Electrical Insulation

in Power Systems
 POWER ENGINEERING

Series Editor

H. Lee Willis
ABB Power T&D Company Inc.
Cary, North Carolina

1. Power Distribution Planning Reference Book, H. Lee WUHs

2. Transmission Network Protection: Theory and Practice, K G.
Paithankar
3. Electrical Insulation in Power Systems, N. H. Malik, A. A. Al-Arainy,
and M. L Qureshi
4. Electrical Power Equipment Maintenance and Testing, Paul Gill
5. Protective Relaying: Principles and Applications, Second Edition, J.
Lewis Blackburn

ADDITIONAL VOLUM ES IN PREPARATION

Electrical Insulation
in Power Systems

N. H. Malik
A. A. Al-Arainy
M. I. Qureshi
King Sand University
Riyadh, Saudi Arabia

Taylor &. Francis

Taylor Francis Group

Boca Raton London New York Singapore

A C R C title, part of the Taylor Sr Francis im print, a member of the

Taylor Sr Francis Group, the academic division of TSrF Informa pic.
Library of Congress Cataloging-in-Publication Data

Electrical insulation in power systems / by Nazar Hussain Malik, A. A. Al-Arainy,

Mohammad Iqbal Qureshi.
p. cm. - (Power engineering ; 3)
Includes bibliographical references and index.
ISBN 0-8247-0106-2 (hardcover)
1. Electric insulators and insulation. 2. Electric power systems—Protection.
3. Dielectrics. I. Malik, Nazar Hussain . II. Al-Arainy, A. A. III. Qureshi,
Mohammad Iqbal. IV. Series
TK3401.E42 1997
621.319'37-dc21 97-34427
CIP

The publisher offers discounts on this book when ordered in bulk quantities. For
more information, write to Special Sales/Professional Marketing at the address
below.

This book is printed on acid-free paper.

Copyright © 1998 by MARCEL DEKKER All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or
by any means, electronic or mechanical, including photocopying, microfilming,
and recording, or by any information storage and retrieval system, without
permission in writing from the publisher.

MARCEL DEKKER
270 Madison Avenue, New York, New York 10016
http://WWW.dekker. com

Current printing (last digit):

10 9 8 7 6 5 4 3 2

PRINTED IN THE UNITED STATES OF AMERICA

Series Introduction

Power engineering is the oldest and most traditional of the various areas within
electrical engineering, yet no other facet of modem technology is currently
undergoing a more dramatic revolution in both technology and industry
stmcture. As these changes take place, many of the traditional and fundamental
areas of power engineering become more important than ever. Certainly one
such area is insulation. Always an essential element for satisfactory power
system operation, its correct interpretation and application becomes more
critical at a time when deregulated power grids are utilizing ever higher voltage
levels to move greater amounts of power over longer distances.
Electrical Insulation in Power Systems, by Drs. Malik, Al-Arainy, and
Qureshi, is a particularly useful book, because it combines a very
comprehensive coverage of insulation methods with a consistent attention to
detail and practical application. All of the insulation approaches used in modem
power systems are presented in a straightforward and thorough manner,
including vacuum, air, gas, liquid, solid and composite dielectric technologies.
Within each ^rea, the authors address the traditional fundamentals completely,
and provide a good discussion of recent developments and their applications.
Equally important are the final three chapters of the book, which provide a very
cogent and well-organized presentation on testing and diagnostic procedures and
the interpretation of their results.
As the editor of the Power Engineering Series, I am proud to include
Electrical Insulation in Power Systems among this important group of books.
Like all the books in this series, this volume presents modem power technology
in a context of proven and practical application. It is useful as a reference book,
as a text in an advanced power systems curriculum, or for self-study and tutorial
III
iv Series Introduction

application. The Power Engineering series will eventually include books

covering the entire field of power engineering, in all of its specialties and
subgenres, all aimed at providing practicing power engineers with the
knowledge and techniques they need to meet the challenges of the electric
industry in the 21st century.

H. Lee Willis
Preface

Electrical insulation is the backbone of all modern power system networks.

Different types of dielectrics that constitute these insulating elements are
subjected to a variety of stresses during their life span. A thorough knowl­
edge of their fundamental properties is therefore essential for their appro­
priate and optimal design as well as their pre- and postinstallation testing
and reliable operation.
The rapidly growing demand for electrical energy in the world today
necessitates its transmission at extra high voltage levels. Electrical power
engineers must efficiently tackle the complicated problems of insulation
and overvoltages arising in power networks. To train and better equip these
engineers now and for the future, it is necessary that they be provided with
a relevant specialized background. At present, only a few books are avail­
able that address this problem, particularly when compared with the num­
ber of those published in other areas of electrical engineering. Even among
these few, no single book covers the entire range of essential topics. Some
are almost two decades old, and others are limited to monographs on spe­
cific individual topics such as switchgear, corona, insulating oils and gas
insulated systems (GIS). There is therefore a great need for a book that
covers in a single volume the topics that are pertinent to the electrical
engineer working with high voltage insulation.
Electrical Insulation in Power Systems is an attempt to fill this gap. It
is expected to be of considerable importance not only for engineers in­
volved with power utilities and the insulation industry, but also for senior
undergraduate and postgraduate students. We are associated with the sub­
VI Preface

ject of high voltage insulation as researchers and as teachers, at both the

undergraduate and graduate levels, of specially tailored courses for prac­
ticing engineers working in industry and electric utilities. Most of the ma­
terial presented in this book is the outcome of lectures prepared for uni­
versity and industry audiences. The book covers in detail the fundamental
properties of dielectrics and their desirable properties for pertinent power
applications as well as the means of assessing the performance of materials
and equipment in the laboratory and in service. This book will certainly
enable the reader to understand high voltage insulation better and apply
that knowledge more effectively.
This book comprises 12 chapters. Each chapter deals with a specific
aspect of the subject and can be read as a self-contained unit with inclusive
references. In Chapter 1, the reader is introduced to the importance and
development of high voltage insulation technology through a brief descrip­
tion of important individual elements of modern power system networks
together with a general perspective of the important properties of insulating
materials. Chapter 2 is devoted to basic gas discharge mechanisms; Chapter
3 describes the air clearances necessary for transmission network and sub­
station equipment, plus the factors that conspicuously control these clear­
ances. No networks exist that do not employ sections of SF^ CIS. Chapter
4 describes the fundamental mechanisms that control breakdown in SF^
gas and offers a detailed account of SF^ circuit breakers and CIS techniques
generally encountered in modern power networks. Chapters 5 and 6 de­
scribe the importance and classification of liquid and solid dielectrics. We
deal with their electrical properties in detail and then cover their conduction
and breakdown aspects. Vacuum as a dielectric is described in Chapter 7,
whereas Chapter 8 is concerned with composite insulation. Polymeric and
oil-filled cables are an integral part of a power network, so Chapter 9 has
been devoted solely to the treatment of this topic. Chapter 10 contains a
iiscussion of the methods for generation and measurement of high test
oltages.
With the advent of digital/opto electronics and microprocessors, most
of the measurement techniques in almost every branch of technology have
been modified. This applies to the high voltage measurements and diag­
nostic techniques as well. In the past decade, a variety of such techniques
have been introduced and today many power stations are equipped with
this equipment. Until now, details of such methods were available only in
scattered form in the literature. Chapter 11 is fully devoted to a detailed
survey of these techniques.
It is very important, for design engineers as well as practicing engi­
neers, to test the ability of an insulation system or apparatus to meet the
guaranteed specifications. For this purpose, testing prior to and after in­
Preface vii

stallation is generally necessary. Chapter 12 deals with high voltage testing

techniques, including their internationally accepted classifications, test pro­
cedures and standards. Newcomers to this field as well as experienced
designers of equipment and practicing engineers will find this vital infor­
mation presented in a systematic way to make them aware of the latest
techniques in their chosen field.
We welcome constructive suggestions for future editions of this book.

N. H. Malik
A. A. Al-Arainy
M. /. Qureshi
Contents

Series Introduction H. L ee W illis III

Preface V

1 Introduction to Electrical Insulation in Power Systems

1.1 Introduction 1
1.2 Properties of Dielectrics 2
1.3 Classification of Insulating Materials 5
1.4 Applications of Insulating Materials 8
1.5 Electric Fields 10
1.6 Design Parameters of High Voltage Equipment 18
References 19

2 Gas Dielectrics 21
2.1 Introduction 21
2.2 Gas Behavior Under Zero Electric Field 21
2.3 Generation of Charged Particles 23
2.4 Deionization Processes 28
2.5 Uniform Field Gas Breakdown 29
2.6 Nonuniform Field Gas Breakdown 36
2.7 Time to Breakdown 42
2.8 Discharges Under Nanosecond Pulse Voltages 43
2.9 Gap-Type Discharge 44
2.10 Choice of Dielectric Gases 46
References 48

IX
 Contents

3 Air Insulation 49
3.1 Introduction 49
3.2 Air Insulation Applications and Modeling 49
3.3 Voltage Stresses 51
3.4 Impulse Breakdown Probability 51
3.5 Breakdown Voltage Characteristics 52
3.6 Volt Time Curve and Insulation Coordination 65
3.7 Phase to Phase Breakdown Characteristics 69
3.8 Arc Discharge 71
3.9 Undesirable Effects of Corona 72
3.10 Television Interference 80
References 80

4 SFg Insulation 83
4.1 Introduction 83
4.2 Basic Properties of SF^, Gas 83
4.3 Breakdown Processes in SFg 85
4.4 Uniform Field Breakdown 86
4.5 Nonuniform Field Breakdown 87
4.6 Estimation of Minimum Discharge Voltages 90
4.7 Factors Affecting Discharge Voltages 91
4.8 Arc Interruption in SF^^ 96
4.9 Gas Insulated Switchgear 98
4.10 Compressed Gas Insulated Cables 107
4.11 Other Applications of SF^ 108
4.12 SFft Gas Handling 108
References 109

5 Liquid Dielectrics 111

5.1 Introduction 111
5.2 Classification of Insulating Oils 112
5.3 Essential Characteristics of Insulating Oils 119
5.4 Streaming Electrification 123
5.5 Reconditioning of Insulating Oils 124
5.6 Electric Conduction in Insulating Liquids 125
5.7 Breakdown in Insulating Liquids 129
References 143

6 Solid Dielectrics 147

6.1 Introduction 147
e n ts XI

6.2 Solid Insulating Materials 148

6.3 Dielectric Loss in Solid Insulating Materials 167
6.4 Breakdown in Solid Insulation 173
References 186

Vacuum Dielectrics 188

7.1 Introduction 188
7.2 Prebreakdown Electron Emission in Vacuum 189
7.3 Eactors Affecting Breakdown Voltage in Vacuum 192
7.4 Breakdown Mechanisms 195
7.5 Arc Interruption in Vacuums 198
7.6 Vacuum Circuit Breaker 202
References 206

Composite Dielectrics 209

8.1 Introduction 209
8.2 Dielectric Properties of Composites 210
8.3 Edge Breakdown 213
8.4 Cavity Breakdown 214
8.5 Breakdown Due to Surface Erosion and Tracking 217
8.6 Chemical and Electrochemical Deterioration and
Breakdown 218
8.7 Materials of Outdoor Insulators 220
8.8 Oil-Impregnated Insulation 224
8.9 Elexible Laminates 234
References 238

High Voltage Cables 241

9.1 Introduction 241
9.2 Cable Materials 241
9.3 Types of Cables 244
9.4 Cable Constants 245
9.5 Electric Stress in Cables 249
9.6 Cable Losses 250
9.7 Cable Ampacity 251
9.8 Partial Discharges in Cables 253
9.9 Treeing in Cables 255
9.10 Cable Aging and Life Estimation 263
9.11 Cable Accessories 266
9.12 Cable Eault Location 270
9.13 Recent Advances in Cable Technology 272
Xll C on ten ts

References 273

Generation and Measurement of Testing Voltages 276

10.1 Introduction 276
10.2 High Voltage DC Generation 276
10.3 High Voltage AC Generation 279
10.4 High Voltage Impulse Generation 282
10.5 Nanosecond Pulse Generation 288
10.6 Spark Gaps as a Voltage Measuring Device 289
10.7 Potential Dividers for High Voltage Measurement 294
10.8 Other High Voltage Measuring Devices 295
10.9 Measurement of Corona and Gap Discharge Currents 302
References 304

New Measurement and Diagnostic Technologies 306

11.1 Introduction 306
11.2 Digital Impulse Recorders 306
11.3 Digital Techniques in HV Tests 311
11.4 Testing Automation 315
11.5 Electric Field Measurements 316
11.6 Electro-Optic Sensors 317
11.7 Magneto-Optic Sensors 325
11.8 Measurements of Very Fast Transients in GIS 328
11.9 Space Charge Measurement Techniques 329
11.10 Electro-Optical Imaging Techniques 335
References 339

Insulation Testing 342

12.1 Objectives of Testing 342
12.2 HV Test Classification 343
12.3 Test Voltages 345
12.4 Test Procedures and Standards 349
12.5 Testing of HV Measuring Devices 351
12.6 Partial Discharge Test 355
12.7 Dielectric Loss Test 358
12.8 Testing of HV Apparatus 361
12.9 Electrostatic Hazards 375
References 378

Indexc 381
Introduction to Electrical Insulation
in Power Systems

1.1 INTRODUCTION

The economic development and social welfare of any modern society de­
pends upon the availability of a cheap and reliable supply of electrical
energy. Extensive networks of electrical power installations have been built
in industrialized countries and are being constructed in developing coun­
tries at an ever-increasing rate. The major function of such power systems
is to generate, transport and distribute electrical energy over large geo­
graphical areas in an economical manner while ensuring a high degree of
reliability and quality of supply.
The transmission of large amounts of electrical power over long dis­
tances is best accomplished by using high voltage (HV), extra high voltage
(EHV) or ultra high voltage (UHV) power lines (see Table 1.1 for voltage
classification). Thus, high voltage equipment is the backbone of modern
power systems. Besides generation, transmission and distribution of elec­
trical energy, high voltages are also extensively used for many industrial,
scientific and engineering applications such as:
1. Electrostatic precipitators for the removal of dust from flue gases
2. Atomization of liquids, paint spraying and pesticide spraying
3. Ozone generation for water and sewage treatment
4. X-ray generators and particle accelerators
5. High-power lasers and ion beams
6. Plasma sources for semiconductor manufacture
7. Superconducting magnet coils

1
 Chapter 1

Table 1.1 Voltage Classifications (V = Rms, Line to

Line Voltage)

Voltage class Voltage range

Low voltage (LV) V < IkV

Medium high voltage (MHV) 1 kV < V < 70 kV
High voltage (HV) 110 kV < V < 230 kV
Extra high voltage (EHV) 275 kV < V < 800 kV
Ultra high voltage (UHV) 1000 kV < V

In all such applications, the insulation of the high voltage conductor is of

primary importance. For proper design and safe and reliable operation of
the insulation system, knowledge of the physical and chemical phenomena
which determine the dielectric properties of the insulating material is very
important. In addition, the basic processes which lead to degradation and
failure of such materials and appropriate diagnostic techniques are of prime
importance since any such failure can cause temporary or permanent dam­
age to the system, thereby influencing its reliability and cost. Considering
the high cost and comparatively long life span (20-40 years) of high volt­
age equipment, every effort should be made to select the most appropriate
materials as well as the design, installation and operational parameters for
such apparatus. This book attempts to provide an overview of different
aspects of electrical insulation as practiced mainly in high voltage power
systems.
This chapter outlines some basic definitions and fundamental concepts
which are essential for proper understanding of the electrical insulation
behavior. It further provides a brief overview of general categories of in­
sulating materials, a short description of their applications and some de­
sirable properties of various classes of dielectric materials. Subsequent
chapters provide the detailed properties, applications, failure modes and
diagnostic techniques used for evaluating and testing different insulating
materials.

1.2 PROPERTIES OF DIELECTRICS

There are several properties of a dielectric which are of practical impor­

tance for an engineer. The most important of these properties are briefly
defined here.
Introduction 3

1.2.1 DC Conductivity
DC conductivity, a, is defined as cr = J/E where J is current density (in
A/m^) resulting from the application of a direct electric stress E (in V/m).
It is related to the bulk resistivity p of the dielectric by cr == 1/p and is
calculated from measured values of the insulation resistance. Alternatively,
if p and geometry are known, insulation resistance can be calculated. In­
sulation resistance is used as an indication of conduction behavior of in­
sulating materials in many practical applications, such as in hi-pot testing.
For most insulating materials, a depends upon the material purity, its tem­
perature T and electric stress E. It generally increases as the ionic impu­
rities in the insulation system increase. Similarly a also tends to increase
with T and E in most cases following a relationship of the type
o-(T) - Ae- ( 1. 1)

where k is the Boltzman constant, A is a constant, and cr(T) is the value

of a at temperature T. In addition, due to polarization effects, a also de­
pends upon time of application of the stress. It influences the power losses
in a dielectric and controls the electric stress distribution under direct volt­
age applications.

1.2.2 Dielectric Permittivity

Dielectric permittivity, relative permittivity or dielectric constant, of an
insulating material is defined as = C/C^, where C is the capacitance
between two parallel plates having the space between them filled with the
insulating material under discussion, and is the capacitance for the same
parallel plates when these are separated by vacuum. Generally is not a
fixed parameter but depends upon temperature, frequency and molecular
structure of the insulating material.

1.2.3 Complex Permittivity, Loss Angle and Dissipation Factor

In order to determine the response of dielectrics to alternating voltages, it
is traditional to model the dielectric by a parallel RC network. Such a
network along with its phasor diagram is shown in Figure 1.1. Here R
represents the lossy part of the dielectric taking account of losses resulting
from electronic and ionic conductivity, dipole orientation and space charge
polarization, etc., and C is the capacitance in the presence of the dielectric
as defined earlier. If an AC voltage, v = V 2 V sin cot, is applied then the
capacitive component of current is 1^ = while the resistive compo­
nent of current is Ir = - j l tan 8. Since loss angle 3 is usually very small.
 Chapter 1

I.

(a)
Figure 1.1 (a) Parallel equivalent circuit of a dielectric material and (b) corre­
sponding phasor diagram.

Ic ~ I and If. = - jlc tan 8. Hence, total current I = 1^ + 1^ can be

expressed as:
I = ju;C,V(8, - js, tan 6) = jcuC.Vs* (1.2)
where e* = complex relative permittivity having a real part equal to the
dielectric constant, 8^, and an imaginary part equal to the loss factor,
8^ tan 8. The loss factor differentiates the losses in one dielectric material
from those in the other one. Tan 8 is commonly known as loss tangent or
dissipation factor or sometimes as power factor (cos 6) of the dielectric. It
usually depends upon frequency and may also be influenced by the applied
electric stress as well as the temperature. The power loss in the dielectric
is given as:
Power loss = coCY^ tan 8 = wC^Y^Sj. tan 8 (1.3)
Furthermore a and tan 8 are related as:
tan 8 = (1.4)
i08oSr
where 8^ = 8.85 X 10“*^ F/m is the permittivity of free space or vacuum.

1.2.4 Polarization
Unlike conductors where free electrons are easily available, most of the
electrons in insulating materials are bound and not free to move. Under
Introduction 5

the influence of an applied electric field, the resulting electrostatic forces

create some level of polarization forming dipoles. It is this electronic po­
larization which results in relative permittivity of more than 1 for most
dielectric materials. In some crystalline dielectric materials, relative dis­
placement occurs between positive and negative ions such as in Na^ and
Cl“ producing atomic polarization. In another mechanism which operates
in organic substances including many polymers, permanent molecular di­
poles are reoriented in electric field. The last type of mechanism to be
considered is interfacial polarization which is observed for heterogeneous
materials. In this case, mobile conduction charges are held up at some
boundary within the dielectric (see chapter 6, section 6.3 and chapter 8,
section 8.2.1 for more details). Electrolytic capacitors where charge is held
up at the electrode surface double layer is an example of this type of
polarization [1].

1.2.5 Dielectric Strength

This is defined as the maximum value of applied electric field at which a
dielectric material, stressed in a homogeneous field electrode system,
breaks down and loses its insulating property. It is given in V/m. The
breakdown strength of most dielectrics depends upon the purity of material,
time and method of voltage application, type of applied stress as well as
other experimental and environmental parameters. Although in many cases,
it may be difficult to assign a unique value of dielectric strength to a given
material, a range of values can be found and used for application purposes.
Although by definition dielectric strength refers to a uniform field system,
in many applications breakdown strength under inhomogeneous field con­
ditions needs to be defined and is sometimes referred to as the nonuniform
field dielectric strength.

1.3 CLASSIFICATION OF INSULATING MATERIALS

There are virtually hundreds of insulating materials which are used in the
electrical power industry. All such materials can broadly be classified into
different categories: gases, liquids, solids, vacuum and composites. Some
of the materials commonly used under each category and their desirable
properties are summarized next.

1.3.1 Gases
In normal state most gases are good insulators. Consequently, overhead
lines and open air circuit breakers using air insulation are in service since
6 Chapter 1

the early days of the electrical power industry. More recently, metalclad
switchgear and gas-insulated cables filled with compressed sulfur hexa­
fluoride gas, SF^, have made their entry into the electric power systems.
In special applications, as for instance in Van de Graaf accelerators or in
measurement capacitors, other gases or mixtures of SF^ with gases such
as N2 , O2 , CO2 , air, and N2 O are also used.
An ideal gaseous insulator should be cheap, chemically and thermally
stable, and should not form toxic, corrosive or flammable products under
prolonged electrical stress. It should have good heat transfer and arc
quenching properties, and exhibit low condensation temperature even when
pressurized. Most importantly, it should have high uniform and nonuniform
field dielectric strengths under DC, AC and impulse voltages and should
not suffer any loss of these values under prolonged use.

1.3.2 Vacuum
The absence of any residual gas in the interelectrode gap space results in
a vacuum medium which has excellent insulating and arc quenching prop­
erties. A true vacuum is very difficult to achieve and residual gas pressure
of the order of 10“^ to 10“’^ bar may exist in vacuum insulated equipment.
In such equipment, material, shape and surface finish of electrodes, residual
gas pressure and contaminating particles are important factors. Vacuum
insulated medium voltage switches and circuit breakers are being used
more frequently.

1.3.3 Liquids
An ideal liquid insulant should have high values of dielectric strength,
volume resistivity, specific heat, thermal conductivity and flash point plus
low values of loss factor, viscosity, pour point and density. Furthermore,
the liquid should be noncorrosive, nonflammable, nontoxic and chemically
stable having good arc quenching as well as gas-absorbing properties [2].
No single liquid is available which possess all of these properties, and
compromises usually have to be made. Mineral oil having alkanes, cyclo­
alkanes and aromatics as the main constituents has been used since the last
century. Another class of liquids which have been used for transformers
and capacitors are chlorinated aromatics or askrals, also called PCBs. They
exhibit excellent flame resistance and very good electrical properties and
are derived from benzene or from biphenyl. In the 1970s, it was found that
these compounds exhibit a considerable health hazard and most countries
have now legally banned the production and use of these liquids. Many,
Introduction 7

new liquids have since been developed which are without adverse ecolog­
ical effects and also possess properties which are comparable with those
of PCBs. These include silicone oil, synthetic hydrocarbons and fluorinated
hydrocarbons.

1.3.4 Solids
An ideal solid dielectric must have some of the properties mentioned earlier
for gases or liquids. In addition, it should have good mechanical and bond­
ing properties. Inorganic as well as organic solid insulating materials are
widely used in electrical power components. The most important inorganic
materials are ceramic and glasses which are used to manufacture insulators,
bushings and other high voltage components. The most prominent organic
materials are thermosetting epoxy resins or thermoplastic materials such
as polyvinylchloride (PVC), polyethylene (PE) or cross linked polyethylene
(XLPE). Thermoplastic materials are mainly used for manufacture of ex­
truded dielectric power cables. Kraft paper, natural rubber, ethylene poly­
propylene rubber (EPR), silicon rubber and polypropylene are some of the
other solid dielectric materials which are widely used.

1.3.5 Composites
In many engineering applications, more than one class of insulating ma­
terials are used together, giving rise to a composite or a hybrid type of
insulation system. Examples of such systems employing solid/gas insula­
tion are transmission line insulators and solid spacers used in gas insulated
switchgear (GIS). In solid/gas composites, the solid/gas interface usually
represents the weakest link and has to be carefully designed. Similarly, in
vacuum insulated systems, the interface of solid insulating spacer and vac­
uum proves to be a weak link. Examples of solid/liquid composite insu­
lations are oil impregnated paper tapes used in high voltage cables, trans­
formers, capacitors and bushings. Similarly oil impregnated, metallized
plastic films used in power capacitors also belong to this category.
In the applications of composites, it is important to ensure that both
components of the composite should be chemically stable and not react
with each other under combined thermal, mechanical and electrical stresses
over the expected life of the equipment and should have nearly equal di­
electric constants. Furthermore, the liquid insulant should not absorb any
impurities from the solid which may adversely affect its resistivity, dielec­
tric strength, loss factor and other properties.
 Chapter 1

1,4 APPLICATIONS OF INSULATING MATERIALS

An electric power system has many high voltage components such as gen­
erators, transformers, circuit breakers, cables, bushings, overhead lines,
surge arresters, GIS, capacitors, protective gaps and rotating machines. All
such components need proper insulation. There are four principal areas
where insulation must be applied [3]. They are (1) between coils and earth
(phase to earth insulation), (2) between coils of different phases (phase to
phase insulation), (3) between turns in a coil (inter-tum insulation) and (4)
between coils of the same phase (inter-coil insulation). A brief introduction
to the use of insulating materials in major high voltage system components
is provided next.

1.4.1 Transformers
Present high voltage power transformers use enameled conductors, paper,
glass or thermoplastic insulating tape, pressboard, glass fabric, porcelain
and mineral or silicone oil. The windings are insulated by tape, held in
place over the iron core by pieces of pressboard, glass fabric or porcelain,
and impregnated with an insulating fluid which also acts as the cooling
medium. Various designs of windings and oil cooling medium are em­
ployed [4,5]. In small power transformers as well as in current and voltage
measurement transformers, the insulating materials used are thermosetting
resins, insulating tapes, SF^ gas, etc. In such cases, pressurized SF^ gas
provides insulating as well as cooling functions. Fire resistant transformers
use insulating fluids such as high flash point mineral oil, chlorofluorocar-
bons or perchloroethylene.

1.4.2 Circuit Breakers

High voltage circuit breakers use ceramics, epoxy resins, epoxy resin
bonded glass fiber, polyester resins, vulcanized fiber, synthetic resin
bonded paper, SF^ gas, air, vacuum and mineral oil. The air, oil, SF^ gas
or vacuum serves as the main insulation and arc quenching medium
whereas ceramic or epoxy resin parts are used for mechanical support, bus
bar insulation and arc chamber segments, etc. In low voltage breakers,
synthetic resin moldings are used to carry the metallic parts.

1.4.3 Power Cables

The insulating materials used in power cables are paper or plastic tape,
thermoplastic materials (such as PE, XLPE or PVC), silicon rubber, EPR,
Introduction 9

thermosetting resins, SF^ gas and mineral oil. In oil filled cables, the inner
conductor is insulated by lapped paper tape and impregnated with mineral
oil. In polymeric insulated cables, the conductor and the insulating mate­
rials are extruded jointly and then insulation is cured and crosslinked. In
gas-insulated cables, the inner conductor is held concentrically in a metallic
tube by insulating spacers made of thermosetting resins and the tube is
filled with pressurized SF^ gas. Low voltage cables employing PVC, PE
or XLPE insulation are normally without the other screen.

1.4.4 Bushings
Bushings are made of porcelain, glass, thermosetting cast resin, air, SF^
gas, paper tape and oil, etc., and are constructed such that the feed-through
conductor is insulated by paper tape and oil and is housed in a porcelain
tube that enters the enclosure of the high voltage equipment. Two types of
construction are normally used resulting in noncondenser and condenser
graded bushings. Condenser graded bushings are used for rated voltages
of over 50 kV, whereas noncondenser bushings are preferred for lower
voltage applications. The paper tape used in bushings is usually resin
bonded paper, oil impregnated paper or resin impregnated paper.

1.4.5 Overhead Lines

Overhead power lines use porcelain, glass, thermosetting resin, and air as
the main insulation where the conductors are suspended via insulator
chains from towers. Insulators are made out of porcelain or hard glass.
Plastic insulator chains employing fiber glass and cast resins are also being
used more in recent years. Room-temperature vulcanized rubber (RTV)
coating is also being employed to improve the ceramic insulator’s perform­
ance in polluted environments.

1.4.6 Gas Insulated Switchgear

Gas insulated switchgear (or GIS) use SF^ gas, thermosetting resins and
porcelain as the main insulating materials. In GIS construction, different
components such as bus bars, interrupters and earthing switches are located
in adjacent cylindrical compartments which are air tight, sealed and contain
compressed SF^ gas as the insulation medium. The inner live conductors
are separated, at regular intervals, from the grounded enclosure by insu­
lating spacers made of epoxy resins.
10 Chapter 1

1.4.7 Surge Arrester and Protective Gaps

Lightning or surge arresters and protective gaps consisting mainly of air,
SF^, porcelain and metal oxide resistors are used to limit the transient
overvoltages caused by lightning or switching actions in high voltage sys­
tems. In the simplest form, air insulated rod-rod chopping gaps are used
across bushings, insulator chains, cable terminations or live conductors,
etc. Alternatively, nonlinear resistors made of metal oxide (such as zinc
oxide) with or without series spark gaps are used. In most recent designs,
metal oxide surge arresters (MOAs) are used without any series spark gap.
Ceramic or porcelain housing is used for mechanical support and for pro­
tection from the exposure to the environment.

1.4.8 Power Capacitors

Modem power capacitors consist of metallized polypropylene film, alu­
minum foil and polypropylene film, or metallized paper electrodes and
polypropylene or other film and the impregnation fluid. The metallized foils
represent the capacitance, the fluid minimizes the voids, increases the di­
electric strength and sometimes the capacitance via its dielectric constant.
The fluids used in recent years are isopropylbiphenyl, phenylxylylethane
and silicone liquid imprégnant. The whole assembly is housed inside a
container with appropriate terminals or bushings for external connections.

1.4.9 Rotating Machines

Generators use mica tape system on conductors impregnated with either
an epoxy or polyester resin. Other insulation materials normally used in
rotating machines are polyvinyl acetal, polyester enamel or bonded fiber
glass for inter-turn insulation; bakelized fabric, epoxy fiber glass, mica
glass sheet, epoxy impregnated mica paper and varnished glass for inter­
coil or phase-to-earth insulation; and bakelized fabric or epoxy fiber glass
strips for slot closure. The impregnation treatment normally consists of
alkyd phenolic estermide or epoxy based varnishes.

1.5 ELECTRIC FIELDS

Proper design of any high voltage device requires a complete knowledge

of the electric field distribution and methods to control this field. Further­
more, for an understanding of the insulation failure modes, some knowl­
edge of the electric field concepts is a prerequisite. A very brief introduc­
Introduction 11

tion to this subject is provided here; more details can be found in the
references cited.
The electric field intensity E at any location in an electrostatic field is
related to force F experienced by a charge q as F = qE. Moreover the
electric flux density D associated with E is given as D == eE where c =
is the absolute permittivity of the medium in which the electric field
exists. If the medium is free of any space charge, the electric field is
obtained from the solution of the Laplace equation:
- 0 (1.5)
where the operator is called the laplacian and is the potential which
is related to E and path i through which the charge is moved by

-I E • di ( 1.6)

If the field medium has a space charge of density p, then the field is
governed by the solution of the Poisson’s equation:

V" (/> = — (1.7)

£
When the medium under discussion is gaseous or vacuum, s =

1.5.1 Field Distribution Types

Broadly speaking, the field distribution in a region may be classified as
homogeneous (or uniform) and nonhomogeneous (or nonuniform). In a
homogeneous field, E is the same throughout the field region; whereas in
a nonhomogeneous field, E is different at different points in the region. In
the absence of space charges, the electric stress, E, in a nonuniform field
gap usually obtains the maximum value at the surface of the conductor
which has the smallest radius of curvature and achieves the minimum value
at the conductor having the largest radius of curvature or the earth. In this
case, the field is nonhomogeneous as well as asymmetrical. Most of the
practical HV components used in electric power systems have nonhomo­
geneous and asymmetrical field distributions. Uniform or approximate uni­
form field distributions exist between two infinite parallel plates or two
spheres of equal diameters with a gap spacing which is smaller than the
sphere radius. Sphere electrodes are frequently used for high voltage mea­
surements and in impulse voltage generation circuits. Similarly “profiled”
parallel plates of finite sizes are also used to simulate homogeneous fields.
In some gaps, which produce nonhomogeneous fields, the field along the
12 Chapter 1

gap axis may be symmetrical towards both electrodes with respect to the
gap center. Examples of such nonuniform symmetrical fields are fields
produced by similar diameter rod-rod or sphere-sphere gaps (with large
distance between the spheres). When one of the electrodes of such sym­
metrical field gaps is earthed, the field symmetry may be disturbed due to
the earthing effect. Consequently, the high voltage electrode has somewhat
higher electric stress than the grounded electrode.

1.5.2 Methods of Field Estimation

In simple physical systems such as a single conductor above ground, two
parallel conductors above ground, two equal diameter spheres, two infi­
nitely long parallel plates, coaxial cylinders, and concentric spheres, it is
possible to find an analytical field solution. However, for most high voltage
components, the physical systems are so complex that it is extremely dif­
ficult to find an analytical field solution. In such cases, numerical methods
are employed for electric field calculations. The existing methods include
the finite difference method, the finite element method, the Monte Carlo
method, the moment method, the method of images, the charge simulation
method, the surface charge simulation method and combinations of these
methods. Details of some of these methods have been discussed in the
literature [5-10] and will not be repeated here. For complex field problems,
experimental modelling using electrolytic tank, semiconducting paper or
resistive mesh analog are also useful. Computer software packages are now
available to carry out most of the field calculations.

1.5.3 Field Enhancement Factor

Whereas any designer of the high voltage apparatus must have a complete
knowledge of the electric field distribution, for a user of the system the
knowledge of the maximum value of the electric field to which the
insulation is likely to be subjected and the location of such a maximum
gradient point is generally sufficient. Consequently, the concept of field
enhancement factor or simply field factor f is of considerable use. This
factor is defined as:

^ _
F max
( 1. 8)

where E^^ is the average field in the gap and is equal to the applied potential
difference divided by the gap separation between the electrodes. Values of
f for most geometries of practical interest to a power system engineer are
Introduction 13

summarized in Table 1.2. These equations can either be derived analytically

or are empirical equations derived from numerically computed stress values
for a single homogenous dielectric medium [1,5,6,11,12,15-17]. From
these equations one can easily estimate the maximum stress to which
a given dielectric material may be subjected when a voltage V is applied

Table 1.2 Field Enhancement Factor for Some Common Configurations

Configuration f value Comment

(a) Parallel plates 1 Uniform field

(b) Concentric spheres R /r R = outer sphere radius
(c) Coaxial cylinders (R - r)/(r In R/r) R = outer cylinder
radius
(d) Hyperbolic point- 2 x /ln (l + 4x)
plane
(e) Equal diameter 0.25[B + VB^ + 8] Symmetrical voltages
spheres and B = X + 1
(f) Equal diameter 0.943 0.458X + X < 2 and one sphere
spheres 0.121x2 earthed
(g) Sphere-plane X + 0.55 X > 2 and plane earthed
(h) Equal diameter . X A A = Vx2 + 4x
A / 2 In
parallel cylinders
(i) Conductor-plane 2 + 0.32x X< 5
2.642 + 0.216X - 5 < X < 2000
0.0002x2
(j) Cylinder surrounded 0.433 - f - 0.307 (R/r) -H R = torus radius
by a torus 0.0095 (R /r)2 r = cylinder and torus
cross-sectional radius
(k) Hemispherical rod- 0.85 (1 + x) X< 3
plane 0.45x In (6x) 3 < X < 500
fiToo
(l) Bipolar DC line with d = distance between
- + i |/ln (4hVrD)
2 subconductors r D/ bundle centers
(m) 3-phase line with 2 d = distance between
d/ 2r In
subconductors V rD adjacent phases
(n) Two hyperboloidal x V l + 2/x
points
tan h
IX + 2
d = gap length, r = high voltage eleetrode radius, x d/r, h = height above ground, and
D = bundle diameter.
14 Chapter 1

across the two conductors with a gap spacing of d since = f*V/d.

Instead of field enhancement factor f, sometimes the field utilization factor,
/Xf = 1/f, is used, which gives more insight about the effective use of the
dielectric space since a larger value of represents a more compact
equipment.
In a multidielectric media, the field computations become complicated,
since in addition to the Laplace or Poisson equations, the boundary con­
ditions (see section 1.5.6) must also be satisfied at the interface of the two
dielectrics. Therefore, in such cases, except for some very simple config­
urations, numerical computations are essential. General values of f for such
cases, therefore, cannot be given.

1.5.4 Field Intensification at Protrusions

Protrusions at the electrode surfaces are created during manufacture, in­
stallation or operation of many high voltage components. At the tip of such
a protrusion, the field lines converge. Consequently, the microscope field
at this tip, Ep = fE, becomes greater than the macroscopic gap field E.
The field enhancement factor f in this case depends upon the shape and
size of the protrusion. Equations for stress concentration have been derived
for hyperbolic, spheroidal and ellipsoidal points. The stress enhancement
factor at a conducting sphere surface in a uniform field (either an isolated
sphere or a spherical boss on a conducting plane surface) is around 3. For
a conducting ellipsoidal boss on a conducting plane or an isolated con­
ducting ellipsoid in a uniform field f is given by [11]:
2n^
f - (1.9)
m + n
m In - 2n
m
where m is the ratio of the major to minor axis of the ellipsoid and n =
V(m^ - 1). For a hyperbolic point to plane geometry, factor f is as given
in Table 1.2. For needle-like geometries which have A > 10, f can be
approximated by [12]:

f - ( 1. 10)
In A - 0.3
where A = protrusion height/protrusion base radius. For some other ge­
ometries, the value of f depends upon the protrusion shape, its base height
(h) as well as its radius of curvature or its base radius (r). As a limiting
case, if h = r, i.e., a hemispherical protrusion, the value of f is approxi­
mately 3. The protrusions and surface defects play a prominent role in the
Introduction 15

initiation of partial discharges and ultimate breakdown of air and SF^ in­
sulation, polymeric insulated cables and vacuum insulated equipment, etc.
Figure 1.2 shows values of f as a function of protrusion parameters for a
number of protrusion configurations [7,20].

1.5.5 Field at the Interface of Composites

At the interface of two different dielectric materials (A and B) having
dielectric constants of and , the tangential electric field and the normal
flux density must be continuous. Therefore, in reference to Figure 1.3, the
following conditions must be satisfied:

-^At ( 1. 11)

E ^n ~ E en ( 1. 12)

If there is a surface charge at the interface, equation (1.12) is modified as

follows:

7
■^ ;a
Spheroid Cone
1
Sphere Cylinder
CATHODE

(b)

Figure 1.2 Typical protrusion geometries are shown in (a) and the resultant field
enhancement factors for various values of h/r are shown in (b). Curve A is for a
sphere or for a cylinder. Curve B is for a cone. Curve C is for a spheroid. (From
Ref. 20.)
16 Chapter 1

0^]
Boundary
-Bn

^Bn Ps (1.13)
where is the surface charge density. It is important to note that for
spacers and insulators used in air or SF^ (A = spacer, B = air or SF^),
is typically 2 to 4 whereas = 1. Therefore, application of equations
(1.12) and (1.13) show that the field is distorted at the solid/gas interface
and the net field on the air side of the interface becomes larger, which
makes such an interface the weakest link in the system. For DC voltage
applications, the insulating materials can get charged. Similarly, there may
be charging of insulator due to corona or other types of discharges giving
rise to a surface charge density which can enhance the total surface field
on the gas side. Similar arguments apply for voids or gas cavities inside
solid or liquid insulating materials as discussed next. Therefore, each of
such solid/gas, solid/vacuum, solid/liquid or solid/solid interfaces needs
careful consideration.

1.5.6 Field Inside Cavities

Cavities can be generated inside solid or liquid dielectrics during manu­
facture or operation. Such cavities are usually spaces of lower density
material or a gas pocket in a surrounding solid or liquid dielectric. Such
gas spaces experience higher electric stress than the bulk liquid or solid
media as a result of their lower dielectric constant. A cavity whose diameter
perpendicular to the electric field is appreciably larger than its depth in the
direction of the electric field, and when its depth is small compared to the
total insulation spacing, experiences an electric field under AC voltages
of:
Introduction 17

E - - E, (1.14)

where and are the relative permittivities of the dielectric and the cavity
respectively, and E^ is the electric stress in the dielectric medium. For a
gas cavity whose diameter is quite small in proportion to its depth in the
electric field direction, the electric field in the cavity approaches nearly
equal to that in the surrounding medium, i.e., E^ = E^. For a spherical
shaped gas cavity having cavity depth equal to cavity diameter, E^ is given
as:
38h
E. = (1.15)

The electric stress in cylindrical cavities of various depths and diameters

is given in [13]. For prolate and oblate spheroidal cavities, the exact electric
stress in the cavities can be calculated [14].
The stress distribution under DC voltages will be dependent on time.
As the voltage is raised, the initial distribution is determined as for AC
voltages. However, it subsequently assumes a distribution dependent on the
surface and volume resistivities of the cavity and the volume resistivity of
the dielectric in series with the cavity. Consequently, the mean stress in
the cavity under DC conditions can be determined from the calculations
described above for AC voltages and by considering the relative permittiv­
ity of the solid dielectric to be very large, i.e., tends to oo [15]. Hence
generally large stress values occur inside cavities under DC voltages than
under AC applied voltages. Insulation is sometimes subjected to both AC
and DC voltages simultaneously, or the DC voltage can contain an AC
ripple component. In such a case, the cavity stress could be estimated by
superposition principle. For fast impulse voltages, the stress distribution in
cavities can be found by using the AC method discussed above.

1.5.7 Fields at Free Particles

In liquids and gaseous as well as vacuum insulation systems, free con­
ducting or insulating particles may be introduced during manufacture, in­
stallation or operation. These particles can acquire some charge as a result
of various mechanisms and may drift in the insulating medium. The electric
stress at the ends of such particles can be enhanced, and consequently these
particles can trigger breakdown of the insulation medium. The field en­
hancement factor for cylindrical metallic particles of radius r and length a
is given as [18]:
18 Chapter I

2 + 1 - (1.16)

where d = gap separation. For free spherical metallic particles, the field
factors are given in Figure 1.2. In the presence of ionic or electronic space
charges and for particles of significant resistivity, the field factors are sig­
nificantly influenced by the experimental parameters.

1.5.8 Electric Stress Control

Electric stress values have to be controlled in the design of high voltage
equipment since a higher value of electric stress may trigger or accelerate
the degradation and failure of the insulation. Thus, electrical stresses are
controlled in cable terminations, high voltage bushings, potential trans­
formers, etc. Methods have been suggested and employed with the aim of
optimizing the stresses throughout different parts of the high voltage equip­
ment in order to arrive at the most economical design. In addition, corona-
free connections are desired during the testing of some high voltage equip­
ment, thereby limiting the stress to values below critical values. Thus
electrical stress control method are an important area and its details are
available elsewhere [3,5,6,8,10,11].

1.6 DESIGN PARAMETERS OF HIGH VOLTAGE EQUIPMENT

Insulating materials play a critical role in the design and performance of

high voltage power equipment. These materials must not only meet the
dielectric requirements, but must also meet all other performance specifi­
cations including mechanical and thermal requirements, reliability, cost,
ease of manufacture as well as environmental concerns. Table 1.3 sum­
marizes the typical operating electric stress values of various types of
power equipment and compares the mechanical forces experienced in op­
eration, the relative complexity of the insulation and relative automation
used in manufacture [191. It is interesting to note that the highest operating
electrical stresses are for capacitors where there is the simplest insulation
configuration employing nearly uniform field distribution, highest manu­
facturing automation and where mechanical forces are fairly low. The de­
sign electrical stresses values given in Table 1.3 are determined by equip­
ment test or basic impulse level (BIL) requirements. Generally, for
economic reasons, the operating electric stress increases with system volt­
age whereas the design electric stress decreases with system voltage. Thus,
Introduction 19

Table 1,3 Relative Dielectric/Design Parameters of Typical High Voltage Power

Equipment

Operating Design Mechanical Insulation Manufacture

stress stress force complexity automation
Equipment (kV,„,/cm) (kVpeak/cm) (relative) (relative) (relative)

Generators 25 130 1.0 1.0 0.2

Transformers 15 115 1.0 0.9 0.3
SF^ equipment 40 180 0.2 0.3 < 0.1
Capacitors 600-1000 2000-3000 0.1 0.2 0.8

Source'. Ref. 19.

as the system voltage increases, the equipment operates closer to its limits
and presence of any defects can have serious consequences for the equip­
ment’s life expectancy.

REFERENCES

1. A. Bradwell (ed.), E le c t r i c a l in s u la tio n , Peter Peregrinus Ltd., London, Eng­

land, 1983.
A.C. Wilson, I n s u la tin g L iq u id s — T h e ir U ses, M a n u fa c tu r e a n d P r o p e r tie s ,
Peter Peregrinus Ltd., London, England, 1980.
3. M.S. Naidu and V. Kamaraju, H ig h V o lta g e E n g in e e r in g , Tata McGraw-Hill,
New Delhi, India, 1982.
4. A. White, “Design of High Voltage Power Transformers” in H ig h V o lta g e
E n g in e e r in g a n d T e s tin g by H.M. Ray an (ed.), Peter Peregrinus Ltd., London,
England, 1994.
5. T.J. Gallagher and A.J. Pearmain, H ig h V o lta g e M e a s u r e m e n t, T e stin g a n d
D e s ig n , John Wiley, New York, 1983.
6. M. Khalifa (ed.). H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c tic e , Marcel
Dekker, Inc., New York, 1990.
J.M. Meek and J.D. Graggs (eds.). E l e c t r i c a l B r e a k d o w n in G a s e s , John Wi­
ley, New York, 1978.
8. E. Kuffel and W.S. Zaengl, H ig h V o lta g e E n g in e e r in g F u n d a m e n ta ls , Perga-
mon Press, New York, 1984.
9. M. Chari and P. Silvestor (eds.). F in ite E le m e n ts in E l e c t r i c a l a n d M a g n e tic
F ie ld P r o b le m s , John Wiley, New York, 1980.
10. N. H. Malik, IEEE Trans, on Elect. Insul., Vol. 24, No. 1, pp. 3-20, 1989.
11. H. Bateman, P a r tia l D if f e r e n tia l E q u a tio n s o f M a t h e m a tic a l P h y s ic s , Cam­
bridge University Press, New York, 1944.
20 Chapter 1

12. R.V. Latham, H ig h V o lta g e V acu u m I n s u la tio n : T h e P h y s i c a l B a s is , Academic

Press, San Diego, 1981.
13. H.C. Hall and R.M. Russek, lEE Proc., Vol. 101, p. 47, 1954.
14. C.J. Böttcher, T h e o r y o f E l e c tr ic P o la r iz a tio n , Elsevier Publishing Co., Am­
sterdam, The Netherlands, pp. 52-54, 1952.
15. R. Bastnikas (ed.). E n g in e e r in g D i e le c t r i c s — V o l 1: C o r o n a M e a s u r e m e n t a n d
I n te r p r e ta tio n , ASTM Press, Philadelphia, 1979.
16. Y. Qiu, IEEE Trans, on Elect. Insul., Vol. 21, No. 4, pp. 673-675, 1986.
17. L.L. Alston (ed.). H ig h V o lta g e T e c h n o lo g y , Oxford University Press, New
York, 1968.
18. H.C. Miller, J. of Appl. Phys., Vol. 38, No. 11, pp. 4501-4504, 1967.
19. A.H. Cookson, IEEE Elect. Insul. Magazine, Vol. 6, No. 6, pp. 7-1 0 , 1990.
20. PA. Chatterton, Proc. Physical Society London, Vol. 88, pp. 231, 1966.
Gas Dielectrics

2.1 INTRODUCTION

Gases are the simplest and the most widely used dielectrics. In order to
utilize these dielectrics efficiently, it is necessary to know their electrical
behavior, especially the physical processes which lead to ionization and
breakdown under different practical electrode systems. Each gas will break­
down at a certain electric stress. The breakdown voltage is defined as the
peak value of the applied voltage at the instant of a spark discharge. This
chapter highlights the basic mechanisms of gas breakdown in order to
provide the base for understanding its engineering implications.

2.2 GAS BEHAVIOR UNDER ZERO ELECTRIC FIELD

In the absence of an electric field, a gas obeys Boyle’s law, i.e.

Pv = C (2.1)
where P and v are gas pressure and volume, respectively, and C is a con­
stant which depends on the absolute temperature T and the mass m. How­
ever, V varies with T according to Gay Lussac’s law, i.e.
V
z ( 2 . 2)
V. T

21
22 Chapter 2

where and are the initial values of volume and temperature, respec­
tively. From equations (2.1) and (2.2) it follows that
Pv = n.RT (2.3)
where n^ is the number of kilomoles of gas and R is the universal gas
constant, i.e., 8314 J/°K. Assume Nq is the number of gas molecules in a
mole (Avogadro’s number) = 6.02 X 10^^ molecules/mole, and N' is the
number of total molecules in the gas which is equal to Nv, where N is the
gas number density. If n^ in equation (2.3) is replaced by N 7 Nq, and N'
by Nv, it can be written as:
N' Nv
(2.4)

Thus,
P - NkT (2.5)
R/N^ = k = Boltzmann constant = 1.3806 X 10“^^ K.
w h e re
Using classical mechanics, with some assumptions, the kinetic energy
of a gas can be related to its thermal energy by:

- mi r = - kT ( 2 . 6)
2 2
where v is the molecular velocity. In the absence of an electric field, the
gas molecules will have random velocities ranging from zero to infinity.
Maxwell derived the distribution function of molecular velocities and
proved that this function is unique for a fixed gas temperature. Figure 2.1
shows this function, which can be expressed as [ 1 ]:

f(i^r) = exp(-i^) (2.7)

7T

where the relative velocity is given as = v/v^ and is the most

probable velocity.
The average velocity v and the effective or rms velocity can be
related to as [ 1 ]:
- 1.128 ( 2. 8)

= 1.224 Vp (2.9)
Since the gas molecules move randomly they will collide with each other
and with the walls of the container. The distance a particle traverses be-
Gas Dielectrics 23

M olecular relative velocity ^Vr

Figure 2.1 Maxwell distribution function for molecular velocities. (From Ref. 1
© Wiley, 1971.)

tween two successive collisions is called its free path (A). ObviouslyJ'ree
path values differ greatly, and hence the concept of mean free path (A) is
more practical. If there are N pairs of particles of radii and r 2 then:
1 1
( 2 . 10)
7T(r, + r^)^ N, 8
where 8 is the effective collision cross-section. For a certain gas, A is
directly proportional to gas temperature and inversely proportional to gas
pressure. Table 2.1 shows values of A for some selected gases.

2.3 GENERATION OF CHARGED PARTICLES

If a voltage is applied across two electrodes separated by an insulating gas,

the conduction current will flow only if there are charged particles in the
gas media. Such particles may include:
1. An electron e (negative charge)
2. A positive ion (neutral atom missing one electron) A^ = A - e
3. A negative ion (neutral atom with one excess electron) A“ = A
+ e
24 Chapter 2

Table 2.1 Values of the Mean Free Path A of Some Gas Molecules Calculated
from the Kinetic Theory of Gases at T = 288 K and P = 1013 mb

Diameter
Gas Molecular weight A (10-^ m) m)
(10-^ ^

2.016 11.77 2.74

He 4.002 18.62 2.18
H2 O 18.000 4.18 4.60
Ne 20.180 13.22 2.59
N, 28.020 6.28 3.75
O, 32.000 6.78 3.61
Ar 39.940 6.66 3.64
CO2 44.000 4.19 4.59
Kr 82.900 5.12 4.16
Xe 130.200 3.76 4.85

These particles can be generated through various processes. Figure 2.2

shows the main processes which result in the generation of such particles
in a gas discharge. Before discussing some of these processes, the Bohr’s
theories concerning the atomic structure and the energy levels are briefly
summarized as follows.
1. The electrons can exist only in discrete stable orbits around the
nucleus without radiating any energy. These stable orbits are located at a
distance r from the center of the nucleus, where:
qh
r = ( 2 . 11)

where q = quantum number (an integer), h = Plancks constant = 6.6257

X 10“^"^ J-s, = electron velocity, and m^ = electron mass.
2. When the energy of an atom changes from a value Wj to a lower
value W2 , the excess energy is emitted as a quantum of radiation (photon)
whose frequency (fp) is related to Plancks constant by:
hfp = - W2 ( 2 . 12)

Generation of free electrons can result from ionization of neutral atoms or

from detachment of negative ions. The ionization process needs a specified
minimum amount of energy. If the energy absorbed by an atom is below
this specified amount, it may lead to excitation, where the electron will
not leave the atom but go to a higher energy level (or an outer orbit).
Normally, excited states are not stable and the atom may absorb more
energy to become ionized or it may go back to its original stable state by
Gas Dielectrics 25

c c co
o co
0
C co
c oZ3 00
0
E •D COw ,N
‘c C O CO
SI O N .W o N N
*c
Ü C
O E0 I E 'cg o
CO
tü To 'o
< 'O :g I ^c E Ü
(0
CL O o Ü o o 0
C 0 0 Q.
O f ii- LU E

Figure 2.2 The main processes responsible for the production of the charged
particles in a gas discharge.

radiating the excess energy. A brief description of the main ionization

processes is given below.

2.3.1 Ionization by Collision (Impact Ionization)

When an electric field is applied between two electrodes and a free electron
is present in the gap, the electron gains energy as it travels in the field
towards the anode. While it traverses, it also collides with neutral gas
molecules. If the electron energy is less than the ionization energy of the
gas atom or molecule, an elastic collision may result. Alternatively, the gas
atom may be excited. However, when the electron energy is larger than
the ionization energy of the gas, an inelastic collision may occur, causing
the ionization of the gas atom or molecule. In such inelastic collisions,
each impact produces a positive ion and an extra electron. The positive ion
is attracted to the cathode. Now these two electrons will gain energy from
the field and may ionize further gas molecules. Depending upon the phys­
26 Chapter 2

ical conditions, this process will either reach a stable condition, where a
certain number of charged particles are generated and some conduction
current flows, or the ionization process progresses to avalanche and then
breakdown (see section 2.5). The following equations outline the ionization
progression by this process:
A + e —►A'^ + e + e or
A + le —►2e + A”^
2A + 2e —>■4e + 2A"^
4A + 4e —►8 e + 4A^ and so on (2.13)

2.3.2 Photoionization (Ionization by Radiation)

An excited atom generally has a lifetime in the range of 10~^ to 10“^ s.
When it returns to its ground state, it emits extra energy as a photon. Such
low-energy photons may lead to ionization as follows:
A* ^ A + hfp
B -h hf + (2.14)
Here A* represents the excited state of an atom A and hfp is the photon
energy which is assumed to be more than that the ionization energy of
atom B. External sources such as x-rays, nuclear radiation or cosmic rays
may cause photoionization from much deeper energy levels within a mol­
ecule. Photoionization is an important process, especially in the breakdown
of gas mixtures composed of the rare gases whose excited states can have
long lifetimes.

2.3.3 Thermal Ionization

A sufficient increase in gas temperature will cause its particles to move
faster, and consequently may cause ionization on collision between gas
atoms or molecules. Thermal energy (W^) can cause ionization by itself at
high temperature like in flames and arcs as follows:
A -h W, ^ A+ + e (2.15)

2.3.4 Electron Detachment

An electron may detach from negative ion as follows:
A—A + e (2.16)
Although the number of charged particles do not increase in this process.
Gas Dielectrics 27

detachment can be considered as an ionization process since in this process

the slowly moving negative ions are converted to fast-moving electrons.
Since the kinetic energy is directly proportional to mass and to the square
of the speed, the lighter but fast-moving electrons will have much higher
kinetic energy compared to the heavier but slow-moving negative ions.
Therefore electrons can cause further ionization more effectively than the
negative ions.

2.3.5 Cathode Processes

The charged particles can also be supplied from the electrodes especially
the cathode. At normal state, the electrons are bound to the solid electrode
by electrostatic forces between electrons and ions in the lattice. For the
electron to leave the cathode, a minimum specified energy, known as work
function, is required where its value depends on the material. The source
of the energy required can be one or combination of the following.

Positive Ion and Excited Atom Bombardment

When a positive ion has an impact on the cathode, an electron is released
provided the impact energy is equal to or more than twice the cathode
work function. At least two electrons will be released; one will neutralize
the positive ion and the other will be ejected to the gas medium. An elec­
tron might also be emitted as a result of the bombardment of cathode by
neutral excited atoms or molecules.

Photoemission
If the energy of a photon striking the cathode surface is higher than the
cathode work function, an electron may be ejected from the cathode.

Thermionic Emission
Raising the cathode temperature to a very high value (around 2000 K) will
lead to some electrons leaving its surface since the violent thermal lattice
vibrations will provide the electrons with the required energy. The therm­
ionic emission process has been widely used since the early days of
electronics.

Field Emission
A high electrostatic field may overcome the binding force between elec­
trons and protons and lead to the liberation of one or more electrons from
the cathode. This takes place when the electric field value is of the order
28 Chapter 2

10^-10^ V/cm. Most of power system components do not operate at such

high stress values. However, conditions for field emission can exist at elec­
trode protrusions and microdefects.

2.4 DEIONIZATION PROCESSES

Deionization is the process by which the number of charged particles in a

gas volume, especially the electrons, decreases. Since these processes op­
pose the ionization, in some applications they are desirable, e.g., to prevent
the avalanche growth or to quench an arc. The main deionization processes
are briefly described here.

2.4.1 Diffusion
Charged particles move from the region which has a higher concentration
to the region which has a lower concentration. The general diffusion equa­
tion is given as:
J - -DVn (2.17)
where J = the rate of charge flow, Vn = charge concentration gradient,
and D = diffusion constant, which is expressed as [1]:

D (2.18)

For electrons D will be three orders of magnitude higher than that of ions
due to their higher mean velocity v. When time is taken into account, the
rate of change of ion density n is given as [ 1 ]:
dn
= -V • J = D • V^n (2.19)
at

Solution of equation (2.19) will give the concentration of ions (n) at any
time and at any point. If this equation is solved for the case of diffusion
from a cylindrical concentration, the average displacement (r^) will be
given as [ 1 ]:
r<, = V W t (2.20)

2.4.2 Recombination
Positive and negative ions combine to form neutral atom as:
A+ + B- AB + hf ( 2 . 21)
Gas Dielectrics 29

The rate of recombination is directly proportional to the concentration of

both positive n+ and negative n_ ions, i.e.
dn+ dn_
-pn+n_ ( 2 .22)
^ ~dT
where p is the recombination constant. Since in general n+ = n_ = n, it
follows that:
dn .
(2.23)
di =
If this equation is integrated with respect to time, the density of charged
particles at any given time will be:

n(t) (2.24)
1 + n, pt
where m is the initial concentration.

2.4.3 Electron Attachment

Some electronegative gases such as O2 , CO2 and SF^ attach slow moving
free electrons to neutral gas molecules and form heavy negative ions.
Hence electron attachment, which constitutes a deionization process, can
be written as A + e —►A. This process is opposite to the detachment
process described in section 2.3.

2.5 UNIFORM FIELD GAS BREAKDOWN

The application of a specific electric field E across a dielectric gas will

result, at first, in its ionization. As the time and/or the applied field in­
creases, the ionization may lead to other physical processes, as shown in
Figure 2.3. The exponential growth of ionization usually leads to avalanche
formation. Such avalanches may, in turn, result in the formation of
streamer, leader, spark, arc or plasma. The electric field can be uniform or
nonuniform depending on the electrode arrangement. In case of a uniform
field gap, the electric stress is the same everywhere and hence the ioni­
zation and deionization parameters are constant. Therefore, the physics of
uniform field breakdown can be understood more easily. The two most
widely accepted theories of breakdown are the Townsend theory (1910)
and the streamer theory (1940). These are briefly described next.
30 Chapter 2

Leader

Ionization Avalanches Streamer

Spark

Arc

Figure 2.3 The various possible discharge process in a gaseous insulation.

2.5.1 Townsend Theory of Breakdown

Townsend investigated the ionization and breakdown under DC voltage
conditions, and proposed a theory to explain the experimental observations.
Considering the circuit of Figure 2.4, he assumed that n^ electrons are
being emitted from the cathode per second by the ultraviolet light. The
Townsend’s first ionization coefficient a is defined as the number of ion­
izing collisions made by an electron in per unit distance as it travels in the
direction of the applied field. It depends on gas pressure and electric field.
The number of electrons at a distance x from the cathode will be n^ and
is given by:

Figure 2.4 Experimental circuit for the study of Townsend discharge.

Gas Dielectrics 31

—a (2.25)
dx
and hence
n^ - n^ exp (ax) (2.26)
Similarly, n^, the number of electrons reaching the anode placed at distance
d, will be given by:
n^ = n^ exp (ad) (2.27)
The number of new electrons created on the average by each primary
electron leaving the cathode is:
n. - n^
= exp (ad) - 1 (2.28)

The average current in the gap, which is numerically proportional to the

number of electrons traveling per second, will be:
I = lo exp (ad) (2.29)
where Iq is the initial current at the cathode. The above constitutes a single
avalanche process. During the amplification of electrons in the field by a
process, additional electrons are being liberated in the gap by other (sec­
ondary) processes as well. The secondary electrons thus produced create
their own avalanches. Secondary processes include positive ion bombard­
ment on cathode and photo-ionization and detachment. Townsend second
ionization coefficient y is defined as the net number of secondary electrons
produced per primary electron leaving the cathode, y is a function of E/P
as well as electrode material, etc. The influence of secondary process on
the current growth can be considered as follows. Let n, = number of
secondary electrons produced at cathode per second, and n^ = (n^ + n j =
total number of electrons leaving the cathode per second. Therefore, the
total number of electrons reaching the anode becomes
n^ = nj exp (ad) = (n^ + n j exp (ad) (2.30)
However, by definition
«s = y(nd - n,) (2.31)
Thus,
Hs = y[nd - (Ho + n,)] (2.32)
Rearranging equations 2.30 and 2.32, we get:
32 Chapter 2

exp(ad)
(2.33)
1 - y[exp(ad) - 1 ]
Thus, current growth in the presence of a and y processes is given as:
Iq exp (ad)
I (2.34)
1 - y[exp (ad) - 1 ]
At breakdown I = oo, since the current is only limited by the resistance of
the external circuit. This condition is called Townsend breakdown criterion
and can be written as:
y[exp (ad) - 1 ] = 1 (2.35)
Normally exp (ad) > > 1, therefore the above equation becomes:
y exp (ad) == 1 (2.36)
Since a and y are dependent on E and P, thus for a certain value of d there
will be a value of E and hence V which will satisfy the Townsend break­
down criterion. The voltage V which satisfies the breakdown criterion is
called the sparkover or breakdown voltage and the corresponding dis­
tance d is called the sparking distance.
For electronegative gases where electron attachment takes place in ad­
dition to a and y processes, the attachment coefficient is also considered
in the current growth equation. Consequently, Townsend breakdown cri­
terion is also modified for such cases [1]. Townsend mechanism explains
breakdown phenomena only at low pressures corresponding to Pd < 1.45
bar-cm for air. For gaps with larger Pd values or for breakdown under fast
surge voltages, usually the streamer theory of gas breakdown applies.

2.5.2 Streamer Theory of Breakdown

Townsend theory fails to explain some experimental observations such as:
the zigzag and branched paths of the spark channel, and short breakdown
times when gaps are overstressed or have large Pd values. Due to these
and other limitations, streamer theory was proposed. This theory can be
briefly summarized qualitatively as follows.
1. In the uniform field gap of Figure 2.5, an electron avalanche con­
sisting of fast moving electrons and slow moving positive ions is generated
as discussed earlier. Thus, a space charge field opposing the main field is
formed. Soon the electrons will be absorbed by the anode leaving behind
an accumulation of positive ions. Excitation of atoms also takes place as
a result of collisions during the primary avalanche and photons are emitted
from the excited atoms.
Gas Dielectrics 33

Figure 2.5 The development of avalanche to a streamer in a uniform field (+ =

positive ions, - = electrons, ph = photons emitted from the avalanche).

2. These photons will be absorbed by gas atoms, and photoelectrons

are produced in the gas at various distances from the avalanche (Figure
2.5a). If the space charge field mentioned earlier is of the same order of
magnitude as the original applied field, then a second generation of aux­
iliary avalanches will be started by some of the most suitably located
photoelectrons.
3. As auxiliary avalanches are formed, further photons and conse­
quently photoelectrons are produced. Thus new third generation auxiliary
avalanches will be created (Figure 2.5b). The electrons do not follow the
original field lines any longer because space charge field distorts the orig­
inal applied field. Besides, many avalanches may be created almost si­
multaneously. This is the cause of the observed branching and the zigzag
paths of breakdown channels (Figure 2.5c). The auxiliary avalanches will
be continuously absorbed by the primary avalanche and positive ions space
charge will grow towards the cathode. Hence the ionized channel extends
from anode to cathode. The ionized channel is called a streamer. The
streamer tip forms branches that grow as a result of the incoming ava­
34 Chapter 2

lanches (Figure 2.5b). Those electrons at the tips will soon be absorbed by
the streamer and move in the channel towards the anode by virtue of a
potential gradient within the streamer channel.
4. The propagation of one streamer tip continues while the others
stop advancing due to the lack of avalanches feeding into them as shown
in Figure 2.5c. If this process continues, a final streamer channel will be
formed between the anode and the cathode, causing a complete breakdown.
This channel will be similar to the one sketched in Figure 2.5d with nu­
merous “incomplete” branches.
Once a streamer is formed, it usually leads to breakdown quickly.
Mathematically an empirical streamer breakdown criterion for uniform
field gaps can be formulated as:

a • dx = n (2.37)
f
where n^ is the critical number of electrons or ions in an avalanche when
it transforms into a streamer. Usually it is believed that n^ ~ 10^ for air
and other gases.

2.5.3 Paschen Law

Based on the Townsend breakdown criterion, a relation between breakdown
voltage and the product of pressure and gap spacing can be established.
This relation can be deduced as follows.
Both a and y depend upon electric field E and gas pressure P. There­
fore we can write
E
- f, (2.38)

V
and since in a uniform field, E = —,
d
y_ y_
= f. (2.39)
Pd Pd
Substituting equation (2.39) in equation (2.35) leads to:
y_
exp = 1 (2.40)
Pd
There is only one value of V for a particular Pd value which satisfies
equation (2.40). This value of V is the breakdown voltage which can
be written as:
Gas Dielectrics 35

Figure 2.6 Paschen curve for air.

V, = f(Pd) (2.41)
Equation (2.41) shows that for a particular gas, the breakdown voltage is
a unique function of the product of pressure and gap length. This relation
is known as the Paschen law. The Paschen curve for air is shown in Figure
2 .6 , which shows a minimum value of V, around a particular value of the

product Pd. Table 2.2 shows minimum V, values (V^ for some gases
along with the corresponding Pd values [2].

Table 2.2 Minimum Sparking

Potential for Various Gases

Pd at V,
Gas (V) (Pa-cm)

Air 327 75.6

Ar 137 120
H, 273 153.3
He 156 533
CO2 420 68
N. 251 89
N2O 418 66.65
O2 450 93.3
SO2 457 93.3
H^S 414 80
36 Chapter 2

2.6 NONUNIFORM FIELD GAS BREAKDOWN

Any gas can withstand a certain electrical stress, and when the stress ex­
ceeds this value, a discharge will ensue. Since the stress in a uniform field
gap is equal everywhere, discharge in such gaps usually takes the form of
a complete breakdown. However, in non-uniform field gaps, the discharge
will take place only in the areas where the stress is higher than the dielec­
tric strength of the gas. This is known as a partial discharge (PD), and
when it occurs at electrodes in air or other gases it is called “corona.” In
high voltage systems, often it is not economical to design equipment that
is free of corona at nominal working voltages. The knowledge of corona
onset voltage and the physical damage caused by corona is therefore
important.

2.6.1 Corona Inception

Corona onset field is the critical stress value at the conductor surface cor­
responding to the corona inception. For power frequency applications, the
following relations are generally applicable for corona inception.
1. For a single conductor above ground, the critical field (E^) which
will cause stable corona on the conductor is given as:

E. = 30 m RAD 1

where r = radius of the conductor, RAD = relative air density and m,. =
surface irregularity factor (nij, = 1 for smooth conductors whereas for rough
conductors its value is less than 1). The corona onset voltage in this
case can be written as:
m
V. = E.r In (2.43)
r
where H is the height of the conductor above ground.
2. For coaxial cylinders of inner and outer radii, r, and rj, respec­
tively, E(, and V^, are given as:
0.308 ^
E. = 31 m. RAD 1 (kVpeak/cm) (2.44)
Vr, • RAD/

Vp = Ep r, In (2.45)

In the above equations, relative air density (RAD) at pressure P (mbar) and
temperature T (K) is given as:
Gas Dielectrics 37

293
RAD = (2.46)
1013 T

2.6.2 Corona Discharges

Corona discharges are best investigated using a sphere (or rod) to plane
electrode configuration where the sphere (or rod) radius is chosen accord­
ing to the field non-uniformity desired. Depending on the applied voltage
and the shape of the electrode, there are six possible modes of corona:
three for the positive DC or +ve half cycle of the AC, and three for the
negative DC or -v e half cycle of the AC voltage. Table 2.3 and Figure
2.7 illustrate the occurrence of various corona modes in air, as they appear
with an increase in the applied voltage.
The voltage values for positive and negative corona onset are approx­
imately similar, however the transition between different modes occurs at
different voltages for the two polarities. Depending on the rod radius and
the interelectrode gap spacing, one or more of the above-mentioned modes
may be absent for some electrode arrangements. In addition, two modes
may occur simultaneously in some cases. Brief descriptions of different
corona modes are given next.

Negative Corona
Trichel Pulses
As the voltage is raised up to the critical field intensity, electron avalanches
are formed and propagate towards the anode leaving behind positive ions.
When the electrons enter the low field region, they form negative ions.
Thus, a space charge field is formed as shown in Figure 2.8. Here is
the applied electrostatic field and is the field due to space charges. The
discharge stops when the effective field (E^ - EJ in the vicinity of the
sharp electrode drops below the critical field value. After the space charges
clear the gap, the field in the vicinity of the rod recovers and the process

Table 2.3 Corona Modes

Negative corona Positive corona

Trichel pulses (TP) Onset pulses (OP)

- v e pulseless glow ( - v e G) + ve Herrnstein glow (+ v e G)
- v e prebreakdown streamer ( - v e st) + ve prebreakdown streamer (+ v e st)
Complete breakdown Complete breakdown
38 Chapter 2

Figure 2.7 AC cycle with the various possible corona modes.

Avalanche in progress

z> Avalanche stops

W hen(Eo- Es)< Ec

Figure 2.8 N egative co rona avalanche.

Gas Dielectrics 39

Figure 2.9 Negative corona current (Trichel pulse).

is repeated again. This leads to the formation of corona current pulses as

shown in Figure 2.9. Upon raising the applied voltage, the number of
pulses per second generally increases. In addition, the pulse amplitude
decreases.
Negative Glow
Upon increasing the voltage further, at a certain voltage the space charge
field becomes incapable of reducing the effective electric field intensity in
the vicinity of the cathode below the critical value. Hence the discharge
becomes continuous causing pulseless current. This mode is known as glow
corona (Figure 2.10).
Negative Prebreakdown Streamers
Upon increasing the voltage further, the discharge occupies a major portion
of the interelectrode spacing and incomplete streamers are formed. Such
streamers are similar to the incomplete streamers already discussed under
uniform field breakdown (section 2.5). This corona mode is known as
prebreakdown streamer corona. Upon increasing the voltage further, a com­
plete breakdown will take place.

Figure 2.10 N egative glow corona current.

40 Chapter 2

Positive Corona
Onset Streamers
As the voltage is increased till the critical field is established at the anode,
electrons are accelerated from the low field region towards the anode and
cause ionization in the high field region. At anode, the electrons will be
absorbed quickly whereas the positive ions accumulate around the anode
(Figure 2.11). The discharge stops when the effective field near the anode
drops below the onset conditions. However, the discharge restarts when the
positive ions are cleared away from the anode towards the cathode and the
field in the vicinity of anode recovers to a higher value. In some special
cases, streamers extending tangentially onto the anode can be formed.
These are called burst pulse streamers. Such pulses are characterized by
slow rise time and small magnitudes.
Positive Glow
As the voltage is increased further, the field at the anode will be high
enough to cause discharge even when there is a positive space charge near
the anode. In this case a continuous current will flow from glow corona.
Positive Prebreakdown Streamers
Upon raising the voltage further, incomplete streamers are formed resulting
in prebreakdown streamer mode of corona. As the voltage is further in­
creased, complete breakdown occurs in the gap.

(a)

(b)

Figure 2.11 Positive corona avalanche: (a) avalanche in progress; (b) avalanche
stops.
Gas Dielectrics 41

In power systems, corona occurs on overhead transmission lines and

can display all of the modes discussed earlier due to sinusoidal nature of
the AC voltage. Such corona increases the attenuation of high voltage
surges as they propagate on the line conductors. However, it causes radio
interference, audible noise and power losses. Corona also finds many in­
dustrial applications, such as in high-speed photocopying machines, in pre­
cipitators used for gas pollution control, in Van de Graff HVDC generators
and in discharging undesirable charges from airplanes and plastics, etc.
Recently pulsed corona is being used for air and water purification and for
many other industrial applications.

2.6.3 Breakdown in Nonuniform Field Gaps

In a nonuniform field gap, a and y are no longer constants. These vary
with the field between the two electrodes and hence equations for current
growth have to account for such a position related dependence of a and
y. If the field factor, f = is less than five, the field is called quasi­
uniform and the discharge phenomenon would be similar to that in a uni­
form field gap. However, if f > 5, the field is called nonuniform. In non-
uniform field gaps, the corona plays an important role in the final
breakdown of the gap and hence breakdown voltage is strongly influenced
by the presence/absence of corona as well as by the prevailing corona
mode(s).
Once the corona starts, the applied electric field becomes distorted by
the space charge field, and hence the breakdown process becomes much
more complex. If the highly stressed electrode is positive, the space charge
acts as an extension of the anode. On the other hand, if the highly stressed
electrode is the cathode, the space charge acts as a shield that decreases
the field in its vicinity and thus such a configuration needs a higher voltage
for complete breakdown. Consequently, in nonuniform field gaps, break­
down voltage is lower for positive polarity as compared to the negative
one. Therefore, the breakdown voltage characteristics for positive polarity
direct voltage applied to the sharp point is usually more important for
practical applications. Figure 2.12 illustrates schematically the dependence
of the breakdown voltage on the gap length and the sphere diameter for
sphere-plane air gaps under positive DC voltage. There are three main
regions where the transition between them is sphere diameter dependent:
Region I. For short gaps, the field is almost uniform and the break­
down voltage depends mainly on the gap length. There is no co­
rona in this case.
42 Chapter 2

Figure 2.12 Corona and breakdown voltage as a function of gap spacing for
sphere to plane geometry.

Region II. For moderate gap lengths, the field shows moderate non­
uniformity. No corona occurs in this region also and the break­
down voltage increases with the sphere diameter as well as with
the gap length.
Region III. If d > 2D, the field is highly nonuniform and the break­
down is preceded by corona. The corona onset voltage depends
mainly on the sphere diameter while breakdown voltage is gap
length dependent.

2.7 TIME TO BREAKDOWN

The breakdown process develops over a certain period of time like any
other natural phenomenon. Generally this time is very short and not no­
ticeable under DC or AC applied voltages. However, under impulse volt­
ages, this time becomes important, since it may be comparable with the
impulse voltage front time or its duration. Impulse voltages are generally
used to simulate the lightning or switching surge overvoltages that occur
in power systems. In most cases, the insulation design of power system
components is based mainly on the magnitude of surge overvoltages.
Gas Dielectrics 43

The time lag between the instant when the applied voltage is of suf­
ficient magnitude to cause breakdown and the actual event of breakdown
can be divided into statistical (tj and formative (tf) time lags. The former
is the time required for an initiatory electron to appear in the highly
stressed region of the gap after the application of an impulse. The formative
time lag is the time necessary for the breakdown process to be completed
after the initiatory electron becomes available (see Figure 2.13). Thus the
impulse breakdown voltage is higher than static or DC breakdown volt­
age V^. The total time lag, t = k + tf, will depend on the overvoltage
value, AV = Vj - V^, and it will decrease as AV increases. The statistical
nature of time lag leads to a probabalistic variation of the breakdown volt­
age. Consequently, varying shapes of volt-time characteristics are noticed
for different surge voltages. Its engineering significance will be highlighted
in the next chapter.

2.8 DISCHARGES UNDER NANOSECOND PULSE VOLTAGES

In the presence of electric field, the number of electrons grow exponentially

since n^ = n^ exp (ax). At the critical avalanche length x^, the number of
electrons at the avalanche head becomes n^ = n^, where n^ ~ 1 0 ^ for air.
The avalanche space charge field becomes comparable to the applied elec­
trostatic field and conditions for the formation of a streamer are satisfied.
Once a streamer forms, breakdown may take place. For the most conser­
vative case, n^ = 1 , and the critical avalanche length is given as

Figure 2.13 T im e lag com ponents u nder a step voltage.

44 Chapter 2

In n
X c- (2.47)

If Xc > d (gap length), then the primary avalanche is insufficient to trans­

form into a streamer and cause a breakdown. Therefore, subsequent ava­
lanches (2nd, 3rd, etc.) are necessary to cause a breakdown. If < d, a
primary avalanche can become a streamer provided the avalanche generates
enough photons to ionize the gas molecules near the avalanche head. Such
photons are produced when excited molecules go to the ground state in
about 10"^-10“^ seconds [1]. This de-excitation time (T^^^) should be more
or at least comparable with the electron drift time T^, which is given as:
In m
(2.48)
a K
where is electron drift velocity in the avalanche. Thus, from equations
(2.47) and (2.48) the conditions for streamer formation can be summarized
as:
In m In m
Xc = -----^ < d; Te = -----^
a
> Th (2.49)

These conditions can be achieved if a high voltage pulse of nanosecond

duration is applied to the electrode gap. To complete a discharge under
such voltage conditions, a large number of electron avalanches must be
produced by photoelectrons near the cathode. In multi-avalanche initiation,
a space charge current comparable with the maximum circuit current is
possible. To reach this condition, an electric field must be applied with a
value much higher than that associated with DC, AC and switching or
lightning impulse voltages. Using nanosecond high voltage pulses, very
intense corona can be produced with high applied electric fields. One of
the advantages of nanosecond pulsed discharge is the low energy con­
sumption, since only electrons are accelerated while there is insufficient
time for ions to speed up. In such circumstances the gas temperature re­
mains low and the spark is avoided except at extremely high applied field
values. Nanosecond pulse discharges are finding increasing applications in
recent years. The generation of high voltage pulses of nanosecond duration
and their applications will be outlined in chapter 1 0 .

2,9 GAP-TYPE DISCHARGE

A gap-type discharge is a complete electrical breakdown of the insulation

between two charged surfaces. It occurs in the form of a spark at microgaps
Gas Dielectrics 45

( < 1 mm) where the potential gradient is large enough to initiate a dis­
charge. Either both electrodes of the gap or at least one electrode is ca-
pacitively coupled to a voltage source or to ground. Figure 2.14 shows a
typical arrangement for gap-type discharge where one of the electrodes is
capacitively coupled to the high voltage source. Associated with this dis­
charge is a current pulse characterized by a very high crest value (from a
few amperes to several tens of amperes) and extremely short rise time
(—1.0 ns). This discharge could be repetitive depending on the gap loca­
tion. This type of discharge is a broad spectrum electromagnetic noise
source (up to hundreds of MHz) which can be radiated or conducted over
long distances causing severe interference in nearby computer, communi­
cation and control systems [6,7].
The gap-type discharge is not restricted to EHV lines but it could also
occur on an improperly designed or constructed distribution system pro­
vided that the stress in the gap reaches a certain critical value. The follow­
ing are some examples of gap-type discharges where the spark may occur
between bad contacting metallic parts exposed to high voltages: ( 1 ) be­
tween caps and pins of an insulator, (2 ) between cross-arms loosely at­
tached to the wooden distribution towers, (3) at the junctions of insulator
strings and the transmission line towers and (4) between the spacers and
subconductors of a transmission line’s bundle conductors. Bad contacts
between metallic parts can be caused by corrosion, dust or dry pollution.
Other factors causing miscontact are vibrations due to wind, insufficient
mechanical loading and the changes in dimensions due to temperature var-

HV

=i=Ci

<^ m m y
C able Z q
Measuring
i Zo TT instruments

Figure 2.14 G ap discharge arrangem ent.

46 Chapter 2

iations. A microdischarge may also occur during rain in a microgap be­

tween a water drop and the transmission line, or between two water drops
close to the transmission line [8,9]. Figure 2.15 displays a typical discharge
current waveform resulting from a gap-type discharge.

2,10 CHOICE OF DIELECTRIC GASES

The desirable properties of an insulating gas were summarized earlier in

section 1.3.1. SF^ and air are the most commonly used gases for insulation
in high voltage power systems as both of these gases possess most of the
desired properties. Although SF^ has been used satisfactorily and almost
exclusively for pressurized gas-insulated high voltage transmission and
substation equipment, there have been many investigations into alternative
gases. The gas with higher dielectric strength offers the possibility of op­
erating existing designs at higher voltages or reducing the equipment size.
Reduction in size means less quantity of gas and lesser amount of sheathing
and other insulating materials, etc. The other properties considered for a
better gas are its immunity toward particle-initiated breakdown, to surface
roughness and area effects [9]. Similarly a gas to be used in circuit breakers
as a replacement for SF^ must possess better arc interrupting properties as
well.

Figure 2.15 Typical gap discharge current.

Gas Dielectrics 47

The dielectric strength of a gas depends on its ability to attach free

electrons over a wide energy range. However most dielectric gases capture
electrons effectively only in a very restricted low energy range; e.g., SF^
attaches electrons at energies < 0.4 eV. Beyond this energy range, SF^
electron attachment cross section (crj decreases rapidly [ 1 0 ], so electrons
that possess higher energies escape the attachment. Behavior of of SF^
also explains the considerable decrease in the dielectric strength of SF^
under nonuniform fields, surface roughness, and particle contamination
(see chapter 4 for details) [11]. In new gases, a higher cr^ is sought and
considerable research has resulted in a variety of new insulating gases.
Table 2.4 shows few properties of some dielectric gases. It is clear that
insulating gases such as perfluorocarbons that possess higher cr^ exhibit

Table 2.4 Relative Dielectric Strength of Some Gases in Uniform Fields at 1.5
bar (SF, = 1)

Boiling point Dielectric

Name Formula (°C) strength

Hexafluoro-2-butyne C4F, -2 5 2.2

Trifluoromethyl sulfur CF3SF5 -2 0 1.55
pentafluoride
Trifluoromethane sulfonyl CF3SO2F -2 2 1.49
fluoride
Thionyl fluoride SOF2 -4 4 1.42
Trifluoronitromethane CF3NO2 -3 1 1.34
Perfluoro-n-butane f-4Fs 2 1.5
Chloropentafluoroethane (F-115) C2F5C1 -3 9 1.13
Perfluoro-butane C4F.0 -2 1.06
Sulfur dioxide S02 -1 0 1.0
Sulfur hexafluoride SF, -6 4 1.0
Dichlorodifluoromethane (F-12) CCI2F2 -3 0 0.99
Octofluoropropane (F-218) CsFs -3 7 0.98
Perfluorodimethyl ether C^F.O -5 9 0.84
Hexafluoroethane (F-116) C2F, -7 8 0.79
Bromotrifluoromethane (F-1381) CBrF3 -5 8 0.75
Chlorotrifluoroethylene (CTFE) C2F3CI -2 8 0.69
Nitrous oxide N2O -8 9 0.5
Carbon tetrafluoride (F-14) CF4 -1 2 8 0.42
Air N2 + O2 0.37
Nitrogen N2 -1 9 6 0.37
Carbon dioxide CO2 -7 9 0.32

Source: Ref. 9.
48 Chapter 2

higher breakdown voltages. Besides satisfactory dielectric strength and

cost, properties like toxicity, low vapor pressure, boiling point, tendency
to carbonize and environmental hazards are also important. As shown in
Table 2.4, perfluoro-n butane (C4 F 8 ) possesses a value of dielectric strength
which is higher than SF^ by a factor of 1.5. But its high boiling point
(+2.0°C) as compared to -64°C for SF^ prevents its application in cold
climates or in a compressed state. The greatest potential for new practical
gaseous insulators therefore lies in the development of synergistic multi-
component mixtures, the components of which are selected on the basis
of their individual physico-chemical properties [12,13].

REFERENCES

E. Naser, F u n d a m e n ta l o f G a s e o u s I o n iz a tio n a n d P la s m a E le c tr o n ic s , Wiley

Interscience, New York, 1971.
2. M.S. Naidu and V. Kamaraju, H ig h V o lta g e E n g in e e r in g , Tata McGraw-Hill
Publishing Company Ltd., New Delhi, India, 1982.
3. J.M. Meek and J.D. Craggs (eds.). E l e c t r i c a l B r e a k d o w n in G a s e s , John Wiley
and Sons, New York, 1978.
G.A. Mesyata, Y.I. Bychekov and U.V. Kremnev, Soviet Physics WS Pekhi,
Vol. 15, No. 3, pp. 282-297, 1972.
W. Janischewskyj and A.A. Al-Arainy, Proceedings of the 1988 U.S.-Japan
Seminar, “Electromagnetic Interference in High Advanced Social Systems,”
Honolulu, Hawaii, August 1-4, 1988.
6. A.A. Al-Arainy, “Laboratory Analysis of Gap Discharges on Power Lines,”
Ph.D. thesis. University of Toronto, Canada, 1982.
7. W. Janischewskyj and A.A. Al-Arainy, IEEE Trans, on PAS, Vol. 100, No. 2,
pp. 539-551, 1981.
8. A.A. Al-Arainy, “The Effects of Rain on Electromagnetic Characteristics of
Corona,” M.Sc. thesis. University of Toronto, Canada, 1977.
9. L.G. Christophorous, G a s e o u s D i e le c t r i c s , Vol. II, Pergamon Press, New York,
1980.
10. L.G. Christophorou, lEE Conf. Publication, No. 165, pp. 1-8, 1978.
11. R.E. Wooten, S.J. Dale and N.J. Zimmerman, Proc. of the Second Int. Symp.
on Gaseous Dielectric, Knoxville, Tennessee, 1980.
12. L.G. Christophorou and L.A. Pinnaduwage, IEEE Trans, on Elect. InsuL, Vol.
25, No. 1, pp. 55-74, 1990.
13. L.G. Christophorou and S.J. Dale, E n c y c lo p e d ia o f P h y s i c a l S c ie n c e a n d T ech ­
n o lo g y , R.A. Meyors (ed.), Vol. 4, pp. 246-262, Academic Press, New York,
1987.
Air Insulation

3.1 INTRODUCTION

The basic processes which lead to the electrical breakdown of gases were
summarized in chapter 2. Air is the most commonly used gaseous insula­
tion medium in high voltage power networks because it is free, is abundant
and becomes self-restoring after a breakdown. Thus, electrical breakdown
behavior of air is very important for designers and operators of high voltage
equipment. For this reason, electrical breakdown and prebreakdown of air
gaps have been thoroughly investigated since the start of this century and
a vast amount of literature and data are available on this subject. Based on
such information, international recommendations for air clearances have
been established and are being used for the design of air insulated, high
voltage power lines and other equipment. This chapter provides a brief
summary of the breakdown characteristics of air gaps and the most im­
portant factors which can influence air insulation characteristics from the
power system engineer’s point of view.

3.2 AIR INSULATION APPLICATIONS AND MODELING

Air is used as an insulant for outdoor as well as indoor high voltage power
networks. For insulation purposes, it is used to provide the phase-to-phase
as well as phase-to-ground insulation. In addition, air is also used in chop­
ping, spark and measurement gaps. Due to a wide range of such applica-

49
50 Chapter 3

tions, the electrodes normally used in air insulated components have a great
variety. However, most of these cases can be modeled by some simple
electrode configurations. Table 3.1 summarizes the most common sections
of air insulated power network and their commonly adopted electrode mod­
els that are used for the evaluation of these component’s/section’s dielec­
tric behavior.

Table 3.1 Applications of Air Insulation in Power Network and Laboratory

Models Used for Investigations of Their Dielectric Characteristics

Network section Function Model used for simulation

Phase to phase insulation To provide insulation Conductor-conducter, i.e.,

between two phases of parallel cylinders or
an AC transmission rod-plane electrodes to
line or between simulate the worst case
opposite poles of a of a sharp point on one
bipolar DC line conductor opposite to
the other conductor
Phase to tower or phase To provide insulation Conductor-plane or rod-
to ground insulation between the phase plane electrodes
conductor and the
grounded tower or the
ground itself
Small diameter conductor To provide insulation Point-plane or rod-plane
to grounded object between a sharp electrodes
insulation conductor and a flat
grounded object in
front of it. This type
of geometry has the
lowest air breakdown
strength
Sphere-sphere gaps High voltage Sphere-sphere electrodes
measurements and
high voltage switches
in impulse generators
High voltage protective To bypass high voltage Rod-rod electrodes
or measuring gaps surges to ground by a
spark discharge. Also
used for impulse
chopping and for high
direct voltage
measurements
Air Insulation 51

3.3 VOLTAGE STRESSES

High voltage power network components, besides normal operating volt­

ages, are subjected to temporary power frequency overvoltages caused by
network faults. In addition, transient voltage surges caused by atmospheric
lightning phenomena (lightning surges) and system switching operations
(switching surges) overstress the insulation. The dimensioning of the sys­
tem insulation is dictated by the air gap’s breakdown behavior when sub­
jected to standard lightning or standard switching impulses depending upon
the rated voltage of the component. For equipment with a line voltage
rating of <300 lightning impulses are more critical, whereas for
higher voltage ratings, switching impulses assume a greater importance.
Therefore, for the design of air insulated components, lightning and switch­
ing impulse breakdown data are very important.

3.4 IMPULSE BREAKDOWN PROBABILITY

As discussed in chapter 2, the AC or DC breakdown voltage of a uniform

field gap depends upon the gas pressure (P) and the gap distance (d),
provided the temperature is kept constant. When the field distribution is
nonuniform, the breakdown voltage is also influenced by the presence of
corona and space charges in the gap, and gaps with asymmetrical field
distribution exhibit pronounced polarity effect. Consequently, when posi­
tive polarity voltages are applied to the stressed electrode, these result in
a lower breakdown voltage value than the corresponding breakdown volt­
age value for the negative polarity. Thus, positive polarity breakdown data
assume greater importance in such cases.
Under impulse voltage applications, breakdown voltage of an air gap
exhibits statistical variations due to time lag effects discussed in chapter
2. Consequently, when a given gap is repeatedly subjected to impulses of
a fixed amplitude, there exists a certain probability of breakdown. For
impulse voltages, the probability function, which shows variation of the
breakdown probability with applied impulse voltage magnitude, is very
important. From such a function, the following parameters can be
determined:
V5 0 = 50% breakdown voltage level, i.e., the impulse amplitude
which exhibits 50% probability of breakdown.
Vq = 0 % breakdown voltage level, i.e., the highest withstand voltage,
or the maximum impulse amplitude at which no breakdown
occurs.
52 Chapter 3

Vjoo =1 0 0 % breakdown voltage level, i.e., the lowest impulse ampli­

tude at which a breakdown is always ensured.

a - standard deviation of the breakdown voltage.
COV = 0 -/V5 0 == coefficient of variation of the breakdown voltage.
Air gaps mostly display a Gaussian or normal probability distribution,
and thus the probability versus voltage graph is a straight line when plotted
on a probability graph paper as shown in Figure 3.1. Sometimes the break­
down probability function is not normal due to a change in either the
breakdown mechanism or some experimental variables, and a mixed prob­
ability distribution function is observed. If the probability function is nor­
mal, a can be calculated from:

V s 4 - V50 = V30 - V, (3.1)

Moreover, Vq V5 0 - 3(t whereas V, (V5 0 + 3cr). Thus, for an air
gap’s breakdown data, V5 0 values are usually quoted.

3.5 BREAKDOWN VOLTAGE CHARACTERISTICS

Table 3.1 shows that different electrode configurations used for simulating
practical air insulated apparatus. For most applications, the gaps employed
have nonuniform field distribution and, therefore, breakdown data of such
gaps are very important. In addition to the electrode configuration, the other
factors which can influence the breakdown behavior of an air gap are the
voltage waveform, voltage polarity, air pressure, temperature and humidity
as well as presence of atmospheric pollution. The influence of such factors
on breakdown characteristics of air gaps is described next.

3.5.1 Basic Electrode Shapes

Since the laboratory measurements are carried out to simulate the actual
electrodes used in the high voltage system, the electrode arrangements that
are most commonly used for air gap breakdown studies are:
1. Sphere-sphere electrodes which form a symmetrical and a uniform
field distribution gap. Sphere-sphere gaps have the highest average
breakdown strength. The breakdown voltages for such gaps depend
upon sphere diameter, gap length, voltage polarity and voltage
waveform and are discussed in chapter 1 0 in detail.
Air Insulation 53

Figure 3.1 Normal breakdown voltage probability distribution.

2. Rod-rod electrodes which form a symmetrical but a nonuniform

field distribution in the gap and have several applications. These
are discussed in Table 3.1.
3. Rod-plane electrodes which form an asymmetrical and a nonuni­
form field distribution in the gap. Such electrodes have generally
the lowest average breakdown strength.

3.5.2 Breakdown Characteristics of Rod-Plane Gaps

The conservative approach in dimensioning the phase to phase and phase
to tower clearances of a transmission line is to use the rod-plane gap’s
breakdown data. For medium and high voltage lines, the minimum air
clearance is determined by AC voltage stress, insulator flashover problems,
corona effects and lightning impulse level. However, for extra high voltage
lines, positive switching impulses are the most critical. Therefore, break­
down of rod-plane gaps under positive switching impulse voltages have
been studied by many investigators. Several researchers have proposed em­
pirical formulas for the relation between breakdown voltage (V5 0 ) and gap
length (d) for rod-plane geometry under positive switching impulse volt­
54 Chapter 3

ages. Table 3.2 presents several of these formulas and the range of the gap
length for which they are applicable. Figure 3.2 shows the relation between
V5 0 and gap spacing using the above mentioned formulas. Obviously there
is no single formula that can cover the whole range of the tested gap of
up to 30 m. For gaps employed in power equipment, most of these for­
mulae give similar results. It is interesting to see some saturation in the
breakdown voltage as gap length increases. The scientific community is
not sure about the behavior of longer rod-plane gaps, i.e., d > 30 m, unless
measurements for such long gaps are carried out. The saturation tendency
shown in Figure 3.2 suggests an upper limit on the possible transmission
voltage level for overhead UHV lines.

3.5.3 Effect of Electrode Shape

Paris [1] showed that V5 0 of a given air gap geometry stressed by positive
switching impulses is proportional to the V5 0 value of a rod-plane geometry
of the same gap length. This behavior can be expressed as follows:
V^o = KV-P (3.10)
where V5 0 = V5 0 value of any air gap configuration x, = V5 0 value
of a rod-plane air gap of the same gap length and K = gap factor.
Several other researchers investigated this subject further in order to
establish the limitations of the gap factor K and the ranges where it can
be applied [2-5]. Based on the above references, and the present authors

Table 3.2 Breakdown Voltage Formulae for Rod-Plane Gap under Positive
Switching Impulse

Range of Equation
V50 (kV) d (m) Reference No.

500 (doo) 2 < d < 8 Paris [1] (3.2)

3400/(1 + 8 /d) d < 15 Gallet et al. [6] (3.3)
450 [1 + 1.33 In (d - Ind)] d < 10 Lemke [7] (3.4)
[1.5 X lO® + 3.2 X 10=* d]«' - 350 d < 20 Waters [8] (3.5)
1260r[l - (r/d)]o^ tanh-‘V l - (r/d) d < 20 Aleksandrov [9] (3.6)
1400 + 55 d 13 < d < Pigini et al. (3.7)
30 [10]
[[(1556 + 5 0 d )/(l + 3.89/d)] + 78] d > 4 Rizk [11] (3.8)
1080 In (0.46 d + 1) d < 25 Kishizima et al. (3.9)
[12]

d = gap length, r == rod radius (both in meters).

Air Insulation 55

Figure 3.2 V 50 breakdown voltage as function of gap length using the various
formulae mentioned in Table 3.2. Numbers within parentheses indicate reference
number.

verifications, Table 3.3 was constructed which gives the values of K for
most practical electrode configurations. This table shows that the gap factor
increases as the electrode geometry departs from the most divergent case,
i.e., rod to plane geometry to a less divergent field. Normally if K > 1.6,
the withstand voltage for the negative polarity switching impulses becomes
less than that for the positive polarity impulses [5]. Therefore, in such a
case, the negative polarity breakdown data will assume greater significance
and will form the design basis.

3.5.4 Influence of Voltage Waveshape

The magnitude of the sparkover voltage also depends on the shape of the
applied voltage waveform since the amount of injected space charge in the
gap in the prebreakdown period depends on the time taken by the voltage
to reach its crest value. The charges accumulated in the gap during the
predischarge process modify the electric field distribution in the gap, and
hence influence the sparkover mechanism.
Figures 3.3 and 3.4 show the breakdown voltage as function of gap
length for the different voltage waveforms for rod-plane and rod-rod elec­
trodes, respectively. It is clear that for the same gap spacing, the rod-plane
gaps have similar or somewhat lower breakdown voltages than rod-rod
56 Chapter 3

Table 3.3 Gap Factor K for a Number of Practical Electrode Configurations

Gap configuration Gap factor (K)

Rod-plane 1
Conductor-plane 1.12 - 1.25 (depends on d)
Horizontal rod-rod above ground 1.35 - (d/H , - 0.5)
Hj = Height of rods above ground
Vertical rod-rod H
1 + 0.6 H + d,
Conductor-rod (1.1 to 1.15) exp [0.7 H /(H /H + d)]
Parallel conductors 1.6 - 1.75 (depends on d)
Conductor-rope 1.4
Conductor-cross arm 1.45 (typical value)
Conductor windows 1.25 (typical value)

H = length of grounded rod, d = gap length.

configuration. Moreover, the positive impulse breakdown voltages are gen­

erally lower than the corresponding negative polarity values. In some rare
cases, especially in the presence of high humidity, an air gap may break­
down at a lower negative voltage than the positive one [ 1 1 ].
The results of Figure 3.3 also show the existence of critical crest time
corresponding to the minimum breakdown voltage for positive rod-plane
gaps. With an increase in the impulse front time, the breakdown voltage
generally decreases up to a certain critical value. When the front time is
further increased, the breakdown voltage starts to increase. Thus a U-
shaped curve (or a U curve) type of relation between the breakdown voltage
and impulse front time is obtained. Typical U curves are shown in Figure
3.5. From air insulation applications point of view, the worst combination,
i.e., the combination which produces the lowest breakdown voltage, is the
rod to plane electrodes and positive switching impulse wave having a crit­
ical front time The critical front time (t^j.) can be related to the
gap spacing and electrode shapes as follow [13]:
t,, - [50 - 35(K - l)]d (3.11)
where K is the gap factor. In the design of EHV and UHV lines, the
impulses are assumed to have rise time equal to t^j.* The reported research
regarding the influence of impulse tail time (to 5 ) (see chapter 1 0 for details
of to 5 ) on the breakdown voltage value is scarce. One such study [14]
showed that V5 0 and Vjo increase with wavetail reduction provided t^^. is
kept more or less constant. In general if ^0.5 kr’ the breakdown voltage
becomes less sensitive to the changes in to5 values.
Air Insulation 57

Figure 3.3 Breakdown voltage characteristics of rod to plane gaps.

3.5.5 Influence of Atmospheric Parameters

Atmospheric parameters such as temperature, pressure, humidity, rain (or
snow) and wind can influence the breakdown voltage of a gap. It is found
that an increase in the absolute temperature (T) which causes an increase
in the distance between molecules has the same effect as a decrease in
pressure (P) and vice versa. These two variables, i.e., P and T, can be
combined in the relative air density (RAD) which is defined as:
273 + 20
RAD = — -T )/ = i—
\1013/) 273 + T„
(3.12)

where = standard atmospheric pressure at sea level = 1013 mbar,

= standard ambient temperature = 293°K (or 20°C), = actual ambient
temperature in °C and P = actual pressure in mbar.
58 Chapter 3

Figure 3.4 Breakdown voltage characteristics of rod to rod gaps.

In general, an increase in RAD causes an increase in the breakdown

voltage. The air humidity influences some physical discharge parameters
such as the ionization and the attachment coefficients. Consequently, the
breakdown voltage values are influenced by the changes in the absolute air
humidity (H), expressed as the weight of water molecules present per unit
air volume (g/m^). Generally the breakdown voltage increases with H over
the practical range of 2 < H < 25 g/m^. The extent of this increase
depends upon the field configuration, gap length, waveform and polarity
of the applied voltage, and the value of H. Although the values of humidity
correction factor k^ are required by the standards [15,16], there is still some
disagreement between researchers about the value of this factor [3,17,18].
According to lEC 60-1 [15], the following correction factors are applicable
for breakdown voltages measured at nonstandard atmospheric conditions:
Air Insulation 59

Figure 3.5 Influence of impulse front time on the breakdown voltage of positive
rod to plan gaps. (From Ref. 6 © IEEE, 1975.)

1. Air density correction factor (kj), which is given as:

k, - (RAD)"^ (3.13)
where m is an exponent which depends on the electrode geometry,
gap length and the applied voltage waveform.
2. Humidity correction factor (k2 ). When the absolute humidity, H,
is different from the standard humidity of 1 1 g/m^, then the fol­
lowing humidity correction factor should be applied:
k2 = (3.14)
where w is an exponent similar to m and k is given by:
k = 1 + A[H/RAD) - 1] (3.15)
60 Chapter 3

Here A = 0.01, 0.012 and 0.014 for impulse, AC and DC applied

voltages, respectively.
Figure 3.6 shows the values of exponents m and w as a function of g,
where g is given by:

(3.16)
500dk(RAD)
Thus, the actual air gap breakdown voltage is related to the breakdown
voltage at standard atmospheric conditions (at T = 20°, P = 1013 mbar
and H = 11 g/m^) and the above mentioned correction factors by the
following relation:
V3 = V, • k. = V /k, • k,) (3.17)
Table 3.4 shows some typical values of T, P and H in different geo­
graphical regions and their influence on the breakdown voltages. From
Table 3.4 it can be seen that P, T and H play an important role in the
design of air insulation. For example, a certain air gap located in region 3
will breakdown at a voltage which is only —65% of the voltage needed
for the breakdown of the same gap when located in region 2 .
Wind has some influence on corona characteristics but in general has
little effect on breakdown voltage values provided the gap distance remains
unchanged. However, the wind can cause motion of transmission line con­
ductors and can bring the two phases of a transmission line closer, thereby
increasing the possibility of breakdown. Consequently, the sparkover volt-

1.0 "T 1— "1— I— r

1 \
m / 1
w 1 \

/ \
/ \

/ \
ij
/— '
0.5 !
m rw / \

/ \
i
/ \
/ \
.f
✓ \
__ 4 9 .J __^__L
g 3
Figure 3.6 The value of exponents m and n in equations (3.13) and (3.14) as per
lEC 60-1.
Air Insulation 61

Table 3.4 Examples of Influence of , P and H on Impulse Breakdown

Voltages of Rod-Plane Air Gap

T,(°C) P(mbar) H(g/m^) K, V, = KiK^V, Region

20 1013 11 1 1 1.0 V. 1. Standard

conditions
-3 0 1013 5 1.2 0.93 1.125 V3 2. Very cold,
dry area at
sea level
40 850 3 0.795 0.928 0.738 V, 3. Hot dry
area at high
altitude
40 1013 5 0.936 1.16 1.08 V, 4. Hot humid
area at sea
level
-3 0 850 5 0.94 0.92 0.865 V, 5. Cold dry
area at high
altitude

age can be considerably reduced by the wind. Hence, the influence of

swing angle on the switching and lightning impulse breakdown voltages
should be included in the transmission line design criteria.
The rain can also affect the breakdown voltage characteristics of open
air gaps. It can considerably reduce the breakdown voltages of open air
gaps employing large area electrodes. The rain droplets form sharp protru­
sions over such electrode surfaces, thereby changing the electrode field
configuration from a quasi-uniform field gap to close to a nonuniform field
point-plane gap, accompanied by an associated decrease of the breakdown
voltage.

3.5.6 Influence of Sand/Dust Particles

Airborne particles resulting from sand and dust storms may also influence
the insulation behavior of open air gaps. Most of the earlier studies on this
regard were confined only to small gaps that are of limited practical value.
However, recently authors [19-23] designed and used an environmental
chamber to simulate the natural sand and dust storms that are frequent in
the desert regions. Various types of voltages and electrode shapes were
employed in order to investigate the breakdown characteristics. The break­
down behavior of polluted rod-plane, rod-rod, sphere-plane and sphere-
sphere gaps were studied and typical results are summarized here. These
62 Chapter 3

findings are applicable to lightning impulses of =1000 kV and switching

impulses of <850 kV with H < 11 g/m^

Rod-Plane Gaps
The dust and sand particles slightly reduce the breakdown voltages of rod-
plane gaps under positive impulse voltages. The highest measured reduc­
tion for positive impulses was about 3%. However, the atmospheric pol­
lution has a major effect on the breakdown voltages of such gaps under
negative impulses. The effect depends on gap length, cathode radius (r)
and the impulse voltage waveform, and can best be summarized by Figure
3.7 for lightning impulses [20]. In this and the subsequent figures of this
section Vp = V5 0 value when the gap is contaminated with sand and dust
particles, whereas = V5 0 value for the same gap in the absence of any
contamination, i.e., clean gap. It is clear from this figure that in small gap
lengths, V5 0 decreases by up to 35, whereas in medium gaps V5 0 increases

Figure 3.7 The relative value of polluted gap V50 (Vp) to clean gap V 50 (V J as
function of gap length under negative lightning impulses for asymmetrical geom­
etries studies. (From Ref. 20 © IEEE, 1991.)
Air Insulation 63

by up to 75% due to atmospheric pollution. For larger gaps, atmospheric

pollution has minimal influence. The results also show that for most of the
gap lengths employed in power lines, the breakdown voltages under neg­
ative polarity impulses are either higher or practically similar to the values
observed under positive polarity impulses even under extreme particulate
pollution. Only in very short contaminated gaps, negative polarity has
lower breakdown voltages than the corresponding positive polarity values.
The outdoor substation equipment often uses large surface area electrodes
to minimize corona effects and/or to grade voltage distribution uniformly.
Such large area electrodes result in air gaps with quasi-uniform field dis­
tribution. As shown in Figure 3.7 particle contamination can significantly
reduce the negative lightning impulse breakdown voltages of such quasi­
uniform field gaps. Similar results were also found for switching impulses
[21]. Consequently, in polluted environment, the breakdown voltages under
negative polarity impulses may become even lower than the corresponding
positive polarity values. Therefore, in the selection of proper clearances
for station equipment, negative impulses may become more critical than
the positive ones, and if the equipment has to operate in an environment
where sand and dust storms are frequent, the influence of such contami­
nation must be carefully considered.

Rod-Rod Gaps
An important aspect of the insulation design of overhead lines and substa­
tion equipment is to ensure that the flashovers associated with overvoltages
are restricted to protective gaps. Rod gaps of various configurations are
widely used for this purpose. The most commonly adopted configurations
are either square cut or hemispherically terminated rod-rod electrodes.
The studies under lightning and switching impulses show that dust and
sand pollution has a considerable effect on the average breakdown voltage
gradients of rod-rod gaps. The magnitude of this effect is dependent on
polarity, electrode shape and gap length. Figure 3.8 compares the V5 0 val­
ues of rod-rod gaps under clean and polluted conditions when subjected
to lightning impulses. It is clear from this figure that the pollution effect
can be divided into three distinct gap regions: small, medium and large.
Small gap region exhibits pollution related reduction in V5 0 . In the medium
gap region, the pollution increases the V5 0 , and at larger gaps, pollution
has no significant influence on V5 0 . Similar results were found for switch­
ing impulses as well [21]. The gap range in which pollution displays severe
effects is confined to impulse voltage of <450 kV. Thus, the systems op­
erating at medium voltages of 33 and 6 6 kV are most susceptible to pol­
lution related deviations in the protective gap’s performance characteristics.
64 Chapter 3

Figure 3.8 The relative value of polluted gap V50 (Vp) to clean gap V50 (V J as
function of gap length under lightning impulses for rod-rod gaps. (From Ref. 19 ©
IEEE, 1991.)

It has also been observed that the square cut rod gaps have more immunity
towards the influence of contamination than the hemispherical rods [2 1 ].
For hemispherical and square cut rods, the V5 0 values under polluted
conditions do not vary by more than ± 2 % as compared to the clean gap
values, provided the ratio d/r is kept >30 where d = gap length and r =
rod radius. This is equally applicable for both polarities and both types of
impulse voltages. Similar to V5 0 studies for gaps where d/r is kept >30,
the scatter in the values of breakdown time (T^) is considerably reduced
for polluted gaps while the mean T^ values do not deviate more than ± 1 0 %
as compared to clean gaps T^ values.
In rod-rod gaps, most of pollution related effects can be attributed to
surface adhering dust particles. American standard CD801-1968 (1973)
gives V5 0 values for 20 ~ 2400 mm rod-rod gaps when subjected to light­
ning impulse with an accuracy of ± 8 %. Similarly results of V5 0 from
several of the European high voltage laboratories for rod-rod gaps give
differences as high as ± 10% [24]. Therefore, for longer gaps, changes of
up to ± 2 % caused by sand and dust storms can be considered practically
insignificant, and protective rod gaps can be safely designed for sand and
dust storm hit areas, based on the clean gap criteria, provided the square
cut rod electrodes are selected with d/r > 30.
Air Insulation 65

Sphere-Sphere Gaps
Sphere gaps are commonly used for measurements of peak values of high
voltages with a measurement accuracy of ±3%. The presence of sand/
dust pollution in the air gap or on the sphere surface can significantly
influence its breakdown behavior. Figure 3.9 displays the (Vp/V^) ratio as
a function of d/D for spheres of two different diameters where D = sphere
diameter [23]. This figure shows that for the gaps which are used in high
voltage measurements (d/D < 0.5) air pollution causes a reduction in
breakdown voltages. Similar results were also obtained for lightning im­
pulses. It is well known that certain sphere gaps should be adequately
irradiated to get reproducible breakdown voltages with an accuracy of
±3% [15,25]. Al-Arainy et al. [23] observed that if small diameter spheres
are clean but kept “hidden” from the light generated by the impulse gen­
erator spark gaps, the breakdown voltages can be up to 1 0 0 % higher than
those given in the standard tables [see Ref. 25]. If the gap receives enough
ultraviolet light, the presence of dust pollution does not have any major
influence on the breakdown voltages of such gaps. On the other hand, in
unirradiated gaps, the sand/dust pollution reduces V5 0 significantly.

3.6 VOLT TIME CURVE AND INSULATION COORDINATION

In order for an insulating media to breakdown the following requirements

are needed:
1. The existence of at least one free electron to start the ionization
process
2. Enough voltage to cause ionization and avalanche formation
3. Enough time to complete the breakdown process
In nature, free electrons are available from sources such as photo­
ionization and cosmic rays. The number of free electrons in a specific
volume is a statistical matter depending on many factors. It is well known
that the breakdown time lag depends on the amount of overvoltage. The
time lag can be defined as the time elapsed between the moment the voltage
reaches the breakdown level and the completion of the breakdown process.
Thus, the breakdown voltage value is related to the time taken for break­
down to materialize. This relation is called volt-time characteristic or, in
brief, V-t curve. This curve depends mainly on the insulating material,
voltage waveform and the electrode shape, in addition to pressure, humid­
ity, etc. The V-t curve for a high voltage device can be established by
66 Chapter 3

d/D

Figure 3.9 The relative value of polluted gap V50 (Vp) to clean gap V50 (V^) as
function of d /D under switching impulses (D = sphere diameter, d = gap length).

applying a standard impulse of fixed shape but with different peak values
and then finding for each application, the breakdown time and the highest
voltage to which the gap was stressed up to the breakdown instant. From
these data V-t curve can be constructed.
The importance of V-t curve comes when designing an overvoltage
protective scheme. Figure 3.10a shows a protective device in parallel with
the protected object. Their V-t curve must be in the fashion shown in Figure
3.10b. This means that the protective device, like the rod-rod air gap, will
always breakdown before the protected object such as insulator, bushing,
transformer, etc. gets damaged by the overvoltage. If the two curves inter­
sect, a protection coordination problem arises and the protective gap only
provides a partial overvoltage protection over a certain range of voltage
(or time).
In high voltage power networks, rod-rod air gaps or “arcing horns” of
various types are used for overvoltage protection of different pieces of
equipment such as transformers, bushings, insulators, etc. Since these are
usually located outdoors, they are subjected to the variations in atmospheric
parameters and pollution levels. In addition, the rod tips may be covered
by a film of fine sand and dust particles for extended time periods if the
rain is rare like in the desert areas. Several studies have been reported for
clean bi-rod air gaps and are summarized by IEEE Committee reports
[26,27]. As alternative to bi-rod air gaps, the multiple rod gaps employing
(three rod system) are also being used as protective apparatus for medium
Air Insulation 67

Protected
object

(a)

Figure 3.10 Insulation coordination arrangement: (a) protection schemes, (b) V-t
curves for good insulation condition.

and high voltage power networks in many parts of the world. These are
preferred, since they can be made to sparkover more consistently and in
lesser time than single long gap, and comparatively at a lower voltage.
They also interrupt power follow current with greater reliability [28].
In selecting gaps for overvoltage protection, the rod end profile, its tip
radius, gap length and gap configuration (e.g., single gap or multiple gaps)
and horizontal or vertical gaps are the main design parameters. For polluted
environment the choice should be of rod end shapes which show minimal
influence toward dust pollution. As dust pollution has similar affects for
single or multiple gaps as well as for horizontal and vertical gaps [29], the
main design parameters are the rod radius, its end profile and gap length.
For gap configurations employing rods of bigger diameters and smaller gap
lengths, there may be significant differences between the V-t curves under
clean and polluted conditions even for smaller breakdown time values as
shown in Figure 3.11. It is found that rods with square cut ends and smaller
diameters are preferable since these offer immunity toward dust pollution
68 Chapter 3

(a)

(b)

Figure 3.11 Influence of dust particles on V-t characteristics for multiple rod air
gaps (gap length = 2 cm + 2 cm).

related influences [29]. Therefore, the rod radius should be as small as

possible to minimize the pollution effect on the rod gaps overvoltage pro­
tection performance. However, it should not be so small that the gap is
always coronating under normal AC voltages.
Table 3.5 shows a typical example of a protective gap. This table in­
cludes the recommended gap lengths for single as well as multiple practical
rod gaps used on 13.8 kV and 33 kV systems located in semi-arid regions.
The rods are assumed to have cylindrical shape with square cut end profiles
and rod diameters are not greater than 8 mm. The extreme range of break­
down voltages shown in Table 3.5 includes the range of breakdown prob­
ability from 5% to 100%, while considering the temperature variations
from 0 to 50°C, pressure variation from 925 to 940 mbar and the influence
Air Insulation 69

Table 3,5 Recommended Gap Lengths for Single and Multiple Rod Gaps Used
Across Transformer Bushings Installed in Semi-Arid Region

Extreme Extreme
Nominal range of range of
system Recommended breakdown Recommended breakdown
voltage BIL single gap voltage multiple rod voltage
(kV) (kV) (cm) (kV) gap (cm) (kV)

13.8 95 6.5 58-83 1.5 + 1.5 64 -8 6

33 170 16 106-145 5.5 + 5.5 110-150

Source: Ref. 28.

of dust particles. Since the above variations will be different for different
regions, the protective gap spacing for each regions need to be adequately
adjusted.

3.7 PHASE TO PHASE BREAKDOWN CHARACTERISTICS

In the early days of UHV line design, the phase to phase spacing was
determined by corona and not by the flashover voltage requirements. The
higher transmission voltages have resulted into higher possible switching
surges. In addition, the economizing of the transmission line design has
reduced the interphase clearances. The air clearance between two phases
of a UHV line are in the range of several meters and the breakdown pro­
cesses in such long gaps have some differences with those corresponding
to short or medium gaps.
The discharge associated with positive polarity switching impulses for
such long gaps may be divided into three stages: corona formation, leader
propagation and the final jump [30]. In most practical geometries, where
the field is very divergent, corona inception voltage is much lower than
the breakdown voltage. The space charge resulting from corona formation
significantly influences the gap’s electric field value. Depending on the
applied voltage waveshape, the ionization may continue for some time
(dark period), where at the end of this period a secondary (burst) corona
occurs. At the positive electrode, corona is generally followed by the ini­
tiation of a highly ionized channel called leader. The leader’s behavior in
the gap depends on the geometry of the electrodes and on the shape of the
applied voltage. If the voltage is not sufficient, the leader stops propagating
and the gap does not breakdown. However, if the voltage is high enough.
70 Chapter 3

the leader approaches the grounded electrode where its velocity suddenly
increases and with the final jump a complete breakdown of the gap takes
place. The critical time to crest (t^j.)’ discussed earlier, is directly linked to
the duration of the leader propagation before the final jump [5]. In this
regard (t^j,)“ < (t^j.)^ since negative leader propagates faster than the pos­
itive one. Also t^^ increases linearly with gap length and is influenced by
the electrode shape, especially the cathode.
In most cases the breakdown probability of long gaps under positive
switching impulses can be approximated by normal distribution function.
However, in cases where two or more modes of breakdown may occur,
abnormal breakdown probability distribution may take place [30]. Time to
breakdown Tg^ also has, in general, normal distribution for long gaps since
the leader is dominant at such gaps. Abnormal distribution may occur when
d < 5 m where the leader corona inception time lag is significant in pro­
portion to Tgj) [30].
The interphase stress is more complex than between one phase and the
ground due the unpredicted time and locations of surges at the two phases.
The strength of phase-to-phase insulation is function of the total voltage
between the phases and the individual phase-to-ground voltage [3]. The
phase-to-phase insulation is a system of two energized electrodes and the
ground to which these two different voltages to ground are applied. Phase-
to-ground (earth) breakdown voltage is always higher than the phase-to-
phase flashover voltage, thus the former is protected by the latter. This is
because the spacing between phases and the ground is higher than the
interphase gap. In addition the magnitude of overvoltage between phase to
phase is 1.4 to 1.8 times higher than the voltage between a phase and the
ground [31].
In studying the phase to phase breakdown, the relative values of pos­
itive (V^) and negative (V“) impulses on the two phases is important. Let
a be:
|V-
(3.18)
|V"| + |V-
Assuming the two impulses are synchronized then the breakdown voltage
increases approximately linearly with a [32]. Generally a ranges from 0.3
to 0.7, but for practical purposes, its range can be limited to 0.3 to 0.5
because if a > 0.5 then V“ > and it is not of much interest to the
system engineer since — Vio-
The relative occurrence time of the two impulses in the opposing
phases is important since the interphase withstand voltage can be reduced
appreciably if the negative surge of V“ proceeds the positive surge of
[31]. This variable time delay is denoted by At = t^ - t^ (when the neg-
Air Insulation 71

ative surge preceeds the positive surge). The gap insulation is strongly
influenced by the space charge created during the negative impulse appli­
cation. The pre-existing negative space charge field assists the positive
leader inception and propagation. For 0.2 < At < 300 ms, the breakdown
voltage is lower than when only the positive impulse occurs.
In addition to the single surge or double surges of “standard” shape
discussed above, there exist in practice other types of surges such as double
or multiple peak surges and irregular surge shapes which result from sud­
den waveshape changes during the propagation of the surges [31]. The
detailed knowledge of the above subjects are important when designing
and operating live line maintenance schemes [33].
The bundles in the phases can be arranged in optimal way to strengthen
the phase to phase insulation strength. Alexandrov et al. [32] found that
the bundle optimization can result in average interphase dielectric strength
of 5.4 kV/cm and 4.5 kV/cm for interphase spacings of 1 m and 8 m,
respectively.

3.8 ARC DISCHARGE

Spark is defined as a transient plasma channel which bridges two electrodes

separated by an insulating gas. If this spark is continuous, then it is called
an arc discharge. Arc discharge produces very intensive heat and light and
thus it is useful in some industrial applications such as heating, welding
and arc discharge lamps. In power systems, the arc discharge is generally
undesirable. However, it does take place during circuit breaker opening
and during short circuits. There are many types of arc discharges which
can be classificated according to the method of electronic emission from
cathode or according to gas pressure used. In power systems, the arc is
generally classified according to the conditions associated with the arc such
as:
1. High short circuit current arc
2. Short gap arc, where arc column is effected by the electrodes
3. Long gap arc, where the arc is independent of the electrodes such
as in a lightning discharge
When analyzing a network containing an arc an accurate equivalent
circuit should be used to represent the arc. One way of representing an arc
is shown in Figure 3.12 where arc resistance R, inductance L, and the back
voltage E are functions of arc length, arc current and the rate of current
change with respect to time [34]. The above parameters are generally found
experimentally.
72 Chapter 3

Stray
Cap/

(b)

Figure 3.12 Arc equivalent circuit: (a) DC and AC arc during burning period, (b)
AC arc during the extinction period. (From Ref. 35.)

Circuit breakers are designed to interrupt the AC current at the moment

of current zero. Such an interruption leads to transient recovery voltage
TRV across the open breaker contacts. If TRV is high, arc may reignite
and the breaker may fail to interrupt the current. Usually the arc quenching
process is affected by the heat generation and heat dissipation rates. In an
arc, the heat is dissipated mainly by diffusion. The heat dissipation can be
increased by arc movement caused by electromagnetic forces. This tech­
nique is employed in the interruption of arcs in circuit breakers, gap-type
surge arresters and chopping gaps or arcing horns. Another technique
which is successfully used in air blast circuit breakers to interrupt arcs is
the use of air blast to elongate the arc and increase the arc heat losses.

3.9 UNDESIRABLE EFFECTS OF CORONA

Corona in air has the advantage of attenuating the high voltage surges
propagating along the transmission line since the transmission line is usu­
ally highly coronating at such high voltages. However, corona on power
network has major undesirable effects such as corona loss, audible noise,
radio interference and chemical decomposition of air. A brief discussion
of some of these effects is presented below.

3.9.1 Corona Loss

In transmission lines, the copper losses (PR) at rated current ranges be­
tween 20 and 200 kW/km. The fair weather corona loss which is inde-
Air Insulation 73

pendent of load current is generally designed to be around 10% of the PR

losses. In foul weather, the corona loss increases by at least ten-fold. Beside
the weather influence, the corona loss also depends on many other factors
such as the difference between the operating voltage (V) and corona onset
voltage V^, i.e., (V-V^), conductor surface gradient, conductor surface
smoothness, system frequency and conductor sizes. Fair weather corona
loss can be estimated based on experimental results of similar lines. Corona
loss can be neglected if the conductor gradient is low, and the conductors
have acceptable audible noise performance. EPRI proposed the following
equation to determine foul weather total local corona loss Pl for three
phases (in kW/km) for the UHV range [3]:

Pl Ppw Jr2 ln(l + lOR) (3.19)

where Pp^ = total three phase fair weather corona loss (kW/km), V ==
rms line voltages (kV), J = loss current constant (7.04 X 10“^^ for 400
kV lines; 5.35 X 10"^® for 500 and 700 kV lines), r == conductor radius
(cm), n = total number of such conductors in the bundle, E; = maximum
gradient for subconductor i and R = rain rate (mm/h).
In some special cases such as very dirty transmission line conductors
or transmission line passing through forest having high population of in­
sects, the corona loss may become a significant part of the total power
delivered.

3.9.2 Audible Noise

A broad band audible noise (AN), in the human hearing range, is generated
during the corona discharge. In addition to the lateral distance from the
transmission line, audible noise level depends on the factors mentioned
above for the corona loss. Generally audible noise is negligible during fair
weather, but during heavy rain it could reach above 60 dB at locations
close to the transmission line and it could become very annoying. It is a
normal practice to limit the AN at the right of way to around 52 dB. The
following empirical formula can be used to calculate the AN level for 3
phase bundled conductor lines during the heavy rain [3]. The AN level is
given by dB with 20 juiPa. sound pressure level as the reference:
AN = 20 log n + 44 log D, - 665/E

+ K„ - 1 0 log - 0 .0 2 Dj^ + An + B„ 22.9(n - 1) ^ (3.20)

where n == number of subconductors in the bundle, = subconductor

74 Chapter 3

diameter (cm), D = bundle diameter, E = conductor surface gradient

(kV/cm), Djj^ = distance from line to the measurement point (m), K„ =
7.5 for n = 1, K„ = 2.6 for n = 2 and K„ = 0 for n > 3, == 75.2 for
n = 1 and 2 and Aj^ = 67.9 for n > 3, and == 0 for n = 1 and 2 and
Bxt = 1 for n > 3.

3.9.3 Radio Interference

The current produced by the pulsative corona is responsible for electro­
magnetic interference (EMI). The current injected in the transmission line
conductor due to corona can be approximated by the following analytical
expression [30]:
i(t) = Ue-/^- -t/ T-,
^) (3.21)
where Tj and T2 are the front and tail time constants, respectively. The
maximum pulse amplitude 1^^ is about 0.6 Iq. The approximate values of
Ti, T2 and for the negative and positive polarity corona pulses are shown
in Table 3.6. These current pulses may be represented by a Eourier fre­
quency spectrum with spectral intensity 8 ( 0)) given as:

S(w) r i(t) e“j^' dt (3.22)

When the corona pulse is injected into the line, it splits into two equal
parts which propagate along the conductor on both sides of the injection
point. Each spectral component of such pulses will behave differently de­
pending upon its wavelength. In the frequency range of 0.15 to 30 MHz,
the direct electromagnetic radiation from the corona pulses does not con­
tribute much to the EMI which is caused primarily by the propagation
along the conductor of various spectral components of the current pulses.
When the wavelength of a spectral component is long, a system of two
orthogonal fields, one electrical and one magnetic, associated with this
spectral current component propagate along the line. This constitutes a
guided plane wave with a relatively low attenuation. Therefore, the inter-

Table 3.6 Corona Current Parameters

Corona pulse polarity T ,(n s) T2(ns) I„(mA)

Negative 6 45 2.7
Positive 30 180 60
Air Insulation 75

ference field is dominated by the aggregation of the effects of all discharges

spread over some tens of kilometers on both sides of the measuring loca­
tion. For spectral components above 30 MHz, the wavelengths are close
to, or less than the line clearances and the noise propagation is primarily
by radiation.
The quality of radio reception depends upon the signal to noise ratio.
To characterize this ratio correctly, it is essential to define noise by a meas­
urable quantity. In general, the instantaneous intensity, S(t), of corona gen­
erated noise varies continually and in an erratic manner; but if its average
energy for a long time, e.g., one second is constant, then this noise is
called stationary random. The RMS value (RN) of the part of noise con­
tained in a narrow frequency band Aco centered at co^ (rad/s) is expressed
as:

S^(co) Aco
RN = (3.23)
2 tt
Therefore, a radio noise measuring set is basically a selective voltmeter
characterized by a passband with bandwidth equal to {AodU tt) which can
be tuned to a center frequency f^ = {ojJ I tt). The radio noise is usually
measured in the 0.15-30 MHz frequency range. The measured noise level
is proportional to the square root of the bandwidth. Different types of
detector weighing circuits such as average, peak, and quasi peak are used
for RI measurements.
In radio noise meters complying with international standards [35,36],
it is preferred to express the “quasi-peak” value rather than average, peak
or true rms value. This type of detector leads to a more realistic measuring
device as it represents the psychological effects of the EMI experienced
by the listener. The quasi-peak (QP) detector is basically a diode which
charges a capacitor placed in parallel with a resistor. When noise is applied
to this device, after filtration by the passband of the receiver and suitable
amplification, the voltage on the capacitor floats at a value a little lower
than the peak value of the noise signal. The charge time constant of the
detector is ~ 1 ms whereas the discharge time constant is either 160 ms or
600 ms as per ANSI or lEC (CISPR) detector specifications, respectively.
Based upon extensive EMI data from a large number of lines, a simple
formula was proposed by CIGRE which has a good applicability to many
types of lines [35]. The range of parameters for which CIGRE formula
was derived are as follows:
Nominal line voltage: 200 - 765 kV,^,
Maximum electric gradient: 12 to 20 kY^^Jcm
Conductor radius: 1 to 2.5 cm
76 Chapter 3

Number of subconductors per bundle: 1 to 4

Subconductor spacing: 10 to 20 times the conductor diameter
This formula gives a noise figure for the most probable interference level
using CISPR detector in dB above 1 /xV/m, for dry, aged and moderately
dirty line conductors at 2 m above the ground level at a horizontal distance
of 15 m from the external conductor, at a measurement frequency of 0.5
MHz. In this method, first the interference level due to each phase is cal­
culated as:

NPj = 3.5 g^, + 12r, - 33 log - 30 (3.24)

2 0

p,
NP2 = 3.5 g^, + 12r, - 33 log - 30 (3.25)
2 0

NP3 = 3 .5 g^ 3 + 1 2 r3 - 33 log ( ^ - 30 (3.26)

where g^j = maximum conductor gradient at the surface of phase i and D,

= the distance between phase i and the reference measuring point.
Then the summation of these three fields is made in the following way.
If one of the fields is at least 3 dB greater than the others, then the other
fields are neglected. Otherwise, we use:
NP, + NPb
NP = — + 1.5 dB (3.27)

where NP^ and NP^ are the two highest among the 3 values obtained from
equations (3.24), (3.25), and (3.26).
For double circuit lines, the interference fields produced by each of
the 6 conductors are calculated at the measuring point. The fields produced
by the phases corresponding in time are added quadratically and then the
summation is made on the three resulting fields.
To obtain the noise level NP at a frequency different from 0.5 MHz,
it is sufficient to apply the correction given by equations (3.28) or (3.29)
to the value of NP„, i.e., NP calculated at 0.5 MHz as follows:
Four double circuit and triangular lines:
NP = NP, - (18 log F + 10 log" F + 4.3) (3.28)
For horizontal lines:
NP = NP, - (23 log F 4- 12 log" F + 5.8) (3.29)
Air Insulation 77

Figure 3,13 Predictions of transmission line EMI for average types of weather.
(From Ref. 41 © IEEE, 1973.)

Likewise, the interference level at a lateral distance D other than the

reference distance of 15 meters, i.e., D^, will be obtained by:

NP = NP, - 20 Kj log (3.30)

where Kj = 1.4 for triangular, vertical and double circuit lines, and Kj =
1.6 to 1.9 for horizontal lines depending upon the voltage level.
The EMI level, for atmospheric conditions other than mean dry fair
weather, can be estimated using Figure 3.13. Figure 3.14 shows the mea­
sured RI levels from 380 kV lines operating at =12.1 kV/cm located
in a semi arid dry region [37].
The radio noise problems are not merely restricted to extra and ultra
high voltage transmission lines. In many instances distribution lines oper­
ating at medium high voltages and much reduced surface gradients can
78 Chapter 3

Figure 3.14 Frequency spectra of EMI from 380 kV lines (maximum conductor
gradient = 12.1 kV/cm ). (From Ref. 37 © IEEE, 1989.)

F(MHz)

Figure 3.15 All weather frequency spectra for EMI radiated from 33 kV distri­
bution lines (based on over 100 measurements made under different lines). (From
Ref. 38 © IEEE, 1989.)
Air Insulation 79

Figure 3.16 Frequency spectra for the gap-type discharge generated from geom­
etry shown in Figure 2.14.

Figure 3.17 Frequency spectra of EMI from a 132 kV line (maximum conductor
gradient = 14.4 kV/cm ). The measurements were made near a substation (upper
curves) and 300 m away from the substation (lower curves). (From Ref. 37 © IEEE,
1989.)
80 Chapter 3

also exhibit significant levels of radio interference. The possible noise

sources on such lines are corona from hardware and poorly constructed
joints, gap-type discharges and corona on insulations. Since the height of
such lines is usually much smaller as compared to EHV and UHV lines,
the listener is “closer” to the “noise source,” Figure 3.15 (page 78) shows
typical frequency spectrum of radio noise level measured on 33 kV distri­
bution lines located in arid dry regions [38].

3.10 TELEVISION INTERFERENCE

The outdoor high voltage network may cause severe television interference
(TVI) at nearby locations. TVI is not caused by corona but by the gap-
type discharge as discussed in chapter 2. The spectrum of the generated
noise is not well defined like RI from corona, but varies significantly de­
pending on geometrical and operating conditions [39]. The level and band­
width of the gap-type discharge interference depends mainly on voltage
level, microgap dimensions and impedance of the circuit external to the
gap discharge [39,40]. Figure 3.16 (page 79) shows noise generated from
a gap-type discharge model [25] using peak detector [40]. Outdoor mea­
surements show similar results. As an example. Figure 3.17 (page 79)
shows FMI measurements using a quasi peak detector for a 132 kV double
circuit line located in semi-arid land [37].

REFERENCES

1. L. Paris, IEEE Trans., Vol. PAS-86, No. 8, pp. 936-947, 1967.

2. Y. Aihora, IEEE Trans., Vol. PAS-97, No. 2, pp. 342-347, 1978.
3. T r a n s m is s io n lin e R e fe r e n c e B o o k , EPRI, California, 1982.
4. K. Schneider and N. Week, Electra, No. 35, pp. 25-48, 1974.
5. CIGRE Task Force, “Guidelines for the Evaluation of the Dielectric Strength
of External Insulation” , CIGRE 33.92, (WG-07) 2 IWD, Paris, France, 1992.
G. Gallet, G. Leroy, R. Lacey, and I. Kromer, IEEE Trans., Vol. PAS-94, No.
6, pp. 1989-1993, 1975.
7. E. Lemke, Z. Electr. Inform. Energetechnik, Leipzig, Germany, Vol. 3, No. 4,
pp. 186-192, 1973.
8. R.T. Waters, “Spark Breakdown in Nonuniform Field” , in E l e c t r i c a l B r e a k ­
d o w n o f G a s e s , by J.M. Meek and J.D. Craggs (eds.), John Wiley & Sons,
New York, pp. 510-513, 1978.
9. G.N. Aleksandrov, Zumal Tecknicheskoi Fiziki, Vol. 9, pp. 744-756, 1969.
10. A. Pigini, G. Rizzi, R. Barmbilla, and E. Garbagnati, “Switching Impulse
Strength of Very Large Gaps” , 3rd ISH Milan, Paper No. 52.15, 1979.
Air Insulation 81

11. F. Rizk, IEEE Trans, on Power Delivery, Vol. 4, No. 1, pp. 596-606, 1989.
12. I. Kishizimask, Matsumoto and Watanabe, IEEE Trans., Vol. PAS. 103, No.
6, pp. 1211-1216, 1984.
13. L. Thione, Electra, No. 94, 1983.
14. M. Dietrich, J. Wolf, E. Lemke and J. Kurcera, “Influence of the Tail Duration
on the Positive Switching Impulse Breakdown of Large Air Gaps” , Proc. 4th
ISH, Athens, Greece, 1983.
15. IEC-60-1,2,3,4, “High Voltage Testing Techniques” , 1989.
16. “IEEE Standards Techniques for High Voltage Testing” , IEEE Std. 4-1978.
17. A. Albert, IEEE Trans., Vol. PAS-100, No. 7, pp. 3666-3672, 1981.
18. P. Zacke, IEEE Trans., Vol. PAS-96, No. 2, pp. 701-708, 1977.
19. M.I. Qureshi, A. Al-Arainy and N.H. Malik, IEEE Trans, on Power Delivery,
Vol. 6, No. 2, pp. 706-714, 1991.
20. A. Al-Arainy, N.H. Malik and M.I. Qureshi, IEEE Trans, on Elect. Insul., Vol.
27, No. 2, pp. 193-206, 1991.
21 . M.I. Qureshi, A. Al-Arainy and N.H. Malik, IEEE Trans, on Power Delivery,
Vol. 8, No. 3, pp. 1045-1051, 1992.
22 . A.A. Al-Arainy, N.H. Malik and M.I. Qureshi, IEEE Trans, on Dielectrics
and Elect. Insul., Vol. 1, No. 2, pp. 305-314, 1994.
23. A.A. Al-Arainy, N.H. Malik and M.I. Qureshi, The Arabian Journal for Sci­
ence and Engineering, Vol. 20, No. 3, pp. 495-509, 1995.
24. H. Batz, “Comparative Impulse Tests with Impulse Voltage on Rod Gaps” ,
CIGRE Report No. 325, Paris, France, 1962.
25. lEC Publication 52, “Recommendation for High Voltage Measurement by
Means of Sphere-Gaps (One Sphere Earthed),” Geneva, Switzerland, 1960.
26. IEEE Committee Report, IEEE Trans., Vol. PAS-86, pp. 1432-1437, 1967.
27. IEEE Working Group Report, IEEE Trans., Vol. PAS-93, pp. 196-205, 1974.
28. Ohio Brass Company, “Hi-Tension N ew s” , Vol. 48, No. 9, pp. 2 -3 , 1979.
29. A.A. Al-Arainy, Journal of King Saud University, Engineering Science, Vol.
9, 1997.
30. Les Renardières Group, CIGRE Publication No. 53, pp. 31-153, Paris, France,
1977.
31 Special issue on “UHV Air Insulation; Physical and Engineering
Research— Part 1” , lEE Proceedings, Vol. 133, Part A, No. 7, 1986.
32. G. N. Alexandrov, G.V Podporkgn, Yu. G. Seleznev and A.D. Sivayev, 4th
ISH, Athens, Greece, Paper No. 44.11, 1983.
33. WG No. 7 of Committee 33, D i e le c t r i c S tr e n g th o f E x te r n a l I n s u la tio n S y s ­
te m s U n d e r L iv e W o rk in g , CIGRE, Paris, 1994.
34. M. Khalifa, (ed.). H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c tic e , Marcel
Dekker, New York, 1990.
35. CIGRE Committee Report, “Interference Produced by Corona Effects of Elec­
tric System: Description of Phenomena, Practical Guide for Calculation” , CI­
GRE (Electra), Paris, France, pp. 89-97, 1971.
36. ANSI, “Specifications for Radio Noise and Field Strength Meters 0.015 to
30 m egacycles/second” . New York, 1963.
82 Chapter 3

37. A. Al-Arainy, N.H. Malik and L.N. Abdulal, IEEE Trans, on Power Delivery,
Vol. 4, No. 1, pp. 532-538, 1989.
38. N.H. Malik and A. Al-Arainy, IEEE Trans, on EMC, Vol. 31, No. 3, pp.
273-279, 1989.
39. W. Janischewskyj and A. Al-Arainy, “Statistical Characteristics of Microgap
Discharge” , U.S.-Japan Seminar on EMI in Highly Advanced Social Systems,
Honolulu, August 1988.
40. A. Al-Arainy, “Laboratory Analysis of Gap Discharge on Power Lines” ,
Ph.D. thesis, University of Toronto, Canada, 1982.
41. IEEE Radio Noise Subcommittee Report, IEEE Trans, on PAS, Vol. 92, pp.
1029-1042, 1973.
4
SFg Insulation

4.1 INTRODUCTION

It is well known that heavy gases belonging to the seventh group of the
periodic table (fluorine, chlorine, etc.) have a considerably higher dielectric
strength compared to air under similar experimental conditions. The high
breakdown strength depends mainly on their capability of taking up free
electrons, thereby forming heavy negative ions. Gases having these prop­
erties are called electronegative. Of the many available electronegative
gases, sulfur hexafluride, SF^, has especially gained importance because of
its chemical stability as well as its high breakdown strength. It is therefore
the dielectric gas of choice. Furthermore, due to its effectiveness in the
extinction of arcs, it is used extensively in circuit breakers as well. This
chapter provides a comprehensive review about the basic properties, die­
lectric characteristics and applications of SF^ insulation.

4.2 BASIC PROPERTIES OF SFg GAS

SF^ is a colorless, odorless, nontoxic, nonflammable and inert gas. Some

of its important physical properties are summarized in Table 4.1 [1]. At
about 14 bars, SF^ liquifies at 0°C; whereas at -40°C, the gas liquifies at
about 3.5 bars [1]. Thus, in extremely cold regions, SF^ liquefaction is a
potential problem if high gas pressures are used in such regions.

83
84 Chapter 4

Table 4.1 Physical Properties of SF^

Molecular weight 146.6

Melting point -50.8°C
Liquid density (at 25°C/50°C) 1.33/1.98 g/m l
Gas density (at 1 bar and 20°C) 6.2 g /1
Relative density 5.1
Critical density 0.735 g/m l
Critical temperature 45.6°C
Boiling point -63.0°C
Sublimation temperature -63.9°C
Critical pressure 35.56 bar
Specific heat (at 30°C) 0.599 J/g
Thermal conductivity 0.1407 W/m°C
Dielectric constant 1-1.07
Vapor pressure (at 20°C) 10.62 bar

At atmospheric pressure, SF^ is chemically stable up to at least 500°C.

At higher temperatures, it can decompose into several species. The gas
itself is inert and does not cause corrosion to most common metals used
in electrical equipment such as copper, aluminum and steel within the usual
temperature range of operation of such equipment. However, SF^ will un­
dergo some degree of decomposition and oxidation in an electrical dis­
charge, particularly when oxygen and water vapor are present. Some of
the byproducts of SF^ decomposition are quite toxic or corrosive, and there
has been increasing concern about the influence that these byproducts can
have on system reliability and safety. When SF^ is dissociated in an elec­
trical discharge, the products of dissociation tend to recombine at a rapid
rate to reform SF^ and thus the dielectric strength of the gas quickly re­
covers to its original level. It is this characteristic of SF^ that helps to make
it a good arc-interrupting medium.
SF^ decomposition in electrical discharges (see Table 4.2 for some
reactions) produces species such as SF5 , SF4 , S2 F 1 0 , SOF2 , SOF4 , SO2 , HF,
S2 OF2 , SF2 , S2 F2 , SFg, SiF4 and metal fluorides. Some of these products
are formed under high power arc discharges [3] while others are produced
even under low power corona discharges in the presence of oxygen and
water [2]. SF^ itself is nontoxic and a human can survive indefinitely in a
mixture of 20% O2 and 80% SF^. However, SF^ will not support life and
can cause suffocation. Being extremely dense, it will accumulate in low-
lying areas, requiring care if being exhausted to atmosphere. Moreover,
some of the discharge byproducts of SF^, such as S2 F,o, SOF4 , and HF,
SF^ Insulation 85

Table 4.2 Some Possible Reactions in

SF,

SF^ + e SF4 + 2F + e
SF^ + e SFo + 4F + e
SF^ + e SF5 + F + e
SF5 + SF5 S2F'i O
SF4 + H2O SOF2 + 2HF
SF4 + O SOF4
SOF2 + H2O SO2 + 2HF

can be very toxic, having working day exposure limits in the range of a
few ppm. Therefore, it is prudent to assume that any SF^ equipment which
has been in service for some time may contain some toxic species, thereby
requiring extreme care in its handling. It is generally believed that if mois­
ture is absorbed by molecular sieves and oxygen can be prevented from
being present in SF^ equipment, the metal fluorides which constitute the
main discharge byproducts are relatively harmless [4]. Thus chemical ad­
sorbents such as sodalime and activated alumina are used in some SF^
equipment to keep the harmful byproduct concentration within tolerable
limits.
In general SF^ has good heat transfer characteristics. Considering spe­
cific heat, thermal conductivity, molecular weight, and viscosity of SF^, it
can be shown that SF^ has better heat dissipation than that of air. For a
cooling gas flow at the sonic velocity, as is typical in circuit breakers, SF^
has a convective cooling efficiency of about four times that of air. Thus,
SF^ has superior cooling characteristics in an arc environment.

4.3 BREAKDOWN PROCESSES IN SF«

As discussed in chapter 2, a gas is normally an almost perfect insulator.

Like other gases, the primary ionization process in SF^ is by electron/gas
molecule collision and is represented by the Townsend first ionization co­
efficient a. The most important secondary processes in SF^ are photoion­
ization of gas, photoemission from electrodes and thermal ionization in
case of arcs. In SF^, negative ions are formed by the direct or dissociative
attachment processes. The attachment processes are described by the elec­
tron attachment coefficient rj.
86 Chapter 4

Both coefficients a and t] are strongly dependent on the applied electric

field E (in kV/cm) and gas pressure P (in kPa). The net ionization coef­
ficient, a = a - 7], can be expressed as [5]:
a
= K (cm kPa) (4.1)
P
where K = 27 and (E/P),,^^ = 877.5 V (cm kPa)" Here is the critical
or limiting field strength. If applied field E > E^^it ionization
will take place leading to discharges. On the other hand, if E < E^j.jp then
a < 7] and ionization and consequently discharges cannot occur. Therefore,
from theoretical considerations, electrical breakdown in SF^ gas should not
occur if the applied electric field is less than E^^jj = (E/P)^j.it * P- Conse­
quently, SF^ has a dielectric strength of ~ 90 kV/cm at 1 bar. At a pressure
of 3.5 bars, SF^ will have a dielectric strength of more than 300 kV/cm.
For pressures of technical interest, electrical breakdown in SF^ occurs
as a result of streamer formation. In the streamer mechanism (see chapter
2 for details), it is assumed that the growth of a single electron avalanche
becomes unstable before reaching the anode, which results in the formation
of streamers. The field in a streamer is —E^^if nonuniform field gaps,
strong streamers can transform into highly conducting leader channel with
electric field of only a few kV/cm, which ultimately causes the collapse
of the applied voltage. Mathematically, the streamer breakdown criterion
as proposed by Pedersen [6] can be expressed by equation (4.2).

(a - 7])dx = ln(NJ= M (4.2)

r
where is the critical number of ions in an avalanche which has travelled
a distance x^ (i.e., the critical avalanche length), in the direction of the
applied field when it transforms into a streamer. M is a constant which is
considered to have a value between 10 and 20 [7]. The gas volume where
an avalanche can grow into a streamer is called the critical volume. Thus,
the applied field should be such that equation (4.2) is satisfied for streamer
formation and hence for breakdown to occur. A critical analysis shows that
equation (4.2) can predict measured breakdown voltages in a uniform field
gap with good accuracy in the range of 0.6 < Pd < 100 kPa cm [7].

4.4 UNIFORM FIELD BREAKDOWN

To obtain minimal insulation spacing in SF^ equipment, wherever possible,

uniform or slightly non-uniform field gaps with field utilization factor
SF^ Insulation 87

(juif = greater than 0.2 are used. Therefore, uniform field break­
down data are very important. Uniform field breakdown characteristics are
normally represented by the Paschen curve and the ranges in which Pas-
chen’s law is valid. Paschen’s curve for SE^ gas for AC voltages of 50 or
60 Hz [8] and DC applied voltages [5] have been reported and can be
approximated by the equation [9];
Vb - 1.321 (Pd)"" (4.3)
where the breakdown voltage is in kV and Pd is in kPa cm. At pressures
above 200 kPa, deviations from Paschen’s law are generally observed [5].

4.5 NONUNIFORM FIELD BREAKDOWN

Whereas in uniform and moderately nonuniform field gaps an electron

avalanche usually results in the formation of a streamer which leads to a
spark breakdown of SF^, in highly nonuniform field gaps the situation is
somewhat different. In this case, the field near the sharp electrode is much
higher than its corresponding value in the rest of the gap. Therefore, the
resulting streamers are confined only to the highly stressed region near the
sharp electrode causing corona discharges. Thus, the steamer breakdown
criterion gives corona onset voltage in highly nonuniform field gaps
whereas the breakdown spark takes place at a higher voltage. This type of
breakdown is called corona-stabilized breakdown. As the gas pressure is
increased, the breakdown mode changes and at a certain pressure, the first
avalanche can transform into strong enough streamers (and a leader dis­
charge) which can lead to a spark breakdown without any stable corona.
This type of breakdown is called direct or corona-free breakdown. These
phenomena lead to nonlinear breakdown voltage-pressure characteristics
which are influenced by the radius of the sharp electrode, the gap sepa­
ration, the voltage polarity as well as the voltage waveform. Figure 4.1
shows breakdown and corona onset voltage characteristics for a rod-plane
gap under positive and negative direct voltages. Impulse and AC applied
voltages also produce such nonlinear breakdown voltage-pressure charac­
teristics. Similarly, practical coaxial electrodes with metallic contaminating
particles also exhibit this type of breakdown behavior. Under certain ex­
perimental conditions, the transition from corona-stabilized to corona-free
breakdown is very sudden [10]. Moreover, the pressure where such a tran­
sition occurs is strongly influenced by the gas composition [11]. Another
important factor which can influence the nonuniform field breakdown char­
acteristics is voltage waveform. Thus, the breakdown voltage characteris­
tics of practical systems are influenced by several parameters and an in-
88 Chapter 4

Figure 4,1 Effect of SF^ gas pressure on the breakdown voltage behavior o f rod-
plane gap under direct voltage. (From Ref. 10 © IEEE, 1979.)

crease in gas pressures does not always result in an increase in the

breakdown voltage. The mechanism of leader formation and breakdown in
nonuniform electrode system with stress enhancement is described next
[ 12].
If a small stress enhancement on the conductor results in a small crit­
ical volume of gas stressed above —90 kV/cm-bar and a free electron
occurs within that volume, streamers will be generated and fill the critical
volume (see Figure 4.2a). Once the volume is filled, discharge activity stops
and the positive and the negative ions separate in the field. As the field at
the boundary of the critical volume was, by definition, the maximum sus­
tainable in the gas, any ion separation at this boundary results in an in­
crease of the field above the critical value. Thus charge separation results
in a breakdown from the critical volume boundary back to the stress en­
hancement. This breakdown develops into a leader, with a field along its
length of only a few kV/cm (as opposed to 90 kV/cm-bar for a streamer).
Thus, the leader represents an extension of the stress enhancement into the
volume of the gas (Figure 4.2b). The end of the leader creates a new critical
SF^ Insulation 89

(a)

(b )

(c)
Figure 4.2 Corona discharges in SF^-N2 mixtures: (a) streamer corona; (b)
streamer and leader corona; and (c) streamer and stepped leader corona.
90 Chapter 4

volume which fills with streamers, and the process repeats. Through this
stepwise breakdown process (Figure 4.2c), a small stress enhancement
which creates only a small critical volume can lead to total breakdown of
the system [13,14]. Figure 4.2 shows still photographs of streamer, leader
and stepped leader corona discharges under positive impulse voltages in
SF6 -N2 gas mixture.

4.6 ESTIMATION OF MINIMUM DISCHARGE VOLTAGES

The estimation of minimum discharge voltages, i.e., corona inception or

breakdown voltage, is of considerable importance from engineering point
of view. A method capable of predicting discharge voltage with reasonable
accuracy can save considerable amount of time and money required in the
design and development of new SF^ equipment and can provide some in­
sight into the insulation behavior of SF^ gas. Existing methods of discharge
voltage calculations can broadly be classified into two categories, namely:
those based on streamer breakdown theory and those based on concept of
critical field strength. Both methods give reasonably close results to ex­
perimental values in most cases and are summarized here.
The streamer criterion, as given in equation (4.2), offers a very prom­
ising approach for the solution of engineering problems. This criterion
states that the streamer formation results in partial or total breakdown of
the insulation depending on the degree of nonuniformity of the electric
field in the gap. In a uniform field gap, streamer formation will directly
lead to a breakdown while in a nonuniform field it may cause a corona or
a complete breakdown. Application of this criterion shows that the mini­
mum discharge voltage (in kV) in SF^ insulated systems can be ex­
pressed as [15]:
Vb - 0.8775 AifSfQ Pd (4.4)
where ptf = the field utilization factor; P = gas pressure in kPa; d = gap
length in cm; < 1 is the electrode surface roughness factor, which de­
pends upon the conductor surface conditions and gas pressure and can be
determined from experience or experiments; and > 1 is the curvature
factor, which depends upon the shape and size of electrodes as well as gas
pressure.
As mentioned earlier, if E < and a < iq, then avalanches cannot
form. Hence is the lowest value of stress at which SF^ gas can break­
down. In the critical field intensity method, the breakdown voltage of SF^
is related to E^j-iP length, gap pressure and field utilization factor. Using
SF^ Insulation 91

this approach, empirical equations have been proposed to estimate dis­

charge voltages in many configurations [1]. Under impulse voltages, the
probability of breakdown is also influenced by the statistics of the initiatory
electrons. Such electrons are mostly produced by detachment of negative
ions. A method capable of predicting the inhomogenous field breakdown
for a variety of electrode configurations is discussed by Wiegart et al.
[13,14].

4.7 FACTORS AFFECTING DISCHARGE VOLTAGES

The level of dielectric strength of compressed SF^ outlined in the foregoing

sections may not be fully attainable in reality. Several factors are known
to have an effect on the dielectric strength of SF^ when used in practical
systems. The most important of these factors are discussed here.

4.7.1 Contamination
Any fixed or free metallic particle present in SF^ gas can lower its corona
onset and breakdown voltage considerably. The breakdown voltage in con­
taminated gas can be as low as 1 0 % of the corresponding clean gas value.
The breakdown voltage depends on the particle shape, size, material, lo­
cation, motion, gas pressure and nature of the applied field [16]. Figure
4.3 shows the effect of some of these parameters on breakdown voltage
[17]. Free particles present in the SF^ system can gain charge and move
between the electrodes under applied stress. During the motion, the particle
can also have corona at its sharp edges [18]. Gas breakdown usually occurs
when a metallic particle moves very near to the opposite electrode. The
breakdown is triggered as a result of the high pulsed field which appears
when a charged particle approaches the opposite electrode, and creates a
microdischarge
Insulating particles such as glass and fibers do not have any significant
effect on the breakdown voltage. However, dust particles can lower the
breakdown voltage by as much as 30%. The effect of dust particles is
particularly pronounced under AC voltages. Therefore, cleanliness is very
important in SF^ insulated equipment. Particle traps are usually provided
in large SF5 insulated equipment to capture and immobilize the free con­
ducting particles which can pose a serious risk to the insulation integrity.
Another approach which may offer some immunity from free conducting
particle-related dielectric strength reductions is the use of insulating coat­
ings on conductors [1,19].
92 Chapter 4

Figure 4.3 Breakdown in SF^ initiated by free copper wire particles in a 150/
250 mm coaxial electrode system. Particle diameter is 0.4 mm. (From Ref. 17 ©
IEEE, 1973.)

4.7.2 Electrode Factors

The condition, area and material of an electrode surface can also influence
dielectric strength of SF^. When the applied field exceeds —100 kV/cm,
SF^ breakdown strength decreases with increasing electrode area. This in­
fluence is more pronounced at higher field gradients and also depends on
the voltage waveshape and polarity. The negative switching impulse break­
down gradients are less influenced by the electrode area as compared to
the power frequency breakdown gradients. In SF^, the breakdown voltage
at higher pressures obeys the extreme value distribution, i.e., the breakdown
is governed by the weakest point in the gap. Since in large electrode area
systems such points are expected to increase in numbers, lower breakdown
gradients are expected for such systems.
Electrode surface roughness can cause a large reduction of breakdown
voltages in SF^ insulated apparatus. Surface roughness can cause localized
microscopic regions near rough (or sharp) points where the electric field
is much higher than the macroscopic average field in the gap. Depending
on gas pressure and applied voltage, such regions of enhanced field can
SF^ Insulation 93

satisfy streamer criterion at lower average stress values, resulting in a large

reduction of the breakdown gradient. For a hemispherical protrusion of
radius R above a flat plane, Pedersen [20] has shown that the surface
roughness will not affect the threshold of breakdown if PR < 0.4 kPa cm.
However, as the product PR increases above 0.4 kPa cm, the average break­
down field can become as low as 30% of the value for the clean and smooth
electrode system. For practical GIS, the operating pressures are usually in
the range of 3-4 bars. If P = 4 bar, R must be less than 10 fim to fully
achieve the dielectric strength of SF^. Such a good surface finish is usually
very difficult, if not impossible, to achieve in practical systems.
Although over the range where Paschen’s law is valid, the electrode
material has no influence on the dielectric strength of SF^ the electrode
material does affect the high pressure breakdown when the applied stress
exceeds —200 kV/cm. Under such conditions, stainless steel electrodes
exhibit a higher dielectric strength than copper electrodes. The dependence
of breakdown strength on electrode material and electrode area is espe­
cially evident in uniform and quasiuniform field electrode systems.

4.7.3 Insulating Spacers

Insulating spacers are used to support the high voltage conductors within
the earthed casing in SF^ insulated equipment and a flashover along the
solid/gas interface has a high probability. The spacer flashover voltage is
usually lower as compared to the corresponding value for gas gap break­
down. Thus, the spacer efficiency—which is defined as the flashover volt­
age with spacer present divided by the breakdown voltage for same gas
gap but without the spacer—is usually less than one. This efficiency gen­
erally decreases with increasing gas pressure and depends upon the design
of spacer (shape, material, etc.), its contact with the electrodes, applied
voltage waveform and polarity as well as presence of contaminants, e.g.,
dust, metallic particles and moisture [21]. A less than perfect contact be­
tween the spacer and the electrode greatly affects the flashover voltage.
The presence of moisture adversely affects the dielectric strength of the
spacer/gas interface, which can drop by more than 50% when the moisture
is increased, as shown in Figure 4.4 [22]. Generally, the “triple” junction
between electrode, spacer and gas is very important and is designed to
optimize the field distribution in this region. Presence of particles on the
spacer are extremely dangerous for integrity and life expectancy of SF^
equipment. Once a particle is deposited onto a spacer surface, it is likely
to be held on this surface by electrostatic charges. Corona from the parti­
cle’s tip may cause SF^ decomposition to highly corrosive species such as
hydrofluoric acid (HF), which can attack the spacer surface and cause
94 Chapter 4

P cirtial pressure of water ( torr)

Figure 4.4 Relation between spacer flashover voltage and partial pressure of
water. (From Ref. 22 © IEEE, 1971.)

eventual failure. Similar situation arises when corona occurs on any floating
component. Under DC stresses, the spacer can get charged which can com­
plicate things further. Therefore, extreme care is required in the design and
installation of spacer within SF^.

4.7.4 Gas Purity

Whereas gas purity might affect the chemical and biological properties of
SF^, small percentages of common gases such as air, N 2 and CO2 when
mixed with SF^ do not greatly reduce its dielectric strength. Figure 4.5
shows the variation of relative breakdown voltage with SF^ content in SF^-
N2 mixtures [23]. The results for SF^-air, SF6 -CO2 and SF6 -N2 O mixtures
are also similar [23-25]. Thus, whereas a small percentage of air or N2
contaminants in SF^ insulated equipment will not adversely affect its di­
electric strength, the influence of moisture on insulators and the reaction
of moisture and O2 with SF^ decomposition byproducts can, in the long
run, affect the integrity of SF^ insulated apparatus and should be avoided.

4.7.5 Gas Temperature

SF^ insulated equipment may be subjected to extremely hot or cold ambient
conditions. In some areas, the outdoor equipment may face temperatures
as low as -50°C, which can cause liquefaction of the gas at normal pres-
SF^ Insulation 95

Figure 4.5 Variation of breakdown voltage with SF^ percentage in SF6- N 2

mixtures. (From Ref. 23 © IEEE, 1979.)

sures that are commonly employed in such equipment. Therefore, low tem­
perature breakdown behavior of SF^ is of significant importance.
In 20 to 50°C temperatures, SF^ exhibits a constant dielectric strength
as long as gas has a constant density. In the temperature range of 50 to
325°C, density (N) reduced critical field intensity of SF^, i.e., (E/N)^,^^
increases by about 10%. This increase has been attributed to an increase
in the electron attachment at higher temperatures [24]. In another study,
low temperature DC breakdown of SF^ insulated system was investigated
for temperatures ranging from -50°C to 24°C. It was found that, although
nonuniform breakdown was barely affected, the uniform field breakdown
strength of SF^ insulated system was temperature dependent and decreased
by about 10% as the temperature was reduced. The decrease appeared at
a temperature threshold between -2 5 and -30°C, and remained constant
down to -50°C [26]. Thus, even when constant density is maintained, the
operational safety factor of HV equipment using SF^ insulation may be
reduced if the equipment temperature drops below -20°C.
96 Chapter 4

4.8 ARC INTERRUPTION IN SFg

The arc interruption behavior of SF^ gas is of considerable practical interest

for the use of this gas in circuit breakers. A circuit breaker is required to
isolate the faulted part of a network from the rest of the system (Figure
4.6). The breaker is tripped by the protection system when it detects the
presence of high fault current If flowing through the network. The breaker
contacts, which are initially closed, start opening upon the activation of
the trip mechanism and an arc is formed between the contacts. The arc
current is equal to If and depends upon the system configuration, type and
location of the fault as well as the fault impedance. The electrical power
input to the arc depends upon arc current and arc resistance. Thermal losses
occur in the arc by diffusion, conduction, radial and axial convection and
radiation processes, and depend upon the gas pressure, temperature and
turbulence in the arc region as well as the arc length. Depending upon the
balance between the electrical input power and the thermal losses, the arc
is extinguished usually at current zero.
Upon current interruption, transient oscillations are generated in the
network. These oscillations produce transient recovery voltage (TRV)
across the open breaker contacts. For successful interruption of the arc, the
dielectric strength of the gas should recover very quickly. Moreover, the
residual conductivity should quickly decrease after the current zero to keep
the arc extinguished. If these conditions are not met, the arc is re­
established and the breaker fails to interrupt the current. For SF^ circuit
breakers, two types of failures with different behavior have been noticed
[27,28]. Immediately after current zero, if the rate of rise of TRV [i.e.,
RRRV (rate of rise of recovery voltage)] is greater than a critical value,
the decaying arc channel is re-established by ohmic heating caused by
residual conductivity. This period, which consists of the first 4 to 8 yas
after current zero, is mainly controlled by the arc thermal energy balance
and is known as the thermal interruption mode. If the thermal interruption
is successful, the TRV can reach to such a high peak value Vp, that the
breaker sometimes fails through the dielectric breakdown of the open con­
tact gap. This is known as the dielectric failure mode and is usually 50 yas

Figure 4.6 Isolation of a faulted part of the network by a circuit breaker.

 Insulation 97

after current zero. Thus, the behavior of SF^ and its byproducts in thermal
and dielectric recovery phases determines the arc interruption characteris­
tics. From a wide variety of possible fault locations and network conditions,
it is found that the terminal fault, i.e., a fault that occurs at the terminal of
the circuit breaker, produces the highest stresses for the dielectric failure
mode [23]. For the thermal failure mode, the critical fault is the one that
occurs on a line some distance (a few km) from the circuit breaker.
The thermal recovery characteristic is usually expressed in the form
of a critical boundary separating fail and clear conditions on a
RRRV-dIf/dt or RRRV-If diagram. Typically, the boundary obeys the re­
lationship of the form:

RRRV = A (4.5)
dt
where A is a constant and n = 1-4.6 [27]. The thermal recovery perform­
ance can be improved by increasing SF^ pressure (P) in the arc chamber
since RRRV is proportional to (P)"" where m = 1-3, and by a suitable
design of the interrupter head geometry.
For the dielectric recovery regime, the characteristic is also represented
by the critical boundary separating successful clearance m d fa il on a Vp-If
diagram. The dielectric recovery performance may be improved by increas­
ing the number of interrupter units in series. By combining the thermal
and dielectric recovery characteristics, the overall limiting curves for the
circuit breaker performance are obtained as shown in Figure 4.7, where

lo g (RRRV)

Figure 4.7 Superposition of thermal and dielectric limiting curve diagrams along
with network responses. (From Ref. 27.)
98 Chapter 4

the network response under both thermal and dielectric modes is also given.
The maximum current which can be interrupted according to thermal mode
is Ith, whereas the corresponding value under dielectric mode is shown as
Idiei- Figure 4.8 shows the overall circuit breaker performance characteris­
tics for one and two interrupters in series.
In SF^ breakers, the arc control is usually achieved by gas blast or
electromagnetic methods [28]. In gas blast circuit breakers, the contacts
are separated along the axis of a gas flow guiding nozzle so that the arc
is subjected to the convective effects of the gas flow. The thermal recovery
performance of such breakers depends upon the fault current level as well
as gas pressure and gas flow parameters. In electromagnetic circuit break­
ers, the arc is moved by the action of Lorentz force produced as the fault
current flows through a magnetic field which is produced by a coil. The
resulting arc motion can be used to control the thermal recovery perform­
ance of the arc. Therefore, for improving the thermal and dielectric recov­
ery characteristics, the choice of SF^ pressure, number of interrupters in
series and the methods of arc control and arc movement are important
parameters. The other factors which can influence the arc interruption char­
acteristics are nozzle geometry and materials of interrupter contacts, mois­
ture and foreign particles as well as effects of high frequency transients
[28].

4.9 GAS INSULATED SWITCHGEAR

SF^ is used in metal enclosed gas insulated switchgear (GIS) ranging from
medium voltages to 800 kV. In GIS, different components of the switchgear
are enclosed in metallic housing in adjacent compartments and insulated
with compressed SF^. This arrangement offers the following advantages:
1. GIS are compact as compared to open air substations. Compared
to conventional open air station, GIS needs only about 10-15%
floor area.
2. They provide total protection from rain, fog, atmospheric pollu­
tion, chemicals, etc.
3. They are safe, noise-free, reliable and require minimum mainte­
nance.
4. They are prefabricated and are of modular construction, thereby
allowing easier installation and flexible design even under adverse
site conditions.
However, GIS requires continuous gas monitoring, gas tight construction
and pressure relief devices. Moreover, due to sensitivity of the dielectric
SF< Insulation 99

Figure 4,8 Overall circuit breaker performance. (From Ref. 27.)

strength of SF^ gas to several factors discussed in section 4.7, extreme care
is required in manufacture, installation and maintenance. GIS can be lo­
cated indoors or outdoors. Though basic design of both types is the same,
the equipment for outdoor GIS requires additional weatherproofing to suit
the climatic conditions.

4.9.1 General Design

There are two basic types of designs of GIS. At low-rated voltage of

<200 kV, all three phases have usually a common enclosure, i.e., 3-in-l
construction; whereas at higher-rated voltages, each phase uses its own
separate enclosure, i.e., single-phase construction. If single-phase auto/
reclose is a requirement, then for the circuit breaker at least, single-phase
construction is preferable even at lower voltages. In a 3-in-l construction,
a single-phase fault within the GIS usually transforms into a phase-to-phase
fault and consequently an enclosure bum-through is avoided. Sometimes a
single-phase arrangement is adopted for the circuit breaker while a 3-in-l
arrangement is used for the ‘‘back parts,” i.e., busbars, connectors, etc.
The whole GIS is divided into a number of adjacent gas-tight com­
partments. Basic components of a GIS are circuit breakers, disconnectors,
earthing switches, busbars, connectors and current and voltage transform­
ers, etc. Each component is designed to withstand the expected electrical,
thermal and mechanical stresses under normal and fault conditions. Figure
4.9 shows typical feeder bay for a double-busbars GIS.
100 Chapter 4

Figure 4.9 Typical feeder bay for double-busbar GIS: 1 = voltage transformer,
2 = busbars, 3 = circuit breaker, 4 = cable box, 5 = hydraulic mechanism, 6 =
gas pressure (or density) meter, 7 = disconnector and earthing switches, 8 =
SF^-air bushing and 9 = barrier insulator.

Usually bus conductors are made of aluminum tubes whereas copper

is used for switch and circuit breaker contacts. The tube diameter and wall
thickness are determined by voltage and current ratings of the equipment.
The outer enclosure is made from aluminum or steel, the choice depending
upon economic and technical considerations. The insulators are usually
made of filled epoxy resins and are of conical, disc or tube designs. They
are carefully designed to provide enough creepage length and suitable
stress distribution along the surface.
The GIS designs are in modular form and various components can be
assembled as desired. SF^ gas pressure between 2 and 5 bars is used and
is determined based on voltage rating, cost, equipment size and reliability
considerations. Systems operating at low pressures will have relatively
larger size, but will be more defect tolerant and comparatively more reli­
able than the high pressure systems. Typical service stress levels used for
GIS are about 7 to 8 per mm per MPa of gas pressure. However,
the barrier insulators are restricted to stress levels of less than 4 kY^^^ per
mm. Thus, the working stress of GIS is significantly lower than the break­
down strength of SF^. By this approach, the manufacturers hope to ensure
the long-term dielectric integrity of SF^ equipment, provided normal qual­
ity and service procedures have been maintained.
SF. Insulation 101

Provision is generally made for a solid grounding of the entire GIS

enclosure. In addition, automatic grounding switches are provided for
grounding of the cables whenever the isolators are opened.

4.9.2 Circuit Breakers

When a circuit breaker starts operating, the arc is established through a
nozzle by the separation of the contacts and/or gas flow and is subjected
to a gas blast which abstracts energy from the arc. In the early SF^ circuit
breakers, the gas flow was achieved by the use of a two pressure system,
whereby the operation of a blast valve allowed gas to flow from a high
pressure reservoir through a nozzle into a low pressure reservoir. After arc
extinction, the gas was recycled back into the high pressure reservoir. This
system required two separate gas reservoirs with associated seals, a com­
pressor and gas handling plant, and heaters to prevent gas liquefication at
low temperatures. Furthermore, in such a system, there was the necessity
of synchronizing the blast valve and the contact driving system.
The majority of interrupters today are of much simpler single-pressure
“puffer” type design. In this design, during the tripping operation, gas in
the arcing chamber is compressed to a high pressure within a piston-and-
cylinder arrangement, and then directed across the arc to effect the extin­
guishing process, finally expanding back to its normal pressure. Depending
upon the method by which the gas is directed to the arc the puffer breakers
are classified as (see Figure 4.10):
1. Mono-blast, where the gas is forced in a single direction
2. Duo-blast, where the gas is forced in two opposite directions,
through two nozzles of equal diameter
3. Partial duo-blast, where the gas is exhausted through two or more
unequal diameter nozzles
The interrupters are accommodated in aluminum or steel tanks with the
number of interrupters in series depending on the voltage and short circuit
current ratings. To ensure equal interrupting duty, voltage grading capaci­
tors are fitted across multibreak units. Puffer type interrupters having 63
kA rating at 520 kV with one break per phase are common now. The SF^
interrupter can be incorporated in either live tank or dead tank configura­
tion. The type chosen will depend on economics and/or the type of appli­
cation. For example, the dead tank construction in which all interrupters
are enclosed within an earthed pressure vessel is essential for use in com­
plete metalclad installation, although the same breaker may also be used
with terminal bushings in open air type layouts. The circuit breaker tank
may be mounted in a horizontal or a vertical configuration.
102 Chapter 4

(C)
Figure 4.10 Types of gas blast in interrupters: (a) monoblast; (b) partial duo blast;
and (c) duo blast.

A high performance puffer system requires a powerful operating mech­

anism to ensure a high velocity for the contact/cylinder arrangement.
Motor-wound spring, pneumatic or hydraulic power units are usually used
for this purpose. Arcing contacts are generally fitted with copper-tungsten
alloy tips to reduce erosion. Interrupter nozzles are made of polytetrafluoro-
ethylene (PTFE). The breakers are designed for long life and maintenance-
free operation. Whenever maintenance is required, the dismantling of the
breaker should be carried out in a clean and dry environment. The moisture
reacts readily with arc left out byproducts causing formation of hydro­
fluoric acid, which reduces the insulating capacity of dielectric materials.
During arcing, some metallic fluorides are formed and settle on breaker
parts as a fine nonconducting dust. This must not be inhaled and must be
carefully cleaned off with a dry cloth as soon as the breaker is opened.
The dielectric integrity of the equipment is not influenced by these
compounds.

4.9.3 Disconnectors
Disconnectors are located on each side of the circuit breaker and are usu­
ally the off-load type. They, however, should be able to break small charg­
SF^ Insulation 103

ing currents without generating excessive overvoltages. Usually, discon­

nectors are designed such that they can be accommodated in busbars or
connectors. Their operating mechanisms are hydraulic, hydrostatic or mo­
tor operated; the last type being the most common, in which case the motor
can be directly mounted on the chamber.

4.9.4 Earthing Switches

Earthing switches are either slow or high speed. Slow speed or off-load
earthing switches are used for protection during maintenance and are only
operated when the high voltage is off. These are either accommodated
within the same chamber as the disconnector, or are housed in a separate
compartment. They are normally equipped with a manual operating mech­
anism, though power drives are also possible. For applications where in­
terlocking with the remote end is not possible, fast earthing switches are
required. Such switches can be operated under full voltage and short circuit
power and are required to interrupt induced currents from energized par­
allel lines as well. These switches have motor-wound spring operating
mechanisms to produce the required speed of operation.

4.9.5 Voltage and Current Transformers

Voltage transformers can be mounted external to the GIS. In the metalclad
construction, electromagnetic type of transformers with secondary wind­
ings accommodated in a separate SF^ insulated chamber and a multi-turn
primary connected to the HV bus conductor through a sealed joint are
normally used. Alternatively, a capacitive voltage transformer (CVT) can
be designed by adding a measuring electrode near the GIS enclosure.
Current transformer can be located near or away from the circuit
breaker. The secondary windings can be accommodated over the primary
conductors in the circuit breaker tank, in turrets attached to the circuit
breaker casing, within standard GIS chambers or externally in modules of
metal or glass fiber concentrically mounted over connection chamber [4].

4.9.6 Surge Arresters

These are normally located external to the GIS in open terminal pattern.
For cable connected installations and for transformers and reactors directly
coupled to GIS, a metal enclosed design is used. Only available for single­
phase construction, these comprise SF^-insulated metal enclosures in which
the active components (nonlinear resistors) are installed and connected
through a sealed joint to the main conductors. In modern designs, metal
104 Chapter 4

oxide nonlinear resistor blocks, without series spark gaps, form the active
component resulting in a very compact design and fast operation. A hood
and graded electrodes may be required to achieve an acceptable voltage
distribution across the active element.

4.9.7 Busbars and Connectors

Busbars and connection chambers of various configurations, such as
straights, elbows, tees or angles, are bolted together with 0 -ring seals.
Joints of the conductors have provisions for thermal expansion. Similarly
bellows are provided to accommodate for changes in the length of the
compartments and variation of dimensions due to manufacturing toler­
ances. The whole GIS assembly is supported on ground mounted steel
frame work.

4.9.8 External Connections

SF^-air bushings are used to connect a GIS to the overhead lines, trans­
formers and other air insulated components. These consist of a porcelain
enclosure through which the HV conductor passes. The bushings are nor­
mally divided into two compartments and filled with SF^ gas at different
pressures. A bushing can be bolted directly onto the GIS and can be sup­
ported on the wall or roof of an indoor station. Alternatively, the primary
conductor is encased with solid insulation and sealed in a porcelain enclo­
sure filled with a nonsetting compound of SF^ gas.
If a GIS is to be connected to a cable network, the cable connections
are brought to the SF^ filled cable end boxes or terminating chambers
where these are bolted to the GIS. The interface seal has to be carefully
designed so that SF^ does not penetrate the oil or vice versa. The failure
of such a seal can lead to serious consequences.

4.9.9 Gas System

The whole GIS installation is filled with SF^ gas at the desired pressure.
Since different compartments of a GIS are fitted with gas-tight seals, any
gas leak within one compartment does not affect the performance of other
compartments. Similarly, arcing byproducts within circuit breaker and dis­
connector chambers do not enter the main busbars. Furthermore, busbar
and connectors, etc., are broken into a number of discrete sections to in­
crease reliability and to minimize effect of internal faults. Each gas-tight
section is provided with gas filling and emptying stop valve, a pressure
SF^ Insulation 105

relief device, and a pressure gauge or a gas density meter. The density
meters also provide alarm and lockout signals.

4.9.10 Testing
All GIS components are subjected to type and routine tests. Type tests for
equipment of <245 kV rating require lightning impulse and AC voltage
tests. For system voltage >300 kV, switching impulse test is an additional
requirement. Before shipment to construction site, each transportable as­
sembly is subjected to an AC voltage routine test. During assembly on the
shop floor, stringent cleanliness and inspection procedures must be main­
tained to prevent any contamination. In addition, all seals are tested with
a leak detector. Usually an internal arcing test is also done to ensure the
proper working of pressure relief devices and to check the possibility of
“bum through” of the enclosure wall. Each unit of circuit breaker is sub­
jected to synthetic testing to verify its performance.
At the installation site, tests are performed to check for leaks at every
flanged joint; moisture and air content of SF^ gas in the system; continuity,
resistance and voltage drop tests on conductor joints; and functional checks
on various GIS components, such as breakers, isolators, earthing switches
and control instrumentation. In addition, HV and partial discharge (PD)
tests may also be performed. For GIS operating at EHV levels, techniques
for PD monitoring are now well established and can detect free and fixed
particles, electrode protmsions, floating components, loose nonfloating
parts and voids in insulators [30,31]. On line or periodic PD monitoring
of the equipment is therefore recommended for GIS equipment operating
at EHV levels, since any failure of the equipment can have very serious
economic and technical consequences.

4.9.11 High Frequency Transients

The breakdown process in SF^ gas is very fast and voltage collapse can
occur in a few nanoseconds. During the switching operation of a discon­
nector, prestrike or restrike can occur several times before the switching
operation is finally completed. Similarly, during a phase-to-earth fault,
breakdown takes place in SF^ equipment. During such breakdowns, very
fast transients with rise time of 4-15 ns are generated which last for several
hundred milliseconds. Such transients have very high voltage rate of rise
(100 kV/ns) and distance rate of rise (5 kV/cm) [32]. Since GIS is a low
loss coaxial medium, there is very little attenuation. Hence various reflec­
tions of the transients can take place that could lead to overvoltages in
106 Chapter 4

excess of basic impulse level (BIL) of the system. Also when the transients
travel to system terminations, they couple to the outside world through
SF^-air bushings, cable sealing ends and current and voltage transformers.
Consequently, the grounded enclosure of a 500 kV, SF^-insulated system
near gas to air termination, can rise to tens of kilovolts during switching
operations.
Although such switching transients are usually of modest magnitudes
(<2.5 per unit), these do cause significant number of failures of all GIS
failures in the EHV range due to lack of corona stabilization under fast
surges, due to their frequent occurrence and due to a reduced BIL-to-
system voltage ratio at the EHV range [32]. Such transients can also cause
failure of transformers connected to GIS. Major factors which influence
the magnitude and frequency of these very high frequency (VHF) transients
in GIS are disconnector contact configuration, SE^ gas pressure, speed of
contacts, capacitive current being switched and trapped charge at the in­
stant of prestrike or restrike [4]. During the disconnector operation, re­
peated sparking between contacts may lead to phase to earth breakdown
also.
Besides the influence of these VHF transients on the performance of
GIS, these transients can pose potential hazards to the personnel who come
into contact with the GIS enclosure during switching operations as a result
of transient ground potential rise. In addition, voltages induced on control
and secondary wiring systems in the vicinity of the GIS installation can
cause failure of electronics and relays. Several means are used to limit the
adverse affects of these transients and achieve electromagnetic compati­
bility in GIS environment [19,32]. Techniques to measure such VHF tran­
sients are described in chapter 1 1 .

4.9.12 Overvoltage Protection

The switching and re-energization of long EHV transmission lines can
generate very high switching overvoltages. These overvoltages can be
reduced by pre-insertion resistors and by point on wave switching. SE^
breakers with pre-insertion resistors have complex drive mechanisms, are
more expensive than standard breakers and are less reliable. An alternative
method of reducing the switching overvoltages is to use metal oxide surge
arresters (MOA). These have increased reliability, trouble free service and
are more economical. When the HV side of a transformer is directly con­
nected to SF^ circuit breaker by the GIS busbar and disconnectors, a surge
arrester is connected near the transformer for overvoltage protection against
VHF and other transients. However, even the fast acting MOA cannot fully
SF^ Insulation 107

cope with VHF transients, which can overstress the transformer winding
even in the presence of MOA.
The outdoor, open-terminal substation equipment can also be subjected
to pollution problems, lightning strikes and other forms of transient over­
voltages. Such equipment can be protected using rod gaps or arcing horns
and surge arresters. The indoor GIS equipment is normally equipped with
MOA and surge diverters [33].

4.10 COMPRESSED GAS INSULATED CABLES

SF5 is also used in gas insulated cables. Such cables have the potential of
transmitting powers in the range of 1000-5000 MVA as compared to lower
ratings of 200-1500 MVA for conventional cables. SF^ insulated cables
have:
1. Lower capacitance, charging current and reactive power
2 . Lower resistive and dielectric losses, better heat transfer and ther­
mal performance and hence higher ampacities
3. MVAR losses, which are function of current—and for a certain
value of current, total reactive power losses can be made equal to
zero thereby operating the cable at a unity power factor
4. The potential of increasing the operating voltage of cable by
increasing gas pressure without changing the other design param­
eters
Such cables are of single phase coaxial design or three phases in a common
grounded enclosure. The cross-sectional area of conductor is much bigger
in such a cable as compared to conventional cables. The other design fea­
tures are similar to the busbars of the GIS. Typical gas pressures used are
about 3-4 bars. Flexible cable designs also have been developed. Spacer
flashover, contaminants, and spacer life are generally the limiting factors
in the performance of such cables. For best designs, the ratio of the enclo­
sure diameter to the conductor diameter is kept near 2.72. Typically, for
420 kV system with a BIL of 1425 kV operating at SF^ gas pressure of
3.5 bar, the minimum possible diameters of outer and inner conductors are
approximately 14 and 5 cm, respectively. Generally for a given voltage,
the surge impedance of a gas insulated cables is about 60 and therefore
surge impedance loading of a compressed gas cable is about 3 to 4 times
that of a corresponding overhead line. Cables are manufactured in about
5-10 m segments that are joined in the field, requiring considerable ex­
108 Chapter 4

cavation and field work that makes such cables expensive. At present, such
cables are used in comparatively shorter segments for bulk power transfer.

4.11 OTHER APPLICATIONS OF SFg

Besides various applications of SF^ insulation described in the previous

sections, this gas is also used in high voltage gas-insulated capacitors, fast­
acting high voltage switches, power transformers, van de Graff generators,
high fidelity electrostatic loudspeakers, the semiconductor industry as well
as sealed multipan windows.

4.12 SF. GAS HANDLING

Since SF^ condenses to a liquid at room temperature at about 20 bars, it

is stored in liquid form out of direct sunlight in a safe place. Its low storage
pressure greatly improves handling safety and reduces storage costs relative
to other commercial gases. Storage carts are available with compressors,
vacuum pumps, oil separators, desiccants, storage vessels with refrigeration
to limit the pressure during storage and heaters to help in evaporation of
the liquid SF^. Instruments are available for measuring the moisture content
of SF^, the amount of air in SF^, gaseous decomposition byproduct con­
centration, etc. The gas may be withdrawn from the cylinder in either the
gaseous or liquid phase. When the cylinder is in upright position, SF^ will
issue out as a gas. When the cylinder is inverted, SF^ discharges as a liquid.
As mentioned earlier, SF^ itself is nontoxic but can generate toxic
byproducts. Therefore, in some power applications, working with an SF^-
insulated apparatus requires use. of a respirator, filter gas mask, and/or
protective clothing. Some of the toxic byproducts of SF^ may be absorbed
by a 50-50 mixture of soda lime (NaOH + CaO) and activated alumina,
i.e., especially dried AI2 O3 . The preferred granule size is between 8 and
12 mesh. Generally the weight of the absorbent should be equal to about
10% of the weight of the gas. The absorbent should be placed so as to
provide maximum contact with the gas which may contain the decompo­
sition products.
Sometimes it is necessary to fill SF^ in the field. This involves initial
evacuation down to less than 1 mbar. Then, the equipment is filled with
dry N2 at 3-4 bar and left in this state for some time. The system is then
evacuated again and carefully filled with dry SF^ gas. After some period,
the gas samples are taken to check moisture level which should be kept
SF^ Insulation 109

below —30 mg of water per kg of SF^. Higher moisture levels can lead to
tracking and failure of insulators.
SF^ gas leaks from GIS and cables usually can be detected by using
tracer gases. The commonly used tracer gases are halogen gases (freon),
helium or some radioactive gas. The tracer gas is fed into the system and
the leak is located by detecting the presence and measuring the concentra­
tion of the tracer gas along the equipment route by a variety of methods.
In case of a major leak, all personnel should be evacuated from the area.
Good ventilation should be provided in the area and the gas should be
removed by pumping it through the absorbent and stored in cylinders.

REFERENCES

V. N. Mailer and M. S. Naidu, A d v a n c e s in H ig h V o lta g e I n s u la tio n a n d A r c

a n d V acu u m , Pergamon Press, New York, 1981.
I n te r r u p tio n in
2. R. J. Van Brunt and J. T. Herron, IEEE Trans, on Elect. Insul., Vol. 25, No.
I, pp. 75-94, 1990.
G. D. Griffin, I. Sauers, L. G. Christophorou, C. E. Easterly and P. J. Walsh,
IEEE Trans, on Elect. Insul., Vol. 18, No. 5, pp. 551-552, 1983.
4. S. M. Ghufran Ali and W. D. Goodwin, Power Engineering Journal, Vol. 2,
No. 1, pp. 17-26, 1988.
5. N. H. Malik and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 13, No. 3,
pp. 135-145, 1978.
6. A. Pedersen, IEEE Trans, on PAS, Vol. 89, No. 8, pp. 2043-2048, 1970.
7. N. H. Malik, IEEE Trans, on Elect. Insul., Vol. 16, No. 5, pp. 463-467, 1981.
8. T. W. Dakin, G. Luxa, G. Oppermann, J. Vigreux, G. Wind and H. Winkeln-
kemper, Electra, No. 32, pp. 61-82, January 1974.
9. R. S. Nema, S. V. Kulkami and E. Husain, IEEE Trans, on Elect. Insul., Vol.
17, No. 1, pp. 70-75, 1982.
10. N. H. Malik and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 14, No. 6,
pp. 327-333, 1979.
11. N. H. Malik and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 14, No. 1,
pp. 1-13, 1979.
12. S. A. Boggs, IEEE Elect. Insul. Magazine, Vol. 5, No. 6, pp. 16-21, 1989.
13. N. L. Wiegart, F. Niemeyer, F. Pinnekamp, W. Boeck, J. Kindersberger, M.
Morrow, W. Zaengl, M. Zwicky, I. Gallimberti and S. A. Boggs, IEEE Trans,
on PWRD, Vol. 3, No. 3, pp. 923-930, 1988.
14. N. L. Wiegart, F. Niemeyer, F. Pinnekamp, W. Boeck, J. Kindersberger, M.
Morrow, W. Zaengl, M. Zwicky, I. Gallimberti and S. A. Boggs, IEEE Trans,
on PWRD, Vol. 3, No. 3, pp. 931-938, 1988.
15. A. Pedersen, IEEE Trans, on Elect. Insul., Vol. 24, No. 5, pp. 721-739, 1989.
16. J. R. Laghari and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 16, No.
5, 1981, pp. 388-398, 1981.
lio Chapter 4

17. A. H. Cookson and O. Parish, IEEE Trans, on PAS, Vol. 92, pp. 871-876,
1973.
18. A. H. Mufti, A. A. Arafa and N. H. Malik, IEEE Trans, on Dielectrics and
Elect. Insul., Vol. 1, No. 3, pp. 509-519, 1994.
19. M. Khalifa (ed.), H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c tic e , Marcel
Dekker, Inc., New York, 1990.
20 . A. Pedersen, IEEE Trans, on PAS, Vol. 94, pp. 1749-1754, 1975.
21 . J. R. Laghari and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 16, No.
5, 1981, pp. 373-387.
22. T. Ushio, I. Shimura, and S. Tominaga, IEEE Trans, on PAS, Vol. 90, No. 5,
pp. 2166-2174, 1971.
23. N. H. Malik and A. H. Qureshi, IEEE Trans, on Elect. Insul., Vol. 14, No. 2,
pp. 70-76, 1979.
24. L. G. Christophorou and R. J. Van Brunt, IEEE Trans, on Dielectrics and
Elect. Insul., Vol. 2, No. 5, pp. 952-1003, 1995.
25. M. Akbar and N. H. Malik, IEEE Trans, on Elect. Insul., Vol. 20, No. 3, pp.
581-585, 1985.
26. M. E. Erechette, D. Roberge and R. Y. Larocque, IEEE Trans, on Dielectrics
and Elect. Insul., Vol. 2, No. 5, pp. 925-951, 1995.
27. K. Ragaller (ed.), C u r r e n t- I n te r r u p tio n in H ig h V o lta g e N e tw o r k s , Plenum
Press, New York, 1978.
28. H. M. Ryan (ed.), H ig h V o lta g e E n g in e e r in g a n d T estin g , Peter Peregrinas
Ltd., London, England, 1994.
29. A. Bradwell (ed.). E le c t r i c a l I n s u la tio n , Peter Peregrinas Ltd., London, Eng­
land, 1983.
30. J. S. Pearson, O. Parish, B. F. Hampton, M. D. Judd, D. Templeton, B. M.
Pryer and I. M. Welch, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 2,
No. 5, pp. 893-905, 1995.
31. R. Baumgartner, B. Fruth, W. Lanz and K. Pettersson, IEEE Elect. Insul.
Magazine, Vol. 8, No. 1, pp. 16-27, 1992.
32. S. A. Boggs and H. H. Schramm, IEEE Elect. Insul. Magazine, Vol. 6, No.
I. pp. 12-17, 1990.
33. J. R. Simms, lEE Power Engineering Journal, Vol. 1, pp. 215-222, 1987.
Liquid Dielectrics

5.1 INTRODUCTION

Electrical insulating liquids are used abundantly in a variety of components

of power system networks, such as transformers, power cables, circuit brea­
kers, power capacitors, bushings and switches. Here they serve various
functions either alone or in combination with solid insulating materials,
such as to insulate components from each other or the ground, to impart
cooling action, to fill up voids in composite dielectrics, to impregnate thin
sheet insulation, to control and provide high capacitance in power capac­
itors, for arc extinction in high voltage switches/circuit breakers and also
to act as lubricants in tap changers and circulating pumps.
A wide variety of natural and synthetic oils are being used in the
electric power industry. Mineral oil is the cheapest and most commonly
used oil. However, selection of a typical dielectric fluid depends on its
physico-chemical and electrical properties besides the nature of the service
it will perform. This chapter provides details of various important aspects
of insulating liquids. These include general classification of liquids, essen­
tial characteristics of most commonly used and recently introduced oils,
reconditioning of service aged oils beside details on conduction and break­
down mechanisms. The references selected for different topics are chosen
from a vast literature to assist both research scholars and engineers in
gaining in-depth knowledge of the topic concerned.

Ill
112 Chapter 5

5.2 CLASSIFICATION OF INSULATING OILS

Insulating oils can be broadly divided into two categories: organic and
inorganic. The most commonly used organic liquid dielectrics for electrical
power equipment are mineral oils. Beside the naturally occurring oils, a
variety of synthetic organic insulating liquids are also available in the mar­
ket. Most common of these are silicone oils and chlorodiphenyles. Recently
some new synthetic organic oils such as high molecular weight hydrocar­
bons and tetrachloroethylene have also been introduced. They possess ex­
cellent dielectric and thermal properties but are expensive. Among inor­
ganic insulating liquids, highly purified water and its aqueous solution are
being used for pulsed power applications, such as capacitors and modu­
lators. At present, water-filled discharge lines for pulsed power are under
intensive investigation and development [1]. Similarly, liquefied insulating
gases are commonly employed for cryogenic applications. A brief descrip­
tion of most commonly used insulating liquids is given below.

5.2.1 Petroleum Oils

The main elemental constitutes of petroleum are carbon and hydrogen to­
gether with trace amount of sulphur, oxygen and some metals. These are
joined together to form three commonly occurring structures/compounds:
paraffins, naphthenes and aromatics. Examples of some of these are shown
in Figure 5.1. These compounds are treated and mixed to form insulating
oils to meet the desired physio-chemical and electrical properties. Some­
time they are also treated with additives. An additive is defined as a sub­
stance not normally present in petroleum but its blending becomes neces­
sary to influence the oxidation, gas absorption and pour point
characteristics of oils. Antioxidants are widely used in some countries,
whereas gas-absorbing additives are used for high voltage cable oils. Pour
point depressants are necessary for waxy oils of paraffinic type.
An oil is considered as naphthenic or paraffinic based, if the content
of either of them exceeds the other. Furthermore, an oil is considered
weakly aromatic if the presence of aromatics is less than 5% and highly
aromatic when these exceed 10%. Most of the mineral oils that are being
used today in high voltage apparatuses are low pour point naphthenic based
because the paraffinic based oils have the tendency of wax formation when
operated at low temperatures.
Liquid Dielectrics 113

(a) CHa- (CHa)n -C H 3

(b) CHa- (CH2)n -C H -C H a

H,C
(C)
H,C

H H
.C,
^CH
(d ) II I
H H
Figure 5.1 Molecular structure of some basic constituents of mineral oils: (a)
straight chain paraffin; (b) branched chain paraffin; (c) naphthene; (d) aromatic ring.

5.2.2 Synthetic Hydrocarbons

Among synthetic liquid dielectrics, polyolefins are the dielectrics of choice
for applications in power cables and some other electrical systems. Over
55% of synthetic materials produced worldwide today are polyolefins [2].
Most commonly used olefins are poly-butylene and alkyl-aromatic hydro­
carbons (e.g., alkyl-benzene). Their composition is simpler than mineral
oils, but the general characteristics are very similar.

Poly-butylenes
These are produced by the polymerization of short chain length hydrocar­
bons. Figure 5.2 illustrates their structure. The carbon chain extends in the
range Cg to C2 5 . Their properties are given in Table 5.1. They possess low
pour point and high viscosity index, i.e., small change in viscosity with
temperature. Their gas absorption characteristics and dielectric properties
are slightly better than petroleum oils.

Alkyl-benzenes
These are available in a range of mixtures composed of compounds having
alkyl chains up to 25 carbon atoms attached to a benzene ring. Table 5.2
114 Chapter 5

CH3 CH3 CH3

I I
C H 3 -C • C~CH2- CH2—c
I I
CH3 CH3 CH2

Figure 5.2 Structure of branched chain poly-iso-butylene.

shows their properties. They are unstable in presence of O2 and are there­
fore not suitable with breathing transformers. However, they possesses high
gas absorption characteristics as shown in Figure 5.3 and also lack in
reaction with copper. They are therefore employed extensively in EHV and
UHV oil filled cables either alone or mixed with mineral oil.

5.2.3 Chlorinated Hydrocarbons

Two aromatic hydrocarbons, benzene and diphenyle, are chlorinated to
produce chlorinated aromatic compounds called askarels or simply poly­
chlorinated biphenyl (PCB). They posses high fire point and excellent elec­
trical properties. In recent years their use has been banned throughout the
world, because once they are discharged in the environment they exhibit a
strong resistance to biodegradation. They accumulate in biological organ­
isms and end via the food chain in human body, thus posing a serious
health hazard.

5.2.4 Silicone Oils

Silicone oils represent an alternative to PCBs but they are rather expensive.
Their molecular structure’s main chain consists of silicone and oxygen and

Table 5.1 Chemical and Electrical Properties of Poly-butylenes

Characteristic lEC ASTM

Breakdown voltage, kV (2.5 mm) 40 > 35

Dielectric dissipation factor at 90°C <0.0005 0.0003
Resistivity ilm at 90°C 1.5 X 10'2 >10^"^
Relative permittivity at 90°C 2.2 2.2
Neutralization number mg KOH/g <0.03 < 0.04
Water content (ppm) — 40
Liquid Dielectrics 115

Table 5.2 Essential Properties of Alkyl-Aromatic Hydrocarbons

Gassing characteristics (hydrogen atmosphere) ml min“ ‘ at 80°C <70

Breakdown voltage, kV (2.5 mm) >60
Impulse breakdown voltage, kV (25 mm)
Positive 92
Negative 312
Dielectric dissipation factor at 90°C 0.0004
Resistivity, H-m at 90°C 10^2
Relative permittivity at 90°C 2.15-2.5

organic groups constitute the side chain. The chemical configuration of

polymethylesiloxane is shown in Figure 5.4. Chain lengths of up to 800
siloxane units and relative molecular weights up to 60,000 can be found.
Table 5.3 compares some important properties of commonly used min­
eral and silicone oils. The dissipation factor of silicone oils is independent

Figure 5.3 Hydrogen gas absorption of (1) alkylbenzene and (2) cable mineral
oil. (From Ref. 12.)
116 Chapter 5

C H 3 CH3 CHa
I I I
CH j— Si - 0_Si - 0 — Si — CHa
I I
CHs CHa \ CHa

Figure 5.4 Molecule of silicone oil; n is in the range of 10 to 1000.

of frequency and temperature. Their high long-term thermal stability at

150°C is particularly notable. Silicone oils are resistant to most chemicals.
They are oxidation resistant, even at higher temperature. On thermal dis­
sociation in an arc, nonconducting silicone dioxide (quartz) is formed from
the main chain. The higher admissible working temperature compared with
mineral oil allows volume-saving designs. Silicone oil is an acceptable
substitute for PCBs in transformers despite its slightly inferior nonflam­
mable properties. Its major drawbacks are its poor gas absorption and large
water content absorption. However, it is compatible with mineral oil and
PCBs, which make it well suited for retrofilling contaminated transformers.

Table 5.3 Properties of Commonly Used Insulating Oils

Property Unit Mineral oil Silicone oil

Breakdown field strength kV/m m 28 10

Volume resistivity (p) il-m 10“ ~ 10i^ 10*3
Dielectric constant s^. — 2.2 2.8
Dissipation factor at 25°C (1 MHz)
tan 8 — 0.001 0.0002
Density g/cm^ 0.91 0.96
Thermal conductivity W /K°m 0.14 0.16
Specific heat (Cp) cal/g/°C 0.53 0.36
Thermal stability limit °C 90 150
Flash point °C 145 > 300
Neutralization number (acidity) mg KOH/g <0.03 —

Pour point °C -4 0 -5 5
Dielectric impulse breakdown,
negative needle to sphere (25.4
mm gap) kV 145 —

Water content ppm 25 50

Source: Refs. 3-5 and 12.

Liquid Dielectrics 117

5.2.5 Esters
Natural ester such as castor oil has been used as a capacitor impregnant
for many years, but currently two types of synthetic esters are being used:
organic esters and phosphate esters. Others are under investigation.
Organic esters have high boiling points in relation to their viscosity
and, therefore, have high fire points. They have a good viscosity-
temperature relationship, and respond to inhibitor treatment to produce
products of excellent stability. The permittivity of selected esters is higher
than that of mineral and the silicone oils, but is lower than that of the
askarels. They are used extensively in capacitors.
The phosphate esters have a better fire resistance than silicone oil, but
not as good as the askarels. They have poor viscosity-temperature char­
acteristics, relatively poor electrical properties, but a higher permittivity
than mineral and silicone oils. They also generally have a poor hydrolytic
stability. Due to their high boiling point and low flammability, they are
therefore potential candidates both for transformers that are to be installed
in hazardous areas as well as in switchgear, where arc suppression in a
fluid system is of major importance.

5.2.6 Some Recently Introduced Oils

At present there is considerable short supply of naphthethic-based crude
oils. Moreover, in a continuing effort to replace PCBs and continue the
search for better quality oils, some new oils have been introduced in recent
years. These are being marketed under different commercial names, such
as high temperature or high molecular weight hydrocarbon oil, tetrachlo-
roethylene, perfluoropolyether and many others [6].

High Temperature or High Molecuiar Weight Hydrocarbon Oils

These are alternative to PCB fluids and are called HTHs. They have good
electrical insulating properties and adequate heat transfer properties. These
are chemically similar to regular mineral transformer oils, but they possess
higher boiling points and higher fire points. However, they have higher
viscosity which reduces heat transfer capability. General properties of
HTHs are given in Table 5.4.

Tetrachloroethyiene (C2CIJ
This is also a nonflammable insulating fluid. It can be used in mixtures
with mineral oil. It has very low viscosity and therefore gives excellent
heat transfer properties. Its mixtures with mineral oils also remain inflam-
118 Chapter 5

Table 5.4 Some Typical Properties of HTH and C2CI4 Fluids

Property HTH C2CI4

Flash point, °C 285 none

Fire point, °C 312 none
Expansion coefficient cc/cc/°C at 25°C 0.0008 0.00102
Pour point, °C -3 0 -2 2
Viscosity, cSt
100°C 16 0.36
50°C 85 0.42
25°C 350 0.55
Specific gravity, g/cm^ 0.877 1.620
Dielectric strength, ASTM D-877, kV 43 43
Impulse breakdown, kV
Negative polarity 118 —

Positive polarity 85 —

Dielectric constant, 2.38 2.365

Dissipation factor (%)
100°C 0.4 —

50°C 0.4 0.05

25°C < 10“^ —

Source: Ref. 6 © IEEE, 1992.

mable. However some toxic effects of this insulating fluid are also reported
[5].

Perfluoropolyether
This has recently been introduced in the European market with the trade
name Galden HT40 as replacement for PCBs. It is nonflammable oil, as
its boiling point exceeds 400°C. It possess low vapor pressure, thus it can
be used as a good heat transfer medium. Its molecular structure is given
in Figure 5.5. Since it has no C-H bonds and C-F bonds are much stronger
(single bond energy 4.6 eV), bond scission does not take place at normal
stress. Therefore no gaseous products are formed. Another interesting fea­
ture of HT-40 is the lack of moisture sensitivity, as it has no affinity for
water and it does not absorb or mix with water.
If any fluorine evolves during service it does not react with water to
form any acid (unlike PCBs which produce highly corrosive HCl). Because
of their higher density, water accumulated in transformer will float on top
Liquid Dielectrics 119

0 — CH — CF,
0 — CF,
m
CF,

Figure 5,5 Molecular structure of perfluoropolyether.

of the oil surface rather than at the bottom, as is the case in conventional
petroleum mineral oils. However, its viscosity and cost are high.

5.3 ESSENTIAL CHARACTERISTICS OF INSULATING OILS

Table 5.5 outlines a list of the most important characteristics of insulating

oils along with their typical accepted values. However, for the purpose of
design and operation, the following are very essential and are therefore
dealt with in some detail:

Table 5.5 Properties of Mineral Insulating Oils Considered Prior to, During and
After Their Long Use

ASTM test lEC

Characteristic method publication Typical values

Dielectric strength D-877 156 > 3 0 kV (1.0 mm)

D-1816 > 28 kV (1.02 mm)
Dissipation factor D-924 247 0.1-0.5% (90°C)
Impulse strength D-3300 897 145 kV
Dissolved-gas content D-831 567
Density, specific gravity D-1298 296 825 ~ 890 kg/m^
Viscosity D-445 296-A 3 ~ 16 cSt (+40°C)
Pour point D-97 296-A -30°C
Neutralization number D-974 296 < 0.5 mg KOH/g
Antioxidant content D-2668 666 <0.3%
Water content D-1533 733 < 8 0 ppm
Gassing characteristic D-2300 (AB) 628(A,B) - 3 5 to +35 mm^/min
Interfacial tension D-971 296 40 ~ 60 m N /m
Resistivity D-1169 247 3 — 10 G ilm
120 Chapter 5

1. Thermal transfer characteristics

2. Chemical stability against electrical stresses
3. Dielectric properties

5.3.1 Thermal Transfer Characteristics

In a liquid filled system (transformer, cable, circuit breaker, etc.), heat is
transferred mainly by convection. Under natural atmospheric cooling con­
ditions convection (N) is given as [7]:
AC
N (5.1)

where K = thermal conductivity, A = coefficient of expansion, C = spe­

cific heat per unit volume, v = kinematic viscosity, and n = 0.25 ~ 0.33.
It is clear that heat transfer is strongly dependent on K and to a lesser
degree on A and C, whereas it inversely varies as viscosity. Generally in
liquids, A and C do not vary much, but v varies greatly. Therefore, the
main factors that control heat transfer are K and v. Obviously, an increasing
value is preferable for systems likely to operate continuously at a high
temperature. On the other hand, a low value of K and high viscosity can
lead to localized overheating or even electrical “burn out.” Figure 5.6
shows the variation of viscosity with temperature. It is an important feature

Figure 5.6 Effect of temperature on viscosity in two commonly used insulating

oils.
Liquid Dielectrics 121

of the liquid that as temperature increases its viscosity falls off. This greatly
enhances the cooling effect.
Silicone oils do not exhibit this property appreciably, therefore this
viscostatic property coupled with poor heat capacity and falling thermal
conductivity can pose severe overheating problems at high temperature in
systems based on such insulants. Thermal conductivity and specific heat
values are therefore much more useful for evaluating temperature gradients
and possible hot-spot formation in places where the oil is trapped. Inside
oil-impregnated paper insulation, for instance, a temperature gradient of
IS^'C has been reported across a layer of only 0.4 mm of oil-impregnated
paper covering a copper conductor [8].

5.3.2 Chemical Stability Under Electrical Stress

In service, insulating liquids are subjected to thermal and electrical stresses
in the presence of materials like O2 , water, fibers and decomposition prod­
ucts of solid insulation, as well as oil soluble constituents of impregnating
varnishes and resins. These, either singly or in combination, promote deg­
radation of the liquid with the result that soluble solid and gaseous products
are formed, which can result in corrosion, impairment of heat transfer,
deterioration of electrical properties, increased dielectric losses, discharges
and arcing. In the absence of any remedial action, this cycle continues and
produces an ever-worsening liquid and equipment condition.
Degradation due to electrical and severe thermal stresses is indicative
of an equipment design or fault conditions, while the oxidation constitutes
the most important “in-service” deteriorating property of insulating liquids
in equipment subject to breathing. The rate of oxidation is increased by
the presence of materials acting as catalysts. Copper is one of the most
active of such materials [9]. Improper impregnation, however, results in
enhanced catalysis due to attack on the conductors by varnish acids. It also
produces heavy sludge deposits when incompletely cured impregnants dis­
solve in the liquid filling and precipitate after polymerizing or curing.
To retard oxidation, oils are inhibited with additives called antioxi­
dants. A widely used antioxidant additive is DBPC (2,6 di-tert-butyl per-
esol) and, to a lesser extent, DBP (2,6 di-tert-butyl phenol). Antioxidants
do not prevent copper dissolution in oil (Figure 5.7) but as long as they
are present (induction period) they prevent the formation of free-radical
peroxides, especially from the saturated hydrocarbon fraction, and the sub­
sequent chain oxidation reactions. During this period they themselves are
slowly oxidized and deactivated [8] and, when they have been almost com­
pletely consumed (end of the induction period), the above protection is no
longer available. Accelerated formation of peroxides is then observed, fol-
122 Chapter 5

Figure 5.7 Oxidation of insulating oil versus time: I, induction period; II, accel­
eration period; III, saturation period; (1) antioxidant content, (2) copper content,
(3) interfacial tension, (4) peroxide content, (5) acidity, (6) sludge content. (Re­
printed with permission from Ref. 1 © ASTM.)

lowed by acids and polar oxidation compounds and, finally, by oil-soluble

and -insoluble sludges. The relative saturation in the final stage of oxidation
results from the buildup of inhibiting phenolic forms of the aromatics. The
electrical properties (tan 8 and conductivity) are closely related to the ob­
served oxidation steps and the mechanism applies to paraffinic as well as
naphthenic oils [11].
The electrical characteristics such as electric strength, impulse strength,
DC volume resistivity, dielectric dissipation factor and relative permittivity
are generally of much importance. The maintenance of electrical properties
at an acceptable level ensures satisfactory equipment performance, reduces
ohmic losses and limits discharge inception within the liquid. The level of
test results is affected by the chemical constitution of the liquid, but with
the exception of impulse strength and permittivity, is predominantly due
to the presence of conducting contaminants such as fibers, water, particu­
late matter (dirt) dipolar and ionic or dissociated compounds.
Liquids can vary widely in electric strength. Power frequency break­
down depends considerably on the level of moisture and contaminating
particles. Similarly these factors also effect the dielectric loss and resistivity
of insulating oil. However, with filtration and removal of moisture these
can be considerably improved.
Liquid Dielectrics 123

5.4 STREAMING ELECTRIFICATION

Transformer failures due to streaming electrification have been reported

worldwide to occur within few years after commissioning [13,14]. It has,
therefore, received considerable attention since 1980. Charge separation at
interfaces between a moving fluid and solid insulation boundary can give
rise to the generation of substantial electric fields. Either alone or in com­
bination with the already existing charges in the fluid produced due to
aging of insulation and/or due to energization of the equipment can thus
lead to insulation failure.
Figure 5.8 illustrates the formation of a double layer at solid interface
(e.g., transformer pressboard, paper) due to ionic charges resulting from
dissociable impurities in the liquid. Enforced motion of the fluid strips
away the ions from this double layer and transports them downstream. The
resulting charge separation between the bound charges on the solid surface
and convected opposite-polarity charge can generate dangerous potentials.
Streaming electrification therefore depends mainly on the ionic species that
get adsorbed at the solid surfaces. In transformers, the -OH groups asso­
ciated with the pressboard/paper cellulose are the likely sites for the at­
traction of ionic species from the bulk of oil. Their accumulative effect
increases with aging and polymerization of cellulose, since as a result the
leeching of ionic species is enhanced. Increase in operating temperature

Fluid Flow

©0©
¡O il i
©el© ®
©0|© © © ©
©
©01© 0 © ©
© ©
© © ¡© (g © © ©
©01© (y. © © ©
© ® ©
001© ®
100® ®
Fixed D iffu sed
Layer Layer

Figure 5.8 Electrical double layer and ion distribution at the insulated paper/oil
interface.
124 Chapter 5

and presence of moisture also play a detrimental part to increase it further

[14,15]. Beside these factors, the velocity of the fluid is yet another con­
trolling parameter, since the shearing of charges at the solid/liquid inter­
face is intimately related to the rate of flow of the fluid. The transition of
liquid flow into turbulence itself becomes another source of charge gen­
eration. In transformer, the turbulence in liquid flow can occur in ducts,
windings or circulating pumps.
Currently a great deal of effort is being made, both in industry as well
as in several research centers, to understand these problems and to identify
the practical solutions. The initial industrial response so far has been to
reduce design velocities and curtail the use of pumps. The most innovative
approach has been adopted in Japan, where the use of charge suppressant
additives is widespread. Materials such as alkylbenzene and 1,2,3-
benzotriazole (BTA) are successfully being used to suppress the effects of
streaming electrification [14,17].

5.5 RECONDITIONING OF INSULATING OILS

Insulating liquids can be kept in service as long as their properties have

not deteriorated beyond specified levels, as described earlier (for details
see Table 5.5). For instance, their breakdown voltage strength should not
decrease below 30 kV. The limits for the resistivity of the liquid are 3-10
Gil • m, and for the dissipation factor 0.001 ~ 0.005 at 90°C. In more
highly stressed equipment such as cables, the limit for the dissipation factor
is set as low as 0.001 [12]. Similarly the level of acidity neutralization
factor of insulating liquids could be measured as the amount of potassium
hydroxide sufficient to neutralize the acids in 1 g of the liquid and its
acceptable level is <0.5 mg KOH/g. The acceptable level of water content
in power equipment is 15-80 mg/kg. The level is set even lower (0.1
mg/kg) for EHV cables, as their working electrical stresses are very high.
Similarly, diagnosis of the lifetime of transformer oil based on furfural
dissolved in oil has recently received much attention [8,18]. (For more
details see chapter 8, section 8.8.6.) The characteristics of the quality of
oil being used should be checked periodically. The frequency of testing
depends upon the atmospheric/ambient conditions, power rating, nature
loading, construction of the equipment and manufacturer’s
recommendations.
As the level set for any of the foregoing characteristics is approached,
measures for reconditioning the liquid should be taken. There is portable
equipment now available for reconditioning insulating oils while in service.
A method well known and in use is that of filtering and vacuum drying.
Liquid Dielectrics 125

While under a vacuum of about 1 kPa, the liquid is heated to about

30-60°C, which is above the boiling point of water at such a reduced
pressure. With large oil surfaces exposed to the vacuum, it becomes freed
from its moisture content, dissolved gases, as well as acids and particulate
matter. The process could include replenishing inhibitors and other addi­
tives in the liquid, as shown in Figure 5.9. Reconditioned oils usually have
about half the life of new ones [19].

5.6 ELECTRIC CONDUCTION IN INSULATING LIQUIDS

In an effort to understand the breakdown mechanism better, the study of

conduction in insulating liquids is being carried out since the period they
are serving as insulators or dielectrics in electrical equipment. Many books
and reviews published during the last 50 years reflects this aspect [20-27].

5.6.1 Polar and Nonpolar Liquids

Insulating liquids, in general, can be characterized as polar or nonpolar.
The dielectrics that have permanent dipoles are known as polar. In such
dielectrics, the asymmetry in their molecular structure leads to a permanent
displacement of positive and negative charge centers even in the absence
126 Chapter 5

of an external field. Typical examples of polar liquid dielectrics are water,

propylene carbonate and nitrobenzene. However they dissolve and disso­
ciate impurities so efficiently that they are difficult to keep in purified state
so that they exhibit high levels of resistivity. However, highly pure polar
liquids are attractive by virtue of their large permittivities, e.g., pure water
(Cj. = 80) and propylene carbonate (Cj. = 69).
The dielectrics that do not form dipoles appreciably in the absence of
external electric field are known as nonpolar. The dipoles formed in such
dielectrics, on the application of electric field, are not permanent and their
atoms return back to their original state once the electric field is removed.
Most of the insulating liquids used in power system applications are of
nonpolar nature.
When subjected to electrical stress both polar as well as nonpolar liq­
uids are characterized by a universal conduction current characteristic. Pig­

ia )

Electric field strength IkV/cmJ

(b)

Figure 5.10 Conduction current as a function of electric field strength: (a) trans­
former oil, test gap = 6 mm (from Ref. 1 © ASTM), and (b) idealized schematic
representation of different regions.
Liquid Dielectrics 127

ure 5.10a depicts current-field variation in transformer oil, whereas Figure

5.10b shows the general trend in liquids in which three distinct regions are
observed. Regions I and II appear under electric stresses that do not exceed
around 2 X 10^ Vm~* and therefore are considered to represent conduction
at low fields; whereas region III appears when these stresses exceed 10^
and represents the case of conduction under high field. In this sit­
uation the conduction current increases rapidly with slight increase in elec­
tric field strength. A further increase in the electric field strength leads
finally to breakdown (region IV).
In liquids, unlike in gases where natural radiation plays important role,
the major role is played by the dissociation of ionic impurities and solid
impurity particles. The host liquid may dissociate the impurity molecules
into ion pairs. These positive and negative ions, which escape recombi­
nation, drift to the electrodes and thus give rise to a flow of current in the
ohmic region. The ionic drift velocity and therefore the current increases
with increasing field, as illustrated in Figure 5.10. At intermediate fields,
the charge carriers are produced entirely by electro-chemical processes at
the interface between liquids and metal. A so-called double layer is formed
at the electrode [1]. Its significance is that positive ions preferentially, but
negative ions also, are created at this junction. At certain field strength the
ions will move to the electrodes at a rate faster than their generation rate
in the bulk of the liquid. The current then depends on their resultant effect.
This effect causes growth of space charge at electrodes [20] and thus leads
to the situation where the current tends toward saturation, as shown in
region II.

5.6.2 Conduction Current at High Fields

When the field applied to insulating liquids is sufficiently high, the current
tends to increase sharply for comparatively small increments of field. A
number of factors govern this current. Following are the most prominent
among them:
1. When the concentration of undissociated impurity molecules is
sufficiently high, the increase in electric field will lead to an enhancement
of the dissociation process. As a result, more free ion pairs are produced
and conductivity increases.
2. Reduction or oxidation of impurity molecules at the liquid/metal
interface may yield excess negative or positive ions. Such electrochemical
reactions are controlled by the double layer potential which in turn is de­
termined by the density and the nature of absorbed ions.
3. Solid impurity particles are also considered to contribute exten­
sively toward high field conduction. Since these particles may also be of
128 Chapter 5

semiconducting or insulating nature, prolonged action of electric stress at

the electrodes will either lead their removal from interelectrode gap or
neutralization at the electrodes, thus reducing the conduction current. The
temporal dependence of current after the application of stress at a fixed
value is found to decay in a manner which can be expressed as [21,28,29]:
i(t) "" io f (5.2)
where n decreases with time. In commercial liquids, additional conduction
current will be caused by the impurities beside particles, like water drop­
lets, acids, resins and cellulose fibers.
4. High field can lead to emission of electrons at cathode, whereas
at anode, field ionization can be initiated. According to the model proposed
for nonpolar liquids by Schmidt et al. [25], the field strengths required for
these processes to materialize are of the order of 1.5 ~ 2.0 X 10^ Vm“k
Recently, Denate et al. [30] demonstrated the occurrence of field emission
current in the form of short-duration pulses in cyclo-hexane. Similarly,
when the electrode tip is positive, the positive charge carrier formation due
to field ionization has been observed both in aromatic liquids as well as
in silicone oils [30,31]. Due to the injected space charge under high fields,
the convection currents are established which augment the ion drift veloc­
ities and hence the conduction. If, at a given condition the injected charge
has a density q, then it will give rise to a coulomb’s force of qE. Due to
action of this force on the bulk of liquid, hydrodynamic instability will be
caused. It will develop convective motion in the liquid which is found to
be always directed toward the opposite electrode [32]. This interelectrode
turbulent motion of liquid is known as electrohydrodynamic (EHD) motion
and the mobility of charge earners in this situation is given as [27,33]:
A^ehd = V s^sT p (5.3)
where p is the density of the liquid. The transport of ions in this situation
is enhanced by a factor V m , where M is the ratio of 1^ the true
mobility of the ions. For some polar and nonpolar liquids, M can vary in
the range of 2 ~ 100 [27].
5. Under high fields, conduction currents are usually accompanied
with randomly occurring pulses or bursts of currents. In the past they have
been reported to be either due to particles or ionization of microscopic gas
bubbles in the bulk of the liquid [20,22]. However, with the aid of elegant
ultrafast optical techniques, it has been demonstrated recently that these
pulses are associated with bubble generation process [34] and the devel­
opment and propagation of discharges called trees [35].
However, the dynamic nature of the process remained obscure due to
the random nature of these events and the inherent delay in the detection
Liquid Dielectrics 129

system. These difficulties were overcome by using more viscous liquids,

since the occurrence of prebreakdown pulses at a point cathode under DC
are observed to be strongly influenced by insulating liquid viscosity [36].
Their onset is always initiated at a critical threshold voltage level which is
a function of liquid structure and cathode tip radius. Their repetition rate
increases in an exponential manner with the increase in voltage. At elevated
stress levels, besides the single pulse activity, a regular burst activity also
occurs. Each burst is composed of a well-structured set of pulses which
increase in amplitude, initially in a linear fashion with time, before becom­
ing irregular. Increasing the viscosity results in an increase in the duration
of the burst and the time interval between the pulses within a burst of
current. Using the onset of each pulse or burst to trigger the capture of
shadowgraph always produces an associated discharge. Its size is found to
increase with burst duration. Figure 5.11 demonstrates bursts of current
pulses and associated shadowgraphs in silicone oil [37]. These studies also
support the view that current pulses encompassed in a burst of conduction
current are the consequence of periodic spark discharge inside a vaporized
cavity which is formed by the initial current pulse through the process of
Joule heating. At elevated stresses these discharges expand into well-
defined tree-like structures which span around 2/3 of the gap before the
breakdown ensued [36]. Under point anode, steady current is observed but
only with occasional random pulses of small amplitude. However, at ele­
vated stresses a sudden rise in current is associated with filamentary dis­
charge, which once initiated always leads to complete breakdown [35,36].
Individual branches of these discharges are filamentary, unlike the dis­
charge patterns observed under cathode which possess a thicker structure
and therefore resemble the bush-type and tree-type discharge patterns ob­
served under short-duration high voltage pulses.

5.7 BREAKDOWN IN INSULATING LIQUIDS

Efforts to understand breakdown mechanisms in a variety of liquid insu-

lants have been continuing for many decades. The pertinent features
of these studies have been summarized in a number of reviews
[1,20,22,24,38]. However, the picture is not impressive as compared to the
state of knowledge achieved for gases and solids. This is because the mo­
lecular structure of liquids is not simple and not so regular as compared
with solids and gases. For instance, transformer oil alone contains well
over 100 chemical compounds, and the fact that liquids tend to be contam­
inated with various impurities is a serious problem for fundamental studies.
Moreover, the transition from liquid to gas phase, which takes place during
130 Chapter 5

(C )

Figure 5.11 (a) Onset current pulse in 0.65 cSt silicone oil. Horizontal axis = 20
ns/div, vertical axis = 4 m A/div; 0.03 pC charge, (b) Regular current burst and
the effect of oil viscosity upon pulse repetition rate (i) 0.65 cS oil 40 ns/div, (ii)
10 cS oil, 200 ns/div, (hi) 100 cS oil, 400 ns/div, (iv) 1000 cS oil, 4 ms/div. (c)
Conduction current burst (time scale = 400 ns/div) with shadowgraph of associated
discharge in 100 cS oil. Arrow indicates instant of light flash. (From Ref. 36.)

the development of breakdown, still further complicates the phenomena

and hence their interpretations. Several promising hypotheses of break­
down based on particles and bubble effects were advanced in the late
1960s, but it has been necessary to modify and sometime even reject them
with the emergence of new experimental evidence.

5.7.1 Study of the Breakdown Process Without Optical

Techniques
Based on these studies, the breakdown strength of liquids is influenced by
various factors such as experimental procedure, electrode material and sur­
Liquid Dielectrics 131

face state, electrode geometry, presence of chemical impurities, presence

of physical impurities, molecular structure of oil, temperature and pressure
[24,38]. From such studies, breakdown theories like electronic theory, sus­
pended particle theory, cavitation theory and bubble theory were
postulated.

Electronic Breakdown Theory

This theory was extended at the early stages of investigations and used the
concept of electron avalanche operative in gases. Field emitted electrons
from a cathode were assumed to collide with the atoms of liquid molecules.
If enough energy is transferred during such collisions, some electrons
would be knocked off their atoms and drift with the original electrons
toward the anode. Thus electron avalanches similar to those in gas dis­
charge develop in the liquid and finally lead to breakdown [39-41]. This
model appeared quite reasonable, but since 1960 it has been rejected on
the following grounds:
1. The mean free path of electrons in liquids is very short (of the
order of 1 0 “^ cm) to enable them to acquire ionization potential
of 10 eV necessary for liquid molecules. Moreover, there was no
direct experimental evidence for an «-process.
2. It failed to explain pressure dependence of the breakdown process.
At pressures of 25 atm breakdown strength increases by 50%,
whereas at this pressure the mean free path of the electron is
hardly altered [43].

Suspended Particle Breakdown Theory

Suspended particles are always an integral part of liquids. In spite of rig­
orous cleaning techniques imparted both on liquids as well as test cells,
submicron sized particles cannot be removed from the system. The relative
permittivity of these particles, 8 2 , is higher than that of oil, e j. If we assume
them to be spherical of radius r, and if the applied field is E, then the
particles experience a force F such that:

1 3 (fi2 - Si)
F = ::: r --------- grad (5.4)
2 2 8j + 82

This force is directed toward areas of maximum stress. Particles will thus
align on the high stressed electrode and start forming a bridge, which could
lead to gap breakdown. Similarly if particulate matter is fiber it will get
polarized due to the presence of moisture on its surface and move along
132 Chapter 5

converging fields. Assuming hemispherical tips with radius r, the charge q

at either end of fiber would be [19]:

± q = TTx\e2 ~ fiiKE (5.5)

where Cq is the permittivity of free space. If the field in the gap is non-
uniform and 6 2 » 6,, the driving force will be
F = r^e^ grad E (5.6)
When a fiber reaches either electrode, its outward tip would act as exten­
sion of the electrode and cause field intensification and thus attract more
fibers, thereby forming a bridge in the gap. This can lead to breakdown
via joule heating of the bridge and its surrounding liquid.
Although this theory did explain the strength of liquids containing
large amounts of particles, it is unlikely to be extended to pure liquids.
Moreover, particles have been seen on several instances to bridge the gap,
while discharge occurs in a different region and still at higher voltages
[43]. This means breakdown involves some other mechanism. Nevertheless,
particles may be instrumental as an aid in the process of breakdown.

Cavitation Theory o f Breakdown

This theory was proposed by Krausucki [44], and is based on the concept
that whenever a particle comes in the high field region, the presence of
enhanced field on its surface will generate an electromechanical pressure
(P^) tending to lift the liquid off the particle surface against the opposing
hydrostatic pressure (P^) and pressure due to surface tension (P^^). This
action will develop a region of zero pressure, thus forming a vacuous
cavity. Electron bombardment of the walls of cavity will sustain its growth,
which will eventually lead to breakdown. The critical condition for zero
pressure generation and vacuous cavity formation is given as:

Pm - Ph + Pst (5.7)
If m is the field intensification factor at the surface of particle, then P^ =
1/2 e(mE)^ and P^t = 2i///r; where if/ = surface tension and r = particle
radius. Putting these values in equation (5.7) leads to the expression:

I (mE)= - P, + ^ (5.8)

Krasucki selected m = 4.2, therefore breakdown strength in this case

becomes:
Liquid Dielectrics 133

Eb - 0.337 Vm- (5.9)

Using particle radii of 100 A and 250 A (particle size usually produced in
spark breakdown), Krasucki showed that the pressure variation of the di­
electric strength measured by Kao and Higham [42] is contained within
the theoretical estimates as shown in Figure 5.12. However this theory fails
to explain the effects of electrode material, electrode separation and dis­
solved gases.

Bubble Theory o f Breakdown

According to this theory [45], a low density vapor bubble is generated in
the liquid by the injection of large leakage currents at the cathode protru­
sions. By this process local vaporization can occur in a few milliseconds.
If H is the heat needed to vaporize a liquid from ambient temperature T^,
then for a unit mass of liquid:
H = C( T , - T J + L (5.10)
where C = specific heat at constant pressure, L = latent heat of vapori-
zation. Tb = boiling temperature, and == ambient temperature. Near

Figure 5.12 Variation of breakdown strength of n-hexane with hydrostatic pres­

sure (a) r = 100 A°; (b) r = 250 A°; (A) experiments of Kao and Higham (Ref.
42).
134 Chapter 5

breakdown the emission current from the cathode is space charge limited
and is given as:

I = AV^ (5.11)

where x is in the range of 1.5 ~ 3 and A is a constant. It follows that the

local energy input during the applied high voltage pulse duration can
be expressed as:

H = A 'W t (5.12)

where A' is constant. Combining equations (5.10) and (5.12) yields the
relationship between E and H as:

A'E^ CAT, - T J + L (5.13)

This is the thermal breakdown criterion and exhibits a marked pressure

and temperature dependence since T, increases with pressure. It also ex­
plains the effect of molecular structure of the liquids on breakdown. Figure
5.13 exhibits the fidelity of equation (5.13) when tested with respect to the
experimental data of Kao and Higham [42]. However, the main objection
to this model has been the simple heat transfer treatment based on the
steady state equation for a phenomena which needs to be described by
transient heat fiow dynamics.

Figure 5.13 Pressure dependence of breakdown strength of n-hexane. Solid line

is due to equation (5.13) with x = 1.5, whereas data points are from Ref. 42.
Liquid Dielectrics 135

5.7.2 Study of the Breakdown Process Using Optical

Techniques
It is clear that no single concept so far proposed can explain all experi­
mental observations in a unified manner. However, with the advent of fast
electro-optical techniques, our understanding of breakdown in liquids has
been advanced tremendously. With these techniques, once a voltage pulse
is applied any perturbations occurring in the electrode gap can be easily
visualized under magnification by taking a photograph of each fast-
occurring event. Verification of the bubble concept was first put to test
using ultra high speed photography by Hakim and Higham [46] and later
by several others [47-53]. They all confirmed that near breakdown, stream­
ers emerge from the high voltage needle electrode that resemble in struc­
ture like bush or tree as shown in Figure 5.14. If the field at the originating
electrode is critical than they grow out in the liquid toward the opposite
electrode. Actual breakdown is preceded by the formation of secondary
streamers which grow much faster than the primary ones.
To study the phenomena in more detail and also to see photographi­
cally the effects of various experimental parameters on the breakdown cri-

Figure 5.14 (a) Cathode initiated bush type and (b) anode initiated filamentary
discharge in n-hexane.
136 Chapter 5

teria (such as chemical additives, particulate additives, viscosity and rate

of rise of the applied voltages), liquids varying from simple hydrocarbons
and silicone oil to more complex technical fluids were used. Various so­
phisticated electroptical devices were employed. However, the most pop­
ular diagnostic methods presently being used are the shadowgraph and
Schlieren techniques of photography. Salient features of these are explained
next in brief [36].

Schlieren and Shadowgraph Optical Systems

The main elements of these two systems are shown in Figure 5.15. The
light source at A can be a spark in a gas or a nanosecond light pulse laser
along with a collimating lens, providing an intense parallel light beam. In
case of the Schlieren system, the probing light beam passes through the
liquid between the electrodes at B before being focused on the knife edge
at C. The knife edge can be moved with a micrometer so as to cut off all
or part of the light that would pass C. Any light that might skirt around
the knife edge would be focused on to a still or moving film at D. If the
knife edge is critically adjusted so that it just cuts off all the light, then on
the occurrence of a discharge in the gap there is an associated temperature
and density change in those parts of the liquid affected by the discharge.
The change in the refractive index of effected part will cause the beam to
be deflected. It may then pass beneath the knife edge and be focused at
D. The effect is as if the low density region emits light, whereas in reality
the light is derived from the illuminating beam and refracted by the dis­
charge. Therefore, this technique is more useful for studies of problems in
which a precise density gradient is to be probed, since the angle of refrac­
tion of the ray passing through the object is directly linked to the density
gradient.
In the case of the shadowgraph system of photography, the knife edge
is not used and the probing light rays are either absorbed, reflected or
refracted out of the low density region (the discharge) and therefore do not
reach the image plane. The region in which this phenomena occurs is thus
represented by its shadow in the image plane at D. Comparatively, sha-
dowgraphy is the diagnostic method which is easy to set up and gives an
economical and qualitative description of shape, size and different density
regions and of strong gradients occurring in the liquid under the application
of high voltage.
To capture an event associated with the applied voltage pulse in the
electrode gap, synchronization of voltage pulse output at B, the light pulse
at A and camera at D is essential. Alternatively, if events associated with
current pulse are to be observed, then the synchronization of this pulse
Liquid Dielectrics 137

(b)
Figure 5,15 Essential elements of schlieren/shadowgraph system with synchro­
nization of (a) applied voltage pulse and (b) induced current pulse.

with light source and camera is essential. Pictures taken at D could be

single-frame or multiframe photography. For single-frame operation, use
is made of regular cameras; however, for the purpose of multiframing, ultra
high speed image converter cameras (ICCs) are being used with or without
image intensifiers [1]. By multiframe photography, usually 10-12 frames
138 Chapter 5

are obtained with a speed which varies over 2 X 10^ ~ 10^ frames per
second and with exposure time ranging from 100 ns to 10 ms per frame
[30]. More recently, digital frame stores are also being used, where the
events captured by a CCD camera are converted into a digital form and
can be viewed on a TV monitor, and saved or processed by the computer
and printed at will [36,54,55].

Experimental Results
All of the results compiled so far confirm the role of the electrode/liquid
interface, the discharges (steamers) produced in the prebreakdown regime,
and their fast propagation in the interelectrode gap which leads to final
breakdown.
Earlier workers were mostly confined to visualization of the failure
mechanisms in liquid under unrealistic overvoltages. However, recent in­
vestigations [36,54,56] deal in detail with the initial stage of discharge
growth and its relation to the molecular/liquid structure and other associ­
ated characteristics. Based on these and many other findings [49-56], mod­
els of cathode and anode initiated breakdown mechanisms have been
forwarded.
Breakdown Processes at Cathode
With the increase in voltage to a critical threshold level, either a thin or
spherical image of low density region appears at cathode which expands
with time. Experiments carried out in low viscosity fluids and both under
DC fields [57] as well as short duration voltage pulses [36,58] have shown
that this low density region consists of a vaporized bubble which expands
and collapses following Raleighs theory [59]. According to this theory, the
time taken for a cavity to collapse (r) limited by inertia and driven by the
ambient pressure is given as:

T - 0.915 P (5.14)
LPaJ

where R^ = cavity maximum radius, p^ = ambient pressure and p = liquid

density. If the cavity is formed from an initial impulse of energy (i.e., in
a very short time), then the growth and collapse will be symmetrical in
time.
Figure 5.16a illustrates the growth and collapse of cavity produced in
0.65 cSt silicone fluid following the application of a single 632 ns high
voltage pulse. The cavity takes several microseconds to grow to R^ before
it starts to collapse in size with almost at similar rate to that of its expan­
sion. The cavity size rebounds after reducing down to 25 pum and the
Liquid Dielectrics 139

(a)

( b)

Figure 5.16 (a) Cavity radius versus time in 0.65 cSt silicone oil, under the ap­
plication of a 632 ns duration 15 kV pulse, (b) Experimental collapse and expansion
time of cavity versus theoretical time according to equation (5.14). (From Refs. 36
and 58.)

process is repeated in several cycles of rapidly decreasing amplitudes, until

the cavity comes to an equilibrium state with surrounding liquid. Figure
5.16b compares the variation of theoretical value of r estimated from
and based on equation (5.14) to the experimentally observed values of both
the contraction as well as expansion of the cavity. It is clear that the ex­
perimental values conform in an excellent manner with the theoretical val­
ues and strongly exhibit the inertia limited expansion and contraction of a
bubble.
Figure 5.17 depicts growth of cavities with time for 0.65, 10 and 100
centistoke (cSt) viscosity silicone oil. These images cover the initial growth
140 Chapter 5

MU

Ml
0.5
0.2 0.3

1.2
0.5 1.4

0.8
4 1.5
1.8

2.8
1.0 1.5

#
3.4
1.8 2.6

3.8
2.1

(a) (b) (c)

Figure 5.17 Shadowgraphs illustrating the growth of the cavity with time, and
the development of wave-like instability in silicone oil, at 7.5 kV (50 p m scale bar
on each picture). Delay times are marked in ^ts. Oil viscosity (a) 0.65 cSt; (b) 10
cSt; and (c) 100 cSt. (From Refs. 36 and 54 © IEEE, 1991.)

of cavity, the formation of the instabilities and then propagation of these

instabilities deep into the gap. With the increase in voltage or with the
increase in pulse duration more instabilities appear and they elongate into
filaments. Thus the cavity develops into a bush-type structure, and these
figures are referred in literature as primary streamer elements. If the voltage
magnitude is small, these streamers detach from the cathode and start frag­
Liquid Dielectrics 141

menting into bubbles. However, if voltage is significant, these streamer

elements propagate with even larger velocity toward the anode. At break­
down voltage level, when primary streamer elements cross 1/3 to 3/4 of
the electrode gap, then secondary streamers emanate from primary ones
and finally impinge at the anode, causing breakdown. The secondary
streamers possess thinner filaments and their growth rate approaches sonic
speed and is almost an order of magnitude higher than that of primary
elements. These structures have been found to be conductive and are also
found to be suppressed with the increase in ambient pressure. Light pulses
are also reported to occur in synchronism with the current pulses that
appear as the initial cavity begins to expand. The following description
represents a generally accepted mechanism for initiation and propagation
of negative streamers.
Energy for cavity generation and expansion comes from a current pulse
from the cathode point. Microbubbles (<10 nm in diameter) are created
by rapid localized injection of current pulse from the point cathode, and
this energy is converted into heat, which causes evaporation of liquid and
provides driving force. Beyond kinetic limits, the bubble expands by the
dual action of internal vapor pressure and the force generated by the elec­
trostatic field. Inside the cavity the pressure will decrease as the structure
expands. As the pressure drops, the mean free path of electrons being
injected from the cathode increases and they are thus able to gain energy
from the field and produce impact ionization and/or breakage of C-C and
C-H bonds of oil. Eventually the Paschen’s minimum limit will be reached
and a sudden discharge will occur. The external circuit will recognize it as
a current pulse. This way the internal field in the cavity controls charge
injection while external field, largely determined by the shape of cavity,
appears to control its growth.
From this cavity surface appear the streamer filaments. Watson [60]
has treated this problem as the Raleigh-Taylor instability. According to
this theory if one medium expands within another medium of different
density then hydrodynamic instabilities appear at the interface. In the pres­
ent case, the field-driven interface with laterally mobile charges corre­
sponds to an EHD instability. The amplitude of instabilities grows as the
field at the outward tip of the elements increases. Fields in excess of 10^
V/cm have been calculated to exist in front of these elements. This is how
the primary branches of streamers grow at rates that are of the order of 1
km/s or less. The secondary streamers emanate from primary filaments
and grow at a rate in excess of 10 km/s. At present, little is known about
the cause of these secondary streamers except that presence of impurities
may be involved [56].
142 Chapter 5

Breakdown Processes at Anode

Generally the positive streamer growth rates are consistently higher than
those for the negative streamers. The structure of the two types are also
different. The positive streamers are mostly filamentary as compared to
thicker and bushier patterns of negative streamers as shown earlier in Fig­
ure 5.14. Initiation voltages are always higher with positive polarity than
with negative polarity. Once initiated, provided that the voltage is main­
tained, all but the smallest positive discharges inevitably lead to
breakdown.
There are three stages of growth for streamers that originate at the
point anode. The first is the bushy, subsonic structure which occurs mostly
near the onset voltage levels or when the insulants are subjected to elevated
atmospheric pressure. The second is a thin filamentary structure which
propagates near sonic velocity. The third mode is an order of magnitude
faster than the second [56,61].
Light emission in the form of pulses also occurs with the growth of
positive streamers. Light emission is similar for anode and cathode stream­
ers, but the growth of the anode streamer is much smoother than for the
cathode streamer.
Various models have been proposed to explain the propagation of an­
ode-initiated streamers. However each is supported by a limited set of
experimental results, and none of them gives an adequate explanation for
all the experimental phenomena reported so far. According to the model
of Devins et al. [38], the positive steamers grow by a field ionization
process. Thermalized electrons present in the liquid (due to the liquid’s
contact with the anode) are focused near the anode surface, where favorable
sites exist for electron transmission that are characterized by a large en­
hancement factor and a low work function. Such a site then enhances the
local field and causes field-induced ionization of the liquid column existing
between the focused charge and the anode spot. The conductive streamer
can be represented as an extension of the electrode in liquid. The tip of
the streamer replaces the surface spot on anode and the field ionization
process can repeat itself in any direction in which focused charges can
accumulate near the tip. Overall, the streamer grows steadily but in short
spurts.
Field ionization of liquid at the head of the streamer filament has also
been put forward by Chadband [62]. According to this model, propagation
of the streamer tip depends upon the local tip field. The local field is taken
as the vector sum of the applied (Laplacian) field and the Coulomb field
arising from the positive charge at the streamer tip. Growth occurs through
the action of the local tip field on the liquid ahead of the tip. Negative
Liquid Dielectrics 143

charge is released and falls into the tip, leaving behind a net positive charge
which becomes an extension of the tip. This process causes both heating
and dissociation of liquid, making the filament visible optically.
Watson [63] considers that positive streamers propagate as columns of
hot liquid that convert to high-pressure vapor channels in volatile fluids,
whereas Feleci [64], on the basis of energetic considerations, concludes
that steamer filaments probably contain an ionized gas phase. Lesaint [65],
through meticulous experimental technique, showed that these filaments
expand and collapse like cavitation bubbles as they are composed of vapor
whose pressure varies with time and space. This vapor is created mainly
from energy dissipated at the fast-moving filament tip.
Postbreakdown Events
When a streamer bridges the gap, an arc begins to develop. This devel­
opment is characterized by a rapid increase in current, a rapid decrease in
the voltage and the emission of visible light. During this stage the resis­
tance of the electrode gap approaches the wave resistance of the connecting
conductors [1]. It has been found that once the breakdown is initiated in
n-hexane, the current rises to values of more than 100 amperes in 4 ns
[66]. The establishment of the arc deposits significant energy in a small
volume in the fluid and thus raises that volume to a high temperature.
Upon arc initiation, an intense emission of light occurs and a shock wave
is produced which propagates away from the arc at the sonic velocity.
Liquid vaporizes quickly to make a gas column surrounding the arc. This
column is quickly transformed into a bubble which expands in time due
to the heat energy. The bubble then collapses in a few seconds by disin­
tegrating into smaller bubbles that eventually lead to the liquid surface or
are finally absorbed by the liquid.

REFERENCES

1. R. Bartnikas (ed.), E n g in e e r in g D i e le c t r i c s Vol. 11: E l e c t r i c a l I n s u la tin g L iq ­

u id s , ASTM, Philadelphia, 1994.
2. B. Seymour (ed.), H a n d b o o k o f P o ly o le fin s , Marcel Dekker, Inc., New York,
1993.
3. A.C.M. Wilson, I n s u la tin g L iq u id s , Peter Peregrinus Ltd., London, England,
1980.
4. D. Kind and H. Karner, H ig h V o lta g e I n s u la tio n T e c h n o lo g y , Fried, Vieweg
and Sons, Braunschweig, Germany, 1985.
5. W.T. Shugg, H a n d b o o k o f E le c t r i c a l a n d E le c tr o n ic - I n s u la tin g M a te r ia ls ,
IEEE Press, New York, 1995.
144 Chapter 5

6.R.C.M. Pompilli and C. Mazetti, Electrical Insulation Magazine, Vol. 8, No.

1, pp. 28-32, 1992.
7. A. Bradwell, E l e c t r i c a l I n s u la tio n , Peter Peregrinus Ltd., London, England,
1983.
8. M. Daval, J. Aubin, Y. Gigurre, G. Pare and Y. Langhame, IEEE Trans, on
Elect. Insul. Vol. EI-17, No. 5, pp. 414-422, 1982.
9. T. Tanaka, T. Okamoto, K. Nakanishi and T. Miyamote, IEEE Trans, on Elect.
Insul. Vol. 28, No. 5, pp. 826-844, 1993.
10. D.W. Murray, Proceedings of World Petroleum Congress, Vol. 11, No. 4, pp.
447-457, 1983.
11. C. Lamarre, J.P. Crine and M. Duval, IEEE Trans, on Elect. Insulation, Vol.
EI-22, No. 1, pp. 57-62, 1987.
12. T. Tanaka and A. Greenwood, A d v a n c e d P o w e r C a b le T e c h n o lo g y , CRC Press,
Boca Raton, Elorida, 1983.
13. T. Tanaka, T. Okamoto, K. Nakanishi and T. Miyamoto, IEEE Trans, on Elect.
Insul. Vol. 28, No. 5, pp. 826-844, 1993.
14. J.K. Nelson, IEEE Elect. Insul. Magazine, Vol. 10, No. 3, pp. 16-28, 1994.
15. LA. Metwally, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 3, No. 2, pp.
307-315, 1996.
16. M. leda, Y. Yanari, T. Miyamoto and M. Higaki, EPRI Workshop on Static
Electrification, EPRI, San Jose, California, 1992.
17. M. leda, K. Goto, H. Okubo, T. Miyamoto, H. Tsukioka and Y. Kohno, IEEE
Trans, on Elect. Insul., Vol. 23, No. 1, pp. 153-157, 1988.
18. lEC-Draft 1198, “Methods for Determination of 2-Furfural and Related Com­
pounds” , lEC, 1991.
19. M. Khalifa, H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c ti c e , Marcel Dekker,
Inc., New York, 1990.
20. A.A. Zaky and R. Hawley, C o n d u c tio n a n d B r e a k d o w n in M in e r a l O ils , Peter
Peregrinus Ltd., London, England, 1973.
21. I. Adamczewski, I o n iz a tio n , C o n d u c tiv ity a n d B r e a k d o w n in D i e le c t r i c L iq ­
u id s, Taylor and Francis Ltd., London, England, 1969.
22. T.J. Gallagher, S im p le D i e le c t r i c L iq u id s , Calendren Press, Oxford, England,
1975.
23. A. Gemant, I o n s in H y d r o c a r b o n s , Interscience Publishers, New York, 1962.
24. A.H. Sharbaugh, J.C. Devins and S.J. Rzad, IEEE Trans, on Elect. Insul., Vol.
EI-13, pp. 249-276, 1978.
25. W.F. Schmidt, IEEE Trans, on Elect. Insul., Vol. E l-19, pp. 389-418, 1984.
26. K.C. Kao, IEEE Trans, on Elect. Insul., Vol. EI-11, pp. 121-128, 1976.
27. N. Felici, IEEE Trans, on Elect. Insul., Vol. EI-20, No. 2, pp. 233-238, 1985.
28. T. Matuszewski, Acta Physica Polonica, Vol. A48, pp. 861-869, 1975.
29. A.A. El-Sulaiman, A.S. Ahmad and M.I. Qureshi, IEEE Trans, on Elect. In­
sul., Vol. E l-16, No. 5, pp. 453-457, 1981.
30. A. Denat, J.P. Gosse, IEEE Trans, on Elect. Insul., Vol. 23, No. 4, pp.
545-554, 1988.
Liquid Dielectrics 145

31. J.G. Rabe and W.F. Schmidt, Journal of electrostatics, Vol. 7, pp. 253-266,
1979.
32. T. Takashima, R. Hanaoka, R. Ishibashi and A. Ohtsubo, IEEE Trans, on
Elect. Insul., Vol. 23, No. 4, pp. 645-658, 1988.
33. R Atten, IEEE Trans, on Dielectric and Elect. Insul., Vol. 3, No. 1, pp. 1-17,
1996.
34. M.I. Qureshi and W.G. Chadband, IEEE Trans, on Elect. Insul., Vol. 23, No.
4, pp. 715-722, 1988.
35. O. Lessaint and R. Tobazeon, IEEE Trans, on Elect. Insul., Vol. 23, pp.
941-954, 1988.
36. M.I. Qureshi, Ph.D. thesis. University of Salford, U.K., 1992.
37. M.I. Qureshi and W.G. Chadband, Proc. of 10th ICDL, Grenoble, France,
IEEE Conf. Reed. 90CH 2812-6, 1990.
38. J. C. Devin, S.J. Rzad and R.J. Schwab, Journal of Applied Physics, Vol. 52,
pp. 4531-4545, 1981.
39. A. von Hippel, Journal of Applied Physics, Vol. 8, pp. 815-832, 1937.
40. T.J. Lewis, Journal of Applied Phys. Vol. 27, pp. 645-650, 1956.
41. D. W. Swan, Proceedings of the Physical Society, Vol. 78, pp. 423-432, 1961.
42. K. C. Kao and T.B. Higham, Journal of the Electrochemical Society, Vol. 108,
pp. 522-528, 1961.
43. A. Ruffini, D.A. Hoch and J.P. Reynders, Proc. 6th International Symposium
on High Voltage Engineering, New Orleans, Louisiana, Paper No. 13.14,
1989.
44. Z. Krausucki, Proceedings of Royal Society London, Vol. A-294, pp.
393-404, 1966.
45. PK. Watson and A.H. Sharbaugh, Journal of Electrochemical Society, Vol.
107, pp. 106-112, 1960.
46. S.S. Hakim and J.B. Higham, Nature, Vol. 189, pp. 966-996, 1961.
47. W.G. Chadband and T.G. Wright, British Journal of Appl. Physics, Vol. 16,
pp. 305-313, 1965.
48. W. Hauschild, Ph.D. thesis. Technical University of Dresden, Germany, 1969.
49. W.G. Chadband and J.H. Calderwood, Journal of Electrostatics, Vol. 17, pp.
75-91, 1979.
50. E. O. Foster and PP. Wong, IEEE Trans, on Elect. Insul., Vol. E l-12, pp.
183-187, 1977.
51. H. Yamashita and H. Amano, IEEE Trans, on Elect. Insul., Vol. EI-20, pp.
247-255, 1985.
52. R.E. Hebner, Annual Report CEIDP, pp. 26-34, 1983.
53. O. Lessaint and P. Goumey, IEEE Trans, on Elect. Insul., Vol. 1, No. 4, pp.
702-708, 1994.
54. 5. M. Arghi, M.I. Qureshi and W.G. Chadband, IEEE Trans, on Elect. Insul.,
Vol. 26, No. 4, pp. 663-672, 1991.
55. M.I. Qureshi and W.G. Chadband, Proc. of the International Conference on
Partial Discharge, lEE Publ. No. 378, pp. 9-1 0 , 1993.
56. E.O. Foster, Journal of Physics D. Appl. Phys., pp. 1506-1514, 1990.
146 Chapter 5

57. R. Kattan, A. Denat and N. Bonifaci, IEEE Trans, on Elect. Insul., Vol. 26,
No. 4, pp. 656-662, 1991.
58. M. I. Qureshi, W.G. Chadband and RK. Watson, Proc. of 6th lEE Conf. on
Dielectric Materials and Measuring Apparatus, pp. 89-92, Manchester, Eng­
land, 1992.
59. L. Rayleigh, Phil. Magazine, Vol. 34, pp. 94-98, 1917.
60. RK. Watson, W.G. Chadband and M.S. Araghi, IEEE Trans, on Elect. Insul.,
Vol. 26, No. 4, pp. 543-559, 1991.
61. G.J. Fitzpatrick, Proc. of CEIDP, 1985, IEEE Conf. Record 85CH2I6J-9, pp.
26-32, 1985.
62. W.G. Chadband, IEEE Trans, on Elect. Insul., Vol. 23, pp. 697-706, 1988.
63. RK. Watson, T. Sufian, W.G. Chadband and H. Yamashita, Proc. of 11th IEEE
Inti. Conf. on Conduction and Breakdown in Dielectric Liquids, Baden, Swit­
zerland, pp. 234-238, 1993.
64. N. Felici, IEEE Trans, on Elect. Insul. Vol. 23, pp. 497-503, 1988.
65. O. Lesaint and P. Goumey, Journal of Physics D: Appl. Phys., Vol. 27, pp.
2111-2116, 1994.
66 . J. Fuhr and W.F. Schmidt, Journal of Applied Physics, Vol. 59, pp.
3702-3708, 1986.
Solid Dielectrics

6.1 INTRODUCTION

Solid insulating materials are used extensively in all types of electrical

power networks, devices and substation equipment. The equipment size
and operating limitations are dictated by the type and amount of material
required for insulation. In the early days of the electrical industry, engineers
had to adapt varnishes, natural resins and petroleum residues as saturents
and coatings for tapes used to wrap coils and cables. Today, with the advent
of polymers, as well as the availability of a variety of glasses and ceramics,
an almost unlimited number of insulating materials is available, and the
problem becomes one of selection rather than adaption.
Solid dielectrics have high breakdown strength as compared to liquids
and gases. A good dielectric should, besides having high dielectric strength,
have low dielectric loss and high mechanical strength and stiffness, be free
of gaseous inclusions and moisture, and be resistant to thermal and chem­
ical deterioration. It should also be insensitive to ambient condition at its
site of application. Ozone resistance, impermeability, hygroscopic resil­
ience, low water absorption, radiation stability, among others, are the ad­
ditional requirements. Efficient utilization and improvement in the perform­
ance of such materials, therefore, calls for not only a detailed knowledge
of their essential properties but also the mechanisms which cause their
degradation and failure. This chapter illustrates both of these essential pros­
pects of solid dielectrics.

147
148 Chapter 6

6.2 SOLID INSULATING MATERIALS

A larger proportion of commercial insulating materials that are used in

electrical power apparatus are solids. With the advent of new insulating
materials in the past three decades, many earlier conventional materials
that were commonly used in apparatuses like electrical machines, cables
and capacitors are being discarded now. The most common method of
classifying solids is by using their chemical composition, such as organic,
inorganic and synthetic polymers. Some examples are outlined in Table
6.1. From their application’s point of view, they can be classified as: (1)
thermoplastic compounds, (2) thermosetting compounds, and (3) embed­
ding and jacketing compounds.
Organic materials are derived from either vegetable or animal matter.
They possess good insulating properties. However, they deteriorate rapidly
if their operating temperature exceeds 100°C. They are mostly employed
after treatment with varnishes or impregnation in oil. Examples of such
insulators include paper and pressboard that are commonly used in oil-
filled equipment such as cables, capacitors, panel boards and transformers.
Inorganic materials are distinctly different from organic substances. As
a rule, they do not show any appreciable fall in either mechanical or elec­
trical quality at 100°C and may retain their properties up to a working
temperature of ~250°C. Because of their compact physical structure, they
do not absorb oil or varnish, with the exception of the fibrous asbestos
material. Inorganic solids are difficult to fabricate but they are very good
dielectrics. The most important members of this group are glasses and
ceramics. Glass serves as a material for power-line insulators whereas por­
celain, in the form of a bushing, was the first ceramic material to be used
by the electrical industry [ 1 ].

Table 6.1 Classification of Some Commercial Solid Dielectrics

Synthetic polymers

Organic Inorganic Thermoplastic Thermosetting

Amber Ceramics Perspex Epoxy resins

Paper Glass Polyethylene Phenolics
Pressboard Mica Polypropylene Melamine
Rubber Fiber glass Polystyrene Urea formaldehyde
Wood Enamel Polyvinyl chloride Crosslinked polyethylene
Resins Polyamid Elastomers
Polycarbonate
Solid Dielectrics 149

Synthetic polymers include all types of polymeric materials that have

been produced by various industrial processes. Polymers are generally di­
vided into two groups: thermoplastic and thermosetting. The former have
low melting temperatures, in the range of 100-120°C. However, they are
flexible and can be molded and extruded at temperatures below their melt­
ing points. These properties make thermoplastics extremely desirable as
insulants for high-voltage cables. Thermosetting polymers are heat-curing
solids. On heating they acquire substantial mechanical strength and hard­
ness. Epoxies, introduced in 1947, are predominant in embedding com­
pounds. In combination with phenolics they make useful thermosetting
compounds. Polycarbonate was introduced in 1950. It has excellent di­
electric properties and is also corona resistent. Polyethersulfone resins were
introduced in 1973. They possess a temperature index rating of 180°C,
which is the highest for any thermoplastic. Specialty resins introduced
since 1980 include poly acrylate and polyetherether ketone (PEEK), which
can be applied up to a temperature of 200°C and are also highly resistant
to hydrolysis [2]. Some important types of insulating materials that are
commonly used as high voltage insulants along with potential areas of their
applications are described next.

6.2.1 Dielectric Paper and Boards

Dielectric paper and boards are produced from a variety of materials, in­
cluding wood, cotton, organic fibers, glass, ceramics and mica. The dis­
tinction between paper and board is not specific but paper is generally
<0.8 mm thick, whereas boards are >0.8 mm thick. For more than 6 -mm
thickness, boards are laminated with adhesive to get the desired thickness.
The boards are also referred to as pressboard, transformer board or fuller
board.
The paper normally employed for insulation purposes is a special va­
riety known as Kraft paper. The thickness and density of papers vary de­
pending on their application. Low-density paper (0.8 g/cm^) is preferred
in high-frequency capacitors and cables, while medium density paper is
used in power capacitors. High-density papers are preferred for energy
storage capacitors and for the insulation of DC machines.
Paper is hygroscopic, therefore, it has to be dried and impregnated
with mineral oil, synthetic oils or vegetable oils. The relative dielectric
constant of impregnated paper depends upon the permittivity of cellulose
(the base material) and the permittivity of the imprégnant and the density
of the paper. Paper is also used in the form of hardboard and pressboard.
Hardboard is produced by compression of paper with epoxy or phenolic
resins. It is used as supporting material and insulating barrier. On the other
150 Chapter 6

hand, soft paper or pressboard is used with impregnation in transformers

and bushings.

6.2.2 Mica and Its Products

Most of the mica used for insulation purpose is a naturally occurring in­
organic substance. It occurs in the form of crystalline mineral silicates of
alumina and potash. Mica can be split into very thin flat laminae. It has a
unique combination of electrical properties, such as high dielectric strength,
low dielectric losses, resistance to high temperatures and good mechanical
strength. Due to these properties, it is used in many electrical apparatus
where high temperatures are experienced. The grading of muscovite mica,
the most widely used for insulation purposes, is covered by ASTM stan­
dard D351 [3].
Mica is also grown electrothermally with a composition similar to that
of natural mica. Its splittings are used extensively for glass-bonded mica
insulation. Mica is built into sheet form by bonding together with a suitable
resin or varnish. Depending on the type of application, mica can be mixed
with a required type of resin to meet the operating temperature require­
ments. Micanite is another form of mica which is used extensively for
insulation purposes. Mica splittings and mica powder are used as fillers in
insulating materials, such as glass and phenolic resins. The use of mica as
a filler results in improved dielectric strength, reduced dielectric loss and
improved heat resistance and hardness of the material. Table 6.2 gives
electrical properties of mica used for electrical insulation purposes.

Table 6.2 Electrical Properties of Mica

Property Natural mica Synthetic mica

Dielectric strength (at 30°C) - 1 0 0 0 kV/m m - 1 0 0 0 kV/m m

Dielectric constant (tan 5) 6.5-8.T 6.5
(1 kH z-3 GHz)
Loss tangent
50 Hz 0.03 —

1 MHz 0.001 0.0002

Surface resistivity (at 60% humidity) lO'^-lO“* n-m —

Volume resistivity (constant up to lO'^'-lO'" fl-m lO^^-lO^"" il-m

200°C)
Maximum operating temperature 540°C 980°C
Solid Dielectrics 151

Glass-bonded mica is produced by bonding finely grounded (natural

or synthetic) mica with low temperature melting electrical grade glass, to
form a ceramic that is both machinable as well as moldable. Its important
properties are its high service temperature (650°C) and high mechanical
strength, and that it is nonflammable, nonarcing, nontracking or outgassing
and impervious to moisture. It is being used extensively in circuit breakers,
switchgears, arc barriers and bushings.

6.2.3 Glass
Glass is a thermoplastic inorganic material comprising a complex systems
of oxides (Si0 2 ). Glass is defined as a liquid which has cooled to a rigid
solid without crystallization. At temperatures below the glass transition
temperature, glass is rigid and displays properties of the crystalline state.
At the temperatures above glass transition temperature, glass is plastic and
viscous. Glasses of interest in electrical insulation are primarily silicate
based. The so-called E-glass is used for producing fiber glass which is
used for reinforcing plastic materials to obtain high mechanical strength.
The dielectric constant of glass varies from 3.7 to 10, whereas the density
varies from 2.2 to 6 g/cm^. At room temperature, the volume resistivity
of glass varies from 10’"^ to 10^^ ohm-m. The dielectric loss factor of glass
varies from 0.004 to 0.02 depending on the frequency. The losses are high­
est at lowest frequencies. The dielectric strength of glass varies from 3 to
5 MV/cm, which decreases with increase in temperature, reaching half the
value at lOO^'C. At present the most common application of glass on power
networks is in the form of fiber glass, which is used ( 1 ) in bandaging core
packets of transformers, (2 ) as resin-impregnated fiberglass cores for com­
posite insulators, (3) as resin-impregnated fiberglass mats and insulating
plates, and (4) as fiberglass reinforced plastics in the form of tapes in
electrical machines.
Glass in the form of paper is also used for insulation purpose. It is
composed of glass microfibers. Its outstanding feature is its thermal sta­
bility up to 538°C. Other attributes include high thermal conductivity, low
moisture adsorption and good chemical resistance [2 ].

6.2.4 Ceramic Insulating Materials

Ceramics are inorganic materials produced by consolidating minerals into
monolithic bodies by high-temperature heat treatment. Ceramics can be
divided into two groups depending on the dielectric constant. Low-
permittivity ceramics < 1 2 ) are used as insulators, while the high-
 152 Chapter 6

premittivity ceramics (Sr > 1 2 ) are used in capacitors and transducers.

Table 6.3 gives selected dielectric properties of some ceramics commonly
used for electrical insulation purposes.
Porcelain and steatite are gas tight, corrosion proof, chemically inert
to all alkalies and most of acids and therefore resistant to contamination.
Steatite possesses higher mechanical strength and low and is also suitable
for making electronic components.
Alumina (AI2 O3 ) has replaced quartz because of higher mechanical
strength, good insulating properties and good thermal conductivity. Ther­
mal conductivity of plastics range between 0.15 and 0.3 W/mK, in por­
celains and glass it ranges between 1.2 and 1.7 W/mK whereas for alumina
it stands around 35 W/mK [4]. Its dissipation factor of 0.0002 at 1 MHz
is the lowest among all ceramics, although other dielectric properties are
not outstanding. Overall, alumina is one of the best ceramic insulations
available. It is used to fabricate high-current vacuum interrupters, and for
a variety of electrical/ceramic components. Its powder is being used to

Table 6.3 Dielectric Properties of Electrical Insulating Ceramics

Dielectric Dielectric
strength V /m il constant Dissipation
(ASTM D 149) 1 MHz factor 1 MHz
Ceramic (kV/mm) (ASTM D 150) (ASTM D 150)

Alumina (99.9% AI2O3) 340 (13.4) 10.1 0.0002

Aluminum silicate 150 (6.0) 4.1 0.0027
Beryllia (99% BeO) 350 (13.8) 6.4 0.0001
Boron nitride 950 (37.4) 4.2 0.00034
Cordierite (2Mg0 *2A l203*5Si02) 200 (7.8) 4.8 0.0050
Magnesia (MgO) — 5.4 < 0.0003
Porcelain (4K20*Al203*3Si02) — 8.5 0.005
Quartz (Si02) — 3.8 0.0038
Sapphire — 9.3-11.5 0.0003-0.00086
Silica (fused) — 3.2 0.0045
Steatite — 5.5-7.2 0.001
Zircon (ZrSi04) — 5.0 0.0023
Magnesium metatitanate (MgTi03) — 16 0.0002
Strontium zirconate (SrZr03) — 38 0.0003
Titanium oxide (Ti03) — 90 0.0005
Calcium titanate (CaTi03) — 150 0.0003
Strontium titanate (SrTi03) — 200 0.0005
Barium titanate (BaTi03) — 1500 0.015
Solid Dielectrics 153

produce various composite insulators and sheets. Similarly, feldspar coarse

grain porcelain insulators have been replaced by fine-grain alumina-
porcelain. At present, alumina based porcelains are widely being used for
suspension-type insulators, station post insulators, and so on, which de­
mand a high mechanical strength.

6.2.5 Polymers
Polymers consist of long-chain macromolecules with repeating monomer
(or mer) units. A polymer is usually named by putting the prefix poly- in
front of the name of the monomer from which it is derived. For example,
the monomer ethylene is the repeated monomer in polyethylene:
H H H H

C = C C - C ( 6 . 1)

H H [H H
ethylene polyethylene
A selection of polymers which are commonly used in electrical insulation,
together with their monomers is given in Table 6.4. Different molecular
units are attached to the ends of the chains (e.g., CH3 in polyethylene).
However, since n (the degree of polymerization) is very large in this case
(typically in the range 1 0 ^ - 1 0 ^), the end units do not usually influence the
physical properties of the polymer.
Table 6.4 includes some important polymers which are based on

(—C—C—) linkage along the length of the polymer “backbone.” These

are known as homopolymers. The other form is hetrochain polymers, in

which carbon atoms in the backbone have been replaced by other elements
and are placed into categories depending on their characteristic chemical
linkages. Simple polymer chains may form branches off the main chain;
this is commonly found in polyethylene as shown in Figure 6.1 (on page
155). Such branches can occur every 30-100 monomer units along the
backbone and result in side branches which can be short (e.g., up to several
monomer units long) or which can be long (e.g., as long as the main chain).
Branching can be produced or inhibited to a large extent by altering the
polymerization conditions. Branching reduces the potential for regular mo­
lecular packing and so lowers the density; for instance, in the case of
polyethylene (PE), it produces what is commonly called low-density pol-
154 Chapter 6

Table 6.4 Chemical Structure of Commonly Used Polymers

Generic structure Name (abbreviation)

X X
i I X= H polyethylene (PE)
-c-c~ X= F polytetrafluoroethylene (PTFE)
Ii
X X
H X X= CH3 polypropylene (PP)
i I X= CEL poly(vinyl chloride) (PVC)
-C -C - XsCeHs polystyrene (PS)
I i
H H XsOCOCHg poly (vinyl acetate) (PVA)

H X
I I x=ce poly (vinylidine chloride) (PVDC)
-c-c~ X=: F poly (vinylidine fluoride) (PVDF)
I I XrCHa polyisobutylene (butyl rubber)
H X
H X
II X=CH3 poly (m ethyl m ethacrylate) (PM M A)
-c-c- Y=C00CH3
;
X= H polybutadiene (BR)
c=c XrCH3 polyisoprene (natural rubber)
-CH2 ^CH2~
CH
polycarbonate (PC)

CH 3
Q— poly (ether ether ketone) (PEEK)

0 ^
-(C H 2>„- N-C-(CH2)„ - c - n- m = 4 , n = 6, polyam ide 6.6 (PA6.6, nylon 6.6)

H H

yethylene (LDPE). LDPE is mechanically inferior to high-density polyeth­

ylene (HDPE), its nonbranched counterpart, but nonetheless it has excellent
dielectric properties and is commonly used in high voltage power cables.
Beside the branches, polymers also have “cross-links” in which the pol­
ymer chains are joined by short, long or even polymeric molecules which
effectively form connecting branches. A cross-linked polymer is therefore,
in principle, essentially one gigantic molecule and so above its melting
point (if it is semicrystalline) or above the so-called glass transition tem­
perature (if it is amorphous) it becomes rubber-like rather than a liquid.
Solid Dielectrics 155

H H H H H H H H H
1 1
1 1 I1 I1 1 1 1 1 1 1
1 1
— C - C - c — C— c — c — C—C - - c —
1 1 1 1
1 1 1 1 1
1 1
I I 1
H H H H H H H H

H —C — H
I
H— C —H
I
H— C— H
I
Figure 6.1 Side branches in the molecular structure of polyethylene.

Cross-linking Techniques
Cross-linking is usually achieved in three different ways that are known as
curing techniques', catalyst curing, curing with chemical hardener, and ra­
diation curing.
1. In the case of catalyst curing, a catalyst (also called an initiator)
is mixed into the polymer. After this mixture has been molded or cast into
the required geometry, it is subjected to heat and/or pressure to initiate a
crosslinking reaction. Such polymeric products are called thermosets, since
their shape, once set, becomes irreversible. Polyethylene cables are usually
crosslinked by incorporating 1 - 2 % peroxide (e.g., dicumyle peroxide)
which does not react when the polymer is extruded but reacts later when
the cable is either ( 1 ) heated by super-heated steam, (2 ) heated by high-
pressure nitrogen gas maintained at an elevated temperature in a long tube,
or (3) kept in silane solution. In the latter case, the cables, once formed,
are placed in a “sauna” for cross-linking to take place. Polyethylene cross­
links by means of polymer-free radicals produced via the initiation process.
Since these are usually generated in low concentration, the product is nor­
mally only lightly cross-linked. For compounding in line during the extru­
sion process, there are advantages to using a liquid instead of a powder.
Di-tert-butyl peroxide, which decomposes more slowly, is also frequently
adopted.
2. Curing with hardener. In this case a chemical (hardener) is mixed
with the base polymer for its cross-linking. Electrical grade epoxies, and
elastomers are the examples of this type of curing technique.
3. Curing with radiation is also used to promote cross-linking. How­
ever, this process is used only on thin sheets. As the equipment required
 156 Chapter 6

in this case is quite expensive, this technique finds only limited applications
[ 2].
Polymers which are not cross-linked can be remolded to other shapes
and hence are called thermoplastics. Polyethylene, polycarbonate and ac­
etal copolymers are examples of thermoplastics commonly employed in
electrical applications. Table 6.5 summarizes salient dielectric properties
of some thermoplastic materials that are commonly used for power system
applications [2 ].

Polyethylene
Polyethylene (PE) is a thermoplastic semicrystalline polymer which is
widely used in the cable industry. It is produced by polymerization of
ethylene C2 H4 as shown in equation (6.1). The long chain molecules of PE
are not chemically bonded to each other, therefore, it is hard at room
temperature but shows viscous flow at high temperatures, since these mol­
ecules start sliding on each other. It is an amorphous solid with maximum
crystallinity of 95%. Higher crystallinity increases its tensile strength, ri­
gidity, chemical resistance and opacity while it reduces its permeability to
liquids and gases. Its properties may be modified by additives and fillers.

Table 6.5 Selected Properties of Thermoplastics Used for Electrical Insulation

Dielectric Dielectric Dissipation Maximum
strength"* constant factor service Arc
Volts/mil temperature resistance
Polymer (kV/mm) 60 Hz lO^ Hz 60 Hz 100 Hz °C (seconds)

Polyphenylene 380 (15.0) 3.1 3.2 0.0003 0.0007 205 34

sulphide
Polyether 400 (15.7) 3.5 3.5 0.001 0.004 180 70
sulphone
Poly 380 (15.0) 3.2 3.0 0.009 0.01 130 120
carbonate
Acetal homo 380 (15.0) 3.7 3.7 — 0.0048 90 220
polymer
Acetal 380 (15.0) 3.7 3.7 0.001 0.006 105 240 (burns)
copolymer
Acrylic 500 (19.7) 3.7 2.2 0.05 0.3 95 No tracking
(PMMA)

mm (125 mil) thick sheets tested as per ASTM-D149.

Source: Ref. 2.
Solid Dielectrics 157

which are usually mixed with PE granules during the extrusion process.
Antioxidants are important to prevent its degradation during manufacture
and in service. These include amines, hindered phenols and phosphites.
Cable-grade PE also includes metal deactivators, since PE reacts with
copper conductor and screens. It is inhibited by adding polymerized
l,2-dihydro-2,2,4-trimethylquinoline. Degradation by ultraviolet light is in­
hibited to a remarkable degree by incorporating <3% of carbon black.
Inflammability is reduced significantly by additives such as antimony ox­
ide, aluminum trihydrate and halogenated compounds. Depending upon the
reaction process, PE can be produced either as LDPE or HDPE. The crys­
tallinity of LDPE is typically 45-55% while that of HDPE is 70-80%.
As a raw material, PE is available in the form of granules which can
be extruded with conductors to form cylindrical cable insulation. HDPE
has greater breakdown strength and higher dielectric constant than LDPE.
For thin films, breakdown field strengths of more than 200 kV/mm are
achieved.
PE does not contain any polar groups and therefore has a low dielectric
constant and a very low dissipation factor. Its volume resistivity is extraor­
dinarily high (lO’"^ flm), and thus stationary space charges can occur in
the material which can produce undesirable field distortions. PE can be
employed in the temperature range of -50°C to 75°C. It is combustible,
and its chemical resistance is good except to chlorine, sulfur, nitric acid
and phosphoric acid. Under the influence of oxygen, its surface becomes
brittle.
Whereas LDPE and HDPE are being produced by reaction processes
introduced several decades ago, a new family of reaction processes is pres­
ently being used to produce linear low-density polyethylene (LLDPE) and
medium-density polyethylene (MDPE). As shown in Figure 6.2, LLDPE
and HDPE (also MDPE, not shown) molecules are generally linear in struc­
ture. The linear resins exhibit short-chain, rather than long-chain, branch­
ing. This fundamental difference in molecular structure accounts for many
of the major differences in mechanical and dielectric properties as com­
pared to LDPE [5]. LLDPE has become the most prominently used plastic
in underground power cable jacketing because of its excellent mechanical
properties, installation temperature range, vapor transmission resistance
and environmental stress cracking resistance (ESCR). The other two linear
polyethylenes, HDPE and MDPE, also have excellent properties but are
stiffer because of their higher densities.
158 Chapter 6

Figure 6.2 Structure of LDPE, XLPE, LLDPE, and HDPE.

Cross-linked Polyethylene
Cross-linked polyethylene (XLPE) is obtained by cross-linking PE as de­
scribed earlier. Cross-linking renders PE infusible and suitable for service
temperatures up to 125°C, thus permitting cables to carry higher current
densities. It also for a short time withstands fault temperatures of around
250°C. Its resistance to cold flow and abrasion are superior to conventional
PE, while its dielectric properties are comparable. Another advantage of
cross-linking is that it makes possible higher filler loading without signi­
ficant loss of physical properties. This is why XLPE cables have almost
completely outclassed compound filled cables in the medium voltage range.
However, the main difficulties of PE and XLPE insulation are their sen­
sitivity to partial discharge and the associated question of life time. Cavities
of 1-30 fim are unavoidable during manufacture and are potential source
of commonly observed partial discharge activity. Table 6 . 6 gives compar­
ative properties of commonly used cable insulating materials [2,36].
 Co

Table 6.6 Comparative Properties of Some Cable Insulating Materials

Polyethene Impregnated
PVC kraft paper
Low High Cross- (0.125 mm
Property Flexible Semiregid Irradiated density density linked thick)

Specific gravity 1.37 1.31 1.34 0.92 0.95 0.92 0.65

Elongation (%) 300 300 150-200 600 400 500 2.6
Maximum operating 60-105 80 105 80 90 125 —
temperature (°C)
Flame resistance Good Good Good Poor Poor Poor-Good —
(comparative)
Dielectric constant 6.2 max 4.3 2.7 2.28 2.34 2.3 3.3-3.9
(1 MHz)
Dissipation factor 0.1 max 0.0005 0.0007 0.0003 0.0026-0.003
(1 MHz) 20°C (0.14 at
80°C)
Volume resistivity 5 X 10" 2 X 10'- 2 X 10"’ 10''' 10^4 10'“* 10'^
(O-m)
Dielectric strength 550 500 550 1884
V /m il (kV/m m) (21.6) (19.7) (21.6) (74)

Source: Refs. 2 and 36.

so
160 Chapter 6

Polyvinylchloride
Polyvinylchloride (PVC) has been in use since 1930 as cable insulation
and recently as jacketing material. It is produced by the polymerization of
vinyl chloride. There are several commercial polymerization processes, all
carried out in pressure vessels. PVC typically contains 56.8% chlorine
which can be increased to 67%. It’s chemical structure is as follows:
H H H H H H

C— C — C— C —

H Cl H Cl H Cl
PVC resins alone are hard and brittle up to 82°C. To be made useful for
cable applications, they are compounded with plasticizers and additives. A
typical electrical grade formulation contains 50% PVC, 25-35% plasticizer
and the rest other additives. Trimellitats are favored for cable applications
up to 105°C. To achieve balance in properties, more than two plasticizers
are generally added. Following is the list of commonly used additives and
their functions [2 ]:
• Heat stabilizers, frequently organometallic compounds, retard ther­
mal degradation, prolonging useful life.
• Impact modifiers improve impact resistance. These include chlo­
rinated polyethylene, acrylonitrile-butadiene-styrene (ABS), methyl
methacrylate-buta-diene-styrene and ethylene-vinyl acetate (EVA)
polymers.
• Fillers, such as calcium carbonate, reduce compound cost and raise
heat deflection temperature. Calcined clay improves dielectric
properties.
• Pigments impart color and opacity and improve weatherability.
• Flame retardants, such as antimony oxide, significantly increase
the oxygen index (OI) of PVC compounds.
PVC is also irradiated to improve its properties such as mechanical
strength, abrasion, cut through resistance and temperature resistance (which
is increased up to 105°C in case of cable grades).
PVC has good resistance to aliphatic hydrocarbons, oils, fats, waxes,
alcohols, concentrated inorganic acids and bases, dilute organic acids, salts
and most solvents. PVC is attacked by aromatic and halogenated hydro­
carbons, phenols, ketones, esters, ethers and concentrated organic acids.
PVC insulation possesses only average electrical and dielectric prop­
erties. Its breakdown strength is <30 kV/mm, whereas dissipation factor
Solid Dielectrics 161

at room temperature, which depends upon the filler grade, varies in the
range 0.002-0.1. This poor characteristic of PVC restricts its application
in medium-voltage cable insulation to only around 10 kV. However, it is
very widely used for low-voltage cables operating up to 1.0 kV. Beside
that, its irradiated version is also commonly used for providing inner and
outer protective sheathing of high voltage cables.

Elastomers
Elastomers are polymeric materials which exhibit elastic properties similar
to rubber. The most commonly used elastomers on power system network
are silicone rubber, ethylene propylene rubber (EPR) and ethylene propyl­
ene dien monomer (EPDM).
Elastomers are generally prepared by vulcanization with sulfur, sulfur
compounds, or other polymerizing agents. The compounding of elastomers
and blends of elastomers are very complex, which often involves around
10-15 ingredients. Depending on the elastomer and curing system, these
ingredients may include [2 ]:
• The elastomer(s), up to 50% by weight.
• Sulfur or sulfur compounds as crosslinking agents, <2%.
• A peroxide curing agent, such as dicumyle peroxide, <3%.
• A rubber accelerator, such as dipentamethylene thiuram hexasul-
fide, < 2 %.
• A curing promoter such as N,N'-m-phenylenediamaleide, to im­
prove scratch and insulation resistance, < 1 %.
• A metal oxide acid acceptor and vulcanizing agent, such as sub­
limed litharge (PbO) or dibasic lead phthalate, up to 5%,
• A hydrogenated wood resin to activate compounds containing lith­
arge, < 1 %.
• A stabilizer to improve heat resistance, such as nickel dibuthldi-
thiocarbonate, < 2 %.
• Carbon black to improve weathering properties, up to 25%.
• A lubricant, such as paraffin or petrolatum, up to 12%.
• A hindered phenol antioxidant, <3%.
• A filler, such as calcined clay or kaolin, up to 50%.
• A plasticizer, such as phthalate ester, chlorinated paraffin or alkyl
substituted trimellitate.
Once prepared, the compound is subject to curing. There are three types
of common curing techniques: ( 1 ) curing using vulcanizing agent such as
sulphur or its compounds, (2) lead sheath curing and (3) radiation curing.
162 Chapter 6

Silicone rubber is a high-temperature insulation applied to specialty

type of cables and high voltage insulators. It is prepared from dichlorosi-
lane, which is crosslinked or vulcanized, by the action of heat in the pres­
ence of a vulcanizing agent (a peroxide). It is therefore called high-
temperature vulcanized (HTV) silicone rubber. Its physical properties are
enhanced by compounding with fillers such as silica and diatomaceous
earths. Silicon rubber insulation can safely operate in a temperature range
of -55°C to 200°C. It has good resistance to ozone, corona and weathering.
It also exhibits, good resistance to alcohols, dilute acids, alkalis, salts and
almost all types of oils and waxes. Silicone rubbers are, however, attacked
by halogenated hydrocarbons, aromatic solvents, concentrated acids and
steam.
Besides cable insulation, HTV silicone rubber is being used for the
manufacture of outdoor high voltage insulators and also in the form of
additional extended sheds on ceramic insulators to enhance their dielectric
integrity and their performance under polluted atmosphere [9]. Fiberglass
reinforced rod insulators with silicone rubber sheds are accepted as stan­
dard overhead line insulators with a good experience for voltage ratings
up to 765 kVac and ±500 kVdc [6,7]. Table 6.7 summarizes important
properties of HTV silicone rubber.
The other type of silicone rubber that has gained popularity for its
applications as outdoor high voltage insulators is room temperature vul­
canized (RTV) silicone rubber. RTV is being used in the form of coatings
on ceramic insulators. Commercial RTV coating consists of a poly-
di-methylesiloxane (PDMS) polymer, plus two fillers such as fumed sil-

Table 6.7 Comparative Properties of Silicone Rubber and EPDM

Property HTV silicone rubber EPDM

Dielectric strength (kV/mm) 20.0 19.7-31.5

Dielectric constant (1 MHz) 3.0-3.6 2.5-3.5
Dissipation factor (1 MHz) 0.005 0.007
Volume resistivity (il-m) 10^^ 10'^
Specific gravity 1.15-1.55 0.85
Elongation, % (ASTM-D412) 200 200
Comparative abrasion resistance Fair Good
Water resistance Good Excellent
Maximum operating temp. 200°C l ir e
Elame resistance Poor Poor
Resistance to ozone — Excellent

Source: Ref. 2.
Solid Dielectrics 163

ica and aluminum trihydrate (ATH), beside a colorant pigment and a

crosslinking agent [8 ]. This coating may also include PDMS fluid, addi­
tional fillers, a condensation catalyst and an adhesion promoter for its im­
proved bonding to ceramic surfaces. The driving motivation for their in­
creased acceptance and use on high voltage insulators is their unique
characteristic of hydrophobicity, which provides a high surface resistance,
even in the presence of moisture and contamination, thereby suppressing
the leakage current to low values as shown in Figure 6.3 [8-10]. Uncon­
trolled leakage current promotes intense dry band arcing, which on poly­
mers, can ultimately lead to material degradation in the form of tracking
and corrosion and/or flashover even at operating voltages [11]. Unlike most
polymeric insulating materials, silicone elastomers are able to maintain
their low surface energy (contact angle >90°). This property causes water
to “bead up” rather than form a continuous film of moisture. This is the
critical property of silicone elastomer which causes suppression of leakage
current. Moreover unlike EPRs, silicone elastomers are immune to the
sun’s ultraviolet rays. Today the trend of the utilities worldwide is toward
using silicone rubber for the sheds of all types of overhead line insulators
[ 12].
Ethylene-propylene rubber (EPR) is primarily an extruded dielectric
used in medium and high voltage power cables. Its extensive use is based
on superior electrical properties. It has attractive features like wet electrical
stability, flexibility, water tree and corona resistance.

Figure 6.3 Suppression of leakage current for a RTV coated insulator tested in a
salt-fog chamber. (From Ref. 8 © IEEE, 1995.)
164 Chapter 6

For the manufacture of cable insulation, ethylene and propylene are

mixed with fillers such as clay, talc, whiting, silica and alumina. Other
ingredients typically compounded with EPR are metal oxides, plasticizers,
antioxidants and curatives. Curing (crosslinking) of this compound is car­
ried out in the presence of peroxides. One of the curing methods— steam,
dry nitrogen and pressured salt solution—is selected. However, at present,
a debate continues in the industry as to whether microvoids formed during
steam curing result in a more rapid degradation of the insulation. Salt
solution is preferred in the case of large diameter cables. For better cable
performance, choice of filler and its content are of key importance [ 1 0 ].
Today, three types of EPRs are in common use for overhead electrical
insulation; namely, ethylene-propylene monomer (EPM), ethylene-
propylene diene monomer (EPDM), and a copolymer of ethylene-
propylene and silicone (ESP). All three types are highly filled with alumina
as well as other types of fillers as described above. A critical filler level is
considered important for an optimum end result. Table 6.7 compares the
key properties of EPDM and silicone rubber. Early EPRs suffered from
tracking and poor resistance to ultraviolet radiation. However, formulations
of present-day EPRs are somewhat immune to tracking and exhibit good
resistance to ultraviolet radiation, showing only surface ablation. These are
being employed successfully for distribution and transmission class insu­
lators of up to 765 kV. However, their long-term performance under pol­
luted environment has been unsatisfactory. The ESP insulators are still
relatively new and will require a few more years in use for their potential
weaknesses to be discovered [ 1 2 ].

Epoxy Resins
Epoxy resins are a family of thermoset polymers in which two components
are mixed to eventually form a glassy product at room temperature which
has reasonable electrical insulating properties and is also highly imper­
meable to water. Epoxies are polymers in which the end groups contain
the three-membered epoxide ring as shown in Figure 6.4a, whereas Figure
6.4b depicts the common configuration of the diepoxide structure. Many
different curing agents are used to bring about crosslinking of this resin.
The action of the curing agent or “hardener” is to open and join into the
epoxide rings. Diamine compounds (H2 N—R '—NH2 ) are commonly used
which provide four sites for attachment as illustrated in Figure 6.4c.
Because of the high chemical reactivity and potentially large number
of epoxide rings, crosslinking in cured epoxies can be very high and an
extensive network of connections with high mechanical rigidity is pro-
Solid Dielectrics 165

Figure 6.4 (a) Epoxide ring and (b) diapoxide structure, where R is commonly
bisphenol-A and n is typically about 10; (c) structure of epoxy resin compound.

duced. Once the compound is cured, the material becomes epoxy polymer
[13].
To improve the physical and mechanical properties of the end product
and also to control the cost, the epoxy resins are loaded with fiberglass,
fumed silica and other inorganic particulate fillers. Cast resin polymers are
compounds that are formulated by mixing resin with, hardener, filler, plas­
ticizer, and coloring pigments. These fillers may constitute 50% or more
of the compound weight. Table 6 . 8 summarizes essential properties of some
commonly used glass-filled epoxy compounds. The highest recommended
temperature for continuous use of nonreinforced epoxies is usually 130°C,
but some of the glass-filled compounds can be used up to 250°C. Water
absorption of epoxy polymers is generally low, ranging from 0.05 to 0.5%.
The higher values apply to epoxy polymers that are reinforced filled. Their
relative permittivity lies between 3.5 and 5.0 which is increased to between
4 and 8 for glass-filled epoxies as shown in Table 6 .8 .
The dissipation factor of epoxy increases only slightly with tempera­
ture, as shown in Figure 6.5. However, an initial hump generally appears
 166 Chapter 6

Table 6.8 Properties of Thermosetting Molding Compounds

Dielectric Dielectric Dissipation Maximum
strength"" constant factor service Arc
Volts/mil temperature resistance
Compound (kV/mm) 60 Hz 10*’ Hz 60 Hz 10*' Hz (°C) (seconds)

Glass-filled allyle 400 (15.7) 4.2 3.5 0.004 0.01 260 140
Glass-filled 380 (15.0) 6.0 5.0 — 0.02 232 180
phenolic
Glass filled epoxy 390 (15.4) 5.0 4.6 0.01 0.01 204 187
(electrical
grade)
Glass-filled 340 (13.4) 8.0 6.2 — 0.02 204 180+
melamine
Glass-filled alkyd/ 375 (14.8) 5.3 4.6 0.1 0.02 204 180+
polyester

''3.175 mm (125 mil) thick sheets tested as per ASTM-D149.

Ref. 2.
S o u rce'.

0 20 AO 60 80 100 120 1A0 160

Temperature (C)
Figure 6.5 Variation of dissipation factor with temperature of solid dielectrics at
50 Hz AC voltage. (From Ref. 14.)
Solid Dielectrics 167

around 130°C. This is attributed to the dipole orientation. Once the polymer
is warmed up to allow ionic conduction, the movement of the ions is
enough to contribute to heating, and the runaway thermal heating shown
by the increasing tan 6 values at ~150°C [14,15]. Figure 6.5 also illustrates
a much lower dissipation factor of epoxy resin as compared to PVC, por­
celain and paper.
Epoxy resins are extremely useful for their high mechanical strength,
good adhesion to materials and metal (except polythene), and resistance to
moisture. They are being used extensively in high voltage switchgear and
electrical machine insulation as well as for the “potting” of electrical com­
ponents. They are also finding applications in low and high voltage trans­
former winding encapsulation, as well as spacers for SF^ GIS installations.
Glass-fiber-reinforced epoxy rods for overhead line polymeric insulators
are another field of their application.
Bisphenal-A-based epoxy resins (highly filled with quartz) were used
in the first generation of polymeric outdoor insulators. During long service,
however, cracks were reported in insulators, and were found to be due to
low tracking resistance and poor ultraviolet radiation resistance of bis-
phenal-A. At present it has been replaced by cycloaliphatic-type epoxy
resin which contains a hydrated aluminum filler [7]. Today, a large variety
of these insulators are in use at voltage levels of <69 kV. Their long-term
performance in normal atmosphere has been quite successful. However, in
polluted atmospheric conditions, their performance has been far from sat­
isfactory. Therefore, their present use is mostly limited to apparatus bush­
ings and bus bar insulators [ 1 2 ].

6.3 DIELECTRIC LOSS IN SOLID INSULATING MATERIALS

It was shown earlier that relative permittivity (s^.), dissipation factor (tan
5), and hence the loss factor or the loss index tan 8) are the character­
istics of an insulating material pertaining to its molecular and atomic struc­
ture, which play a significant role in its power loss. Before considering this
dielectric loss, some basic concepts will be reviewed here.

6.3.1. Electric Polarization

Let us consider a parallel plate capacitor having surface charge density q
on its opposite plates of area A and separation d, then the resultant electric
field in vacuum is:
4 7 rq (6 .2 )
168 Chapter 6

If we insert a dielectric material of permittivity e between the plates, it

will cause a decrease in the electric field intensity between the plates hav­
ing a fixed surface charge density. This is because the dielectric material
tends to neutralize the charges at the electrode surface due to the formation
of dipole chains of the molecules, as shown in Figure 6 .6 . This phenom­
enon is called dielectric polarization (P). The charge density in this case
becomes where e == Hence

477 4778

(6.3)

These equations provide a direct relationship between P, s and E, and

emphasize that polarization is increased with permittivity and applied
stress.

6.3.2 Polarizability of Molecules

In the absence of applied electric field, the net charge on molecules of a
dielectric material is zero. But on the application of external field, the
electrons are slightly displaced with respect to the nuclei; induced dipole
moments result and cause the so-called electronic polarization of materials.
For example, if in a neutral H2 atom the charge of an electron and that of
a proton are displaced distance d, then it results as a net-induced dipole
moment, = ed, where e is the charge of electron. The resultant dipole
moment due to electrons in an insulating material can therefore be ex­
pressed as:
= a^E (6.4)

© (EI±)(E3G 3C 3 ©
© (- +) (- GZ3
■>•) ( - ©

© GZ±) (EZ±) (E© <EZ© 0

© (EZ3 (EZ5>(Z© GZ© ©

© G© (E© (E©> (E© 0

© c- 4 ) (E Z ± ) t ) ( E Z i) ©

© GZZ±) (E Z Z ±) d ± ) C- ©

© C- -0 C~ •<•) ©

Figure 6.6 Electric polarization in a dielectric.

Solid Dielectrics 169

where is the electron polarizability of atoms, which depends on E. For

all materials, the value of is constant with frequencies up to the ultra­
violet spectrum (—2.5 MHz).
When atoms of different types form molecules, they will normally not
share their electrons symmetrically, as the electron clouds will be displaced
eccentrically toward the stronger binding atoms. Thus atoms acquire
charges of opposite polarity, and an external field acting on these net
charges will tend to change the equilibrium positions of the atoms them­
selves. This displacement of charged atoms or groups of atoms with respect
to each other results into atomic polarization of materials. In this case,
“atomic or ionic polarizability” is given as a^. The value of is generally
constant, with frequencies up to the infrared spectrum. For nonpolar ma­
terials, the polarizability is mainly determined by as generally does
not exceed 0.1 [15].
Polar substances, on the other hand, possess permanent dipole mo­
ments in their molecules even in the absence of an applied field. In this
case, the centroids of positive and negative charge distributions are per­
manently separated by fixed distances, thus resulting in permanent dipoles.
Such dipole moments experience a torque in an applied field that tends to
orient them in the direction of the field. This is called orientation {or
dipole) polarization, P^. This is of particular significance, since most pol­
ymeric insulating materials have permanent dipoles in their molecular
structure.
When a field is applied suddenly across a dielectric, it will attain finite
polarization in a very short time provided it is only due to or How­
ever, in the case of slower movement of orientation dipoles, the polarization
will attain saturation only with some time lag (also called relaxation time,
Tp) due to slower migration of ions. When the applied field is alternating,
the dipoles or the charges must change their direction every half cycle. If
the frequency is very low, or when the duration of half cycle is longer than
Tp, the polarization will attain its maximum value. But at higher frequen­
cies, the duration of half cycle could be shorter than Tp, therefore in this
case the dipoles cannot follow the change in the field intensity. This results
into reduction in polarization, which will ultimately diminish to zero at a
very high frequency. Figure 6.7 illustrates variation of polarization (and
hence e) as a function of frequency [15]. The different relaxation times
result in frequency limits beyond which the respective mechanisms no
longer exist, because the corresponding dipole movement does not occur.
This is why the e must also decrease. At each transition (sudden drop in
Pd) of the dielectric constant, the dissipation factor (tan &) has its maximum
value. But only the transition region controlled by P^, which represents the
frequency range in which dipole orientation vanishes, is of interest for
170 Chapter 6

Figure 6.7 Schematic plot of polarization as a function of frequency. (From Ref.

15.)

technical insulation systems. In this transition domain of frequency, which

is also known as the dispersion domain, the relative permittivity e, depends
strongly on eigenfrequency = 1 /^p)^ which is a characteristic property
of a material and varies from one dielectric material to another. For ex­
ample, for hard pressed board it is 10 kHz, whereas for pure water it is
—100 MHz. It is also a function of temperature.

6.3.3 Losses Under High Voltages

In practical insulating materials, the losses are not confined to polarization
processes. The presence of ionic impurities and voids in the bulk of in­
sulating material as well as the presence of trapped space charge may result
as ionic flow of current under the application of high electric stress. It will
increase with the increase in electrical stress as well as in temperature. In
turn, the dissipation factor tan 3 = a l cos will also follow suit, since the
AC conductivity a follows this trend while e remains constant with electric
field. Similarly in the vicinity of breakdown stresses, increased losses could
also result due to electronic emission from electrodes, which can arise
either from the Schottky effect [16] or from the “tunneling effect,” where
electrons would be tunneled into the bulk of the dielectric material through
thin (<100A) dielectric oxide layers that cover the metallic electrodes [15].
In ceramics and glasses, the electronic conduction (even at low tem­
peratures) is mainly governed by the “electron hopping effect” through
transition metal oxides. This refers to the oxides in which either all or a
Solid Dielectrics 171

portion of the d-shell electrons have been given off by the transitional metal
atom to oxygen atom [17].
The above briefly outlines the theory of dielectric loss and the various
processes that control it. For more in-depth treatment, the reader is referred
to Bartnikas and Eichhorn [15] and O’Dwyer [18]. It is clear that dielectric
loss depends, in principle, on the molecular structure of the insulating
material, its homogeneity and the included impurities. Efficient utilization
of the dielectric in electrical power apparatus, therefore, requires the
knowledge of dielectric loss behavior under specific voltage, frequency and
temperature conditions. It is much more helpful—and expedient—to com­
pare the loss behavior in different generic materials such as polymers,
glasses and ceramics.
In practical polymers, the losses are mainly dominated by dipole ori­
entation or ionic conduction due to contaminants. Exception to this general
rule may be observed only in polymers which possess an exceptionally
high degree of purity, which leads to lower conductivities and dielectric
constants. In the electrical power industry, the most commonly used pol­
ymers are XLPE and EPR. Figure 6 . 8 compares the 60-Hz dielectric con­
stants of commercially available XLPE, filled XLPE and EPR over a prac­
tical temperature range [19]. It is clear that of XLPE is considerably
lower than that of filled XLPE over this temperature range. This shows
that the addition of “fillers” in polymers leads to “interfacial polarization,”
which will lead to higher dielectric constant and dielectric losses. If these
fillers contain polar species, it will further enhance the loss. The decrease

Figure 6.8 Dielectric constant as a function of temperature at 60 Hz for XLPE

and EPR. (Erom Ref. 19.)
172 Chapter 6

of gj. with temperature for all three materials results directly from a de­
crease in the density with temperature and hence the number of molecules
per unit volume that contribute to orientation polarizability. The sudden
drop in e^. for both XLPE samples at 110°C is due to the vicinity of melting
transition temperature.
Figure 6.9 illustrates tan 8 values as a function of temperature for the
same materials. The losses in EPR and filled XLPE are higher than in
XLPE. The increase in tan 8 in the case of XLPE beyond 70°C with a
maximum at ~120°C is due to the a-type relaxation process that is attrib­
uted to the vibrational and reorientation molecular motions within the crys­
talline regions [20]. For filled XLPE and EPR, the a-type process is also
operative but it is rather dispersed and its peak occurs at a considerably
higher temperature. It is interesting to note that in extruded XLPE cable
material, no distinct a-peak is observed [19]. If moisture is absorbed by
the polymers, it leads to the formation of a double phase resulting in in­
terfacial polarization loss. Likewise, any ionic impurities contained in water
will give rise to a higher overall conductivity and thus a higher loss.
Whereas the dominant loss portion in polymers is due to orientational
polarization, the dielectric loss in glasses and ceramics is mainly electronic
and ionic in nature. Electrical grade glasses consist, to a large extent, of

Figure 6,9 Tan 5 as a function of temperature for XLPE and EPR at 60 Hz. (Prom
Ref. 19.)
Solid Dielectrics 173

SÌO2 , B2 O3 or phosphoric anhydride (P2 O5 ) structures. These are suffi­

ciently open to permit ionic diffusion and migration. The conduction losses
in this case mainly result from sodium (alkali) impurity ions. Ceramics are
composed of various materials that are formed permanently into durable,
hard dielectrics by either firing or sintering processes. Various clays are
also used as fillers. These are the source of ionizable impurities that give
rise to significant losses. The main charge carriers responsible for gener­
ating dielectric losses may be either electrons or ions, or both types of
species may contribute. However, because of the complex structure of
ceramic materials, it has not always been possible to delineate clearly
which charge carrier process may be responsible for the observed dielectric
loss [15].

6.4 BREAKDOWN IN SOLID INSULATION

Matter in the solid state consists of atoms held together in a condensed

phase. The most striking feature of the majority of solids is that their atoms
(or atom groups) are arranged with a high degree of order in some regular
repetitious pattern in three dimensions; such solids are called crystalline.
Solids whose atoms are arranged in an irregular and more or less random
fashion are called noncrystalline or amorphous. Since a large proportion
of commercial insulating systems are solids, studies of the breakdown of
solid dielectrics are therefore of extreme importance in insulation studies.
The application of a strong electric field to a solid dielectric material
can result in motion of free charge carriers, charge injection from the elec­
trodes, charge multiplication, space charge formation and dissipation of
energy in the material. Depending on the circumstances, many of these
may occur alone or in combination, thus finally leading to the electrical
failure of the material, which is also called its breakdown. The theoretical
aspects of different mechanisms advanced to explain breakdown have been
reviewed by O’Dwyer [18], while breakdown of polymeric insulation has
been reviewed recently by leda et al. [2 1 ].
When breakdown occurs, solids get permanently damaged while gases
fully recover their dielectric strength, and liquids partly do, after the ap­
plied field is removed. The mechanism of breakdown is a complex phe­
nomena in the case of solids, and depends on the time of application of
voltage, as shown in Figure 6.10. It is unlikely that any single mechanism
can be responsible for the many diverse phenomena generally associated
with the breakdown of solids. In the following sections the principles of
various breakdown mechanisms are described in brief. These are:
174 Chapter 6

Figure 6.10 Variation of breakdown strength and the operating breakdown mech­
anism with time of stressing.

1. Electronic breakdown mechanisms

2. Thermal breakdown mechanisms
3. Electromechanical breakdown mechanisms

6.4.1 Electronic Breakdown Mechanisms

In such mechanisms, the field causes either the number of electrons or
their energy to reach unstable magnitudes such that they rise catastrophi­
cally, causing, at least locally, the destruction of the lattice of the solid
material. Electronic breakdown is essentially of two types: intrinsic break­
down and avalanche breakdown.

Intrinsic Breakdown
If the material under test is pure and homogeneous, the temperature and
environment conditions are carefully controlled, and the sample is so
stressed that there are no external discharges, then under voltages applied
for short time the electric strength increases to an upper limit which is
called the intrinsic dielectric strength. The intrinsic strength is a property
of the material and its internal temperature only.
In pure homogenous dielectric materials the conduction and the va­
lence bands are separated by a large energy gap, and at room temperature
the electrons cannot acquire sufficient thermal energy to make transitions
from valence to the conduction band. The conductivity in perfect dielectrics
Solid Dielectrics 175

should therefore be zero. In practice, however, all crystals contain some

imperfections in their structures due to missing atoms, and more frequently
due to the presence of impurities. The impurity atoms may act as traps for
free electrons in energy levels that lie just below the conduction band.
At low temperatures the trap levels will be mostly filled with electrons
caught there as the crystal was cooled down during its manufacturing. At
room temperature some of the trapped electrons will be excited thermally
into the conduction band, because of the small energy gap between the
trapping levels and the conduction level. An amorphous crystal will there­
fore have some free conduction electrons. Such electrons would be accel­
erated in applied field and, provided the density of such carriers is low,
energy would be lost to the lattice of the crystalline material by phonon
interaction. In steady-state conditions, the electron temperature will be
nearly equal to the lattice temperature. Breakdown for this model is deemed
to take place when average ratio of energy gained from the field is greater
than that lost in collisions with the host material. This condition is called
von Hippel low field breakdown criterion [22].
Unlike in crystalline solids, in amorphous or impure solids the trapped
impurity electrons below the conduction band have a higher concentration.
In this case. Fröhlich [23,24] postulated that electron-electron interactions
may take place between the conduction band and trapped electrons so that
they form part of the same energy distribution that is defined by the elec­
tron temperature. The electron system will gain energy from the field
through the acquisition of kinetic energy by the free electrons and lose it
by lattice scattering primarily via the far more numerous trapped electrons.
For this energy loss to take place, electron temperature must be greater
than the lattice temperature. The increase in field will increase electron
energy more rapidly than the electrons can transfer it to the lattice, so the
electron temperature will exceed the lattice temperature. The effect of the
increased electron temperature will be a rise in the number of trapped
electrons reaching the conduction band. This increases the material’s con­
duction and, as the electron temperature continues to increase, a complete
breakdown is eventually reached. This breakdown mechanism is normally
referred to as electronic thermal breakdown.

Avalanche Breakdown
In its simplest form, avalanche breakdown may be visualized in a similar
way as the Townsend a process in gases. An electron released at the cath­
ode will gain energy from the applied field. This high energy electron may
collide with a bound electron thereby resulting in a pair of free electrons.
In the presence of a high field these electrons acquire sufficient energy to
176 Chapter 6

produce two more free electrons. Repetition of this process increases the
number of free electrons, and since it is only free electrons which can
acquire energy from the field, the avalanche can lead to very high local
energy dissipated into the lattice, causing its disruption after a sufficient
number of generations.
The critical number of generations was calculated as —40 by Sietz
[25], therefore the critical ionization rate per unit length is given as:
a. = 40/d (6.5)
where d is the material thickness. The field dependence of a at constant
pressure may be expressed as:
a = A' exp(-B 7E ) ( 6 . 6)

where A' and B' are constant. B' depends on both the energy for impact
ionization and mean free path for phonon collision.
Fowler-Nordheim field emission process [26] at the cathode is invoked
to take care of the injection carriers which take part in the electron ava­
lanching in the applied field, leading to failure. The expression for current
J reaching the anode may be shown to be:

exp + ad (6.7)
4>
where kj and k2 are constants and O is the energy barrier for electrons
escaping from cathode. At the instant of breakdown, the exponent term
becomes zero. Therefore breakdown field can be given as:

Eb = ( 6. 8)
ad
Similarly, conduction current effects can also take place due to field-
dependent emission from traps in the bulk of the insulator (Pool-Frenkel
effect [15,18]).
It is interesting to note that since the development of an avalanche is
dependent on the availability of an initiating electron, a statistical time lag
is also predicted by this theory. The field for avalanche breakdown is ex­
tremely well defined experimentally in crystalline systems. For example,
in Zener diodes the current may rise by an order of magnitude for about
0.1% rise in the field. Kitani and Arii [27] have extensively studied time
lags to breakdown in polymer films using nanosecond pulses, and consider
that initiating electrons may be provided from the electrode or the bulk,
depending on the material and the temperature.
Solid Dielectrics 177

The electron avalanche breakdown is purely electronic breakdown pro­

cess usually observed in the low-temperature region for polymers. Typical
characteristics of avalanche breakdown are a negative thickness depend­
ence and a positive temperature dependence of the electric strength and
the breakdown time lag of < 1 . 0 ns [28].

6,4.2 Thermal Breakdown

When an insulation is stressed, heat is generated within it because of con­
duction currents and dielectric losses due to polarization. In general, the
conductivity (cr) increases with temperature, conditions of instability are
reached when the rate of heating exceeds the rate of cooling and the spec­
imen may undergo thermal breakdown. To obtain the basic equation for
thermal breakdown, let us consider a dielectric cube of face area A(m^).
Assume that the heat flow in the x-direction is as shown in Figure
6 . 1 1 , then the

dT
Heat flow across face (1) = KA (6.9)
dx
where K is the thermal conductivity of the material.
I f -r i 1 /

Heat flow across face (2) = K - — h KA ) Ax (6.10)

dx dx \d x /
The second term in equation (6.10) represents the heat input into the block.
Hence

Heat flow/volume = K ( ^ ) ^ div(K grad T) (6.11)

dx \dx 7

Figure 6.11 Heat input and output from a cubical specimen.

178 Chapter 6

The conservation of energy requires that heat input into the element must
be equal to the heat conducted away, plus the heat used to raise the tem­
perature T of the solid, or heat generated = heat absorbed + heat lost to
surroundings, hence:

o-E" div(K grad T) ( 6 . 12)

dt
where == thermal capacity of dielectric, a = electric conductivity, T
temperature and t = time over which heat is dissipated.
Experimental data on breakdown with temperature shows that the volt­
age at which thermal breakdown can occur depends on the time of voltage
application. Therefore two limiting cases can be considered: impulse ther­
mal breakdown and steady-state thermal breakdown. Both of these repre­
sent breakdown that takes place on a broad front across the dielectric. Later
we will also consider breakdown in filament through the dielectric.

C ase I. Impulse Thermal Breakdown

In this case, the heat builds up very rapidly, therefore the heat lost to
surrounding can be neglected. Therefore, equation (6.12) reduces to:

(6.13)
dt " dE dt
To obtain breakdown field in time t,,, we apply a ramp function field;
then

E = (6.14)

Also the conductivity a shall vary as:

W
a = <T^ exp (6.15)
k jj
where W = the activation energy to surmount the potential barrier, cr^ =
conductivity at ambient temperature T^ and k^ is the Boltmann constant.
Substituting equations (6.14) and (6.15) into (6.13) yields
Eb
^ E2 dE = r (
Eb C,, Jt„
When breakdown occurs, then T^ —►T^, i.e., the critical temperature and
W ^ k^T Solution of this equation gives the value of E^ as:
Solid Dielectrics 179

w
Eb = exp (6 .1 6 )
J UkbToJ
T h is shows th a t u n d e r im p u ls e th e rm a l stress the a p p lie d c ritic a l fie ld is
re la te d in ve rse ly to the tim e o f the a p p lic a tio n o f fie ld ; m o re o v e r it is
a p p ro x im a te ly in d e p e n d e n t o f the c ritic a l tem perature.

Case //. Long-T im e (Steady State) Therm al B reakdow n

F o r th is case w e assume the d ie le c tric slab is c o n ta in e d w ith in b ig elec­
trodes, as show n in F ig u re 6.12. In th is case the m e ta l electrodes w ill act
as a strong heat sin k. T h e re fo re heat p ro d u ce d in the m id d le o f the slab at
a tem perature T j w i l l be con d u cte d to w a rd the electrodes a fte r a s u ffic ie n tly
lo n g tim e lag. T h e re fo re in th is case w e can n e g le ct the “ heat absorbed”
te rm (C ^ d T /d t) in eq u a tio n (6 .1 2 ) and the expression reduces to

o-E^ = d iv ( K grad T ) (6 .1 7 )

A t b re a kd o w n vo lta g e , E = E^. S ince crE = J and E = - d V / d x , th e re fo re

J = - c r d V / d x . S u b s titu tin g th is value in eq u a tio n (6 .1 7 ) y ie ld s

T^=KA
dx dx Vdx

In te g ra tin g b o th sides, w e get

Figure 6.12 Arrangement for testing a solid dielectric under steady thermal stress.
180 Chapter 6

V2
K p ,d T
2
w here = Ma

Vg = Kp,dT (6 .1 8 )
<
C ritic a l c o n d itio n s are reached w h e n T j == T^, i.e., the c ritic a l te m perature
at w h ic h b re a kd o w n sets in . T h is shows th a t b re a k d o w n vo lta g e u n d e r
steady-state tem perature c o n d itio n s fo r a th ic k d ie le c tric slab is in d e p e n d ­
ent o f its thickness. F u rth e rm o re , m a te ria ls o f h ig h e r th e rm a l c o n d u c tiv itie s
are essential fo r a b e tte r pe rfo rm a n ce . H ig h th e rm a l c o n d u c tiv ity , K , helps
to co n d u ct unw ante d heat aw ay fro m h o t spots w ith in the d ie le c tric m o re
e ffic ie n tly . G e n e ra lly, a 10°C d ro p in h o t spot tem p e ra tu re results in d o u ­
b lin g the th e o re tic a l lif e o f in s u la tio n [2 9 ].
T he re la tio n (6 .1 8 ) does n o t h o ld fo r th in sheets o f th ickn e ss d. A
th e o re tic a l analysis y ie ld s an a p p ro x im a te re la tio n in th is case as [3 0 ]:

Eh = C V d (6 .1 9 )

w here C is a constant w h ic h depends on th e rm a l c o n d u c tiv ity .

T he losses are m u ch h ig h e r u n d e r A C voltages than u n d e r d ire c t fie ld s.
C onsequently, is g e n e ra lly lo w e r un d e r a lte rn a tin g fie ld s . T h e rm a l
b re a kd o w n is a w e ll-e s ta b lis h e d m echanism , th e re fo re the loss fa c to r (s
tan 8) is an essential para m e te r fo r a p p lic a tio n o f d ie le c tric s . E x p e rim e n ta l
data illu s tra te s th a t the m a te ria ls in w h ic h loss fa c to r increases w ith te m ­
perature g ive rise to a q u ic k th e rm a l ru n w a y.

Case III. Filamentary Thermal Breakdown

In th in d ie le c tric film s the b re a k d o w n does n o t u s u a lly o c c u r on a broad
fro n t across the in s u la tio n area, as assum ed ea rlie r, b u t g e n e ra lly occurs at
w e a k spots. T he tem p e ra tu re o f a w e a k spot (w ith h ig h e r lo c a l cr, lo w e r
K , fie ld -e n h a n c in g in c lu s io n s o r the lik e ) reaches the c ritic a l tem perature
b e fo re the rest o f the in s u la tio n . S everal exp e rim e n ts have been re p o rte d
in support o f such a fila m e n ta ry th e rm a l b re a kd o w n . F ig u re 6.13 illu s tra te s
lo c a liz e d heat gene ra tio n in L D P E film u sin g a tw o -d im e n s io n a l v is u a l
th e rm o g ra p h [3 1 ]. P re b re a kd o w n c u rre n t m easurem ents are also ca rrie d o u t
to c o n firm th is m echanism . T h e b re a k d o w n fie ld can be re la te d to c u rre n t
d e n sity and the tim e as [3 2 ]:
Solid Dielectrics 181

2 9 - 30 C
2 7 -2 8 C ^ 2 8 -2 9 ® C

2 6 - 27C

Figure 6.13 Thermograph of PE film before breakdown and breakdown point at

24.5°C. (From Ref. 31 © IEEE, 1990.)

kJlC ^D
( 6 . 20)

w here D is the m a te ria l d e n s ity and t^ is the tim e m easured ju s t b e fo re the

b re a kd o w n , a fte r the a p p lic a tio n o f vo lta g e . To a cco u n t fo r the c u rre n t in
a c y lin d ric a l fila m e n t o f ra d iu s r^, s u b s titu tio n o f I f = Trr^J tu rn s e q u a tio n
(6 .2 0 ) in to

( 6 . 21)
7rk,T^Cn
values o f E^, I f and t^ p r io r to b re a k d o w n are m easured e x p e rim e n ta lly ,
and fo r a fix e d m a te ria l a p lo t o f rf versus t^^ has in deed been v e rifie d
[32].

6.4.3 Electromechanical Breakdown

W h e n s o lid d ie le c tric s are subjected to h ig h e le c tric fie ld s , fa ilu re occurs
due to e le ctro sta tic com pressive forces (é le c tro s tric tio n ) w h ic h can exceed
182 Chapter 6

the m e ch a n ica l com pressive strength. I f the in itia l th ickn e ss o f the speci­
m en is d^, w h ic h is com pressed to a thickness d u n d e r an a p p lie d v o lta g e
V, then the e le c tric a lly d evelop ed com pressive stress is in e q u ilib riu m if:

eofiv ^ ( 6 . 22)

w here Y is the Y o u n g ’ s m o d u lu s. U s u a lly , m e ch a n ica l in s ta b ility occurs

w h e n d /d o = 0.6 o r d ^ /d = 1.67. C onsequ ently, fro m e q u a tio n (6 .2 2 ), the
hig h e st apparent e le c tric stress b e fo re b re a kd o w n is g iv e n as:

E 0.6 (6 .2 3 )
"" d.
T he above equatio n is o n ly a p p ro x im a te as Y depends on m e c h a n ica l
stress. W h e n a m a te ria l is subjected to h ig h stress the th e o ry o f e la s tic ity
does n o t h o ld w e ll and p la s tic d e fo rm a tio n has to be considered . F ig u re
6.14 shows p lo t o f e x p e rim e n ta l data o b ta in e d fo r va rio u s sam ples o f p o l­
ye th yle n e [3 3 ]. I t is cle a r th a t the cu rve (s o lid lin e ) p lo tte d based on equa­
tio n (6 .2 3 ) does n o t g iv e a goo d fit to the m easured values.
R ecently, D issa d o and F o th e rg ill [3 4 ] re fin e d th is th e o ry based on the
concept o f fra c tu re m echanics and proposed a new m e ch a n ism in w h ic h
fila m e n ta ry-sh a p e d cracks propagate th ro u g h a d ie le c tric , re le a sin g b o th
ele ctro sta tic energy and e le c tro m e c h a n ic a l strain energy stored in the m a ­
te ria l due to the a p p lie d e le c tric fie ld . T h is m e ch a n ism has a n a lo g y w ith

Figure 6.14 Breakdown voltage as a function of Young’s modulus for various

polymeric sheets. (From Ref. 33.)
Solid Dielectrics 183

c o n ve n tio n a l m e ch a n ica l cra c k p ro p a g a tio n in b rittle m a te ria ls, w h e re b y

cra ck propagates spontan eously i f the stra in energy released is greater than
that re q u ire d to ove rco m e the toughness o f the m a te ria l (G ).
L e t a' = stress and = m e ch a n ica l strain, then Y o u n g ’ s m odules
Y = c r'/e ,. T h e stra in energy d e n s ity can be w ritte n as c r'^ /2 Y D u rin g a
cra ck an e xtra area is created, w h ic h is p ro d u ce d b y energy p e r u n it area,
c a lle d toughness (G ). T h e re fo re , fo r a c ra c k to de ve lo p the c o n d itio n is

cr'^
(6 .2 4 )

In fila m e n ta ry e le c tro m e c h a n ic a l b re a k d o w n m echanism , each spark fila ­

m e n t is envisaged as a crack, as show n in F ig u re 6.15. A t the h e m is p h e ric a l
tip o f the fila m e n t there w i l l be an enhanced e le c tro s ta tic fie ld due the
a p p lie d voltage. T h is w i l l g iv e rise to e le ctro sta tic energy d e n s ity ==
l/2e,8,E2.
S im ila rly , m e ch a n ica l stress a in d u ce d b y the e le c tro s ta tic fie ld =
re s u ltin g in a strain energy d e n s ity per u n it v o lu m e (W^^) o f
c r'^ /2 Y = s I e^E'^/SY is ind u ce d , th e re fo re to ta l stra in energy (W) released
(per u n it v o lu m e ) is m ade up o f e le ctro sta tic (W ^) and e le ctro m e ch a n ica l
(Wj^) com ponen ts. T h e re fo re , W =
L e t the le n g th o f the tu b u la r cra c k = d l, then its v o lu m e = vrTf d l:

W - r^dl (6 .2 5 )
8Y

Figure 6,15 A filamentary crack emanating from an electrode.

184 Chapter 6

If we consider the usual breakdown field of ~10^V/m and Y = 3.10^ Pa

(for polyethylene), we find that the contribution of in the electrofractal
process for polymeric insulation is much larger than W^. Hence, the above
relation reduces to
e^eVr^
^EMl (6.26)
8Y
When a crack occurs it has to overcome the crack surface energy W, and
crack deformation energy Wf, therefore W > (W^ + Wf) where
W , = 27rrfGdl (6.27)
W f = 77r?Y dl (6.28)
Combining equations (6.27) and (6.28), we get electromechanical break­
down criterion as

g4 ^> {iT T xf} + Trr^Y) (6.29)

8Y
or
8Y(2G + Yrf)

The value of rf depends on what initiates the crack. For example, it may
be an impurity particle, microvoid, protrusion or extending element of elec­
trical tree. The value of rf never exceeds 10 pm. For polyethylene, G >
6500 J • m“^ and Y = 3 x 10^ Pa, then 2G > (Y • rf), which means W^
> Wf. Thus
16GY
(6.30)

so
Fb 0^ Y®-25
The results of Figure 6.14 have been replotted in Figure 6.16 based on
equation (6.30). It is clear that these give a good fit on this theory.
The speed of the crack propagation process is given by (Y/D)^^, where
D == density. In polyethylene, D = 930 Kg m“^, which will result in a
maximum speed of 180 ms“ ^ However, actual speeds measured in PF are
much higher, i.e., 380 ms“^ (negative point) and 1700 ms“^ (positive point)
[35]. This implies that the electromechanical process only plays role in
Solid Dielectrics 185

Figure 6.16 A log-log plot of data taken from Figure 6.14. Slope of the line
0.25. (From Ref. 34.)

tree propagation once a faster electronic process has already generated low
density region.
Experimental evidence shows that, in practice, the failure of solid di­
electrics is a complicated process which often involves combination of
several breakdown mechanisms. Figure 6.17 depicts schematically the tem-

Figure 6.17 Schematically illustrated temperature dependence of various electri­

cal breakdown processes operating in polymers. (From Ref. 21.)
186 Chapter 6

perature dependence of the various breakdown processes operating in pol­

ymers. The minimum Eb-temperature curve shown is the usually observed
behavior in polymers and includes the influence of various factors such as
applied voltage (impulse, DC, AC), electrode conditions (material, metal-
insulator interface, edge effects) and the dielectric itself involving impu­
rities, defects, etc., together with secondary effects like space charge and
the local heating [21]. In thicker insulation over a service life time, the
breakdown is likely to be due to the progressive degradation caused by
thermal aging, partial discharges and treeing. Also, mechanisms such as
edge breakdown, chemical erosion, tracking and surface discharges contin­
uously contribute toward the degradation of insulation. All of these mech­
anisms and their deleterious impact on insulation are dealt with in detail
in chapters 8 and 9.

REFERENCES

T. J. Gallagher and A. F. Pearmain, H ig h V o lta g e M e a s u r e m e n t, T estin g a n d

John Wiley and Sons, New York, 1983.
D e s ig n ,
2. W. T. Shugg, H a n d b o o k o f E l e c t r i c a l a n d E le c tr o n ic I n s u la tin g M a te r ia ls ,
IEEE Press, New York, 1990.
3. ASTM Standard D-351, Natural Muscovite Block Mica and Thins Based on
Visual Quality, 1990.
4. R. C. M. Pompilli and C. Mazetti, IEEE Elect. Insul. Magazine, Vol. 8, No.
I, pp. 28-32, 1992.
5. G. Graham and S. Szaniszeo, IEEE Elect. InsuL Magazine, Vol. 11, No. 5,
pp. 5-12, 1995.
6. E. Bauer, H. Karner, K. H. Muller, and P. Verma, CIGRE Report 22-11, Paris,
Prance, 1980.
7. R. V. D. Huir and H. C. Karner, 6th ISH. New Orleans, 1989, Paper No.
21.03.
E. A. Cherney, IEEE Elect. Insul. Magazine, Vol. 11, No. 6, pp. 8-14, 1995.
J. W. Chang and R. S. Gorur, IEEE Trans, on Dielectrics and Elect. Insul.,
Vol. 1, No. 6, pp. 1039-1046, 1994.
10. R. J. Arhart, IEEE Elect. Insul. Magazine, Vol. 9, No. 6, pp. 11-14, 1993.
11. R. G. Hougate and D. A. Swift, IEEE Trans, on Power Delivery, Vol. 5, No.
4, pp. 1944-1955, 1990.
12. E. A. Cherney, IEEE Elect. Insul. Magazine, Vol. 12, No. 3, pp. 7 -1 5 , 1996.
13. R. W. Sillars, E le c t r i c a l I n s u la tin g M a t e r i a ls a n d t h e ir A p p l ic a t io n s , Peter
Peregrinus Ltd., London, England, 1973.
14. R. Arora and W. Mosch, H ig h V o lta g e I n s u la tio n E n g in e e r in g , Wiley Eastern
Ltd., New Delhi, India, 1995.
Solid Dielectrics 187

15. R. Bartnikas and R. M. Eichhorn (Eds.), E n g in e e r in g D i e le c t r i c s V o l IIA ,

E l e c t r i c a l P r o p e r t ie s o f S o lid I n s u la tin g M a te r ia ls : M o l e c u la r S tr u c tu r e a n d
ASTM Publication, Philadelphia, 1983.
E le c t r i c a l B e h a v io r ,
16. C. Wagner and W. Schottky, Zeitschrift fur Physikalische Chemie, Vol. 11,
Part B, pp. 163-210, 1930.
17. A. Adler, P h y s ic s o f E le c tr o n ic C e r a m ic s , Part A. L. L. Hench and D. B.
Dove (Eds.), Marcel Dekker, Inc., New York, 1971.
18. J. J. O’Dwyer, T h e T h e o r y o f E l e c t r i c a l C o n d u c tio n a n d B r e a k d o w n in S o lid
D i e le c t r i c s , Clarendon Press, Oxford, England, 1973.
19. H. St. Onge, H. Bartnikas, R. Braunovic, M. de Tourreil, M. Duval, EPRI
Research Report No. EL-938 (RP-933-1), EPRI, Palo Alto, California, 1978.
20 . C. R. Aschcraft, R. H. Boyd, J. Polymer Sei., Vol. 14, pp. 2153-2193, 1976.
21 . M. leda, M. Nagao and M. Hikita, IEEE Trans, on Dielectrics and Elect.
Insul., Vol. 1, No. 5, pp. 934-945, 1994.
22 . A. von Hippel, J. Appl. Phys. Vol. 8, pp. 815-832, 1937.
23. H. Erohlich, Proc, of the Royal Society of London, Vol. A -160, pp. 230-241,
1937.
24. H. Fröhlich, Proc, of the Royal Society of London, Vol. A -188, pp. 521-532,
1947.
25. F. Seitz, Physical Review, Vol. 76, pp. 1376-1381, 1949.
26. R. H. Fowler and L. Nordheim, Royal Society Proceedings, Vol. A l 19, pp.
173-181, 1928.
27. I. Kitani and K. Arii, IEEE Trans, on Elect. Insul. Vol. 17, pp. 571-576, 1982.
28. K. Arii, K. Kitani and Y. Inushi, Trans, of Japan Inst, of Elect. Eng. Japan,
Vol. 94-A, pp. 251-258, 1974.
29. G. C. Zguris, IEEE Elect. Insul. Magazine, Vol. 4, No. 6, pp. 26-29, 1988.
30. M. M. Khalifa (Ed.), H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c ti c e , Marcel
Dekker, Inc., New York, 1990.
31. M. Nagao, T. Kimura, Y. Mizuno and M. Kosaki, IEEE Trans, on Elect. Insul.,
Vol. 25, No. 4, pp. 715-722, 1990.
32. T. Mizutani, I. Kanno, M. Hikita and M. leda, IEEE Trans, on Elect. Insul.
Vol. 22, No. 4, pp. 432-477, 1987.
33. M. Hakita, Japan J. Appl. Phys., Vol. 24, pp. 988-996, 1985.
34. L. A. Dissado and J. C. Fothergill, E l e c t r i c a l D e g r a d a tio n a n d B r e a k d o w n in
P o ly m e r s , Peter Perigrinus Ltd., London, England, 1992.
35. Y. Yamada, S. Kimura and T. Sato, Proc. 3rd Int’l. Conf. on Conduction and
Breakdown in Solid Dielectrics, Trondheim, Norway, pp. 87-91, 1989.
36. T. Tanaka and A. Greenwood, A d v a n c e d P o w e r C a b le T e c h n o lo g y , CRC Press,
Boca Raton, Florida, 1983.
Vacuum Dielectrics

7.1 INTRODUCTION

Compared to other dielectric materials, an ideal vacuum has, in principle,

the highest possible dielectric strength, because there are no carriers to
conduct the current. In a vacuum below 10“^ Pa, less than 3 X 10^^ mol­
ecules per cubic centimeter are present and the length of the free path is
in the order of meters. In such a vacuum, an electron may cross the elec­
trode gap of few centimeters without any collision with particles. There­
fore, in the absence of the multiplication process, a vacuum behaves as an
ideal insulator. Nevertheless, in actual practice the existence of metallic
and insulating surfaces within the vacuum and the presence of adsorbed
gases and oil vapors contaminates the vacuum. Therefore, the vacuum has
a breakdown threshold, since the electric field intensity ultimately generates
charge carriers by literally pulling them from surrounding electrodes and
contaminants. However, with a proper design, vacuum insulation can still
exceed the breakdown strength of most dielectrics. This is why vacuum
insulation has found widespread use where very high electric field inten­
sities are needed.
Applications of vacuum insulation include electron microscopes.
X-ray tubes, particle accelerators and electrical insulation in space. Its most
important application in power systems is in the form of a vacuum inter­
rupter, which is the central unit of a vacuum circuit breaker. Both the
vacuum and SF^ circuit breakers have increased their share in the power

188
Vacuum Dielectrics 189

utilities worldwide and this trend is expected to grow even faster in the
21st century.

7.2 PREBREAKDOWN ELECTRON EMISSION IN VACUUM

When voltage across a small gap (<2 mm) encapsuled in vacuum is stead­
ily increased to higher values, a relatively small steady current begins to
flow which has been found to consist mainly of electrons. For longer gap
spacings (>10 mm), small pulses of the order of microcoulomb charge
values (microdischarges) can also occur independently or superimposed on
the quasi-steady current. With a further increase in voltage, these micro­
discharges disappear and give rise to a steady current. For both small
and larger gaps, a subsequent increment in voltage causes breakdown of
the gap. With the interelectrode distance of 2 ~ 10 mm, breakdown may
be caused by both the prebreakdown phenomena acting together or separ­
ately [1].
From decades of experimental work it has been established that the
field-assisted electron emission process originates at the metallic protru­
sions on the electrodes and finally leads to breakdown of the gap when
the field is sufficiently enhanced to a critical value. However, recent ex­
perimental evidence has explained a new mechanism, according to which
electron emission originates from insulating/semiconducting surface ox­
ides or impurity particles adhering to electrodes as well. Therefore, a form
of nonmetallic emission mechanism has also been used to explain pre­
breakdown conduction [2]. Both of these mechanisms are the achievement
of meticulous experimental work carried out almost from the onset of this
century.

7.2.1 Electron Emission Mechanism from Metallic Surfaces

Several mechanisms can produce electron emission from metallic surfaces
under different experimental conditions. These are thermionic emission,
Schottky emission, Fowler-Nordheim cold field emission, photoelectron
emission and secondary emission processes [3]. Today the most widely
accepted model of electron emission is the cold field emission model that
was originally presented by Fowler and Nordheim [4]. According to this
model, the current density J from a field emitting sharp electrode as a
function of electric field E at the tip is given as:
B0^M y)
J = AE^ exp A/m^ (7.1)
190 Chapter 7

where A = (1.54 X 10“^)/[(/)P(y)], B = -6.831 X 10^^ and cf) = work

function of the metal.
The expressions t(y) and v(y) are slow varying functions which are
frequently regarded as constants. The above equation can also be expressed
as:
B(/>^My)(i/E)
log -log (7.2)
R2 2.3026
Since A, and essentially v(y) are constants, a plot of log J/E^ against
the reciprocal of E produces a straight line having a negative slope. This
straight line is often used to assess the applicability of Fowler-Nordheim
relation to the experimental data. However, in the case of large area elec­
trodes, the protrusions and uneven surfaces become emitters of unknown
geometry. Therefore, to take account of these difficulties, Alpert [5] mod­
ified equation (7.2) to:
1 B(/)‘My)[d/V]
log = -lo g (7.3)
2.3026/3
Here A' is the electrode area and d is the gap length. Obviously J has been
replaced by (I/A'), whereas the modified field is written as a product of
local field enhancement factor ¡3 and the average field E (= V/d). The
actual emitting area is replaced as A'/3^. Equations (7.1)-(7.3) have been
experimentally confirmed by many investigators. If the enhancement factor
(3 of the electric field due to protrusion is high enough and the critical
threshold value of macroscopic field is exceeded in the gap, the emitting
site explodes within 2 fis [6]. Field-assisted collisional ionization process
then becomes active in the metal vapor thus formed. If sufficient charge
carrier multiplication takes places, then the breakdown of the vacuum gap
through this ionized vapor cloud becomes inevitable.
Other sources of field electron emission have been explained to occur
from oxide films and trapped positive surface charges, absorbed particles
on cathode surface, local hot spots and field-assisted thermionic emission
from low work function regions and cathode microprojections [3]. How­
ever, under clean conditions, the prebreakdown current has been estab­
lished to be due to only the field emission.

7.2.2 Electron Emission Mechanism from Nonmetallic Surfaces

Prebreakdown electronic currents and pulses have also frequently been
found to originate from nonmetallic surfaces [7]. These are associated with
Vacuum Dielectrics 191

some form of insulating/semiconducting oxide layers on the surfaces, or

of impurity concentration. Microinclusions present on electrode surfaces
can stimulate strong electron emission and can thus reduce breakdown
voltage of the vacuum gap [8].
Large area electrodes in vacuum are generally contaminated with ox­
ides, adsorbates and dust. Oxidation in vacuum takes place much more
rapidly than under normal pressure. At a pressure of 10“^ Pa, the oxide
layer on an electrode is formed within 60 ms. Halbritter [9] has shown that
an oxide layer thickness of 5 nm and adsorbates of 10 nm are common.
In the presence of the high electric field, these layers undergo chemical
changes due to the impact of electrons, photons and ions. Presence of
adsorbates and dust enhances the field emission of electrons, whereas ox­
ides and adsorbates are found to enhance the secondary electron emission.
Electrode surfaces in vacuum, also have an adsorption layer which
consists of atoms and molecules from gases and vapor of the pumping oil.
Dominant species of adsorbates have been investigated and are found to
consist of H2 , He, O2 , N2 , CO/N 2 , H2 O, CO2 and certain hydrocarbons.
Under a high electric field, these adsorbed particles are released from sur­
faces due to charge exchange phenomena and cause a week discharge
known as “microdischarge” [10,11]. To account for such electronic emis­
sion from insulating structures, a “field induced hot electron emission
(FIHEE) mechanism” has been proposed by Latham [3]. It involves a com­
posite of metal-insulating emission regime to explain the physical origin
of prebreakdown burst current.
At present, it is a well-established fact that prebreakdown electron
emission can also originate from submicron particulate structures that are
randomly placed on the metallic electrode surfaces. These emission sites
become a potential source of breakdown as soon as they become unstable
in the presence of electric field.

7.2.3 Microdischarges
As explained earlier, the field electron emission is caused by the electric
field at microprotrusions of cathode surface or at nonmetallic contaminants
of the cathode. Beside the prebreakdown field current, the other phenom­
enon linked with electrode surfaces in long gaps is the onset of low power
pulses called microdischarges. These are self limiting with durations of 0.1
to 100 ms, frequency of 0.1 to 100 s“ ^ and amplitudes of <10 mA [1].
They may be caused by (1) small particles of electrode material that are
pulled out from one electrode and strike the other, (2) the beam of electrons
from a cathode that can vaporize a small quantity of material from the
192 Chapter 7

surface of anode or of the cathode, or (3) positive and negative ions ejected
from contaminant films on electrode surfaces through an ion exchange
mechanism.
Microdischarges occur at critical voltage thresholds that depend on the
state of electrode contamination. At a given pressure, the frequency of
occurrence of microdischarges increases with increase in voltage, until the
eventual occurrence of breakdown.

7.3 FACTORS AFFECTING BREAKDOWN VOLTAGE IN VACUUM

Electric strength of vacuum is defined in different ways. In the case of

vacuum insulated switchgear, it is the value of the voltage to cause the first
breakdown that is important. However, when a gap is caused to breakdown
repeatedly, the breakdown voltage increases with the number of breakdown
shots until it reaches a steady or “conditioned” value. This^alue is often
taken as the breakdown strength of the vacuum gap. Figure 7.1 illustrates
this characteristic. It has mainly been attributed to the removal and/or
smoothing out of microprotrusions or due to the change in the work func­
tion of the electrodes because of adsorption processes [13]. In general the
relative effects of various parameters affecting the breakdown voltage of
the vacuum gap are always determined based on its conditioned value.
These various factors include electrode separation, electrode material, re­
sidual pressure and the type of the applied voltage waveform.

150
kV
120

90
Vb
60

30

0 AO 80 120 160 200

Figure 7.1 AC breakdown voltage as a function of number of sparks (n). Gap

spacing = 3.00 mm. (From Ref. 12 © IEEE, 1993.)
Vacuum Dielectrics 193

7-3.1 Effect of Electrode Separation

In a vacuum gap, as the voltage is raised above the onset of prebreakdown
phenomena, various conduction processes intensify, as described earlier. In
small gaps (<2 mm) and with extremely clean electrode surfaces, bright
spots of light appear at the anode. At still elevated stresses, these hot
regions that emerge due to impinging of electron beam from cathode can
start radiating thermally, eventually leading to spark breakdown. The break­
down voltage Vb is linear function of gap spacing d such that:
Vb-kd (7.4)
where k is a constant. For such gaps the breakdown stress is relatively
high and is of the order of 10^ V/cm. Linear relation indicates that field
emission of electrons plays a dominant role in the breakdown process in
this region.
In longer gaps (>1 cm), field emission currents are low while the
breakdown still occurs with a localized spark channel. However, in this
case breakdown stress is considerably decreased, e.g., for a gap of 1 cm it
is of the order of 10"^ V/cm only. It obeys a nonlinear relationship with
gap length:
Vh = kd" (7.5)
where, for a given electrode surface conditions, k and n are constants for
a given range of d. In this case, generally the microsized particulate struc­
tures lying on the electrode surfaces are believed to dominate the break­
down process [15].

7.3.2 Effect of Electrode Material

The breakdown strength of a vacuum insulated gap is also a strong function
of electrode material. Contact material exerts a strong influence on nearly
every phase of the behavior of vacuum gap because the discharge depends
entirely on the contact metal vapor, whereas the contact properties are
directly influenced by the microstructure and the physical properties of the
contact metal. Several investigators have shown that there exists a critical
field for each material, at which breakdown occurs due to field resistive
heating of the protrusions [3]. Table 7.1 gives a partial list of electrode
materials and the values of the breakdown voltages associated with these
electrodes when they are placed 1 mm apart. It clearly illustrates that the
correct choice of electrode material is of vital importance if a high insu­
lating capability is to be achieved for a given gap design.
194 Chapter 7

Table 7.1 Experimentally Measured Critical Breakdown

Voltage Across 1.0-mm Vacuum Gap

Electrode material Voltage (kV)

Stainless steel 179

Chrome plated copper (baked at 500°C) 143
Chrome plated copper (unbaked) 89.4
Nickel 89.5
Aluminum 57
Silver 27

Source: Ref. 14.

Electrode geometry with regard to size and shape also plays a signif­
icant role. The larger the area of the electrode, the lower is the breakdown
voltage. Similarly, electrodes of dissimilar material exhibit a strong polarity
effect.
Surface finish is yet another factor associated with electrodes, which
can be manipulated to get higher breakdown voltage. Besides this, there
have been investigations into the effectiveness of the evaporated metal films
as well as the insulating films which can give improvements in breakdown
voltages by a factor of 2 to 3. The increase in breakdown voltage with
insulating film has been attributed to the suppression of field emission by
dielectric film.

7.3.3 Effect of Pressure

The influence of residual gas pressure on the breakdown voltage depends
on the gap length. For small gaps (<1 mm), the improvement of the vac­
uum beyond 10“^ Pa shows only insignificant changes in breakdown volt­
age [16]. However, longer gaps exhibit a definite pressure effect and an
anomaly is observed such that with the increase in pressure, the breakdown
also increases at first, but then decreases sharply with subsequent increase
in its value. This phenomena has been confirmed by several investigators.
Figure 7.2 illustrates this effect [17].

7.3.4 Effect of Voltage Waveform

For its application in power systems, broad area electrodes and longer gaps
(>1 cm) are required in vacuum interrupters. However, most of the results
reported for power frequency applications give breakdown data in gaps
Vacuum Dielectrics 195

Figure 7.2 Effect of pressure on the breakdown voltage in a stainless steel sphere-
plane electrode gap of 20 cm. (From Ref. 17.)

that are confined only to a few millimeters [16,18]. Comparatively little

work has been done under both AC and lightning/switching impulses in
longer gaps. Under AC, due to the reversal of polarity under each half
cycle (making of anode region to cathode under each polarity switchover),
the breakdown voltage level remains almost the same as reported for DC
voltage [15]. Contrary to that, the breakdown values in 1.5 cm gaps under
standard lightning impulse (1.2/50 jus) have recently been reported to be
30-40% higher than AC breakdown values (see Figure 7.3) [20]. It is also
demonstrated that processes with time lags of >50 ¡jiS play an important
role in the breakdown of vacuum gaps of >10 mm [20,21].

7.4 BREAKDOWN MECHANISMS

As a result of intense theoretical and experimental investigations over the

last several years, different theories have been put forward in an attempt
to explain quantitatively the mechanism that initiates breakdown in vac­
uum. Generally, these theories fall into four main categories [22].

7.4.1 Clump Theory

This theory postulates that breakdown is initiated by the removal of a
charged particle or aggregate of material (clump) from one of the elec-
196 Chapter 7

Figure 7.3 Breakdown voltage under AC and impulse voltage stress in a stainless
steel electrode gap. (From Ref. 20.)

trodes, which is then accelerated across the gap to the opposite electrode.
The impact energy causes a localized temperature rise that is high enough
to free some other material from the electrode surface as a result of the
impact. This mechanism was forwarded by Cranberg [23]. He postulated
that when the energy deposited on impact by microparticles reaches a crit­
ical value, the breakdown occurs. A consequence of this hypothesis is that
breakdown voltage is proportional to the square root of the gap length.
This hypothesis has been reasonably substantiated as a contributing but not
as a sole cause of vacuum breakdown [24].

7.4.2 Interaction Theory

According to this mechanism, an electron present in the gap would be
accelerated by the voltage and impinge on the anode, where it will produce
positive ions and photons due to ionization of absorbed gases. They, in
turn, would be accelerated back to cathode, resulting in emission of sec­
ondary electrons. These would then be accelerated to anode, releasing fur­
ther positive ions and photons. When such a chain reaction of electrons,
ions and photons reaches a critical level, then the breakdown of the gap is
triggered. A qualitative explanation of this mechanism is as follows:
Let Aq = average number of electrons produced by one electron
Bq = average number of secondary electrons produced by one of
these electrons
Vacuum Dielectrics 197

= average number of photons produced by one electron

Dq = average number of secondary electrons produced by a
photon
The breakdown will occur if the coefficient of production of secondary
electrons exceeds unity, i.e., (A^Bq + Q D J > 1.

7.4,3 Cathodic Theory

Cathodic theory assumes field emission of electrons from protrusions at
the cathode, as explained earlier as due to Fowler-Nordheim mechanism.
These protrusions can give rise to an enhanced electric field (j8E) of the
order of 200, which, when introduced in the Fowler-Nordheim equation,
predicts sufficient current density to contribute to vacuum breakdown. Mi­
croprotrusions on the cathode surface are heated, evaporated or exploded,
due to the emitted current, liberating the cathode material, which enters
the interelectrode region and is subsequently ionized. This results in what
is known as cathode-induced breakdown.

7.4.4 Anodic Theory

Anodic theory assumes involvement of both cathode and anode in the
breakdown process. It assumes the existence of a beam emitted from the
cathode that impinges on the anode causing a local rise in temperature and
release of gases and vapors. Additional electrons ionize the atoms of gas
and produce positive ions. At the cathode, the effect of these ions is two­
fold: (1) increased primary electron emission due to space charge formation
and local field enhancement, and (2) electron emission by bombardment
of the cathode surface. This process continues until sufficient gas is gen­
erated to give rise to a low-pressure discharge in the stressed gap.
Of all these theories, the interaction theory (see section 7.4.2) has been
ruled out as applied to clean vacuum gaps and electrode surfaces. Similarly,
no single theory seems to explain all breakdown events reported in the
literature. Depending on gap geometry, electrode material, and voltage
waveform, one can establish a situation in which a clump, cathode or anode
process will be the primary cause of breakdown. However, a major volume
of the present and past work reported in the literature tends to support the
field-emission-initiated vacuum breakdown theory [24]. This field emission
may be followed by, or may occur together with, other mechanisms dis­
cussed here. At low stress values, evidence supports a heating of the emis­
sion areas followed by melting of the emitter. At higher stresses, explosive
destruction of the emitter occurs. In either case, the liberation of metal
198 Chapter 7

leads to plasma production in the near-cathode space. This copious pro­

duction of electrons by the plasma may produce an anode plasma through
deposition of energy at the anode surface. However, this anode plasma is
not produced under all experimental conditions. In all cases the produced
plasma can be further ionized to breakdown condition by electrons stream­
ing from the cathode emission sites [24,25].

7.5 ARC INTERRUPTION IN VACUUMS

7.5.1 Vacuum Arc
An electric arc is basically a type of discharge between two electrodes. It
differs from the general gaseous discharge, since in this case (1) the elec­
trode current density is extremely high, of the order of kiloamperes per
cm^; (2) a greater part of discharge current near the cathode is carried only
by the electrons and not by the ions formed during the ionization of me­
dium; and (3) the potential difference between electrodes is small, i.e., only
of the order of a few tens of volts. In the vacuum, arcing is established
when the current carrying electrode contacts separate, melting or explosive
vaporization of the last metallic point of contact occurs initially only with
a single melting point on the cathode, known as the cathode spot. The
cathode spot is a small, limited region of high temperature and pressure
from which ions, electrons and neutral particles are emitted. At this spot,
the current density is >10^ A/cm^, whereas surface temperature is close
to the boiling point of the contact material. The peripheral areas of this
spot represent an intense source of neutral metal vapor, which is ionized
in the form of a discharge cone with its apex at the cathode. Contrary to
arc discharge in a gaseous medium, more than 90% of the total current in
the vacuum arc is transported by the electrons, whereas positive ions cause
a neutralization of the negative space charge produced by the electrons.
If the arc current exceeds a limit that is found to be a function of
contact material, the cathodic melted point splits into a number of parallel
cathode spots. Studies show that without the influence of external magnetic
field, these spots move in random directions around the entire cathode
surface at high speeds, while the adjacent spots repel each other. This
process leads to further production of copious metallic vapor. This type of
discharge is called “diffused arc discharge” and is shown in Figure 7.4a.
During diffused discharge, only the cathode is the active electrode, whereas
the anode forms a condensation surface for vapor particles and charge
carriers. Thus it functions merely as a collector. At current levels >10 kA
(with cylindrical copper electrodes), the arc is constricted at the anode as
shown in Figure 7.4b. Anode spots are formed while cathode spots con-
Vacuum Dielectrics 199

(a)

Diffused

(b ) (c)

Anode C o nstrictio n at
Constriction Anode and
CATHODE B Cathode

Figure 7.4 Schematic diagrams of various modes of the vacuum arc: (a) diffused
arc; (b) arc constriction at anode; and (c) arc constriction at both electrodes (anode
and cathode).

verge to a single spot. The constricted vacuum arc possesses in it a high-

pressure metal vapor and also causes a voltage drop at the anode. The
constriction of arc at both electrodes as shown in Figure 7.4c forms a
thermal column, and in this case erosion of anode becomes inevitable.
Therefore, in the case of high short-circuit current interrupters, local over­
heating of electrode contacts must be avoided. This is achieved by constant
forced motion of the arc spots. For this purpose, spiral electrodes are used
as shown in Figure 7.5 [39]. The current path induced by this spiral shaping
of contacts generates a magnetic field whose tangential force vectors act
on the spots forcing them to move around on the contact surfaces at a high
velocity. This action not only avoids burning of the contacts, but also helps
in the handling of large interrupting currents. The question of arc contrac­
tion and arc motion and the role of the magnetic field is a subject of
continuing research [28].

7.5.2 Vacuum Gap After the Arc

In AC circuit breakers, the arc is interrupted at a current zero. At current
zero, the interelectrode space is quickly deionized, and the space surround­
ing the electrodes is filled with a residue consisting of desorbed gas, neutral
electrode vapor and plasma. The process, the time in which this residue
200 Chapter 7

Figure 7.5 Spiral contacts of an interrupter. (From Ref. 39.)

decays and the result of the vacuum gap regaining its dielectric strength
are referred to as an arc recovery phenomena.
If one considers the stresses on the interrupter contacts after the current
zero, one can see that power frequency voltage exists at the generator side
contacts, whereas the voltage of the faulted resonant circuit exists on the
load side contacts. The vector sum of these voltages forms the stress on
the contact gap. This total is known as transient recovery voltage (TRV).
The rise of the dielectric strength in the contact gap (dielectric recovery
voltage) must take place more rapidly after quenching of the arc than for
recovery voltage, otherwise “restriking” of the arc will take place. Under
real power network fault conditions, the TRV consists of many high-
frequency oscillations, one superimposed on the other. This results in a
very high rate of rise of recovery voltage (RRRV), and thus dictates the
stress factor such that at any instant during arc interruption, it should not
exceed the dielectric recovery of the vacuum gap.
It therefore becomes apparent that both the maximum RRRV and the
peak TRV are important parameters for circuit breaker specifications and
testing. In general, the dielectric recovery speed of separating contacts in
an interrupter highly depends on (1) the prior thermal stresses on the con­
Vacuum Dielectrics 201

tact gap, which are controlled by the arc voltage, the breaking current and
the arcing time, and (2) the arc quenching medium, which also plays a
significant role. In Figure 7.6, tests with nitrogen reflect the conditions in
a compressed air breaker, and tests with hydrogen approximately reflect
the conditions in minimum oil circuit breakers. It is clear that dielectric
recovery takes place slowly in circuit breakers in which an arc quenching
medium is used [29]. Contrary to that, the dielectric recovery in vacuum
is accomplished within —10 ^ts [30]. In practice, single units have shown
the rate of rise of recovery voltage of the order of 20 kV/jas, which is at
least an order of magnitude higher than in other types of interrupters.
Similarly, the arcing time in a vacuum is extremely small. Interruption is
usually accomplished at the first current zero after the contacts part. Arc
duration is therefore always much less than one cycle.

7.5.3 Current Chopping and Contact Material

Below a certain minimum current, a metal vapor arc is interrupted prior to
its sinusoidal zero. This phenomena is called current chopping [29]. It can
lead to generation of high overvoltages on the power network or other
devices connected across the vacuum circuit breaker (VCB), caused by the
magnetic energy still trapped in the circuit’s inductance. Studies have
shown that chopping is not only dependent on the breaking current but
also on the contact material [32,33]. This shows that at each instant during
arc interruption, there must be exactly as much metal vapor available as is

10 100 1000 10000

T im e(/is) _ i ^

Figure 7.6 Dielectric recovery voltage as a function of time elapsed after arc
extinction at current zero (rms current = 1600 ampere). Gas pressure 1.0 bar with­
out switching media flow. Electrode spacing = 6.35 mm. (From R ef 29.)
202 Chapter 7

required to conduct the current then flowing, otherwise the current will be
interrupted with a high rate of di/dt, resulting in a buildup of unnecessary
voltage oscillations in the inductive circuits (high Ldi/dt) connected on the
load side. While the interrupter may successfully withstand this voltage
without breaking down, the voltage may cause failure in other parts of the
connected circuit.
During the 1960s, tungsten alloys were used as contact material be­
cause they possess a high melting point. A great amount of energy is
required to vaporize this material. With small currents approaching natural
zero, this energy is no longer available, thus causing a forced quenching
of the arc (its chopping). This problem impeded the introduction of VCBs
in the market. However, in the 1970s this problem was solved with the
introduction of a low-surge-type contact material consisting of a copper-
chromium (CuCr) alloy. It combines good electrical properties with arc
erosion and good welding resistance [34]. With these contacts the average
value of chopping current has been contained to around a 3.0 A level
[32,35]. More recently, a silver-tungsten carbide (AgWC) contact material
has been introduced [33,36]. In this case, the base contact material is WC,
and the high pressure metal component is silver. With these contacts the
small current arc can be maintained to just before the current reaches its
natural zero, because the vapor from highly molten silver metal is injected
into the interelectrode space. In this case, the value of chopping current
lies close to that of puffer type SF^ breakers and minimum oil circuit
breakers [30]. Other studies show that in some circuit conditions, even SF^
breakers can have higher chopping currents than those exhibited by vacuum
interrupters [37].
There are, however, some high surge impedance circuits where surge
protection is required in parallel with VCBs [38]. For repeated switching
of medium voltage motors using vacuum contractors, a high vapor pressure
material such as bismuth is added to the CuCr base to lower the value of
the chop current, which is generally not severely affected by repeated
switchings [26,34]. Table 7.2 compares the chopping currents of various
materials [28].

7.6 VACUUM CIRCUIT BREAKER

As discussed earlier, the vacuum is an excellent insulator and arc extin­

guisher. Extensive research and development effort over the past three dec­
ades in vacuum arcs, arc interruption processes and high voltage design
have lead to the production of vacuum circuit breakers (VCB) that exhibit
an outstanding interruption performance. Their short-circuit interruption
V acuum D i e le c t r i c s 203

Table 7.2 Average Chopping

Current of Pure Metals and
Some Alloys at 45 A, 50 Hz and
400 V

Cu 18 A
Cr 6.5 A
W 9 A
Ca 8 A
Bi 0.3 A
Sb 0.4 A
Cu Cr25 6 A
Cr Cr50 5 A
Cu Cr25 Sb9 3.9 A
Cu Cr25 ZnlO 3.2 A
W Cu30 6.2 A
W Cu30 Sb2 2.8 A
WC Ag 40 1.7 A

Surge impedance = 1 kil.

Source: Ref. 28 © IEEE, 1993.

capability has been pushed to currents of 50 kA and voltages of up to 36

kV. The applications of VCB over the past two decades have demonstrated
its strength over the other technologies. It now enjoys more than 50% of
the world market share, which is continuously growing [26]. Both, the
vacuum and SF^ technologies provide advantages over minimum oil and
magnetic air circuit breakers. The user can employ either SF^ or vacuum
interrupter in metal clad switchgear and have circuit protection that will
satisfy the required operational requirements.

7.6.1 Construction of Vacuum Interrupter

Central to the design of a VCB is the vacuum interrupter. Figure 7.7 il­
lustrates the internal components of a typical vacuum interrupter. The am­
bient gas pressure within the evacuated envelope is kept in the range of
lO^^-lO^'^Pa. Under normal conditions the contacts are closed and elec­
trodes are in contact with each other.
When the bellowed contact moves away from the stationary contact,
the arcing is established as a result of vapor evaporated from the local hot
spots on the surfaces of the contacts. There is a continuous supply of this
metal vapor, which is distributed on the contact surfaces, whereas the re­
maining vapor condenses on the surrounding metal vapor condensation
204 Chapter 7

shield. The latter is usually isolated from both contacts and serves to pro­
tect the glass or the ceramic external insulator from the deposition of the
vapor. When the current wave reaches its natural zero, the vapor production
stops and the original dielectric strength of the vacuum is rapidly restored,
which is above the transient recovery voltage applied on the contacts due
to the external circuit. While the contacts remain isolated, they are held
externally by the insulating envelope. The metal bellow enables the moving
contact stem to carry out its mechanical stroke, which varies according to
the rated voltage of the interrupter.
External insulating envelopes are generally made of glass or metallized
aluminum-oxide-based ceramics, which permits them to be permanently
brazed to metal. Therefore, there are no replaceable seals, because the
interrupter has a permanent sealed construction. Field data gathered by
almost all manufacturers of vacuum interrupters have confirmed that the
probability of vacuum loss is negligible [26]. Moreover, if these units are
not mishandled, their service life with full current interrupting ratings is
much longer than 20 years.

7.6.2 Limitations of VCB

Figure 7.8 shows the dielectric strength of different insulating mediums.
For a gap of —7.0 mm, the vacuum interrupter has higher breakdown
Vacuum Dielectrics 205

Figure 7,8 Lightning impulse withstand capability of various insulating media as

a function of gap spacing in homogeneous field. (From Ref. 19.)

strength than the SF^ breaker operating with 5.0 bar pressure. With the
increase in contact gap, the breakdown strength of SF^ gas increases lin­
early, whereas in a vacuum it tends toward saturation. For a gap of 16 mm,
the corresponding strength of VCB is 180 kV, which is slightly higher than
basic impulse level (BIL) of 170 kV recommended for 36 kV power fre­
quency operations. It is clear that VCB has considerable advantage over
an SF^ breaker in the medium voltage range. To go for higher voltage
levels, therefore, two or more of such interrupters will be required in series.
This is always not very economical. It can thus be deduced that VCB is
advantageous up to rated voltages of 36 kV, but beyond that level, the SF^
breaker takes the lead.

7.6.3 Merits and Demerits of VCB

Merits
1. In a vacuum interrupter, the electrical contacts remain unaffected
by the ambient internal environment, so there is very little change
in its contact resistance throughout its life. In SF^ breakers, on the
other hand, arc by-products continuously react with internal com­
206 Chapter 7

ponents, and if there is an ingress of moisture, it will produce

hydrofluoric acid, which will react with contacts and glass filled
insulators, which can affect the equipment integrity. Generally,
VCB remains pollution free.
VCB is self contained and does not need gas or oil filling. There
is no need for an auxiliary air system or oil handling system.
The necessary maintenance is modest, thus it is economical in the
long term.
4. There is no risk of explosion or leakage of toxic byproducts.
5. VCBs have silent operation due to short stroke and light mechan­
ical loadings and shorter contact parting time. Moreover, a smaller
number of components are needed.
6. Since contact erosion is small, the normal current can be inter­
rupted up to 30,000 times, and rated short-circuit current interrup­
tions up to 100 times are generally quoted—more than all other
types of circuit breakers.

Demerits
1. In the event of loss of vacuum, the unit cannot be repaired at the
site.
2. In some applications where low magnetizing currents are in­
volved— such as furnace transformers, reactors and some high
voltage motors—the additional surge suppressors are required in
parallel with the VCB in each phase [32]. These are installed as
close as possible to the load and are also permissible in the switch-
gear panels.

REFERENCES

1. R. V. Latham, IEEE Trans, on Elect. Insul., Vol. 23, No. 1, pp. 9 -16, 1988.
2. W. Ziomek and H. M. Grzesiak, IEEE Trans, on Elect. Insul., Vol. 28, No. 4,
pp. 481-487, 1993.
3. R. V. Latham, H ig h V o lta g e V acu u m I n s u la tio n , Academic Press, London,
England, 1981.
4. R. H. Fowler and L. Nordheim, Royal Society Proceedings, Vol. A 119, pp.
173-181, 1928.
5. D. Alpert, J. Vacuum Sei. and Technol., Vol. 1, No. 2, pp. 35-50, 1964.
6. E. Dullni, IEEE Trans, on Elect. Insul., Vol. 28, No. 4, pp. 454-460, 1993.
7. G. A. Farrall, IEEE Trans, on Elect. Insul., Vol. 20, No. 5, pp. 815-841, 1985.
8. B. M. Cox, J. Phys. D, Vol. 8, No. 17, pp. 2065-2073, 1975.
9. J. Halbritter, IEEE Trans, on Elect. Insul., Vol. 20, No. 4, pp. 671-681, 1985.
Vacuum Dielectrics 207

10. M. Wurtz, H. Adam and W. Walcher, T h e o r y a n d P r a c ti c e o f V acu u m T ech ­

n o lo g y , Freid Viewveg and Sohn, Braunsweig, Germany, 1989.
11. V. A. Nevrovskiy, J. Phys., Vol. 14, pp. 215-220, 1981.
12. J. Ballat, D. König and U. Reninghaus, IEEE Trans, on Elect. Insul., Vol. 28,
No. 4, pp. 621-627, 1993.
13. M. G. Danikas, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 3, No. 2,
pp. 320-321, 1996.
14. J. B. Birks (Ed.), P r o g r e s s in D i e le c t r i c s , Vol. 7, Hey wood Books, London,
England, 1967.
15. V. N. Mailer and M. S. Naidu, A d v a n c e s in H ig h V o lta g e I n s u la tio n a n d A r c
I n te r r u p te r in SF^ a n d V acu u m , Pergamon Press, Oxford, England, 1981.
16. R. Hackam and L. Altech, J. Appl. Phys., Vol. 46, pp. 627-636, 1975.
17. K. W. Arnold, Proc. 6th International Conference on Ionization Phenomenon
in Gases, 2, 101, 1963.
18. C. C. Erven, IEEE Trans, on PAS, Vol. 91, pp. 1589-1596, 1972.
19. R. Hawley, “Vacuum as an Insulator” , Vacuum, Vol. 10, pp. 32-41, 1961.
20 . F. Schneider, H. C. Karner and M. Gollar, Proc. 6th International Symposium
on High Voltage Engrg., New Orleans, 1989, Paper 48.07.
21 . H. G. Bender and H. C. Karner, IEEE Trans, on Elect. Insul., Vol. 23, No. 1,
pp. 37-42, 1988.
22 . H. C. Miller, IEEE Trans, on Elect. InsuL, Vol. 26, No. 5, pp. 949-1043,
1991.
23. L. Cranberg, J. Appl. Phys., Vol. 23, No. 5, pp. 516-522, 1952.
24. J. E. Thompson, Proc. 6th International Symposium on High Voltage Engrg.,
New Orleans, 1989, Paper 48.01.
25. G. A. Meayats and D. I. Proskurovsky, P u l s e d E l e c t r i c a l D is c h a r g e s in V ac­
u u m , Springer Verlag, Berlin, Germany, 1989.
26. P. G. Slade and R. W. William, Power Technical International, pp. 171-175,
1993.
27. F. G. Rowland, Electronics and Power, Vol. 21, pp. 496-499, 1975.
28. K. Erohlich, H. C. Kamer, D. König, M. Lindmayer, K. Möller and W. Rieder,
IEEE Trans, on Elect. InsuL, Vol. 28, No. 4, pp. 592-596, 1993.
29. T. V. Armstrong and P. Headley, Electronics and Power, Vol. 20, pp. 198-201,
1974.
30. E. Dullni and E. Schade, IEEE Trans, on Elect. InsuL, Vol. 28, No. 4, pp.
607-620, 1993.
31. J. M. Lafferty, V acu u m A r c s , T h e o r y a n d A p p lic a tio n , John Wiley and Sons,
New York, 1980.
32. D. L. Swindle, IEEE Trans, on Industrial Applications, Vol. 20, No. 5, pp.
1355-1363, 1984.
33. E. Kaneko, K. Yokokura, M. Hommn, Y. Satoh, M. Okawa, J. Okutami and
I. Ohshima, IEEE Trans, on Power Delivery, Vol. 10, No. 2, pp. 797-803,
1995.
34. P. G. Slade, Proc. 18th International Conference on Electrical Contacts, 1992.
35. S. S. Rao, S w itc h g e a r a n d P r o te c tio n , Khanna Publishers, Delhi, India, 1993.
208 Chapter 7

36. R. P. R Smeets, E. Kaneko and I. Ohshima, IEEE Trans, on Plasma Sci., Voi.
2, No. 4, pp. 439-445, 1995.
37. J. D. Gibbs, D. Koch, P. Malkin and K. J. Comick, IEEE Trans, on Power
Delivery, Voi. 4, No. 1, pp. 308-316, 1989.
38. J. F. Peridne and D. Bheeavanich, IEEE Trans, on Industrial Applications,
Voi. 19, No. 5, pp. 679-686, 1983.
39. ABB-Calor Emag, Germany, Cat. VD4-Vacuum Circuit Breaker, drawing no.
V0003 Sp/E.
8
Composite Dielectrics

8.1 INTRODUCTION

The insulation requirements of a power network rarely consist of a single

material. The use of two or more insulating materials becomes necessary
due to design considerations or due to practical difficulties of fabrication.
These different materials may be in parallel with one another such as an
air gap in parallel with solid insulation, or oil in parallel with pressboard.
Similarly they may be in series with one another, such as laminates. Apart
from these, filled or reinforced insulating materials may have microscopic
volumes of another phase of different materials present in the bulk, e.g.,
in polyethylene or in a mixture of two granulated substances in epoxy
resins or impregnated solids, etc. Therefore, all of these forms of insulation
are composite in nature.
In certain cases the behavior of a composite insulation system can be
predicted by the behavior of individual components. But in most cases, the
system as a whole has to be considered. The performance of such a system
can be evaluated only by considering the factors such as: (1) the stress
distribution at different parts of the insulation system, (2) breakdown char­
acteristics at the surface, as they are effected by the boundaries of the
composites, and (3) partial discharge products and/or chemical aging prod­
ucts of one component which may react with the other component. Simi­
larly the economic life of the composite system will have to be considered,
since under long-term operation, several types of breakdown mechanisms
(other than electronic, electromechanical or thermal breakdown mecha-

209
210 Chapter 8

nisms, as explained in chapter six) may be in progress either independently

or in combination. This chapter is concerned with composite dielectric
systems that are generally encountered on a high voltage power network.

8.2 DIELECTRIC PROPERTIES OF COMPOSITES

Figure 8.1 illustrates slabs of two dielectrics placed perpendicular (in se­
ries) to the axis of an applied AC electric field E. If we assume that neither
one of the dielectrics affects the properties of the other and no moisture,
gas inclusion, etc., exist, then the ratio of electric fields in the two mediums
varies inversely with their respective dielectric constants, i.e., E , 8 i =
E2 8 2 •
Similarly, under a direct applied voltage the respective field ratio will
vary directly with the resistivity ratios of the two mediums. Since the
composite system shown in Figure 8.1 can be considered as two capacitors
in series, the voltage distribution across the upper dielectric of thickness
d, can be given as:
V
V. - ( 8. 1)
1 + (8 jd2 / 8 2 dj)
where V is the total voltage applied across the composite and d2 is the
thickness of the lower dielectric. If the upper dielectric is air (8 , = 1) in
series with a solid insulating sheet or barrier, and suppose dj = 4 d2 and
8 2 = 3, the evaluation of stress using equation (8.1) shows that the stress

in the air gap has increased by 15% due to the presence of the insulating
barrier. This stress in the air gap will further increase if the barrier thick­
ness is increased and will approach 300% of the stress value without the

Figure 8.1 Two-layer composite dielectric.

Composite Dielectrics 211

barrier when (I2 is the only marginally smaller than d, the total gap length.
If the air gap is marginally designed then there is a chance that it will
breakdown and full voltage will appear across the barrier which is then
likely to fail. Therefore, in such a situation, one must ensure that the air
space is capable of withstanding the increased stress caused by the pres­
ence of the solid insulating material.
The other commonly used composite dielectric is generally a two-
phase mixture of either reinforced fibers with resins, or inorganic filler in
a polymeric resinous matrix as shown in Figure 8.2. Thin cavities may be
present due to inefficient manufacturing process in such systems. The in­
terface therefore exists around every minute fiber or particle of filler and
cavities. The difference in coefficients of thermal expansion and volume
shrinkage can create free volumes in the bulk. Similarly, many of the re­
inforcing materials have great affinity to absorb moisture. If a chemical
bond does not exist at the interface of solids, then water molecules can
easily diffuse into the composite and form water layers [1]. Presence of
water will reduce the DC resistivity and the dielectric strength, whereas it
will slightly increase the permittivity of the composite. However, the dis­
sipation factor will suffer an appreciative increase. This implies that the
dielectric properties of a composite measured at a certain time mainly
depend on the history of the specimen.
The dielectric constant of a composite is generally based on the di­
electric constants of the individual components and their volume fractions
{v). If a composite is formed of only two components, neither of them

§§o°<i=o°o°cS§O o°o°°
O O ^ O O o u W ^ O O rw Q
ROq QQ9^ 9 P Q g
(b )
Figure 8.2 Internal bonding of composite mixtures: (a) mica platelets and
(b) parallel rods of fiberglass.
212 Chapter 8

affects the properties of the other and provided that moisture, gas inclusions
or similar other factors are absent, then the value of s and tan 8 for the
composite can be given as:

S = -- ^ ( 8 . 2)
8 , V2 + £ 2 ^1

tan 6 j + (8 j V2IS2 ^\) * ^2

tan 8 = (8.3)
1 + (Sj V2I ^2 )
Here the prefix numbers represent the individual values of the two com­
ponents used to formulate a composite. The representation of such a com­
posite as a simple two-layer dielectric is no doubt a simplification of a
complex problem. A full account of various theories and experimental re­
sults related to this topic have been given by van Beek [1] and Bartinkas
and Eichhorn [2].

Interfacial Polarization in Composites

If an insulating material is composed of two or more different phases that
contain dispersed macroscopic impurity regions, then on the application of
an electric field a space charge builds up at the microscopic interfaces as
a result of differences in conductivities and permittivities of its individual
components. This is known as interfacial polarization. It will lead to in­
creased dielectric losses and also cause field distortion in the material [2,3].
When one of the composite components has significant ionic conduc­
tivity, then mobile charges (usually impurity ions) diffuse under the influ­
ence of the applied field across the more conducting component up to the
interface of the less conducting component. At this new interface they will
become stationary and thus build up a surface charge, as shown in Figure
8.3. This effect will remain there until the applied field reverses, as in the
case of alternating voltage. In the case of DC voltage and longer stressing
time, a much larger space charge buildup continues until a very significant
reverse electrical field develops from this surface charge, eventually ar­
resting the current flow to a value that is determined by the less conducting
component. This type of polarization is very common in composite insu­
lation, since such interfaces are distributed throughout its structure. A re­
cent study [4] has established that the use of DC voltage testing of poly­
meric cables is harmful for their life, which depends on several factors;
the most important of these is due to the space charge buildup in semi­
crystalline structure of crosslinked polyethylene. The high concentrations
of traps are found at the crystalline-amorphous interfaces [5]. The study
Composite Dielectrics 213

| - oi>a,

i± ___

1 “ o-,>aj

Figure 8.3 Interfacial polarization in a composite dielectric.

of space charge accumulation in polymeric cables and solid insulation is

a topic of current research worldwide [6,7].

8.3 EDGE BREAKDOWN

In practical systems electrodes may be embedded on the surface of insu­

lation as shown in Figure 8.4. It is clear that at the edges of electrode a

Figure 8.4 Breakdown of solid specimen due to edge effect.

214 Chapter 8

composite insulation is formed, since the ambient medium becomes in

series with the insulating slab under test. Assuming homogeneous field,
consider elementary cylindrical volume of area dA spanning the electrode
at distance x, then on application of voltage V across the electrodes, its
fraction Vj [as given in equation (8 . 1 )] shall appear across the ambient air
medium. For most of solid dielectrics used in power apparatus, = 2 --
3. Hence there will be substantial enhancement of the field in the air me­
dium. The stress will further increase as x is decreased. Consequently, the
ambient medium will breakdown at relatively low voltage as d, becomes
very small. The charge at the tip of the resulting discharge will make the
electrode arrangement highly nonuniform. Therefore, local breakdown at
the tip of discharge impinging at the solid interface will result in a dis­
charge channel which will erode its surface. Occurrence of many such
breakdown channels will result in the formation of a tree in the slab which
will extend step by step through the whole thickness, ultimately causing a
complete breakdown of the slab insulation. However, it should be noted
that the tree-like pattern of discharge is not limited specifically to the elec­
trode edge effect but may be observed in other failure mechanisms in which
nonuniform stresses predominate. For details, see Chapter 9.

8.4 CAVITY BREAKDOWN

Solid insulations, and to a lesser extent, liquid dielectrics, contain voids or

cavities within the medium or at the boundaries between the dielectric and
the electrodes. Similarly, in amorphous polymers, there are vacant spaces
or holes between the molecules which are called free volume. In polyeth­
ylene, below the glass transition temperature Tg, the free volume is ap­
proximately 2.5% of the total volume, which increases with temperature
above Tg[8 ]. Similarly, epoxy resin composites are prepared by mixing
reinforcing fillers. Generally there exists between them a chemical or quasi­
chemical adsorption-type bonding at the interface. But if care is not taken
in fabrication, many areas of interface may have no bonding but a gap of
varying thickness. These voids are generally filled with medium of lower
dielectric strength and lower permittivity. Hence the electric strength in the
voids, as shown earlier, will be higher than that across the dielectric. There­
fore, even under normal working voltages the field in the voids may exceed
their breakdown value, and breakdown may occur in these small entities.
Figure 8.5 shows a cross-section of a dielectric of thickness d, con­
taining a cavity in the form of a disc of thickness t, together with its
corresponding “abc” equivalent circuit. In this circuit the capacitance
corresponds to the cavity, corresponds to the capacitance of the dielec-
Composite Dielectrics 215

trie that is in series with and is the capacitance of the rest of the
dielectric. For t < < d, which is usually the case, and assuming that the
cavity is filled with a gas, the field strength across is given by the
expression where is the relative permittivity of the dielectric.
For the simple case of a disc-shape dielectric in solid shown in Figure
8 .5 , the discharge inception voltage can be expressed in terms of the cavity

breakdown stress. Assuming that the gas-filled cavity breakdown stress is

E^f,, then treating the cavity as series capacitance with the healthy part of
the dielectric, we may write:
A

C. =

The voltage across the cavity is:

V
V = V =
c. + c.
1 . 1 ^ - 1

fir \ t
Therefore the voltage Vj across the dielectric which will initiate discharge
in the cavity will be given by

(8.4)

Generally, a cavity in a material is often nearly spherical, and for such a

case the internal field strength can be given as:
216 Chapter 8

3g,E
(8.5)
rr + 2 e,

where e^c is the dielectric constant of the cavity, and E is the average stress
in the dielectric. Since therefore E^ 3/2 E. Hence, the stress
inside the cavity becomes much greater than the average stress in the
dielectric.
The sequence of breakdown under sinusoidal alternating voltage is
illustrated in Figure 8 .6 . The dotted curve qualitatively shows the voltage
that would appear across the cavity if it did not breakdown. During the
positive half cycle, as reaches the value a discharge takes place,
the voltage collapses and the gap extinguishes. The voltage across the
cavity then starts again increasing until it reaches when a new dis­
charge occurs. Thus several discharges may take place during the rising
part of the applied voltage. Similarly, during the negative half cycle of the
applied voltage, the cavity discharges as the voltage across it reaches
In this way, groups of discharges originate from a single cavity and give
rise to positive and negative current pulses on the positive and negative
cycles of the voltage wave. The discharge in such an insulating medium
can be measured with partial discharge (PD) detectors. For details refer to
section 1 2 .6 .

Va

Figure 8.6 Sequence of cavity breakdown under alternating voltages.

Composite Dielectrics 217

When the gas in the cavity breaks down, the surface of the insulation
acts as cathode and anode. Some of the electrons impinging upon the anode
are sufficiently energetic to break the chemical bonds of the insulation
surface. Similarly, bombardment of the cathode by positive ions may cause
damage to the internal surface of the cavity by increasing the surface tem­
perature and producing local thermal instability. Additional chemical deg­
radation may result from active discharge products, e.g., O3 or NO2 , formed
in air, which may cause insulation deterioration. Whatever the deterioration
mechanism operating, the net effect is a slow erosion of the material and
a consequent reduction in the thickness of the solid insulation. Partial dis­
charges are therefore harmful to the insulation, regardless of the size of
the void in which the discharges take place [9]. However, a larger void is
more dangerous than a smaller one, since the larger void not only dis­
charges at a lower applied stress, but also dissipates more energy in each
discharge [10], and thus causes more severe insulation damage. A survey
has revealed that void size > 50 pm is the main cause of the high voltage
cable-insulation failure [ 1 1 ].

8,5 BREAKDOWN DUE TO SURFACE EROSION

AND TRACKING

Polymeric insulators are preferred in many application as they are easy to

fabricate in complicated shapes and are tough, light in weight and possess
excellent dielectric properties. However, their service on high voltage net­
works is coupled with the processes of degradation due to erosion and
tracking. When the discharges occur on the insulation surface, the erosion
takes place initially over a comparatively larger area. The erosion roughens
the surface and also causes pitting, which aids accumulation of contami­
nation and finally gives way to tracking. Erosion and pitting is also pro­
moted due to the “edge mechanism” explained earlier.
Tracking is the formation of a permanent conducting path, usually
carbon, across the surface of insulation, resulting from degradation due to
continuous discharges and erosion. In service, the surface of insulation is
progressively contaminated. The contamination takes up moisture from the
atmosphere and the wet layer of pollution provides a continuous, con­
ducting path between the HV electrode and the ground. The surface resis­
tance decreases considerably in the presence of pollution and moisture.
Low resistance, in turn, leads to high surface leakage currents and high
power dissipation, causing significant loss of moisture from the surface.
This loss is not uniform, and leads to the formation of dry bands. When a
dry band is formed, the flow of surface leakage current is interrupted. The
218 Chapter 8

inductance of the system generates HV surges so that the effect is similar

to a circuit breaker in which the contacts sustain the system voltage plus
a high frequency transient reaching up to twice the supply voltage. In the
case of dry banding, almost all of the applied voltage is concentrated across
these dry bands and electrical discharges occur, causing the insulation sur­
face to reach high local temperatures at the arc root, leading to gradual
erosion and the formation of carbonaceous residues which become the
focus of further action. Complete breakdown of the insulation surface gen­
erally follows when a conducting carbon path propagates to the extent that
the remaining insulation is incapable of withstanding the system voltage.
A number of international standard testing methods have been estab­
lished in order to evaluate the relative tracking resistance of different in­
sulating materials [12]. At present, the generally accepted method is the
liquid contaminant inclined plane tracking method, described in IEC-587
(1984) and ASTM-D 2303 (1984) [13,14]. The lEC method offers two test
procedures for the time to track ( 1 ) under a constant voltage, and (2 ) under
a stepwise voltage application. The lEC method is preferred because it has
been observed that two materials with the same initial tracking voltage
(ITV) but having a different resistance to tracking (due to filler content)
may give very different times to track [15]. Another method widely used
in many European countries and in the United States is described in the
ASTM Test Method D 495 [16]. In this method, voltage is applied between
two chisel-shaped tungsten rods as shown in Eigure 8.7. No contaminant
is used in this test method. The numerical value of the voltage that initiates
tracking is called the “tracking index” and is used to quantify the surface
properties of the dielectric under test.

8.6 CHEMICAL AND ELECTROCHEMICAL DETERIORATION

AND BREAKDOWN

Since insulating materials are either composed of differing chemical sub­

stances or come in contact with materials of different composition, chem­
ical reactions between various materials are inevitable. Application of con­
tinuous electric stress in the presence of high temperature may act as
catalyst and enhance the rate of chemical reactions. Thus such composites
may undergo chemical deterioration leading to reduction in electrical and
mechanical strengths. The most deleterious effects are caused by O2 in the
presence of ultraviolet radiation and contact with moisture/water.
In the presence of air (O2 ), materials such as rubber and PE undergo
oxidation, giving rise to surface cracks. During electric discharge in air,
nitrous oxide (NO) is produced, which in the presence of moisture forms
Composite Dielectrics 219

Tungsten rod

Figure 8.7 Dry arc electrode assembly.

nitric acid ( H N O 3) which can degrade the dielectric. Similarly, in the pres­
ence of moisture, paper, cotton tape, cellose, etc., deteriorate due to
hydrolysis.
Individual composite material such as epoxy resins, may have exten­
sive interface area between the two mixed solids throughout the composite
(see Figure 8.2). In such cases, thermal expansion coefficient of the two
solids may have significant difference. For example, glass in reinforcing
fibers has a linear thermal expansion coefficient (a) of about 0.5 to 0.7 X
10“^ cm/cm/°C, while the resins used in composites with the glass have
a = 2 ~ 9 X 10“^ cm/cm/°C, which is nearly tenfold higher [17]. With
an increase of temperature, this may result into breakage of the interfacial
chemical bonds, thus leaving behind cavities or microcracks. Since most
of the reinforcing materials have a great affinity for water, water molecules
can diffuse into these free sites and replace them with water layers. Pres­
ence of such a water film along the interface will further accelerate the
process of electrochemical deterioration. Moreover presence of such a film
will increase the electrical conductivity and the dielectric losses.
Similarly, water in contact with insulator surface (or absorbed mois­
ture) has been found to decrease significantly the surface resistivity. Study
on the cellulose-based laminates has shown a decline of about 65% in the
dielectric strength, in the presence of 90% humidity in the air [17]. In case
of composites containing hydrophilic fibers, if both sides of one of the
internal fibers extends to the wet external surface, it will form an internal
wet path. The conduction along the internal electrolyte path can develop
sufficient heat to vaporize the water at the hottest point, thereby breaking
the electrical circuit and forming a small arc. This will thus result into
carbonized tracks. This mechanism is similar to external tracking, but in
this case there is less cooling by evaporation and convection, so the internal
arcs may develop at lower voltage gradients. Such an internal arc tracking
220 Chapter 8

has been found to cause carbon tracks in resins, which do not normally
carbonize under external arcing conditions [17].
It should be particularly recognized that the power frequency short-
time dielectric strength values usually quoted are not usable in service,
since all organic resin composites decline very significantly in dielectric
strength with the time of voltage application. The time to failure seems to
vary as an inverse exponential of the electric stress, as shown in Figure
8 .8 . The decline in strength is due to partial discharges occurring internally

or externally at the surface. There is also a decline in dielectric strength

with thermal aging, which degrades the organic resins and the chemical
bonds of the organic reinforcement [18].

8.7 MATERIALS OF OUTDOOR INSULATORS

Porcelain and glass have been the most widely used materials for outdoor
HV insulation applications. Extensive service experience has shown that

Log [Time to breakdown (seconds)]

Figure 8.8 Voltage endurance of mica insulation. (From Ref. 2 © ASTM, re­
printed with permission.)
Composite Dielectrics 221

these materials are very reliable and cost effective for a majority of outdoor
HV applications. However, since the early 1960s, alternative materials,
namely polymers, have emerged and presently they are being used exten­
sively for a variety of outdoor HV applications. The chief advantages of
polymers are their light weight, superior vandal resistance and better han­
dling of mechanical shock loads.
Initially, polymeric insulators were considered as replacements for por­
celain and glass for special applications such as areas with a high incidence
of vandalism, urban locations with limitations on right of way, and areas
of severe contamination problems. However, many difficulties were en­
countered regarding their performance in actual service during the fist two
decades of operation. Typical of these were tracking and erosion of sheds,
chalking and crazing of sheds, which lead to increased contamination col­
lection; arcing and flashover; bonding failures and electrical breakdown
along the rod-shed interface, corona splitting of sheds and water penetra­
tion due to hot line water washing which also resulted in electrical failures
[21]. However, through continuous effort and research in polymeric ma­
terial development and improved manufacturing technology, much im­
proved versions of insulators have been introduced. Their comparable per­
formance with porcelain and glass have lead to their widespread use even
for routine outdoor HV insulation applications.
Today, polymeric line insulators are in use on lines operating up to
765 kV. However, they are more popular on transmission levels from 69
through 345 kV. A worldwide survey carried out by CIGRE [19] explains
that there are several thousand polymeric insulators in service at all distri­
bution voltage levels. An EPRI survey of polymeric insulators in the United
States has reported that 78% of the utilities reported good performance,
18% reported acceptable, and only 4% were found unsatisfied with the
performance of polymeric insulators.

8.7.1 Basic Polymeric Insulator Design

The basic construction of polymer insulators for overhead line applications
consists of a core, weathersheds and metal end fittings. Two types are in
common use. The suspension/dead-end type (Figure 8.9a) is used where
line loads subject the insulator core to tension forces; the post type (Figure
8.9b) is used where line loads subject the core to appreciable bending
forces in addition to the tension forces. The only significant differences
are in the design of the attachment hardware and in the size of the core,
which is much larger for posts [ 1 0 ].
The core of a nonceramic insulator has the dual burden of being the
main insulating part and of being the main load-bearing member, be it in
222 Chapter 8

CORE WEATHERSHEDS

(a )

(b )
Figure 8.9 (a) Suspension/deadend-type and (b) post-type polymer insulators.
(From Ref. 10 © IEEE, 1991.)

suspension, cantilever or compression. For suspension and line post insu­

lators, the core consists of axially aligned, glass fiber-reinforced resin con­
taining 70-75% by weight of E-type fiberglass. The fiber diameter ranges
from 5 to 20 pum. The resin system can be of polyester or epoxy and the
rod is formed by the pultrusion process. Although epoxy resin is considered
to be the better of the two, because of lower cost the core used today is
usually polyester resin.
The end seal is considered to be the most important element of the
design of a nonceramic insulator. Field failures have occurred due to brittle
fracture of the fiberglass rod due to breach of the end seal, thereby allowing
the rod to come into contact with atmospheric pollutants and moisture.
Tracking of the fiberglass rod leading to failure has also been observed in
nonceramic insulators.
Nonceramic insulator end seals have three basic types: glued, friction
and boned types. Glued-type seals that are made using a sealant material
such as an RTV silicone rubber applied between the sleeved core and the
hardware are not permanent, generally because of poor adhesion. Friction-
Composite Dielectrics 223

type seals in which the sleeved core fits into the hardware are quite effec­
tive, as long as the dimensional tolerances are maintained; they do not
cause any problems, provided that no movement of the fitting occurs. End
seals that are made by molding the sleeved core material onto the end
fitting are by far the best because of the physical bond obtained during
molding [2 1 ,2 2 ].

8.7.2 Development of Composite Materials

Sheds made from various nonceramic materials compounded for electrical
applications are shaped and spaced over the rod in various ways to protect
the rod and to provide maximum electrical insulation between the attach­
ment ends. It is quite clear that with such a diversity of constructions
possible, the performance of nonceramic insulators depends on the choice
of materials, the design and the construction of the insulator. A variety of
polymer insulating materials has been developed for overhead electrical
insulation. This includes PTFE (Teflon), epoxy resins, polyethylene, instant
set polymers based on urethane chemistry, polymer concretes, various co­
polymers, ethylene-propylene elastomers and room temperature and high
temperature vulcanizing silicone elastomers. Each material offers particular
characteristics. However, only the elastomeric materials have shown suc­
cess in outdoor electrical insulation applications, with silicone meeting
all of the requirements for long-term performance in practically all
environments.
Today, only three classes of materials are in any significant use: epoxy
resins, hydrocarbon elastomers and silicone elastomers. Properties of these
have been described in detail in section 6.2.5. The polymers have the ability
to interact with pollutants and reduce the conductance of this pollution
layer [23,24,48]. This is illustrated in Figure 8.10. The important charac­
teristic of this polymeric insulator which controls the conductance is due
to hydrophobicity or water repellency of its surface. It has been observed
that the hydrophobicity is maintained in silicone rubber materials even after
many years in service, and it is this attribute that is responsible for the
superior contamination performance of silicone rubber family of materials
when compared to other polymers. The recovery of hydrophobicity is
mainly due to ( 1 ) a diffusion process whereby low molecular weight
(LMW) polymer chains with only a fraction (—20%) from within the bulk
of material migrate to the surface, where they form a thin layer of silicone
fluid, and (2 ) reorientation of surface hydrophillic groups away from the
surface. These processes are temperature dependent. Higher temperature
causes more rapid recovery. Moreover, regeneration of the LMW chain
224 Chapter 8

Figure 8.10 Surface resistance of bare and silicone-coated porcelain insulators

under salt fog conditions (1.0 inch-25.4 mm). (From Ref. 23 © IEEE, 1991.)

occurs even after the initial supply of LMW polymer chains has been
depleted, but this recovery from within the main material causes weakening
of its mechanical properties [49].

8.8 OIL-IMPREGNATED INSULATION

The use of oil-impregnated cellulose paper in high voltage equipment has

persisted to the present day despite the competition from a variety of syn­
thetic materials. Although synthetic materials such as polypropylene (PP)
possess lower tan 6 , lower moisture absorption and increased dielectric
strength, their capability of impregnation and mechanical strength are much
inferior to that of cellulosic paper, generally known as “kraft paper.” How­
ever, new materials that are in fact a sandwiched composite of paper and
PP films are now available. Similarly, a variety of impregnants are also
currently in use that range from low-cost petroleum oil to specially syn­
thesized products.

8.8.1 Liquid-Solid Film Interaction

Impregnation can be regarded as a physical interactive process controlled
predominantly by the surface tension force. The rate determining factors
Composite Dielectrics 225

include the viscosity of the imprégnant, its surface tension and the contact
angle with the dielectric. A smaller contact angle (ff) between the liquid
and the solid causes rapid penetration of the liquid from the insulation
surfaces into the pores of the solid. The surface tension force is basically
equal to r] cos 6, where r] is the surface tension of the liquid and is there­
fore dependent on the nature of the imprégnant. In other words, impreg­
nation, which is a very important process to fill the cavities of solid films,
is dependent on the cavity size and the nature of the imprégnant. On the
other hand, excessive fluid absorption by the film results in its swelling,
which produces several disadvantages, such as drop in electrical strength
and reduction in tensile strength.
Chemical interaction between solid and liquid dielectrics is generally
confined to various leaching processes whereby low molecular weight or
ionic components migrate from the solid to the liquid or vice versa. The
presence of ionic contamination in the liquid can be detected by an increase
in conductivity or loss angle, as shown in Figure 8.11. This increase can,
in turn, result in thermal limitations on AC equipment, and can also lead
to further chemical deterioration at a rate which will depend on current

Figure 8.11 Loss tangent as a function of temperature for paper and various syn­
thetic polymers. (From Ref. 44.)
226 Chapter 8

density and chemical activity of ions liberated at the electrodes. It can also
be enhanced by the presence of moisture and other contaminants that can
dissociate in the dielectric. Electrochemical deterioration is a major cause
of failure in liquid-impregnated power capacitors [25].

8.8.2 Effect of Multiple Layers

The simplest form of composite insulation system consists of two layers
of the same material. This system is extensively used especially in cable
and capacitor construction, where advantage is taken of the fact that two
thin sheets have a higher electric strength than a single sheet of the same
total thickness. Research work carried out on thin sheets has illustrated that
this phenomenon is mainly due to the loss of energy from the discharges
or partial breakdown channels at the interfaces. A discharge having pene­
trated one layer could not enter the next layer of material until the spot on
the interface centered on the channel had been charged to a potential which
could produce a field comparable with that of the channel at the level in
question.

8.8.3 Application in HV Cables

In practice, cables have to experience mechanical stresses such as bending
that could cause wrinkles or creases at the butt joints. It has been dem­
onstrated experimently that the formation of such deformation reduces
breakdown strength. Therefore, cable insulation requires impregnated in­
sulation with high mechanical and electrical strength. Impregnated paper
is used in the form of tapes. This is lapped onto the cable conductor and
then impregnated; it may or may not be pressurized. Such cables exist in
two categories, self-contained oil-filled (OF) cables and pipe-type OF
(POF) cables. The bulk of the underground cable systems installed in
United States are high-pressure, POF-type cables. In contrast, self-
contained OF cables are commonly used in Europe and Japan [8 ].
Experiments have shown that the thinner the paper sheet, the higher
its dielectric strength. This arises from the resulting reduction in the oil-
gap length, and increases the barrier effect and the impermeability. The
barrier effect is caused by the arrangement of the stacked fine fiber. Figure
8 . 1 2 shows a linear relationship between impulse breakdown strength and
the logarithm of paper thickness [8 ].
Impregnated-paper insulation in cable consists of simple layers of pa­
per tapes. Each tape from 50 to 200 ^im in thickness is wrapped helically
around the conductor. The resulting insulation is comparatively homoge­
neous. One of the most important features of pressurized oil-impregnated
Composite Dielectrics 227

2 3 4 5 8 10 ' 2
Thickness of Insulating Pokier (mm)

Figure 8.12 Thickness dependence of impregnated paper on impulse breakdown

voltage. (From Ref. 8.)

paper is its high resistance to and/or freedom from corona or internal

discharges. This provides the stability needed in cable insulation, especially
for solid impregnated cables, and it is a main reason for its use in such
applications.
Unfortunately, its electrical performance can be easily affected by a
small amount of water. Figure 8.13 shows the variation of tan 8 due to the
addition of water. Thermal degradation is also enhanced by a small amount
of water. For this reason, it is a standard practice to dry paper in a vacuum
before oil impregnation and to protect a finished cable from water per­
meation with a lead sheath. In cables, since lower permittivity of insulation
is required, mineral oils or alkylbenzene synthetic oil or their mixtures are
used.

8.8.4 High Voltage Capacitors

Capacitors were introduced in HV power apparatus about 75 years ago
when reliable and economical oil-impregnated paper was developed. Today
HV capacitors are found in precision devices such as voltage dividers and
in high kvar banks for power-factor correction on inductively loaded cir­
cuits. The dielectric is invariably an impregnated polymer of natural or
synthetic origin. High voltage measuring divider systems generally contain
a dielectric of mineral-oil-impregnated paper, but for power-factor correc-
228 Chapter 8

Figure 8.13 Effect of water content on loss tangent of oil-impregnated paper.

(From Ref. 8.)

tion capacitors, the trend is towards an increasing use of polyproplyene

film. The lower loss angle of dielectric materials in capacitors gives a
reduction in dielectric heating and lower power consumption. Polypropyl­
ene film has been found to be a superior dielectric for capacitor application
because of its reduced dielectric losses and higher dielectric constant. The
volume of the dielectric in a capacitor is inversely proportional to the
product of its Sj. and square of its working stress E, i.e., [47]. There­
fore, use of PP film reduces both the size and weight of the capacitor.
However, mixed paper/PP film dielectric also continues to be used in stan­
dard tank-type capacitors, since they enable a more complete impregnation
of oils as compared with all-film-type capacitors. The latest designs of
capacitors have loss tangents only one-tenth of the value for the early paper
types.
The impregnants of power capacitors should have high dielectric
strength, high dielectric constant, good insulation affinity with paper, chem­
ical and thermal stability and adequate viscosity for efficient conduction
of heat. Mineral oils with high aromatic content for high gas-absorbing
Composite Dielectrics 229

ability have traditionally been used. However, they have been replaced with
a variety of synthetic oils such as di-isopropylnaphthalene (DIPN), mono
isopropyl biphenyl (MIPB) and phenylxylyl ethane (PXE), because of their
high gas-absorbing ability, high dielectric strength and dielectric constant.
Dimethyl siloxane oil that is used as a replacement fluid in transformers
instead of PCBs was found unsuitable for capacitor application because of
its gas-evolving character under high electric stress. However, phenyl
methyl silicone oil (PMS) possesses high gas-absorbing ability and is also
used as nonflammable impregnan! for HV capacitors. Blends of (BO) tri-
cresyl phosphate (TCP) with DIPN and/or PXE are also being used as
self-extinguishing capacitor impregnants.
Some modified aromatic hydrocarbons such as cumylphenylethene
(CPE) and diphenylethane (DPE) have recently been introduced in Japan
for capacitor impregnation [46]. Properties of some widely used high volt­
age capacitor impregnants are summarized in Table 8.1.

8.8.5 HV Bushings
Bushings are used at the interfaces between various power distribution
equipments. The high voltage versions contain fluid- or resin-impregnated
insulation which is operated at a fairly high stress level. The construction
is typically a porcelain housing containing a paper element which is wound
and lapped around a central conductor. Coarse textured papers are often
used to facilitate drying and impregnation. If the bushing is integral with
other equipment, a common impregnan! is used, otherwise a good-quality
mineral oil is suitable. A well-known design is the capacitor bushing, which
contains concentric layers of paper dielectric interlayered with aluminum
foils where each foil forms an equipotential screen. In service, bushings
often experience the full extent of fast-rising transients and it is normal for
the bushing insulation to be rated in terms of its ability to withstand im­
pulse breakdown.

8.8.6 Impregnated Insulation in Transformers

During the past 30 years, many new types of synthetic varnishes and solid
insulations have been used in transformers. Nevertheless, the mechanical
stability and load-bearing properties of cellulose are still highly valued by
design engineers. Much of the insulation system used in oil-filled trans­
formers consists of oil-impregnated pressboard placed around the low and
high voltage windings and supported by the iron core. The thin oil layers
within the multilayer pressboard structure result in high dielectric strength,
which increases with pressboard density. In addition, the oil passing be-
 O
boo

Table 8.1 Properties of Imprégnants of High Voltage Power Capacitors

Impregnating oil

Property MO TCD^ DIPN PXE CPE MIPB PMS BO

Specific gravity 0.88 1.40 0.96 0.99 0.96 0.99 0.97 1.07
Viscosity 30°C (cSt) 11.0 25.0 8.5 8.0 5.2 8.3 35 37
8, (60 Hz, 80°C) 2.18 5.20 2.48 2.51 2.44 2.55 2.65 4.61
Tan 5 (60 Hz, 80°C) 0.01 0.04 0.005 0.005 0.005 0.005 0.08 0.61
Breakdown voltage 70 65 80 80 80 80 58 60
(kV /2.5 mm)
Gas generating 46 — 70 70 72 71 — 60
voltage (kV/mm)
Fire point (°C) 135 170 144 148 156 140 317 148
Pour point (°C) -3 2 -1 7 -4 7 .5 -4 7 .5 -6 0 -5 0 -7 7 -4 0

"*Use is restricted in many countries due to environmental concerns.

MO = mineral oil, TCD = trichlorodiphenyl, DIPN = di-isopropylnaphthalene, PXE = phenylxylyl ethane, CPE = cumylphenylethene,
MIPB = mono isopropylbiphenyl, PMS = phenylmethyl dimethyl siloxane, BO = blend of DIPN or XPE and TCP.
Composite Dielectrics 231

tween the individual pressboard components of the insulation structure al­

lows the required heat dissipation in order to achieve the necessary elec­
trical and thermal performance.
In transformers, the average working electric stress is usually low and
is typically less than one-fifth of the stress value used in impregnated paper
cables and capacitors. This is essential, as there is a need to maintain a
high impulse strength over complicated insulation geometry. Various tech­
niques are used to determine the field distribution in transformers under
both impulse and steady-state conditions. Similarly, thermal degradation of
the fluid and paper has to be minimized and the operating temperature for
the fluid is usually restricted to an upper limit of 60°C. The fluids which
are chosen for use in larger power transformers generally have a high
oxidation resistance and may contain inhibitors.

Aging of Paper/Pressboard Insulation

The lifetime of transformers is usually considered to be around 30 years.
A substantial number of transformers were installed worldwide in the
1950s, a period of sustained economic growth. Such units are now antic­
ipated to have reached the end of their life and therefore have been sub­
jected to intensive investigation and diagnostics [26]. Various test proce­
dures have emerged from these investigations.
Diagnosis of Lifetime by Degree of Polymerization
Insulating paper used in an oil-immersed transformer decomposes gradu­
ally due to heat dissipation. Figure 8.14 shows some of the decomposition
products for cellulose paper. The number of cellulose molecules, m, indi­
cates the degree of polymerization. It is usually 1000-1100 for the new
transformer kraft paper. A critical value of end for lifetime estimation is
considered to be around 500, as at that level the paper becomes extremely
brittle and cracks upon bending. The degree of polymerization (DP) of a
small amount of transformer paper sample is measured using IEC-450 [43].
Insulating paper located at the highest temperature point, such as the upper
part of winding inside the transformer, is usually evaluated. The lifetime
of the transformer is then predicted from equation (8 .6 ) [27]:
D = D,(l - r)- ( 8.6)

where D is the degree of polymerization, is the initial degree of polym­

erization, r is a constant and t is the operation period.
Diagnosis by this method, however, has recently been subjected to
criticism due to poor reproducibility of results. Investigations have shown
that this method gives only a crude average of molecular chain size and is
therefore a coarse measure of transformer degradation [28]. It is also re-
232 Chapter 8

H OH CH2OH H OH
HO,
’v /a i h\ j * \/^
*jt/ a N
h ^ _ _ o/^OH
CH2OH H OH CH2OH
C ellu lo se m

Products of Alcohol

CH 3 OH C2 H5 OH
i HO CH 2 t )
Products of Aldehyde
r T
CH3CH0 C2H5CH0
O -C -H H0CH2-1[3c-H
0
F u r fu r a l
i'
Products of Acid
r I----------
CH3C00H C2H5C00H

1
CO 2 . CO
COOH

Figure 8.14 Degradation products of cellulose. (From Ref. 26 © IEEE, 1993.)

ported that due to long exposure to high temperature, although kraft paper
becomes brittle and looses its tensile strength, yet the dielectric strength
does not vary much with the reduction in degree of polymerization, as
shown in Figure 8.15 [29]. This shows that transformer insulation is still
good although it has in fact aged considerably, since it is ultimately the
loss of tensile strength that determines the life expectancy of the insulation.
Diagnosis of Lifetime by Dissolved Gas Analysis
The diagnosis of oil-immersed transformers using the dissolved gas anal­
ysis (DGA) technique has been widely accepted internationally and is be­
ing used by laboratories, transformer users and manufacturers since the
introduction of the IEC-599 [30]. This method is effective in the diagnosis
of faults in transformers, such as arc discharge, partial discharge and over­
heating. The deterioration of paper insulation in this method uses the
CO2 /CO ratio and other hydrocarbon gases dissolved in transformer oil,
and not by directly sampling the paper or pressboard. According to this
method, the formation of fault involving paper insulation is probable when
Composite Dielectrics 233

- 1 0 0 1 100
X •/. V.
I 80
•3
80
<u
S 60 60

AO AO
Bi)
UJ
2 0 - 20

Figure 8.15 Tensile strength, elongation and dielectric strength as a function of

an average degree of polymerization. (From Ref. 29 © IEEE, 1996.)

the CO2 /CO ratio is <3. It is based on the fact that the generation of CO
increases faster than that of CO2 as the decomposition temperature of cel­
lulose increases. This method was first used by Tamara et al. [31] and is
based on the fact that there exists a good correlation between the amount
of dissolved CO2 gas and the degree of polymerization of insulating paper
in oil-immersed transformers.
The continuous use of this method has shown some difficulties in in­
terpreting DGA results as well. Lifetime diagnosis is comparatively diffi­
cult for N2 -sealed-type oil-immersed transformers, because dissolved CO2
escapes from oil into the nitrogen cushion. The concentration of CO2 gas
strongly depends upon the temperature at which the oil is sampled. It is
often found that concentration of CO and CO2 dissolved in transformer oil
is higher in summer and lower in winter, which may result in misleading
conclusions. Moreover, it is also found that CO2 , once decomposed from
pressboard, can be reabsorbed by the paper if the operating temperature is
below 80°C [32].
Diagnosis by the Measurement of Furfural Dissolved in Oil
As shown in Figure 8.14, insulating paper is decomposed into alcohol,
aldehyde, acid and finally into carbon dioxide based on the degree of aging.
At the aldehyde decomposition stage, 2-furaldehyde (commonly known as
furfural) is generated. Abnormal operation conditions such as overheating
and electrical discharge would generate CO2 and furfural in abundance.
Aging phenomena related to furfural have been reviewed in by Schroff and
Stannett [33]. It was found that the amount of furfural in oil is proportional
234 Chapter 8

to the log (DP), i.e., the degree of paper polymerization. A method for
furfural analysis has been proposed in lEC Publication 1198 [34], which
utilizes high-performance liquid chromatography (HPLC). More recently,
techniques such as “ size exclusion chromatography” (SEC) of the paper
to measure changes in its molecular weight distribution, which is related
to the degree of polymerization, have also been introduced [35,36].
In fact, the whole are of transformer insulation aging, monitoring and
life prediction has recently been reviewed [37,38], and a comparative study
of different methods that are in common use shows that HPLC analysis of
oil can potentially be better used to monitor in service degradation of paper.
Normal levels of 2-furaldehyde are generally in the range of 100-1000
ppb. However, levels of about 1 ppm have also been measured in “normal”
transformers, rising to 5 ppm in a transformer running hot and to over 10
ppm in transformer that was overheated due to failure in its cooling system.
However, this is a subject of current research in a number of centers
throughout the world.

8.9 FLEXIBLE LAMINATES

Elexible laminates are used extensively in electrical equipment. Often an

application demands insulation that may have high dielectric strength and
high mechanical strength at high temperatures. Usually a single dielectric
does not possess all these characteristics. Therefore, two materials pos­
sessing different characteristics can be combined in the form of sheets so
that one can take advantage of best of both types of materials.

8.9.1 Composite Laminate Classifications

Elexible laminates are manufactured by combining selected dielectrics with
appropriate substances using various types of adhesives as bonding agents.
Outer layers of these laminates may, in some cases, be further treated with
various resinous compounds to provide surface or volume characteristics
required for some applications. lEC Publication 626 [40,41] is a guide for
the identification of these complex laminates.
Use is made of internationally recognized abbreviations for the various
elements of these laminates as shown in Table 8.2. Eurther, a system of
classification of the finished laminates has been developed which not only
greatly simplifies their description (Table 8.3), but also provides a conven­
ient and foolproof means for purchasing them internationally.
Composite Dielectrics 235

Table 8.2 Designation of Flexible Laminate Components

Form of Code Code
component designation Nature of component designation

Film F Cellulose acetate CA

Cellulose triacetate CTA
Polyethylene terephthalate PETP
Polyimide PI
Polycarbonate PC
Polypropylene PP
Paper and non- P Cellulose paper or presspaper C
woven fabric Polyamide (aromatic) paper PAa
Polyethylene terephthalate PETP
Woven fabrics C Cotton or viscose C
Glass fabric G
Polyethylene terephthalate PETP
Adhesive A Pressure-sensitive, thermoplastic Tp
Pressure-sensitive, thermosetting Ts
S o u rce: Ref. 45 © IEEE, 1991.

Table 8.3 Laminate Descriptions

Performance Properties Parameters

Structural Dimensional stability Thickness, total and individual ply

Deformation resistance Water absorption
Rigidity Flexural modulus
Mechanical Ability to withstand Tensile strength elongation
fabrication stress Edge tear resistance
Resistance to service Interply adhesion
stresses, thermal
expansion, etc.
Electrical Insulation Dielectric breakdown voltage
Arc resistance
Surface and volume resistivity

S o u rce: Ref. 45 © IEEE, 1991.

236 Chapter 8

(a) j Film Total

10.023 -a050 0.10-0

if5
' hi!
Iq. H
z
” Id1
Q
H-.'
UJ ^ U
ui a.
J
0. 2:
01,
u i - ;•n
I
Q. Q

THICKNESS (mm)
( b)
Type Film Total

1 0.023 -0 .1 2 5 0 .1 2 -0 .5 0
2 0.050 - 0.35 0 .1 5 -0 .4 5
3 0.023 -0 .3 5 0.20 - 0.55
(0.050) (0.30)^
4 0.023 - 0.050 0 .2 0 -0 .3 3

Figure 8.16 Combined flexible laminates, types and thickness range (abbrevia­
tions as per lEC 626-1): (a) double combination and (b) triple combination. (From
Ref. 45 © IEEE, 1991.)

This classification system employs the abbreviations of Table 8.2, ar­

ranged sequentially in the order found in the actual laminate, with the
various plies separated texturally by means of a slash (/). Thus, a laminate
formed by bonding glass fabric to polyester film by means of a thermo­
plastic pres sure-sensitive adhesive would be designated as:
C-G/A-Tp/F-PETP
Various types of combined flexible laminates comprising double, triple and
quadruple plies are commercially used and are identified in the various
specification sheets of lEC Publication 626 Part 3 [41]. Figure 8.16 shows
a schematic representation of typical double- and triple-layer laminates.
The outer layers may be coated or impregnated with selected resins that
may be either fully cured to provide enhanced thermal capabilities or semi-
Composite Dielectrics 237

P -P A a /F -P I/P -P A a

P-PAo/F-PETP/P-PAa
P-PETP/F-PETP/P-PETP
impregnated
outer layers
P ^ T P /F -E T P /P -P E T P
P - C / F - P E T P /P -C
F - P E T P / P -C

P -C

Limiting Temperature :
Th erm al Class

Figure 8.17 Thermal classes and limiting temperatures. (From Ref. 45 © IEEE,
1991.)

cured so that the laminates can be further cured by heating after insertion
in the electrical apparatus, thus assuring improved mechanical strength and
integrity of the unit.

8.9.2 Selecting Flexible Laminates for Thermal Requirements

At present there are no standard laboratory methods for evaluating the
thermal endurance of flexible laminate insulation. Design engineers usually
rely on materials known to perform satisfactorily under certain accelerated
thermal conditions. lEC publication 85 [42] is a compilation of insulating
materials which experience has shown to be suitable choices for specific
temperature exposures.
Using lEC Publication 85 as a guide, and supported by available ser­
vice data, the combinations shown in Figure 8.17 have been used to in­
sulate electric equipment operating continuously at the temperatures indi­
cated. It will be noted that application of selected resins to the outer layers
of these laminates may in some cases upgrade the thermal capability. In
the particular case of P-PETP/F-PETP/PETP, for example, such treatment
has been shown to raise the insulation thermal classification from Class B
238 Chapter 8

(130°C) to Class F (135°C). A commonly used saturant for this particular

insulation is a suitable epoxy resin. It should be noted here that both the
choice of resin and the degree of impregnation usually depend on the
nature of the particular laminate, the needs of the apparatus to be insulated
and the thermal capability sought.

REFERENCES

L.K.H. van Beek, “Progress in Dielectrics” , J.B. Birks (Ed.), Chemical

Rubber Co. Press, Ohio, Vol. 7, pp. 69-114, 1967.
2. R. Bartnikas and R.M. Eichhorn, E n g in e e r in g D i e le c t r i c s Vol. 1, “Corona
Measurement and Interpretation,” ASTM Publications, Philadelphia, 1983.
3. C.J.E. Böttcher and P. Bordewijk, T h e o r y o f E le c tr ic P o la r iz a tio n , Elsevier
Press, New York, 1979.
4. B.S. Bernstein, IEEE Elect. Insul. Magazine, Vol. 10, No. 4, pp. 33-38, 1994.
5. L. D. Dissado and J.C. Fothergill, E l e c t r i c a l D e g r a d a tio n a n d B r e a k d o w n in
P o ly m e r s , Peter Perigrinus Ltd., London, England, 1992.
6. S. H. Lee, J.K. Park, J.H. Han and K.S. Suh, IEEE Trans, on Dielectrics and
Elect. Insul., Vol. 2, No. 6, pp. 1132-1139, 1995.
D. Malec, R. Essolbi, H. Giam, B. Ai and B. Garros, IEEE Tran, on Dielectric,
and Elect. Insul., Vol. 3, pp. 64-69, 1996. '
T. Tanaka and A. Greenwood, A d v a n c e d P o w e r C a b le T e c h n o lo g y , CRC Press,
Boca Raton, Florida, 1983.
9. M. G. Danikas and T. Tanaka, IEEE Trans, on Dielectrics and Elect. Insul.,
Vol. 1, No. 3, pp. 548-549, 1994.
10. E. A. Chemey, IEEE Elect. Insul. Magazine, Vol. 7, No. 2, pp. 28-32, 1991.
11. T.D. Mintz, IEEE Trans, on PAS, Vol. 103, pp. 3448-3453, 1984.
12. A.S.G. Alghamdi and A.H. Mufti, Proc. 5th GCC-CIGRE Symposium on
High Voltage Overhead Lines and Underground Cables, Qatar, 1994, Paper
No. 12.
13. lEC Publication-587, “Standard Method for Evaluating Resistance to Track­
ing and Erosion of Electrical Insulating Materials used under Severe Ambient
Conditions” , lEC, Geneva, Switzerland, 1984.
14. ASTM Test Method D-2303, “Standard Test Method for Liquid Contaminant,
Inclined Plane Tracking and Erosion of Insulating Materials” , Philadelphia,
1984.
15. A.S.G. Alghamdi, D.W. Auckland, A.J. Risino and B.R. Varlow, IEEE Trans,
on Dielectrics and Elect. Insul., Vol. 3, No. 3, pp. 439-443, 1996.
16. ASTM Test Method D495, “Standard Test Method for High Voltage, Low-
Current, Dry Arc resistance of Solid Electrical Insulation” , 1973.
17. R. Bartnikas and R.M. Eichhorn (Eds.), E n g in e e r in g D i e le c t r i c s V o l 11A , E l e c ­
tr i c a l P r o p e r t ie s o f S o lid I n s u la tin g M a te r ia ls : M o l e c u la r S tr u c tu r e a n d E l e c ­
tr i c a l B e h a v io r , ASTM Publication, Philadelphia, 1983.
Composite Dielectrics 239

18. J.C. Bolts, Proc. 5th Elect. Insulation Conference, Chicago, pp. 202-209,
1963.
19. CIGRE Committee 22, Sub-Working Group 03-01, Electra, Paris, Prance, No.
130, 1989.
20 . H.M. Schneider and J.P. Hall, IEEE Trans, on Power Delivery, Vol. 4, No. 4,
pp. 2214-2221, 1989.
21 . E.A. Chemey, IEEE Elect. Insul. Magazine, Vol. 12, No. 3, pp. 7 -1 5 , 1996.
22 . J. S.T. Looms, I n s u la to r s f o r H ig h V o lta g e s, Peter Peregrinus Ltd., London,
England, 1988.
23. R. S. Gorur and T. Orbeck, IEEE Trans, on Elect. Insul., Vol. 26, No. 5, pp.
1064-1072, 1991.
24. S. H. Kim, E.A. Chemey and R. Hackam, IEEE Trans, on Power Delivery,
Vol. 5, No. 3, pp. 1491-1499, 1990.
25. A.C.M. Wilson, I n s u la tin g L iq u id s , Peter Peregrinus Ltd., London, England,
1980.
26. T. Tanaka, T. Okamoto, K. Nakanishi and T. Miyamoto, IEEE Trans, on Elect.
Insul., Vol. 28, No. 5, pp. 826-844, 1993.
27. C. R. Acker, Proceedings of lEEE-PES Winter Meeting and Tesla Symposium,
IEEE, New Jersey, pp. 25-30, 1976.
28. M. Ali, C. Eley, A.M. Emsley, R. Haywood and X. Xiao, IEEE Elect. Insul.
Magazine, Vol. 12, No. 3, pp. 28-34, 1996.
29. K. Giese, IEEE Elect. Insul. Magazine, Vol. 12, No. 1, pp. 29-33, 1996.
30. lEC-Publication 599, “Interpretation of the Analysis of Gases in Transformers
and Other Oil Pilled Electrical Equipment in Service” , lEC, Geneva, Swit­
zerland, 1978.
31. T. Tamara, H. Anetai, T. Ishii and T. Kawamura, Trans, of lEE Japan, Vol.
lOlA, No. 1, pp. 30-36, 1981.
32. H. Kan and T. Miyamoto, IEEE Elect. Insul. Magazine, Vol. 11, No. 6, pp.
15-21, 1995.
33. D. H. Shroff and A.W. Stannett, lEE Proceedings, Vol. 132, Part C, No. 6, pp.
312-319, 1985.
34. lEC-Draft Publication 1198, “Methods for the Determination of 2-Furfural
and Related Compounds” , lEC, Geneva, Switzerland, 1991.
35. J. Unsworth and F. Mitchell, IEEE Trans, on Elect. Insul., Vol. 25, No. 4, pp.
737-942, 1990.
36. M. Darveniza, T. Saba, D.J.T. Hill and T.T. Lee, Proc. 6th lEE Inti. Conf. on
Dielectrics Materials, Measurements and Applications, Manchester, England,
1992.
37. A.M. Emsley and G.C. Stevens, lEE Proc. Part A, Vol. 5, No. 141, pp.
324-333, 1994.
38. A.M. Emsley, Polymer Degradation Stability, No. 44, pp. 343-349, 1994.
39. M. Schaible, IEEE Elect. Insul. Magazine, Vol. 3, No. 1, pp. 8-12, 1987.
40. lEC Publication 626 Part 1; Specification for Combined Flexible Materials
for Electric Insulation, Definitions and General Requirements” , lEC, Geneva,
Switzerland, 1979.
240 Chapter 8

41. lEC Publication 626 Part 3: “Specification for Combined Flexible Materials
for Electric Insulation, Specifications for Individual Materials” , lEC, Geneva,
Switzerland, 1988.
42. lEC Publication 85: “Thermal Evaluation and Classification for Electric In­
sulation” , lEC, Geneva, Switzerland, 1984.
43. lEC Publication 450, “Measurement of the Average Viscometeric Degree of
Polymerization of New and Aged Electrical Papers” , lEC, Geneva, Switzer­
land, 1982.
44. A Bradwell (Ed.), E l e c t r i c a l I n s u la tio n , Peter Peregrinus Ltd., London, Eng­
land, 1983.
45. K. Giese, IEEE Elect. Insul. Magazine, Vol. 7, No. 4, pp. 27-30, 1991.
46. A. Sato, S, Kawakami, K. Endo and H. Dohi, CIGRE Conf., Paper 15-05,
Paris, France, pp. 1-5, 1985.
47. Y. Yoshida and T. Muraoka, IEEE Elect. Insul. Magazine, Vol. 11, No. 1, pp.
32-45, 1995.
48. J.W. Chang and R.S. Gorur, IEEE Trans, on Dielectric and Elect. Insul., Vol.
1, No. 6, pp. 1039-1045, 1994.
49. R.S. Gorur, J. Mishra, R. Tay and R. McAfee, IEEE Trans, on Dielectrics and
Elect. Insul., Vol. 3, No. 2, pp. 299-306, 1996.
High Voltage Cables

9.1 INTRODUCTION

Cables have been used in transmission and distribution networks since the
early days of the electrical power industry. Generally, long-distance power
transmission is carried out through overhead lines. However, transmission
and distribution in densely populated urban areas mostly uses underground
cables. Although significantly more expensive than the overhead lines, the
cables are preferred in urban areas due to safety, reliability and aesthetical
considerations. As a result of developments in insulating materials and
manufacturing techniques, high voltage cable technology has improved sig­
nificantly over the years. With a continuous increase in the overall length
of cable networks, questions regarding reliability, failure modes and di­
agnostics of such cables have assumed greater significance. This chapter
briefly discusses various aspects of high voltage power cables with em­
phasis on polymeric insulated cables, which are almost exclusively being
used in distribution networks in many countries. The insulation testing of
high voltage cables is discussed in Chapter 12.

9.2 CABLE MATERIALS

9.2.1 Conductors
Cables are constructed using a variety of materials for conductors, insu­
lation, screening and armoring. The most common conductor materials are

241
242 Chapter 9

copper or aluminum of high purity (>99.5% pure), since the resistivity of

such materials significantly increases with impurity content. Sometimes,
sodium is also used as the conductor material [5]. In recent years, the
possibility of using high-temperature superconducting materials for power
applications is also being examined. The choice between aluminum and
copper is normally based on resistivity (p), cost as well as mechanical and
manufacturing considerations. Table 9.1 compares resistivity values for a
few materials used in cables. For a given current rating, aluminum requires
a larger conductor cross-sectional area than copper. Both solid and stranded
conductors are used; the choice depends upon total cross-sectional area,
flexibility and manufacturing considerations. For larger cross sections,
stranded construction with alternate layers spiraled in opposite directions
is generally preferred.

9.2.2 Insulation
The cable insulation usually consists of (1) impregnated paper, (2) syn­
thetic polymers and (3) compressed gases. Early cables mostly used im­
pregnated paper insulation. Next stage was the introduction of oil-filled
paper insulation. At present, polymers are widely used. The polymers most
often used are polyvinyle chloride (PVC) for low-voltage cables; polyeth­
ylene (PE), cross-linked polyethylene (XLPE) and ethylene propylene
rubber (EPR) for medium-voltage cables; and XLPE and EPR for high-
voltage cables. In addition, high-density polyethylene (HDPE), high-
molecular-weight polyethylene (HMWPE), tree-resistant (or retardant)
XLPE (TRXLPE) and terpolymer ethylene propylene diene monomer
(EPDM) are also used for medium-voltage underground distribution (URD)
system cables.

Table 9.1 Resistivity of Some

Conductors at ~20°C

Material p (p.il-cm)

Copper 1.73
Aluminum 2.83
Sodium 4.68
Lead 21.5
Steel 10.20
High Voltage Cables 243

Sometimes compressed gases such as SF^, CCI2 F2 (Freon-12) and N2

are also used for cable insulation. Compressed-gas-insulated cables em­
ploying SF^ have already been described in Chapter 4. Other types of
compressed gas cables employ paper and compressed gas insulation [5].
The use of plastic insulated cables started in the early 1950s and now
these cables are the most common choice in URD systems. However, at
EHV and UHV levels, oil-filled paper-insulated (OF) cables are still ex­
tensively used. In recent years, XLPE insulated cables with mean AC work­
ing stresses approaching those used in OE cables (~15 kV/mm) have been
developed [1], and it appears that in the future such cables will become
more common at transmission voltages as well.

9.2.3 Screens and Jackets

In paper-insulated cables, conductor screens in the form of lapped metallic
(e.g., aluminum) foils or semiconductive carbon paper tapes are used to
relieve the stress concentrations which arise due to conductor stranding.
Moreover, a metallic sheath from a lead alloy or aluminum is also used
for mechanical protection of the insulation which also helps to confine the
electric stress within the insulation. Such a sheath also minimizes surface
discharges, limits electrostatic and electromagnetic interference and pre­
vents the ingress of moisture to the cable insulation. In polymeric insulated
cables, conductor and insulation are provided with semiconducting screens.
These screens have the following functions: (1) to equalize the electric
field around the conductor periphery, (2) to maintain a contact between the
conductors and the insulation throughout the expansion and contraction
caused by load cycling and (3) to prevent extraneous damage to the in­
sulation. Normally the two screens and the insulation are extruded si­
multaneously. The screens are formed from materials consisting of a
polyethylene-based resin mixed with conductive carbon black and small
amounts of antioxidants [1]. The level of impurities in the screens and their
smoothness have a very strong influence on the cable performance. Su­
persmooth, extra-clean conductor screens can increase the cable’s life [2].
The copper neutral wires or tapes that surround the insulation screen
are usually covered with a polyethylene jacket for mechanical protection
and to reduce the moisture intrusion into the cable insulation. It has been
found that the life of URD cables can be increased by using a jacket of
linear low-density polyethylene (LLDPE) [3]. Sometimes cables are jack­
eted with a semiconductive material. Such cables are finding increased
applications when commonly shared tunnels are used for power and com­
munication cables.
244 Chapter 9

9.2.4 Armors
Armoring is required for cables subjected to mechanical stress. In such
cases, a PVC sheath or a suitable bedding acts as a mechanical cushion
and chemical insulation between the metal sheath and the armor. Armoring
usually consists of steel tapes or steel wires. However, sometimes bronze
wires are used instead to minimize hysterisis and eddy current losses in
the armor. An outer layer of PVC sheath is normally provided to protect
the armor against corrosion. Moreover, such a sheath is also used to in­
scribe information about the cable.

9.3 TYPES OF CABLES

Cables can be classified based on several parameters, such as:

1. Voltage rating of the cable, e.g., low voltage, medium voltage, high
voltage and EHV cable.
2. Number of cores of the cable, e.g., single core, two core and three
core.
3. Insulation of the cable, e.g., XLPE, PVC, EPR and oil-paper cable.
4. Oil-pressure level in case of oil filled cables, e.g., self-contained
(OF) and pipe-type (POE) cable.
5. The presence or absence of the metallic shield over the cable in­
sulation, e.g., shielded and nonshielded cable.

Furthermore, in the case of multicore cables, each core may be individually

shielded thereby forming a coaxial cable or all three cores may have one
common shield as in the belted cables, as in Figure 9.1. Usually, belted
cables are restricted to voltage ratings of less than 33 kV. In general, shield­
ing should be considered for nonmetallic covered cables operating at circuit
voltages of above a few kV.
Generally, low and medium voltage URD cables are impregnated paper
or polymeric insulated, single or multicore, and coaxial or belted cables.
Figure 9.2 shows the cross section of a typical medium- and high-voltage
polymeric-insulated three-core cable. For three-core construction, three
cores (without armor and external jacket) are put together and enclosed
within a common underarmor jacket, an armor and an outer jacket, in
addition to necessary fillers, as shown in Figure 9.1b. Figure 9.3 shows
typical cross sections of single- and three-core OF cables. Figure 9.4 shows
typical cross-sectional details of a pipe type, oil-filled paper-insulated
(POF) cable.
High Voltage Cables 245

Conductor

Core insulation

Rller Insulation

Beit insulation

Common shield

(a )
Conductor

Core insulation

Filler insulation

Core shield

Binder

Common shield

Figure 9.1 Three-core cables: (a) belted, (b) screened or H-type.

9.4 CABLE CONSTANTS

The most important cable constants include resistance, capacitance, induc­

tance and insulation resistance which are briefly described next.

9.4.1 Conductor Resistance

The DC resistance of a conductor depends upon its resistivity, its length,
its cross-sectional area as well as its temperature. The resistance, in terms
of temperature, is given as:

(9.1)
R, T. + T.
where Rj = conductor resistance at temperature Tj in °C, R2 = conductor
resistance at temperature T2 in °C, and T^ = constant varying with con­
ductor material and is equal to 234.5, 241 and 228 for annealed copper,
hard-drawn copper, and hard-drawn aluminum, respectively.
The AC resistance is higher than the DC resistance due to skin effect.
The difference between the two values depends upon the frequency and
the conductor cross-sectional area [5]. Sometimes, if conductor area is
246 Chapter 9

Figure 9.2 Typical medium and high voltage three-core cable cross-sectional de­
tails: 1 = copper or aluminum conductor, 2 = inner semiconducting screen (con­
ductor screen), 3 = XLPE insulation, 4 = outer semiconducting screen (insulation
screen), 5 = copper tape, 6 = filler, 7 = mylar tape or swellable material bedding
layer, 8 = PVC bedding, 9 = steel wire armor, 10 = PVC or PE sheath.

large, segmented or hollow conductors are used to reduce the skin effect.
In oil-filled cables, hollow conductors are preferred as these contain oil for
better cooling and dielectric characteristics.

9.4.2 Cable Capacitance

Consider the single core coaxial cable of Figure 9.5. The capacitance of
this cable is given by:
I ttsqE^
C = F/m (9.2)
ln(b/a)
where = 8.854 X 10“ F/ m and is the relative permittivity of the
insulation. For three-core screened cables, each core will have a capaci­
tance given by equation (9.2). For three-core belted cables, which can be
modeled by the network of Figure 9.6, the capacitances between cores (C^)
High Voltage Cables 247

INSULATION
CONDUCTOR
LEAD SHIELD

REINFORCEMENT

OIL CHANNEL

EXTERN AL
PROTECTION

(a )
METALLIC
CONDUCTOR

EXTERNAL
PROTECTION

INSULATION

REINFORCEMENT
LEAD SHEATH

( b)
Figure 9.3 Cross-sectional views of typical oil-filled cables: (a) single core and
(b) three core.

Figure 9.4 O il-filled, paper-insulated, three-phase pipe-type cable.

248 Chapter 9

Figure 9.5 Cross section of a coaxial cable.

and from each core to the sheath (C,) can either be measured or determined
by numerical computation or approximate formulas [5-7].

9.4.3 Cable Inductance

For the coaxial cable of Figure 9.5 having a solid inner conductor and a
perfect metal screen, the total inductance L is given as:

L = 2 X 10- - + In - H/m (9.3)

4 \a
For belted cables, the term (b/a) in equation (9.3) is replaced with the term
S/a, where S is the spacing between two cores. For practical power cables,
the inductance also depends on screening and armoring materials as well
as the proximity of the cable to other conductors and ferrous objects, and

Figure 9.6 C apacitances o f a three-core belted cable.

High Voltage Cables 249

therefore its accurate calculation is difficult. However, it can be accurately

determined from measurements [4].

9.4.4 Insulation Resistance

If the cables of Figure 9.5 have bulk insulation resistivity of (in flm),
then its per meter insulation resistance Rj is given as:

R, - ^ l n ( b / a ) ilm (9.4)

Since Pi is usually very large for cable dielectrics (see Table 6.6), Rj
is hundreds of Milm for healthy high voltage cables. However, after a
partial or complete insulation breakdown, Rj may decrease substantially.

9.5 ELECTRIC STRESS IN CABLES

The electric stress E in the coaxial cable of Figure 9.5 varies only radially
and its value at a distance x from the cable center is given by:
V
E(x) (9.5)
xln(b/a)
where V is the applied voltage across the cable insulation. The insulation
next to the inner conductor surface is subjected to the maximum stress
having a value of V/(aln(b/a)). If the conductor surface is not perfectly
smooth, this value is increased due to the field intensification effect dis­
cussed in Chapter 1. Consequently, the maximum stress at a rough point
will become fV/(aln(b/a)) where f is the field intensification factor given
in Table 1.1 for some selected protrusion shapes. Such rough points can
cause insulation deterioration and act as potential tree sites (for details see
section 9.9) and significant efforts are being made by cable manufacturers
to use smooth and defect-free conductor and insulation screens in order to
reduce field intensifications.
It can be shown from equation (9.5) that for specified voltage and
conductor radius values, the conductor stress is minimum when b/a =
2.781, i.e., the insulation thickness is —1.72 times the conductor radius.
However, since this thickness depends upon the cable dielectric material
used and the cable’s voltage rating, the above condition for optimum stress
design cannot always be realized in practice.
For belted cables, the electric stress is not merely confined to cable
cores and its calculation requires use of simplifications or analysis using
250 Chapter 9

numerical methods. The charge simulation method can perform very ac­
curate field calculations for such cables [8-10].

9.6 CABLE LOSSES

Cable losses include losses in conductor, insulation, sheath, screens and

armors. Conductor losses (PRac losses) depend upon the rms current I and
effective AC resistance of the cable conductor Dielectric losses com­
prise of losses due to leakage through the cable insulation and losses
caused by dielectric polarization under AC stresses. The net dielectric
losses are given by equation (1.3) and depend upon cable voltage, its fre­
quency as well as the permittivity and loss tangent of the cable dielectric
material. Although for low and medium voltage power cables, dielectric
losses are insignificant, these assume importance for HV and EHV cables.
This is specially the case for oil paper cables. For instance, the dielec­
tric losses in a 400 kV OF cable may reach about 20% of the total cable
losses [4].
Generally, tan 6, which partially controls the dielectric losses, is sig­
nificantly higher for oil-paper insulation as compared to XLPF insulation.
For most of the dielectric materials used in cables, tan 8 depends upon
temperature, applied stress and supply frequency. For oil-paper insulation,
tan 8 is also strongly influenced by moisture content. Therefore, in high
voltage cables, a moisture level of less than 0.05% is desirable in order to
keep dielectric losses within acceptable limits [11]. Table 6.6 gives typical
8^, tan 8 and pj values for some dielectrics used in power cables. The
presence of voids and microcracks can also influence dielectric losses. Any
partial discharge in such voids increases the effective tan 8 value for the
insulation. Consequently, when the applied voltage is raised above the dis­
charge inception threshold, the dielectric losses exhibit a distinct increase.
Similarly, impurities in the cable insulation and screening materials can
also increase dielectric losses.
The AC current flowing along each cable conductor induces emf in
the metallic sheaths of the cable. Without grounding, such sheaths would
operate at a potential above the ground potential and can pose a safety
hazard. Furthermore, it will accelerate degradation of the jacket and other
materials, thereby affecting the cable’s life and reliability. When the sheaths
are bonded, circulating current flows in them causing power losses. How­
ever, for three-core cables such losses are negligible. In addition to circu­
lating currents, eddy currents are also induced in sheaths of both single
and multicore cables causing additional losses which usually are of small
magnitudes.
High Voltage Cables 251

Single-conductor cable sheaths may be open or short-circuited. When

the sheaths are short-circuited, they are usually bonded and grounded at
every manhole. This makes sheath voltage equal to zero, but allows the
flow of sheath currents. When the sheaths are open-circuited, no current
flows but sheaths can assume certain potential. In order to keep such a
potential within a safe value, i.e., 25 V in exposed positions, cable sheaths
are usually cross-bonded. Figure 9.7 shows a typical cross-bonding and
grounding arrangement to minimize sheath losses for single conductor ca­
bles. Alternate methods of operating sheaths in an open circuit include
grounding the sheath at one point only, or bonding sheaths through im­
pedances or transformers. The grounding conductor and its attachment to
the sheath must be designed such that it can safely carry fault and circu­
lating currents without overheating.
Losses also occur in armor and in any metallic pipe housing the cable.
For the armors made of magnetic materials, additional losses due to mag­
netic hysterisis can be of significant magnitudes. Thus, all such losses must
be considered in the design of a cable system.

9.7 CABLE AMPACITY

The ampacity or current carrying capacity of a cable is defined as the

maximum current which the cable can carry continuously without the tem­
perature at any point in the insulation exceeding the limits specified for
the respectively material. The ampacity depends upon the rate of heat gen­
eration within the cable as well as the rate of heat dissipation from the
cable to the surroundings. The rate of heat generation within a cable de­
pends upon various losses (discussed in section 9.6), whereas the rate of
heat dissipation depends upon the thermal resistances of different cable

r 1

Figure 9.7 C ross bonding o f single-conductor cable sheaths.

252 Chapter 9

materials and the media surrounding the cable. Hence, the cable ampacity
is influenced by numerous factors, such as the materials and dimensions
of the cable conductor, insulation, shields and armors; the method of cable
installation; the thermal characteristics of media surrounding the cable; the
characteristics of other heat sources/sinks in the vicinity of the cable (e.g.,
other cables, hot/cold water and steam pipes); and general environmental
conditions (e.g., solar radiation level, ambient temperatures) in the area
where the cable is installed.
Neher and McGrath’s method of calculations [12] forms the basis of
ampacity related Association of Edison Illumination Companies (AEIC)
specifications as well as IEC-287 [13]. Eor practical cables these calcula­
tions are fairly complex and are normally carried out using software pack­
ages [14] and numerical techniques such as the finite element method [15].
The calculations are based on the principle that the heat generated within
a cable should be dissipated to the ambient without exceeding the maxi­
mum allowable conductor temperature which depends upon cable’s di­
electric material. PE and XLPE insulated cables have ~ 75°C and 90°C
under normal conditions and 90"’C and 130°C, respectively, under emer­
gency operating conditions.
For a coaxial cable buried in a homogeneous earth, the thermal circuit
of Figure 9.8 can be used to evaluate the cable ampacity. In Figure 9.8,
W^, Wj, W, and are, respectively, the conductor, the dielectric, the
shield and the armor losses (W/m), whereas R^, Rb, Rs and R^ are the
thermal resistances of the cable dielectric, the bedding, the serving and the
earth surrounding the cable, respectively, in m°C/W. T^, and are the
conductor, the sheath and the ambient earth temperatures, respectively. The
conductor losses, are calculated at the maximum operating
temperature. W, and are expressed in terms of by factors Aj and
À2 such that = AjW^ and - The cable ampacity is found by
applying the thermal form of Ohm's law to the circuit of Figure 9.8. Con­
sequently, we can write:

0.5 0.5 R^ R. R. R.

’c "’ s ”a

Figure 9.8 Thermal equivalent circuit of a single coaxial cable buried in earth.
High Voltage Cables 253

T, - T, = PR,,[R, + (1 + A,)R,
+ (1 + Aj + A2)(Rs + Re)] + ATj (9.6)
where AT^ is the cable conductor temperature rise caused by the dielectric
losses and is given as:

Rd
AT, = W, ^ + R, + R, + R, (9.7)

Thus, ignoring the skin and proximity effects, the ampacity of a coaxial
cable is given by:
T - T AT,
I = (9.8)
Rac(Rd + (1 + >^l)Rb + (1 + Al + A2)(R^ + Re))
For a coaxial cable buried in a uniform earth medium, the thermal resis­
tances can be easily evaluated [12]. However, the evaluation of such resis­
tances for belted cables and for nonhomogeneous multilayer earths are
more difficult, and thus the subject of several reports [12-16].

9.8 PARTIAL DISCHARGES IN CABLES

Gas-filled cavities or voids are formed in the insulation or at the screens/

insulation interfaces during manufacture, installation or operation. For ex­
ample, voids may be formed in polymeric cables during the extrusion pro­
cess. In paper-insulated cables, such voids may form during the impreg­
nation cycle. Voids may also form in cables by the differential expansion
and contraction of cable materials due to cyclic loading or short-circuit
conditions.
Generally, voids have a higher electric stress as compared to the bulk
of insulation. However, the gas inside a void usually has a lower breakdown
strength as compared to the main insulation. When the electric stress in
the void exceeds the breakdown strength of gas within the void, partial
discharges (PDs) may occur. Such discharges, if sufficiently intense, will
gradually degrade and erode the insulation, eventually leading to its break­
down. Therefore, the inception voltage for the onset of PDs, their magni­
tudes, their dependence on voltage, and methods of measurement and lo­
cation of such discharges are of significant importance. Consequently, all
solid dielectric high voltage power cables are tested for PDs.
For the coaxial cable of Figure 9.5, it can be shown that the theoretical
discharge inception voltage (DIV) depends upon the radius of conductor;
thickness and permittivity of the insulation; shape, size and location of the
254 Chapter 9

void; as well as the gas pressure inside the void [17,18]. Generally, the
minimum theoretical DIV can be expressed as [17]:
325
= — a ln(b/a) (9.9)

where is in volts, C (the cavity size is in meters) and the factor K =

E^/Ej depends upon cavity shape and of the insulation [see equations
(1.14) and (1.15)]. In practice, cavity sizes are usually small and PD pulses
may not be detectable in a reasonable amount of time due to a lack of
initiatory electrons when the applied voltage is equal to V^in- Conse­
quently, a somewhat higher applied voltage, a longer duration of applied
stress or some external source of irradiation may be necessary to detect
the PD initiation in small-size cavities.
When a PD occurs inside a cavity, the actual PD pulse thus generated
has a duration of only a few nanoseconds. However, as the pulse propagates
away from the source, it suffers frequency-dependent attenuations and,
therefore, the signal detected outside the test object depends upon the na­
ture of connection between the point of pulse generation and the external
circuit. Moreover, for a given void size, the detected PD magnitude gen­
erally varies inversely with cable voltage rating (or cable size). Thus, the
measurement sensitivity is the lowest for PD testing of the highest voltage
cable where the room for measurement error is the minimum [19]. Standard
methods exist for measurements and interpretation of PDs in power cables
[21-23]. In addition to measurements of PD magnitudes, PD source lo­
cation is also possible [24]. However, the accuracy of PD source location
in power cables is limited by the high frequency attenuation of PD pulses
as they propagate through the cable [20].
PDs cause cable insulation degradation through a variety of electronic,
chemical and mechanical processes which have been the subjects of sig­
nificant research in recent years. In many cases, PDs lead to initiation and
growth of trees which ultimately cause failure of the cable. In addition to
cavities, other defects which may ultimately lead to PD and tree initiation
include protrusions and stress enhancements at sheaths, insulating or con­
ducting contaminants in the insulation, microcracks and other defects. It is
important to note that a PD test has limitations when used as a diagnostic
tool for detection of electrical trees and some other defects [25]. Never­
theless, techniques to measure PD magnitudes of energized XLPE cable
operating at EHV levels have been reported [26].
High Voltage Cables 255

9.9 TREEING IN CABLES

Treeing is an electrical prebreakdown phenomenon. This name is given to

any type of damage which progresses through a stressed dielectric so that,
if visible, its path resembles the form of a tree. Tree-like discharge patterns,
sometimes leading to total breakdown of the insulation, have been observed
for many years in oil-impregnated pressboard and in oil-impregnated paper
insulated cables. Treeing can occur in most solid dielectrics including glass
and porcelain but it is a serious problem in polymers, rubbers and epoxy
resins, etc. However, since rubbers and resins are often pigmented or min­
eral filled, the existence of tree-like channels may go unnoticed in such
materials.
Treeing may or may not be followed by complete electrical breakdown
of the insulation; but in organic extruded dielectrics, it is the most likely
mechanism of dielectric failure which is the result of a lengthy aging pro­
cess [27,28]. Electric stress and stress concentration are always required
for the initiation and growth of trees. Treeing can progress rapidly under
high electric stresses in dry dielectrics by periodic partial discharges or
more slowly in the presence of moisture at lower electric stresses without
any detectable PD. Treeing can occur under DC, AC and impulse voltages.
The trees can be considered in two broad classifications: electrical trees
and water trees. All trees are initiated at sites of high and divergent electric
stresses and their growth rate may be aggravated by the presence of mois­
ture, chemicals, contaminants and other defects in the dielectric as dis­
cussed next.

9.9.1 Electrical Trees

Electrical trees initiate and propagate due to high and divergent electric
stress at metallic or semiconducting contaminants and/or voids, etc., by
partial discharges occurring in a dry dielectric [28]. Such trees consist of
hollow channels resulting from decomposition of dielectric material by the
PDs. The tree shows up clearly in PE and other translucent solid dielectrics
when examined with an optical microscope and transmitted light. Electrical
tree channels are permanently visible and there is a great variety of the
visual appearance of stems and branches of such trees as well as the cir­
cumstances in which initiation and growth of such trees occur. Eigure 9.9
shows the typical appearance of some electrical trees [29]. Many names
such as dendrites, branch type, bush type, spikes, strings, bow-ties and
vented trees have been used in the literature to describe such trees. Trees
which start to grow from within the insulation and progress symmetrically
outwards from the electrodes are called bow-tie trees because of their ap-
256 Chapter 9

(a) (b)

Figure 9,9 Typical patterns of electrical trees in PE: (a) bush-like and (b) tree­
line patterns. (From Ref. 29 © IEEE, 1978.)

pearance. On the other hand, trees which initiate at an electrode (or sem-
iconductive screen) insulation interface and progress towards the opposite
electrode are called vented trees.
Access to free air is an important factor in the growth of a vented tree.
Such trees are capable of growing continuously and long enough to bridge
the electrodes or cause a dielectric failure. Bow-tie or nonvented trees do
not have a free supply of air to support continuous PDs. Therefore, the
growth of such trees is intermittent and discharge occurs with longer pe­
riods of extinction, which is believed to be due to an increased void pres­
sure resulting from ionization. During the extinction period, gas pressure
in the tree channel is reduced by diffusion and conditions become favorable
for occurrence of another PD causing further growth of the tree. Usually
vented trees do not grow long enough to bridge the entire insulation thick­
ness or cause a failure [27,28].
There are two distinct periods in electrical treeing. The first is an in­
cubation period during which no measurable PD can be detected, but at
the end of which a tree-like figure is first observed. The second is a prop­
agation period during which a tree-like figure grows in the insulation and
significant PD magnitude can be measured. Usually, the PD level fluctuates
over a wide range with the growth of an electrical tree, as shown in Figure
9.10 [30]. The incubation period depends upon the stress level and its
distribution at the initiation site, the composition and properties of the
dielectric and the environmental conditions. Generally, at low stress levels,
cumulative processes are proceeding and eventually foster conditions
High Voltage Cables 257

rim e
(a)

(b) Tim e

Figure 9.10 Typical evolution of PD level and electrical tree length. (From Ref.
25 © IEEE, 1992.)

which initiate treeing. The mechanisms which have been proposed to ex­
plain the initiation of electrical treeing include electron injection and ex­
traction at the tree site, mechanical fatigue cracking due to periodic Max­
well stresses, heating, thermal decomposition and PDs in voids [27,28].
The mechanical effects including the buildup of strain, fatigue failure and
fracture due to shock waves involved are believed to play an important
role in the initiation and growth of electrical trees [30].
Aging of low density PE under uniform and divergent fields has shown
that electrical trees do not develop at fields of up to 20 kV/mm. Local
258 Chapter 9

fields of —100 kV/mm are required for tree initiation. Such fields can
occur at cable working voltages due to various stress enhancement mech­
anisms, as discussed earlier in Chapter 1. For example, a conductive pro­
trusion with a 10 to 1 ratio of axis will have a local field at its tip which
is about 50 times the macroscopic average field. The stress at which trees
initiate in a given polymer depends upon the waveform, frequency, mag­
nitude and time of test voltage, and on whether the voltage is applied
continuously or is interrupted periodically.
After initiation the tree growth proceeds by a series of sporadic bursts
of activity. Consequently, tree branching becomes more frequent and the
rate of tree growth slows down. Two of the most important factors that
influence tree propagation rate are the development of internal gas pressure
due to PDs and the shielding effect of adjacent tree branches on the electric
field. Tree channels are generally hollow where conducting carbon particles
may also be found. The channel diameter can vary from a few up to several
tens of microns.

9.9.2 Water Trees

The tree-like figures which appear in water-exposed polymer-insulated
stressed cables are named water trees. Water treeing occurs in the presence
of moisture. As compared to electrical treeing, water treeing usually starts
at lower electric stress values and progresses more slowly without any
detectable PDs.
Water (or wet) trees are different from electrical (or dry) trees. Con­
trary to electrical trees, water trees do not exhibit measurable PD levels of
>0.1 pC. The propagation time of water trees is measured in years,
whereas once initiated an electrical tree can very quickly propagate through
the insulation, e.g., under the influence of a surge overvoltage. The ap­
pearance of the two types is usually different from each other as water
trees do not exhibit much branching. However, sometimes the two types
are difficult to distinguish. A practical method for differentiating between
the two types is to examine them after drying the insulation. Unstained
water trees become invisible after the insulation is dried while electrical
trees are clearly visible even under dry conditions. However, it is possible
to make water trees permanently visible by the use of methylene blue with
a base or acidic rhodamine dye [27,28]. Unlike electrical trees, water trees
do not consist of permanent hollow channels. Instead, water tree channels
usually consist of fine filamentary paths between small cavities through
which moisture penetrates under the action of a voltage gradient.
Similar to electrical trees, there are two basic types of water trees,
namely, bow-tie trees and vented trees (Figure 9.11) [31]. Vented trees are
High Voltage Cables 259

■■ lll llll

(a)

III
Figure 9.11 (a) Bow-tie water tree (total length = 200 fjon) initiated from one of
the impurities in the insulation, and (b) vented water tree bent near the insulation
outer surface. (From Ref. 31 © IEEE, 1990.)
260 Chapter 9

initiated at the insulation surfaces whereas bow-tie trees are initiated in the
insulation volume. Both types have different growth behavior and pose
different levels of danger. Both types grow from points having high electric
stress values which are also moisture or moisture vapor sources. The mois­
ture source may consist of condensed water or water vapor of —65-70%
relative humidity. The water vapor may become available from external
sources or may be contained in the dielectric during the cable manufacture.
The concentration of vented trees is often low compared to that of
bow-tie trees, and at the beginning of their growth, the propagation rate of
vented trees is normally lower than that of bow-tie trees. However, at a
later stage, the opposite may be true since the growth of bow-tie trees is
strongly reduced after a certain time and, therefore, their total length is
restricted. Consequently, a bow-tie water tree is seldom the origin of cable
breakdown. However, vented water trees usually have access to water and
are capable of growing long enough to bridge the dielectric. Alternatively,
such trees may grow long enough to reduce the effective insulation thick­
ness below that required to support the electric stress, after which failure
may occur by electrical treeing. The growth of electrical trees near or at
the tip of water trees have often been observed [27-30].
Water trees contain water. If this water is evaporated, e.g., by heating,
the tree channel becomes invisible. Usually the tree absorbs water again if
the insulation is exposed to water or water vapor afterwards. Near the
initiation spot, a vented tree column can contain up to 10% water. At a
certain distance from the tree site, the water content may be up to 1-2%
[31]. Though water trees weaken the dielectric, these do not totally damage
the insulation and the tree channels exhibit properties of a poor dielectric
material. There is a clear relation between the size of the water trees and
the electric breakdown strength, as shown in Figure 9.12 [31]. It has been
observed that water trees crossing the entire dielectric section do not cause
immediate breakdown under service conditions. Such dielectrics often still
have a breakdown strength above the service stress level of ~2 kV/mm,
as evident from Figure 9.12. Thus, breakdown in cables containing even
large water trees may be initiated by some kind of transient surge or tem­
porary overvoltage. In order to determine the level of degradation of aged
cables by water treeing, a characterization test has been proposed [32].
There is a hypothesis that undetectable PDs of <0.02 pC magnitude
accompanied by light emission do occur during water treeing. Recently, it
has been shown that long vented water trees could generate measurable
PDs when excited with an AC voltage of moderate magnitude [33]. The
tree propagation rate increases with applied voltage and is also influenced
by supply frequency, ambient temperature, mechanical stress, water con­
ductivity and the nature of salts and chemicals present in the water.
High Voltage Cables 261

Figure 9.12 Relation between the mean breakdown stress level and the water tree
size. (From Ref. 31 © IEEE, 1990.)

Though a definite mechanism of inception and growth of water treeing

is not agreed upon, it appears that the capillary action, osmosis. Coulomb’s
forces, dielectrophoresis, thermal degradation, partial discharges and chem­
ical degradation all play some part in water treeing. Various proposed
mechanisms of water treeing are summarized in the literature [27,28,
32,34].

9.9.3 Electrochemical Trees

Electrochemical trees are the water trees which are stained permanently
during their growth period due to the presence of certain minerals or ions
in the water. Such ions penetrate the dielectric material under an applied
stress. Examples of electrochemical trees include sulfide trees caused by
H2 S and green, blue and brown trees caused by sulfur, iron, copper or
aluminum ions [27,28]. Such trees are permanently visible, like electrical
trees, and can have a color depending upon the chemistry of water soluble
ions and the cable dielectric material.

9.9.4 Techniques to Suppress Trees

Although there is no definite theory to explain all aspects of treeing in
dielectrics, it is generally accepted that the trees start from regions of high
and divergent electric stresses. Such regions are usually caused by defects
262 Chapter 9

such as loose and/or rough semiconductive screens, cavities, defects and

inclusions of foreign particles in the insulation. Therefore, the first step to
suppress trees is to minimize such defects by employing suitable materials
and manufacturing techniques. In recent years, smoother semiconducting
screens with improved bonding to the insulation have been developed. In
addition, the following four principles may be used to reduce treeing [27]:
Principle I: to fill voids with some suitable material
Principle II: to coat the internal surfaces of voids with semicon­
ducting compounds
Principle III: to relax locally intensified electric stresses
Principle IV: to trap or decelerate high energy electrons
The treeing problems in oil-paper cables were overcome by careful drying
and degassing of both the paper and the oil and by the development of OF
and POF cables (principle I). In this case, the original gas in the void is
replaced by a gas, a liquid or even a solid of either higher dielectric
strength and/or higher permittivity in order to increase the discharge in­
ception voltage and to reduce the PD magnitude. This is the basic philos­
ophy of principle I.
Lowering the surface resistivity of a void by suitable additives can
suppress internal discharges because, in this case, the void is virtually
shorted out. Various additives having the required characteristics are avail­
able and used. When mixed with the dielectric, they bleed out to the surface
of the insulation, including the void surfaces. There are two groups of
additives; one has low resistivity by nature, while the other exhibits low
resistivity only after being subjected to partial discharges [27].
Field grading with additives or voltage stabilizers is applicable to prin­
ciple III. Voltage stabilizers act to soften sharp electrode profiles and reduce
electric stress around them. The addition of such semiconducting materials
increases tan 6, which in some cases may cause thermal runaway of the
insulation and therefore can be harmful. One example of this principle is
the replacement of the conductive tapes by extruded layers of semicon­
ductive materials.
Principle IV is to add various organic materials to the dielectric which
can quickly absorb injected electrons before they could react with the in­
sulation. Cable manufactures have patented different additives in this cat­
egory [27].
Cable performance against water treeing could be improved by reduc­
ing the tree initiation sites. In addition, water-tight construction and suitable
curing methods are also useful. Moreover, metallic or nonmetallic water
impervious sheath and radial and/or longitudinal barriers can also be used
to minimize the availability of water for rapid tree growth. The use of
High Voltage Cables 263

laminate sheaths as moisture barriers is also being explored [35]. Further­

more, tree retardant additives may be introduced in the cable dielectric.
References [27,28,31] provide more details about the role of such additives
in suppressing water tree growth in cables. Very low density polyethylene
(VLDPE) has shown some resistance to water treeing due to its crystallinity
and the additives used [36].

9.10 CABLE AGING AND LIFE ESTIMATION

All cables are subjected to simultaneous electrical, thermal and mechanical

stresses due to high voltages and high currents. In addition to aging of the
cable dielectric under the above mentioned stresses, there are normal chem­
ical changes occurring in the dielectric material as well. Since it is impor­
tant to estimate the lifetime of a given dielectric from accelerated life tests,
the usual voltage, frequency and temperature aging tests as applied to
power cables are briefly outlined here.

9.10.1 Voltage Aging

Partial discharges and treeing can reduce the cable life appreciably. The
material degradation by PDs is strongly influenced by two important fac­
tors: applied voltage and frequency. In the presence of cavities and PDs,
the lifetime (t) of a dielectric is strongly affected by the applied voltage V
and follows a relationship of the form:
V"t = D. (9.10)
where is a constant and the voltage life factor, n, depends upon the
dielectric material, method of cable manufacture, cable size and type of
applied voltage. For cables, n is between 5 and 25 [1,27]. For medium
voltage XLPE cables, n = 9 is considered suitable. Eigure 9.13 shows a
voltage life diagram of unscreened, three-phase, belted-type XLPE and
PVC cables [37]. Equation (9.10) is used to predict cable life under voltage
stress alone and it does not take into account other stresses which may
result under actual service conditions.

9.10.2 Frequency Aging

When the applied voltage is held constant but supply frequency is in­
creased, the insulation degradation due to partial discharges increases. Con­
sequently, the time to failure is related to the frequency F by:
264 Chapter 9

Figure 9.13 Voltage-life curve of unscreened three-phase belted, XLPE and PVC
cables with PDs. (From Ref. 37 © IEEE, 1989.)

F"^t = Df (9.11)

where is a constant and the frequency life factor, m, is ~ 1 for surface

discharges and for electrical tree inception. For void discharges and for
tree propagation, m < 1 [27]. Combining both voltage and frequency ac­
celerated aging, the dielectric life time can be expressed as:

t = Df, F "^ V (9.12)

where is the constant for simultaneous voltage and frequency aging.

This equation can be used for life estimation based on accelerated life
testing of cable samples and materials.
High Voltage Cables 265

9-10-3 Thermal Aging

At elevated temperatures, insulation deteriorates more rapidly leading to a
reduction in its useful life period. Since the cables operate at elevated
temperatures, it is important to consider the effects of thermal stresses on
cable aging as well. Dakin [38] proposed a chemical rate theory model for
thermal aging and suggested that the internal chemical changes in a die­
lectric material depend upon the temperature. Such changes influence the
dielectric properties and hence the expected life of the material. Conse­
quently, the life is related to temperature by:
t = G exp (H/T) (9.13)
where G and H are constants determined by the activation energy of the
reaction, which influences the dielectric behavior, and T is temperature in
°K. Thus, if the logarithm of life is plotted against 1/T, a straight line is
usually obtained. This is known as the Arrhenius relation and is used to
estimate the thermal aging behavior of cables. Generally, if temperature is
increased by 8-10°K, the life of a given cable insulation is reduced by
about one-half. In order to use equation (9.13), some suitable parameter
such as tensile strength, breakdown strength or tan 8, must be identified
along with suitable end point values (e.g., life is reached when the dielec­
tric strength is reduced to one-half of the original value) which are mon­
itored at different temperatures to estimate the life.

9-10-4 Multifactor Stress Aging

Cables are usually subjected to electrical, thermal, mechanical, radiation,
environmental and chemical stress aging simultaneously. With one aging
stress at a time, the lifetime results are quite different from those obtained
when several aging parameters are applied simultaneously. Therefore, con­
siderable research effort is being exerted in understanding multifactor stress
aging [39]. However, so far, no general quantitative model capable of pre­
dicting the aging behavior and life expectancy under multifactor stress
aging of insulation exists. At present most of the accelerated aging studies
concentrate on simultaneous voltage, temperature and environmental (water
and chemicals) effects. Figure 9.14 shows a voltage-life curve for XLPE
and PE cables with accelerated aging tests along with cables removed from
service [40]. This figure shows that due to water treeing, the life of a cable
is shortened considerably. However, without treeing, the cable’s life points
lie on the characteristic aging curve of the respective material as given by
equation (9.10).
266 Chapter 9

Figure 9.14 E-t curve of XLPE cable and cables with water trees. (From Ref. 40
© IEEE, 1976.)

9.11 CABLE ACCESSORIES

Limitations of manufacture, transportation and installation require cables

to be jointed in the field when the total length exceeds a certain value.
Moreover, terminations are required where the cables are connected to
overhead lines or electrical apparatus. Thus, joints and terminations are an
integral part of any cable system and should possess the same integrity as
their associated cables. Various types of joints and terminations can broadly
be classified as shown in Table 9.2 along with their potential applications
[27].

9.11.1 Joints
Straight joints splice two cables of the same kind in a straight line. Besides
the normal joints commonly used, there are other types such as insulated
joints, stop joints, semistop joints and gas-stop joints. Insulated joints are
constructed such that the conductors are joined while the sheaths are in­
sulated from each other and are used for cross-bonding purposes (Figure
9.7). Beyond a certain optimum length, OF cables are sectionalized for the
purpose of oil feeding and maintenance and stop joints are used in such
cases. These connect cables electrically but block oil flow. In gas-filled
cables, gas-stop joints perform a similar function. Semistop joints are used
to permit or stop the flow of oil in POF cables. Heterojoints connect two
High Voltage Cables 267

Table 9.2 Classification of Splices and Terminations

Joints Comments

A. Straight Joints
1. Normal joints Taped and extruded cables.
2. Insulated joints Shields insulated for cross bonding.
3. Stop joints To stop oil flow in OF cables.
4. Semistop joints To stop oil flow in POF cables.
5. Gas-stop joints To stop gas flow in GF cables.
B. Terminal joints
1. Termination in air Used for taped and extruded cables. For taped
cables, porcelain bushing is used. For extruded
cables, porcelain, epoxy resin or elastomer mold
bushings are used.
2. Termination in oil Porcelain bushing are used.
3. Termination in SF^ Epoxy-resin bushing are used.
4. Direct connection With circuit breakers, transformers, etc.
C. Heterojoints Used to connect OF-XLPE, OE-POF, cables, etc.
D. Branch joints Used for making Y (or T branch) and X (or +
branch).

Source: Ref 27. Reprinted with permission.

different types of cable systems whereas branch joints are used when a
cable is to be connected to multiple cable systems.
The basic concept of jointing of a single conductor cable is outlined
in Figure 9.15. Here the joint is made up of (a) a conductor compression
sleeve which joins together the conductors of the two cable ends, (b) a
tapering down of the insulation on each cable, (c) the joint insulation which
is applied over the conductor sleeve and (d) the stress of relief cone.
The tapering of the insulation is called stepping for taped cables and
penciling for extruded cables. Stepping consists of a set of steps having
risers and treads from the level of the conductor surface to that of the cable
insulation surface. Penciling, however, gives a smooth surface. The insu­
lation applied over the conductor sleeve should be well blended into the
cable insulation so as to make the overall cable insulation as homogeneous
as possible. Generally a tangential component of electric stress is intro­
duced at the end of conductor sleeve in the tapered insulation. Usually the
joint insulation is covered with a suitable protective layer. Since the joint
insulation is normally built up to some diameter greater than that of the
cable insulation, thermal discontinuities may arise at joints because the
268 Chapter 9

1. Conductors 5. Sheath
2. Connector 6. Enclosure
3. Penciling of insulation 7. Joint insulation compound
4. Core insulation 8. Stress control electrode

Figure 9.15 Basic components of a cable joint.

thermal resistance tends to be higher at the joints than in any other part of
the cable. In addition to these simple elements, a Joint should also have
conductor and insulation shields and a sheath or a protective covering. The
arrangement employed depends upon the type of the joint as well as the
type of the cable. Due to the availability of premolded, heat-shrinkable
components, the jointing of extruded cables is much simpler than that of
the taped cables. Table 9.3 summarizes the types of splices available for
extruded dielectric cables [27]. The detailed design principles and perform­
ance requirements of joints are discussed by Tanaka and Greenwood [27].

Table 9.3 Types of Extruded Cable Splices

Types Insulating materials

Tape wrap splices Self-adhesive tapes, pressure sensitive tapes or heat

shrinkable tapes are used with silicone oil, etc.
Field molded splices Involves taping, heating and cross-linking using EPR, PE
vulcanizable tape or injection mold (PE).
Prefabricated splices EPR, silicone rubber, epoxy resin and insulating oil are
normally used in prefabricated splices.
Semiprefab splices Uses EPR mold or tape and partial replacement of tape
with mold.

S ou rce: Ref. 27. Reprinted with permission.

High Voltage Cables 269

9.11.2 Terminations
Terminations are required where cables are connected to overhead lines or
other electrical equipment that may be air, oil or SF^ insulated requiring
the use of different terminations, as mentioned in Table 9.2. In some cases,
direct cable connection to the apparatus is used instead. Figure 9.16 shows
the basic structure of a single-core cable termination. It consists of a con­
ductor lead-out rod, an insulation reinforcing layer with a stress relief cone,
and a casing or a bushing. The end of taped cables is usually encapsulated
with a porcelain bushing which acts as an external insulation. The space
between cable core and the inner surface of the bushing may be filled with
oil, SFg or other suitable insulation compound. The external insulation may
be made of plastic tape, a porcelain bushing, an epoxy resin bushing or a
rubber molded bushing. The end of the termination is often open to air for
connection to an overhead line, but sometimes it is connected to an ap­
paratus in oil or SF^. Terminations in oil or SF^ usually need less space.
For SF^ applications, epoxy resin bushings are preferred.

Figure 9.16 Basic components of a cable termination: 1 = conductor lead-out

rod, 2 = penciling of insulation, 3 = cable conductor, 4 = insulating compound,
5 = insulation reinforcing layer, 6 = stress relief cone, 7 = bushing, 8 = cable
sheath.
270 Chapter 9

The electric field in a coaxial cable is purely radial and the axial stress
is zero. At a joint or termination, the stress distribution is no longer com­
pletely radial as joints cannot, in general, be made without introducing an
increase in diameters of both the conductor and the insulation shield. At
these transitions, an axial or longitudinal component of stress is introduced.
Similarly, at a termination, a longitudinal stress component is introduced
between the high voltage terminal and the cable shield. This longitudinal
stress is controlled by the accessory geometry, i.e., the profile of the stress
relief cone and the pending or stepping of the joint insulation. Normally
stress relief cones are used to reduce stress concentrations that occur at the
sheath’s edge. For higher voltage ratings, even such a cone is not sufficient
and a capacitively graded termination is usually preferred [27]. For such
applications, capacitor bushings made of cylindrical laminates of metal foil
and insulating paper are designed to get a uniform stress distribution.
Problems can arise at joints and terminations due to voids and due to
the use of inhomogeneous dielectrics. Other problems such as corona deg­
radation, surface flashover and tracking are caused by higher longitudinal
and sheath edge stresses. Any such problem can lead to a premature failure
of the respective cable splice. In addition, when a termination is made up
of different materials, the different rates of aging, expansion/contraction
or other inhomogeneous properties can also cause premature failure.

9.12 CABLE FAULT LOCATION

All types of cables are subjected to electrical faults. It is important to locate

the point of such a fault as soon as possible. OF, POF, and gas-filled (GF)
cables can also experience other types of faults which may be followed by
electrical faults. Oil leaks in case of OF and POF cables and gas leaks in
case of GF cables are examples of such faults. A brief summary of different
fault location techniques is given in this section.

9.12.1 Detection of Oil Leaks

An oil-leak point can be located by visual inspection, by oil-flow behavior
or by the frozen cable method. It usually involves judging the oil flow
direction from axial variation of temperature on the pipe when part of the
pipe is heated. When the oil leak is caused by external forces, the protective
covering of the cable is damaged and it may also be earthed. In this event,
it may be possible to utilize a ground point detection technique. Details of
various methods for detection and location of oil leaks are reported by
Tanaka and Greenwood [27].
High Voltage Cables 271

9.12.2 Detection of Gas Leaks

Gas leak detection methods include (1) the bubble observation method,
which is usually suitable for joint boxes and other cable accessories, and
(2) gas flow, tracer gas and acoustic detection methods, which are suitable
for main cables. The gas flow method estimates the position of a gas leak
by measuring the pressure gradient caused by the leak. In the tracer gas
method, a certain amount of tracer gas (e.g., a halogen gas) is injected in
the cable and conditions are monitored along the cable route to detect the
leak. Acoustic methods detect the leak through acoustic waves which are
generated at the point of a gas leak [27].

9.12.3 Fault Location

Electrical faults in multiconductor cables may be classified as:
1. High- or low-resistance earth faults involving one or more
conductors
2. Open-circuit faults
3. External flashover or self-healing faults
The cable fault location is accomplished in three basic steps. These are:
(1) recognition of a fault condition, (2) estimation of the point of fault and
(3) confirmation of the point of fault. Eault presence is indicated by the
operation of relays. The faulty conductor and fault type are identified by
simple tests at both cable ends. Such tests include measuring the conductor
resistance and measuring the insulation resistances between the cable cores
and between the cores and the ground. From such measurements, the type
of fault, the faulted phase and the approximate fault zone are identified.
Terminal measurement methods are used to measure the distance D from
either end of the cable and then locate the fault more accurately. The
distance D can be measured by either a pulse method or a bridge method.
A DC bridge is suitable for low-resistance faults, while AC bridges are
more appropriate for open-circuit faults such as broken conductors. Murray
loop method is useful to find D for low-resistance single line to ground
fault. However, it is not suitable for open conductor faults or for three-
phase faults. If the fault has a high resistance to ground, usually a high
voltage is first applied to “bum out” the fault and reduce the fault resis­
tance. A Murray-Fischer bridge is useful to locate a fault when no healthy
cable is available [27].
Reflection or pulse methods, which can be used for precise fault lo­
cation for a variety of cable faults [27], are based on the behavior of
traveling waves. For an open or a short circuit, cable impedance changes
272 Chapter 9

are very sudden and traveling waves are reflected back with or without a
complete change in the incident pulse polarity. In pulse methods, a pulse
is sent from one cable end and its reflection is monitored. The polarity of
the reflected pulse, the time interval between the incident and the reflected
pulse, and the velocity of wave propagation are used to estimate the fault
location. Moreover, if a train of high-frequency pulses is sent along the
cable and the resulting magnetic field is detected, there would be a de­
tectable signal all along the cable route up to the point of the fault, beyond
which it would cease. A high-resolution radar method to locate cable faults
as well as underground splices and feed-through devices with good accu­
racy has been described by Banker et al. [41].

9.13 RECENT ADVANCES IN CABLE TECHNOLOGY

Present OF and POF cables are the outcome of fairly mature and proven
technology. Recent years have seen increasing interest in the development
of paper-polypropylene-paper insulation (PPP) for cable applications. Con­
sequently, a 345-kV underground/underwater PPP insulated pipe-type ca­
ble was successfully deployed [42]. However, the major developments are
taking place in the materials, manufacturing techniques, applications and
diagnostics of extruded dielectric power cables. In the past two decades,
XLPE and EPR have been extensively investigated in order to develop
cable insulations with improved resistance to initiation and growth of elec­
trical and water trees. Consequently, TRXLPE and EPR insulations have
shown improved characteristics in this regard. Considerable improvements
have also been made in the extrusion processes, curing methods and use
of semiconducting screens [1,43]. In addition to improved smoothness of
these screens and better bonding between the screens and the insulation,
the number and size of impurities and other defects in the insulation are
also being controlled.
The use of solid dielectric cables have been extended to EHV range
and 500-kV, XLPE insulated cables have been in service since 1987 with
good performance records [44,45]. EPR cables are also being manufactured
for medium and high voltage applications. Moreover, XLPE cables are
increasingly being used in DC and underwater applications. Along with
such developments in extruded cables, accessories for such cables have
also been developed for voltages of up to 500 kV [46].
In the medium voltage URD systems, utilities have gained from their
earlier experience and have modified the cable specifications for improved
performance against treeing and premature failures. Consequently, besides
High Voltage Cables 273

other factors, a water impervious jacket is recommended for URD cables

as there is an overwhelming evidence that the use of moisture barriers on
medium and high voltage cables can solve the problem of water treeing
[3,47]. Metal-plastic laminates are also being developed for such applica­
tions [35]. For existing cables that have suffered degradation due to treeing,
use of silicon fluids to extend cable life has been proposed [48].
A significant research effort has also gone into understanding the aging
and degradation mechanisms as well as into developing diagnostic tech­
niques. There is a general consensus that DC testing of service aged ex­
truded dielectric cables can cause premature cable failures after the cables
are returned to service, and the DC test should be replaced with a very
low frequency AC test [49]. Impulse overvoltages can also affect the re­
maining life of extruded cables by aiding tree propagation. PD and leakage
current methods have been proposed to monitor the deterioration of XLPE
cable insulation [26,50].
An important area in which progress is desirable is cryogenic super­
conducting cables. Some of the cryogenic cables tested up to the present
time have demonstrated an acceptable short-term performance and the op­
erating parameters achieved should permit economical superconducting
systems to be developed [51]. However, a considerable research and de­
velopment effort will be required before such systems, using either con­
ventional or high temperature superconductors, are commercially used.
Further details of various aspects of high voltage cables can be found in
cable handbooks, e.g., Bungay and McAllister [52].

REFERENCES

1. T. Fukuda, IEEE Elect. Insul. Magazine, Vol. 4, No. 5, pp. 9 -16, 1988.
2. N.M. Bums, R.M. Eichhorn and C.G. Reid, IEEE Elect. Insul. Magazine, Vol.
8, No. 5, pp. 8-24, 1992.
3. G. Graham and S. Szaniszlo, IEEE Elect. Insul. Magazine, Vol. 11, No. 5,
pp. 5-12, 1995.
4. M. Khalifa (ed.). H ig h V o lta g e E n g in e e r in g : T h e o r y a n d P r a c tic e , Marcel
Dekker, Inc., New York, 1990.
5. P. Graneau, U n d e r g r o u n d P o w e r T ra n s m is sio n : T h e S c ie n c e , T e c h n o lo g y a n d
E c o n o m ic s o f H ig h V o lta g e C a b le s , John Wiley & Sons, New York, 1979.
6. N.H. Malik and A.A. Al-Arainy, Int. J. Elect. Engineering Education, Vol. 25,
No. 1, pp. 27-32, 1988.
7. T. Gonen, E le c tr ic P o w e r T r a n s m is s io n S y s te m E n g in e e r in g : A n a ly s is a n d
D e s ig n , John Wiley & Sons, New York, 1988.
8. N.H. Malik, IEEE Trans, on Elect. Insul., Vol. 24, No. 1, pp. 3-20, 1989.
274 Chapter 9

9. N.H. Malik and A.A. Al-Arainy, IEEE Trans, on Power Delivery, Vol. 2, No.
3, pp. 589-595, 1987.
10. N.H. Malik and A.A. Al-Arainy, IEEE Trans, on Elect. Insul., Vol. 20, No.
3, pp. 499-503, 1988.
11. H.M. Ryan (ed.). H ig h V o lta g e E n g in e e r in g a n d T e s tin g , Peter Peregrinus Ltd.,
London, England, 1994.
12. J.H. Neher and M.H. McGrath, AIEE Trans, on PAS, Vol. 76, pp. 752-772,
1957.
13. lEC Standard Publication 287, 2nd Edition, 1982.
14. G.J. Anders, M.A. El-Kady, R.W. Ganton, D.H. Horrocks and J. Motlis, IEEE
Trans, on PWRD, Vol. 2, No. 2, pp. 289-295, 1987.
15. M. A. El-Kady, IEEE Trans, on PAS, Vol. 103, pp. 2043-2050, 1984.
16. B. M. Weedy, T h e r m a l D e s ig n o f U n d e r g r o u n d S y s te m s , John Wiley & Sons,
New York, 1988.
17. N. H. Malik, A.A. Al-Arainy, A.M. Kailani and M.J. Khan, IEEE Trans, on
Elect. Insul., Vol. 22, No. 6, pp. 787-793, 1987.
18. J.C. Chan, P. Duffy, L.J. Hiiala and J. Wasik, IEEE Elect Insul. Magazine,
Vol. 7, No. 5, pp. 9-20, 1991.
19. S.A. Boggs, IEEE Elect. Insul. Magazine, Vol. 6, No. 4, pp. 33-39, 1990.
20. S. A. Boggs, A. Pathak and P. Walker, IEEE Elect. Insul. Magazine, Vol. 12,
No. 1, pp. 9-16, 1996.
21 . ICEA Publication T-24-380, “Guide for Partial Discharge Test Procedures” ,
1980, USA.
22 . lEC Publication 270, “Partial Discharge Measurements” , lEC, Geneva, Swit­
zerland, 1981.
23. lEC Publication 885, “Partial Discharge Tests” and “Test Method for Partial
Discharge Measurements on Lengths of Extruded Power Cables” , Parts 2 and
3, lEC, Geneva, Switzerland, 1987 and 1983.
24. F. H. Kreuger, M.G. Wezelenburg, A.G. Wiemer and W.A. Sonneveld, IEEE
Elect. Insul. Magazine, Vol. 9, No. 6, pp. 15-23, 1993.
25. C. Laurent and C. Mayoux, IEEE Elect. Insul. Magazine, Vol. 8, No. 2, pp.
14-17, 1992.
26. G. Katsuta, A. Toyo, K. Muraoka, T. Endoh, Y. Sekii and C. Ikeda, IEEE
Trans, on PWRD, Vol. 7, No. 3, pp. 1068-1074, 1992.
27. T. Tanaka and A. Greenwood, A d v a n c e d P o w e r C a b le T e c h n o lo g y , Vol. I and
II, CRC Press, Boca Raton, Florida, 1983.
28. R. M. Eichhorn, “Treeing in Solid Organic Dielectric Materials” in E n g in e e r ­
in g D i e le c t r i c s , Vol. IIA, R. Bartnikas and R.M. Eichhorn (eds.), ASTM Press,
Philadelphia, 1983.
29. S. Grzybowski and R. Dobroszewski, Proc. IEEE Int. Symp. on Elect. Insul.,
pp. 122-125, 1978.
30. B.R. Varlow and D.W. Auckland, IEEE Elect. Insul. Magazine, Vol. 12, No.
2, pp. 21-26, 1996.
31. E.F. Steennis and F.H. Kreuger, IEEE Trans, on Elect. Insul., Vol. 25, No. 5,
pp. 989-1028, 1990.
High Voltage Cables 275

32. E.F. Steennis and A. van der Laar, Electra, No. 125, pp. 89-101, 1989.
33. M.S. Mashikian, R. Luther, J.C. Mclver, J. Jurdsin and RW. Spencer, IEEE
Trans, on PWRD, Voi. 9, No. 2, pp. 620-628, 1994.
34. J.J. Xu and S.A. Boggs, IEEE Elect. Insul. Magazine, Voi. 10, No. 5, pp.
29-37, 1994.
35. K.E. Bow, IEEE Elect. Insul. Magazine, Voi. 9, No. 5, pp. 17-28, 1993.
36. A. Asano, T. Takahashi, K. Maeda and T. Niwa, IEEE Trans, on PWRD, Voi.
9, No. 1, pp. 553-558, 1994.
37. F.H. Kreuger, P.H. Morshuis and A.M. van der Laar, IEEE Trans, on Elect.
Insul., Voi. 24, No. 6, pp. 1063-1070, 1989.
38. T.W. Dakin, Trans. AIEE, Voi. 67, pp. 113-122, 1948.
39. VK. Agarwal, H.M. Banford, B.S. Bernstein, E. Brancate, R.A. Fouracre,
G.C. Montanari, J.L. Parai, J.N. Seguin, D.M. Ryder and J. Tanaka, IEEE
Elect. Insul. Magazine, Voi. 11, No. 3, pp. 37-43, 1995.
40. T. Tanaka, T. Fukuda and S. Suzuki, IEEE Trans, on PAS, Voi. 95, No. 6, pp.
1892-1900, 1976.
4L W.A. Banker, PR. Nannery, J.W. Tarpey, D.W. Meyer and G.H. Piesinger,
IEEE Trans, on PWRD, Voi. 9, No. 2, pp. 1187-1194, 1994.
42. J. Grzan, R.V Casalaina, E.I. Hahn and J.O. Kansog, IEEE Trans, on PWRD,
Voi. 8, No. 3, pp. 750-760, 1993.
43. K.B. Mueller, U. Trefow, B. Dellby and G. Hjalmarsson, IEEE Trans, on
PWRD, Voi. 5, No. 4, pp. 1660-1668, 1990.
44. K. Ogawa, T. Kosugi, N. Kato and Y. Kawawata, IEEE Trans, on PWRD,
Voi. 5, No. 1, pp. 26-32, 1990.
45. K. Kaminaga, M. Ichihara, M. Jinno, O. Fujii, S. Fukunaga, M. Kobayashi
and K. Watanabe, IEEE Trans, on PWRD, Voi. 11, No. 3, pp. 1185-1194,
1996.
46. T. Kubota, H. Takaaki, H. Noda, M. Yamaguchi and M. Tan, IEEE Trans, on
PWRD, Voi. 9, No. 4, pp. 1741-1759, 1994.
47. J.H. Dudas, IEEE Elect. Insul. Magazine, Voi. 10, No. 2, pp. 7 -16, 1994.
48. PR. Nannery, J.W. Tarpey, J.S. Lacenere, D .F Meyer and G. Berlini, IEEE
Trans, on PWRD, Voi. 4, No. 4, pp. 1991-1996, 1989.
49. H.R. Gnerlich, IEEE Elect. Insul. Magazine, Voi. 11, No. 5, pp. 13-16, 1995.
50. T. Nakayama, IEEE Trans, on PWRD, Voi. 6, No. 4, pp. 1359-1365, 1991.
51. E.B. Forsyth, IEEE Elect. Insul. Magazine, Voi. 6, No. 4, pp. 7-16, 1990.
52. E.W.G. Bungay and D. McAllister (eds.). E l e c tr ic C a b le H a n d b o o k , BSP Pro­
fessional Books, Oxford, England, 1990.
10
Generation and Measurement of
Testing Voltages

10.1 INTRODUCTION

High voltage testing is the final step in ensuring the dielectric quality of
the developed insulation material. Similarly, the complete HV system or
device is tested to ensure its integrity and performance. Chapter 12 will
deal with various aspects of high voltage testing. This chapter discusses
the basic circuits used for generating the test voltages and the traditional
measuring techniques. The next chapter will outline more recent advances
in high voltage measurement techniques. High test voltages normally re­
quire only moderate currents (from few mA up to a few A), and thus the
generation schemes discussed in this chapter will generally fall in this
category.

10.2 HIGH VOLTAGE DC GENERATION

High DC voltages (HVDC) are mainly used in scientific research, in testing

equipment for HVDC transmission systems, and sometimes for high volt­
age cables and insulation testing. HVDC can be generated by either direct
method, i.e., changing mechanical energy directly to HVDC, or by indirect
method, i.e., changing mechanical energy to high AC voltages (HVAC) and
then converting it to HVDC. The direct generation is obtained by van de
Graff generators, but this method suffers from low kVA output, limiting
its use to some special applications in physics. In the indirect method.

276
Generation and Measurement of Testing Voltages 277

HVAC is rectified using HV diodes and capacitors. Ripple factor (5V) and
voltage drop (AV) are the two important parameters to be considered when
dealing with rectification of HVAC to get HVDC. In reference to Figure
10.1, 6V is defined as half the difference between the maximum and the
minimum voltage values, i.e.:

Figure 10.1 Voltage doubler: (a) circuit, (b) output voltage at no load and
(c) output voltage at load.
278 Chapter 10

SV = 0.5(V, Vm •m)/ ( 10. 1)

In most applications, the percentage ripple allowed is between 3% and 5%.

The voltage drop AV is defined as:
AV = V,,. - Vp, ( 10. 2)

where and Vp.L are average values of no-load and full-load DC volt­
ages, respectively. The voltage doubler circuit shown in Figure 10.1a is
widely used to generate a DC voltage of ± 2 Vp from an AC voltage of
Vp sin o)t The voltage waveshape at no load is shown in Figure 10.1b.
The output voltage with load is depicted in Figure 10.1c, showing AV as
well as 8V. The maximum voltage across the diodes or the capacitor is
2Vp. Since the costs of these components increase at much higher rates
than the increase in their voltage rating, it is not economical to increase
their voltage rating to generate higher voltages. Instead, two or more dou­
bler circuits are connected in cascade to form a voltage multiplier circuit.
Figure 10.2 shows a multiplier circuit consisting of three stages (n == 3)
with maximum output voltage of 6Vp. For such a circuit, 8V and AV are
given as [1]:

8V = ¿ „ ( „ + I, (10.3)

AV = (10.4)
3fC 5
where I = DC load current, f = AC supply frequency, C = stage capaci­
tance and n = number of stages. It can be seen from the above equations
that SV is proportional to n^ while AV is proportional to n^. Thus, the

Figure 10.2 Multiplier circuit (n = 3) for HVDC generation.

Generation and Measurement of Testing Voltages 279

number of stages in such a circuit is the limiting factor as it influences

adversely the values of SV as well as AV.
The above limitations can be overcome by energizing each stage by a
transformer comprising a tertiary low voltage winding which excites the
primary winding of the next upper stage. In this way, the problems of
and AV are solved; however, there are extra costs involved for the addi­
tional transformer windings. The limiting factor in such circuits employing
cascaded transformers is the loading of the first stage transformer’s primary
which has to supply the current for all n stages. In addition to the above
circuits, there are other HVDC generating arrangements like Deltatron cir­
cuit and voltage multiplier circuit with individual transformers for each
stage [1].

10.3 HIGH VOLTAGE AC GENERATION

There are three main methods for HVAC generation, which are briefly
described next.

10.3.1 Single Testing Transformer

This is similar to a single-phase step-up transformer with emphasis on the
insulation of the HV winding. The windings are generally designed for
low current ratings. The per unit impedance is kept less than 0.05 and the
magnetizing current is also kept low to minimize the harmonics in the
output voltage. Furthermore, care must also be taken to ensure that the
field distribution over the HV winding is uniform. Single test transformers
are economical for voltages of <300 kV. Above this level, a single test
transformer becomes quite expensive due to the added insulation require­
ments and due to the burden in transportation and erection of large trans­
formers. Thus, for higher voltages, cascaded transformers are the main
choice.

10.3.2 Cascaded Transformers

The idea of cascading transformers is based on connecting the secondary
windings of various transformers in series while their primary windings
are energized in the normal manner. Figure 10.3 shows three stages of a
cascaded transformer. The second- and third-stage transformers are isolated
from the ground by insulators capable of withstanding voltages of V and
2V, respectively. Here the total output voltage is 3V and is taken across
280 Chapter 10

31

Pi

Figure 10.3 Cascaded transformer for HVAC generation.

the high voltage terminal of the third transformer and the ground. The two
main limitations of increasing the number of stages in this scheme are the
high total internal impedance and the high loading of the primary of the
first stage. As can be seen from Figure 10.3, the current in the first trans­
former’s primary winding is 31. The net internal impedance of this circuit
can be derived as follows.
The total reactive power is given as:
Q = P(X„ + + X ,; + P(X,3 + Xp,)
+ (2I)2(X,3 + Xp^) + (3I)^Xp, (10.5)
If X„ = X,, = X h3 = X „ X, 3 = X, 3 = X„ and X^, = X^^ = X^ 3 = X^,
then:
Q = 3XhP + 5X,P + 14XpP ( 10.6)
If the whole arrangement is modeled by one impedance X^^ and one current
I then:
Q = = P[3X, + 5X, + 14Xp] (10.7)
Thus,
X,^ = 3X, + 5X, + 14Xp (10.8)
In general, for an n stage cascaded transformer, the total impedance is
given as:

Xe, = S [Xhj + (n - j ) % 3 + (n + 1 - j)^Xpj] (10.9)

j=i
Generation and Measurement of Testing Voltages 281

10.3.3 Resonance Circuit

One of the problems associated with the HV test transformer is the reso­
nance of its inductance with the load capacitance. Figure 10.4a shows a
simplified transformer equivalent circuit with single inductance (it is
assumed that R j = 0). If the load is purely capacitive (C J, such as a long
cable, then the load voltage is given as:
V,
( 10. 10)
(x)

where co is the supply frequency. Since the denominator is less than 1,

Vl > Vq and could be several times depending upon oj, L j and Cl and
could lead to the destruction or breakdown of the tested load. If this
phenomenon is used in a controlled manner, output voltage Vl of up to 50
times Vq could be obtained from a supply voltage V^. This is achieved by
using adjustable inductance L (where L > > and a capacitance C such
that C > > Cl (see Figure 10.4b). This type of circuit is called a resonance
circuit and it is used to generate HVAC especially when the required output
current is low. The main advantages of such a circuit are:
1. Less power requirement from the source (<10% of kVA required
for testing)

Lx

Vl

Lx

C T ^L
(9

(b)
Figure 10.4 Series LC circuit: (a) simplified HV transformer equivalent circuit
(R = 0) with capacitive load and (b) resonance circuit for HVAC generation.
282 Chapter 10

2. Pure sinusoidal output waveforms due to resonance

3. No arcing in case of a short circuit since the condition of reso­
nance ceases to exist with low shunt resistance of the arc
4. Simple and compact system
5. More than one resonance circuit can be cascaded to obtain even
higher voltages
Such systems are very suitable for testing capacitive loads such as cables.

10.4 HIGH VOLTAGE IMPULSE GENERATION

It is well known that power system components are subjected to severe

overvoltage due to internal switching surges or external lightning surges.
Consequently, the integrity of the individual components, devices and sub­
systems must be checked through high voltage surge testing. High voltage
surges that are simulated in laboratories are commonly known as high
voltage impulses. An impulse is defined as a unidirectional voltage (or
current) rising quickly to its peak value and then decaying slowly to zero.
Thus, an impulse has two parts: a rising part, which is usually realized by
charging a capacitor, and a decaying part, which is realized by discharging
a capacitor. Figure 10.5a shows a capacitor C2 which is charged through a
resistor Rj from a much larger capacitor Cj. In Figure 10.5b the two ca­
pacitors are discharged through a resistor R2 . Combining the two circuits
and replacing the mechanical switch by a HV switch, e.g., a sphere spark
gap, the simple impulse generating circuit of Figure 10.5c is obtained. The
impulse voltage waveform can be modeled mathematically by a double
exponential wave defined by the equation:
Vo(t) - A [exp(-at) - exp(-iSt)] (10.11)
Generally (3 » a, where /3 corresponds to the charging or the rising
portion of the voltage while a corresponds to the discharging or decaying
portion of the waveform. Figure 10.6 shows the typical impulse waveshape,
which is characterized by its front time and the tail time Tail time or
time to half of the voltage peak value (k) is the time between tg = 0 and
the time where the impulse decays to half of its peak value. Front time tf
is defined as peak time (tp) or [(ig - 1 3 ) 7 0 .6 ] for switching and lightning
impulses, respectively. Equation (10.11) can be derived from Figure 10.5c
using either time or frequency domain circuit analysis and is written as:
V
Vo(t) = [exp(-at) - exp(-/3t)] ( 10. 12)
k(l3 - a)

where
Generation and Measurement of Testing Voltages 283

t=o Ri
----------w v -----
=—
_+
Cl V C2^7k (a)

t=ti

R2 (b)

Ri
pO 0 -----W v —

_+ _Vo
V C2? R2 (c )

Figure 10.5 Principle of HV impulse generation: (a) charging a capacitor,

(b) discharging capacitors and (c) basic impulse generating circuit.

a
a,(3 b (10.13)
2
1 1 1 1
a = + ,b = , k = R,C2 (10.14)
R|Cj Ri^2 ^2^1 R jR2C]C
and V = initial voltage across Cj. The impulse generator circuit can also
be analyzed using computer software such as PSPICE, where the adjust­
ment of parameters becomes simpler. If the circuit components are known,
tf and k can be found by sketching V(t) versus t or by solving equation
(10.12) using the definitions of tf and k. If R2 > > R i, and Cj > > C2 , then
tf and k can be found approximately by the following formulas:
284 Chapter 10

Figure 10.6 Impulse waveshape parameters, (tf)^ = front time for lightning im­
pulses, (tf)s = front time for switching impulses and = time to half value.

tf = 3R, (10.15)
c, + c.
t, = 0.7(R, + R^XC, + C^) (10.16)
These simplified expressions are based on the assumption that a capacitor
can be fully charged in about where is the changing time constant,
i.e., T, = R,C,C 2 /(C, 0 ^
2 ^ Ftirthermorc, it is assumed that the capacitor

is discharged to its half voltage when the elapsed time is —0 . 7 where

is the discharge time constant and is given as = (Rj + R2 )(Ci + C2 ).
When tf and h are known and it is required to design the generating
circuit elements, we need to find a, ¡3, and then Rj, R 2 , Cj and C2 . Both
exact as well as approximate methods are available. For either case, it is
a normal practice to specify the values of two circuit elements and then
calculate the values of the remaining two elements. Usually Cj and C2 are
specified such that C2 < < Cj. C2 can also be used as the voltage divider
for measuring and recording the output voltage. The values of R^ and R2
can be found accurately by substituting the values of h and k in equation
(10.12) and finding a and (3 using numerical techniques. Then the values
of Rj and R2 can be evaluated by using equations (10.13) and (10.14) and
Generation and Measurement of Testing Voltages 285

assumed values of Cj and C2 . The values of a and ¡3 and consequently Rj

and R2 can also be found approximately by using the following formulas
[2]:
1 1
and (3 = (10.17)
RiQ C2 R 1
Furthermore, Table 10.1 can be used to find approximate waveshape or
circuit parameters. When using Table 10.1 or equation (10.17) to find the
circuit parameters, the calculated values can be verified by evaluating tf
and f and comparing them with specified values.
The single stage impulse generator circuit shown in Figure 10.5 is one
of the several circuits in which either the position of R2 is changed, or it
includes some additional resistors or inductors [3]. When very high impulse
voltages are needed, multistage impulse generators are used since the costs
of a single stage circuit’s elements become prohibitive as the voltage rating
becomes high. The basic idea of a multistage circuit is to charge several
stage capacitors in parallel and then discharge them in series. Figure 10.7a
shows such a circuit; Figure 10.7b shows the same circuit during the stage
capacitor charging period; and Figure 10.7c shows the same circuit during
the discharge period. Figure 10.7d shows the equivalent single stage circuit
which can be used to analyze the multistage circuit. In this circuit, only
Cj is distributed throughout all the stages. In other multistage impulse
circuits, R, and R2 can also be distributed throughout the different stages.
However C2 is normally left as a single unit which normally works as a
voltage divider or represents the load capacitance. There are several details
concerning the operation and control of this type of circuit. These can be
found in manufacturer’s manuals or in various references [1,2].
Such single and multistage impulse generation circuits are used to
generate lightning and switching impulses. For some applications, special

Table 10.1 Impulse Parameters for Various

Standard Waveshapes

- ( ms) ‘ ¿ (M S )-
tf/t, a 7] = V p/nV

1.2/50 68 0.4 0.96

1.2/200 284 0.38 0.99
170/1700 2200 43.5 0.91
250/2500 3160 62.5 0.9
650/2600 2500 250 0.75
286 Chapter 10

Ri

V— — w\ ----------- W
--------- v ------------ W v ----------
----------
+ + + +

V ^ ^ nCI V ? ^ nC1 V ? ^ nC1 M r

— —
f wv —
------------ W
----------
------------ v ----------

(b)

Ri

(d)
Figure 10.7 Multistage impulse generation: (a) main circuit, (b) stage capacitor
charging part, (c) capacitor discharging part and (d) equivalent single stage circuit.
 Generation and Measurement of Testing Voltages 287

kinds of switching impulses, such as impulses with very long tail times (in
ms) or impulses with some controlled oscillations on their tail, are required.
Table 10.2 shows a few circuits along with their output voltage waveforms
and mathematical representations of such waveforms.
When a full wave surge occurs on a power network and a flashover
takes place, e.g., across a bushing or an insulator, the voltage instantane­
ously falls to zero, resulting in a chopped wave. The voltage chopping can
take place either on the front, at the peak or on the tail of a surge. To
simulate a chopped surge wave, a rod-rod chopping gap is normally placed
in parallel with the tested object. The distance of the chopping gap can be
adjusted to control the width of the applied chopped wave during the

Table 10.2 Special Impulse Generating Circuits and Their Output Voltage Waveforms
Output Voltage
Circuit diagram waveform Output Voltage formula in s domain

V /K
Vo(s) =
L Ri Vo^ s^ + as^ + bs + c
K - R,C2
ci R2 ^ Vo R, 1
a = -— h
________ L C2R2
1 2 3 t(ms) RjCj + R2C2 + 1^2^i
b =
LC1 C2 R2
1
c =
C.C2R2L
V
Voi Vo(s) = ^K ■
s^ + as^ + bs^ + cs 4- d

K = LA. a = ^
__-L-J—^ ^ 1 1
■8 1 1.5 2 t(ms) b == +
C,L, C A
R 1
C = , d =
CTT C 1C 2 L 1L 2

Ri = 0,L2 >> L,,R2 = R

Va
^o(s) - s^ +---- TT
as + b
Vo
288 Chapter 10

chopped impulse testing. Triggered chopping gaps are often used to control
the chopping time. Chopped impulse testing is required in some applica­
tions as discussed in Chapter 12.

10.5 NANOSECOND PULSE GENERATION

In Chapter 2 the gas breakdown under high voltage pulses of nanosecond

duration was discussed. Due to an increasing number of applications of
discharges under nanosecond voltage pulses, several electric circuits have
been developed to produce such voltage pulses. The most common circuit
used for this purpose is based on the discharge of a capacitor into a low
inductance through a spark gap switch. However, modifications have been
incorporated in this basic circuit by several investigators for achieving a
better performance. Masuada [4] reported a simple circuit to generate nan­
osecond voltage pulses. A capacitor is rapidly charged and discharged util­
izing a rotary spark gap switch. Rea [5] analyzed the characteristics of a
nanosecond pulse generator circuit of Figure 10.8, which is based on Mas-
uada’s idea. In this case, the pulse rise rate can be controlled by R and
Li, while the pulse repetition rate is controlled by the rotation speed of the
spark gap. To improve the circuit performance, a DC supply, is added
and coupled to the circuit through C. It is found that the intensity of the

20KP 2 0 Y .f i R Ls Li C
-<wx— —Tjir-

I
i-E =r0.1Ai F
_.^p
5.7m;?

HV

1 1 1
[-*_50 0 _ ^5 [,---- 3100.

Figure 10.8 Nanosecond pulse generator circuit; dimensions are in mm. (From
R ef 5 © IEEE, 1995.)
Generation and Measurement of Testing Voltages 289

corona discharge produced by voltage pulses of this circuit depends on

total voltage magnitude, pulse voltage rise rate, value of Cp and the shape
of corona electrode. Bajeu et al. [6] used a four-stage Marx generator with
rotating spark gaps to generate nanosecond voltage pulses. The main prob­
lems encountered in this circuit are relatively longer rise times (tens to
hundreds of nanoseconds), the energy loss in the sparking of the rotary
gaps and the erosion of the spark gap. Kamase et al. [7] tested several
types and shapes of materials for the rotary gaps and found that flat elec­
trodes made from sintered tungsten with thorium dioxide coating had the
longest life. Recently a specially made HV semiconductor switching ar­
rangement was introduced to overcome the problems of mechanical switch­
ing where the pulse repetition rate can be increased to very high values
[8]. A typical wave shape produced by this circuit is shown in Figure 10.9
where the rise time is around 35 ns.
The main applications of nanosecond pulsed corona are the removal
of harmful impurities from air [5,6], purification of water from volatile
materials [8] and the destruction of weeds and some other harmful insects
and bacteria in the soil [9,10].

10.6 SPARK GAPS AS A VOLTAGE MEASURING DEVICE

10.6.1 Sphere Gaps
The sphere-sphere spark gap is one of the oldest and most reliable devices
for high voltage measurements. A roughly uniform field distribution is
achieved in the sphere-sphere gap when the gap length (d) is equal to or
less than the sphere radius (D/2). It is well known that under known en-

Figure 10.9 Nanosecond pulse voltage generated by ns pulse circuit using

semiconductor switching: y-axis = 5 kV/div and x-axis = 20 ns/div.
290 Chapter 10

vironmental conditions a uniform field air gap has a reasonably well-

defined breakdown voltage. Thus, such a gap can be used for the mea­
surement of the peak voltage value. Since the breakdown voltage
characteristics of air gaps at standard atmospheric conditions are reasona­
bly well defined, it is possible to establish tables (or figures) relating the
gap spacing with the breakdown voltage. Table 10.3 lists the sparkover
voltages of sphere-sphere gaps for various values of D and d at standard
environmental conditions of temperature = 20°C, pressure = 1013 mbar,
absolute humidity = 1 1 g/m^ and clean air. For other atmospheric condi­
tions, the sparkover voltage will vary and can be found using the procedure
discussed in Chapter 3. Consequently, for a gap distance d between two
spheres of diameter D, the actual sparkover voltage when P = P j, T =
Tj and H = Hj is given as:
(10.18)
where Kj is the atmospheric correction factor described in Chapter 3, and
Vj is the breakdown voltage value corresponding to d and D given in Table
10.3.
On the other hand, if one needs to adjust a gap spacing between two
spheres of diameter D such that the spheres sparkover at a certain pre­
specified voltage V^, then first Vj is calculated using equation (10.8). Then
from Table 10.3 the required value of d can be found which exhibits a
sparkover voltage of corresponding to the sphere diameter D. When
d < D/2, the error in the voltage values measured by sphere gaps is within
±3% [15]. However for D/2 < d < D, the error is within ±5%. The
breakdown voltage values are given for such large gaps in parenthesis in
Table 10.3.
Measuring spheres can be arranged horizontally or vertically as shown
in Figure 10.10. Moreover, for an accurate measurement scheme, there are
certain requirements regarding the minimum clearances to the nearby
earthed objects as well as to the ground. These requirements are specified
in lEC 52 [11]. Table 10.4 summarizes the required minimum clearances.
For uniform field air gaps, the breakdown voltage (in kV peak) can also
be estimated by the Bruce formula [12]:
V3 = 24.22d + 6.08Vd.RAD (10.20)
where RAD = relative air density (for details see Chapters 2 and 3) and
d = gap length in cm.
For the measurements of lightning impulses, breakdown time lags can
significantly influence the measurement accuracy when spark gaps are used
as the voltage measuring device. Irradiation increases the number of free
electrons needed for avalanche initiation. Studies show that external irra-
Table 10.3 Peak Values of Breakdown Voltages of Sphere Gaps with One Sphere Grounded at Standard Atmospheric
Conditions (P = 1013 mbar, T = 20°C)
s

Sphere Peak Voltage kV 5'

s
Gap Sphere dia D in cms a
s
Spacing s
d(mm) 12.5 25 50 75 rv
100 150 200 C<3
s
a b a b a b a b a b a b a b
10 31.7 31.7
20 59 59 59
30 85 85.5 86 86
40 108 no 112 112 S'
q
O
50 129 134 137 138 138 138 138 138 138 138 138 138 138
75 167 (181) 195 199 202 203 203 202 203 203 203 203 203 203
100 (195) (215) 244 254 263 263 265 265 266 266 266 266 266 266 S'
Oq
125 (214) (239) 282 299 320 323 327 327 330 330 330 330 330 330
150 (314) (337) 373 380 387 387 390 390 390 390 390 390
175 (342) (368) 420 432 443 447 443 450 450 450 450 450
200 (366) (395) 460 480 492 505 510 510 510 510 510 510
250 (400) (433) 530 555 585 605 615 620 630 630 630 630
300 (585) (620) 665 695 710 725 745 745 750 750
350 (630) (670) 735 770 800 815 850 858 855 860
400 (670) (715) (800) (835) 875 900 955 965 975 980
450 (700) (745) (850) (890) 945 980 1050 1060 1080 1090
500 (730) (775) (895) (940) 1010 1040 1130 1150 1180 1190
600 (970) ( 1020) ( 1110) (1150) 1280 1310 1340 1380
'toO
 iNJ
Table 10.3 Continued
Sphere Peak Voltage kV
Gap Sphere dia D in cms

Spacing
d(mm) 12.5 25 50 75 100 150 200

a b a b a b a b a b a b a b

700 (1025) (1070) ( 1200) (1240) 1390 (1430) 1480 1550

750 (1040) (1090) (1230) (1280) (1440) (1480) 1540 1620
800 (1260) (1310) (1490) (1530) 1600 1690
900 (1320) (1370) (1580) (1630) 1720 1820
1000 (1360) (1410) (1660) (1720) 1840 1930
1100 (1730) (1790) (1940) (2030)
1200 (1800) (1860) (2020) (2120)
1300 (1870) (1930) (2100)
1400 (1920) (1980) (2180)
1500 (1960) (2020) (2250)
1600 (2320)
1700 (2370)
1800 (2410)
1900 (2460)
2000 (2490)

"^For ac, ± d c, - v e lightning and - v e switching impulses.

^For + ve lightning and + ve switching impulses.
Generation and Measurement of Testing Voltages 293

(a)

Figure 10.10 Sphere gaps for HV measurement: (a) horizontal arrangement and
(b) vertical arrangement. 1 = insulating support, 2 = sphere shank, 3 = operating
gear, 4 high voltage connection with series resister, 5 = stress distributor, P =
sparking point of HV sphere, A = height of P above ground, B = radius of space
free from external structures, X = item 4 not to pass through this plane with a
distance B from R

diation is necessary for measuring low impulse voltages of <50 kV or for

measurements using spark gaps having spheres of diameter <125 mm if
an acceptable accuracy of ±3% is to be maintained [1,11].
A dust film on the surface of the unirradiated spheres may also influ­
ence the voltage measurement accuracy. Experiments show that the pres­
ence of dust film reduces the scatter of impulse breakdown voltages and
significantly improves the measurement accuracy for lightning and switch­
ing impulses which can further be improved by providing enough external
294 C h a p te r 1 0

Table 10.4 Clearances for Sphere Gaps

Value of A

D (cm) Maximum Minimum Value of B (min)

up to 6.25 7D 9D 14d
10 to 15 6D 8D 12d
25 5D 7D lOd
50 4D 6D 8d
100 3.5D 5D 7d
150 3D 4D 6d
200 3D 4D 6d

A and B are shown in Figure 10.10; D = sphere diameter, d = gap length, d < D/2.

irradiation [13,14]. However, for measurements of AC and DC voltages,

the error caused by dust particles on the sphere surfaces may reach up to
6% [13]. Thus, the presence of dust particles on sphere electrodes should
be considered when such gaps are used for high voltage measurements.

10.6.2 Rod Gaps

Rod-rod gaps are used for measuring HVDC with an error of less than 3%
[15,16]. The used rods should have a square cross section with each side
between 15 and 25 mm. For gap length in the range of 25 to 250 cm and
for absolute humidity H such that 1 g/m^ < H/5 < 13 g/m^, the break­
down voltage (in kV) as function of gap spacing d (in cm) can be
expressed as [16]:
Vb = 2 + 5.34d (10.21)
In this case, the applicable air density and humidity correction procedures
are similar to those discussed in Chapter 3. Moreover, these apply only to
the above specified gap lengths where the breakdown voltage-gap spacing
relationship is linear. Beyond these limits, the rod gaps cannot be used for
HVDC measurements, since the influence of environmental parameters es­
pecially humidity becomes nonlinear [17].

10.7 POTENTIAL DIVIDERS FOR HIGH

VOLTAGE MEASUREMENT

For the human safety and for the protection of measuring instruments,
potential dividers are used to decrease the high voltages. Two basic types
of dividers, i.e., resistive and capacitive dividers are commonly used for
Generation and Measurement of Testing Voltages 295

HV measurements. However each type has certain limitations when used

for measurements of very high transient voltages. Such limitations arise
from the stray capacitances and series inductances of divider elements.
Table 10.5 shows the two basic dividers, their inherent problems when used
for EHV and UHV measurements and the possible techniques to overcome
such problems, in addition to the conditions which must be satisfied to
ensure the divider transfer function is frequency independent. The divider
is usually connected to a low voltage measuring instrument (e.g., peak
voltmeter or oscilloscope) through a coaxial cable. In order to have an
accurate measuring system, the divider, the cable, and the low voltage
measuring and recording instruments must have adequate bandwidths suit­
able for the measurement of the required transient voltage. In addition there
should be a complete matching between different components in order to
avoid reflections and hence measuring waveform distortions. There are
certain test procedures to determine the divider ratio, and to assess the
divider suitability for high voltage and fast impulse measurements (see
section 12.5).

10.8 OTHER HIGH VOLTAGE MEASURING DEVICES

Table 10.6 summarizes the basic devices used for high voltage measure­
ments. The operating principles of some of these devices are described
here.

10.8.1 Electrostatic Voltmeter

When a voltage (V) is applied across a pair of parallel plate electrodes of
area A, separated by a spacing S and having a uniform electric field in­
tensity, E, the energy stored W is given as:

W = i sE^AS ( 10. 22)

2
where e = absolute permittivity of the medium between the plates. If the
spacing changes by dS, then the change in energy dW is given by:

dW = ^ sE^AdS (10.23)
2
Thus, mechanical force experienced by the plates is given by:
dW
F = 1 eE"A (10.24)
dS 2
296 Chapter 10

Table 10.5 Potential Dividers Used for HV Measurement

VI V2
2,
Vi Z/ + Zi
Z1 Z2
ViA^i must be independent o f frequency (f).

Resistive divider Capacitive divider

Ideally Ideally
V1 V2 VI V2
Yj . = ^2 •— VSAr-^-VSA/— ► Cl
Vj^ ^ ____
Vi R i^ R i R1 R2 Vi Ci ^C2 C1 C2

Prokteni Problem
With high V , R i becomes long and then stray When increasing V, C i becomes
cap Cs is not negligble, Cs in pF is long and series inductance (L)
given by: becomes important. V tÍ V i is
dependent on frequency.
_ 27tSoI
US/
In
2r
Cl + C 2 ~ 0) L C i C 2
/ = resistor length (m) and r = resistor radius
(m). V t/VI is dependent on f It vdll cause
distortion and errors.

Remedies Remedy
A) Add capacitors in parallel Add series resistor to both LV & HV sides
with L V & H V sides. o f the divider (damp the oscillations).

Y l. = R2+
V, ZI +Z2 J(^2
Ri
Zi = J+JcoCjRj Ri + . ^
jcoCj ^
jaC,
R2
Z, = J+JC0C2 R2 l + jw C ,R ,
= _____ Ri J^jCùC^Rj_
f', l-^jcùC^R2
R2+Ri^1^-JcoC^Rt if C iR , = C2R2
1 Cl
i f R iC I = R2C2
P, R1 + R2 Y j+ C1 +C2
Cl
select CiRi = C2R2
B ) screen the resistor, i.e.,make the stray
capacitance large but across the whole divider.
 ~
;::;
Table 10.6 High Voltage Measurement Systems and Devices "'
~
:::t.
System/ device Type of voltage measured Main advantages Main drawbacks a
;::;
:;::,
;::;
1. Sphere gaps Peak value of AC, DC and Simple, reliable, cheap Sensitive to gap adjustment, :;::,..,
impulse need correction for
environmental conditions.
~
:;::,
;:::
Continuous arc for AC "';;:
and DC: Accuracy ;:::
= ±3% "'~
2. Rod gaps 25 em < d < Peak value of HVDC Similar to sphere gaps Similar to sphere gaps. In ~
250 em addition, the influence of ~
humidity outside the ;:;.
~·
assigned range is not
defined ~
3. Electrostatic voltmeter AC (rms), DC (mean) Extremely high input Expensive and bulky ~
()Q
impedance, very accurate especially for very high "'"'
(error< .25%) voltages
4. Peak voltmeter Peak value of AC Simple, can be constructed There are many sources of
and calibrated in the error (frequency,
laboratory capacitors, diodes, LV
meters)
5. Resistive divider, a cable AC, DC and impulse Easily constructed in the The presence of stray
and LV meter or an lab, waveshape can be capacitance especially for
oscilloscope displayed. The frequency EHV and UHV dividers
of AC, and the front and and fast transient (see
tail times of impulses can section 10.7)
be measured w
\Q
'-1
 to
Oo

Table 10.6 Continued

System/device Type of voltage measured Main advantages Main drawbacks

6 . Capacitive divider, cable AC and impulse Similar to resistive dividers The presence of series
and LV meter or an inductance especially for
oscilloscope EHV and UHV dividers
and fast transients can
cause errors (see section
10.7)
7. Voltage transformer and AC Accurate The transformer is costly
LV meter especially for EHV and
UHV ranges
8. Series impedance and LV AC and DC Simple, can be constructed Not highly accurate
meter in the lab especially for AC, loads
the source, and the
reading is sensitive to
changes in temperature
9. Generating voltmeter AC and DC Does not load the supply, Requires separate drive, and
can also measure electric needs constant calibration
field directly since any disturbance in
position or mounting
needs new calibration
Generation and Measurement of Testing Voltages 299

Hence, the mechanical force experienced by the parallel plates can be used
for voltage measurement purposes since it is proportional to or
Electrostatic voltmeters usually have one fixed plate; the other is move-
able within a fraction of a millimeter in order not to disturb the original
electric field too much. Since the mechanical movement is very small, its
effect is amplified through electrical or optical means so that a reasonable
scale can be obtained. Figure 10.11 shows the arrangement of plates and
the use of light reflection for voltage measurement. The logarithmic scale
can be used so it can give the values of directly. The detailed con­
struction of electrostatic voltmeters vary depending on the manufacturer,
the measured voltage range and the insulation medium used.

10.8.2 AC Peak Voltmeter

The 90° phase shift between the current and the voltage in a capacitor is
utilized in peak AC voltmeter based on the Chubb and Fortesque circuit.
Figure 10.12 shows the basic circuit which includes a standard capacitor
C, two diodes and a low voltage ammeter. This ammeter will read the
average value of the current I through the diode Dj which is given as:
1 ^ P dV(t)
I = dt CdV(t) (10.25)
0 dt
For an AC voltage, v(t) = Vp sin cot. Therefore, equation (10.25) leads to:

Lam p

Figure 10.11 Basic principle of electrostatic voltmeter.

300 Chapter 10

C --

(a) (b)

Figure 10.12 AC peak voltmeter: (a) circuit and (b) i^t and v(t) waveshapes.

2C
I = ^nr VpP (10.26)

Since T = 1/f, equation (10.26) can be rewritten as:

V = — (10.27)
P 2fC
Thus, by measuring I and knowing f and C, Vp can be determined. This
method is accurate provided the positive and the negative half cycles of
AC voltages are symmetrical and equal.
The above principle can be modified to have digital peak voltmeters
for measuring AC as well as impulse voltages. This is achieved by passing
the rectified current through a resistor and then transferring this voltage to
a digital meter using (A/D convertor) and electronically controlled gates.

10.8.3 Generating Voltmeter

It is a variable capacitor electrostatic voltage generator which generates
current I that is proportional to the electric field E or to the applied voltage
V. Figure 10.13 shows a schematic diagram for a rotating-vane-type gen­
erating voltmeter. The measured high voltage (V) is connected to disc (D^),
while is being rotated by a motor at a suitable constant speed, and is
located at distance S from D^; is connected directly to the ground, while
is connected to the ground through an ammeter in order to measure
the current I(t), which is given as:
Generation and Measurement of Testing Voltages 301

Dh Dr Do D

HV

Dh

Figure 10.13 Rotating-vane-type generating voltmeter.

(10.28)

For a DC voltage, dV/dt = 0. Thus, equation (10.28) becomes:

dC
I(t) = V (10.29)
dt
When the motor is at a standstill, dC/dt = 0 and no current flows. When
the motor is rotating, C changes since the exposed area of to electric
field is changing with time from zero when it is completely covered, to
when the plates of disc are completely exposed to E. Thus:

C (.) = . ^ (10.30)

Hence,
Vc dA(t)
I(t) = (10.31)
S dt
Figure 10.14 shows the variation of C(t) and I(t) with t. The variation of
A(t) can be made linear or sinusoidal. This voltmeter can be used for AC
voltage measurements as well provided the speed of the drive motor is half
of the frequency of the measured voltage. For a four-pole motor, a speed
of 1800 or 1500 rpm is suitable for measuring 60 or 50 Hz voltages,
respectively. For peak value measurements, the phase angle of the motor
must also be controlled so that coincides with .
302 Chapter 10

Figure 10.14 The interdise capacitance and generated current as function of time
of generating voltmeter.

Field measurement is also possible by using this meter provided the

field between the HV disc and ground disc is uniform (other methods of
field measurements are given in Chapter 11). Precautions must be taken to
avoid breakdown between Dj, and D^. Thus, the voltage limit of such an
instrument should be clearly identified.

10.9 MEASUREMENT OF CORONA AND GAP

DISCHARGE CURRENTS

The fast current pulses resulting from corona and gap discharges are nor­
mally measured by placing a resistive shunt of low ohmic value between
the grounded electrode and the earth. It must be ensured that the inductance
of this shunt is very small. One method of generating smooth and true
corona current pulses is the use of coaxial cylindrical electrode geometry
where the high voltage is applied on the outer cylinder while the inner
cylindrical (pipe) is grounded [18,19]. A small protrusion (corona point)
is introduced in the center of the cylinder where it is isolated from the
grounded pipe by a small (i.e., 50 il) resistor. A 50-il cable is used to
carry the voltage drop across the resistor to the oscilloscope or any other
measuring instrument. Figure 10.15 shows the schematic diagram for the
above arrangement and the negative corona resulting from this corona
point.
There are other ways of reducing the inductance of the detection re­
sistance, such as placing several resistors in parallel to obtain the exact
value of the required resistance but reducing the resultant inductance by
paralleling the individual resistor’s inductances. Figure 10.16 shows such
an arrangement, where six 300-H resistors are arranged in parallel to give
a 50-il shunt while reducing its inductance to l/6th the value for individual
resistor element [20,21]. This shunt can be used for detecting corona or
Generation and Measurement of Testing Voltages 303

p n i
■■■

(b)

Figure 10,15 (a) Experimental setup for the generation and measurement of co­
rona using cylindrical geometry, (b) Resulting corona current pulse (time scale =
20 ns/div).

gap discharge current. Figure 2.14 shows typical gap discharge current
measured using such a detection resistor [21].
The suitability of the resistive shunts for measurements can be checked
either by applying and measuring a known signal or by finding the spec­
tral bandwidth of the measuring system, i.e., resistor, cable and the
oscilloscope.
304 Chapter 10

instrument

Figure 10.16 Current shunt resistors with low equivalent inductance used for co­
rona and gap-type discharge measurements.

REFERENCES

1. E. Kuffel and W. Zaengle, H ig h V o lta g e E n g in e e r in g F u n d a m e n ta ls , Perga-

mon Press Ltd., Oxford, England, 1984.
2. M. Naidu and U. Kamaraju, H ig h V o lta g e E n g in e e r in g Tata McGraw-Hill,
New Delhi, India, 1982.
3. M. Khalifa (ed.). H ig h V o lta g e E n g in e e r in g T h e o r y a n d P r a c ti c e , Marcel Dek-
ker Inc., New York, 1990.
4. S. Masuada and S. Husakawa, “Submicrosecond Pulse Energization for
Retrofitting Applications” , Proc. 2nd Int. Conf. on Electric Precipitation,
Kyoto, Japan, p. 613, 1984.
5. M. Rea and K. Yan, IEEE Trans, on Industry Applications, Vol. 31, No. 3,
pp. 507-512, 1995.
6. G. Bajeu, I. Mustata, C. Lungu and G. Musa, Proc. IEEE Industry Application
Society’s Annual Meeting, Vol. 3, pp. 1583-1586, 1994.
7. Y. Kamase, M. Shimiza, T. Nagahama and A. Mizuno, IEEE Trans, on In­
dustry Applications, Vol. 29, No. 4, pp. 793-797, 1993.
8. A. Al-Arainy, S. Jayaram and J.D. Cross, Proc. of 12th Int. Conf. on Cond.
and Breakdown in Diel. Liquids (12-ICDL 96), Rome, Italy, pp. 427-431,
1996.
9. A. Mizuno and Y. Hori, IEEE Trans, on Industry Applications, Vol. 24, No.
3, pp. 387-394, 1988.
10. A. Mizuno, T. Inoue, S. Yamaguchi and K. Sadamoto, Proc. IEEE Industry
Application Society’s Annual Meeting, pp. 713-727, 1990.
11. lEC 52, “Recommendation for Voltage Measurement By Means of Sphere
Gaps (One Sphere Earthed)” , Geneva, Switzerland, 1960.
12. C. Wadhwa, H ig h V o lta g e E n g in e e r in g , Wiley Eastern Limited, New Delhi,
India, 1994.
13. N.H. Malik, A. Al-Arainy and M.I. Qureshi, Journal of King Saud University
(Engineering Sciences) Vol. 9, No. 1, 1996.
14. A.A. Al-Arainy, N.H. Malik and M.I. Qureshi, The Arabian Journal for Sci­
ence and Engineering, Vol. 20, No. 3, pp. 495-509, 1995.
Generation and Measurement of Testing Voltages 305

15. N. Allen, IEEE Trans, on Elect. Insul., Vol. 28, No. 2, pp. 183-191, 1993.
16. lEC 60-1, “High Voltage Test Technique” , lEC, Geneva, Switzerland, 1989.
17. CIGRE Report #72, “Guide for the Evaluation of the Dielectric Strength of
External Insulation,” Working group 07 of Study Committee No. 33, CIGRE,
Paris, France, 1992.
18. W. Janischewskyj and G.L. Ford, IEEE Int. Symp. on Electromagnetic Com-
patability, IEEE Publication No. 70C28-EMC, pp. 436-441, 1970.
19 W. Janischewskyj and A. Al-Arainy, IEEE Trans., Vol. PAS-100, No. 2, pp.
539-551, 1981.
20 . R. Dobroszewski and W. Janischewskyj, IEEE Trans., Vol. PAS-100, No. 5,
pp. 2695-2702, 1981.
21. A. Al-Arainy, “Laboratory Analysis of Gap Discharges on Power Lines,”
Ph.D thesis. University of Toronto, Toronto, Canada, 1982.
11
New Measurement and
Diagnostic Technologies

11.1 INTRODUCTION

High voltage components and apparati are tested for a variety of reasons
(see chapter 12). During such tests, measurements of voltages, currents,
partial discharge levels, etc. are made to ensure that required test voltage
and/or current waveforms and amplitudes have been applied and the tested
equipment complies with the test requirements and specifications. Chapter
10 briefly reviewed the methods of generation of high test voltages and
their measurements. Due to great advances made during the last two dec­
ades in the fields of computers, electronics and optics, interest in the ap­
plications of such advanced technologies for the purpose of high voltage
measurements, testing and diagnostics have been growing. Consequently,
electro-optical and digital techniques are finding increasing use in the area
of high voltage, high current measurements and diagnostics. This chapter
provides a brief introduction to such new measurement and diagnostic
technologies.

11.2 DIGITAL IMPULSE RECORDERS

The conventional method of measuring high test voltages involves a suit­

able voltage divider, with a well-defined and frequency-independent divider
ratio, and a suitable low voltage measuring instrument. For high DC and
AC test voltages, a suitable analog or a digital reading peak voltmeter is

306
New Measurement and Diagnostic Technologies 307

normally sufficient since breakdown of most insulating materials is related

to the peak instantaneous applied voltage. For transient or impulse voltages,
an impulse oscilloscope or some other suitable recording device is used,
since in addition to the peak amplitude the voltage waveform is also re­
quired. A storage oscilloscope or an oscilloscope equipped with a camera
is usually used for this purpose. Alternately, a digital recorder may be used.
Commercial storage oscilloscopes can record waveforms with rise times as
short as a few nanoseconds and camera-equipped oscilloscopes can even
record pulses of subnanosecond rise times. Moreover, the amplitude can
be measured with —1% accuracy.
Digital recorders have several advantages over conventional oscillo­
scopes. For example, waveform data can be easily recorded, stored, re­
trieved, processed, compared and transmitted to other locations at great
speed and are, therefore, becoming increasingly popular for measurement
and recording of transient voltages or currents. For HV testing, these re­
cording instruments may be called upon to record standard lightning and
switching impulses, chopped and steep fronted impulses, as well as very
fast transients generated during the operation of SF^ and vacuum circuit
breakers, testing of surge arresters, insulators and breakdown of some in­
sulating materials. Therefore, such recorders should not only be fast
enough for required applications, but should also be accurate and reliable
to capture random transients. For high voltage measurement and recording
applications, the requirements for digital recorders are specified in the rel­
evant lEC and IEEE standards [1,2].

11.2.1 Recorder Parameters

In a digital recorder, an analog-to-digital converter (ADC) samples the
input signal at regular but discrete time intervals, which are defined by a
clock. The samples are then converted to digital codes and stored in suc­
cessive memory locations. Once recorded, the contents of the memory can
be transferred to a computer for analysis, display and printout or can be
displayed repetitively on a slow conventional oscilloscope with the aid of
a digital-to-analog converter (DAC). Thus, there are two important param­
eters which control the performance limits of a digital recorder. These are
the sampling rate (R) and the number of ADC bits (N) [3]. The amplitude
resolution commonly known as quantization error or code bin width (W)
is given as:
FSV
W = ( 11. 1)
308 Chapter 11

where FSV is the allowable full scale voltage which is usually —100 V.
Existing high speed digital recorders typically have 8, 10 or 12 bits, re­
sulting in quantization errors of 0.39%, 0.1% and 0.025%, respectively.
The sampling rate (R) defines how many times per second the input analog
signal will be converted to a digital code. According to information theory,
a sinusoidal signal of frequency F can be reconstructed if R ^ 2F. However,
for single-shot recording of voltage surges which are not bandwidth limited
and which provide only one opportunity to capture the waveform, it is
desirable to have as high a sampling rate as possible. Figure 11.1 shows
the errors caused by slow sampling rates when measuring the amplitude
of a linearly rising surge voltage. Recorders with R ranging from 20
MSa/s (or megasamples per second) to 5 GSa/s are available. R is
generally related to T^, the time to be measured (e.g., front time of an
impulse) by:

r. 30
( 11.2)
T
A third important specification of a digital recorder is the memory length
and its speed, since the memory must be able to store all the magnitude
codes as fast as generated by the ADC. The readout speed of the memory
may be slower.

11.2.2 Measurement Errors

As discussed above, the number of ADC bits N and the sampling rate R
set the theoretical limits on the precision of amplitude and time measure­
ments by digital recorders. Such recorders normally employ ADC tech­
nologies such as scan converters, charge coupled devices, flash converters
and ribbon-beam digitizers [3,4], each of which exhibits nonideal charac­
teristics. In such recorders, dynamic measurement errors can occur which
are quantified by differential nonlinearity (DNL), integral nonlinearity
(INL) and aperture uncertainty [4]. DNL is the difference between the
measured code bin width and the average code bin width (W) divided by
the average code bin width. For an ideal recorder, all of the code bin widths
are equal. However, DNL introduces a local variation in the code bin
widths. INL gives an estimate of the deviation of the recorders quantization
characteristics from its ideal counterpart. Thus, INL gives difference be­
tween the corresponding points on the measured quantization characteristic
and the quantization characteristic calculated using the static scale factors.
Aperture uncertainty is the standard deviation of the time intervals between
adjacent instants of sampling. Internally generated noise is another source
New Measurement and Diagnostic Technologies 309

Time (seconds)

Figure 11.1 Error due to slow sampling rate of a digitizer.

of error with real digital recorders. High resolution recorders are generally
more susceptible to such an error due to noise. For very accurate mea­
surements, the recorder errors have to be quantified, and therefore, tech­
nical procedures and computer models have been proposed to assess the
characteristics of digital recorders [5,6].

11.2.3 Technical Assessment

The traditional approach of testing the recorders uses sine wave inputs and
is known as an equivalent or effective bit (EB) test [3]. This test provides
a relation between the effective resolution of the recorder and the frequency
of the input signal. In this test, pure sine wave signals are digitized and
the digital record is fit with the best-fit sine wave function. The difference
between the digitized value and best-fit sine function value is taken as the
instantaneous error from which rms error is calculated. Then the rms error
is evaluated for an ideal digitizer, i.e., a digitizer where the only error
considered is due to quantization. The number of equivalent bits (EB) is
calculated from the relation:
EB = N - log2 RATIO (11.3)
where RATIO is the rms error of the real digitizer divided by the rms error
of an ideal digitizer. The shape of EB versus frequency characteristic for
recorder looks somewhat similar to the frequency response of an oscillo-
310 Chapter 11

scope as shown in Figure 11.2 [3]. The frequency response gives an exact,
well-defined and fully accurate account of error introduced in the signal
when recording it with a particular analog instrument. However, unlike the
frequency response of an analog oscilloscope, the EB characteristic only
provides statistical representation of the random errors that may be intro­
duced by the digital recorder. Thus, when the same signal is measured
repeatedly with the same digital recorder, there will be some differences
between the successive records. The EB characteristics sheds some light
on the possible magnitudes of such errors. On the other hand, when the
same signal is measured repeatedly using the same analog oscilloscope,
the records obtained will always be the same and will have the same dif­
ferences from the true signal. Technical procedures and computer models
have been proposed to assess the error bounds of such recorders using
input signals such as ramps, etc. [5,6]. Techniques for calibration of digital
recorders, dynamic testing of such recorders and methods used for evalu­
ation of amplitude, time parameters and overshoot for impulses and the
associated errors in such measurements have been discussed in literature
[7,8].

(0
ffl
•¥*
c
o
m
>
'5
cr
LU

Figure 11.2 Equivalent bit characteristic for a 100 MHz, 8 bit recorder. (Data
from Ref. 3.)
New Measurement and Diagnostic Technologies 311

11.2.4 Applications
Digital recorders are used for impulse voltage and current measurements
during high voltage testing of insulation. For standard lightning impulse
tests, the recorders can be separated into the following two groups based
on the performance requirements [8]:
1. For testing self-restoring insulation where only the evaluation of
impulse amplitude is required with a resolution of 0.4% and a time
uncertainty of 1%, an 8 bit recorder with a minimum sampling
rate of 40 MSa/s is sufficient.
2. For tests on equipment containing non-self-restoring insulation
such as transformers, records of two impulses and their resulting
neutral currents are compared. For such cases full and chopped
impulses including tests with front-chopped waves are sometimes
required. Therefore, a 10 bit recorder (which will allow an ampli­
tude resolution of 0.1%) with 60 MSa/s sampling rate is required.
Standard switching impulses can be accurately measured by using sam­
pling rates of as low as 120 kSa/s with recorder of 10 bits. Digital re­
corders can also be used to record switching surges on power system net­
works [9]. The rise time of such surges can vary from 50 ns to hundreds
of microseconds depending on system components. An 8 bit, 200 MSa/s
digitizer can record such surges with an acceptable accuracy. A restrike
during switching operations can also be recorded provided that either
enough memory is available or the digitizer’s clock speed is dynamically
changed during the record [9].
Partial discharges having rise and fall times of a few nanoseconds can
be digitally recorded if the sampling rate is either high enough (greater
than 5 GSa/s) or the PD pulses are slowed down by filtering [3]. Digital
recorders with high resolution (12 bits) and high sampling speeds (10
GSa/s) are now available and can be used for measurements associated
with short circuit testing as well [7].

11.3 DIGITAL TECHNIQUES IN HV TESTS

In addition to the use of digital recorders for impulse and other measure­
ments, discussed in the previous section, there are several other areas in
which digital techniques, involving both hardware and software, are being
applied successfully. Some of these applications and their advantages are
briefly outlined here.
312 Chapter 11

11.3.1 Improving Measurement Accuracy by Deconvolution

It is known from the convolution theorem that output, y(t), from a mea­
suring system is related to its input, x(t), by the impulse response, h(t), of
the system as follows:

y(t) x(r)h(t - T)dr (11.4)

If y(t) and h(t) were known analytic functions, deconvolution could be used
to determine x(t). This allows for improving the measurement accuracy by
determining the errors caused by the nonideal characteristics of voltage
dividers and measuring circuitry, etc. Therefore, the method of deconvo­
lution has been used by several researchers to improve the accuracy of
measuring systems with nonideal impulse responses. However, since y(t)
and h(t) can only be measured with limited resolution and may also be
contaminated with noise, the deconvolution technique should be used with
great care and must not be applied to compensate for an unsuitable mea­
suring system in standard impulse measurements [7]. However, in routine
measurements, deconvolution represents a useful tool to reduce errors
caused by measuring system limitations.

11.3.2 Impulse Testing of Power Transformers

Impulse testing of high voltage transformers is based on the comparison
of the voltage and the neutral (or ground current) transients when a stan­
dard lightning impulse is applied to the transformer’s HV winding at the
basic insulation level (BIL) and at a reduced voltage level (usually 50 to
75% of BIL) (for details see chapter 12). This is a nondestructive test in
which it is assumed that the test and measuring circuit remains essentially
the same during the reduced and the full wave tests and, therefore, any
differences in voltage and current records are caused only by partial or
complete insulation failure. In the conventional impulse tests, oscillo­
graphic records of transients are generally obtained by camera-equipped
oscilloscopes, and in some cases it is difficult to conclude if the insulation
has failed or passed the test, due to limited resolution of the oscillographic
records of the transients. Using digital recorders, these quantities can be
recorded in a digital format and, consequently, the differences between
voltages and currents at reduced and at full voltage levels can be calculated,
compared, magnified and displayed for a better judgement and evaluation
of the state of the tested insulation. Thus, the use of a digital recorder in
the time domain mode improves the transformer’s insulation diagnostics.
New Measurement and Diagnostic Technologies 313

In addition to the time domain applications, digital data of the applied

voltage and the resulting current records can also be manipulated to de­
termine the transfer function of the tested winding in the frequency domain.
The transfer function, TF(iL>), is defined as:

TF(oj) = i M (11.5)
V(co)
where V(a>) and I(cl>) = input line voltage and the output neutral current
functions in the frequency domain (oj). Since voltage and current values
are sampled in a digital format, such data can be converted to V(co) and
I(w) by fast fourier transform (FFT) computations with a digital processor.
The state of the transformer winding’s insulation is determined by com­
paring the transfer function obtained at full and at reduced test voltages.
TF(w) versus o) plots have been found to reveal that the local breakdowns
in the winding insulation between two turns can shift the resonance fre­
quencies in the transfer function thereby causing a shift in the transfer
function poles (Figure 11.3b). However, partial discharge between two
windings causes reductions in the peak value of the transfer function cor­
responding to some resonance frequency (Figure 11.3a). It has been further
observed that small changes in the applied impulse waveform or chopping
times do not significantly influence the shape of the transfer function. Thus,
the transfer function method permits an unambiguous acceptance or rejec­
tion of the transformer. Therefore, along with time domain analysis, this
technique is also finding increasing acceptance by utilities, transformer
manufacturers and test laboratories, thus resulting in improved diagnostics
of the winding insulation [7,8,10,11].

11.3.3 Partial Discharge Measurements

The occurrence of partial discharges (PDs) inside high voltage components
(see also chapters 8 and 12) aids the degradation of insulation leading to
its premature failure. Therefore, most power apparati have traditionally
been tested for PD using a particular type of measuring circuit and con­
ventional analog instruments [12]. The quantities usually measured involve
discharge inception voltage (DIV), discharge extinction voltage (DEV) and
apparent magnitude of the PD pulses. These quantities can also be deter­
mined using digital recorder as mentioned earlier in section 11.2.4.
In today’s environment of inexpensive high speed ADC and micro­
computers, it is possible to quantify different parameters of PD pulses with
relative ease. A tremendous amount and variety of PD data are now being
gathered by many researchers and new methods for the diagnostics of the
314 Chapter 11

Figure 11.3 Changes in transfer function due to (a) a discharge between two
windings and (b) a breakdown between two adjacent turns. (From Ref. 11 © IEEE,
1994.)

insulation integrity are being proposed. A comprehensive bibliography and

a brief review of the applications of digital techniques for partial discharge
measurements has been prepared by an IEEE working group [13].
Using digital techniques, pulse-height distributions, pulse-phase distri­
butions and pulse-time distributions can be obtained to statistically quantify
the partial discharge parameters [13,14]. In recent years, many advances
have been made in developing sophisticated hardware and software for
better quantification of partial discharges which were previously performed
only partly by using multichannel analyzers [15]. Consequently, in order
to assist engineers in the assessment of the quality and the condition of
the insulation, digital techniques have helped to improve PD measurement
sensitivity and to capture the stochastic properties of PD pulses. Some of
the areas where digital techniques are providing useful data include on­
line PD measurements in real-life, noisy environments through the use of
digital filtering, pattern-recognition and noise-reduction techniques, PD
New Measurement and Diagnostic Technologies 315

source identification and PD site location [8,13]. In addition, digital anal­

ysis of calibration pulses for PD measurements have demonstrated that the
calculated charges agree with the nominal measured values from 5 pC to
500 pC with ± 10% error and that variations of the test circuits influence
such values by less than 5% [7]. Despite all such advances, the persisting
general lack of knowledge on the aging rates of various insulating materials
exposed to partial discharges has largely prevented the use of computer-
controlled pulse-distribution methods on a routine test basis for electrical
apparatus [16].

11,4 TESTING AUTOMATION

High voltage testing is essential to ensure the insulation quality of the

manufactured component. High voltage tests can be separated into three
main parts (see chapter 12 for details): routine tests during the manufac­
turing process for quality assurance, type test as part of research and de­
velopment quality assurance, and on-site tests to ensure that proper instal­
lation and commissioning steps are followed. In addition, sometimes high
voltage tests are also required for periodic assessment of the insulation
quality.
Routine tests, which are the most frequent, offer the greatest potential
for testing automation. Computer-controlled test procedures and voltage
sources lead to a reduction in staff, testing time and error probability. In
addition, test results can be stored in digital form, a test report can be
prepared and printed automatically, and the results can also be transmitted
to a host computer which collects all the relevant data for a given customer
order, so that high voltage tests can become part of the computer integrated
manufacturing [8]. In automated testing, the operator can input the data
for the required test and the computer will select the test program, perform
the required calculations (e.g., for air density and humidity correction fac­
tors) and adjust the generator parameters. These adjustments can then be
confirmed by the operator and the test sequence is completed along with
the recording and printout of the test results. Computer-controlled impulse
testing systems are available from high voltage test system manufacturers
[17].
Other applications where computer-controlled testing and data-
acquisition systems are useful include long-term testing and monitoring of
insulation in multistress aging tests which are conducted to assess the suit­
ability of new materials and to estimate the expected life of high voltage
components.
316 Chapter 11

11.5 ELECTRIC FIELD MEASUREMENTS

Electric field strength (E) is one of the most important parameters in the
insulation design of high voltage apparati. A knowledge of E is also a
prerequisite to assess the biological effects of electromagnetic radiation and
the electromagnetic compatibility of electronic devices. The capacitive
probe, flux meter, and dipole antenna all have been successfully used for
measuring electric fields. Some of these sensors are metallic which limits
their use to the field measurement at the surface of metallic electrodes
only. In addition, the measured field value may be affected by the presence
of the probe in the interelectrode area. In recent years, new field sensors
based on electro-optic effects have been developed which overcome most
of these disadvantages and such sensors will be discussed in the forthcom­
ing sections. This section outlines the basic principle of a commercially
available, capacitive probe-based, electric field measurement system
[8,18].
The measuring system consists of a spherical (80 mm in diameter)
field sensor, two transmission cables, a receiver and a recording instrument.
The probe measures electric field components in two orthogonal directions,
which are transmitted by two fiber optic cables and then converted to an
electrical signal by the receiver. A capacitive probe usually consists of a
sensing electrode of area A immersed in a medium of absolute permittivity
e. The probe is connected to a reference electrode through a measuring
resistor It can be represented by the equivalent circuit of Figure 11.4,
where C is the probe capacitance and is the measuring voltage. For
this circuit [8]:

io(t) + Xu C + ¡R ( 11.6)
A dt dt R.

Figure 11.4 Equivalent circuit of an electric field measuring probe.

New Measurement and Diagnostic Technologies 317

If the sensor is terminated with the characteristic impedance of the mea­

suring cable, then i^ < < ip>, and therefore:

Vm = R„£ A ^ (11.7)

Since the measuring voltage is proportional to the derivative of the applied

field, E can be measured by integrating the measuring signal V^. On the
other hand, if the sensor is terminated with a high impedance (i.e., >
1 Mil), iR < < ic. the measuring signal is proportional to E since equation
(11.6) can be expressed as:
EA
( 11. 8)

Both principles are used in practical applications and several designs of

such probes exist.
If the distance between a high voltage electrode and the field probe is
larger than the probe diameter, electric field can be measured near the
electrode with good accuracy as long as no corona occurs at the high
voltage electrode. Using such a probe, high voltage can also be measured
if the probe is calibrated at a fixed position. Capacitive probes are useful
for measuring AC and impulse fields and have been successfully used for
such purposes [8].

11.6 ELECTRO-OPTIC SENSORS

Due to rapid advances in optical telecommunications technology, practical

applications of electro-optic transducers are increasing for the measure­
ments of voltages, currents as well as electric and magnetic fields associ­
ated with high voltage power networks. Consequently, a variety of electro­
optic measuring sensors have been developed. For the measurement of high
voltages and electric fields, such sensors have the following advantages
[19]:
1. They directly measure the electric field in space
2. They contain no electronic circuit and no power sources
3. They respond to changes in the electric field in the frequency range
from DC to GHz
4. The field distribution is disturbed only marginally by the sensors
itself, thereby resulting in better accuracy
318 Chapter 11

Furthermore, such sensors are very small, and therefore easy to transport
and install [20]. There is an increasing interest to integrate electro-optic
and magneto-optic transducers for monitoring electrical quantities in power
apparati such as circuit breakers, transformers, GIS, etc. Fiber optic sensors
can also be used to measure temperature, pressure, vibrations, gas density,
etc., and are being used for power equipment monitoring as well [8].
Electro-optic sensors for electric field intensity and voltage measurements
can broadly be classified into the following two main types: (1) sensors
based on the Pockels effect and (2) sensors based on the Kerr effect. A
brief introduction about the principle, characteristics and applications of
these sensors is provided next.

11.6.1 Electro-Optic Effect

The velocity of light depends on the refractive index of the material
through which the light passes. The refractive index of some materials has
different values for two mutually orthogonal polarizations of the light
wave. This property, whereby a material possesses two refractive indices,
is known as birefringence. In some materials, birefringence occurs when
an electric field is applied across the material. The birefringence, therefore,
causes the orthogonal components of polarization of the light to travel at
different velocities, which causes a phase shift between the waves. This is
the electro-optic effect. The refractive index n for each orthogonal com­
ponent of the optical polarization is related to the applied electric field E
by [19]:

n = n^ + aE + bE^ + (11.9)

where n^ = normal refractive index in the absence of any applied field and
a and b = coefficients for the electro-optic effects [19]. The term aE shows
that n varies linearly with E, i.e., a linear electro-optic effect commonly
known as the Pockels effect. However, the term bE^ shows a quadratic
electro-optic effect which is referred to as the Kerr effect. The higher-order
terms including and above E^ normally contribute very little to the refrac­
tive index changes and, therefore, can be usually ignored. Generally, in a
given material, the Pockels and Kerr effects do not appear simultaneously,
but one effect becomes dominant depending on the symmetry and the
electrical-polarization structure of the material. The Pockels effect is ex­
pected to appear in solid dielectric materials and the Kerr effect mainly in
liquid insulants [19].
New Measurement and Diagnostic Technologies 319

11.6.2 Pockels Sensors

Consider the light wave e which is linearly polarized and therefore has two
orthogonal components e^ and ey, and propagating along the z-axis, as
shown in Figure 11.5. The phase velocities of two components will vary
in inverse proportion to the refractive indices n^ and ny for each polarization
component of the light. So, when birefringence occurs, n^ Uy, and there­
fore the two components will travel at different velocities. After the light
passes through the electro-optic material, i.e., the Pockels crystal, the two
components are out of phase due to different velocities and have a phase
shift AO induced by the electric field. When the Kerr effect and other higher
order terms are neglected, this phase shift is given as [19]:
AO = AO^ + AE ( 11. 10)
where AO^ = phase shift corresponding to the natural birefringence in the
absence of any electric field (AO^ is usually ~ zero for most crystals) and
A = Pockels cell’s sensitivity coefficient, which is given as:
2 tt
A = nl rX ( 11. 11)

where A = wavelength of the incident light, r = Pockels coefficient and

ELLIPTICAL
POLARIZATION

LINEAR
POLARIZATION

ANALYZER
(<t>=-7r/A)
POCKELS CRYSTAL

POLARIZER
(0 = 7r/^)

Figure 11.5 Principle of electro-optic effect. (From Ref. 19 © IEEE, 1996.)

320 Chapter 11

L = length of Pockels crystal along the z-axis. The phase shift A0 causes
varying degrees of elliptical polarization of light, and consequently the
linear polarization of the incident light results in an elliptically polarized
state at the output of the Pockels crystal, as shown in Figure 11.5. If the
output light passes through a polarizer oriented orthogonally to the input
polarizer as in Figure 11.5, the output light intensity P^ varies nonlinearly
with E. However, if an optical bias of ttH radians is added to the original
phase shift of equation (11.10), P^ is related to the input light intensity Pj
by:

P, - ^ (1 + sin ^e) ( 11. 12)

When A6 is relatively small, sin ~ ~ + AE ~ AE, since A^^

is normally zero in most cases. Thus, P^ varies linearly with applied electric
field E. The crystal producing Pockels effect is called a Pockels crystal, a
Pockels device or a Pockels cell. Table 11.1 shows some important prop­
erties of a few Pockels crystals [19]. Crystals having a large Pockels effect
(A values) are desired to sense electric fields. In addition, the Pockels
crystal should have low dielectric constant, high resistivity, small piezo­
electric constant and no natural birefringence in order to minimize mea­
surement errors.
A practical Pockels sensor system consists of a coherent light source,
an optical conversion device for producing circular polarization, a Pockels
crystal, a polarizing plate and a photodetector. Figure 11.6 shows two typ­
ical Pockel sensor systems, whereas Figure 11.7 shows the relationship

Table 11.1 Selected Properties of Some Pockels Crystals

Sensitivity coefficient Resistivity

Material A /L (x lO -V V ) Relative permittivity (il cm)

LiNb03 40-100 50% 100" 10'*’

LiTa03
ADP, KDP 30 20", 50" lO'O-lO'''
Quartz 1 4 > 10'«’
Bij2SiO20 40 -7 0 50 10'“
Bi i2fje02o
ZnS, ZnTe 30-80 7 -1 0 10" -10"
Bi4Ge30i2 10 16 10'“

^These values are for two orthogonal directions.

Source: Ref. 19.
New Measurement and Diagnostic Technologies 321

INCIDENT BEAM
(CIRCULAR POLARIZATION)
I OUTPUT
I r u I DBEAM
C M |V | 1 ^
^ j PHOTODETECTDR
POLARIZING PLATE
^
^ k s "'
\ ^ " ^ - ^ C K E L S CRYSTAL

PO LA R IZER (0=7r/4)
(a)
POLARIZED
BEAM SPLITTER 1
POLARIZING PLATE
QUARTER WAYE PLATE
LASER POCKEL5 CRYSTAL

OPTICAL FIBER

LEO OR LO POLARIZED
BEAM SPLITTER 2

PHOTO DIODE (b)

Figure 11.6 Two typical Pockels sensor systems employing (a) free space trans­
mission of laser and (b) optical fiber sensor. (From Ref. 19 © IEEE, 1996.)

Applied Electric Field (MV/m)

Figure 11.7 Variation of optical sensor output with applied electric field. (Data
from Ref. 19.)
322 Chapter 11

between optical output of the sensor and the applied electric field for two
Pockels crystals [19]. A laser, a light-emitting diode (LED) or a laser diode
(LD) is usually used as a source of light. All optical devices in which
polarization of light is modified are assembled into a single unit. After
passing through this unit, the light has an intensity that depends on the
applied electric field, and thus the electric field can be measured by mea­
suring the light intensity. Pockels sensors for measuring electric field or
voltage have been developed and used by many investigators [19-23].
Some of such sensors are commercially available as well. For a Pockels
sensor to be used in the measurement of voltage, the sensor should be
calibrated by a standard voltage transformer or a calibrated voltage divider.
The factors which should be considered in the choice of a Pockels sensor
are response nonlinearities due to Kerr and higher-order effects, influence
of temperature and vibrations on sensor performance, frequency response,
and oscillations introduced in the output signal of the Pockels crystal by
the piezoelectric effect, etc.
In many countries, Pockels voltage sensors have been incorporated into
electric power networks and power apparati such as GIS. Pockels field
sensors are applied to the measurement of not only the electrostatic field,
but also the space charge modified fields in electrical discharges due to
DC, AC, lightning impulse or switching impulse voltages. This has helped
in a better physical understanding of the space charge and discharge phe­
nomena in gas insulation [19]. The sensors are being developed for mea­
suring field vectors and are being miniaturized with the introduction of
optical waveguide technology and are showing improved performance. It
is anticipated that the applications of Pockels sensors in high voltage com­
ponents and insulations systems will increase.

11.6.3 Kerr Sensors

Kerr-effect-based electro-optic sensors are also used for electric field and
voltage measurements and work on the same electro-optic principle as the
Pockels sensor. However, in this case the phase shift A0 between the two
orthogonal components of a light wave after it travels through an electro­
optic Kerr cell is proportional to W. If we assume no Pockels effect and
no natural birefringence (i.e., in the absence of an applied field = 0),
then A6 is given as:
= BE^ (11.13)
where r,^ = Kerr constant and L = effective length of the optical path in
the Kerr cell. The Kerr cell sensitivity coefficient B is equal to 27rrj^L, as
opposed to Pockels cell sensitivity coefficient A given by equation (11.11).
New Measurement and Diagnostic Technologies 323

Since it is difficult to measure AO directly, it is calculated from the mea­

sured relative transmitted light intensity ratio i.e. Pq/P í, where P^ is the
transmitted light intensity and P^ the incident light intensity.
Liquid dielectrics usually exhibit Kerr effect because their molecules
have different polarizabilities in at least two different perpendicular mo­
lecular axis and are, therefore, used in Kerr cells. Table 11.2 gives values
of Kerr constant rj^ and breakdown strength for a few dielectrics which
have large Kerr constants [24]. Kerr sensors are used to measure either
high voltages or to measure electric field values in stressed liquid dielec­
trics. For voltage measurements, application of linearly polarized light at
an angle of tt/4 radians to the direction of the applied electric field gives
equal polarization components parallel to and perpendicular to the exter­
nally applied field. This light is elliptically polarized by the Kerr cell. After
exiting from the cell, the light is passed through a second polarizer which
is oriented orthogonally to the input polarizer. Then output optical intensity
is given as:
AO T
P: sin^ = P; sin^ (11.14)

where V == applied voltage between the cell electrodes, and d = effective

distance across which V is applied. This type of arrangement uses crossed
polarizers. Other optical arrangements are also used for E field measure­
ments. Equation (11.14) can also be used to measure unknown voltage V.
Kerr cells employing nitrobenzene are used for high voltage measurements
and as standard voltage dividers [25-27].
Kerr electro-optical transducers have also been used to measure the
spatial field distribution in stressed dielectric liquids in order to understand
the role of space charge in their breakdown. The measurement range of
the electric field magnitudes is limited by the Kerr constant (rj^) of the
liquid dielectric and the length L of the electrode for lower limit, and by
the breakdown strength Eg of the dielectric medium for the upper limit.
For such measurements, the applied field magnitude E should also be much
higher than the field magnitude E^^ necessary to reach the first light maxima
to produce some bright and dark isochromatic fringes. Table 11.2 also
shows E^ values for a fixed value of L for some liquid dielectrics. Figure
11.8 shows optic measurement systems using linearly and circularly po­
larized light [24]. For both arrangements, either cross polarizers or aligned
polarizers can be used resulting in different forms of the Pq/P ì relationships
[24]. In recent years, the field measurement sensitivity has been greatly
improved by the use of electric field modulation and elliptically polarized
incident light [28]. Moreover, the dynamic behavior of two-dimensional
electric field distributions has been observed because of advanced tech-
 324 Chapter 11

Table 11.2 Kerr Constant r,,, Field Magnitude and Breakdown Strength Eg
of Some Dielectric Liquids

E , [V/m ]
Liquid specimen r, (m/V^) (L = 3 cm) Eb [V/m ]

Nitrobenzene 20 X 1 0 -‘" 2.3 X 10'’ 8.5 X 10'’

Purified water 3.4 X 7.7 X lO" 14.0 X lO*-
Transformer oil 3.5 X 10-'^ 55.1 X lOo 20.0 X 10'’

S o u rce: Ref. 24.

niques such as two-dimensional lock-in amplifier, high quality optic de­

vices, highly sensitive charge coupled device cameras and computer image
processing techniques [24,29]. A method to measure three-dimensional
electric field distributions in liquid dielectrics using the Kerr effect com­
bined with AC field modulation and circularly polarized light has been

Voc
* e
s' m
Voc
V I
Film
He-Ne
Laser

P o la rize r
BW E
Analyzer
Cam era

Test cell

(O) Voc ,

I I „

Polarizer IM w ave 1/4 wave Analyzer

plate plate

(b)
<t^=+;r/4 <l>Q,=o ‘I>Q2=*” 72 <t>A=-n’/A

Figure 11.8 Optic measurement system for nonuniform electric field in the plane
perpendicular to the light propagation: (a) using linearly polarized incident light
and (b) using circulary polarized incident light. (Erom Ref. 24 © IEEE, 1996.)
New Measurement and Diagnostic Technologies 325

discussed by Shimizu et al. [30]. The applications of Kerr effect for di­
agnostics of liquid dielectrics are expected to grow in the future.

11.7 MAGNETO-OPTIC SENSORS

Magneto-optic sensors are used for measuring current and magnetic field
intensity in high voltage power networks. Current measurement is required
during high power testing in addition to its use for metering and relaying
purposes. Magnetic field measurements are also essential because of in­
creasing concern about the possible health hazards of power frequency
electro-magnetic fields. Conventional methods of current measurements
employing bulky current transformers are being questioned by a large num­
ber of electric power utilities due to cost, reliability and safety considera­
tions. Consequently, interest is growing in the development and applica­
tions of integrated electro-optic and magneto-optic sensors. Such integrated
electro- as well as magneto-optic sensors offer several advantages over
their conventional counterparts, i.e., potential and current transformers (PTs
and CTs). For example integrated optic sensors are nonintrusive, inherently
insulating, corrosion resistant, compact, lightweight, flexible and immune
to electromagnetic interference. The output signals of these sensors are
compatible with the signal levels required for digital protection systems.
Therefore, such magneto-optic transducers are briefly described next.

11.7.1 Faraday Effect

The magnetic field and current measuring optical sensors are usually based
on magneto-optic effect which was first observed by Michael Faraday 150
years ago and is commonly known as the Faraday effect. According to this
effect, when a linearly polarized light ray propagates through a magneto­
optic medium in the presence of an external magnetic field, it undergoes
a rotation of the plane of polarization proportional to the strength of the
magnetic field (H). The rotation angle Ad (radians) is related to the mag­
netic field intensity H (ampere-turn/m) and interaction length L (m) by the
following equation:
Ad = acK, / H • dL (11.15)
where = Verdet constant of the material (radians/ampere-turn), ¡ = lj

relative permeability, and H and dL = components in the direction of light

propagation. The Verdet constant varies with light wavelength and is
material as well as temperature dependent. When a linearly polarized light
326 Chapter 11

encircles a current carrying conductor, by using Ampere’s law, equation

(11.15) can be rewritten as:
= iuiK,N,l (11.16)
where = number of turns of the optical path around the current carrying
conductor. Thus, from equation (11.16), the angle of rotation A6 is directly
proportional to the enclosed electric current I. This method rejects the
magnetic field signals due to external currents which are also quite strong
in power systems. Equation (11.15) can also form the basis of magnetic
field measurements using magneto-optical effects. Table 11.3 shows Verdet
constant values for a few magneto-optic materials.
Usually Faraday crystals show sensitivity to ambient temperature since
the Verdet constant is sensitive to the crystal temperature. In addition,
some of the magneto-optic materials also exhibit electro-optic (Pockels)
effect. Therefore, in the choice of a suitable Faraday material, these factors
have to be considered. Usually materials with large constant but having
less sensitivity to the temperature and smaller rp (Pockels coefficient) co­
efficients are preferable. Thus, several factors such as frequency response,
temperature effect, effect of noise and vibrations on the sensors, and errors
caused by neighboring conductors on the measured values are some of the
important considerations in the selection and use of magneto-optic trans­
ducers [22,23,31,32].

11,7.2 Magnetic Field Sensors

A practical magnetic field sensor will have polarized light going through
a Faraday cell which is subjected to the magnetic field to be measured.
The optical output is then passed on to an analyzer. A linear relation be­
tween sensor output voltage and the applied magnetic field can be produced

Table 11.3 Verdet Constant for Some

Magneto-Optic Materials

Material K, (rad/A)

Optical glass (SF-57) 3.32 X 10-=

Bismuth silicate (BSO) 3.66 X 10-=
Bij2Si02o
ZnSe 5.48 X 10-=
Terbium gallium garnet (TGG) 8.2 X 10-=
Y3Fe50„ (YIG) 3 X 10-=
New Measurement and Diagnostic Technologies 327

as clear from equation (11.15). Figure 11.9 shows the structure of a prac­
tical magnetic field sensor which utilizes two Faraday crystals, i.e., a BSO
crystal with a Verdet constant of K^j and a ZnSe crystal with a Verdet
constant of K^ 2 [22]. The overall rotation angle NO in this case will be
given by:
NO = (11.17)
where Lj, L2 = lengths of the two crystals and Hj and H2 = associated
magnetic field values. This sensor provides a linear output with applied
magnetic field over the range of 0 < H < 8 kA/m with a sensor output
voltage of —0.6 V per 1 kA/m of the applied magnetic field. Since BSO
crystal is also a Pockels sensor, such an arrangement can lead to devel­
opment of an integrated voltage and current measurement system.

11.7.3 Optical Current Transducers

Optical current transducers (OCTs) are finding extensive applications in
power systems because of their several advantages over the conventional
current transformers (CTs). There is a large variety in the available OCT
designs, ranging from conventional CT with optical readout to the all-
optical sensors. The IEEE Working Groups on Emerging Technologies and
Fiber Optic Sensors [33] provide a comprehensive review about the various
available instruments and their experience with their use in the power sys­
tem. It has also provided a list of key references in OCT technology.
There is a growing interest to develop a single optical transducer for
both voltage and current measurements in high voltage networks [22,23].
In future, it is expected that such transducers will be used extensively in
key high voltage components for monitoring and diagnostic purposes.

M agnetic-field

Figure 11.9 Structure of a magnetic field sensor. (From Ref. 22 © IEEE, 1987.)
328 Chapter 11

11.8 MEASUREMENTS OF VERY FAST TRANSIENTS IN GIS

Electrical transients in gas-insulated switchgear (GIS) result from either

breakdown or partial discharge. Phase-to-ground or phase-to-phase break­
down can occur during the operation of disconnect switches or circuit
breakers. The breakdown generated transients have rise times of the orders
of 5-10 ns. Due to the coaxial construction and low loss attenuation char­
acteristic of a GIS (10 dB/km at 1 GHz), these internal surges can prop­
agate around the GIS installation with very little attenuation [34]. Hence,
various reflections of the transients can occur which can lead to overvol­
tages. In addition, at the station terminals, the transients can couple into
the GIS enclosure ground system, secondary wiring as well as the control
circuitry. Therefore, there is a strong interest to measure and investigate
such transients. Partial discharges in GIS can result from a number of
sources such as free conducting particles, solid dielectric components,
floating components and conductors with surface defects, all of which gen­
erate transients with <1 ns rise times. Since partial discharge induced tran­
sients are indicative of incipient faults, their early detection and location
is of considerable importance.
For the measurement of voltage induced transients, external voltage
dividers are usually inadequate and specially designed probes have to be
included in the metal enclosure of the GIS. Such a probe is normally a
capacitive voltage divider which is placed close to the point where the
transient voltage is to be measured, since the voltage may differ along the
length of the GIS due to propagation effects [8]. The probe normally con­
sists of a plane metal electrode which is placed close to, but is insulated
from, the grounded enclosure. The stray capacitance of ~1 pF between the
line conductor and the probe forms the high voltage arm of a voltage
divider. The low voltage arm of the divider consists of some tens of nan­
ofarads capacitance arranged coaxially between the probe and the enclo­
sure and can be obtained by different methods. The probe is placed in a
region of low electric stress to maintain the dielectric integrity of the GIS
and is connected to the recording instrument by a coaxial cable matched
at both ends. The cable length is kept as small as possible [8]. It is properly
calibrated to ensure divider ratio is accurately known at high frequencies
as well [35].
For partial discharge measurement, coaxial probes or coaxial capacitive
voltage dividers, with flat response from 10 Hz to more than 1 GHz, have
been successfully designed and built with a wide dynamic range (divider
ratios ranging from 3:1 to 2000:1) in order to investigate various sources
of partial discharges in GIS [36,37]. Such dividers can also be used to
record breakdown induced transients of modest magnitudes.
New Measurement and Diagnostic Technologies 329

11.9 SPACE CHARGE MEASUREMENT TECHNIQUES

Space charge is formed in stressed dielectrics and plays an important role

in the breakdown and DC leakage current behavior in polymers. Further­
more, the formation of space charge is believed to have a strong influence
on electrical trees initiation under various conditions such as DC voltage
application, polarity reversal and short circuits. The formation and accu­
mulation of space charge in XLPE power cables may distort the electrical
stress distribution in the insulation, thereby causing excessive stresses that
may result in breakdown under some adverse conditions. Therefore, the
effect of space charge must be considered in the design and testing of high
voltage cables and other solid insulation. For this reason, many efforts have
been made during the last decade to develop nondestructive techniques of
space charge measurements. The aim of such measurements is to under­
stand the processes that take place in the dielectric under study and make
it possible to select materials and interfaces that minimize the risk of break­
down in HV applications, thereby improving the performance and relia­
bility of the insulation system.
Previously, thermally stimulated currents (TSCs) were measured to
gain an insight into the space charge accumulation in dielectrics. The TSC
technique involves the polarization of the dielectric by a static field at a
high temperature. The sample is then cooled quickly to a much lower
temperature in the presence of the external field. The external field is then
removed, the electrodes of the sample are short circuited to allow it to
discharge isothermally for a suitable period of time, and then the sample
is warmed at a constant rate. A current corresponding to dipole relaxation
is recorded as a function of temperature [38]. A plot of this current with
temperature is the thermogram which may contain current peaks related to
molecular relaxation process and delocalization of charges injected in the
dielectric during the polarization process (Figure 11.10). The TSC spectra
thus obtained is then used to estimate the space charge distribution. Though
useful, TSC method is not sufficient to determine space charge distribution
fully.
To overcome the limitations of the TSC method, pulse methods for
space charge measurements were developed. Pulse methods of measuring
space charge are useful to find the spatial and temporal variation of space
charge in a stressed dielectric and have seen significant advances in recent
years. These methods can be broadly classified as follows [39]:

1. Pressure wave propagation (PWP) methods. The methods based

on this principle include piezoelectrically induced pressure step
(PIPS) method and laser induced pressure pulse (LIPP) method
330 Chapter 11

Figure 11.10 Thermally stimulated discharge current thermogram.

2. Pulsed electro-acoustic (PEA) method

3. Laser intensity modulation method (LIMM)
4. Thermal pulse (TP) and thermal step pulse (TSP) methods
Some of these techniques will be briefly discussed next.

11.9.1 Piezoelectric Transducers

Piezoelectric transducers (or electrets) used for space charge measurements
should have the piezoelectric characteristics such that when exposed to an
electric field, acoustic waves are generated in the transducers; by contrast,
when the pressure is applied, surface charges are induced on the transducer.
Thus, when an electric field e(t) is applied to the transducer of Figure
11.11a, the resulting acoustic wave p(t) is generated which is given as:

i(t)

g(z) h(z) q(t)

?(t)@
P(t) p( t ) D
transducer transducer

(a ) (b)

Figure 11.11 Piezoelectric transducers.

New Measurement and Diagnostic Technologies 331

P(t) g(z == vr)e(t - t) c1t (11.18)

where g(z) = électrostriction constant of the transducer, v = acoustic ve­

locity of the pulse and K = a constant. When a pulsed acoustic wave p(t)
is propagated through the transducer of Figure 11.11b, surface charges q(t)
are induced in the transducer which are given as:

q(t) = K f h(z = vr)p(t -

Jo
T)dr (H.19)

where h(z) = piezoelectric constant of the transducer, i.e., transfer function

of the transducer.

11.9.2 Charged Dielectrics

Polymer insulating materials possess piezo strain and stress characteristics,
though only with very small values. However, when a space charge is
accumulated in polymeric materials, the piezo strain and stress character­
istics of the charged material becomes large, as in an electret transducer.
Therefore, when a dielectric with trapped space charge p(z) is exposed to
an electric field e(t) in Figure 11.12a, the acoustic pulse p(t) thus generated
becomes [40]:

p(t) - K f p(z = vr)e(t - T)dr ( 11. 20)

Jo
where v = acoustic velocity of the pulse p(t).
Similarly, when a charged dielectric is exposed to an acoustic pulse
p(t), the position where the space charge is accumulated is compressed,
and a current i(t) flows through the external circuit of Figure 11.12b. This
current is given by [40]:

i(t)

Figure 11.12 Piezoelectric effects in charged dielectrics.

332 Chapter 11

i(t) = K f p(z = vr)p(t - r)dr ( 11.21)

Jo

If the duration of the acoustic wave is much narrower than the travel time
of the acoustic wave through the sample, i(t) is proportional to the accu­
mulated space charge. Thus, the current waveform shows the space charge
distribution.

11.9.3 Pressure Wave Propagation Technique

In this method, a pressure pulse p(t) is impressed upon the charged di­
electric, and the resulting current i(t) is measured as in Figure 11.12b. The
techniques used for generation of p(t) differ between the PIPS and the
LIPP methods. In the PIPS method, the pressure pulse is generated by
stressing a piezoelectric transducer with a narrow voltage pulse, as shown
in Figure 11.11a. However, in the LIPP method, a laser pulse of subna­
nosecond duration is irradiated to the target electrode of the dielectric sam­
ple. The evaporation of the electrode material under laser irradiation gen­
erates a pulsed acoustic wave which is used for charge measurements.
Using the PIPS method, the space charge distributions in dielectric
films of 50 ~ 200 /xm can be directly obtained if the duration of the applied
voltage pulse is about 1.25 —5 ns. Thus, nanosecond-duration high voltage
pulses are required for such measurements and circuits have been built to
generate such pulses [39,40] (see also chapter 10). The LIPP method is
applicable to both thin (10-100 jum) and thick (1 to —20 mm) specimens.
The time scale of the space charge measurements is related to the acoustic
wave transit time across the specimen. The duration of the pressure wave
(or that of the laser pulse) determines the resolution of spatial charge dis­
tribution. The main drawbacks of the LIPP method are the size and the
cost of the laser system required for pressure pulse generation.

11.9.4 Pulsed Electroacoustic Method

The pulsed electroacoustic (PEA) method has been used by many research­
ers in recent years and involves the applications of a pulsed electric field
across the dielectric with accumulated space charge. This causes a pulsed
force that generates a pulsed acoustic wave that is detected by using a
piezoelectric transducer. The induced charge q(t) on the surface of the
transducer is proportional to the space charge distribution in the dielectric
provided the duration of applied pulsed electric field At is much shorter
than the propagation time of the acoustic wave through the sample.
The space resolution, Al, in the PEA method is given as:
New Measurement and Diagnostic Technologies 333

A1 - vAt ( 11. 22)

where At = duration of the externally applied field, and v = velocity of

propagation of the acoustic wave in the sample and is —2000 m/s for low
density polyethylene (LDPE). For example, if a voltage pulse of 30 ns
duration is used in LDPE, the theoretical resolution of space charge mea­
surements is —60 /mm. In practice, since the acoustic wave broadens as it
propagates towards the transducer, the measurement resolution is —100 ¡jun
[41]. Thus, for better resolution a smaller value of At is required. Further­
more, in order to obtain q(t), which is linearly proportional to p(t), the
propagation time At for the detection transducer should also be much
shorter than the wave propagation time through the sample. The acoustic
wave can be detected using a polyvinylidene fluoride film (PVDF) trans­
ducer having 10 — 30 /mm thickness.
The PEA method has been used to study the formation of space
charges in dielectric films as well as in coaxial cable samples [39-42].
Figure 11.13 shows an example of output signal waveform showing spatial
charge distribution in LDPE obtained by the PEA method [40]. The pos­
itive space charge is distributed at the interfaces of the electrodes and the
sample film (thickness is 600 /¿m). These results were obtained from a
gamma-ray-irradiated sample after a DC voltage of 15 kV was applied.
As the negative charge is swept out, only the positive charge is left in the
sample. A 1 kV pulse voltage of 15 ns duration was also applied and the
output signal includes both the surfaces charges induced by the applied
DC voltage and the space charges [40]. Figure 11.14 shows the effect of
polarity of DC voltage on the space charge distribution in XLPE cable

Figure 11.13 Signal waveform showing spatial charge distribution in low density
PE. (From Ref. 40 © IEEE, 1994.)
334 Chapter 11

Figure 11,14 Polarity effect of DC voltage on space charge accumulation in

XLPE cable insulation. (From Ref. 41 © IEEE, 1995.)

sample [41]. In Figure 11.14a, 1 is the negative surface charge induced on

the cathode, 2 is positive surface charge induced on the anode, whereas 3
and 4 are positive and negative space charges accumulating in the XLPE.
When the voltage polarity is reversed, all these charges reverse polarities
as shown by the numbers 5, 6, 7 and 8 (marked on Figure 11.14b).

11.9.5 Other Techniques

Other techniques such as the laser intensity modulation method (LIMM)
or thermal methods (TP and TSP methods) can also be used to measure
space charge or electric field distribution in polarized dielectrics. TP and
TSP methods utilize the thermal expansion induced by illumination of a
laser or a flash lamp of one of the electrode surfaces. This expansion moves
through the dielectric specimen and induces a charge signal which is mea­
sured to determine the field distribution in the sample, the charges induced
on the electrodes and the total space charge [39]. The signals for thermal
methods are inherently slower than the acoustic-based method discussed
New Measurement and Diagnostic Technologies 335

earlier and, therefore, such methods are not suitable for studying transient
phenomena. Furthermore, these methods have less spatial resolution.
LIMM [43] is a useful method that has been successfully employed to
determine the spatial distribution of polarization and space charges in
PVDF electret and feroelectric ceramics but is more complicated as com­
pared to the PEA method.

11.10 ELECTRO-OPTICAL IMAGING TECHNIQUES

Photonic emissions associated with the initiation and propagation of elec­

trical discharges are often used to investigate the breakdown and prebreak­
down mechanisms in various dielectrics. The electro-optical devices com­
monly used for such applications include photomultipliers, image
intensifiers, and high speed image converter and streak cameras. The use
of such advanced technologies in discharge studies have improved our
understanding of the breakdown phenomena in long air gaps, compressed
gases, vacuum and liquid dielectrics. Laser-Schleiren and Shadowgraphic
techniques (see chapter 5) along with image intensifiers and high speed
cameras have been used to understand the development of streamers in
liquid dielectrics. In gaseous and vacuum dielectrics, use of such electro­
optic devices is helping researchers to investigate fast breakdown and flash-
over processes. Some recent examples of the uses of such electro-optic
imaging techniques are provided in the literature [44-48]. Figure 11.15
shows an experimental arrangement for photography of fast (ns) discharges
with speeds of —10^ cm/s on spacer surfaces in gases [46]. An example
of the use of photomultipliers in the early detection of discharges in SF^
insulated equipment is provided in Binns et al. [47]. Electro-optical and
commercial imaging equipment (camera, video recorders and monitors)
have been applied to study the growth of electrical trees in XLPE under
mechanical stress and strain [48]. A brief introduction to some of the elec­
tro-optical imaging devices commonly used in the investigations of dielec­
trics is provided next.

11.10.1 Photomultipliers
The photomultiplier is a very sensitive detector of radiant energy in the
ultraviolet, visible and near infrared regions of the electromagnetic spec­
trum. Figure 11.16 shows a schematic representation of a typical photo­
multiplier tube (PMT) and its operation. The basic radiation sensor is the
photocathode, which is located inside a vacuum envelope. On the incidence
of light at photocathode, the photo-electric emission occurs at the photo-
336 Chapter 11

HV IMPULSE GENERATOR

Figure 11.15 Schematic diagram for streak photography of spacer flashover.

PHOTOCATHODE

ANODE
New Measurement and Diagnostic Technologies 337

cathode resulting in the production of photoelectrons. These photoelectrons

are directed by an electric field to an electrode or dynode within the PMT
envelope. Dynode provides current amplification by emitting a number of
secondary electrons for each impinging primary photoelectron. These sec­
ondary electrons, in turn, are directed to a second dynode and so on, until
a final gain of —10^ is achieved. The electrons from the last dynode are
collected by an anode that provides the current signal which is then mea­
sured. Amplification ranging from 10^ to 10^ with extremely fast time
response, with rise times as short as a fraction of a nanosecond, makes
photomultiplier a useful detector. Important parameters of a PMT are its
spectral response, luminous responsivity (given in mA/lumen), gain, rise
time, maximum average anode current and anode dark current. For appli­
cations in the studies of breakdown and prebreakdown events, a broad
spectral response and a fast rise time (<1 ns) are desirable. Photomulti­
pliers are generally most suitable for detecting photon emission associated
with corona and prebreakdown types of partial discharges.

11.10.2 Image Intensifiers

An image intensifier directly intensifies electronically, images at extremely
low light levels. It is made up of a photocathode, an electron lens, micro-
channel plates (MCP) and a phosphor screen. When an optical image is
focused onto the photocathode (Figure 11.17), it emits electrons in ac­
cordance with the intensity of the input optical image, thereby converting
the input optical image into an electronic image. The electronic image is
then focused on the MCP where it is intensified, and then strikes the phos­
phor screen of the intensifier where an intensified optical image is repro­
duced. The result of this process is —10"^ to 10^ intensification of the orig­
inal incident light. The MCP normally used acts as a secondary electron

INPUT
IMAGE
INTENSIFIED
IMAGE

PHOSPHOR SCREEN
PHOTOCATHODE {ELECTRON»LIGHT)
MCP
(LIGHT^ELECTRON)
(ELECTRON
AMPLIFICATION)

Figure 11.17 Operation of an image intensifier.

338 Chapter 11

multiplier and consists of an array of millions of glass capillaries fused

into the form of a disk. When an electron enters and hits the capillary wall,
secondary electrons are produced from the wall. These electrons are then
accelerated by an electric field and strike the opposite wall to produce
additional electrons, thereby providing the overall image amplification. Be­
sides gain, spectral response and photocathode luminous sensitivity, the
other parameters that influence the performance of an image intensifier are
phosphor screen material, which influences phosphor’s spectral emission
and decay characteristics, image magnification, limiting resolution and dis­
tortion. For investigations of fast electrical discharge phenomena or to im­
age weak intensity discharges, image intensifiers play an important role.

11.10.3 Image Converters

An image converter is a device for converting an image formed by infrared
or ultraviolet radiation, which are invisible to the human eye, into a visible
light image. It employs a photocathode that is sensitive to infrared or ul­
traviolet radiation, and a phosphor screen that emits visible light. The spec­
tral response of the photocathode determines whether the image converter
is suitable for infrared or for ultraviolet radiation. If the photocathode has
a broad frequency spectral response, it may image cumulative effects of
ultraviolet, visible as well as infrared radiations. Image converters can be
used for a wide variety of diagnostics such as monitoring of power equip­
ment to locate hot spots, detecting ultraviolet radiations and studying gas
discharges.

11.10.4 High Speed Cameras

The streak camera is a device that measures ultrafast light phenomena and
delivers light intensity variation with time as well as with position and is
extremely useful for the investigation of discharge phenomena. Thus when
high speed cameras are used in combination with proper optics, it is pos­
sible to measure time variation of the incident light with respect to the
position of light, thereby enabling the measurements of leader and streamer
discharge speeds in liquids, gases or vacuum insulants. Such cameras usu­
ally offer streak and multiframing functions at high framing speeds.
In the streak mode, pictures are possible with a spatial resolution of
—0.5 n s /15 mm. In multiframing mode, —10^ frames per second allow
investigations of discharges with temporal resolution of the order of some
tens of nanoseconds. Both streaking as well as multiframing modes are
used in insulation related studies.
New Measurement and Diagnostic Technologies 339

In recent years there has been increasing tendency to use high speed
cameras in combination with charged coupled device (CCD) cameras along
with digital video and personal computers, as shown in Figure 11.15, which
allow digital image storage and retrieval. The use of digital imaging tech­
niques is expected to grow in the coming years for monitoring as well as
investigations of dielectrics.

REFERENCES

lEC Publication 1083-1, “Digital recorder measurements in high voltage im­

pulse tests. Part 1: Requirements for digital recorders,” 1991.
2. lEEE/ANSI Standard 1122-1987, “Digital recorders for high voltage impulse
measurements,” 1987.
3. G. C. Stone and J. Kuffel, IEEE Elect. Insul. Magazine, Vol. 5, No. 3, pp.
9-17, 1989.
T. R. McComb, J. Kuffel and R. Malewski, IEEE Trans, on PWRD, Vol. 2,
No. 3, pp. 661-670, 1987.
5. T. R. McComb, J. Kuffel, R. Malewski and K. Schon, IEEE Trans, on PWRD,
Vol. 5, No. 3, pp. 1256-1265, 1990.
6. J. Kuffel, R. Malewski and R. G. van Heeswijk, IEEE Trans, on PWRD, Vol.
6, No. 2, pp. 507-515, 1991.
T. R. McComb, C. Fenimore, E. Gockenbach, J. Kuffel, R. Malewski, K.
Schon, L. Van der Sluis, B. Ward and Y. X. Zhang, IEEE Trans, on PWRD,
Vol. 7, No. 4, pp. 1800-1804, 1992.
8. H. M. Ryan (ed.). H ig h v o lta g e e n g in e e r in g a n d te s tin g , Peter Peregrinas Ltd.,
London, England, 1994.
9. B. A. Lloyd, B. K. Gupta, S. R. Compbell and D. K. Sharma, IEEE Trans,
on PWRD, Vol. 4, No. 2, pp. 932-936, 1989.
10. R. Malewski and B. Poulin, IEEE Trans, on PWRD, Vol. 3, No. 2, pp.
476-489, 1988.
11. E. Hanique, IEEE Trans, on PWRD, Vol. 9, No. 3, pp. 1261-1266, 1994.
12. F. H. Kreuger, P a r tia l d is c h a r g e d e te c tio n in h ig h - v o lta g e e q u ip m e n t. Butter-
worths & Co. Publishers, London, England, 1989.
13. B. H. Ward, IEEE Trans, on PWRD, Vol. 7, No. 2, pp. 469-479, 1992.
14. R. J. Van Brunt, IEEE Trans, on Elect. Insul., Vol. 26, No. 5, pp. 902-948,
1991.
15. N. H. Malik and A. A. Al-Arainy, IEEE Trans, on Elect. Insul., Vol. 22, No.
6, pp. 825-829, 1987.
16. R. Bartnikas, Discussion published in IEEE Trans, on PWRD, Vol. 7, No. 2,
pp. 477-478, 1992.
17. “Computer aided control system for impulse test plant,” Bulletin E l37,
Haefely, Basel, Switzerland.
18. “Field measuring systems,” Bulletin E142, Haefely, Basel, Switzerland.
340 Chapter 11

19. K. Hidaka, IEEE Elect. Insul. Magazine, Voi. 12, No. 1, pp. 17-28, 1996.
20. N. A. Jaeger and E. Rahmatian, IEEE Trans, on PWRD, Voi. 10, No. 1, pp.
127-134, 1995.
21 . 5. J. Huang and D. C. Erickson, IEEE Trans, on PWRD, Voi. 4, No. 3, pp.
1579-1585, 1989.
22 . T. Metsui, K. Hosoe, H. Usami and S. Miyamoto, IEEE Trans, on PWRD,
Voi. 2, No. 1, pp. 87-93, 1987.
23. A. Cruden, Z. J. Richardson, J. R. McDonald and I. Andonovic, IEEE Trans,
on PWRD, Voi. 10, No. 3, pp. 1217-1223, 1995.
24. T. Takada, Y. Zhu and T. Maeno, IEEE Elect. Insul. Magazine, Voi. 12, No.
2, pp. 8-20, 1996.
25. T. J. Englert, B. H. Chowdhury and E. Grigsby, IEEE Trans, on PWRD, Voi.
6, No. 3, pp. 979-985, 1991.
26. T. R. McComb, E. A. Chagas, R. C. Hughes, G. Rizzi and K. Schon, Electra,
No. 161, pp. 105-119, 1995.
27. R. E. Hebner, R. A. Malewski and E. C. Cassidy, Proc. of IEEE, Voi. 65, No.
11, pp. 1524-1548, 1977.
28. T. Maeno and T. Takada, IEEE Trans, on Elect. Insul., Voi. 22, No. 4, pp.
503-508, 1987.
29. M. Zahn, IEEE Trans, on Dielectrics and Elect. Insul., Voi. 1, No. 2, pp.
235-246, 1994.
30. R. Shimizu, M. Matsuoka, K. Kato, N. Hayakawa, M. Hikita and H. Okubo,
IEEE Trans, on Dielectrics and Elect. Insul., Voi. 3, No. 2, pp. 191-196, 1996.
31. J. Song, P. G. McLaren, D. J. Thomson and R. L. Middleton, IEEE Trans, on
PWRD, Voi. 10, No. 4, pp. 1764-1771, 1995.
32. Y. Yamagata, T. Oshi, H. Katsukawa, S. Kato and Y. Sakurai, IEEE Trans,
on PWRD, Voi. 8, No. 3, pp. 866-873, 1993.
33. IEEE Working Groups on Emerging Technologies and Eiber Optic Sensors,
IEEE Trans, on PWRD, Voi. 9, No. 4, pp. 1778-1788, 1994.
34. S. M. Ghufran Ali and W. D. Goodwin, lEE Power Engineering Journal, Voi.
2, No. 1, pp. 17-26, 1988.
35. “Monograph on GIS very fast transients,” CIGRE, Paris, Prance, 1989.
36. S. A. Boggs, G. L. Pord and R. C. Madge, “Measurements of transients
potentials in coaxial transmission lines using coaxial dividers,” in M e a s u r e ­
m e n t o f e le c t r i c a l q u a n titie s in p u l s e p o w e r s y s te m s , NBS, Washington, DC,
1981.
37. S. A. Boggs, G. L. Pord, and R. C. Madge, IEEE PES Winter Meeting Paper
81 WM 139-5, IEEE, New York, 1981.
38. C. Lavergene and C. Lacabanne, IEEE Elect. Insul. Mag., Voi. 9, No. 5, pp.
5-21, 1993.
39. T. Mizutani, IEEE Trans, on Dielectrics and Elect. Insul., Voi. 1, No. 5, pp.
923-933, 1994.
40 . Y. Li and T. Takada, IEEE Elect. Insul. Mag., Voi. 10, No. 5, pp. 16-28,
1994.
New Measurement and Diagnostic Technologies 341

41. X. Wang, D. Tu, Y. Yanaka, T. Muronaka, T. Takada, C. Shinoda and T.

Hashizumi, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 2, No. 3, pp.
467-474, 1995.
42. Y. Li, M. Yasuda and T. Takada, IEEE Trans, on Dielectrics and Elect. Insul.,
Vol. 1, No. 2, pp. 188-195, 1994.
43. D. K. Das Gupta and J. S. Hornsby, IEEE Trans, on Elect. Insul., Vol. 26,
No. 1, pp. 63-68, 1991.
44. B. M. Coaker, N. S. Xu, R. V. Latham and F. J. Jones, IEEE Trans, on
Dielectrics and Elect. Insul., Vol. 2, No. 2, pp. 210-217, 1995.
45. O. Linhjell, L. Lundgaard and G. Berg, IEEE Trans, on Dielectrics and Elect.
Insul., Vol. 1, No. 3, pp. 447-458, 1994.
46. S. Y. Li, K. D. Srivastava and G. D. Theophilus, IEEE Trans, on Dielectrics
and Elect. Insul., Vol. 2, No. 1, pp. 114-120, 1995.
47. D. F. Binns, A. H. Mufti and N. H. Malik, IEEE Trans, on Elect. Insul., Vol.
25, No. 2, pp. 405-414, 1990.
48. E. David, J. L. Parpal and J. R Crine, IEEE Trans, on Dielectrics and Elect.
Insul., Vol. 3, No. 2, pp. 248-257, 1996.
12
Insulation Testing

12.1 OBJECTIVES OF TESTING

Electrical insulating materials are used in various forms to provide insu­

lation for high voltage power networks. These materials should possess
good insulating properties over a wide range of operating parameters since
the reliability of the system depends on their insulation quality. High volt­
age tests are applied to determine the ability of the insulation to meet its
design requirements. High voltage tests can be destructive or nondestruc­
tive. The destructive tests are used to measure the dielectric strength of
insulation and are made on sample pieces of the material or apparatus.
Nondestructive tests are carried out to ensure that the quality of insulation
is adequate for the required service conditions. In this case, the insulation
is not exposed to excessive test voltages. The customer views the testing
as an examination of the ability of the system or apparatus to meet the
guaranteed specifications of dielectric properties. Through the insulation
testing, a manufacturer wants to check the design and to ensure the suit­
ability of the selected materials. A researcher may use these tests and others
to develop better insulating materials and superior equipment design.
Normal operating voltages overstress the system insulation only under
special circumstances, such as contamination of external insulation. Power
system equipment has to withstand its operating voltage in addition to
occasional transient overvoltages. These transients may have external or
internal origins. External overvoltages are associated with lightning dis­
charges and are not dependent on the system’s operating voltage level. As

342
Insulation Testing 343

a result, the importance of stresses produced by lightning decreases as the

operating voltage increases. Internal overvoltages are generated by switch­
ing operations and due to occurrence of faults on the power system net­
work. These are called switching surges, and their magnitude depend on
the system rated voltage. Basic insulation level (BIL) and basic switching
level (BSL) depend on the levels of lightning and switching surges, re­
spectively. The designed values of BIL and BSL, which depend on the
system operating voltage, determine the insulation dimensions of the
equipment.
In the insulation design, the areas of specific importance are (1) de­
termination of the voltage stresses that the insulation must withstand and
(2 ) determination of the response of the insulation when subjected to such
voltages. The relationship between the electric stresses imposed on the
insulation and the dielectric strength forms the framework of insulation
coordination. Another requirement for the satisfactory performance of in­
sulation is the minimization of internal voids, as their presence may lead
to internal discharges and an ultimate breakdown. Partial discharge (PD)
tests are carried out to help in a proper design and selection of void-free
insulating components for power system applications.
Sometime tests are also required to ensure the safety of operating per­
sonnel in areas where a buildup of hazardous high voltages is possible.
Potential examples of such cases are found in oil industry and as streaming
electrification in oil-filled transformers.

12.2 HV TEST CLASSIFICATION

There are several ways of classifying the high voltage tests. The tests may
be destructive or nondestructive. They may be named in association with
type of the HV used, i.e., alternating voltage, direct voltage or impulse
voltage tests. These may be further classified according to the type of
measurement, such as a volt-time characteristic. Some of the tests are re­
quired during the production of insulating materials or equipment, whereas
the others are carried out after equipment production and/or installation.
These include the routine, type and special tests. Figure 12.1 shows a
schematic diagram of the main HV test classifications.
Destructive testing is associated with non-self-restoring solid insulat­
ing media. The nondestructive tests can be performed to find the insulation
quality of any material, even after it forms part of an equipment. Such
tests are mainly carried out to assess the electrical properties, such as DC
resistivity, the dielectric constant and loss factor over a wide range of
operating parameters. The quality of insulation is usually assessed by mea-
344 Chapter 12

Figure 12.1 Main high voltage (HV) test classifications.

suring the loss factor at high voltages and also by conducting partial dis­
charge tests to detect any deterioration or internal faults in the insulation
of the apparatus.
Standard tests can be classified as follows:
1. Routine test: This test is carried out by the manufacturer on every
unit produced. It can be part of the manufacturing process. An
example is the power frequency overvoltage test for insulation,
i.e., 1 min power frequency voltage withstand test.
2. Type test: This test is performed by the manufacturer for a few
samples to check if the new product meets certain specifications.
Users may ask for such a test on selected samples. An example is
the lightning impulse voltage test on oil-filled distribution trans­
formers.
3. Special test: As the name implies, this is a special test which is
performed to check insulation behavior under certain extreme con­
ditions, e.g., “measurement of acoustic sound level” for oil-filled
distribution transformer.
The range of high voltage tests depends on the nature of the equipment or
system being tested, but generally they might involve the following:
1. Short-time dry/wet withstand/flashover test
Insulation Testing 345

2. Induced overvoltage withstand test

3. Full/chopped wave impulse voltage withstand test
4. tan 8 (or loss angle) test
5. Temperature rise test
6 . Electrochemical test
7. Electromechanical test
8 . Mechanical test
9. Puncture test
10. Porosity test
11. Aging test
12. Pollution test
13. Radio influence voltage (RIV) measurement or corona related
tests (e.g., radio interference, corona loss and audible noise)
Usually standard tests and test procedures are defined by national and
international standards organizations for each equipment or component of
the HV network.

12.3 TEST VOLTAGES

The test voltages normally used can be divided into three main groups: (1)
direct voltages, (2 ) power frequency or low frequency alternating voltages,
and (3) impulse voltage, which are further divided into lightning impulses
and switching impulses. The generation and measurement techniques of
such test voltages were presented earlier in chapter 10. Tables 12.1 and
1 2 .2 list the recommended test voltages adopted for testing equipment hav­
ing rated AC voltage ranging between 1 and 765 kV [1]. The system
voltage levels can be broadly classified as shown in Table 1.1. The per­
formance of equipment rated at MHV, HV and EHV are verified by the
following types of tests:
1. The system behavior under AC operating voltages, temporary
overvoltages and switching overvoltages is generally checked by
a short duration (usually 1 min) power frequency voltage withstand
test.
2. For insulation aging and external insulation contamination, the
performance is generally checked by long-duration power fre­
quency voltage test.
3. The performance under lightning impulses is checked by lightning
overvoltage tests of equipment rated up to 300 kV. For equipment
with > 300 kV, such a test is complimented with or replaced
by the switching impulse voltage test.
346 Chapter 12

Table 12.1 Standard Insulation Levels for 1 kV < < 300 kV.
Insulation Testing 347

Table 12.2 Standard Insulation Levels for > 300 kV

12.3.1 Tests with Direct Voltages

For direct voltage tests, the test voltage should not contain AC components
corresponding to a ripple factor of more than 3% at rated current. It is
required that the rate of voltage rise above 75% of the estimated final test
voltage value, be about 2% per second [2]. The value of the test voltage
should be maintained within ± 1 % of the specified level thorough the test
if duration is < 60 sec, otherwise the tolerances are ± 3%.

12.3.2 Tests with Alternating Voltages

The alternating test voltage should be sinusoidal with a frequency normally
in the range 45-60 Hz. Its peak/V2 value should not differ from its true
rms value by more than 5%. Partial discharge should not significantly re­
duce the test voltage. This is usually achieved if the total HV circuit ca­
pacitance is of the order of 500 to 1000 pF and the current, with the test
object short circuited, is at 1A or higher [2]. For dry tests on small samples
of solid insulation or insulating liquids, a short-circuit current of 0.1 A
can be sufficient. However, for tests under artificial pollution, the re­
quired short-circuit current will be higher (as high as 15A) and depends
on the ratio of series resistance to the steady-state reactance of the
348 Chapter 12

voltage source, including the generator or supply network, at the test fre­
quency [2 ].

12.3.3 Tests with Impulse Voltages

The standard lightning and switching impulse voltages were defined in
chapter 10. According to IEC-60 [2], the rated impulse voltage withstand
tests are as follows:
1. For tests on non-self-restoring insulation, three impulses of the
specified polarity are applied at the rated withstand voltage level.
The requirements of the test are satisfied if no failure occurs.
2. For withstand tests on self-restoring insulation, the following two
methods are applicable:
Fifteen impulses of specified shape and polarity having peak
value equal to the rated withstand voltage are applied. The requi­
rements of the test are satisfied if not more than two failures occur.
The 50% breakdown voltage (V5 0 ) is determined. The test re­
quirements are satisfied if V5 0 is not less than 1/(1 - 1.3cr) time
the rated impulse withstand voltage, where a is the per unit stan­
dard deviation of the breakdown voltage. The value of the V5 0 and
a could be determined using either probability or up and down
methods [2]. Test procedures of these methods are described next.

P ro b a b ility M e th o d
In this method, n impulses (e.g., n = 20) of a fixed peak value are applied
at each test voltage level. The value of V5 0 is obtained from a curve of
breakdown probability versus corresponding applied voltage. Several volt­
age levels should be selected to cover the probability range and to ensure
acceptable accuracy which increases with an increase in the number of
voltage applications at each level.

Up a n d D ow n M e th o d
One impulse voltage having an amplitude V close to the estimated V5 0 is
applied. If this impulse cause a breakdown, the next applied impulse should
have the voltage V - AV, where AV is the voltage step and is approxi­
mately equal to 3% of V. If there is no breakdown at the level V, the next
impulse voltage should have a value of V + AV. This procedure is contin­
ued until 20-30 impulse voltages have been applied. If there are a total of
Insulation Testing 349

N applications with voltage for ith application, where i = 1, 2, 3, . . .

N, then:

( 12. 1)

and for normal distribution:

a = S (V, - V3o)V(n - 1) ( 12.2)

If a cannot be calculated, it can be assumed as 0.03 and 0.06 p.u. for

lightning and switching impulses, respectively.

12.4 TEST PROCEDURES AND STANDARDS

In the early days of the power industry, only national standards existed.
However, as time passed, it became necessary to establish international
standards to regulate the dealings with such apparati. Setting international
agreements on acceptance tests based on similar requirements can consid­
erably simplify approval of new equipment and enhance international co­
operation leading to mutual benefits to the industry and users in different
countries:
1. The manufacturer can produce HV apparatus in accordance with
international standards, and then they can easily sell their products
in any country
2. Equipment can be produced in larger quantities having unified
standard, and hence cost per piece is reduced
3. Users in any country can purchase equipment conforming to in­
ternational standards at competitive prices in the open market
In order to perform a certain test on a device or a component, a compre­
hensive standard procedure should be followed in order to ascertain the
reliability of the test. This includes:
Choosing the sample
Preparing the sample for test
Specifying the condition at which the test should be performed and
the correction procedure if different conditions prevail
350 Chapter 12

The detailed steps of executing the test

Data collection
Results, analysis and reporting
The above procedures are described in general standards and specific ap­
paratus standards. Several national and international standards exist for HV
testing. International standards are more general, because they try to ac­
commodate most of the possible cases and leave some of details to the
local or national standards. For example, the international standard may
leave the specifications of minimum and maximum ambient temperature
to the local conditions. The ambient temperature may range from -4 0 to
+ 30°C in Sweden, and from - 7 to +55°C in Saudi Arabia.
The following are some of the standard organizations which deal with
high voltage apparatus and systems.
1. International Electrotechnical Commission (lEC). This is the most
recognized international body dealing with standards in electrotechnology.
It publishes many standards on different electrical apparati, and devices.
lEC may have, for each HV apparatus, several standards outlining its spec­
ifications and the applicable methods for testing. In the next section, some
of these standards will be highlighted when discussing examples of HV
testing of power apparati. lEC, like any standards organization, updates its
standards over the years. There are several special committees within lEC
dealing with special subjects. One example of such committees is the In­
ternational Special Committee on Radio Interference (CISPR).
2. American National Standard Institute (ANSI). This is an American
body dealing with standards. In some cases, other countries may also have
to deal with ANSI standards due to their importance and common use of
the American manufactured products. In the range of medium and high
voltages, there are some common areas and some differences between
ANSI and lEC standards, and for this reason individual apparatus standards
should be consulated. On the other hand, there are commonalities between
the two organizations regarding EHV and UHV equipment standards,
since:
Such equipment and techniques were developed recently when the
Americans and the Europeans started to cooperate in these matters
The cost of manufacturing at these voltage levels is very high, therefore
such equipment manufacturing is limited to a few companies world
wide, and hence it becomes easier to correlate their activities
3. Institute of Electrical and Electronics Engineers (IEEE). The IEEE
was formed in the United States more than 100 years ago to enhance the
development of electrical engineering. Besides publishing technical jour-
Insulation Testing 351

nais and magazines, it also publishes several standards concerning electri­

cal apparatis. Since it is an international organization located in the United
States, it tries to adopt a “compromise approach” between ANSI and lEC.
4. British Standard Institute (BSI). This may be the oldest standard
organization that still exists. With the advent of lEC, which is European
in specihc and international in general, the publicity and importance of
British standards have started to decline—like that of other individual Eur­
opean standards.
5. Other standards-making institutes:
Verband Deutscher Elektrotechniker e.v. (VDE), Germany
Association Française de Normalisation (AFNOR), France
Canadian Standards Association (CSA), Canada
Japanese Industrial Standards (JIS), Japan
Table 12.3 lists the latest editions of some lEC standards dealing with HV
equipment.

12,5 TESTING OF HV MEASURING DEVICES

When performing various tests, a measuring system is employed to quan­

tify the measured items. A common practice for HV measurements is to
use a potential divider, connecting cable and LV measuring instrument (see
section 10.7 for details). The complete measuring system includes ah the
components and their interconnections and any auxiliary equipment con­
nected with the measuring system.
The LV meter must reproduce faithfully the required HV parameters
that are being measured. This is achieved if the potential divider is fre­
quency independent, the cable has low attenuation over a wide frequency
range, the LV meter has a suitable bandwidth and there exists good match­
ing between these three components. In order to check the suitability of
the measuring system, a test may be performed in order to determine the
following:
1. The maximum impulse amplitude and steepness, in addition to the
impulse duration for which the system is suitable
2. The accuracy with which the system can measure the impulse
amplitude, its front time and its tail time; in addition to the ability
of the system to detect and measure any oscillations of the impulse
3. The influence of the proximity of objects on the accuracy of the
measured values
Table 12.3 S om e lE C S tandards R elevant to T esting o f P ow er S ystem C om ponents

Standard no. Uj
IX )
and year of Impression (Imp.) or
HV apparatus latest edition Title amendment (Amd.)

1. General 60 High voltage test technique

60.1 (1989) General definition and test requirement 2nd Imp. (1991)
60.2 (1994) Test procedure
60.3 (1994) Measuring devices
60.4 (1977) Application guide for measuring devices 2nd Imp. (1987)
2. Transformers 76 Power transformers
76.1 (1993) General
76.2 (1993) Temperature rise
76.3 (1980) Insulation level and dielectric tests
76.3.1 (1987) External clearances in air 2nd imp. (1991)
76.4 (1976) Tapping and connections 3rd Imp. (1991)
76.5 (1976) Ability to withstand short circuit 3rd Imp. (1990), Amd. 1 (1979)
Amd. 2 (1994)
3. Cables 55 Paper insulated cable
55-1 (1978) General Amd. 1 (1989)
55-2 (1981) General and construction requirement Amd. 1 (1989)
141 Tests on oil-filled and gas pressure cables and their
accessories
141.1 (1993) Oil filled with up to 400 kV Amd. 1 (1995)
141-2 (1963) Internal gas pressure cables with up to 257 kV Amd. 1 (1967)
141-3 (1963) External gas pressure cables with up to 275 kV Amd. 1 (1967)
141-4 (1980) Oil, high pressure cables with up to 400 kV Amd. 1 (1990)
183 (1984) Guide to the selection of HV cables Amd. 1 (1990) s
230 (1966) Impulse tests on cables t
502 (1994) Extruded solid dielectric insulated power cables
(1-30 kV) IX )
840 (1988) Tests for power cables with extruded insulation Amd. 1 (1993)
30 kV < V < 150 kV
4. Circuit Breakers 56 (1987) HV AC circuit breaker Amd. 1 (1989), Amd. 2 (1995)
427 (1989) Synthetic testing of HV AC circuit breakers Amd. 1 (1992), Amd. 2 (1995)
5. Switchgear 694 (1980) Common clauses for HV switchgear and control Amd. 1 (1985), Amd. 2 (1993),
s
gear standards Amd. 3 (1995)
6. Switches 265 HV switches
265-1 (1983) HV switches for 1 kV < V„ < 52 kV Amd. 1 (1984), Amd. 2 (1994) Oq

265-2 (1988) HV switches for V„ > 52 kV Amd. 1 (1994)

420 (1990) HV AC switch fuse combination
7. Contactors 470 (1974) HV AC contactors Amd. 1 (1975)
8. Surge 99 Surge arrestor (SA)
Arrestors (SA) 99-1 (1991) Nonlinear resistor type gapped SA for AC system
99-2 (1962) Explusion type lightning arrestor
99-3 (1990) Artificial pollution testing of SA
99-4 (1991) Metal-oxide SA without gap for AC system Amd. 1 (1993)
9. Insulators 383 (1983) Tests on insulator of ceramic or glass for overhead
lines with V^^ > 1 kV
506 (1975) Switching impulse tests on HV insulators
507 (1991) Artificial pollution tests on HV insulators to be
used on AC system
10. Bushings 137 (1995) Bushings for AC with V„ > 1 kV
11. Fuses 282 HV fuses
282-1 (1994) Current limiting fuse Amd. 1 (1995)
282-2 (1995) Explusion fuse
12. HV 143 (1992) Series capacitors for power systems
capacitor 143-2 (1994) Protective equipment for series capacitor banks
13. Digital recorders for 1083 Digital recorders for measurements in HV impulse
measurement in HV tests
impulse test 1083-1 (1991) Requirements for digital recorders
Uj

V„ = rated systems voltage.

354 Chapter 12

Most of the factors affecting the performance of a measuring system can

be evaluated from the step response of the system. Step response is the
output of the system as a function of time when a step voltage is applied
between its high voltage and ground terminals. Once the step response of
a measuring system is known, the output of the system for any other shape
of input voltage can be determined.
The step generator must have zero impedance to obtain the true
step response of the measuring system. This is impossible because of the
large distance between the two input terminals. To minimize the two
input terminals are brought close together as much as possible and then
the step response is taken. This is not the true step response of the mea­
suring system, but in most cases will give a response very close to the true
one.
The quality of the measuring system is checked by the value of the
response time. When the response time is short, the system can measure
fast impulses accurately. From step response it is possible to determine
both the error in the measurement of the impulse peak amplitude and the
error in the measurement of the chopping time. The response time (T) is
essentially the area under the unit step minus the area under the normalized
unit step response. It has the dimension of time, since the X-axis is time
and the Y-axis is simply a relative numerical value. The response time can
be either positive or negative, depending on the shape of the step response.
If the step response is oscillatory or has an overshoot above the unit am­
plitude line, the response time will be negative if the sum of the negative
areas is greater than the sum of the positive areas. In the case of damped
type, the response time is simply the area between the step response and
the step input. Figure 12.2 shows damped and oscillatory types of step
responses. The algebraic sum of all of the areas in case of oscillatory type
is the response time, which in this case is negative. A negative response
time means that the measurement of a front-chopped impulse or a chopped
ramp will show a higher peak voltage and higher chopping time than the
actual values of these parameters.
If the rate of rise of impulse voltage is denoted by S and the chopping
time by T^, the true peak amplitude is S • T^. Provided that the chopping
of the impulse occurs in the region where the response is rising at the same
rate as the input, the measured value of the amplitude will be [S(T^ - T)],
since (T^ - T) is the measured time to chop. This quantity is greater than
T^ when the response time is negative, as in Figure 12.3 and therefore the
measured value of the peak amplitude will be larger than the true value.
The error in the measured value is clearly [S(-T)].
Insulation Testing 355

(a)

T=T i -T2+T3-T4+T5..

Figure 12.2 Evaluating the response time from a step response oscillogram: (a)
damped and (b) oscillatory response.

12.6 PARTIAL DISCHARGE TEST

Before discussing testing of various power apparati, a brief description will

be first presented for the two main nondestructive testing techniques, i.e.,
partial discharge (PD) and dielectric loss or tan d measurements. The basic
mechanism of PD and PD measuring schemes will be discussed here, and
tan 8 will be described in the next section. These nondestructive tests are
usually used to assess the electrical properties of dielectric materials over
a wide range of operating and environmental conditions. The term partial
discharge means the discharge is not complete between the two conducting
electrodes. If such a discharge occurs in the air, it is called a corona. Many

t(ns)
Figure 12.3 Integrals of the step response and unit step input of Figure 12.2.
356 Chapter 12

solid insulating materials contain gas voids (cavities) which may be bridged
by local discharges causing a PD. PD is considered one of the major rea­
sons behind the aging and eventual failure of solid insulating materials;
albeit there is no direct relation available between PD level and the ex­
pected life of the insulation as yet.

12.6.1 Partial Discharge Modeling

Development of partial discharge in a cavity enclosed in a dielectric slab
was explained earlier in section (8.4) with the help of Figure 8.5 and Figure
8 .6 . It was illustrated that when V is the applied voltage across the dielec­

tric, the voltage across the cavity will be given as:

V = V (12.3)
Q + Q
This voltage may cause void breakdown depending on the instantaneous
value of and the dielectric strength of air inside the void, since the
stress in the void (E^) may become much higher than the average stress
in the dielectric. The energy released in this discharge will cause deterio­
ration of the dielectric.

12.6.2 Partial Discharge Detection System

The observed effects of a PD—which include the energy loss, electric
current pulse, sound and electromagnetic radiation and changes in the ma­
terial properties—can be used for PD detection. Each effect is useful in
certain applications depending on the apparatus and the type of information
needed. In this section, details will be presented concerning the subject of
electric pulse detection since it is the most commonly used method for the
PD detection. For others the reader can refer to Krueger [3].
The electric pulse generated from a PD can be detected easily using
electrical detection arrangements. Figure 12.4 shows the two basic detec­
tion circuits which are termed as straight and balanced methods. The bal­
anced method is advantageous since in this case the external interference
is eliminated as it produces identical signals at the input of the differential
amplifier. The specimen capacitance in Figure 12.4 is assumed to be
lumped, and this assumption is valid for most of the power apparati. How­
ever, in cases like long cable and transformers, is a distributed parameter
and a special measure should be considered.
The detection system can be calibrated by injecting a pulse of known
charge magnitude to simulate the actual charge. The linear relation between
Insulation Testing 357
Filter
HV 1

Figure 12.4 Partial discharge detection circuits: (a) straight detection circuit and
(b) balanced method of discharge detection. = coupling capacitor and Z^,
and Z^2 ^re measuring impedances.

this known charge and the measured one is the calibration factor for the
instrument. If a square wave generator with an output is coupled to the
detection system through capacitor C, then the calibrating charge is
given as:
qo = CV„ (12.4)
There are other ways of calibrating the detection system, like measuring
well-known signals such as charges from point-plane corona or charges in
an artificial cavity [3].
Traditionally, the value of apparent charge (q^) is used as an indicator
for the PD level, q^ is measurable and it is related to the real charge q^ by
[3]:
Ch
fia "" fie (12.5)
Ch + c

Recently, the use of q^ as an accurate indicator for a quantitative evaluation

of PD has been questioned. Instead it is proposed that the discharge energy
be used as an appropriate measurable quantity to characterize PD inside
the dielectrics [4].

12.6.3 Partial Discharge Classification

In addition to PD detection, its source (e.g., cavity discharge, surface dis­
charge, corona) can be known through the PD classification. Research has
been carried out on various PD sources in order to identify the PD source
through the evaluation of typically measured discharge parameters. These
include pulse height distributions, PD frequency and location on the AC
cycle and statistical nature. From such measurements, PD classification can
358 Chapter 12

be attempted. In earlier studies, classification was performed by studying

the discharge patterns on the oscilloscope [5]. Table 12.4 shows typical PD
discharges, their sources, and the influence of time and voltage.
In recent years, digital signal processing techniques have been advan­
tageously used to automate the PD pattern recognition [6 - 8 ]. Krivda [8 ]
used the “fingerprints” derived from various PD data to discriminate be­
tween different PD sources, and then to classify these PD sources accord­
ingly. Commercial applications of such PD classification techniques are
expected in the future.

12.7 DIELECTRIC LOSS TEST

When a dielectric material is placed between two electrodes, a capacitor

is formed, and when a voltage is applied across this dielectric, a certain
amount of current (I) flows. Thus, the dielectric is represented by an
RC equivalent circuit. If a parallel RC equivalent circuit is used, tan 8 =
l/ioCpRp (ideally, Rp = oo). On the other hand, if a series equivalent circuit
is used, tan 8 = wC^R^ (ideally, R^ = 0). The energy dissipated as heat in
the dielectric is proportional to tan 5, and thus the value of tan 8 is an
important measure of the insulation quality. The maximum value of tan 8
is normally given in component or system specifications. The variation of
tan 8 with the applied voltage is an important relation which gives useful
information about the quality of the insulation and if an imperfection exists
in the insulation or not. As the voltage across the dielectric increases, the
value of tan 8 is roughly constant in the beginning; then, at some specific
voltage, it starts to increase appreciably with voltage indicating the incep­
tion of a PD. Similarly, the properties of the dielectric material change
with time and tan 8 measurements give some measure of this change. There
are many causes of changes in solid insulation, including changes due to
variations in temperature, mechanical stress, chemical reactions, water ab­
sorption and surface state deterioration. Similarly, changes in oil properties
take place due to gas and water absorption, chemical changes and intro­
duction of impurities. Periodic measurements of tan 8 can provide an in­
dication of the status of the insulation quality in such cases.
Capacitance and tan 8 are measured using bridges having two fixed
arms, consisting of the test object and a standard capacitance, and two
adjustable arms. The sensivity of the measuring bridge is the most impor­
tant and perhaps the limiting factor in low-frequency applications. If high-
frequency AC is used, the sensitivity is limited by the frequency-dependent
error of the resistance elements and the stray capacitance. Schering bridge.
 Table 12.4 Partial Discharge Display Patterns, Minimum Detectable Discharge (MDD), Discharge Extinction Voltage (DEV) and
Discharge Inception Voltage (DIV)
Pulse amplitude and inception voltage Effect of increasing the
Discharge source Discharge display characteristics voltage Effect of time

1. Corona from sharp point in Pulses have fixed amplitude and are No. of pulses increase, No effect.
air uniformly distributed around - v e then same will appear
voltage peak around the + ve peak.

2. Internal discharge in a solid DIV is well defined and above MDD, Little or no effect. The time of voltage application
dielectric bounded cavity pulses have similar amplitude al­ has little effect on the dis­
though differences of 3:1 from one charge pattern during single
side to the other are normal. test. Pulse amplitude and lo­
cation have some
randomness.

3. a. Several internal cavities Pulses are the same for +ve and - v e Pulses increase till becom­ No effect (at least up to 10
b. Discharge between two di­ half cycles, 1:3 differences in mag­ ing unresolved. min).
electric surfaces nitudes is normal. DIV is distinct
c. Surface discharge and above MDD, DEV^DIV.

4. Luminar cavity Pulses are the same in + ve and - v e Pulses are resolved at in­ If voltage is high, pulses mag­
half cycles. ception becoming rap­ nitude gradually increases
idly unresolved with in­ but becomes stable after
crease in voltage. The —10 min.
magnitude increases
steadily with increase in
voltage.

5. Gas bubbles in an insulating These can be random in magnitude Well-defined DIV with If voltage is fixed above DIV
liquid in contact with moist and location on the waveform. The steep rise in magnitude the magnitude of pulses in­
cellulose (e.g., oil impreg­ bubbles are generated by electric as voltage is increased. crease by 100 times in few
nated paper) field in moist cellulose. They in­ minutes. With extinction and
crease by discharges but dissolve in reinception DIV is about 3
liquid and disappear upon removing times less than the original
stress for a period of time. value. If restarting after 1
Oo
day, high DIV will be
restored.
Oo
Os
O

Table 12.4 Continued

6. Cavity between metal or Pulses are asymmetrical, i.e., unequal DIV is well defined and The time application (up to 10
carbon and dielectric in number and magnitude in the two the magnitude remains min) has little effect on the
half cycles. When the metal face at unaltered as voltage is display.
the cavity is grounded, a small num­ increased.
ber of pulses with large magnitudes
at —ve peak and large number with
low magnitudes at -hve peak occur.
When the metal is at HV, the situa­
tion will be reversed.

7. Gap-type discharge in Symmetrical pulses in magnitude and The pulses cover larger No effect.
nearby objects number. portion of the waveform
and no change in magni­
tude. DIV is well de­
fined above the noise
level and DIV = DEV

8. Tree growth Symmetrical pulses, the location can The pattern can be re­ If display is similar to A the
be erratic as can be the magnitude, solved at low voltage magnitude is likely to be
DIV is variable and higher than and become unresolved stable for minutes or longer.
DEV. Large and unrepeatable at high voltage. Pulse If it is similar to B then the
changes in magnitude may occur magnitude increases rap­ pattern is likely to be erratic
particularly at high voltage. idly with voltage. with rapid and substantial
changes in magnitudes in
few minutes.

S o u rce: Ref. 5.
Insulation Testing 361

shown in Figure 12.5, is the most widely used bridge for tan 8 measure­
ments. From Figure 12.5, it can be deduced that at balance:

C. = C ( 1 2 .6 )

C0 C4 R4 (12.7)
A variety of bridge circuits and accessories are available to suit the various
measuring conditions. To overcome the stray capacitance of the high volt­
age arm, a Wagner earth device is used in connection with the bridge [26].
In recent years, the measurements using the Schering bridge have been
automated to save time for the repetitive kind of measurements.

12.8 TESTING OF HV APPARATUS

In this section, the main HV tests performed on some of the power apparati
as per lEC standards will be outlined. First, the various tests for the indi­
vidual apparatus will be mentioned and then a brief description of some
of these tests will be provided.

12.8.1 Distribution Transformers

The following tests are required on most power transformers, as recom­
mended by relevant lEC publications [9,10]:

Figure 12.5 Schering bridge with Wagner earth arrangement to eliminate stray
capacitance. G = galvanometer, Z = additional arm impedence.
362 Chapter 12

1. Routine tests
a. Measurement of winding resistance
b. Measurement of voltage ratio and check of voltage vector
relationship
c. Measurement of impedance voltage, short-circuit impedance
and load loss
d. Measurement of no-load loss and current
e. Test on tap changers
2. Type tests
a. Dielectric tests
b. Temperature rise tests
Some special tests such as a short-circuit test, measurement of zero-
sequence impedance on three-phase transformers and measurement of
acoustic sound level may also be required. The two type tests mentioned
above will be discussed.

Insulation Level a n d D ie le ctric Tests

A transformer has both internal as well as external insulation. A failure in
the non-self-restoring internal insulation is catastrophic and normally leads
to the transformer being withdrawn from service for a long period. If the
transformer’s rated voltage is less than 300 kV, then only lightning and AC
voltage tests are required. However, switching voltage tests are also re­
quired when the system voltage is higher than 300 kV.
Impulse Testing of Oil-Filled Transformers
The purpose of the impulse tests is to determine the ability of the trans­
former insulation to withstand the transient voltages due to lightning. Since
the transients are impulses of short rise time, the voltage distribution along
the transformer’s windings is not uniform. The equivalent circuit of a trans­
former winding for impulses is shown in Figure 12.6. If an impulse wave
is applied to such a network, the voltage distribution along the element
will be uneven due to the existence of the capacitances Cg and C^. The
voltage stress across the winding is characterized by the following equation
if the neutral is isolated [ 1 1 ]:
ax
cosh
V(x) ( 12. 8)
cosh a
where a = VCg/C,. Thus the voltage distribution along the transformer is
not uniform, and the initial part of the winding (close to the line side) will
Insulation Testing 363

Line end Neutral

o
Cs
Hf- Hf- Hf-
/
Cg

Figure 12.6 Equivalent circuit of transformer winding under fast impulses.

have higher stress and thus it will be the first to breakdown in case of high
voltage surges.
Impulse test of an oil-filled transformer is usually performed using both
the full-wave and the chopped-wave impulses with chopping time from 2
to 6 p.s. To prevent large overvoltage from being induced in the windings
that are not under test, they are short circuited and connected to ground
through low impedance.
The schematic diagram showing connections for the impulse testing of
a three phase A/Y distribution transformer is given in Figure 12.7. Here,
winding UW is under test with full impulse voltage whereas windings UV
and VW are subjected only to half of the applied voltage. In the case of
Y/Y winding connections, each HV winding is tested separately. In trans­
former testing, it is essential to record the waveforms of the applied voltage
and the resulting current through the winding under test. Sometimes, the
transferred voltage in the secondary winding and/or the neutral current is
also recorded.
Following sequence of impulse voltage applications is used for oil-
filled transformers:

1. One full wave impulse voltage of reduced (50-75% of BIL)

magnitude
2. One full wave impulse of 100% BIL
3. One or more reduced chopped impulses
4. Two chopped impulses at 100% BIL
5. Two full wave impulses of 100% BIL

The fault in winding insulation is detected by general observations of noise,

smoke, etc., during voltage application. The inspection of voltage and cur­
rent oscillograms give more accurate indication of the failure, especially
364 Chapter 12

Impulse Generator

C R O -1

RC Voltage Chopping
Divider gap

Figure 12.7 Arranagement of transformer for impulse voltage test on terminal U

and winding UW.

the partial one. A partial or complete failure of winding appears as a partial

or complete collapse of the applied impulse voltage. However, the impulse
voltage may not show a small partial failure since the sensitivity of the
voltage waveform method is low and this method does not detect faults
which occur on less than 5% of the total winding. The failure detection is
enhanced by a current oscillogram, which usually shows a record of the
impulse current flowing through a resistive shunt connected between the
neutral and the ground or between the low voltage winding and the ground.
The current oscillogram usually consists of a high frequency oscillation, a
low frequency disturbance and a current rise due to reflections from the
ground end of the windings. When a major fault such as breakdown be­
tween turns or between one turn and the ground occurs, high frequency
pulses are observed in the current oscillogram and the waveshape changes.
For local failure such as a partial discharge, only high frequency oscilla­
tions are observed without a change of waveshape. To detect any failure,
voltage and current oscillographs for the full wave impulses are compared
with the initial reduced wave records. In addition, chopped impulses are
compared as well. The lEC test criterion is met if there are no significant
differences between the oscillograms corresponding to reduced and full
Insulation Testing 365

voltage applications. Figure 12.8 shows examples of the voltage and cur­
rent oscillograms corresponding to reduced and full voltage impulses. In
this figure, the three possible cases are displayed, i.e., no failure, partial
winding failure and complete failure (complete winding, external flashover
or breakdown between winding and the transformer tank).
Power Frequency Test
A voltage of specific value (see Table 12.1 and 12.2) is applied across each
of the transformer windings. The transformer is considered to pass the test
if no breakdown takes place during the test.

Tem perature Rise Test

The temperature rise values for the windings, core and oil of transformers
designed for operation at normal altitude (<1000 m) are specified by lEC
according to the cooling medium employed [8 ]. A reduced temperature
rise may be considered for transformers located at altitudes higher than
1000 m. The temperature rises are checked by subjecting the transformer
to simulated normal load condition for a specific time period. Details of
the measurement procedures and interpretation of the test results are spec­
ified in relevant standards [9].

12.8.2 HV Cables
Cables are subjected to electrical and thermal stresses while in service.
During their transportation, installation and repair, they are usually sub­
jected to mechanical stresses also which may cause insulation cracking or
produce voids, which in turn can act as sites of PD that may lead finally
to the breakdown of the cable insulation. Therefore, HV cables are sub­
jected to mechanical and electrical tests [12]. The following are the main
electrical tests as recommended in lEC standard.
1. Routine tests
a. Electrical resistance of the conductor
b. AC voltage withstand test
c. Partial discharge tests
2. Type tests. The following electrical tests are specified by lEC in
the sequence indicated:
a. Insulation resistance (R-J at room temperature
b. PD test
c. Bending test plus PD test (to check if the bending creates void
or not)
366 Chapter 12

Figure 12.8 Oscillograms of transformer windings subjected to impulse testing.

Upper traces represent applied voltage pulse, whereas lower traces are for current
flow. Cases (a) and (b) represent no failure; (c) and (d) indicate partial failure
whereas in (e) and (f) winding insulation has totally collapsed. Oscillograms (a),
(c) and (e) are for reduced full wave; whereas (b), (d) and (f) are for full wave.
Time scale = 1 |xs/div in (a)-(d) and 10 }xs/div in (e) and (f).

d. Tan 8 as function of voltage and capacitance measurements.

e. Tan 8 as function of temperature
f. Insulation resistance at operating temperature
g. Heating cycle test plus PD test
h. Impulse voltage withstand test followed by AC voltage test
i. HV alternating current test
In addition to the above, there are several nonelectrical type tests as well.
3. Special tests. There are several special tests such as conductor
examination, check of dimensions and hot set test.
For PD measurements and voltage withstand tests, samples have to be
prepared and terminated carefully. Otherwise, excessive leakage, corona or
end flashover may occur during testing. The normal length of the cable
Insulation Testing 367

samples used varies from 5 to 10 m. The terminations are usually made

by shielding the end conductor with stress shields to relieve the excessive
high electric stresses which would occur otherwise.

12.8.3 Circuit Breakers

Circuit breakers are tested to evaluate their constructional and operational
characteristics, in addition to check their performance in making and break­
ing the expected high currents. The main properties of a circuit breaker
are:
1. Electrical properties. These include the arc characteristics, the
current chopping characteristics and the shunting effects in inter­
ruption.
2. Nonelectrical properties. These include the type of media in which
the arc is extinguished, operation time, the number of operations
and the size of the arcing chamber.
In addition, the circuit performance characteristics should also be specified.
These include the degree of electrical loading, rated voltage, the type of
fault in the system which the breaker has to clear, the circuit time constant,
the time of interruption, the rate of rise of recovery voltage and the re­
striking voltage.
The following are the main tests specified for the circuit breakers [13]:
1. Routine tests
a. AC dry test of the main circuit
b. Voltage tests on control and auxiliary circuits
c. Measurement of the resistance of the main circuit
d. Mechanical operation tests
e. Design and visual checks
2. Type tests
a. Dielectric tests
b. Radio influence voltage (RIV) test
c. Temperature rise test
d. Measurement of the resistance of the main circuit
e. Mechanical and environmental tests
f. Making and breaking tests
g. Short circuit tests
Testing of circuit breakers requires highly equipped laboratory and sophis­
ticated testing procedures. A brief description is presented here for some
of the circuit breaker tests.
368 Chapter 12

M ech a n ica l Test

More than 80% of the total failures in circuit breakers are associated with
mechanical failures. To check its mechanical performance, the circuit
breaker is subjected to 2 0 0 0 operating cycles with no voltage or current in
its main circuit. During the test, replacement of any part and mechanical
adjustments are not permissible, though lubrication of the mechanical parts
is allowed according to the manufacturer instructions.

Tem perature Rise Test

When a circuit breaker is in service, the temperature of any part of the
breaker should not exceed the specified limits of temperature rise. These
limits vary for individual parts and for circuit breaker types. This test is
carried out in conditions similar to the normal operating conditions where
the normal rated current flows and the mounting is under normal service
conditions. If the maximum observed temperature rise is within the spec­
ified limits, the breaker is deemed to have passed the test.

D ielectric Test
This test is done to ensure circuit breaker can withstand the expected ov­
ervoltages within the power system. AC and impulse voltages are used for
the tests. The breaker under test should withstand the specified test voltages
without flashover or puncture. The specified AC voltage is applied for 1
min to indoor and outdoor breakers under dry or wet conditions. Positive
and negative polarity impulse voltages of specified amplitudes are applied
15 times. The test criterion is satisfied if the number of flashovers at either
polarity does not exceed 2 .

S h o rt-C ircu it Tests

Short-circuit tests are usually carried out according to test duties that spec­
ify the test current, the percentage of DC component, and the transient and
the power frequency recovery voltages. The purpose of the test is to ensure
that the circuit breaker is capable of making and breaking the circuit under
short-circuit conditions without damage to its components. This test is
subdivided into many tests, including the following:
1. Breaking and making current tests
2. Short-time current test
3. Operating sequence test
4. Single-phase short-circuit test
Insulation Testing 369

5. Out-of-phase switching test

6. Short-line fault test
7. Capacitor charging current breaking test
8. Small inductive current breaking test
The short-circuit tests need high-power testing plants since the circuit
breaker needs to break high current while the high voltage is applied. Such
high-power test facilities are very costly. An alternative solution is to use
synthetic method of circuit-breaker testing. It basically consists of two
independent voltage and current sources (Figure 12.9). The idea is to syn­
chronize momentarily the injection of high current and application of high
voltage. The current source injects the current into the circuit breaker under
test at a relatively reduced voltage. The voltage source injects a high volt­
age across the circuit-breaker contacts at the moment it is made to interrupt
the high current. The high voltage source usually contains a large capacitor
bank. The actual synthetic testing circuits would also Include elaborate
control and instrumentation schemes, in addition to the protection arrange­
ments in case of a breaker failure.

12.8.4 Switchgear
Switchgear normally contains circuit breakers, switches, busbars and other
metering and monitoring instruments. In addition to the tests performed on

Figure 12.9 Synthetic testing circuit for circuit breakers: C.B. = circuit breaker
under test, M.B. = main breaker, S.S. = synchronizing switch and O.R = over­
voltage protection.
370 Chapter 12

the c irc u it breaker, there are several o th e r tests c o n c e rn in g the p e rfo rm a n c e

o f the w h o le sw itchgear. T h e fo llo w in g are the m a in tests sp e cifie d fo r the
sw itch g e a r [1 4 ].

1. R o u tin e tests
a. A C d ry test
b. V oltag e tests on c o n tro l and a u x ilia ry c irc u it
c. M easu re m e n t o f the resistance o f the m a in c irc u it

2. T ype tests
a. D ie le c tric tests, w h ic h in c lu d e tests w ith lig h tn in g and s w itc h ­
in g im p u ls e voltages, A C vo lta g e , a rtific ia l p o llu tio n , P D and
A C tests on a u x ilia ry and c o n tro l c irc u its
b. R a d io in flu e n c e vo lta g e ( R IV ) test
c. T em perature rise test
d. M ea su re m e n t o f the resistance o f the m a in c irc u it
e. S h o rt-tim e w ith s ta n d c u rre n t and peak w ith s ta n d c u rre n t test.

A b r ie f d e sc rip tio n o f some o f the tests fo r s w itc h g e a r is p ro v id e d next.

Radio Interference (or Influence) Voltage (RIV) Test

A volta g e o f I . I V / V S sh a ll be a p p lie d to the s w itc h g e a r and m a in ta in e d
fo r at least 5 m in ( V is the rated vo lta g e ). T h e v o lta g e sh a ll then be de­
creased b y steps d o w n to 0 . 3 V / V 3 . T h e n the v o lta g e is raised again b y
steps to the in itia l value and fin a lly decreased b y steps to 0 . 3 V / V 3 . A t
each step, R IV m easurem ents are m ade and then R IV is p lo tte d against
the a p p lie d voltage. T h e e q u ip m e n t meets the test c rite rio n i f R IV at
1.1 V / \ ^ does n o t exceed 2500 pM.

Dielectric Tests
Tests are m ade on s w itc h g e a r and c o n tro lg e a r c o m p le te ly assem bled, as in
service and m o u n te d w ith m in im u m clearances and h e ig h t as s p e cifie d b y
the m anufacturer. L ig h tn in g im p u ls e , s w itc h in g im p u ls e ( i f rated vo lta g e
is m ore than 300 k V ) and p o w e r fre q u e n c y voltages are a p p lie d in a
m eth o d s im ila r to th a t m e n tio n e d e a rlie r fo r the c irc u it breakers.

12.8.5 Surge Arresters

Surge d iverters o r lig h tn in g arresters are used to p ro te c t the p o w e r system
com ponents against tra n sie n t o ve rv o lta g e due to lig h tn in g and s w itc h in g
surges. To ensure th e ir w o rk in g p e rfo rm a n ce , surge arresters m u st be tested
u sin g standard test procedures. Id e a lly , lig h tn in g arresters sh o u ld have in ­
Insulation Testing 371

fin ite resistance fo r o p e ra tin g p o w e r fre q u e n c y voltages and zero resistance

fo r tra n sie n t o ve rvo lta g e , thus d is c h a rg in g the heavy tra n sie n t c u rre n t, and
then re c o v e rin g its in s u la tio n a fte r the surge has fin is h e d w ith o u t a llo w in g
the p o w e r fre q u e n cy c u rre n t to co n tin u e . Surge arresters are id e n tifie d b y
th e ir rated volta g e , rated fre q u e n cy, n o rm a l discharge cu rre n t and class o f
lo n g -d u ra tio n discharge [1 5 ]. T h e fo llo w in g are the m a in tests sp e cifie d
fo r surge arresters.

1. R o u tin e test: D ry p o w e r fre q u e n c y sp a rko ve r v o lta g e test

2. T ype tests
a. P ow e r fre q u e n c y sp a rko ve r v o lta g e test
b. L ig h tn in g (and s w itc h in g ) im p u ls e sp a rko ve r vo lta g e test
c. V o lt-tim e cu rve fo r lig h tn in g im p u lse s
d. R e sid u a l vo lta g e test
e. C u rre n t im p u ls e w ith s ta n d test
f. O p e ra tin g d u ty test
g. Test o f arrester d isconne ctors

T he standard acceptance tests are d ry A C vo lta g e and lig h tn in g im p u ls e

tests. In some cases, re sid u a l v o lta g e test is added. These tests sh o u ld be
m ade on a n u m b e r o f sam ples equal to the h ig h e r w h o le n u m b e r to the
cube ro o t o f the n u m b e r o f arresters to be supplied . Som e o f the above
tests are described b rie fly below .

Power-Frequency Sparkover Tests

D ry and w e t tests are m ade on co m p le te arresters. T h e v o lta g e in itia lly
a p p lie d to the arrester s h o u ld be o f a lo w value in o rd e r to a vo id sparkover
o f its series gaps. T h e p e rm is s ib le tim e in w h ic h the a p p lie d v o lta g e m ay
exceed the rated v o lta g e o f the arrester is in the range 2 - 5 sec. A fte r
sparkover, the test vo lta g e is sw itc h e d o f f b y a u to m a tic trip p in g w ith in 0.5
sec. T he p o w e r-fre q u e n c y spa rko ve r v o lta g e is the average o f fiv e test
results.

Impulse Sparkover Tests

These tests are p e rfo rm e d on the same sam ples as those used fo r p o w e r-
fre q u e n cy sparko ve r tests. T h e proce d u re o f test fo r lig h tn in g im p u lse s is
the same as fo r c irc u it breakers discussed earlier.

Impulse Current Withstand Tests

T he rated vo lta g e o f these tests m u st be in the range o f 3 - 6 kV . Each
sam ple is subjected to tw o c u rre n t im p u lse s o f the standard shape w ith
J7 2 Chapter 12

peak im p u ls e values depe n d in g on the arrester class. B e fo re each test, the

sam ples m ust be at a p p ro x im a te ly the a m b ie n t tem perature. Subsequently,
the p o w e r-fre q u e n c y sp a rko ve r vo lta g e is de te rm in e d . I t sh o u ld n o t change
b y m ore than 10% . A ls o , no evidence o f p u n c tu re o r fla s h o v e r o f the
n o n lin e a r resistors o r s ig n ific a n t dam age to the series gaps o r g ra d in g c ir ­
c u it should occur.

12.8.6 Ceramic and Glass Insulators

In su la to rs are nam ed as ty p e A i f the p u n ctu re path is at least h a lf o f the
creepage le n g th , i.e., the shortest fla sh o ve r path th ro u g h a ir on in s u la to r
surface. O th e rw ise , these are ca lle d typ e B in su la to rs. In s u la to rs used on
overhead tra n sm issio n lin e s and in substations are subjected to ro u tin e
tests, typ e tests and sam ple tests, as described n e x t [1 6 ].

1. R o u tin e tests
a. T h e rm a l shock test
b. V is u a l e x a m in a tio n
c. M e c h a n ic a l test
d. E le c tric a l test (on some in s u la to r types)
2. T ype tests
a. D ry lig h tn in g im p u ls e vo lta g e w ith s ta n d test
b. W et s w itc h in g im p u ls e v o lta g e w ith s ta n d test
c. W et A C v o lta g e w ith s ta n d test
3. S am ple tests
a. V e rific a tio n o f lo c k in g system and d im e n sio n s
b. M e c h a n ic a l and e le c tro m e c h a n ic a l fa ilin g lo a d test
c. T h e rm a l shock test
d. P unctu re test
e. P o ro s ity test
f. G a lv a n iz in g test

Som e o f the above tests are described b rie fly below .

Dry Lightning Impuises Withstand Test

T he d ry in s u la to r is tested u n d e r p o s itiv e and negative lig h tn in g im pulses.
I t shall be adequate to test w ith one p o la rity i f it is e v id e n t th a t th is p a r­
tic u la r p o la rity w i l l g iv e lo w e r fla s h o v e r vo lta g e . T w o test procedures are
in c o m m o n use fo r the lig h tn in g im p u ls e w ith s ta n d test, i.e., the w ith s ta n d
procedure w ith 15 im p u lse s and the 50 % fla sh o ve r v o lta g e p ro ce d u re d is ­
cussed e a rlie r in section 12.3.3.
Insulation Testing 373

Wet Switching Impuise Withstand Test

T he in s u la to r is tested u n d e r p o s itiv e and negative 2 5 0 /2 5 0 0 ¡ms s w itc h in g
im p u ls e v o lta g e waves. T h e w e ttin g pro ce d u re is sp e cifie d in the standard.
T h e 50% fla sh o ve r pro ce d u re is n o rm a lly used. I f it is n o t po ssib le to use
th is procedure, the w ith s ta n d p ro ce d u re can be used.

Wet Power-Frequency Withstand Test

T h is test is a p p lic a b le o n ly to in s u la to rs fo r o u td o o r use. W ith in su la to rs
u n der w e t c o n d itio n s , a vo lta g e o f about 75 % o f the test v o lta g e sh a ll be
a p p lie d and then increased g ra d u a lly w ith a rate o f rise o f about 2 % o f
th is v o lta g e p er second. T h e test v o lta g e is m a in ta in e d at th is value fo r
one m in u te . I f no fla s h o v e r occurs, the in s u la to r is deem ed to m eet the test
requirem ents. F o r w e ttin g , the spray is arranged such th a t the w a te r drops
fa ll a p p ro x im a te ly at an angle o f 45° to the v e rtic a l. T h e test o b je c t is
sprayed fo r 15 m in in c lu d in g the a d ju stm e n t tim e b e fo re the vo lta g e ap­
p lic a tio n and the spray is c o n tin u e d fo llo w in g the v o lta g e a p p lic a tio n . T he
characteristics o f the p re c ip ita tio n are:

V e rtic a l and h o riz o n ta l com ponen ts = 1 to 1.5 m m /m in

L im its fo r in d iv id u a l m easurem ents = 0.5 to 2.0 m m /m in
T em perature o f c o lle c te d w a te r = a m b ie n t te m perature ± 15°C
C o n d u c tiv ity o f w a te r co rre cte d to 20°C = 100 ± 15 /x S /c m

12.8.7 Bushings
A b u shing is a stru ctu re c a rry in g one o r several con d u cto rs th ro u g h a
p a rtitio n such as w a ll o r ta n k and in s u la tin g it (th e m ) th e re fro m , in c o rp o ­
ra tin g the means o f a ttachm ent to the p a rtitio n . T h e b u sh in g in s u la tio n is
s o lid , liq u id , gas o r co m p o site . T h e fo llo w in g are the m a in tests sp e cifie d
fo r H V bushings [1 7 ]:

1. R o u tin e tests
a. Tan 8 and capacitance m easurem ents
b. D ry A C test
c. P D test
d. Tape in s u la tio n test
e. Pressure test fo r gas typ e bushing s
f. Pressure test fo r gas o r liq u id typ e bushings
g. T ig h tn e ss at fix in g devices
2. T ype tests
a. W e t A C test
374 Chapter 12

b. D ry lig h tn in g im p u ls e test
c. D ry o r w e t s w itc h in g im p u ls e test
d. T h e rm a l s ta b ility test
e. Tem perature rise test
f. T h e rm a l s h o rt-tim e cu rre n t w ith s ta n d test
g. D y n a m ic c u rre n t w ith s ta n d test
h. C a n tile v e r lo a d w ith s ta n d test
i. Tightness test fo r liq u id type

W et A C test, d ry lig h tn in g im p u ls e test and s w itc h in g im p u ls e tests are

p e rfo rm e d in m anners s im ila r to those discussed e a rlie r fo r in su la to rs.

12.8.8 High Voltage Fuses

A h ig h vo lta g e fuse is in tro d u c e d in the c irc u it to in te rru p t the c u rre n t in
case o f a fa u lt s im ila r to the L V fuse. H o w e ve r, it co n ta in s H V in s u la tio n
to isolate the thread o f the fuse fro m the o th e r parts o f the c irc u it. There
are several standard tests to be c a rrie d o u t to ensure the p ro p e r o p e ra tio n
and the q u a lity o f in s u la tio n o f H V fuses. T h e fo llo w in g are the typ e tests
recom m ended b y the re le va n t lE C standard [1 8 ]:

1. D ie le c tric test
2. T em perature rise test
3. B re a k in g test
4. Tests fo r tim e /c u rre n t cha ra cte ristics

T he m a in H V test is the d ie le c tric test w here the standard test voltages are
a p p lie d successively w ith one te rm in a l o f the o u tp u t o f the im p u ls e gen­
erator o r one p o in t o f the p o w e r fre q u e n c y source connecte d to the earth.
T he vo lta g e is a p p lie d :

1. B etw e e n the te rm in a ls and a ll earthable m e ta l parts

a. W ith the fuse in c lu d in g the fuse lin k and its fuse c a rrie r c o m ­
p le te ly assem bled and ready fo r service
b. W ith the fuse lin k and its fuse c a rrie r re m o ve d
2. B etw een te rm in a ls
a. F o r d ro p o u t fuses, the fuse c a rrie r s h o u ld be in the d ro p o u t
p o s itio n
b. F o r d is c o n n e c to r fuses, the fuse c a rrie r o r the fuse lin k sh o u ld
be rem o ve d fro m the fuse base

T he tests re q u ire d are lig h tn in g im p u ls e tests u n d e r d ry c o n d itio n s and

p o w e r fre q u e n cy tests u n d e r d ry and w e t c o n d itio n s . T h e procedures fo r
these tests are s im ila r to those d escribed fo r o u td o o r insu la to rs.
Insulation Testing 375

12.9 ELECTROSTATIC HAZARDS

T he p ro b le m o f e le c tro s ta tic charge b u ild u p concerns a w id e v a rie ty o f

hum an a c tiv ity ra n g in g fro m large-scale in d u s trie s to d o m e stic a c tiv itie s .
T he effects o f such a b u ild u p can cause m a jo r hazards w ith p o te n tia l loss
o f life and m oney. A fte r d e s c rib in g h o w the e le c tro s ta tic charges are gen­
erated and m easured, som e p ra c tic a l exam ples o f th e ir hazards w ill be
g ive n in c o n ju n c tio n w ith the means o f c o n tro llin g and p re v e n tin g such
hazards. Som e ty p ic a l cases o f the e le c tro s ta tic b u ild u p in p o w e r system
com ponents w ill be m e n tio n e d as w e ll.

12.9.1 Electrostatic Charge Generation and Measurement

E le c tro s ta tic charge b u ild u p and its p o te n tia l hazards can o c c u r i f the f o l­
lo w in g three c o n d itio n s p re v a il:

1. C h a rg in g o f the m a te ria l o r nearby structures occurs

2. T he leakage o f such charge is so s lo w th a t charge a c c u m u la tio n
takes place
3. T he c o n d itio n s fo r e x p lo s io n o r shock are fa vo ra b le

T he presence o f h ig h h u m id ity reduces the surface re s is tiv ity , and hence

the re la x a tio n tim e constant and co n se q u e n tly the danger o f e le ctro sta tic
hazards decrease w ith h u m id ity . A static charge is due to an excess o r a
d e fic ie n c y o f electrons. S tatic e le c tric ity is generated b y the co n ta ct and
separation o f d iffe re n t m a te ria ls. W h e n m a te ria ls are in contact, electrons
fro m one m a te ria l can m o ve across the in te rfa c e to the surface o f the o th e r
m a te ria l w here th e y w ill a lig n them selves since the firs t m a te ria l w ill n o w
have a s lig h t p o s itiv e charge due to the loss o f the electrons. In o rd e r fo r
the e le ctro sta tic charge b u ild u p to take place, one o r b o th m a te ria ls m u st
be in su la to rs. M a te ria ls d iffe r w id e ly in th e ir a b ility to g iv e o r receive
electrons on co n ta ct w ith o th e r objects. T y p ic a l trib o e le c tric series is show n
in Table 12.5, w h ic h serves as a u s e fu l ru le o f th u m b to p re d ic t the b e h a v io r
o f one substance w h e n ru b b e d w ith another. W h e n any tw o o f the m a te ria ls
on the lis t are ru b b e d together, the one h ig h e r in the lis t becom es p o s itiv e
w ith respect to the one lo w e r on the lis t. H o w e ve r, it ca nnot be used in ­
d is c rim in a te ly [1 9 ]. T h e nature o f the charges p ro d u ce d on a surface de­
pend on the e n v iro n m e n ta l c o n d itio n s , the p h y s ic a l c o n d itio n s o f the sur­
face and the presence o f im p u ritie s as w e ll. T h e value o f the discharge w ill
depend, in a d d itio n to the above, on the area o f co n ta ct and speed o f
separation.
376 Chapter 12

Table 12.5 Triboelectric Series

Positive End

Asbestos
Glass
Mica
Wool
Cat fur
Lead
Silk
Aluminum
Paper
Cotton
Wood
Sealing wax
Ebonite
Ni, Cu, Ag, brass
Sulphur
India rubber

Negative End

S o u rce: Ref. 19.

T he m echanism b y w h ic h the m a te ria ls re ta in the charge depend on

the value o f its re la x a tio n tim e w h ic h depends on the re s is tiv ity p and
the d ie le c tric constant e, o f the m a te ria l:

= ^o^rP (12.9)

T he value o f is d e te rm in e d b y the leakage paths a v a ila b le to the charge.

T he leakage path th ro u g h a ir depends on the presence o f io n s and the
possible presence o f io n iz in g m a te ria ls. T h e leakage p ath th ro u g h p ow ders
in v o lv e s the re s is tiv ity o f the pow ders, w h ic h is dependent on h u m id ity
and p a ckin g . A fte r the e le c tro s ta tic charge is generated and retained, a
p o ssible spark and thus fire a n d /o r e x p lo s io n can take place.
T he am o u n t o f charge o r the value o f the v o lta g e develop ed on the
surface o f an in s u la tin g b o d y can be m easured b y u sin g special m eters.
T he m ost co m m o n devices in th is regard are e le ctro sta tic v o ltm e te r, ele c­
troscope and the d iffe re n t k in d s o f tube-based m eters. T h e m a in c o m m o n
p ro p e rty o f these m eters is the extre m e h ig h value o f in te rn a l im pedance,
w h ic h is necessary to a vo id charge leakage d u rin g the m easurem ent.
Insulation Testing 377

12.9.2 Examples of Electrostatic Hazards

T he fo llo w in g are som e exam ples o f hazards w h ic h e x is t in in d u s try [1 9 ].

1. Aviation Industry. S ta tic charges are develop ed on an a irc ra ft due

to the p h y s ic a l co n ta ct o f the a irc ra ft in flig h t w ith a irb o rn e dust
and w a te r p a rticle s. S everal accidents have o ccu rre d d u rin g fu e l­
lin g op eration s because o f static discharges re s u ltin g fro m in a d e ­
quate b o n d in g and g ro u n d in g .
2. Flour and Grain Industry. M a te ria l m o v e m e n t b y means o f c o n ­
v e y o r belts and elevators can be re sp o n sib le fo r charge a ccu m u ­
la tio n . F in e p a rtic le s o f g ra in dust suspended in the a ir can act as
sources o f e xp lo sio n s.
3. Gas Industry. M o v e m e n t o f a gas th a t is c o n ta m in a te d w ith m e ­
ta llic oxid e s, o r liq u id p a rtic le s can pro d u ce e le c tro s ta tic charge.
4. Paper and Printing Industries. T h e e le c tro s ta tic charges can b u ild
up b y the m o v e m e n t o f the paper its e lf o v e r the va rio u s ro lls and
the m a c h in e ry o f m an u fa ctu re . In the presence o f fla m m a b le in k s
and solvents used in the p rin tin g process, e le c tro s ta tic charge b u il­
dups have caused m a n y fires and an o cca sio n a l e x p lo s io n .
5. Refining Oil Industry. E le c tro s ta tic charge has caused in the past
m any hazardous fires in re fin e rie s. E x te n s iv e p re ca u tio n s against
this hazard are im p o rta n t to a vo id its occurrences.

12.9.3 Electrostatic Charge Control by Grounding and Bonding

M a n y “ s ta tic ” p ro b le m s can be solved b y b o n d in g the va rio u s parts o f the
e q u ip m e n t tog e th e r and g ro u n d in g the e n tire system . H o w e v e r g ro u n d in g
in some cases is n o t enough. I f the m a te ria l b e in g processed is ra th e r b u lk y
and has h ig h re s is tiv ity , the charge on the up p e r p o rtio n o f the m a te ria l
w ill be e ffe c tiv e ly in su la te d fro m g ro u n d and m a y re s u lt in a spark d is ­
charge. G e n e ra lly the area o f co n ta ct w ith g ro u n d is an im p o rta n t param eter
in th is regard. T h e fo llo w in g m ethods are used to m itig a te the p ro b le m o f
e le ctro sta tic charge b u ild u p .

1. H u m id ity c o n tro l. W h e n the h u m id ity is increased the e le ctro sta tic

charge w i l l n o t be re ta in e d since charge leakage w i l l be increased
as e xp la in e d earlier. H o w e ve r, the h u m id ity sh o u ld be increased
o n ly to the le v e l w here i t does n o t h a rm the concerned process.
2. S tatic co lle c to rs v ia m e ta llic com bs.
3. N e u tra liz in g the e le ctro sta tic charges b y th e ir opp o site charges
p roduce d b y special devices c a lle d n e u tra lize rs.
378 Chapter 12

W h e n usin g one o f the above m ethods, care m u st be exercised b y m a in ­

ta in in g the relevan t in s tru m e n ta tio n and co n n e ctio n s th ro u g h p ro p e r p re ­
ve n tive m aintenance schedule. F a ilin g to p e rfo rm such a m aintenan ce m a y
lead to unpleasant consequences. A t present there are som e standards such
as lE C 801-2 [2 0 ] w h ic h o u tlin e the procedures fo r e le c tro s ta tic discharge
requirem ents o f in d u s tria l processes.

12.9.4 Electrostatic Buildup in Power Systems

In h ig h vo lta g e p o w e r system s, e le c tro s ta tic charge b u ild u p p ro b le m s are
in v o lv e d in some apparati such as sw itc h g e a r and large p o w e r tra n sfo rm e rs.
F o r exam ple, lE C 2 5 5 -2 2 -2 deals w ith e le c tro s ta tic discharge test o f ele c­
tro n ic re la y [ 2 1 ].
T he ele ctro sta tic charge can b u ild up in flo w in g in s u la tin g liq u id s as
described e a rlie r in section 5.4. In the e le c tric a l p o w e r in d u s try , these
charges can be fo u n d in larg e p o w e r tra n sfo rm e rs w ith fo rc e d o il c irc u la ­
tio n . T he need fo r h ig h e r p o w e r ra tin g s le d to the a p p lic a tio n o f new in ­
s u la tin g m a terials and increased o il flo w rates and hence m o re e le ctro sta tic
b u ild u p . T akagi et al. [2 2 ] re p o rte d th is p ro b le m in 1978 fo r the firs t tim e ,
w h ile C ro fts [2 3 ] e xa m in e d the tra n sfo rm e rs fa ilu re s caused b y e le ctro sta tic
charge b u ild u p due to o il c irc u la tio n . K e d z ia [2 4 ] used the e le ctro sta tic
charge b u ild u p as an in d ic a to r fo r the le v e l o f o il aging. M o re research
sho uld be d ire cte d to the p ro p e r m easurem ent o f the e le c tro s ta tic charge
b u ild u p and c o n tro l and m itig a tio n in large p o w e r tra n sfo rm e rs in p a rtic u la r
and p o w e r apparati in general. Peyraque et al. [2 5 ] suggested the use o f
the leakage c u rre n t values on the ta n k and in the w in d in g s as a means o f
m easuring the le v e l o f e le c tro s ta tic charge b u ild u p on tra n sfo rm e rs. H ence
the dangerous le v e l o f the charges can be a voided b y p ro p e rly m o n ito rin g
the leakage curren t. T h e y show th a t th is c u rre n t is s tro n g ly in flu e n c e d b y
tem perature, a p p lie d v o lta g e and o il flo w rate.

REFERENCES

lEC 71, “Insulation Coordination, Part 1: Terms, Definitions, Principles and

Rules and Part 2: Application Guide” , lEC, Geneva, Switzerland, 1976.
lEC 60, “High Voltage Testing Techniques” , Part 1: General Definitions and
Test Requirement, Part 2: Test Procedures” , lEC, Geneva, Switzerland, 1989
and 1994.
3. E. Kreuger, P a r tia l D is c h a r g e D e t e c ti o n in H ig h V o lta g e E q u ip m e n t, Butter-
worths, London, England, 1989.
4. J. Sletbak, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 3, No. 1, pp.
126-130, 1996.
Insulation Testing 379

5. D. Nattrass, IEEE Elect. Insul. Mag., Vol. 4, No. 3, pp. 10-23, 1988.
6. E. Kreuger, E. Gulski and A. Krivda, IEEE Trans, on Elect. Insul., Vol. 28,
No. 6, pp. 917-931, 1993.
R. Van Brunt, IEEE Trans, on Dielectrics and Elect. Insul., Vol. 1, No. 5, pp.
761-784, 1994.
A. Krivda, “Recognition of Discharges, Discrimination and Classification” ,
Ph.D Thesis, Delft University, Delft, The Netherlands, 1995.
9. lEC 76 Power Transformers, Part 1: General, Part 2: Temperature Rise, Part
3; Insulation and Dielectric Tests, and Part 5: Ability to Withstand Short
Circuit, 1993, 1993, 1980, and 1976.
10. lEC 76-3-1, “Power Transformers: External Clearances in Air” , 1987.
11. A. Greenwood, E l e c t r i c a l T r a n s ie n ts in P o w e r S y s te m , Wiley Interscience,
New York, 1971.
12. lEC 502, “Extruded Solid Dielectric Insulated Power Cables (1-30 kV)” , lEC,
Geneva, Switzerland, 1994.
13. lEC 56-4, “High Voltage Alternating Current Circuit Breakers,” lEC, Geneva,
Switzerland, 1987.
14. lEC 694, “Common Clauses for High Voltage Switchgear and Controlgear
Standards” , lEC, Geneva, Switzerland, 1980.
15. lEC 99-1, “Non Linear Resistor Type Gapped Surge Arrester for AC Sys­
tems” , lEC, Geneva, Switzerland, 1991.
16. lEC 383, “Tests on Insulators of Ceramic Material or Glass for Overhead
Lines with Nominal Voltage Greater than 1000 V ” , lEC, Geneva, Switzerland,
1983.
17. lEC 137, “Bushings for Alternating Voltages above lOOOV” , lEC, Geneva,
Switzerland, 1995.
18. lEC 282, “High Voltage Puses, Part 1: Current Limiting Fuse, and Part 2:
Expulsion and Similar Fuses” , lEC, Geneva, Switzerland, 1994 and 1995.
19. IEEE STD 142-1972, “IEEE Recommended Practice for Grounding” , 1972.
20. lEC 80-1, “Electromagnetic Compatibility for Industrial Process, Measure­
ment and Control Equipment, Part 1: Electrostatic Discharge Requirement” ,
lEC, Geneva, Switzerland, 1991.
21. lEC 255-22-2, “Current Relays, Part 22: Electrical Disturbances Tests for
Measuring Relays and Protection Equipment, Section Two: Electrostatic D is­
charge Tests” , lEC, Geneva, Switzerland, 1989.
22 . T. Takagi, I. Ishii, T. Okada, K. Kurita, R. Tamura and H. Murata, “Reliability
Improvement of 500 kV Large Capacity Power Transformer” , CIGRE Paper
No. 12-02, Paris, France, 1978.
23. D. Crofts, “The Static Electrification Phenomena in Power Transformers” ,
Proceedings of CEIDP, Claymont, Delaware, pp. 192-199, 1986.
24. J. Kedzia, IEEE Trans, on Elect. Insul., Vol. 24, No. 2, pp. 175-178, 1989.
25. L. Peyraque, C. Boisdon, A. Berouel and F. Buret, IEEE Trans, on Dielectrics
and Elec. Insul., Vol. 2, No. 1, pp. 40-49, 1995.
26. B. Hague, A lte r n a tin g - C u r r e n t B r id g e M e th o d s , 5th edition. Pitman and Sons,
London, England, 1959.
Index

Accelerated aging, 2 6 3 -2 6 6 A rc , 30, 71, 72

frequency, 263 industrial applications, 71
m u ltifa c to r stress, 265 interruption, 72, 9 6 -9 8 , 198
therm al, 265 re ig n itio n , 72, 9 6 -9 8
voltage, 263 A rc interruption in 9 6 -9 8
Acceptance tests, 344 A rc interruption in vacuum,
A F N O R , 351 197-202
A g in g o f insulation, 2 6 3-265 A rc in vacuum, 198, 199
A ir clearances, 69 anode spots, 198
conductor-conductor, 50, 56 cathode spots, 198, 199
conductor-cross arm, 56 diffused arc discharge, 198
conductor-ground, 50, 56 vacuum gap after the arc, 199
conductor-rod, 56 Arresters (see Surge arresters)
conductor-rope, 56 Askerels (PCBs), 6 , 114
conducto r-w in dow , 56 A tm ospheric conditions, effects on
A ir insulation, 4 9 -8 2 discharges, 51-6\(see also
applications, 5, 9, 10, 49, 50 Gas parameters)
m odeling, 49, 50 Attachm ent coefficient, 24, 83, 85,
A ir gaps, 9, 10 95
A lk y l benzene, 113, 116, 124 A u d ib le noise, 41, 73, 74
A N S I, 9 6 ,3 5 1 -3 5 2 Avalanche, 3 0 -3 4 , 37, 44, 8 6 , 87
A n tio x id a n t additives, 121 A vogadro number, 22
D B F , 121 Basic im pulse insulation levels
D B P C , 121 (B IL , B S L ), 106, 107, 343

38]
382 Index

B o h r’ s theories, 25 Breakdow n o f SF^, 85-95

B o ilin g point, 47 breakdown voltage calculation,
B oltzm ann constant, 3, 22 8 6 , 9 0 ,9 1
B oyle's law, 21 discharge parameters, 85, 8 6 ,
B reakdow n o f composite factors effecting breakdow n
dielectrics, 93, 94, 2 1 3 -2 2 0 voltage, 9 1 -9 5
cavity breakdown, 214 n o n u n ifo rm fie ld breakdown,
edge breakdown, 213 87-91
surface erosion, 217 u n ifo rm fie ld breakdown, 8 6 , 87
tracking, 217 Breakdow n in solid dielectrics,
B reakdow n fie ld strength, 5 -7 , 173-186
4 6 -4 8 , 8 6 , 95 avalanche breakdown, 175
B reakdow n in gases, 2 9 -4 6 , 8 5 -9 0 electrom echanical breakdown,
in long air gaps, 5 3 -5 9 174, 181
at m in im u m voltage, 35, 36 electronic breakdown, 174,
under nanoseconds pulses, 43, filam entary therm al breakdown,
44, 288, 289 180
in n onun iform fields, 3 5 -4 2 , im pulse therm al breakdown,
5 3 -57, 293 178, 180
in u n ifo rm fields, 29, 34, 35, in trin sic breakdown, 174,
8 6 , 87, 2 89-29 2 steady state therm al breakdown,
B reakdow n in liq u id dielectrics, 179
129-143 therm al breakdown, 177-181
breakdow n process at anode, B reakdow n strength, 5 -7 , 4 6 -4 8 ,
142, 143 8 6 , 95
breakdow n process at cathode, B reakdow n in vacuum, 192-198
138-141 factors effecting breakdown,
bubble theory o f breakdown, 192
133 vacuum breakdow n theories,
cavitation breakdow n theory, 195
132 anode theory, 197
electronic breakdow n theory, clum p theory, 195
131 cathodic theory, 196
post breakdown events in interaction theory, 196
liquids, 143 Breakdow n voltage, 21, 32, 35, 41,
suspended particle breakdown 4 2 ,5 1 -6 6 , 69, 87, 8 8 ,9 1 -9 5 ,
theory, 131 114, 1 1 5 ,2 8 9 -2 9 3
w ith optical techniques, w ith 0 % p ro b a b ility , 51
135-138 w ith 1 0 % p ro b a b ility, 51
w ith o u t optical techniques, w ith 50% p ro b a b ility, 51, 6 2 -6 6
131-134 w ith 100% p ro b a b ility , 52
B reakdow n p ro b a b ility , 51-53 Breakers {see C irc u it breakers)
Index 383

Bridges, 357, 360 [Cables]

Bruce breakdow n form ula, 291 sheath phenomena, 250, 251
B S I, 350 superconducting, 242, 273
Bubble theory, 133 synthetic insulation, 242, 243
B undled conductors, 13, 71 term inations, 269, 270
B urst pulse corona, 70 testing of, 351, 365, 366
Busbars, 99, 103, 104, 106 types, 244
Bushings, 9, 100, 104, 106, 229, C alculation o f electric fields, 11,
269, 352 1 2 ,36
testing of, 372, 373 Capacitors, 7, 353
Capacitive dividers, 293, 294
Cables, 6 - 8 , 241-273 Capacitive voltage transform er,
accelerated aging, 2 6 3 -2 6 6 103
accessories, 2 6 6 -2 7 0 Cascaded re c tifie r circuits (see
armors, 244 H ig h voltage D C generators)
belted, 244-24 6 , 248, 249, 253, Cascaded transform ers, 279, 280
263, 264 Cathode processes
capacitance, 246, 248 in gases, 27
compressed gas, 6 , 8 , 107, 242, in vacuum, 189, 190, 197
2 4 3 ,2 6 6 , 267, 271 Cavities in insulation, 16, 17, 105,
constants, 24 5 -2 4 9 214, 2 5 3 ,2 5 4 , 258
construction, 2 4 1 -2 4 4 Ceram ic insulating materials, 151
cross-bonding, 250, 251 Charged dielectrics, 3 3 1 -3 3 2
current-carrying capacity Charged particles generation, 23, 24
(am pacity), 251-253 Charge suppressant, 124
D C testing, 273 benzotriazole (B T A ), 124
electric fields in, 249, 250 Chem ical reactions in oils (see
inductance, 248, 249 Insulating oils)
insulation, 242, 243 C hlorinated hydrocarbons, 6 , 114
insulation resistance, 249 C irc u it breakers, 8 , 71, 100-102,
internal gas pressure, 351 351
jo in tin g , 266-2 6 8 arc interruption in, 9 6 -9 8
life (see Accelerated aging) R R R V (rate o f rise o f restrikin g
locating faults, 2 7 0 -2 7 2 voltage), 9 6 -9 7 , 200
losses, 250-253 Sp 6, 1 0 1 - 1 0 2
materials, 24 1 -2 4 4 short-line faults in, 97
o il-fille d , paper-insulated, term inal faults in, 97
241-244 , 250, 255, 262, testing of, 105, 351, 367 -3 6 9
266-267 , 2 7 0 -2 7 2 ,3 5 1 types of, 8 , 1 0 1 , 10 2
resistance, 245, 246, 250 C IG R E, 76
screens and jackets, 243 CISPR, 76, 350
selection, 351 C o e fficie n t o f attachment, 24, 83,
sheath grounding, 251 85, 95
384 Index

C oe fficie n t o f d iffu s io n {see [Corona]

E lectron d iffu s io n ) prebreakdown streamer, 3 7 ,3 8 ,4 0
C o e fficie n t o f ionization, 3 0 -3 2 , in SF6, 8 7 -9 0
85, 8 6 , 90 trich e l pulses, 3 7 -3 9
C oe fficie n t o f variation, 52 w ater p u rific a tio n , 41
C o llisio n cross-section, 23 Corona audible noise, 41, 73, 74
Com posite dielectrics, 7, 15-17, Corona current pulse, 37, 74
2 0 9 -236 Corona-free breakdown, 87
chem ical deterioration, 218 Corona onset voltage, 36, 87, 8 8 ,
dielectric constant o f 90
composites, 2 1 1 Corona pow er loss, 41, 73
electrochem ical deterioration, Corona radio interference, 41,
218 7 3 -8 0
interfacial polarization, 2 1 2 average level, 75
materials, 223 peak level, 75
m u ltip le layers, 226 quasi-peak level, 7 5 -8 0
properties, 2 1 0 rms level, 75
tracking, 217 CSA, 350
tracking index, 218 Corona stabilized breakdown, 8 8 ,
Compressed-gas cables {see Cables) 106
C onduction in liquids, 125 C ross-linking techniques, 155
conduction under high fields, C uring techniques, 155
127 catalyst curing, 155
conduction under lo w fields, hardener curing, 155
126,127 radiation curing, 155
C o n d u ctivity {see Properties o f C urrent pulses in liq u id dielectrics,
dielectrics) 129,130
Conductor bundles, 13, 71 Current transducers, 327
Contactors, 352 Current transform ers, 103, 327
Contam ination o f air gaps {see C urrent-carrying capacity o f cables
Sand/dust particles influence) {see Cables)
Corona, 35, 69, 70, 72, 289, 358
A C , 37 -3 8 Damped capacitive divider, 294
advantages, 41 D e-excitation tim e, 44
applications in industry, 41, 289 D eionizatio n o f gases, 28
current measurement, 301, 302 b y electron attachment, 83
generation, 301, 302 b y recom bination, 28
glow , 37, 38, 40 D eltatron c ircu it, 279
inception, 36 Diagnosis o f transform er life tim e
nanosecond pulsed, 44, 288, 289 by degree o f polym erization ,
negative DC , 37, 39, 74 231
onset streamer, 3 7 -3 9 b y dissolved gas analysis, 232
positive DC , 37, 39, 40, 74 by measurement o f fu rfu ra l, 233
Index 385

D ielectric [Elastomers]
constant (o r p e rm ittiv ity ), 3 ethylene-propylene m onom er
paper and board, 149 (E P M ), 161
liquids, 111 (see also L iq u id ethylene-propylene rubber
dielectrics) (EPR), 161, 163, 171,242,
loss factor, 4, 6 , 7 268, 272
relaxation tim e, 375, 376 H T V rubber, 162
D ie le ctric gases, 2 1 -4 8 , 83, 84 R T V rubber, 162
characteristics, 6 -7 , 83, 84 silicone rubber, 162, 164
choice of, 5, 6 , 4 6 -4 8 E lectric fie ld , 10-18
D ie le ctric losses in solids, 3, 4, in cable insulation, 249, 250
167, 170, 2 5 0 -2 5 3 ,3 5 7 in cavities, 16, 17, 2 15-217 ,
D ielectric loss test, 358, 361 253
D ielectric strength, 5 -7 , 4 6 -4 8 , 8 6 , com putation, 11, 12, 36
95 control and optimization of, 18, 93
D ig ita l recorders, 306-311 enhancement factor, 1 2 -1 4 ,4 1 ,
applications, 310, 311 190
measurement errors, 308, 309 estim ation, 12
parameters, 307, 308 at free particles, 17,18
technical assessment, 309 -3 1 0 intensificatio n at protrusions,
D ig ita l techniques in H V tests, 14, 15, 8 8 ,9 1 -9 3 ,2 4 9 , 258
3 1 1 - 315 at interfaces, 15, 16
deconvolution, 3 1 1 ,3 1 2 measurements, 315-323
partial discharge measurements, in m u ltid ie le c tric media, 15, 93
31 4 ,3 1 5 n onun iform , 5, 11, 239
transfer functio n m ethod, 3 1 2 - types, 11
314 u n ifo rm , 5, 11, 2 8 9 -2 9 2
D ipole polarization, 169 u tiliz a tio n factor, 14, 8 6 , 87, 90
Discharges in voids, 359, 360 (see E lectric conduction in liquids,
also Partial discharge) 125-129
Disconnectors, 99-103 E lectric polarization, 3 -5 (see also
D issipation factors, 3, 4, 250, 265 Polarization)
D is trib u tio n transform er testing, 8 , E lectric stress control, 18, 93
312- 3 1 4 ,3 6 1 -3 6 6 E lectrical treeing, 2 5 4 -2 5 8
D ividers, 293, 294 E lectrochem ical treeing, 261
Dust and sand effect on breakdown Electrode effects on breakdown,
(see Sand/dust particles) 5 4 -5 8 , 92, 93
Electrode optim ization, 18, 107
E arthing switch, 101, 102 E le c tro ly tic tank, 12
Elastomers, 161 Electrom agnetic c o m p a tib ility ,
ethylene-propylene diene 106,31 6
m onom er (E P D M ), 161, 162, Electrom agnetic interference, 41,
164, 242 45, 46, 7 3 -8 0 , 243
386 Index

E lectron Esters, 115, 116

attachment, 24, 83, 85, 95 External insulation {see O utdoor
attachment cross-section, 47 insulators)
avalanche, 3 0 -3 4 , 37, 44, 8 6 , 87 E xtra high voltages (E H V ), 1, 2,
critic a l avalanche length, 43, 86 4 5 ,4 6 , 80, 105, 106, 243,
detachment, 26 254, 272, 350
diffu sio n , 28
d rift tim e, 44 Faraday effect, 3 2 4 -3 2 6
d rift ve locity, 44 F ie ld emission o f electrons, 24, 27,
emission, 24, 27, 128, 131 128, 131
energy, 25 Field factor, 1 2 -1 4 ,4 1 , 190
excitation, 25 Field inside a cavity, 16, 17,
free path, 2 2 2 1 5 -2 1 7 , 253
Electron p o la riza b ility , 169 F ield measurement, 3 1 6 -3 2 4
Electronegative gases, 32, 83 {see F le xib le laminates, 2 3 4 -2 3 6
also S ulfur hexafluoride gas) classification, 234
E lectro-optical high voltage components, 235
measurements, 317, 318, 322 designation, 235
E lectro-optical im aging techniques, properties, 235
3 35-339 selection, 236
E lectro-optic effect, 318 Form ative tim e lag, 43
E lectro-optic sensors, 317, 318, Front tim e o f im pulse, 2 8 2 -2 8 7
324 Fuse, 352, 374
K e rr sensors, 322 -3 2 4 F ull-w ave rectifiers {see H ig h
Pockels sensors, 318 -3 2 4 voltage D C generators)
Electrostatic generators {see H ig h
voltage D C generators) Gap factor, 54
Electrostatic hazards, 343, 3 75-378 Gap-type discharge, 4 4 -4 6 ,3 0 2 ,
b uildup in pow er system, 378 360
control of, 377-3 7 8 current w aveform of, 46
generation of, 375 interference fro m , 79, 80
in industry, 377 measurement, 302
measurement of, 375 sources, 45
Electrostatic precipitators, 1,41 Gas bubbles, 359 {see also V o id s in
Electrostatic voltm eters, 295, 296 insulation)
Em ission {see E lectron emission) Gas density m onitor, 98, 100, 105
E poxy resins, 7 -9 , 100, 149, 164, Gas dielectric {see D ie le ctric gases)
166 Gas insulated switchgear (G IS ), 6 ,
EPRI, 73, 74 7, 9, 9 8 -1 0 7 ,3 1 6 , 327, 328
E rosion o f dielectrics, 253, 254 components, 9 9-103
{see also Partial discharges) conductor systems, 9 9 -1 0 0 , 104
E stim ation o f m in im u m discharge dimensions, 98
voltage in SF^, 90 earthing switches, 101, 103
Index 387

[Gas insulated switchgear] H ig h voltage bushings, 9, 100, 104,

effect o f m oisture on, 84-8 5 , 106, 229, 2 6 9 ,3 5 2
93 -9 4 , 98, 102, 108 H ig h voltage D C generators,
enclosure configurations, 99, 2 7 6 -2 7 9
100 cascaded m u ltip lie r circuits, 278
gas system, 100, 101, 104 deltatron circu it, 279
insulation coordination, electrostatic generator, 276
105-107 H ig h voltage dividers, 293, 294
layout, 99, 100 H ig h voltage equipm ent design
particle contam ination, 91,92 parameters, 18, 19
pressure effect, 8 7 -9 3 , 97, 98, H ig h voltage fuse, 352
100-102 testing, 374
spacers, 7, 93, 94, 100 H ig h voltage im pulse generators,
testing, 105 2 8 2 -2 8 8
very fast transient, 327, 328 (see circuits, 283, 285, 287
also H ig h frequency control of, 283, 284
transients) fo r lig h tn in g impulses, 285, 286
Gas m ixtures, 6 , 94, 95 m odeling, 282
Gas parameters, 2 1 -2 3 m ultistage circuits, 285, 286
effects on breakdown, 87-90, fo r sw itching impulses, 2 8 5 -2 8 8
94, 95 (see also Atm ospheric H ig h voltage measurements, 352
conditions) deconvolution, 311 -3 1 2
Gay Lussac’ s law, 21 by d ig ita l recorders, 306, 307,
Generation o f high voltages (see 3 0 8 ,3 1 1 ,3 5 3
H ig h voltage im pulse by electro-optical methods,
generators) 3 2 3-324
Generating voltm eter, 297, by electrostatic voltm eter, 295,
299-301 296
Glass, 151 b y generating voltm eter, 297,
electric, 173 299-301
G lo w discharge by peak voltm eter, 295, 296,
negative, 38 298, 299
positive, 40 by rod gaps, 293, 296
by series impedance, 297
H ig h frequency transients, b y sphere gaps, 2 8 9 -2 9 2 , 296
105-107, 282, 327, 328 by voltage dividers, 293,29 4,297
H ig h m olecular w e ig h t by voltage transform ers, 297
hydrocarbon o il, 116-118 H ig h voltage pow er capacitors, 7,
H ig h speed cameras, 338, 339 353
H ig h voltage A C generators, 2 7 9 - imprégnants, 228, 230
282 (see also Testing H ig h voltage test techniques, 352
transform ers) H u m id ity effect
H ig h voltage applications, 1, 2 on breakdown, 58-61
388 Index

[H u m id ity effect] [Insulating oils]

correction factor, 60, 61 service aged, 121, 124
electrostatic b u ild up, 375, 377 synthetic, 113, 114
H ydrocarbon oils, 7 therm al heat transfer, 12 0
h igh m olecular w eight, 116-118 Insulation coordination, 6 6 -6 9
Insulation level, 106, 107, 343
lE C , 76, 252, 3 0 7 ,3 6 1 -3 7 4 Insulation losses, 3, 4, 167, 170,
IE E E , 307, 314, 350 2 5 0 -2 5 3 , 357
Image converters, 338 Insulation resistance, 3, 366
Image intensifiers, 337, 338 Insulator p o llu tio n , 222, 223
Im pact ionization, 2 4 -2 6 , 85 Insulators {see O utdoor insulators)
Im pregnated insulation, 7 -1 1 , 224, In te rfa cia l polarization, 171,212
242, 243, 267 Internal discharges, 359, 360
Im pregnated paper properties, 159 Internal overvoltages, 42, 69, 9 6 -
Im pulse breakdow n p ro b a b ility , 97, 105-106, 282, 2 4 1 ,3 4 3
51-53 In trin sic breakdown, 174
Im pulse chopping tim e, 355 Io n -p a ir production, 24
Im pulse fro n t (o r crest) tim e, 2 8 2 - Ionization , 25, 30, 70
287 b y electron detachment, 26
critical, 56, 57, 70, 71 b y electron im pact, 2 4 -2 6 , 85
Im pulse generators {see H ig h b y photons, 24, 26, 85
voltage im pulse generators) co efficient, 3 0 -3 2 , 85, 8 6 , 90
Im pulse recorders {see D ig ita l therm al, 24, 26, 85
recorders)
Im pulse sparkover, 5 6 -5 9 JS I,351
Im pulse ta il tim e, 57, 2 8 2 -2 8 7
Im pulse voltage dividers, 293, 294 K e rr effect, 318, 3 2 2 -3 2 4
Im pulse withstand voltages, 106, K in e tic energy, 22
107, 343 K in e tic theory o f gases, 2 1 -2 3
Insulating liquids {see Insulating oils)
Insulating materials applications, Laplace’ s equation, 11, 14
8 -1 0 , 9 8-1 08 L a yo u t o f G IS, 99, 100
Insulating materials classification, Leader, 30, 69-71
5 -7 Leader channel, 70, 8 8 , 89
Insulating oils, 6 , 112, 117-126 Leakage current, 273
characteristics, 119-122 L ig h tn in g impulse, 51, 285, 286,
chem ical sta bility, 1 2 1 , 1 2 2 310,31 1
classification, 1 1 2 L iq u id dielectrics, 6 , 111-128 {see
constitutes, 6 also Insulating oils)
natural, 1 1 1 electrohydrodynamic m otion, 128
petroleum , 1 1 2 nonpolar liquids, 125-126
properties, 119 polar liquids, 125,126
reconditioning, 124, 125 streaming electrification, 123,124
Index 389

L iv e line maintenance, 71 N o n -p o la r liq u id s (see L iq u id

Loss angle, 3, 4 dielectrics)
Losses in dielectrics, 3, 4, 167, 170, N onuniform fields, 5 ,1 1 ,2 3 9
2 5 0 -2 5 3 ,3 5 7 N o n u n ifo rm fie ld gaps, 5, 11,
3 5 ^ 2 , 239
M agnetic fie ld sensors, 326, 327
M agneto-optic sensors, 324 -3 2 7 O il, insulating (see Insulating oils)
M a rx generator (see H ig h voltage O il-im pregnated insulation, 224
im pulse generators) O il-fille d cables (see Cables)
M ean free path, 23, 24 Onset voltage, 36, 87, 8 8 , 90
Measurement O ptical current transducers (O C T),
by d ig ita l methods, 306 -3 1 2 327
by electro-optic methods, O rientation polarization (see
317-324 P olarization)
o f electric fields, 3 1 6-324 O scilla to ry sw itching impulses (see
o f high voltages (see H ig h H ig h voltage im pulse
voltage measurements) generators)
o f m agnetic fields, 3 2 4-327 O utdoor insulators, 2 1 7 -2 2 4 , 353
by m agneto-optic methods, composite insulators, 217
3 24-327 p o lym e ric insulators, 217, 2 2 1
o f very fast transients, 327, 328 hyd ro p h o b icity, 223
M easurement system testing, 351, testing, 353, 372, 373
354,3 5 5 O vervoltage, 260 (see also Surge)
M edium h igh voltage, 1, 2 lig h tn in g , 42, 282, 342-343
M etal oxide arresters (see Surge protective devices, 10, 53,
arresters) 63-65, 6 8 , 69, 106
M ethods to suppress trees, 261-263 switching, 4 2 ,6 9 ,2 8 2 , 342,
M ica and its products, 150 343 (see also Internal over­
M in e ra l oils (see Insulating o ils) voltages)
M oisture effect on breakdown (see
H u m id ity effect on Partial discharge, 35, 216, 2 5 3 -
breakdow n) 258, 260, 262, 263, 273,
M o le cu la r velocities, 22 3 0 6 ,3 1 1 ,3 4 2 ,3 5 5 -3 5 8
M u ltifra m e photography, 137, 334, apparent charge, 357
338 classification, 357, 358
M u ltip le rod gaps, 6 7 -6 9 degradation caused by, 217,
M u ltip lie r circ u it (see H ig h voltage 2 5 5 -2 5 8
D C generators) detection system, 356, 357
extin ctio n voltage, 359, 360
Nanoseconds pulse generation, inception voltage, 253 -2 5 4 ,
288, 289 359, 360
N egative corona (see Corona) m in im u m detectable discharge,
Negative ion, 23, 24, 38, 88 359, 360
390 Index

[Partial discharge] Polarization, 3 -5 , 167

measurements of, 105, 254, 306, dipole, 169
3 1 1 ,3 1 4 ,3 1 5 electronic, 167
m odeling, 353 interfacial, 171, 2 1 2
o rig in , 2 5 3 -254 orientation, 169
recognition, 358 P o llu tio n , 222, 223
significance, 253, 254 Poly-butylenes, 113, 114
P hotom ultipliers, 3 35 -3 3 7 P o ly v in y lc h lo rid e (P V C ), 7 -9 , 160,
Piezoelectric transducers, 330, 331 166, 242, 244, 2 6 3 ,2 6 4
Particles in liq uid s {see Insulating Polyethylene (PE), 7 -9 , 153, 156,
oils) 242, 252, 257, 265, 268, 333,
Paschen’ s law, 34, 35, 87 334
Peak voltage measurement {see cross-linked (X L P E ), 79, 154,
H ig h voltage measurements) 158, 171, 172, 2 4 2 -2 4 4 , 250,
Peak voltm eter, 295, 296, 298, 299 252, 254, 2 6 3 -2 6 7 , 272, 273
Perfluorocarbons, 47 high density (H D P E ), 154, 157,
P erfluoro-n butane, 48 242
P erfluoropolyether, 118, 119 linear lo w density (L L D P E ),
P e rm ittiv ity 157, 243
com plex, 3 -4 lo w density (L D P E ), 153, 263
relative, 3 -4 , 15-17, 84, 246, m edium density (M D P E ), 157
250, 253 P olym erization, 153
Petroleum oils, 112 Polymers, 153-167
additives, 112, 121, 124 hom opolym ers, 153
aromatics, 1 1 2 therm oplastic, 7, 8 , 156
naphthenes, 1 1 2 thermoset, 155
paraffins, 1 1 2 Polym ethylsiloxane, 114, 115, 117,
pour point, 1 1 2 120, 121
Phase to ground insulation, 49, 50, Polypropylene, 244
5 3 ,7 0 Potential dividers, 293, 294
Phase to phase breakdown, 69-71 Potential transform ers, 103,277,
Phase to phase insulation, 49, 50, 35 2 ,3 7 8
53 Porcelain, 152, 166
P hotoionization, 24, 26, 85 Prebreakdown streamer
Photocopying machine, 41 negative, 37, 38
Photoelectric emission, 24, 27, 128, positive, 37, 40
131 Properties o f composites, 7, 209,
Plancks constant, 25 210
Point-plane gaps {see Rod-plane Properties o f gases, 6 , 4 6 -4 8 , 8 3 -
gaps) 85
Pockels effect, 318 Properties o f liquids, 6 , 114, 115,
Poisson’ s equation, 11, 14 117
Polar liquids, 125, 126 Properties o f solids, 7
Index 391

Properties o f vacuums, 6 Sand/dust particles influence

Precipitators, 1,41 on air breakdow n voltage,
Properties o f dielectrics, 2 -8 6 2 -6 6
com plex p e rm ittiv ity , 3, 4 on measuring spheres, 62, 65,
D C conductivity, 3 6 6 , 291
dielectric p e rm ittiv ity , 3 on protective gaps, 6 2 -6 5 , 6 8 ,
dissipation factor, 3 69
insulation resistance, 3, 365 on transmission line design, 62,
loss angle, 3 63
re sistivity, 3, 6 , 7, 249, 250 Sobering bridge, 357, 360
Protective gaps {see R od-rod gaps) Schlieren optical system, 136, 137,
Pulse generators {see H ig h voltage 335
im pulse generators) Secondary avalanches, 33
Q uasi-uniform fie ld , 61, 62 Secondary electrons, 31, 32
Sem iconductor sw itching fo r
Radio influence voltage (R IV ), 345 nanosecond pulse generators,
Radio interference {see Corona 289
radio interference) Sp 6 equipm ent {see Gas insulated
Rate o f rise o f im pulse voltage, 354 sw itchgear)
Rain effect on discharge, 61 Sp 6 insulation, 8 3-110, 243, 269
R aleigh T a ylo r in s ta b ility in Sp 6 gas {see S u lfu r hexafluoride)
liquids, 141 Shadowgraphic optical system,
Rate o f rise o f restrikin g voltage 136, 137,335
{see A rc interruption ) Silicone oils, 114, 117, 120, 121
R ecom bination o f ions, 28, 29 Single-stage generator {see H ig h
R elative air density (R A D ), 36, voltage im pulse generator)
58-61 Skin effect, 245, 246
correction factor, 59, 61, 290, S olid dielectrics, 7, 147-186
2 9 1,2 9 3 alum ina, 152
Relaxation tim e constant, 375, 376 ceramic insulating material, 7, 151
Resistive divider, 293, 294 classification, 148
Resistive capacitive divider, 294 d ielectric board, 149
Resonant transform ers, 281 dielectric paper, 149, 231
Response tim e o f measuring glass, 7, 151, 173
system, 351, 354, 355 inorganic insulating materials, 7,
R ipple 148
factor, 279, 280 m ica products, 150
voltage, 279 organic insulating materials, 7,
Rod-plane gaps, 13, 37, 50, 53, 56, 148
6 1 -6 3 , 88 porcelain, 152
R od-rod gaps, 10, 50, 53, 56, 58, steatite, 152
6 3 -6 9 , 293 therm al classes, 236
R od-structure, 55 treeing in, 255 -2 6 3 , 360
392 Index

Space charge measurements, [S u lfu r hexafluoride]

328-335 storage and handling of, 108,
laser intensity m odulation 109
method, 330, 334 to x ic ity of, 84, 85
pulsed electroacoustic method, Surface discharge, 358
330-3 34 Surface irre g u la rity factor, 36
pressure wave propagation Surge arresters, 10, 103, 104, 106,
method, 330, 332 107, 352
therm al pulse method, 329, 330, metal oxide, 10, 103, 106, 107
334 testing, 353, 370-372
therm al step pulse method, 330, Surge
334 lig h tin g , 42, 2 8 2 ,3 1 0 ,3 1 1 ,3 6 2 ,
Space charge fie ld , 38, 41, 69, 370
322-324, 333, 334 sw itching, 42,69, 282, 287, 302,
Space charge in polym ers, 328, 3 1 1 ,3 7 0
329, 332-334 w ith double peak, 71
Spacers {see Gas insulated w ith long duration, 287
switchgear) w ith m u ltip le peaks, 71
Spark discharge, 30, 71 w ith oscillations on ta il, 287
Spark gaps {see also Sphere gaps), Surge voltage, 42, 306, 310, 311
288, 289 Switches, 353
Sphere gaps, 52, 65, 6 6 , 2 8 9-292 , Switchgear, 353, 358 {see also Gas
297 insulated sw itchgear)
Sphere to plane gap, 37, 41, 42, 50, Switchgear testing, 105, 369, 370
63 Sw itching impulses, 51, 2 8 2 -2 8 8 ,
Standards, 345, 349-351 311
Standard deviation, 52, 348 Synthetic hydrocarbons, 113 {see
Statistical tim e lag {see Tim e to also L iq u id dielectrics)
breakdow n) Synthetic insulating materials, 149
Step response, 351, 354, 355 Synthetic testing o f c irc u it
Streamer, 30, 33, 34 breakers, 105, 352, 369
Streamer breakdown criterion, 34,
8 6 , 90,9 1 T a il tim e, 57, 2 8 2 -2 8 7
Streamer mechanism o f Tape insulation, 373
breakdown, 3 2 -3 4 , 8 6 - 8 8 T elevision interference (T V I), 48,
Streaming electrification, 123-124, 50, 302
S ulfur hexafluoride (SF^), 6 , 8 -1 0 , Tem porary overvoltage, 51, 260
16, 46, 83-1 09, 243, 269 Test
breakdown characteristics, 87, 88 aging, 345
breakdown processes, 85, 86 acceptance, 3
decom position of, 84, 85, 94, alternating current (A C ), 273,
1 0 2 , 108 306, 343, 345 -3 4 8 , 366
properties of, 83-8 5 bending, 366
Index 393

[Test] [Test]
breaking, 372 types, 343, 344, 362, 366, 367,
cantilever load w ithstand, 374 370 -3 7 4
current impulse, 371 w et w ithstand, 344, 372 -3 7 4
direct current (D C ), 273, 306, Test voltages
343, 345, 347 generation, 276-288
dielectric, 352, 362, 367, 368, measurement, 289-301
370, 374 Testing, 2, 105
dry, 344, 367, 372-374 autom ation, 315
dynam ic current withstand, classification, 342-345
374 destructive, 342, 343, 348
electromechanical, 345, 372 nondestructive, 342-344,348, 355
galvanizing, 372 p ro b a b ility m ethod of, 348
im pulse, 345-349 , 353, under im pulse voltages, 343
362-366, 371 up-and-down method of, 348,349
induced overvoltage, 345 Testing transform ers
m echanical, 345, 367, 368, 372 cascaded, 279, 280
operating duty cycle, 371 resonant, 281
partial discharge, 253, 254, 366, single-unit, 279
373 Tetrachloroethylene, 117, 118
p o llu tio n , 345 Therm al breakdow n {see
porosity, 345, 372 Breakdow n in solid
pow er frequency, 365, 366, 371 dielectrics)
pressure, 373 Thermal classes o f insulating
procedures, 34 9-351 , 352 materials {see Solid dielectrics)
puncture, 345, 372 Therm al energy, 22
radio influence voltage (R IV ), Therm al ionization, 24, 26, 85
367, 370 Therm al rating o f cables, 2 5 1-253
residual voltage, 371 Therm al resistance, 2 5 1 -2 5 3 , 268
routine, 343, 344, 361, 362, 366, Therm al step methods, 3 2 9 -3 3 0 ,
367, 370-373 334
sample, 343, 372 T h e rm a lly stim ulated currents, 329
short circu it, 367 -3 6 9 , 374 Tim e to breakdown, 43, 64, 70, 71,
special, 344, 362, 366 291
standards, 349-351 form ative, 43
tan 5, 345, 366, 373 statistical, 43
tape insulation, 373 Townsend breakdow n criterion, 32
temperature rise, 345, 352, Townsend firs t ionization
365-368 , 374 co efficient, 3 1 ,4 1
therm al shock, 372 Townsend second ionization
therm al stability, 372 co e fficie n t, 3 1 ,4 1
tightness, 373, 374 Townsend theory o f breakdown, 30
tim e/current, 374 Tracking, 217
394 Index

Transform er testing, 8 , 3 12-314 , Vacuum dielectrics, 6 , 188-207

361-366 m icrodischarge, 191
Transient current measurements, nonm etallic emission, 189, 190
301,302 m e ta llic emission, 189
Transient overvoltage, 42, 306, 310, prebreakdown emission, 189
311 V an de G ra a ff generator (see under
Transient recovery voltage, 7 2 ,9 6 -9 7 H ig h voltage D C generators
Treeing {see Solid dielectrics) electrostatic)
T riboelectric series, 3 74-375 V D E ,3 5 1
T rich e l pulses {see Corona) V oids in insulation, 16, 17, 105,
214, 2 5 3 ,2 5 4 , 258
U curve, 56, 59 V o lt-tim e characteristics, 43,
U ltra -h ig h voltages (U H V ), 1, 2, 6 6 -6 9 , 343,37 1
54, 56, 69, 73, 80, 243, 350 V oltage classification, 1, 2
U ltra v io le t lig h t, 66 V oltage dividers, 293, 294
U n ifo rm fie ld , 5, 11, 2 8 9 -2 9 2 V oltage drop, 2 7 7 -2 7 9
U n ifo rm fie ld breakdown, 29, 34, V oltage gradient {see E lectric
35, 8 6 , 87, 2 8 9 -2 9 2 fie ld )
U niform field gaps, 11, 13,29-35, 86 Voltage m u ltip lie r circuits {see
U niversal gas constant, 22 H ig h voltage D C generators)
U p-and-dow n m ethod {see Testing) V oltage ripple, 279
V oltage stresses, 51
V 50, 348 V oltage transform ers, 103, 277,
V acuum c irc u it breakers (V C B ), 3 5 2 ,3 7 8
199, 2 0 2 -2 0 6 ,2 9 2 W agner earth device, 361
construction, 203 W ater in insulating liquids, 119,
contact m aterial, 2 0 1 124, 227
current chopping, 2 0 1 W ater treeing, 2 5 8 -2 6 3 , 273
lim itations, 204 Waveshape o f impulse, 282-285,
R R R V in V C B , 200 287
m erits and demerits, 205 W et w ithstand test {see Test)
spiral contacts, 2 0 0 W in d effect on discharge, 61