.2 Edgeeffects 64
3.2.3 Multipleelectrodeconfigurations 66
3.3 Electricstresscontrol 68
3.4 Summary 69
3.5 References 69
3.6 Problems 70
4 Insulation defects in power-system equipment: Part 1 71
4.1 Suspensionandpost insulators 71
4.1.1 Suspension(string) insulators 71
4.1.2 Post insulators 73
4.2 High-voltagebushings 74
4.3 High-voltageinstrument transformers 77
4.3.1 Oil-impregnatedcurrent transformers 78
4.3.2 Dry-typecurrent transformers 80
4.3.3 Capacitor-typevoltagetransformers CVT 81
List of contents vii
4.4 High-voltagepower capacitors 82
4.5 High-voltagesurgearresters 83
4.6 High-voltagecircuit breakers 84
4.7 Gas-insulatedsystems(GIS) 86
4.8 High-voltagecables 87
4.8.1 Oilpaper cables 87
4.8.2 Extrudedcables 90
4.9 Summary 93
4.10 References 93
4.11 StandardsrelatedtoChapter 4 95
4.12 Problems 96
5 Insulation defects in power-system equipment: Part 2 97
5.1 Electrical rotatingmachines 97
5.1.1 Low-voltagemotors 97
5.1.2 High-voltagemachines 98
5.1.3 Possibleinsulationfailuremechanismsinrotating
machines 100
5.1.4 CIGRE summaryof expectedmachineinsulation
degradation 103
5.1.5 Futureof machineinsulation 103
5.2 Transformersandreactors 104
5.2.1 Windings 104
5.2.2 Transformer insulationstructures 106
5.3 Summary 118
5.4 References 118
5.5 Problems 120
6 Basic methods for insulation assessment 121
6.1 Generationandmeasurement of test highvoltages 122
6.1.1 Power-frequencyvoltages 122
6.1.2 High-frequencyvoltages 127
6.1.3 Very-low-frequencyvoltages(VLF) 128
6.1.4 Direct voltages 128
6.1.5 Hybridtest circuits 129
6.1.6 Lightningimpulsevoltages 129
6.1.7 Switchingsurgevoltages 133
6.1.8 High-voltageequipment for on-sitetesting 133
6.2 Non-destructiveelectrical measurements 135
6.2.1 Insulationresistance(IR) measurements 135
6.2.2 Measurementsof thedielectricdissipationfactor
(DDF) 137
6.2.3 Measurement of partial dischargesbyelectrical
methods 140
6.2.4 Dielectricresponsemeasurements 147
viii Condition Assessment of High-Voltage Insulation
6.3 Physical andchemical diagnosticmethods 150
6.3.1 Indicatorsof in-serviceconditionof oilpaper
systems 150
6.3.2 Analysisof SF
6
samplesfromGIS 153
6.3.3 Surfacedeteriorationof compositeinsulators 153
6.3.4 Water treeinginXLPE cableinsulation 153
6.3.5 Ultrasonicmethodsfor detectionof partial
discharges 154
6.3.6 Miscellaneoustechniques 154
6.4 Summary 154
6.5 References 154
6.6 Standardsrelatedtobasictest methods 157
6.7 Problems 158
7 Established methods for insulation testing of specific equipment 159
7.1 Overheadlineandsubstationinsulators 160
7.1.1 Porcelainandglassinsulators(overheadlines) 161
7.1.2 Ceramicandglassinsulators(post type indoor and
outdoor) 161
7.1.3 Compositeinsulatorsfor overheadlines(stringandpost
units) 162
7.2 Overheadlineandsubstationhardware 162
7.3 Surgearresters 163
7.4 Switchgear 164
7.4.1 Circuit breakers 164
7.4.2 Self-protectedswitchgear 166
7.4.3 Disconnectors(isolators) 166
7.4.4 Metal-enclosedswitchgear 166
7.4.5 Transformer tapchangers 167
7.5 Bushings 167
7.6 High-voltageinstrument transformers 168
7.6.1 Current transformers 168
7.6.2 Inductivevoltagetransformers 169
7.6.3 Capacitor voltagetransformers 170
7.7 High-voltagepower capacitors 171
7.8 High-voltagerotatingmachines 171
7.8.1 Stator bars 172
7.8.2 Assembledmachine 172
7.9 High-voltagecables 173
7.9.1 Oil-impregnatedcables 173
7.9.2 Extrudedcables 173
7.10 Distributionandpower transformers 175
7.10.1 Power-frequencyovervoltagewithstandtests 175
7.10.2 Partial-dischargetests 177
7.10.3 Summaryof transformer HV test requirements 180
List of contents ix
7.10.4 Additional tests 182
7.11 Dielectrictestingof HVDC equipment 182
7.12 Miscellaneousitems 184
7.13 Summary 184
7.14 References 184
7.15 StandardsrelatedtoChapter 7 185
7.16 Problems 188
8 Sensors for insulation condition monitoring 189
8.1 Ultra-high-frequencysensors 189
8.2 Optical-fibresensors 190
8.2.1 Basicphysicsof optical-fibresensing 193
8.2.2 Optical-fibrePDsensors 194
8.2.3 Optical-fibretemperaturesensors 196
8.2.4 Advantagesanddisadvantagesof optical-fibre
sensors 199
8.3 Directional sensorsfor PDmeasurements 200
8.3.1 Directional coupler sensor 200
8.3.2 Directional fieldsensor 201
8.4 Summary 203
8.5 References 203
8.6 Problems 205
9 Online insulation condition monitoring techniques 207
9.1 Themainproblemswithofflineconditionmonitoring 207
9.2 Noise-mitigationtechniques 208
9.2.1 Noisegating 209
9.2.2 Differential methods 211
9.2.3 Noiseidentificationbysignal waveformanalysis 214
9.2.4 Multipleterminal PDmeasurements 215
9.3 Non-electrical onlineconditionmonitoring 219
9.3.1 Temperaturemonitoringof theinsulations 219
9.3.2 OnlineDGA 219
9.3.3 Acoustic-basedtechniquesfor PDdetection 222
9.4 Onlineacoustic/electricPDlocationmethodsfor transformers 224
9.4.1 Acoustictransducersandwindingterminal
measurements 224
9.4.2 Applicationof internal combinedacousticand
VHF/UHF transducers 224
9.5 Electrical onlineconditionmonitoring 225
9.5.1 Onlinedielectricdissipationfactor andcapacitance
measurements 227
9.5.2 Onlineleakagecurrent measurement 228
9.5.3 Electrical onlinePDdetection 230
9.6 Summary 236
x Condition Assessment of High-Voltage Insulation
9.7 References 236
9.8 Problems 239
10 Artificial-intelligence techniques for incipient fault diagnosis and
condition assessment 241
10.1 Databasefor conditionassessment 241
10.1.1 A computer databaseanddiagnosticprogram 242
10.1.2 A combinedmethodfor DGA diagnosis 243
10.2 Fuzzy-logicfault diagnosis 244
10.2.1 Theconventional methods 245
10.2.2 A fuzzy-logicmethod 245
10.3 Asset analysisandconditionranking 255
10.3.1 Equipment rankingaccordingtotheinsulation
condition 255
10.3.2 Insulationhealthindex 255
10.3.3 Membershipfunctionsof fuzzyset 256
10.3.4 Exampleof fuzzylogicconditionranking 258
10.4 Summary 262
10.5 References 263
10.6 Problems 264
Appendix 1 List of Abbreviations 265
Appendix 2 Major standards organizations 266
Appendix 3 Answers to problems 267
Index 270
Preface
Theneedforincreasedreliabilityandoptimumeconomicperformanceof high-voltage
power systems has become of greater importance in recent years. A major factor
in achieving theseobjectives is theprovision of efficient maintenanceof thewide
range of equipment. This applies especially to the assessment of the condition of
theinsulating materials, many of which aresubjected to high electrical stresses in
critical locations. Theratesof deteriorationof thesematerialsaredependent onthe
operatingconditionsand, insomecases, materialsareexpectedtoretaintheir useful
propertiesforfortyyears. Inordertomonitoranydangerouschangesintheinsulating
materials, muchwork isbeingcarriedout worldwideintheuniversitiesandsimilar
establishments, aswell asbyutilitiesandplant manufacturers.
Thisbookintroducesthereadertothemannerinwhichthemoreimportantcompo-
nentsinapowersystemareinterrelated. Thevariouselectrical insulatingmaterialsare
reviewedandparticular properties identifiedas beingsuitablefor conditionassess-
ment and monitoring. A guideis given as to how electric stress calculations may
assist in explaining insulation failures. Analyses areincluded of someof thefault
scenariosoccurringinhigh-voltagepower-systemequipment. Thesecondhalf of the
book is devotedto presentationof awiderangeof insulation-conditionassessment
techniques. Recentadvancesintheapplicationof digital techniquesformeasurement
andanalysisof partial dischargesarediscussed. Descriptionsaregivenof thehigh-
voltagetestapparatusnecessaryfor applyingwithstandtestsaccordingtothevarious
equipment standards. In thelast threechapters newcondition monitoring methods
inuseor under development arepresented. Theseincludeapplications of newsen-
sors, onlineproblems withparticular solutions andtheuseof artificial-intelligence
techniquesfor incipient fault diagnoses. Extensivereferencesareincluded.
The subject matter of the book is suitable for final-year courses in electrical
power engineering, for short courses on insulation-condition assessment and for
postgraduate programmes involved with the study of insulating materials. Power-
systemengineers associated with high-voltageequipment should find thebook of
valueinrelationtofaultinvestigations, maintenancerequirements, insulationtesting
andconditionmonitoring.
xii Condition Assessment of High-Voltage Insulation
Bothauthorshavesignificantindustrial experienceintheUnitedKingdom(REJ )
andChina/Singapore(QS) andinteachingandresearchat PortsmouthPolytechnic
(REJ ), Universityof NSW(REJ , QS) andMonashUniversity(QS). Duringthelatter
periodsmany consultanciesconcernedwithindustrial high-voltageinsulationprob-
lems wereundertaken. Wewish to acknowledgethevalueof our association with
manyex-colleaguesinindustryandtheuniversitiesandthoseinthevariousutilities
withwhomwehaveworked.
Our thanks areespecially dueto our wives andfamilies Felicia(REJ ), Liling
anddaughter Shirley (QS) for their patienceandunderstandingduringwritingof
thebook.
R.E. James
Q. Su
Chapter 1
Introduction
Power systemcomponents
Insulationcoordinationconceptsandhigh-voltagetest levels
Needfor insulationconditionmonitoring
A high-voltagepower systemconsistsof acomplexconfigurationof generators,
long-distancetransmissionlinesandlocalizeddistributionnetworkswithabove- and
below-groundconductors for deliveringenergy to users. This introductory chapter
indicates the wide range of high-voltage components whose successful operation
dependsonthecorrectchoiceof theelectrical insulationfortheparticularapplication
andvoltagelevel. Theconditionof theinsulatingmaterialswhennew, andespecially
astheyage, isacritical factorindeterminingthelifeof muchequipment. Theneedfor
effectivemaintenance, includingcontinuousinsulationmonitoringinmanycases, is
becominganimportantrequirementintheassetmanagementof existingandplanned
power systems.
As thevoltages and powers to betransmitted increased over thepast hundred
yearsthebasicdielectricsgreatlyimprovedfollowingextensiveresearchbyindustry
andinspecializedlaboratories, wheremuchof thisworkcontinues. Itisof interestto
notethat paper, suitably driedandimpregnated, is still usedfor many high-voltage
applications. Newdielectricsarebeingintroducedbasedonmany yearsof research
and development and arebecoming morewidespread as operational experienceis
obtained. In order to ensure an economic power-supply systemwith a high level
of reliability, it is important to beableto monitor thedielectric parameters of the
variousinsulationsbeingutilized whennewandinservice. Later chaptersdescribe
thematerialsandtheir applications, includingexamplesof possiblefault scenarios,
dielectric testing techniques for completed equipment, newand existing condition
monitoringsystemsand, finally, theapplicationof artificial intelligenceinincipient
fault diagnosisandconditionassessment.
Present power systemsareageingsignificantlyandinmanycases40per cent of
theequipmentisolderthantheconventional designlife of 25years. Thisfigurewas
probably chosen becauseof theuncertainties in estimating theanticipated lives of
2 Condition Assessment of High-Voltage Insulation
thepractical insulationstructuresandfor commercial reasons. Infact, manysystem
componentsarestill functioningsatisfactorilyafter muchlonger periods. Thisispos-
siblyduetotherelativelylowaverageelectricstressvaluesusedtoallowfor inherent
inaccuraciesincalculationsof maximumvalueswithinthecomplex structures. The
development of suitablecomputer programshasenabledmuchimproveddesignsto
beachieved. Also, in many systems thecircuits wereoperated in parallel to cater
for overloadingandpossiblefailureof onelineor unit. Thisconfigurationprobably
resultedintheaveragedielectric temperatures beingbelowtheallowablemaxima.
Thesituationischangingwiththeneedfor themanagedassetstorealizemaximum
economicreturns. It isonlybyeffectiveconditionmonitoringover longperiodsthat
datacanbeacquired, thusenablingtherateof deteriorationof theinsulationstructures
to bedeterminedinservice. This wouldnaturally includetheinfluenceof possible
genericmanufacturinganddesignfaultsaswell asinappropriatemaintenance. Trends
insuchdataassist inthemorereliablepredictionof theremaininglifeof equipment,
possiblyincludingtheapplicationof probabilistictechniques.
1.1 Interconnection of HV power system components
Contemporarysystemvoltagesrangeupto1000kV(RMSthree-phase) orhigherand
600kV(DC), although themoreusual AC values are500/750kV and below. Bulk
powers greater than 1000MW may betransmitted by a singlethree-phasecircuit
over longdistances, insomecasesfor morethanseveral hundredkilometres. Local
deliveryratingsmaybeof manytensof MVA downtoafewkVA.
The application of renewable sources for example solar devices, wind gen-
erators, biomass generationandsmall hydro-plants is becomingmoreimportant.
Withintenyears it might beexpectedthat embeddedgenerationfromsuchsources
couldcontributebetween10and20per cent of thetotal power insomecountries,
althoughcommercial problemsmaylimitthedevelopments[1]. Theformof theexist-
ingpower systeminfrastructures wouldprobably not changesignificantly for such
conditions, especially wherehighlevels of energy arerequiredat aparticular loca-
tion. Thenewer sourceswill operatelocallyatlowvoltagesandincludeconventional
step-upsystemswheretheyarecoupledtothemaindistribution/transmissionsystem.
Special insulationproblems will beinvolvedbut theseareoutsidethescopeof the
conditionsconsideredinthisbook. Descriptionsof howrenewablesourcesarebeing
developedandthepossibleeffects of their dispersionwithintheestablishedpower
systemsarediscussedintheliterature.
Although the majority of power systems transmit at alternating frequencies, a
significant number incorporatedirect voltages. Thisrequiresspecial equipment and
introducesdifferent insulationproblems, someof whichareconsideredlater.
1.1.1 Alternating voltage systems
Themajor componentsof asystemwiththeir possiblerelativelocationsat apower
stationandinsubstationsareindicatedsimplyinFigure1.1.
Introduction 3
G C1 CB1
TML2 T3 SE
Substation SBS 3 Wood pole line and cable circuit Substation SBS 4 Cable to CC To local customers
INS TML3 and TML4 INS SE T4 CBX T5 T6 OH T7 CC
Power station Local substation SBS1 Transmission line SBS 2
SA
Fuses
Autocloser
Earthing devices
T1 SA VT ISO and E CB2 CT INS INS T2 TML1
Figure 1.1 Basic system for generation, transmission and distribution of AC power
The systems are based on three-phase configurations, although many of the
individual elements are single-phase. Each device must have appropriate elec-
trical insulation for its particular structure. Many of the methods by which this
is achieved are discussed in Chapters 35 and techniques for assessing the con-
dition of the materials when new and in service are described in subsequent
chapters.
At thepower stations thegenerators (G) may bedrivenby diesel (oil) engines,
gas turbines, water turbines or steamturbines thelast of thesebeing most usual
for thelarger machines. Generationvoltages inlargesystems rangefrom12kV to
24kV (perhaps up to 33kV in a few cases) with current ratings of 1500A up to
16,000A orlarger. Thesehighcurrentsarefedthroughcables(C1), ormetal-enclosed
bus conductors of largecross sections, to thelow-voltagewindings of thestep-up
generator transformers(T1). High-current circuit breakers(CB1) maybeinstalled
between the generator and transformers. The conductors required fromthe high-
voltage terminals of T1 are of reduced dimensions, thus allowing power transfer
by theuseof bareoverheadcablesthroughalocal substation(SBS1) andthenover
longdistances(TML1) or, withincities, throughfullyinsulatedundergroundcables
(TML4).
At thereceivingendof thevariouslines, astep-downtransmission transformer
(T2) is connected. Such units are often wound as autotransformers, especially if
thelower voltageis at an intermediatelevel (e.g. 145kV) for secondary transmis-
sion(TML2) or for supplyingacitys major distributionsystem. Thesystemfeeds
double-woundtransformers(T3) withoutputsof theorder of 6633kV (TML3 and
TML4) for reduction of thevoltages (T4T7) to customer operating levels in the
range 12kV to 415V/220V/110V. A cable-fed control cubicle (T7CC) is shown
for undergroundsupply to anumber of domestic customers. Largeindustrial orga-
nizations may purchase power at the higher voltages and install their own local
substation. The choice of voltage ratios, and the required transformer impedance
4 Condition Assessment of High-Voltage Insulation
valuesbetweenwindingsdependonmanyfactorsrelatedtotheparticular supplyand
loadconditions. Numerousbooksandtechnical papershavebeenpublishedonthis
subject [2].
At the major changes in voltage where primary lines (or generator(s)) feed a
numberof otherlinesasubstation(SBS1SBS4) isconstructedforcontrol of theindi-
vidual circuits: for monitoringthereal andreactivepower flowspossibly including
an optical fibre-coupled thyristor firing systemfor operation of static VAR com-
pensators (perhaps of the relocatable form[3]) and for protection of the system
when subjected to faults and overvoltages. The various devices, some of which
are represented in Figure 1.1, must be insulated for the different service voltages
including surges to ground and between phases. Switching and isolation are
provided by circuit breakers and air isolators (CB2 and ISO). Thecurrent magni-
tudes and steady state voltages are monitored by current (CT) and voltage (VT)
transformers of various designs. Surgevoltages dueto lightningandswitchingare
limitedbysurgearresters(SA)andairgaps forexampleacrosstransformerbushings
(T2), circuit-breaker insulationandattheentrytoasubstation. Whereahigh-voltage
conductor passes through an earthed tank abushing is required as in power trans-
formers, deadtank instrument transformers, someolder oil circuit breakersandin
gas-insulatedsystems(GIS). Theoverheadlines(TML1TML3) mustbesupported
withinsulator stringsor similar (INS) capableof withstandingthevariousvoltages
andadverseweatherconditions againrodgapsandsurgearrestersmaybeutilizedfor
protection.
Themachinefloors of asteam-turbine-generator and ahydro-generator power
stationaredepictedinFigures 1.2and1.3respectively. Thecomplexity of outdoor
substationsisindicatedinphotographs, Figures1.41.6, inwhichmaybeidentified
many of theitems in Figure1.1. Thecomponents in substation SBS1 arepresent
inoneformor other inall levels of high-voltagesubstations oftenincludingcable-
sealingends (SE) as inSBS3andSBS4. A largesystemwouldinvolvemany lines
andplant items.
Followingthedevelopmentof GIS, ithasbeenpossibletodesignandbuildcom-
pactsubstationsforveryhighvoltages. Manyexamplesof thisapplicationexistwhere
spaceislimited near or inmajor cities.
At thelower voltages muchmaintenanceis necessary to ensurehighreliability
of supply inthelocal distributionsystem. InFigure1.7areshowndifferent aspects
of suchasysteminabuilt-uparea. The415V housesuppliesarefedfromthe11kV
overhead lines through apole-mounted transformer, Figure1.7(a), and Figure1.1
(T6), by means of either overhead wires (Figure1.7(a)) or a three-phasecableto
cubicleslocatedseveral hundredmetresawayinahousingcomplex(Figure1.7(c)).
Eachunit (Figure1.7(c) andT7CC inFigure1.1) containsastep-downtransformer
withappropriateprotectionfor theoutgoing415V cablecircuits. Theitemsof par-
ticular interestinrelationtoinsulationarethesurgearresters, cableandsealingends,
11kV fuses, thevariouslineandstand-off insulators, theoil-filledtransformer and,
of course, thewoodenpoles. ThepoleinFigure1.7(b) wasbeingreplacedbecause
of termitedamagebutthisoperationcouldhavebeennecessaryfollowingalightning
strikeor evenabushfire.
Introduction 5
Figure 1.2 Steam turbine-generator [4] [reproduced by permission of CIGRE]
Figure 1.3 Hydro-generator [5] [reproduced by permission of CIGRE]
6 Condition Assessment of High-Voltage Insulation
Figure 1.4 330kV substation. Note (from left to right) current transformers, SF
6
circuit breakers, support insulators for air isolators/automatic earthing
arms [reproduced by permission of TRANSGRID, New South Wales]
Figure 1.5 330kV substation. Note
insulator strings and
corona rings [repro-
duced by permission of
TRANSGRID, New South
Wales]
Figure 1.6 132kV isolator with good
corona design. Note auto-
matic earthing arms in fore-
ground
Introduction 7
11 kV lines
11 kV cap and pin insulators
(b) (c)
(a)
415 V lines
11 kV fuses
Pin insulators
Surge arresters
House supply
Sealing ends
3-phase 11 kV cable
Figure 1.7 11kV and 415V local supply systems. (a) 11kV/415V pole-mounted
transformer; (b) 11kV overhead line to 11kV three-phase cable. Pole
maintenance; (c) A 415V cubicle substation.
Note surge arresters, 11kV cable sealing ends, 11kV fuses, line
insulators and 415V wires to houses.
8 Condition Assessment of High-Voltage Insulation
AC filter
Converter
transformer
Shunt capacitor
Converter
DC filter
Smoothing reactor
Figure 1.8 Principle of an HVDC transmission scheme
1.1.2 Direct-voltage systems
In effect a direct-voltage system is a hybrid circuit incorporating AC and DC
components (see Figure 1.8). The incoming power is froman alternating source,
whichisrectifiedandfilteredbeforetransmissionthroughtheDC system, inversion
takingplaceat thereceivingendinorder toprovidetheusual ACsupplyconditions.
TheharmonicsproducedbytheconvertersarereducedbyfilterscomprisingR, L and
Celements. Theearlier significantsystemsincludedthatfromtheSwedishmainland
to Gotland (150kV, 1954), theoriginal cross-Channel connection between France
and England (100kV, 1961), thecrossing between theNorth and South Islands
of NewZealand(250kV, 1965), the50Hz/60Hz tieinJ apan(125kV, 1965), the
link between Sardinia and the Italian mainland (200kV, 1967), the overhead line
fromVolgograd to Donbass (400kV, 1965) and the Pacific Intertie in the USA
(400kV, 1970).
Thesegroundbreakingsystems(andafewothers) incorporatedmercury-arctech-
nology, whichtendedtoreducetheattractionof HVDC transmissionduetovarious
operatingproblems. However, withthedevelopment of reliablehigh-voltage, high-
power thyristors, thesituationchangedandtherearenowmanysystemsworldwide.
Such schemes are well established, transmitting 60GW or more of the worlds
power [6]. Typical voltagelevels, powerstransmittedandlinelengths, together with
commissioning dates, are included in Table 1.1. The number of such schemes is
probablyapproachingonehundred.
Modern systems use two 6-pulse bridges giving a 12-pulse converter bridge.
Onevalve consistsof anumber of thyristors perhaps100series-connectedunits
for 600kV, each of which may be rated at about 8.5kV maximumpeak voltage
withstand capability [7]. Thenumber of thyristors required for 100MW is quoted
as18(comparedwith234thirtyyearsago) inReference8. Thecomplexityof these
structureshasresultedinrigorousinsulationtestingprocedures(seeSection7.11).
Theadvantagesinrespectof lowercoronanoiseandlosses, smallerwayleavesand
thecapabilityof beingabletoutilizecablesfor longlengthsbecauseof thereduction
inlosses comparedwiththethree-phaseAC equivalent may, insomeapplications,
Introduction 9
Table 1.1 Examples of HVDC transmission schemes: thyristor valves
System Voltage Year Capacity OHline Cable
(kV) (MW) (km) (km)
Skagerrak 250 1976/77 500 113 127
Vancouver 280 1977/79 476 41 33
NelsonRiver BP2 250 1975/85 2000 930
HokkaidoHonshu 250 1979/93 600 124 44
China 500 1987/98 1200 1100
ItaipuBP 600 1987 3000 783
CrossChannel 2 270 1986 1000 72
USSR 750 1985 1500 2400
RihandDadri (Delhi) 500 1991 1500 840
EWMalaysia 350 1995 1000 600
Garabi, Brazil BtB Converters 2000/2002 2200
ThreeGorges 500 2003 3000
Changzhou, China
offset theincreasedcosts of converter stations comparedwithacorrespondingAC
system. Thesefeatures will, of course, also beadvantageous whereenvironmental
requirementsareatapremium. Thereliabilitiesof asignificantnumberof theschemes
aremonitoredregularlybyWG14.04of SC 14of CIGRE [9]: thisreport covers28
thyristor valveand5mercury-arc valvesystems operatingduring1997/1998. Data
wereobtainedinitiallyin1968.
Of special interest in respect of insulation assessment and possiblemonitoring
aretheconverter transformers, whichmaybesubjectedtocombinedalternatingand
directvoltages, thesmoothingreactors, theoverheadlineinsulators, thebushingsand
especiallyanycables/accessories, particularlyasusedfor underwater crossings.
Withthenewsystemsutilizingvoltage-sourcedconverters(seeSubsection1.4.4)
itappearsthattheinsulationof equipmentmaybesubjectedtoperiodicimpulse-type
voltages[8], theeffectsof whichhavenot beenextensivelyinvestigated.
1.2 Insulation coordination
Insulationcoordinationdesignof powersystemsaimsatminimizingoutagesof major
itemsof plant andcritical circuitscausedby switchingor lightningsurges. Thetra-
ditional protectivemethodsusevariousformsof air gapsconnectedacrossparticular
equipmentortransmission-linecomponents. Becauseof thelackof matchingbetween
theV-T (volt-time) characteristics of thegaps andthoseof thenon-restoringinsu-
lationsin, for example, power transformers, theapplicationmay not beaseffective
as required. Also the gaps may allow the passage of high-value power frequency
follow-througharcs.
10 Condition Assessment of High-Voltage Insulation
Theselimitationshavebeenovercometoalargeextentbytheintroductionof surge
arresters(seeChapters4and7) incorporatingnonlinear resistors. Theunitsaremore
complexthangaps, buthavebetterresponsetimesandcansuppresspotential arcs. The
reliability of surgearresters has increasedgreatly, especially withthedevelopment
of the gapless type, which has raised confidence in their performance. Condition
monitoringunder steady-stateconditionsissometimesconsiderednecessary.
Inorder to assist intheplanningof insulationcoordinationof apower system,
international standards havebeen produced for determining appropriateinsulation
levels inrelationto theoperatingvoltages. Theselevels arebasedontheexpected
overvoltages that might beproduced by theoccasional power-frequency fault and
surgesduetoswitchingandlightning. Duringthedetaileddesignof thepowersystem,
estimatesof suchdisturbancesmustbemade. Thisisaverycomplexprocess, depend-
ingonmanyfactors. Technical discussionsandexchangeof datahavetakenplaceover
manyyearsthroughCIGREandtheIECworkinggroups, enablingagreedlevelstobe
setandmakingamajorcontributiontothedesignandconstructionof reliableandsafe
systems[10, 11]. Inthecaseof HVDCsystemssuchstandardizationisnotcomplete.
1.3 High-voltage test levels
Thetestvoltagesforpower-frequencysystems short-durationandsurge standard-
izedbytheIEC for preferredvaluesof U
m
arelistedinIEC 60071-1. It shouldbe
notedthatU
m
isthehighestoperatingvoltageclassificationof equipment(kV-RMS)
betweenphases, althoughthemajorityof testsaretoground. Thebasisfor thechoice
of thetest levelsandtheassociatedvoltageformsarediscussedbelow.
Thevoltageschosenfor agivenlevel of U
m
will dependonlocal conditions, the
typeof line, themethodof protectionadoptedfor surgesuppressionandanypossible
pollutionproblemsthat might affect thepower-frequencyperformance. Thevarious
choicescanbecomplex, requiringextensiveanalyses. Someguidanceregardingthe
conceptsandproceduresaregivenintheIECdocuments60071-2and60071-3[10].
The situation for HVDC transmission systems is not well established and the
requiredtest levelsaredeterminedby theuser andmanufacturer. IEC 60071-5[11]
does not includepreferredstandardizedlevels. IEC publication61378-2covers the
applicationof converter transformersinHVDC supplysystems[12].
Awiderangeof testsisappliedtotheindividual componentscomprisingthepower
supplysystems. Themainprovinghigh-voltagetestsfornewpower-frequencyequip-
ment involvetheapplicationof overvoltages. Thesetests includepower frequency,
lightningimpulseandswitchingimpulsedependingonthechosenvoltageclass, as
indicatedinTable1.2. Thevaluesinrelationtotheoperatingvoltageswerechosen
followingagreementwithintheindustrybasedonlong-termresearchandexperience.
Tablesarealsoavailablefor wet testswiththedifferent voltageforms.
Theformsof thevoltagesthat might exist inapower-frequencysystemaresum-
marizedinTable1.3togetherwithpossibletestshapeswhereapplicable. Therelative
breakdownstrengthsof non-restoringinsulationwhensubjectedtodifferentformsof
voltageareindicatedinFigure1.9for asimplesamplearrangement.
Introduction 11
Table 1.2 Possible test levels for particular system voltages. See IEC 60071-1 for
details and Table 1.3 for test voltage shapes
Highest system Short-durationpower Lightningimpulse
voltage, U
m
frequencywithstand withstandtest
(phasephase) test voltagetoearth voltagetoearth
kV(RMS) kV (RMS) kV (peak)
12 28 95
36 70 170
72.5 140 325
145 230 650
245 395 1050
Switchingimpulsewithstandtest voltages
Phasetoearth Phasetophasevalue
test voltage referredtophaseto
kV peak earthtest voltage
300 850 1.5 1050
362 950 1.5 1175
420 950 1.5 1425
525 1175 1.5 1550
765 1550 1.6 2100
Extensivedevelopmenttestingandexperiencewithprototypescontributedtothe
choiceof therelativetest levels finally selected for equipment-proving tests for
exampleas inTable1.2. Inaddition, it is vital that theequipment insulationstruc-
tureswithstandthetestvoltageswithaneconomicallyacceptablesafetyfactor (SF).
Thisfactor may besimply definedastheestimatedbreakdownvoltage specified
testvoltageforeachvoltageform. Thechosenvalueof SFallowsformanyconditions,
includingcomplexinsulationconfigurations, manufacturingvariations, non-uniform
electricstressdistributionsandthenatural (statistical) scatter inbreakdownstrength
of thematerialsof liquids, solidor gas/air. Thechoiceof safetyfactorsvariesamong
manufacturers andusers andis acritical part of thedesignandmanufacturingpro-
cesses. Ascondition-monitoringtechniquesareimprovedfor applicationduringthe
high-voltage test procedures, it may be possible to use lower SF values, thereby
resultinginamoreeconomicproduct.
Inpractical equipmenttheratioof theone-minute testvaluetooperatingvoltage
is as high as 2.83.5, for example transformers, bushings and switchgear. These
values haveservedtheindustry well andhaveensuredthat equipment designedto
withstand such test levels will operatesatisfactorily for 25years or more, sustain-
ing overvoltages caused by lightning, earth faults and someswitching events. If a
12 Condition Assessment of High-Voltage Insulation
Table 1.3 Shape of AC and impulse test voltages
Class Shape Frequencyand
timeduration
AC tests
1/ f
T
t
(1) f = 50or 60Hz
(2) VLF, e.g. 0.1Hz
(3) Resonancevoltage
(20300Hz)
(4) Inducedvoltagetests
(100/120400/480Hz)
(5) T
t
= 10secondsto
60minutes
Switching-impulsetests
1.0
0.5
T
p
T
2
T
p
= 250s
T
2
= 2500s
Lightning-impulsetests
1.0
0.9
0.5
0.3
T
1
T
2
T
1
= 1.2s
T
2
= 50s
Fast-impulsetests
1/f
2
1/f
1
T
f
T
t
100ns T
f
> 3ns
0.3MHz <f
1
< 100MHz
30kHz <f
2
< 300kHz
T
t
3ms
highstandardof maintenanceandin-servicemonitoringisincorporatedintheoper-
ational programmes, lives of 40 years arenowbeing predicted. FromTable1.2 it
will beseenthat for valuesof U
m
= 300kV andaboveashort-time-durationpower-
frequency test is not specified, aswitching surgetest (seeSubsection 1.3.3) being
considered moreappropriate. However, steady-state-test overvoltageproving tests
areretainedinsomeformor other. Inthecaseof transformers this includes apar-
tial dischargetest (seeChapter 7) at, perhaps, 1.5 U
m

3for 60minutesor more

for U
m
300kV units. In many cases such tests arecalled for on lower-voltage
equipment. Themethodsof producingthevoltagesfor test purposesandprocedures
Introduction 13
%
500
Impulse
Electrode arrangement
126 kV crest
60 Hz
1 second 1 min 2780 hrs
42 kV crest
300
200
100
0
Figure 1.9 Volt-time curve for treatedpressboardinoil 1/16-inch-thick pressboard
at 25

C [13]
for determininginsulationpropertiesaredescribedinChapter 6andapplicationsto
particular apparatusinChapter 7.
1.3.1 Power-frequency voltages
A power-frequencyovervoltagetest level of thesameformastheoperatingvoltage
hasbeenthetraditional methodfor provingtheintegrity of equipment insulationas
indicatedinTable1.3. Anincreaseinfrequencymaybenecessaryforinducedtesting
of transformersandinresonance-typetestsasincludedinthetable.
1.3.2 Lightning-impulse voltages
Theneedforteststoreducefailuresduetolightningwasappreciatedveryearlyinthe
development of transmissionsystems. Thisappliedparticularly tothetransformers,
as apower-frequency-induced test does not provetheintegrity of complex induc-
tive/capacitivestructureswhensubjectedtosurges. Theimplicationsareanalysedin
Chapter 5, whereit isshownthat thebehaviour under suchconditionsisadominant
featureintheinsulationdesignof largetransformer windings.
Inthe1920sand1930sitwasrealizedthatthereductionof theearthing(footing)
resistanceof thesupport towersof theoverheadlinesandtheprovisionof overhead
earthwires couldgiveamoderatedegreeof protection. It was possibleto measure
lightning currents during astriketo thetowers and to develop scenarios for deter-
minationof themagnitudes andforms of thesurges to beexpected. Theseinclude
inductionfromnear strikes, back flashoversduetohighearthresistancesanddirect
strikestothepower conductors. Thevariousmeasurements, andbasicobservations,
of strikes to lightning conductors for example at the Empire State Building in
NewYorkandinSouthAfrica enabledaconsensustobeestablishedregardingthe
shapeandmagnitudeof thestrokecurrents: intherangeupto150300kA withan
averageof perhaps25kA. Usingthisinformationinconjunctionwithcalculatedline-
surgeimpedances, theshapes andmagnitudes of expectedvoltages wereestimated
14 Condition Assessment of High-Voltage Insulation
andagreementreachedastotheformof standard wavestobeappliedtoequipment
for particular voltagesystems. Thebasicshapeis1.2/50microsecondsasdefinedin
Table1.3andselectedlevelslistedinTable1.2. Itwill benotedthatimpulse-voltage
magnitudes areof theorder of 3.5to 5times thepeak of theoperatingvoltages to
ground.
1.3.3 Switching surges
Asthevoltagelevelsof thetransmissionsystemsincreased, itbecameapparentthatthe
surgesproducedbyswitchingcouldbemoresignificantthantheshort-durationpower-
frequency overvoltages caused by faults and other operational abnormalities. The
magnitudesof thesesurgesarelessthanthelightningdisturbancesbut areof slower
rise-time and oscillatory in form. Thus, it was considered necessary to introduce
a test to cover this condition. Although the surges are, in effect, oscillatory with
frequenciesintherangeuptotensof kHz, thisprovedadifficultconditiontosimulate
asahigh-voltagetest. Becauseaninsulationfailureor flashover especially across
air clearances in thetransmission system wouldbeexpectedto occur duringthe
first peak, it was agreedwithin theindustry that awavehavingarisetimeof 250
microsecondsandadecayof 2500microsecondswouldberepresentativeandshould
be adopted as a standard test (see Table 1.3). Some of the recommended values
aregiven in Table1.2. Themagnitudes areof thesameorder as thepeaks of the
short-duration power-frequency test voltages specified in previous versions of the
standards.
1.3.4 Very fast transient tests (VFTT)
Followingtheextensiveapplicationof gas-insulatedswitchgear it becameapparent
thatveryfasttransientswerebeinginjectedintothepowersystemduringoperationof
disconnectors. Thesesurgeshaverisetimesof tensof nanosecondswithsuperimposed
oscillationsintherangeof tensof kHzupto100MHzasdepictedinTable1.3. Such
disturbancescanbedangerousif thegas-insulatedswitchgear equipment isdirectly
connected to, for example, a transformer. The problems are under close scrutiny
withinCIGRE andelsewhere(seeChapter 8of Reference14).
1.3.5 Direct-voltage tests
Direct-voltageacceptancetestsfor HVDCequipmentareof theorder of twotothree
timesthesystemoperatingvalues. Thevoltagesareappliedfor longperiodsof, per-
haps, 30minutestoensuretheinsulationsystemisinastable, chargedconditionand
thestressdistributionacrossmultipledielectricsisidentical totheservicesituation.
Allowancemust bemadefor systemswhereservicepolarity reversalsareinvolved.
Levels for converter transformers areincluded in IEC Standard 61378-2 [13] and
a wide range of tests for the complex components of the thyristor valves in IEC
Standard60700.
Under certain conditions direct voltages are used for testing AC equipment,
e.g. when withstand voltage testing of large rotating machines and cables where
Introduction 15
thechargingcurrentsaretoohighfor normal ACtestequipment. HV DCtestsources
arealsoincorporatedinHV impulsegenerators. AtlowervoltagesDCtechniquesare
appliedinthemeasurementsof anumber of material dielectricpropertiesasrequired
for monitoringtheinsulationcondition.
1.4 Power system developments
Duringrecentyears, inordertoachieveincreasedutilizationof theexistingpowersys-
teminfrastructure, therehasbeenatrendtowardsoperatingplantandlinecomponents
muchnearer totheir maximumratingsandfor longer periodsthanpreviouslyconsid-
eredadvisablefromanengineeringaspect. Suchdevelopmentstendthermallytostress
theequipment morehighlyandprobablyresult inthespecificationof fewer planned
time-based routine maintenance outages. The success of the changes will depend
greatlyonmaintainingtheinsulatingmaterialsandstructuresingoodcondition. This
applies to themorehighly stresseddesigns nowbeingintroducedas well as to the
ageingequipment that isrequiredtoremaininservice.
1.4.1 Reliability requirements
Anacceptablyreliablesystemoperatingatornearfull loaddemandsthatexistingand
newcomponentsmustbemonitoredeffectively. Themannerinwhichthisisachieved
inrelationtothesystemdevelopmentanditsoverall operatingandreplacementcosts
issubject tomuchplanning, necessitatingcomplex decisionsby theasset managers
of present-daypower supplysystems.
Thereliabilityrequirementsexpectedinasystemdependonthecontinuousavail-
ability of adequategenerationandtheefficient maintenanceof thesupply network
incorporatingoverheadlines andcables, together withthevarious associatedplant
items.
Statutory regulations specify that thevoltagelevels must bemaintainedduring
changesintheloadconditions. Failuretomeetthecontractual requirementscanresult
inhigheconomicpenalties. Thiswill applyif ashutdownisnecessaryandmayinclude
liabilityfor environmental damageduetomalfunctioningof plant. Asfar aspossible
thesafetyof personnel must beensured.
Someapplicationsof artificial intelligencefor incipient fault diagnosisandcon-
ditionassessment arediscussedinChapter 10. It is clear that effectivemethods for
monitoringof thevariousinsulatingmaterialsmust beincludedintheeconomicand
technical development of existingandnewpower systems.
1.4.2 Condition of present assets
Inmany organizations systemdevelopment includes thedeterminationof thevalue
of theexistingassetsintermsof predictedremaininglife. Thisisadifficult process
with plant ranging fromages of morethan 25 years, perhaps on theupturn of the
bathtub curve, through to modern equipment of high capital cost incorporating
updateddesignsandsomenewtypesof materials.
16 Condition Assessment of High-Voltage Insulation
Again, assessmentof theconditionof theinsulatingmaterialscanbeanextremely
useful tool in acquiring datafor theoverall costing procedures. This is especially
effectiveif methodsfor continuousmonitoringfromnewareincorporatedincritical
plant items, therebyenablingtrend statisticstobeobtainedfor theequipment type
andinsulationsystem.
1.4.3 Extension of power system life
Aspartof theproceduresrelatedtothedevelopmentof asystemitishighlydesirableto
considertechniquesforextendingthelivesof equipmentbeyondtheirexpectedretire-
mentdate. Muchworkisproceedingwiththereconditioningof plant, includinglarge
generatorsandtransformers. Therestringingof overheadlinesandthereplacementof
itemssuchascables, instrument transformersandswitchgear becomenecessary for
lifeextension. Theappropriatetimeforcommencingsuchworkdependsonmanyfac-
tors, thestateof theinsulationbeingof major significance. Reliableandinformative
data for insulation condition monitoring are essential as part of the decision-
making process. A range of established and newer techniques are described in
Chapters610.
1.4.4 New systems and equipment
Intheimmediatefuture, power systemtechnological developmentsinvolvinginsula-
tionwill includemoreextensiveuseof compositeinsulatorsfortheoverheadlines, the
increasedapplicationof gas-filledequipment possiblywithminimizationof theSF
6
content, utilizationof fibre-opticinstrumenttransformersandinstallationof plastic
inparticular XLPE cablesuptothehighest voltages.
Some of the newer developments could include the application of cable
wound generators, motors and transformers for direct connection to the transmis-
sion/distribution systems [15] and the use of high-temperature superconductivity
(HTS) cables [16] for large power transfers in metropolitan areas. HTS research
anddevelopment arewell advancedintheUSA, EuropeandJ apan.
Considerableactivityiscontinuinginthedevelopmentof HVDCasamajor con-
tributor to the overall expansion in developing areas and as a backup to existing
networks. Onasmaller scale, thetechnologyisbeingappliedfor local interconnec-
tions. A particular exampleof asmall-scaledevelopment (80kV, 50MW, 70km
XLPE cables) basedonvoltagesourceconverter (VSC) technology is describedin
Reference7. The220MW VSC Murraylink interconnection between Victoriaand
SouthAustraliaincorporatesa176kmlandcableandthe330MWVSCschemefrom
theUSA mainlandto LongIslanda40kmcable[17]. A comprehensivereviewof
thecomponentscomprisingaVSCschemeandpossibleoperatingcharacteristicsare
describedintheCIGRE BrochureNo. 269[18] preparedbymembersof WGB4.37.
Although experience with VSC systems is limited, the report is optimistic about
thefutureof thetechnology. Thehighest ratings of systems inservice(2004) were
150kV and350MVA.
Thevariousdevelopmentswill inevitablyrequirenewinsulationassessment and
monitoringtechniques.
Introduction 17
1.5 Future insulation monitoring requirements
Theabovesurvey of someof thecomplexities of apower supply systemhas high-
lighted anumber of components in which themaintenanceof electrical insulation
ingoodconditionis essential inorder toachieveefficient andsafeoperation. Such
objectivescanberealizedat theengineeringlevel only by theapplicationof appro-
priatemonitoringtechniques, inparticular thoseassociatedwithassessingthestate
of theinsulating materials in equipment when newand during lives of up to forty
years or longer. It must beacceptedthat insulatingmaterials inevitably deteriorate
withtime theratebeingvery dependent onusageandthequality of maintenance
achieved.
The choice of whether or not to incorporate simple or advanced monitoring
instrumentationwill dependonmany factors includingthereplacement/repair cost
of theparticular equipment and, probably moreimportantly, theoverall economic
effectsandassociateddisruptionof thepower systemfollowingamajor problemor
failure.
Atthelowervoltagestheapplicationof periodicsteeppulses[10] in, forexample,
HV motor-control devicesmayrequirespecial insulationmonitoringsystems. Also,
sensitivelocal distributionnetworkscouldjustifythedevelopmentof newtechniques.
Descriptions andanalyses of someof themoreadvancedindustrial monitoring
methods now in use or under development are described in later chapters. These
follow areview of thematerials applied, their location in particular power equip-
ment, possible fault conditions and details of established insulation assessment
techniques.
1.6 Summary
In this introductory chapter the principle of an AC high-voltage power systemis
presentedtogether withanindicationof thetypesof equipment involved. Theorder
of magnitudeof theoperatingandassociatedtestvoltagesarereviewed. Theconcept
of insulationcoordinationfor protectingpower equipment insulationfromdamage
duetolightningandswitchingsurgesisdescribed. Inaddition, thedevelopmentsin
HVDC transmissionareconsidered, includingrecent progressinlocalizedschemes.
Appropriate references are given. The need for future insulation assessment and
monitoringisemphasized.
1.7 References
1. J enkins, N., Impactof dispersedgenerationonpower systems, CIGRE Electra
199, December 2001, pp. 613
2. Weedy, B.M., andCory, B.J ., Electric Power Systems (J ohnWiley&Sons, New
York, 1998)
3. Luckett, M., HowVARscantravel, IEE Review, September 1999, pp. 20710
18 Condition Assessment of High-Voltage Insulation
4. Poidvin, D., SteamTurbine-Generator at Penly, Important Achievements of
CIGRE, December 1998(bypermissionof CIGRE)
5. Berenger, P., Hydro-Generator at Pouget, Important Achievements of CIGRE,
December 1998(bypermissionof CIGRE)
6. Asplund, G., HVDC using voltage source converters a new way to build
highlycontrollableandcompactHVDCsubstations, CIGRE2000, Paper P2-04
(onbehalf of StudyCommittee23)
7. Andersen, B., andBarker, C., A newerainHVDC?, IEE Review, March2000,
pp. 339
8. Lips, H.P., Voltage stresses and test requirements on equipment of HVDC
converter stationsandtransmissioncables, CIGRE 2000, Panel 2, Paper P2-06
(onbehalf of StudyCommittee14)
9. A surveyof thereliabilityof HVDCsystemsthroughouttheworldduring1997
1998, CIGRE 2000, Paper 14-02 (on behalf of Study Committee 14 WG
14.04)
10. IEC 60071InsulationCoordinationParts13
11. IEC 60071 Insulation Coordination Part 5: Procedures for HVDC Converter
Stations
12. IEC 61378-2Converter Transformers Part 2: Transformersfor HVDC Appli-
cations
13. Montsinger, V.M., Breakdowncurvefor solidinsulation, Electrical Engineer-
ing, December 1935;54:1300
14. Ryan, H.M. (ed.), High Voltage Engineering and Testing, 2ndedition(IET, UK,
2001)
15. Leijon, M., Dahlgren, M., Walfridsson, L., Ming, L. andJ aksts, A., A recent
developmentintheelectrical insulationsystemsof generatorsandtransformers,
IEEE Electrical Insulation Magazine, May/J une2001;17(3):1015
16. Superconductivity makes its power transmission debut, general review in
Engineers Australia, J uly1999, pp. 2630
17. Graham, J ., Biledt, G., and J ohansson, J ., Power SystemInterconnections
using HVDC Links, IX Symposiumof Specialists in Electrical Operational
andExpansionPlanning(IX SEPOPE), 237May2004, RiodeJ aneiro, Brazil,
SP151
18. BrochureNo. 269, CIGRE VSCTransmission, WGB4.37. SeealsoSummary
inElectra 219, April 2005, pp. 2939
1.8 Problems
1. Frompersonal observationsanduseof theWeb, identifyandrecordexamplesof
thepower systemcomponentsmentionedinthechapter. Takeadvantageof visits
topower stationsandHV substations.
2. What is meant by embedded generation and how is it utilized? Describe at
least twosuchpower sourcesandthemethodsfor connectingthemtothemain
high-voltagetransmissionsystem.
Introduction 19
3. Discussthereasonsfor thevarioushigh-voltagetestsasappliedtopower-system
equipment. Indicatewhy andonwhat bases weretherelativelevels andforms
establishedbytheindustry.
4. Describe the advantages and disadvantages of HVDC transmission systems,
including their relationship to theAC systems. Discuss theapplication of the
newer developmentsinvolvingVSC technology.
Chapter 2
Insulating materials utilized in power-system
equipment
Themaininsulatingmaterials
Characterizationof insulationcondition
Modesof insulationdeteriorationandfailure
Electrical operatingstresses
Thesuccessful operationof high-voltagepower-systemequipmentisverydepen-
dent on the correct choice of insulating materials and maintaining themin good
conditionthroughout their life. This requires knowledgeof thetypes of traditional
andcontemporarymaterialsavailableandhowtheywouldbeexpectedtobehavein
theparticular operatingenvironment, especiallyover longperiods.
Theacceptanceby theindustry of new materials is aslow process becauseof
cost restrictions, changes inproductiontechniques andtherequirement that ahigh
probability of reliablecontinuous performancefor periods exceeding 25years can
beachieved. Suchestimates arebasedonexperience, experimental test results and
statistical analyses. Numeroustest specificationshavebeenwritteninorder toassist
indeterminingsuchreliabilities. Someof themethodsarediscussedinChapter6. The
mostimportantdevelopmentswiththenewermaterialshavebeenassociatedwiththe
useof SF
6
gas in switchgear andtransformers, plastics in high-voltagecables and
differentformsof syntheticpolymersandglassfibresinmachines, power/instrument
transformersandinsulators.
Inthischapter arangeof insulatingmaterialsandtheir special areasof applica-
tioninthepower systemarereviewed. Thisincludesthewell-establishedmaterials,
as these still form much of the insulation, i.e. air, hydrogen, wood, porcelain,
glass, hydrocarbon oil, oil-impregnated paper, oil-impregnated pressboard, wrap-
pings of synthetic-resin-bondedpaper, resin-bondedwoodlaminates, resin-bonded
paper laminates, andthenewer materials.
In the review of insulating materials (Section 2.1) reference is made to the
electrical parameters permittivity (), resistivity (), dielectric dissipation factor
22 Condition Assessment of High-Voltage Insulation
(DDF) andpartial dischargecharacteristics(PD) andtodissolved-gas-in-oil anal-
yses (DGA), all of whicharesignificant inthemonitoringof insulationsystems in
power equipment. ThesequantitiesaredefinedinSection2.2andanindicationgiven
of howthey canbeutilizedto characterizetheconditionof thematerials. Someof
thepossibledeteriorationandfailuremechanismsof practical materialsarepresented
in Section 2.3, followed by consideration of therelationships between themagni-
tudesof breakdownstressesfor samples/prototypesandarangeof operatingstresses
(Section2.4).
An understanding of these various factors and the expected behaviour of the
materials enables the most appropriate techniques for insulation-condition assess-
ment tobechosenandassistsininterpretationof thecomplex output datarecorded
by the measurement and monitoring systems. Such appreciation contributes to
improved operation and maintenanceof thepower system, both economically and
technologically.
2.1 Review of insulating materials
Thematerialsarereviewedunder thegeneral headingsof gases, liquidsandsolids;
thelast of theseincludes composites andconditions whereanimpregnant is neces-
sary in order to obtain therequired dielectric strength. Detailed properties arenot
covered, asthesecanbefoundinvariouspublications for examplereferences1to3
andinspecificationsof insulationmanufacturers. Toassistinthechoiceof appropri-
atematerialsfor specific temperatures, aclassificationguidewasintroducedwithin
theIEC standardssystemin1957. A number of GuidelinesandStandards([S2/1] to
[S2/4]) havebeenissuedsincethattimecoveringdeterminationof thermal endurance
properties, identificationof insulationsystemsfor particular temperatureconditions,
associatedageingmechanisms anddiagnostics, appropriatestatistical methods and
functional tests for evaluating theexpected performancein service. In IEC 60085
[S2/1] theinsulation thermal classifications aretabulated according to therecom-
mendedoperatingtemperatures. Alsoincludedistheearlier methodof alphabetical
classificationasthisisstill usedinthepower industry.
Theprimeobjectiveof thepresentreviewistooutlinethemajorcharacteristicsof
theinsulatingmaterialsandindicateinbroadtermshowtheparametersrelatetopower
systemapplications. Somepractical configurationsaredescribedinChapters35.
2.1.1 Gases
Inadditiontoair, anumberof othergasesareusedasinsulantsinpowersystems. These
includesulphur hexafluoride(SF
6
), nitrogenandFreon(e.g. C
2
F
6
andC
2
F
5
C
1
) for
applicationsinequipment suchasswitchgear, cablesandtransformersandhydrogen
inlargeturbine-generators.
Theelectrical properties of air arewell documented, as it comprises themajor
insulationinmanycomponentsof theoverheadpower system. Insomecasesthepre-
dictablebreakdownstrengthandself-restoringpropertiesof air areusedinprotective
Insulating materials utilized in power-system equipment 23
1500
200/3000 s
neg.
1.2/50 s
neg.
1.2/50 s
pos.
200/3000 s
pos.
50 Hz
kV
p
1000
500
0
1 2 3 4
Metre
Figure 2.1 Relative air gap flashover voltages in non-uniform fields
devices, for examplerodgaps andgap-typesurgearresters. At very highvoltages,
and thereforelong gaps, it is found that switching surges will causeflashovers at
relatively lowvoltagesif themorehighly stressedelectrodeisat positivepotential.
This factor is of importanceinthedesignof transmissionlines andsubstations for
thehigher systemvoltages[4]. Theorder of magnitudeof thebreakdownstrengthof
anon-uniformfieldinair isindicatedinFigure2.1. Theactual flashover valueswill
dependontheconfigurationandtheatmosphericconditions.
Knowledgeof thecharacteristicsof air inthepresenceof moistureisextremely
importantbecause, incontrasttothemajorityof insulations, itmustperformsatisfac-
torilyunder adverseclimaticconditions. Adjustmentsof flashover voltagelevelsare
madewithrespecttohumidityvalueswhentestinginthehigh-voltagelaboratory. The
insulationcharacteristics of air at interfaces, as associatedwithporcelainandsyn-
thetic polymer surfaces, areof great importancein practical equipment. Flashover
curvesfor variousoverheadlineconfigurationsmaybefoundinReference5.
Thebehaviour of SF
6
hasbeenstudiedingreat detail, theearlier work beingin
the1950s. Investigationshaveincludedbreakdowntestsat pressuresupto0.8MPa
(8bar) for awiderangeof electrodeconfigurations relatedto switchgear, current
transformers, power transformers, substationhardwareandcables. Thishasbeenin
additiontostudiesof thechemical structuresandlong-termstabilityof theelectroneg-
ativegas. Itisnecessarytoachievedewpointsof 20

orbetteratworkingpressures
inorder toavoidmoistureproblems.
Inpractical applicationstheeffectof particleswithintheSF
6
gasof gas-insulated
systems(GIS) canbeverysignificant[4, 6]. Followingaflashover withinGISequip-
ment or disconnector operationdangeroussurgeswithrisetimesof theorder of tens
of nsmaybeinjectedintothelocal power system. Thevolt-timecharacteristicstend
tobeflat after about 10s(4).
Descriptions of the processes involved in air and SF
6
breakdowns have been
published widely in many research papers and a number of books. Both of these
24 Condition Assessment of High-Voltage Insulation
1000
100
Breakdown
voltage-
kV peak
10
SF6
Air
H2
1.0
0.1
2
0.001 0.01 0.1 1.0
Pressure spacing product (bar-mm)
10 100 1000
4 6 8 2 4 6 8 2 4 6 8 2 4 6 8 2 4 6 8 2 4 6 8
Figure 2.2 Approximate breakdown strengths of gases in a uniformfield Paschen
Curves
gases exhibit increased electric breakdown stresses at small spacings and in near
uniformfields[7]. TheapproximaterelationshipsbasedondatapublishedinElectra
Nos. 32and52aredepictedinFigure2.2, whichalsoincludesresultsfor hydrogen.
In thesetypes of fields onceapartial dischargeevent occurs completebreakdown
wouldbeexpected. Thisisespeciallysignificantwithdirectvoltages, as, forexample,
withintheair gapsof impulsegenerators(seeChapter 6).
Therelationshipfor air [8] canbeapproximatelyrepresentedbytheequation
V
peak
(kV) 6.72

pd+24.36pd (2.1)
where(pd) isinbar-cm. Theequationisvalidfor pd>0.1bar-mm.
Thereductioninstrengthfor non-uniformfieldconditionsfor air isindicatedin
Figure2.1. Insuchconfigurationshighstressesareproducedat sharpelectrodesand
will trigger theultimateflashover. However, partial breakdown can bemaintained
without failureinsomegaptypesresultinginthephenomenonof corona. Thismay
beobserved visually in air, for exampleon overhead lines and insulators, and has
beenstudiedvery extensively frombasic physical characteristics[8] throughtothe
effect of theassociatedfinitelossesontheperformanceof thelines.
Nitrogenis usedat pressures upto 1.0MPainstandardcapacitors andinsome
formsof cables, whilethelowdensityof hydrogenisexploitedinlargewater-cooled
turbo-generators. As shown in Figure2.2, thebreakdown strength of hydrogen at
atmospheric pressureis about half that of air. Operatingpressures areintheregion
of 0.4/0.5MPa with moisture contents corresponding to dew points of the order
of 20

C.
Insulating materials utilized in power-system equipment 25
2.1.2 Vacuum
Inits purestateahighvacuumis anideal dielectric over short distances, sinceno
electron multiplication is possible. However, in practical equipment such as high-
voltage circuit breakers, contamination fromthe metallic and insulation surfaces,
together withresidual oil andgases, limitsthevoltagestressesthat canbeachieved
[9]. Withgooddesignandappropriateelectrodematerials, vacuumcircuit breakers
arenowusedincircuitsuptoandincluding36kV.
2.1.3 Liquids
Theuseof oil asaninsulant isverycommon, either onitsownor asanimpregnant
for achievingthegoodproperties of alaminatedor porous solid material inthe
caseof transformersandsomedesignsof high-voltagecablesitactsasaheat-transfer
mediumbetweentheactiveconductorsandwater or air coolers.
Thetypeandqualityof oil requiredisdependentontheparticularapplication. The
specificationsrangefromnormal hydrocarbonoilsassuppliedfor useinswitchgear
andtransformersthroughtospecial typesfor cablesandcapacitors[10]. Whenused
as impregnants, the liquids are carefully dried, degassed and filtered to produce
structuresof highdielectricstrength.
Themotivation for using paraffinic-based oils in placeof naphthenic types
seemstohavediminished. Muchworkwascarriedoutontheformer, especiallywith
respect toviscosityat lowtemperaturesandageingcharacteristics[11].
Thequalitiesof thevariousoilsarecheckedbytestslaiddowninspecifications
[e.g. S2/5S2/7]buttheresultsdonotnecessarilyindicatehowtheliquidswill behave
whenbuilt intoacomplexstructureover alongperiod.
A simpleexampleis thereduction in partial dischargeandbreakdown stresses
withincreaseinspacingor volumeinauniformfieldgapasindicatedinFigure2.3
for 50Hz andimpulsevoltageconditions. Thegraphs arerepresentativeof awide
spreadof resultsfromseveral sources, asreviewedinreference[12], andareincluded
hereonlyasanindicationof animportanttrendwhenapplyingdatafromaparticular
sampleto aconfiguration of different dimensions. Theabsolutestress magnitudes
will dependontheconditionof theoil testedor beingusedintheequipment.
During recent years concern has arisen regarding theeffect of particles on the
strength of oil as utilized in high-voltagepower transformers. Someearlier exper-
imental results [13] showed that an increase in the density of suspended particles
(>5m) from2000/100cm
3
to 12000/100cm
3
reducedthebreakdownvoltageof
largeoil volumes by theorder of 40 per cent. Thetest electrodes wereconcentric
cylinders and theoil volumeapproximately 4 10
5
cm
3
. Theaveragebreakdown
stresses for low-particlecontents wereof thesameorder as indicatedinFigure2.3
(b). A later report by CIGRE [14] containsdatafrom15laboratoriesandanindica-
tionof theparticleconcentrationstobeexpectedinpractical transformers. Suchdata
areappliedindesign, althoughfurther informationis probably requiredrelatingto
partial-dischargeinceptionstresses, especiallyunder surgevoltages.
Themacroscopic behaviour of oil insmall andlonggaps has beenextensively
researchedandisstill theobject of experimental studies, e.g. References15and16.
26 Condition Assessment of High-Voltage Insulation
70
50
30
20
E

k
V
(
r
m
s
)
/
m
m
10
5
0.1 0.5 1 5
50Hz
1013
50
(b) (a)
(d) (c)
30
20
E

k
V
(
r
m
s
)
/
m
m
10
5
.01 .1 1 10 10
2
10
3
10
4
10
5
50Hz
200
100
E

k
V
(
p
e
a
k
)
/
m
m
50
40
30
20
0.1 0.5 1 5
1.2/50 s
10 13
100
70
50
40
30
E

k
V
(
p
e
a
k
)
/
m
m
20
10
.01 .1 1 10
Oil volume, c.c. Oil gap, mm
10
2
10
3
10
4
10
5
1.2/50 s
Figure 2.3 Reduction in oil partial-discharge inception and breakdown stresses
in uniform fields range of experimental results [12]. (a) 50Hz PD
inception stresses; (b)(d) Breakdown stresses
This work is aimedat understandingthebreakdownmechanisms and, moreimpor-
tantly, thepartial-dischargephenomenawithandwithoutcontaminants inparticular
moisture, conductingandnon-conductingparticlesandair.
Asthegas-absorbingcharacteristicsof oilscanvary, somedifficultiesmayarise
in theinterpretation of gas-in-oil analyses. Thesegas-absorbing oils arenowused
in transformers as well as cables andcapacitors. In thelatter cases they helpmin-
imize the formation of bubbles in the tightly packed insulation structures. A new
test incorporatingapoint-sphereelectrodesystemfor checkingthepartial discharge
characteristics of oils has been developed by CIGRE (SC 15) and is now an IEC
Standard[S2/8].
Anoil characteristicthatcontinuestobeof interestisconcernedwiththeelectro-
static chargingeffect producedby flowrates of, perhaps, 1.5metres/secondwithin
theinsulationconfigurationof certaindesigns of high-voltagepower transformers.
Extensivestudies havebeen madein J apan, theUSA and Europe. It appears that
localizedchargescanbebuilt uponinsulationsurfacesthat mayproducedangerous
partial discharges andevenflashover at theinterfaceor inthebulk oil. Techniques
havebeendevelopedfor assessingtheelectrostaticchargingtendencies(ECT) of the
Insulating materials utilized in power-system equipment 27
different oils (newand aged) on their own [17] or in conjunction with pressboard
surfaces[18], asalsoinvestigatedbyStudyCommittees12/15of CIGRE [19].
Syntheticliquidswereintroducedfor useindistributiontransformersmanyyears
ago in order to overcome the fire hazards associated with hydrocarbon oils. One
of thecommoner liquids consisted predominantly of polychlorobiphenyl (PCB), a
substancethat isnowunacceptablefor environmental andhealthreasons. Extremely
lowconcentrationsareallowedinexistingequipment for example<0.5ppm.
Thedestructionof PCB liquids has requiredthedevelopment of special collec-
tion and waste-disposal techniques, at considerable cost to the electrical industry.
An advantageof PCB liquids was their high permittivity when used in capacitors.
Replacementoils, includingsiliconeliquids, arenowavailableforuseinsmallertrans-
formersand, forpowercapacitors, awiderangeof individuallydesignedsyntheticoils
hasbeendeveloped[15, 16]. A well-establishedsynthetic liquidisdodecylbenzene
(DDB) for impregnationof wrappedinsulationinhigh-voltagecables. It isclaimed
that theliquidhas better ageingandgas-absorptioncharacteristics thanthenatural
oils. If low-temperaturecablesprovetobecommercially viable, it appearsthat liq-
uidnitrogenand/or heliumwill beconsideredaspossiblefluidsfor impregnationof
lappedplasticdielectrics.
2.1.4 Solids
Clearlythestructuresof all thepower-systemequipmentmustincludesolidinsulating
materialscapableof efficientlysupporting, andisolatingfromeachother, conductors
atdifferentpotentials. Suchmaterialsmusthaveadequatepunctureandcreep/tracking
strengths. They must be able to withstand the expected thermal, mechanical and
chemical conditionsandmaintainelectrical stressestogiveeconomicandtechnically
acceptabledesigns. Theordersof magnitudeof achievablestressesaresummarizedin
Section2.4forsampleandequipmentconfigurations. Thematerialsreviewedinclude
alimitednumber of themoreimportant syntheticpolymersappliedinpower system
engineering. The industrial development of these materials is continuous but the
insulationengineer must becautiouswhenassessinganewmaterial for aparticular
application. Thewell-establishedmight bethebest solution, especiallyinrelationto
long-serviceperformance.
2.1.4.1 Wood
Woodis oneof theoldest insulations usedby electrical engineers anddespitelim-
itations in the natural formit is widely applied. The outstanding application is in
overhead line systems, where its relative cheapness and insulation properties are
attractive[20]. It is alsoutilizedintransformers, someolder switchgear andgener-
atorsinalaminatedformsuitablydriedandglued/impregnatedwithresinstogivea
highmechanical strengthandacceptableelectrical properties.
2.1.4.2 Porcelain
For manyyearsporcelain, andtoalesser degreeglass, hadnocompetitor asaninsu-
lationfor overheadlineinsulators. It weatherswell, evenunder moderatepollution,
28 Condition Assessment of High-Voltage Insulation
and has good flashover characteristics, and methods havebeen developed to meet
thestringent mechanical requirements. A disadvantageisitsdensitywhenit isbeing
handledandthesusceptibilitytocrackor fracturewhensubjectedtosuddenphysical
shocks.
Althoughit hasarelativelylowpuncturestrength, porcelainisusedextensively
in bushings, high-voltageinstrument transformers, stand-off insulators andsimilar
components (seeChapters 4and5). Inmost of theseapplications theresistanceof
thematerial to atmospheric conditions and possibleflashover without catastrophic
failureistheinsulatingpropertybeinginvoked.
2.1.4.3 Glass
Inadditiontoitsapplicationfor shedsof overheadlineinsulators, glassisusedinthe
power industryinfibreformfor theproductionof insulationcomponents, including
tapes, tubes, boardsandthetierodsincompositeinsulators.
2.1.4.4 Rigid laminates
A largegroupof insulatingmaterials may beclassedas laminates. Theseinclude
paper andclothbasedboardsgluedwithresin, special plywoodseither fully or par-
tiallyimpregnatedwithresin, layersof gluedpressboardfor oil impregnationandthe
high-quality materials of glass fibrelays impregnated, for example, with silicone,
epoxyor polyester formulations[1, 3].
For external use, and in somedry-typeequipment, thechoiceof material will
oftendependonitstrackingpropertiesandresistancetodeteriorationduetomoisture
anddirt.
For applications wheregoodmechanical strengthor support is requiredinhigh
electricfields, asinpower transformers, thepresenceof voidsbetweenlayersof the
laminates can result in partial discharges. Also, thelosses in unsuitableresins can
produceexcessivelocal dielectricheatingandanyimpuritiesinlayersparallel tothe
fieldmaygivealowbreakdownstrength.
2.1.4.5 Sheet composites and tapes
Polymers areusedintheproductionof sheet composites, whichmay beformedas
requiredandcuredin situ, e.g. for machineslotinsulationandinterwindingwrapsin
dry-typetransformers. Theapplicationof Melinex/Mylar(polyethyleneterephthalate
film) withpolyester fibremat produces amaterial suitablefor Class 130(Class B)
temperatures. At higher temperatures, Nomex (polyamidepaper) for Class155and
Kapton (polyamide film) for Class 180 (Class H) and above (Class 200) may be
used for specific applications. Nomex is a sheet material also applied in dry-type
transformers impregnatedwithresin, usually under vacuum, operatingintheClass
130(ClassB) range.
These types of material may be used in tape form, especially for machines.
Research continues in order to develop tapes for insulating conductors and leads
suitable for operating at higher temperatures and more onerous conditions. Mica-
basedresinsystemsinformedor tapeconfigurationsarewidelyusedinhigh-voltage
Insulating materials utilized in power-system equipment 29
rotatingmachines bothmotorsandgenerators. Ananalysisof thebehaviourof mica
insulationsystemsispresentedinReference21. Thetechniquesarespecializedbut
areof considerableinterestinpowerstationoperations. Studiescontinueinrespectof
thepartial dischargecharacteristics. Theactionof themica-resinsysteminresisting
highPDlevelsisnotfullyunderstood. Thousandsof picocoulombsmaybewithstood
for manyyearsof operation.
2.1.4.6 Epoxy resins
Theuseof castresinsinpower engineeringiswell established. Manufacturersoffer a
widerangeof componentsinwhichdesignsincorporatingtraditional materialshave
been modified to exploit theadvantages of cast resins, e.g. thermoset epoxy resin
systems. Thefinal characteristicsareinfluencedby thetypeandquantity of filler
for example, glassfibre, silicaor other inorganic materials. Dielectric constantsare
intherange3.0to4.0dependingontheepoxyfiller.
Apart fromthermal difficulties dueto thedifferences inexpansioncoefficients
betweentheresinandconductors, amajor developmentproblemwastheelimination
of partial dischargesinvoids, asit wasfoundthat theresinsweresusceptibletoPDs
of lowvalue perhapstensof pCs possiblyleadingtoservicefailures. Goodquality
control isessential, inparticular themonitoringof partial discharges.
Onemajor restrictionwasthelimitationof cast resinsfor indoor useonly. After
muchR&Deffort, thecycloaliphaticepoxyresinswithappropriatefillerswereintro-
ducedforoutdoorapplications. Anumberof currenttransformershavebeeninservice
for many yearsandlong-termtestshavebeencarriedout onlineinsulators, but the
newer polymericcompositesseemtobesuperior inthislatter application.
2.1.4.7 Elastomers
Elastomers are a group of polymers that have rubber-like mechanical properties.
Theycanbemouldedintosuchcomponentsasinsulator shedsandextrudedtoform
medium-voltageHV cables. Inall applicationsawiderangeof fillersandchemical
modificationsarerequiredtoobtainsatisfactoryperformancecharacteristics. Thisis
especiallyimportant for outdoor insulatorsandshells(seeChapter 4). Therearetwo
major systemsinuseinthesupplyindustry.
Ethylene propylene rubber (EPR). For medium-voltage cables extruded EPR
is widely applied. It is flexible, does not formwater trees, is resistant to internal
partial discharges and has good electrical properties. Ethylene propylene hexadi-
enemonomer (EPDM) isapreferredformfor outdoor insulatorsbecauseit ismore
resistant to UV radiation and corona than EPR, although its long-termbehaviour
under pollutionisstill beingmonitored. Other typesareethylidiene-norborneneand
dicyclopentadien[22].
Siliconerubbers (HTV, LSR, RTV). Thehigh temperaturevulcanized formof
silicone rubber (HTV) is used for outdoor insulators as it has good resistance to
corona, ozone, UV andmanyformsof pollution. Thematerial ishydrophobic, which
resultsinsurfacewater formingasdroplets, andconsequentlyahigh-resistancepath
30 Condition Assessment of High-Voltage Insulation
for leakagecurrents, thusminimizingthepossibility of dry-bandarcing. Siliconeis
probablytheonlypolymer havingthisverydesirableproperty.
Liquidsiliconerubber (LSR) appearstobeadvantageouswherecomplexshapes
arerequiredandavoidanceof voidsisessential [23]. Room-temperature-vulcanized
siliconerubber (RTV) has beenappliedto porcelaininsulators inorder to improve
their wet andpollutionperformance[24]. A comparisonof thepropertiesof EPDM
andHTV siliconerubber is giveninReference1. Dielectric constant values areof
theorder of 2.53.5and3.34.0respectively.
2.1.4.8 Heat-shrinkable materials
Animportant development wastheintroductionof heat-shrinkablepolymericmate-
rials[25]. Thisledtochangesinthetechniquesadoptedfor 11kV (andabove) cable
terminations at switchgear andsimilar locations. Muchtestinghas beencompleted
inthelaboratoryandat outdoor test sites. Assessment includedthedeterminationof
thebehaviour of theshrunken material when subjected to thermal cycling, as in a
cable. Air gapsmust not appear betweenthesleeveandthecableinsulation(plastic
or oil-impregnated), asthiscouldresult inpartial dischargeswithsubsequent failure
(seeChapter 4).
2.1.4.9 Polyethylene (including XLPE)
Theuseof polyethyleneas an insulating material is attractive, becauseit has low
lossesandmoderatelyhighelectricstrength. Initially, itsthermal stabilitywasunac-
ceptableatthetemperaturesrequiredinpower engineeringandnotuntil cross-linked
polyethylene(XLPE)wasintroduceddiditfindwideapplication[1]. Thematerial can
beextrudedandwasfoundsuitableforcablemanufactureonceanumberof problems
weresolved. Theseincluded development of methods for curing and cooling long
lengths, atthesametimeeliminatingvoidsinwhichPDsmightbeinitiated. XLPE is
susceptibletodischargesof theorder of tensof pCsbut isnowbeingappliedasthe
major insulationincablesover awiderangeof voltage, afewashighas500kV (see
Chapter 4). Thedielectricconstant is2.3.
2.1.4.10 Polyvinyl chloride
Polyvinyl chloride (PVC) is used extensively for the insulation of wires and as a
sleevematerial for low-voltageapplications. Inmany power situationstheintegrity
of thesecondary wiringis vital. PVC has athermal ratingof 105

C, whichcanbe
increasedby suitableformulations, andadielectric constant of 3.04.0, depending
ontheform. Thematerial isresistant toarangeof liquidsbut isattackedby others,
inparticular aromatichydrocarbonspossiblyassociatedwithmigrationof plasticizer
fromthePVC [1].
2.1.4.11 Polytetrafluoroethylene
Polytetrafluoroethylene(PTFE)extrusions, mouldsandfilmsareusedwheredemand-
ing dielectric, mechanical, chemical and thermal conditions are encountered.
Specialist applications include insulators, cables, wires and windings where the
Insulating materials utilized in power-system equipment 31
high cost can be justified. The material is susceptible to corona and radiation.
Its dielectric constant is 2.0. Fluorinated ethylene propylene copolymer (PEP)
Teflon has similar properties to PTFE but is not as tough and has a more lim-
itedtemperaturerange. Itmaybeprocessedbyconventional extrusionandmoulding
methods.
2.1.4.12 Polypropylene
Polypropylene(PE) filmisoneof thematerialsusedasthedielectricinpower capac-
itors, having high electric strength and lowlosses. It has alowdielectric constant
andtheappropriategrademust beselectedtominimizeswellinginsomedielectric
liquidsasusedincapacitors.
Initsbulkformthematerial isutilizedfor mouldingof componentsandextruded
asinsulationforcablesoperatingatlessthan5kV [1]. Polypropylenetapeslaminated
withpaper (PPL) havebeenusedfor high-voltagecableinsulationfor tenyears or
more for examplebytheNational GridintheUK.
2.1.4.13 Liquid-impregnated insulation systems
At thehigher voltagesit seemsthat liquid-impregnatedsystemswill continuetobe
usedsuccessfullyatextra-highvoltagesincables, bushings, instrumenttransformers,
power transformers, distribution transformers and power capacitors, although gas
systems(SF
6
) arenowwell developedandarecompetitiveinanumber of applica-
tions. Despiteitsapparent disadvantages, oil-impregnatedpaper (OIP) hasproveda
reliableandeconomicinsulantinmanyapplications. Theachievementof theefficient
utilizationof natural materialshasresultedfromR&Deffortovermanyyears inpar-
ticular, thedeterminationof conditionsnecessary towithstandhighelectric stresses
for theexpectedlifetimesof at least 20to30years. Of special interest tothedesign
andoperatingengineersaremoisturecontent, gascontent of theimpregnant, losses
atoperatingstresses, partial-dischargeinceptionstresses, thelocationof anyPDsand
theageingcharacteristics.
Initial drying and impregnation conditions necessary to avoid failure are well
established and formpart of theproduction know-how. Vacuumovens capableof
pressures down to 0.1 torr or lower and temperatures of 100

130

are common.
Theprocessingtimes of 23weeks for alargetransformer may beshortenedvery
considerablybyinstallationof vapourphaseheatingequipment, forexamplebyusing
keroseneforheattransferintheearlystagesof dryout. Itiscommonfordrynesslevels
of lessthan0.1%tobeachievedinsuchprocessesbeforeimpregnation. Theoil would
beexpectedtohaveamoisturecontent of afewppm(e.g. <2ppm).
Thelossesmayberelatedtomoisturecontentaswell astothequalityof themate-
rial: alow-losspaper isusedincablemanufacture. Aneffect foundwhenattempting
toestablishbreakdowncriteriafor newOIP isindicatedinFigure2.4. It isessential
thatlowmoisturecontentsbemaintainedinthepractical systemsif thermal runaway
conditions andPDs areto beavoided(seeSection2.3). This is particularly impor-
tant withcables inwhichtheconductor losses pass throughtheOIP andaddto the
dielectriclosses.
32 Condition Assessment of High-Voltage Insulation
100
80
60
Range of
breakdown values
Moisture content (%)
0 1 2 3 4 5
R
e
d
u
c
t
i
o
n

i
n

e
l
e
c
t
r
i
c

s
t
r
e
n
g
t
h

(
%
)
40
20
0
Figure 2.4 Effect of moisture on electric strength of oil-impregnated paper
insulation
In high-voltagecables and somecurrent transformers, arrangements aremade
to seal thesystemagainst theatmosphere, maintainingamoisturelevel of 0.5%or
less, whilein power and distribution transformers thesituation is less critical, the
stresses being lower and theinsulation structurenot so compact. This also means
that gas-absorbent oils arenot usually requiredintransformers, althoughthereare
indicationsthatsuchoilsarebeingprovidedasthenorm. Incables, powercapacitors,
instrument transformersandcertainregionsof power transformerstheexistenceof a
gasbubblecanbeverysignificant, asitmayleadtoPDsthatareconfined, eventually
leading to deterioration of the tape or sheet material. In the degradation process
it appears that low-level ionization can producea formof cross-linking in theoil
impregnantresultingintheformationof thewell-knownX-waxcompoundcontaining
carbonandhydrogen. After disassemblythepresenceof thecompoundisoftentaken
as anindicationthat discharginghas occurredinthepaper/oil system. X-wax may
be detected by magenta dye or ultraviolet light. It can be observed as a smooth
cheeselikesubstancewithinthesurfaceof thepaperinsulation. Theeffectisdiscussed
inReferences26and27.
Aspecial situationinoperatingoil-filledpowertransformersiswherethemoisture
content has risen to perhaps 2.54 per cent and during an overload theconductor
temperatures areallowed to increaseto 140

(IEC 60354) or 180

, as specified in
theIEEE/ANSI standards for higher-temperaturematerials. Under suchconditions
it seems possiblethat gas bubbles may bereleased resulting in partial discharges.
Theeffect of this will dependontheir locationandtheelectric stress values. As it
was suspectedthat anumber of transformer failures may havebeen dueto bubble
formation during overload conditions [28], thephenomenon has been investigated
in some detail, including a joint project between Monash University and Electric
Power Research Institute. It was demonstrated in a separate research programme
Insulating materials utilized in power-system equipment 33
that problems in the transformer interwinding insulation might arise at the higher
temperatures[29].
Extensivedevelopmentsinpowercapacitorinsulationsystemsincludedtheintro-
ductionof polypropylenefilminconjunctionwiththemoretraditional paper, thelatter
actingasanimpregnatinginterface. However, for anumber of years, all-filmcapaci-
tors, inwhichahazy orsimilarpolypropylenefilmisusedtoenableimpregnationto
takeplace, havebeeninservice. Capacitorinsulationisverysusceptibletoimpurities
becauseof thethicknessof only10micrometres(m) andtheveryhighstresses(tens
of V/m) used. Thedielectriclossesinthenewdesignsarevery lowandnolonger
amajor limitationindesign, givinglowDDF values. Thedevelopment of insulating
materialsfor high-voltagecapacitorscontinuestobeactive.
2.2 Characterization of insulation condition
Anumberof measurableparameterscanbeemployed directlyorindirectly tochar-
acterizetheconditionof insulatingmaterialswhenbuilt intosystemequipment. The
predominant electrical characteristicsarethevaluesof permittivityandcapacitance,
resistivity andinsulationresistance, insulationtimeconstants, dielectric dissipation
factor andpartial dischargestatus, all of whicharedefinedinSubsections2.2.15. In
oil-insulatedapparatusthelevelsof moisturecontent, dissolvedgasvolumesandvar-
iouschemical quantitiesareindicativeof theoil/solidcondition. Thesearediscussed
inSubsection2.2.6andinlater chapters.
2.2.1 Permittivity () and capacitance (C)
Therelativepermittivity,
r
, istheratiobetweencapacitancesof identical electrode
systems with and without a dielectric present. The relationship is the capacitance
C = k
r

o
, wherekisaconstantrepresentingthegeometrical structureof thesystem
and
o
= 8.85410
12
. Thevaluesof
r
rangefrom1for air throughto5.5or sofor
porcelain. Capacitancevalues rangefromafewpF (10
12
farads) for cap-and-pin
insulatorsthroughtoF (10
6
farads) for cables, generatorsandpower capacitors.
In addition to the requirements of measurement systems (see Chapter 6) relative
permittivity magnitudes areimportant inthealternatingvoltagedesignof multiple
dielectricstructures(seeChapter3). Atpowerfrequencieslittlevariationoccursinthe
valueof
r
withappliedstressor temperaturefor thematerialsusedinpower system
practice. Thepresenceof moisture(
r
80) or trapped gas might beexpected to
increaseor decreaserespectivelythemeasuredvalues.
2.2.2 Resistivity () and insulation resistance (IR)
Volume resistivity, , isdefinedastheresistancebetweenoppositefacesof a1-metre
cubeof theinsulation.
TheresistanceR = L/A, whereL is thespacing between electrodes and A
is thearea of theelectrodes. Theunit of volumeresistivity is theohm-metre, the
34 Condition Assessment of High-Voltage Insulation
absolutevaluesof whichrangefrom10
9
to10
10
for oilsthroughtotheorder of 10
14
for materialssuchasXLPE.
Surface resistivity maybedefinedastheresistancebetweentwooppositeedges
of a unit square of material. It is an important quantity for assessing the state of
surfacesusedinair andgassystems.
For a given material the values vary widely depending on impurity content
(especially moisture for hygroscopic dielectrics) and temperature. In addition the
initial insulation resistance will change with time of application for a constant
direct voltage stress. This property is related to the time constant (see Subsec-
tion 2.2.3 and Chapter 6). Therelativevalues of areimportant in thedesign of
direct-voltagemultiple-dielectric systems suchas transformers, bushings andcable
terminations/joints. Thedifferencebetweendirect-voltageelectricstressdistributions
andthosewithalternatingvoltagescanbeverysignificantduringtestingof equipment.
2.2.3 Time constants
Another parameter that can berelated to someaspects of theinsulation quality is
the time constant, as measured by applying a direct voltagepulseto thesystem,
measuring the rise time required to reach a predetermined value, discharging the
sample(or equipment) for aknowntimeandmeasuringtherecovery voltage. Such
characteristics are influenced by the complex impedance and frequency response
of theequivalent capacitanceandresistancenetwork. Theeffect was studiedinthe
1950s, beingdesignatedthedispersion effect, andwasappliedtothedetermination
of moisturecontent ininsulatingmaterials inparticular power transformers. More
recent work hasresultedinthedevelopment of improvedtechniquesasdiscussedin
Chapter 6.
Thetimeconstantchangesalsobecomeapparentwhenmeasuringtheinsulation
resistance.
2.2.4 Dielectric dissipation factor
Anideal dielectric or insulatingmaterial may beconsideredas anon-conductor of
electricityactingasapurecapacitancewhenbuiltintoequipment. However, inprac-
ticeconductionoccursandfinitelossesareproducedthatmaybeof greatimportance
duringtest andoperatingconditions. Under alternatingvoltages thelosses arerep-
resentedby aslight changeinthecurrent phaseinrelationtotheappliedvoltageas
depictedinthephasor diagramof Figure2.5.
Angle represents theusual power factor value, I
R
/I =cos , whileangle is
designatedtheloss angle. I
R
representsthelossesandI
C
thechargingcurrent. Both
of thesecanbemeasuredasdescribedinChapter 6. Theratioof I
R
/I
C
=tan andis
definedasthedielectric dissipation factor (DDF).
For good insulation, tan is of the order of 13 10
3
, even for dry oil-
impregnatedpaper atroomtemperature. Thevaluefor XLPE ismuchlower, perhaps
0.310
4
. Withinthisrangeof values, tan , enablingtheDDF magnitudetobe
conveniently quotedinmilliradians (mR) as measured. Thecosineof thedielectric
Insulating materials utilized in power-system equipment 35
I
C
I
I
r
V

Figure 2.5 Phasor diagram for practical (lossy) dielectric

phaseangle(orthesineof thedielectriclossangle)maybedesignatedasthedielectric
power factor bysomeauthorities.
2.2.5 Partial discharges (PD)
Accordingto IEC Publication60270[S2/9], apartial dischargemay bedefinedas
localizedelectrical dischargethatonlypartiallybridgestheinsulationsystembetween
conductors, andthatmayormaynotoccuradjacenttoaconductor. TheStandardstates
that PDsareusually duetolocal electrical stressconcentrationsintheinsulationor
onitssurface, thepulsedischargeshavingdurationsof muchlessthan1sinmost
cases. Thepulse-less dischargesingasesareof amorecontinuousformandwould
not bedetectedbythetechniquesdescribedinIEC 60270. Thelatter alsonotesthat
coronaPD disturbances areassociatedwithconductors inagaseous mediumaway
fromsolidor liquidinsulants. Thetermis not to beappliedto other forms of PDs.
It is observedthat sound, light, heat andchemical reactions areoftenproducedby
partial discharges.
Inelectrical termsPDsmanifestthemselvesattheterminalsof samplesandequip-
mentasverysmall-magnitudepulses(V tovolts) of risetimesinthenstosrange.
Theintensitiesof thePDsareusuallyrepresentedbythechargelevel inpicocoulombs
(pC) ornanocoulombs(nC) butotherparametersarealsoof significanceinestimating
possibledamagingeffects(seeSubsection2.3.3).
Acceptablemagnitudesproducingminimal deteriorationinservicerangefroma
fewpCs(10
12
C) inXLPE andGISthroughtotensof pCsinoil-impregnatedpaper
andresins, toseveral thousandsinmica-basedconfigurations. Higher valuesmaybe
allowedinshort-durationovervoltagetests.
2.2.6 Physical and chemical changes
Manydeteriorativeprocessestakeplaceduringtheoperationof insulation. Someof
theseresult inby-products, whichmay beusedfor identificationandmonitoringof
36 Condition Assessment of High-Voltage Insulation
degradation. This is especially thecasefor oil-immersedsystems, wheredetection
of combustible gases and of furans in the oil has resulted in the development of
specialized techniques for assessing the condition of the materials. The methods
utilizeDGA andtheapplicationof liquidchromatography(HLPC).
An important practical insulation condition indicator associated with oil-
immersed equipment is the analysis of the gases and their relative concentrations
asproducedbyparticular typesof faultinthestructure. Possiblediagnosesarequan-
tified in the revised IEC Publication 60599 [S2/10] and discussed by Duval (see
Chapter 6). Theanalysesgivesignaturesindicatingthepresenceof low-energypar-
tial dischargesthroughtohigh-energyfaultsduetoarcing. Interpretationof thedata
obtainedbysuchtechniquesisconsideredinChapters9and10.
Thephysical ageing of paper as represented by thereduction in thedegreeof
polymerization(DP) hasbeenrelatedtotheamount of furaniccompoundsproduced
at hot-spot temperatures and absorbed in the oil of transformers in service [30].
A reduction in the magnitude of the DP froma value of >1000 (2-Furaldehyde
00.1ppm) whennewto250(2-Furaldehyde>10ppm) isconsideredtobetheend
of life for oil-impregnated cellulose by some authorities. (See also Reference 39
of Chapter 6.) Much work is being carried out in this field, including studies for
assessing the characteristics of naturally aged oil-impregnated materials aimed at
estimatingprobablelifetimes for example, Reference31. Theformationof water
under certainconditions, suchasinsealedunits, maybeof considerableimportance
inthepredictionof therateof ageinginoil paper systems.
Conventionally, oil acidity, interfacial tensionandsludgingtendencyareusedas
physical andchemical guidesfor monitoringtheconditionof transformer oil [S2/6].
Determinationof thephysical changesassociatedwiththevariousformsof treeing
andtrackinginXLPEandresinshasbeentheaimof muchresearch. Theunderstanding
of themechanismof water treeinginXLPE isof considerablepractical significance
andremainsanimportant researchtopic. Thisworkrequirestheremoval of samples
fromequipment suspectedof deteriorationandnolonger suitablefor normal service
andthereforeisonlyanindirect monitoringtechnique.
2.3 Modes of deterioration and failure of practical insulating
materials
Theconceptof insulatingmaterialshavinganintrinsicbreakdownstrengthwaspostu-
latedmanyyearsago. Itwasfoundthat, for thinspecimensof certainpuredielectrics
andbyminimizationof edgeeffectsinthetestconfiguration, theimpulsebreakdown
stressescouldbeincreasedtoperhapstentimesthatachievablewithnormal materials.
However, inpractice, impurities andmanufacturingvariations prevent exploitation
at alargescaleof thesepurelyelectronicprocesses.
Themajordeteriorativeandfailuremodesassociatedwithpowerequipmentinsu-
latingmaterials, whichoftencomprisecombinationsof solidandgas/air or solidand
liquid, arelistedbelow. Thisisfollowedby considerationof someof theprocesses
Insulating materials utilized in power-system equipment 37
involved. Inthecaseof thephysical andchemical changes, thepresent treatment is
limitedtomentionof thoseaspectsconsideredhelpful inunderstandingthecondition
monitoringrequirements.
(a) Dielectriclossescausingthermal instability, or runaway, inthebulkof thesolid
material.
(b) Partial dischargesrepresentativeof
partial breakdown in voids and gaps enclosed within the solid insulation
producinglocal erosionof adjacent material;
partial breakdown in theambient medium(oil, gas or air) at an interface
betweenanelectrodetodielectricor adielectrictodielectric, therebyinitiat-
ingflashover or creepand, perhaps, localizedpuncturethroughanadjacent
surface.
(c) Ageingduetothermal, electrical andmechanical stressing, includingtheeffects
in(a) and(b).
(d) Long-termchemical changes produced by incompatibility between materials,
resultinginthecreationof dangerousby-products. Thechangesarealsorelated
to(c).
(e) Deteriorationof surfacematerial byexternal pollutants, leadingtoreductionin
trackingstrength.
2.3.1 Dielectric losses
Thefailureof insulatingmaterialsduetohighdielectriclossesisawell-knownphe-
nomenon. Inpracticetheeffect is dueto theuseof, for example, unsuitableresins
andoftentothepresenceof moisture. Thelatter maybeproducedbychemical age-
ing, in somecases becauseof poor equipment sealing and, in others, by incorrect
processingprocedures inthefactory. Additionally, thesummationof theeffects of
energydissipatedbyanyhigh-valuepartial dischargescancontributesignificantlyto
theoverall dielectriclosses.
2.3.1.1 Determination of loss relationship
A measureof thetotal lossesmaybeobtainedbyconsideringtherelationshipbetween
theappliedvoltageandthecurrentflowinginasimplecapacitor. Therelevantphasor
diagramisgiveninFigure2.5asdefinedinSubsection2.2.4. Assuminganangular
frequencyof thelossunder alternatingvoltageconditionsis
W= VI
R
= VI
C
tan
= V
2
C tan watts (2.2)
whereV isthevoltageinvolts(RMS) andC isthecapacitanceinfarads.
38 Condition Assessment of High-Voltage Insulation
tan
0.08
0.01
20 90
Temperature (C)
Figure 2.6 Possible variation of tan with temperature for multi-layer clamping
board
If C is assumed to be a cube of metre length sides then C =
r

o
farads and
equation2.2becomes
W= V
2

o
tan watts/cu. metre
= E
2

o
tan watts/cu. metre (2.3)
whereE =volts/metre,
o
= 8.85410
12
and
r
= relativepermittivity.
Inastabledirect-voltagesystemthelosses wouldbesimply representedby the
I
2
Rvalues. Losseffectscanbedeterminedfromcalculationssimilartothefollowing.
2.3.1.2 Example of loss calculation
The relationship given in (2.2) may be applied for determining possible thermal
instabilityinequipment. Instudiesrelatedtotheoperationof oil-impregnated-paper
power cables it was shownmany years ago that theconductor losses, addedto the
dielectric losses, couldleadto acontinuous increaseinthepaper temperature. The
calculationassumesareferencetemperatureontheconductor, e.g. 90

, animposed
outerscreentemperature, knowledgeof thevariationof tan withtemperatureandthe
dielectric thermal conductivity. Anallowanceis madefor thelogarithmic variation
of stress(E) throughtheinsulation[26].
A simpler calculationhas beenmadefor anoil-filledtransformer withwooden
laminatedwindingclampingboards. Inthiscaseitwasrequiredtodeterminetheorder
of magnitudeof electricstressthat wouldbeexpectedtoproducethermal instability
in5cm- and10cm-thick boards, withtheoil at atemperatureof 90

. Thevariation
of tan withtemperaturewasassumedtobeof theforminFigure2.6. Thecalculated
temperaturedistributions throughtheboards for arangeof electric stresses for the
twocasesareindicatedinFigure2.7. Thetemperatureriseduetoaknownlossinthe
dielectricmaybedeterminedfrom(2.4).
= (1/
T
)(d/A)W (2.4)
Insulating materials utilized in power-system equipment 39
90
Surface
temperature
to give 90 C
at centre
80
5cm thick
10cm thick
Temperature at centre
Temperature drop
0 0.8 1.6 2.4
Average stress (kV/mm)
3.2 4.0

T
1
A
W =

E, I
70
60
50
40
30
20
10
C

T
= 0.24 W/ C/m
Figure 2.7 Calculated temperature distributions through insulation boards for
variation in electric stress
where = temperaturedrop through adielectric with thickness of d metreand
cross-sectional area of A square metres. A thermal conductivity of
T
= 0.24
watts/

/metreisassumed.
Asanexample, thedielectricmaybedividedintotensectionsfor thecalculation
andthemagnitudeof determinedforeachusingthevalueof internal lossdissipated
andthat duetothermal flowthroughthesection.
2.3.2 Partial discharges sources, forms and effects
Inordertoappreciatethesignificanceof partial dischargesandhowtheycaninfluence
thelifeof theinsulatingmaterialsit isnecessarytounderstandwhereandhowthey
might occur and the manner in which they can lead to deterioration and ultimate
failure. Thesefactorsareconsideredbrieflyinthefollowing.
2.3.2.1 Sources of partial discharges
Under alternating voltage conditions the PD sources are usually associated with
trapped gas or the ambient medium. Such components of a combined insulation
systemhave lower permittivities and breakdown strengths than the solid materi-
als andthereforetendto betheinitial locationof thelocal partial discharges. This
aspectisconsideredinChapter 3together withdescriptionsof someof theinsulation
configurations in various equipment. Despitethedifferences in thestructures it is
40 Condition Assessment of High-Voltage Insulation
5
2
1
(i) Types of internal
discharges in solid materials
( ) = Idealised shapes
(iii) Internal discharges due
to metal surface/particles
embedded in insulation structures
(v) Surface discharges due to
sharp electrodes at dielectric
(vi) Discharges (corona) from
uninsulated highly stressed
conductors
(vii) Discharges created by changes in
potential differences e.g. bushing outer
foil and transformer winding shield
VV
1
C
2
V
Foil or
shield
Break
Earth
Flexible
connection
joint to foil
within
insulation
As designed V
1
=0
After break
V
1
=VC
2
/(C
2
+C
1
)
C
1
(iv) Discharges in conductor/
surface and dielectric/surface
wedges (A)
(ii) Internal discharges in
laminated materials and
barriered structures
3
(a)
(1) (2) (3)
E
HV
HV
Earth Earth
PDs at
negative
peak in
air
PDs at
positive
peak in
air
(b) (a) (b)
4
E
E
E E E
(2) (2)
e
Foil

Gas/liquids: Thin ducts or gaps(2)

Ambient
A
Solid
A
Figure 2.8 Examples of some partial discharge sources
possibleto identify anumber of common situations whereelectrical overstressing
mightoccur resultingintheinitiationof partial discharges. InFigure2.8aredepicted
somepossibilities. Theseexamplescover internal voidsor cavitiesinsolidandlam-
inatedinsulation, relatively highstressinginambient mediaat electrodeor support
interfaces, sharpedgesatsurfacesandinfreespace(gas/air or liquid) andconditions
whereapotential differencemight becreatedduetoabrokenor poor connection. It
shouldbenotedthatPDscanexistatearthaswell asathighvoltage. Inair-insulated
systems, suchas overheadlines, discharges areusually designatedcorona or even
radio interference, and includethosefromhardwarecomponents in, for example,
substations.
Insulating materials utilized in power-system equipment 41
Underdirect-voltageconditionsthepotential distributionsacrossinsulatingmate-
rials aregovernedby therelativevalues of theresistivities that in oil-impregnated
systemscanresult intheoverstressingof thesolidinsulantsandnot theliquid. The
complexchargingstateswithdirectvoltagesmayproducepartial dischargesthatneed
tobemonitored.
2.3.2.2 Forms of partial discharges
Themodel normallyusedfor partial-dischargeanalysisisdepictedinFigures2.9(a)
and2.9(b). Thecavity, theinsulationbetweenthecavity andtheelectrodesandthe
remaining bulk insulation are represented as capacitances C
c
, C
b
and C
a
respec-
tively thea, b, c representation. Theconfigurationwas suggestedby A. Gemant
andW.V. Phillippoff in1932andanalysedfurther as discussedby Whitehead[32]
andothers[33, 34].
The representation is oversimplified, as it does not allow for such factors as
the existence of by-products, the finite leakage resistances (especially in the cav-
ity), thechanges dueto pressurebuild-upandthevariationof dischargingsurfaces
withinthecavity. However, theconceptishelpful inunderstandingthepulsechanges
produced. Following breakdown of the gas in the cavity due to the instantaneous
appliedvoltage, asteeppulseisproducedacrossC
c
andmaybedetectedat theelec-
trodes (or terminals). In Figure 2.9(c) is shown the effect of multiple discharges.
The distribution of pulses within the power-frequency cycles will depend on the
valueof theinstantaneous inception voltage(V
i
) in relation to that of theapplied
voltage (V
pk
). Thus if inception is near the peak, the pulses occur only on the
rising quadrants, or near the peak; but, if the applied voltage is raised above the
inceptionvalue(V
i
V
pk
) as is oftentheconditioninpractice they wouldbe
expectedto spreadacross thezeros as indicatedinFigure2.9(c). This concept was
described earlier by Austen and Whitehead [33] and is discussed in, for example,
Reference34. However, theresults of practical measurements oftendonot support
thesehypothesesasnotedbyWhiteheadwhenobservingthat, althoughinternal dis-
charges tendto congregatetowards thezeros withincreaseinvoltage, they do not
readilyextendintotheprecedinghalf-cycle. Anexampleof thistendencyisgivenin
Figure2.10, whichistheoutput of acomputer-basedpartial-dischargemeasurement
system[35].
Theeffectsdonotdevelopsimilarlywithhighlystressedelectrodes(seeFigure2.8
(vi)) locatedinself-sealingdielectricssuchasair andliquids. Under theseconditions
the pulses are often very distinctive, tending to cluster at the voltage peaks. The
special caseof coronainair wheredisturbancesoccur atthenegativepeakfor ahigh-
voltage point may be explained by considering the gaseous ionization conditions
around thehighly stressed electrode[34]. An equivalent diagramin which theair
gapis assumedtoberesistiveandinseries withacapacitance, C
c
, associatedwith
thespacechargeat thepoint electrode, is suggestedby Kind[36]. This hypothesis
representsthepulsesandtheir phaseshift inrelationtothepower-frequencyvoltage
peak. Thedifferenceinformandlocationinthe50Hzcycleof thepartial discharges
mayassist indiscriminatingbetweenexternal andinternal discharges.
42 Condition Assessment of High-Voltage Insulation
Void
V
C
a
C
c
C
b
V
v
Insulation
Voltage
Applied voltage V
Void Voltage
(no discharge)
Void voltage
(partial discharges)
Vv
0
(a)
(c)
(b)
Figure 2.9 Simplified model of sample containing partial discharges [32, 33].
(a) Physical model; (b) Equivalent capacitance network; (c) Sequence
of PDs for inception well below alternating voltage peak
4.0
3.2
2.4
1.6
0.8
0 90 180 270 360
Degrees
nC
Figure 2.10 Typical PD pattern from a wet pressboard sample
Insulating materials utilized in power-system equipment 43
InFigure2.9themagnitudeof theterminal pulse, V
x
, dependsontherelative
values of thecapacitances C
c
andC
b
, whichareusually unknown, andthat of C
a
.
Theapparent chargetransfer (partial discharge) isnormallytheparameter measured
at theterminals. Therelationshipbetweenthecavity chargetransfer andthat at the
terminal maybedetermined.
ReferringtoFigure2.9bthechangeof chargeacrossthevoid
q
c
= V
c
{C
c
+C
b
C
a
/(C
a
+C
b
)}
V
c
{C
c
+C
b
} as C
a
C
b
(volumeof C
b
is small)
V
c
C
c
, as C
c
C
b
(area of C
b
is small and is length of sample)
Theterminal voltagepulseduetothedischargeinthecavityis
V
x
= V
c
C
b
/(C
a
+C
b
)(capacitivedivision)
Theapparent dischargeq
x
detectedat theterminalsis
q
x
= V
x
C
x
= V
x
{C
a
+C
b
C
c
/(C
b
+C
c
)}
V
x
(C
a
+C
b
) as C
c
C
b
V
c
C
b
Thus q
x
q
c
C
b
/C
c
(2.5)
whichisthequantitynormallymeasuredattheterminals. Itwill benotedthatitwould
beexpectedtobemuchlessthanthatdischargedinthevoidas, ingeneral, C
c
C
b
.
However, for agiven systemand approximately known location thevalueof q
x
isagoodmeasureof potential damage. It isoftendescribedastheapparent partial
discharge.
Thepartial dischargesthatcanoccurwithintheambientnearinsulationsurfaces
asinregionsA of Figure2.8(iv) areoftenof larger valuesthandisturbanceswithin
thebulkmaterial. Inair-insulatedsystemsitmaybepossibletocalculatetheorder of
magnitudeof theinceptionstresscontainingawedgeadjacenttothesurface[37]. For
asharp-edgedelectrodeonasurfaceinoil or air (Figure2.8(v, 1) and(v, 3)), Kind
[36] quotestheempirical relationship
V
i
= K(s/
r
)
0.45
(2.6)
whereV
i
is in kV, s is thespacing in cm, and thevalueof K is approximately as
follows:
For corona dischargeinception K = 8 for metal edgein air
K = 12 for graphiteedgein air
K = 30 for metal or graphiteedgein oil
For brush dischargeinception K = 80 for metal or graphiteedgein air or oil
44 Condition Assessment of High-Voltage Insulation
2.3.2.3 Other characteristics
Becauseof theunknowndifferences intherelativechargevalues andthephysical
changesoccurring, it isconsideredby many investigatorsthat energy valueshavea
greater significance. It caneasily beshown(for example, inReference34) that for
theidealizedcapacitivenetworkandasinglepulsetheenergydissipatedinthecavity
is approximately equal to themeasured terminal valueof 0.7 q
x
V
i
, whereV
i
is
theRMSvalueof theterminal inceptionvoltage. Inorder toobtainabetter estimate
of theenergy being dissipated, it is necessary to makeacumulativemeasurement
of q
x
v
i
, where q
x
is the terminal partial discharge value and v
i
is the instan-
taneous value of the applied high voltage. Such measurements are possible using
specially designed instruments or a general computer-based measurement system
withappropriatesoftwareanalysisprograms.
Thesenewer techniques allowmoreaccuratedetermination of such significant
PDcharacteristicsastimeof occurrencewithinthepower-frequencycycle, repetition
rate, polarity, cumulativeenergy dissipation and various statistical analyses of the
data. Themeasurementsareanextensionof manyearlier studiesusinglesssophisti-
catedequipment aimedat recognizingpartial dischargesunder particular conditions.
Moderndigitizedsystems havealso enabledthedetectionandanalysis of theindi-
vidual pulsewaveshapes, whichmay haverisetimesof afewnsandadurationof
tensof ns. Insomecaseswiththepresenceof aninsulationsurfacethesetimescanbe
of theorder of amicrosecond. Researchinsuchareasisnowbeingpursuedinmany
laboratoriesworldwide. Referenceshouldbemadetopublishedpapers, for example
thepublicationsof theDEIS of theIEEE andthoseof theInternational Conference
onHighVoltageEngineering(ISH) conferences.
In thetraditional measurement systems (seeChapter 6), which arestill widely
appliedinindustry, andassumingtheinceptionvaluesrecordedarenear thepeak of
thehigh-voltagewave, energy levelstendtobecomparable. Thisisespecially soif
theresultsarefor testsonsimilar plant.
2.3.2.4 Damage due to partial discharges
It isapparent that certaincombinationsof stressandcontaminationarerequiredfor
theonset of partial dischargesunder operatingconditions. Duringovervoltagetests
PDsmaybeinitiatedbyincorrectdesignand/ormanufacturingerrors. Anydamaging
effectswill dependonthematerialsinvolvedand, inmanycases, whether or not the
by-products, especiallygases, arecontainedlocallyinthestructure.
Insolidmaterialssuchasresinsandpolyethyleneaformof internal trackingor
treeingoccurs, thestressatthetipsof thetreestendingtodeterminetheprogressof the
branches. Thiseffect iswell documentedwithdangerouslevelsbeingaslowastens
of pCsinsomecases. Withmicaceous-basedresinsystemsinternal PDsareprobably
associatedwiththeorganicmaterial buttreeingisrestrictedbythemicaflakes, which
areabletowithstandthousandsof pCsfor manyyearsat operatingstresses.
Inlaminatedmaterials resin- or oil-impregnated containinggas-filledvoids
trackingtends to developalongthesheet, or tape, material andmay exist for long
Insulating materials utilized in power-system equipment 45
periods of time if the direction of the primary component of stress is across the
laminationsof thebuilt-upstructure(Figure2.8(ii)). Experienceindicatesthaterosion
candevelopwithout failurewithhundredsof pCspresent whenthestressisnormal
to the laminations or barriers. However, failure would be expected as the effect
of thetrappedgases increases andhigh values of several thousandpCs developat
operatingstresses. Numerous132kV synthetic-resin-bondedpaper(SRBP) bushings
havesurvivedformanyyearsundersuchconditions. Insomeoil-impregnatedsystems
moderate values of PDs of, perhaps, 10002000pC can be withstood for many
months.
A situationof practical importanceis wheredischarges areinitiatedonacreep
surfaceduetothestressthroughthematerial asindicatedinFigure2.8(iv) and(v). In
suchcasesflashover maytakeplacealongthesurfaceasatrackdevelops. PDvalues
canvary fromtens of pCs to thousands of pCs onoil-immersedpressboardandon
pollutedair surfacesinoutdoor substations.
Withvery intensefields as inFigure2.8(v, 2) tens of thousands of pCs canbe
withstood for long periods at an oil-impregnated pressboard surface with average
stresses as high as 78kV/mm. Theconcentration of energy produces pitting and
degradation of thefibres [38]. Theenergy required to produceagiven volumeof
gas fromoil-impregnated pressboard was estimated in the experiments by Fallou
et al. [39].
TheconditionshowninFigure2.8(vi) isunlikely toexist under oil unlesssome
extraneousmetal object, frayedlead, perhapsagrosserror indesignor amanufactur-
ingfaultispresent. PDlevelsof 100pCor somaybedetectedatnear inceptionrising
to several thousands at higher stresses, depending on theconfiguration. If remote
fromground, suchdisturbancescancontinuewithout failure, becausetheoil isself-
sealing. Thisisalsothecaseforairwheretheinitial valuesmaybeaslowas1020pC
but canbemanythousandsof pCsat higher stresses asinair-insulatedequipment
of thepower system. ThedifferenceinPD polarity betweenoil andair at inception
isnotableandsometimesuseful ininterpretation. Air coronainanenclosedspace
suchasahigh-voltagemachine canhaveadeteriorativeeffect duetothecreation
of ozone.
Althoughanunusual sourceof partial discharging, theconsequenceof badcon-
nections can be very significant. Figure 2.8 (vii) indicates the effect of a broken
connectionwheredischargingoccurs across thebreak dueto capacitancecoupling
tothehigh-voltageelectrode. Themagnitudeof thePDswill bepartlydependent on
suchcouplingandmay beof theorder of thousands of pCs under adversephysical
conditions.
Muchof theearlier workrelatedtoidentificationof PDsandassociatedpossible
damagewas developed following theintroduction of commercial dischargedetec-
tors inthe1950s. Many of theresults werecollatedandpresentedinthereport by
CIGRE [40].
The manner in which the discharge sources might be identified using modern
measurement techniques is anareaof activeresearch, someof whichis detailedin
CIGRE papersandinparticular abrochurepreparedbyWGD1.11[41].
46 Condition Assessment of High-Voltage Insulation
2.3.3 Ageing effects
The various long-termchanges that occur in service are often described generi-
callyasageing. It isreasonabletoconsider ageingasacompositeprocessincluding
voltagestressinadditiontothedegradationcausedbythermal stressesonly. A pos-
siblemethodologyaimedat determiningtheconditionof materialsnaturallyagedin
serviceis described in Reference31. Theresults of such techniques may indicate
themost appropriatein-servicemonitoring procedures for theparticular operating
conditions.
Mechanical stresseffects, especiallyformachineinsulation, areimportantbutare
not includedindetail inthepresent discussion.
2.3.3.1 Electric stress ageing
Earlier breakdownvoltagev timetest results reportedby, for example, Montsinger
(Figure1.9) indicatedthemanner inwhichthestrengthof insulationdependsonthe
durationof theappliedvoltage. Presumably, suchdatacontributedto thechoiceof
impulseand50Hztestlevelsasrelatedtotheoperatingvoltages relativelevelsthat
arestill applicable.
Results obtained by a number of investigators led to the establishment of the
life/voltage(stress) relationship, especiallyapplicabletodry-typeinsulation.
Life(L) = K/V
n
(2.7)
whereK isaconstantforthematerial, nistheinverseslopeof thelife-lineonloglog
paper andV theappliedvoltage.
Inaparticular investigation, thevalueof nwasfoundtobebetween9and20for
older stator barsbut for other insulationsit maybemuchsmaller. Theinterpretation
of thedatafromexperimental testsisnotsimple, particularlyforplasticcablesamples
wherethescatter of results canbelarge. Insuchcases Weibell statistics [42] have
beenappliedfor estimatingtheprobability of failureafter agiventime. Allowance
canbemadefor thelengthanddiameter of thepractical cablesystemassumingcare
isexercisedintheuseof thestatisticsasdiscussedinReference43. Evaluatingthe
resultsof voltageendurancetestsoninsulatingmaterialsisdiscussedinIEC 61251
[S2/4] together withexplanationsof thefactorsinvolvedinsuchtests. Belowcertain
values degradation due to electric stress would not be expected for good-quality
materials withinnormal equipment lives. If this is acorrect assumptionit must be
consideredinthestatistical analysesof test results[43].
2.3.3.2 Thermal stress ageing
Thedegradationdueto hightemperatures onthelifeof insulationwas appreciated
intheveryearlydaysof electrical engineering. Thisledtoproposalsfor classifying
materials according to their expected performanceat particular temperatures, irre-
spectiveof electricstressing. InIECstandard60085[S2/1] referenceismadetoIEC
60216-6[S2/2], inwhichproceduresfor thedeterminationof theRelativeThermal
EnduranceIndex(RTE) of amaterial aregiven. Thematerial isthenclassifiedaccord-
ingtothevalueof theRTEinterval anditsthermal class. ForexampleanRTEinterval
Insulating materials utilized in power-system equipment 47
100000
10000
M
e
a
n

l
i
f
e

(
h
o
u
r
s
)
1000
100
10
60 100 150 200
Temperature (C)
10 years
1 year
0.1 year
Impregnated paper
Epoxy glass matt
300
Figure 2.11 Examples of the form of thermal ageing lines
of 105120

isallocatedathermal classof 105

, equivalent tothepreviousdesigna-
tionof ClassA. AnRTE of 130155

correspondsto130

andClassB. Thestandard
tabulationcovers thetemperaturerangeof 70

to250

andincludes abibliography
of relevant IEC andISOstandards.
Theearlier experimental testsindicatedthat material life=A exp(-mt) whereA
andmareconstantscharacteristicof thematerial andt isthetemperatureindegrees
Celsius. Thisworkformedthebasisforthewell-knownrule of an810

risehalving
theexpectedlifeof aninsulatingmaterial.
Theseempirical resultswerelater showntobeexplainablebyapplicationof the
Arrheniusdescriptionof arateof chemical reaction
K = C exp (E/RT) (2.8)
whereCisaconstant, Etheactivationenergy, RisthegasconstantandT theabsolute
temperature. ThetheorywasdevelopedbyDakin(1948) andgavetheequation
Life= A exp(B/T) or1n (Life) = 1n A +B/T (2.9)
whereA andB areconstants determinedby theactivationenergy of theparticular
reaction and T is theabsolutetemperature. A brief reviewof thedevelopments is
givenbyBrancato[44].
Strictly, therelationshipappliesonlytoconditionswhereonefirst-orderchemical
changedetermines theageingprocess. Plots of 1n(Life) v 1/T shouldbelinear for
this condition. A hypothetical exampleis giveninFigure2.11. Results for estima-
tionof lifeareoftenpresentedinthisform. Statistical analysesfor interpretationof
measurementsareincludedinIEC Document 60216[S2/2].
48 Condition Assessment of High-Voltage Insulation
2.4 Electrical breakdown and operating stresses
AnindicationhasalreadybeengiveninFigures2.12.3of thebreakdownstressesin
gasesandtransformer oil for simpleelectrodeconfigurations. It isdifficult todefine
or specify theelectrical breakdown strength of individual materials, although this
quantityisclearlyof major interest for designpurposes. Thedifferencesinstrengths
betweenmaterials arelargewiththevalues for similar samples of aparticular type
of dielectric varyingfor aspecifiedtest condition. Thisvariationrequirestheintro-
duction of probability concepts when choosing test and operational safety factors.
Figure1.9 is an exampleof how themean breakdown strength of somematerials
might beexpectedtoreducewithincreaseof thedurationof theappliedvoltage.
In Table 2.1 are listed the order of magnitude of the 50Hz operating stresses
for someof thematerials utilizedinthevarious items of power-systemequipment.
Precisevalues cannot bequoted, as theseareusually confidential, oftendependon
thelocal insulationconfigurationandvaryaccordingtothepracticeof theparticular
manufacturer. A number of typical samplebreakdownstressvaluesarealsoincluded
inthetable. Themajority of suchresults areobtainedfromshort-termtests onthin
materialsusingthemethodsspecifiedinaseriesof standards for exampleIECand
ASTM. Usuallythesetestconfigurationsarenotrepresentativeof theoverall working
conditions, althoughaparticularparametermaybesimulated. Thetestscomprisepart
of thequality-control processof thematerialsasmanufacturedby thesupplier. The
samplebreakdownresultsarequotedinthespecificationfor theparticular material
(or liquid) but must not beusedassuchinthedesignof full-scaleequipment.
Theelectrical stress magnitudes at failurefor new materials areinfluenced by
manyparameters, includingthethicknessof thematerial, thetypeof electrodesystem
andtheformanddurationof theappliedvoltage. Therelationshipbetweenthebreak-
downstressinasimpletestsampleandthatinafull-scalestructureof similarmaterials
isdifficulttodetermine. Also, theratioof thepower-frequencyshort-termequipment-
test valueandthemuchlower operatingstressnecessarytogiveanacceptableprob-
abilityof lifeisnoteasilyestimated. Thevariousequipmentmanufacturersandusers
haveestablishedratiosthroughtherelevantnational andinternational standardscom-
mittees. Ratioshavealsobeenselectedtoincludeassociatedlightningandswitching
impulsetest levels. Somedetailsof test proceduresarediscussedinChapter 7.
Itshouldbenotedthattheoperatingstressesareoften50per cent, or evenlower,
of thevaluesappliedduringtheshort-termpower-frequencytestsandmuchlessthan
thoseintheimpulsetestswhenthesearerequired. Theestimatedbreakdownstresses
for apractical configuration installed as part of power-systemequipment must be
significantlyhigher thanthetest levels. Thisisnecessaryinorder tomaintainasafe
margin during high-voltage routine and type tests. The value chosen is based on
experimental tests withstatistical analyses of theresults if sufficient informationis
available, possiblysupportedbytheuseof advancedconditionmonitoringtechniques.
Thechoiceof safetyfactorsisinfluencedalsobytheexperienceandknow-howof the
manufacturer anduser, especially inrespect of thelong-termoperatingstresses. An
important factor istheneedtoappreciatethat breakdowntestingof full-scalepartial
structures may not necessarily allowfor problems associatedwiththemanufacture
andassembly of thefinal equipment. For items manufacturedin largequantities a
lower safetymarginmaybeacceptable.
Insulating materials utilized in power-system equipment 49
Table 2.1 Examples of operating stresses and breakdown values of associated
samples of insulating materials used in power-system equipment
Equipment Typeof Rangeof Selectedsample50Hz short-term
insulation 50Hz breakdownstresses(average)
operating
stresses kV(RMS)/mm Reference
kV(RMS)/mm
Rotating Resin-micatape 34.5
machines
Power OI paper 33.5 75(ramp), 300m [45], [46]
transformers 37(40min), 1mm
OI pressboard 34.5 40 IEC 60243[47]
Oil 3(5mm) 12(5mm), Figure2.3[13]
<1(large) 3.9(15cm)
Oil-solidcreep 0.42.0 78(6mm) [46]
surface
Inter-phaseair For ex.300kV IEC 60076-3
Class
0.133(2.250m) 50cm. SG@300kV IEC 60052
2.35(180mm)
Rod-plane@ Figure2.1
530kV
0.265(2.00m)
Bushings OI paper 34.5
SRBP 2
Resin- 3.5
impregnated
paper
Cast resin 3
Porcelain
radial 416(3mm)
surface(oil) 0.5 ASTM D149[1]
surface(air) 0.2
Current OI paper 3.54 Asabove
transformers Cast resin 3
Capacitors Oilpaper 1025
Filmtype 50
Cables OI paper
(pressure) 1215
joints 7
XLPE 1015 48(XLPE) [48]
(newtypes)
48
EPR 3.5 2032(EPR) ASTMD149[1]
Polypropylene
paper-fluid(PPL) 10
Continued
50 Condition Assessment of High-Voltage Insulation
Table 2.1 Continued
Equipment Typeof Rangeof Selectedsample50Hz short-term
insulation 50Hz breakdownstresses(average)
operating
stresses kV(RMS)/mm Reference
kV(RMS)/mm
Insulators Porcelain(creep) 20mm/kV with
mediumpollution
Composite 0.5(near
(creep) termination)
OHlinesand Air small gaps 2.1(10mmUF gap) Figure2.2
air clearances Air longgaps 1 1.06(1.50m UF IEC 60052
gap)
Air - non-uniform 0.28(1.50mgap) Figure2.2
2.5 Development of insulation applications
In conclusion of this chapter it is interesting to note that there appears to be an
upsurgeinactivityrelatedtotheapplicationof insulatingmaterialsinpower-system
equipment, especially in respect of themost efficient useof present plant and the
incorporationof moderninsulationsinnewequipment.
This ranges fromattempting to estimatetheremnant lifeof existing materials
throughtoapplicationsof thenewer materials, e.g. XLPE high-voltagecables(upto
500kVRMSor higher), SF
6
for insulationof highlyratedtransformers, low-lossand
lesshazardousliquids, metallizedfilmsfor power capacitorsandmaterialsfor useat
superconductivitytemperatures.
As discussed in later chapters, many techniques are available or under devel-
opment for monitoringthecritical insulatingmaterialsasinstalledinpower-system
equipment.
2.6 Summary
Thereviewof insulatingmaterialsincludestraditional andnewformsasappliedin
power-systemequipment. Theelectrical and physical properties of significancein
characterizingandassessingtheconditionof thematerials areintroduced. A num-
ber of thedeteriorativeandfailuremechanisms associatedwithpractical insulating
materials aredescribed, including an indication of themagnitudeof theelectrical
breakdown stresses of samples/prototypes compared with the actual operating
stresses. An understanding of thesevarious factors and theexpected behaviour of
Insulating materials utilized in power-system equipment 51
thematerialsenablesthemostappropriatetechniquesforinsulation-conditionassess-
ment tobeselectedandassistsininterpretationof thecomplexdatarecordedbythe
monitoringsystems. Thelatter areconsideredinlater chapters.
2.7 References
1. Shugg, W. Tillar, Handbook of Electrical and Electronic Insulating Materials,
2ndedn(VanNostrandReinhold, NY, 1995)
2. Engineering Dielectrics, Vol. I Corona, Vol. IIA Solids, Vol. III Liquids ASTM
3. Bradwell, A. (ed.), Electrical Insulation (Peter PeregrinusLtd, London, 1983)
4. Ryan, H.M. (ed.), High Voltage Engineering and Testing, 2ndedn(IET, London,
2001)
5. EPRI AC Transmission Line Reference Book - 200kV and above, 3rdedn(EPRI,
PaloAlto, 2005)
6. Holmberg, M.E., and Gubanski, S.M., Motion of metallic particles in GIS,
IEEE Electrical Insulation Magazine, J uly/August 1998: 514 (also Septem-
ber/October 2000)
7. Dakin, T.W., Luxa, G., Oppermann, G., Vigreux, J ., Wind, G., andWinkelnkem-
per, H., Breakdown of gases in uniformfields: Paschen Curves for nitrogen,
air andsulphur hexafluoride, Electra, J anuary 1974;32:6182(seeElectra 52,
for hydrogenvaluesandHeylen, A.E.D., IEEE Electrical Insulation Magazine,
May/J une2006;22(3):2535for formulae)
8. Kuffel, E., Zaengl, W.S. andKuffel, J ., High Voltage Engineering Fundamentals,
2ndedn(Butterworth-Heinemann, UK, 2000)
9. Latham, R.V., High-Voltage Vacuum Insulation: The Physical Basis (Academic
Press, London, 1981)
10. Wilson, A.C.M., Insulating Liquids: their uses, manufacture and properties
(Peter Peregrinus, Stevenage, UK andNY, 1980)
11. Rouse, T.O., Mineral insulatingoil intransformers IEEE Electrical Insulation
Magazine, May/J une1998: 616
12. J ames, R.E. Behaviour of oil immersedsurfaces whensubjectedto tangential
electrical stressat highvoltage, PhDthesis, Universityof London, 1974
13. Samat, J ., and Lacaze, D. Micro-particles in transformer oil and dielectric
withstandeffects, Alsthom Review 1988;(11):4757
14. CIGRE Brochure157, Effect of particles on transformer dielectric strength,
Electra, J une2000;(190):13540(WG12/17)
15. Beroual, A. et al., Liquid Dielectrics CommitteeInternational Study Group,
IEEE Electrical Insulation Magazine, March/April 1998: 617
16. Dielectric Liquids, special issue, IEEE Transactions on Dielectrics and
Electrical Insulation, February2002, vol. 9
17. Oommen, T.V., andPetrie, E.M., Electrostaticchargingtendencyof transformer
oils, IEEETransactions onPower Apparatus andSystems, 1984;103:19231931
52 Condition Assessment of High-Voltage Insulation
18. Peyraque, L., Beroual, A. andBuret, F., Static electrificationof pressboard/oil
interface and transient phenomena, IEEE Transactions on Dielectrics and
Electrical Insulation, J une1998;5: 4439
19. CIGRE Brochure 170, Static Electrification in Large Power Transformers,
Electra, J une2000;(191):1014(J WG12/15.13, TF 01& 02)
20. Darveniza, M., Electrical Properties of Wood and Line Design (University of
QueenslandPress, Brisbane, 1980)
21. Hepburn, D.M., Kemp, I.J . andShields, A.J ., Mica IEEE Electrical Insulation
Magazine, September/October 2000;16(5):1924
22. Allen, R.D., Improving the high-temperature performance of EPDM in
Cheremisinoff, N.P., andDekker, Marcel (eds.), Handbook of Polymer Science
Technology (Marcel Dekker, NewYorkandBasel, 1989), pp. 12741
23. Roby, D., Liquid silicone technology for outdoor insulation, presented at
Electrical EngineeringCongress, Sydney, Australia, 1994, pp. 2430
24. Cherney, E.A., RTVSilicone Ahightechsolutionfordirtyinsulatorproblems,
IEEE Electrical Insulation Magazine, November/December 1995;11:814
25. Mackevich, J .P., and Hoffman, J .W., Insulation enhancement with heat-
shrinkable components Part III, IEEE Electrical Insulation Magazine,
J uly/August 1991;7(4)
26. Brazier, L.G. The breakdown of cables by thermal instability, JIEE,
1935;77:10415
27. Black, R.M., andReynolds, E.H., Ionizationandirradiationeffectsondielectric
materials, Proc. Inst. Electr. Eng;112:1231J une1965
28. CIGRE WG12.09, A surveyof factsandopinionsonthemaximumsafeoper-
atingtemperaturesof power transformersunder emergencyconditions, Electra,
1990;(129)
29. J ames, R.E., Liu, Z., Phung, B.T. and Blackburn, T.R., Effect of moisture
and temperature on the PD behaviour of an oil-impregnated insulation sys-
tem, Proc. AUPEC97, Universityof NSW, Australia, September/October1997,
pp. 60714
30. Wang, W., Vandermaar, A.J ., andSrivastava, K.D., Reviewof conditionassess-
ment of power transformers inservice, IEEE Electrical Insulation Magazine,
November/December 2002;18(6): 1225
31. Allan, D., andBlackburn, T.R., Serviceagedinsulation Guidelinesonmanag-
ingtheageingprocess, CIGRE (WGD1.11) Brochure228, Electra, J une2003;
(208)
32. Whitehead, S., Dielectric Breakdown of Solids (OxfordUniversityPress, 1953)
33. Austen, A.E.W., andWhitehead, S., Dischargesininsulationunder alternating
currentstresses, Proc. Inst. Electr. Eng., PartII, Vol. 88, p. 88, 1941(seeArman,
A.N., andStarr, A.T., Themeasurementof dischargesindielectrics, Journal of
Institution of Electrical Engineers, 1936;79:6781)
34. Kreuger, F.H., Partial Discharge Detection in High-Voltage Equipment (Butter-
worth& Co., London, 1989)
Insulating materials utilized in power-system equipment 53
35. J ames, R.E., andPhung, B.T., Development of computer-basedmeasurements
andtheir applicationtoPDpatternanalysis, IEEE Transactions on Dielectrics
and Electrical Insulation, October 1995;2(5) 83856
36. Kind, D., An Introduction to High-Voltage Experimental Techniques (Vieweg
andSon, Braunschweig, 1978)
37. Mason, J .H., Discharges, IEEE Transactions on Electrical Insulation, August
1978;EI13:21138
38. J ames, R.E., and White, P.A., Relationships between fast and slow risetime
partial discharges, Proceedings of the 4th International Symposium on High
Voltage Engineering, Athens, September, 1983, Paper 22.05
39. Viale, F., Poittevin, J ., Buccianti, R. et al., Studyof acorrelationbetweenenergy
of partial dischargesanddegradationof paper/oil insulation, Paper1512, 1982,
sessionof CIGRE
40. Recognitionof Discharges, Electra, December 1969;(11), p. 61
41. Knowledgerulesforpartial dischargediagnosisinservice, CIGRE(WGD1.11),
Brochure226, Electra, 2003; (207)
42. Hauschild, W., and Mosch, W., Statistical Techniques for High-Voltage Engi-
neering, IET Power Series, no. 13(IET, London, 1992)
43. Simmons, M., High Voltage Cables in Ryan, H.M. (ed.), High Voltage
Engineering and Testing (IET, London, 2001)
44. Brancato, E.L., Insulating aging A historical and critical review, IEEE
Transactions on Electrical Insulation, August 1978;EI-13:30817
45. Fofana, I., Wasserberg, V., Borsi, H., and Gockenbach, E., Challenge of
mixedinsulatingliquidsfor useinhigh-voltagetransformers, Part 2: Investiga-
tionsof mixedliquidImpregnatedpaper insulation, IEEE Electrical Insulation
Magazine, J uly/August 2002;18(4):515
46. J ames, R.E., Dischargedetection in high-voltagepower transformers, Proc.
Inst. Electr. Eng., J uly1970;117(7):135262
47. Moser, H.P., and Dahinden, V., Transformerboard II, (H. Weidmann AG,
Rapperswil, 1987)
48. Vahedy, Vahdat, Polymer insulated high voltage cables, IEEE Electrical
Insulation Magazine, May/J une2006;22(3): 1318
2.8 Standards related to insulating materials
S2/1IEC 60085(Ed. 3.0, 2004): Electrical insulation Thermal classification
S2/2 IEC 60216: Electrical insulating materials Properties of thermal endurance
(thermal enduranceproperties)
Part 1(Ed. 5.0, 2001): Ageingproceduresandevaluationof test results
Part2(Ed. 4.0, 2005): Determinationof thermal endurancepropertiesof electrical
insulatingmaterials Choiceof test criteria
Part 3 (Ed. 1.0, 2002): Instructions for calculating thermal endurance
characteristics
54 Condition Assessment of High-Voltage Insulation
Part 4-1(Ed. 3.0, 1990): Part 4-2(Ed. 1.0, 2000) andPart 4-3(Ed. 1.0, 2000):
Ageingovens
Part 5(Ed. 2.0, 2003): Determinationof relativethermal enduranceindex(RTE)
of aninsulatingmaterial
Part6(Ed. 1.0, 2003): Determinationof thermal enduranceindices(TI andRTE)
of aninsulatingmaterial usingthefixedtimeframemethod
S2/3IEC60505(Ed. 3.0, 2004): Evaluationandqualificationof electrical insulation
systems
S2/4 IEC/TS 61251 (Ed. 1.0, 1993): Electrical insulating materials AC voltage
enduranceevaluation Introduction
S2/5IEC60156(Ed. 2.0, 1995): Insulatingliquids Determinationof thebreakdown
voltageat power frequency test method
S2/6IEC 60296(Ed. 3.0, 2003): Fluids for electrotechnical applications unused
mineral insulatingoilsfor transformersandswitchgear
S2/7 IEC 60247 (Ed. 3.0, 2004): Insulating liquids measurement of relative
permittivity, dielectricdissipationfactor (tan) andDC resistivity
S2/8IEC/TR61249(Ed. 1.0, 1993): Insulatingliquids Determinationof thepartial
dischargeinceptionvoltage(PDIV) Test procedure
S2/9 IEC 60270 (Ed. 3.0, 2000): High-voltagetest techniques partial discharge
measurements
S2/10IEC 60599(Ed. 2.0, 1999): Mineral oil-impregnatedelectrical equipment in
service guidetotheinterpretationof dissolvedandfreegasesanalysis
2.9 Problems
1. How can moisture and/or air (gas) affect the electrical properties of extruded
XLPE, oil-impregnated paper and resin mouldings as applied in high-voltage
equipment? Outlinetheprincipal methods that might beusedto determinethe
presenceof suchcontaminants.
2. DeterminetheAC breakdownstressesinasimpleuniformfieldinsulationsys-
temwithanoverall spacingof 12mmandaninsulationresincoveringof 2mm
thicknessononeelectrodewhenimmersedin(a) air, (b) SF
6
(bothatatmospheric
pressure) and(c) drytransformer oil. Calculatestressineachambient medium
andtheresin.
resin
= 3.5,
oil
= 2.2
3. An air-filled cavity of 3mmdiameter and thickness of 1mmis suspected of
existingwithin thelaminations of an insulatingboardof 40mmthickness and
largearea. Estimatethemagnitudeof partial discharges that might exist inthe
cavity onapplicationof ahighvoltage, V, across theboard. What valueof PD
shouldbedetectableattheelectrodes?Determinethevalueof theappliedvoltage
V requiredtoinitiatedischargesinthecavity. = 4.5
Chapter 3
I ntr oduction to electr ical insulation design
concepts
Basicrequirementsfor insulationdesign
Electricstressdistributionininsulationsystems
Electricstresscontrol
Duringtheprocessof choosinganappropriateinsulationmonitoringorassessment
systemit isof valuetobeawareof thedesign, manufacturing, electrical testingand
operatingrequirements of thepower equipment to bemonitored. Inthis chapter an
indicationis givenof someof themoreimportant generic factors to beconsidered
whenselectinginsulatingmaterialsfor practical designs. Theapplicationsof suitable
materialsfor arangeof plant itemsaredescribedinChapters4and5. Thematerials
arereviewedinChapter 2.
Theinterpretationof recordeddatageneratedby potential faultsandchangesin
the characteristics of the insulating materials may be aided by analysis of simple
electric fieldconfigurationsrepresentingtheoriginal system. For complex arrange-
mentsit isoftenpossibletosimplify anareaof interest toassist withapreliminary
analysis. Several simplified cases are described in this chapter. Detailed analyses
involvecomputerizedfielddeterminationsbasedonaccurateknowledgeof thepar-
ticular design. A special featureof electric fielddesign is theuseof stress-control
techniquesanumberof whicharelistedtogetherwithreferencetoexamplesincluded
inother chapters. Theincorrect applicationof thesemethodscanresult ininsulation
problemsdetectablebyappropriatemonitoringsystems.
3.1 Over view of insulation design r equir ements
Thefollowingpresentationaimsat givinganoverviewof thecomplexitiesinvolved
inproducingasatisfactory insulationstructurecapableof operatingreliably for 25
to 40 years or more. Such lifetimes can beachieved only if efficient maintenance
programmes, includinginsulationmonitoring, arerigorouslyimplemented.
56 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3.1.1 El ectr i cal requi rements
Theelectrical test voltage/operating voltageratio varies according to thesystems
basic insulationlevel, typeof equipment anditslocation. Thevalueisspecifiedby
theuser, guidedbythedatainIEC60071-1(Chapter 1). Dependingontheparticular
circumstances, partial discharge, radiointerferencevoltageanddielectriclosslevels
must bebelowspecifiedvalues as manufactured. To ensuretheserequirements are
met, appropriatetestsareappliedasincludedonthediagram. Special arrangements
mayberequiredfor in-servicemonitoringof, for example, PDsandDDF.
Electrical requirements
Tests
DDF IR PD
and
RIV
AC
High or
low
frequency
DC Impulse
and
switching
surge
Surges-
lightning
and
switching
AC
or
DC
Corona
RI
PD Dielectric
losses
Operating conditions
3.1.2 Physi cal l i mi tati ons
Awarenessof thephysical limitationsimposedbynon-insulationfactorsisimportant.
Therearemanyoccasionswhensuchrestrictionsexistandonlyminor modifications
are possible to improve the overall dielectric performance. Some of the physical
changestotheinitial overall designproposalsthat might beconsideredareindicated
below.
Can critical
clearances be
increased?
Is a change in the
ambient medium
possible?
Are alternative
electrode support
systems allowable?
May the
electrode
shapes be
modified?
Physical layout proposals
3.1.3 Wor ki ng envi ronment
No matter howwell designedthestructureandhowsatisfactory thetest laboratory
results, thesemeasurements areof littlevalueunless attempts havebeen madeto
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 57
simulateand/or allowfor possibleinsulationdeteriorationduetopoor workingcon-
ditions. If theenvironment resultsincontrol equipment becomingcoveredwithdust
inadampatmosphere, or arotatingmachinecontaminatedwithgreaseanddirt, or
anoverheadlineinsulator pollutedwithsalt andindustrial fumes, thenthesefactors
must beconsideredat thedesignstage.
A major deteriorativeeffect is dueto thetrappingof air or thebuild-upof gas
leading to partial discharges and ultimate failure. This can occur in oil-insulated
transformers dueto poor impregnation, failureto top upcorrectly, or not follow-
ingspecifiedprocedures wheninstallingabushing. Theingress of moistureover a
longperiodcanincreasethedielectric losses inmany materials, againresultingin
breakdown. This applies to solidmaterials suchas epoxy resins andXLPE as well
as oil-impregnated paper, pressboard and wood. Acceptanceand monitoring mea-
surements havebeendevisedtocover thevarious conditions. Thesetechniques are
describedinlater chapters.
Access of
dirt or other
pollutant
Moisture
access
Unusual
temperature
range
Working environment
Air or gas
access
External
pressure.
Atmospheric
pressure
3.1.4 Mechani cal requi rements
As in much of electrical engineering, it is important to appreciate that allowance
for the mechanical requirements in the application of insulation is essential. This
is oftentheprimefactor inancillary lowvoltagesystems includingcoaxial cables,
wiring, plugs and sockets and somewindings. At high voltages, in such items as
rotatingmachines, suspensioninsulators, switchgear, transformers, compressed-gas
or vacuumequipment andcablesthemechanical performanceof theinsulationisof
major consequence.
Compression
and tension
e.g. transformers,
insulators
Workability Vibration
e.g machines
Mechanical requirements
Impact
effects
e.g. switchgear
Bending
e.g. cables
58 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3.1.5 Ther mal condi ti ons
In addition to knowledge of the thermal performance of the chosen insulations,
the design must allow for the imposed thermal conditions and any built-in cool-
ing systems that exist or might beadded. Thermal conditions in somecases limit
the application of large amounts of insulation and structures have to be devised
to allow the heat from an inner winding or conductor, and the losses in the
dielectric itself, to be dissipated satisfactorily in order to avoid overheating of
the insulating materials. An outstanding example of such a problem is a high-
voltagepower cableusingoil-impregnatedpaper asitsinsulationwherearunaway
condition is possible. Such conditions have also occurred in power and current
transformers in which the insulating materials were not correctly dried. The use
of internal cooling of conductors has eased thethermal problemconsiderably in
certain cases, as demonstratedin largegenerators andin anumber of highly rated
cables.
Imposed
Temperature
limits
Any thermal
cycling
Total losses
Thermal conditions
Losses
in dielectric
Losses
in conductors
(and core)
Heat
dissipation
paths
Method
of
cooling
Heat flow
3.1.6 Processi ng
The behaviour and expected life of the insulation structures depend critically on
theprocessingprocedures adoptedinthefactory. Theseincludelargevacuumdry-
ingovens through to small resin vacuum-pressure-impregnation (VPI) systems. In
all cases the requirements and possible limitations must be clearly defined. An
insulation changefromatraditional material to anewplastic or resin may appear
attractivetechnologically but proveuneconomic becauseof thecost of investment
in additional manufacturing plant. For example, such decisions werenecessary in
the cable industry when introducing plastic systems, in transformer manufacture
during development of cast-resin-type distribution units, in overhead line insula-
tor developments, replacing conventional assembly methods with VPI processes
for somelarger rotatingmachines andreplacement of oil withSF
6
inanumber of
plant items.
Theconsequenceof manufacturingchangesorapplicationof newtechniquesmust
becarefullyassessedfollowinganunexpectedinsulationproblem.
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 59
Processing limitations
Components
throught-put
rate
Availability of
expertise for
any major
changes
Overall size
of units
Integration
with overall
manufacturing
programme
Cost of
any
process
changes
3.1.7 Rel i abi l i ty
The above factors influence, directly or indirectly, the ultimate reliability of the
insulationstructuresand, therefore, of thehigh-voltageequipment, componentsand
associated systems. As reliabilities of possibly better than 99 per cent based on
averagetimesof outagesoverthewholesystem arebeingclaimedbysomeutilities,
and areexpected by customers, periodic maintenanceand monitoring of theinsu-
lationisbecomingof major importance. Inthediagram, stress includeselectrical,
thermal andmechanical conditions. Anattemptismadetoshowhowreliabilitymight
berelatedto condition(age), cost andthelevel of monitoringwhichmight appear
necessary.
Access
limitations
due to
safety
requirements
Materials
compatibilty
Main plant
and components
Factors affecting reliability
Existing
Probably
low stress
Aged Life
assessment
necessary
Effective monitoring
and maintenance
required
Intermediate
reliability
New
High internal
temperatures
near main plant
Poor
mechanical
environment
Electronic devices,
controls, metering
cabling, monitoring
systems and similar
Low stress High stress
Less
material
Lower cost
Much
material
Higher cost
Low level
monitoring
High level
monitoring
Good
reliability
Intermediate
to
good reliability
Surges
Ancillary
equipment
The interpretation of some of the above factors affecting reliability is very
simplisticandisintendedasaguideinrelationtotheinsulationaspects.
60 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3.2 Electr ic str ess distr ibutions in simple insulation systems
Therearemany difficulties inapplyingthedataavailableconcerningthedielectric
characteristics of insulation, particularly theresults fromsmall sampletests where
the concepts involved in volume, area and thickness effects must be considered.
Allowanceneedstobemadeforanyvariationswithtimeof electrodevoltages, ascan
occur whenatransformer windingissubjectedtoalightningimpulseor, sometimes,
wheredirect andalternatingvoltagesareappliedsimultaneously.
Becauseof themany conflicting factors it is often impossibleto configurethe
materialstoobtainanoptimuminsulatingsystem. Insuchcasesitisnecessarytopre-
pareacompromisearrangement withintheoverall equipment designandproduction
limitations, aswell asthecustomer specification. Considerableexpertiseisrequired
inthelayout of insulationtoenableaproposedelectrodesystemof high-voltageand
earthedcomponents to withstandthevarious test andoperatingconditions. To this
endknowledgeof theelectricfielddistributionisessential together withinformation
regardingtheapplicationof sampletestresultstothepractical configuration, electric
strengthdataand, perhapsmostimportantly, previousexperienceaccumulatedbythe
organization.
Designof insulationstructuresisoftenassistedby considerationof simplecon-
figurations, which can be helpful in simulating the region of interest. For some
arrangements it is difficult to interpret a field determined by an advanced three-
dimensional computer programunless prior knowledgeas to theweakest regionin
relationto thelocal stress is known. Thehighest stress may occur inaninherently
stronginsulationstructurebut elsewhereadangerous lower stress might exist ina
locationthat isbasicallyverydifficult toinsulate. Suchfeaturesaremoreobviousin
thewell-established, butnowrarelyused, analoguetechniquessetupfor theplotting
of capacitivefieldsinconductivemedia, thecommonest twobeingaresistivepaper
sheet or anelectrolytic tank [1]. Thechoiceof anappropriatemethodwill depend
ontheparticularcircumstancesincludingtheexperienceandknowledgeof personnel
available, complexityof theproblem, theaccessibilityof hardwareandsoftware, and
whether thesolutionsarefor development, designor urgent investigativepurposes.
Thereis nodoubt that inlargeorganizations computer programs areappliedexten-
sively. Duetocost, however, thesimpler techniques may still bepreferredinsome
cases, especiallyfor two-dimensional problems.
Programs for solvingsymmetrical 3-D problems by theFiniteElement Method
havebeeninuseformanyyears. MorerecentlytheBoundaryElementMethodseems
tohavebeendevelopedfortruly3-Dasymmetricconditions. Inthistechniqueamesh
isrequiredonlyattheelectrodeboundariesanddielectricinterfaces. Thisreducesthe
storageandmemory neededandusually gives stress values directly as well as the
equipotential distribution. Thelatter isoftenof valueintheinitial designof complex
insulationsystems. Largecomputers andcostly programs involvingmany hours of
runningtimearestill necessaryfor thesolutionof practical problemssuchastheGIS
componentsdiscussed, for example, inReference2.
Becauseasignificantnumberof insulationfailuresarerelatedtotheelectricstress
distributions magnitudesandspatial ratesof change considerationisnowgivento
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 61
anumber of commonfieldconditions. SuchanalysesandassociatedPD/breakdown
datacanbehelpful ininterpretingtheresults of monitoringandtestingof specific
designs, includingsomecomplexconfigurations. Stress-enhancementfactors maxi-
mumstressaveragestress havebeendeterminedforanumberof simpleelectrode
systems containingonedielectric, usually air or SF
6
. A summary of afewof these
factorsisgiveninReference3.
3.2.1 Mul ti pl e di el ectr i c systems
The simplest multiple dielectric configurations occur in uniform (parallel) and
concentricfields.
3.2.1.1 Par allel electr odes
InFigure3.1is showntheconditionwheredielectrics arelocatedbetweenparallel
electrodes. Byneglectingtheedgeeffectsthisarrangement canbeusedtodetermine
thestress distribution in aparallel field in which several dielectrics arelocated in
series, includingliquidandgaslayersasinlaminatedstructures.
100
Ratio

1

1

2
30
10
5
2
1
0.01 0.1
=
0.02
0.1
0.2
0.3
0.5
1.0
1.0
10
d
1
d
2
1 +
+
d
1
d
2
(1 + )
d
1
d
1
+ d
2
d
2
d
1
d
2
+
d
1
d
2

2
E
1
E
Thickness ratio
S
t
r
e
s
s

r
a
t
i
o
100
Resistivity
ratio

1
10
1
10
=
100
50
30
20
15
10
100 1000

1
d
1
d
2
E
2
E
Thickness ratio
S
t
r
e
s
s

r
a
t
i
o
V
O
V
d
d
2
d
d
1
= d d
2
V
d
E
1
=
AC conditions
DC conditions
E

=
V
d
V
E(mean) = =
E
2
E
=
Fi gure 3.1 Mul ti pl e di el ectr i cs i n a par al l el fi el d (var i ati on of stress for two
di el ectr i cs (n = 2, d
1
, d
2
))
62 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
If n dielectricsarerepresentedthenthestressinlayer i is
E
i
=
V

i
(
d
1

1
+
d
2

2
+. . . . +
d
n

n
)
(3.1)
andthecapacitance
C = 2
0
1
d
1

1
+
d
2

2
+. . . .. +
d
n

n
farads/metre
2
(3.2)
Fortheparticularcaseof twodielectrics(n = 2) forexamplegas(air)/porcelain,
gas (air)/pressboard, gas (air)/resin, oil/pressboard(or paper), paper/resin nondi-
mensional curves givingtheratioof thestress (E
1
) indielectric n
1
andtheaverage
stress(E) may becalculatedasinFigure3.1. Thesecover therangeof permittivity
ratiosof 0.02to0.5andresistivityratiosof 100to10forarangeof dielectricthickness
ratiosof d
1
/d
2
. Applicationsof thesesimplebutuseful conceptsmaybedemonstrated
quantitativelyasfollows.
(a) If theelectrodespacingd= d
1
+d
2
= 10mm, theair gapd
1
= 3mm(
1
= 1),
thebarrier thickness d
2
= 7mm(
2
= 3.5), theappliedvoltageV, thenfrom
(3.1) theair stressE
1
= 0.2V/mm.
FromFigure 2.2 the breakdown of a 3 mm-thick air gap in a uniformfield
is 2.59kV/mm, thus theconfiguration would beexpected to dischargewhen
E
1
= 2.59kV/mmor V = 13kV(RMS). Thecalculation is an indication of
howtheintroduction of abarrier of inappropriatethickness between HV and
earth or even between phases may producecorona(partial discharges) in
servicefor certain field conditions, thereby reducing thestrength of thesys-
tem. Failurebypuncturemayoccur duetoerosionof thesurfacebycoronaor,
possibly, aflashover aroundthebarrier eventhoughthestressacrossthebarrier
islow 0.74kV/mmat coronainceptioninthiscase.
Withabarrier of 4mmthicknessplacedcentrallybetweentheelectrodestheair
spaced
1
= 6mmisdividedintotwo3mmgapsandd
2
= 4mm. Theairstresses
computeas0.14V/mm, givingadischargeinceptionvoltageof 18.5kV (RMS)
assumingthestressesinthegapsare2.59kV/mm.
Althoughtheinceptionvoltageisincreasedbyreducingthebarrier thicknessit
is still belowthat of 21.9kV for agapof 10mmwithout abarrier. Thelatter
wouldbeunstablewithanear-uniformfieldcontainingparticles, moistureand
other pollutants as inapractical system for example, bus-bar arrangements
inmetal-cladswitchgear. Equipment test stresseswouldbechosentobemuch
lower thanthosededucedfromthePaschencurves(seeFigure2.2), butthedata
areof valueinchoosingtheoptimumconfigurationandinanalysingfailures.
(b) For theabovecondition, wheretheductor gapisfilledwithoil (
1
= 2.2, 3mm
thick), thebarrier ispressboard(
2
= 4.4, 7mmthick), (3.1) givesthevalueof
E
1
= 0.153V/mm. FromFigure2.3a3mm-thick oil gapwouldbeexpected
tobreakdownat approximately14kV/mm(RMS) correspondingtoanapplied
voltageof V = 91kV (RMS) withastress of only 7kV/mminthesolidand
anaveragestressof 9.1kV/mm(RMS). Thisaveragestresscanbeincreasedby
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 63
reducingthebarrier thicknessandforming, for example, two3mm-thickducts.
By a procedure as in (a) the voltage for inception becomes 112kV (RMS),
althoughthegapstill limitstheoverall strength. Thesestressvaluesmaynotbe
achievedinpracticeintransformerswheretheallowableinterwindingoil stress
duringshort-termtestingmightbeof theorder of 10kV/mm. Thisiswell below
theuniformfieldconditions representedinFigure2.3. If thestress-v-volume
graphinFigure2.3hadbeenappliedfor alargevolume, alower voltagewould
beexpectedto producepartial discharges. Suchconsiderations aretakeninto
account whendesigninglargeinsulationstructuresasinpower transformers.
It will benotedthat asd
1
0(or d
1
/d
2
0) thestressE
1
E
2
/
1
.
Similar calculations may becarriedout for DC conditions inoil. Insuchcases
thestress in thesolid material will behigher than in theoil. For example, with a
resistivity ratio of 100 (oil 10
11
m, impregnated paper/pressboard 10
13
m)
andd
1
/d
2
= 1/9thestress intheoil is very small (seeFigure3.1). If theoil gapis
relativelylargethesolidmaybeoverstressed. Duringtheapplicationof directvoltage
teststoACequipment thisconditionmust berecognizedwhenchoosinglevelstobe
applied. Withanair gap(highresistivity), partial discharges caneasily beinduced
but areusuallyof alowrepetitionrate. Thehissing fromhigh-voltageDC test sets
andchargingunitsistypical.
3.2.1.2 Concentr ic electr odes
Anothersimpleconfigurationthatcanoftenbeof assistanceininterpretingtestresults
isthat of concentriccylinders, Figure3.2(a).
Inthiscasethefielddistributionislogarithmic, themagnitudesdependingonthe
radial locationandthevalueof thepermittivity (for AC conditions) or conductivity
(for DC conditions). Thestressat radiusr
i
isgivenbythefollowing, where
x
isthe

2
HV rod. tube
(a) (b)
2
r
0
2
r
1
2
r
2
Outer screen
(earth)
100
AC
DC
5 10 20
Radius mm
0
S
t
r
e
s
s

%
+
Fi gure 3.2 Mul ti pl e di el ectr i cs i n a concentr i c fi el d. (a) Confi gur ati on for two
di el ectr i cs; (b) Var i ati on of stress across two di el ectr i cs
r
0
= 5mm, r
1
= 10mm, r
2
= 20mm.
1
= 3.5,
2
= 2.2

1
= 10
13
m,
2
= 10
11
m
64 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
permittivityof theparticular section:
E(r
i
) =
V

x
r
i
(
1

1
ln
r
1
r
0
+
1

2
ln
r
2
r
1
+. . . .. +
1

n
ln
r
n
r
n1
)
(3.3)
for acapacitivedistribution. Notethat thetotal capacitance
C = 2
0
1
1

1
ln
r
1
r
0
+
1

2
ln
r
2
r
1
+. . . .. +
1

n
ln
r
n
r
n1
farads/metre (3.4)
Substitutingconductivitiesfor permittivitiesgivestheequivalentDCconditions.
Thestressdistributionacrossatwo-layer systemfor particular valuesisindicatedin
Figure3.2(b).
Thesuddenchangeinelectric stress dueto different permittivities andconduc-
tivities at the insulation interface is apparent. The possibility of maximumstress
reduction is sometimes exploited in high-voltage oil-impregnated cables by using
higher-densitypapersat theinner conductor.
Thesignificant differencein theAC and DC stress distributions can introduce
problemswhentestingpowerfrequencyequipmentwithdirectvoltages inparticular
cablesealingendsandthroughjoints.
3.2.2 Edge effects
Edge effects at the interface or terminations of the electrodes and the insulating
materialscanintroduceproblemsrequiringspecial arrangementstoprevent PDsand
avoidbreakdown. Thedeterminationof theelectric stressdistributioninsuchcases
isessential inorder toensureoptimumperformanceontest andinservice.
A condition that can arisein bushings, somecurrent transformers, possibly in
capacitors, in incorrectly terminated cables and on the stator bars of high-voltage
machinesisindicatedinFigure3.3(a). Herethethinfoil or screenendisnot stress-
relieved. Althoughsomestressrelievingby inclusionof acurvedelectrodemay be
effectedasinFigure3.3(b), partial dischargesmightstill occurintheambientmedium
withinthewedgeattestandoperatingvoltages. Someindicationof thevoltagelevels
allowablecanbedeterminedbycalculationof thestressesintheambient insulation
of air/gasor insulatingoilswithinthewedgeformedat thesurface.
By assuming a capacitivedivision across distances d
1
and d
2
with a common
localizedarea, theinceptionvoltageV
i
V
1
{
1
d
2
/
2
d
1
+ 1} whereV
i
is the
inceptionvoltagetoearth. For arangeof valuesof d
1
theexpectedinceptionvoltages
(V
1
) maybeestimatedfromthePaschencurvesinFigure2.2for gasesandthedata
inFigure2.3for oil.
Thisapproachfor determiningtheinceptionvoltageinthegap, andthereforethat
of asystemcontainingsuchawedge, was describedby Mason[4] andis analysed
further byKreuger [5].
Thepositionof thePDsourceintheoil or air next totheelectrodesurfacemay
betransferredto asafer regionby additionof insulationontheelectrodeincon-
figurationsasshowninFigure3.3(b) seealsoFigure2.8(iv, b). Thisisaformof
stressreductionalongtheinsulationsurface. Theparticular caseof aspheretouching
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 65
Wedge
V
1
d
1
d
2
Insulation
Earth
80
90
70
60
(a)
(b)
50
40
30
20
10
Earth
100% HV conductor
Insulation
surface
Insulation
surface
Fi gure 3.3 Effects at edge of el ectrodes l ocated on an i nsul ati on sur face.
(a) Vol tage di str i buti on at edge of thi n ear thed el ectrode; (b) Model
for deter mi nati on of PD i ncepti on at a cur ved el ectrode
or beingembeddedinaninsulationfor arangeof permittivityratioswasanalysedby
Binns[6]. Thisincludesdeterminationof maximumstressesandtangential valuesat
theinterfacebetweenthetwoinsulatingmaterials. Equipotential plotsshowinghow
thefield along theinsulation surfaceadjacent to apartially embedded electrodeis
influencedbythepermittivityratiosof thesolid( = 4) andambientmedium( = 1)
aregiven as an examplein Figure3.4. Thereduction in stress for condition (b) is
significant. A number of practical applications, in particular for GIS, aregiven in
References2and3.
Thesupportingof insulatedhigh-voltageleadsinoil adjacent toaplaneearthed
surfaceoftenrequiresacreepsurfacebetweenthetwo, asinFigure3.5(a). Inorder
toreducethemaximumtangential stress(at thepaper surface) alongthesupporting
boardtoanacceptablevalueitmaybepossibletoincreaselocallythediameter of the
paperonthelead. If theoverall spacingislarge, anindicationof thediameterrequired
canbeestimatedby assumingaconcentric cylinder configurationeventhoughthe
earthelectrodeisaplane. ThisisacasewheretheinsulationisbeingusedtoeffectAC
stresscontrol inacritical regionaswell asprovidingincreasedpuncturestrength. For
moreaccurateestimates, afieldanalysiswouldberequired. Thevoltagedistribution
across thesystemis of theformin Figure3.5(b). As in aprevious example, for a
paper/oil configuration thedirect voltagedistribution is primarily across thepaper
andnot theoil. Thisisnot thecasefor agas/air system, inwhichtheconductivityof
66 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
HV
100%
20%
10% Division
(a)
(b)
Earth
10%

1
= 1

2
= 4
HV
100%
30%
10% Division Earth
20%
10%

1
= 4

2
= 1
Fi gure 3.4 Equi potenti al pl ots for par ti al l y embedded el ectrode. (a)
1
/
2
= 1/4;
(b)
1
/
2
= 4/1
theambient is very low. Theconcepts also apply if theambient is agas as inGIS,
dry-typetransformers, air-insulatedswitchgear andhigh-voltagemachines.
ThecurvesinFigure3.5(b)forthemaingaparebasedonpermittivity, conductivity
and spacing values for practical systems. The principles can be applied to other
insulationarrangementsincorporatingdifferent materials.
3.2.3 Mul ti pl e el ectrode confi gur ati ons
During the investigation of certain types of fault it is often unclear as to why a
particulartrackingorflashover/puncturepathwasfollowed. Suchasituationcanarise
wherethreeor moreelectrodesareinvolved. Stressinginawell-insulateddirection
can producePDs that may causeafailurealong an unrelated path. This has been
demonstratedexperimentally at highvoltagewiththeelectrodesystemrepresented
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 67
H.V.
r
1
r
3
S
(a)
(b)
r
1
r
2
r
3
100
Insulation
interface
Capacitive
distribution
Resistive
distribution
80
60
V
o
l
t
a
g
e
,

%
40
20
0
r
2
Creep
surface
Oil
Solid
Radius
Fi gure 3.5 I nsul ati on suppor t for i nsul ated hi gh vol tage l ead. (a) Basi c confi gu-
r ati on; (b) Vol tage di str i buti on i n mai n gap for AC and DC condi ti ons
(sol i d/oi l system)
d
S
12
S
13
High voltage V
1
Wedge
Low voltage or earth V
2
V
3
Fi gure 3.6 Effect of nor mal and tangenti al stressi ng i n a mul ti pl e el ectrode system
in Figure 3.6. This is an example of a practical test in oil in which the flashover
voltagestrengthalongtheinsulationboardwasreducedby anincreaseinthestress
valueacrosstheinsulationthickness.
Althoughpaths S
12
andS
13
may beconsideredindependently, theinitiationof
PDs intheoil wedgedueto thenormal stress betweenV
1
andV
2
canreducethe
breakdown strength alongthetangential path S
13
between theelectrodes at volt-
agesV
1
andV
3
. Similar situationscanoccur indry-typeequipment. Suchascenario
requires careful analysis of theelectric field in relation to theoriginal design and
howthis may haveinfluencedthefailurepattern. A morecomplex situationexists
intransformer windings, whereeachturnmay betreatedas aseparateelectrodeat
adifferent potential fromtheothers and fromadjacent windings. In dry-typecon-
figurations, wheredischargesinthewedgemay beinducedmoreeasily, suchstress
conditionscanbeverysignificant.
68 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Simultaneous testing between phases and to ground in three-phase equipment
mayproducesimilar situations.
3.3 Electr ic str ess contr ol
Inorder toensurethemosteconomicuseof spaceandmaterials, insulationdesigners
haveemployedarangeof techniqueseffectivelytoreducethemaximumtoaverage
stressratiointheconfigurationsutilizedinhigh-voltageequipment. Undersomecon-
ditions suchmethods may not behelpful. For example, duringtests withinsulated
rods toaplaneinoil it was foundthat themoreuniformfields gavelower average
stressesat breakdown. Thiswasduetothevolume effect associatedwiththeten-
dency for particlestobeuniformly distributedthroughout theoil space. Inthecase
of thelessuniformsystemsit canbepostulatedthat theparticlesareattractedtothe
morehighlystressedregions, therebyproducinganoverall oil gapof greater strength
inserieswiththeinsulatedhigh-voltageelectrode.
The outstanding examples of stress control are in cable terminations, bush-
ings, instrument transformers, insulator strings, stator bars, switchgear and power
transformers. Inmanysituationstheseinsulationsystemsmust bedesignedtowith-
stand impulse voltages of twice or more the magnitude of the RMS value of the
correspondingpower frequencytest voltage.
Many methods of stress control arepractised as described in numerous papers
andbooks. A number of areasof applicationof thetechniquesaresummarizedbelow
and, whereappropriate, referencemadetodiagramsinthisbook.
(a) Reductionof stressat high-voltageexternal terminationsby
useof ringsandsimilarlyshapedelectrodesinsubstationsonsupport struc-
tures, surgearresters, circuit breakers, disconnect switches (isolators) and
other devices(Figures1.41.6);
application of continuous shields, or possibly cages, for bushing
(Figure5.11) andsealingendterminations.
(b) Reductionof stressat high-voltageconductorsinair by
configuringtwoor four conductorsinparallel asontransmissionlines.
(c) Reductionof stressat internal support structuresby
shapingof cast-resinspacerstoimprovetheequipotential distributionat the
conductorsof SF
6
busbars;
dimensioning of insulation to reduce critical stress at HV lead supports
(Figure3.5).
(d) Reductionof stressat theedgesor cornersat earthedexitsby
contouring earthed foils in bushings (contoured or part of capacitor foil
system), andinsealingends(Figures4.9and4.12);
adding semiconducting tapes or similar compounds to plastic cables
(Figure4.12) andtostator bars(Figure5.2).
(e) Reductionof stressexternal tooil-immersedtransformer windingsby
includingmetallizedendringsindisc-typewindings(Figure5.10) andouter
shieldsinlayer-typewindings(Figure5.5(c)).
I ntroducti on to el ectr i cal i nsul ati on desi gn concepts 69
(f) Reductionandcontrol of stresswithinthemainstructuresof devicesby
windinginfoils withincurrent transformers (Figure4.4) andincapacitive
typebushings(Figure4.3);
adjustingtheturns arrangement togivecapacitivecontrol of surgevoltage
stressesintransformer windings(Figure5.4);
fittingof metallicshieldsaroundline-endtapchangers;
inclusion of shields in through joints of high-voltagecables (Figures 4.10
and4.13) anddry-typeterminations.
Someof theaboveaspects will behighlighted in Chapters 4 and 5 during the
descriptionof possiblefaultsinparticular equipment andcomponents.
3.4 Summar y
Thechapterreviewsmanyof thefactorsinvolvedinthedesignof electrical insulation
as applied in power-systemequipment ranging fromphysical, mechanical and
ageingaspectsthroughtothedeterminationof electricstress.
Stressanalysesof simpleinsulationconfigurationsarepresented. Suchconcepts
arehelpful inthedeterminationof theorderof magnitudeof electricstresses, perhaps
after afailureontest or inservice, andinchoosingthemost appropriatecomputer-
basedprogramtoinvestigatedetails. Thesebasicanalysesalsoassistinunderstanding
theadverseconditionsthatcanbepresentorarise. A numberof electricstress-control
methodsasappliedinpracticearelistedat theendof thechapter.
Anappreciationof theproblemsassociatedwiththeinsulationdesignof particular
power equipment isdesirablewhenchoosingor developingaconditionmonitoring
system.
3.5 Refer ences
1. Vitkovich, D., Fi el d anal ysi s: Exper i mental and Computati onal Methods
(D. VanNostrand, NewYork, 1966)
2. deKock, N., Mendik, M., Andjelic, Z., andBlaczczyk, A., Applicationof the3D
boundary element methodinthedesignof EHV GIS components, I EEE DEI S
El ectr i cal I nsul ati on Magazi ne, May/J une1998;14(3):1722
3. Ryan, H.M. (ed.), Hi gh Vol tage Engi neer i ng and Testi ng, 2nd edn (IET,
London, 2001)
4. Mason, J .H., Discharges, I EEE Tr ansacti ons on El ectr i cal I nsul ati on, August
1978; EI 13:21138
5. Kreuger F. H., Par ti al Di scharge Detecti on i n Hi gh-Vol tage Equi pment
(Butterworth& Co., London, 1989)
6. Binns, D.F., andRandall, T.J ., Calculationof potential gradientsfor adielectric
slabplacedbetweenasphereandplane, Proc. I nst. El ectr. Eng., 1967;114:15218
70 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3.6 Pr oblems
1. Identify the major electrical, thermal, mechanical and manufacturing require-
ments for electrical insulating materials as applied in the following AC
power-systemequipment. Indicatepossiblematerialsfor eachcase.
(i) Outdoor insulators(6.6kV andabove).
(ii) HV power transformers(132kV andabove).
(iii) HV cables(11kV andabove).
Refer toChapters2, 4and5.
2. For thecasesinQ1suggesttheinsulationparametersthatmightbemonitoredin
service.
Refer toChapters2, 4and9.
3. A pieceof oil-impregnatedpressboardisplacedbetweentwoparallel plateelec-
trodesintransformeroil. Theeffectiveareaof theelectrodesis1010cmandthe
spacing1cm. Thethicknessof thepressboardis3mmanditsarea1010cm.
Therelativepermittivitiesof theoil andpressboardare2.2and4.4respectively.
Assumethecorrespondingresistivitiesare10
11
and10
12
ohm-metres.
Calculatetheelectrical stressesintheoil andboard:
(i) For avoltageof 100kV (RMS) appliedbetweentheelectrodes. Assumea
parallel field.
(ii) Repeat (i) for adirect voltageof 100kV applied.
What arethepractical implicationsof thedifferenceinthefielddistributions?
4. For theconditions in Q3 determinethecapacitances of theoil and pressboard
thicknesses and the total capacitance. Calculate the corresponding resistance
values. Assumea1010cmarea. (
o
= 8.85410
12
)
5. For thedimensions in Q3 determinetheexpected terminal voltageat onset of
partial dischargesor breakdownintheoil withthepressboardlocated(i) at one
of theelectrodesand(ii) inthemiddleof thegap.
Refer toFigure2.3of Chapter 2.
Chapter 4
I nsulation defects in power-system equipment:
Par t 1
Insulatorsandbushings
Capacitor voltagetransformers
Power capacitors
Surgearresters
Circuit breakersandgasinsulatedsystems
Power cables
Inthischapter arediscussedtheinsulationconfigurationsof anumber of power
systemcomponents that may beconsidered as behaving predominantly as capaci-
tances. RotatingmachinesandtransformersarereviewedinChapter 5. Descriptions
andexamplesaregivenof insulationconditionsthatmightresultindeteriorationand
possiblefailure.
4.1 Suspension and post insulator s
4.1.1 Suspensi on (str i ng) i nsul ator s
Theconstructionof oneformof conventional cap-and-pin-typesuspensioninsulator
asusedinstringsisdepictedinFigure4.1(a). Theunitsconsist of porcelainor glass
as themajor insulation andareconnectedin series accordingto theparticular line
voltage. For example, a330kV linemay require20individual units dependingon
theexpected pollution at thetower site. It is notablethat in 1996 glass insulators
representedincost terms 19per cent of theworlds exports of porcelainandglass
units [1]. Thesurfaceprofiles of theinsulator sheds aredesigned to minimizethe
effect of pollution, rain, fogandiceasspecifiedfor thelocal application. Inextreme
conditionssuchasnear thecoast, indust or desert areasandinlocationsadjacent to
someindustrial installations, flashoversmayoccur duetothecreationof conduction
paths. A particular electrical dischargephenomenonisassociatedwiththeformation
72 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Concrete
(a) (b) (c)
Porcelain
Metal cap
Metal pin (HV)
Polymer
protection and
sheds
(i)
(ii)
Metal
(HV)
Fi gure 4.1 Exampl es of suspensi on and post i nsul ator s. (a) Porcel ai n suspensi on
i nsul ator ; (b) Composi te pol ymer suspensi on i nsul ator ; (c) I nsul ator
shel l s (i ) porcel ai n (i i ) composi te
of dry bands ontheporcelainsurface. Thebandsarecreatedby theelectric stress
inlocalizedregions onthesurfacebeingof sufficient magnitudeto producelosses
inthepollutant, whichmaydryout becausetheambient condition(e.g. themoisture
evaporationrate) doesnot allowdissipationof theheat. Thiscanresult inabuild-up
of voltageacrossthebandwithsubsequent dischargesor arcing[2].
Another fault that mayor maynot betroublesomewithceramicinsulatorsisthe
presenceof partial discharging(corona). AlthoughthePDs, usually associatedwith
themetal work, rarelycausedamagetothematerials, theycanresultinunacceptable
radioandTV interference.
Failuresof insulatorswerecausedby excessivedepositionof iceandsnowdur-
ing storms in theUSA and Canada. Thesetypes of failurearesurveyed in a2005
CIGRE report [3], which contains a number of references related to the condi-
tion. EPRI investigations have indicated that a number of cap-and-pin insulators
wouldbeexpectedtolosemuchof their insulatingpropertiesafter 60yearsinnor-
mal service [4]. Despite the excellent performance of the older units, cases have
beenreportedof catastrophic mechanical failuresof recently manufacturedceramic
insulators.
Long-rodcompositepolymericinsulatorsareincreasingly beingapplieddespite
the naturally cautious approach of utilities. The advantages include reduction in
weight givingeasier handling andsomewhatimprovedperformanceunder heavy
pollutionconditions. Howeverthereisstill someconcernregardingchemical changes
duetodry-band-arcing, coronaandweathering. Laboratoryandfieldtestshavebeen
carriedout for manyyearsbut thereappearstobeawiderangeof opinionandexpe-
rienceastothelong-termeffects. IntheUnitedStatesahighpercentageof all new
insulatorsarecomposites, whileinother countriestheintroductionof polymer units
isbeingcarriedout withcautionor possiblyonatrial basis.
I nsul ati on defects i n power -system equi pment: Par t 1 73
Theprincipleof theconstruction of long-rod compositeinsulators is shown in
Figure4.1(b) astressringat thelower endisnot included. Theload-bearinginner
rod consists of fibreglass-reinforced polymer (FRP) epoxy resin probably being
themost usedpolymer terminatedwithmetal ends. Theseareshapedtominimize
corona, whosepresencecancontributetobrittlenessandcrackingof theglassresin
composite. Thepolymer seal must bemaintainedinorder toprevent accessof mois-
ture. Thecomplex processesassociatedwithbrittle failureshavebeenstudiedfor
manyyears. Partof theresearchintheUSA supportedbytheElectricPowerResearch
Institute(EPRI) isreportedinsomedetail inReference5. Thecatastrophic failures
develop radially across the glass fibres, probably in the high-voltage region, and
appear tobeduetothecombinedinfluenceof theworkingmechanical stress, acidic
attack due to the interaction of the materials (resin and glass), the electric stress,
coronaeffects (ozone) andmoistureingress/diffusion. Theuseof polyester resinis
consideredtobeunsuitablebytheEPRI investigators. Althoughthenumber of fail-
uresislow(200) comparedwiththosenowinservice(>10
6
) [5] thelong-termeffects
arenotknown. However, variousdesignandmanufacturingchangeshavebeenmade
basedonthestudies so far, includingrecommendations regardingcorrect handling
procedures.
Incompositeinsulators, theweather sheds areformedover therodandusually
consistof siliconerubber (SIR). Chemical bondingiseffectedbetweentheshedsand
therodandbetweenthesheds themselves. As theelectric stress acts alongtherod
fibres and thelongitudinal interface, it might beexpected that any manufacturing
or serviceproblemsintheseregionscouldresult inelectrical aswell asmechanical
weaknesses[4].
SIRisprobablythepreferredmaterial atthepresenttime, becauseitshydrophobic
properties(minimumaffinityfor water) aregoodandithasahigher resistancetoUV
radiationthansomeother materials andalso to dry-bandarcing. After exposureto
intensedischarges, SIR forms silica, whichis non-conducting. Other materials that
have been used and are in service include ethylene propylene rubber (EPR) and
cycloaliphaticepoxy.
For many yearsroom-temperature-vulcanized(RTV) SIR high-voltagecoatings
havebeenusedbysomeutilitiesforimprovingthepollutionperformanceof porcelain
insulators. Thebehaviour of suchsystemsisdiscussedby Cherney et al . [6], where
it is emphasized that, unlikethecasewith porcelain, low-energy coronafromend
fittings candegradetheRTVSIR coating. This canbeasignificant problemif the
original source of corona is not removed by modification of rods, end rings and
similar.
Themonitoringof potential faultsassociatedwithinsulatorsmaybedetectedby
thermal (or acoustic) scanning, asdiscussedinChapter 9.
4.1.2 Post i nsul ator s
Post insulators may consist of a solid core, a shell of porcelain or, in the case of
composites, an inner, fibre-wound epoxy cylinder with metal terminations and a
protectivepolymer covering. For thelastof these, shedsareprovidedof acompatible
74 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
material e.g. siliconerubber. Identical faultstothosefor thesuspensionunitshave
not been reported [4], although it is possible that high crimping pressures of the
terminationsmay havecausedapparently similar mechanical failuresat theearthed
end, as well as at thehigh-voltageconnection. Larger-sizeshells arealso used in
bushings, instrument transformersandsurgearresters.
4.2 High-voltage bushings
Bushingsintheir variousformsareusedwherever it isnecessarytopassalivecon-
ductorthroughametal plateortankatadifferentpotential usuallyatearthpotential.
ThebasicrequirementsareindicatedinFigure4.2(a). A bushingconsistsof ametal
rod(or tube) supportedbyasimpleinsulationstructureatthelower voltagesthrough
to higher-voltage units containing many stress-grading foils within the insulation.
Commercial bushingsareclassifiedinStandardssuchasIEC 60137[S4/1].
Themethodof improvingstresscontrol bymeansof isolatedconductingfoilsis
showninFigure4.2(b), whereacomparisonismadebetweenanungradedandgraded
configurationof ahypothetical bushing. It isapparent that thevoltagedistributions
axiallyalongtheinsulationsurfacesandradiallythroughthestructurearemuchmore
uniformincondition(ii) thanintheuncontrolledcase(i).
Inpracticeit is not possibletoachievetheideal of uniformdistributioninboth
axes by adjustingtherelativecapacitances [7]. Computer programs areutilizedfor
solvingthecomplexcalculationsrequiredtodeterminethenumberanddimensionsof
foilsnecessarytogiveacceptablestresseswithinthestructure. Thevaluesarerelated
tothepunctureandcreepstrengthsof thevarioussolidinsulationsandtheassociated
To
minimise
D-d and L
(a) (b)
High
surface
stress
75%
25%
40%
50%
60%
75%
Surface stress
50%
40%
High
stress
No foils
(i)
Grading foils
(ii)
Surface stress
reduced
Creep
Creep
d
D
Earth
L
Fi gure 4.2 Basi s of hi gh vol tage bushi ngs. (a) Desi gn requi rements; (b) Equi po-
tenti al di str i buti on (i ) No foi l s (i i ) Gr adi ng foi l s
I nsul ati on defects i n power -system equi pment: Par t 1 75
ambient mediumof air, SF
6
, or oil. According to thevoltageclass thenumber of
foils varies fromzero at 11kV to two or threeat 33/66kV throughto perhaps one
hundredor moreatthehighestlevels. Caremustbetakenthatoverstressingdoesnot
occur atthefoil terminations. Theouter foil of themajorityof high-voltagebushings
of 72kV classandabovemaybefloated inorder toprovideameansof monitoring
theinsulationconditioninthefactory andinservice. It is normally earthed. If not
earthed, or connectedthroughanappropriateimpedance, thefoil will risetoahigh
voltageandprobablycauseaninsulationfailure.
Thetypesandformsof theinsulatingmaterialsinbushingsdependonthemanu-
facturerspractice, thevoltageclassand, tosomeextent, theageof theproduct. Older
unitswereinsulatedwithSRBPandthesolidstructureimmersedinoil withinaporce-
lainshell. Suchdesigns canoperatesuccessfully upto 300kV but therehavebeen
problemsregardingpartial dischargingwithintheresinpaper wraps. Bothlossangle
andPDmeasurementsareappliedtoassessthestateof, forexample, SRBP66kV and
132kV bushings, manyof whichof theolder unitshaveinternallyerodedinsulation,
oftenproducedover longperiodsof time. Insomecasesreplacementprogrammesof
upto30-year-oldunitshavebeeninstitutedbasedonsuchmeasurements[8]. Some
of thesebushingshavebeenoperatingwithPDlevelsof several hundredpC. High-
voltageresin-vacuum-impregnatedpaper bushingshavebeenproducedfor anumber
of years includingunits upto500kV. However, it appears that theoil-impregnated
paper typeisstill preferredfor manyapplicationsat thehigher voltages.
Oil-impregnatedpaper bushingsarebuilt into, for example, power transformers,
shunt andseriesreactors, switchgear andenclosuresat substationsuptothehighest
voltages. In thesedesigns it is essential that themulti-foil paper structurebewell
driedandefficiently impregnatedwithdry oil under lowvacuumwhenmountedin
theporcelainhousing(shell). Morerecently, compositeshells possiblywithsilicone
sheds andprotection havebeenmanufactured. Lowpartial-dischargelevels must
beachievedinall high-voltagebushingsinordertopreventproblemsinservice. Also
lowvalues arenecessary in any bushing typeinstalled in equipment in which PD
levelsof tensof pCsor lessarerequiredtobemeasured. A stateof drynessmust be
maintainedduringoperation, otherwisethedielectric losses may risesignificantly,
resultinginthermal deteriorationof theinsulationandpossiblytheinitiationof high-
level PDs(hundreds/thousandsof pCs). Bothchangescanresultinfailureandinternal
explosion. Sucheventshaveresultedinintensefiresandwidespreaddamagewithin
largepower transformers. Theapplicationof compositeshells reduces thedamage
anddanger topersonnel shouldanexplosionoccur.
An example of external flashovers of wall bushings due to salt pollution is
described in Reference 9. The failure resulted in the outage of a complete power
station. The bushings were replaced with another design and a leakage-current-
monitoringsysteminstalledhavinganinitial sensitivity level of 13mA. Automatic
washingequipment wasalsoinstalledasatrial.
Typical equipotential distributionsaroundtheoil endof aconventional andare-
entranthigh-voltagetransformerbushingmountedinaturretaredepictedinFigure4.3
[10]. Themanner inwhichthefieldiscontrolledbythecapacitivegrading(foils) and
the stress distributor at the point of entry of the high-voltage lead is apparent for
76 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Foils
Foils 10%
20
30
40
50
60
70
80
90
100%
90% 60% 30% 10% 0%
(a) (b)
100%
Porcelain
1
0
0
%
70%
20%
Paper
Turret
(earth)
0%
10%
20
30
40
90
80
Oil Oil
Stress
distributor
(HV)
Fi gure 4.3 Typi cal equi potenti al fi el d pl ots around the oi l ends of hi gh-vol tage
bushi ngs wi thi n an ear thed tur ret [ 10] . (a) Conventi onal type; (b) Re-
entr ant type [ Reproduced by per mi ssi on of CI GRE]
theconventional arrangement. Thefield is moreuniformfor there-entrant design
andrequireslessspacewithintheturret but reducestheallowablemovement when
installingthebushing. Duringtestingof thistypeit isessential toprovideatail to
ensurethefielddistributioniscorrectalongtheinnersurfaceof thelowerporcelain. In
earliertypesfailureshaveoccurredduetogasbecomingtrappedbetweentheleadand
there-entrantsectionof theporcelain. Methodsarenowincorporatedtopreventsuch
acondition. Failureshavebeenreportedinconventional-typebushingsduetodefects
inthefoilsandpoor connectionof theinternal stressshieldatthelower endterminal.
Thelatter condition can producePD disturbances of hundreds of picocoulombs in
service.
Inthecaseof bushingsfor direct-voltageapplications, thefieldplotsaresignifi-
cantly different fromthoseinFigure4.3becauseof theeffect of thehighresistivity
of thesolid compared with that of theoil seeChapter 3. Thedifferences within
abushing arediscussed in Reference10. It is shown how theincreased electrical
stressesunder DCconditionsmaybecontrolledbythejudiciousapplicationof insu-
lationbarriers aroundtheendof thelower porcelainandpaper insulatedshieldas,
for example, inconverter transformer practice.
At the lower voltages extensive use is made of casting resins, especially for
switchgear applications. In such units it is possible to improve the electric field
I nsul ati on defects i n power -system equi pment: Par t 1 77
conditions at theouter diameter by modifyingthecontour of theearthelectrodeor
includinganearthedfoil of appropriateshape, bothwithinthebodyof theinsulation.
However theproductionof thesebushings must beof highquality, as any voids or
trapped air can result in partial discharging and subsequent failurein service. PD
levelsof below200pC areknowntohavecausedfailuresinservice. Valuesfor new
componentsareinthetensof pCs. AcceptancelevelsarespecifiedinIEC Publica-
tion 60137 [S4/1]. User measurements areessential beforecommissioning critical
equipmentcontainingbushingsof thistype, particularlywhereunusual materialsare
incorporatedinthestructure.
With some of the older 11kV bushings difficulties have arisen during the PD
testing of high-voltage power transformers because of internal discharging within
the LV conductor to porcelain radial oil, or even air, gap. Also, problems have
beenexperienceddueto coronaaroundtheunshieldedearthclampingsystem(see
Figure4.2(b, i)).
Itshouldbeappreciatedthatsomeof theelectrical faultsassociatedwithinsulators
canalso occur withporcelainandresins/polymers utilizedinbushings. Inall cases
thisappliestoanyattachedhardware particularlyrodgaps.
4.3 High-voltage instr ument tr ansfor mer s
Insulationproblems associatedwithinstrument transformers haveassumedconsid-
erable importance during the past ten to twenty years. This applies particularly
to high-voltagecurrent transformers (HVCT) of thehairpin design. Much effort
hasbeendevotedtodevelopingefficient andeconomicinstrument-transformer site-
monitoringsystems. Althoughthecostof anindividual transformerisrelativelysmall,
thefailureof such aunit can result in aviolent explosion with high consequential
damage. Thisisinadditiontotherisk of injurytopersonnel duetofracturedporce-
lain, sections of which can be projected many metres. Valuable information was
collected through international surveys by CIGRE WG 23.07 (197085) and WG
12.16(198595). Inthelatter theHVCT (>60kV) samplesizewas approximately
89200, including24000combinedCT/VT units. Of thistotal 95per cent wereoil-
paper insulationsystems, 3per cent resinandtheremainder SF
6
. Theoverall major
failurerateper year for oilpaper HVCTswas0.035per cent (60kV to>380kV) of
thatpopulation, averagedover theten-year period; thecorrespondingvaluefor resin-
insulatedunits(60kV to200kV) was0.175per cent and0per cent for SF
6
designs
uptoandabove380kV. Major year l y failureratesfor magneticvoltagetransformers
(MVT) andcapacitor voltagetransformers(CVT) wereof theorder of 0.044per cent
and0.026percentrespectively. Althoughthesemajorfailureratesarelow, eachevent
representsapotentiallycatastrophicsituationand, therefore, theyareof considerable
concerntotheindustry.
Theproblemisbeingaddressedinmanyways, inparticular bydesignchangesin
newunits, includingtheapplicationof polymercompositehousingsinordertoreduce
thedangerousconsequencesof anexplosivefailure, theuseof SF
6
insulationsystems
and, in the longer term, the limited application of fibre-optic-type units [11,14].
78 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
However, many thousands of existing oilpaper-insulated instrument transformers
will continueto operatefor many years to come[8]. It is important that users be
awareof possiblefailurescenarios, especially for theoil-impregnatedunits, andof
thesite-monitoringmethodsnowavailableor under development (seeChapter 9).
4.3.1 Oi l -i mpregnated cur rent tr ansfor mer s
Theconstructions of HVCTs vary considerably, as indicated in Figure4.4. These
may be designated (a) hairpin, (b) eye bolt and (c) bar primary. Earlier designs
were reviewed in a report prepared for Study Committee SC 23 of CIGRE by
WG23.08.
Incases(a) and(b) theoilpaper insulationstructuresareequivalent toelectro-
staticallygradedbushingswiththeinner electrodeat highvoltageandtheouter foil
at earth, adjacent totheearthedcorearoundwhicharelocatedthesecondary low-
voltagewindings. A 330kV CT might incorporate30mainfoils anda66kV unit,
only two or three, each with end rings. In some very high-voltage designs short
intermediatefoilsmaybeusedbetweentheendsof themainfoilsinorder toimprove
theendstresseswithout incorporatinglocal stressrings. Thestructuresareenclosed
inahousingmountedonametal tank, thewholebeingdriedat temperatures of up
to120

C andvacuumsof lessthan0.02torr andthenfilledwithprocessedoil under

vacuum. Becauseof thelargediameters required, eachlegof somehairpindesigns
is accommodated in its own housing. It is noted that in J apan aporcelain bushing
housing(shell) of 1.5metrediameter and11.5metreslengthhasbeendevelopedfor
1000kV AC operationbut suchsizesareveryexpensive.
Thebar primaryarrangement (c) is, ineffect, aninvertedbushingwiththeinner
electrodetubecontainingtheconnections fromthesecondary windings, whichare
mounted on the earthed core, also insulated fromthe high-voltage primary bar at
thelineend. Themajor internal insulationfromthehigh-voltageouter to theinner
Fi gure 4.4 Exampl es of cur rent tr ansfor mer (HVCT) confi gur ati ons. (a) Hai r pi n
type; (b) Eye bol t type; (c) Bar pr i mar y type
I nsul ati on defects i n power -system equi pment: Par t 1 79
groundedtubeisprovidedbyanaxiallyandradiallyvoltage-gradedinsulationsystem.
A housingisprovidedfor containmentof theoil andfor supportingtheoverall struc-
ture. Theheadof thetypein(c) may bemetal or of resin. Suchunits areavailable
for operationin765kV systems. Thecapacitancesof thevariousHVCTsareinthe
rangeof afewhundredtopossiblyathousandpicofarads.
Most of themajor HVCT failures havebeen associated with paper/oil hairpin
units. The breakdown modes are difficult to identify but probably involve one or
moreof thefollowing:
(a) Degradation of the paper systemdue to ingress of moisture through leaking
seals. A faultcoulddevelopasthedielectriclossesincrease, resultinginthermal
instability at the operating stresses of 34kV/mm. This, together with surge
effects, mayexplainfailuresafter relativelyshort periodsof 10to15years.
(b) Development of trappedgasinthebodyof theinsulationduringtheprocessin
(a) resulting in theinception of partial discharges, initially of low levels and
increasing to values of thousands of pCs beforefailureby puncture. X-wax
andsmall black spotsinnon-failedregionsareindicativeof long-termthermal
instability followedby partial discharging. Sucheffects may arisedueto poor
dryout andimpregnationduringmanufacture.
(c) Overstressingatafoil terminationproducedbydeteriorationorphysical damage
couldresult inPDsleadingtopuncture.
(d) Inconsistencies in the taped paper structure allowing partial discharges to
develop dueto reduction in thickness and, also, thecreation of oil gaps (see
Figure4.5).
Fi gure 4.5 Fai l ure of HV cur rent tr ansfor mer oi l paper tapi ng due to di el ectr i c
over stressi ng and par ti al di scharges [ 26] [ reproduced by per mi ssi on of
CI GRE]
80 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(e) Long-termdisintegrationof theconnectionbetweentheinternal DDF terminal
(normallyearthed) andthefirstfoil usedforthemeasurement. Thismayproduce
gas, whichresultsinaxial flashover throughtheoil. Thefault couldbeduetoa
cumulativeeffect producedby thesurgecurrentsandassociatedarcingcreated
duringswitchingof local air isolators. Anoptional special testisincludedinthe
revisionof theinstrument transformer standard[S4/2, Annex B] requiringthe
applicationof aminimumof 100choppedwaves.
(f) Badconnectionsorfloatingmetal attheDDFtap, whichmayresultinacondition
asin(a) (Figure2.8(vii)).
Condition(d)mayexplainsomefailuresafterlessthanayearsservice. Thefailure
modeprobably involves paper burningbetweenseveral foils andtheproductionof
PDs possiblyincreasingtotennCsor morenear failure. A commonfeatureseems
to be that the final scenario develops quickly perhaps in days or even hours
makingpredictionof breakdownby meansof thecommonest monitoringtechnique
of periodicoil samplingof DGAineffective. Also, ithasbeenfoundinsomecasesthat
asignificant changeintheDDF valuewouldnot necessarily ariseif, for example,
considerable burning between two foils was present [13]. In other instances low
partial-dischargevalues havebeenrecorded, eventhoughtrackingwas established.
Nevertheless, continuous monitoring of DDF and, where practical, PDs is highly
desirable for critical installations. It is noted that in these types of failure a high
internal pressureisbuiltupduetoarcgasesbeingproducedwithinthesmall volume
of theCT housing. This can, anddoes, result inexplosions, causingtheshattering
of the porcelain with dangerous consequences. Other anomalies are described in
Reference13.
Partial-dischargelevelsmayrangefromhundredstomanythousandsof pCsina
dielectricstress typefailure. DDF valuespossiblyincreasetohundredsof mRsina
thermal runaway breakdownmode.
4.3.2 Dr y-type cur rent tr ansfor mer s
Inthesurveys quotedaboveit was reportedthat major failures of cast-resinCTs in
thevoltagerange>60kV to <200kV were0.175 per cent of their population. It
is not known whether thereis adominant causefor thesefailures. However, with
such components good quality control is essential in order to ensurethat no voids
are produced during processing. An example of such a problemwas a 33kV/

3
bar primary unit that, onreceipt fromareputablemanufacturer, was foundto have
a PD level of 200pC at operating voltage. Due to operational pressures the CT
was installedandfailedafter oneyears service. Pre-serviceand, perhaps, periodic
PD measurements may bejustified for this typeof construction. Another example
of a poor-quality product was an 11kV CT designed for mounting on a bushing.
In this case a low-quality plastic insulating tube and associated compounds were
used, resultinginanunacceptably highDDF value. Theexcessivedielectric losses
could have produced a service failure. Again, the unit was froman established
manufacturer.
I nsul ati on defects i n power -system equi pment: Par t 1 81
Atthehighervoltages, livetankdry-typeunitsarenowcommon, utilisingSF
6
gas
astheinsulantbetweentheprimarybarandtheearthedcomponentslocatedinsidethe
head. Thesupportstructuremaybeof porcelainorpolymericmaterial. Anappropriate
gaspressuremust bemaintainedduringthelifeof theCT anditslevel monitoredin
service.
4.3.3 Capaci tor -type vol tage tr ansfor mer s CVT
Therearetwobasicformsof voltagetransformerforuseathighvoltage: thecapacitor
divider type(seeFigure4.6) andthemagnetictype MVT (seeChapter 5).
Thecheaper, butpossiblylessaccurate, capacitor-type(CVT)designiseffectively
avoltagedivider withatunedlow-voltagearm. Thevoltageacrossthisarmwouldbe
approximately19kV withacapacitanceratioof, for example, 3800/38000pF for a
330kV unit and, possibly, 12kV (5600/30000pF) for a132kV CT. Thecapacitors
consist of manyserieselementssubjectedtoaslight positivepressurewithasteady-
state voltage of, typically, 2kV across each. The design of the elements and the
operatingstresses(20kV/mm) aresimilartothoseusedinpaper/oil powercapacitors.
Wherepaperisincluded, theinsulationsystemisdriedandvacuumimpregnatedwith
processedoil.
Inthecaseof thecapacitor units(Figure4.6) major failureshavebeenrelatedto
dischargingat theendof foilsof individual elements(Figure3.3), thermal runaway
andoil leakageexposingtheupper elements. It ispossiblefor twoor threeelements
tobeshortedandamajor failurenot tooccur immediately. Someoperatorsconsider
that physical inspection, e.g. for oil leakage, can beavery useful indicator of the
internal stateof aninstrument transformer. Suchaprocedureislabour-intensiveand
may be hazardous in the case of suspect units. Because the loss of elements will
vary theoutput voltage, theinspectionof CVTs canbeminimizedby continuously
monitoringanysmall changesintheoutput voltage.
Insulation housing
Tuned LV arm
HV terminal (sealed)
Capacitor units in series
Low voltage tap
Earthed case
Fi gure 4.6 Capaci tor vol tage tr ansfor mer (CVT) confi gur ati on
82 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
4.4 High-voltage power capacitor s
Theapplicationof capacitorsforpower-factorcorrectionathighvoltagesisbecoming
moreextensivewith theincorporation of static VAR compensators (SCV) into the
supply systems. The usual design is to mount outdoors in the substation a large
number of individual units in a series/parallel configuration at the LV side of a
step-down coupling transformer. The capacitance values of the units range up to
hundredsof microfaradsand10to20kV, dependingontheparticular internal design
andapplication.
Therearevarious forms of power capacitor, thesimplest of whichcomprises a
series/parallel arrangement of elements mounted within a metal tank or casehav-
ingdimensionsof perhapstheorder of 500 200 750mmhigh, but thesevalues
are very dependent on the particular application. In a particular design, each ele-
ment consists of acompressed roll of continuous lengths of two metal foils the
capacitor electrodes separatedbythinsheetsof paper, paper/plasticfilmor special
filmonly, acting as the insulation. The whole is dried and filled with an insulat-
ingliquid natural or synthetic under highvacuum. A simplelayout is indicated
inFigure4.7.
Thefoilsof thevariouselementsareinterconnectedwithinthetank beforepro-
cessinginorder togivetherequiredcapacitanceat theterminals, whichareusually
outdoor-typebushings. Theolder unitswereimpregnatedwithlow-viscositymineral
oil but, later, PCBliquidwasalsoused(seeChapter2) becauseof itshighpermittivity
(
r
= 5) and, incidentally, itshigher flashpoint. However, capacitorscontainingPCB
areunacceptableforenvironmental reasonsandtheyhavenowbeenreplacedbyunits
impregnatedwithoneof themany other synthetic dielectric liquidsdevelopedover
thepast twenty years or more. Thefoilfoil operating voltages areof theorder of
2kV, givingstresses of 1318V/mfor paper, 40V/mfor mixeddielectrics and
Foil terminals of
one element
Insulation
barrier
Metal case
Capacitor elements
(interconnected)
Note: External bushings
not shown
Foils separated by
insulation
Fi gure 4.7 Possi bl e confi gur ati on of foi l s and i nsul ati on i n a power capaci tor
I nsul ati on defects i n power -system equi pment: Par t 1 83
>50V/mfor all film. Inthelast casespecial materialsarenecessarytoensureeffi-
cient vacuumimpregnationandtominimizeswelling. Researchiscontinuinginthis
area.
Because of the thin insulating materials used in the capacitors, great care is
requiredtoensurenoparticlesbecometrappedduringmanufactureandthat thefoil
endsarelocatedcorrectlywithreliableconnections. Also, at thesehighstressesit is
essential that theprocessingeliminates thepossibility of voids, otherwisePDs will
causefailure. Thisappliesparticularlywiththefilmmaterials, wheredry areasmay
becreated. Ultrasonicmonitoringdeviceshavebeenappliedasaqualitycheck dur-
ingproduction. Someproblemsinservicehavebeenassociatedwithdeteriorationof
theinsulationbetweenthecapacitor elements andthetank side. Suspect units with
theolder paper/oil systemwerefound to havePD levels of several thousand pCs
after removal fromthehigh-voltagestructure. Theseweredetectedby direct obser-
vationof distortionof thetank dueto gas build-upor offlinechecks of units from
asuspect capacitor bank. Other difficultieshaveinvolvedthebushingsandinternal
fusemountings. Thedetectionof partial discharges inpower capacitors tends tobe
insensitivebecauseof thelargecapacitancevaluesand, possibly, theresonanteffects
of theconnections. Traditionally, capacitors aremonitored by DDF measurements
but thistechniquetendstobelessuseful, becausethelossesintheall-filmunitsare
very low. It is notedthat thefailurerates for thenon-paper designs andthemixed
insulationtypesaresignificantly lower thanfor thepaper/oil systems. A number of
standardizedrequirementsaregiveninIEC Publication60871, Parts1and2[S4/4].
4.5 High-voltage sur ge ar r ester s
Surgearrestersareusedfor limitingthemagnitudeof lightningandswitchingsurge
voltagesonoverheadlinesandequipmentoperatingintherange6.6kV upto500kV
AC andabove, andonDC transmissionsystems. Their applicationaspart of system
insulationcoordinationisdefinedinIEC 60071, Parts2and5[S4/5].
Themajor components arenonlinear siliconcarbidediscs withseries gaps and
metal oxidewithout series gaps [S4/6] andanappropriatehousingof porcelainor,
morerecently, polymer. Thelayout for alow-voltageunit is shown in Figure4.8.
High-voltagearrestersareincludedinFigure5.11(a).
Thesedevices areanessential part of theinsulationcoordinationdesignof the
power systemand therefore must be reliable under steady-state as well as surge
conditions. Failureshaveoccurredthataresuspectedof beingdueto, for example,
(a) ingressof moisture goodsealingisessential; thisisaproblemwiththeolder
gaptypeunits;
(b) localizedlossesanddischargingcausedbypoor inter-disccontact;
(c) deterioration of grading resistors in the gapped type, leading to sparking at
operatingvoltage;
(d) housingdeterioration(or pollution) changingthevoltagedistributionalongthe
stack;
84 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Porcelain or
polymer housing
HV terminal
Series gap
Non-linear resistor
multiple discs
LV terminal
Fi gure 4.8 Layout of a si ngl e-stage gap-type surge ar rester
(e) mechanical fractures inthemetal-oxidematerial (gapless type) duetothermal
runawayafter ahighcurrentsurge; alsodamageduetosurgecurrentconcentra-
tionat theedgeof theelectrodecanresult infailure[15];
(f) resultant damage to the discs created by previous multiple-stroke lightning
surges; thisisaconditionnot coveredbyroutineor typetestsandmaybevery
significant [16].
Metal-oxide arresters are inherently faster-acting than the gapped type, since
thereisnotimedelay duetoseriesair gapsextinguishingthecurrent. It seemsthat
themetal-oxideunitsareprovingmorereliablethanthepreviousdesigns.
Incritical locations, e.g. adjacenttoamajortransmissionorgeneratortransformer,
it maybejustifiabletomonitor arrester conditionscontinuously seeChapter 9.
4.6 High-voltage cir cuit br eaker s
Therearemany types of circuit breaker installed in power supply systems. These
rangefrom11kV air breakmetal-clador resin-insulatedsystems, bulkoil unitsfrom
11kV upto275kV, low-oil-content livetankdesigns, air blast of deadandlivetank
configurationsupto400kV and, inrecent years, thewidespreadapplicationof SF
6
units. Different designsof thelast of theseareappliedover thewholevoltagerange.
Themajorityof newbreakersareof theSF
6
typeandincludeliveanddeadtanktypes.
Examplesof SF
6
unitsproducedbysixmanufacturersarepresentedinChapter 8of
Reference18. However, manyof theexistingunitsatthelower voltagesareoil-filled
andanumber of air-blast typesremaininserviceat 132kV andpossibly275kV, all
of whichwill needregular periodicinspectionsuntil replaced.
For someapplications 11kV and 33kV vacuuminterrupters areinstalled. The
switches are vacuum-sealed and should not require maintenance throughout the
I nsul ati on defects i n power -system equi pment: Par t 1 85
specifiedlife. Thesebreakerscanproducesteeptransients, ascanSF
6
circuitbreakers
at thehigher voltages. Protection of thecircuits and connected plant can often be
effectedbycorrect useof metal-oxidearresters(MOA).
Insulation failures in circuit breakers would beexpected to beassociated with
deteriorationof thearc-suppressingmediumortheinsulationof componentsbetween
highvoltageandground. Accordingtotheageandtypeof breaker, thecomponents
might includebushings, laminatedwoodenoperatingarms inolder designs, SRBP
and resin-impregnated glass-fibreoperating rods, tubes of resin-impregnated glass
fibrefor compressed-air or SF
6
gasfeedsinlivetankconfigurations, outer shellsand
support insulatorsof porcelainor polymers.
Insulation breakdowns within SF
6
circuit breakers can bedueto chemical by-
productsproducingconductiveareasoninsulatorsurfaces, protrusionsonconductors,
conductiveparticles, faultsinthesoliddielectricsincludingvoids, trappedmoisture
and poor adhesion of parts. Dielectric overheating produced by bad contact in the
current circuit canalsoleadtobreakdown. Inpractice, failureratesduetoinsulation
deteriorationarelowcomparedwithotherfactorsassociatedwithbreakeroperations.
Internal flashover duetotrackingalonglaminatedcomponentsinolder bulk oil
breakerscanresultintheproductionof gasesandsubsequentexplosionandfire[17].
Aninterestingexampleof afailuretogroundwasduetoawoodenoperatingrod
withinanoil-filledporcelainshell supportingthelive-endtank of a330kV breaker.
Due to moisture ingress and the incorrect location of the rod with respect to the
shell, it was concludedthat flashover hadprobably occurredfromthehigh-voltage
terminationover theouter porcelainsurfacetoanintermediatemetal flange, across
the reduced oil gap to the rod and then along the rodoil interface to ground. In
addition to visual examination theprobablesequenceof events was deduced from
measurementsof theinsulationresistance(IR) valuesof thevariouspathsinvolved.
Anothercaseinvolvingmoistureingressmeasurementsonlaminatedcomponents
fromoil-filledswitchgear enabledacriterionbasedonIR values to bedetermined
for thecontinuedoperationof similar equipment thusactingasasimplecondition
assessment for theparticular insulationsystem[18].
Theapplicationof unsatisfactoryinsulationasspacersbetweenbusbarsinmetal
clad11kV switchgear createdmajor problems inanumber of substations. Follow-
inganinter-phasefailureat onesite, diagnostic techniques, includingDDF andPD
measurements with ultrasonic detectors, were applied. These procedures allowed
surveillanceof operatingsubstationswhileotherswerebeingmodified. Several years
wererequiredto updatethemany substations involved. Thesuccess of theproject
showedthevalueof havingavailablearangeof monitoringsystems together with
thenecessaryspecializedengineersandtest personnel.
Theapplicationof partial dischargetestinginanopen, conventional, high-voltage
switchyardisoftenrestrictedbythepresenceof air coronafromthebus-bar system.
Obvious sources arerodgaps andfaulty insulators. Siteexperiencehas shownthat
manydesignsof airisolatorsusedforprotectionduringmaintenanceprocedureshave
sharp-edgedguidesfor mechanicallylocatingthehorizontal armasitclosesontothe
line. Coronafromtheseguides canbeeasily identifiedusingahandheldultrasonic
detector and may be of several thousand pCs magnitude. In many cases it would
86 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
appear possibleto reducethedisturbances by modificationof theconfigurations to
allowmeasurementsonadjacentequipment. Othercoronasourcesnotedincludedthe
cornersof lugsprovidedfor temporaryearthingpurposeson132kV busbarsand, in
onecase, thelackof bondingbetweena66kV barandthemetal capof thesupporting
insulator. Thisfaultproducedadisturbanceof greater than10,000pC, preventingthe
completionof anyuseful PDmeasurementsinthesubstation. Experiencehasshown
thata6mm-diameterrodmountedona132kVbushing(76kVtoground)will produce
aPDof 10002000pC, dependingontheweather conditions.
Routineperiodic monitoring (DDF and IR) of circuit-breaker insulation would
probablyincludethebushingsandanyassociatedbusbarsasappropriate. Undersome
conditionsit maybepossibletocarryout onlinePDmeasurementsutilisingahigh-
voltagereferencecapacitor (seeChapter 9) or high-frequency probes(capacitiveor
current transformer Chapter 8). By applicationof aseparatesourcesupply anda
differential circuit measurement, sensitivities of 10pC or less have been reported
duringsitetestsinhigh-voltageswitchyards.
4.7 Gas-insulated systems (GI S)
Inordertoovercomethespacelimitations, andsomeenvironmental problemscreated
by conventional open-typesubstations, gas-insulatedsystems havebeendeveloped
inwhichthecircuit breakers, other switchgear, disconnectors(isolators), instrument
transformers, surgearrestersandbusbarsareall enclosedinmetal earthedchambers.
Epoxy insulators areused for supporting thestructurein thecompressed SF
6
gas.
Thesystemsaredesignedtokeepthegasdryandoxygen-free, thusminimisinglong-
terminsulationdeterioration. Sincenomaintenanceisrequiredexceptforinspections
after aspecifiednumber of switchingoperations, GISarenormallyconsideredtobe
maintenance-free. Completesystemsareavailablefor voltagesuptoat least 765kV.
Anearlier 132kV designwas installedinSydney, Australia, inthe1970s for con-
nectingtheoutputsof 330/132kV transformerstothelocal distributionsystem. The
conventional transformersweresuppliedfromoutsidethecitythrough330kV single-
phasecables. Muchinformationis availableregardingthedesignandperformance
of GISsubstations[18].
Special diagnosticmeasurement techniquesfor GIShavebeendevelopedover a
longperiodof timebutarenotyetstandardized(seeChapters6and9). Themajority
of themethodsareappliedonlineandincludetheelectrical detectionof low-level PDs
of lessthan10pCatveryhighfrequenciesandbyultrasonicsystems. Lowlevelsare
alsorequiredat overvoltagesduringcommissioningfollowingassemblyof thecom-
plexstructures. Becauseof therelativelylowcapacitances, separate-sourcePDtests
arepractical. By-productsdetectedduringanalysisof SF
6
gassamplesmayrepresent
afaultaswell asindicatingthepossibilityof damagetotheinsulationandothermate-
rialsinthechambers. Themajor causesof deterioration, or flashover, areassociated
with thepresenceof small metal particles, loosebolts, bad contacts, faults devel-
opingintheepoxy spacersduetodecompositionof thegas, andperhapsvoidsthat
wereundetectableduring commissioning. Inappropriateelectric field distributions
I nsul ati on defects i n power -system equi pment: Par t 1 87
alongspacersmayoccasionallyproducesurface-creepproblems. Detailedcomputer
programs havebeendevelopedby manufacturers to achieveoptimumelectric field
conditionsaroundtheelectrodesandalonginsulationsurfaces.
4.8 High-voltage cables
Muchinformationis availabledescribingthemany designs andinsulationsystems
of high-voltagecablesandaccessories. Electrically, mechanicallyandthermally, the
configurations aresimpleconcentric cylinders, except for theolder, lower-voltage,
three-phasebeltedcables, wherenoouter screenwasprovidedfor eachcore. When
practical, thethreescreenedcoresof athree-phasesystemarecontainedinacommon
cablebut, at thehigher ratings andvoltages, it is necessary tomanufactureandlay
themseparately.
Theelectricfieldinthebodyof thecablefollowsthat givenbyEquation3.3for
onedielectric andthat at theendby thetypeof distributioninFigure3.3(a), where
veryhighstressingoccursattheedgeof theouter sheath. Inthisregionstresscontrol
iseffectedbyuseof oneof themanyterminatingmethodsavailable, thecomponents
of whicharecritical inthedesignandassemblyof acablesystem.
Themanytypesof high-voltagecablerequirehighlyspecializedproductionand
installationtechniques. At thepresent time, inrespect of insulation, therearethree
major classesof cableinservice.
4.8.1 Oi l paper cabl es
Oilpaper cablesmaybeclassifiedaccordingtothefollowing:
(a) Older oilpaper solid systemsfilledwithoil/rosin operatingupto33kV and
possibly66kV;
(b) Oilpapersystemsimpregnatedwithlow-viscosityoil athighvacuumsandsealed
inserviceunderpositivepressure installeduptothehighestvoltagesandratings
(hollowconductors areutilizedfor coolinginsingle-phasecables andchannels
or apipe-typeenclosureinthree-phasedesigns);
(c) Oilpaper systembut filledwithdrynitrogenunder pressure;
(d) Oilpolypropylenepaper (PPLP) systems for very high-voltageapplications
thisisarelativelynewtechnology[19].
4.8.1.1 Faults in the bodies of oilpaper cables
(a) Sheathdamage accidental or duetochemical erosion allowingtheingressof
moisture, andinpressurecablesthelossof oil, withsubsequent degradationof
thedielectric. Erosionproblemswouldappear likelyindistributioncablessome
of whicharemorethan40yearsold.
(b) Thermal runaway causedby overloadingbeyondthespecifieddesignlimitsor
the lack of adequate heat dissipation around the cable. The latter is a major
factor in the installation of underground cables. The failures in Auckland in
88 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
1998[20]werebelievedtobeduetoacombinationof thesefactors unavoidable
overloading in order to maintain supply and insufficient thermal dissipation.
Temperaturemonitoringsystemsarenowavailablefor installationat important
locations(seeChapter 8).
(c) Partial dischargesleadingtotrackingandwaxingasdescribedinSection2.1.3.
4.8.1.2 Faults in the oilpaper cable accessor ies
Faults in the cable accessories are more likely than in the main cable due to the
requiredmanual additionof insulationonsiteandbecauseof thecomplexinsulation
structuresinsuchregions.
Sealing ends(ter minations): Atthemediumandhighervoltagesthesheathedge
stressandlongitudinal fieldsinoilpaper sealingendsarecontrolledbystressreliev-
ingconesasindicatedinFigure4.9. Suchconesmay beformedby paper tapingon
siteor bepreformedwithinasynthetic resinmouldingfedover theendof thecon-
ductor. Incorrectstressdistribution, moistureingress, voidformationpossiblydueto
oil leakage, or overheatingdueto combinationof loadandexcessiveambient tem-
peratures asproducedbydirectsunlight mighteachcreateaninsulationproblem.
Inonecaseabadcontact problemat thetopof a132kV sealingendwasassociated
withthecap. Theintermittentpartial dischargeswereof sufficientmagnitudetomask
PDmeasurementsbeingmadeonthecabletermination. Thefault wasdetectedand
locatedbyuseof ahandheldremoteultrasonicdetector.
Special careinthestress distributiondesignis necessary, especially if different
insulating materials areutilized. In thenon-pressurized systems theoil may drain
downwardsover alongperiodof timeproducingoil-dryregionsthatcoulddischarge
withresultant punctureof thepaper or flashover alongthesurface. Suchfaults can
occur in the cable boxes or similar of 11/22kV equipment, many of which were
Cable screen
(earthed)
Insulator
Rod gap
HV conductor
Insulation
(tapered)
Stress distributor
Fi gure 4.9 Pr i nci pl e of medi um vol tage oi l paper seal i ng end
I nsul ati on defects i n power -system equi pment: Par t 1 89
filled with bitumastic material or oil. Thelatter systemhas now been replaced in
many cases by theapplication of plastic heat-shrink coverings which may include
someelectric stressrelief without theneedfor abulk insulationfiller. Thesenewer
techniquesrequirecareinorder toavoidthetrappingof air at theinterface.
PDsinsuchregionsmaybedetectablebytheuseof localizedultrasonicdetection
techniques(seeChapter9). Exceptforthelargerdischargesthisisnotpossiblethrough
ametal sheath.
Joints: Theformationof jointsinoilpaper cablesisaskilledandlengthy pro-
cedure. Inprincipletheprocess consists of scarfingor steppingof paper wraps on
thesections to bejoined and, after connection of theconductors, filling theinter-
space with layers of pre-dried, oil-impregnated paper tapes and wraps. For the
higher-voltagecables, allowanceismadefor accommodationof ahigh-voltagestress
distributor containedinsyntheticresinandconnectedtotheconductor. Thepaper is
builtuptosupportastress-relievingconeatthesheathedge. Thisworkiscarriedout
withthesystemexposedto air, thereby introducingthepossibility of voidcreation
and moistureingress. Special precautions and speed of application areessential
for example, under temporarycover andusingdryair duringaperiodof perhaps12
hoursisrequiredfor alargehigh-voltagejoint.
Theconfigurationfor anolder designof high-voltagejoint withaninternal oil
ductisshowninprincipleinFigure4.10. Alatersystemwithcone-shapedductsforoil
input failedinserviceduetopartial dischargesprobablycausedbytheaccumulation
of small metal particles fromthe conductor wires which then generated gas.
Failureoccurredinthethinoil duct betweentheresincastingcontainingthestress
distributorandthepapertape[21]. Otherdesignsarebasedonsimilarprinciples[19].
Theprincipleof atransitionjointdesignbetweenafed oil- (or gas-) filledpaper
cableandanextrudedpolymericcableisdescribedinSection1.2.4of Reference24.
At thelower voltagessuchajoint may haveanoilpaper cablewithPD magnitude
levels of 200300pC operating satisfactorily, but thevalues in theextruded cable
wouldbeexpectedtobemuchless perhapsinthetensof pCs. Thismeasurement
problem, andalsohowitrelatestolocationerrors, ishighlightedinCIGRE Brochure
297[22].
Lapped cable
paper (O.I.P.)
Cable core (H.V.)
Duct in core Creep stress relieved
O.I.P. coils
H.V.
Resin
Conductor
joint well
screened
Stepped creep paths along oil duct from H.V. to earth
Oil
Screen (EARTH)
Fi gure 4.10 Pr i nci pl e of an ear l i er hi gh-vol tage j oi nt wi th oi l feed
90 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Whereasolidinsulatingmaterial of different conductivitytothepaper isusedin
suchjoints andinother high-voltagecomponents thedesignmust allowfor the
changedfielddistributionif direct-voltagetestsaretobeappliedinthefactory and
duringcommissioningof theAC cable. Previousdamagewassuspectedduetosuch
atest followingthefailureof a132kV cablejoint inservicesomeyearsago.
4.8.2 Extr uded cabl es
Duringthepastfiftyyearsormore, extensiveresearchintothedevelopmentof plastic
materialshasresultedintheirsuccessful applicationasinsulationforcables initially
for low-voltagewiring(e.g. PVC coverings) throughto modernsystems of XLPE,
someof whichareoperatingat 500kV (AC). Test requirementsfor thevariousvolt-
agelevelsaregivenintheStandards[S4/7S4/9]. Thewidestapplicationof extruded
cables has beenfor mediumtohighvoltages inthedistributionfield: rangingfrom
11kV through to 33kV/66kV and afew systems at 132kV and 500kV. Thepro-
duction techniques havevaried but thecommonest, based on much experience, is
thetriple-extrusionmethod. Theconstructionof atypical singlecore22kV cableis
depictedinFigure4.11.
During theprocess an inner semiconductivelayer, themajor insulation and an
outer semiconductivelayer areextrudedsimultaneously, thus producing, ideally, a
goodconcentricfield. Inearlier methodstheouter layer wasaddedseparately, which
sometimes created voids at theinterfaceand subsequent failuredueto partial dis-
charges. XLPEinparticularisverysusceptibletolow-level PDs. Theusual insulation
isXLPE, althoughEPRisfavouredbysomemanufacturersfor specificapplications.
Apart fromavoidingtheformationof voidsduringthemanufacturingprocess, a
major problemwithXLPE has beenthetendency for water trees to developinthe
material. Theeffect has beenminimizedby replacingtheearlier pressurizedsteam
cross-linkingtechniques with, for example, thesiloxaneprocess or thecontinuous-
gas-vulcanizationdry-curingmethodinpressurizednitrogenand, whenconsidered
necessary in service, the use of a water barrier on the outside of the cable. It is
notedthat EPRandtree-retardant cross-linkedpolyethylene(TRXLPE) appear tobe
thepreferredmaterialsfor medium-voltage(1535kV) cablesintheUSA [23]. The
applicationdoesnotrequiretheuseof water barriers. However, for 69kV andabove,
barriersareoftenspecified.
Water swellable tape
XLPE insulation
Semi-conductor screen
HV conductor
Outer semi-conductor
screen
Copper wires
Lead sheath
Serving
Fi gure 4.11 Constr ucti on of a typi cal XLPE 22 kV (RMS) si ngl e core cabl e
I nsul ati on defects i n power -system equi pment: Par t 1 91
XLPE and EPR cables are now very competitive with oilpaper types at the
medium-voltagerangefor anumber of reasonsincludingthepossibilityof operating
at somewhat higher temperatures. At thehigher voltages theapplication of plastic
cabledesignsisincreasing.
Althoughthemanufacturingandacceptancetestsfor plasticcablesareverystrin-
gent, anumber of potential faultsarepossibleinservice. Theseincludethefollowing
inthebulkof thecable.
(a) The development of carbonized trees in the main insulation due to voids or
particlesintroducedbyincorrectmanufacturing. Theinitial faultsmaynothave
beendetectableduringthefactoryPDmeasurements.
(b) Localizedseparationbetweenthemaininsulationandsemiconductor outer lay-
ersresultinginPDs, possiblyformedafter manufactureduetolack of bonding
andaccentuatedbybendingof thecable.
(c) Thecreationof water treesinXLPE formedduringmanufacture but nowless
likely andinservicefollowingingressof moisture. Failuresintheoldercables
wererelatively frequent dueto this cause[23]. Thetrees cannot bedetected
easily in the final product and become apparent only when almost bridging
the dielectric. Advanced detection techniques are being developed including
dielectricresponsemeasurements(seeChapter 6).
4.8.2.1 Ter minations and j oints for extr uded cables
Manyformsof terminationsandjointshavebeendevelopedfor extrudedpolymeric
cables. Someof thesearedescribedinprincipleinthereportbyWG21.06of CIGRE
[24]. Thisincludesoutlinesof techniquesfortransitionjointsforconnectingoil-filled
andpolymericcables.
Oneof thespecial problemsassociatedwithpolymericterminationsandjointsis
thepossiblecreationof faultsattheinterfacebetweenthecableinsulationandtheelec-
trical stress-relievingmaterial or semiconductor coneconnectedtotheearthscreen.
Theobjectiveistoimprovethelongitudinal (axial) stressalongtheinterfaceandthe
cableinsulation. TheexamplesinFigure4.12indicatehowthismay beeffectedby
applicationof (a) aheat-shrink stresscontrol sleeve, or (b) asemiconductingstress
cone, or (c) asemiconductingconewithaconical thermosetcastingandcompression
springs. At thehigher voltages thelongitudinal stress canbeimprovedby winding
layersof conductingfilminterleavedwithpolymericfilmaroundthecableinsulation
toformacapacitor-typebushingconstruction[24].
Thenecessityfor verygoodphysical contact betweeninterfacesurfacesisanal-
ysedinCIGRE Brochure210[25]. Amongthefindings, it is concludedthat (i) the
surfacesmust besufficiently smoothtoreducevoidsizestoamicrometreor soand
(ii) sufficient pressurebeappliedby for example heat shrinking(Figure4.12a),
theelasticityof therubber material or springpressureforcingarubber insert against
anepoxyconical casting(Figure4.12c). Themechanical andthermal characteristics
must besuchthat nosignificant gapsor voidsareformedwheninstallingonsiteor
duetomovement inservice. Althoughthemaximumelectrical stresswill beradial,
it isthelongitudinal component that maycausefailure. Thisisespeciallythecaseif
92 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Conductor stalk
Corona
shield
Insulator or
cast resin
Liquid or gas
Conical thermoset
casting
Insulating elastomer
Semiconducting
elastomer
-both in a moulded
stress cone
Compression
device
Fluid or
gas seal
Liquid or
gas level
Sheds
Insulating
liquid or
gas
Stress cone
(taped or
moulded
elastomer)
Liquid or gas
seal
Insulator
Polymeric
extruded cable
Plumb
Waterproof seal
Heat shrink sheds
Heat shrink sleeve
Heat shrink stress
control sleeve
Polymeric extruded
cable
Waterproof seal
(a) (b) (c)
Fi gure 4.12 Possi bl e outdoor ter mi nati ons for pol ymer i c extr uded cabl es [ 24]
[ reproduced by per mi ssi on of CI GRE]
voidshavebeenformedandtheradial stressvalueissufficientlyhightoinducepartial
discharges. If themagnitudeof thelatter isof theorder of onlyafewpicocoulombs
this might leadto failurewithinafewhours insomejoint structures [25]. Outdoor
terminations may havelower stresses andsimilar (or higher) PD values but failure
wouldbeexpectedtooccur inthelonger term. Inorder tominimizepossibleprob-
lems, emphasisisgiventocareful preparationof thesurfaces, theprecisefollowing
of theaccessorymanufacturersinstructionsandtheemployment of skilledjointers.
Inolder designsafailureduetoarcingcanoccur if thesemiconductor elastomer
of aconeisincorrectlyassembledover theearthedscreenandisnot makingcontact.
Such acondition has resulted in thesemiconductor rising in potential aboveearth
withsubsequent dischargetothescreenandfailure.
Theprinciples discussedabovefor terminations areapplicabletothenumerous
formsof joints[24, 25]. Inmanyrespectstheinstallationandthelong-termcondition
of thejointsaremorecritical becauseof their complexityanddifficultaccess. A par-
ticular designof joint is showninFigure4.13[24]. InReference26theseparation
or relaxationof thestress-control materials inservicewas foundtobeasignificant
causeof failure.
Muchattentionisbeinggiventothedevelopmentof techniquesfor thecondition
monitoringof extrudedcablesand, inparticular, their accessories(Chapter 9).
4.8.2.2 Gas-insulated tr ansmission lines (GI L )
Gas-insulatedtransmissionlineshavebeendevelopedovermanyyearsforinstallation
inlocations whereaccess, for example, toacity centreby ahigh-voltagesupply is
required. Thelinescomprisearigidmetal outertubewithaninnerconductormounted
onspeciallydesignedepoxyresininsulators. Thelinescanbeburied, aboveground
I nsul ati on defects i n power -system equi pment: Par t 1 93
Stress control profile screen
Insulation screen
Insulation
HV electrode
Elastomeric moulding
(one piece)
Ferrule
Joint shell Plumb
Fi gure 4.13 Exampl e of dr y-type pre-moul ded j oi nt [ 24] [ reproduced by per mi ssi on
of CI GRE]
or mounted in existing/new tunnels. An overview is presented in El ectr a [27], in
whichsystems of upto 550kV and3kminlengtharedescribed. Thegas mixture
usedisquotedas80per centN
2
, SF
6
20per centatapressureof 7bar, thusreducing
thequantityof SF
6
required.
Inanearlier 275kV systemof 3.3kmlengththediameter of theouter aluminium
tube is 420mmand that of the high-voltage conductor tube 170mmwith an SF
6
gas pressure of 0.44MPa. These systems would be expected to require minimal
maintenance. The commissioning checks include partial-discharge measurements,
probablyutilizingUHF technology[27] (Chapter 9).
4.9 Summar y
Thechapter reviews howthevarious insulatingmaterials areappliedinarangeof
power-systemequipment excluding rotating machines and transformers. Although
thenumber of examplesof possiblefailuresislimited, thosechosenindicatewhere
problems might arise. Knowledgeand understanding of thebehaviour of newand
agedinsulationstructures is essential whenconsideringtheneedor otherwisefor a
particularcondition-monitoringsystem. Thevaluesof thepartial dischargesexpected
for thedifferentoperatingconditionsgiveanindicationof thesensitivityrequiredfor
theconditionmonitoringsystems.
4.10 Refer ences
1. InsulationNewsandMarket Report, 1998;6(1):54
2. Williams, D.L., Haddad, A., Rowlands, A.R., Young, H.M., andWaters R.T.,
Formation of dry bands in clean fog on polluted insulators, I EEE Tr ans. on
Di el ectr i cs and El ectr i cal I nsul ati on, October 1999;6(5):72451
94 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3. Yoshida, S., and Naito, K., Survey of electrical and mechanical failures of
insulators caused by ice and/or snow, CIGRE WG B2.03, El ectr a, October
2005;(222):2226
4. EPRI, Electrical andmechanical propertiesof ceramic insulators, Report No.
1000505, December 2000
5. Kumosa, M., Kumosa, L., andArmentrout, D., Failureanalysesof non-ceramic
insulators: Part 1 Brittle fracture characteristics, I EEE El ectr i cal I nsul a-
ti on Magazi ne, 21(3):1427; Part II The brittle fracture model and failure
prevention, I EEE El ectr i cal I nsul ati on Magazi ne, 2005;21(4):2840
6. Cherney, E.A., and Gorur, R.S., RTV silicone rubber coatings for outdoor
insulators, I EEE Tr ans. on Di el ectr i cs and El ectr i cal I nsul ati on, October
1999;6(5):60511
7. Zaengl, W.S., and Kuffel, E., Hi gh Vol tage Engi neer i ng Fundamental s,
(PermagonPress, Oxford, 1984), Chapter 4.
8. Allan, D.M., J ames, R.E., McMullan, P.J ., et al ., AustralianandNewZealand
experiencein theapplication of diagnostic techniques for assessment of insu-
lation condition in high voltage apparatus, CIGRE Paper 15/33-10, Session
1990
9. Gilbert, R., Gillespie, A., Eklund, A., Ericsson, J ., Hartings, R., J acobson, B.,
Pollutionflashoversonwall bushingsat acoastal site, CIGRE Paper 15-102,
1998
10. Bossi, A., andYakov, S., Bushingsandconnectionsfor largepower transform-
ers, CIGRE Paper 12-15, 1984
11. Cease, T.W., andJ ohnston, P., Amagneto-opticcurrenttransducer, I EEE Tr ans.
on Power Del i ver y, 1994;9 (4):177888
12. Dubanton, C., CIGRE WG 23.07, Instrument transformer technology and
servicebehaviour designandconstruction, 1988/89
13. J ames R.E., BlackburnT.R., CottonM., et al ., Challenges for advanceddiag-
nostictechniques Faultsundetectablebyexistingelectrical methods, CIGRE
Paper 15-306, Session2000
14. Yakuymyshyn, C.P., Brubakar, M.A., J ohnston, P.M., and Reinbold, C.,
Manufacturingchallengesof optical currentandvoltagesensorsforutilityappli-
cations, Proceedi ngs of SPI E, 3201:219, Sensors andControls for Advanced
Manufacturing01/1998
15. Andoh, H., Nishiwaki, S., Suzuki, H., Boggs, S., andKuang, J ., Failuremech-
anisms and recent improvements in ZnO arrester elements, I EEE El ectr i cal
I nsul ati on Magazi ne, J anuary/February2000;6 (1):2531
16. Darveniza, M., Roby, D., andTumma, L.R., Laboratory andanalytical studies
of theeffects of multipulselightning currents on metal oxidearresters, I EEE
Tr ans. Power Del i ver y, April 1994;9:76471
17. Balcombe, H.G., The assessment and management of older oil-filled
switchgear, Power Engi neer i ng Jour nal , J uly1977;11 (6):2348
18. Ryan, H.M. (ed.), Hi gh Vol tage Engi neer i ng and Testi ng, 2ndedn(IET, London,
2001)
I nsul ati on defects i n power -system equi pment: Par t 1 95
19. Tsurga, Masato, Numato, Naohisa, and Nagano, Hideaki, 400kV
Polypropylene-insulated oil-filled cable for power grid, Singapore, Hi tachi
Cabl e Revi ew, October 1999;18:912
20. Kent, H., andBucea, G., InquiryintoAucklandpower supplyfailure. Technical
Report onCableFailures, I ntegr al Energy May 1998[also online]. Available
fromwww.med.govt.NZ/upload/23747/failtech.pdf [accessed20May2007]
21. Gibbons, J .A.M., Saunders, B.L., andStannett, A.W., Roleof metal debrisinthe
failureof stop-jointsasusedin275kV and400kV self-containedoil-filledcable
circuits, I EE Proc. C, Gener. Tr ansm. Di str i b., November 1980;127(6):40619
22. Practical aspects of thedetection and location of partial discharges in power
cables, El ectr a, J une2006;(226):6369, TF D1.02.05(CIGRE Brochure297)
23. Dundas, J .H., and Cochran, W.H., Technical advances in the underground
medium-voltage cable specifications of the largest investor-owned utilities
in the U.S., I EEE El ectr i cal I nsul ati on Magazi ne, November/December
1999;15(6):2937
24. Accessories for HV cables with extruded insulation, El ectr a, February
2001;194:313(CIGRE Brochure177, Section2, Part 1)
25. Interfacesinaccessoriesfor extrudedHV andEHV cables, J TF 21/15, El ectr a,
August 2003;203:539(CIGRE Brochure210)
26. Serviceagedinsulation. Guidelinesonmanagingtheageingprocess (D. Allan
and T.R. Blackburn, WG D1.11), El ectr a, J une 2003;208:6972 (CIGRE
Brochure228)
27. Koch, H., and Schoeffner, G., Gas insulated transmission line (GIL). An
overview, El ectr a, December 2003;211:817(CIGRE Brochure218)
4.11 Standar ds r elated to Chapter 4
S4/1IEC 60137(1995): Bushingsfor alternatingvoltagesabove1000(AS1265)
S4/2IEC 60044-1(2003): Instrument transformers Part 1: Current transformers,
AnnexB (Informative)
S4/3 IEC 60044 (2004): Instrument transformers Part 5: Capacitive voltage
transformers(AS60044.5, 2004)
S4/4 IEC 60871-1 (2005): Shunt capacitors for AC power systems having arated
voltageabove1000V (AS2897)
S4/5IEC 60071-2& 5: Insulationco-ordination(AS1824)
S4/6 IEC 60099-1: Surge arresters Part 1: Non-linear resistor type gapped surge
arrestersfor ACsystems(1999). Part4: Metal-oxidesurgearresterswithoutgapsfor
AC systems(2004)
S4/7IEC60502-2(2005): Powercableswithextrudedinsulationandtheiraccessories
for rated voltages from1kV (Um= 1.2kV) up to 30kV (Um= 36kV). Part 2:
Cablesfor ratedvoltagesfrom6kV (Um= 7.2kV) upto30kV (Um= 36kV). See
alsoAS/NZS1429.1:2000
96 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
S4/8IEC 60840(2004): Power cableswithextrudedinsulationandtheir accessories
for ratedvoltagesabove30kV (Um= 36kV) upto150kV (Um= 170kV) Test
methodsandrequirements(AS1429.2:1998)
S4/9IEC 62067(2001): Power cableswithextrudedinsulationandtheir accessories
forratedvoltagesabove150kV (Um= 170kV) upto500kV (Um= 550kV) Test
methodsandrequirements
4.12 Pr oblems
1. Capacitivegradingis to beincludedinanoil impregnatedbushing(
r
= 3.5).
Thedetailsareasfollows:
Lengthof foil adjacent toHV conductor = 400mm(l
1
).
Theradial spacingsbetweenconductor/foils/earthedflangeare6mm.
The conductor radius (r
o
) = 20mm. The earthed flange inner radius (r
4
) =
44mm.
Themaximumradial stressesintheOIP of eachlayer aretobeequal.
Determine
(i) thefoil lengths(l
1
. . . . . .l
4
)
(ii) thecapacitancevalues betweenfoils (C
1
. . . . . .C
4
) andthat betweenHV
andearth(C)
(iii) themaximumradial stresses
(iv) themaximumradial stresswithout foils
Thetest voltageis60kV (RMS)
AssumeQ= CV = C
1
(V
0
V
1
) = C
2
(V
1
V
2
) = . . . . . . = C
4
(V
3
V
4
).
2. What arethelimitationswhenmeasuringthepartial-dischargecharacteristicsof
an AC extruded power cable on site (a) using a separate-source high-voltage
supplyand(b) under operatingconditions?Assumeaterminationandajointare
inthesectionunder test. Refer toChapters6, 7and9.
3. Anextrudedcableisterminatedwithasemiconductingstresscone. Byassuming
a radial operating stress at the cone and cable insulation of 2kV/mm(RMS)
calculatethethicknessof avoid/air gapwithintheinterfacethatmightdischarge
under this condition. Also, determine the maximumgap thickness allowable
to avoidpartial discharges duringatest voltageapplicationof 1.5U
o
. Refer to
Chapter 3 (Section 3.2.1) and Chapter 2 (Figure 2.2). Assume the insulation
thicknesstheair gapandthatparallel fieldconditionsareapplicable
r
= 2.2.
Chapter 5
I nsulation defects in power-system equipment:
Par t 2
Low- andhigh-voltagemotors
Largegenerators
Power, distributionandmagneticvoltagetransformers
Thischapter considerstheconstructionof rotatingmachinesandtransformersin
sufficientdetail toidentifyanumber of possiblefailuremodes. Several examplesare
included.
5.1 Electr ical r otating machines
In thesupply systems, awiderangeof rotating machines is utilized in addition to
thepower generators. Thisisespeciallythecaseinthepower stationswheremotors
fromratingsof lessthanakWat 415V throughtothosefor blower andpumpdrives
of uptotheorder of 10MWor moreat 11kV might beinstalled.
5.1.1 Low-vol tage motor s
With thewider application of power electronics for adjustablespeed drives, there
seems to beaneed for morecareful design of theinsulation systems even for the
lower-voltagemotors. Thepresenceof partial discharges has beendetectedinran-
domwound stators operating at 440V supplied fromPWM converter circuits [1].
In this areaspecialized tests arerequired to provetheturn-to-turn strength. A dif-
ferential 0.2srise-timesurgetest [2] asdefinedfor formedwoundmachinesmay
beapplied for assessing theinter-turn wirecovering. However, this does not nec-
essarily provetheintegrity of awindingsubjectedto less steeppulses but of much
higher repetitionrate. Thecomplexityof theconditionsasimposedontheinsulation
areconsideredinReference3. Followingfailuresof motorssuppliedbyPWM con-
trol systems, investigationsweremadeof theeffectsof voltage/frequencyvariations,
98 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
pulserisetimes, pulsewidth/duty cycles and overshoot/resonancevoltages on the
electrical insulation. It ispossiblefor partial dischargestodevelopduringthepulses
under adverseconditions. Theincorporationof sine-waveoutputfiltersmayresultin
improvements but their sizeandcost canbeprohibitive[4]. Thework of IEC TC2
WG27inpreparingthenewIEC SpecificationIEC 60034-18-41[5] ispresentedby
Wheeler inReference4.
5.1.2 Hi gh-vol tage machi nes
Inprinciple, theelectrical insulationsystems of thestator windings of themajority
of high-voltagemachinesmaybeconsideredtobesimilar [6]. Insulatedconductors,
usuallyof rectangular section, areassembledintheslotsof thelaminatedcylindrical
core aprocessdescribedaswi ndi ng. AsdepictedinFigure5.1(a), theconnections
betweentheactiveparts (straight sections) of theinsulatedconductors aremadein
theoverhang region, enablingtheformationof series/parallel configurations with
theappropriatecoil pitch(spread) toproducethedesiredthree-phaseoutput voltage.
Theoverhangissupportedbyvariousinsulationstructures, possiblyinvolvinglarge-
diameter insulationrings, cords(strings) andadditional insulationtaping.
5.1.2.1 M otor s
For motorsintherangeof 3kV to11kV anduptotensof MWsratings, itispossible
topreformandinsulatethetwo-sideddiamond-shapedcoilscontainingmultiplerect-
angular turnswiththeappropriatepitchandoverhangcontour suitablefor assembly
Stator core
(a)
(b)
(c)
End winding
overhang
Stator cooling
water manifold
Slot wedge
Core
plate
Conductor
Conductor
Insulation
Slot
Winding
water
boxes
PTFE cooling tubes
Fi gure 5.1 Layout of coi l s/bar s and i nsul ati on i n AC stator wi ndi ng. (a) Core and
end wi ndi ngs (over hang); (b) Tur ns/bar s i n sl ot; (c) Large water -cool ed
tur bi ne gener ator stator
I nsul ati on defects i n power -system equi pment: Par t 2 99
(winding) in thecoreslots. As an example, theconductor dimensions for athree-
phase, 11kV, 3MW motor might beof theorder of 5 3mmcross-section, 4 in
parallel with5turns/coil and30coilsper phasehavinganaxial corelengthof about
1metrewithoverhangsof lessthan0.5metre. Thecoreinner diameter might beof
theorder of 0.75metre.
5.1.2.2 Gener ator s
The large low-speed (a few hundred rpm), air-cooled, salient-pole, vertical-shaft
hydro-generators may haveratings upto 500MVA at 15kV andthesteam-turbine
generators up to 1600MVA at, possibly, 27kV and 1800rpm(60Hz). Thelatter
typesformthemajor power sourcesfor thesupplysystemsof theutilitiesandmore
usually have outputs of 500660MVA operating at 3000rpm(50Hz) and 22kV.
Eachstator bar (twoper slot) is manufacturedseparately withthecorrect overhang
configuration and, after assembly in thecore, is connected to its associated turn
probably by meansof aflexiblelink. Withinternally cooledgeneratorsthedistilled
water issuppliedthroughplastic tubes typically of PTFE feedingwater boxes
at theendsof thebars(Figures5.1(b) and(c)).
Theconductor of anolder 500MVA, 22kV steamturbine-generator might have
an overall cross-section of 66 26mmcomprising many strands in parallel each
insulatedfromtheotherstoreduceeddycurrentlossesandhollowfor water cooling.
Thebar lengths might beof theorder of 65 metrewithin thecoreand 08 metre
for eachoverhang. Twoparallel paths per phasearerequiredwith16bars inseries
in each, depending on thenumber of slots. Full linevoltageappears between the
lineendbars(conductors) whichmaybephysicallyadjacent intheoverhang. Phase
voltagetogroundispresent fromthelineendbarstothecorewithintheslots. The
wholesystem core, rotor andwindings operatesindryhydrogenat apressureof
about6bar, thegasactingasacoolantandinsulant. SeeFigure2.2forthebreakdown
characteristicsof hydrogen.
5.1.2.3 M achine insulation systems
Intheolder machines, manyof whicharestill inservice, theturninsulationconsisted
of thermoplastic systems. Following many years of research and development,
micapaper (incorporatingsmall micasplittings), synthetic thermosettingresinsand
higher-temperaturebackingtapes wereintroducedinthe1960s andcontinueto be
theconfigurationsnowusedbutwithimprovedtechniquesandmaterial formulations.
Thethreebasicmethodsarebrieflyoutlinedinthefollowing.
(a) Shel l ac/bi tumen bi ndi ng var ni shes
Earlier systems comprised mica splittings, cellulosic paper and binders of shel-
lac/bitumenvarnish. After wrappingof theinsulatedconductorsandformingof the
coil contour two-sided for smaller machines and half-coil/bars for largegenera-
tors thewall insulationwasadded, thestraight sectionscompressedtothecorrect
thicknessfor fittinginthecoreslots, theoverhangtapingaddedandthewholecon-
solidated at an appropriatetemperature. If not continuously hand-taped, ascarfed
100 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
joint was includedbetweenthestraight sectionprojectingbeyondtheslots andthe
evolutedsections.
(b) Resi n-r i ch systems
Theconductors in multi-turn coils areinsulated with enamel and glass or amica-
based tape. After formation into thecorrect shapeand overall dimensions, ground
wall insulation consisting of film- or glass-backed epoxy-resin-bonded micaceous
tapeis applied. Thetapes aresuppliedwiththeresinpartially curedto aB-stage
conditionandcontainsufficient excessresintoeffect adequateconsolidationduring
thesubsequentheatandpressuretreatmentnecessarytoproduceavoid-freeinsulation
configuration. Handormachinewrappingmaybeuseddependingonthemanufacturer
andequipment size.
Theend-windinginsulationmayconsistof thesametypeof tape, thereforeneces-
sitating special heat treatment or, alternatively, afully cured tapethat will enable
manipulationof thecoilsduringwindingwithoutdamage. If ascarf jointisincluded,
additional resinmayberequiredalongthelengthof thejointwithsubsequentlocalized
curing.
(c) Vacuum-pressure-i mpregnati on (VPI ) process
Inthis process thecoils andbars areinitially insulatedwithdry micaceous tapes
containingaminimumamount of resin. Thestructure coil or bar isfittedintoa
mouldwithinatank, subjectedtoahighvacuumtoextractair andmoistureandthen
flooded (impregnated) with low-viscosity resin under apositiveair pressure. This
procedure(VPI) shouldresultinavoid-freeinsulationsystem, whichissubsequently
cured to thermoset the resin. The inclusion of some flexible tapes in the evolute
sectionsmakesit possibletowindthecoilsor barsintothestator slotsandformthe
correct overhangconfiguration. Thelatter is maintainedby useof cords andtapes
and, for thelargegenerators(singlebars), theinclusionof asupportingstructureof
treatedwoodor fibreglass.
Inthecaseof medium-sized, high-voltagemachines, this wholeprocess canbe
carriedout withthecoilsalreadywoundintothestator andtheoverhangcompleted.
Plantsarenowavailabletoaccommodatemachinesuptoratingsof 200MVA [7].
As required, insulation systems operating with Class 130

C and Class 155

C
havebeendeveloped. Moredetailsof windinginsulationanditshistorical develop-
ment aregiveninReference7.
5.1.3 Possi bl e i nsul ati on fai l ure mechani sms i n rotati ng machi nes
Insulationfailuresinhigh-voltagemachinesarepredominantlyassociatedwithcom-
binationsof thermal andelectrical stressesfor themediumsizesandmechanical and
electrical stressesfor thelarger units. Temperaturesrangefrom105

C for theolder
machinesuptoClass155for thenewer insulationsandaverageelectrical operating
stresses areintherange14kV/mm, dependingonageandtypes of material. It is
possibleto identify, ingeneral terms, theregions whereinsulationproblems might
arise, dependingonthesize, insulationsystemandageof themachine.
I nsul ati on defects i n power -system equi pment: Par t 2 101
5.1.3.1 M ain wall insulation
A simplifiedarrangement of themainwall insulationisshowninFigure5.2. Voids
(A) maybeformedintheinsulationduetodelaminationinthethermoplasticsystems
or inefficient resinimpregnationinboththeresin-richandpossiblyVPI processes.
Theair-gapthickness betweenthecoil insulationandthestator coremay beof
theorder of 1mm, allowingfor assemblyclearances. Insuchgapspartial discharges
wouldbeproducedfor unscreenedconductor insulationasmay becalculatedusing
theparallel fieldrelationshipsandPaschensLaw(Chapter 3). Toprevent suchPDs
thewall insulationiscoveredwithacontinuoussemiconductinglayerof resistanceof
perhaps1k/square. Thissurfacemaybedescribedasacoronascreenandif earthed
atonelocation(thestator)shouldlimitthevoltagedropalongthelengthtoafewvolts.
If ascreen-freeareaof onlyafewmmdiameter (D) causedbymechanical vibration
isformed, thendischargesmightoccur tothecoreacrossthegasor air gapand, more
significantly, alongthebareinsulationarea(D) [8]. Thesesurfacedischargesat the
edgeof thescreenmaybeof highintensity. Thesituationisparticularlydangerousfor
anisolatedsectionof screenasindicatedatEinFigure5.2. Anadditional problemmay
developif theerodedareasoccur near thecorner of acorecoolingduct (C). Failures
haveoccurredinsuchregionsandmuchinsulationandcoredamagefoundthoughtto
beduetoverylargedischargesorlow-level arcing. Someof theseeffectsarereported
andanalysedbyWilson[8] basedontenyearsof measurementsandinvestigationsat
theNational GridR&DCentre, Leatherhead, UK. It isnoted[Reference2of 8] that
withhydrogen-cooledgeneratorsthephenomenamay not bepresent becauseof the
higher gaspressures, althoughproblemsmight still occur.
B
Laminations
Core cooling ducts Corona screen
(earthed)
E
Wall insulation
Conductor
Bare insulation
Conductor insulation
screen (side)
Isolated
screen
section
E
C
A
D
Void
Overhang
Fi gure 5.2 Si tes of possi bl e par ti al di scharges rel ated to the wal l i nsul ati on of a
stator wi ndi ng
5.1.3.2 I nter-tur n insulation
Inthelarger form-wound(pre-formedcoils) motors theinsulationsystems may be
subjectedtosteepsurgesproducedinunprotectedinputcircuitsincluding, for exam-
ple, high-voltagevacuumcircuitbreakers. Thisisawell-knownproblembutcanstill
arisedueto poorly designedmotor supply circuits, including, for example, signif-
icant lengths of cable. With multipleturns, as shown in Figure5.3, asurgeat the
high-voltageterminal cancausebreakdownover agroupof turns(Figure5.3(a)) or
102 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Critical flashover distances
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fi gure 5.3 Mul ti -tur n motor wi ndi ng. (a) Si ngl e secti on; (b) Doubl e secti on;
(c) Tapi ng i n end-wi ndi ng of fai l ed 11 kV motor
betweensectionsif wound, asinFigure5.3(b), duetothenon-uniformvoltagedis-
tributionacrosstheline-endturns. ThephotographinFigure5.3(c) showspartof the
overhangsectionof aVPI coil theimpregnationof whichwasinsufficient. A failure
occurredinasimilar manner tothat inFigure5.3(a), duetoswitchingof avacuum
circuitbreaker attheendof acable. Protectivedeviceswereaddedafter thisincident.
In largemachines theinteraction of thedistributedcapacitanceandinductance
is important whenconsideringtheanalysis of partial-dischargemeasurements. The
resultsrelatetosuchparametersasthenon-uniformpulsedistribution, thewinding
pulsetransit timeandthecouplingbetweenphases. Thesefactorsalsoinfluencethe
designsof thetest systemsfor monitoringthepartial discharges.
5.1.3.3 End-winding insulation
Theinsulationof thewindingoverhangcanbecomplex. Thisisassociatedwiththe
necessary arrangement andshapeof theconductors together withtheasymmetrical
electromagnetic forces set upby thestator windingcurrents. Inthecaseof motors
many starts per day areoftenrequired. Continuous vibrationcanleadtothespacer
blocksbecomingloose, resultinginfrettingof theconductor insulation. Dust dueto
thisprocessisevidenceof insulationdeterioration. Althoughtheelectrical stressesto
groundaremuchlower thanintheslots, therearelocationsinmany designswhere
line-to-lineconductors areadjacent and insulation-creep surfacestrengths must be
maintained particularly under poor environmental conditions (e.g. moisture and
dust). Failures intheseregions havebeenreported. Thepresenceof creepsurfaces
canrepresentadesignproblemforlargehydrogen-cooledgenerators, whicharetested
inthefactory andat siteinair at the2U
N
level. It must beacceptedthat thenewer
resin-impregnated materials (e.g. glass-fibresystems), as well as thenatural types
such as impregnated laminated wood, will absorb a limited amount of moisture,
whichcouldresult inadegradationof electrical properties.
I nsul ati on defects i n power -system equi pment: Par t 2 103
A one-minutetest inair onahydrogen-cooledgenerator at approximately 2U
N
(AC) resulted in discharges, initiated at the semiconductor edges, flashing across
theendinsulationsurfaces. Thiswasprobablyduetolocalizedchargedareasonthe
high-resistivity insulatingmaterialsused. Theseeffectsarenot present inhydrogen
at theoperatingpressureof 6bar, asthegasbreakdownstressesaremuchhigher for
thiscondition.
To minimize surface PDs at the exit point of the coils/bars (Figure 5.2 B), a
semiconductingsurfaceisaddedonmachinesof about6kV andabove. If deteriorated
or incorrectlyappliedthiscanresult inhigh-energypartial dischargesincludingair-
corona at operating voltages. This was observed for an 11kV pump motor where
excessivecoronawas locatedaroundtheendwindings usingahandheldultrasonic
detector. Thelevelsweretoohightocarryout significant PDtests. Suchastatemay
havebeencausingenvironmental problemswithinandexternal totheunit.
Also in these exit regions it may be necessary to incorporate a scarf joint
(Figure5.2B) forintegratingtheslotwall insulationandtheoverhangexternal tapes.
Inanother typeof 11kV motor semi-curedtapeshadbeenappliedintheformationof
thejoint inorder toassist inwinding. Duringassembly it wassuspectedthat undue
force was applied, resulting in a lack of bonding during the final cure. A failure
occurredat suchalocationinservice. Subsequent testingof theother phases indi-
catedpartial dischargelevels of several thousandpicocoulombs at belowoperating
voltage. Subsequently, failureoccurredat thecorrespondingjoint onincreasingthe
test voltage. Themicapaper wasfoundtobepoorlybonded.
5.1.4 CI GRE summar y of expected machi ne i nsul ati on degr adati on
It isof valuetonotethesummaryincludedinCIGRE General Reportsfor 1998[9].
Thecommonopinionisthat modernepoxy-micabasedinsulationsystemswill not fail for
electrical reasons. What will affect theinsulationis mechanical malfunctionthat may in
timeleadtoelectrical failure. Suchfailuresinclude:
Excessivecoil endvibrations
Looseningof thebarsintheslots
Looseningof theendwindingbracing
Water leakageresultingindelaminationof theinsulationof water cooledbars
Poor manufacturingqualityfor earlier installations(prior to1970)
Failureof thesemi-conductinglayer of thestator bars
Thus, for modernsystems, theinsulationinitself cannotbeexpectedtobearootcause
of future failures. This is in contrast to the experience with many older systems where
electrical problemsareacommoncauseof failure.
Asthefleet of operatinggeneratorsincludesmanyoldaswell asnewmachines, it can
beexpectedthat alot of electrical as well as mechanically inducedfailures will occur in
thefuture.
5.1.5 Future of machi ne i nsul ati on
Thedevelopmentof improvedinsulating-tapesystemscontinues, asdoestheadvance-
mentof stator insulationmanufacturingtechniquesfor largeunits for exampleVPI
processes. A newconcept utilizinghigh-voltageXLPE cableas thestator winding
104 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
thusallowingoperationatmuchhighervoltagesisdescribedinReference10. Thepro-
totypewascommissionedin1998, havingaratingof 11MVA at45kV and600rpm.
Thefirstcommercial unitsincludedaturbinegenerator ratedat136kV, 42MVA and
3000rpmandahydro-generator of 155kV, 75MVA and125rpmrating. Applying
thesameprinciple, high-voltagemotorswithintheratingrange4MWto70MWare
available.
Monitoringof thelargemachineinsulationsystemsisexpectedtoexpandasthe
measurementtechniquesbecomemoreadvanced. Somepossibilitiesarediscussedin
Chapter 9.
5.2 Tr ansfor mer s and r eactor s
Theinsulationproblemsassociatedwithpower-systemtransformersandreactorsare
especiallyrelatedtothecomplexityof thevariouswindings, themethodsof cooling,
theneedfor efficient clampingtechniques toresist short circuits andtransportation
forces, thelocation and mounting of interconnecting leads, theaccommodation of
bushings and tap changers, and theformand condition of theinsulating materials
themselves. All thesefactors influencetheforms of thethree- dimensional electric
fields and the magnitude of stresses in the insulation structures as set up by the
steady-statevoltagesand, critically, thesurgesduetolightningandswitching.
Insulation deterioration in serviceoccurs naturally but theratecan bereduced
by good and well-planned maintenance. Although thesituation is improving with
theintroductionof moreextensivemonitoring, transformers aresometimes treated
as the sleeping partners in the system, even though their continuous service is
vital to thesuccessful operation both technologically and economically and to
theversatility of thenetworks. Thelatter includeDC transmissionschemes where
converter transformersformthelinkwiththeAC generationsource(Chapter 1).
5.2.1 Wi ndi ngs
5.2.1.1 Gener al consider ations
Inprincipletherearetwo basic forms of transformer winding: (i) helical andlayer
types withturns woundaxially insingleor interconnectedmultiplelayers; and(ii)
disc(core-typetransformers)orpancake(shell-typetransformers)typeswiththeturns
woundradiallyandthediscscontinuousor coilsinterconnectedif woundseparately.
There are many winding variations depending on the voltage and current ratings,
thetappingrangespecified, thereactancerequiredbetweenwindings(animportant
parameter insystemdesign), anyswitchingsurgetests(seeChapter 1), thelikelihood
of very fast transients(VFT) associatedwithSF
6
switchgear and, very importantly,
thelightningimpulsetest (Chapters1, 6and7).
5.2.1.2 L ightning impulse design
Theimpulsetest requirement has amajor influenceon thedesign of high-voltage
windingsasthevoltagedistributionbetweensections, layersandturnsisnotidentical
I nsul ati on defects i n power -system equi pment: Par t 2 105
to that for thepower-frequency condition. This resulted in theuseof shielding to
reducetheeffectiveshuntcapacitanceC
g
, interleaving[11,12]toincreasetheeffective
seriescapacitanceC
s
or judiciousdesignwithlargeratingdiscwindingstoincrease
the radial depth, thereby producing a natural increase in C
s
. The effect of these
parameters on the initial distribution of the surge voltage represented as a unit
functionmaybedemonstratedbyconsideringtheequivalentcapacitancenetworkfor
auniformwinding(Figure5.4(a)). Thissimplificationassumestheseriesinductances
of the winding are open circuit at zero time. The final voltages are theoretically
proportional totheresistivecomponentsandarethusidentical totheturnsdistribution.
InFigure5.4(b) areplottedvoltagesfor different valuesof where =

C
g
C
s
inthe
equation v
x
= V{sinh (x/L)}/sinh . Notethat C
g
and C
s
aretheeffectivetotal
valuesandthatthewindinghasanearthedneutral. Byachievingalow valueabetter
balancebetweentheinsulationrequirementsfor steady-stateandimpulseconditions
ispossible. Atsomelocationsspecial precautionsarenecessarytoallowfor thesurge
Incident
wave
(a)
(b)
C
S
C
g
=
= 0
1
2
4
6
12
24
100 90 80 70 60 50
Percent winding length
from neutral
= 100
X
L
40 30 20 10
Neutral Line
V
o
l
t
a
g
e

t
o

e
a
r
t
h

a
s

p
e
r
c
e
n
t
o
f

a
p
p
l
i
e
d

v
o
l
t
a
g
e
0
=
C
g
C
S
Neutral
Shunt capacitance
100
90
80
70
60
50
40
30
20
10
Series capacitance
E
Line
L
X
=

1
0
0
s
i
n
h
s
i
n
h

x
L
Fi gure 5.4 I ni ti al di str i buti on of i mpul se vol tage i n a uni for m wi ndi ng wi th ear thed
neutr al . (a) Equi val ent capaci tance networ k negl ecti ng i nductances and
l osses; (b) I ni ti al di str i buti ons for di fferent val ues of factor
106 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
condition particularlyalongtheaxial insulationpathsatthesurfacesof discwinding
stacks.
Inpracticethevaluesof rangefrom5or 6for anuncompensatedwindingto1
or 2for thehigh-voltagelargeMVA units; thepower frequencydistribution( = 0)
isuniform. For agivenlocation(L = x) thedifferencebetweentheinitial andfinal
voltages is anindicationof thepossiblemagnitudeof theoverswingor oscillation
occurringlater intimeatthatposition. Althoughtheseconceptsarerelativelysimple
theyformthebasisof theadvancedcomputerizeddesignprogramsthat mayinclude
seriesandmutual inductanceparameters. Additionallytheycanbeof assistancewhen
carryingoutafirstassessmentof afailureduringtestorperhapsasuspectedlightning
strikeinservice.
5.2.1.3 Par ticular winding configur ations
A number of manypossiblewindingconfigurationsaredepictedinFigure5.5. These
rangefromthelower voltage(6.6 11kV) low-kVA unitswithsimplelayer (wire
orfoil) windingsthroughratingsof tensof MVA at66132kV withdischigh-voltage
coilstothemajor generator andtransmissiontransformersof 660MVA or higher at
voltagesupto550kV as3-phaseunitsand750kV 1000kV single-phaseunits.
Althoughmany designsof core-typetransformerswithlayer windings[13] and
shell-typetransformerswithpancakewindings[14] areinservice, thecoretypewith
discwindingsareprobablydominant withnormal, interleavedorintershieldedcoils
andappropriateendradial shieldsasshowninFigure5.5(e, f2, f3).
Thesketchesindicatealsothelocationsof tappingwindings. Theseareessential
forvoltageregulationbutcanintroduceunacceptablereactancechangesif positioned
inthemaingap. This restrictionledtotheintroductionof external taps (Figure5.5
(e, f2, f3)). Theincreases inMVA andloadcurrent enabledcentreentry windings
(twohalvesinparallel) tobedevelopedeconomically andallowedthehigh-voltage
input to belocated away fromtheyoke(earthed). Such aconfiguration simplifies
theelectricstressconditionsattheendsof thewindings, althoughwithouter tapping
windingsother clearanceproblemsareintroducedaroundthehigh-voltagelead.
Reactorsarenotincludedinthefigure. High-voltagepowersysteminter-buscou-
plersandfilter reactors(DCsystems) areseries-wound, beingdesignedtowithstand
afull-impulsetest withanappropriateequivalent surgeimpedanceat thenon-tested
terminal, andalsotowithstandaseparatesourcetest toground. Thelatter canbea
searchingtest for oil-insulatedwindings becauseof thelargeareas andvolumes of
oil, whicharesubjectedtoafairlyuniformstress. SeeChapter 2.
5.2.2 Tr ansfor mer i nsul ati on str uctures
An outlineis given in thefollowing of themoreimportant aspects of transformer
structures, includingindicationsof whereproblemsmightoccur. Examplesaregiven
of particularfaultsandhowthesemightbedetectedormonitored. Theconfigurations
may beloosely dividedaccordingto their mainfunctions, namely windings, inter-
winding space, winding end clearances, interconnecting leads, bushings, internal
terminationsandtapchangerswithleadsandmountingboards.
I nsul ati on defects i n power -system equi pment: Par t 2 107
V
ph
V
ph
C
L
C
L
C
L
C
L
C
L
C
L
A3 HV bushing
(a) (b)
(d)
(f)
(e)
(c)
A2 A1 neutral
a22
Foil H-L
clearance
Gaps for approximately
50% induced test
Small gap
on insulation for
LV and neutral
a12 a11 a21
L
V
L
V
L
V
L
V
L
V
L
V
HV
1 2 3
H
V
T
T
M
V
HV
T
T
H
V
T
A
P
S
HV
shield
20%
40%
60%
80%
H
V
H
V
T
HV(A)
HV(B)
Fi gure 5.5 Exampl es of the l ayout of di fferent types of tr ansfor mer wi ndi ngs. (a)
Low kVA (l ayer wi re or foi l ); (b) Medi um kVA (di scs and taps) ful l y
i nsul ated; (c) Layer wi ndi ngs core-type power tr ansfor mer s; (d) Pan-
cake wi ndi ngsshel l -type power tr ansfor mer s; (e) End connected di sc
wi ndi ngs core-type power tr ansfor mer s; (f) Centre entr y di sc wi ndi ngs
core-type power tr ansfor mer s
Themajorityof transformersareinsulatedwithcellulose-based, oil-impregnated
materials. However, withinthedistributionrange possibly upto afewMVA, for
example22kV/415V (:Y) cast-resinhigh-voltagecoilsareinservice. Unitsfor
108 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
indoor applicationsinthe150300kVA, 17kV/415V rangearenowcommon. Many
of thesmall 510kVA, 11kV/415V pole-mountedtransformers areinsulatedwith
oil-basedsystemsbut thesewill probablybereplacedbydry-typedesigns.
Althoughnotconsideredhere, anumber of SF
6
gas-insulatedpower transformers
areinservice. A notableexampleisthesubstationinthebusinesscentreof thecity
of Sydney (Australia), where three 400MVA, 330/132kV units were installed in
2003[22].
Overall conditionsthat cancontributetooil-filledtransformer insulationfailures
are(i) thepresenceof moistureduringtest causedbyinadequateprocessingand, on
site, duetoinappropriatemaintenanceorlong-termageing, and(ii) thetrappingof air
causedbyincorrect vacuumfillinginthefactoryandonsite, thelatter perhapsafter
loweringof theoil for inspection purposes or abushingchange. In somedry-type
designs, airclearancesandterminationconfigurationsmaybesuchastocausecorona
andconsequential erosionof theadjacent insulatingmaterials.
5.2.2.1 W indings
(a) Low-kVA oi l -i mpregnated wi ndi ngs
The high-voltage windings of the low-kVA units (11kV systems) may consist of
resin-covered, small-diameter wirewoundinlayerswithalow-voltage, single-layer
winding of enamelled rectangular strip. In some cases the LV winding comprises
layers of aluminiumfoil separatedby oneor two sheets of presspaper of, perhaps,
0.12mmthick. Although thevolts/turn arevery low perhaps five afewcases
have been reported of inter-turn service failures, probably due to metal particles
trapped in the rolls of foil puncturing the paper during operation but not on test.
In another case, froma different manufacturer, an 11kV foil winding of a well-
established design failed during impulse test. The inter-foil (inter-turn) insulation
was 0.2mmthickness (several tens of volts/turn) specially preparedto allowresin
curingandoil penetrationduringsubsequent vacuumfilling. Thefailurewasdueto
unusual mechanical distortionof thefoil duringwinding. Foil thicknessesseemtobe
intherangeof 0.4to0.6mm, dependingonthecurrent rating. A moreconventional
11kV, 7MVA three-phasewindingmight haveeight layersof copper, resin-coated,
rectangular conductor withdimensions of theorder of 11mm 4mm 4strands
inparallel andinter-layer presspaper containingvertical coolingduct(s). Thesystem
wouldbedriedandoil-filledunder vacuum.
A magnetic-typevoltagetransformer (MVT) consists of abushing, acoreand
windingswithhighturnsratios. Thehigh-voltagewindingconductor isof finewire,
perhapsof theorder of 0.5mmdiameter, usuallywoundinconcentriclayers. A good
impulse-voltageresponseisachievedbecauseof thehighcapacitancesbetweenlayers
andbytheprovisionof inner andouter electrostaticshields. For thehighest voltages
acascadearrangement ispreferred.
Resin- andSF
6
-insulatedunitsareinservice. Efficientresinimpregnationaround
alargenumber of turnsof finewirerequiresverycareful processing.
A number of major failures of oilpaper MVTs havebeen associated with the
terminationof theendsof theelectrostatic shieldontheoutsideof thehigh-voltage
I nsul ati on defects i n power -system equi pment: Par t 2 109
HV terminal
(a) (b)
HV
shield
Interlayer
insulation
Earth
LV shield
Secondary
windings
Fi gure 5.6 Si ngl e-stage magneti c vol tage tr ansfor mer (MVT). (a) Layout of dead
tank desi gn wi th l ayer wi ndi ng; (b) Equi val ent ci rcui t representi ng many
concentr i c l ayer s
winding. Thedryingandvacuumoil impregnationof theductless paper insulation
between winding layers must be efficient to prevent voids and subsequent partial
discharges. Theformof aparticular single-stagestructureisshowninFigure5.6. In
somecasesductsandlayer endinsulationmaybeincorporated
(b) Dr y type wi ndi ngs
Small, dry-typewindingsfor special applicationsareinsulatedwithmaterialssuchas
polyester film(Class130, e.g. Mylar andMelinex) andcompositesandaramidpaper
(Class 220, e.g. Nomex) andthenimpregnatedor simply dippedinresins followed
by curing. Thelatter procedures can lead to thetrapping of air with resultant par-
tial discharges of many thousands of picocoulombs intheinter-windinginsulation.
A number of suchunitsfailedinserviceduetothiscause. PDtestingenabledrecom-
mendationstobemadefor improvementsinthedesignandprocessingmethods. For
larger ratingsof, perhaps, 150kVA (three-phase) at22kV, VPI techniqueshavebeen
developed. This allows completeencasement of eachphaseof theHV windings in
theresinusingamouldgivingasmoothouter finishwithappropriateconnections.
Eachof thewindings is loweredover theparticular LV windingonthecorewitha
corona-freeinter-windingairgap. PDlevelsof lessthan10pCareachievablealthough
connectingleadsandmountingshaveproducedmuchhigher values aslocatedby
ultrasonicdetectors.
(c) Hi gher -r ated oi l -i mpregnated wi ndi ngs
Low-voltage windings (1122kV) for oil/paper/pressboard-insulated transformers
of higher ratings are usually of the helical formin one layer with many strands
in parallel. Problems have occurred where the enamelling on the conductors was
110 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
relieduponto provideinsulationbetweenstrands but after several years of service
abrasivemechanical movement betweentheconductorsproducedshort circuitswith
subsequent circulatingcurrents andoverheating. This requiredcompleterewinding
withpaper-insulatedconductorsat considerablecost. Suchfaultsmight bedetected
bymonitoringof gasesandfuransproducedbyoverheatingof theenamel, pressboard
spacersandadjacent oil.
Thelarger layer-typehigh-voltagewindingsincoretypetransformersfor 220kV
andabove(Figure5.5(c))mightconsistof 57layersof paper-insulated, continuously
transposedcable. Extralayersareincludedfor tapsandthelow-voltagesectionof an
autotransformer. Theturnsarewoundonwraps(sheets) of paper uptoathicknessof
possibly 12mmandcooledby oil ductsmaintainedwithaxially locatedpressboard
spacersof 6to12mmthickness, therelativedimensionsdependingontheelectrical
andthermal designrequirementsfor theparticular unit. Withanearthedneutral, the
layersandinsulationaregradedasinFigure5.5(c). Theinter-layer paper isflanged
(petalled) at right angles to formtheendinsulationincludingacontinuationof the
vertical oil ducts. In one arrangement the layer ends are brought out through the
horizontal ductsandconnectedinseries toptobottom bytheprovisionoutside
thewindingshieldof accuratelylocatedinsulatedleadsof specifieddimensions[13].
Theinner andouter shields arenecessary toensureagoodimpulsedistributionfor
thesimplearrangement shown. Theinner shield may not berequired under some
conditions. Insulationfaultsinthesewindingsincludedeteriorationof thewinding-
to-shieldconnections, blockingof ductsduetobulgingof paperattheendof thelayers
andleadconnectionproblems causinglocal overheating. Gas analysis is themajor
monitoring techniqueused, although somefaults havebeen located ultrasonically.
Trackingduetoelectrostaticcharginginthelowerinletandcoolingductswasreported
by onemanufacturer. Tests onmodels andin-servicetransformers werecarriedout
[15]. Theproblemwaslocatedbypartial-dischargeultrasonicmethodsandconfirmed
onunwinding. It isconsideredthat thelayer-typewindinghasbeensuccessful with
unitsoperatingfor40yearsormore. However, itappearsthattheinternational mergers
of manufacturersandtheassociatedeconomicevaluationswill seethedemiseof the
designinfavour of themoreversatilediscwinding.
Shell-typetransformercoils(Figure5.5(d))havegoodsurgeresponseandarewell
supportedmechanically[14]. Theductsbetweenpancakes arerelativelythinandthe
coolingoil-flowratestendtobehigh. Electrostaticchargingeffectswerebelievedto
haveresultedinanumberof servicefailuresindifferenttypesof transformersinwhich
local oil velocitiesseemedtobeexcessive. ExtensiveinvestigationsintheUSA and
J apanandby CIGRE Study Committees 12and15werecarriedout [16,17]. Some
monitoring is possibleby low-voltageDC measurements and ultrasonic detection.
Theadvantagesof shell-typetransformersarebeingappliedbysomemanufacturers.
Themajorityof high-voltagewindingsfromtensof MVA (66kV) uptothelargest
ratings and highest voltages consist of discs of paper-insulated conductors wound
continuouslyorinpairs(coils). Horizontal andvertical ductsareformedbypressboard
spacers. Withdirectedflowunits, ducts ontheinsideandoutsidediameters of the
windingareformedbyvertical barriers(cylinders) andinsulationwashersprovided
at critical locations inthestack inorder to direct theoil flowhorizontally between
I nsul ati on defects i n power -system equi pment: Par t 2 111
(a) (b)
Fi gure 5.7 Exampl e of the sur face of a medi um-si ze di sc wi ndi ng (10 MVA, 66 kV).
(a) Vi ew of l eads from tappi ng wi ndi ngs l ocated wi thi n the 66 kV stacks.
(b) Di spl acement of 66 kV wi ndi ng tur ns due to ser vi ce faul t
discs. Theoil ispumpedintothebottomof thewindingfromtheexternal air or water
coolers.
Thefailureratesof windingsarelow, asreportedinCIGRE surveysandpapers.
Problems with theolder medium-sizeunits areprobably associated with lightning
strikeswheretheprotectionwasinadequateor thedesignnot appropriate. Windings
may fail also dueto tap changer maloperation (Figure5.7). Occasionally, unusual
faults occur as in a 66kV winding where a nut had lodged between a disc and a
vertical barrier. Thisprobably occurredduringaninspectionthreeyearspreviously,
but was not detected by theovervoltagetests applied at that time. It was deduced
that the nut was dislodged into a more critical position during service and partial
dischargeswereset up, whicheventually ledtoawindingfailure. Suchacondition
mayhavebeendetectablebygasanalysesor, possibly, byPDmeasurementsonsite.
It would havebeen very difficult, or impossible, to detect by ultrasonic methods,
asthefault waswithinthewinding. Therelocationof aged, moderate-sizeor large,
transformerscanproduceinternal movementandconsequentinsulationdamage. This
problemhas beenreportedfor newtransformers, whereinsufficient allowancehad
beenmadefor theforcesproducedduringtransport byship, rail androad.
Thedesirability of continuous onlinemonitoring, especially for largegenerator
andtransmissiontransformers, isindicatedbyanumberof failuretypeswhosecauses
aredifficulttodeterminewithcertainty. Theseinclude(i)manufacturingfaultsrelated
tothewindinginsulation, whichdevelopintoadangerousconditiononlyafter some
timeinservice, e.g. mechanical failuresof conductorjoints; (ii) electrostaticcharging
112 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
effectscausedbyhigh-velocityoil flowsinwindingducts; (iii) contaminationcaused
by theproductionof finemetal particlesintheoil circuit; (iv) reductioninwinding
clamping pressures, allowing movement of the coils and consequential insulation
damage, duringhighcurrent throughfaultsand, possibly, normal switching; and(v)
theeffectof high-frequencytransientsproducedbyalocal GISflashover. Anexample
of theconditionin(iv) wasbelievedtobethecauseof theshortingof thestrandsof
aturninadiscatthebottomof athree-phase, 400MVA, 362kV windingafter many
years in service. This fault probably produced gas bubbles, which developed into
anaxial flashover alongthevertical surfaceof thehigh-voltagewinding. Themajor
failureof onelegof onephaseof a990MVA, 21.45/500kV generator transformer
resultedinadetailedinvestigationof thepossiblecauseasdescribedinReference18.
After extensivemeasurements, modellingandanalysesitwasconcludedthatthefail-
urewas possibly dueto excessivevoltagestresses across theline-enddiscs. These
stresses were assumed to be created by several SF
6
bus flashovers at the end of
the 600moverhead transmission line connected to the transformer. It was noted
that duringaperiodof sixmonthsthetransformer wasnot connectedtothegenera-
tor, thus creatingaconditionwheretheLV windingcaninfluencetheHV winding
stresses under surgevoltages. Thepaper concludes that thetransformer designers
requiredetailedinformationof probableservicetransient conditionsnot coveredby
thenormal factorysurgetests.
Thewindings LV, HV andtaps of a300MVA, 400kV, three-phaseunit may
haveanexternal diameter of 2.5metres, aheight of >3metresandweigh8tonnes
(eachphase) whenassembledonthecore. Thetotal weightof suchatransformerwith
thecoreandcoilsassembledinthetank andfull of oil ismorethan250tonnes. The
height of thetankmight be4.6metresandthebushingsafurther 2.5.metres.
The core and coils of a 220kV substation transformer with high-voltage disc
windingsinthefinal stagesof assemblyisdepictedinFigure5.8. Inthisunitthecon-
nectionsfromtheoutsidetappingwindingsareconnectedtotheon-loadtapchanger
located on the left-hand side of the structure. A series reactor in an 11kV delta
connectionismountedontheright.
5.2.2.2 I nter-winding insulation
Inoil-core-typetransformers theinter-windinginsulationusually consists of wraps
or cylinders of pressboard, thelatter having avertical split overlap with ascarfed
joint to prevent radial build-upandto allowtolerances onthediameters. Oil ducts
are maintained by vertical pressboard spacers in the thickness range of 6mmor
less, up to 12mmor more. Theboard thicknesses are36mmwith lengths of the
order of 3 metres. The spacers should be attached with non-contaminating glue
and in the lower electrically stressed areas where possible. Adjacent to the high-
voltagewindingincorrect procedures canleadtoanaxial failure, especially during
animpulsetest or lightningstrike. Thedamagetoadisccoil followinganaxial fail-
ureonsiteis indicatedinFigure5.9. Anexampleof theequipotential distribution
andpossiblebarrier configurationat theendof ahigh-voltagewindingis givenin
Figure5.10.
I nsul ati on defects i n power -system equi pment: Par t 2 113
Fi gure 5.8 Core and coi l s of a 225 MVA, 220/66/11 kV substati on tr ansfor mer wi th
outsi de taps and centre-entr y hi gh-vol tage l eads [ photogr aph by AREVA
T& D, Br i sbane, Austr al i a]
Fi gure 5.9 Exampl e of wi ndi ng damage due to i nter nal i nsul ati on breakdown
It is very unusual for inter-winding insulation systems to fail, particularly in
puncture. Oneconfigurationthathasbecomeaprobleminafewinstancesaftermany
yearsof operationisastructurewheretheinterwindinginsulationcomprisescontin-
uouswrapsof presspaper. Thiswasintroducedtoproduceamoreeconomicsystem
andexploit theadvantagesof anewdesignof winding. Theunitsweresuccessfully
tested (300kV class) but in serviceit becameapparent that thematerials may not
havebeen completely dried out or efficiently impregnated under vacuumbecause
of thelargeradial thicknessandaxial lengthof theductlesssystem. Failuresdevel-
opeddueto highlocal partial discharges andassociateddielectric losses. Puncture
occurred fromthehigh-voltagewinding to theneutral axial shield in afewcases.
Disassembly of non-failedunitsshowedagradual erosionintothewrappingsat the
insidediameter(centreentry)of thehigh-voltagecoils. Otherdesignswithcontinuous
114 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a) (b)
Fi gure 5.10 Typi cal i nsul ati on l ayout and equi potenti al fi el d of an end-connected
hi gh-vol tage power tr ansfor mer [ di agr ams by Wei dmann Tr ansfor mer
Systems AG, Rapper swi l ]
paper wrapshaveprobablyexhibitedsimilar problems. Bytheinclusionof thinducts
(seeSection3.2.1) it ispossibletoachievehighaverageoperatingstressesif thisis
requiredintheproposeddesign. Otherfaultsdetectedwithininter-windinginsulation
includeabrokenconnectionbetweentheinner axial shieldandneutral.
A potential fault conditioninanewer transformer was thepresenceof contam-
inationby finemetal particles producedby afaulty cooler. This particular unit had
not failedbut ondisassemblingwasfoundtocontainpressboardandpaper surfaces
coveredwiththeparticulate. On-sitePDtestsshowednoapparentlysignificantvalues
and, again, after a factory impulse test during which failure occurred well below
theexpectedwithstandlevel. Physical examinationdetectedafinetrack mark along
thesurfaceof thebarrier formingtheduct next to theinsidediameter of thehigh-
voltagewinding(centreentry). Inthis caseconventional electrical monitoringwas
unabletopredict apossibleservicefailureduetotheparticlesor atrackpathcreated
byalow-energysurgeflashover [19].
Metal particles of millimetredimensions withinastructurehavebeendetected
duringPD testsinthefactory. Suchaparticlewill producePDsonly if it islocated
andtrappedinapositionwhereitsshapeenhancestheelectricstress. PDmagnitudes
of the order of hundreds of pCs might be expected [19]. A number of failures in
serviceas well as duringovervoltagetests arereportedintheCIGRE review[20].
Theseproblemsaremorelikelytobeassociatedwithlongoil distances, e.g. abushing
terminationtoturretortheoutsideof awindingtoground, ratherthanwithinthecoils
themselves.
5.2.2.3 End insulation
Withend-connectedwindings(Figures5.5(e) and5.10(a)), careful layoutof theinsu-
lationisnecessarytopreventflashoveralongsurfacestoground(includingtheearthed
core) and to adjacent windings such as outer taps and to ensuregood oil flowfor
cooling of thewindings. Theuseof aradial shield connected to thelineend disc
I nsul ati on defects i n power -system equi pment: Par t 2 115
(Figure 5.10(a)) [21] and the contoured cylindrical pressboard barriers following
the equipotential lines (Figure 5.10(b)) provide an efficient insulation system. In
Figure5.10(b) theequipotentials represent 5per cent intervals. Thecorresponding
electricstressesmaybecalculatedbymeansof anassociatedfieldcomputerprogram.
Ideally, thespacings betweenbarriers may bestaggeredinorder to prevent partial
discharges in the particular oil gap (Figure 2.3) during testing. This is especially
important adjacent to themaximum-stress areas at thecorners of thehigh-voltage
andlow-voltagewindingsandthetapwindings. InFigure5.10(b) shieldsat thecor-
ners of theLV andtapwindings areat, or near, earth potential. At theposition of
connectionof theHV leadtheconfigurationbecomesmorecomplex. Preformedbar-
riersandcomponentsfor leadinsulationareavailablefromspecialistmanufacturers.
Thefiguresplot isfor aseparatesourcetest. Duringaninducedor impulsetest con-
ditionsintheendinsulationdonotchangesignificantlyexcept, perhaps, atthelower
cornersof theradial shield.
Inaparticular caseomissionof horizontal pressboardwashersabovethecooling
gapthroughtheradial shield(Figure5.10(a)) resultedinanimpulsetestfailurealong
atortuouspressboardpathtoearth.
A problemthathasarisenistheproductionof partial dischargesattheedgesof the
metal (earthed) windingclampingstudswherethesewerenot of correct dimensions
andinsulated. Anunsatisfactory arrangement canresult inPDs of many thousands
of pC leadingtoflashover toanadjacent high-voltageleador similar duringfactory
tests(seeFigure3.6). Dischargesat thesurfaceof earthedclampingstudshavebeen
detectedbyDGA on-siteandlocatedusingultrasonicmethods. Duringmaintenance,
theupper insulating board (laminated wood or pressboard) may beexposed when
lowering the oil, thus allowing ingress of moisture if precautions are not taken.
FailuresduetodielectriclossesinsuchboardsarepossibleasconsideredinChapter2.
5.2.2.4 I nsulation to tank and cor e
Itisusual inthelargertransformerstoprovidepressboardbarriersinordertobreakup
thelongoil pathsbetweenthewindingsandtank, betweenphases(three-phase) andto
theouter coreleg(5-limbdesigns), especiallywhereparticlesmight bepresent [20].
Thedistancesinvolvedarepossiblyintherangeof 50250mmdependingonthetest
voltagesandthechosensafety factor. Inolder transformerstrackingonbarriershas
beenfoundalthoughnodirect feed point wasinvolved. Theeffect isprobablydue
tomoisturedevelopmentor ingressandwouldnotbedetectablebyexistingelectrical
monitoring[19]. SomechangeinDGA samplevaluesmaybemeasurable.
5.2.2.5 Bushing inter nal ter minations and leads
Bushingdesigns havebeendiscussedinSection4.3. Themethodof connectionof
aconventional bushingwithinapower transformer at highvoltages is indicatedin
Figure5.5(e) andtheassociatedfieldplotsfor bothconventional andre-entranttypes
mountedinaturret inFigure4.3. PlotsarealsopresentedinReference13, including
theequipotential distributionfor are-entrant bushingwithout ahigh-voltageleadin
place. BushingsinstalledintransformersareshowninFigure5.11.
116 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a) (b)
Fi gure 5.11 Bushi ngs i nstal l ed i n hi gh-vol tage power tr ansfor mer s. (a) 100 MVA,
11/230 kV gener ator tr ansfor mer [ photogr aph by Wi l son Tr ansfor mer
Company] ; (b) 133 MVA, 330/

3/138/

3kV auto tr ansfor mer [ by

per mi ssi on of Tr ansgr i d NSW]
Flashovershaveoccurredbetweenbushingsandturretsduetocontaminatedoil
andparticlesandwherealocal barrier hasbeenexposedfor toolongduringinstalla-
tion. Inanumberof casesthestressdistributor(Figures5.5(e) and4.3(a)) hasbecome
loose, resultingindischarging(sparking) andgasproduction, whichmighthavetrig-
geredanaxial flashover. SuchdischarginghasbeendetectedbyDGA measurements
andsuccessfully locatedby ultrasonic techniques. Problems inre-entrant bushings
can arisedueto incorrect location of thepaper-insulated lead within theporcelain
(Figure4.3(b)). For example, air might betrappedwithinthepaperporcelaingapin
someconfigurationswhentoppinguptheoil onsite.
Thedesign and insulation of leads frombushing ends to thewindings require
special care to ensure safe stresses at the paper surfaces. By use of prefabricated
pressboard systems, concentric cylinders and angles can be built in as shown in
principleinFigure5.5(f, 2/3). A critical regioninsuchstructuresiswhereascarfed
joint is required at theexit point. Creep along an incorrectly madescarf can lead
toamajor flashover withpre-dischargingof thousandsof pCsduringAC tests. The
dischargesmaybedetectedbyPDmeasurementsandlocatedbyultrasonicmethods.
Failures to thetopof unshieldedtapwindings without barrier protectionhavealso
occurredontest, theconditionbeingexacerbatedbyther adi al stressingproduced
bytheadjacent high-voltagewinding(Figure5.5(e) and(f, 2/3)).
Thelocation, thethicknessandthediameter of insulationaroundhigh- andlow-
voltageleadsaredeterminedbythelocal fielddistortionaswell astheoil andcreep
paths. Fieldanalysesarenecessarytoensurethestressesareacceptable.
I nsul ati on defects i n power -system equi pment: Par t 2 117
5.2.2.6 Tap changer s
Althoughmost tap-changer failurescanbetracedtomechanical maloperation, some
arerelatedtotrackingalongtheinsulationsurfacesor adirect oil flashover perhaps
associatedwithcontaminationcreatedbyincorrect wearingof themetal contacts.
Inaparticular66kV transformer, distortionof thetapwindingsinthecentreof the
high-voltagewinding(Figure5.5(b)) wasduetoaflashover alongtheinner cylinder
of the20-year-oldtapchanger. Somedamagewasalsopresentinthemainwindings,
thusrequiringarewindof thecompleteleg.
As conditions within a tap changer are quite onerous, monitoring of the gas
levelsandregular maintenanceareessential. Newtechniquesarebeingdevelopedto
improvesurveillance. Careisnecessary wheninterpretingtransformer DGA results
if thetapchanger isinthemaintank or thereisapossibilityof leakagefromthetap
changer/diverter enclosures.
5.2.2.7 Cor e and magnetic shields
Overheating due to circulating currents in cores and laminated magnetic shields
causedby low-level interlaminar insulationdegradationhasbeendetectedby DGA
sampling. Thetemperaturesproducedcanresult inbreakdownof theoil. Theloca-
tions havebeen identified using ultrasonics and units subsequently kept in service
until remedial actioncouldbetaken. If allowedtobuildup, thedepositsat thefault
aredifficulttoremoveasexperiencedinanold275kV, 120MVA transmissiontrans-
former. Withmagnetic-shieldfaults, continuedoperationmaybepossibleatreduced
loadcurrent.
5.2.2.8 M aj or tr ansfor mer failur es
Servicefailures of oil-filled transformers can result in completedestruction of the
systembyfireandirreparabledamagetocoreandtankaswell asexpensiveoutages.
For example, seeFigure5.12. If thesupply-systemprotectionoperatesquickly, dam-
agemaynotinvolvethewholeunit, althoughcarboncontaminationcanbeextensive,
especially if pumpedoil coolers areinstalled. It is anticipatedthat development of
continuousmonitoringsystemswill helpeliminatesuchcatastrophicfailures.
5.2.2.9 Power tr ansfor mer summar y
Thesituationsdescribedaretypical of someof theinsulationfaultsthatmightdevelop
inpower transformers. Althoughaparticular failurecanbeverysignificant it should
beappreciatedthat theproblems involveasmall percentageof thetotal number of
unitsinservice. However, itisprobablethatthepercentagewill increaseastheassets
age, full-load(or more) currentsaredemandedfor longer periodsandhigher electric
stressestendtobeappliedinnewdesignstomeet economicdemands. It isessential,
therefore, that testing, insulation-condition assessment and monitoring continueto
beimproved, especiallyfor thegenerator transformersandthemedium- tolarge-size
transmissiontransformers. Itwouldbeexpectedthatif largeSF
6
-insulatedunits[22]
becomemorecommonsomeof theexistingoperational problemswill bereduced.
118 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Fi gure 5.12 Fi re fol l owi ng fai l ure of a 330 kV tr ansfor mer i n a tr ansmi ssi on
substati on [ photogr aph by Brendan Esposi to/Sydney Mor ni ng Her al d]
5.3 Summar y
A number of possibleinsulationproblemsassociatedwiththelargest unitsof equip-
ment generatorsandpower transformers arediscussedinthechapter. Although
faults areinfrequent, their influenceon theoperation of thepower systemcan be
verysignificant, botheconomicallyandtechnically. Thepreventionof lengthymajor
outagesaswell asdamagetocritical plantitemsjustifiesconsiderationof theapplica-
tionof thenewcondition-monitoringtechniquesastheyaredeveloped. Thisapplies
especiallytoonlinemethodsbut alsoincludesthecontinuationandimprovement of
offlineprocedures. Anunderstandingof theequipment insulationconfigurationsas
considered in this chapter assists in interpretation of failure scenarios [23] and in
applyingthemost appropriatemonitoringsystem.
5.4 Refer ences
1. Yin, W., et al ., Improvedmagnet wirefor inverter-fedmotors, Proceedi ngs of
the I EEE El ectr i cal I nsul ati on Conference, Chicago, September 1997
2. IEEE 522-2004, IEEE Guide for Testing Turn-to-turn Insulation on Form-
WoundStator Coilsfor A.C. RotatingElectricMachines.
3. Hains, A.J ., Insulationperformanceunderswitchedvoltagewaveforms, Power
Engi neer i ng Jour nal , August 2000:15863
I nsul ati on defects i n power -system equi pment: Par t 2 119
4. Wheeler, J .C.G., Effectsof converter pulsesontheelectrical insulationinlow
andmediumvoltagemotors, I EEE El ectr i cal I nsul ati on Magazi ne, March/April
2005;21(2):229
5. IEC/TS 60034-18-41Ed.1.0, 2006Rotatingelectrical machines, Part 18, Sec-
tion41: Qualificationandtypetestsfor Type1electrical insulatingsystemsused
inelectrical machinesfedfromvoltageconverters.
6. Say, M.G., Al ter nati ng Cur rent Machi nes, 5thedn(Pitman, London, 1990)
7. Boulter, E.A., andStone, G.C., Historical developmentof rotorandstatorwind-
ing insulation materials and systems, I EEE El ectr i cal I nsul ati on Magazi ne,
May/J une2004;20(3):2539
8. Wilson, A., Slot dischargedamageinair cooledstator windings, I EE Proc. A,
Sci . Meas. Technol ., May1991;138(3):15360
9. CIGRE Group11RotatingMachinesReport, El ectr a, February1999;(182):9
10. Leijon, M., Dahlgren, M., Walfridsson, L., Ming, L., andJ aksts, A., A recent
developmentintheelectrical insulationsystemsof generatorsandtransformers,
I EEE El ectr i cal I nsul ati on Magazi ne, May/J une2001;17(3)
11. Chadwick, A.T., Ferguson, J .M., Ryder, D.H., and Stearn, G.F., Design of
power transformers to withstand surges due to lightning with special refer-
enceto anew typeof winding, Proc. I nst. El ectr. Eng., December 1950;97,
part II:73750
12. Pedersen, A., On response of interleaved transformer windings to surge
voltages, I EEE Tr ansacti ons on Power Appar atus and Systems, 1963;82:34956
13. McDonald, D., Theelectrolytic analogueinthedesignof high-voltagepower
transformers, Proc. I nst. El ectr. Eng., April 1953;100 part II:14566
14. Bean, R.L., ChackanJ r., N, Moore, H.R., andWenz, E.C., Tr ansfor mer s for the
El ectr i c Power I ndustr y (NewYork, McGrawHill, 1959)
15. Poittevin, J ., andSapet, J ., Preventivedetectionof electrostatic phenomenain
transformers, Paper 15/12-01, CIGRE, Session1992
16. Lindgren, S.R., Washabaugh, A.P., Zahn, M., Brubaker, M., vonGuggenberg,
P., and Nelson, J .K., Temperature and moisture transient effects on flow
electrificationinpower transformers, Paper 15/12-02, CIGRE Session1992
17. CIGRE J WG 12/15, Lindgren, S.R. (convener), Static electrification in large
power transformers, El ectr a, August 2000;(191):1014 (Technical Brochure
170)
18. Morched, A.S., Marti, L., Brierley, R.H., andLackey, J .G., Analysis of inter-
nal winding stresses in EHV generator step-up transformer failures, I EEE
Tr ansacti ons on Power Del i ver y, April 1996;11(2)
19. J ames, R.E., Blackburn, T.R., Cotton, M., et al . Challengesfor advanceddiag-
nostic techniques faults undetectable by existing electrical methods, Paper
15-306, CIGRE Session2000
20. CIGRE WG 12.17, Aubin, J . (convener), Effect of particles on transformer
dielectricstrength, El ectr a, J une2000;(190):13540(Technical Brochure157)
21. Moser, H.P., and Dahinden, V., Tr ansfor mer board I I (H. Weidmann AG,
Rapperswil, 1987); seealsoTr ansfor mer board (1977)
120 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
22. Ebb, G.M., and Spence, G.S., Gas insulated transformers for Haymarket
Substation, Proceedi ngs of the I EEE/PES Conference on Tr ansmi ssi on and
Di str i buti on, J apan, October 2002, vol. 1, pp. 51116
23. CIGRE WG D1.11 Allan, D. (convener), Service aged insulation guidelines
on managing the ageing process, El ectr a, J une 2003;(208):6972 (Technical
Brochure228).
5.5 Pr oblems
1. (i) Differentiatebetweentheconventional insulationandcoolingsystemsof
alargesteamturbinegenerator andthat of ahydro-generator.
(ii) Howarethesurfacedischargescontrolledattheexitof thestatorbarsfrom
thecoreslots?
(iii) Describe techniques suitable for the monitoring of possible partial dis-
chargesinthetwotypesof machine(refer toChapter 9).
2. In an air-cooled, medium-size rotating machine the outer earthed conducting
layer onastator bar hasbeenerodedbyvibrationwithinacoreslot. Determine
whether air dischargeswouldbeexpectedfor aninsulationthicknessof 3mm, a
spacingtothecorelaminationsof 0.5mmandavoltageof 11/

3kV. If similar
dimensions wereusedinahydrogen-cooledmachineat apressureof 5bar, at
what voltagemight PDsbeinitiatedfor asimilar condition?Refer toFigure2.2.
Assumeapressboardpermittivityof 3.5.
3. (i) Describe the major differences between the insulation systems of large
core- andshell-typetransformers.
(ii) Thespacing (duct) between adjacent discs (7per cent of winding turns)
at the line end of a uniformly wound core type transformer is 8mmof
oil and thetwo-sideconductor insulation thickness is 1.5mm. Calculate
thetheoretical lightningimpulsebreakdownstrengthsof twowindingsof
similar configurationswith valuesof 4and1. Assumeauniformfield(a
bigapproximation) andthatfailurewouldbeexpectedtooccur intheoil at
thelineendtotheadjacentdisc. Thepermittivitiesof theoil andconductor
paper are2.2and3.2respectively. Determinefor whichvoltageclasseach
windingwouldbesuitableassumingaminimumtest safety factor of 1.3.
Refer toEquation3.1, Figure2.3(c), Figure5.4andTable1.2.
4. Fromatransformer DGA results it was found that theoil hydrogen level was
1200ppmandthetotal combustiblegas 3000ppm. This indicatedasourceof
PDswithintheunit. Describemethodsfor detectingandlocatingthesiteaswell
as confirming that the disturbance was probably a partial discharge. Refer to
Chapters6, 9and10.
Chapter 6
Basic methods for insulation assessment
High-voltagetest supplies
Electrical non-destructivetest methods
Physical andchemical assessment techniques
Inthischapteranoverviewisgivenof themoreimportantmethodsemployedinthe
supplyindustryfor assessingtheconditionof insulationinpower-systemequipment
before leaving the factory, on commissioning, during service and when under-
going major maintenance or repair. The particular requirements for equipment of
differentinsulationstructuresaredescribedinChapter7, includingreferencetoStan-
dardswhereappropriate. Moreadvancedrecently developedmethodsarepresented
inChapters810.
Theearliest methods for determiningwhether or not aninsulatingmaterial was
suitablefor aparticular usageincludedtheapplicationof steady-statevoltages for,
perhaps, oneminuteat twiceor moretheoperatingstress. Later, this was followed
bythedevelopmentof surgevoltageteststosimulatelightningandsystem-switching
effects. Muchresearchwasnecessaryinthedesignandconstructionof high-voltage
test equipment andassociatedtest/laboratoryareas.
Inrecentyearstherehasbeenatendencytoclosehigh-voltagelaboratoriesinthose
countrieswithdevelopedpowersystems. However, thismaybeashort-sightedpolicy,
sinceit isessential toprovidelocal referencemeasurement systemsandcapabilities
for checkingandprovingageingequipment, aswell asnewitemsof plant inwhich
unproveninsulatingmaterialsmaybeincluded.
In thefollowing, thevarious methods of assessment aregrouped according to
whether theapplicationis:
(i) anovervoltagewithstandtest;
(ii) anon-destructiveelectrical measurement; or
(iii) aphysical/chemical determinationthat may requireaccess toasampleof the
insulatingmaterial.
122 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
6.1 Gener ation and measur ement of test high voltages
Theformof thesourceandmethodof voltagemeasurementarebrieflydescribedfor
eachtypeof test voltage. Moredetailedtreatmentswill befoundinthebooksby, for
example, Kuffel andZaengl [1], Schwab[2], Kind[3] andHylton-Cavallius[4].
6.1.1 Power -frequency vol tages
Power-frequencyvoltagesforinsulationwithstandtestingareproducedbyspecialised
transformerswithoutputswithintherangeof tensof kVtoafewhundredkVinasingle
unit anduptoamillionkV or moreincascadearrangements. Thevoltagesrequired
forroutinetestingaretabulatedinChapter1. Thecurrentratingsvaryaccordingtothe
capacitancesof theparticularequipmenttobetestedandthekVA ratingsfrom10kVA
(e.g. 0.1A at100kV) upto5001000kVA atthehigher voltages. Theratingsquoted
aremaximaandthecurrent values must not beexceeded for example, a100kV,
10kVA transformer can only operate at 5kVA (0.1A) at 50kV unless otherwise
specified. Such transformers are often designed for short time ratings at full-load
current and arenot overvoltage-tested at thesamemargins as power transformers.
Usuallytheyarenot subjectedtolightningandswitchingtypesurges. Modernunits
aresuppliedwithalowpartial-dischargeguarantee, typicallylessthanafewpCsat
full voltage. In addition to units produced by test equipment specialists, anumber
of power and distribution transformer manufacturers have built their own testing
transformers.
6.1.1.1 Single-unit testing tr ansfor mer s
Thesimplesttestingtransformerconsistsof aclosedcorewithalow-voltagewinding
(240/120V), over whichiswoundagradedlayer windingof many turns, themajor
insulationcomprisingoil-impregnatedpaper. Thehigh-voltagewindingisprovided
withanouter shield(seeFigure5.5(c)) inorder toensureauniform voltagedistri-
butionfollowingaflashoverattheHV terminal duetobreakdownof atestobjectora
measurement spheregap. Thecoreandcoilsaremountedinasteel tankwithahigh-
voltagebushing(Figure6.1) or inmanyapplications, for indoor use, inaninsulating
cylinder withaspinningastheHV terminal mountedonthetopof thecylinder. The
unitinFigure6.2hasadiameter of 1metre. Inonecaseacentre-entrydesignwithan
open coreandouter insulatingcylinder was successfully usedfor separatesource
testingupto voltages of 70kV. Singleunits for productionof amillionvolts were
builtinthe1930sbutsomeeventuallyfailed, probablyduetopartial discharges. The
morecommonsingle-unit designsareprobablyintherangeof 200to600andafew
upto750kV [4].
6.1.1.2 M ultiple-unit (cascade) testing tr ansfor mer s
Production of voltages above 350500kV is probably achieved more reliably by
cascadingtransformersof lower voltages, asindicatedinFigure6.3. Theindividual
units aresimilar to singletransformers but havean exciting winding supplying an
upper transformer, whichisinsulatedfromearthfor theoutput of thelower unit.
Basi c methods for i nsul ati on assessment 123
Fi gure 6.1 200 kV testi ng tr ans-
for mer metal tank
Fi gure 6.2 250 kV testi ng tr ansfor mer
and capaci tor di vi der i nsu-
l ati on cyl i nder s as shel l s
Aninsulatingshell withtheindividual units stackedaboveoneanother toform
acompositetransformer may beused for examplethe2.4MV unit at IREQ[1,4]
or, morecommonly, stand-off insulators supportingthetanks of theupper units. It
isimportant toallowfor thedifferent operatingcurrentsintheunits thelower unit
supplyingthepowerfortheothertransformers andalsoforthenon-uniformvoltage
distributioncreatedacrossthevariouswindingsonflashover at theoutput terminal.
Compensatingreactors aresometimes includedinparallel withthewindings when
testinglargecapacitances, suchascables, wherecurrentsof several amperesmaybe
required. Analysesof thevariousconditionsarepresentedinReferences1and4.
Although cascaded transformers may occupy considerable floor space, this is
offset by theversatility of having availabletwo or threeindividual units and ease
inhandlingwhentests arerequiredat lower voltages. Insomeapplications lower-
voltageunitsmight beutilized(e.g. 100kV each) inorder toformaversatilesource
for siteapplications.
6.1.1.3 Oper ation of testing tr ansfor mer s
Thesimplest supplyfor atestingtransformer consistsof avariablevoltage50/60Hz
sourcesuchasa240V variac transformerforsmall unitsupto5kVA. Analternative
124 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
P
E
H
P
H
P
E
H
Fi gure 6.3 Cascade ar r angement of testi ng tr ansfor mer s: P pr i mar y wi ndi ngs; H
hi gh-vol tage wi ndi ngs; E exci ti ng wi ndi ngs
isatappedregulatingtransformerwithtwomovingcontactsbetweenwhichisaresis-
tor or reactor toavoidtheproductionof switchingsurgesinthetestingtransformer
duringtapchanging. If manytappingsarebuilt-inthismethodisefficientandgiveslit-
tlewaveformdistortion. Probablythecommonestregulatorforcontrol of higher-kVA
transformersistheinductiontype, inwhicharotorissuppliedfromthesourceandthe
stator isconnectedinserieswiththetransformer primaryalsoacrossthesource. The
output of thestator canbevariedover thefull voltagerangebyrotationof therotor
[5]. Themajordisadvantageof theseregulatorsisthewaveformdistortionbutthiscan
beminimisedbyappropriatedesign. Insomecasesamotor-drivensynchronousgen-
erator mayprovidethevariablevoltage. Nowaveformproblemswouldbeexpected
insuchasystem, althoughinterferenceof partial-dischargemeasurementshasbeen
experienced.
If ahigher frequencyisbeingapplied for examplewhensimulatinginsulation
structurestobeusedduringpower-transformer-inducedtestsat200300Hzorduring
accelerated lifetests caremust betaken that series resonanceconditions do not
produceunexpectedvoltagesacrossthetesttransformerwindingduetothecapacitive
natureof theloads. A speciallydesignedtransformer maybenecessaryfor thehigher
frequencies. Such a situation can also arise at normal frequencies with 50/60Hz
units, wherearunawayconditionispossible. If atestingtransformer isswitchedinat
above50percentof itsratingtheoutputcanoscillatetoabovethemaximumnominal
voltage[1]. A reactor of appropriatevalueat theinput (seeFigure6.4) improves
thepower factor of thesupply, thereby reducingthecurrent required. Quantitative
determinationsarenecessaryshouldtherebedoubt regardingthestabilityof thetest
circuits. As with all testing circuits, it is good practice to include interlocking to
prevent closure of the switch to the transformer unless the regulator is set at the
zero-output position.
A conditionthat may arisewhenflashover of alargenon-uniformair gapwith
acapacitiveloadoccurs probablyonthepositivevoltagepeak isfor oscillations
intheoutput circuits todevelop, resultinginanincreasedvoltageappearingacross
thetest transformer. Theeffect was first mentionedby Parnell [6] andis described
Basi c methods for i nsul ati on assessment 125
1 2 3 4 5 7
6
Fi gure 6.4 Typi cal power -frequency test ci rcui t: 1. Regul ati ng tr ansfor mer ;
2. Compensati ng reactor ; 3. Test tr ansfor mer ; 4. Test obj ect; 5. Peak
and RMS vol tmeter s; 6. Dampi ng resi stor ; 7. Sphere gap
inReference4. A series resistanceinthehigh-voltagelinewill prevent theexcess
voltagebeingappliedtothetransformer(suchresistorsarenotpractical atthehighest
voltages) or it may bepossibleto includeathyristor configuration to short-circuit
thesupply at theinstant of theflashover [7]. This phenomenonhas beentermeda
snapover [4].
Thecircuit in Figure6.4 includes acapacitivedivider for measurement of the
high-voltageoutput. C
1
usuallyhasavalueof 100pF andconsistsof several sections
for thehigher voltageswithlarge-diameter ringsattheterminal for theeliminationof
coronaandeffectinggoodvoltagedistributionalongthetopsectionsof thedivider.
Thelow-voltagearmconnections may bearranged to measureboth peak and rms
values [1,2]. If thetest transformer has abushingwithalow-voltagetap, thelatter
may beusedfor measurement, assumingasatisfactory calibrationis provided. The
accuracymaybeestablishedbycomparisonwiththespheregapflashover valuesfor
theparticular loadingor by acalibratedhigh-voltagestandarddivider. Theuseof a
spheregapisdefinedinStandardS6/1(seeSection6.6below) andStandardsrelated
tohigh-voltagetestingtechniquesinIEC Publications60060-1to60060-3[S6/2].
Atthelower voltagesahighresistanceinserieswithafull waverectifier feeding
amovingcoil meter issometimesused for example, a50kV, 33Mresistor. Such
aresistor couldalsoact asthehigh-voltagearmof aresistivedivider withitsoutput
measured by ahigh-impedancevoltmeter. Thethermal stability and freedomfrom
partial discharges andcoronaof thehigh-voltagecomponents must beguaranteed.
For voltages up to 100kV a fully insulated electrostatic voltmeter was produced.
Althoughsomewhatfragileforregularindustrial use, theinstrumentisvaluablewhere
minimumcircuit loadingisrequiredandincaseswherethedifferencebetweentwo
voltages, bothaboveearth, istobemeasured.
Therequiredaccuracyof thehigh-voltagemeasuringsystemsandtheirmethodof
calibrationarespecifiedintheStandards. Aspheregapcalibrationif effectedcorrectly
[S6/1] givesanaccuracyof 3per cent but commercial instrumentsincludingtheir
high-voltagecomponentswouldbeexpectedtogiveamuchimprovedvalue. These
systemsarecalibratedagainst standarddividerswhoseratioshavebeendetermined
inanational or international standardslaboratory.
126 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Itisimportanttonotethatintestcircuitsinvolvingtestingtransformerscomplete
relianceshouldnever begivento avoltagereadingonthelow-voltagewindingas
ameasureof thehigh-voltageoutput. Theapparent turns ratio may not becorrect
or constant under different loadconditionsduetoresonanceeffects. A high-voltage
devicemustbeincludedand, preferably, asecondsystem e.g. aspheregap usedas
acrossreference. Partial failuresinthemeasurementsystemmaynotbeimmediately
apparent perhapsresultinginanexcessivevoltagebeingappliedtothetest object or
anunexpectedair flashover.
6.1.1.4 Ser ies r esonant cir cuits
Theapplicationof tunableseries resonant circuits for high-voltagetestingof large-
capacitiveobjectssuchascables, powercapacitors, generatorsandhigh-voltageshunt
reactorsiswell established. ThebasictestcircuitsareshowninFigure6.5(a) andthe
twopossibilitiesfor testingcapacitive-typeequipment inFigure6.5(b) and(c). The
natural frequenciesof thecircuitsaregivenbyf
n
= 1/(2

LC), whereL istheHV

inductanceandC thetotal capacitance.
Thesupply circuitsareonly requiredtoprovidetheresistivelossesinthehigh-
voltagecomponents. A major advantageof theconfigurationsisthat they aretuned
specificallytothetest frequency, therebyavoidingthepossibilityof unwantedreso-
nances, whichcanproduceovervoltagesandevenfailuresintransformer systems
aswell aswaveformdistortion. Additionally, if breakdownof thetest object occurs
the high-voltage collapses, thus minimizing the damage and reducing the risk of
explosion.
Thedisadvantageof thetechniquesfor general high-voltagetestingisthelackof
versatilitycomparedwithatransformer circuit. A givenresonantcircuitislimitedby
therangeof loadvaluesfor whichit istuneable.
Commercial seriesresonant test setsparticularlyaimedat cabletestinghaverat-
ingsof uptotensof MVA andvoltagesof hundredsof kilovoltsincorporatingseveral
reactors in series. In the usual series circuit (ACRL) the reactor has a split core
(Figure6.5(b)) whichmaybevariedmechanicallytogiveanoverall changeininduc-
tanceof 10to20[1]. Qvaluesof theorder of 30or higher arepossible, requiringan
inputof afewpercentof themaincircuitratingandatransformer voltageof only1/Q
togivethedesiredoutput. Individual units of 350kV havebeenstackedinvertical
assembliestoproduceupto2.1MV (sixunits) [1].
Inonelocationatuningcapacitor withavoltageratingof 1000kV wasusedfor
testinghigh-voltageshunt reactors, e.g. a33MVAR, 500/

3kV unit.
A morerecentdevelopmenthasbeentheintroductionof variable-inputfrequency
systemswithfixedvaluesof seriesinductance(Figure6.5(c)). Therequiredtestfre-
quencies arewithin the20300Hz rangeas specified and discussed in IEC 62067
[S6/3] for tests on extruded cables above 150kV. It would be expected that the
weight/powerratioforsuchsystemswouldbelessthantheequivalentvariableinduc-
tanceconfiguration[8]. Theratios vary from1to 8kg/kVA for ACRL circuits and
0.6kg/kVA for ACFL circuits. Qfactorsfor thesesystemscanbe>100depending
onthetest frequencyandthelossesintheseriesreactor/transformer/loadcapacitors
Basi c methods for i nsul ati on assessment 127
L
L
L
C
C
C
(c)
(b)
(a)
Fi gure 6.5 Pr i nci pl e of ser i es resonant test ci rcui ts: (a) basi c ci rcui t; (b) tuneabl e
reactor ci rcui t (ACRL) fi xed frequency; (c) tuneabl e frequency ci rcui t
(ACFL) fi xed ser i es i nductance
at theparticular f
n
value[9]. For PDtestingapplicationstheIGBT switchingpulses
inthevariablefrequencysourcecanbesuppressed.
Parallel resonantcircuitsmaybeutilizedfor cableandgenerator testingonsitein
order toreducethesupplycurrent. However, thisarrangementrequiresatransformer
of the full test voltage but not full current as well as a tuneable high-voltage
reactor. Tuned circuits arenot suitablefor tests during which excessivelosses can
develop, asinpollutiontesting[4]. Anappreciablevoltagedropmaybeproducedby
theincreaseinresistivecurrent near flashover.
6.1.2 Hi gh-frequency vol tages
High-frequencyvoltagesaresometimesusedfor testinglow-capacitancesamples, in
particularoverheadlineinsulators(seeSection7.1.1). Suchvoltagesmaybeproduced
by a Tesla coil circuit in which a charged capacitor is discharged by an air gap
throughtheprimaryof anair-coredstep-uptransformer. Thesecondaryistunedbya
secondcapacitor torespondtotheprimarycurrentoscillations. Thusaseriesof high-
frequency, high-voltageoscillationsareproduced[10]. TheHV windingcomprises
manyturnsof thinwire. Frequenciesinthehundredsof kilohertzrangeandvoltages
up to a million are achievable with low-capacitive loads. The application of the
circuitisverylimitedbecauseof theoscillatorypulsewaveformsandtheinterference
createdby therepetitivefiringof theair gap. Thesystemis animpressivetool for
demonstrationof thegenerationof veryhighvoltagesinthelaboratory continuous
sparkingcanbemaintained!
128 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
6.1.3 Ver y-l ow-frequency vol tages (VLF)
FollowingvariousinvestigationsattemptingtorelatethestrengthandPDcharacter-
isticsof cableandmachineinsulationat50/60Hztothoseatlowfrequency, anumber
of test supplieshavebeendevelopedoperatingat 0.1Hz.
VLFtestinghasbeenusedintheworldformanyapplicationsrequiringACtesting
of high-capacitanceloads. SinceHV DCtestscancausespacechargeanddamageto
XLPEinsulation, VLF hasbecomeawidelyadoptedtestingmethodforXLPEpower
cables. It isalsoaneffectivetool for burningdowncabledefectsintheinsulationas
well asaccessories.
It seemsthat thelow-frequencyPDcharacteristicsmaynot alwaysbesimilar to
thesystemoperatingfrequencies. Thiscouldbeduetothelargedifferenceinrateof
riseof theappliedvoltages, thehigher stress requiredfor inceptionandthelonger
timeof applicationtorecordsignificant voltagereversals. VLF testinghasprovedto
beof valuefor detectingelectrical treegrowthinXLPE insulation.
Differentprinciplesareinvokedfor productionof VLF voltagesincludinggener-
ationof cosine-squarewavesfollowedby polarity reversal givingacosinefunction
or, morerecently, VLF sinewavevoltagesgeneratedbyelectronicinverters.
6.1.4 Di rect vol tages
Direct-voltagetestingcircuitsarenecessaryfor checkingtheintegrityof AC paper-
insulated cables and generator insulation in addition to that for DC transmission
schemes. Thesimplest arrangement for outputsinthe100200kV rangeandtensof
milliampsisahalf-waverectifier (Figure6.6(a)) and, commonly, avoltagedoubler
(Figure6.6(b)) inwhichtheoutput istwicethepeak of theAC voltage. If required,
a smoother output can be obtained by incorporating a centre-tapped transformer.
For higher voltages, several stagesareincludedinmultiplecolumnscomprisingthe
capacitors, interconnected by solid-staterectifiers. A two-stagesystemwith asso-
ciated circuitry is depicted in Figure6.6(c). Theoutput is four times theAC peak
voltage.
Cascaded systems have been built for testing and research up to the voltages
requiredfor thevariousDCtransmissionschemes. Analysisof thesesystemsiscom-
plex andit is difficult to predict fromtheoretical considerations thepreciseoutput
voltageandripplecontent [1,4].
With large-capacitive loads a smoothing (reservoir) capacitor is probably not
required, as thetest object becomes thedominant component. A series-protective
resistor is provided in caseof external flashover or failureof thetest object. The
voltageoutput is measuredby means of ahigh-valuestableresistor connectedto a
meter calibratedinkilovolts. If theconfigurationisenclosedinaninsulatingcylinder
of high-resistivity material, careis necessary toavoidstatic chargingeffects onthe
surfaces. Inall direct-voltagetestingthemethodof earthingthetest object after dis-
connectingthesupplyisverycritical. For largecapacitancesitisprobablynecessary
todischargethetest object througharesistor of sufficient thermal capacitytoavoid
injurytopersonnel andtoprevent damageduetoasuddencurrent surge.
Basi c methods for i nsul ati on assessment 129
(a)
(b) (c)
Fi gure 6.6 Di rect vol tage testi ng ci rcui t: (a) hal f-wave recti fi er and wavefor m;
(b) vol tage doubl er ci rcui t; (c) vol tage quadr upl er ci rcui t
Portable DC test sets are available for low-current outputs at up to 100kV or
more with a low ripple and high stability. Such units incorporate high-frequency
(e.g. 80kHz) suppliesthroughasmall transformer.
6.1.5 Hybr i d test ci rcui ts
Asapossiblealternativetoresonanttestsetsandtoenableaformof partial discharge
measurements to be made on large capacitances, e.g. cables, hybrid circuits have
beendevelopedutilizingdirect-voltagechargingandfastswitchingtosimulatepower
frequencywaveshapesintheinsulationsystemunder test.
One method involves the slow DC charging (few seconds) of the test object
throughaseries inductanceandwhenat therequiredtest level shortingthesource
with a high-speed switch. Due to the low loss of the reactor and cable a number
of oscillationscanbeproducedwithout appreciabledecay at frequencieswithinthe
range50Hz to1kHz, dependingonthecircuit L andC values[11].
A similar systemalsoincreasesthedirect voltageslowly (within10seconds) to
2U
0
andsuddenly discharges thetest object to producetherequiredoscillations of
about 10msper half-period[12].
Inbothtechniques thetest voltageis appliedseveral times to obtainsignificant
partial-dischargeresults.
6.1.6 Li ghtni ng i mpul se vol tages
Themagnitudesandwaveshapesof thetransientvoltagesrequiredfortestingpower-
systemequipment arespecifiedinIEC 60071, asdescribedinChapter 1. Thebasic
130 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
V
o
(a)
(b)
C
1
C
2
V(t) R
2
V
o
C
1
C
2
V(t) R
2
V
1.0
0.9
0.5
0.3
0
B
A
T
T
1
T
2
T
1
= 1.67T
T = 0.3T
1

= 0.5T
O
1
T
Front
Tail
C
R
1
R
1
t
Fi gure 6.7 Basi c i mpul se gener ator ci rcui ts: (a) equi val ent ci rcui t; (b) defi ni ti on
of wave shape
requirements for producing, applyingandmeasuringsuchvoltages areincludedin
IEC 60060[S6/2].
In thecircuit of Figure6.7 thecapacitor C
1
is charged fromaDC sourceto a
preset value,V
0
, andthendischargedthroughanair gapintotheRC network of R
1
(frontresistor), R
2
(tail resistor) andC
2
, representingtheloadcapacitance, including
acapacitivedivider if connected. Thepositionof R
2
withrespect toR
1
modifiesthe
voltageefficiency. For thelightning impulsewaveR
2
R
1
and C
1
C
2
. The
output voltageV
t
isgivenby(6.1), inwhichK, and areconstantsdependingon
thecircuit parameters[1].
V
t
= V
0
K(e
t
e
t
) (6.1)
It will be noted that the equation represents the difference of two decaying
exponentials. Thestandardizedlightningimpulsewaveis includedin Table1.3of
Chapter 1.
Basi c methods for i nsul ati on assessment 131
6 4 4
C
C
R5
R3
4
R5
R3 R5
R3
7 8
R5
C
C
6
6
6
R
C
1
2
2 4
4
R3
6
4
9
Impulse generator Test
object
Voltage
divider
Measuring/
chopping
gap
Fi gure 6.8 Four -stage i mpul se gener ator and associ ated test l ayout modi fi ed
Mar x (Goodl et) ci rcui t: (1) DC suppl y wi th var i abl e i nput; (2) tr i gger
el ectrode and i nput; (3) chargi ng resi stor s; (4) front resi stor s; (5) tai l
resi stor s; (6) ai r gaps; (7) exter nal tai l resi stor ; (8) l oad capaci tance;
(9) protecti ve resi stor
Inpractice, multistagegenerators, asproposedbyE. Marxin1923, arenecessary
inordertoachievetheveryhighvoltagesrequiredfromrelativelylow-voltagesources
intherange100to400kV.Thelattervoltagetendstobethelimitbecauseof corona
andclearanceproblems. For simplicityanimpulsegenerator withonlyfour stagesis
includedinthetestcircuitinFigure6.8. Thefourcapacitorsarechargedinparallel and
thendischargedinseriesthroughtheair gaps. Theseriesconnectionisequivalent to
C
1
inFigure6.7. Thephysical configurationof thegapsandcapacitors togetherwith
thechargingresistors issuchthat theswitchingof all gapscanbeeffectedwithin
nanosecondtimes by triggeringthefirst gap, theothers beingof equal dimensions
butof slightlylarger spacings. Thisoccursbecausegapovervoltagesdevelopandare
maintainedfor tensof nsbythestraycapacitancestoground, betweencapacitorsand
acrossthegaps[1]. Theoperationof thegapsisfacilitatedif theyaremountedinline
of sight, allowingthelower gapdischargetoirradiatethoseabove.
Intheoriginal Marx circuit thelowest capacitor was not earthed, thus allowing
onesideof thefirst gapto beearthed. This simplifiedthelater development of the
triggeringarrangementsbutresultedinareversal inpotential betweeninputandoutput
voltages. Subsequentlyit wassuggestedbyB.L. Goodlet (J IEE, 1929) that oneside
132 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
of thefirst-stagecapacitor beeartheddirectly(notthrougharesistor), thusgivingthe
samevoltagepolaritiesand, incidentally, reducingthegenerator height (Figure6.8).
Such areduction may beof significant advantagein asmall test area, whereonly
threeor four stages might beinstalled. Very highvoltages of upto several million
areavailableinanumber of factoriesandlaboratoriesworldwide[4]. Thismight be
achieved, forexample, by28stageswithachargingvoltageof 100kVandcapacitors
of theorderof tensof microfaradstogiveamaximumoutputof 2.4MVatanefficiency
of approximately80per cent. Thecapacitorsmaybeinporcelainshellsseparatedby
similarspacersandstackedincolumnsorinmetal cansmountedwithintheinsulating
framework. Oneschemefor distributingthefront resistorsandtherebyreducingthe
voltage stresses across themis given in Figure 6.8. In addition, the tail resistors
aremoreeasily supportedwithinthegenerator structurethanexternally. Theuseof
distributedresistorsisrestrictiveif frequent wave-shapechangesarenecessary, but
thisdisadvantageisusuallyacceptablebecauseof insulationrequirements, clearance
reasonsandthelimitationof theeffectsof internal oscillations. External components
canbeaddedif necessary.
Insomeapplicationsthecircuitisallowedtooscillatetoproducehigher voltages
and/orawavewithspecifiedoscillations. A serieshigh-voltageinductorisconnected
intheoutput tocreateanoscillatinglightningimpulse(OLI) [1].
ThecircuitinFigure6.8includestheDCsupplyasdescribedinSubsection6.1.4.
Thismust besuitablefor polarity reversal by meansof abuilt-inswitchingsystem.
It is essential that anautomatic earthingswitchbeprovidedat thesupply terminal
direct voltages canlurk wherenot expected! Thetrigger electrodeis firedby an
external pulse, whichalsoallows synchronismwiththerecordingdevices. Thelat-
ter shouldcomply withtherequirementsof IEC Publication61083-1[S6/4]. CROs
and/or digital recordersmonitor theoutputof theHV divider andother transducersif
inuse. ThedividermaybearesistivetypewithanHV armof perhaps1020kilohms.
Thelowerarmmustbecorrectlymatchedwiththecoaxial cableandinputimpedance
of the recording instrument. Capacitance dividers of 1001000pF are often used
but allowanceis necessary for possibledistortionof theincomingsignal duetothe
LV capacitance/coaxial cable surge impedance [1,2,4]. Many investigations have
been carried out regarding theperformanceof dividers, especially with respect to
their responseto thehigh-frequency components of theimpulsewaves. No divider
is a pure resistance or capacitance. Various compensating techniques including
shielding and additional damping areused at thevery high voltages to obtain the
requiredaccuracies. AtUHV of several megavoltsthetall columnsandlongconnec-
tions may introduceadditional errors inthemeasurements. Combinedresistiveand
capacitivedividersareutilizedunder someconditionsinorder toobtainacceptable
responses[1,4].
A spheregapcanbeusedfor measurementor for chopping theoutputwaveata
predeterminedtime. Synchronizationbetweenthegenerator andmaingaptriggering
is necessary perhaps fromless than 0.5s (front of wavetesting) to tens of s.
Chopping times repeatableto within 0.1s or less may berequired. For voltages
aboveabout 500kV multiplechoppinggapsareavailablegivingmorereliablefiring
thanonelargegap.
Basi c methods for i nsul ati on assessment 133
A major problemwithimpulse-voltagetestingsystemsisthepossibilityof inter-
ferencecurrentpulsesbeinginduced, orflowing, intheearthcircuitsandmeasurement
cables. TheconfigurationindicatedinFigure6.8assistsinminimizingtheseeffects.
Thelatter arediscussedinIEC 60060-2[S6/2].
Theobjectiveof suchtestlayoutsistolimitthemaindischargecurrentof, perhaps,
thousandsof ampstoaverylowimpedancepathandtoarrangethedividersoitsearth
connectionsdonotdirectlyinvolvethehigh-currentpath. Thisconfigurationmaynot
besuccessful inparticular situations. Additional copper matsor thincopper foilsof
10cmwidthforearthconnections evenwithanunder-floorsystem maybehelpful.
Doublescreeningof cablestogether withpower isolationandgoodshieldingof the
control desk instrumentationisoftenessential. Experiencerelatedtothedesignand
operationof aparticular ultra-high-voltagelaboratoryisreviewedinReference4.
6.1.7 Swi tchi ng surge vol tages
Fortestingof themajorityof purelycapacitive-typeequipmentwithstandardswitch-
ing surges as specified in Table1.2 (Chapter 1), aconventional impulsegenerator
may beused, withappropriatemodificationof theresistivecomponents of theRC
circuit. The slow front of 200s requires an increased value of the front resistor
and/or anadditional capacitanceacrossthetest object. Caremust betakentoensure
thattheenergyratingsof thechosenresistorsaresufficienttoavoidfailurebythermal
overloading, whichmight result inflashover alongtheir length.
An alternative method [S6/2] of producing a switching surge is to induce the
required voltage in the high-voltage winding of a testing transformer, or a trans-
former to betested, by application of an impulseto thelow-voltagewinding. For
slow-impulserisetimesthetransferredvoltageinatransformer isapproximatelythat
of theturnsratio.
A warningintheStandardisgiventhatnon-disruptivedischargescanoccurinthe
test object duringswitchingsurgeapplicationsandmay result inappreciablewave-
shape distortion. The effect can be very significant in pollution tests on external
insulationat highvoltages. If thesourceisof highimpedancethedesireddisruptive
dischargemaynot develop.
6.1.8 Hi gh-vol tage equi pment for on-si te testi ng
On-sitehigh-voltagetestsareappliedduringthecommissioningstagetoensureequip-
ment isundamagedandcorrectly assembledwhennewandafter locally completed
repairsor invasiveinspections.
SuchHV tests arealsorequiredfor various diagnostic procedures, inparticular
partial-discharge and dielectric-dissipation-factor measurements. Some of the site
diagnosticrequirementsforindividual itemsof equipmentarediscussedinChapters8
and9.
All the sources considered in Sections 6.1.16.1.7 may be required on site.
However, thenormal laboratory/factory testingequipment is usually unsuitablefor
transportation, andspecial designsof portablesystemsarenecessary. Thesemaytake
134 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Fi gure 6.9 Exampl e of hi gh-vol tage test vehi cl e (250 kV testi ng tr ansfor mer and
anci l l ar y equi pment) [ photogr aph by Ener ser ve (Energy Austr al i a)]
theformof custom-built vehiclescontaining, for example, HV testingtransformers
andstandardcapacitors(Figure6.9), resonanttestsets, generator andLV transformer
for induced-voltagetestingof power transformers, direct-voltagesuppliesfor cable
testing(excludingtestsonextrudedcables) andsmallerunitsforseparatesourcetests
(50Hz, VLF andhybridsupplies) for thelower-voltagesystemsof 11/33kV.
It isnotedthat VLF techniquesareusedonsitefor withstandvoltagetestingof
oilpaperandextrudedcablesandforrotatingmachines, thusreducingthehigh-value
chargingcurrentsrequiredbyconventional non-resonantpowersources. Voltageout-
puts are of the order of 50120kV, although systems of higher value have been
described[13]. Thechoiceof frequenciesfor sitetestsisconsideredinReference9.
Impulse generators of modular design are required for testing GIS and some
transformersonsite. Itispossiblealmosttodoublethegenerator outputbyproviding
aseriesinductancebetweenthegenerator andloadandallowingdampedoscillations
todevelop bothfor lightning(OLI) andswitching(OSI) surges. Frequencyranges
for OLI andOSI voltagesaresuggestedinReference8. Inthisreport of CIGRE TF
33.03.04, recommendations aremadeas to howtheStandardrequirements of IEC
Publications60060-1and60060-2, 1994[S6/2], might bemodifiedtoallowfor on-
siteconditions. A number of references of developments arequoted in thearticle.
TheStandardspecifyingdefinitions andon-sitetestingrequirements was issuedin
2006[S6/2]. ThedetailsinTable6.1areextractedfromReference8andareaguide
astothetypesof on-siteteststhat maybespecifiedfor different itemsof equipment.
Clearly, animportant factor inthedesignof on-sitetestingcomponents is theneed
for greater mechanical strength than their counterparts in the laboratory and also
increasedawarenessregardingsafetyaspectswhensettinguptemporarytest areas.
Basi c methods for i nsul ati on assessment 135
Tabl e 6.1 Vol tage types and sources used for si te testi ng [ 8] [ reproduced wi th
per mi ssi on of CI GRE]
ACTC transfomer circuit for AC voltage generation
ACRL Inductance-tune resonant circuit for AC voltage generation
ACRF frequency tuned resonant circuit for AC voltage generation
applied in the past and no longer recommended 1)
mainly for PD diagnostics 2)
W withstand voltage test
PD voltage test with PD measurement
DM voltage test with dielectric measurement (mainly tan)
6.2 Non-destr uctive electr ical measur ements
Theprinciplesinvolvedinthenon-destructiveelectrical test methodsfor monitoring
andprovingtheconditionof insulatingmaterials insampleformandinequipment
areoutlinedbelow. Themajorityof thesemethodsarecalledfor invariousStandard
specifications. AlsoincludedinSection6.2.4aremorerecent, well-developedtech-
niques, whicharebeingappliedtodeterminetheir viability, especiallyfor assessing
theconditionof insulationonsite.
6.2.1 I nsul ati on resi stance (I R) measurements
The method measures the resistance of a dielectric or insulation structure when a
high-valuedirectvoltageisapplied. ThecommonestinstrumentistheMegger, which
consistsof ahand- or motor-drivenDCgenerator (D) upto5000V andabuilt-in
movingcoil instrument for measuringtheratio(resistance) of theappliedvoltage
andthecurrentflowing. TheprincipleisdepictedinFigure6.10, inwhichareshown
thecurrent coil (1) andvoltagecoil (2). Thetorquesproducedact inoppositesenses
suchthatthemeterdeflectionisdependentontheratioof voltagetocurrent. Thescale
isnon-uniform, beingmoreopenat thehigher-resistancevalueswherethecurrent is
small.
Although low-resistance paths in series with good insulation are not easily
detectableandparallel-patheffectscannotbeeliminatedwithinstructures, thissimple
test hasbeendevelopedby theindustry asavaluableassessment tool for indicating
insulationconditioninmanyapplications[14]. It doesnot correctlymonitor theAC
operatingconditions.
136 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
E
L 2
1
D
Fi gure 6.10 Pr i nci pl eof Megger ci rcui t: D gener ator ; 1 cur rent coi l ; 2 vol tage
coi l
100
10
1
2
3
4

1
0.1 1 10
Time (seconds)
Fi gure 6.11 Hypotheti cal cur rent - ti me cur ves dur i ng i ni ti al vol tage appl i cati on
A number of measurableresistanceparametersbasedonthevariationincurrent
following application of thevoltagemay beinterpreted as representing significant
changesintheinsulationcondition. Thetotal current (1) maybeconsideredascom-
prising three components the absorption current (2), the conduction (leakage)
current (3) andthecapacitance-chargingcurrent (4), asshowninFigure6.11.
Theteststhat might beappliedcanbedefinedasfollows.
6.2.1.1 Shor t-time test
Theshort-timetest is usually appliedfor 60seconds andrepresents onepoint ona
curveof increasingresistance, onlygivinganindicationof theinsulationcondition
(capacitancechargingeffect).
6.2.1.2 Time-r esistance test
For goodinsulationfreeof moistureanddirt, theresistancevaluewouldbeexpected
toincreasewithtimeuntil reachingafinal steady(IR) value. Thetime-resistancetest
Basi c methods for i nsul ati on assessment 137
iscarriedout for aperiodof 10minutesor more. Thisalsoallowsdeterminationof
thedielectricabsorptionratio(60-secondvaluedividedby30-secondvalue) andthe
polarizationindex(10-minutevaluedividedbythe1-minutevalue).
A satisfactory dielectric isoneinwhichthedielectric absorptionratioisgreater
than1.4 1.5andPolarizationIndex is greater than2.0. Correspondingvalues for
doubtful insulatingmaterialsareof theorder of 1.0or less.
6.2.1.3 Step-voltage test
Inthestep-voltagetest measurementsaremadeat twovoltagelevels, preferablyina
ratioof 1:5. If theIR valueat thehigher voltageislessthan75per cent of that at the
lower voltage, thentheinsulationissuspect. Thetest maybeconsideredsignificant
for timesasshortas60seconds, eventhoughtheabsorptioncurrentmaynotbezero.
6.2.2 Measurements of the di el ectr i c di ssi pati on factor (DDF)
Themeasurement of thedielectricdissipationfactor (formerlydescribedastheloss
tangent or dielectric lossangle DLA) iswell establishedastheprimary test for
diagnosingpossibledegradationanddeteriorationof equipment insulation inthe
factory, laboratoryand, insomecases, onsite. Thetechniquegivesameasureof the
bulklossesproducedunder AC conditions. It alsoenableschangesincapacitanceto
berecordedandcanbeappliedatoperatingandtestvoltagesif anappropriatesupply
andstandardcapacitor isavailable.
6.2.2.1 HV Scher ing Br idge
Traditionally, inthemajorityof high-voltagelaboratoriesand, forexample, inbushing
andcablefactories, theScheringBridge(devisedby H. Scheringin1919) hasbeen
usedfor measurement of DDF. Thebasiccircuit isgiveninFigure6.12andconsists
of two RC arms with a sensitive electronic detector, the latter type replacing the
much earlier vibrating galvanometer tuned to thetest frequency. Derivation of the
balanceequation[1,2] assumesthatC
X
(thecomponentunder test) hasanequivalent
serieslosselementandthatthestandardcapacitor C
N
isloss-free. Equivalentcircuits
andcorrectionsfor high-lossangles(>310
2
) areconsideredbySchwab[2] and
Zaengl [1]. For thelow-losscasetheunknowncapacitance
C
X
= C
N
R
4
/R
3
(6.2)
andtheDDF
tan = C
4
R
4
(6.3)
TheDDFvaluesforthemoretraditional insulatingmaterialsareintherangedown
to0.0010.003andaremeasurablebyanormal ScheringBridge. However, thismay
notbethecasefor thenewer materials, suchasXLPEandpolypropylenefilm, which
haveDDF values <0.001 (seeReference1, Chapter 2). Such sensitivities may be
achievableunder good conditions in thehigh-voltagelaboratory. This assumes no
corona, andprobablyrequiresdoublescreening, includinganinter-windingshieldin
138 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
HV
C
x
R
3
R
4
C
N
C
4
Fi gure 6.12 Pr i nci pl e of the hi gh-vol tage Scher i ng Br i dge
thetestingtransformer, careful screeningof thelow-voltagecomponentsandcables,
aswell asefficient earthing. For detaileddiscussionsconsult References1and2.
6.2.2.2 Tr ansfor mer r atio-ar m br idge
Forthirtyyearsormorecommercial transformerratio-armbridgeshavebeenavailable
for measuringtheDDF of insulation. Theseinstrumentsaremoresensitive, easier to
screenagainst stray effects andmoresuitablefor useonsitethantheHV Schering
Bridge. Automatedversionsareofferedfor applicationswheremanymeasurements
aretobemade, e.g. for oil testing. A circuit diagramof theprincipleof thebridgeis
giveninFigure6.13.
Themainpart of thebridgeis athree-windingtransformer or comparator. This
hasvery lowlossesandleakage(high-permeability core) andiswell shieldedfrom
stray magnetic fields. A major advantageisthat nonet mmf existsacrosswindings
W1andW2at balance. Also, it isunnecessarytoallowfor thestraycapacitancesof
thelow-voltageleadsandthoseof thewindingsasnovoltageappearsacrossthem
thusallowingtheuseof longleads[1]. AswiththeScheringBridgeahigh-voltage
standardcapacitor C
N
isrequired.
Allowance may be made for testing earthed and unearthed equipment. In the
former caseit must bepossibleto float thelower end of thesupply transformer.
For thegroundedconditiontwomeasurementsarerequired onewiththetestobject
disconnected(i.e. measurement of thestray capacitancetoearth, C

X
) andtheother
withC

X
andarmC
X
/R
X
inparallel. Thebalanceconditionisachievedbyvaryingthe
turnsratioof windings1(N
1
turns)and2(N
2
turns)togivezerooutput. Atampere-turn
balance, I
1
N
1
= I
2
N
2
, by adjustingR andC allowancecanbemadefor thephase
Basi c methods for i nsul ati on assessment 139
R
x
C
N
C
x
R
2 1
I
x
I
N
C
Fi gure 6.13 Tr ansfor mer r ati o-ar m br i dge for unear thed test obj ect
shift inI
X
duetothelossesinthetest object C
X
. Thelatter isrepresentedasaseries
circuit of R
X
andC
X
.
At balance with an ungrounded test object the values of C
x
and tan may be
derived. The equations for determining the magnitudes of the relevant currents,
neglecting the impedances of 1 and 2 and assuming an applied voltage of V, are
asfollows.
I
N
= V jC
N
(1+jCR)/{1+j(C+C
N
)R} (i)
Bydivisioninparallel paths:
I
2
= I
N
/(1+jCR) = V jC
N
/{1+j(C+C
N
)R} (ii)
I
1
= V jC
x
/(1+jC
x
R
x
) (iii)
At balanceI
1
N
1
= I
2
N
2
(iv)
Fromequations(ii), (iii) and(iv) andbyequatingreal andimaginaryparts:
C
x
= C
N
N
2
/N
1
(6.4)
tan = R(C+C
N
) (6.5)
Becauseof thevery lowimpedanceof thewindingsit ispossibletomakemea-
surements with a significant capacitanceacross winding 1. A valueof 2000pF is
allowableinaparticular test set. If thisisunspecified, appropriatechecksshouldbe
made, for examplewhenmeasuringviatheDDF tapof abushing.
Thestandardcapacitor, C
N
, isusuallyagastype(e.g. SF
6
), havinginsignificant
losses at full voltage and probably a capacitance of 100pF. These units are very
140 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Capacitance range:
10pF or less to
4F or more
V
a
to provide
electrical
stressing
Voltage range:
hundreds of volts
to hundreds of kV
q
a
range:
0.1pC to
100.000pC
V
a
- Voltage range:
V --- mV --- 100V pulses
V
a

q
a
C
a
Test
object
as C
a
A
B
Detector
to measure
discharge
q
a
= V
a
C
a
Fi gure 6.14 Bl ock di agr am showi ng PD measurement requi rements
expensive. For conditions withanearthedtest object, referenceshouldbemadeto
themanufacturersinstructionmanual.
6.2.3 Measurement of par ti al di scharges by el ectr i cal methods
Thepartial-dischargemeasurement requirementsareindicatedintheblock diagram
shown in Figure6.14. For simplicity, it is assumed that theapparent chargepulse
magnitudeis theparameter to bemeasuredThehighvoltage, V
a
, is appliedto the
test object representedbyanideal capacitance, C
a
. Partial dischargescreatedbythe
stressingduetoV
a
will appearacrosstheterminalsABasasmall pulsevoltageV
a
.
Therangeof pulsevoltages to bemeasuredmay beestimatedapproximately using
thesimplerelationship(neglectinganyparallel capacitance),
V
a
= q
a
/C
a
(6.6)
whereq
a
representsthedischargemagnitude(0.1pCupto100000pC) expectedin
thetestedequipment, thelatterperhapshavingcapacitancesintherangeof 10pF(cap
andpininsulators) to tens of microfarads or larger ingenerators, power capacitors
andcables. Thepulsestobemeasuredhaverisetimesrangingfromnstoseveral s
dependingontheparticular insulationstructure.
Thevaluesof pulsevoltages(V
a
) for arangeof magnitudesfor C
a
= 100pF
andaninceptionvoltageof V
a
= 10kV (RMS) aregiveninTable6.2. Theequivalent
energylevelsarealsoincluded. ItshouldbenotedthatC
a
representsthetotal effective
capacitanceacrossA andB.
Itisclearthatalarge-ratiovoltagedividerisneededatthedetectorinputif hightest
voltagesaretobeapplied. Usingasimpledividerwouldproduceanunacceptableloss
insensitivityduetoattenuationof thesmall pulses. Fortunately, theappliedvoltage
andthepulsestobemeasuredhavewidelydifferentfrequencyspectra 0/50/300Hz
Basi c methods for i nsul ati on assessment 141
Tabl e 6.2 I deal PD energy l evel s and pul se vol tages for
appl i ed i ncepti on vol tage of 10 kV (RMS)
For q
a
C
a
V
a
(rms) w(J oules) V
a
1pC 100pF 10kV 0.710
8
0.01V
100pC 100pF 10kV 0.710
6
1V
10,000pC 100pF 10kV 0.710
4
100V
comparedwithhundredsof kHzuptoseveral hundredMHz. Byuseof ahigh-voltage
couplingcapacitor this frequency differenceenables alow-impedancepath for the
pulsestoberetainedbetweentheterminalsAB andthedetector. At thesametimeit
providesahigh-ratiodivider for reductionof theappliedvoltagebytheadditionof a
high-passfilter at thedetector input.
Theformof thedetector varies accordingto theparameter to bemeasuredand
many papers on the subject have been published. Probably the more important
techniquescanbegroupedasfollows:
(a) measurement of individual apparent dischargemagnitudeusingaresistanceor
tunedcircuit input;
(b) useof radio interferencevoltage(RIV) techniques withresistiveinput giving
anintegratedcharge valueasameter reading;
(c) measurement of aquadraticmean, themeter responsebeingproportional tothe
integral of thesquaresof thedischargemagnitudes;
(d) measurement of losses due to multiple-discharge sites (e.g. dielectric loss
analyser).
Of these methods the first is probably the most widely used and accepted as
astandard test; higher sensitivity is achieved using atuned circuit input, because,
amongother advantages, the50Hz component canbereducedtoalower level. This
techniqueanditsapplicationtopractical caseswill bedescribed. Variouscommercial
systemsarenowavailablebut thesediffer only intheir versatility andsensitivity of
detection. Thecircuit developed by G. Moleof theERA [28] will beused as the
basis for thepresent discussion, but many of theconcepts havebeenusedby other
investigators. Muchof thereviewby Masonpublishedin1965is still of relevance
andincludesanumber of earlier references. Theuseof RIV measurementsmay be
retainedfor somecasesbutitisclear thatgrosserrorscanariseandthereadingsmay
havelittlequantitativevalueas PD measurements. Moreadvanced PD techniques
developed for particular conditions aredescribed in later chapters. Thesemethods
arenotincludedintherevisedversionof IEC60270[S6/5]. TheIEChascommenced
(2006) thepreparationof anewStandardcoveringunconventional partial-discharge
measurementsincludingUHF andacoustictechniques[S6/12].
142 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
LV HV
Supply
voltage
External
calibrator
(U
o
, C
o
)
C
a C
k
= 1000pF
C
q
PC
magnifier
and
analyser
C
q
Z
Input unit PD detector
Amp.
Int. cal.
Fi gure 6.15 Basi c par ti al -di scharge detecti on ci rcui t
6.2.3.1 Basic test cir cuit
Thebasiccircuit isshowninFigure6.15andassumesasimplecapacitiveload(test
object)withacouplingcapacitance, C
k
, forconnectiontotheimpedance-formingpart
of thehigh-passfilter andtunedcircuit intothedetector. Insomecasesthedetecting
impedancemay bemountedat thehigh-voltageterminal of C
k
andcoupledto the
detector throughanoptical link. Thedetectingimpedancemay belocatedinseries
withtheloadunder someconditions but this is not desirablewithlow-capacitance
samples as stray effects canbesignificant andit is impossibleinsomeother cases
becausethebottomendof theequipment ispermanentlyearthed.
The order of measurement frequencies of such general-purpose instruments is
for narrow-bandtypespossiblyof 500kHz 4kHz, andfor wide-bandsystems15
300kHzwithcapabilitiesof measuringmoderate-magnitudedischargesinverylarge
capacitances (upto 250F). A resolutionof 135discharges per 5ms is achievable
inaparticular equipment [15]. Theuseof oscillographic recordingis essential for
assistanceininterpretation, evenincommercial instrumentsusedforindustrial routine
work. Detailedvariationsof thebasiccircuitsmaybeascertainedfromtheliterature,
includingtheapplicationof computer-basedsystems for monitoringandanalysing
theoutputsof theanaloguedetectioncircuits. Oneof thefirstcommerciallyavailable
computerizedsystemswasdesignedasanadd-ontoexistingdetectorsandisdepicted
inFigure6.16[16].
6.2.3.2 Calibr ation pr ocedur es
Intheearliest detectionsystemsit wasnecessary toincludeacalibrationmethodin
ordertoattempttoquantifytheparameterbeingmeasured theapparentpartial charge
for arangeof differentconditions. Initiallytheconceptof aninternal calibrator was
introducedasindicatedinFigure6.15. Inthisarrangementaknownstepvoltagepulse
(intherangemV to 100V), e
q
, witharisetime<60ns is injectedthroughasmall
capacitor C
q
anditsresponsecomparedwiththat duetoalivepartial dischargefrom
theequipmentundertest(C
a
). Itmaybeshownthatforanidealizedcapacitivecircuit
(Figure6.17(a)) theapparent changeinchargeshouldbe
q
a
= e
q
C
q
(1+C
a
/C
k
) = e
q
C
q
if C
k
C
a
(6.7)
Basi c methods for i nsul ati on assessment 143
Fi gure 6.16 Ear l i er par ti al -di scharge detector wi th computer -based recordi ng and
anal ysi s system[ 16] [ pi cture by Uni ver si ty of NSW]
Howeverduringtheearliertestsonlargepowerequipmentdoubtsaroseaboutthe
possibleerrorsproducedbyavarietyof factors, includingtheinfluenceof samplesize,
straycapacitanceand, especially, thecomplexeffectsproducedbytheinteractionof
thefrequencycharacteristicsof thetestobjectandthemeasuringsystem. Theproblem
becameparticularlyacutewhentestinglargetransformersinthe1950s. Thisresulted
intheapplicationof thetechniqueof directcalibrationasstandardizedinIEC60270.
Themagnitudesof possibleerrorareconsideredinReferences17and18. Forthecase
of thesimplecapacitivenetworkandareferencepulsee
q
(Figures6.15and6.17(b)),
alow-valuecapacitor, C
0
, isconnectedat thetest object terminal andastepvoltage
pulse, U
0
, of risetime<60nsinjected.
For the direct calibration condition (Figure 6.17(b)) the apparent measured
dischargewithC
0
incircuit is
q
0
= U
0
C
0
= e
0
C
q
(1+(C
a
+C
0
)/C
k
) (6.8)
If C
0
/C
a
1( 0.1isspecifiedinIEC 60270)
q
0
e
0
C
q
(1+C
a
/C
k
) or q
0
ke
0
or k = q
0
/e
0
(6.9)
Thus, for thehigh-voltagetestq
a
= ke
q
, wheree
q
isthedetector readingandk
isthescalefactor. Inpracticekismorecomplexandsimplyrepresentsthefactor by
whichtheinstrument readingmust bemultipliedtodeterminetheliveapparent PD
magnitude, q
a
. Thesimplecapacitivenetworkindicatesthatfor C
k
C
a
thevalue
of kissmall andthereforethesensitivityof detectionincreases. C
0
mustberemoved
forthehigh-voltagetestsif notratedforthefull voltage, asisusuallythecase. During
suchcalibrations, whichshouldbecarriedout for eachloador circuit change, it is
144 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Applied
voltage
(if energised)
(a)
(b)
Low
internal impedance
Pulse
genr.
Pulse
generator.
Low impedance
V
s
- to discharge
detector
V
s
- to discharge
detector
C
q
C
K
C
a
V
C
q
C
K
C
o
C
a
V
a
C
o
C
a
e
q
e
q
U
o
Fi gure 6.17 Cal i br ati on of i deal i zed capaci ti ve ci rcui ts: (a) i ndi rect (reference)
cal i br ator ; (b) di rect cal i br ator
essential that theconnectionbetweenthecalibrator capacitor C
0
andthetest object
beshort andthat theleadfromthepulsegenerator tothecapacitor bescreenedand
correctly matched. This requirement may introducedifficulties when testing large
(tall) componentssuchastransformers, switchgear andbushingsbut isnecessaryin
order to obtain, as far as practicable, accurateand repeatablemeasurements under
adverseconditions. Theexternal calibrator is usually battery-drivenandshouldbe
capableof calibratingthesystemfor theexpectedhigh-voltagetestPDvalues. Fixed
magnitudesof 5pC, 10pC, 100pC andpossibly500pC for transformer testingmay
beappropriate. Therequirement that C
0
/C
a
0.1must bemet.
Theconfiguration in Figure6.16 allows immediateresponseto changes in PD
levelsduringhigh-voltagetestsbyadjustingtheamplifiergainandinternal calibrator
settings, therebyretainingtheoriginal calibration. Linearityover theexpectedrange
shouldbecheckedusinganexternal calibrator. Thealternativeprocedure(circuit) is
todeterminetheamplifier output intermsof pC/mm, pC/volt or apC meter during
thecalibrationandmaintainafixedgainfor thehigh-voltagetest. It is possibleto
Basi c methods for i nsul ati on assessment 145
10mV/div 200pVs/div
5ns/div
Pulse waveform
5ns/div
Integration result
Fi gure 6.18 Exampl e of cal i br ator check by i ntegr ati on method
usefixed attenuation ratios e.g. 1, 10, 100. In this systemthereis no direct
comparisonbetweenthelivedischargeandtheinternal reference(e.g. e
q
).
If adigital measuringsystemisinusethecalibrator mustbecapableof producing
pulses of adjustablerepetition rateup to limits imposed by theresolution timeof
thesystem. Also thecalibrator must beableto generatepulses at, say, 100Hz for
aspecified timeperiod in order to check that thedigital instrument will record an
equal number a deviation of 2 per cent is allowed in IEC 60270. The latter
contains aninformative Annex (E) givingguidelines for digital acquisitionof the
analoguepartial dischargesignals. A comprehensivereviewof thevariousproblems
andrequirementsispresentedinReference19.
Calibrationof thecalibrator is essential inorder to ensureacceptableaccuracy
during equipment tests. In Annex A of IEC 60270 it is specified that the charge
shall becompared with areferencecalibrator traceableto national standards. The
comparisonmay bemadewithaPD measurement systemor adigital oscilloscope
with integration capabilities. A typical response is depicted in Figure 6.18 for a
calibrator withC
0
= 1pF, voltage20V andloadingresistor 50. Inthiscasethere
isgoodcorrespondencebetweenthepC magnitudeof thepulseinjected, theareaof
themeasuredpulse(approximately) andtheintegratedvalueindicatedbythedigital
oscilloscope.
Forsomeyearsdevelopmenthastakenplaceintheapplicationof inbuiltcapacitor
probesfor measurementof PDsinGISequipment. Thetechniquehasbeenvalidated
inanEl ectr a report [20] andisincludedasacalibrationmethodinIEC 60270. See
Chapter 9.
A simplemethod for determining theapparent measurement circuit sensitivity
under live conditions is to connect across the test object a very low-capacitance
PD source. Various techniques are applied including the temporary connection of
a thin wire spike at the HV terminal. As suggested by Kreuger [17] consistent
discretedischargesat low-voltagecanbeobtainedwithasimpleconfigurationof a
needlemountedwithitstiplocatedat thecentreof anearthedhollowhemispherical
electrodeof 25mminternal radius. Dischargemagnitudes of 20pC or so at 1.7kV
canbeachieved, dependingonthepointradius. Pulsesof magnitudesof 25to250pC
146 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
for different point radii for avoltageof about 3kV arequoted. Thisarrangement has
beenof valueinmakingcomparisonsbetweendifferent systems. PDsinducedwith
thinwireor sharppointsalsoallowconfirmationof thepositionof thenegativepeak
of thepower frequencywavesontheinstrumentationdisplay.
6.2.3.3 Reduction of inter fer ence
Thereductionof interferenceoccurringat thelineterminal of atest object or plant
itemduringaPD test cansometimes beminimizedby useof abalancedcircuit or
apolaritydiscriminationcircuit. BothconfigurationsareincludedinIEC 60270and
areavailablecommercially. Theconventional ERA detector depictedinFigure6.16
is designed to operate froma centre-tapped transformer with dual inputs suitable
for connectioninabalancedcircuit andiscapableof effectingrejectionratiosof 30
to100. Anotherwell-establishedequipmentdevelopedbyKreugerandusedformany
yearshastwoseparateadjustableimpedancesallowingrejectionratiosof upto1000
underfavourableconditions[17]. ThediscriminationcircuitdevelopedbyBlack[21]
recognises thepolarity of thepulses at thedetectingimpedance. If thoseoccurring
simultaneously are of identical polarity they are rejected. These instruments have
beenappliedinswitch-yardsupto300kV. Anearlier CIGRE publication[22] may
beof assistanceinunderstandingsomeof theinterferenceproblemsassociatedwith
PDdetection. Moreadvancedtechniquesandrecentdevelopmentsfor eliminatingor
reducinginterferencearediscussedinChapter 9.
6.2.3.4 Pr esentation of PD measur ements
Intheearlier detectors for exampleasdevelopedbyMoleandlater byKreuger a
visual oscillographicdisplaywasincorporatedtoallowobservationandmeasurement
of theamplifieroutputandfordeterminingtheapproximatetimeof occurrencewithin
the power-frequency cycle. Additionally a meter reading of the largest pulse was
included.
Thisrelativelysimpleconceptandeffectivesystem withsomeminor changes
hasbeentheindustrynormfor 50years. However, inthe1960sminicomputersmade
it possibletorecordandanalysedata[23] that, previously, wererecordableonly by
bulkyPHA systemsor directlyfromtheoscilloscopescreen.
Thenewsystemsbecamemorepractical followingdevelopmentof microcomput-
ersuntil, inthe1980s, thenewPCsresultedinanupsurgeof portablecomputer-based
systemsinthe1990s. Thelatterinstrumentationallowscomplexanalysestobecarried
out althoughthegreatest valueisprobably theeasewithwhichtrendingof changes
inmaximumandaveragecharge, repetitionratesandpulse-power variationscanbe
recordedduringatestandfor futurereference. Manyattemptsatrelatingthemeasur-
ablePDcharacteristicstotheconditionswithinparticular insulationsystems asin
thepast[24] havebeenmade[25,26] butitisnotclear thatthiscanbeachievedfor
all circumstances[27]. Thisreferencealsosupportstheviewthatfor oil-impregnated
materialsthemagnitude, phaserelationshipsandrepetitionratesof thePDsmay be
moresignificant thanthemorecomplex parametersthat canbecalculatedfromthe
basic data. An exampleof acomputer record for adischarging sourceis given in
Basi c methods for i nsul ati on assessment 147
Figure2.13. CIGRE Brochure226 [26] presents theprobableresponsefor awide
rangeof conditionsinvolvingpartial dischargesinsamplesandequipment. Interest-
ingly, anumberof faultsinsubstationshavebeenidentifiedbyexcessiveinterference
onportabletelephones.
6.2.4 Di el ectr i c response measurements
Theapplicationof polarization/depolarizationmeasurementsforassessingthequality
of insulatingmaterialsisarelativelysimpleprocedurebutanalysisof thedataisvery
complex. Inthefollowinganindicationis givenof thetimeandfrequency domain
methodscurrentlyavailable. Detailsof thevariousanalysesrelatingmeasurementsto
moisturecontent, ageingandpossiblecontaminationwill befoundinthereferences.
A number of significantparameterscanbemeasuredatwell belowoperatingvalues.
Oneof theearliest instruments was that developedby Mole[28] at theERA inthe
early1950s. Thedi sper si on meter wasusedformanyyearsforindicatingthemoisture
levels in oilpaper systems, especially transformers. The advances in technology
and interpretation of dielectric theory haveallowed thedevelopment of improved
methods.
Possibilitiesfortime- andfrequency-domainmeasurementhavebeeninvestigated
indetail duringthepasttenyearsandanumber of instrumentsandexperimental sys-
temsdeveloped. Manytestsarebeingcarriedoutonequipmentwithoil-impregnated
insulationstructures, someonXLPE cablesandafewonsurgearrestersassociated
with HV proving tests [29]. A review of the theory and background of dielectric
response measurements is given by Zaengl [30], including 103 references. The
particular casefor transformersisconsideredinareportbyCIGRETF 15.01.09[31].
Inthefollowingareoutlinedtheparametersinvolvedandtheprinciplesof mea-
surement. Thevariousinstrumentsandspecial test proceduresarenot described, as
thenewer techniques arebeingtrialledandassessedwithrespect to their valuefor
monitoringparticular apparatus.
6.2.4.1 Time-domain measur ements
Low-valuedirect voltages areappliedto thetest object for aspecifiedtimet
c
fol-
lowed by achosen short circuit period beforeisolating thesample. Depending on
thetest procedureit is possibleto measurethevoltageor currents as indicated in
Figure6.19. Manyon-sitetestshavebeencarriedout attemptingtorelatethesevari-
ationstotheinsulationconditionof oilpaper transformers, currenttransformersand
cables(oilpaperandXLPE)usingexperimental methodsandcommerciallyavailable
instruments. Thetechniquesarenotuniversallyacceptedandcautionininterpretation
of resultsisnecessary, especially for transformerscontainingoil gapsinserieswith
thecellulosicinsulations[31].
Anadvantageof theproceduresisthat for knownconditionssignificant datacan
beobtainedfor lowelectric stressesoncompletedequipment. Thetestsareapplied
offlineandareverylengthy, andcaremust betakentominimiseinterference, asthe
quantitiesbeingmeasuredaresmall.
148 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3
2
1
U
1. Charging 2. Discharging 3. Recovery voltage
DC V.I
V/M
I
U
(i) (ii)
(iii)
Vr(V)
1000
(a)
100
10
1
0.01 0.1 1 10 100 1000
T(s)
(v)
4%
3%
2%
(iv)
Charging
Grounding
Return voltage
t t
d
t
c
t
c

t
c
log t
c
1
2
I
I
Polarisation
current
I
DC
U
(b)
E/m
1. Charging
2. Discharging
(i) (ii)
Depolarisation
current
Fi gure 6.19 Ti me-domai n measurements. (a) Recover y vol tage measurements
(RVM): (i ) Ci rcui t; (i i ) Vol tages; (i i i ) V
m
val ues for r ange of chargi ng
ti mes t
c
; (i v) Val ue of V
m
at t

c
. Sl ope of recover y vol tage for t

c
; (v)
Possi bl e var i ati on of char acter i sti cs wi th apparent moi sture content
i n oi l -i mpregnated paper. (b) Pol ar i zati on and depol ar i zati on cur rents
and vol tages (PDC): (i ) Ci rcui t; (i i ) Cur rents
Basi c methods for i nsul ati on assessment 149
(a) Recover y (retur n) vol tage measurement (RVM)
In theRVM mode, Figure6.19 (a, i and ii), aDC charging voltageis applied for
time t
c
, then usually short-circuited for time t
c
/2, followed by a recording of the
recovery voltagefor apredetermined time. Theparameters that may berelated to
themoisturecontent arethemaximumrecoveryvoltage, V
m
, thetimeof occurrence
of V
m
and the slope of the initial voltage rise. In a particular instrument this
procedureisrepeatedfor manyvaluesof t
c
andthecritical valueof t
c
(t

c
) determined
for themaximumrecovery voltageV

m
, together withtheslope

at thevalueof t

c
(Figure6.19(a, iiiv). Therangeof chargingtimes is from0.02seconds to 10,000
seconds. Theresultsaredescribedasapolarizationspectra [32].
Experimental testsindicatedthat arelationshipexistsbetweenthemeasuredand
deducedparameters andthemoisturecontent inoil-impregnatedpaper/pressboard.
However, various investigations haveshown that for transformers this assumption
mayintroduceerrorsinthemoistureestimatesbecauseof thedifferencebetweenthe
conductivities and dielectric constants of theoil within theducts and thoseof the
impregnatedpaper/pressboardsolids. Also, thephysical structurecangiveincorrect
estimates.
Initsoriginal formtheRVMtechniquehasbeenappliedtocables[30, Part2]. This
is especially thecasefor XLPE systems, wherehigh-voltagedirect voltagetesting
isnot allowedbecauseof possibledamagetotheinsulation. Theaimof theRVM is
todetect thepresenceof water treesbeforetheformationof tracking. Inthecaseof
shorter lengthsof thesehigh-resistivitycablesitmaybenecessarytoreducetheload
resistanceacrossthemeasurement devicebyuseof parallel resistorsof theorder of
tensof G. Another possibleproblemwhencomparingcableresultsistheinfluence
of the cable length on the time constants. It has been suggested that charging at
different voltagesmaygivesignificant informationastothepresenceof water trees.
(b) Pol ar i zati on and depol ar i zati on cur rent (PDC) measurements
Inthismethod(PDC) astepvoltageof perhaps1000voltsisappliedtothetestobject
for possibly 10000secondsfollowedby ashort-circuit withinashort timeinterval.
Thepolarizationanddepolarizationcurrentsarerecordedbyanelectrometer asindi-
catedinFigure6.19(b). Theprocedureenablesestimatestobemadeof thecondition
(moistureandageing) of twodielectrics withdifferent conductivities anddielectric
constants. AnanalysisispresentedinReference33andmoregeneral observationsin
theCIGRE TF 15.01.09report [31].
(c) I sother mal rel axati on cur rent measurements (I RC)
Another recently developedassessment methodis theisothermal relaxationcurrent
measurements (IRC) mostly for polymer cables. The IRC method determines the
changes of the polarization processes inside the insulation. The cable is normally
chargedtoaDC voltageof 10per cent or possibly higher of theoperatingvoltage.
Thevoltageisthenremovedandthecableshort circuitedfor fiveseconds, followed
bymeasurementof thedepolarizationcurrentdecreasewithtime, whichmaylastfor
thirtyminutes. Theproductof theinstantaneouscurrentandtimecanbeanindication
150 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
of thenumber of voids involvedandtheageingstageof theinsulation, but carein
interpretationisessential [30]. A formof thismethodisbeingappliedcommercially.
6.2.4.2 Dielectr ic fr equency domain spectr oscopy measur ements (FDS)
Anumberof investigationshavebeencompletedinordertodeterminewhethercapac-
itanceand loss changes over arangeof frequencies (e.g. 0.00011000Hz) might
indicatethepresenceof moistureintransformersandalsoinXLPE cablesIt ispos-
siblethistechniquemaybeapplicabletoAC cablesat moderatelyhighstresses. An
applicationof thetechniqueisgiveninReference34, whereafrequencyrangeof 0.1
to10Hz wasusedupto30kVpk.
6.3 Physical and chemical diagnostic methods
Therearemany non-electrical techniques for determinationof theconditionof the
insulatingmaterialsusedinpower equipment. Thepredominantapplicationisfor oil
andcellulosicmaterials, althoughnewtechniquesarebeingdevelopedfor assessing
thestateof, forexample, compositeinsulators, theby-productsof SF
6
decomposition
andagedXLPE cables. A major restrictiontoapplicationof themethodsisthelack
of access to thematerials whenbuilt into equipment. Samples of oil inthecaseof
oil-filledtransformers, switchgear, cables, bushings andof gas fromGI equipment
may beobtainedwithout harmful effects. However, theremoval of solidinsulation
isnormallyimpossiblebecauseof resultingdamagetotheinsulationstructureof the
apparatus.
6.3.1 I ndi cator s of i n-ser vi ce condi ti on of oi l paper systems
In-serviceindicatorsof theconditionof theoil-impregnatedmaterialswithintrans-
formers, tap changers, bushings and cables include the levels of concentration of
moisture, gas, furans and particles in theoil as well as its overall quality. During
refurbishment of power transformersit maybepossibletoremovepaper/pressboard
samples fromnon-critical locations for determinationof their moisturecontent and
degreeof polymerization.
6.3.1.1 M oistur e content in oil-impr egnated systems
A significant factor intheoperationof oil-impregnatedpaper systemsistheneedto
maintainlowmoisturecontents. Thesimplest techniqueistomonitor oil samplesby
means of thewell-establishedKarl Fischer method(IEC 60814, S6/6). This publi-
cationincludes tests for oil-impregnatedpaper andpressboard. Caremust betaken
whenremovingtherequiredoil samples. Theoil temperatureat thetimeof removal
shouldbenoted, together withthetypeof oil-coolingsystemif operative. Preferred
techniquesaredescribedinIEC publications60475[S6/7] andIEC 60567[S6/8].
Theoil-impregnated cellulosic material moisturelevels areestimated fromthe
oil valuesasindicatedinFigure6.20. Theseequilibriumcurvesweredevelopedby
Oommen in 1983. A goodreviewof thevariables andhistory of thederivation of
Basi c methods for i nsul ati on assessment 151
5.0
4.0
3.0
%

M
o
i
s
t
u
r
e

i
n

p
a
p
e
r
2.0
1.0
0
0 5 10 15 20 25 30
PPM moisture in oil
100C
90C
80C
70C
6
0
C
5
0

C
4
0

C
3
0

C
2
0

C
1
0

C
0

C
35 40 45 50
Fi gure 6.20 Oi l paper moi sture equi l i br i um cur ves (after Oommen) [ 35] [ repro-
duced by per mi ssi on of the I EEE]
relationshipsusedbytheindustry, includingtheoriginal researchbyFabreandPichon
(1960), arepresentedinreference[35]. Moisturecontents aremuchless than1per
cent at operating temperatures for new transformers through to 23 per cent for a
well-maintainedunit(withsilicabreather) andperhaps45per centafter thirtyyears
of service. Thelast valuecanleadtorapidageingandpossiblefailure.
If themoisturein thepaper and oil arenot in equilibriumand not at thesame
temperature, errorsintheestimatewill arisebut inpracticethetechniqueisfoundto
bevaluable.
Direct measurement (IEC 60814, S6/6) of paper/pressboardsamplesfromfailed
units e.g. current transformers areoftencarriedout duringpostmortem investi-
gationsandrefurbishments. Aswithoil samples, caremust betakenduringremoval
andrecordsmadeof thelengthof timeof exposuretoair (anditshumidity), theunit
historyandthepreciselocationwithinthestructure.
6.3.1.2 Dissolved gas in oil measur ements
DGA isprobablytheprimeonlinemonitoringtechniquefor detectionof developing
faults in power transformers and has been applied for morethan thirty years. The
gases considered to beof significancearehydrogen (H
2
), methane(CH
4
), acety-
lene(C
2
H
2
), ethylene(C
2
H
4
), ethane(C
2
H
6
), carbon monoxide(CO) and carbon
dioxide(CO
2
). Values vary fromafewppmto hundreds for thehydrocarbons and
fromseveral hundredstothousandsfor thecarbonmonoxide/carbondioxides. These
levelsaremeasuredinthelaboratory or possibly withinamobileunit usinggas
chromatographsmanufacturedfor thisspecificfunction. Theproceduresarerecom-
mendedinIEC 60567[S6/8]. Inorder toobtainreliableandrepeatableresults, it is
essential thattheoil samplingbecarriedoutaccordingtoaprescribedprotocol [S6/8]
152 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
but related to the particular equipment design: power transformers, tap changers,
bushings, instrument transformersor cables.
A number of onlinemonitoringdevicesareavailable, inparticular for detecting
changesinthehydrogenlevelsinpower transformers(Chapter 9).
Over theyearsvariousmethodsof interpretationhavebeenproposedfor identi-
fyingfaults intheinsulationstructure inparticular theratio system, oneformof
whichispresentedinIECPublication60599[S6/9]. Thecodedlistof faultsrelatedto
particular practical situationsisdiscussedbyDuval anddePabloinReference36and
extendedin37. Thedatabases andanalyses arebasedonextensivestudies by IEC
TC10andCIGRE TaskForcesof SC15. Thisworkwascontinuedtocover casesfor
whichexistingcriteriamaygiveanunsatisfactoryinterpretation. References38and
39includeextractscovering179caseswherefaultswereidentifiedbyinspectionand
relatedtothegaslevels. Alsotabulatedarenormal valuesfor gasestobeexpected
insatisfactorypower andinstrumenttransformers. Itappearsthatthemagnitudesare
considerablyhigherthanthoseallowedbymanyutilitiesbeforeadditional monitoring
isrequired. Althoughtherearesomedifficultiesininterpretation, thetechniquegives
agoodindicationof possibleproblems, especially if appliedperiodically asatrend
measurementbasedontheexperienceof theparticular authority. Examplesfromone
sourcearepresentedin Chapter 9andan advancedformof dissolved-gas analysis
utilizingfuzzy-logicmethodsdescribedinChapter 10.
6.3.1.3 Fur an measur ements
Thetechniqueutilizinghigh-performanceliquidchromatographyfordetectingfurans
in insulating oils is well established and is being applied by many utilities and
manufacturers worldwide. Thelevel of concentrations of different forms of furans
has been shown to be indicative of the deterioration of cellulose at normal oper-
ating temperatures and may berelated to anticipated lifeunder certain conditions.
Theconcentrationsof significanceareof theorder of 0.05mg/L(ppm) throughto
5mg/L(ppm) or higher for badlydegradedinsulation[38,39].
6.3.1.4 Par ticle counts
Theconcentrationof different typesanddimensionsof particlesinoil canbequan-
tified by use of such procedures as those specified in Standard S6/10 (see under
Section6.6below).
6.3.1.5 Oil quality
Many tests arespecified for ensuring insulating oils areof thequality required. A
number of Standards within the IEC describe the requirements and test methods,
inparticular IEC 60296[S6/11] andassociatedpublications. Themethods covered
include:
breakdowntests
oxidationdetermination
permittivity, DDF andresistivitymeasurements
Basi c methods for i nsul ati on assessment 153
samplingmethods
water determination
ISOStandardsincludedeterminationof
viscosity
flashpoint
pour point
In transformer practice the more important characteristics as indicators of oil
deteriorationareoftenconsideredtobeacidity, DDF andinterfacial tension[14].
6.3.1.6 Degr ee of polymer ization
Oil-impregnatedpapersamplesremovedduringrefurbishingcanbecheckedfortheir
degreeof polymerization. Thetemperatureprofilewithinatransformermaybedeter-
minedfromsuchmeasurementsaswell asdeterminingtheprobablelifeexpectancy
of theinsulation. DP values arein therange1000 for new units to 200 for those
near their endof life[38,39]. Problemshaveariseninobtainingcomparableresults
betweenlaboratories.
6.3.1.7 Other techniques
Theconditionof samplesof deterioratedmaterialsmaybeexaminedbymanytech-
niques including IR, X-ray, SEM, UV (for detecting waxed state) and molecular
changes[40].
6.3.2 Anal ysi s of SF
6
sampl es from GI S
Samples removedfromaGIS areanalysedby gas chromatography andmass spec-
trometry. Theobjectiveistodetect anySF
6
decompositionproductsthat might have
beencreatedby, for example, low-level partial discharges.
6.3.3 Sur face deter i or ati on of composi te i nsul ator s
Althoughtheusual methodof determiningpossiblesurfacedeteriorationof compos-
iteinsulatorsisby visual observationandperhapsleakage-current measurements, a
techniquehasbeendeveloped[41] inAustraliainwhichsliversof thesurfacemate-
rial areremovedfor detailedquantitativeanalysis. Byuseof aspecial hot stick the
methodhasbeensuccessfullyappliedunder live-lineworkingconditions.
6.3.4 Water treei ng i n XLPE cabl e i nsul ati on
Application of the newer non-destructive techniques such as described in Sec-
tion 6.2.4. has increased the possibility of detecting water trees in plastic cables.
Followingfailuresor suspectedfaultstheonlyeffectivemethodof investigationisto
cutoutlengthsof thecableandfromsuspectedregionscarefullypreparedthinwafers
for physical inspection. Thesemethodsarestill requiredtoconfirmthepresenceof
treesandthestateof thedielectric, althoughnewtechniquesarebeingdeveloped.
154 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
6.3.5 Ul tr asoni c methods for detecti on of par ti al di scharges
Althoughusuallynot consideredquantifiableintermsof electrical chargeor energy
parameters, thepresenceandrelativeintensityof partial dischargesincertaininsula-
tionstructurescansometimesbeidentifiedbynon-electrical techniques inparticular
themeasurementof thelevel andformof thesoundwavesemittedfromvarioustypes
of PDsources.
Thesimplest exampleis theapplication of ahandheld parabolic dish in which
is mounted a piezoelectric transducer (e.g. 40kHz) connected to a battery-driven
amplifier andsmall loudspeaker. Suchdeviceshavebeencommerciallyavailablefor
many years and arevaluablefor determining, for example, thelocation of corona
dischargesinsubstations, onoverheadlinehardwareandinhigh-voltagetest areas.
Techniques utilizing several transducers on the external surface of a power
transformer tank for detection and location of internal partial discharges are well
established for factory and siteapplications [42]. Extensions of thesemethods are
describedinChapter 9.
6.3.6 Mi scel l aneous techni ques
Thereareanumber of physical inspectionscarriedout onequipment duringmainte-
nanceroutinesthatcanbecategorizedasassessmentof theconditionof theinsulating
materials. Suchinspectionsincludenotingtheoil pressure(expansionof bellows) in
HVCTs, theodour of oil inanolder transformer, oil leakagefromtank gasketsand
fromvalves, sludgeformationontransformer windings, thesurfacestateof outdoor
insulatorsandthepresenceof fretting (dustproduction) inlargerotatingmachines.
6.4 Summar y
Thechapter givesanoverviewof themoreimportant methodsemployedinindustry
and laboratories for assessing the condition of insulation in power-systemequip-
ment as manufactured, on commissioning, during serviceand when undergoing
maintenanceor repair.
The coverage includes the techniques for generation and measurement of the
requiredhighvoltages, arangeof electrical non-destructivemethodsandanumber
of physical and chemical diagnostic procedures. The chapter content forms the
basis for the testing and more advanced condition monitoring systems discussed
inChapters710.
6.5 Refer ences
1. Kuffel, E., Zaengl, W.S., andKuffel, J ., Hi gh Vol tage Engi neer i ng Fundamental s
(NewnesPress, Oxford, 2000)
2. Schwab, A.J ., Hi gh Vol tage Measurement Techni ques (MIT Press, Mas-
sachusetts, 1972)
Basi c methods for i nsul ati on assessment 155
3. Kind, D., An I ntroducti on to Hi gh-Vol tage Exper i mental Techni ques (Vieweg
andSon, Braunschweig, 1978)
4. Hylton-Cavallius, N., Hi gh Vol tage Labor ator y Pl anni ng (Emil Haefely & Cle
AG, Basel, 1986)
5. Gallaghar, T.J ., and Pearmain, A.J ., Hi gh Vol tage Measurement, Testi ng and
Desi gn (J ohnWileyandSons, NewYork, 1984)
6. Parnell, T., Theinfluenceof circuit parameters onpower frequency flashover
tests, Tr ansacti ons of the I nsti tuti on of Engi neer s, Australia, vol. EM3, no. 2,
1961(and1964)
7. Train, D., and Giano Trink, N., Prevention of recovery overvoltage on high
voltage testing transformers, IEEE PES Winter Meeting, Paper T 73 064-1,
1973
8. WG 33.03 TF 04 CIGRE, Proposed requirements for high voltagewithstand
testsonsite, El ectr a, December 2000;(193):1321
9. Gockenback, E., and Hauschild, W., The selection of the frequency range
for high-voltage on-site testing of extruded insulation cable systems, I EEE
El ectr i cal I nsul ati on Magazi ne, November/December 2000;16(6):1116
10. Kuffel, E., and Abdullah, M., Hi gh Vol tage Engi neer i ng (Permagon Press,
London, 1970)
11. Gulski, E., Wester, F.J ., Smit, J .J ., Seitz, P.N. andTurner, M., Advancedpar-
tial dischargediagnostic of MV power cablesystemusingoscillatingwavetest
system, I EEE El ectr i cal I nsul ati on Magazi ne, March/April 2000;16(2):1725
12. Lemke, E., andSchmiegel, P., Complexdischargeanalysing(CDA) analter-
nativeprocedurefor diagnosis tests onHV power apparatus of extremely high
capacity, Proceedi ngs of the 9th I nter nati onal Conference on Hi gh Vol tage
Engi neer i ng, Paper 5617, Graz, Austria, 1995
13. Gronefield, P., andvonOlshausen, R., Averylowfrequency200kVgeneratoras
apreconditionfortestinginsulatingmaterialswith0.1ACvoltage, Proceedi ngs
of the 4th I nter nati onal Conference on Hi gh Vol tage Engi neer i ng, Paper 21.02,
Athens, 1983
14. Allan, D.M., J ames, R.E., McMullan, P.J ., et al ., AustralianandNewZealand
experienceintheapplicationof diagnostictechniquesforassessmentof insulation
conditioninhighvoltageapparatus, Paper 15/33 10, pp. 19, CIGRE, Paris,
1990
15. Mole, G., and Robinson, F.C., A versatile discharge detector Co-operative
Electrical Research, No. 21
16. J ames, R.E., Phung, B.T. and Blackburn, T.R., Computer-aided digital tech-
niques for partial measurements and analysis, Int. Symposium on Digital
TechniquesinHV Measurements, Toronto, Canada, October 1991, pp. 2.72.11
17. Kreuger, F.H., Par ti al Di scharge Detecti on i n Hi gh-Vol tage Equi pment (Butter-
worth& Co., London, 1989)
18. Liu, Z., Phung, B.T., J ames, R.E., Blackburn, T.R., and Ariastina, W.G.,
Optimizationinpartial dischargemeasurements, Proceedi ngs of Austr al asi an
Uni ver si ti es Power Engi neer i ng Conference (AUPEC), Curtin University of
Technology, Perth, Australia, 2001, pp. 40712
156 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
19. CIGRE SC 33 TF 33.03.05, Calibration procedures for analog and digital
measuringinstruments, El ectr a, October 1998;(180):12344
20. CIGRE SC15.03TF 15.05.02, Partial dischargedetectionsystemfor GIS: Sen-
sitivityverificationfortheUHF methodandtheacousticmethod, El ectr a, April
1999;(183):7587
21. Black, I.A., A pulsediscriminationsystemfordischargedetectioninelectrically
noisy environments, Proceedi ngs of the I nter nati onal HV Symposi um, Zrich,
September 1975
22. CIGRE, Elimination of interference in discharge detection, El ectr a,
1977;(21)
23. Austin, J ., andJ ames, R.E., On-linedigital computer systemfor measurement
of partial discharges ininsulationstructures, I EEE Tr ansacti ons on El ectr i cal
I nsul ati on, December 1976;11(4):12939
24. CIGRE WorkingGroup21.03, Recognitionof discharges, El ectr a, December
1969;(11):6198
25. J ames, R.E., andPhung, B.T., Development of computer basedmeasurements
andtheir applicationtoPDpatternanalysis, I EEE Tr ansacti ons on Di el ectr i cs
and El ectr i cal I nsul ati on, October 1995;2(5):83856,
26. CIGRE WGD1.11, Knowledgerulesfor partial dischargesinservice, El ectr a,
April 2003;(207):636(CIGRE Brochure226)
27. Lundgaard, L.E., et al ., Partial discharges in transformer insulation, CIGRE
TaskForce15.01.04, Paper 15-302, Paris, 2000
28. Mole, G., Improved test methods for theinsulation of electrical equipment,
Proceedi ngs of the I nsti tuti on of El ectr i cal Engi neer s, Section IIA, no. 3,
1953
29. Mardira, K.P., Saha, T.K., andSutton, R.A., Investigationof diagnostic tech-
niques for metal oxidesurgearresters, I EEE Tr ansacti ons on Di el ectr i cs and
El ectr i cal I nsul ati on, February2005;12:5059
30. Zaengl, W.S., Dielectric spectroscopy intimeandfrequency domainfor high
voltagepower equipment I EEE El ectr i cal I nsul ati on Magazi ne, 2003: Part 1,
September/October;19(5):519; Part 2, November/December;19(6):922
31. CIGRE Task Force 15.01.09, Dielectric response methods for diagnostics
of power transformers, I EEE El ectr i cal I nsul ati on Magazi ne, May/J une
2003;19(3):1218(seealsoEl ectr a, J une2002;(202):2536)
32. Bognar, A., Kalocsai, L., Csepes, G., Nemeth, E. andSchmidt, J ., Diagnostic
tests of h-v oilpaper insulating systems (in particular transformer insulation)
usingdcdielectrometrics, CIGRE Paper 15/33-08, Paris1990
33. Saha, T.K., and Purkait, P., Investigation of polarization and depolarization
current measurements for theassessment of oilpaper insulationof agedtrans-
formers, I EEE Tr ansacti ons on Di el ectr i cs and El ectr i cal I nsul ati on, February
2004;11:14454
34. Hvidsten, S., Holmgren, B., Adeen, L., andWettstrom, J ., Conditionassessment
of 12- and 24-kV XLPE cables installed during the80s. Results froma joint
Norwegian/Swedish research project, I EEE El ectr i cal I nsul ati on Magazi ne,
November/December 2005;21(6):1723
Basi c methods for i nsul ati on assessment 157
35. Du, Y., Zahn, M., Lesieutre, B.C., Mamishev, A.V., andLindgren, S.R., Mois-
tureequilibriumin transformer paper/oil systems, I EEE El ectr i cal I nsul ati on
Magazi ne, J anuary/February1999;15(1):11 20
36. Duval, M., and dePablo, A., Interpretation of gas-in-oil analysis using new
IEC Publication60599andIEC TC 10databases, I EEE El ectr i cal I nsul ati on
Magazi ne, March/April 2001;17(2):3141
37. Duval, M., A reviewof faultsdetectablebyGas-in-Oil analysisintransformers,
I EEE El ectr i cal I nsul ati on Magazi ne, May/J une;18(3):917
38. dePablo A., Interpretation of degradation models of furanic compounds,
CIGRE WG15.01TF 03(1997) (seealsoEl ectr a, 1997;(175))
39. Oommen, T.V., andPrevost, T.A., Celluloseinsulationinoil-filledpower trans-
formers: Part II Maintaining insulation integrity and life, I EEE El ectr i cal
I nsul ati on Magazi ne, March/April 2006;22(3):514
40. Hill, D.J .T., Le, T.T., Darveniza, M., andSaha, T.K., A study of degradation
of cellulosic insulation materials in a power transformer Part 1: Molecular
weight studyof celluloseinsulationpaper, Pol ymer Degr adati on and Stabi l i ty,
1995;48(1):7987
41. Birtwhistle, D., Blackmore, P., Krivda, A., Cash, G., and George, G.,
Monitoring thecondition of insulator shed materials in overhead distribution
networks, I EEE Tr ansacti ons on Di el ectr i cs and El ectr i cal I nsul ati on, October
1999;(6):61219
42. Harrold, R.T., Acoustic techniques for detecting and locating electrical dis-
charges, Engi neer i ng Di el ectr i cs, vol. 1, Chap. 10(ASTMTechnical Publication
669, 1979) (seealsoI EEE Tr ans. El ectr. I nsul ., vol. 11, 1976)
6.6 Standar ds r elated to basic test methods
S6/1IEC60052(Ed. 3.0, 2002): Voltagemeasurementbymeansof standardair gaps
S6/2IEC 60060: High-voltagetest techniques:
Part 1(Ed. 2.0, 1989): General definitionsandtest requirements
Part 2(Ed. 2.0, 1994): Measuringsystems
Part 3(Ed. 1.0, 2006): Definitionsandrequirementsfor on-sitetesting
S6/3 IEC 62067 (Ed. 1.1 2006): Power cables with extruded insulation and their
accessories for ratedvoltages above150kV (U
m
= 170kV) upto 500kV (U
m
=
550kV) Test methodsandrequirements
S6/4IEC 61083-1(Ed. 2.0, 2001): Instrumentsandsoftwareusedfor measurement
inhigh-voltageimpulsetests Part 1: Requirementsfor instruments
S6/5 IEC 60270 (Ed. 3.0, 2000): High-voltagetest techniques Partial discharge
measurements
S6/6 IEC 60814 (Ed. 2.0, 1997): Insulating liquids Oil-impregnated paper and
pressboard Determinationof water byautomaticcoulometricKarl Fischer titration
S6/7IEC 60475(Ed. 1.0, 1974): Methodof samplingliquiddielectrics
158 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
S6/8IEC 60567(Ed. 3.0, 2005): Oil-filledelectrical equipment Samplingof gases
andof oil for analysisof freeanddissolvedgases Guidance
S6/9 IEC 60599 (Ed. 2.0, 1999): Mineral-oil-impregnated electrical equipment in
service guidetotheinterpretationof dissolvedandfreegasesanalysis
S6/10ASTMD6786-02: Standardtestmethodforparticlecountinmineral insulating
oil usingautomaticoptical particlecounters
S6/11 IEC 60296 (Ed. 3, 2003): Fluids for electrotechnical applications Unused
transformersandswitchgear
S6/12 IEC 62478 (Ed. 1.0): High-voltagetest techniques: Measurement of partial
dischargebyelectromagneticandacousticmethods(newproject, 2006)
6.7 Pr oblems
1. Discuss the accuracy problems when applying (i) resistive and (ii) capacitive
dividersfor measurement of high-voltagelightningimpulses.
A resistiveunit hasahigh-voltagearmof 100k andalow-voltagearmof
50. Itisconnectedtomeasuringinstrumentationbya75coaxial cable. How
shouldthecablebeterminatedandwhat istheHV/LV measurement ratiofor a
1.2/50simpulsewave?
For acapacitivedivider of 100pF HV and0.2F LV arms, determinethe
ratioandappropriatecableterminations.
2. (i) Indicate why care must be taken when interpreting the readings froma
Megger-typeIR measurement.
(ii) Outlinetheadvantages of atransformer ratio armbridgecompared with
a Schering Bridge for measurement of insulation loss angles on power
equipment.
3. Derivetherelationshipsforthelossangleandcapacitancevaluesof theunknown
capacitor C
x
intheScheringBridgecircuit inFigure6.12.
During ahigh-voltagetest on newoil-impregnated bushing thepermanent
values of theBridgecomponents wereC
N
= 100pF, R
4
= 318. At balance
R
3
= 700 andC
4
= 0.086F. Thetest frequencywas50Hz.
Determinethebushing capacitanceand loss angle. Comment on thetan
valueif thetest wascarriedout at 1.05 operatingvoltagetoground. Refer to
S7/18of Chapter 7. What further actionmight betaken?
4. High voltage partial discharge tests are to be applied to a range of power
capacitors. Detail thetest andmeasuringequipment requiredincludinga100pF
couplingcapacitor.
Determine the relative sensitivities for tests on the following capacitors.
(i) 50pF, (ii) 1000pF and(iii) 10,000pF. If thedesiredsensitivityfor (iii) cannot
beachievedhowmight thetest circuit bemodified?
Chapter 7
Established methods for insulation testing of
specific equipment
Typetests, routinetestsandspecial tests
Insulators, surgearresters, switchgear, bushings, capacitors, cables, HV rotating
machinesandtransformers
Insulationtestsfor HVDC components
This chapter is concerned with existing acceptance and testing procedures for
assessingtheconditionof theinsulatingmaterialsandinsulationstructuresincom-
pletedhigh-voltagepower-systemequipment. Thetechniques(seeChapter6) chosen
forprovingthesatisfactorystateof aparticularitemwill dependontheformandmag-
nitudeof thetestvoltagerequired, thephysical sizeof thecomponent, itscomplexity,
theloading on thetest supply circuits and themonitoring/fault-detection methods
adopted.
Tests may bedesignatedas type (design), sampl e or routi ne, dependingonthe
formandnumber of itemsmanufactured. Insomeinstancesaprototype isproduced,
whichissubjectedtoawiderangeof testsbeforefull-scaleproductionbegins. Testing
scenarios vary widely withthetypeof product. For example, cap-and-pininsulator
samples of four or fiveunits may bechosen for test fromabatch of 2000 while,
with largeplant such as generators and transformers, aseries of type, routineand
commissioningtestsarecarriedout ontheone-off completedunit.
Thetestmethodsandthecharacteristicstobecheckedforeachtypeof equipment
areusually based on international and national standards formulated by experts in
theparticular fieldandsubsequentlyapprovedbyrepresentativesof users, manufac-
turers, testing laboratories and research organizations. Many of thesepublications
contain valuabletechnical information and application guides. Information on the
widerangeof Standards availablemay beobtainedby contactingIEC andvarious
national organizationssuchastheIEEE, ASTM, BSI, VDEandSA (seeAppendix2).
Insomefieldsnewnonstandardizedtechniquesarebeingtrialledbytheindustry.
160 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Inthefollowingaregivenexamplesof anumber of themoreimportantdielectric
test procedures coveredby Standards. Ingeneral terms, thedescriptions arerelated
toIECandSAA publicationsandareintendedonlyasaguideastotherequirements
for particular classes of apparatus. The aimof the presentation is to indicate the
complexityandtypesof extensivetestingnecessaryinorder tomaintainahighlevel
of operational reliabilityof thepower system. Itisessential torefer totheappropriate
Standard(s) listedattheendof thechapterinordertodeterminetheprecisedetailsfor
practical implementation of aparticular procedure. TheStandards areperiodically
updated.
Withtheexceptionof theinduced-voltagetestingfromthelower-voltagewinding
of transformers, separate-source high-voltage supplies of sufficient VA power are
necessarytoprovethepower-frequencywithstandstrengthsof theinsulationsystems.
In themajority of cases, thelatter behaveas predominantly capacitiveloads. The
capacitancealsodeterminestherequiredstoredenergyof thetest equipment needed
for carryingout theassociatedhigh-voltageimpulsetests. Special situationsinclude
wet testsandtestingof nonlinear devices. Anydirect-voltagesupplywill belimited
by theconductivity of theinsulation load, including current leakageover adjacent
surfaces.
Thepossiblerequirementsfor thevariousinsulationtestsareconsideredinrela-
tionto particular equipment. Thesequencebegins withthesimpler capacitive-type
systemsthroughtothemorecomplex involvingdistributedcapacitances, travelling
wavesandcapacitive/inductivedevices.
7.1 Over head line and substation insulator s
Thesimplestcapacitive-typecomponentsof thesupplysystemarethevastnumbersof
overheadlineandsubstationinsulatorsintheir different formsasdescribedinChap-
ter 4. Theseincludestring(cap-and-pin TypeB; long-rod TypeA), pin, linepost
andstationpostinsulatorsasspecifiedinthevariousStandards. Compositeinsulators
aredefinedinseparatepublications. All of thesecomponentshavelowcapacitance
values. Dueto thenumerous designs, thefollowingcangiveonly anindicationof
thetests that may bespecified. Becauseof thedevelopments in requirements and
materials, theupdatingof Standardsisacontinual process.
Theelectrical assessmentproceduresareslightlydifferentfor thevariousclasses
of insulator but oneor moreof thefollowingtestsarespecifiedasappropriate. The
basictest proceduresandwaveshapesaredefinedinIEC 60060[S7/1].
drypower-frequencywithstandvoltage;
drypower-frequencyvoltagewithsporadicflashovers;
high-frequency(100500kHz) flashover voltage;
drylightning-impulsewithstandvoltage;
50per cent lightningflashover voltage;
dryswitching-impulsewithstandvoltage;
wet power-frequencywithstandvoltage;
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 161
wet lightning-impulsewithstandvoltage;
wet switching-impulsewithstandvoltage;
power-frequencypuncturewithstandvoltage(andbreakdownvalueif required);
lightning-impulsepuncturewithstandvoltage
High-voltagepower-frequency and impulsetest resources arerequired to meet
theseconditions. Theusual testsarefor individual componentsbutshortstringsmay
beusedtoverifythecharacteristicsof unitsthat aretobeconnectedinalongstring.
Extensivemechanical strengthtestsaredescribedintheStandardsbut, sincethese
aremoreconcernedwiththephysical structure e.g. metal-to-insulationbondingand
interfacestrengthfor composites theyarenot consideredhere.
Type tests are carried out on one or more insulators in order to confirmthe
characteristicsof thedesign.
Sample tests areappliedto samples inagroupof, for example, four andthree
items selectedfromabatch(lot) of 2000after completionof theroutinetests. The
sampletestsarespecifiedfor insulatorswherethesolidinsulationpuncturedistance
is 50per cent of theflashover or creeppath(TypeB insulators). For stationpost
insulators1per cent maybechosenfromalot of 100to500.
Routine tests arefor checkingeachinsulator oncompletion.
Thephysical arrangements for testingthedifferent insulator configurations are
outlinedintherelevant Standard.
7.1.1 Porcel ai n and gl ass i nsul ator s (over head l i nes)
Theseinsulators includecap-and-pin and long-rod string types, pin typeand line-
post (Australia) type. Appropriateelectrical insulation tests for insulator units and
insulator strings/setsareincludedinReferencesS7/2andS7/3intheStandardsatthe
endof thechapter.
Type tests include dry impulse (1.2/50s) withstand voltage and wet power-
frequencywithstandvoltageapplications. Theimpulsetestsarecarriedoutaccording
toIEC 60060-1, preferably usingtheupanddown procedurefor determinationof
theflashover voltage. Bothpolarities areappliedunless it is evident that onegives
alower withstandvoltage. A wet withstand-voltagetest is appliedandif flashover
occursaretest isallowed.
Sampl e testsconsistof power-frequencypuncturewithstandvoltageapplications
withtheinsulator immersedinadielectricfluid.
Routi ne electrical testsarerequiredonlyfor TypeB insulators. Power-frequency
or high-frequency sources may beused. For thesetests thetesting transformer or
high-frequency sourcemust bestableanddesignedtowithstandthesuddencurrent
andvoltagesurgesproducedduringthemanyroutinetestscarriedout.
7.1.2 Cer ami c and gl ass i nsul ator s (post type i ndoor and outdoor )
Postinsulatorsaredesignatedaccordingtowhethertheirconstructionincludesasolid
cylindrical core(puncture-proof), acavitycorewithinternal integral ceramicbarrier
or acylindrical coreandinternal metal fittings withthesolidinsulatingmaterial of
162 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
at least half thelengthof theexternal flashover path(puncture-proof). For unitswith
thesolidlengthlessthanhalf of thetotal, andfor thepedestal-post typeinwhichthe
thicknessof thesolidissmall comparedwiththeexternal distance, electrical sample
androutinetestsarespecified. ReferencesS7/4andS7/5arerelevant.
Type testsareappliedtooneinsulator only. Theproceduresareasin7.1.1.
For indoor applications, dry lightning impulse, switching impulse and power-
frequency withstand-voltage tests may be required. For outdoor applications, wet
switchingimpulseandpower-frequencytestsmayberequired. Therelativetestlevels
for particular lightning-impulsevaluesaredetailedinReferenceS7/4. Becauseof the
variability in thebasic insulation levels thedry lightning-impulsevalueis chosen
as thereferenceand not thesystem-highest voltage. Notethat switching impulses
(250/2500s) wouldusuallybespecifiedfor unitsintendedfor operationinsystems
withhighest voltagesof 300kV andabove.
Sampl e puncturetestsmayberequiredonunitswherethesolidinsulatingmaterial
distanceishalf or lessthanthat of theflashover path.
Routi ne electrical testsarespecifiedfor thesamedesignsasthesampletests.
7.1.3 Composi te i nsul ator s for over head l i nes (str i ng and post uni ts)
Extensivedesi gn tests arecalledfor intheassessmentof compositeinsulators. These
involvedrypower-frequencyflashover testsfor boththeline[S7/6] andpost [S7/7]
insulators, in addition to steep fronted impulse (1000kV/s) applications for the
special conditions specified. Tracking tests are also required. The comprehensive
sequenceof mechanical, thermal andmoisturetests(includingleakagecurrent mea-
surements) maybefoundinreferencesS7/6andS7/7. Type tests includedrylightning
impulses, wetswitchingimpulsesandwetpower-frequencywithstandvoltages. These
areappliedtooneinsulator or insulator unitonly. Noelectrical testsarespecifiedfor
samplechecksandfor routineinspections.
WithintheStandard, voltagelevels aregivenfor anoptional radio-interference
voltagetest(RIV). NoRIV valuesarequotedbutthetestvoltageindicatesthevalueat
whichcoronawouldbeexpectedtoextinguishunder dryconditions. Acceptablelev-
elsareagreedbetweenthepurchaserandmanufacturer aguideforestablishinglimit
values is giveninCISPR 18-2, Amendment 1[1]. Detailedprocedures for obtain-
ingRI characteristicsfor dryhigh-voltageinsulatorsaregiveninIEC 60437[S7/8].
The specified reference frequency given in CISPR 18-2 is (0.5 0.05) MHz but
others in the range 0.52MHz may be selected based on previous practice. It is
expected that RI disturbances frominsulators will not normally affect television
reception.
7.2 Over head line and substation har dwar e
Although theStandards recommend air clearances for thedesign and operation of
overhead line systems and substations, it is often necessary to prove the perfor-
manceof aparticular configuration, especially at thehigher voltagelevels, where
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 163
switching-impulses or unusual pollutionconditions exist. Suchtestingincludes the
determinationof coronainceptionlevelsinordertoensurethatunexpectedflashovers
donotoccur andthatinterferencedoesnotaffectlocal radiotransmissionsandrecep-
tions. Thisinterferencecaninvalidatesomeof themonitoringsystemsbeingusedon
sitefor thedetectionandmeasurement of partial discharges.
In thelaboratory, levels can bequantified by PD measurements and theRadio
InterferenceVoltagemethods describedinCISPR 18-2. Usually, insulators andthe
HV hardwarewill becheckedsimultaneously. Themeasurementsaremadeutilizing
astandardCISPR test set as specifiedinCISPR 16-1:1993[2]. Thetechniquehas
beenappliedfor many years andis sometimes usedas apartial-dischargetest (see
IEC 60270).
Thediametersof overheadconductorsfor agivenvoltageandconditionarewell
defined, as in Reference 3. Sometimes, however, it is necessary to determine the
self-screening efficacy of multipleconductors and thesatisfactory performanceof
anewdesignof spacer (or spreaders), asinquadarrangements. Proposedelectrode
configurationsattheterminalsof bushings, switchgearandarrestersaswell asinsula-
torswill requirecheckinginthehigh-voltagelaboratoryat aboveoperatingvoltages
requiringappropriatecorona-freesupplysources.
7.3 Sur ge ar r ester s
Surgearrestersareavital part of thepower system. They must beshowntobeable
towithstandthesteady-statevoltageswithout deterioration, torespondcorrectly by
reducinganexcessivelightningor switchingsurgetoasafevalueandthentorecover
their previouspower-frequencystrength.
Theearlier unitsincludedsparkgapsandsiliconcarbideelementsrequiringtests
asdetailedinIEC60099-1[S7/9]. Thetestsforthelaterdesignsof metal-oxidegapless
unitsaredescribedinPart4of theStandard. Thetestingproceduresarecomplexand
extensive. In thefollowing aresummarized someof thetypeand routinetests for
provingof theinsulationrequirementsandV-I performance.
Typetests on thearrester housings includealightning-impulsedry withstand-
voltage test and a similar switching-impulse test both above the corresponding
protectionlevel. Inthelatter caseawettestisrequiredfor outdoor units. Thepower-
frequency withstandtest is appliedfor oneminutewithapeak valueof just below
thelightning-impulseprotectionlevel.
Residual-voltage-typetestsonthreecompletesamplearrestersdeterminethehigh-
est residual voltageat therateddischargecurrent. Fromtheresultsof other teststhe
switchingimpulseprotectionlevel for specifiedcurrentsmaybedefined.
Type tests are carried out in order to simulate current flow froma recharged
transmissionline. Long-durationcurrent-impulsewithstandswithvirtual lengthsof
the wave peak within the range 5003000s are applied. A simplified circuit is
describedinAppendixJ of IEC 600994.
Operating-duty-typetestsincorporatingthesimultaneousapplicationof impulse
andpower-frequency voltages arealso includedintheStandard. A major aimis to
164 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
provethat thermal runaway does not occur andthat theunit cancool downduring
thepower-frequencyperiods. A special multi-pulseoperating-dutytest representing
multiplelightningstrokeshasbeenproposed[4, S7/10]. Thishasbeenshowntofail
someunitsthat wouldhavepassedthenormal requirements.
Routine reference voltage tests require determination of the power-frequency
voltagenecessarytoproducethereferencecurrent asmeasuredat thebottomof the
unit. Thevoltagevaluemust bewithintherangespecifiedbythemanufacturer. This
isaparameter that maybeusedfor monitoringinservice.
Routinelightning-impulseresidual-voltagetestsarecarriedoutat, preferably, the
nominal dischargecurrent. Thevoltagemust not exceed theappropriatespecified
valuesfor acompletearrester typicallyof theorder of 3theratedlinevoltage.
Aspartof theroutinetestsapartial-dischargecontactnoisemeasurementismade.
Theinternal partial discharges, including disturbances dueto bad contacts, should
not exceedtheequivalent of 10pC. It may bepractical to monitor theconditionof
suspect unitsinservicebymeansof PDmeasurements.
7.4 Switchgear
Inthissectionswitchgear isdeemedtoincludecircuitbreakersfor on-loadswitching
and fault protection, disconnectors (operating off-load) for isolation of the high-
voltage circuits, metal-enclosed switchgear and on-load tap changers installed in
transformersfor changingthesystemvoltagelevels. For all of thesedevicesaspecial
requirementistheprovingof thedielectricstrengthacrossopencontacts, inaddition
totheusual insulationovervoltagecheckstogroundandbetweenphases.
7.4.1 Ci rcui t breaker s
Extensiveelectrical, thermal andmechanical testsarespecifiedfor typeandroutine
testingof circuitbreakers[IEC60694, Ed. 2.2B,and62271-100, Ed.1.0B,RefsS7/11
andS7/12]. Theprimarydielectrictestsaresummarizedbelow. Thepower-frequency,
lightning-impulseandswitching-impulsewithstandtest voltagesaretabulatedinthe
specifications.
For 245kV classthephase-to-phase, phase-to-earthandopen-switchwithstand
testvoltages(one-minutepowerfrequencyand1.2/50simpulse) areequal, withthe
valuesfor theisolatingdistancesapproximately1015per cent higher.
Above245kV class thepower-frequency test valueacross theopencontacts is
higherthanthephase-to-earthlevels; theswitching-impulsetestlevel betweenphases
is greater thanthevalueto earth; and, also, acombinedswitching-impulse/power-
frequency test is required for checking the isolating distances. A combined
lightning-impulse/power-frequency withstand-voltage test is applied across open
contacts/isolatingdistances.
Onemethodfor applyinglongitudinal testvoltageswithoutexceedingthewith-
standlevel toearthis by applicationof twoseparateout-of-phasepower-frequency
sources and, for impulse tests, an impulse of the withstand value to earth to one
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 165
terminal withacomplementaryvoltageof oppositepolaritytotheother terminal. If a
secondimpulse-voltagesourceisnotavailable, thepeakof thepower frequencymay
beutilized. ThetechniqueisdescribedinIEC 60060-1.
If asecondvoltagesourceis not availablefor open-switchor isolatingdistance
tests, onevoltagemaybeapplied, but thevaluefromthetestedterminal totheframe
of thedevicemust not exceedthenormal withstandlevel. Therequiredlongitudinal
testvoltageisachievedbyincreasingtheframepotential toaboveearthandearthing
theother terminal of theopencontacts.
Thehigh-voltagelaboratory or factory test facility must becapableof carrying
out thesevariousprocedures, themajorityof whicharetypetestsonly.
7.4.1.1 Type tests
For 245kV-ratedvoltageandbelow:
power-frequency withstand-voltage tests for one-minute dry (indoor), wet
(outdoor);
lightning-impulse withstand-voltage tests with voltages of both polarities
(1.2/50s) drytestsonly.
For above245kV-ratedvoltage:
power-frequencyvoltagewithstandtestsfor one-minutedrytestsonly;
switching-impulse withstand-voltage tests with voltages of both polarities
(250/2500s) dry(indoor), wet (outdoor);
lightning impulse withstand-voltage tests with voltages of both polarities
(1.2/50s) drytestsonly;
Partial-dischargetests;
(PD tests may bespecifiedas agreedbetweenmanufacturer andcustomer. The
tests areof importancefor GIS equipment for which thestandard method (see
Chapter 6) or theUHF measurement systemsmaybeused(seeChapter 9).)
radiointerferencevoltagetest (RIV).
(RIV tests may be specified for equipment rated at 123kV and above. The
measurements aremadewith thecontacts in both open and closed states with
all other terminals earthed. Theprocedures for switchgear aredetailed in IEC
60694. An RIV versus applied voltage characteristic curve is recorded. The
acceptance level is 2500V at 1.1 phase-to-earth voltage. No guidance
is given as to possible equivalent levels measured by conventional PD test
methods. Such information could be of value when monitoring equipment in
substations.)
7.4.1.2 Routine tests
Short-termpower-frequencywithstandtestsarespecifiedfor newdryconditionson
thecompleteapparatus or individual terminals. If solid-coreinsulators providethe
major insulation, thehigh-voltagetestsmaynot benecessary.
166 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
7.4.2 Sel f-protected swi tchgear
Switchgear containing built-in surge arresters cannot be tested at the full power-
frequency withstand voltages. Methods for determining appropriate test levels as
relatedtooperatingvoltagesaregiveninIEC60694, AnnexF. Theimpulsetestsare
appliedwithprospectivevoltagesaslistedinthenormal withstandvoltagetables. No
disruptivedischarges shall occur andrecords areto beprovidedof thewaveshapes
of thevoltagestoearth.
7.4.3 Di sconnector s (i sol ator s)
Thedielectric withstandtests for disconnectors andearthingdevices aresimilar to
thosespecified for circuit breakers. Somedifferences in detail areincluded in the
relevantstandard[S7/13]. Problemsthathavearisenconcerningtheflashover of SF
6
units haveresulted in thedesign of test circuits for simulating thehigh-frequency
overvoltagesproducedbymakingandbreakingoperations.
7.4.4 Metal -encl osed swi tchgear
Specificationsareavailableforthedielectrictestingof switchgearassembliesinwhich
thevariouscomponents for exampleanumber of circuit breakers, bushings, cable
terminations, instrument transformers, busbars, protective devices and associated
control gear aretotallyenclosedbyanearthedmetal enclosure. Accessisprovided
forhigh-voltageandotherconnections. Forsuchequipmentfactoryandsitetestsmay
berequiredand, whereappropriate, individual itemstestedbeforeassembly. Thefinal
testsonthecompletedenclosureincludethefollowing.
Type tests to check the insulation level of the complete systemand partial-
discharge measurements to evaluate the quality of the inbuilt insulation. IEC
62271-200[S7/14] details thevarious high-voltagetests necessary to cover the
manytypesof ACdesigns, toearth, betweenphasesandacrossisolatingdistances.
ImpulsetestsarerequiredaccordingtoReferenceS7/11.
Routinetestsatthemanufacturersworksshouldbecarriedoutoneachassembly
andoncomponentsinstalledat siteasagreed. Thepower-frequencytest voltage
istobelessthanthefactoryvalueunlessaroutinetestwasnotappliedpreviously.
A PDtest mayalsobenegotiated.
InAnnex B of ReferenceS7/14aPD test accordingto IEC 60270is specified.
Thedifficulties inselectingapass PD valueareemphasizedand, again, shouldbe
agreed. Suggestedacceptancevalues are10pC at 1.1U
r
/

3whereU
r
is therated
linevoltage. Asanexampleof thevalueof PDmeasurementsonsiteafaultduetothe
incorrect assembly of acurrent transformer was locatedina33kV metal-enclosed
switchgear unit. PDvalueswereof theorder 100pC or more.
ItisnotedthatIECStandardshavebeenissuedforinsulation-enclosedswitchgear
[S7/15] and for gas-insulated metal-enclosed switchgear for rated voltages above
52kV [S7/16].
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 167
7.4.5 Tr ansfor mer tap changer s
An oil-immersed tap changer consists of many contacts mounted on an insulation
boardor cylinder, whichmaybenear earthpotential (ClassI) or connectedatahigh-
voltagepositioninthetransformer winding(ClassII). Inadditionadiverter switchis
requiredforchangeoverbetweentheselectorcontacts. Thetestvoltagestobeapplied
betweenthevariouselectrodesarespecifiedbythetransformer designer.
The dielectric-type tests according to IEC 60214-1 [S7/17] consist of power-
frequency(oneminute) andlightning-impulse(threeof positiveandthreeof negative
polarity) voltagewithstandtests betweenall liveparts andearthfor ratedvoltages
below300kV. For voltagesof 300kV andabove, switching-impulseapplicationsare
alsospecified, consistingof threefull wavesasinthetransformer standard(7.10).
Thewithstandvoltagelevelsbetweenpartsmust besuitablefor thevoltagedif-
ferences power frequencyandimpulse appearingacrossthetransformer tapping
leads. Duringtheappropriatetestsallowancemustbemadeforthelimitationsimposed
byanytransient voltagecontrol deviceswhichmaybeincludedinthetapchanger.
For Class II units for operationintransformers withaU
m
121kV, apartial-
dischargetest between liveparts and earth is required as atypetest. Thedetailed
procedureis specified in thestandard and is similar to thepower-transformer test
requirements. TheacceptablePDvalueat 1.5ratedvoltagetoearthis50pC.
Norouti ne high-voltagetestsarespecified.
7.5 Bushings
Theinsulationsystemsof high-voltagebushingsvaryaccordingtothematerialsand
configurationadopted. Withstand-voltagetestsarerequiredasroutineteststogether
withDDFandPDmeasurements.. Suchdataareusedforqualitycontrol andforfuture
referenceinservice. Thetest voltagemagnitudesaredeterminedfromIEC 60071-1,
dependingontheparticular application. Refer toIEC 60137[S7/18] for details.
For outdoor bushings of lower voltageratings awet power-frequency voltage-
withstand-type test is applied for one minute. Dry lightning-impulse voltage-
withstand-typetestsarecarriedoutonall bushingswith15impulsesof bothpolarities.
Dry-switchingimpulse-voltage-typetestsarerequiredfor indoor unitsof thehigher
voltageratingsandsimilar wet testsfor outdoor bushingsof theseratings. Thecon-
dition of the insulation is monitored by DDF, capacitance and PD measurements
after testingand, also, byrecordingof impulsewaveshapesduringtestsif considered
necessary.
DDF measurementsaremadeasroutinefor voltagesupto1.5ratedvoltageto
ground. Valuesat2.5kV and10kV arerecordedasbeinguseful duringlater service-
maintenanceprocedures. MaximumallowableDDFvaluesat1.05ratedvoltageare
quotedaccordingtothevariousmaterials, i.e. foroil-impregnatedpaper, resin-bonded
paper, resin-impregnatedpaper, cast resinandinsulatinggas incapacitance-graded
bushingsandcast resininnon-capacitance-gradedunits. AllowablechangesinDDF
andcapacitancemagnitudesarespecified. Withcompositebushingsvaluesaretobe
provided.
168 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Routinedry, power-frequency, one-minute, withstand-voltagetestsareappliedto
all bushingsandpartial-dischargetestsat just aboveratedvoltage. If requiredfor a
transformerinstallationthevoltageisincreasedto1.5ratedvoltage. PDacceptance
valuesare10pCfortheseconditionsexceptforresin-bondedandcast-resininsulation
systems, wherelevels of 100/300pC and 100pC respectively areusually taken as
reasonablelimits.
Theinsulationof thebushingtest tapmust becheckedby anappliedvoltageof
2kV andtheDDF andcapacitancevaluesmeasured. Thesevaluesshouldbe100mR
and5000pF or less. Thelower thetapcapacitancethemoresensitiveisthePDtestas
carriedoutaccordingtoIEC60270fortransformers(7.10.2)andforsomeswitchgear.
Whentestingfor special applications it is important to usecorrect terminations
(e.g. for re-entranttypes) andtoensurethatclearanceswithinthetesttankaretypical
of thoseintheequipment inwhichthebushingis to beinstalled. Air terminations
may need to bescreened, rod gaps removed and attention begiven to themethod
of mounting/clampingtheearthedflangetothetest tank. Temporary clampscanbe
a source of corona. The measurement of low-value PDs requires the usual good
housekeeping andpreferably, for routinetests, adedicatedareawherethePDcircuit
layout canbepermanent.
Guidance is given in IEC 61464 [S7/19] regarding the interpretation of DGA
resultsfor in-serviceoil-impregnated-paper bushings. Normal sitetestsincludeDDF
andIR measurements.
7.6 High-voltage instr ument tr ansfor mer s
The tests required for the wide range of instrument transformers applied in the
power systems aredetailed in IEC 60044 Parts 18, which includes current trans-
formers (CTs), voltage transformers (MVTs and CVTs), combined transformers
(CV/VT), andelectronic types. Thedevelopmentsininstrument transformer design
havenecessitatedupdatesof thestandard.
7.6.1 Cur rent tr ansfor mer s
Thehigh-voltagewithstandtestsonCTsareappliedtotheparalleledprimarywinding
terminalswiththeouterscreensandsecondarywindingsearthed. Themajordielectric
testsaresummarizedbelow. ThetestvoltagevaluesarebasedonthoseinIEC60071
andtabulatedinIEC 60044-1[S7/20].
7.6.1.1 Type tests
L ightning-impulse voltage-withstand test: For U
m
< 300kV. A maximum
of 15 standard impulses of each polarity is to be applied and for U
m
300kV
onlythreestandardimpulsesof eachpolarityarerequired. Waveshapechangesmay
indicatefailure.
Switching impulse-voltage withstand test: For U
m
300kV 15 switching
impulsesof positivepolarityaretobeapplied. Thetest istobeunder wet conditions
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 169
for outdoor units. The failure criteria are similar to those for the U
m
< 300kV
lightningimpulsetests.
Wet power-fr equency withstand-voltagetest: ForU
m
< 300kVthetestvoltage
istobeappliedfor 60seconds.
Radio-inter fer ence voltage measur ement: RIV testsarecarriedout onCTsfor
useinair-insulatedsystemswithU
m
123kV.Duringthetesthardwareisrequiredto
simulateoperatingconditions. RefertoIEC60044, Part1, fordetails. Byagreementa
PDtestmaybesubstitutedduringwhichthemeasuredvaluemustnotexceed300pC.
It shouldbenotedthat theRIV andPDmagnitudesarenot directlyrelated.
7.6.1.2 Routine tests
Dr y power-fr equency withstand voltage: The test voltage is applied for 60
seconds between the shorted primary terminals and earthed shorted secondary
windings/frame. Theappropriatetest levelsaretabulatedinIEC 60044-1.
Par tial-dischar ge measur ement: Partial-dischargetests areapplicabletounits
of 7.2kV. Thetest methodisspecifiedinIEC 600270. A PDlevel of 5pC should
bedetectable. Thetest maybeperformedwithoneof thetwoprocedures: (i) during
reductionof thewithstandvoltage; (ii) after thewithstandtestbypre-stressingfor 60
secondsat 0.8withstandvoltagefollowedbyreductiontothePDtest voltage.
ThePDtestvoltageismaintainedfor 30seconds. TheallowablelevelsatU
m
and
1.2U
m
/

3are10pCand5pCrespectivelyfor liquid-immersedunitsand50pCand
20pC for solidtypes. Thesystemneutral istobeearthed.
7.6.1.3 Special tests
Chopped impulse tests on pr imar y winding: Negative standard lightning
impulse test voltages with chopping times between 2s and 5s are applied for
U
m
< 300kV andU
m
300kV conditions. Variationsinthefull-wave(FW) shapes
beforeandafter thechop-wave(CW) applicationsmayrepresent aninternal fault.
M ultiple chopped impulse test (U
m
300kV): Note: This test is includedas
aninformativeAnnex(B) of IEC 60044-1proposingapossibleprocedurefor deter-
miningtheresponsetohigh- frequencystressesasmaybeproducedby, for example,
isolator switching. Dissolved-gas-in-oil analyses (IEC 60599) beforeand after the
test, withathree-dayinterval, aresuggestedfor monitoringfailure.
Capacitance and dielectr ic dissipation factor (DDF) measur ements: These
measurements apply only to liquid-immersed units for U
m
72.5kV. Values in
therange10kV to U
m
/

3 areto berecorded. Theaimis to check thequality of

production. Measurementsaretypically<0.005, possibly3mR or less.
7.6.2 I nducti ve vol tage tr ansfor mer s
The dielectric test requirements for inductive-type transformers are considered
separately from the capacitive configuration. The relevant standard is IEC
60044-2[S7/20]. All thespecifications in relation to insulation tests aresimilar to
thoseintheCurrent Transformer specificationwiththefollowingexceptions.
170 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
7.6.2.1 I mpulse withstand tests
During the impulse tests records are required of earth currents and of voltages
transferredto thesecondary windings. Thesearenecessary inorder to improvethe
sensitivityof faultdetectionasintheproceduresfor power transformer testing. Also,
it maybeessential toreducetheeffect of coresaturationbetweentheapplicationof
switchingimpulses.
7.6.2.2 Power-fr equency withstand tests
Inorder to avoidsaturationof thecoreduringthe60seconds overvoltage-induced
tests, the test frequency may require increasing to more than twice the rated
value. In this case the length of the application may be reduced as for power
transformers.
7.6.2.3 Par tial-dischar ge tests
For identical test conditionstheallowablePDlevelsaresimilar tothecurrent trans-
former values. Phasephase values at 1.2U
m
are 5pC for liquid-immersed units
and20pC for soliddesigns. Thevarious quotedvalues arealsovalidfor non-rated
frequencies.
7.6.3 Capaci tor vol tage tr ansfor mer s
AsaCVT consistsof avoltagedivider andanelectromagnetic unit, theassessment
of theinsulationconditionrequiresseparationof thecomponentsfor particular tests.
TheappropriateproceduresaredescribedinIEC/PAS 60044-5:2002[S7/20]. Some
of themoresignificant testsaresummarizedbelow.
7.6.3.1 Type tests
Monitoringof theinsulationstateiseffectedbycapacitanceandDDF measurements
and, inthecaseof impulsetests, by thecomparisonof waveshapesbeforeandafter
thetest sequence. Thetypetestsincludethefollowing.
Wetpower-frequencywithstand-voltagetestsonoutdoorunits(<300kV system).
Lightningimpulsewithstand-voltagetestscomprising15consecutiveapplications
of eachpolarity for aU
m
< 300kV systemunit andthreeof eachpolarity for a
U
m
300kV systemunit. A standardwaveis to beusedbut if loadingonthe
generator istoolargeafront risetimeof upto8smaybeapplied.
Twochoppedimpulsewavesof negativepolarityappliedtobothclassesof CVTs.
Full wavesarerequiredbeforeandafter thechoppedwaves.
Wetswitchingimpulsetestsof 15positivepolaritywavesappliedtooutdoorunits
for U
m
300kV.
RIV tests if specified. Conditions as for inductivecurrent voltagetransformers.
A PDtest maybeusedasanalternative.
Ferro-resonancetestsif required.
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 171
7.6.3.2 Routine tests
As aninsulationmonitor capacitanceandDDF measurements aremadebeforeand
after thehigh-voltagetests.
Drypower-frequencywithstand-voltagetest for oneminute.
Partial-dischargetesttobeapplied. Alternativemethodsareallowable, including
useof balancedcircuit asinIEC 60270. Permittedlevelsfor earthedneutral are
10pC at U
m
and5pC at 1.2U
m
/

3.
Power-frequencywithstandtest istobeappliedtotheelectromagneticunit.
7.7 High-voltage power capacitor s
Thedielectricassessmentof power capacitorsrangesfromtestsoncompleteunitsof
tensof microfaradsfor operationatapproximately7kV (RMS) throughtoendurance
tests(overvoltagecyclingandageing)onindividual elementsratedat, perhaps, 1.8kV.
Manyof thecapacitorsaremountedoutdoorsinlargebanksforoperationatthehigher
systemvoltages. They must beabletowithstandadverseload-switchingconditions
andovervoltagesandremainthermallystableunder theveryhighinternal electrical
stresses. Thetestingof power capacitorsrequiresdedicatedsuppliescapableof pro-
vidingseveral MVAR at test voltagesof theorder of 20kV andfor somecasestobe
effectivelyPD-freeupto10kV. Testsfor confirmingtheinsulationperformanceare
describedinIEC 60871-1and-2[S7/21].
Typetestsincludethermal stabilitymeasurements, DDFdeterminationatelevated
temperature, an AC voltage withstand test and a lightning impulse-withstand test
betweenterminalsandcontainer.
Routine testsarerequiredformeasurementof capacitanceandDDF, forchecking
the AC withstand voltage and DC withstand voltage between terminals, and AC
withstandvoltagebetweenterminalsandcontainer.
Special test. Endurance tests may be required on elements typical of the par-
ticular designandmanufacturingprocess. Procedures andconstructionof elements
for ageing and overvoltage cycling tests are specified in IEC 60871-2. Although
partial-dischargemeasurements arenot normally specifiedbecauseof testingprob-
lems, facilities may berequired to carry out such tests on completecapacitors. A
PD-freesourceof upto, perhaps, 3MVARand15kV (RMS) maybenecessary. With
thelargest-ratingcapacitors, sensitivities of only 50pC areachievable, whichmay
not beconsideredsignificant by someauthorities. A possiblePDtest sequencewas
describedinAS2897, 1986[S7/21].
7.8 High-voltage r otating machines
Thequalitycontrol of theinsulatingmaterialsandinsulationstructuresof largeHV
rotatingmachines duringmanufacture, assembly andcommissioningis effectedby
application of a number of tests, some of which are listed below. The stator bars
172 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
aretestedbeforeassembly inthecore, probably withproceduresdeterminedby the
manufacturer and approved by the purchaser. The required measurements on the
completemachinearespecifiedinStandards suchas IEC 60034-1andIEC 60034-
15[S7/22]. Thehigh-voltagetestscheck clearancesbetweenphasesandtoground.
During tests in air corona may be produced in the overhang regions of the stator
windings.
7.8.1 Stator bar s
Possibletestsonindividual barsmight include:
insulationresistanceandpolarizationmeasurements;
capacitanceanddielectricdissipationfactor measurements;
power-frequencywithstandvoltageat aboveratedvaluefor oneminute;
partial-dischargetestsat agreedvoltages passlevelsareprobably of theorder
of 1500/2000pC at 1.2operatingvoltage.
7.8.2 Assembl ed machi ne
Possibletestsoncompletedwindingsandmachinesinclude:
insulationresistanceandpolarizationmeasurements;
capacitanceanddielectricdissipationfactormeasurementsoverarangeof voltage
(seeIEEE Standard286, 2000[S7/23]);
power-frequencywithstandvoltagefor oneminutebetweenphasesandtoground
at1000V +2ratedvoltage(line-linevoltage); all stator barsareattheapplied
voltageduringthistest;
additional, oralternative, HVDCwithstand-voltagetestas, forexample, specified
inIEEE 95, 1977[S7/24] maybeapplied;
(Thistest isnot consideredsuitablebysomemanufacturersandpurchasers. The
recommendedDC test voltageis1.7power-frequency(RMS) acceptancetest
valueapplied for oneminuteat thefull value failureis taken as acomplete
breakdown. Controlled overvoltagetests arealso described during which the
leakagecurrent ismonitoredasanindicator of possiblefailure.)
offlinepartial-dischargetestover arangeof voltagesusingtraditional procedures
(IEC60270); thestatorbarsof thephaseundertestarefloated atthetestvoltage
withtheother phasesgrounded;
online partial-discharge monitoring after commissioning at normal voltage
includingfull loadconditions.
(Thevarious techniques arediscussedinChapter 9andoutlinedinIEEE 1434,
2000 [S7/25]. The measured PD values for an assembled machine tend to be
significantlyhigherthanforindividual bars perhaps5000pCormoredepending
onthedesignandlocal conditions. Acceptablelevelsarebasedonexperiencewith
previous structures and operating machines. Such PD results formareference
for theestablishment of trendcharacteristicsasmonitoredduringthelifeof the
machine.)
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 173
Becauseof thelargestructures involvedmany units haveto bereassembledon
site, requiring testing of thecompletemachinei n si tu. Under theseconditions the
overvoltagepower-frequency tests requirealargekVA supply sourceperhaps con-
sistingof avariable-tunedcircuit at 50/60Hz, or avariablefrequencysystem(upto
300Hz) or alowfrequencytest set seeChapter 6.
Inmanymachinedesignsasurgetest isspecifiedinorder tochecktheinter-turn
andgroundinsulationinapplicationswheretheunitmaybesubjectedtolightningor
switchingimpulses. Therecommendedtestsforform-woundwindingsaredetailedin,
for example, IEC60034-15[S7/22]. Faultdetectionisbyobservationof therecorded
voltagewaves. Duringtheovervoltagetestsnopunctureor flashover isacceptable.
7.9 High-voltage cables
7.9.1 Oi l -i mpregnated cabl es
The insulation testing of AC high-voltage oil-impregnated cables in the factory
requirestheapplicationof power-frequencyandimpulse-withstandvoltagesandthe
measurement of DDF over arangeof voltages, thelevels dependingonthepartic-
ular installation. Typetests includethermal stability measurements at highvoltage.
Detailsfor lower-voltagemass-impregnatedcablesaregiveninIEC 60055[S7/26].
IEC 60141 (Parts 1, 2 and 4) [S7/27] includes tests for oil-filled and gas-pressure
cables upto andincluding500kV AC. For checkingtheterminations andjoints of
such oilpaper cables on siteit is usual to apply direct voltage, thus avoiding the
needfor largeVA test supplies. For exampleit isrecommendedthat a75kV DCtest
voltagebeappliedfor15minutesbetweenconductorsandfromeachtothescreenfor
an18/30kV AC cable. Nobreakdownisallowable. Thecablesystemsaredesigned
towithstandtheDC stressesproduced.
7.9.2 Extr uded cabl es
Followingthewidespreadintroductionof extrudedcablesthreeIEC Standardshave
been prepared covering theoperating voltageranges 630kV [S7/28], 30150kV
[S7/29] and 150500kV [S7/30]. The need for detailed insulation recommenda-
tions within the standards is partly due to the known susceptibility of XLPE to
partial discharges and, also, guidanceas to thevalues of lowlosses to beexpected
in such materials. An indication of the more important dielectric tests required
as manufactured are summarized below. For precise details, refer to the relevant
Standard.
7.9.2.1 Type tests for extr uded cables
PD measurements are specified after the bending and heat cycling tests. The
allowablePDmagnitudesat 1.5operatingvoltageare 5pC, thelatter being
therequiredbackgroundnoise.
174 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Fi gure 7.1 Equi val ent ci rcui t of PD defect i n a power cabl e [ 5] [ reproduced by
per mi ssi on of CI GRE]
DDF measurements at operating and ambient temperatures are specified. For
XLPE theexpectedvalues are 10 10
4
andinthetens of mRs for EPR at
operatingvoltages.
Power-frequencywithstand-voltagetestsarerequiredfor longandshort periods
dependingonthethermal conditions.
Lightning-impulsewithstand-voltageapplications at temperatures of 95100

C
are followed by a 15-minute power-frequency withstand voltage. Switching
impulsesarespecifiedfor theheatedsystemat voltages 300kV.
7.9.2.2 Routine tests for extr uded cables
Power-frequencywithstandtestrequirementsrangefromfiveminutesatapprox-
imately 3.3 operatingvoltageto earthfor thelower voltagesystems to twice
for 60minutesat thehighest voltages[S7/30].
PDtestsarespecifiedat1.5operatingvoltagetoearth. Themethodof calibrating
whentestingalong cableusingtheconventional detectionsystem(IEC60270)
isdescribedinReferenceS7/31.
Thewell-knowncircuitrepresentationof acableasatransmissionlinecontaining
aPD defect is depicted in Figure7.1 [5]. This referencepresents agood analysis
of thevarious developments in cablePD measurements. Figure7.2 indicates how
reflections areproduced by asingledischargein along cable. Theexampleis for
apartial dischargeat 450 metres of an open-ended 480-metre-long cable[5]. The
first pulseat themeasurement endis followedby areflectionfromthefar endand
aseriesof pulsesset upasshown. Thedoublepulsesareduetothesmall 30-metre
reflectiondistanceat thefar end. Theintervalsbetweenthepulsepairsrepresent the
transittimesfor theforwardandreflectedwaves. Analysisof suchdata probablyin
thepresenceof noise enablestheestimationof locationof dischargesandpossible
attenuationeffects[5].
7.9.2.3 I nstallation tests for extr uded cables
Thecommissioning of AC polymeric cables using HVDC for testing was not rec-
ommendedby SC21(B1) of CIGRE (1990), sinceit isseenasbeingineffectiveand
possibly causingdamageto theinsulation, includingtheaccessories assembledon
site. AC tests of a60-minutedurationfor voltagelevels intherangeof 2U
0
at the
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 175
1.98
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
0.38
Time 2.5 microsecs./div.
V
o
l
t
a
g
e

(
V
)
Fi gure 7.2 Attenuati on of a mul ti pl e refl ected PD pul se i n a power cabl e [ 5]
[ reproduced by per mi ssi on of CI GRE]
lower voltages downto 1.1U
0
for 500kV systems [S7/30] weresubsequently pro-
posed[CIGREBrochure173, Reference6]. Twenty-four-hourtestsatavoltageof U
0
maybeemployedif agreed, or higher if thesupplyislessthanthepower frequency.
A test-frequency rangeof 20Hz to 300Hz is quotedin S7/30. Such power supply
sourcesarenowavailable, asreviewedinChapter 6.
Thespecial caseof longextrudedAC submarinecables(36170kV) isreported
in El ectr a [7]. IEC revisions of theparticular standards arerelated to thevarious
proposals. A later CIGRE report [8] givescollationof practical test resultsobtained
oninstallationsusingthelevelsrecommended. Theresultsconfirmtheconclusions
of theearlier work.
7.10 Distr ibution and power tr ansfor mer s
Thedielectric testsondistributionand, especially, power transformersarecomplex
astheturn-to-turn, inter-windingandwinding-to-earthinsulationsarerequiredtobe
proved for two or morewindings under steady-stateand surgeconditions [S7/32].
These requirements often necessitate the application of special arrangements in
designandwhentesting inorder toachievethespecifiedtest levels.
7.10.1 Power -frequency over vol tage wi thstand tests
In three-phase units, particularly with three-limb cores, the single-phase induced
power-frequencytestcanresultinlargevoltagedifferencesbetweenphasesbutthese
may be reduced as indicated in Figure 7.3. A number of other configurations are
possible[IEC 60076-3]. Therecommendedinter-phaseair clearancesarealsogiven
inIEC 60076-3.
Forwindingswithoneendnominallyearthed(non-uniforminsulation)itisusually
impossibletoinducethespecifiedturn-to-turnvoltageof twicenormal andthecorrect
HV terminal-to-earthvoltagesimultaneously, unlesstheneutral terminal issupplied
176 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
U
(a) (b)
(c) (d)
U/3
Auxiliary transformer
U 0.5U 0.5U
U1
U2
Ut
U
~ ~
~
Fi gure 7.3 Si ngl e-phase i nduced-vol tage tests to ground on three-phase and 1-
phase tr ansfor mer s wi th non-uni for m i nsul ati on [ S7/32] [ reproduced
by per mi ssi on of SAI GLOBAL] : (a) fi ve-l i mb core (wi thstand over vol t-
age test and PD test); (b) three-l i mb core (wi thstand over vol tage test).
Extr a i nsul ati on at neutr al ; (c) three-l i mb core (wi thstand over vol tage
test and PD test). 1.5U between phases; (d) 1-phase auto-tr ansfor mer
(wi thstand over vol tage test). Auxi l i ar y testi ng tr ansfor mer mai ntai ns
cor rect vol tage r ati o between HV and Common ter mi nal s
fromanindependentsource forexample, foranauto-transformerasinFigure7.3(d).
Forsuchconditionstheneutral endbushingandthetransformerinsulationlevelsmay
needtobeof ahigher valuethanrequiredfor normal tests.
Previously, thedetectionof failureduringpower-frequencyshort-termtestsof 60
seconds orlessathigherfrequenciestopreventsaturationof thecoreduringinduced
tests wassimplytakenascollapseof thetest voltage. If agreed, theStandardnow
includesaspecificationfor partial-dischargemeasurementsasamonitor of possible
failureduringtheshort-duration(ACSD) test sequence. TheHV withstand-voltage
test durationfor frequencies greater thantwicenormal is givenby therelationship
120 (rated frequency/test frequency) seconds withaminimumof 15seconds. In
somecasestheterminal voltagemagnitudeandwaveformmaybemonitoredviathe
bushingtapor anHV capacitor divider usedfor measuringtheoutput voltage.
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 177
7.10.2 Par ti al -di scharge tests
If a partial-discharge test is specified, the conventional technique applied is that
describedinChapter 6utilizingtheHV bushing(andtap) asthecouplingcapacitor
for thehigher-voltageunits. For lower-voltageunits, e.g. dry-typedistributiontrans-
formers, separatecoupling capacitors arerequired [S7/32]. ThePD test is usually
carriedoutasasingle-phasetest. Thetestsequencesfor boththeshort-termduration
(ACSD) andthelong-termduration(ACLD) testsaresimilar. Thedurationof each
andthevoltagemagnitudesaregivenintheStandard. TheACLDtest isconsidered
tobeaquality-control test andnot for provingthedesign.
Inbothtypesof testafive-minuteconditioning periodat1.5U
m
/

3kVtoground
isinducedbeforeincreasingthevaluetothewithstandtestvoltagefor theACSDtest
and to 1.7U
m
/

3 for theACLD test. After lowering thevoltageto 1.5U

m
/

3kV,
partial discharges are measured continuously for assessment during a five-minute
periodfor theACSDconditionandduringasixty-minuteperiod(U
m
> 300kV)/30-
minute period (U
m
< 300kV) for the ACLD test. Also, values in the final five-
minuteperiods at 1.1U
m
/

3areincludedintheassessment. Other PDs measured

throughout the sequences are not considered significant in judging the success of
thetests.
Single-phase tests on three-phase transformers are preferably made with con-
nections as in Figure7.3 (c). Themaximumvoltageis applied for theone-minute
equivalentasdeterminedfromthetestfrequency. Theother timesareindependentof
frequency. Allowancemust bemadefor synchronizingthedischargedetector with
thetest frequencytoassist withinterpretationof thePDs.
7.10.2.1 I nter pr etation of PD measur ements
Theintroductionof PD measurements duringtheshort-durationover-potential test
(ACSD) may enablethedetectionof theinceptionandextinctionvoltages andthe
dischargemagnitudestobedeterminedif adisturbanceispresentduringtheapplica-
tion. Duetotherapidriseof voltageit may bedifficult toestimatetheprecisetime
andvalues. Anautomateddetectionsystemwouldbeof assistance. Becauseof the
variabilities this part of thetest is not usedfor PD assessment purposes. However,
thedataobtainedmight beof assistancetodesignersandthemanufacturer if afault
issuspected.
In both types of test acontinuous PD valueof >300pC after thehigh-voltage
application is not acceptable. The value at the conclusion is to be <100pC with
the voltage 10 per cent above operating voltage. This is based on a background
interferenceof 100pC that might beconsideredtobeahighvalue. Insomespecifi-
cations theallowablevalueis 50pC at 1.2 maximumphaseoperatingvoltageto
ground. It is also notedthat bursts of apparent discharges higher than300pC may
beignored. Researchhasshownthat suchunsustainabledischargescanbeproduced
by sometypes of fault and, perhaps, should berecorded for futurereference. The
acceptance or otherwise of PD values near the limit can be difficult as the error
in measurement may beas much as 20 per cent, depending on theconditions and
calibration.
178 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
7.10.2.2 Calibr ation and fault location
Theexecutionof partial-dischargetests inthefactory environment canbedifficult
dueto physical conditions and thesizeof thelarger oil-impregnated high-voltage
units. Apart fromthegoodhousekeeping required, aparticular problemistheneed
toincludesatisfactoryproceduresforinjectingcalibrationpulsesatthebushingtermi-
nals, possiblyat 7metresabovegroundlevel seeIEC 60270andS7/32. If several
bushing taps are available for example at the HV and common terminals of an
autotransformer andat theterminal of atertiary winding aprofileof theexpected
PD outputs for injections at the three terminals can be set up. By comparing this
distributionwiththat measuredduringthehigh-voltagePD test it may bepossible
todeterminetheelectrical locationof aparticular disturbance, especiallyif real-time
comparisonsby meansof simultaneousrecordingareincludedinthedetectionsys-
tem. Thisconcept isdiscussedinIEC 60076. Incaselargedischargesaredetected,
most manufacturers (andmany utilities) haveavailableultrasonic PD detectors for
physical locationof thedisturbances. Moreadvancedtransformer PDdetectionand
locationtechniquesarediscussedinChapter 9.
Withinthetestingscenariothedecisionastothecorrectproceduretofollowafter
thedetectionof anapparentlysignificantdischargecanbeverycomplex. Thesimplest
but expensiveactionistodraintheunit andrefill withdryoil under vacuum even
thisprocessmight takeseveral daysbeforeretestingispossible. Removingthecore
andcoilsfromthetankisamajor decisionevenif thePDlocationiswell definedby
theelectrical/ultrasonicmeasurements. Damageduetoalarge-magnitudedischarge
possibly causedby theshort-timeovervoltagetest anddevelopedduringtheACSD
test maynot bevisible. Theinspectionwouldbeaimedat lookingfor manufacturing
faultsor designproblemsintheareaindicated.
Monitoringof theinsulationconditionmay alsobeeffectedby measurement of
DGA levelsof oil-immersedunits. IEC61181[S7/33] coverstheapplicationof such
techniquesfor factorytesting. Moredetailedguidancefor serviceconditionsisgiven
inIEC 60599.
Forthree-phasedry-typetransformerstheroutinepartial-dischargetestconsistsof
apre-stressinterphasevoltageof 1.8ratedvoltageU
r
followedbyPDmeasurements
at1.3U
r
. TheallowablePDvalueinthiscaseisonly10pC, whichiscomparablewith
thevaluesfor other typesof equipment withsimilar insulationsystems.
7.10.2.3 L ightning- and switching-impulse voltage tests
Thedetection of internal failureduring lightning-impulsetests is difficult and has
resultedinthedevelopment of sensitiveproceduresenablingsmall changessuchas
turn-to-turn disturbances to beobserved by means of oscillographic and/or digital
records.
Except for major failuresperhapsinvolvingaline-to-earthbreakdownor short-
ingof asignificant fractionof thewinding, theappliedvoltagerecordsgiveminimal
indicationof aproblem, particularlyduringlightning-impulsetesting. Moresensitive
detectioniseffectedbyrecordingthewaveformsof theneutral currentand, forexam-
ple, thecurrentandvoltagetransferredtoanadjacentwinding. If thetransformertank
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 179
Z
1
(C
1
)
Z
2
(C
2
)
1 2 3 4 6
5
Fi gure 7.4 Vol tage measurement and faul t detecti on dur i ng l i ghtni ng-i mpul se tests
on tr ansfor mer s [ S7/32] [ reproduced by per mi ssi on of SAI GLOBAL]
1. Vol tage measur i ng ci rcui t; 2. Tank cur rent; 3. Neutr al or wi ndi ng
cur rent; 4. Capaci ti vel y tr ansfer red cur rent; 5. Cur rent shunts; 6. Vol t-
age measurement and tr ansfered vol tage Z
1
(C
i
), Z
2
(C
2
) i mpedances
(capaci tances) i n the vol tage di vi der
canbeinsulatedfromearthitmaybeadvantageoustorecordthetank-to-earthcurrent.
Possiblelocations of themeasurement impedances areshowninFigure7.4. These
testsarecarriedout under single-phaseconditionswiththeother phasesearthed.
Theexpected waveforms at thevarious measurement positions aredetermined
in thefactory by injection of low-voltageelectronic pulses fromarecurrent-surge
generator (RSG), or its equivalent, at the HV terminals to be tested. This allows
correct choiceof thecurrent shunts anddeterminationof transferredvoltages. The
latter must not beallowedto exceedthedesignedinsulationlevels. Conversely the
simpleshortingand/or earthingof other windingscanproduceunacceptableinternal
stressesduringtheimpulsetest. Of special interestarevoltagestressesinthetapping
windings. Although not strictly accurate, because of the omission of the oil, the
windingimpulsedesigncalculationscanbecheckedandsimulationof faultsmaybe
possibleonthetransformer initsuntankedstateusingRSGtechniques.
Anindicationof thetypes of waveforms to berecordedandinterpretedis pre-
sented in Figures 7.5 and 7.6 [S7/32]. Theexamples wereselected frommany in
theStandardandshowthedifferencebetweenalargefailureinatappingwinding
and a small breakdown between turns after a CW. There is no simple method of
determining preciseacceptancecriteria. Therequirement of no change(excluding
180 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
proportionatemagnitudelevels) intherecordedwaveformat reducedvoltageandat
full testvoltagesometimesnecessitatescareful considerationand, perhaps, agreement
ininterpretationbetweenthepurchaser andmanufacturer. This may beparticularly
relevantfor chopped-wavetestswheresuddenchangesoccur atthetimeof chopping
(effectedafter afews) tendingtomaskanyhigh-frequencychangesduetoafault.
Dependingonthenumber of recordingchannels available, andtheparticular case,
timebases chosenareintherange010/25s for theearlier part of thewaveand
0100sfor thelater times.
The detection of failure during switching-impulse tests is simpler than for
lightning-impulsetests, asthevoltagedistributionwithinthewindingsandphasesis
effectivelylinearandsimilartopower-frequencyconditions. Recordsaremadeof the
linevoltageand, probably, of theneutral currentwaveshapes. Therequiredwaveform
must haveavirtual risetimeof at least 100s withatimegreater than90per cent
of thetest valuefor 200sandatimefromvirtual zerotothefirst zerocrossingof
500s. Therecordingtimebasesrequiredarefrom100/500sto1000/5000s.
Typical examples aregiven in IEC 60076-4. Thefront must besuch as to givean
approximatelyuniformdistributionwithinthewinding.
Amajorfactorwhenapplyingswitchingsurgesisthepossibilityof coresaturation,
resultinginwaveformdistortion. Inorder to allowcomparisonbetweenrecords at
differentlevelsitmaybenecessarytoapplylower-level impulsesof oppositepolarity,
or evendirectvoltage, toestablishacceptablecoreconditions. Thewithstandvoltage
may beapplieddirectly to thewindingunder test or inducedfromalower-voltage
winding. Thevoltagesareapproximatelyintheratioof theturns.
Duringswitchingsurgetestsonthree-phasetransformersitisimportanttoensure
theclearancesbetweenphasesareadequate[IEC 60076-3]. Testsareusuallycarried
out with negative impulses, as air flashover voltages are greater than for positive
waves. Thenon-restoring internal insulation is assumed to havethesamestrength
for bothpolarities. Superimposedoscillationsduetointer-phasecapacitivecoupling
mayproducehigherthanexpectedvoltages[IEC60076-4], whichmustbecontrolled,
possiblybyresistivedamping.
Thetestissuccessful if nosuddenchangeoccursintheshapeof thelinevoltage.
Corresponding changes in the neutral current, if recorded, may also indicate a
problem.
7.10.3 Summar y of tr ansfor mer HV test requi rements
Theappropriatehigh-voltagetestsforvoltage(kV) classesU
m
72.5, 72.5< U
m

170, 170< U
m
< 300andU
m
300aretabulatedinIEC60076-3[S7/32]. Thetests
relatetouniformandnon-uniforminsulatedtransformers. Theappropriatewithstand-
voltagemagnitudes(one-minute test) andimpulselevelsfor eachoperatingsystem
aregivenintheStandardTables.
Switching-impulse withstand voltage (SI ): If theACSDtest isnot appliedthe
SI test isspecifiedfor thoseunitssubjectedtotheACLDtest.
L ightning-impulse withstand voltage (L I ): When specified this test covers
choppedwaves(seeFigure7.6) asaspecial test.
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 181
(1)
(2)
(3)
IEC 1421/02 IEC 1422/02
Reduced full wave (62.5%) without fault Reduced full wave (75%) with fault
(Amplitudes not equalized)
1 Applied impulse, 100 s sweep
2 Capacitively transferred current from the shorted adjacent winding to earth, 100 s sweep
3 Neutral current, 100 s sweep
Note: Failure indicated after 30 s in voltage, capacitively transferred current and neutral current oscillograms
Fi gure 7.5 Li ghtni ng-i mpul se ful l wave fai l ure i nter l ayer breakdown i n coar se-
step tappi ng wi ndi ng of a 400/220 kV tr ansfor mer [ reproduced by
per mi ssi on of SAI GLOBAL]
(1)
(2)
Reduced chopped wave (70%) without fault
1 Applied impulse, chopped wave, 50 s sweep
2 Capacitively transferred current from the shorted adjacent winding to earth, 50 s sweep
Note: Failure indicated immediately after chopping in both the voltage and capacitively transferred current oscillograms
Chopped wave (115%) with fault
Fi gure 7.6 Li ghtni ng-i mpul se chopped-wave fai l ure between tur ns breakdown i n
fi ne-step tappi ng wi ndi ng of a 400/220 kV tr ansfor mer [ reproduced by
per mi ssi on of SAI GLOBAL]
182 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Separ ate-sour ce AC withstand voltage: Thetestcheckstheinsulationfromthe
neutral to ground for non-uniformwindings and thewholeof theinsulation to
groundfor uniformwindings.
Shor t-dur ation induced AC withstand voltage (ACSD): A PD test may be
appliedif agreedbetweenmanufacturer anduser.
L ong-dur ation induced AC voltage (ACL D): Thetest includes PD measure-
mentsfor 30/60minutesatovervoltagestoconfirmtheabsenceof damagingPDs
duringservice.
Incaseswherelightning-impulsechoppedwavesarespecifiedapossiblesequence
might be 1FW (5075 per cent), 1FW (100 per cent), several CW for reference
(5075per cent), 2CW(100per cent), 2FW(100per cent), 1FW(5075per cent).
Thechoppingtimeshouldbeintherange26s. Failureis determinedby careful
comparisonof recordstakenduringthereducedchoppedwavesandthoseat thefull
level (Figure7.6). It isassumedthat thechoppingtimesat reducedandfull voltage
areapproximately equal. A slight differencein chopping times (perhaps <0.1s)
mightresultinchangessimilar tothosecausedbyafaultduringapplicationof aCW.
7.10.4 Addi ti onal tests
In addition to the well-established tests reviewed above, a number of newer
tests are being trialled especially for in-service monitoring. These include more
advanced PD systems, DC charging and relaxation measurements and frequency
responseanalysis(FRA). Detailsof someof thevarioustechniquesareconsideredin
Chapters6, 8and9.
7.11 Dielectr ic testing of HVDC equipment
Thechoiceof lightningandswitchingimpulsesforHVDCequipmentdependsonthe
insulationcoordinationlevelsof thesystem. Therelevantdirectvoltagetestsasrelated
tothecontinuousDCoperatingvoltagesare, ingeneral, agreedbetweenmanufacturer
anduser. A discussionof voltagestressesandtestrequirementsfor converter stations
and HVDC cables is presented in References 9 and 10. Guidanceon coordination
proceduresisconsideredinIEC 60071-5[S7/34] for converter stations.
Animportant aspect of theprotectivemethodsisthecorrect applicationof surge
arresters, especiallyrelatedtothethyristor valves. Thetestingof suchvalvesisspec-
ifiedinIEC60700-1[S7/35], includingmethodsfordeterminingthetestlevelsbased
ongiventestsafetyfactors. Forexampletheseare, respectively, 1.6and1.3duringthe
one-minuteandthree-hour DC tests of external insulationto earth. Thethree-hour
valueis reduced to 0.8 for avalvetest. Partial-dischargetests arerequired during
theseapplications, thePDvaluesandrepetitionratesbeingrecordedduringthelast
hour. Thenumber of pulses greater than 300pC must be 15/minute aspread
of allowablevaluesof upto2000pC (1/minute) isallowed. Themeasurementsare
accordingto IEC 60270andmust incorporatemethods for countingandrecording
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 183
thepulses. AC partial-dischargetests arealsorequired, thelimit being200pC dur-
inga30-minutetest at avoltagebasedontheequivalent peak-to-peak, steady-state
operatinglevel andasafetyfactor of 1.15. Thestandardizedlightningandswitching
surge(threeof eachpolarity) testlevelsare1.15surgeprotectivelevel of thevalve
arrester and1.2for steepsurges. Thevarious complexities intestingthedielectric
strengthof thyristor valves whensubjectedtothecombinedeffects of DC, AC and
surgeconditionsaredetailedinthestandard.
Recommendationsfor testsonpaper-insulatedcablesfor DC transmissionvolt-
agesupto800kV areincludedinaCIGRE report [11]. It isrecommendedthat the
factory high-voltageacceptancetest beanegativeDC voltageof 1.8 U
0
applied
for 15 minutes. The allowable power factors of the manufactured cable lengths
are specified for example, for an oil-filled cable rated at 400kV DC, the value
shouldnot exceed46 10
4
at amaximumRMS stress of 20kV/mm. Typetests
includecombinedpolarityreversal/loadcycleswithavoltageof 1.4U
0
. Impulse-
(lightning- andswitching-) superposedtestswiththecableat anegativeDC voltage
of U
0
aredescribedfor eachcase. Positiveandnegativeimpulses aretobeapplied
withrecommendedminimumvaluesof 1.15lightningimpulseprotectivelevel and
1.15 switchingsurgelevel of theparticular installation. Detailsof theprocedures
aregiveninthereport. Thesuggestedsitetest is anegativeDC voltageof 1.4U
0
appliedfor 15minutes.
Extrudedinsulationfor DC transmissioncableshasbeenlimitedtocommercial
applicationsuptovoltagesof 80kVand150kVasintheHVDCLightVSCprojectsin
Australia(2000/2002) includingextrudedcablelengthsupto180km. A fewfailures
haveoccurred in other schemes, possibly dueto space-chargeproblems following
polarityreversals. Material developmentsandtheintroductionof thenewconverter
technology(voltagesourceconverter, VSC), whichdoesnotreversethepolarity, may
giveimpetustothewider useof extrudedHVDCpower cables[12]. Thisisreflected
in the proposals for establishing test procedures for such cables in the range 80
250kV includedinthereport byCIGRE WG21.01[13]. A number of theproposals
aresimilar to thoseinReference11, althoughallowanceis made, for example, for
thelonger timeconstants. Thetypeloadtestsat1.85U
0
include24+24hoursload
cyclesanddifferencesinimpulsetestsfor aVSCsystem. ACtestsarerecommended
wherepractical for monitoringinsulationquality. Thefactory-acceptanceandpost-
installationtestsaretobeHVDC.
Bushings for theconverter transformers and reactors must betested separately
underHVDCconditions, asthedesignsarenotsimilartoACunits, thevoltagestresses
beinggovernedbytheresistivitiesandtemperaturesof theinsulatingmaterials. Long-
termPDtestswouldprobablyberequired.
Thetestingof theconverter transformersrequiresapplicationof HVDCvoltages
tothevalvewindingsinadditiontoappropriateACandimpulseteststoall windings.
Theprocedures aimto cover thecondition of superimposed AC and DC voltages.
TheseincludePDtestsunder DCconditionsrequiringlong-termtestingtoallowthe
correctvoltagedistributionstobeestablished. After 50yearsof testingexperiencein
theindustry, IEC has publishedastandard[IEC 61378-2, S7/36]. A Trial UseTest
CodewaspublishedbytheIEEEin1999[C57.129]. Theseriesreactorsaresubjected
184 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
onlytoDC andimpulsehigh-voltagetests, thelatter checkingthedielectricstrength
acrossthewindingaswell astoground.
Other components requiring HVDC tests include the surge arresters and insu-
lators. A standard for ceramic and glass string insulators was issued in 1995 [IEC
61325, S7/37].
7.12 M iscellaneous items
Aspart of themonitoringandmaintenanceproceduresonsite, it isessential tohave
availablesafeandreliableequipment for usebypersonnel, especiallywhencarrying
out high-voltagetests andlive-lineworking. Therearemany rules andregulations
issued by responsible authorities covering the formand application of items such
as earthing sticks, live-line working tools, cherry-picker buckets and associated
vehicles (see Figure 1.7), insulated gloves and insulation mats. All these devices
must be subjected to routine high-voltage tests and, usually, insulation resistance
measurements at regular intervals possibly every six months insomecases. IEC
Standardsinclude60855, 60900and60903[S7/3840].
7.13 Summar y
Inthischapter arepresentedtheformsof standardizedteststhathavebeendeveloped
bytheindustryovermanyyearsinordertoproducereliableandeconomical equipment
forapplicationinthepower-supplysystems. Althoughnewtestsarebeingcontinually
introducedtoallowforchangesininsulatingmaterials polymers, castresins, XLPE,
SF
6
andinsomedesignconcepts, thebasicmethodsareusuallyretained.
It isinterestingtonotethedifferencesinchoiceof test sequencesasdetermined
bythetypeof equipment. For exampleimpulsetestsoncables, insulators, bushings
andswitchgear requiretenor moreapplicationswhilefor transformersonlytwoFW
shotsareusuallyspecified. Surgearrestersareaspecial caseinvolvingsimultaneous
applicationof power-frequencyandimpulsevoltages.
7.14 Refer ences
1. CISPR 18-2: Admnt. 1, 1993, Radio interference characteristics of over-
head lines and high-voltageequipment; Part 2: Methods of measurement and
procedurefor determininglimits
2. CISPR16-1: 1993, Specificationforradiodisturbanceandimmunitymeasuring
apparatus and methods Part 1: Radio disturbance and immunity measuring
apparatus
3. Guidelinesfordesignandmaintenanceof overheaddistributionandtransmission
lines, HB C(b)1-1999AustralianStandards
4. Darveniza, M. et al . A multiplepulselightningimpulsegenerator, Proceed-
i ngs of the 6th I nter nati onal Conference on Hi gh Vol tage Engi neer i ng, Paper
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 185
47.07, Aug/Sep1989, NewOrleans(seealsoI EEE Tr ans. Power Del i ver y, April
1994;9:76471, Ref. InAS1307.2)
5. CIGRE WGD1.33TF 02.05, Practical aspectsof thedetectionandlocationof
partial dischargesinpower cables, 2006, Brochure297
6. CIGRE WG 21.16, Partial discharge detection in installed extruded cable
systems J uly2000, Brochure173
7. CIGRE WG21.02, Recommendationsfor testingof longAC submarinecables
with extruded insulation for systemvoltage above 30(36) to 150(170) kV,
El ectr a, April 2000;(189):2937
8. CIGRE TF 21.05, Experiences with AC tests after installation on the main
insulationof polymeric (E)HV cablesystems, El ectr a, December 2002;(205):
2735
9. Lips, H.P., Voltage stresses and test requirements on equipment of HVDC
converter stationsandtransmissioncables, 2000, Paper P2-06CIGRE
10. CIGRE WG 14.32, HVDC converter stations for DC voltages above +/
600kV, El ectr a, December 2002;(205):51, Brochure215
11. CIGREWG21.02, Recommendationsfortestsof powertransmissionDCcables
for arated voltageup to 800kV (El ectr a 72, 1980 Revision), El ectr a, April
2000:3955(seealsoEl ectr a, February2005; (218))
12. CIGRE WG B4.37, VSC Transmission, El ectr a, April 2005;(219):2939,
Brochure269
13. CIGRE WG21.01, TestingDC extrudedcablesystemsfor power transmission
upto250kV, Electra, February2003;(206):4753, Brochure219
7.15 Standar ds r elated to Chapter 7
The reference of the nearest equi val ent Austr al i an Standard i s i ncl uded where
appropr i ate.
S7/1IEC 60060: High-voltagetest techniques(AS1931.1, 1996)
S7/2IEC 60383: Insulatorsfor overheadlines:
Part 1: Ceramicor glassinsulator unitsfor AC systems.
Part2: Insulator stringsandinsulator setsfor ACsystems(AS2947, Parts1, 2, 4)
S7/3IEC 60433: Insulatorsfor overheadlines Ceramicinsulatorsfor AC systems
Characteristicsof insulator unitsof thelongrodtype
S7/4IEC 60273: Characteristics of indoor andoutdoor post insulators for systems
withnominal voltagesgreater than1000V (andfor indoor organicunits) (AS4398,
2005, Parts1, 2)
S7/5IEC 60168: Testsonindoor andoutdoor post insulatorsof ceramicmaterial or
glassfor systemswithnominal voltagesgreater than1000V (AS4398, 2005, Part2)
S7/6 IEC 61109: Composite insulators for AC overhead lines Definitions, test
methodsandacceptancecriteria
S7/7IEC 61952: Insulators for overheadlines Compositelinepost insulators for
AC withanominal voltagegreater than1000V (AS4435.4, 2005)
186 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
S7/8IEC 60437: Radiointerferencetest onhigh-voltageinsulators(ASIEC 60437
2005)
S7/9IEC 60099: Surgearresters:
Part 1: Non-linear resistor typegappedarrestersfor AC systems(AS1307.1)
Part 4: Metal-oxidesurgearresterswithout gapsfor AC systems
S7/10AS1307.2: Surgearresters Part2Metal-oxidearresterswithoutgapsfor AC
systems
S7/11IEC60694(Ed. 2.2, 2002): Commonspecificationsforhigh-voltageswitchgear
andcontrolgear standards
S7/12 IEC 62271-100 (Ed. 1.1, 2003): High-voltageswitchgear and controlgear
Part 100: High-voltagealternating-current circuit-breakers(AS62271.100, 2005)
S7/13 IEC 62271-102 (Ed.1.0, 2003): High-voltage switchgear and controlgear
Alternatingcurrent disconnectorsandearthingswitches(AS62271.102, 2005)
S7/14IEC62271-200(Ed.1.0, 2003): High-voltageswitchgearandcontrolgear Part
200: AC metal enclosedswitchgear andcontrolgear for ratedvoltages above1kV
anduptoandincluding52kV (AS62271200, 2005)
S7/15IEC62271-201(Ed.1.0, 2006): High-voltageswitchgearandcontrolgear Part
201: ACinsulation-enclosedswitchgearandcontrolgearforratedvoltagesabove1kV
anduptoandincluding52kV
S7/16 IEC 62271-203 (Ed. 1.0, 2003): High-voltageswitchgear and controlgear
Part 203: Gas-insulated metal-enclosed switchgear for rated voltages above52kV
(AS62271.203, 2005)
S7/17IEC 60214-1(Ed.1.0): Tap-changers Part 1: Performancerequirements and
test methods(AS60214.1, 2005)
S7/18IEC 60137(Ed. 5.0, 2003): Insulatedbushingsfor alternatingvoltagesabove
1000V (AS1265, 1990)
S7/19IEC61464TR2(Ed. 1.0, 2003): Insulatedbushings Guidefor theinterpreta-
tionof DGA inbushingswhereoil istheimpregnatingmediumof themaininsulation
(generallypaper) (ASTR 61464, 2006)
S7/20IEC 60044: Instrument transformers(AS60044Parts1, 2, 5)
60044-1(Ed. 1.2, 2003) Part 1: Current transformers
60044-2(Ed. 1.2, 2003) Part 2: Inductivevoltagetransformers
60044-5(Ed. 1.0, 2004) Part 5: Capacitor voltagetransformers
S7/21 IEC 60871: Shunt capacitors for AC power systems having arated voltage
above1000V (AS2897, 1986, withdrawn)
60871-1(Ed. 3.0, 2005) Part 1: General
60871-2Ed. (2.0, 1999) Part 2: Endurancetesting
S7/22IEC 60034: Rotatingelectrical machines
60034-1(Ed. 11.0, 2004) Part 1: Ratingandperformance(AS1359.101, 1997)
60034-15(Ed. 2.0, 1995) Part 15: Impulsevoltagewithstandlevelsof rotatingAC
machineswithform-woundstator coils
S7/23IEEE Std. 286, 2000, IEEE: Recommendedpracticefor themeasurement of
power factor tip-upof electricmachinerystator coil insulation
Establ i shed methods for i nsul ati on testi ng of speci fi c equi pment 187
S7/24IEEEStd. 95, 2002, IEEE: Recommendedpracticeforinsulationtestingof AC
electricmachinery(2300V andabove) withhighdirect voltage
S7/25IEEE Std. 1434, 2000, IEEE: Guideto measurement of partial discharges in
rotatingmachinery
S7/26IEC 60055-1(Ed. 5.1, 2005): Paper-insulatedmetal-sheathedcablesfor rated
voltages upto 18/30kV (withcopper or aluminiumconductors andexcludinggas-
pressure and oil-filled cables) Part 1: Tests on cables and their accessories (AS
1026, 2004)
S7/27IEC 60141: Testsonoil-filledandgas-pressurecablesandtheir accessories
60141-1(Ed. 3.0, 1993) Part 1: Oil-filled, paper or polypropylenepaper laminate
insulated, metal-sheathedcables andaccessories for alternatingvoltages upto and
including500kV
60141-2(Ed. 1.0, 1963/67) Part 2: Internal gas-pressurecablesandaccessoriesfor
alternatingvoltagesupto275kV
60141-4(Ed.1.0, 1980/90) Part 4: Oil-impregnatedpaper-insulatedhighpressure
oil-filledpipe-typecablesandaccessoriesforalternatingvoltagesuptoandincluding
400kV
S7/28 IEC 60502: Power cables with extruded insulation and their accessories for
rated voltages from1kV (U
m
= 1.2kV) up to 30kV (U
m
= 36kV) (AS 1429.1,
2006)
60502-2(Ed. 2.0, 2005) Part2: Cablesfor ratedvoltagesfrom6kV (U
m
= 7.2kV)
upto30kV (U
m
= 36kV)
60502-4(Ed. 2.0, 2005) Part 4: Test requirements onaccessories for cables with
ratedvoltagesfrom6kV (U
m
= 7.2) upto30kV (U
m
= 36kV)
S7/29IEC 60840(Ed. 3.0, 2004): Power cables withextrudedinsulationandtheir
accessories for rated voltages above 30kV (U
m
= 36kV) up to 150kV (U
m
=
170kV) Test methodsandrequirements(AS1429.2, 1998)
S7/30IEC 62067(Ed. 1.1, 2006): Power cables withextrudedinsulationandtheir
accessories for ratedvoltages above150kV (U
m
= 170kV) upto 500kV (U
m
=
550kV) Test methodsandrequirements
S7/31IEC 60885-3(Ed. 1.0, 1988): Electrical test methodsfor electriccables Part
3: Test methods for partial-dischargemeasurements on lengths of extruded power
cables(AS1660.3, 1998)
S7/32IEC60076-1(Ed. 2.1, 2000): Power transformers Part1: General (AS60076
Parts1, 3, 4, 11)
60076-3(Ed. 2.0, 2000): Power transformers Part 3: Insulationlevels, dielectric
testsandexternal clearancesinair
60076-4(Ed. 1.0, 2002): Power transformers Part 4: Guideto lightningimpulse
andswitchingimpulsetesting Power transformersandreactors
60076-11(Ed. 1.0, 2004): Power transformers Part 11: Dry-typetransformers
S7/33IEC 61181(Ed. 1.0, 1993): Impregnatedinsulatingmaterials Applicationof
dissolved-gasanalysis(DGA) tofactorytestsonelectrical equipment
S7/34IEC/TS60071-5(Ed. 1.0, 2002): Insulationcoordination Part 5: Procedures
for high-voltagedirect-current (HVDC) converter stations
188 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
S7/35IEC 60700-1(Ed. 1.1, 2003): Thyristor valvesfor high-voltagedirect-current
(HVDC) power transmission Part 1: Electrical testing
S7/36IEC 61378-2(Ed. 1.0(2001): Converter transformers Part 2: Transformers
for HVDC applications
S7/37IEC61325(Ed. 1.0, 1995): Insulatorsforoverheadlineswithanominal voltage
above1000V Ceramic or glassinsulator unitsfor DC systems Definitions, test
methodsandacceptancecriteria
S7/38IEC 60855(Ed. 1.0, 1985/1999): Insulatingfoam-filledtubes andsolidrods
for liveworking
S7/39IEC 60900(Ed. 2.0, 2004): Liveworking Handtoolsfor useupto1000V
AC and1500V DC
S7/40IEC 60903(Ed. 2.0, 2002): Liveworking Glovesof insulatingmaterial
7.16 Pr oblems
1. Analyse and discuss the function and influence of Standards on the applica-
tionof insulationinthehigh-voltagepower industry. Highlight their importance
economically andtechnically andtheneedtostandardizeassessment/condition
monitoringmethodsfor equipment andmaterialswhenever appropriate.
2. Determine the PD pulse separation at the measurement end of the cable in
Section7.9.2 for asourcelocated at 100 metres fromthefar end. Why is the
apparent pulsevelocitylessthanthevaluebasedonacable-relativepermittivity
of 2.2?
Chapter 8
Sensor s for insulation condition monitor ing
Ultra-high-frequencysensors
Fibre-opticsensors
Directional sensors
Sensorsplay animportant roleininsulationconditionmonitoring. For different
insulationstructuresandmaterialsusedinvariousequipment, itisnecessarytoselect
themost suitablesensors. Theenvironmental noisealso affects theselectionof the
sensor. Thischapterpresentssomeadvancedsensorsrecentlydevelopedforcondition
monitoringof electrical insulation.
Asdiscussedinpreviouschapters, insulationof HV equipment canbeexamined
fromthemeasurement of someimportant parameters suchas insulationresistance,
dielectric dissipation factor, partial discharge and dissolved gases in oil. A rapid
change of these parameters can be seen as the symptomof incipient faults and
should be closely monitored. Other parameters such as temperature, dew point
and water in oil are also important because they have significant impacts on the
servicelifeandelectrical and/or mechanical strengthof aninsulationsystem. Accu-
rately monitoringtheseparametersisabasic requirement for conditionmonitoring.
The conventional PD sensors such as wide-band and radio-interference sensors
(refer to Chapter 6) are useful, but may not be satisfactory in some cases, which
has resulted in the development of new sensors as an important area for world
researchers.
8.1 Ultr a-high-fr equency sensor s
PD electrical detectors have different frequency bands. As discussed in previous
chapters, conventional detectorsarecategorizedasnarrow-band(e.g. radio-frequency
detectorwith1020kHzbandwidthandcentrefrequencyupto50MHz)orwide-band
(e.g. theERA impedancesensor from10kHz upto1MHz bandwidth). SinceaPD
pulseatthedischargesitecoversaverywidefrequencyspectrum, thefrequencyband
190 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
of adetector can significantly affect therecorded signal, which may bedisplayed
on the screen or transferred to an A/D converter for digitising. Both waveshape
and magnitude will be changed after passing through the detector of a certain
bandwidth.
During thelast decade, ultra-high-frequency (UHF) sensors havebeen widely
usedfor PD detectionontransformers, generators, GIS andcables. Thefrequency
bandrangesfrom100MHz to2GHz. Withsuchawidefrequencybandwidth, much
energyof aPDisreceivedinthedetector, givingahigher signalnoiseratio. There-
fore, theoverall sensitivityisimproved. Theresultingsignal waveshapesof various
detectorsaregiveninFigure8.1for comparison.
Wide-banddetection, asshowninFigure8.1(b), iscommonly usedincommer-
cial PD detectors for offlinePD detections on HV equipment. With thesetypes of
detector, apparent dischargescanbemeasuredfairly accurately, whichisimportant
for insulation-conditionassessment.
TheUHF sensorsarewidely usedinGIS andcableterminationsbecauseof the
highsensitivity[3]. Duetothesignificant attenuationof high-frequencypulsestrav-
ellingalongcables, stator windingsandbusbars, only localizedPDsaredetectable.
The typical installation and structure of a UHF sensor installed on a 230kV GIS
are shown in Figure 8.2 (a). The sensor is basically a capacitive coupler with a
detectionfrequency between500MHz and2GHz. For generators, anantenna-type
UHF sensor is inserted in the stator bar slots (under the wedge) to detect PDs in
theconductor [4]. UHF detectionis alsoappliedtopower transformers by locating
antennaeinvalveor preformedopenings inthetank, as illustratedinFigure8.2(b)
and(d) [5,6,7]. Details of PD measurements usingthesesensors will bediscussed
inChapter 9.
Figure8.3showsaportablePDdetector withexchangeablecapacitiveandinduc-
tivesensors, whichdetect theelectromagnetic transientsof PD events. Inthemode
for differential measurement, it may pinpoint thelocationof aPDsourceif it isnot
completely screenedelectromagnetically. Thesensorsareof widefrequency bands.
Using two capacitive sensors connected to the inverting and non-inverting inputs
respectively, thePDmaybelocatedbyaPDpulsepolaritymethod.
8.2 Optical-fibr e sensor s
Insulationconditionmonitoringreliesonsensorstodetectelectrical or non-electrical
parameters of an insulation system. Current methods for onlinemonitoring of HV
equipment oftenhaveproblemsinaccessinghighfieldareasandinreliably detect-
ingelectrical signals innoisy environments. Optical-fibresensors offer aradically
different approach to theseproblems and aregaining much attention in thepower
industry. Thetechnological basestemsfromanintensiveeffort inoptical-fibrecom-
munications. Their ruggedness and flexibility make it possible for themto reach
previously inaccessibleareas. Theeffects of external variables on thefibrecan be
sensedviaanumber of very precisechanges onthelight beams travellingalongit.
Thereisthereforeagreatpotential foroptical fibrestoreplacetheroleof conventional
Sensor s for i nsul ati on condi ti on moni tor i ng 191
Coupling
capacitor
Impedance
Amplifier
DDA
V
1 V
2
Filter
f from 10 kHz up
to 1 MHz
f
fm
f
t t
6dB
t t t
Coupling
capacitor
(a)
(b)
(c)
Impedance
Amplifier
Rectifier Meter or
DDA
V
1 V
2
V
3
Filter
f = 10-20kHz
fm = 1MHz-
50 MHz
f
fm
f
6dB
UHF
sensor
UHF
amplifier
DDA
V
1
V
2
Filter
f from 100 MHz
to 2 GHz
f
f
t t
6dB
Fi gure 8.1 The PD waveshapes detected by di fferent detector s wi th frequency
pass bands of: (a) nar row band of 1020 kHz, (b) wi de band of
100 kHz1 MHzand (c) ul tr a-wi deband of 100 MHz2 GHz, whereDDA
di gi tal data acqui si ti on system, f frequency bandwi dth and f
m
centre frequency [ 1,2]
electrical devicesinanumber of sensingapplications. However, it will benecessary
todevelopcalibrationmethodsindifferent applications.
Optical-fibre sensors are particularly well suited for condition monitoring
applications becausethey arelightweight, durable, sensitive/insensitiveto several
192 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(d)
(b) (a)
(c)
Fi gure 8.2 UHF sensor used for PD detecti on on var i ous HV equi pment. (a) UHF
sensor for GI S; (b) UHF sensor i nstal l ed on a 400 kV GI S; (c) UHF sen-
sor for tr ansfor mer s; (d) The UHF sensor i nstal l ed on a 66 kV/10 MVA
tr ansfor mer through the oi l outl et val ve [ 5,6]
Fi gure 8.3 A por tabl e PD detector wi th exchangeabl e capaci ti ve and i nducti ve
sensor s [ cour tesy of Lemke Di agnosti cs AG, URL: http: //www.l di c.ch,
http: //www.hvdi agnosti cs.de]
Sensor s for i nsul ati on condi ti on moni tor i ng 193
parameters, and can be incorporated within the equipment itself without being
intrusive. The parameters that can be monitored range from chemical species
(degradation by-products, contaminants and wear debris), internal/external strain
fields, pressure, fracturing and fatigue to vibrations, acoustic emission and
temperature.
Thehighexpectationsof optical fibresasinformationcarriersincommunication
systemshasbeenjustifiedby their performanceover thepast threedecades. Dueto
thehighbandwidthandlowattenuationproperties, eachfibreiscapableof replacing
1000copper wires. Asasensor, anoptical fibrecancarry muchmoreinformation.
Light that islaunchedintoandconfinedtothefibrecorepropagatesalongthefibre
unperturbedunlessexperiencinganexternal influence. Any disturbancetothefibre
altersthecharacteristicsof theguidedlight, whichcanthenbemonitoredandrelatedto
themagnitudeandlocationof thedisturbinginfluence. Thecharacteristicsof thelight
thatmaybemonitoredinsensingapplicationsincludeamplitude, phase, wavelength,
polarization, modal distributionandtime-of-flight.
8.2.1 Basi c physi cs of opti cal -fi bre sensi ng
Therearemany different types of optical-fibresensor. Accordingtothemethodby
which the light is modulated in the fibre, they fall into four main groups [8], as
follows.
I ntensiometr ic: Intensiometricoptical-fibresensorsarebasedonthemodulation
of light intensity inthefibreandaregenerally configuredas adistributedsen-
sor. Thelight intensity may changeintheformof fractureloss, time-of-flight,
refractiveindex, amplitudeandwavelength.
I nter fer ometr ic: Optical-fibre interferometers are generally intrinsic sensors.
Withthelight fromasourceequallydividedanddirectedthroughtwo(or more)
fibre-guidedpaths, thebeamsarethenrecombinedtomixcoherently. Fingerprint
typescanbeestablishedinrelationtotheoptical phasedifferenceexperiencedby
thetwolightbeams. Single-modefibresareusedandveryhighsensitivitycanbe
achieved.
Polar imetr ic: Theprincipleof polarizationsensingisbasedonthebirefringence
withinthefibre. Whenlaunchingapolarizedlightbeamalongtheprincipal axisof
asingle-modefibreat45

, thetwoorthogonal polarizationeigenmodesareequally
excitedanddegenerated. Thedegeneracyisincreasedif thefibreissubjectedtoan
external forceandaphasedifferenceisinducedinthetwoeigenmodes. Thestate
of polarizationinthefibrecanthenberelatedto themagnitudeof theexternal
force. Thesetypesof sensorarenotassensitiveasinterferometers, buthavelonger
sensinglengths.
M odalmetr ic: Optical fibres are microphonic and the detection sensitivity is
based on themodulation in thedistribution of modal energy in thefibre. This
typeof sensor is sensitive, but themodulationof themodal patternis normally
nonlinear in relation to all disturbances, resulting in fading and drifting in the
outputsignal. Usually, thesesensorsareusedinconjunctionwithothertechniques
suchasmicrobendingandpolarimetry.
194 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Pulsed
laser
(a)
(b)
Data
acquisition
and signal
processing
Optical
filter
Laser light travelling along the fibre
Incident light
Brillouin line
Rayleigh band
Brillouin line
Wavelength
"Stokes"
Raman band
"Anti-Stokes"
Raman band
Optical
Rx
Optical
coupler
Optical
coupler
Fi gure 8.4 The bl ock di agr am of an i nter ferometer opti cal -fi bre sensor and the
back-scatter spectr um [ 8,10]
Theblockdiagramof aninterferometeroptical-fibresensorisshowninFigure8.4.
Optical-fibre sensors are rapidly proving to be a vital new tool for insulation
researchers. Thedeviceshavealreadyfoundusesintemperature, pressureandvibra-
tion measurements. Speciality fibre technology is opening up more opportunities
in insulation condition monitoring such as PD measurements and transformer oil
moisture detection. A few typical applications are listed below, although further
investigationsarenecessarybeforetheycanbefullyutilizedintheindustry.
8.2.2 Opti cal -fi bre PD sensor s
8.2.2.1 Par tial dischar ge measur ements on gener ator s and GI S
Theacousticemissionsfrompartial dischargescanbedetectedbyoptical-fibresen-
sors. Researchers haveinvestigateddifferent types of suchsensors andhavefound
that achievingsufficient sensitivityisthemost difficult problem. InaGISchamber,
thetypical acoustic vibrationgeneratedby PDsis1 10
5
gto1 10
2
g, where
g = 9.81m/s
2
. Most of theacoustic energy fromPDs is at ultrasonic frequencies,
Sensor s for i nsul ati on condi ti on moni tor i ng 195
100 mV
50
0 1 2 3 4
ms
Fi gure 8.5 The typi cal responses of an opti cal -fi bre sensor to PDs i n a gener a-
tor stator bar tested i n the l abor ator y. The di scharge magni tude was
approxi matel y 500 pC [ 8]
i.e. around 10kHz. At this frequency, the acoustic vibration acceleration can be
translatedintodisplacementsintheorder of 10
14
mto10
12
m. For generators, the
magnitudeof PDsismuchhigherthaninGIS, butevensomoresensitivityisrequired
beforeareliablesensor canbeused. ThePDenergydetecteddependsonthelength
of thesensingfibre, theacousticpower andthenumerical apertureof thefibre.
MonashUniversityandFFT LtdhavejointlydevelopedtheFopticStrainsensor
andassociatedinstrument [8]. Thetechniqueis basedonawell-knownfibre-optic
interferometric principle that can achieve a wide-frequency bandwidth of 1MHz.
Thesensorislesscomplexandof lowercostthanotheroptical-fibresensingsystems.
Onlyashortsensinglengthof 6mmisnecessary, whichmakestheinstallationeasier.
Thesensor hasbeentestedonagenerator stator bar inthelaboratoryandthetypical
responseof thissensor toPDsinthegenerator stator bar isshowninFigure8.5.
8.2.2.2 Par tial dischar ge measur ements on tr ansfor mer s
PDmeasurementsontransformersusingelectrical methodsareverydifficultbecause
of thelowlevel of dischargeto bemeasuredandthelargeinterferenceat site. Ina
330kV substation, coronadischargesfromtheoverheadtransmissionlinescanreach
3000pC or more, whilethePD level tobemeasuredcouldbeless than100pC. In
suchanoisy environment to detect PDs smaller than5per cent of noisewouldbe
impossiblebyconventional methods. Optical-fibresensorsareimmunefromelectri-
cal interferenceandcanbeinstalledaroundthewinding, whichis excellent for PD
monitoringontransformers. However, thesensitivity of optical-fibresensors is not
sufficientlyhighfor suchanapplication.
A novel optical sensor wasdeveloped[9] that usesanoptical-fibreheadinstead
of apiezo-electricsensor for thedetectionof acousticemissionfromthePDsource,
asshowninFigure8.6. Theacousticpressurechangestheshapeof theoptical fibre,
whichcarrieslaser lightinjectedfromterminal 1. Asaconsequence, thephaseof the
light will changeaccording to thepressure. When reflected fromthemirror at the
fibrehead3, thechangewill bedetectedandconvertedto anelectrical signal. The
magnitudeof thesignal will vary accordingtotheacoustic wavepressurefromthe
PDsource.
196 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
f
1
f
2
1
Fibre
coupler
4
Photo diode
Fibre sensor head
3
Sensing arm
Down-lead fibre
Reference arm
2
Fi gure 8.6 The pr i nci pl e of an opti cal -fi bre system for acousti c detecti on of PDs i n
a tr ansfor mer [ 9]
8.2.3 Opti cal -fi bre temper ature sensor s
Since the maximumoperating temperature is critical for HV equipment such as
transformersandpower cables, itisimportanttoknowtheir internal temperaturedis-
tributionandthermal environment. Thecontinuous, cyclic andshort-timeoverload
capacitiesof apieceof HV equipmentareverydependentonitshotspottemperature,
excessivevaluesof whichcauseincreasesininsulationdeterioration.
8.2.3.1 Hot-spot detection in power tr ansfor mer s
Transformer loading is mainly determined by thewinding temperatureand it is of
vital interest tothedesignersanduserstoassesscorrectlythehot-spot temperatures,
excessivevaluesof whichcauseincreasesininsulationdeterioration. Becauseof the
complicatedandcompactstructure, atransformercanhaveverydifferenttemperature
distributionsinsidethetank. Theconventional temperaturesensorscanmeasureonly
thetank surfaceandoil temperatures, whichtoalargeextent areonly averagetem-
peratureindications. Optical-fibresensorsmakeit possibletodetect thetemperature
of hot spotsdeepinsidethewindings. Electrical isolationbetweenthewindingsand
thetemperaturemeasurement is systeminherent intheuseof anoptical-fibrecon-
nectingcableof gooddielectriccharacteristics. Thesmall sizeof thesensorsmakesit
possibletoreachmanylocationsinthewindings. Anoptical fibrecanalsobewound
with thetransformer winding, so enabling thetemperaturedistribution to beread-
ily measuredandhotspotsidentifiedusingadistributedtemperature-sensing(DTS)
technique. In such asensing system, theoptical fibreacts as both thesensing and
thedata-transmittingmedium. Thetemperature-distributionprofileof atransformer
windingdetectedbyaDTSsystemisdepictedinFigure8.7.
8.2.3.2 Hot-spot detection in power cables
Figure8.8showsthetemperatureprofileof part of a10km, 230kV cablemeasured
usingaDTSdetector. Thereareanumber of spikesanddipsalongthecablewiththe
maximumdifferenceof 20

C. Thetemperaturevariationisassociatedwiththelocal
thermal environment aroundthecable. Weather conditionsandthethermal conduc-
tivity of themediumindirect contact withthecablesurface(soil, air induct, etc.)
Sensor s for i nsul ati on condi ti on moni tor i ng 197
Temperature in C 21.8 MVA-transformer
140 100 60 20
1
0
0
0
P
o
s
i
t
i
o
n

i
n

m
0
Fi gure 8.7 The temper ature di str i buti on al ong a tr ansfor mer wi ndi ng detected
by an opti cal -fi bre sensi ng system [ 10] [ cour tesy of Sensa Ltd,
http: //www.sensa.org]
52
C
47
42
37
32
5500 6000 6500 7000 7500 8000
Distance from terminal substation
m
Fi gure 8.8 Temper ature profi l e of a 230 kV underground cabl e wi thi n a secti on of
about 3 km l ength
haveasignificant effect onheat dissipationfromthecable. Thethermal capacityof
thecableitself alsoneedstobeinvestigated, especiallywhenconsideringthedynamic
temperaturechange. Lackof understandingof theexternal thermal environmentmay
present further risktotheperformanceof undergroundcables.
Withtherecentdevelopmentof optical-fibre-basedDTS, remotecontinuousmon-
itoringisnowimplementedinanumber of cablesaroundtheworld. Theadvantages
of DTSareasfollows.
Hot spotsalongacablecircuit canbeaccuratelylocated[11].
Thecablecapacitycanbemoreeffectivelyutilized.
198 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a) (b)
(c)
Fi gure 8.9 The DTS system made by Sensa Ltd and the cabl es wi th fi bre sensor s
i ntegr ated (b) or l ai d on the sur face (c) [ 10] [ cour tesy of Sensa Ltd,
http: //www.sensa.org]
Thetemperaturerecordprovidesimportantinformationaboutthethermal history
of acablethat canthenbeusedto analyseinsulationageingfor lifeprediction
andmaintenanceplanning.
Itassistsinsystemoperationbygeneratinganalarmwhenthetemperatureexceeds
thelimit.
The real-time temperature measured is important for cable emergency load
management [12].
Figure 8.9 shows the DTS system widely used in underground HV power
cables [12]. Theoptical fibrecanbeintegratedinsidethecable(under thesheath),
attachedtothesurfaceor intheduct next tothecable.
Therearevarious optical-fibretechniques for temperaturesensing. Figure8.10
shows thesensingsystemusingoptical timedomainreflectometry. Thelaser light
isinjectedthroughtheoptical coupler totheoptical-fibresensor andreflectedback
at the mirrored end. While travelling along the fibre, the light is distorted by the
temperatureinacertainpattern. Thereflectedlight is thenreceivedafter coupling
backtothedetector andcomparedwiththeinjectedlightinitsphaseandmagnitude,
whichcanthendeterminethetemperatureat different locationsalongthefibre:
emission-to-samplingtimedefinesthemeasurement position;
thelight intensityat aparticular locationgivesthetemperature.
Sensor s for i nsul ati on condi ti on moni tor i ng 199
Wavelength
flattened
coupler
Sensing probe
(optical fibre)
Mirrored
end
Diffraction
grating
Broad
wavelength
laser
source
Data
processing
Detector
Optical fibre
Fi gure 8.10 The bl ock di agr am of a di str i buted opti cal fi bre sensi ng system [ 8]
Animportant characteristic of DTS is thetemperaturesensitivity, whichcanbe
increasedbytheuseof Ramanscatteredlight.
Pointsensorsarealsoavailabletodetecttemperatureatimportantpartsof acable,
e.g. cablejointsandterminations. Thedatacanbecollectedlocally andtransmitted
tothecontrol centreor laboratoryfor continuousmonitoringandanalysis.
8.2.4 Advantages and di sadvantages of opti cal -fi bre sensor s
Theprincipal advantagesof optical fibresover conventional sensorsare:
small size, lowweight androbustness;
lowunit cost for sensor;
corrosionresistance, high-tensilestrengthandhighfatiguelifeof optical fibre;
non-conductivityandimmunitytoelectromagneticinterference;
verywide-frequencybandwidthandfast responsetimes;
abilitytomeasureverywidetemperaturerange;
highspatial resolutionandhighsensitivityfor someevents;
simultaneoussensingof morethanoneparameter.
Themajor disadvantagesof optical-fibresensorsarethat:
theoptical fibremaybesensitivetodifferent influences, whichcouldrequirethe
isolationof unwantedparameters;
thedetectionandsignal-processinginstrumentscanbecomplicatedandcostly;
thelong-termstability of anoptical-fibresensor inparticular power equipment
environmentsneedsexamination;
it hasalowsensitivitytopartial dischargeat thisstageof development.
200 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
8.3 Dir ectional sensor s for PD measur ements
For HV equipment with earthed metal sheaths or enclosures such as transformers,
power cables and HV motors, electrical noise sources are mainly external for
example, coronaon theconductor connected to theHV terminal. On such acon-
ductor, PD pulsestravel fromtheequipment terminal outwardsandelectrical noise
travelsinwards, asshowninFigure8.11for ashieldedpower cable. Thepropagation
directionof thepulses is thereforeauseful indicator for identifyingthenoise. The
conventional methodfordeterminingthedirectionof atravellingwaveonaconductor
istoconnecttwoHV capacitorstotheconductor atacertaindistanceapart. External
noisewill arriveat theouter sensor first andat theother after ashort timedelay. The
distancehastobe>3mtomakethedelaysufficientlylongfor theelectroniccircuit
tomakethecomparison. Theuseof HV capacitorsfor thispurposeisexpensiveand
cancauseadditional riskstotheHV network.
8.3.1 Di recti onal coupl er sensor
EHVundergroundpowercablesarecritical forthereliabilityof electrical transmission
systems. It hasbeenfoundthat cableaccessoriessuchasjointsandterminalsarethe
weakest partsintermsof theinsulationstrength. For thevoltageabove220kV, it is
required that thePD level should bevery lowor undetectableunder theoperating
voltage. Thedirectional coupler system(LDS-5, LemkeDiagnosticsAG) shownin
Figure 8.12 provides an effective means to discriminate external interference and
coronanoisefromthetestingvoltagesupply.
Thedirectional couplersareinstalledbetweenthecablesheathandthesemicon-
ductivelayer. Therearefour outputs fromthecouplers, whicharemarkedA, B, C
andDinFigure8.12. Thenoisetravellingfromleft toright will bepickedupby A
andC outputsandfromright toleft pickedupby B andD. However, PDsfromthe
PD pulse comes
from the cable
Discharges
inside the cable
Noise travels along the
conductor to the cable
Fi gure 8.11 The PD pul se from a cabl e and the exter nal noi se tr avel l i ng i n opposi te
di recti ons
Sensor s for i nsul ati on condi ti on moni tor i ng 201
Forward Backward
Directional coupler I
Semicon layers
Conductor
Cable joint
A B C D
Directional coupler II
Outer metallic
cable sheath
Direction of signals
Fi gure 8.12 The di recti onal coupl er s i nstal l ed on both si des of a cabl e j oi nt for PD
detecti on at the hi gh measurement sensi ti vi ty [ cour tesy of Lemke Di ag-
nosti cs AG, URL: http: //www.l di c.ch, http: //www.hvdi agnosti cs.de]
cablejointwill bepickeduponlybyBandC, whicharedistinguishedfromthenoise
originatedonbothsides.
8.3.2 Di recti onal fi el d sensor
Thenew directional sensor developed jointly by Monash University and Insultest
AustraliaLtd does not requireaconnection to theHV conductor [13]. It consists
of a 1.2-metre-long rectangular detector that picks up the electrical and magnetic
fieldsassociatedwiththePDs. Thefieldsreflect thevoltageandcurrent of apartial
discharge. Withthecorrect configuration, thevoltageandcurrent signals produced
by thePDs fromtheHV equipment will havethesamepolarity, but thosedueto
external noisewill beof oppositepolarities. By comparingthepolaritiesof thetwo
signals, thedirectionof travellingpulsescanbedetected. Theprincipleof thesystem
isshowninFigure8.13.
Becauseof itshighsensitivitythesensor canbeinstalledat groundlevel, witha
distanceof 200300mmfromtheHV conductor (busbar), asshowninFigure8.14.
Thereisnoneedtoconnect thesensor totheHV conductor. Thesensorshavebeen
installedat twopower stations givingsomeencouragingresults. A ring-typedirec-
tional sensor was also developed, which can easily be installed around the cable
terminal or transformer bushingwithaclearanceof 100300mm. Investigationsare
beingcarriedout toimprovefurther themeasurement sensitivityfor thedetectionof
PDsintransformersandXLPE cables[14,15].
Outputs of thedirectional sensor of Figure8.14(b) for PDs travellingalongthe
busbar indifferent directions aregiveninFigure8.15. It canbeseenthat for aPD
travellingfromlefttoright, thefirstpeaksof theelectricandmagneticprobeoutputs
havethesamepolarity, whereas, for PDs travellingfromright toleft, thepolarities
areopposite.
202 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
F
1
(+)
E
1
(+)
F
1
()
E
1
()
A
(a)
(b)
B
E
2
()
F
2
(+)
E
2
(+)
F
2
()
A
B
Fi gure 8.13 El ectr i c and magneti c fi el ds gener ated by PD pul ses A and B
tr avel l i ng i n opposi te di recti on al ong a conductor. (a) Posi ti ve i mpul se.
(b) Negati ve i mpul se, where E el ectr i c fi el d; F magneti c fi el d. The
posi ti ve di recti on i s from l eft to r i ght [ 14]
The ring-shape
directional sensor
Discharges
inside the cable
The rectangular-shape
directional sensor
(a)
(b)
Fi gure 8.14 The new di recti onal sensor j oi ntl y devel oped by I nsul test Austr al i a and
Monash Uni ver si ty HVI CM Group: (a) the schemati c di agr am; (b) the
di recti onal sensor tested i n an HV l abor ator y under a 15 kV busbar [ 13]
Sensor s for i nsul ati on condi ti on moni tor i ng 203
300
(a)
(b)
200
100
0
m
V
100
0 20 40 60 80 100
200
300
6
4
2
0
m
V
2
0 20 40 60 80 100
ns
ns
300
200
100
0
m
V
100
0 20 40 60 80 100
200
300
400
ns
5
4
3
2
m
V
1
0 20 40 60 80 100
0
1
2
3
ns
4
6
Fi gure 8.15 The outputs of el ectr i c (l eft tr aces) and magneti c (r i ght tr aces) fi el d
probes for a par ti al -di scharge pul se tr avel l i ng al ong an HV busbar :
(a) from l eft to r i ght; (b) from r i ght to l eft [ 13]
8.4 Summar y
Thereisapressingneedtomonitor accuratelytheinsulationconditionof HV equip-
mentintheindustry. Conditionmonitoringof in-serviceequipmentiscomplexdueto
thenoiseproblemsandaccessdifficulties. Newsensorsforonlineconditionmonitor-
ingarepromisingandarethecoreof anynewtechniques. However, muchexperience
is still requiredandfurther improvements necessary beforethenewsensors canbe
widelyacceptedbyutilities.
8.5 Refer ences
1. Kreuger, F.H., Par ti al Di scharge Detecti on i n Hi gh Vol tage Equi pment (Butter-
worth, London, 1989)
2. Kuffel, E., and Zaengl, W.S., Hi gh-Vol tage Engi neer i ng: Fundamental s
(PergamonPress, OxfordandNewYork, 1984)
3. deKock, N., Coric, B., and Pietsch, R., UHF PD detection in gas-insulated
switchgear suitability and sensitivity of the UHF method in comparison
with the IEC270 method, I EEE El ectr i cal I nsul ati on Magazi ne, 1996;12(6):
206
204 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
4. Sedding, H.G., Campbell, S.R., Stone, G.C., and Klempner, G.S., A new
sensor for detecting partial discharge in operating turbine generators, I EEE
Tr ansacti ons on Energy Conver si on, December 1999;6(4):7006
5. J udd, M.D., Yang, L., and Hunter, I.B.B., Partial discharge monitor-
ing for power transformers using UHF sensors, Part 1: Sensors and signal
interpretation, I EEE El ectr i cal I nsul ati on Magazi ne, May/J une 2005;21(2):
514
6. J udd, M.D., Yang, L., andHunter, I.B.B., Partial dischargemonitoringforpower
transformers using UHF sensors, Part 2: Field experience, I EEE El ectr i cal
I nsul ati on Magazi ne, May/J une2005;21(3):513
7. J udd, M.D., Farish, O., Pearson, J .S., andHamptonB.F., Dielectric windows
for UHF partial discharge detection, I EEE Tr ansacti ons on Di el ectr i cs and
El ectr i cal I nsul ati on December 2001;8(6):9538.
8. Tapanes, E., Oanca, I., Katsifolis, J ., andSu, Q., Theinnovativeuseof optical
fibresforconditionmonitoringof highvoltageequipment, Proceedi ngsof the5th
I nter nati onal Conference on Opti mi zati on of El ectr i c and El ectroni c Equi pment,
Brasov, Romania, 1517May1996, pp.124
9. Blackburn, T.R., Phung, B.T., J ames, R.E., Optical fibre sensor for partial
dischargedetectionandlocationinhigh-voltagepower transformer, Proceed-
i ngs of the 6th Annual Conference on Di el ectr i c Mater i al s, Measurements and
Appl i cati ons, 710September 1992, pp. 336
10. Nokes, Geoff, Optimisingpower transmissionanddistributionnetworksusing
optical fibre distributed temperature sensing systems, Power Engi neer i ng
Jour nal , December 1999:2916
11. Su, Q., Li, H.J ., Tan, K.C., Hotspot location and mitigation for under-
ground power cables, Proceedi ngs of the I EE on Gener ati on, Tr ansmi ssi on
and Di str i buti on, November 2005;152(6):9348
12. Li, H.J ., Tan, K.C., andSu, Q., Assessment of undergroundcableratingsbased
on distributed temperature sensing, I EEE Tr ansacti ons on Power Del i ver y,
October 2006;21(4):17639
13. Su, Q., Research and development on insulation condition monitoring in
Australia, keynotespeechatthe1stInternational ConferenceonInsulationCon-
dition Monitoring of Electrical Plant, Wuhan, China, 2426 September 2000,
pp. 1924
14. Su, Q., Development of a directional sensor for noise discrimination in par-
tial discharge measurements, Proceedi ngs of I EEE conference on Preci si on
El ectromagneti c Measurements, Sydney, May2000, pp.14
15. Su, Q., andSack, K., Non-contact directional sensors for PD measurements,
Techcom2003, Sydney, May2003, pp. 2631
16. Australian Standard AS 1026:1992, Electrical Cables Impregnated Paper
Insulated, for workingvoltagesuptoandincluding33kV
17. Australian/New Zealand Standard AS/NZS 1429.1:2000 and 1429.2:1998,
Electrical Cables PolymericInsulated
Sensor s for i nsul ati on condi ti on moni tor i ng 205
8.6 Pr oblems
1. What are the advantages of optical sensors for PD measurement on HV
equipment? Whyaretheynot widelyusedinindustry?
2. Thebandwidthof adetector cansignificantlyaffectPDmeasurementsensitivity
and the effectiveness of noise immunity. Give three examples for each case
of applications of narrow, wideand UHF band detectors in PD measurements
respectively.
3. Whydoesthedirectional fieldsensordiscussedinthischapterdetectthedirection
of travellingPD signals alongaconductor? What aretheprinciples of electric
andmagneticfieldmeasurements?
4. Thedirectionof atravellingPDpulsealongaconductor canalsobedetermined
by placingtwocapacitivecouplingsensorsat acertaindistanceapart alongthe
conductor. What is theprincipleof this directional sensor? What arethelimi-
tations of this typeof directional sensor (in frequency band, distancebetween
sensorsetc.)?
Chapter 9
Online insulation condition monitor ing
techniques
Noise-mitigationtechniques
Non-electrical onlineconditionmonitoringtechniques
Electrical onlineconditionmonitoringtechniques
Offlineconditionassessmentof theinsulationinHV equipmentisappliedexten-
sivelyinordertominimizethepossibilityof failureinservice. However, therequired
testing and measurement procedures are sometimes impractical, costly and not
indicativeof operating conditions. This chapter considers thealternativeof online
monitoringby meansof whichmorecontinuousassessment ispossibleunder oper-
ating conditions. Someof thedevelopments during thepast decadearediscussed,
including a number of new techniques aimed at overcoming many difficulties in
implementingin-servicemeasurements. Thecost, reliabilityandconvenienceof the
newsystemsneedtobebalancedagainstthesavingseffectedbyreductioninoutages
andtheextensionof lifeof theinsulatingmaterials.
9.1 T he main pr oblems with offline condition monitor ing
Althoughofflineinsulationtests, either destructiveor non-destructive, arevaluable
for theassessment of HV equipment condition, thereexist several disadvantages.
1. Theequipment hastobetakenout of service, whichmay causeanunnecessary
outage or reduction in electricity supply. For example, a large steamturbine
generator hastoruncontinuously duringthemaintenanceinterval of 34years.
A power transformer at the critical location in a power systemmay need 36
months preparationforloadstobetransferredbeforeitcanbetakenoutof service.
2. The equipment cannot be continuously monitored during operation. A fault
may occur between planned offline tests. Such periodic measurements cannot
guaranteetodetect all developingdefectsof significance.
208 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
3. Theequipmentselectrical, mechanical andchemical stressesmaybeimpossible
toduplicatewhentestingoffline. Someinsulationdefectscannotbeactivatedand
detectedunderofflineconditions. Forexample, duringinsulationtestsonastand-
still generator, mechanical vibrationdoesnotexistandthevoltageandtemperature
distributionscanbeverydifferent fromthosepresent duringoperation.
4. Thecostinvolvedinofflinetestscanbeexpensiveandtime-consuming. Sincethe
equipment hastobetakenout of serviceandall necessary safety issuesneedto
beaddressedbeforethetestingvoltagesourceandinstrumentscanbeconnected,
longpreparationandmeasurement timesarecommon.
Theabovefactors indicatethat it is preferableto monitor insulation condition
online. However, thereareadditional difficulties when testing online, such as the
noiseproblemduringPD measurements. Also, onlineconditionmonitoringresults
may beless accurate, althoughprovidingameaningful continuous record, andthe
instruments moreexpensive. Nevertheless thereis an increasein thedevelopment
of onlinecondition monitoring techniques and this is expected to lead to thewide
applicationof centralized(integrated) onlinecondition-monitoringsystems.
9.2 Noise-mitigation techniques
Noise or interference is one of the difficult problems encountered during online
condition monitoring, especially during electrical sensing such as in online PD
measurements. Many techniques are available for noise suppression, but each
may beeffectiveonly for particular equipment types andunder certainconditions.
Selection of noise-mitigation techniques should bemadeaccording to thespecific
sample/equipment, circuit connectionandnoiseenvironment.
Since1970, PDmeasurementsonHVequipmenthavegainedconsiderableimpor-
tance. Extensive research has been carried out around the world for improving
measurementaccuracyandsuppressingnoise. Noise-discriminationmethodsmainly
fall intotwocategories: softwareandhardware.
Har dwar e: Hardware development in PD measurements has been continuous
for over half a century. Various sensors and test circuits have been developed to
discriminate against noise. Although some techniques need further improvement,
theyareeffectiveandmaybeusedtoestablishareliablemeasurementsystemfor PD
onlinedetection. Themaintechniquesinhardwaredevelopment include:
1. new sensors such as optical-fibre sensors, directional sensors and VHF/UHF
sensors, whichwerediscussedinChapter 8;
2. noise-gatingtechniques;
3. differential discriminationcircuits;
4. multipleterminal measurementsandcomparison.
Softwar e: Advanceddigital signal-processingtechniqueshavebeenusedininsu-
lationconditionmonitoringareasfor aboutthreedecades[15]. Thedevelopmentof
fast digital A/Dconvertersalsomakesit possibletoprocessnoise-contaminatedsig-
nalsmorerapidlyandaccurately. Althoughmanynewandadvanceddigital processing
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 209
techniques have been used in communication, robotics and other areas, the main
techniquesadoptedinPDmeasurementsareprobablythefollowing:
1. digital band-passfiltersandadaptivefilters;
2. digital signal averaging;
3. statistical analysis, fingerprint analysisandpatternrecognition;
4. cepstrumanalysis;
5. linear prediction;
6. wavelet analysis;
7. neural networkanalysis;
8. expert systems;
9. fuzzy-logicanalysis.
Somereasonablynewandeffectivetechniquesaredescribedinthissection.
9.2.1 Noi se gati ng
NoisegatinghasbeensuccessfullyusedinPDmeasurementswithpulse-typeinterfer-
ence. Theblockdiagramof agatingcircuitisshowninFigure9.1. Themeasurement
circuitof thePDdetectorincludesfastanalogueordigital switchingcircuits/gates. The
gateiscontrolledbyatriggeringcircuit that isactivatedwhenever noiseisdetected.
Theswitch is then kept open for acertain period, e.g. 150s, depending on the
behaviour of thenoiseanditsoscillatingnature. ThePDmeasurement circuit isthus
temporarilyblockedandnonoisecanenter themeasurement system. After thenoise
pulsefinishes, theswitchis closedandready to measurethesubsequent PDs. The
noisefromcoronadischargesusuallylastsfor 15sfor high-frequencybandmea-
surements(upto10MHz) and1020sforlower-frequencybandmeasurements(up
to500kHz). Inordertoblocknoisefromthyristorexcitersof agenerator, theblocking
windowshouldbe20100slong. After thehighnoiseisblocked, theamplifier in
thedetector circuit canbeadjustedmanually or automatically toincreasethesignal
magnitudebeforesending it to theA/D converter. This can normally improvethe
measurement accuracyandsensitivity.
The noise-blocking process can also be illustrated in Figure 9.2, in which the
noisepulsesareblockedinseveral windowsections, leavingonlyPDsdetectableby
thesystem.
For thedetection of airbornenoiseand that entering theHV terminals, an RF
antenna, Rogowski coil or other typesof couplingsensorsmaybeusedtodetect the
PD + Noise PD
Noise triggering
circuit
Fi gure 9.1 Bl ock di agr am of a noi se-gati ng system
210 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Signals + noise (1)
(2)
(3)
(4)
Noise gate inputs
Gate operating
Signals output to
A/D converter
Noise detected
by the antenna
Noise detected by
directional sensor
Fi gure 9.2 Noi se gati ng process: (1) si gnal + noi se; (2) noi se detected by var i ous
sensor s; (3) gati ng wi ndows gener ated by noi se; (4) noi se removed by
gate l eavi ng PD si gnal s to be measured
noise. NoiseenteringtheHV terminalsmayalsobedetectedbyadirectional sensor,
asdiscussedinChapter 8.
Noise may also be removed in the software of the measurement system. The
detectednoisecanberegisteredinthecomputer programbyatriggeringchannel and
subsequentlyremoved. For example, inthepeakdetectionandA/Dconverter circuit,
if thenoiseinputchannel showsapositivesignor thenoiseexceedsthetrigger level,
anadditional bitwill beattachedtothedigital signal fromtheA/Dconverterandlater
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 211
C1 C2
HV
Z1
S1 S2
Z2
Fi gure 9.3 PD detecti on usi ng the bal anced ci rcui t
removedintheanalysisprogram. Softwarenoisegatingisveryfastandcaneasilybe
implementedinthesystem, but mayreducethespeedof themeasurement system.
9.2.2 Di fferenti al methods
A differential circuitisdesignedtocompareoutputsof twoparallel circuits. If certain
criteriaaremet, theoutputsareidentifiedasnoiseandcancel eachother. Otherwise
theoutputsareconsideredasPDsintheequipment andmeasuredaccordingly.
9.2.2.1 Balanced cir cuits
With a balanced circuit connection, PDs inside one itemof equipment and noise
entering fromtheterminal will givedifferent responses S1 and S2 at thesensors,
as shown in Figure9.3. Discharges insideC1 or C2 producedifferent amplitudes
and opposite polarities at the sensors, whereas noise fromoutside the equipment
generatespulsesof thesamepolarity, similarwaveshapeandalmostequal amplitudes
under favourableconditions. C1andC2couldbetwosingle-phasetransformers, two
bushingsof acircuit breaker, twoHV current transformersor avoltagetransformer
andadischarge-freecapacitor. Althoughthisisawell-knowntechnique, thefollowing
pointsmaybeignored, resultinginunsuccessful results.
(a) Themeasurementsystemmusthaveenoughresolution, whichisusuallyachiev-
able with wideband detectors. A poor resolution can cause overlapping of
subsequent pulses, making it difficult to identify signal pairs fromthe same
source.
(b) S1andS2causedbyterminal noisemayhavedifferentamplitudesif C1, C2and
associatedcircuitsarenotidentical inall frequencyranges. Terminal connections
mayalsocontributetothedifference. Theproblemcanbealleviatedif thetwo
measurement channels canbecalibratedby injectingat thecommonterminal
andcompensatedbyadjustingZ1andZ2.
(c) AlthoughthetimedelaybetweenS1andS2maycauseinaccuracies, itcangive
additional informationfordiscriminationof noise. Thisdelaymaybedetectable
for cablesandcertaintypesof transformer or generator windings.
(d) If C1 and C2 are items of equipment with HV terminals connected to long
busbars, possiblesignal coupling between thebusbars should beconsidered.
Thecouplingcoefficient varies in different conditions andcareful calibration
wouldbenecessary.
212 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
9.2.2.2 Noise discr imination using additional dischar ge-fr ee capacitor
DaveAllan of PowerLink, Australia[6], has developed asystemto detect online
PDsintransformer bushingsusinganadditional discharge-freecapacitor, asshown
inFigure9.4. Thecapacitor canberaisedto touchtheconductor connectedto the
terminal of thebushing. Signalsaredetectedfromthebushingcapacitor tappingand
fromthePD-freecapacitor for comparison. Noisefromoutsidethetransformer can
thenbeidentified. Thetechniquehas beenusedsuccessfully inseveral substations
upto275kV, especiallyfor CTsinwhichPDs<100pC weremeasured.
HV conductor
Resistor
shorting
switch
HV capacitive
probe
Remotely operated
pneumatic hoist
Mobile
trolley
Clip-on
transducer
Instrument
transformer
(a)
(b)
Detection
equipment
High voltage busbar
Equivalent
losses
Simplified
HV insulation
Test
point
Caplink
L
i
n
k
Phase
information
Earth grid
PD
information
PD
information
HV probe
insulation
Caplink

+
Phase
information
Internal
discharge
Making and
breaking
resistor
L
i
n
k
Fi gure 9.4 Onl i ne detecti on of PDs i n a tr ansfor mer bushi ng usi ng an addi ti onal
di scharge-freecapaci tor [ 6] : (a) pi ctor i al representati on of thedetecti on
system; (b) pr i nci pl e of the techni que [ cour tesy of Dave Al l an]
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 213
9.2.2.3 Par allel cir cuits in HV equipment
ByusingtwoHV devicesof different phaseswherehigh-frequencycouplingexists,
it is possibleto discriminatenoiseby comparingthepolarities of thefirst peaks of
PDs occurringat thesametimeon each phase, e.g. between transformer bushings
andbetweencapacitiveinstrument transformers. However, insomecasesaparallel
circuitmaybefoundinonepieceof HV equipmentfordifferential PDmeasurements.
A circuit suggestedbyMalewski [7] isshowninFigure9.5, wheredischargepulses
aredetectedbytwohigh-frequencycurrent transformersclampedaroundtheneutral
leadandthetankgroundingleadof atransformer. Thepulsesdetectedat theneutral
lead are reversed and added to the pulse fromthe tank grounding lead following
certainattenuation. Thesum(i.e. differencebetweenthetwoCT outputs) isfiltered
andmeasured. Thenoisefromoutsidethetransformer producespulsesof thesame
polarity and are cancelled in the system, whereas internal PD signals of opposite
polarityareenhanced.
Anotherpossibledifferential circuitconsistsof twoshieldspreinstalledinareactor
forPDdetection. Thetwobrassshieldsaresemicylindrical inshapeandlocatedsym-
metricallyinsidethetankwithleadsconnectingtothedetectionterminals. Theshields
actascapacitivecouplerstosensePDsinthewinding. Noiseenteringthereactoratthe
terminal will showthesamepolarityof voltagesat thedetectionterminals, whereas
HV terminal
Cg
PD
HF CT HF CT
INV
ATT
Filter
Peak
detector
+
Cd
Noise
Transformer
core
Fi gure 9.5 A di fferenti al ci rcui t for noi se di scr i mi nati on for PD measurements on
power tr ansfor mer s [ 7] , where C
d
capaci tance between wi ndi ng tur ns
and di sks, and C
g
capaci tance between HV ter mi nal shi el d, wi ndi ng
and tr ansfor mer core. Two HF CTs are cl amped around the ear thi ng
l eads of the core and the wi ndi ng respecti vel y
214 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Channel 1 Channel 2
Fi gure 9.6 The symmetr i cal shi el ds i nstal l ed i n an HV reactor for noi se di scr i mi -
nati on i n PD measurements
PDsinsidethewindinggeneratevoltagesof different polaritiesat theterminals. The
internal configurationof thereactor isshowninFigure9.6.
9.2.3 Noi se i denti fi cati on by si gnal wavefor m anal ysi s
Noise may have spectra different from those of PDs, depending on its origin,
frequency characteristics and location in relation to the detector. The following
characteristicsarehelpful inidentifyingnoisefromparticular sources.
(a) Airbornepulsatingnoisenormally has aspectrumof very highfrequency and
appearsinthedetectorashigh-frequencyoscillationsthatlastforrelativelyshort
times, e.g. pulsesfromcar/truckignitionsandsparkingatlooseelectriccontacts.
(b) Pulses fromagenerator exciter havemuch lower frequency spectrabut large
energy. They normally oscillateat 20kHz200kHz andlast for 10500s in
a wideband measurement system. The low-frequency components could be
filteredout byahigh-passfilter.
(c) Interferences frombroadcast radio andTV stations normally appear as bursts
at acertainfrequency. Theoscillationstartsat alowmagnitude, increasestoa
particular higher valuebeforeslowlyattenuatingtozero. Multipleburstscould
link or overlap, thereby making a morecomplicated envelopeof oscillation.
Themagnitudemay changesignificantly, or jumpupanddown, if magnitude-
modulatedbroadcast isinvolvedintheinterference.
(d) Because of the skin effect, a conductor increases attenuation of the high-
frequencycomponentsof apulsetravellingalongit. Thisisapparent for cables
andtransformerandgeneratorwindings. Aftertravellingacertaindistancealong
acableorwinding, thewaveformcanbesignificantlysmoothedatthewavefront.
This distortion can be useful for the identification of noise or PDs fromfar
distance.
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 215
6
4
2
0
0 200 400 600 800
Microsecond
(a) (b) (c)
1000 1200
m
V
2
4
6
8
10
12
1
0.8
0.6
Microsecond
0.2
m
V
0
0.2
0.4
0.6
1
6
4
2
520
10 20 30 40 50
540 560 580 600
Microsecond
0
m
V
2
4
6
8
10
12
2
1
990 995 1000 1005 1010
Microsecond
0
m
V
1
2
Fi gure 9.7 PDs and noi se detected on an i n-ser vi ce gener ator usi ng an HF
CT cl amped around the neutr al are i denti fi ed from thei r wavefor m:
(a) ai r bor ne noi se from motor spar ki ng; (b) noi se from the exci ter ;
(c) PDs from the gener ator
Figure9.7showsaseriesof pulsesdetectedfromagenerator. Theveryfast and
short oscillations werefromsparkingof pumpmotors inducedthroughtheair. The
longoscillatingpulsesat lower frequencywerefromtheexciter thyristor.
Figure9.8shows anewtechniquedevelopedby Professor G.C. Montanari and
TechImp[9]. Thetechniqueis basedontime-frequency transformationandfuzzy-
logic classification. The equivalent time length T and bandwidth W are extracted
fromeachpulsedetectedduringpartial-dischargemeasurements. SinceT andWare
dependent onthesourceandlocationof pulses, PDs andnoisecancrowdinafew
regionsintheWTplane, asshowninFigure9.8(b). Thepulsesfromdifferentorigins
canthenbeseparatedandanalysedinphase-resolveddistribution(d). Thefeaturesof
PDsor noisefromindividual sourcescanthenbeanalysedinsomedetail withfuzzy
logicandother methods, leadingtotheir identifications(e).
9.2.4 Mul ti pl e ter mi nal PD measurements
Insomeequipment, itispossibletodiscriminateinternal PDsfromthoseoccurredat
theterminalsor outsidenoise. For example, usingamultiple-terminal measurement
technique, PDsinsidetheequipmentcouldbeidentifiedorevenlocated. Thismethod
isespeciallyuseful for equipmenthavingwindingstructuresor distributedparameter
circuitssuchastransformers, generatorsandcables. Theprinciplecanalsobeused
onother typesof equipment.
This techniquehas beenappliedby theauthors for PD detectionontransform-
ers, especiallythosewithinterleavedwindings[11]. Thereusuallyexistsafrequency
216 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a) Original PD pattern
(b) Classification map
(c) Class pulse (d) Sub-pattern (e) Identification
Fi gure 9.8 PD separ ati on and noi se-i denti fi cati on techni que devel oped by TechI mp
[ 9] : (a) or i gi nal PDs and noi se i n the phase resol ved di str i buti on;
(b) PDs and noi se separ ated to three regi ons i n the equi val ent ti me-
l ength/equi val ent bandwi dth pl ane; (c) typi cal PD pul ses and noi se
i n the three regi ons; (d) the phase-resol ved di str i buti ons of PDs and
noi se separ ated i n (b); (e) the type of PD or noi se i denti fi ed by a fuzzy
cl assi fi cati on method [ cour tesy of Prof. G.C. Montanar i ]
rangeinwhichaninterleavedwindingcanbeapproximatedby acapacitiveladder
network[10]. Inideal situations, thepositionof achargemaybeuniquelydetermined
by theratio of thecapacitively transmittedpulses at bothterminals of thewinding.
AsshowninFigure9.9, after filteringwithapass-bandof 300400kHz, whichfalls
withinthecapacitiverangeof themodel transformerwinding, theratioismonotonous
with the injecting position. This suggests a method for locating discharges and
discriminatingnoisefromoutsideof theequipment.
Figure9.10(a) shows themeasurement results ona66kV transformer winding
presented in three-dimensional form[10,11]. Thereweretwo dischargesources in
thewindingandacoronadischargeattheterminal. Theratioof eachpair of terminal
pulseswasusedtodeterminethedischargepositionanddisplayedonthex-axis, the
dischargemagnitudeonthey-axisandthenumber of dischargesonthez-axis. It can
beseenthat thetwodischargesourcescanbeeasilyidentifiedinthe3Dgraphsand
thecoronanoiseidentifiedandremovedfromtheanalysis. ThePDSandcoronaare
separatedaccordingtotheratioof terminal pulsepair andanalysedintheformat of
number-versus-magnitude(pC) andnumber-versus-phasedistributions, asshownin
Figure9.10(b). Thetypeof PD, or otherwiseacorona, may beidentifiedfromthe
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 217
10
100150kHz
300400kHz
400500kHz
0.0 0.2 0.4 0.6
Normalised winding length
0.8 1.0
1
.1
.01
L
o
g

o
f

t
h
e

r
a
t
i
o

(
I
H
/
I
N
)
Fi gure 9.9 Pl ot of the r ati o of ter mi nal vol tages ver sus the posi ti on of si mul ated
di scharges i nj ected at var i ous posi ti ons al ong a model tr ansfor mer
i nter l eaved wi ndi ng after di gi tal fi l ter i ng wi th var i ous pass-bands
100
0
1
0
0
2
0
0
D
i
s
c
h
a
r
g
e

m
a
g
n
i
t
u
d
e

i
n

p
C
3
0
0
50
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
0
1.0
0.75
0.5
Normalised ratio
0.25
0
Fi gure 9.10 (a) A 3D patter n of PD acti vi ti es of two di scharge sources i n a 66 kV
i nter l eaved wi ndi ng and corona at the ter mi nal [ 11] . x-axi s: the r ati o
of ter mi nal di scharge pai r s; y-axi s: the magni tude i n pC; z-axi s: the
number of di scharges.
218 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
30
PD1 - N vers pC PD1 - N vers phase
20
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
10
0
0 100
Discharge magnitude in pC
200 300
30
20
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
10
0
0 60 120 180
Phase position in degrees
240 300 360
20
PD2 - N vers pC PD2 - N vers phase
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
10
0
0 100
Discharge magnitude in pC
200 300
30
20
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
10
0
0 60 120 180
Phase position in degrees
240 300 360
80
60
Corona - N vers pC Corona - N vers phase
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
40
20
0
0 100
Discharge magnitude in pC
200 300
80
60
40
N
u
m
b
e
r

o
f

d
i
s
c
h
a
r
g
e
s
20
0
0 60 120 180
Phase position in degrees
240 300 360
Fi gure 9.10 (b) Based on thei r r ati os of ter mi nal pul se pai r, the two PDs and
corona are separ ated and anal ysed i n number -ver sus-magni tude (pC)
and number -ver sus-phase di str i buti ons. The number -ver sus-phase di s-
tr i buti on of corona shows i ts di sti ngui shi ng char acter i sti c that the
maj or i ty of di scharges occur i n the AC negati ve hal f cycl e.
characteristics, for example, themajorityof coronadischargesareintheACnegative
half cycle. If theterminal disturbanceincludes internal PDs at thelineend of the
winding, softwaremethodssuchaswaveformanalysiscanbeappliedfor separating
thetwotypesof pulse. Thesetechniquescanalsobeusedtoseparatedischargesfrom
thedifferent locations, thusallowingindependent analyses. Themethodisexcellent
for assistingintheassessment of local insulationproblems.
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 219
9.3 Non-electr ical online condition monitor ing
Onlineconditionmonitoringis preferablefor continuous assessment of equipment
integrityandpreventionof incipient faults. Themorewidelyacceptedonlineinsula-
tionmonitoringtechniques for HV equipment rely onnon-electrical characteristics
such as temperature changes, acoustic emission frompartial discharges, decom-
position of theoil, including increasein gas content, and deterioration of thegas
inGIS.
9.3.1 Temper ature moni tor i ng of the i nsul ati ons
Abnormal temperatureon equipment surfacemay indicateadeterioration of insu-
lation losses, an increaseof leakagecurrent or aloosecontact. Theapplication of
optical-fibre sensing techniques for measuring the localized changes of insulation
temperatures within transformers and cables (DTS) is discussed in Section 8.2.3.
Other temperature-detection methods include thermal scanning and irreversible
temperaturelabels, whichareportableor of lower cost.
9.3.1.1 T her mal scanning
Various infraredcameras areavailableonthemarket for remotethermal scanning.
Becausethereisnoneedtoinstall asensor andthemeasurementisatasafedistance,
thethermal scanners arewidely used in theindustry. A typical infrared imager is
showninFigure9.11(a).
9.3.1.2 I r r ever sible temper atur e labels
Irreversibletemperaturerecordinglabelscanaccuratelyandeconomicallysenseand
record surfacetemperatures. Each label contains oneor moresealed temperature-
sensitivechemical indicatorsthatchangepermanentlyandirreversiblyfromsilver to
black at its calibratedtemperature. Theresponsetimeis less thanonesecondand,
dependingonthetemperaturerating, anaccuracyof 1or 2per cent isachievable.
Thelabelsareof miniaturesize, weightandthickness(nominal 0.01

), whichallows
installationinareasandonpartsthatarenotpractical for other instruments. Thelabel
can beuseful for temperaturemonitoring of small transformers, isolators, VT, CT
andswitches. A photoof alabel isshowninFigure9.12.
9.3.2 Onl i ne DGA
Dissolved-gas analysis inoil (DGA) has beenwidely usedfor thediagnosis of HV
equipment withpaper/oil insulationsuchas transformers, reactors andcablejoints
for many years andinmany cases is theprimary methodof monitoring. Although
periodic sampling and tests in thelaboratory (offlinetest) havebeen successful in
diagnosingincipient faults, thereareanumber of problems associatedwithoffline
DGA testing: for example:
(a) thesamplingfrequency hastobeconservatively selectedtobeeffective; even
so, offlineperiodictestsmayfail topredict somefailures;
220 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a)
(b)
Fi gure 9.11 Remote ther mal scanni ng; (a) the i nfr ared i mage of LV connector s
Phase A shows a hi gher temper ature i ndi cati ng a probl em of l oose
contact; (b) an HV ai r bl ast ci rcui t breaker wi th the i mage show-
i ng a l oose contact at the ter mi nal [ cour tesy of Ther mascan Ltd,
http: //www.ther mascan.co.uk/home.html ]
(a) (b)
Fi gure 9.12 The TMC 8 l evel temper ature l abel (a) and i ts appl i cati on to the moni -
tor i ng of a machi ne (b) [ cour tesy of Ther mogr aphi c Measurements Co.
Ltd, http: //www.t-m-c.com]
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 221
(b) excessivenumbersof testsmayincur unacceptablecosts;
(c) careisnecessaryinoil samplingtoprevent contamination.
ThemaingasesproducedbydifferentfaultsareshowninFigure9.13. Insummary,
thecriteriafor keygasconcentrationsaregiveninTable9.1[12].
For someimportantHV transformers, onlinecontinuousmonitoringof keygases
ispreferable. Thismakesitpossibletomonitor fast-developingfaultsandtoevaluate
theirprogresssothatthenecessaryresourcescanbemarshalledfortherefurbishment
or replacement of thetransformer.
C2H2
C2H4
C2H6
CO
CH4
0 20 40 60
Overheated cellulose Overheated oil
Corona in oil Arcing in oil
80
H2
%
100
G
a
s
C2H2
C2H4
C2H6
H2
CH4
0 20 40 60 80
CO
%
100
G
a
s
C2H2
C2H4
C2H6
H2
CH4
0 20 40 60 80
CO
%
100
G
a
s
C2H2
C2H4
C2H6
CO
CH4
0 20 40 60 80
H2
%
100
G
a
s
Fi gure 9.13 The profi l e gases for di fferent faul ts [ 12]
Tabl e 9.1 Cr i ter i a for key gas concentr ati ons [ 12]
Gas Concentration(ppm) Typeof fault
Hydrogen >50 Partial discharges
COandCO
2
>300and3000 Degradationof paper
EthyleneC
2
H
4
>100 Overheatingof oil
AcetyleneC
2
H
2
>2 Arcinginoil
Furan >5ppm Ageingof paper
(andDP >200)
222 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Fi gure 9.14 Onl i ne hydrogen moni tor i ng on a tr ansfor mer [ 13] [ cour tesy of GE Oi l
& Gas, http: //www.gepower.com/prod_ser v/products/substati on_md/
en/moni tor i ng_i nstr _sys/hydr an.htm]
A particular instrument for continuousmonitoringisshowninFigure9.14. Nor-
mally, oneor afew gases aremonitored by theinstrument. TheHydran, madein
Canada, isspecificallydesignedtodetect hydrogenandcarbonmonoxideintheoil.
Other instrumentsmaydetect additional gases, but aremorecostly.
9.3.3 Acousti c-based techni ques for PD detecti on
9.3.3.1 Tr ansfor mer windings
Theultrasonic impulses transmittedthroughthemechanical structureof awinding
associatedwithPDsmaybeusedfor location. If it isimpossibletoutilizeelectrical
pulsesfromthePDandthephysical conditionsarefavourableadischargesourcemay
belocatedusinganumber of appropriatelyplacedacoustictransducers. Reasonable
accuracy is obtained if reflections are not significant and only direct path signals
arepresent [14,15]. Triangulationlocationsystems may also beusedto pinpoint a
location in space, provided it does not liedeep within thetransformer or below a
certainmagnitude. Anadvantageof thesemethodsistheir independenceof electrical
noise. In somecases thedetecting sensitivity has not been satisfactory, especially
when thedischargesources aredeep insidethecoils. Enhancement of thesignals
mayberealizedbydigital signal processingandaveraging. However, if thereismore
thanonedischargesourceandlargeinterferenceispresent, averagingof theacoustic
signalsfromdifferent sourcesislikelytoresult inconfusingresults. Further workis
still beingcarriedout, inparticularwithrespecttoimprovingsensitivityandaccuracy,
locationof inner windingfaultsandinterpretationof characteristicoscillationswith
bothexternal andinternal acoustictransducers.
9.3.3.2 Cable ter minations
AcousticPDmeasurementshavealsobeendevelopedfor cableterminalsandproved
successful. A particular detector consists of afibreglass stick or pipe, asensor and
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 223
Fibreglass stick
Acoustic sensor
Fi gure 9.15 Acousti c PD detector wi th l ong fi bregl ass sti ck used for vol tages
up to 33 kV [ 17]
amonitor, as shown in Figure9.15. Sincetheinsulated stick/pipecan bedirectly
applied to thecableterminal and works as asound-waveguide, thesensitivity of
measurement ishigher thanfor other acoustic sensorsplacedat adistancefromthe
dischargesource.
9.3.3.3 Por table acoustic PD detector
Acousticgunsarealsousedforthedetectionof dischargesaroundequipmentsurfaces
suchastheHVterminals(forcoronadischarges). AsshowninFigure9.16, thebattery-
poweredportablePDdetectorcanbeusedforPDdetectiononHVequipment. Withthe
directional feature, thedetector canhelptolocatePDsourcesandalleviatethenoise
probleminsitemeasurements. However, for internal dischargesintheinsulation, the
Fi gure 9.16 A batter y-powered detector used i n PD detecti on and l ocati on of
HV equi pment [ cour tesy of May El ektroni k GMBH, http: //www.may-
el ektroni k.de/]
224 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
sensitivitymaybereduced. Nevertheless, theyareusedwidelyinthelaboratoryand
at siteto detect coronadischarges dueto insulation design problems, cracking on
porcelaininsulatorsandbadcontact, etc.
9.4 Online acoustic/electr ic PD location methods for tr ansfor mer s
Examples aregiven of theapplication of acoustic transducers used in conjunction
with(i) theassociatedelectrical pulsemeasuredat theterminal (9.4.1) and(ii) the
radiatedelectrical pulsedetectedbyasensor withinthetransformer tank(9.4.2). The
viability of themethodsdependsonknowledgeof thevelocity of soundthoughthe
oil (approximately1.4m/ms) andthefactthattheelectrical pulsesaredetectedalmost
instantaneously.
9.4.1 Acousti c tr ansducer s and wi ndi ng ter mi nal measurements
Theusual combinedtestsarecarriedoutwithanumber of acoustictransducersexter-
nal to thetank and records obtained at theterminal (bushing tap) of theelectrical
pulsesaswell asthosefromthetransducers. Bytriggeringontheacousticpulsesand
determiningtheprior time-delayrangeexpectedfor theelectrical signal estimated
fromthe tank dimensions and the structural layout it is possible to obtain an
indicationof theprobablelocationof aPDsource.
An improvement on the above procedure is to measure electrical PDs at both
theHV terminal andtheneutral andapplythesignal separationmethodsdetailedin
Section9.2.4inorder toachievesignal enhancement. Thisalsoenablesanestimation
of thePDmagnitudeat thesourceif withinthewinding. A newtechniquehasbeen
developedusingacombinedelectric/acousticmethod[15]. Themeasurementcircuit
for a330kV single-phasetransformer isshowninFigure9.17. After digital filtering
of thedischargesignals measuredat theHV terminal andneutral, theratio of each
pair of signalswasusedtoseparatethemaccordingtotheir differentsourcelocations
anddistinguisheachfromexternal interference. Theelectrical signalsthatwerelikely
tobeoccurringat thesamedischargesourcewereaveraged, asweretheir associated
acoustic signals, resultinginasignificant improvement inthedetectionsensitivity
and location estimateas shown in Figure9.18. By suitablecalibration techniques
andknowledgeof theprobablelocationwithinthewinding, anestimateof thePD
magnitudeat itssitemaybepossible.
9.4.2 Appl i cati on of i nter nal combi ned acousti c and VHF/UHF
tr ansducer s
A particular systemfor locationof PDs intransformers has beendevelopedby the
University of Technology, Sydney, and Siemens Ltd [16]. The technique utilizes
speciallydesignedcompositeacoustictransducers, whicharematchedtotheacoustic
impedanceof theoil and haveashort ring-down time. Around each transducer is
locatedacapacitiveringfor detectionof theradiatedRF pulse. Thecapacitivesensor
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 225
To HV busbar
HF CT
Transducer
HF CT
Digital
oscilloscope
Computer
Fi gure 9.17 Ci rcui t connecti on of PD measurement on a 330 kV si ngl e-phase
tr ansfor mer usi ng a combi ned el ectr i c/acousti c PD detecti on method
operatesinatunedcircuitwithin10100MHz. Thecombinedunitisinsertedthrough
thetransformer tankside/lidmountedonaninspectionplateasshowninFigure9.19.
A modifiedarrangement allowsinsertionthroughanoil drainvalve. Thefaceof the
combinedtransducer iskept flushwiththeinner surfaceof thetankthusminimizing
thepossibilityof formingahazardduringthetransformer operation.
Due to the internal mounting, interference effects are reduced and sensitivity
increased compared with externally located transducers. Acoustic matching of the
sensor is good. However, theversatility in positioning is reduced. Also, theVHF
sensor is not calibrated in terms of pCs. The systemcan incorporate continuous
remotemonitoringof anychangesintheoutput of anumber of acoustictransducers,
whichmayidentifyaPDbasedoncorrelationof thesoundandRF signals. Thesignal
processingincludesestimatesof thelocationsinrelationtothesensors.
9.5 Electr ical online condition monitor ing
Althoughofflineinsulationtests, eitherdestructiveornon-destructive, areveryuseful
for the assessment of HV equipment condition, as indicated in Section 9.1, it is
preferabletouseonlinetests. Withelectrical onlinetests, theinsulationdefectscan
bedirectlyassessedfromthevaluesandchangesof theelectrical properties. However,
sinceall HV equipmentisunderhighelectrical stress, theelectrical interferencessuch
as corona and harmonic voltages always make online electrical monitoring more
difficult. Advancedtechniquesinhardwaredesignandsoftwareanalysishavetobe
utilizedinachievingthedesiredresults.
226 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
(a)
(b)
V
V
V
V
ms
ms
ms
ms
0.8
0.6
0.4
0.2
0.0
0.2
0.4
0.6
0.8
1.0
0.0001
0.0005
0.0007
0.0013
0.0019
0.0025
0.7
0.6
0.3
0.1
0.1
0.3
0.5
0.7
0.9
0.002
0.003
0.0 0.5 1.0 1.5 2.0
0.0 0.5 1.0 1.5 2.0
0.0 0.5 1.0 1.5 2.0
0.0 0.5 1.0 1.5 2.0
0.001
0.001
0.002
0.003
Fi gure 9.18 The el ectr i cal and acousti c PD si gnal s si mul taneousl y measured on a
330 kV tr ansfor mer [ 15] : (a) a si ngl e record; (b) after aver agi ng 30
ti mes the di scharges from the same source (X mi l l i second, Y vol t)
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 227
Transformer
tank
Electronic
circuits
Signal outlet
Oil Inspection hole
Transformer core
Acoustic and UHF
transducer head
Fi gure 9.19 Combi ned acousti c and RF tr ansducer s for i nstal l ati on i n i nspecti on
hol e [ 16]
9.5.1 Onl i ne di el ectr i c di ssi pati on factor and capaci tance measurements
AsdiscussedinChapter 2, dielectricdissipationfactor (DDF) or dielectriclossangle
(DLA), andcapacitanceareimportant indicatorsof insulationdeterioration. During
laboratory tests, various instruments can be used to accurately measure DLA and
capacitanceof HV equipment (seeChapter 6). However, accuratemeasurement of
DLA andcapacitanceismoredifficult for equipment inservice.
9.5.1.1 Pr inciple of DL A and capacitance measur ement
For alosslesscapacitor, thecurrentleadsthevoltage90

. Practically, thelossalways
exists and changes the phase difference to slightly smaller than 90

, as shown in
Figure9.20. Thevoltageof anHV devicecanbemeasuredthroughavoltagetrans-
former(VT)andthecurrentflowingthroughthedevicebyacurrenttransformerinthe
earthlead. For mostCTsandtransformer/reactor bushings, theDLA andcapacitance
can bedeterminedby thevoltageandcurrent measuredbut not to theaccuracy as
describedinChapter6. Itshouldbenotedthat. becausethephaseshiftisverysmall, it
isdifficulttomeasureaccuratelytheDLA andcapacitanceusingthismethod. Also, as
itchangeswithtemperatureandother environmental parameters, theDLA of apiece
of HV equipment maychangesignificantlywithtime, whichmakesthedetectionof
equipment insulation deterioration less sensitive. Although this has been themain
disadvantage, fromsophisticatedcalibrationandtemperaturecompensationreason-
ablereliability canbeachievedfor thedetectionof insulationdefectsthroughtrend
analysis.
9.5.1.2 Relative measur ement method
ThedifferenceinDLA or phaseof thecurrent(after ashiftof 120

or 240

) between
different phases of equipment can beused to detect moreaccurately an insulation
228 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
I
c

b
I
a
I
b
V
Fi gure 9.20 Phase di fference between vol tage and cur rent for a l ossl ess capaci tor,
and for phases a and b wi th di fferent l osses
Tabl e 9.2 I denti fi cati on of the faul ty phase from compar i son exampl es

ab

bc

ca
Remarks
Comparisonbetween Larger Larger Small Phasebmayhave
phases aninsulationproblem
Changewithtime Small Larger Larger Phasecmayhave
aninsulationproblem
probleminonephase. Bymonitoringthedifferencebetweenphasesanditschanges
withtime, thefaulty phasecanbeidentifiedandrepairedaccordingly, as shownin
Table9.2.
9.5.1.3 Pr actical online measur ement system
A fewintegratedDLA andcapacitance-monitoringsystemshavebeendevelopedin
theworld for onlinecontinuous condition monitoring of CTs, capacitiveVTs and
transformer bushings. TheonedevelopedbyDaveAllanof PowerLink, Australia, is
showninFigure9.21[6]. Thesystemcomprisescurrent(voltage) sensorstomeasure
thecurrent waveform, phaseencoderstopick upthephaseinformationfromwave-
shapeconditioning, an optical-fibretransmission systemandDLA calculation and
displayunit. Thearrangement hasbeeninstalledinanumber of switchyards.
9.5.2 Onl i ne l eakage cur rent measurement
Most outdoor HV equipment has porcelain housings, e.g. CTs, VTs, bushings and
surgediverters. Duetocontamination, moistureorcrackingof theporcelainhousings,
surfacecreepagedischarges/currentsmay occur, whichcansignificantly reducethe
electric strength. Online monitoring of the leakage current can be used to detect
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 229
HV bus
Test
point
HV
bushing
insulation
Other
CLU
inputs
CLU
Caplink
Bushing
FOL FOL
SR
Modem Modem RDU
Control room environment
FOL = Fibre optic link
CLU = Combination logic unit
PE = Phase encoder
SR = Signal resolver
RDU = Remote display unit
P = Microprocessor
Telephone
network
PE Caplink
HV
environment
V
a
V
b
V
c
P
CLU
Fi gure 9.21 An onl i ne DLA moni tor i ng system for CTs, CVTs and tr ansfor mer
bushi ngs i n a substati on [ 6] [ cour tesy of Dave Al l an]
Metal foil attached
to the surface
To HV busbar
A
Fi gure 9.22 Onl i ne l eakage cur rent moni tor i ng
incipient failure. Thedetectingcircuit isdepictedinFigure9.22. If thelow-voltage
endof theporcelainhousingis directly grounded, anadditional electrodemadeof
metal foil maybeinstalledaroundthehousingsurfacetodetect theleakagecurrent.
Similar systemsareusedfor monitoringhigh-voltagesurgediverters.
230 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
GIS enclosure
Protrusion on
live part
Conductor
Moving
particles
Protrusion on
earth part
Particle
on spacer
Voids
Spacer
Fi gure 9.23 The mai n defects i nsi de GI S resul ti ng i n PDs [ 18]
9.5.3 El ectr i cal onl i ne PD detecti on
9.5.3.1 GI S PD detection
DuetothecompletelysealedHV circuit andtheuseof pressurizedgas(SF
6
) asthe
maininsulation, GIShastheadvantageof acompact structureandlowmaintenance
requirements. AlthoughGISareusuallyveryreliable, onlinePDdetectionisneces-
saryfor commissioningtestsandconditionmonitoringfor detectinganydeveloping
defects.
Metallic-particle contaminants are the most critical defects in GIS. They may
developfrom:
1. mechanical abrasions;
2. movement of conductorsunder loadcycling;
3. vibrationsduringshipment andinservice.
Defectsandparticlesmaybecreatedduringtheproductionof theunitinthefactory
or duringassemblyat site. Innormal operation, particlescouldalsobeproducedby
fast-earthingswitchesordisconnectorswitches. ThemaindefectsinGISaredepicted
inFigure9.23.
Conventional testmethods(IEC60270)areusuallyappliedduringcommissioning
but thesetendtobeinsensitiveduetothepoor couplingbetweenthesensorsandthe
PDsites. Toobtaintherequiredsensitivitiesof <5pC, UHF sensorsandassociated
systems, including remote monitoring, have been developed [18]. The frequency
band might beup to theGHz rangedepending on different design and sensitivity
requirements. Atsuchhighfrequencies, eachsensorcancoveronlyacertaindistance
of thecompartment/busbar. Thereforemultiplesensorshavetobeinstalledtomonitor
thewholeGIS. Typical UHF sensorsusedinGISareshowninFigure9.24.
9.5.3.2 Tr ansfor mer PD detection using VHF and UHF sensor s
Inadditiontoonlineacousticlocationof PDsintransformers, asindicatedearlier, a
numberof organizationshavedevelopedsystemsformeasuringthePDvaluesonline.
Thesearebasedon:
1. single-endedmeasurementsusingthebushingtaponlyandadvancedsignal pro-
cessing including phase resolved analysis for identification and separation of
interference[19];
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 231
Signal
output
~120 mm
(a) (b)
Fi gure 9.24 UHF sensor s used for PD detecti on i n GI S: (a) the sensor attached to
the ear thi ng swi tch i nsi de GI S; (b) the sensor used exter nal l y on the
obser vati on wi ndow [ 18]
2. multiple-terminal measurements includingneutral andtank groundconnections
andappropriatesignal processingtoisolateinterferenceeffects[20].
Recent techniques aimed at reducing interference problems have applied the
knowledgeacquiredfromGISinvestigationsinvolvingUHF detectionmethods. The
KEMA patentedsystem[21] includescapacitor-typesensors, asdoesthatdeveloped
bytheUniversityof Strathclyde[22]. ThelatterstudiesshowthatatUHF itshouldbe
possibletolocatePDsandcharacterizetheminrespectof energyandphaseposition.
9.5.3.3 Online PD detection on power cables
Therearemainlytwotypesof power cable paper/oil andXLPE. Thesehavesignif-
icantly different PD levels. For XLPE cables, theallowedPD magnitudeshouldbe
under550pCdependingonthedifferentvoltagelevels. Ontheotherhand, paper/oil
cables can tolerate larger PDs of even a few hundred pC. Therefore, online PD
measurements at theterminal could besuccessful for paper/oil cables if appropri-
atetechniquesareusedtodiscriminateagainst interference. For XLPE cables, more
advancedtechniquesshouldbeusedtoachieveahigher S/Nratio. Probablythemost
widely acceptedtechniquefor onlinePD detectionis VHF/UHF sensing, whichis
specificallyusedfor PDdetectiononcablejoints. It hasbeenfoundthat cablejoints
andterminationsarestill theweakest part of acablecircuit.
Inorder todetect PDsincablejoints, prefabricatedcapacitivesensorsimbedded
insidethejoint maybeused, givinghigher sensitivityandreducingnoiseproblems.
However, theseinternal sensors tendtomakethecablestructuremorecomplicated
andmoreexpensive.
Sensors can also be installed outside the cable joint, such as the one devel-
oped by someJ apanesecompanies [23]. Theprincipleof this techniqueis shown
inFigure9.25. Twometal-foil electrodesareinstalledaroundthecableoneachside
of thejoint. Thecapacitancebetweenthefoil andtheconductor formsthecoupling
232 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Metal foil Metal foil
Insulation of
the sheath
(a)
(b)
Impedance
Cable condutor
C
1
C
1
Sheath
C
b
C
a
C
b
C
PD current
PD signal
C
Z
Z
PD signal
Fi gure 9.25 Onl i ne PD detecti on on cabl e j oi nt: (a) schemati c ci rcui t connecti on;
(b) the pr i nci pl e and mai n PD cur rent fl ow path [ 23,24] , where: C
capaci tance between cabl e sheath and the metal foi l whi ch i s i nstal l ed
on both si des of the j oi nt; C
1
capaci tance between the conductor and
cabl e sheath; C
a
, C
b
and C

b
capaci tances associ ated wi th a PD i n
the cavi ty i n cabl e i nsul ati on; Z i mpedance for PD detecti on
circuit for PD current to flow through the impedance connected to the two elec-
trodes. A narrowfrequencybandof 100kHz isusedandthecentreisapproximately
25MHz. Usingthisdetector, asensitivity of better than5pC hasbeenachievedfor
PDmeasurementsonin-servicecablejoints.
A similar systemhasbeendevelopedbyLemkeDiagnostics(LDIC) withamax-
imumbandwidthof 2600MHz. Suchmethodsarebeingappliedexperimentallyby
anumber of researchgroups.
9.5.3.4 Online PD detection in lar ge r otating machines
Generatorshavecomplicatedwindingstructures. A simplifiedwindingconfiguration
and terminal connection areshown in Figure9.26. Discharges and interferenceof
variousoriginscanproduceimpulsivecurrentsintheneutral lead.
The high-frequency components of slot discharges (1) and those on the other
endof thegenerator (2) will besignificantly attenuatedbeforereachingtheneutral,
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 233
Stator
(2)
(1) (3)
(5)
Neutral transformer
Neutral RF CT
(4)
A
B
C
Fi gure 9.26 Di scharge sources from i nsi de and outsi de of the gener ator
whereasthelow-frequencycomponentsdonot remarkablyreducetheir magnitudes.
High-frequency components of the discharges on the HV terminal end (3, 4) and
interference(5) may easily becoupledthroughtheoverhangsectionto theneutral,
resulting in a difference in the low- and high-frequency responses in the neutral
currents.
9.5.3.5 RF cur r ent detector
A radio frequency current transformer mounted on theneutral grounding lead has
successfully beenusedto detect internal generator arcing[2527]. A prototypeRF
onlinemonitor wasdevelopedbyEmeryet al . [27], whichcancontinuouslymonitor
theneutral current and alert theoperator if thearcing level becomes destructively
high. Thistechniquemay alsobeusedfor PD detectionif theinterferenceproblem
canbeovercome.
Another methodusingRF techniqueswasdescribedbyMiller andHoggintheir
paper [28]. Their results demonstrated that with theaid of Fourier transformation
techniquesthechangingpulseshapemaybeusedtolocatethesourceandclassifythe
typeof discharges. Furtherinvestigationsarenecessaryfortheapplicationtopractical
generators.
9.5.3.6 T he Ontar io Hydr o Par tial Dischar ge Analyser (PDA)
ThePartial DischargeAnalyser (PDA) developedby Ontario Hydro is atechnique
for the measurement of PD pulse-height distributions in rotating machines in the
presenceof static excitationandother common-modenoise[29,30]. Themeasure-
mentsareperformedby connectiontocouplingcapacitorspermanently installedon
theterminalshigh-voltageconductors. Eachphasehastwocapacitorsconnectedas
directional couplers. Thelengthof thecoaxial cablefromeachcoupler isdetermined
suchthat unwantedpulsesemanatingfromtheoutsidesystemarrivesimultaneously
at thedifferential input of thePDA, resultingincancellationof thepulses, whereas
pulsesfromwithinthegenerator arrivefromeachcoupler withdifferentdelaytimes,
resultinginameasurement. Thecircuit connectionisdepictedinFigure9.27.
234 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
Noise entering HV
terminal
PD in the
winding
Parallel 1 Parallel 2
Differential
amplifier
Coupling
capacitor
Coaxial cable
t
1
t
2

1
Fi gure 9.27 Ontar i o Hydro di fferenti al coupl er i nstal l ati on [ 29] (note: for exter nal
noi se t
1
= t
2
and PD i n wi ndi ng
1
=
2
)
ThePDAdetectsthesteeppulses, whichareconsideredtobecapacitivelycoupled
betweentheendwindingsof eachturn, aslittleof thehigh-frequencycomponentsare
conductedthroughtheslots. Fromregular measurements onanindividual machine
andcomparisonwiththehistorical data, thechangesindischargeperformancecanbe
monitoredandtheinsulationconditionevaluatedfrompreviousknowledgeof itself
andsimilar units.
9.5.3.7 Ontar io Hydr o I RI S Tur bo-Gener ator PD Analyser (T GA)
Due to the different structural configurations of turbo-generators and larger inter-
ference levels, a stator slot coupler for sensing PDs in turbo-generators has been
developed[31]. Thesensor isessentiallyadirectional electromagneticcoupler fitted
intheslot, under thestator windingslot wedges betweenlayers of wedgepacking.
Becauseof its location closeto theslot conductors, thesensor is sensitiveonly to
partial dischargesbut not theinterferencebecauseof thefilteringeffect of thestator
structure. Output voltages of thesensor havevery highfrequency (risetimebeing
withinafewns), whicharemeasuredby aspecially designeddetector. Thelayout
of thesensor isshowninFigure9.28. Thenumber of sensorsisnormallysixinthis
system.
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 235
~300 mm
Insulation board
Signal coupling
conductor
Ground plane
Coaxial cable to detector
Fi gure 9.28 The si mpl i fi ed schemati c of the stator sl ot coupl er [ 31]
9.5.3.8 UNSWSM HEA PD detector
Inthelate1980s, ateamattheUniversityof NewSouthWales, incollaborationwith
engineers fromSnowy Mountains Hydro-Electric Authority developedaPD detec-
tionmethodfor hydraulicgenerators[32,33]. Themethodisbasedonthetravelling
waveandcapacitively coupledPD signalsdetectedat theHV terminal of thestator
winding. Thecapacitivelycoupledpart of aPDinthewindingappearsat thetermi-
nal almost instantaneously, whereas, thetravellingwavepart takes upto 78s to
reachthesameterminal dependingonthePDsourcelocationandthestator winding
structureandsize. Thetimedelay isthenmeasured, whichcanbeusedtodiscrimi-
natePDfromnoiseaswell aslocatethedischarge. Thecircuit connectionisshown
in Figure 9.29 and the PDs detected on a 200MW hydro generator are shown in
Figure9.30.
Generator HV terminal
PDA coupling capacitor
HF CT
Neutral transformer
Digital storage CRO
Computer
Amplifier
To PDA
Two parallel
phase windings
Fi gure 9.29 Test ci rcui t connecti on (onl y one phase wi ndi ng i s depi cted)
236 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
40
30
nC
0.00
N HV Normalised winding length
0.20 0.40 0.60 0.80 1.00
20
10
0
Fi gure 9.30 Apparent charge magni tudes ver sus esti mated l ocati ons of the PDs
detected on a 200 MW gener ator showi ng two possi bl e PD sources at
87.4 per cent and 98.5 per cent of the wi ndi ng [ 33]
9.6 Summar y
Althoughconditionmonitoringonin-serviceequipment isdifficult duetothenoise
problem, new techniques have been developed that make the operation of HV
equipmentsaferandmorereliable. Therearetwoformsof onlinemonitoring: (i) peri-
odical measurementsutilizingbuilt-insensorswithportableinstrumentation; and(ii)
completesystemscontinuouslyrecordingthedata. Onlinecontinuousconditionmon-
itoring should beintegrated into acentralized monitoring system, which needs an
advancedcommunicationsystemusingoptical fibres, digital radioandeveninfrared
techniques. Thishasplacedachallengingtask for futureconditionmonitoringengi-
neers who should be well equipped with knowledge in these areas as well as the
operationandphysical structureof theHV equipment.
9.7 Refer ences
1. Konig, G., and Feser, K., A new digital filter to reduce periodical noise in
partial dischargemeasurements, presentedat the6thInternational Symposium
on High Voltage Engineering, New Orleans, 28 August1 September 1989,
paper 43.10
2. GA-1 Gati ng Ampl i fi er , instrument manual, Insultest Australia, November 1997
3. Borsi, H., andHartje, M., Newmethodstoreducethedisturbanceinfluenceson
thein-situpartial dischargemeasurement andmonitoring, presentedat the6th
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 237
International SymposiumonHighVoltageEngineering, NewOrleans, August
September 1989, paper 15.10
4. Su, Q., Applicationof digital signal processingtechniquesfornoisesuppression
inpartial dischargemeasurements, Proceedi ngs of the 4th I nter nati onal Confer -
ence on Proper ti es and Appl i cati ons of Di el ectr i c Mater i al s, Brisbane, 38J uly
1994, pp. 6025
5. Su, Q., Anadaptivefilteringmethodfor noisesuppressioninpartial discharge
measurements, presentedattheInternational ConferenceonElectrical Insulation
and Dielectric Phenomena, Pocono Manor, Pennsylvania, October 1993, pp.
4816
6. Allan, D.M., Brundell, M.S., and Boyd, K.J ., Newinsulation diagnostic and
monitoringtechniquesforHV apparatus, presentedatthe3rdInternational Con-
ferenceonPropertiesandApplicationof InsulationMaterials, Tokyo, J uly1991,
pp. 44851
7. Malewski, R., Douville, J ., andBelanger, G., A diagnosticsystemforin-service
transformers, CIGRE 27, Paris, AugustSeptember 1986, paper 1201
8. Black, I.A., andLeng, N.K., Theapplicationof thepulsediscriminationsystem
tothemeasurementof partial dischargesunder noisecondition, presentedatthe
IEEE International SymposiumonElectrical Insulation, J une1980, pp. 16770
9. Montanari, G.C., Cavallini, A., and Puletti, F., A new approach to partial
dischargetestingof HV cablesystems, I EEE El ectr i cal I nsul ati on Magazi ne,
J anuary/February2006;22(1):1423
10. Su, Q., andJ ames, R.E., Analysisof partial dischargepulsedistributionalong
transformer windings using digital filtering techniques, I EE Proc. C, Gener.
Tr ansm. Di str i b., September 1992;139(5):40210
11. J ames, R.E., Phung, B.T., and Su, Q., Application of digital filtering tech-
niquestothedeterminationof partial dischargelocationintransformers, I EEE
Tr ansacti ons on El ectr i cal I nsul ati on, August 1989;24(4):657668
12. Duval M., anddePabloA., Interpretationof gas-in-oil analysisusingnewIEC
publication60599andIECTC10database, I EEE Tr ans. Di el ectr. El ectr. I nsul .,
March/April 2001;17(2):3141
13. Ritchie, D., Theopportunitiespresentedbyonlinetransformer gasmonitoring,
presentedat theDobleConference, Boston, April 1999
14. J ones, S.L., Thedetectionof partial dischargesintransformersusingcomputer
aidedacousticemissiontechniques, recordof theIEEEInternational Conference
onElectrical Insulation, Toronto, Canada, J une1990
15. Blackburn, T.R., J ames, R.E., Su, Q., Phung, T., Tychsen, R., andSimpson, J .,
An improved electric/acoustic method for the location of partial discharges
in power transformers, Proceedi ngs of the 3rd I nter nati onal Conference
on Proper ti es and Appl i cati ons of Di el ectr i c Mater i al s, Tokyo, J uly 1991,
pp. 113235
16. UnsworthJ ., BoothN., TallisD., andBall K., Evaluationof novel onlinepartial
dischargemonitor for highvoltagetransformersduringoperation. Presentedat
the16thNordic SymposiumonElectrical Insulation, Copenhagen, J une1999,
pp. 13340
238 Condi ti on Assessment of Hi gh-Vol tage I nsul ati on
17. Kind, D., andKarner, H., Hi gh Vol tage I nsul ati on Technol ogy (ViewegandSon,
Braunschweig, 1985)
18. J udd, M.D., Farish, O., Pearson, J .S., and Hampton, B.F., Dielectric win-
dows for UHF partial dischargedetection, I EEE Tr ansacti ons on Di el ectr i cs
and El ectr i cal I nsul ati on, December 2001;8(6):9538
19. Contin A., et al ., Digital detection and fuzzy classification of partial dis-
chargesignals, I EEE Tr ansacti ons on Di el ectr i cs and El ectr i cal I nsul ati on, J une
2002;9(3):33548
20. Su, Q., andVaillancourt, G.H., Computer-basedmulti-terminal partial discharge
measurementsontransformers, presentedattheIEEEInternational Symposium
onElectrical Insulation, Montreal, J une1996, pp. 614
21. Rutgers, W.R., Ross, R., andvanRijn, TheoG.M., OnlinePD detectiontech-
niquesfor assessment of thedielectricconditionof HV components, presented
at theIEEE 7thInternational ConferenceonSolidDielectrics, Eindhoven, J une
2001, pp. 4814
22. J udd, M.D. et al ., Intelligentconditionmonitoringandassetmanagement par-
tial dischargemonitoringfor power transformers, Power Engi neer i ng Jour nal ,
December 2002;16:297304
23. J inno, M. at el ., Partial dischargemeasurement of underground transmission
cablelines affected by earthquake, Proceedi ngs of I EE Japan on Power and
Energy, 1994, pp. 25560
24. Kakuta, M. et al ., High sensitivity partial discharge detection for ultra-high
voltage XLPE cable line, Proceedi ngs of I EE Japan on Power and Energy,
1994, paper 123, pp. 1316
25. Harrold, R.J ., Emery, F.T., Murphy, F.J ., and Drinkut, S.A., Radio fre-
quencysensingof incipient arcingfaultswithinlargeturbinegenerators, I EEE
Tr ansacti ons on PAS, 1979;PAS-98(3):116773
26. Emery, F.T., andLenderking, B.N., Turbinegenerator onlinediagnosticsusing
RF monitoring, I EEE Tr ans. Power App. Syst., 1981;PAS-100(12):4874982
27. Emery, F.T., and Harrold, R.J ., Radio frequency responseof alargeturbine-
generator stator winding, I EEE Tr ans. El ectron. Comput., 1986;EC-1(2):1729
28. Miller, R., and Hogg, W.K., Pulse propagation of slot and internal partial
discharges in stator windings of electrical machines, Proceedi ngs of the 4th
I nter nati onal Conference on Hi gh Vol tage Engi neer i ng, Athens, September1983
29. Kurtz, M., and Stone, C.C., Generator insulation diagnostic testing, I EEE
Tr ans. Power App. Syst., 1979;PAS-98:1596603
30. Kurtz, M., Lyles, J .F., andStone, C.C., ExperiencewiththeCEA partial dis-
chargetest, Proceedi ngsof the16th El ectr i cal /El ectroni c I nsul ati on Conference,
pp. 658
31. Sedding, H.G., Campbell, S.R., Stone, G.C., and Klempner, G.S., A new
sensor for detecting partial discharges in operating turbine generators, I EEE
Tr ansacti ons on Energy Conver si on, December 1991;6(4):7006
32. J ames, R.E., and Su, Q., Reviewof somerecent developments related to the
locationof partial dischargesingeneratorstatorwindings, presentedattheIEEE
Onl i ne i nsul ati on condi ti on moni tor i ng techni ques 239
PESwintermeeting, NewYork, J anuary1992, IEEE/PESPublication1-800-678-
IEEE, 92THO425-9PWR, pp. 712
33. J ames, R.E., Su, Q., Phung, B.T., Foong, S.C., andTychsen, R.C., Locationof
partial dischargesonan80MW/12.5kV hydro-generator withtheaidof digital
filteringtechniques, Proceedi ngs of El ectr i cal Engi neer s, Australia, November
1990;10(4):33843
9.8 Pr oblems
1. What are the advantages and disadvantages of online PD measurements on a
generator? What arethedifferences of theelectrical, mechanical and thermal
stressesexperiencedbyagenerator duringonlineandofflinePDmeasurements?
2. What arethemaintechniques for noisemitigationinPD measurements? Take
each fromhardware and software approaches and explain the principle and
implementationproceduresfor thesetechniques.
3. Thedielectric loss is usually temperature-dependent, which may affect online
DLA measurement and make it difficult to interpret the measurement results.
What methodscanbeusedtoalleviatetheproblem?
4. A PD exists at theconnection lead of theHV winding of a500kV 600MVA
transformer. Thetransformer isof paper/oil insulationanditsHV bushingshave
capacitivetaps. What techniquesandsensorsdoyouthink will bepromisingto
determinethetypeof fault (PD) and/or locatethedischarge?
Chapter 10
Artificial-intelligence techniques for incipient
fault diagnosis and condition assessment
Databaseandexpert systemsfor fault diagnosis
Transformer DGA fuzzy-logicdiagnosis
Fuzzy-logicconditionassessmentsandranking
Ingeneral, fault diagnosis may bedefinedas aproblemof patternrecognition.
Thepatternvector consistingof detectedparameterscanbecorrelatedwithitscoun-
terpart aset of classes or types of incipient fault andno-fault. Fromtheobserved
symptoms or/and test results (data) of equipment, an expert could makeageneral
reasoningtoderivethecauseorcausesusinghis/herknowledgeandexperience. How-
ever, it maybemoreobjectiveandaccuratewhenartificial intelligenceisemployed
to establish the links between the cause and results behind the observations. An
artificial-intelligence technique could consolidate the knowledge and experiences
froma number of experts. It could even learn and improve itself fromthe cases
it examines. Duringthelast decade, many applications of artificial intelligencefor
incipientfaultdiagnosisof HV equipmenthavebeendevelopedintheworld. Itisnot
withinthescopeof this book to includeall different methods andtechniques. This
chapterconcentratesonlyonsomeof thetechniquesandapplicationsdevelopedbythe
authors. Interestedreaderscanrefertorelevanttechnical papersandbooksfordetailed
information.
10.1 Database for condition assessment
Thenumber of HV devices inanelectrical utility runs into several thousands. The
devicesneedtobetestedperiodicallyforthemonitoringof theirinsulationconditions.
Manydiagnostictestscanbecarriedoutoneachdevice, whichproducesanenormous
numberof data. Withoutasuitablecomputerdatabase, manual storageandanalysisof
thedatawouldbeverydifficultaswell astime-consuming. Also, whenapieceof HV
242 Condition Assessment of High-Voltage Insulation
equipmentisinquestion, itisoftenrequiredtoprovidethemanagementboardwitha
soundandeconomicbasisforthemajordecisionsthatwill havetobemadeconcerning
possiblerefurbishment, lifeextensionorreplacement. Althoughconditionmonitoring
cangiveanearlywarningabouttheinsulationcondition, thereisnosingletechnique
thatalonecanguaranteetodetectall therangeof faultsandgiveareliableestimationof
remnant life. For thisreason, combinedmonitoringsystemsarerequiredtoprovide
a pool of data that will be analysed thoroughly so that an integrated assessment
of equipment condition may be achieved. The accurate assessment also depends
on the historical test record to establish a trend associated with normal operating
conditions.
10.1.1 A computer database and diagnostic program
Inmost cases, accuratediagnosis of apieceof HV equipment or acableneeds the
application of a number of techniques. The analysis of combined diagnostic test
resultsandtheir trendscanbemadeeasier withtheaidof acomputer database.
A database to meet all the above requirements should have the following
features:
canstoreandretrievealargenumber of data;
iseasytouseandfast indataretrieving;
canprocessdataandpresent resultsingraphical andtabular forms;
canbeincorporatedwithinexpert systemsandother artificial intelligence;
isreliableandsecure.
ThedatabasesoftwarepackagedevelopedbyMonashUniversityHVICMGroup
inthe1990sfulfilsmost of theaboverequirements. Thefront pageof theICMT-2.0
programis showninFigure10.1. Someadvantages of thedatabase fuzzy logic
expert systemarelistedbelow.
Ease of use and large storage: All optionsonthescreenareinpull-downform
withappropriateinstructions. Informationaboutall HV equipmentineachpower
stationof autility(ratedvoltage, current, power andyearsinservice, etc.) canbe
stored.
Graphical presentation: Trends of test results can be illustrated graphically,
whichgives abetter assessment of theinsulationcondition. Figure10.2shows
thetrendof dielectric loss angle(tan) of an HV motor presentedin graphical
forminICM-2.0.
Searching for abnormal test results: Thetest resultsover acertainlimit canbe
determinedanddisplayedonthescreenintabular forms. A number of queries
havebeendesigned, e.g. askingtoshowall transformerswithoneor moregases
exceedingtheguidedvalues.
Reporting: Thetest resultscanbeprintedout inareport.
Security: Thereisasecuritysystem. Thelow-level passwordisfor general users
whocanonly view thedata. Toaddtoor changethedata, ahigh-level password
must beused, whichmaybeissuedtoauthorizedpersonnel only.
Artificial-intelligence techniques 243
Figure 10.1 The database and expert system ICM-2.0 developed at Monash
University
0.045
Tangent delta
kV
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
0 1 2 3 4 5 6 7
4/3/90
2/3/92
3/4/94
2/2/95
Figure 10.2 The trend of dielectric loss angle (tan) of an HVmotor against voltage
presented in graphical form in ICM-2.0
10.1.2 A combined method for DGA diagnosis
Largepower transformersareprobablythemostimportantequipmentinanelectrical
transmissionsystem. Correct diagnosisof their incipient faultsisvital for safetyand
reliabilityof anelectrical network. Anin-servicetransformer issubject toelectrical
and thermal stresses, which can break down the insulating materials and release
244 Condition Assessment of High-Voltage Insulation
Expert diagnosis of equipment T845D
Percent failure 0 10 20 30 40 50 60 70 80 90 100
No fault
PD of low energy density
PD of high energy density
Discharge of low energy
Discharge of high energy
Thermal fault T<220 C
Thermal fault T>200 C
Thermal fault T>700 C
Figure 10.3 DGA interpretation using a combined diagnostic method
gaseousdecompositionproducts. Overheating, partial dischargeandarcingarethree
primary causes of fault-relatedgases. Therearemany interpretativemethods based
onDGA to diagnosethenatureof transformer deterioration, suchas theIEC ratio
codes, Rogers and triangle methods [13], which were developed fromintensive
investigationsongasesgeneratedfromindividual faults.
Ithasbeenfoundfor alongtimethattheinterpretationof DGA resultsusingdif-
ferent conventional methodscanbedifferent andconfusing. Eachof thesemethods
may haveitsownfeaturesinfault diagnosis. It isthereforeuseful tocombinethese
methodsandderivetheresultsstatistically. For example, theIEC code, Rogersand
trianglemethodsmaybear 40per cent, 30per cent and30per cent of weight respec-
tively. Theinterpretationwill bebasedonthenormalizedtotal valueas calculated
fromthesethreemethodsbasedonthelastthreetests. Thecombinedmethodcanthen
determinethepossibilityfor eachfaultinanisobar format. AsshowninFigure10.3,
thediagnostic results usingthis methodclearly showthat thetransformer testedis
most likelytohaveafault of low-energydischarge.
10.2 Fuzzy-logic fault diagnosis
AlthoughDGA hasbeenwidelyusedintheindustry, insomecasestheconventional
methodsfail toprovideadiagnosis. Thisnormallyhappenstothosetransformersthat
havemorethanonetypeof fault. Inpractice, theconventional diagnostic methods
arebasedontheratioof gasesgeneratedfromasinglefault or frommultiplefaults
but withoneof dominant natureinatransformer. Whengases frommorethanone
fault in a transformer are collected, the relation between different gases becomes
too complicated and may not match the predefined codes. For instance, the IEC
codes aredefinedfromcertaingas ratios, whichincreaseacross thedefinedlimits
(boundaries); thecodechanges suddenly between 0, 1and2(Table10.1). In fact,
thegasratioboundary may not beclear (i.e. fuzzy), especially whenmorethanone
typeof fault exists. Therefore, betweendifferent typesof fault, thecodeshouldnot
Artificial-intelligence techniques 245
Table 10.1 IEC ratio codes
IEC codes
Sharplydefinedrangesof thegasratio Codesof different gasratios
C
2
H
2
C
2
H
4
CH
4
H
2
C
2
H
4
C
2
H
6
<0.1 0 1 0
0.11 1 0 0
13 1 2 1
>3 2 2 2
changesharplyacrosstheboundaries. Newmethodsbasedonfuzzylogicandneural
networksmayimprovetheaccuracyof DGA diagnosis.
10.2.1 The conventional methods
In dissolved gas analysis, the IEC codes have been used for several decades and
considerable experience accumulated in the world to diagnose incipient faults in
transformers. Early interpretations were concentrated on specific gas components
such as hydrogen andmethanefor thedetermination of discharges in theoil. This
simplistic approachwasrefinedby anumber of researchers[13] whoinvestigated
theratios of certain gases to establish morecomprehensivediagnostic techniques.
ThesetechniqueswerestandardizedbyIECin1978inGuide for Interpretation of the
Analysis of Gases inTransformer andOther Oil FilledElectrical Equipment inService
anditsrevisionIEC60599-1999[8]. Theindividual gasesusedtodetermineeachratio
anditsassignedlimitsareshowninTable10.1. Codesarethenallocatedaccordingto
thevalueobtainedfor eachratioandthecorrespondingfaultcharacterized, asshown
inTable10.2.
AlthoughtheIEC codemethodisuseful for theassessment of transformer insu-
lation, it cannot determinethepossibility of afault. Also, insomecases, theDGA
results cannot bematched by theexisting codes, making thediagnosis unsuccess-
ful. Inmultiple-fault conditions, gases fromdifferent faults aremixed, resultingin
confusingratios betweengas components. This canbedealt withonly withtheaid
of moresophisticated analysis methods such as thefuzzy-logic diagnostic method
presentedinthissection.
10.2.2 A fuzzy-logic method
Duringthelasttwodecades, several methodshavebeendevelopedintheworldtouse
artificial intelligencefor moreaccuratediagnosis [47]. Thesemethods aremostly
suitablefor transformers withasinglefault or adominant fault. Inprinciple, these
methodsarestill basedonthespecificsetof codesdefinedfromcertaingasratios, as
246 Condition Assessment of High-Voltage Insulation
Table 10.2 Fault classification according to the IEC Gas Ratio
Codes
No. Fault type
C
2
H
2
C
2
H
4
CH
4
H
2
C
2
H
4
C
2
H
6
0 Nofault 0 0 0
1 Partial dischargesof 0 1 0
lowenergydensity (but not
significant)
2 Partial dischargesof 1 1 0
highenergydensity
3 Dischargesof lowenergy 1or 2 0 1or 2
4 Dischargesof highenergy 1 0 2
5 Thermal fault of low 0 0 1
temperature, <150

C
6 Thermal fault of low 0 2 0
temperature, 150300

C
7 Thermal fault of medium 0 2 1
temperature, 300700

C
8 Thermal fault of high 0 2 2
temperature, >700

C
showninTable10.1. Inaddition, therearealsosomeother limitationsintheprevious
fuzzydiagnosismethods.
In general, in order to diagnosemoreaccurately theincipient faults in atrans-
former, thekeygasesshouldbeanalysedandthetrendof individual faultsdetermined.
These could be achieved by the fuzzy-logic method presented in this section. It
employs fuzzy boundaries between different IEC codes (i.e. the fuzzy IEC code)
withdemi-Cauchydistributionfunction. Thecritical level of keygasesisalsotreated
withthesamefuzzy methodandcombinedwiththefuzzy IEC codes. Eachfault in
atransformer canthenbeassessedby thefuzzy vector andthetrendof fault devel-
opment withtimecanbecloselymonitored. For multiplefaultsinatransformer, this
techniquecanbeusedtomonitor thetrendof eachfault.
10.2.2.1 Fuzzy IEC codes
AccordingtotheIEC codesinTable10.1, thethreegasratiosr
1
=
C
2
H
2
C
2
H
4
, r
2
=
CH
4
H
2
andr
3
=
C
2
H
4
C
2
H
6
canbecodedas0, 1and2fordifferentrangesof ratios(note, r
1
, r
2
and
r
3
arelarger thanor equal to0). Table10.1isrearrangedtogiveaclear relationship
between therangeof each gas ratio and thecorresponding IEC code, as shown in
Table10.3. Forexample, thecodesforasetof gasconcentration
C
2
H
2
C
2
H
4
< 0.1,
CH
4
H
2
> 1
and1
C
2
H
4
C
2
H
6
3areCode(r
1
) = 0, Code(r
2
) = 2andCode(r
3
) = 1. According
toTable10.2, thetransformer isdiagnosedtohaveaNo. 7fault, i.e. thermal fault of
mediumtemperature300700

C.
Artificial-intelligence techniques 247
Table 10.3 Gas ratios and corresponding IEC
codes rearranged from Table 10.1
RatioCode(r) 0 1 2
r
1
=
C
2
H
2
C
2
H
4
<0.1 0.13 >3
r
2
=
C
2
H
4
H
2
0.11 <0.1 >1
r
3
=
C
2
H
4
C
2
H
6
<1 13 >3
IntheIECcodediagnosis, actuallytheconventional logicANDandORareused.
For example, theseventhfault isrepresentedby
f (7) = code
zero
(r
1
) AND code
two
(r
2
) AND code
one
(r
3
),
wherecode
zero
(r
1
), code
two
(r
2
) andcode
one
(r
3
) arethelogicsof codedvaluesof gas
ratior
1
, r
2
andr
3
respectively. Theyareeither one(true) or zero(false) accordingto
Table3.
For example, code
zero
(r
1
) =

1 for r
1
< 0.1
0 for r
1
0.1
. Therefore, fault f(7)
will be either one (true) or zero (false) by means of the logic operation
code
zero
(r
1
) AND code
two
(r
2
) AND code
one
(r
3
).
Inthefuzzy-diagnosismethoddeveloped, however, theIEC codes0, 1and2are
reconstructedasfuzzysetsZERO,ONEandTWO.Eachgasratior canberepresented
asafuzzyvector
[
ZERO
(r),
ONE
(r),
TWO
(r)],
where
ZERO
(r),
ONE
(r),
TWO
(r) are the membership functions of fuzzy code
ZERO, ONE andTWO respectively. Themembershipfunctionis representedby a
descendingor/andanascendingdemi-Cauchydistributionfunction[11]:

d
(r) =

1 if r A
1
1+(
Ar
a
)
2
if r > A
(10.1)

a
(r) =

1 if r A
1
1+(
Ar
a
)
2
if r < A
(10.2)
where a and A are a pair of parameters that can be selected to give appropriate
membershipfunctions[4]. Indiagnostictests, Aandacanberegardedastheboundary
parameter anddistributionparameter respectively. From(10.1) and(10.2), thefuzzy
IECcodesZERO, ONEandTWOareformed. Foreachsetof ratios, thecorresponding
fuzzyIEC codescanthenbedetermined.
The fuzzy-diagnosis vector F(i) (i = 08) is then determined by replacing
the logic AND with the minimization operation and the logic OR with the
248 Condition Assessment of High-Voltage Insulation
1
1
0.5
(a) (b)
0.8
0.6
0.4
0.2
0
0 1 2 3

ONE
(r
3
)
4 5 6
0 1 2 3 4
IEC code 1
5 6
r r
Figure 10.4 Comparison between IEC code 1 and fuzzy IEC code ONE for the gas
ratio r
3
=
C
2
H
4
C
2
H
6
maximizationoperation. Inother words, thefuzzymultivaluelogicisusedtosubsti-
tutefor theconventional true-falselogic[9]. BasedontheIEC rules, different faults
(fromNo. 0to8inTable10.2) canbediagnosedbythefollowingequations:
F(0) = min[
ZERO
(r
1
),
ZERO
(r
2
),
ZERO
(r
3
)]
F(1) = min[
ZERO
(r
1
),
ONE
(r
2
),
ZERO
(r
3
)]
F(2) = min[
ONE
(r
1
),
ONE
(r
2
),
ZERO
(r
3
)]
F(4) = min[
ONE
(r
1
),
ZERO
(r
2
),
TWO
(r
3
)]
F(5) = min[
ZERO
(r
1
),
ZERO
(r
2
),
ONE
(r
3
)]
F(6) = min[
ZERO
(r
1
),
TWO
(r
2
),
ZERO
(r
3
)]
F(7) = min[
ZERO
(r
1
),
TWO
(r
2
),
ONE
(r
3
)]
F(8) = min[
ZERO
(r
1
),
TWO
(r
2
),
TWO
(r
3
)]
F(3) = max{min[
ONE
(r
1
),
ZERO
(r
2
),
ONE
(r
3
)],
min[
ONE
(r
1
),
ZERO
(r
2
),
TWO
(r
3
)],
min[
TWO
(r
1
),
ZERO
(r
2
),
ONE
(r
3
)],
min[
TWO
(r
1
),
ZERO
(r
2
),
TWO
(r
3
)]}
Theequationsarethennormalizedas:
F
r
(i) =
F(i)
8

j=0
F( j )
, i = 08 (10.3)
A comparisonbetweentheconventional IEC code1code
one
(r
3
) andthefuzzy
membershipfunction
ONE
(r
3
) for gasratior
3
=
C
2
H
4
C
2
H
6
isshowninFigure10.4.
Artificial-intelligence techniques 249
10.2.2.2 Fuzzy key gases
Three key gases are commonly used to identify particular faults according to the
followingrules:
C
2
H
2
Arcing(dischargeof highenergy)
H
2
Corona(partial discharge)
C
2
H
2
Hightemperatureoil breakdown
(thermal fault of hightemperature> 700

C)
Eachfault isindicatedby theexcessivegenerationof therelevant gasaboveits
designated threshold. For different voltagelevels, winding structures and types of
transformer, thethresholdmay bedifferent. Thethresholdcanbedeterminedfrom
previousexperiencefor acertaintypeof transformer. Fromfuzzylogictheories[9],
themembershipfunctionsof thefuzzyset LOW, MEDIUM andHIGH for eachkey-
gascanthenberepresentedasdescendingdemi-Cauchydistributionfunction
L
(x)
or ascendingdemi-Cauchydistributionfunction
H
(x) or their combination
M
(x).
For everygivenkeygasconcentrationx, thecorrespondingfuzzyvector istherefore
[
LOW
(x),
MED
(x),
HIGH
(x)].
Thefuzzy-diagnosis vector F
k
(i)(i = 08) is thendeterminedinthesameway
asthefuzzyIEC code.
10.2.2.3 The fuzzy IEC code key gas method
ThefuzzyIEC code keygasmethod(FIK) developedisacombinationof fuzzy
diagnosesusingtheIECcodesandkeygases. Thecombinedfuzzy-diagnosticvector
isrepresentedby
F(i) = w
1
F
r
(i) + w
2
F
k
(i) i = 08 (10.4)
wherew
1
andw
2
aretheweightstorelatethefaultwiththefuzzyIECcodesandkey
gasesrespectively. w
1
andw
2
canbe0.5/0.5ordeterminedfrompreviousexperience.
F(i) isnormalizedtomakethetotal fuzzycomponent equal to1.
10.2.2.4 Applications
UsingtheFIK method, anumber of 110330kV power transformerswerediagnosed
andsometypical results aregiveninTable10.4. It canbeseenfromsampleNo. 1
that thenewmethodisgenerallyinagreement withtheIECmethodfor transformers
of asingleor adominant fault. Compared with theIEC method, theFIK method
alsohassomeadvantages. For example, duetonomatchingcodes, 13transformers
couldnot bediagnosedbytheIEC methodbut arediagnosedbytheFIK method, as
showninTable10.4, Nos. 23, forsometypical results. Insomecases, thefaultsmay
beonly at theearly stageor intermittent, whichdidnot producesufficient gases to
giveastronger indication, suchasF(2) inNo. 2andF(68) inNo. 3. However, the
informationobtainedshouldbeuseful for futuretrendanalysis. Transformer No. 4
was diagnosed by theIEC method to haveathermal fault of mediumtemperature
250 Condition Assessment of High-Voltage Insulation
T
a
b
l
e
1
0
.
4
C
o
m
p
a
r
i
s
o
n
o
f
d
i
a
g
n
o
s
i
s
r
e
s
u
l
t
s
f
r
o
m
t
h
e
F
I
K
m
e
t
h
o
d
a
n
d
t
h
e
c
o
n
v
e
n
t
i
o
n
a
l
I
E
C
m
e
t
h
o
d
N
o
.
C
o
n
v
e
n
t
i
o
n
a
l
I
E
C
m
e
t
h
o
d
F
u
z
z
y
I
E
C
c
o
d
e

k
e
y
g
a
s
m
e
t
h
o
d
(
F
I
K
)
C
o
m
m
e
n
t
s
I
E
C
c
o
d
e
E
x
p
l
a
n
a
t
o
r
y
n
o
t
e
F
u
z
z
y
f
a
u
l
t
c
o
m
p
o
n
e
n
t
s
E
x
p
l
a
n
a
t
o
r
y
n
o
t
e
1
0
2
0
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
0
)
=
0
.
4
7
9
N
o
r
m
a
l
a
g
e
i
n
g
B
o
t
h
F
I
K
a
n
d
I
E
C
d
i
a
g
n
o
s
e
a
t
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
,
i
n
d
i
c
a
t
i
n
g
a
g
e
n
e
r
a
l
a
g
r
e
e
m
e
n
t
b
e
t
w
e
e
n
t
h
e
s
e
t
w
o
m
e
t
h
o
d
s
.
F
(
1
)
=
0
.
0
0
6
P
D
o
f
l
o
w
e
n
e
r
g
y
F
(
2
)
=
0
.
0
0
6
P
D
o
f
h
i
g
h
e
n
e
r
g
y
F
(
3
)
=
0
.
0
0
0
D
i
s
c
h
a
r
g
e
o
f
l
o
w
e
n
e
r
g
y
F
(
4
)
=
0
.
0
0
5
D
i
s
c
h
a
r
g
e
o
f
h
i
g
h
e
n
e
r
g
y
F
(
5
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
<
1
5
0

C
)
F
(
6
)
=
0
.
4
9
6
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
7
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
8
)
=
0
.
0
0
9
T
h
e
r
m
a
l
f
a
u
l
t
(
>
7
0
0

C
)
2
1
0
0
N
o
m
a
t
c
h
F
(
0
)
=
0
.
5
2
5
N
o
r
m
a
l
a
g
e
i
n
g
I
E
C
c
a
n
n
o
t
d
i
a
g
n
o
s
e
b
u
t
F
I
K
i
n
d
i
c
a
t
e
s
a
h
i
g
h
-
e
n
e
r
g
y
P
D
f
a
u
l
t
w
h
i
c
h
c
o
u
l
d
b
e
a
t
a
n
e
a
r
l
y
s
t
a
g
e
.
F
(
1
)
=
0
.
0
5
3
P
D
o
f
l
o
w
e
n
e
r
g
y
F
(
2
)
=
0
.
2
3
1
P
D
o
f
h
i
g
h
e
n
e
r
g
y
F
(
3
)
=
0
.
0
4
5
D
i
s
c
h
a
r
g
e
o
f
l
o
w
e
n
e
r
g
y
F
(
4
)
=
0
.
0
5
0
D
i
s
c
h
a
r
g
e
o
f
h
i
g
h
e
n
e
r
g
y
F
(
5
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
<
1
5
0

C
)
F
(
6
)
=
0
.
0
4
7
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
7
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
8
)
=
0
.
0
5
0
T
h
e
r
m
a
l
f
a
u
l
t
(
>
7
0
0

C
)
Artificial-intelligence techniques 251
3
1
2
1
N
o
m
a
t
c
h
F
(
0
)
=
0
.
0
0
5
N
o
r
m
a
l
a
g
e
i
n
g
I
E
C
c
a
n
n
o
t
d
i
a
g
n
o
s
e
p
r
o
b
a
b
l
y
d
u
e
t
o
t
h
e
e
x
i
s
t
e
n
c
e
o
f
m
o
r
e
t
h
a
n
o
n
e
f
a
u
l
t
.
T
h
e
f
u
z
z
y
c
o
m
p
o
n
e
n
t
o
f
t
h
e
e
a
r
l
y
t
h
e
r
m
a
l
f
a
u
l
t
i
n
d
i
c
a
t
e
d
b
y
F
I
K
i
s
u
s
e
f
u
l
f
o
r
f
u
t
u
r
e
t
r
e
n
d
a
n
a
l
y
s
i
s
.
A
c
t
u
a
l
f
a
u
l
t
w
a
s
a
n
a
r
c
d
a
m
a
g
e
t
o
t
h
e
c
o
r
e
.
F
(
1
)
=
0
.
0
5
2
P
D
o
f
l
o
w
e
n
e
r
g
y
F
(
2
)
=
0
.
0
5
2
P
D
o
f
h
i
g
h
e
n
e
r
g
y
F
(
3
)
=
0
.
0
0
0
D
i
s
c
h
a
r
g
e
o
f
l
o
w
e
n
e
r
g
y
F
(
4
)
=
0
.
4
0
8
D
i
s
c
h
a
r
g
e
o
f
h
i
g
h
e
n
e
r
g
y
F
(
5
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
<
1
5
0

C
)
F
(
6
)
=
0
.
1
6
1
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
7
)
=
0
.
1
6
1
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
8
)
=
0
.
1
6
1
T
h
e
r
m
a
l
f
a
u
l
t
(
>
7
0
0

C
)
4
0
2
1
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
0
)
=
0
.
0
0
7
N
o
r
m
a
l
I
E
C
d
i
a
g
n
o
s
e
s
m
e
d
i
u
m
t
e
m
p
e
r
a
t
u
r
e
f
a
u
l
t
,
b
u
t
a
c
t
u
a
l
l
y
b
o
t
h
m
e
d
i
u
m
-
a
n
d
h
i
g
h
-
t
e
m
p
e
r
a
t
u
r
e
f
a
u
l
t
s
e
x
i
s
t
e
d
a
s
i
n
d
i
c
a
t
e
d
b
y
F
I
K
.
T
w
o
l
o
c
a
t
i
o
n
s
o
f
o
v
e
r
h
e
a
t
i
n
g
d
a
m
a
g
e
s
w
e
r
e
f
o
u
n
d
d
u
e
t
o
e
d
d
y
c
u
r
r
e
n
t
s
a
n
d
a
b
a
d
c
o
n
t
a
c
t
.
F
(
1
)
=
0
.
0
2
6
P
D
o
f
l
o
w
e
n
e
r
g
y
F
(
2
)
=
0
.
0
2
6
P
D
o
f
h
i
g
h
e
n
e
r
g
y
F
(
3
)
=
0
.
0
0
0
D
i
s
c
h
a
r
g
e
o
f
l
o
w
e
n
e
r
g
y
F
(
4
)
=
0
.
0
3
0
D
i
s
c
h
a
r
g
e
o
f
h
i
g
h
e
n
e
r
g
y
F
(
5
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
<
1
5
0

C
)
F
(
6
)
=
0
.
0
0
3
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
7
)
=
0
.
4
7
7
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
8
)
=
0
.
4
3
1
T
h
e
r
m
a
l
f
a
u
l
t
(
>
7
0
0

C
)
5
B
e
l
o
w
n
o
m
i
n
a
l
v
a
l
u
e
s
N
o
d
i
a
g
n
o
s
i
s
F
(
0
)
=
0
.
4
7
9
N
o
r
m
a
l
A
l
t
h
o
u
g
h
t
h
e
g
a
s
l
e
v
e
l
i
s
b
e
l
o
w
t
h
e
g
u
i
d
e
v
a
l
u
e
,
a
n
e
a
r
l
y
i
n
d
i
c
a
t
i
o
n
o
f
l
o
w
-
e
n
e
r
g
y
d
i
s
c
h
a
r
g
e
b
y
F
I
K
s
h
o
u
l
d
b
e
u
s
e
f
u
l
f
o
r
t
r
e
n
d
a
n
a
l
y
s
i
s
i
n
t
h
e
f
u
t
u
r
e
.
F
(
1
)
=
0
.
0
0
5
P
D
o
f
l
o
w
e
n
e
r
g
y
F
(
2
)
=
0
.
0
5
7
P
D
o
f
h
i
g
h
e
n
e
r
g
y
F
(
3
)
=
0
.
4
4
1
D
i
s
c
h
a
r
g
e
o
f
l
o
w
e
n
e
r
g
y
F
(
4
)
=
0
.
0
1
3
D
i
s
c
h
a
r
g
e
o
f
h
i
g
h
e
n
e
r
g
y
F
(
5
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
<
1
5
0

C
)
F
(
6
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
1
5
0

3
0
0

C
)
F
(
7
)
=
0
.
0
0
0
T
h
e
r
m
a
l
f
a
u
l
t
(
3
0
0

7
0
0

C
)
F
(
8
)
=
0
.
0
0
5
T
h
e
r
m
a
l
f
a
u
l
t
(
>
7
0
0

C
)
252 Condition Assessment of High-Voltage Insulation
Table 10.5 Gas concentration for the transformers listed in Table 10.4
(in ppm)
No. H
2
CH
4
C
2
H
2
C
2
H
4
C
2
H
6
IEC ratio codes
1 95 110 <0.1 50 160 020
2 120 17 4 23 32 100
3 300 490 95 360 180 121
4 200 700 1 740 250 021
5 25 3 0.1 0.1 0.1 belowguidelevel
(300700

C). In comparison, theFIK method indicates that both high (>700

C)
andmediumtemperature(300700

C) faultsexisted. Thelikelihoodof eachfault is

givenbythefuzzycomponentsof 0.477and0.431respectively. Theanalysisresults
of another transformer (No. 5) showthat, althoughthegaslevel isbelowtheguide
value, theFIK methodcanstill beusedandalow-energydischargeisdiagnosed. The
fuzzyvector rangesfrom0to0.441, whichcouldbeuseful for futuretrendanalysis
whenthegaslevel increases.
ThegasconcentrationsforthesamplesinTable10.4aregiveninTable10.5. Most
transformershaveshownamediumor lowlevel of gasesandlower increasingrates,
suchastransformersNo. 1, No. 2andNo. 5. Thereforethey areclosely monitored.
Transformer No. 3wasdismantledandanarcdamageintheinsulationwasfoundin
thecore. Intheinner inspectionof another transformer, No. 4, two locations were
identifiedwithhigh-temperaturedamageduetoeddycurrentsandabadcontact. More
transformers canbeinvestigatedwhencertaincriteriafor individual gases aremet.
Laboratorytestscanalsobecarriedout tofine-tunethefuzzy-diagnosistechnique.
Thedetailsof thetransformersandtheDGA resultsarestoredinthedatabaseand
fuzzy-logicsoftwarepackage, asshowninFigure10.5. Afteratransformerisselected,
bypressingabuttonFuzzyLogicDiagnosis, thelikelihoodof fuzzyelementswill
bedisplayedonthescreengivinganadviceonthepredictedincipient fault. Using
thiscomputer analysistool, anumber of 110330kV power transformersinVictoria,
Australia, werediagnosedandsometypical resultsaregiveninFigure10.6. For the
newtransformer, thefuzzy-logic method indicated an 87 per cent diagnosis index
for nofault, asshownby Figure10.6(a), whereastwofaultswerediagnosedfor the
secondtransformer (Figure10.6(b)). Thetransformer was eventually untankedand
twofaultswerefoundinthewinding.
ThecombinedDGA diagnosismethod(Section10.1.2) isfounduseful because
itusesthreedifferentmethodsandthelastthreetestresults, whichinvolvesthetrend
of faults. Whenthereismorethanonefault inatransformer, thefuzzy-logicmethod
may bemoreeffective. Themethods havebeensuccessfully usedfor thediagnosis
of anumber of transformers inAustralia. It has beenprovedthat, usingthefuzzy-
diagnosismethod, moredetailedinformationaboutthefaultsinsideatransformercan
beobtained. Thisisanimprovement over theconventional IECcodemethod, which
Artificial-intelligence techniques 253
Figure 10.5 The computer program ICM-2.0 works as a comprehensive database
and a fuzzy-logic diagnosis tool for transformers
maybeduetothemorerealisticrepresentationof therelationshipbetweenfaultsand
dissolved-gasratioswiththefuzzy membershipfunctions. Also, usingthismethod,
multiplefaultscanbediagnosedandtheir trendsdetermined, whichisanimportant
feature.
10.2.2.5 Trend analysis for individual faults
InFIK diagnosis, afaultcanbemoreaccuratelydeterminedbyitsfuzzycomponent,
whichindicates thelikelihoodor dominanceof thefault. Deteriorationof thefault
maythereforebecloselymonitoredthroughtrendanalysis. Thistechniquehasbeen
used for a transformer that was tested over a 15-month period. Thermal faults of
mediumandhightemperature(300700

Cand>700

C) werediagnosedbytheFIK
methodandtheir fuzzy componentsagainst thetest timeareplottedinFigure10.7.
Thegraphclearlyshowsthedevelopmentof eachthermal faultinthistransformer. It
canbeseenthat at thebeginningof thismonitoringperiod, themediumtemperature
thermal faultF(7) wasthemainproblemof thistransformerandthefuzzycomponent
of high-temperaturethermal fault (>700

C) was very small, i.e. below0.05. The

high-temperaturethermal fault F(8) wasdiagnosedfromDay114onwardsandthen
becamestableuntil Day 406, whentheoil was degassed. After degassing, because
thethermal faultsremained, thefuzzycomponentsF(7) andF(8) wentupagainfrom
Day453. It took afewweeksfor thegasestobereleasedanddissolvedintheoil to
254 Condition Assessment of High-Voltage Insulation
Fuzzy diagnosis of equipment Healthy Tx
Percent failure
(a)
(b)
0 10 20 30 40 50 60 70 80 90 100
No fault
PD of low energy density
PD of high energy density
Discharge of low energy
Discharge of high energy
Thermal fault T<150 C
Thermal fault T>150C
Thermal fault T>300 C
Thermal fault T>700C
Fuzzy diagnosis of equipment T845D
Percent failure
No fault
PD of low energy density
PD of high energy density
Discharge of low energy
Discharge of high energy
Thermal fault T>150C
Thermal fault T<150C
Thermal fault T>300C
Thermal fault T>700C
0 10 20 30 40 50 60 70 80 90 100
Figure 10.6 Fuzzy-logic diagnosis for transformers using ICM-2.0: (a) a new
healthytransformer; (b) atransformer withtwotypes of incipient fault
0.6
0.5
0.4
0.3
0.2
0.1
De-gassing on the 410th day
F
u
z
z
y

c
o
m
p
o
n
e
n
t
s

F
(
r
)
1 114 147 178 191 218
Day
F(7)-Thermal fault 300700C
F(8)-Thermal fault >700C
406 413 453 469 471
0
Figure 10.7 The trend of two types of thermal fault in a 330 kV transformer
determined by the FIK method
Artificial-intelligence techniques 255
asufficient level for accuratediagnosis. A small fluctuationof F(8) wasrecordedon
Day178, whichmight beduetothelighter loadduringthespecifictimeperiod.
It must benoted that, if atransformer has no fault, thefuzzy component F(0)
always gives alargevalueintherangeof 0.61. For example, theDGA results for
ahealthy transformer are(inppm) H
2
95, N
2
73000, O
2
11000, CO 1000,
CH
4
20, CO
2
8400, C
2
H
4
25, C
2
H
6
45andC
2
H
2
2. Thefuzzy component
of no-fault F(0) = 0.863, whichindicatesthat nofault existsinthetransformer. The
IECcodesare0, 0, 0, alsoindicatingnofault. Fromourexperience, whenthevalueof
F(0) isbetween0.3and0.6, anincipient fault at itsearlier stagemayhaveoccurred.
Whenthefault isgettingworse, F(0) will decreaseto<0.1.
10.3 Asset analysis and condition ranking
Under the current economical climate, no utility could afford to replace all aged
equipment accordingto its designedlifetime. Priority has to begivento thosethat
havealready reachedor areclosetotheir critical point of failure. A thoroughstudy
of theasset condition andcorrect rankingof theequipment wouldbeessential for
thedecisionmaking. This will ensurethat theworst equipment canbereplacedor
refurbishedintime.
10.3.1 Equipment ranking according to the insulation condition
Conditionsof apieceof equipmentcanbeassessedintermsof itsfailureprobability.
Preferably, thewholehistoryof theequipmentsincecommissioningcanberetrieved
andanalysedthoroughly. Thefollowingaresomeimportant factorstobeconsidered
intheequipment ranking:
failurerateat different agesfor aparticular typeof equipment;
trendanddeteriorationrateindicatedbydiagnostictest results;
severityof failureconsequence;
valueandmaintenancecostsof equipment;
workingenvironment, loadingandimpactof electrical, mechanical andchemical
overstresses.
Fromacombinedevaluationof variousfactorswithcertainweightsattached, the
prioritylist for atypeof equipment canbedetermined.
10.3.2 Insulation health index
Theinsulationhealthor failureindex may bedeterminedfromvarious test results,
quality of equipment, operating condition (e.g. temperature), load, overvoltage
impacts and so on. Because there are too many factors that affect the probability
of insulationfailure, it maynot bepossibletoincludeeveryfactor intheevaluation.
Starting fromsomeimportant factors, an evaluation systemusing fuzzy logic has
beendeveloped[1215].
256 Condition Assessment of High-Voltage Insulation
Takingatypeof HV motor as anexample, threeparameters fromconventional
insulationtestswereused: partial discharge(P), dielectricdissipationfactor tan (T)
anditsknee-pointvoltage(K). Theabsolutevalueandincrease/decreaserateof each
parameter wouldformanarrayof sixcomponents(P
a
, P
r
, T
a
, T
r
, K
a
andK
r
) andcan
beanalysedwithafuzzy-logic tool. Accordingtothescalar, thefuzzy setsof every
parameter aredefinedaslow(L), medium(M) andhigh(H):
P
a
partial dischargemagnitude, itsfuzzysetislow(L) for P
a
< 5, 000pC, medium
(M) for 5, 000< P
a
< 50, 000pC andhigh(H) for P
a
> 50, 000pC.
P
r
thetrendof partial dischargerepresentedby theratio betweenPDs measured
insubsequent 6months P
r
=
Q
n+1
Q
n
. It is dividedinto threegroups: low(L) for
P
r
< 1.2, medium(M) 1.2< P
r
< 2.0andhigh(H) P
r
> 2.0.
T
a
thelevel of dielectricdissipationfactor tan. Thethreegroupsarelow(L) for
T
a
< 0.01, medium(M) for 0.01< T
a
< 0.05andhigh(H) T
a
> 0.05.
T
r
thetrendof dielectric dissipationfactor tan representedby theratiobetween
PDs measuredinsubsequent 6months T
r
=
tan
n+1
tan
n
. Thethreegroups arelow
(L) for T
r
< 1.2, medium(M) for 1.2< T
r
< 2.0andhigh(H) for T
r
> 2.0.
K
a
thenormalized knee-point voltage, i.e. V
t
= kV
p
whereV
t
is theknee-point
voltage and V
p
the rated voltage. K
a
is divided into three groups: low (L) for
K
a
< 0.2, medium(M) for 0.2< K
a
< 0.5andhigh(H) for K
a
> 0.5.
K
r
thetrendof knee-pointvoltage(measuredonceayear), K
r
= 1
V
k
(n+1)
V
k
(n)
. K
r
is
dividedintothreegroups: low(L) for K
r
< 0.1, medium(M) for 0.1< K
r
< 0.75
andhigh(H) K
r
> 0.75.
Itshouldbenotedthatthehigh, mediumandlowlevelsof theabovesixvariables
areselectedaccordingtoprevioustestresultsandconditionassessmentexperiencefor
theparticular motor. Theycanbeverydifferentdependingonthetypeof equipment.
10.3.3 Membership functions of fuzzy set
Infuzzydiagnosis, thefuzzysetsof thesixparametersneedtobeestablished. Accord-
ingtothefuzzy-settheoryandthediagnosticexperience, themembershipfunctionsof
fuzzysetsLOW, MEDIUM, HIGH arerepresentedasdescendingdemi-Cauchydis-
tributionfunction
L
(x) or ascendingdemi-Cauchy distributionfunction[4]
H
(x)
or their combination. Thetwofunctions
L
(x) and
H
(x) arerepresentedby

L
(x) =

1 for x A
1
1+(
xA
a
)
4
for x > A
(10.5)

H
(x) =

1 for x A
1
1+(
Ax
a
)
4
for x < A
(10.6)
wherexisthevariableandthevaluesof aandA aregiveninTable10.6, whichare
determinedfrompreviousexperience[4].
Artificial-intelligence techniques 257
Table 10.6 Coefficients a and A of Equations (10.5) and (10.6) for various
fuzzy sets
Fuzzy set P
a
P
r
T
a
T
r
K
a
K
r

LOW
(x) A 5000.0 1.2 0.01 1.2 0.2 0.1
a 1200.0 0.3 0.0025 0.3 0.05 0.025

HIGH
(x) A 50000.0 2.0 0.05 2.0 0.5 1.0
a 8300.0 0.34 0.0083 0.34 0.083 0.17
We take the membership functions of fuzzy sets LOW, MEDIUM, HIGH, as
follows:

LOW
(x) =
L
(x)

HIGH
(x) =
H
(x)

MED
(x) = min[
H1
(x),
L1
(x)]
where

L1
(x) = 1
LOW
(x)

H1
(x) = 1
HIGH
(x)
Thus, every membership function can be constructed fromdescending demi-
Cauchy function
L
(x) and/or ascending demi-Cauchy function
H
(x) given by
(10.3) and(10.4).
For x = P
a
, themembershipfunctionsof fuzzysetsLOW, MEDIUM, HIGHare
showninFigure10.8.
For x = P
r
, themembershipfunctionsof fuzzysetsLOW, MEDIUM, HIGHare
showninFigure10.9.
Fromfuzzy-logic analysis, thetotal failureindex basedonthesetest resultscan
bederived.
F = F1(P
a
) +F2(P
r
) +F3(T
a
) +F4(T
r
) +F5(K
a
) +F6(K
r
)
where
F1(P
a
) = w
1
w
11
[
LOW
(P
a
) +
MED
(P
a
) +
HIGH
(P
a
)]
F2(P
r
) = w
1
w
22
[
LOW
(P
r
) +
MED
(P
r
) +
HIGH
(P
r
)]
F3(T
a
) = w
2
w
11
[
LOW
(T
a
) +
MED
(T
a
) +
HIGH
(T
a
)]
F4(T
r
) = w
2
w
22
[
LOW
(T
r
) +
MED
(T
r
) +
HIGH
(T
r
)]
F5(K
a
) = w
3
w
11
[
LOW
(K
a
) +
MED
(K
a
) +
HIGH
(K
a
)]
F6(K
r
) = w
3
w
22
[
LOW
(K
r
) +
MED
(K
r
) +
HIGH
(K
r
)]
andw
1
, w
2
, w
3
, w
11
andw
22
areweightsfor theparameters.
258 Condition Assessment of High-Voltage Insulation
1
0.8
0.6
0.4
0.2
x (pC)
(a)
(b)
(c)

LOW
(x)
0
0
5
0
0
0
1
0
0
0
0
1
5
0
0
0
2
0
0
0
0
2
5
0
0
0
3
0
0
0
0
3
5
0
0
0
4
0
0
0
0
4
5
0
0
0
5
0
0
0
0
5
5
0
0
0
6
0
0
0
0
1
0.8
0.6
0.4
0.2
x (pC)

MED
(x)
0
0
5
0
0
0
1
0
0
0
0
1
5
0
0
0
2
0
0
0
0
2
5
0
0
0
3
0
0
0
0
3
5
0
0
0
4
0
0
0
0
4
5
0
0
0
5
0
0
0
0
5
5
0
0
0
6
0
0
0
0
1
0.8
0.6
0.4
0.2
x (pC)

HIGH
(x)
0
0
5
0
0
0
1
0
0
0
0
1
5
0
0
0
2
0
0
0
0
2
5
0
0
0
3
0
0
0
0
3
5
0
0
0
4
0
0
0
0
4
5
0
0
0
5
0
0
0
0
5
5
0
0
0
6
0
0
0
0
Figure 10.8 The membership functions of fuzzy sets of P
a
(LOW, MEDIUM, HIGH)
10.3.4 Example of fuzzy logic condition ranking
Inorder tocheckthefailureindexanalysisusingtheproposedfuzzy-logictechnique,
14hypothetical HV motorsandtheir test resultsareused, asshowninTable10.7.
(1) P
a
, P
r
AND THEIR FUZZY SETS
Thelevel of partial discharge(P
a
), itstrend(P
r
) andtheirfuzzysetsLOW, MEDIUM,
HIGHarecalculatedasshowninTable10.8.
(2) T
a
, T
r
AND THEIR FUZZY SETS
Thelevel of dielectricdissipationfactor tan (T
a
), itstrend(T
r
) andtheir fuzzysets
LOW, MEDIUM, HIGHarecalculatedasshowninTable10.9.
Artificial-intelligence techniques 259
1
0.8
0.6
0.4
0.2
x
(a)

L
O
W

(
x
)
0
0
0
.
2
0
.
4
0
.
6
0
.
81
1
.
2
1
.
4
1
.
6
1
.
82
2
.
2
2
.
4
2
.
6
2
.
83
0.4
0.5
0.6
0.7
0.3
0.2
0.1
x
(b)

M
E
D

(
x
)
0
0
0
.
2
0
.
4
0
.
6
0
.
81
1
.
2
1
.
4
1
.
6
1
.
82
2
.
2
2
.
4
2
.
6
2
.
83
0.6
0.4
0.2
0.8
1
x
(c)

H
I
G
H

(
x
)
0
0
0
.
2
0
.
4
0
.
6
0
.
81
1
.
2
1
.
4
1
.
6
1
.
82
2
.
2
2
.
4
2
.
6
2
.
83
Figure 10.9 The membership function of fuzzy sets of P
r
(LOW, MEDIUM, HIGH)
(3) K
a
, K
r
AND THEIR FUZZY SETS
The level of knee-point voltage (K
a
), its trend (K
r
) and their fuzzy sets LOW,
MEDIUM, HIGHarecalculatedasshowninTable10.10.
Usingthefuzzysetsof P
a
, P
r
, T
a
, T
r
, K
a
, K
r
andspecifiedweights, the14hypo-
thetical motorsareexaminedtogivetheir failureindexes, asgiveninFigure10.10.
Theweightsmaybeselectedfromexperienceandcanbetunedupintheanalysis. It
canbeseenthat motorsNo. 8andNo. 12havethehighest failureindicesandshould
260 Condition Assessment of High-Voltage Insulation
Table 10.7 Six important parameters of the hypothetical HV motors
No. P
a
(pC) P
r
T
a
T
r
K
a
K
r
1 3000 1.4 0.01 1.3 0.8 0.2
2 5000 1.3 0.015 1.4 0.75 0.15
3 8000 1.35 0.012 1.2 0.75 0.5
4 10000 1.4 0.015 1.3 0.7 0.65
5 20000 1.6 0.02 1.1 0.7 0.5
6 25000 1.8 0.01 1.2 0.7 0.5
7 10000 1.3 0.008 1.6 0.7 0.75
8 30000 1.3 0.015 2 0.65 0.9
9 40000 1.5 0.01 1.2 0.65 0.5
10 50000 1.4 0.03 1.4 0.65 0.75
11 60000 1.3 0.02 1.3 0.7 0.6
12 10000 2 0.005 1.1 0.7 0.9
13 35000 1.1 0.015 1.2 0.65 0.5
14 80000 1.4 0.012 1.1 0.65 0.8
Table 10.8 P
a
, P
r
and their fuzzy sets of the hypothetical HV motors
No. P
a
(pC) LOW MED. HIGH P
r
LOW MED. HIGH
1 3000 0.999 0 0.001 1.4 0.764 0.151 0.085
2 5000 0.999 0 0.001 1.3 0.938 0.012 0.05
3 8000 0.025 0.974 0.002 1.35 0.88 0.055 0.065
4 10000 0.003 0.995 0.002 1.4 0.764 0.151 0.085
5 20000 0 0.994 0.006 1.6 0.194 0.53 0.277
6 25000 0 0.988 0.012 1.8 0.056 0.101 0.843
7 10000 0.003 0.995 0.002 1.3 0.938 0.012 0.05
8 30000 0 0.971 0.029 1.3 0.938 0.012 0.05
9 40000 0 0.678 0.322 1.5 0.425 0.425 0.15
10 50000 0 0 1 1.4 0.764 0.151 0.085
11 60000 0 0 1 1.3 0.938 0.012 0.05
12 10000 0.003 0.995 0.002 2 0.019 0 0.981
13 35000 0 0.914 0.086 1.1 0.98 0 0.02
14 80000 0 0 1 1.4 0.764 0.151 0.085
beput onthefirst priority for maintenanceor replacement. Thesetwomotors have
thehighest changerateof PDlevel, DDF or knee-point voltage, whichindicatesthat
their rateof insulation condition deterioration is high. Although motor No. 14has
Artificial-intelligence techniques 261
Table 10.9 T
a
, T
r
and their fuzzy sets of the hypothetical HV motors
No. T
a
(pC) LOW MED. HIGH T
r
LOW MED. HIGH
1 0.01 0.998 0 0.002 1.3 0.938 0.012 0.05
2 0.015 0.059 0.938 0.003 1.4 0.764 0.151 0.085
3 0.012 0.708 0.29 0.002 1.2 0.969 0 0.031
4 0.015 0.059 0.938 0.003 1.3 0.938 0.012 0.05
5 0.02 0.004 0.99 0.006 1.1 0.98 0 0.02
6 0.01 0.998 0 0.002 1.2 0.969 0 0.031
7 0.008 0.998 0 0.002 1.6 0.194 0.53 0.277
8 0.015 0.059 0.938 0.003 2 0.019 0 0.981
9 0.01 0.998 0 0.002 1.2 0.969 0 0.031
10 0.03 0 0.971 0.029 1.4 0.764 0.151 0.085
11 0.02 0.004 0.99 0.006 1.3 0.938 0.012 0.05
12 0.005 0.999 0 0.001 1.1 0.98 0 0.02
13 0.015 0.059 0.938 0.003 1.2 0.969 0 0.031
14 0.012 0.708 0.29 0.002 1.1 0.98 0 0.02
Table 10.10 K
a
, K
r
and their fuzzy sets of the hypothetical HV motors
No. K
a
(pC) LOW MED. HIGH K
r
LOW MED. HIGH
1 0.2 0.994 0 0.006 0.2 0.004 0.994 0.002
2 0.25 0.494 0.494 0.012 0.15 0.059 0.94 0.002
3 0.25 0.494 0.494 0.012 0.5 0 0.987 0.013
4 0.3 0.057 0.915 0.028 0.65 0 0.947 0.053
5 0.3 0.057 0.915 0.028 0.5 0 0.987 0.013
6 0.3 0.057 0.915 0.028 0.5 0 0.987 0.013
7 0.3 0.057 0.915 0.028 0.75 0 0.824 0.176
8 0.35 0.012 0.903 0.085 0.9 0 0.107 0.893
9 0.35 0.012 0.903 0.085 0.5 0 0.987 0.013
10 0.35 0.012 0.903 0.085 0.75 0 0.824 0.176
11 0.3 0.057 0.915 0.028 0.6 0 0.968 0.032
12 0.3 0.057 0.915 0.028 0.9 0 0.107 0.893
13 0.35 0.012 0.903 0.085 0.5 0 0.987 0.013
14 0.35 0.012 0.903 0.085 0.8 0 0.657 0.343
thelargest PD level andlowknee-point voltage, becauseits PD level is stableand
knee-point changerateisnot thelargest, it isnot rankedtotheworst motor.
Fromthisexample, it canbeseenthat visual examinationof thetest resultsmay
not besufficient to determinetheconditionrankingof atypeof equipment. Using
fuzzy-logicanalysisandsuitableweightsinthecalculation, thefailureindexcouldbe
moreaccurateintermsof thetotal conditionassessment. Of course, thecoefficients
262 Condition Assessment of High-Voltage Insulation
0.6
Failure index
0.6
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G11 G12 G13 G14
Figure 10.10 The failure index of 14 hypothetical HV motors determined from their
PD and DDF test results using a fuzzy-logic analysis technique
of membershipfunctionandweightsshouldbeadjustedaccordingtotheexperience
of individual utilityandanalysisresultsfromtimetotime.
10.4 Summary
Therapiddevelopment inartificial-intelligencetechniques has ledto awiderange
of applicationsininsulationconditionmonitoringareas. Expert systems, neural net-
worksandfuzzylogichavebeenusedinmaintenance, diagnosisandconditionranking
of electrical plant. Withtheaidof neural networks andfuzzy logic, patternrecog-
nitionisnowusedfor identificationof partial dischargesourcesandsuppressionof
interference. As anexample, thefuzzy-logic DGA andkey gas analysis technique
presentedinthischapter hasbeensuccessfullyusedfor thediagnosisof transformers
inAustralia. Ithasbeenprovedthat, usingthefuzzydiagnosismethod, moredetailed
information about the incipient faults inside a transformer can be obtained. Also,
multiplefaults canbediagnosedusingthis method, whichmay not bepossibleby
other knowntechniques.
Another advantageof thefuzzy-diagnosis method is its quantitativeindication
of thefault likelihood/dominancebymeansof thefuzzydiagnosticvector. Thishas
been used to identify the main faults and determine their severity in comparison
with each other. It has been found fromour experience[12,13] that thelarger the
fuzzy component, themoredominant, activeand severethefault. After a certain
period of monitoring, the trend of each fault in a transformer can also be deter-
minedfromitsfuzzydiagnosticvector. Thisinformationisimportantforanydecision
regardingthereplacement or/andrefurbishment of asuspect transformer. Inthecur-
rent economic climate, thereis anincreasingdemandtoextendtheservicelifeand
reducemaintenancecostsof HV equipment. Withtheaidof moreaccurateinsulation-
condition-assessment techniques, suchasthefuzzy-logic conditionrankingmethod
Artificial-intelligence techniques 263
discussedinthis chapter, themaintenanceschedulecanbeoptimizedandalonger
service life in HV equipment and cables achieved. This would delay the invest-
ment innewequipment and, at thesametime, keepthenetwork performanceat an
acceptablelevel. Thefault-diagnosistechniquehasbeenimplementedinacomputer
programICM-2.0withalargedata-storagecapacity. WithaPCPentiumIV 2.0GHz,
the analysis of each set of DGA results takes about 1 second. The programalso
has auser-friendly interfaceandhas beenusedby several utilities inAustraliaand
NewZealand.
10.5 References
1. Dornerburg, E., Strittmatter, W., Monitoring oil cooling transformers by gas
analysis, Brown Boveri Review, May1974;61:23847
2. Kelly, J .J ., Transformer faultdiagnosisbygas-gasanalysis, IEEE Transactions
on Industry Applications, December 1980;16(4):77782
3. Rogers, R., IEEE and IEC codes to interpret incipient faults in transformer,
usinggasinoil analysis, IEEE Transactions on Electrical Insulation, October
1978;13(5):34954
4. Lin, C. E., Ling, J . M., andHuang, C. L., Anexpertsystemfor transformer fault
diagnosisandmaintenanceusingdissolvedgasanalysis, IEEE Transactions on
Power Delivery, J anuary1993;8(1):23138
5. Tomsovic, K., Tapper, M., andIngvarsson, T., A fuzzy informationapproach
tointerpretingdifferent transformer diagnosticmethods, IEEE Transactions on
Power Delivery, J uly1993;8(3):163846
6. Huang, Y.C., Yang, H.T., and Huang, C.L., Developing a new transformer
diagnosissystemthroughevolutionaryfuzzylogic, IEEETransactions onPower
Delivery, April 1997;12(2):7617
7. Wang, Z., Liu, Y., andGriffin, P.J ., A combinedANN andexpert systemtool
for transformer fault diagnosis, presented at the IEEE PES Winter Meeting,
PE-411-PWRD-0-12-1997, NewYork, February1998
8. IEC 60599-1999, Interpretation of the analysis of gases in transformers and
other oil-filledelectrical equipment in-service
9. Zadeh L., Fuzzy Sets, Information and Control, vol. 8(Academic Press, New
York, 1965), pp. 33853
10. IEEE StdC57.104-1991, IEEE Guidefor theInterpretationof GasesGenerated
inOil-ImmersedTransformers, Transformer Committeeof IEEE PES, 1991
11. Bell, S.R., The Cauchy Transform, Potential Theory, and Conformal Mapping
(CRC Press, BocaRaton, Fl., 1992)
12. Su Q., and Mi, C., Fuzzy diagnosis of transformer and generator faults,
presented at AUPEC96, Melbourne University, Melbourne, October 1996,
pp. 38994
13. Su, Q., Mi, C., et al., A fuzzydissolvedgasanalysismethodfor thediagnosisof
multipleincipientfaultsinatransformer, IEEE Transactions on Power Systems,
May2000;15(2)593598
264 Condition Assessment of High-Voltage Insulation
14. Su, Q., Reliability centeredmaintenanceof electrical plant someimportant
issues, presented at AUPEC97, UNSW, Sydney, SeptemberOctober 1997,
pp. 61520
15. Su, Q., Insulation condition assessment of largegenerators, presented at the
International Power andEnergyConference, Melbourne, 1999, pp. 1238
10.6 Problems
1. Basedonthefuzzy-logicmethodexplainedinthischapter, derivetheIEC ratio
codes and calculate the fuzzy logic elements F(0)F(7) for the transformers
whoseDGA test resultsaregiveninthefollowingtable. Fill inthederivedand
calculatedresultsinthefollowingtables.
No. H
2
CH
4
C
2
H
2
C
2
H
4
C
2
H
6
IEC ratiocodes
1 95 110 <0.1 50 160
2 120 17 4 23 32
3 300 490 95 360 180
No. F0 F1 F2 F3 F4 F5 F6 F7
1
2
3
2. Referringtotechnical papersandbooks, findthedifferencebetweentheIECcode,
Rogersandtrianglemethods. Isthereanyrelationshipbetweenthesemethodsin
termsof thecomparisonbetweendifferent gasconcentrations?
Appendix 1
Abbreviations
CIGRE Conseil International des Grands Reseaux Electriques (International
Council onLargeElectricSystems, Paris, France)
DEIS IEEE DielectricsandElectrical InsulationSociety
EPRI ElectricPower ResearchInstitute(PaloAlto, California)
ERA ERA TechnologyLtd(Leatherhead, Surrey, UK)
IET Institution of Engineering and Technology (incorporating the former
Institutionof Electrical Engineers, UK)
IREQ Institut deRecherchedHydroQuebec(Varennes, Quebec)
ISH International ConferenceonHighVoltageEngineering(bi-annual con-
ference)
Appendix 2
Major standards organizations
ANSI AmericanNational StandardsInstitute
ASTM AmericanSocietyfor Testingof Materials(ASTM International)
BSI BritishStandardsInstitute
CISPR International Special CommitteeonRadio Interference(subcom-
mitteeof IEC)
DKE (DIN/VDE) German Commission of Electrical, Electronic and Information
Technologies(standardsandregulations)
IEC International Electrotechnical Commission
IEEE Instituteof Electrical andElectronicEngineers(USA)
ISO International Organizationfor Standardization
SA StandardsAustralia(SAI GLOBAL)
WSSN World Standards Services Network (www.wssn.net/WSSN/
index.html)
Appendix 3
Answers to problems
Chapter 1
Nocalculations.
Chapter 2
Q.1: SeeChapter 6.
Q.2: (a) BDV 22.4kV (RMS) E
air
= 2.12kV/mm(RMS) E
resin
= 0.6kV/mm
(RMS).
(b)BDV74.75kV(RMS)E
gas
= 7.07kV/mm(RMS)E
resin
= 2.02kV/mm
(RMS).
(c) BDV 84.3kV (RMS) E
oil
= 7.5kV/mm(RMS) E
resin
= 4.72kV/mm
(RMS). SeeFigures2.2and2.3.
Q.3: C
b
= 0.0072pF C
c
= 0.0625pF C
a
C
b
q
c
= 285pC q
a
32pC
V
i
= 31.1kV (RMS).
Chapter 3
Q.1: Refer toChapters2, 4and5.
Q.2 Refer toChapters2, 4, 5and9.
Q.3: (i) E
oil
= 11.8kV/mm(RMS) E
pbd
= 5.9kV/mm(RMS).
(ii) E
oil
= 2.7kV/mmE
pbd
= 27kV/mm. Notehigher stressin solid.
Q.4: C
oil
= 27.8pF C
pbd
= 129pF Total C = 22.9pF.
R
oil
= 710
10
R
pbd
= 3010
10
.
Q.5: (i) 94kV (RMS).
(ii) 127kV (RMS). Reducedoil duct thicknessallowshigher V
i
.
268 Condition Assessment of High-Voltage Insulation
Chapter 4
Q.1: E
m
= (V
0
V
1
)/r
o
lnr
1
/r
o
= (V
1
V
2
)/r
1
lnr
2
/r
1
= . . . . . . . . . C
1
(V
0

V
1
) = Kl
1
(V
0
V
1
)/ lnr
1
/r
o
. AlsoC
1
(V
0
V
1
) = C
2
(V
1
V
2
) = . . . . . . . . .
K = 2
o

r
andl
1
is lengthof first foil (400mm). Fromtheseequations and
E
m
beingequal thefoil lengthxradiusisthesamefor eachcapacitor.
(i) l
1
= 400mm, l
2
= 307mm, l
3
= 250mmandl
4
(outer) = 210mm.
(ii) Capacitance values: K l/ lnr
n
/r
n1
. C
1
= 296pF, C
2
= 288pF, C
3
=
283pF andC
4
= 292pF
Total capacitanceC, HV Earth= 73pF. All inseries.
(iii) Voltages/section= CV/C
n
; C = 73; V = 60kV (RMS); (V
0
V
1
) = 73
60/296= 14.8kV; (V
1
V
2
) = 15.2kV; (V
2
V
3
) = 15.4kV; (V
3
V
4
) =
15.0kV. Small variationsincapacitancecalculations.
Maximumradial stress, E
m
= 2.8kV/mm. Similar inall interfoil insulation.
(iv) Maximumradial stresswithout foils = 60/20ln44/20 = 3.8kV/mm at
HV conductor.
Notes: Thedesignproblemsrelatingtoaxial andradial stressesarecomplex(see
Chapter 12of Reference18 Chapter 4).
Theaveragevaluesbetweenfoil endsinthisproblemrangefrom0.16kV/mm
to0.37kV/mm. Approximatelyaxial uniformdistributionsareoftenrequired.
Q.2: SeeReference5of Chapter 7.
Q.3: For 2.0kV/mm(RMS) applied, possiblePDs for gap of 0.3/0.4mm. PDs for
gapof 0.5mm. For 3.0kV/mm(RMS) appliedPDsfor gap> 0.1/0.2mm.
Chapter 5
Q.1: (i) Speeddifferences. Internal water coolinginlargeturbinegenerators.
(ii) Semiconductingtapesor similar.
(iii) Refer toChapter 9.
Q.2: Air stressduetoappliedvoltagefollowinglocal erosionof theearthedscreen=
6.6kVpk/mm. The0.5mmair gapb/d = 5.4kVpk/mmthusPDsarepossible.
Hydrogenstress ingapfor b/d = 10kVpk/mm, thus appliedvoltagerequired
for PDs= 9.6kV (RMS) toground.
Q.3: (ii) Fromequation(Figure5.4) for = 1voltageacrosslineenddiscs = 11%
of linevoltage. For = 4voltageis26per cent of linevoltage.
Oil b/d stress for 8mmgap = 25kV/mm(Figure2.3(c)) requiring avoltage
betweenthelinediscsof 225kV.
For = 1 linevoltageexpected to causeb/d = 225 100/11 = 2049kV.
Assuming safety factor of approximately 1.3 choose test level <1576kV
(525kV class).
For = 4 line voltage expected to cause b/d = 225 100/26 = 867kV.
Assumingfactor of 1.3choosetest level <667kV (145kV class).
Q.4: Refer totablesinChapters9and10.
Answers to problems 269
Chapter 6
Q.1: Resistivedivider addmatchingresistor of Z
0
at receivingendof cable.
Addseriesresistor at divider endinsomecasestomatchcableandreduce
possiblereflections. Ratio3010
5
if noseriesresistor. Ratio3310
5
withseriesresistor.
Capacitivedivider addseries resistor, Z
0
, at divider 50per cent pulse
input to cablebut maintainingrequiredtimeconstant. Highimpedanceat
receivingendof cableallowsdoublingof thetransmittedsignal. Ratio50
10
5
.
Q.2: (i) Temperaturecritical. Parallel pathsmaymaskresult.
(ii) Ratioarmbridgemoreversatile.
Q.3: Cx = 45pF andtan = 0.0086Asthisvalueisabovetheacceptablevalue
for anewbushingconsider applicationof PDtest.
Q.4: Applying the simplified relationships, the relative values with (i) as
referenceare(i) 1, (ii) 7.3and(iii) 67.
Improvement may beachievedby increasingthecouplingcapacitor value
(e.g. to 1000pF) and/or adjusting the amplifier gain if its noise level is
acceptable.
Chapter 7
Q.2: Pulseseparationbetweendirect andreflectedpulsest =2(L x) / v. For
v =164m/sand(L x) =100, t =1.22s. Thepulsevelocity canbe
determinedfrommeasurements as inFigure7.2. Thereducedvelocity of
about 20%is dueto theinfluenceof theouter andinner semi-conducting
layersinanXLPEcable. Thishasbeenconfirmedinresearchinvestigations.
Chapter 8
Nocalculations.
Chapter 9
Nocalculations.
Chapter 10
Nocalculations.
Index
abbreviations 265
ageing
deteriorationover time 356
fromelectric/thermal stress 467
air asaninsulator
air gapflashover voltages 23, 24
humidityissues 23
alternatingvoltagesystems
basicsystemcomponents 27
generationvoltages 34
switchingandisolation 4
artificial intelligence(AI) 24163
about AI 241, 2623
see also asset analysisandcondition
ranking; databasesfor fault
diagnosis/conditionassessment;
fuzzy-logicfault diagnosis
assessment/testing 12158, 15988
about assessment 121
about insulationtesting 15960
assessment: see insulationmeasurable
parameters; non-destructiveelectrical
measurements; physical/chemical
diagnosticmethodsof insulation
assessment; test voltage
generation/measurement
standardsfor 1578, 1858
testingsee under specific equipments
asset analysis/conditionranking 25562
fuzzylogicrankingexample 25862
insulationhealthindex 2556
membershipfunctionsof fuzzyset
2568
rankingbyinsulationcondition 255
BoundaryElement Methodprograms 60
breakdown, andoperatingstresses 4850
bushings, high-voltage 747
castingresinbushings 767
direct voltageapplications 76
equipotential fieldplots 756
external flashovers 75
insulationtesting 1678
oil-impregnatedpaper bushings 75
older unitswithSRBP 75
stresscontrol 745
stresses/breakdowns 49
cables
distributedtemperaturesensing(DTS) with
optical fibres 1969
stresses/breakdowns 49
see also extrudedcables, high-voltage;
gas-insulatedtransmissionlines;
oil-impregnatedpaper (OIP) cables,
highvoltage
capacitance 33
capacitors: see power capacitors
capacitor-typevoltagetransformers(CVT) 81
routinetests 171
typetests 170
ceramicinsulators, testing 1612
CIGRE onmachineinsulationdegradation
103
circuit breakers, high-voltage 846
breakdowns/failures 85
internal flashover 85
partial dischargetesting 856
Index 271
routinemonitoring 86
testing 1645
routinetests 165
typetests 165
compositeinsulators
surfacedeterioration 153
testing 162
conditionmonitoring
offlineproblems 2078
onlineadvantages 208
conditionmonitoringonline, acoustic/electric
for transformers 2247
acoustic/VHF/UHF transducers 2247
acoustic/windingterminal 224
conditionmonitoringonline, electrical
22536
capacitancemeasurement 2278
dielectricdissipationfactor (DDF) 2278
dielectriclossangle(DLSA) 2278
leakagecurrent measurement 2289
PDdetection
withanRF current detector 233
GISPDdetection 230
largerotatingmachines 2323
withaPartial DischargeAnalyser
(PDA) (OntarioHydro) 2334
IRISTurbo-Generator 2345
onpower cables 2312
transformers/VHF/UHF sensing
2301
UNSW-SMHEA PDdetector 2356
conditionmonitoringonlinenoise-mitigation
20818
differential methods 21114
balancedcircuits 211
discharge-freecapacitor 212
parallel circuits 21314
multipleterminal PDmeasurement
21518
noisegating 20911
withnoiseidentificationbysignal
waveformanalysis 21415
PDmeasurement hardware/software
developments 2089
conditionmonitoringonline, non-electrical
21924
acoustic-basedPDdetection 2224
cableterminations 2223
portablePDdetectors 2234
transformer windings 222
onlineDGA 21922
temperaturemonitoring 219
temperaturelabels 219
thermal scanning 21920
conditionmonitoringsensors 189205
PDdirectional sensors 2003
ultra-high-frequencysensors 18991
see also optical-fibresensors/sensing
conditionranking: see asset analysisand
conditionranking
current transformers 49, 801
databasesfor fault diagnosis/condition
assessment 2414
advantagesof 242
combinedmethodfor DGA analysis
2434
MonashUniversityHVICM Group
software 2423
DDB (dodecylbenzene) insulation 27
design: see Insulationdesign
dielectricdissipationfactor (DDF) 345
measurement 13740, 2278
HV ScheringBridge 1378
transformer ratio-armbridge
13840
dielectriclosses 379
dielectriclossangle(DLA) 2279
losscalculationexample 389
lossrelationshipdetermination 378
dielectricresponsemeasurement 14750
dispersionmeters 147
frequencydomainspectroscopy(FDS)
150
timedomainmeasurements 14750
isothermal relaxationcurrent (IRC)
14950
polarizationanddepolarization
current (PDC) 149
recoveryvoltagemeasurement
(RVM) 148, 149
direct current systems: see highvoltagedirect
current (HVDC) systems
direct-voltagetesting 1415, 1289
disconnector testing 166
dispersioneffect 34
dispersionmeters 147
dissolved-gasanalysisinoil (DGA)
database/combinedmethodanalysis 2434
IEC codesfor 2445
measurements 1512
online 21922
272 Condition Assessment of High-Voltage Insulation
dissolved-gasanalysisinoil (DGA) (cont.)
see also fuzzy-logicfault diagnosiswith
DGA
distributedtemperaturesensing(DTS), cables
1969
EHV undergroundpower cables, directional
PDsensors 2001
elastomersasinsulators 2930
electrical breakdownandoperatingstresses
4850
electrical stressdistributionsininsulation
608
about stressdistribution 601
BoundaryElement Methodprograms 60
edgeeffects 646
electrical stresscontrol 689
FiniteElement Methodprograms 60
multipledielectrics
withconcentricelectrodes 634
inparallel fields 613
multipleelectrodeeffects 668
electricstressageing 46
electrostaticchargingtendencies(ECT) of oils
267
EPDM (ethylenepropylenehexadiene
monomer) insulation 29
epoxyresinsasinsulators 29
EPR (ethylenepropylenerubber) 29
extrudedcables, high-voltage 903
about XLPE andEPR cables 901
insulationtesting
installationtests 1745
routinetests 174
typetests 1734
potential faults 91
terminations 912
water treeproblems 90
FIK keygasmethodwithfuzzyIEC codes
24955
FiniteElement Methodprograms 60
frequencydomainspectroscopy(FDS) 150
Furanmeasurements 152
fuzzy-logicfault diagnosiswithDGA 24455
about conventional DGA 2445
about fuzzy-logicmethods 2456
conventional/fuzzycomparison 2501
fuzzyIEC codes 2469
keygasmethod(FIK) 24953
IEC codesfor DGA, conventional 2445
power transformer application 24953
trendanalysis 2535
fuzzyset membershipfunctions 25662
fuzzylogicconditionranking 25862
gasanalysismonitoring, oil filledtransformers
110
gases, asinsulatingmaterials 224
see also air asaninsulator; SF
6
gas-insulatedsystems(GIS) 4, 234, 867
diagnosticmeasurement 867
optical-fibrePDsensors 1945
UHF sensorsfor PDdetection 1901
gas-insulatedtransmissionlines 923
generators 3, 5
see also motorsandgenerators
glassasaninsulator 28
glassinsulators, testing 1612
heat-shrinkableinsulatingmaterials 30
high-voltagecables: see extrudedcables,
high-voltage; gas-insulated
transmissionlines; oil-impregnated
paper (OIP) cables, highvoltage
highvoltagedirect current (HVDC) systems
89
converter transformer issues 9
dielectrictesting 1824
high-power thyristors 8
principlesof operation 8
systemsinoperation 89
high-voltage(HV) power systems 19
designlife 12
extendinglifeof 16
futureinsulationrequirements 17
interconnectioncomponents 29
newsystemsandequipment 16
present assetscondition 1516
reliabilityrequirements 15
see also alternatingvoltagesystems
high-voltageinstrument transformers 7781
about instrument transformers 7781
capacitor-typevoltagetransformers(CVT)
81
dry-typecurrent transformers 801
insulationtesting 1689
routinetests 169
special tests 169
typetests 1689
oil-impregnated 7880
Index 273
failurecauses 7980
high-voltagetest levels 1015
direct-voltagetests 1415
impulsetest waveshapes 10, 12
lightning-impulsevoltages 1314
one-minutetest values 1112
possibletest levels 1011
power-frequencyovervoltagelevels 13
safetyfactor (SF) 11
switchingsurges 14
U
m
levels 10
veryfast transient tests(VFTT) 14
see also test voltage
generation/measurement
hot-spot detectionintransformers
1967
HTV (hightemperaturevulcanizedsilicone
rubber) insulation 2930
hydrogenasaninsulator 24
IEC codesfor DGA 2445
fuzzyIEC codes 2469
keygasmethod(FIK) 249
inductivevoltagetransformers 16970
impulsewithstandtests 170
partial dischargetests 170
power-frequencywithstandtests 170
insulatingmaterials 2154
about insulatingmaterials 212
deteriorationover time 356
elastomers 2930
electrical breakdownandoperatingstresses
4850
epoxyresins 29
futuremonitoringrequirements 17
glass 28
heat-shrinkablematerials 30
liquid-impregnatedsystems 313
oil-impregnatedpaper (OIP) 31
polyethylene 30
polypropylene(PE) film 31
polytetrafluoroethylene(PTFE)
301
polyvinyl chloride(PVC) 30
porcelain 278
power capacitor insulation 33
rigidlaminates 28
sheet composites 289
standards 534
tapes 289
Teflon(PEP) 31
vacuum 25
wood 27
see also gases...; liquids...
insulationcoordination 910
insulationdefects: see bushings, high-voltage;
circuit breakers; gas-insulated
systems(GIS); high-voltagecables;
high-voltageinstrument transformers;
motorsandgenerators, insulation
defects; post insulators; surge
arresters; suspension(string)
insulators
Insulationdesign 5569
about insulationdesign 55
electrical requirements 56
mechanical requirements 57
physical limitations 56
processingprocedures 58
reliabilityissues 589
thermal conditions 58
workingenvironment 567
see also electrical stressdistributionsin
insulation
insulationhealthindex 2556
insulationmaterial deteriorationandfailure
3647
see also ageing; dielectriclosses; partial
discharges(PD)
insulationmeasurableparameters 336
capacitance 33
dielectricdissipationfactor (DDF) 345
insulationresistance(IR) 334
PD(partial discharges) 35
permittivity 33
physical andchemical changes 356
resistivity 334
timeconstants 34
see also conditionmonitoring...; partial
discharges(PD),
detection/measurement
insulationresistance(IR) 334
insulationresistance(IR) measurement 1357
about IR measurement 1356
Megger circuit 1356
short-timetest 136
step-voltagetest 137
timeresistancetest 1367
insulators, stresses/breakdowns 50
intensiometricoptical-fibresensors 193
interferometricoptical-fibresensors 1934
274 Condition Assessment of High-Voltage Insulation
IRISTurbo-Generator PDAnalyser (TGA)
(OntarioHydro) 2345
isolator testing 166
isothermal relaxationcurrent (IRC) 14950
keygasmethodwithfuzzyIEC codes(FIK)
24955
laminates, rigid, asinsulators 28
lightningimpulsevoltages 1314
generation/testing 12933
liquid-impregnatedinsulation 313
liquidsasinsulators
DDB (dodecylbenzene) 27
PCB problems 27
see also oil asaninsulator
LSR (liquidsiliconerubber) 30
Megger circuit 1356
Melinex/Mylar 289
modalmetricoptical-fibresensors 193
moistureissues
oil-paper systems 1501
transformer insulation 323
MonashUniversityHVICM Groupsoftware
2423
motorsandgenerators, insulationdefects
failuremechanisms 1003
CIGRE degradationsummary 103
end-windinginsulation 1023
inter-turninsulation 1012
mainwall insulation 101
futuredevelopments 1034
generators 99
insulationsystems 99100
resin-richsystems 100
shellac/bitumenbinding 99100
vacuum-pressure-impregnation(VPI)
process 100
motors, high-voltage 989
motors, low-voltage 978
nitrogenasaninsulator 24
Nomex(insulatingsheet) 28
non-destructiveelectrical measurements: see
dielectricdissipationfactor (DDF);
dielectricresponsemeasurement;
insulationresistance(IR)
measurement
oil asaninsulator 257
electrostaticchargingtendencies(ECT)
267
gasabsorbingeffects 26
paraffinicandnaphthenictypes 25
particles, effectsof 25
qualityassessment 1523
oil-impregnatedpaper (OIP) cables, high
voltage 31, 8790
in-serviceconditionindicators 1503
insulationtesting 173
joints 89
sheathdamage 87
terminationfaults 889
thermal runaway 878
typesof 87
online: see conditionmonitoringonline...
operatingstressesandbreakdownvalues
4850
optical-fibresensors/sensing 1909
about optical-fibresensors 1903
advantages/disadvantages 199
intensiometricsensors 193
interferometricsensors 1934
modalmetricsensors 193
PDmeasurement 1946
ongeneratorsandGIS 1945
ontransformers 1956
polarimetricsensing 193
temperaturesensors 1969
power cables 1969
power transformer hot-spots 196
overheadline/substationinsulators
1603
about insulator testing 1601
ceramicandglass(post type) 1612
compositeinsulators 162
porcelainandglass 161
routine/sample/typetests 161
Partial DischargeAnalyser (PDA) (Ontario
Hydro) 2334
partial discharges(PD)
apparent partial discharge 43
damagefrom 445
equivalent capacitancenetwork 423
instantaneousinceptionvoltage 41
simplifiedmodel 412
sources 3941
examples 40
Index 275
partial discharges(PD),
detection/measurement 1407
about PDmeasurement 1401
acoustic-based 2224
basictest circuit 142
calibrationprocedures 1426
calibrator calibration 145
direct calibration 1435
circuit sensitivity 1456
directional sensorsfor cables 2003
coupler sensors 2001
fieldsensors 2013
ideal PDenergylevels 1401
interferencereduction 146
asameasurableparameter 35
measurement presentation 1467
optical-fibresensors 1946
radiointerferencevoltage(RIV)
techniques 141
ultrasonicdetection 154
see also conditionmonitoringonline,
electrical; transformer insulation
testing
PaschensLaw 101
PCB (polychlorobiphenyl) liquids 27
PE (polypropylene) filminsulation 31
PEP (Teflon) insulation 31
permittivity 33
physical/chemical diagnosticmethodsof
insulationassessment 1504
compositeinsulators, surfacedeterioration
153
dissolvedgasinoil (DGA) 1512
oilpaper systemindicators 1503
degreeof polymerization 153
Furanmeasurements 152
moisturecontent 1501
oil qualityassessment 1523
particlecount 152
SF
6
analysis 153
ultrasonicPDdetection 154
water treeinginXPLE cables 153
polarimetricoptical-fibresensing 193
polarizationanddepolarizationcurrent (PDC)
measurements 148, 149
polyester fibremats 28
polyethyleneinsulation 30
polymerizationdegreeof oil 153
polypropylene(PE) filminsulation 31
porcelainasaninsulator 278
porcelaininsulators, testing 161
post insulators 734
power cables: see cables; extrudedcables,
high-voltage; gas-insulated
transmissionlines; oil-impregnated
paper (OIP) cables, highvoltage
power capacitors
insulation 33, 823
stresses/breakdowns 49
type/routine/special testing 171
power transformers: see transformer ...
PTFE (polytetrafluoroethylene) 301
PVC (polyvinyl chloride) insulation 30
recovery(return) voltagemeasurement (RVM)
148, 149
reliability
designissues 589
power systemrequirements 15
renewableresources 2
resistivity 334
rigidlaminatesasinsulators 28
rotatingmachines
insulatingtesting 1713
assembledmachines 1723
stator bars 172
stresses/breakdowns 49
RTV (room-temperature-vulcanizedsilicone
rubber) insulation 30
safetyfactor (SF), test levels 11
ScheringBridge, HV 1378
sensors 189205
ultra-high-frequency(UHF) sensors
18991
see also optical-fibresensors
SF
6
analysis 153
asaninsulator 234
incircuit breakers 84
siliconerubber insulation 2930
spheregaps 132
standardsorganizations 266
substations: see overheadline/substation
insulators
surgearresters, high-voltage
faults/failures 834
testing 1634
suspension(string) insulators 713
cap-and-pininsulators 72
coronadischargeproblems 72
276 Condition Assessment of High-Voltage Insulation
suspension(string) insulators(cont.)
long-rodcompositepolymericinsulators
723
siliconerubber (SIR) weather sheds 73
switchgear insulationtesting 1647
circuit breakers 1645
disconnectors(isolators) 166
metal-enclosedswitchgear 166
self-protectedswitchgear 166
transformer tapchangers 167
switchingsurges 14
tapesasinsulators 289
Teflon(PEP) insulation 31
Teslacoil circuit 127
test levels: see high-voltagetest levels
test voltagegeneration/measurement 12235
direct voltages 1289
high-frequencyvoltages 127
high-voltageon-sitetesting 1335
hybridtest circuits 129
lightningimpulsevoltages 12933
four-stageimpulsegenerator 1312
Marxcircuit 1312
spheregaps 132
operationof testingtransformers 1236
accuracyissues 1256
flashover problems 1245
power-frequencytest circuit 125
transformer regulators 124
power-frequencyvoltages 1227
multiple-unit (cascade) testing
transformers 1223
parallel resonant circuits 127
seriesresonant circuits 1267
single-unit testingtransformers 122
switchingsurgevoltages 133
very-low-frequencyvoltage(VLF) 128
thermal scanning 21920
thermal stressageing 467
timeconstants 34
transformer insulationissues 10618
about transformer insulation 1068
bushing, internal 11516
flashover problems 116
coreandmagneticshields 117
drytypewindings 109
endinsulation 11415
failurerates 111
gasanalysismonitoring 110
insulationtotankandcore 115
interwindinginsulation 11214
contaminationproblems 114
largelayer-typehigh-voltagewindings 110
low-kVA oil impregnatedwindings 1089
major failures 117
oil-filled, moistureissues 323
oil-impregnatedwindings 10912
onlinemonitoring 11112
shell-typecoils 11011
stresses/breakdowns 49
tapchangers 117
transformer insulationtesting 17582
hot spot detection 1967
optical-fibrehot-spot detection 1967
optical-fibrePDmeasurements 1956
partial discharge(PD) tests 17780
calibrationandfault location 178
lightning-impulsetests 17881
measurement interpretation 177
switching-impulsetests 17881
power-frequencyovervoltagewithstand
1756
test requirementssummary 1802
transformer ratio-armbridge 13840
transformer andreactor windings 1047
general considerations 104
lightningimpulsedesign 1046
windingconfigurations 1067
transformers, instrument: see high-voltage
instrument transformers
transformer tapchanger testing 167
trendanalysis 2535
Turbo-Generator PDAnalyser (TGA)
(OntarioHydro) 2345
ultra-high-frequency(UHF) sensors 18990
for PDdetection 1902
UNSW-SMHEA PDdetector 2356
vacuumasaninsulator 25
vacuum-pressure-impregnation(VPI) 58, 100
veryfast transient tests(VFTT) 14
very-low-frequencyvoltage(VLF) testing
128
woodasaninsulator 27
XLPE (cross-linkedpolyethylene) insulation
30, 128, 153