(C. Christopoulos, A. Wright (Auth.) ) Electrical PDF

Electrical Power System
Protection
2nd Edition
Electrical Power System
Protection
2nd Edition
C. Christopoulos
Professor of Electrical Engineering,
University of Nottingham,
Nottingham, UK
and
A. Wright
(deceased)
former Emeritus Professor of Electrical Engineering,
University of Nottingham,
Nottingham, UK
SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Library of Congress Cataloging-In-Publication Data
ChristopouloS, Chrlstos.
Electrical power system protection / C. Christopoulos and A.
Wright.
p. cm.
Wright's name appears first on the earlier edition.
Includes bibliographical references and index.
ISBN 978-1-4419-4734-5 ISBN 978-1-4757-5065-2 (eBook)
DOI 10.1007/978-1-4757-5065-2
1. Electric power systems--Protectlon. 2. Electric power
transmisslon--Equlpment and supplies--Protectlon. I. Wright, A.
(Arthur) II. Title.
TK1055.C474 1999
621.31'7--dc21 99-14201
ISBN 978-1-4419-4734-5
Printed on acid-free paper
All Rights Reserved.

1999 by Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 1999
Softcover reprint of the hardcover 2nd edition 1999
No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical, including
photocopying, recording, or by any information storage and retrieval system,
without prior permission from the copyright owner.
Contents
Acknowledgements xv
Preface to the Second Edition xvii
Preface to the First Edition xviii
List of Symbols xx
1 Fuses 1
1.1 Historical background 1
1.2 Basic requirements 5
1.3 Fuse types and constructions 6
l.3.l Cartridge fuses 7
l.3.2 Semi-enclosed fuses 8
l.3.3 Expulsion fuses 8
1.3.4 Liquid fuses 8
1.4 The behaviour of cartridge fuselinks 9
l.4.l The pre-arcing period 9
l.4.2 The arcing period 10
1.4.3 Determination of fuselink performance 13
l.5 The construction of cartridge fuses 14
l.5.1 Fuse elements 14
l.5.2 Fuselink bodies 16
l.5.3 Filling material 16
l.5.4 Mountings and ratings 16
l.6 Semi-enclosed fuses 18
l. 7 Expulsion fuses 18
1.8 Liquid fuses 21
1.9 The application of fuses 21
l.9.l Time/current relationships 22
l.9.2 J2 t 23
l.9.3 Virtual time 24
1.9.4 Published time / current characteristics 24
l.9.5 Cut-off characteristics 25
l.9.6 Operating frequency 25
vi Contents
1.9.7 Discrimination and co-ordination 26
1.9.8 The protection of power-system equipment 28
1.1 0 The future 36
References 36
Further reading 37
2 Current transformers 39
2.2 Conventional current transformers 42
2.2.1 Equivalent circuits 42
2.2.2 Behaviour under normal steady-state conditions 44
2.2.3 Behaviour under abnormal conditions 49
2.2.4 The effects of core saturation on transformation
behaviour 59
2.2.5 Remanent core flux 62
2.2.6 Operation with a secondary circuit open or of a
high impedance 68
2.2.7 The construction of current transformers 71
2.3 Linear couplers 77
2.3.1 The output of a linear coupler with a burden of
infinite impedance 78
finite impedance 79
2.4 Current transformers with air-gapped cores 81
2.5 Non-conventional current transducers 82
2.6 Specifications and testing 84
2.7 The future 85
References 85
Further reading 86
3 Voltage transformers 87
3.2 Electromagnetic voltage transformers 88
3.2.1 Behaviour during steady-state conditions 89
3.2.2 Behaviour during abnormal conditions 90
3.2.3 Voltage transformer construction 91
3.3 Capacitor-voltage transformers 94
3.3.1 Capacitor dividers 95
3.3.2 Circuit of a capacitor-voltage transformer 96
3.3.3 Steady state behaviour 96
3.3.4 Behaviour during abnormal conditions 97
3.4 Recent developments 98
3.5 Specifications and testing 99
3.6 The future 100
Contents VB
References 100
Further reading 101
4 Overcurrent and earth fault protection 103

4.2 Relay connections and operation 106
4.2.1 The detection of earth faults 106
4.2.2 The detection of overcurrents 107
4.2.3 Electro-mechanical relays lOS
4.2.4 Directional relays 113
4.2.5 Thermal inverse time/current relays 116
4.3 Electronic relays 117
4.3.1 Basic electronic processing liS
4.3.2 Current-operated relay 119
4.3.3 Directional relay 119
4.4 Applications of overcurrent, earth fault and directional relays 124
4.4.1 Current grading 125
4.4.2 Time grading using relays with definite operating times 126
4.4.3 Time grading using relays with inverse time/current
characteristics 127
4.5 The application of directional and current-operated relays 135
4.6 Current and voltage transformers 139
4.6.1 Electro-mechanical relays with a fixed current setting 139
4.6.2 Electro-mechanical IDMT overcurrent relays 139
4.6.3 Electro-mechanical directional relays 140
4.6.4 Electronic relays 140
4.7 Standard specifications 140
4.7.1 IDMT relays 141
4.7.2 Directional relays 142
4.S The future 143
References 143
Further reading 144
5 Current-differential protective schemes 145

5.2 Factors affecting current-differential schemes 148
5.2.1 Current transformer errors 148
5.2.2 Current transformer secondary ratings 150
5.2.3 Interconnecting cables (pilot wires) 150
5.2.4 Symmetry of protective circuits 151
5.2.5 The setting of low-impedance relays 154
5.3 The use of biasing features 156
5.4 Implementation of schemes 158
5.4.1 Units of short physical length 158
viii Contents
5.4.2 Units of considerable physical length 159
5.4.3 Balanced-voltage schemes 162
5.4.4 Schemes to protect zones with more than two ends 165
5.4.5 The protection of bus bars 167
5.5 Earth fault protective schemes 169
5.6 Schemes employing high-impedance relays 171
5.7 Relays used in current-differential schemes 174
5.8 Application of current-differential schemes 177
References 177
Further reading 177
6 The protection of transformers 179

6.2 The construction and behaviour of transformers 180
6.2.1 Construction 181
6.2.2 Operation during normal and external fault conditions 184
6.2.3 Behaviour during internal fault conditions 191
6.2.4 Causes of internal electrical faults 194
6.3 The application of protective schemes and devices to power
transformers with two or more windings per phase 198
6.3.1 Inverse time/current relays 198
6.3.2 Current-differential schemes 198
6.3.3 Restricted earth fault protection 208
6.3.4 Combined differential and restricted earth fault
protection 210
6.3.5 Earth fault protection of delta-connected windings 211
6.3.6 Tank earth protection 212
6.3.7 Overfluxing protection 212
6.3.8 Protection against overheating 213
6.3.9 Buchholz relays 213
6.3.10 The protection of large three-phase transformers 216
6.4 The protection of earthing transformers 216
6.5 Auto transformers and their protection 219
6.5.2 Tank earth protective equipment 223
6.5.3 Other protective equipment 223
6.6 The future 223
References 224
Further reading 225
7 The protection of rotating machines 227

7.2 Protective devices and schemes 230
7.2.1 Fuses 230
Contents ix
7.2.2 Thermal relays 230
7.2.3 Thermal devices 231
7.2.4 Instantaneous electromagnetic relays 231
7.2.6 Current-balance schemes 232
7.2.7 Phase-unbalance relays 233
7.2.8 Voltage-operated relays 233
7.2.9 Control equipment 233
7.2.10 Applications of protective schemes to machines 233
7.3 The protection of motors 234
7.3.1 The protection of small motors 234
7.3.2 The protection oflarge induction motors 235
7.3.3 The protection of synchronous motors 244
7.3.4 The protection of d.c. motors 251
7.3.5 The protection of variable-speed drives 254
7.4 The protection of alternators 255
7.4.1 Alternator construction and behaviour 255
7.4.2 The application of protective equipment to alternators 261
7.5 The future 278
References 278
Further reading 278
8 The protection of busbars 279

8.2 Busbars 281
8.2.1 The construction of enclosed switchgear units 281
8.2.2 Open-type switching sites 283
8.2.3 Further methods to reduce the incidence of busbar
faults 283
8.3 Sectionalization 284
8.4 Faults on or near busbars 286
8.4.1 Internal faults 286
8.4.2 External faults 286
8.5 Positioning of current transformers and incorrectly
protected zones 287
8.6 Protective arrangements for busbars 288
8.6.1 Application to simple single-phase unsectionalized
busbars 289
8.6.2 Application to simple three-phase unsectionalized
busbars 290
8.6.3 Application to complex three-phase sectionalized
busbars 292
8.6.4 Interconnections in current-differential schemes applied
to busbars 296
x Contents
8.7 Relays used in current-differential schemes 296
8.7.1 High-impedance relays 296
8.7.2 Low-impedance relays 297
8.7.3 Low-impedance relays with compensation for
current-transformer saturation 299
8.7.4 The duplication of current-differential protective
schemes 301
8.8 Manual and automatic testing 303
8.9 The future 304
References 304
Further reading 305
9 The protection of overhead lines and cables by

current-differential schemes 307
9.2 Cables and overhead transmission and distribution lines 310
9.3 The application of current-differential protective schemes 312
9.3.1 Comparison arrangements 312
9.3.2 Circulating-current schemes 315
9.3.3 Balanced-voltage schemes 317
9.3.4 Pilot wires 329
9.3.5 Protective schemes which use rented telephone
circuits 330
9.3.6 Monitoring of pilot wires 331
9.3.7 Current-differential schemes incorporating
optical-fibre links 333
9.4 The application of current-differential schemes to
multi-ended circuits 337
9.5 The application of current-differential schemes to
lines and cables terminated with transformers 340
9.6 The future 343
References 343
10 Interlock and phase-comparison schemes for the protection

of overhead lines 345
Introduction 345
10.2 The construction and behaviour of transmission lines 347
10.3 Features of interlock protective schemes 349
10.3.1 Relaying arrangements 349
10.3.2 Interlocking signals 360
10.3.3 Starting relay 361
10.3.4 Signalling channels 363
10.4 Interlock protective schemes 365
Contents xi
10.4.1 Reyrolle interlock protective scheme 365
10.4.2 Modern schemes 365
10.5 Features of phase-comparison protective schemes 367
10.5.1 The phase displacements ofline currents 368
10.5.2 The production of comparison signals 368
10.5.3 The comparison process 369
10.5.4 Comparison signals 371
10.5.5 Signalling equipment 371
10.5.6 Starting relays 372
10.5.7 Current transformers 372
10.6 Phase comparison schemes 372
10.6.1 Telephase protective schemes 373
10.6.2 Contraphase protective schemes 380
10.6.3 Other schemes 381
10.7 Auto-reclosing 381
10.8 The future 383
References 383
11 Distance-type protective schemes for overhead lines

and cables 385
11.2 The behaviour of overhead lines 387
11.2.1 The input impedance of a short single-phase line 387
11.2.2 The input impedance of a long single-phase line 390
11.2.3 The input impedances of a three-phase line 392
11.3 Impedance measurement 393
11.3.1 Relays dependent on the magnitude of impedance 393
11.3.2 Relay characteristics 395
11.3.3 The effects of system transients 400
11.4 Basic schemes 403
11.4.1 Application to three-phase lines 405
11.4.2 The detection of faults close to the input ends
of lines 406
11.4.3 The settings of impedance-measuring relays 407
11.4.4 The operating times of relays 416
11.4.5 Relay performance standards 418
11.5 Conditions when a healthy line is affected by asynchronous
system operation 419
11.6 Schemes applied in the past 422
11. 7 Present-day schemes 423
11.7.1 Optimho static-distance protection schemes 423
11.7.2 Distance protection relay 7SL32 424
11.8 The duplicate protection of transmission lines 425
11.9 The protection offeeder transformers 426
xii Contents
11.10 The protection of cables 427
11.11 The setting of distance type schemes 427
11.12 Accelerated clearance of faults near the ends of lines 428
11.13 The protection of teed lines 429
11.14 The protection of series-compensated lines 430
11.15 The future 434
References 436
Further reading 437
12 Ultra-high-speed schemes for the protection of long

transmission lines 439
Introduction 439
12.2 Travelling waves 441
12.2.1 Initial waves on d.c.lines when short-circuits occur 441
12.2.2 Initial waves on ideal single-phase a.c.lines when
short-circuits occur 443
12.2.3 Initial waves on ideal lines when resistive
faults occur 444
12.2.4 The effects of line resistance and leakage
conductance 444
12.2.5 Later travelling waves resulting from reflections 445
12.2.6 Travelling waves on three-phase lines 452
12.3 Protective schemes which detect travelling waves 456
12.3.1 Directional-comparison schemes 457
12.3.2 Schemes which determine the positions of faults 468
12.4 The application of ultra-high-speed relays to
series-compensated lines 474
12.5 The future 476
References 477
Further reading 478
13 Digital communications 479

Introduction and historical background 479
13.1 Basic communication principles 479
13.1.1 Basic theorems 481
13.1.2 Basic structure of transmission networks 483
13.2 Transmission media and interfaces 485
13.2.1 Transmission media 485
13.2.2 Interfaces for protection and communication 487
13.3 Communication systems and protocols 489
13.3.1 The encoding of data 489
13.3.2 Multiplexing 493
13.3.3 Error detection 494
Contents xiii
13.3.4 Open systems interconnection (OSI) model 496
13.3.5 Message structure and protocols 498
13.3.6 Electromagnetic compatibility 498
13.4 The future 499
References 499
Further reading 500
14 Digital signal processing (DSP) and protection 501

Introduction and historical background 501
14.1 Impact ofDSP on protection 501
14.2 Hardware and software environments 503
14.3 Principles of digital signal processing (DSP) 513
14.3.1 General DSP principles 513
14.3.2 DSP algorithms in protection 519
14.4 The future 523
References 523
15 New relaying concepts 525

Introduction 525
15.1 Expert systems 526
15.2 Fuzzy logic 529
15.3 Artificial neural networks 533
15.4 Adaptive relaying 539
15.5 Integrated protection 541
15.6 The Global Positioning System (GPS) 546
15.7 Genetic algorithms and simulated annealing 547
15.8 Flexible AC Transmission Systems (FACTS) 548
15.9 Embedded generation 550
15.10 The future 550
References 551
Appendix A The testing and application of power-system

protective equipment 555
A.1 Performance requirements 555
A.2 Testing of protective devices and equipment 556
A.2.1 Fuselinks 556
A.2.2 Current and voltage transformers 559
A.2.3 The testing of relays 560
A.2.4 The testing of protective schemes 561
A.3 Automatic testing of relays and protective schemes 567
References 568
Appendix B Percentage and per-unit quantities 569

xiv Contents
Appendix C Transformations of three-pbase quantities 575
C.I Voltage, current and impedance transformation 575
C.2 Transformation into symmetrical components 577
C.3 Other transformations 579
References 581
Appendix D The determination of power-system behaviour

using symmetrical components 583
D.1 Sequence impedances 583
D .1.1 Three-phase synchronous machines 583
D.l.2 Transformers 585
D.l.3 Overhead lines and cables 585
D .1.4 The effects of physical asymmetry 586
D.2 Sequence networks 586
D.2.1 Earthing of neutral points 586
D.2.2 Transformer connections 588
D.3 The interconnection of sequence networks 592
D.3.1 Balanced three-phase normal or fault conditions 592
D.3.2 Conditions when a phase-to-phase fault is present 592
D.3.3 Conditions when a single phase to earth fault
is present 594
D.3.4 Impedances used in sequence networks 595
D.4 Numerical example 595
D.4.1 Short-circuit between phases 'b' and 'c' and earth 597
0.4.2 Short-circuit between phase 'a' and earth 597
Reference 597
Concluding remarks 598
Index 599
Acknowledgements
Figure 4.26 is reproduced from BS 142: Section 3.2: 1990 with the
permission of BSI. Complete copies of the standard can be obtained by
post from BSI sales, Linford Wood, Milton Keynes, MK14 6LE.
Preface to the second edition
The death of Professor Arthur Wright in the summer of 1996 deprived

me of a friend and a colleague whose judgement and experience shaped this
book. I pay tribute to his contributions to protection and electrical
engineering education.
In the five years since the first edition appeared, many developments
have taken place and it is now necessary to update the book. The use of
digital communications and advanced signal processing techniques is now
widespread and several fully numeric relays are available from manu-
facturers. Two new Chapters 13 and 14 have been added to introduce
readers to these concepts and associated techniques. Artificial intelligence
is making its impact in all engineering applications and power system
protection is no exception. Expert systems, fuzzy logic, artificial neural
networks, adaptive and integrated protection, synchronized measurements
using the global positioning system, genetic algorithms, flexible a.c.
transmission systems, are some of the techniques considered in connection
with protection. Although many of these techniques have not yet found
major application in protection, it is nevertheless essential for the educated
protection engineer to have a basic understanding of the underlying
principles and methodology so that he, or she, can evaluate their suitability
for new relaying problems and applications. Chapter 15 was therefore
added to guide readers through this developing area. I have also added
some new material in other chapters to reflect changes over the past years.
The emphasis remains the same as for the first edition and this is the
presentation of the fundamentals of protection illustrated by examples of
current engineering practice.
I wish to thank Professor Shi Shiwen of the Southeast University,
Nanjing, China, for pointing out several errors in the first edition, and my
colleague Miss Ana Vukovic for assisting me in the preparation of this
edition.
Christos Christopoulos
Preface to the first edition
Several books have been produced over the years about the protective
equipment which is incorporated in electrical power systems and
manufacturers continually produce detailed literature describing their
products. Recognizing this situation and accepting that it is no longer
possible in a single volume to provide a complete coverage of the
protective equipment now available and the many factors which have to be
considered when it is being developed and applied, we have concentrated
on basic principles and given examples of modern relays and schemes in
this work.
Chapter 1 deals with electric fuses, which were the earliest protective
devices. The chapter begins with a historical introduction, as do all the
chapters, and then information is provided on the construction and
behaviour of fuses and finally the factors which must be taken into account
when they are to be applied to circuits are examined.
Chapters 2 and 3 deal respectively with conventional current and voltage
transformers and modern transducers. In each case, details are given of
the constructions and behaviours of theses devices, which play important
roles in supplying protective equipment.
Chapter 4 deals with relays which have constant operating times and
those which have inverse time/current characteristics. After tracing their
development, modern relays are described and then the factors which must
be considered when applying them are considered in some detail.
The principles of current-differential schemes are set out in Chapter 5
and the causes of the imbalances which can arise in them when protected
units are healthy are examined. The biasing features provided to enable
satisfactory performance to be obtained are outlined.
The later chapters are devoted to the protection of the main components
of the networks, namely transformers, busbars, rotating machines and
transmission and distribution lines and cables. The presentation is similar
to that in the earlier chapters. In each case information is provided about
the construction and behaviour of the plant being protected and then the
appropriate protective schemes, including current-differential, phase com-
parison, distance and travelling wave, are described and examined.
Appendices dealing with per-unit quantities, symmetrical components
and other modal quantities are included.
Preface to the first edition xix
We express our appreciation of the assistance given to us by Dr
D.W.P. Thomas during the preparation of this book and during research
into travelling-wave protective schemes. We also wish to thank Miss S E
Hollingsworth for typing the manuscript.
We hope that this book will prove of value to those involved in the
study, development, production and application of protective equipment
and that they will enjoy working in a challenging field in which new
problems continuously arise.
Arthur Wright and
List of symbols
A cross-sectional area (m 2)
B magnetic flux density (T)
C capacitance (F)
e instantaneous e.m.f. (V)
E r.m.s. or constant e.m.f. (V)
instantaneous current (A)
I r.m.s. or constant current (A)
L inductance (H)
M mutual inductance (H)
N turns in a winding
R resistance (0)
t time (s)
v instantaneous voltage (V)
V r.m.s. or constant voltage (V)
q> magnetic flux (Wb) or phase angle
~ permeability of magnetic material (Wb/ Am)
SUFFIXES
a, b,c phases of three-phase system
p primary circuit or winding
pk peak value of alternating current or voltage
s secondary circuit or winding
t tertiary circuit or winding
1,2,0 positive, negative and zero-sequence quantities
1
Fuses
Fuses, which were introduced over one hundred years ago, were the first fonn
of protection used on electrical networks. Extremely large numbers of them
have been produced since that time and they are still used extensively in the
lower voltage sections of power systems around the world. Fuselinks are
simple and therefore relatively cheap devices, their cost being very low
relative to that of the plant being protected by them. They thus satisfy a basic
requirement which applies to all protective equipment.
The underlying principle associated with fuses is that a relatively short piece
of conducting material, with a cross-sectional area incapable of carrying cur-
rents quite as high as those which may be pennitted to flow in the protected
circuit, is sacrificed, when necessary, to prevent healthy parts of the circuit
being damaged and also to limit damage to faulty sections of the circuit to the
lowest level possible.
Fuses incorporate one or more current-carrying elements, depending on their
current ratings. Melting of the elements, followed by arcing across the breaks,
occurs when overcurrents flow through them. They can interrupt very high
fault currents and because of the rapidity of their operation in these circum-
stances, they limit the energy dissipated during fault conditions.
1.1 HISTORICAL BACKGROUND

The earliest reference to the use of fuses which the authors have been able to
trace occurred during the discussion which followed the presentation of a paper
by Cockburn [1] to the Society of Telegraph Engineers in 1887 when W. H.
Preece stated that platinum wires had been used as fuses to protect submarine
cables since 1864 and Sir David Solomons referred to the use of fuses in 1874.
Fuses must have been in use in significant numbers by 1879 and there must
have been situations where a simple wire or strip element was not suitable,
because in that year Professor S. P. Thompson produced what he described as
an improved fonn of fuse. It consisted of two wires connected together by a
ball of conducting material as shown in Fig. 1.1. It was stated that the ball
could be an alloy of lead and tin or some other conducting material with a low
melting point. When a high current flowed through the fuse for a sufficient
C. Christopoulos et al., Electrical Power System Protection

Springer Science+Business Media Dordrecht 1999
2 Fuses
Fig. 1.1 Fuse developed by Professor S. P. Thompson. (Reproduced from Wright and
Newbery, 1982, Electric Fuses with the permission of lEE).
period of time, the ball melted and the wires swung apart to form an adequate
break in the circuit.
A variation on the above construction was patented in 1883 by C. V. Boys
and H. H. Cunyngham. In their arrangement the element consisted of two leaf
springs which were soldered together at their tips as shown in Fig. 1.2. Again
the passage of an overcurrent for a sufficient period of time caused the solder
to melt, after which the springs flexed away from each other to provide the
quick and adequate separation needed to ensure current interruption.
Demonstrations of incandescent filament lamps were given by 1. Swan (later
to become Sir Joseph Swan) in Britain in 1878 and at about the same time in
the USA by T. A. Edison. These led to a great demand for the installation of
electric lighting in both public and private buildings. Initially individual con-
sumers had their own generating plants, but shortly afterwards small central
generating stations were provided to supply their own surrounding areas.
Some interesting and detailed information about early installations is given
in letters written by J. H. Holmes to H. W. Clothier in 1932. An excerpt from
one of these letters, which was included in Clothier's book entitled Switchgear
Stages [2], is reproduced below. It clearly indicates that the identity of the
person who first introduced fuses is not known.
Fig. 1.2 Fuse patented by Boys and Cunyngham. (Reproduced from Wright and New-
bery, 1982, Electric Fuses with the permission of lEE).
Historical background 3
Letter from Mr J. H. Holmes:
Regarding the origin of fuses, I have always been uncertain as to who is entitled
to the credit of being the first inventor, and am ofthe opinion that this is very clearly
the case of 'Necessity is the Mother of Invention'.
I have been looking up some records of what was known about fuses in the early
'eighties' (1880's), and in the first volume of 'Electric Illumination', compiled by
J. Dredge and published in August 1882 at the Offices of 'Engineering', on page
630 it is stated that Edison's British Patent of April 1881 appears to have been the
first notification of lead safety wire. It also appears that Edison's device was called
a 'safety guard'.
I think however that Swan used a device for the same purpose and before April
1881, because 'Cragside' near here, the seat of Sir W. G. (afterwards Lord) Arm-
strong was lighted with Swan lamps by the middle of December 1880. Swan used
tinfoil for the fuse and a strip of this was jammed between two brass blocks, so as
to form part of the circuit, by a plug of wood and later of steatite, and I have samples
of a combined switch and fuse, and a fuse only made in this way, and which were
in use at Cragside. In a Swan United Electric Light Co's catalogue dated 1883, I
found such fuses illustrated and called 'safety fusing bridges'.
In the description of the Electric Lighting, on the Swan system, of the Savoy
Theatre in 'Engineering', March 3rd 1882, fusible safety shunts are referred to as
'not intended so much to guard against a danger which is next to impossible to
occur in practical working, but to protect the lamps themselves from destruction
from too powerful a current being transmitted through them. This seems to confirm
what Campbell Swinton says about the Drawing Office at Elswick in 1882*, which
you quote, and I note he also says that at the Paris Exhibition of 1881 there was' a
vast array of switches, fuses, cut outs and other apparatus'.
Factors, including the concern for public safety, the cost and fragility ofthe
lamps, referred to earlier, and the increasing level of available volt-amperes
under fault conditions, made evident the need for protective equipment. As
there were no obvious alternatives to fuses at that time, a number of workers
endeavoured to develop reliable fuses.
Much work was done to understand the processes involved during the melt-
ing of fuse elements, a particularly significant contribution being made by
Cockburn [1]. He attempted to put the design of fuses on a sound engineering
basis. He studied the effects of the heat conducted away from fuse elements
by their terminals and connecting cables and investigated the properties of
conductors in an attempt to select the materials most suitable for use as fuse
elements. He recognized that materials which oxidize significantly would be
It is perhaps not generally known that fuses, as originally introduced by Swan, were designed
not as a safeguard to protect the wires against overloading on short circuits, but in order to prevent
the lamps from over-running. When I went to the Armstrong works at Elswick in 1882, part of
the drawing office had been electrically lighted by the Swan Company, and each incandescent
lamp was fitted with a separate tinfoil fuse for this purpose. The precaution was, perhaps a
necessary one, as the lamps then cost 25 s [1.25] each and were very fragile, while the arrange-
ments for keeping a constant voltage were very crude.' (Campbell Swinton at the lEE Commemora-
tion Meetings, February 1922, lEE Journal, 1922, Vol 60, p. 494).
4 Fuses
unsuitable because the characteristics of fuses containing them would change
with time. Tests which he did on a range of fuses showed that they were not
being applied satisfactorily. He found instances where the minimum fusing
currents were many times the rated currents of the circuits and pieces of
equipment being protected. He suggested that fuses should operate at 150-
200% of the rated current of the circuit being protected.
Most of the early fuses were mounted in wooden boxes, but the individual
elements were not separately enclosed. As early as May 1880, however, T. A.
Edison patented a fuse in which the element was enclosed in a glass tube. This
was done to protect the surroundings from the effects of the rupturing of the
element rather than to affect or control the fuse perfonnance. Undoubtedly the
credit for developing the filled cartridge fuse must go to W. M. Mordey, who
patented the device in 1890. His patent described a fuselink with a fusible
copper conductor, of either thin foil or one or more small diameter wires,
enclosed in a glass tube or similar vessel. It was stated that the tube should be
wholly or partially filled with finely divided, semi-conducting or badly-con-
ducting material, which should preferably be incombustible or non-flammable.
The fuse produced by Mordey is illustrated in Fig. 1.3.
Fuses were the only fonn of protective equipment available during the final
decade of the nineteenth century. That they were produced in large ratings is
evident from Clothier's paper entitled 'The construction of high-tension con-
trol-station switchgears, with a comparison of British and foreign methods'
[3]. A relevant extract from this paper reads as follows:
The High-Tension Fuse most extensively used in Germany is not unlike the
well-known Bates fuse consisting of an open-ended tube of porcelain, ambroin,
Fig. 1.3 Cartridge fuse patented by Mordey. (Reproduced from Wright and Newbery,
1982, Electric Fuses with the permission of lEE).
Basic requirements 5
stabilit, or similar insulating material with plug terminals at each end. The fuse wire
of copper or alloy is threaded through the tube and clamped by screws and plates
or soldered to the terminals. For potentials of 2,000 to 10,000 volts, the length of
these tubes varies between 8 inches and 15 inches. Several fine wires are connected
in parallel for the higher voltages, each wire being enclosed in a separate internal
tube or otherwise partitioned by insulating materials, so that each wire has a column
of air to itself. Unlike the Bates fuse, there is no handle moulded with the tube, but
flanges are provided at the ends, and moreover, in most cases, it is customary to
have a long pair of tongs close by the switchgear with which any fuse can be
clutched while the operator is at a safe distance from it. Considering the massive
tube fuses-from four to five feet long-used at Deptford and Willesden, and also
expensive designs such as the oil-break fuses of home manufacture, it would appear
that either we overestimate the destructive effects caused in breaking high tension
circuits or else the necessity of blowing a fuse without destroying the fuse holder
is not considered a matter of importance in Germany.
During this century, relay-based protective schemes have been produced.

These together with the circuit-breakers controlled by them are now used in
conjunction with all the items of plant on the major generation and trans-
mission networks. Fuses are still used, however, in large numbers to protect
parts of the lower voltage distribution networks as well as the many electrical
appliances and items in use today. As a result, fuses have been continuously
developed to meet new needs and investigators have and are still studying
basic phenomena, such as the arcing process, to enable fuses with the charac-
teristics needed for new applications to be produced.
1.2 BASIC REQUIREMENTS
In 1882 an Electric Lighting Act was passed by the British Parliament and it
was amended six years later to form what was known as the Electric Lighting
Acts 1882 and 1888. A feature of these acts was that they required the UK
Board of Trade to introduce regulations to secure the safety of the public and
ensure a proper and sufficient supply of electrical energy. The early regula-
tions included clauses stating that a suitable fuse or circuit-breaker must be
present in each service line within a consumer's premises. In 1919, British
Standard Specification 88, which covered fuses for rated currents up to 100 A
at voltages not exceeding 250 V, was introduced. It included definitions of
terms such as 'fuse carrier' and 'fusing current' and specified the maximum
short-circuit currents that fuses of various rated currents should be able to
interrupt and also the corresponding minimum currents at which they should
operate.
Over the years there have been revisions of BS88 and other standards have
been produced in Britain, the United States and European countries. To clarify
the situation, the International Electrotechnical Commission (1EC) has pro-
duced standards which require that:
6 Fuses
1. each fuse is so designed and produced that it will, throughout its life, allow
the circuit in which it is included to operate continuously at currents up
to its rated value;
2. each fuse will operate in a sufficiently short time when any current above
a certain level flows, because of an overload, to prevent damage to the
equipment being protected;
3. in the event of a fault developing on a network or piece of equipment,
fuses will operate to limit the damage to a minimum and confine it to the
faulted item.
These requirements, which equally apply to all other types of protective equip-
ment, infer that fuses must have inverse time/current characteristics and that
they must be applied so that discrimination is achieved during fault conditions.
1.3 FUSE TYPES AND CONSTRUCTIONS

Fuses are classifed into three categories, namely high voltage for use in
circuits above 1000 V a.c., low voltage and miniature.
There are a number of basic constructional forms of fuses, namely cartridge,
semi-enclosed, liquid and expulsion.
~~!,,->01I11""_ _.....:...._ _ _ _ _ Fuselink
.Jr--Fuse base
Fig. 1.4 Low-voltage cartridge fuse. (Reproduced from Wright and Newbery, 1982,
Electric Fuses with the permission of lEE).
Fuse types and constructions 7
1.3.1 Cartridge fuses
These are designed for high voltage, low voltage and miniature applications.
The fuselink, which is replaceable, is often fitted into a fuse holder that
consists of a fuse carrier and fuse base, an example being shown in Fig. 1.4.
Fuselinks for low current ratings contain a single element, while those for
higher ratings have a number of parallel-connected elements. The elements
are usually of silver or silver-plated copper, it being necessary to ensure that
oxidation will not occur, as stated earlier. Wire elements are used in fuses with
ratings below 10 A, but for higher ratings strips with one or more sections of
reduced cross-sectional area are used. The elements are attached to plated
copper or brass end caps which together with the body form an enclosure or
cartridge. The bodies, which must be good insulators, must also be robust and
able to withstand the conditions which occur during interruption. They were
Rivet-----"i7rn'l"n'
Asbestos disc _ _ _-{fj~~~~~~~~~r--Outer cap
Inner cap---~n.
Granular quartz _ _ _ __+_'~

I -_ _ Ceramic body
Elements----+~~.:....;...._!_
o I - - - - E n d tag
Fig. 1.5 Cross-sectional view of a cartridge fuselink. (Reproduced from Wright and
8 Fuses
almost exclusively of ceramic or glass in the past but glass-reinforced plastics
have been introduced in recent years.
For high-breaking capacity fuselinks, the space within the body is usually
filled with quartz of controlled grain size and chemical purity. A cross-
sectional view through a typical fuselink is shown in Fig. 1.5.
1.3.2 Semi-enclosed fuses

These fuses are in the low voltage category. They also consist of a fuse base
and carrier, the latter containing the replaceable wire element. They are of low
breaking capacity, being capable of interrupting currents up to about 4000 A.
An example of this type of fuse is shown in Fig. 1.6.
1.3.3 Expulsion fuses

These fuses, which are used for high voltage applications, contain a mechan-
ism to withdraw the fuse link away from one of its contacts when the element
melts. As a result, a long air gap is introduced into the circuit. Satisfactory
interruption is achievable at current levels up to about 8000 A.
1.3.4 Liquid fuses

These fuses are also in the high voltage category. They have a liquid-filled
glass body within which is a short element of wire or notched strip, held in
tension by a spring. During operation, the element melts and a rapid separation
occurs, as the spring contracts. As a result the arc extends and is extinguished
in the liquid filling.
Asbestos liner _ _-+_-++-
"----~- ____ Fuse carrier
_ _ _ _ Fuse base
Fig. 1.6 Semi-enclosed fuse. (Reproduced from Wright and Newbery, 1982, Electric
Fuses with the permission of lEE).
The behaviour of cartridge fuse links 9
1.4 THE BEHAVIOUR OF CARTRIDGE FUSELINKS
These fuselinks incorporate one or more elements which melt and then
vaporize when currents above a certain level flow through them for a certain
time. Thereafter the arc or arcs which result have to be extinguished to
complete the interruption process. The operating time is therefore made up of
two periods, designated the pre-arcing and arcing periods. The behaviour
during these periods is considered in the following sections.
1.4.1 The pre-arcing period

The element or elements and the other conducting parts of a fuse link possess
resistance and as a result a fuselink must absorb electrical power when it
carries current. While the current is constant and below a particular level,
the temperatures of the various parts of the fuse must be at levels above the
ambient value so that power, equal to the input power, is dissipated to the
surroundings and a state of equilibrium exists.
Should the current be increased and maintained above a certain level, a new
state of equilibrium will not be achieved because, although the temperatures
of the fuselink parts will rise, the power dissipated from the fuselink will not
become equal to the power input by the time that parts of the element or
elements melt. Disruption of the element or elements will result and circuit
interruption will take place after a period of arcing. The time from the instant
when the current exceeds the critical value until the melting and initial vapor-
ization of the elements occurs is known as the pre-arcing period. The time
taken thereafter to achieve interruption is known as the arcing period.
The more the current through a fuselink exceeds the maximum value at
which equilibrium can be achieved, the shorter is the time taken before melting
of the element or elements occurs. This is because the power available to cause
the temperatures to rise is equal to the difference between the input power,
which is proportional to the square of the current, and the power dissipated
from the fuselink. The latter quantity is limited because the temperatures of
the fuselink parts cannot exceed the melting point of the element material.
As a result, all fuses have inverse pre-arcing time/current characteristics in
the range of currents above that at which thermal eqUilibrium can be estab-
lished. A current marginally above this latter level would theoretically cause
operation after an infinite time. Clearly this current could not be determined
experimentally and therefore in practice tests are done on a number of similar
fuselinks to determine a current level at which each of them will operate in a
particular time, typically 1-4 hours depending on the fuse rating. This current
is termed the 'conventional fusing current'. Further tests are done on the same
number of fuselinks to determine the current level, termed the 'conventional
non-fusing current', at which none of them will operate in the same time as
that used in the earlier tests. A current between the above two current values
10 Fuses
is then designated as the minimum-fusing current of the fuselink and for
simplicity only this value is referred to in the remainder of this chapter.
It will be appreciated that a fuselink cannot be operated near the minimum
fusing current continuously and that the rated current of the circuit and the
fuse must be at a lower level. The ratio of the minimum fusing current to the
rated current of a fuselink is called the fusing factor.
After a part or parts of a fuse element have melted, the current must continue
to flow in the liquid metal. The situation which then arises is not fully under-
stood but there are a number of conditions which may exist during the initial
stages of vaporization. During the transition period when current levels are
very high, it is possible that the material in the sections of reduced cross-
sectional area boils. There would then be bubbles present in the liquid metal
which would cause the resistance to rise, thus increasing the power input. This
would lead to rapid vaporization of the notched sections. Alternatively, it
could be postulated that the temperatures must be highest at the centres of
restricted sections, causing vaporization to commence at these points. The
vapour escaping through the surrounding liquid could produce hairline cracks
or a gap across the notch.
Whatever process takes place, the vapour in the gaps will not be ionized
initially and capacitance will be present across them. Because of the small
cross-sectional areas of fuselink elements, these capacitances must be small.
They will charge rapidly because of the current flow which will be maintained
through them by the circuit inductance. The resulting voltage will cause the
gaps to break down and thus initiate arcs. This transition period between
melting and arcing must always be of very short duration.
1.4.2 The arcing period

When an arc or arcs have been established they must persist until the current
reaches zero, at which time extinction will occur and it is clearly desirable
that the arc or arcs should not then restrike. This process must always occur
and it is desirable that satisfactory clearance should be obtained at all current
levels. At relatively low currents, the duration of arcing is very short relative
to the pre-arcing period and the effect of arcing on the energy let-through to
the protected equipment is not very significant. This is not the case at very
high current levels however, when the arcing durations may be comparable to
or even greater than the pre-arcing periods.
During the pre-arcing period associated with short-circuit conditions, i.e.
high currents, the current varies with time in a manner determined by the
circuit itself and the resistance of the fuse. It will be marginally lower than it
would have been had the fuse not been present. The current which would have
flowed in these latter circumstances is known as the prospective current and
typical variations which would be obtained in a circuit containing significant
inductance and some resistance are illustrated in Fig. 1.7.
The behaviour of cartridge fuselinks 11
,. .... \ ,. ....
I I \
I \ I \
I \ I \
\ I \ I \ Time
\./ '-.J
Fig. 1.7 Schematic of prospective (broken line) and actual (solid line) currents.
At the instant when arcs are initiated in a fuse link there is a rapid increase
in the voltage drop across it. This voltage then rises as the arcs lengthen due
to more material being vaporized from the element.
Consideration of the circuit shown in Fig. 1.8, assumed to apply for a fault
condition, shows that the following relationship applies:
e =iR + :t (L . i) + Vf
in which e is the source e.m.f., Rand L are the resistance and inductance of
the circuit, Vf is the voltage across the fuselink and i is the current.
When the current is positive as arcing commences, it is necessary that the
rate of change of current (dildt) should become negative so that the current
falls to zero. Clearly this situation will obtain more quickly and more rapid
extinction will be obtained, the greater the voltage across the fuse arcs. These
conditions are illustrated in Fig. 1.9, from which it can be seen that the current
which flows during fault conditions is lower than that which would have
flowed had the fuse not been present. Fuses therefore have the beneficial effect
R L
Fig. 1.8 Equivalent circuit.

12 Fuses
Time
Q)
~ /
o I
> I
I
I
I
I
I
I
Fig. 1.9 Electrical conditions during a short circuit. (a) pre-arcing period; (b) arcing
period; (c) fuselink voltage; (d) source e.m.f. (Reproduced from Wright and Newbery,
of providing current limiting, which assists in reducing damage to the circuits

they protect.
Fuselinks containing notched strip elements may be made to reduce the
arcing period by increasing the number of restrictions and thus the number of
arcs in series. Care must be taken however to ensure that the rate of current
reduction (dildt) is not so great that excessive voltages will be induced in
inductive components. Permissible upper limits for fuselink voltages are in-
cluded in specifications.
The ideal situation, which cannot be achieved in practice, would be for the
fuselink voltage to be at the maximum allowable value throughout the arcing
period. This would give the fastest possible operation.
The behaviour of fuselinks with cylindrical wire elements is less well con-
trolled because such elements do not have well defined restrictions at which
arcs occur during operation. In practice, however, conditions are not uniform
throughout a wire element and distortions occur during the pre-arcing period.
As a result, the cross-sectional area does vary along the element length,
producing thinner and fatter sections, known as unduloids. Gaps ultimately
The behaviour of cartridge fuselinks 13
form at the centres of thin sections, their number not being fixed however, as
in notched elements. In some cases, sufficient arcs may be formed to cause
excessive voltages to be produced in the protected circuits.
1.4.3 Determination of fuselink performance

The performances of fuselinks may be determined by testing in circuits such
as that shown in Fig. 1.10, in which the voltage, current level and waveform
may be controlled. The ability to interrupt very high currents may be ascer-
tained as well as the operating times at lower currents. The minimum fusing
current may also be determined. Because of the possible variability of fuses,
each individual test must be performed on several fuselinks.
When fuselinks are being developed for new applications for which the
required characteristics are known, designers rely on their experience. Proto-
types are produced in sufficient quantities to enable all the necessary tests to
be done. Should the performance not meet the requirements, design modifica-
tions must be made, further prototypes produced and then the tests must be
repeated. This process can be both lengthy and expensive.
There has long been a desire to calculate the performance of fuselinks
so that acceptable and optimum designs can be produced without recourse to
trial and error methods and lengthy testing procedures. In 1941 Gibson [4]
published the results of investigations, done by Professor R. Riidenberg, in
which the high-current pre-arcing behaviour of fuse links with wire elements
was determined. Because the times to reach the melting point of the elements
was very short, heat transfer from the elements was neglected. The current
flow was axial and the current density constant.
With notched elements, the current flow is not axial or of constant density
but nevertheless with various assumptions applicable over certain current
ranges and by employing finite difference or other methods, various workers
[5][6] have modelled pre-arcing performance with high accuracy in recent
years. It is now possible to determine the effects of changing the materials in
fuselinks as well as changing their dimensions and geometry.
Circuit Make
L R breaker switch
X--<>-
Voltaget .... Fuse under
source I test
Fig. 1.10 Schematic diagram of test circuit.

14 Fuses
The arcing process is more complex than that which takes place during the
pre-arcing period and accurate modelling is therefore more difficult. Some
workers have sought from experimental studies to establish empirical relation-
ships between the current through a fuselink and the voltage across it when
arcing occurs. Gnanalingam and Wilkins [7] were able to predict the arcing
performances of certain fuses with reasonable accuracy using the relation-
ship:
n
Uf Xl I
They claimed that the simulation technique is useful for screening prelimin-
ary designs and investigating the effects of system parameters such as fre-
quency.
Such methods are not based on the phenomena taking place and do not assist
in understanding the behaviour. They may not be applicable if the use of other
element or filling materials are being investigated. For this reason Wright and
Beaumont [8] developed a model based on a number of simplifying assump-
tions. They considered it to be superior to earlier models because it was based
on the processes occurring during arcing and took account of energy changes.
Work of this type is still proceeding to enable the amount of testing currently
required to be reduced greatly in the future. Details of work done recently in
this field were given in papers presented at the Fourth International Con-
ference of Electrical Fuses and their Applications [9]. It must be accepted,
however, that it will be necessary to determine fuselink characteristics ex-
perimentally for many years.
1.5 THE CONSTRUCTION OF CARTRIDGE FUSES
As indicated earlier in section 1.3, cartridge fuses are produced in a number

of forms and categories for use in low voltage and high voltage circuits.
Factors to be considered in all cases when such fuselinks are to be designed
or used are dealt with below.
1.5.1 Fuse elements

From theoretical and other studies it is evident that fuse elements should
generally be made from materials with low resistivities. If possible the materi-
als should also possess the following properties:
(a) low specific heat;

(b) low melting and vaporization temperatures;
(c) low latent heats;
(d) low density;
(e) high thermal conductivity.
The construction of cartridge fuses 15
Properties (a) to (d) are desirable when rapid operation at high currents is
necessary. At the rated current level, steady state equilibrium conditions must
be established in which the electrical power input to an element is conducted
away from it and dissipated by the connecting cables and the outer surfaces
of the fuselink. This is the reason for property (e). Of course, not only must
the element material be of high thermal conductivity but so must the surround-
ing filling material and the body. In addition, because the power which can
be dissipated from a particular fuselink is limited, it is necessary to limit the
power input to the element by using elements of adequate cross-sectional area.
After extensive studies of the available materials and taking the above
factors into account, it has been found that silver and copper are the most
suitable materials for elements and they are used in the majority of fuselinks.
Notched elements are produced in various forms, a few being shown in
Fig. 1.11.
The strip thicknesses are usually in the range 0.05--0.5 mm and the widths
seldom exceed 10 mm.
Occasions do arise when the necessary performance cannot be met using
notched elements of a single material with a relatively high melting point. In
these circumstances, advantage may be taken of the so-called 'M' effect in
which a low-melting point material is deposited adjacent to one or more of
Fig. 1.11 Various fuse element designs. (Reproduced from Wright and Newbery.
16 Fuses
======~~~==============~
'M' effect alloy
Fig. 1.12 Fuselink element with 'M' effect alloy. (Reproduced from Wright and
the restrictions on the notched strip, an example being shown in Fig. 1.12.
This method resulted from work done by Metcalf who published an article
entitled 'A new fuse phenomenon' [10] in 1939.
1.5.2 Fuselink bodies

Bodies must possess good electrical insulating properties and should not allow
the ingress of moisture. They should be reasonably good thermal conductors
and have an adequate emissivity constant so that they can transmit and radiate
heat energy emanating from the elements within them. They should also be
physically robust.
1.5.3 Filling material

The filling material, invariably quartz of high chemical purity and grain sizes
in the region of 300 microns, conducts some of the heat energy away from the
fuse element. To ensure consistency of performance it is necessary that the
packing density of the filling material be maintained at a high constant level
during production. It has also been found that this factor may have a very
significant effect on the fuse performance at high current levels. A low pack-
ing density is undesirable as it may allow arcs to expand more rapidly and
thus adversely affect the extinction process.
1.5.4 Mountings and ratings

Low voltage cartridge fuselinks of standardized performance and dimensions
are available with rated currents ranging from 2 to 1250 A to provide over-
current protection on three-phase, 415 V (line), a.c. systems and up to 250 V
d.c. systems. Designs are also available at certain ratings for a.c. circuits up
to 660 V and d.c. circuits up to 500 V.
The construction of cartridge fuses 17
Fuselinks with rated currents up to 200 A are usually fitted into fully-
shrouded fuse holders comprising a carrier and base as described earlier in
section 1.3.1 and illustrated in Fig. 1.4. Such fuses are produced in large
numbers and many are fitted in distribution fuse boards.
Fuselinks with current ratings above about 200 A are large and produced in
small quantities. They are normally installed directly in convenient positions
without their own special or standardized housings. They are often incorpor-
ated in fuse-switch units, an example being shown in Fig. 1.13.
High voltage cartridge fuselinks are produced for use in a.c. systems
operating at frequencies of 50 Hz and 60 Hz with rated voltages exceeding
1000 V. They are of the same basic design as low voltage fuses. Those with
low rated currents have silver wire elements whilst those for higher current
applications have elements of silver strip, with restricted sections along their
lengths. All elements have low melting point metals applied to them to enable
the 'M' effect, referred to in section 1.5.1, to be obtained. The elements are
long, because to achieve satisfactory current interruption, the total voltage
across the fuselink must be high and this is best achieved by having many
restrictions and therefore many arcs in series. To accommodate such elements
in a body of acceptable length, the elements are accommodated in a helical
form on insulating formers with a star-shaped cross section.
Fig. 1.13 Fuse-switch unit. (Reproduced from Wright and Newbery, 1982, Electric
Fuses with the permission of lEE).
18 Fuses
Whilst some high voltage fuselinks are mounted in air, it is nevertheless
common practice for them to be immersed in oil in the pieces of equipment
they are protecting. This is advantageous because the cooling effect of the oil
allows a given current rating to be achieved in a smaller fuselink than that
required for use in air. The smaller fuselink also operates more rapidly at very
high current levels.
Fuselinks suitable for use in systems operating at voltages up to 33 kV (line)
at a range of rated current values are produced. As an example, voltage ratings
of 3.6-7.2 kV with current ratings up to 500 A are available.
1.6 SEMI-ENCLOSED FUSES

These fuses are produced for three-phase industrial applications where the
system voltage does not exceed 240 V to earth and are available with current
ratings up to 100 A. Their breaking capacity is low, being limited to a maxi-
mum of about 4000 A.
They are also used for domestic applications where their breaking capacity
is 1 kA at 240 V a.coo
The construction of these fuses was described earlier in section 1.3.2 and illus-
trated in Fig. 1.6. Because of their cheapness they are likely to remain in demand.
1.7 EXPULSION FUSES

These fuses, which are used in high voltage circuits, contain a short element
of tin or tinned copper wire in series with a flexible braid. These items are
mounted in a fuse carrier incorporating a tube of organic material, usually
closed at the top with a frangible diaphragm, and containing a liner of gas-
evolving material. The fuse is also surrounded by a close-fitting sleeve of
gas-generating fabric. The flexible braid is brought out of the lower end of the
tube and held in tension by a spring attached to the lower end of the fuse base.
Figure 1.14 shows a cross-sectional view of a typical fuse link.
The fuse carrier has pins at the lower end which act as a hinge when it is
positioned in the lower contact of the mount. In the normal service position,
the fuse carrier is tilted from the vertical as can be seen from Fig. 1.15.
When the fuse element melts during operation, the release of the spring
tension disengages a latch, so allowing the fuse carrier to swing down due to
gravity. This provides electrical isolation.
When interrupting relatively low currents the arc extinguishes within the
fabric sleeve around the element, but at high currents the sleeve bursts and the
arc extinguishes within the liner of gas-evolving material. These fuses are thus
able to break a large range of currents.
Fuses of this type are used outdoors in three-phase circuits with current and
voltage ratings up to 100 A and 72 kV (line) respectively. Their breaking
capacity is limited to about 150 MVA.
The previous sections have dealt with the fuses used to protect plant and
Expulsion fuses 19
Flex
Tin or copper element
~~i~~~~~~i:iJcl~opper button head rivet
Solder
\ Washer
Strain tape Copper sleeve
Fig. 1.14 Sectional view of an expulsion fuselink. (Reproduced from Wright and
Fig. 1.15 Expulsion fuse assembly. (Reproduced from Wright and Newbery, 1982,
20 Fuses
connections in power supply networks in Britain and other countries. Vari-
ations do occur in the designs produced in different countries but nevertheless
the practices are broadly similar. Some specialized designs such as the boric-
acid fuses produced in the United States have not been described nor have
miniature and domestic fuses which are used in great quantities. Information
about these and other fuses is available in other works devoted entirely to
fuses [11, 12].
fAiilil!~IiIii~--- Diaphragm
J.::..-...g;;:'------ Element
...g._ _ _ _ Arc-extinguishing liquid
...r:::... . . . g . - - - - Spring
-!-_ _ _ _ Glass tube
Fig. 1.16 Sectional view of a liquid fuse. (Reproduced from Wright and Newbery,
1982, Electric Fuses with the penn iss ion of lEE).
The application of fuses 21
1.8 LIQUID FUSES
In many early fuses the arcs were quenched in liquid and this principle is
used in present-day liquid fuses. In all designs, these fuses have a glass tubular
body which is mounted vertically. The short element of silver wire or strip
is positioned near the top of the tube. The element is held in tension by a
spring anchored to the lower end of the fuse, the tube being filled with an
arc-extinguishing liquid, usually a hydrocarbon. When the element melts dur-
ing operation, the spring collapses and the arc is extinguished in the liquid. A
cross-sectional view of a typical fuse of this type is shown in Fig. 1.16. These
fuses are only used outdoors and provision is made for removing them from
and replacing them into their mountings from the ground. They are mostly
used to protect 11 or 33 kV pole or pad-mounted transformers in rural systems
and also for spurs feeding several transformers.
The breaking capacity of these fuses is lower than that of expulsion fuses
and although they have been produced in signifi~ant numbers and are still
giving good service, they are no longer recommended for new installations.
1.9 THE APPLICATION OF FUSES

Fuses are used for so many different applications in power systems that it is
impossible to deal with all of them and only the more common and important
are considered in this section. There are however some general aims and
requirements which always apply and these are dealt with initially.
A fuselink which is to protect a piece of equipment and/or a circuit should
ideally satisfy several criteria. This may be illustrated by considering an
example based on the simple circuit shown in Fig. 1.17.
The criteria are:
1. The minimum-fusing current of the fuse should be slightly less than the
current which can be carried continuously by the source, the supply and
connecting cables and the piece of equipment.
Supply cable
Fuse A
Fig. 1.17 Circuit protected by a fuse.

22 Fuses
2. The item of equipment should be able to carry currents in excess of its
rated current for limited periods and the fuse should operate, when carry-
ing these overload currents, in times slightly shorter than those which the
equipment can withstand.
3. The cables should also be able to cope with the above overcurrents without
being damaged.
4. High currents may flow due to faults within the item of equipment and in
these circumstances consequential damage should not occur to the remain-
der of the circuit. The extreme case would occur if the input terminals of
the piece of equipment became short-circuited. In this situation clearance
should be effected quickly enough to prevent damage to the cables.
5. A further possibility is a short circuit between the conductors of the
connecting cables. The most severe situation would arise if the fault was
at the input end, i.e. between points A and B. For such faults the fuse
should operate rapidly enough to prevent damage to the source and the
supply cables.
1.9.1 Time/current relationships

To achieve the above criteria, fuselinks should have time/current charac-
teristics which lie close to the withstand curves of their associated circuits as
shown in Fig. 1.18. Obviously the operating time of a fuse should always be
less, at any current level, than the period for which its associated circuit can
withstand the condition.
A very important factor which must be taken into account is that fuse
operating times include the arcing period and a change of state occurs before
clearance is effected. Once arcing has commenced it is clearly impossible for
/ Withstand curve of
/ protected circuit
Fuselink characteristic
"E
;::
Current
Fig. 1.18 Time/current characteristics of circuit and fuse.

an element to return to its original form and even when melting begins,
distortion may result and an element may not return to its original shape on
cooling. Fuselinks with low melting-point materials on their elements will be
subjected to unacceptable and irreversible changes if overcurrents pass
through them long enough to initiate the 'M' effect diffusion process.
There is thus a band below its time current characteristic in which a fuse
should, if possible, not be called upon to perform. While in practice it is not
likely that a high current will flow through a fuselink for a period just less
than its operating time, the possibility nevertheless exists. If such a condition
should arise, then the probable result, which must be accepted, is that the fuse
may operate more quickly than expected on a future occasion.
Difficulties would certainly arise if a fuselink carried a current just below
its minimum fusing value for prolonged periods. To ensure that this situation
is avoided, fuselinks are assigned rated currents somewhat below their mini-
mum fusing levels. The ratio ofthe minimum-fusing current to the rated value,
which is defined as the fusing factor, usually has a value in the range 1.2 to
2. The significance of this factor is that circuits must be able to operate
continuously at levels significantly above the rated current of the fuse protect-
ing them to satisfy criterion 1 above, namely that the continuous rating of the
circuit should exceed the minimum fusing current. This situation arises with
other protective equipment in which current settings above the full load level
are used. Clearly circuits and equipment are not normally designed without
some overload capacity, indeed factors of safety are usual and difficulties are
not encountered in providing adequate fuse protection. Clearly, relatively low
fusing factors are nevertheless desirable on economic grounds.
A further factor which must be recognized is that not all fuselinks are
capable of operating satisfactorily at all current levels between their maximum
breaking capacities and minimum-fusing values. Satisfactory arc extinction
may not take place at certain current levels, usually those just above the
minimum fusing values. Care must be taken to ensure that such fuses are only
used in circuits where currents of these levels will not be encountered or, if
this cannot be guaranteed, then some other protective scheme must be pro-
vided to open the circuit before the fuse attempts to operate. In the USA the
term 'full range' has been introduced to designate those fuses which can clear
all levels of current satisfactorily.
1.9.2 1 2 t
Fuses operate very rapidly under short-circuit conditions when the current
levels are extremely high, clearance times being typically only a few millisec-
onds. Under these conditions, the current wave shapes depend on the para-
meters of the protected circuit, the instant in the voltage cycle at which the
fault occurs and the current-limiting effect of the fuselink itself. Only a small
portion of a cycle of power-frequency current flows before interruption occurs
24 Fuses
and transient components are present. Such currents cannot be assigned a
single value which may be used to determine the corresponding operating time
from the time/current characteristic. Because of this, use is made of a quantity,
termed [2 t, which is the time integral of the square of the instantaneous current
which passes through a fuselink between the incidence of a circuit fault and
the instant at which the fuse arc is extinguished, i.e.
If it could be assumed that the resistances of the components of the protected

circuit were constant throughout the operating period of the fuselink, then the
value of [2 t would be proportional to the energy fed to the circuit and also the
heating produced by it. In practice, however, the resistances must rise, some
significantly. Nevertheless, the [2 t value is often described as the let-through
energy, a term which is not strictly correct because it does not contain a
resistance value and is not therefore in energy units (J) but in A 2 s.
To assist in the application of fuses, many manufacturers of components
determine the [2 t withstand limits of their products and publish them. The
[2 t values needed to operate fuselinks at very high currents tend to be inde-
pendent of the waveshapes and manufacturers publish these values as well as
providing time/current characteristics to enable appropriate fuselinks to be
selected by users.
In practice, fuselink manufacturers provide two sets of [2t values, one set
associated with the total clearance time and the other set, with lower values,
for the [2 t let-through during the pre-arcing period. Both sets are needed to
enable coordination to be achieved between fuselinks which are effectively in
series in networks, as is explained later in section 1.9.7.
1.9.3 Virtual time

The term virtual time is no longer used to any great extent but it may be
encountered in older publications. It was introduced to assist the selection of
fuselinks which would have the required performance under very high current
conditions. It was defined as the [2 t value divided by the square of the r.m.s.
value of the prospective current, the unit clearly being time. Again as with the
quantity [2 t, two values were produced for each fuselink, one associated with
the total clearance period and the other with the pre-arcing period.
1.9.4 Published time/current characteristics

The characteristics published by fuse manufacturers are derived from tests
done on fuselinks which are at a temperature in the range lso-2SoC when
current flow through them is commenced. It must be recognized that slightly
faster operation will be obtained in practice if a fuselink is in an environment
with a high temperature or if a significant load current has been flowing
through it for a period before an overcurrent condition occurs. This behaviour
is somewhat counterbalanced by the fact that the times for which the protected
circuit can carry given overcurrents will be reduced if load currents have been
flowing in it and/or if it is operating in an environment with a relatively high
temperature.
1.9.5 Cut-ofT characteristics

Manufacturers produce characteristics to show the highest possible values of
current which given current-limiting fuselinks will carry for varying values of
prospective current. They are used when checking that the mechanical forces
to which items in the protected circuit will be subjected during short-circuit
conditions are within the permissible limits. A typical characteristic is shown
in Fig. 1.19.
1.9.6 Operating frequency

The characteristics provided for fuselinks are usually based on operation at 50
or 60 Hz. In general, the higher the system frequency the shorter are the
operating times at very high current levels, because of the shorter duration of
the half cycles of the supply voltage. Care must therefore be taken in selecting
fuselinks for use in circuits operating at frequencies below 50 Hz. It must be
realized that their operating times in such circumstances will be longer than
those obtained when operating at 50 Hz. As a result higher arc energies will
be released in the fuselinks, and to ensure that satisfactory clearance will be
obtained, their ratings in terms of operating voltage may have to be lowered.
Characteristics
L:....-------- } for different
L~-------- rated currents
Prospective current r.m.s.
Fig. 1.19 Fuse cut-off characteristics.

26 Fuses
For d.c. applications the voltage rating must usually be reduced very signific-
antly, particularly if the circuit is highly inductive.
1.9.7 Discrimination and coordination

Most power circuits contain many items of equipment and these are both in
series and parallel paths. To ensure a high continuity of supply, several
protective devices must be included to ensure that only the minimum of
interruption will occur to clear any fault condition. The protective devices
must therefore be so coordinated that correct discrimination is achieved.
In practice, some networks are protected by fuses only and in other cases
fuses are used near the loads while feeder connections and equipment are
protected by relays. Methods of achieving correct discrimination in these two
situations are considered separately below using the simple network shown in
Fig. 1.20 as an example.
Networks protected by fuses

It is very common to use the above radial system and to have an upstream fuse
(PDl) in the supply connection and downstream fuses (PD2, 3 and 4) in the
individual load circuits. Each of the fuses PD2, 3 and 4 must have time/current
characteristics which will protect their associated loads. A fault on a particular
load should be cleared by its associated fuse, and although the upstream fuse
(PD 1) will carry the fault current and the other load currents it should not
operate or be impaired.
At medium and low fault-current levels, discrimination will be ensured
provided that the time/current curves of the downstream fuses are to the left
of the curve for the upstream fuse. The time (vertical) displacements between
the curves should nevertheless be sufficient to provide an adequate margin to
PD2
Load 1
PD1 PD3
Supply -F~-+-+--+----l Load 2
PD4
Load 3
Fig. 1.20 A simple network containing fuses.

allow for the manufacturing tolerances of the fuselinks and because the up-
stream fuselink will be carrying current to healthy circuits in addition to that
in the faulted circuit.
At levels of fault current which will result in a downstream fuselink melting
in less than 100 ms, [2 t values must be compared. To ensure correct discrim-
ination, the pre-arcing [2 t of the upstream fuselink should exceed the total
operating [2 t of the downstream fuselinks by a significant margin, say 40%.
Clearly the effects of the load currents of the healthy circuits flowing through
the upstream fuse link are negligible when a very high fault current exists.
Pre-fault conditions must also be considered and the [2 t margin should be
increased if a downstream fuselink may at times be much less loaded relative
to its rated current than the upstream fuselink. A suggested, but rough guide,
for the extreme case where a downstream fuselink may be on no load for
periods during which the upstream fuselink is nevertheless carrying a current
near its rated value is that the upstream fuse rating should be increased by a
further 25%.
Networks protected by fuses and other devices

In such networks, when fuses are used as the downstream devices (PD2, 3 and
4), they must be chosen in the same way as above, their time/current charac-
teristics being such that they will adequately protect their associated circuits.
Thereafter the upstream device (PDt) must have a characteristic which will
provide correct discrimination and there is usually little difficulty in obtaining
satisfactory performance.
In some cases, miniature circuit-breakers incorporating overcurrent protect-
ive features are used as the downstream devices and a fuse link is used as the
upstream device. With this arrangement there is always an upper limit to the
fault current at which discrimination can be achieved. This is because mechan-
ical devices have a definite minimum operating time irrespective of the current
level. The upstream fuse, however, will have an operating time which de-
creases continually as the current increases. As a result, the curves of the
various devices must cross at certain current levels, above which correct
discrimination will not be achieved.
The use of two co-ordinated fuses

As stated earlier, some current-limiting fuselinks are unable to satisfactorily
interrupt currents across the whole range from their maximum breaking capa-
cities to their minimum fusing levels. When this is the case, it is currents
somewhat above the minimum fusing level where operation is unsatisfactory.
In these circumstances, a back-up fuse should be provided, the fuse charac-
teristics being chosen to provide an overall composite characteristic of the
form shown in Fig. 1.21. In this way, the desirable current-limiting property
28 Fuses
1000
100
~
10 IMax breaking
. current of
Iexpulsion fuse
Q)
E
i=
0.1
Current
Fig. 1.21 Combined characteristic of two fuses. (a) back-up fuse characteristic; (b)
current-limiting fuse characteristic. (Reproduced from Wright and Newbery, 1982,
may be obtained under short-circuit conditions as well as satisfactory clear-

ance being achieved over the full operating current range.
1.9.8 The protection of power-system equipment

The preceding treatment has been general, but in practice special requirements
arise when particular items in power systems are to be protected and some of
these are considered below.
Cables
International Electrotechnical Commission (IEC) Publication 364 deals with
Electrical Installations in BUildings. In the UK, the Institution of Electrical
Engineers (lEE) produced the 15th Edition of its Regulations for Electrical
Installations in 1981, this being based on IEC Publication 364. In these regu-
lations, the term overcurrent covers both short-circuit currents and overloads,
an overload being defined as an overcurrent which flows in a circuit which is
perfectly sound electrically.
The current-carrying capacities of cables, which are dependent on many
factors including the environments in which they are to operate, have been
determined under a range of conditions and tabulated in the regulations
referred to above. A cable can, of course, carry currents above its continuous
current-carrying capacity for limited periods and they therefore have inverse
withstand time/current characteristics. Clearly, the characteristic of each fuse
should be such that fault clearance at any current level will be effected within
the period for which the protected cable can carry that current. In addition,
the minimum fusing current should be below the continuous current-carrying
capacity of the cable. To comply with these conditions and taking account of
the fusing factor (say 1.5) would require a cable to operate at only about 66%
of its current capacity. This practice could be unacceptably expensive and
regulations, which are intended to ensure that the life of cable insulation is
not significantly shortened as a result of running the conductors at high tem-
peratures, permit the minimum operating current of protective devices to be
up to 1.45 times the current-carrying capacity of the cable.
To provide adequate protection under short-circuit conditions, fuselinks
should have let-through ]2t values lower than those which can be withstood
by the cables. These latter values can be determined from the recent edition
of the lEE Regulations. In practice it is satisfactory to check whether satisfac-
tory coordination will be achieved using the ]2 t value of the fuselink needed
to provide clearance in 5 s. Further information is available on the above
topics in reference [11].
Motors
It is well known that motors draw surges of current on starting, the magnitude
and duration of the surges depending on the motor, the starting method in use
and the load on the motor. As an example, three-phase a.c. induction motors
up to 2 MW rating operating at line voltages up to 11 kV and employing
direct-on starters may draw 5 to 6 times their full-load currents on starting. In
such applications, current-limiting cartridge fuselinks with rated currents up
to twice the full-load current of the motors to be protected are used, to ensure
that fuse operation does not occur during the starting period. Such highly-rated
fuses provide protection against the very high currents associated with short
circuits, but either air-break or vacuum contactors must also be included with
settings such that they will clear the lower currents associated with overload-
ing. The fuses must be able to carry three times their rated currents during
starting periods and they will reach higher temperatures at these times than
they do when carrying their rated current continuously. The expansion and
contraction which results could cause long elements, such as those used for
high voltage applications, to fail mechanically after a number of starts and
therefore special fuselinks with corrugated elements are produced for motor
protection in these cases. Normal low voltage fuselinks are satisfactory for
this duty, however, because their elements are relatively short and sufficiently
robust.
30 Fuses
When other methods of starting which cause the surges to be smaller are
employed, fuselinks with lower current ratings may be used. Again, these
levels are exceeded during motor starts and fuselinks with special elements
are needed for high voltage applications. It is also necessary for such fuselinks
to have current ratings of at least 125% of the rated current of the protected
motor to limit the temperatures of the elements under normal running condi-
tions to acceptable levels.
To completely protect a motor may require the use of fuses and also several
other devices including inverse-definite-minimum time (IDMT) overcurrent
relays, instantaneous overcurrent and earth fault relays, all the relays operating
a contactor. Clearly these various items must be correctly coordinated.
Power transformers
Many step-down transformers in distribution networks and industrial premises
utilize fuses for their protection on both their high voltage and low voltage
sides.
In Britain high voltage fuses are used in the primary circuits of three-phase,
11 kV/415 V transformers rated up to 1.5 MV A and low voltage fuses are
included in the secondary circuits. High voltage fuses are also included on
both sides of three-phase, 33/11 kV transformers rated up to 5 MV A.
Factors which influence the selection of these fuses are considered
below.
1. Fuses in the secondary circuit of a transformer have to be chosen to protect

the loads and connecting cables while those in the primary circuit must
isolate the transformer if a fault occurs within it.
2. Because the magnetic core of a transformer may go into saturation re-
peatedly over a significant period after the primary winding is connected
to a supply, so-called exciting-current surges of the form shown in Fig.
1.22 may be experienced. These surges may reach peak values many times
the rated current of a transformer even though the normal exciting current
may only be 2 or 3% of the rated current. Clearly they should not cause
the fuses on the primary side to operate. When possible, the magnitudes
and durations of the surges, which depend not only on the transformer to
be protected but also on the high voltage circuit parameters, should be
determined for each application so that fuses with the suitable charac-
terisfics may be chosen. In practice it has been found for distribution
transformers, when actual surge-current information is not available, that
the surge may be taken to be equivalent to 12 times the transformer
full-load current for a duration of 100 ms. This rule, which has been in
use for some time and found to be satisfactory, is specified in the British
Electricity Supply Industry (ESI) Specification 12-8. It is sometimes
supplemented by the requirement that melting of the fuselink elements
1"" __
---
Time
Fig. 1.22 Exciting current of a power transformer. (Reproduced from Wright and
should not begin in less than 10 ms when they are carrying 25 times the
rated current of the protected transformer.
3. A further factor which should be considered is that power transformers
may be deliberately operated above their rated-current levels for predeter-
mined periods, often of several hours, this being allowable because of their
long thermal time constants. To allow this practice, the fuses should have
ratings based on the maximum currents which may flow.
4. In those applications where a transformer is supplied by overhead lines,
the possibility of overvoltages being impressed on its windings in the
event of lightning strikes should be considered. The fuses in the primary
circuit should ideally be able to withstand the resulting high currents
without operating, but this often requires such highly-rated fuselinks that
adequate protection is not provided for other conditions. A compromise
has to be struck and some degree of risk must be accepted in most cases.
5. To enable faults within a protected transformer to be cleared rapidly, the
currents needed for operation in the 10 s region should be as low as
possible.
6. The minimum-fusing current of the primary-circuit fuselinks should be as
low as possible to enable them to clear as many internal faults as possible.
It must be accepted, however, that some conditions, such as interturn
faults, may cause primary currents of less than the full-load value to flow
and clearly these will not cause fuse operation.
7. Correct discrimination between the fuselinks on the high and low voltage
sides of a transformer and between the transformer fuses and other pro-
tective devices on a network should be achieved under all conditions.
Account must be taken of the fact that the transformation ratio is not fixed
if a transformer is fitted with tap-changing equipment and that various
32 Fuses
a b
Q)
E
i=
10~~----------------------
Current
Fig. 1.23 Fuselink characteristics. (a) low-voltage fuse; (b) high-voltage fuse. (Repro-
duced from Wright and Newbery, 1982, Electric Fuses with the permission of lEE).
current distributions are possible in the phase windings when unbalanced

faults occur if the high and low voltage winding connections are different,
for example star-delta.
In addition to the above, account must be taken of the fact that the charac-
teristics of the high and low voltage fuselinks will differ because their designs
will be dissimilar. In general, the high voltage fuselinks will tend, at high
current levels, to have smaller operating-time changes for a given percentage
current change than the low voltage fuselinks and the time/current charac-
teristics will converge at intermediate current levels before diverging again at
lower currents. This situation is illustrated in Fig. 1.23.
To assist users, low voltage fuse links complying with BS88 : Part 5 have
standardized characteristics which provide discrimination with the stand-
ardized characteristics of high voltage fuselinks. This is explained in ESI
Standard 12-8.
Voltage transformers
Electromagnetic voltage transformers are used on systems operating at volt-
ages up to about 66 kV (line) to provide outputs up to a few hundred volt-
amperes to measuring equipment and relays at low voltages, now standardized
at 63.5 V per phase (110 V line) in Britain and many other countries.
Their step-down ratios are high and therefore the primary windings carry
very small currents under normal conditions. Because the transformers are
small physically, compared with power transformers, they have magnetic
cores of small cross-sectional area and consequently they have primary wind-
ings of many turns of fine wire. They are thus more vulnerable than most other
transformers.
Providing protection against all possible faults within a voltage transformer
would necessitate the use of fuselinks, in the high voltage side, with rated
currents so low that their elements would be extremely fragile and liable to
break because of factors such as vibration. This would be completely unac-
ceptable because it would cause vital protective and control equipment to be
de-energized and therefore either inoperative or even worse, causing incorrect
tripping or control of power circuits.
A compromise must therefore be accepted and it is now standard practice
for high voltage fuselinks to have minimum rated currents of 2-3 A at all
voltage levels. Such fuselinks only operate when relatively large faults occur
within a transformer. They do not prevent damage occurring and it is expected
that small internal faults will develop until fuse operation is initiated. Some
users feel that the measure of protection provided by such fuses is not suffi-
cient to warrant their installation and they prefer to rely solely on other
protection in the system.
Fuses are included in the secondary circuits of voltage transformers to
provide protection in the event of faults in the burdens. There is no difficulty
in selecting suitable fuselinks for this duty because the secondary circuits
usually carry currents of a few amperes. Coordination with the protection on
the high voltage side does not cause any problems because the referred rated-
current levels of the secondary fuselinks are very low.
Power-system capacitors
Capacitors are used quite extensively to achieve power factor improvement.
In low voltage situations it is usual to have a single capacitor in each phase.
Fuse manufacturers are usually prepared to provide simple recommendations
in these cases. These are based on past experience and take into account the
high transient inrush currents which may flow when the capacitors are
switched into circuit and also the significant harmonics which may be present
in the capacitor currents because of their reduced reactances at higher frequencies.
As an example, when protecting capacitors of ratings greater than 25 kVA,
using fuselinks satisfying BS88 : Part 2, it is recommended that the fuse
ratings be at least 1.5 times the full-load current of the capacitors.
In large installations for operation at higher voltages, each phase contains
several units made up of individual capacitors connected in series and parallel
groups. The units may also be connected in series and parallel groups. Both
star- and delta-connected three-phase capacitor banks are produced for ratings
up to 1 MV A and for voltages up to 11 kV (line). For higher voltage systems,
however, banks are usually connected in star.
34 Fuses
In most European countries it is the practice to provide a fuse for each of
the individual capacitors forming the units, the fuses having the appropriate
relatively low current-rating and breaking capacity. In Britain however it is the
practice to provide only one fuse for each unit, rather than for each capacitor,
and frequently a line fuse for each phase is also included. Clearly either an
individual capacitor fuse or a unit fuse in' British installations should operate
when necessary to isolate only part of the complete capacitor bank so that the
remainder may be left in service.
Fuselinks, associated with capacitors, must meet the normal requirements
of fast operation in the event of faults and also be able to carry the rated
currents of the capacitors they protect as well as transient currents. The latter
will flow if the voltage across a bank changes suddenly, a condition that may
arise at instants when the supply is connected to a bank or when system faults
affect the network voltages. The magnitudes of these transient charging or
discharging currents and their durations are dependent on the capacitances
and the parameters of their associated circuits. To prevent undesired
fuselink operation under these conditions it is usually necessary to select
fuselinks with rated currents significantly greater than the normal current
through them.
In the event of a breakdown within an individual capacitor or a short circuit
across a unit, current will flow into it from other healthy capacitors and units.
Considering, as an example, a bank of four units connected in series-parallel
as shown in Fig. 1.24, it can be seen that the discharge current (idf) of the
short-circuited unit would not flow through its associated fuse link. Each of
the two lower fuselinks would carry the current (ich) needed to double the
voltage across its associated unit and the upper healthy unit would carry the
current (idh) required to reduce the charge on it to zero. In practice, if each of
the units were of the same capacitance, then the surge currents idf, ich and
idh would be of approximately equal magnitudes and the fuselink associated
with the faulted unit would carry three times the currents flowing in each of
the other fuses. Following the transient surges, the fuse link protecting the
faulted unit would carry four times its normal current to feed twice normal
current to each of the lower units which would then have their voltages
doubled. In these circumstances, correct discrimination would be readily
achieved.
A further factor for which allowance must be made is that when the fuselink
associated with the faulted unit operates, the voltage across the other upper
unit will rise, for this example to 133%, of its earlier value, and the current
through it will also rise similarly. The VAR input would thus rise by 77%.
Clearly similar situations would arise whichever unit in the bank became
short-circuited and all the units would have to be capable of operating above
their normal VAR level. Actual arrangements with other series/parallel com-
binations of the capacitors should be considered in a similar way to determine
the conditions which will arise during and after fuse operation.
Short-circuit
Fig. 1.24 Current flows due to the short-circuiting of a capacitor unit.

isc = short-circuit current
ich = charging current of healthy unit
idh = discharge current of healthy unit
idf = discharge current of faulty unit
isc = 2 ich + idh + idf.
(Reproduced from Wright and Newbery, 1982, Electric Fuses with the permission of lEE).
Other equipment
Fuses are used to protect many items in the load circuits of power systems,
very large numbers being associated with semiconductors employed in the
expanding field of power-electronic equipment. As in all applications, the
characteristics and behaviour of each fuse link should be such that costly items
will not be damaged in the event of faults. In this connection semiconductors
tend to require more rapid fuse operation at high currents than items such as
motors, because their thermal capacity is relatively low. Information about the
withstand abilities of their components is provided by manufacturers and it
should be used to enable the necessary fuselinks to be selected for particular
applications.
Special requirements arise in some applications because of mechanical or
environmental conditions. As an example, the rectifiers in the field-winding
circuits of brushless alternators and their associated fuselinks are mounted on
the rotor, which may revolve at speeds up to 50 or 60 rev/so Special fuselinks
which can withstand the vibration and the large forces encountered in this
situation are necessary.
36 Fuses
Because of the variations in the requirements which must be met by fuses
to be used in the many applications which exist, the IEC has produced applica-
tion guides. These guides explain the basis of the ratings and how to relate
them to practical situations.
Other detailed information on topics such as special applications, testing,
manufacture, quality control and International and National Standards, which
cannot be included in this work, is given in references listed under Further
Reading at the end of this chapter.
1.10 THE FUTURE

The pre-arcing phenomena which occur in fuse links are now well understood
as a result of studies undertaken in recent times. The arcing behaviour is less
clearly understood however, and it is still being investigated both theoretically
and experimentally. The aim is to obtain information on the effects which may
be produced on time/current characteristics and interrupting performance by
changes to the following parameters:
(a) element material

(b) element geometry
(c ) filling material
(d) filler grain size
(e) body material and geometry
Such work should eliminate much of the testing associated with trial and
error methods of producing fuselinks for new applications. It should also
enable better protection to be provided for existing equipment as well as
enabling products such as new semi-conductor devices, to be adequately
protected.
The production and use of fuselinks has risen dramatically over the last
hundred years and it seems likely, because of their attractive qualities, that
they will continue to be used for many years to protect items of equipment in
the lower voltage sections of power systems.
REFERENCES
l. Cockburn, A. C. (1887) On safety fuses for electric light circuits and on the
behaviour of the various metals usually employed in their construction. J. Soc
Te/eg Eng, 16,650-665.
2. Clothier, H. W. (1933) Switchgear Stages, (printed by G. F. Laybourne and Unwin
Brothers).
3. Clothier, H. W. (1902) The construction of high-tension central-station switch-
gears, with a comparison of British and foreign methods, Paper presented to the
Manchester Local Section of the lEE, 18 February, 1902. Printed in reference [2],
pp 1-19.
Further Reading 37
4. Gibson, 1. W. (1941) The high-rupturing-capacity cartridge fuse, with special
reference to short-circuit perfonnance. J. lEE, 88, (1), 2-24.
5. Leach, J. G., Newbery, P. G. and Wright, A. (1973) Analysis of high-rupturing-
capacity fuselink prearcing phenomena by a finite-difference method, Proc. lEE,
120, (9), 987-993.
6. Wilkins, R. and McEwan, P. M. (1975) A.C. short-circuit perfonnance of unifonn
section fuse elements, Proc. lEE, 123, (3), 85-293.
7. Gnanalingan, S. and Wilkins, R. (1980) Digital simulation of fuse breaking tests,
Proc. lEE, 127, (6), 434-440.
8. Wright, A. and Beaumont, K. J. (1976) Analysis of high-breaking-capacity fuse-
link arcing phenomena, Proc. lEE, 122, (3), 252-260.
9. Proc. of Fourth Int Conf on Electric Fuses and their Applications, 22-25 Sept
1991, Nottingham, England, ISBN 0 95 148 28 15
to. Metcalf, A. W. (1939) A new fuse phenomenon, BEAMA. J., 44, 109-151.
11. Wright, A. and Newbery, P. G. (1982) Electric Fuses, Peter Peregrinus.
12. Wright, A. (1990) Construction, behaviour and application of electric fuses,
Power Eng. J., 4, 141-148, (1990) Application of fuses to power networks, ibid.,
4, 298-296; (1991) Application of fuses to power-system equipment, ibid., 5,
129-134.
FURTHER READING
Wright, A. and Newbery, P. G. (1982) Electric Fuses, Peter Peregrinus, ISBN 0
906048788.
Miniature fuses
lEe Publications
127 (1974), 127A (1980), 127-3 (1984); 257 (1968), Amendment No 1 (1980).
British Standards
2950 (1958); 4265 (1977).
Low voltage fuses
lEe Publications
269, Parts 1 (1986), 2 (1986), 2A (1975), 3 (1987), 3A (1978), 4 (1980).
British Standards
88, Parts 1 (1975),2 (1975), 4 (1976),5 (1980); 646 (1977); 714 (1978); 1361 (1971);
1362 (1973); 3036 (1978).
38 Fuses
British Electricity Supply Industry Standards (ES/)
12-10 (1978).
British Ministry of Defence Standards (DEF STAN):

59-96 Part 1 Fuselinks; 59-100 Part 1 Fuseholders.
High voltage fuses

lEe Publications
282, Parts 1 (1985),2 (1970), 3 (1976); 291 (1969), 291A (1975); 549 (1976); 644
(1979); 691 (1980); 787 (1983), Amendment No 1 (1985).
British Standards
2692, Parts 1 (1975),2 (1956).
British Electricity Supply Industry Standards

12-8 (1976).
2
Current transformers
The term current transformer (commonly abbreviated to ct) is used to describe

a piece of equipment consisting of a pair of mutually-coupled windings
mounted around a core, usually of magnetic material. Such transformers are
normally used to step down high currents flowing in their primary windings
to lower levels suitable for feeding to measuring and protective equipment.
Isolation of the latter equipment from the primary circuits is ensured by the
insulation normally associated with the primary windings.
Current transformers have been produced and used in very large numbers
at all voltage and current levels in power systems around the world for most
of this century.

During the last two decades of the nineteenth century, electricity supply
systems were being set up to meet the demands caused by the rapidly expan-
ding use of electric lighting.
Some of the initial installations provided direct voltage supplies, but it was
soon recognized that alternating voltage systems were preferable because
higher transmission voltages could be used and greater efficiency was thereby
achievable. This development, of course, depended on the availability of
suitable power transformers.
The principle of electromagnetic induction on which all transformers are
dependent had been demonstrated by Michael Faraday in 1831 and indeed it
is recorded [1] that by 1838 Joseph Henry was able to change an 'intensity'
current, as he called a current in a high voltage circuit, to a 'quantity' current,
by which he meant a current in a low voltage circuit. He had effectively built
a step-down voltage transformer, which he called a voltage changer, but it was
only used subsequently in laboratory demonstrations.
In 1882 Goulard and Gibbs took out British patents on a transformer de-
signed to, as they said; isolate consumers from the power lines and for reduc-
ing the mains voltage to a safe value. Systems were then developed using these
transformers but they proved unsatisfactory because the primary windings of
the transformers supplying individual consumers were connected in series

40 Current transformers
with each other. In 1885, three Hungarians, Zipernowsky, Beri and Blathy,
realized that parallel connections should be used. In the same year it is re-
corded [2] that Dr S. Z. de Ferranti, who was working on the Grosvenor
Gallery Station in New Bond Street, London, had parallel-connected trans-
formers installed to step down from the 2400 V output voltage of the alterna-
tors then in use.
Thereafter, alternator and transmission voltages rose quite rapidly, trans-
formers being used at first, however, only to step down the voltages supplied
to the consumers.
In 1891, at the Electrical Exhibition in Frankfurt, a transmission line 109
miles long was energized at 25 kV from a transformer to demonstrate that
power could be transmitted over long distances, and by 1897 machines oper-
ating at 13 kV were installed and operating at Paderno in Italy.
At this time, current and voltage transformers could readily have been
produced because of their basic similarity to power transformers and it seems
probable that they were installed to feed measuring instruments associated
with the high voltage circuits then in existence.
Another matter of significance is that doubts were being expressed around
the tum of the century about the use of fuses as protective devices, possibly
because their manufacture was not adequately controlled. Apparently they
were causing unnecessary interruptions of supplies to consumers. This was
referred to by Andrews [3] in a paper presented to the lEE in 1898. The
following two quotations are from that paper:
'Some engineers attribute the immunity of their systems from failures (i.e. inter-
ruptions of supply), to the fact that they use fuses made from copper of the same
cross-sectional area as the mains'.
'If any fuses that it is customary to be used can be omitted, everyone will admit
that they are a source of danger, and consequently better omitted'.
Because of this situation and the fact that systems could not really be left
unprotected, so-called cut-outs, which had contacts capable of interrupting
fault currents, were developed. They incorporated magnetic devices which
operated when overcurrents above certain levels flowed. Subsequently, cir-
cuit-breakers with trip coils were produced. By this time power system volt-
ages and currents had reached levels at which existing fuses could not be used.
One method of dealing with this situation was to connect a fuse, of low current
rating, across the trip coil of a circuit-breaker and to feed this combination
from a current transformer, as shown in Fig. 2.1. The alternative of using a
current transformer to feed a relay, which in tum operated the trip coil, was
also possible. Current transformers therefore became essential components in
power systems, feeding both measuring and protective equipment.
Although current transformers were probably used several years earlier, as
suggested above, the first indication of their use which the authors have been
able to trace was in a paper presented to the lEE in 1902 by Clothier [4]. Figure
Is Trip coil
Fig. 2.1 Fused trip coil.
16 in that paper showed a device labelled as an ammeter transformer. The term

current transformer was used in 1904 in a paper, also presented to the lEE, by
Edgcumbe and Punga [5]. Certainly, current transformers were in wide use by
1910 when Young [6] presented a paper to the lEE on their theory and design.
He described transformers produced by several companies, including a type
suitable for use at voltages and currents up to 7 kV and 500 A respectively,
which was manufactured at that time by British Thomson Houston Ltd of
Rugby, England. It is shown in Fig. 2.2.
Fig. 2.2 Current transformer manufactured by British Thomson Houston Co Ltd.

(Reproduced from Young, 1910, Journal of the lEE, 45, with the permission of lEE).
2.2 CONVENTIONAL CURRENT TRANSFORMERS
Information about the behaviour, construction and use of current transformers
is provided in this section.
2.2.1 Equivalent circuits

Current transformers behave similarly to all two-winding transformers and
may be represented by the well-known circuit shown in Fig 2.3(a) which is
based on the winding directions and current and voltage polarities shown in
Fig. 2.3(b).
Because the connected burdens are normally of low impedance, the second-
ary output VA and voltage are relatively low, typically a maximum of 20 VA
and 20 V for a current transformer with a secondary winding rated at 1 A. The
voltage across the primary winding, because of the turns ratio, is not therefore
likely to exceed a fraction of one volt, a negligible value relative to the rated
voltage of the primary circuit. For this reason, the equivalent circuit of a
current transformer can be simplified to that shown in Fig. 2.4. This circuit
can be made to represent any current transformer operating with any burden
and primary current under either steady state or transient conditions.
Rp Lp Rs
Np Ns
1v, ept tes Burden
- ip s s
(a)
Start __ t.e.r!.o
V'I ~J Ie: lv,

----0
Finish
Finish
(b)
Fig. 2.3 Equivalent circuit of a current transformer.

Conventional current transformers 43
Burden (Zo)
Fig. 2.4 Equivalent circuit of a current transformer referred to the secondary.
It will be seen that the exciting current (ie) is dependent on the exciting
impedance, presented by Re in parallel with Lm and the secondary e.m.f. (e s)
needed to drive the secondary current (is) through the total secondary-circuit
impedance. Because the secondary current of a current transformer may vary
over a wide range, i.e. from zero under no-load conditions to very large values
when there is a fault on the primary circuit, the secondary e.m.f. and excitation
current may also vary greatly and in this respect the behaviour is very different
from that of voltage transformers.
Because of the non-linearity of the excitation characteristics of the magnetic
materials used for transformer cores, the exciting impedance of a given current
transformer is not constant, both the magnetizing inductance (Lm) and loss
resistance (Re) varying with the core flux (<p), needed to provide the secondary
e.m.f. (e s )' Allowance may be made for this non-linearity, if necessary, when
determining the behaviour of a particular transformer under specified condi-
tions, calculation then being done using step by step or other methods. If great
accuracy is not required, however, simplifications can be effected by assigning
constant values, the averages over a cycle, to Re and Lm.
An alternative method of representing a current transformer is to employ
the concept of mutual inductance (Mps )' By definition and based on the con-
ventions used above:
es = - Mps dip/dt
when there is zero current in the secondary winding. It has been shown [7]
that the circuit shown in Fig. 2.5 is equally as satisfactory a model as that
shown in Fig. 2.4.
The self-inductance, Lss of a secondary winding is given by:
Lss = Ls - Mps Ns/Np
Fig. 2.5 Alternative equivalent circuit of a current transformer.
in which Ls is the leakage inductance used in the circuit of Fig. 2.4. It is clear
therefore that the value of Ls for a given transformer may be determined
experimentally by measuring its self (Lss) and mutual (Mps) inductances.
2.2.2 Behaviour under normal steady state conditions

Provided that the impedance of the burden connected to a current transformer
does not exceed the value corresponding to the rated VA output of the trans-
former, and the primary current is sinusoidal and of the rated level or less,
then the e.m.f. induced in the secondary winding will not be great enough to
cause the core to saturate. The exciting m.m.f. of a well-designed transformer
operating under these conditions will be very small compared with the primary
winding ampere turns (ipNp). The exciting current (ie) will nevertheless in
practice be non-sinusoidal.
A fundamental, which applies to all two-winding transformers, with the
conventions adopted in the previous section, is that the sum, at every instant,
of the ampere-turns provided by the primary and secondary windings
(ipNp + isNs) must be equal to the ampere-turns needed to excite the core
(ieN s). Consequently, if the primary current wave is a pure sinusoid, then the
secondary current must contain harmonics of the same size as those present
in the exciting current. The resulting degree of distortion of the secondary
current waveform will nevertheless be very much less than that of the exciting
current because the secondary current will contain a much greater fundamental
component than the exciting current, maybe 100 times. Secondary current
harmonics are therefore negligible and for all practical purposes the secondary
waveform can be regarded as sinusoidal.
On this basis, if the burden and secondary winding impedances are linear
and equal to Zs L~ then an e.mJ. of:
es = lspk Zs sin (rot + c:x + ~)
must be induced in the secondary winding, if the secondary current is given

by:
is =lspk sin (rot + c:x)
To provide the necessary secondary e.m.f. (e s ), the core flux density must
vary as follows:
(2.1)
It is normal when determining the accuracy of transformation to assume that

the core flux alternates symmetrically about zero, in which event the constant
kl in equation (2.1) is zero. The exciting m.mJ. needed for a particular con-
dition could be determined from the excitation characteristic of the core,
which will be of the form shown in Fig. 2.6. It is this need for an exciting
m.mJ. that causes errors in practical transformers and at any instant the
secondary current is given by:
(2.2)
Fig. 2.6 Core-excitation characteristic.

exciting m.m.f.
in which ie = Ns
It is difficult to express the error of a transformer in simple terms if the
excitation non-linearity is taken into account and it is accepted practice to
express errors in current and phase form by considering only the fundamental
components of the exciting current and secondary current. Doing so is equi-
valent to assuming that the B - H characteristic of a core is elliptical. This is
also the case when the components Re and Lm in the equivalent circuits,
described in the previous section, are assumed to be constant. A steady state
phasor diagram of the form shown in Fig. 2.7 may then be produced.
Current error is defined as the amount by which the actual transformation
ratio departs from the desired ratio (Kn). It is therefore given by:
IIp I
---Kn
IIs I
Current error = K per unit (2.3)
n
The phase error is defined as the angle (9) between the phasor representing
the primary m.m.f. IpNp and the secondary m.m.f. reversed (-IsN s). It is
regarded as positive when the secondary m.m.f. reversed leads the primary
m.mJ. It has been shown [7] that current transformers which have a turns ratio
equal to the desired ratio (i.e. NslNp = K n ), have positive current errors, i.e.
IIs I < IIp INp INs, unless the secondary circuit has a leading power factor of
less than cos 0, in which 0 is tan - 1 lIe lIm I and their phase errors are also
positive unless the secondary circuits have a lagging power factor of less than
cos (1t/2 - 0).
Turns compensation
Consideration of the equivalent circuit and the phasor diagram shown in Figs.
2.4 and 2.7 respectively shows that the excitation current (Ie) would, for a
Fig. 2.7 Phasor diagram for a current transformer.

fixed burden impedance, be proportional to the secondary circuit (Is) if the
excitation characteristics of its core were linear. If this were so, both the
current and phase errors would be constant and independent of the primary
and secondary current levels. In these circumstances, the current error could
be reduced to its minimum possible level by using the appropriate number of
secondary winding turns. Clearly if the current error would have been positive,
i.e.IIp INp I
> lIs Ns, with a turns ratio equal to the desired transformation
ratio, then more accurate behaviour could be produced by decreasing the
number of secondary winding turns.
In practice, the current error does vary with secondary current level and
burden impedance, but nevertheless, secondary winding turns may be adjusted
to take advantage of the full range of current errors allowable in the various
classes of accuracy quoted in standard specifications. As an example, Class 1
accuracy in BS 3938 permits current errors up to 1% and the use of extra
secondary winding turns enables the permissible negative error band to be
used.
This adjustment of secondary winding turns from the nominal value is
known as turns compensation and it can be done with considerable precision
on transformers with a large number of secondary turns. For instance, an
adjustment of one secondary tum in the secondary winding of a transformer
with a ratio of 500/1 and a single-turn primary winding would change the
current error by 0.2%, whereas in a transformer of ratio 500/5 the change
would be 1%. It will be evident that turns compensation does not affect the
phase error significantly.
The magnitudes of transformation errors

It is clear from the above that the accuracy obtained in any given operating
condition can only be improved by reducing the ratio of the excitation m.m.f.
needed to support the core flux, relative to the m.m.f. provided by the second-
ary winding. This ratio is affected by the core material and dimensions, the
number of primary and secondary winding turns, the configuration of the
secondary winding and the size of its conductors. These factors are considered
separately below.
Core materials and dimensions The core-flux variation in a given trans-

former supplying a particular burden is proportional to the e.m.f. which must
be induced in the secondary winding and thus the secondary current. For any
value of core flux, the exciting current is proportional to the length of the path
traversed by the flux, and inversely proportional to the cross-sectional area of
the core. Ideally, therefore, a ring core of the smallest possible diameter and
greatest possible cross-sectional area should be used. In practice, however, the
core dimensions are limited by factors such as the size of the primary winding
conductor and the insulation around it.
When reduction of errors is the only important criterion, the core material
should have the highest possible permeability.
The number of primary and secondary winding turns A considerable reduc-

tion of errors can be obtained by providing transformers with multi-tum, rather
than bar or single-tum primary windings. For a given transformation ratio, an
increase in primary winding turns from 1 to n implies a corresponding increase
in secondary winding turns (to nNs). For a given secondary-circuit impedance,
the secondary e.m.f. for a particular condition is not changed, but the core flux
(<p) is reduced because of the extra turns. Further, the exciting current is
reduced to ie /n2 As a result, the errors are reduced to lIn2 times the original
value, so, for example, the use of a two-tum primary winding would reduce
the errors to one quarter of those which would have been obtained with a
transformer having a single-tum primary winding.
Admittedly, the secondary winding impedance of a transformer of a given
size must rise if the numbers of turns in its windings are increased, but unless
the external burden is negligible and the secondary winding impedance is
increased in proportion to the square of the number of turns, then an improve-
ment in accuracy must be obtained.
Secondary windings Whatever burden is to be connected to a transformer, it

is desirable that the secondary winding impedance should be as low as
possible, as this reduces the required driving e.m.f. and core flux to the
minimum attainable values.
To achieve a low-leakage reactance (roL s), the secondary winding should be
toroidal and uniformly wound around the whole of the circumference of the
core. Conductors of the largest possible cross-sectional area should be used
for this winding to keep its resistance low. It is for this latter reason that
secondary winding conductors of current-carrying capacities much in excess
of those needed for transformer current ratings are usually employed, it being
not uncommon to use even 0.055 cm2 conductor in secondary windings rated
at 1 ampere.
Secondary current ratings Current transformers used in power systems are

usually housed in switchgear or other power equipment, which may, because
of the site layout, be a considerable distance from the control and relay rooms
in which the protective and measuring equipment fed by the transformers is
mounted. The interconnecting cables, which typically have conductors of
0.036 cm 2 cross-sectional area, may thus be of appreciable length, typically
100m. Such cables would introduce a resistance of about 0.5 Q into the
secondary circuit of each current transformer. In the case of a transformer with
a secondary winding rated at 5 A, an e.m.f. of 2.5 V would be required to drive
the current through the cables, and therefore an added burden of 12.5 VA
would be introduced. If, however, a transformer with a rated secondary current
of 1 A were to be used, an e.mJ. of only 0.5 V, representing a burden of
0.5 VA, would be added, i.e. the lead burden could be reduced to 4% of the
level of that with a 5 A rated transformer.
A disadvantage of utilizing transformers with low secondary current ratings
is that they are more costly to produce because they have to be wound with
more turns of smaller conductor. It is therefore advantageous to use a higher
current rating (5 A) in applications where the burdens are near the trans-
formers but low ratings (1 A or 0.5 A) for other applications.
2.2.3 Behaviour under abnormal conditions

Current transformers feeding measuring equipment are required to perform
with high accuracy under normal operating conditions, that is up to a level
somewhat above the rated current of their circuits, to allow for possible
sustained overloads. Their performances during the brief periods when very
high currents are flowing because of system faults are not usually of import-
ance.
A transformer associated with protective and control equipment must how-
ever perform acceptably, often in conjunction with others, over the whole
range of currents which may flow in it. These currents are dependent on the
network in which the protective equipment is used and on factors such as the
fault conditions encountered and the instants at which they occur. The mag-
nitudes of these currents may therefore vary greatly and their waveforms will
not always be sinusoidal but may contain harmonics and transient components
when faults occur.
The behaviour of current transformers carrying two common types of tran-
sient waves which arise in power networks are considered below.
Networks with negligible shunt admittance

When the shunt admittances of power lines and other equipment are neglected
most power networks may be reduced to a number of sources of sinusoidal
e.m.f. interconnected via impedances consisting of series resistance and in-
ductance, as shown in Fig. 2.8. Whenever a circuit-breaker in such a network
is closed, the phase currents which will subsequently flow through it will be
of the form:
ip =Ippk {sin (cot + a) - sin (a) . exp (- t/'t) (2.4)
in which a is dependent on the instant of switch closure and t is the ratio of

the inductance to the resistance in the circuit, i.e. t =Lp /Rp.
Normally the currents which flow are relatively small, but very high currents
may flow if a circuit-breaker is closed on to a faulted piece of equipment. The
worst possible condition, illustrated in Fig. 2.9, would be that of closing a
circuit-breaker on to a close-up fault when the busbars were connected to a
Circuit breaker
which is to be
closed
Note: the circuit is closed in single-line form for simplicity
Fig. 2.8 Single-line diagram of a power network.
large (low impedance) source. In this event the value of Ippk in equation (2.4)
would be many times the full-load rating of the circuit and the value of the
time constant of the transient term ('t) could be of the order of 0.1 s or more.
Clearly similar currents would flow in circuits after the incidence of a short
circuit. The behaviour of a current transformer carrying such current waves is
now considered.
Effective source
impedance Busbar
Circuit breaker
which is to be
XT-
/closed
Fig. 2.9 Closure of a circuit breaker when a fault is present.

Current transformer with a purely-resistive burden To simplify the situation
it will be assumed that the transformer has a uniformly-distributed toroidal
secondary winding of negligible leakage inductance and that its core is so
highly permeable that the exciting current is insignificant.
In these circumstances, the transformation would be ideal, i.e.
. Np
Is = - Ns Ip
and the secondary e.m.f. needed would be given by:
in which Rs is the resistance of the secondary circuit.

The core-flux variation would be:
Thus for the current given by equation (2.4), would be:
IppkNp Rs
<1>= (-cos(cot+a)+cotsinaexp(-tlt)}+k (2.5)
coNs2
in which k is a constant of integration.
A limiting case occurs when a = rr./2. For this situation:
IppkNp Rs
 = 2 (sin cot + cot exp (- tit) + k }
coNs
If the core flux is assumed to have been zero at the incidence of the fault,
a reasonable assumption even if load current had been flowing before the fault,
because the associated flux variations would have been small, then the core
flux would be:
IppkNpRs
<1>= 2 (sincot-cot(l-exp-tlt)} (2.6)
coNs
Equations (2.5) and (2.6) show that the core flux has three components, the
sinusoidal one being the steady state variation. The resultant of the constant
and exponentially-decaying components rises and in the limiting case reaches
a value equal to COt times the peak steady state excursion. This is illustrated
in Fig. 2.10 using a time constant (t) of 0.1 s.
The variations of the ratios of the maximum flux level reached, relative to
the steady state peak values with the point on wave of switching or fault
incidence (a), for various values of COt, are shown in Fig. 2.11. It will be seen
that the maximum flux level may lie between twice and (1 + COt) times the
peak steady state value.
I
' ...../ '_/~ \ ..../ ' ... 1
~ Steady-state component
Transient component
-6
-8 Core flux
Note: the above curve is for a circuit in which royRp = 8
Fig. 2.10 Core-flux variation.
This result applies, as stated earlier, for cores of infinite permeability. In

practice the differences produced by taking account of the finite permeabilities
normally encountered are extremely small and may be neglected.
It will be clear that the flux levels reached during normal conditions have
to be very low, when resistive burdens are connected, if saturation of the core
is to be avoided when large transient currents are flowing. For example, if the
incidence of a fault causes the steady state component of the current to be 20
times the rated value, and the on value is also 20, then the peak-flux density
under normal conditions should not exceed 1/(20 x 21), i.e. 1/420 of the satu-
ration flux density of the core, only about 0.003 T, if saturation is not to occur.
Current transformer with an inductive burden To simplify this treatment it

will again be assumed that the transformer has a core of infinite permeability
and also that the impedance in its secondary circuit is purely inductive.
In these circumstances the transformation would be ideal at every instant, i.e.
. Np .
Is =- Ns Ip
and the secondary e.mJ. needed to drive this current through the secondary
circuit inductance (Ls) would be:
N
es =Ls dis/dt =- Ls ~, dip/dt
Ns
f
To induce this e.mJ. the core flux would be:
1
1>=-- LsNp
es dt=--lp+k
Ns N s2
Again, as in the previous section, the flux levels during normal conditions
will be very low, in which event the constant of integration may be taken to
be zero and the required flux variation would be proportional to the primary
current ip, i.e.
LsNp
I>=--i (2.7)
2Ns p
As a result, much smaller flux variations are needed with inductive burdens
than with a resistive burden of the same VA, i.e. of the same impedance, and
10
E
II)
>.
-g
Q)
1ii
i.
~E
e
5
Instant of fault occurrence (a)
Fig. 2.11 Variation of maximum flux with time of incidence of faults and circuit
roL/R ratio.
the maximum flux density will not normally exceed double that of the peak
value of its steady state component.
When secondary circuits contain both inductance and resistance the flux
variations may be determined by summing the fluxes given in equations (2.5)
and (2.7) in the correct proportions. The ratio of maximum flux density to the
peak value of its steady state component will never exceed that which would
be needed for a purely-resistive burden of the same VA.
Current transformer with a capacitive burden As in the previous sections, it

will again be assumed that the transformer has a core of infinite permeability.
In addition it will be taken for simplicity that the secondary circuit is purely
capacitive.
The transformation would again be ideal at every instant, i.e.
. Np .
Is = - Ns Ip
and the secondary e.m.f. needed to drive the secondary current is through the
capacitance (C s) would be:
es =-.L
Cs
f is dt =- ~
Ns C s
f ip dt
This e.m.f. would require a core flux variation given by:
1
cP=--
Ns
f Np
esdt=--
N s2 Cs
f (ipdt)dt (2.8)
In the first integration of ip a constant of integration (k ,), dependent on

earlier conditions, could be introduced, thereby causing the presence of the
terms k1 t + k2 in the second integration. The first of these terms (k1 t) represents
a constant rate of flux change in one direction and if it persisted then the onset
of saturation would be unavoidable.
Careful consideration is therefore necessary if capacitors are to be con-
nected in secondary circuits.
Networks containing components with distributed parameters

As shown on page 49, the currents which flow in power circuits which contain
only lumped, linear impedances usually contain a transient component which
decays exponentially in addition to a sinusoidal component. Should both series
capacitance and inductance be present then a transient component with an
exponentially-decaying sinusoidal waveform can be present.
When a power system contains transmission lines or cables, however, travel-
ling waves are set up whenever there is a disturbance, and these cause the current
waveforms which the current transformers are called upon to carry, to be very
different to those considered in the previous section. Because the subject of
power-line transients is much too vast to be dealt with in this book, only a
brief indication will be given below of some of the currents which may flow.
Line energized from a source providing a direct voltage It is well known that
the input current to a healthy, single-phase, loss-free line when connected to
an infinite source by the closing of a circuit-breaker would be as shown in
Fig. 2.12(a) if the remote end of the line was open-circuited. Should the same
,-----
0 1 2 3 4 5 Time
2 x Ii ne length )
(units =
vela city of light
'---
Note: switching occurs at time t =0
(a)
12
~
:s
.'!l 8
'2
2-
'E
~
:50 4
'S
a.
E
0 L--L-r~----~------~----~----~~----~---:~
Time
2 x line length
. = ______
(Units --':0-_
)
velocity of light
Z source =Zo 12
Note: short circuit occurs at time t =0
(b)
Fig. 2.12 Input currents to an ideal line energized from a source of direct voltage.
!!l
.c;
2-
E Time
~ -1 (ms)
:;
o
"S -2
a.
c
Source voltage = Vpk cos rot ; Zsouree =0
(a)
20 25 Time
(ms)
Fault at instant when voltage at

position of fault is maximum (VFpk)
(b)
2
.~
c
2-
E
~ O~L-------------~--~-------.
5 10 15 20 25
~ Time
"S -1 (ms)
a.
f:
Fault at instant when voltage

at position of fault is zero
(c)
Fig. 2.13 Input currents to an ideal line energized from a source of alternating voltage.
line be suddenly short-circuited at its remote end, however, the input current
would be as shown in Fig. 2.12(b).
Line energized from a source providing an alternating voltage The vari-

ations of the input currents to the above line which would occur under three
different conditions, when connected to an infinite source of alternating volt-
age, are shown in Fig. 2.13. Two important features of these current waves
are that step changes and a unidirectional component may be present.
In practice, the presence of resistance or losses in a line causes the step
changes to attenuate and further effects are produced by the presence of
impedance in the source. These effects are, however, quite small in the case
of modem extremely high voltage transmission lines which have relatively
low resistances and which tend to be connected to large sources.
More details of the current waveforms which may be encountered were
given by one of the authors in an earlier work [7]. The responses of current
transformers carrying such primary currents are now considered. To simplify
the treatment, the behaviour obtained when primary currents of single step and
alternating rectangular waveforms are examined, assuming a current trans-
former with an infinitely permeable core, for which no exciting current is
required.
Under the above conditions, ideal transformation would always be obtained,
i.e. the secondary current would be given by:
. Np .
Is = - Ns Ip
With a secondary circuit containing only resistance (Rs), the secondary

e.m.f. would then be:
and the core flux would be given by:
 =_.l
Ns
f es dt = N~ . Rs
Ns
f ip dt (2.9)
When carrying a step function current, the flux would increase linearly and
clearly the core would inevitably become saturated. To sustain the rectangular
primary current wave shown in Fig. 2.14(a), the flux wave would be triangular
as shown in Fig. 2.14(b) and this clearly indicates that the step-current changes
would be correctly reproduced in the secondary circuit.
In the more practical situations encountered in systems energized from
sources providing sinusoidal e.m.f.s, the current waves tend to contain sinu-
soidally-varying sections and steps. Effectively unidirectional components
may be present during fault conditions and these tend to decrease slowly in
Primary
cu rrent l/pPk
Time
- (a)
Fig. 2.14 Core-flux variation in an ideal transformer carrying a rectangular primary-

current wave.
circuits in which the resistances are low. In these circumstances, the core
fluxes of current transformers would be driven towards saturation. This is clear
from the limiting case, for a loss-free circuit, illustrated earlier in Fig. 2. 13(c).
In this instance, the current would contain a constant direct component requir-
ing a linearly-changing flux component which would ultimately cause satura-
tion to occur.
It was shown earlier (page 53) that the flux variation needed in the core of
an ideal current transformer with a purely-inductive secondary circuit (Ls)
would be:
LsNp
=--.ip+k
N s2
This is clearly a much less onerous condition than that encountered with
resistive secondary circuits because the flux will not tend in practice to contain
a linearly-changing component.
The above treatment has been presented to indicate the core-flux variations
which may be needed in current transformer cores to ensure accurate trans-
formation.
When determining the suitability of current transformers which are to oper-
ate with particular protective schemes, modem computing techniques should
be used to determine the currents which will actually flow during abnormal
conditions and then, using the appropriate secondary-circuit parameters, the
core flux and secondary current variations should be determined, allowance
being made for the core excitation characteristics [8,9].
2.2.4 The effects of core saturation on transformation behaviour
In many applications, as indicated above, impractically large core-flux vari-
ations might be needed under certain circumstances to approach ideal trans-
formation, i.e.
. Np
1 =--1
s Ns p
As a result, saturation of transformers must, on occasion, be accepted and the
associated protective equipment must be so designed that it will nevertheless
function correctly. To enable this to be done it is necessary to determine the
secondary current errors or distortions which will be introduced when core
saturation occurs.
Current transformer with a reslstlve secondary circuit To simplify the

presentation, the approximate excitation characteristic shown in Fig. 2.15 is
assumed. At flux values below the saturation level ( < <l>sat) the permeability
is infinite and the exciting current is zero. For flux variations in this zone,
(- <l>sat < < <l>sat), ideal transformation (is =- Np ip INs) would therefore be
achieved.
When a core is saturated, however, ( = <l>sal or - <l>sal), then the rate of
change of flux will be zero and the secondary e.m.f. (e s ) must then also be
zero. With zero e.m.f., no current will flow in the resistive secondary circuit
cIlsat
Core flux
m.m.f.
-4>sat
Fig. 2.15 Approximate excitation characteristic.

rot
(a)
rot
(rad)
(b)
Fig. 2.16 Behaviour when core saturation occurs in a transformer with a resistive
secondary circuit.
(is=0) and the primary m.m.f. will maintain saturation until the magnitude of
the flux needs to reduce again and this will not occur until the polarity of the
required e.m.f. reverses. When the secondary circuit is resistive this will be at
the next current zero. As an example, if saturation occurs during a positive
half cycle of the secondary current, the e.m.f. will not become negative until
after the next current zero. Considering sinusoidal primary current conditions,
the behaviour would be as shown in Fig. 2.16(a) from which it will be seen
that the tails of the half cycles of secondary current are chopped off, and the
greater the desired flux excursion, the greater the portions of wave lost in each
half cycle.
This pattern of behaviour is repeated whatever the primary current wave-
form including situations where transient components are present, an illustra-
tion being provided in Fig. 2.l6(b).
Clearly the effect of saturation is to distort the waveform of the secondary
current and also to reduce its r.m.s. value.
Current transformer with a purely inductive secondary circuit Under ideal

conditions, the core flux needed when a secondary circuit is inductive is given by:
LsNp .
W=--Ip+k
Ns2
If symmetrical conditions obtain, then the constant of integration term (k)
will be zero and the flux will vary in phase with the primary current.
Saturation, if it is to occur, must therefore take place in the first half of each
half cycle. Thereafter, the secondary e.m.f. (e s ) will be zero which will prevent
the secondary current from changing, i.e. Ls dis/dt =O. This condition will then
be maintained because the primary m.m.f. will exceed that provided by the
secondary winding. Saturation will end when the primary and secondary
m.m.f.s become equal again. Thereafter the magnitudes of the core flux and
secondary current will reduce. This behaviour is shown in Fig. 2.17 and the
general effect is always to cut off the tops of the secondary current waves and,
as with resistive secondary circuits, the r.m.s. values are lower than the ideal
values, i.e. Is < Np Ip/Ns.
Current transformer with a secondary circuit containing resistance and

inductance in series As in the above two situations, ideal behaviour would
be obtained whenever the magnitude of the core flux is below the saturation
level. Once saturation occurs, the flux remains constant and no secondary
e.m.f. is produced. In these circumstances:
es =Ls dis/dt + Rs is =0
The secondary current during a saturated period is therefore given by:
is =A exp(-Rst/Ls)
rol
(rad)
(] f1
[) [)
<llsat = 3/pPk NpLs; <llR = <llsal2

4Ns2
Fig. 2.17 Behaviour when core saturation occurs in a transformer with an inductive
secondary circuit.
in which A is the value of the secondary current at the instant at which

saturati(;>D occurs. Again saturation is maintained because the magnitude of the
primary m.m.f. exceeds that of the secondary m.m.f. Saturation will end once
again when the secondary current reaches its ideal value. An example of a
secondary current waveshape which would be obtained is shown in Fig. 2.18.
The general effects which result from saturation of the cores of current
transformers may be summarized as reductions of the r.m.s. values of their
secondary currents below the ideal values and distortion of the current wave-
forms, the actual form of the distortion depending on the particular secondary
circuit parameters. The importance of these effects in protective applications
and methods of allowing for them will be considered in later chapters.
2.2.5 Remanent core flux

In the previous sections idealized excitation characteristics have been assumed
to simplify the presentation. With practical transformers, however, the char-
acteristics are in the form of symmetrical loops, one of which is shown in Fig.
2.19, and it is therefore possible to have flux in a core when there is zero
exciting current. During a particular normal steady state condition requiring a
sinusoidal flux variation, operation could be around a minor hysteresis loop
displaced along the CI> axis of the CI> - m.m.f. diagram, as shown dotted in Fig.
2.19, because this would not require the exciting current to contain a unidi-
rectional component. This situation could arise due to remanent flux left in
the core at an earlier time.
The amounts of flux which may be retained and their persistence are of
interest to protective-equipment engineers as the presence of remanent flux in
the core of a transformer reduces the available flux swing in one direction,
and makes the avoidance of saturation during fault conditions more difficult.
There are a number of conditions which may leave remanent fluxes in
current transformer cores. The largest levels tend to be left after the clearance
of faults and two possible situations are examined below.
cot
(rad)
Note: this figure is based on the following parameters:

coLJRp= 8; co4/Rs = 1; IIlsat = (/ppkNr7s)/(2co~); IIlR = 0; a-Yl =0
Fig. 2.18 Behaviour when core saturation occurs in a transfonner with a resistive and
inductive secondary circuit.
Fig. 2.19 Core-excitation characteristics.
The clearance of a symmetrical fault current

The passage of a symmetrical sinusoidal fault current through a primary
winding may cause the core flux to vary around a symmetrically-positioned
major hysteresis loop such as that shown in Fig. 2.19. Since circuit-breakers
normally interrupt a fault at a current zero, the operation of the core iron at
the instant of interruption would be at one of two fixed points on the hysteresis
loop, the points depending on the impedance present in the secondary circuit.
With a purely-resistive circuit the flux would have its maximum value at the
instant of clearance, i.e. A or B in Fig. 2.19, whereas with a purely-reactive
circuit, it would be zero, i.e. at P or Q.
For the working point to be at A, say, a direct magnetising m.m.f. would be
required and this could only be provided by a direct current in the secondary
winding after the primary circuit had been opened. To drive this secondary
current, the core flux would have to decrease to induce the necessary e.m.f.
The operating point would therefore move along the hysteresis loop until
equilibrium was established at point C where no secondary current or e.m.f.
would be required.
The maximum possible remanent flux COC) will be left when interruption
occurs at points along the sections AC or BD of the hysteresis loop, conditions
which will occur with inductive secondary circuits of sufficiently high power
factor.
Protective current transformers, because of their need to operate satisfactorily
during faults, usually operate at very small flux levels and the steady state
swings which occur even when large fault currents are flowing may be so small that
only rather insignificant levels of remanent flux may ,be left after clearance.
The clearance of asymmetrical fault currents

The flux variations needed in transformer cores when their primary windings
carry currents containing exponentially-decaying components were examined
on page 51. It was shown that large flux swings are necessary when secondary
circuits are highly resistive. A typical flux excursion for a transformer in a
primary circuit with an roLp/Rp ratio of 8 and a current with a sinusoidal
component starting at zero is shown in Fig. 2.20. This is based on a purely-
resistive secondary circuit and a ratio of exciting reactance (roLm) to secondary
resistance (Rs) of 8. It is the presence of the exciting reactance which is
responsible for the unidirectional flux component falling after rot = 31t to
eventually reach a constant value.
If clearance of this fault current occurred within a few cycles, the core flux
would be at a high value at the instant of interruption and thereafter the flux
would decay, as explained in the preceding section, until the working point
was on the vertical axis of the cl> - m.m.f. diagram, a considerable remanent
flux being left in the core.
The persistence of remanent flux

Because of the non-linear nature of the excitation characteristics of magnetic
materials, it is not possible to determine, simply by quantitive analytical
means, the effects of remanent flux on subsequent behaviour or on the variations
rot
(rad)
-8
Fig. 2.20 Flux excursion when a fault is present.

which will occur in the remanent flux. In practice, numerical step-by-step or
other methods are used to obtain accurate predictions of behaviour.
To illustrate the behaviour which may be obtained, however, a qualitative
presentation of the effects of the passage of sinusoidal primary currents on
remanent flux is provided below and some experimentally-obtained results are
included to illustrate the changes which occur.
When steady state primary current flows again in a transformer after the
clearance of a fault, sinusoidal flux variations are needed to provide the
secondary winding e.m.f. These would start from the remanent value (OR in
Fig. 2.21) and could progress around a minor loop such as RS. Operation
around such a loop could not, however, continue because its offset nature
would require the exciting current to have a direct component. Operation
would therefore continue around minor loops moving progressively to the left
and downwards until the exciting current had no direct component. Equili-
brium would be established at this position, shown as loop TV in Fig. 2.21.
Experimentally-obtained loops of this type for four different current levels are
shown in Fig. 2.22. If the primary current ceased after reaching an equilibrium
state, the remanent flux would be somewhat lower than the earlier value, the
reduction being dependent on the magnitude of the primary current which has
flowed and thus the size of the minor hysteresis loop. For normal current levels
the loops in protective current transformers tend to be small and the resulting
reductions of remanent flux are usually small. It must therefore be accepted
that a large part of the remanent flux left by a fault will tend to remain until
another large fault occurs. The possible effects resulting from this situation
are shown in Fig. 2.23 which includes experimentally-obtained records of flux
density against both time and m.m.f. for two fault conditions. Those shown in
Fig. 2.23(a), which are for a fault requiring a flux swing in the direction of
-----;~'
----""'pS /
/
/
I
I
I
/T I
/ I
I
I
I
I I
I : Major
I I hysteresis
I
I I loop
I
o
Fig. 2.21 Change in remanent flux in a core.

Fig. 2.22 Minor hysteresis loops.
, I'm:t ..
l (SA ) _auuuuuuUUlUUlUlUUUUUUUUUuUlUUUUUUUUlUUlUU
nn"'mnnm..nmn".nrtn"nrtn....nnn".......mm
CD
0.01-------------------
-0.6
Ip (SA r. m.s.) """""u,n""n,n,n","",,,,,n,,n,,n",,"",,n,n,",n,",n

IUUUUUlUUUUUUUUUUUUUUUUUlUUUUUUUUUUUUUUU
06~
o.o~
-0.6 ,WN",,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,ft"""""'''''''''''''''
UUUIUUUUUUuuuuuuuauuuuuuuUUUUUUUlUl
Fig. 2.23 Flux variation when a fault occurs.

Core flux density __ ~L - - 1

I
I
I
I
I
I
I
I
I
I m.m.f.
I
I
I
I
I
I
I
I
Fig. 2.24 Core-excitation characteristics.
the remanent flux, cause the core to saturate. Those shown in Fig. 2.23(b) are
for a fault of the opposite polarity which cause the flux to swing in the opposite
direction to the initial remanent flux.
The deviations of the secondary currents of current transformers from their
ideal values (is =- Np/Ns ip) as a result of saturation of their cores would cause
some protective equipments fed by them to maloperate. This is clearly unac-
ceptable and therefore, in these applications, the flux range between the maxi-
mum possible remanent flux value (OR) in Fig. 2.21 and the saturation value
must be great enough to allow the flux swings needed when large asymmetric
fault currents are flowing. Clearly core materials with high saturation flux
densities and excitation loops with relatively low flux densities at exciting
current zeros, as shown in Fig. 2.24 (curve a) are desirable rather than almost
rectangular loops of the form shown in Fig. 2.24 (curve b). The ratio of
maximum-possible remanent-flux density to the saturation density is referred
to as remanence factor.
2.2.6 Operation with a secondary circuit which is open or of a high

impedance
Current transformers are unlikely to be operated with their secondary windings
on open circuit, because of the widespread knowledge of the high voltages
which may be produced under such conditions. There are, however, cases of
near open-circuiting which occur at times when certain protective schemes are
used. Such situations are possible, for example, when circulating-current
schemes incorporating high-impedance relays are used. An examination of
the way in which high voltages are produced is therefore of interest.
As in earlier sections, a transformer with a particular excitation charac-
teristic will be assumed, to simplify the treatment. In this instance, the char-
acteristic shown in Fig. 2.25(a) will be used. Over the portion Be, the current
I -_ _ _ _D
m.m.f.
(b)
(c)
Fig. 2.25 Behaviour when a secondary winding is an open circuit.

is proportional to the flux and therefore the associated behaviour may be
represented by a constant magnetizing inductance (Lm) in the equivalent cir-
cuit of Fig. 2.25(b). Over the portions AB and CD, no flux change occurs as
the current changes and this behaviour may be simulated by zero magnetizing
inductance. A further assumption can be made that a constant resistance (R 1)
may be included in the equivalent circuit to take account of the hysteresis and
eddy-current effects in the core. This condition is similar to that considered
on page 59 when the effects of core saturation were being examined. It was
shown that saturation ends at primary current zeros and that secondary current
then flows until saturation occurs again. In the case of an open-circuited
secondary winding, the current through the high magnetizing inductance (Lm)
during unsaturated periods is very small and thus the referred primary current
(- Np ip/Ns ) flows through the resistor Re requiring an e.m.f. (e s ) given by:
Np .
es =- Ns Re lp
This variation is shown in Fig. 2.25(c) assuming a sinusoidal primary cur-

rent. In practice saturation occurs after a very short interval of time (ts) and
over this period the primary current may be expressed as:
ip =Ippk rot
Over the period ts the flux changes by 2<1>sat and therefore:
1
2<1>sat =- -
ft s
es dt
Ns 0
From the above three equations, it can be shown that the peak secondary
e.m.f. (Espk ) which occurs at the time ts is given by:
1
Espk = 2 [sat roNp IppkRe P: (2.10)
An alternative treatment which may be used when the hysteresis and eddy
current losses in a core are very low is to assume that the resistor Re has an
infinite value. In these circumstances, the primary referred current must al-
ways flow through the magnetizing inductance (Lm). Whenever the current is
above a certain value, the core will be saturated and as a result the incremental
inductance Lm will then be zero. Near current zeros, however, the magnetizing
inductance, as before, may be taken to have a constant non-zero value. At
these times the secondary e.m.f. is given by:
For a primary current of ip = Ippk sin rot, the secondary e.m.f. would have a
maximum value at the current zeros of:
Stalloy laminations
0.015m thick
Stalloy laminations
0.03m thick
20 50 100 200 500 1000

Ampere turns per
metre of core length
Note: these curves are based on a supply frequency of 50Hz
Fig. 2.26 Voltages during open-circuit conditions.
(2.11)
In practice, the behaviour will lie between the two limiting conditions
described above. More accurate modelling can be performed by using non-
linear components with values related to the core flux and its rate of change.
The normal reason for determining the peak voltages which may be pro-
duced is to ensure that the insulation of a transformer and its burden, including
the connecting cables, is adequate or suitably protected by voltage-limiting
devices. As it is usual to allow a significant factor of safety on insulation
levels, highly-accurate determination of the voltages which may be produced
by current transformers is not justified. It is probably better to make use of
experimentally-derived results such as those shown in Fig. 2.26 which were
obtained by Wright [7].
2.2.7 The construction of current transformers

Each current transformer, as stated earlier, has a core of magnetic material
which is linked with the primary and secondary windings. These three com-
ponents are considered separately below.
A B
A, cross-section; B, full view

(a)
A B
A, cross-section; B, full view

(b)
Fig. 2.27 Current transformer cores.

Magnetic cores
Transformation errors are caused by the need to excite the core of a trans-
former and thereby induce the necessary e.m.f. in its secondary winding.
Clearly, therefore, it is desirable that secondary-circuit impedances should be
kept as low as possible and for this reason the secondary winding leakage
inductances should be reduced to the minimum attainable levels. This may be
done by using ring-shaped cores around which toroidal secondary windings of
one or more layers are uniformly distributed. Cores usually consist of a cy!
lindrical stack of laminations as shown in Fig. 2.27(a) or are made up of strip
wound in spiral form, like a clockspring, as shown in Fig. 2.27(b). The latter
method is much to be preferred when grain-orientated magnetic materials are
being used as this ensures that the flux is able to follow the path of minimum
reluctance.
The core material should ideally be of very high permeability, have a high
saturation flux density and, as shown in the previous section, not be capable
of retaining large levels of remanent flux. Although nickel-iron alloys have
very high permeabilities, they do not have the other desirable properties and
therefore protective current transformer cores are usually made from either
hot-rolled or cold-rolled silicon-iron alloys.
Because it is relatively expensive to produce toroidal secondary windings
on ring cores, cores made up oi' rectangular strips or C-shaped sections are
sometimes employed when a high standard of performance is not required.
This reduces manufacturing costs because the cores can be assembled into
secondary windings which have been produced earlier, in coil form on con-
ventional coil-winding machines. Such an arrangement is shown in Fig. 2.28.
..
:- --~....""""'-::
A, primary winding; B, secondary windings (connected in series); C, built-up core
Fig. 2.28 Current transformer with a built-up core.

Secondary windings
As explained above, secondary windings are usually toroidal and uniformly
spaced around the circumferences of the cores. In applications where the
transformer ratio is low, e.g. 50/5 A, the secondary winding turns will be
relatively small in number and they are then spaced out to occupy a single
layer. When the ratios are high, the numbers of secondary turns will be great
enough to require windings of two or more layers but again the turns will be
spaced to fully occupy each of the layers.
As stated earlier in section 2.2.2, it is usual for secondary windings to be
made with a conductor of a cross-sectional area considerably greater than that
needed to carry the rated secondary current continuously. This practice is
adopted to keep secondary winding resistances low and thus reduce the e.m.f.s
which must be induced.
The windings should be insulated to withstand the voltages which may
appear across them under all normal and abnormal conditions.
Primary windings and positioning
Transformers used in circuits with high rated currents, say 500 A and above,
usually have a single-tum primary winding. For lower current levels, however,
multi-tum windings are commonly used to enable greater transformation ac-
curacies to be obtained. This improved performance, as explained earlier, is
obtained because, for a given ratio, the number of secondary winding turns is
proportional to the primary winding turns, and the greater the number of
secondary turns, the smaller is the flux variation and thus the exciting current
needed to provide a given secondary e.m.f.
Many modem protective schemes monitor the zones between the sets of
current transformers associated with them, and as any fault within a protected
zone is cleared by opening circuit-breakers to isolate the unhealthy section of
the system, discrimination approaching the ideal can only be achieved when
the current transformers are near the contacts of the circuit-breakers. This
topic was treated in detail in reference [8].
When very high voltage switchgear incorporates bushings through which
the main conductors enter and leave, the above desired proximity is achieved
by mounting the secondary wound current transformer cores over these bush-
ings. In addition to the transformers which are needed to feed the main
protective equipment, others are needed to feed back-up protective relays and
indicating instruments and possibly accurate measuring equipment. In prac-
tice, therefore, several wound cores may be stacked together on each phase
bushing, a typical arrangement being shown in Fig. 2.29.
This arrangement is not only ideal operationally, but it is very economic as
it reduces the total amount of costly high voltage dielectric material to a
minimum. It does however, limit the primary winding to only one tum.
A, capacitor bushing; B, tank top; C, current transformer cores and windings;

D, primary conductor; E, clamps and supports
Fig. 2.29 Current transformer mounted around a bushing.
Some forms of high voltage switchgear, such as small-oil-volume and air-

blast designs, do not have suitable bushings over which transformers may be
mounted. Separate current transformers must then be provided adjacent to the
switchgear, unless the latter is associated with and near to power equipment
with bushings on which the current-transformer wound cores may be mounted.
Separate high voltage current transformers are constructed in the post form
shown in Fig. 2.30. Oil-impregnated paper is used to insulate the primary
conductor, which is a U-shaped tube. Metallic foils are inserted in the insula-
tion at a number of levels to control the electric stress distribution and means
must be provided to prevent moisture from entering the housing. The external
diameter of the primary tube must be large enough to keep the electrical stress
in the dielectric to an acceptable value and this usually causes the internal
diameter of the tube to be great enough to permit a lightly insulated cable to
be passed through it. By making suitable connections a two-tum primary
winding can thereby be provided if desired. The secondary wound cores are
fitted around the lower part of the primary tube as shown.
Gas-insulated current transformers have been produced for use in high
voltage systems. In one arrangement the cores and secondary windings are
housed in a tank which is at line potential. It is supported on a porcelain
insulator inside which the secondary circuit connections pass. A 'straight-
through' primary conductor passes centrally through the cores and secondary
windings, from which it is insulated by a gas, usually sulphur hexa-fluoride,
under pressure.
~T~--c
~ ___ B
B---~r-'ffirt.-J~I--- D
~I--__ E
I-I.-k----- F
P==9-- G
H-+---H
A, nitrogen; B, Oil; C, Air; D, terminal chamber; E, porcelain insulator;

F, insulated primary conductor; G, arcing ring; H, cores and secondary winding; J, tank
Fig. 2.30 Posttype current transformer.
In lower voltage metalclad and metal-enclosed iwitchgear, one of the most

satisfactory arrangements is to stack the secondary wound cores and use a
common primary winding with the necessary number of turns. The completed
assembly is then housed in a special transformer chamber, which forms part
of the switchgear. A typical arrangement is shown in Fig. 2.31. In some cases
however, when the necessary performance can be obtained, transformers with
single-tum primary windings (usually called bar primaries) are used. These
transformers are produced in two forms, in one the core and secondaty wind-
ing are fully insulated and in the other the main insulation is on the primary
conductor. For these lower voltage applications, the insulation may be var-
nished tape and cloth or synthetic resin.
In all designs, primary windings must be sufficiently robust and so sup-
ported that they will not be mechanically damaged by the forces which may
Linear couplers 77
"" D
"" E
A, bushbars; B, current transformers; C, operating mechanism;

D, circuit breaker (racked out); E, main frame
Fig. 2.31 Current transformer housed in switchgear.
act on them during short-circuit conditions. In addition they must comply with
regulations concerned with their current-carrying capability. As an example,
BS 3938 requires that a current transformer with an integral primary winding
must be capable of carrying its rated short time current, for its rated time, with
50% of the rated burden connected to the secondary terminals, without suffer-
ing any damage such as would adversely affect it either electrically or mech-
anically.
The rated short-time current of a transformer is the rupturing capacity of the
circuit-breaker with which it is associated and the time rating is the maximum
period for which the transformer may be subjected to the full short-circuit
current, and this depends on the protective equipment fitted to the circuit.
2.3 LINEAR COUPLERS

Although iron-cored current transformers are and have been used in conjunc-
tion with power-system protective equipment for almost a hundred years, they
are not ideal because of the non-linearity of their excitation characteristics and
their ability to retain large flux levels in their cores.
Protective schemes, in which the outputs of non-ideal current transformers
at different points in a power system are compared, could operate correctly
provided that the current transformers each had the same percentage trans-
formation errors at each instant. This situation cannot be achieved in practice,
however, with protective schemes protecting multi-ended units, such as bus-
bars, because the output of a current transformer carrying a large current might
be required to balance with the sum of the outputs of a number of transformers
carrying small currents. The transformers would be operating at different
points on their excitation characteristics and when these are non-linear the
errors of the transformers would differ, preventing perfect balance being
achieved. Even in those cases where the outputs of only two current trans-
formers carrying the same primary currents are compared, unbalance would
occur if they were retaining different remanent fluxes because they would not
then be at the same points on their excitation characteristics and would not
have the same errors.
The above difficulties clearly would not arise if protective schemes were
energized from transformers with cores of unit relative permeability, i.e. of
wood or air, because such transformers are unable to retain flux in their cores
when their excitation is removed and they always behave in a proportional or
linear manner. These devices, which are usually called linear couplers, do not
produce secondary currents approximating to those which would be provided
by an ideal current transformer (is = - ip Np/Ns ) and consequently protective
equipment must be specially designed to operate from them. This may be done
by inserting interface equipment, such as amplifiers, so that an infinite imped-
ance is presented to the outputs of each of the couplers, causing them to
operate as mutual inductors, thus providing output e.m.f.s proportional to the
rates of change of their primary or input currents. Alternatively, the input
impedance of the protective equipment could be matched to that of the
couplers, to enable the maximum possible outputs to be obtained from them.
The procedure for designing linear couplers required to supply particular
outputs was dealt with fully by Mathews and Nellist [10] and therefore this
section will be confined to analysing the general behaviour of linear couplers
during steady state and transient conditions.

infinite impedance
In the limiting case where an infinite impedance is presented to the secondary
winding of a linear coupler, the device operates, as stated above, as a mutual
inductor and the output is in the form of an e.m.f. given by:
es = - Mps dip/dt
Linear couplers 79
in which Mps is the mutual inductance between the windings. Its value is given
by:
in which Np and Ns are the turns of the primary and secondary windings
respectively, Ae is the cross-sectional area of the core and Ie is its magnetic
path length.
During steady state conditions with a sinusoidal primary current ip =
Ippk sin (rot + ex) the output e.mJ. would clearly be:
es = - roMps Ippk cos (rot + ex)

finite impedance
Ideally it should be possible to connect any type of burden to a linear coupler,
and therefore the most general case in which the secondary circuit consists of
resistance, inductance and capacitance in series is examined below.
The equivalent circuit of a linear coupler is the same as that of a normal
current transformer, but because there are no core losses the resistance (Rt) is
infinite and therefore not included. The exciting inductance, designated
MpsNsINp, in the circuit shown in Fig. 2.32 is constant under all conditions.
The basic equation for a coupler is therefore:
-{isRs+Ls dis+-.l
dt e s
f isdt}=Mps'~(ip+
l dt
Ns is]
Np
(2.12)
For the condition of switching a primary circuit containing series inductance

(Lp) and resistance (Rp) on to a fault, or for a fault occurring on such a circuit
when unloaded, the primary current would be:
ip=lpPk{ COS(rot+ex)-cosexex p (- ~: tJ}

The corresponding secondary current would be:
rr
ro Mps Ippk
is = 1 sin (rot + ex - ~)
[Rs2 + {ro (Ls + Mps Z:)- (l/roe s)
+
Mps Ippk cos ex ( R
exp - --.X t
1
Ns Lp
Ls + Mps - - (Lp RsIRp) + Lp/Rp es
2 2
Np
+ ka exp (- mt) + ke exp (- nt) (2.13)
Fig. 2.32 Equivalent circuit of a linear coupler, referred to secondary level.
in which:
r
and m and n are given by:
m =[R,/2 (1, + M,. Z;lH [Ri/4 (L' + M,. Z;Jl-[ lIC, (1, +M" Z;II
n +12 (1, Z;ll-{[
+ M" Ri/4 (1, + M,. Z;Jl- [lIC. (1, + M" Z;II
Examination of equation (2.13) shows, as would be expected, that as the
t
mutual inductance approaches infinity, the secondary current approaches the
value normally regarded as the ideal (is = ip Np/Ns). The low permeability of
the core of a linear coupler prevents a high mutual inductance from being
achieved, however, and there is consequently always a considerable diver-
gence from the performance which is expected of a conventional current
transformer.
The steady state sinusoidal behaviour of a linear coupler can be determined
from the first term in equation (2.13). The ratio of the magnitude ofthis output
to that of the sinusoidal component of the primary current is given by:
. .
Transformation ratIO =-Is =. - - - - - -roMps
-------
Ip [ R; + { ro(Ls + MpsNs/Np) - ro~s }2]!
2
This expression shows that the output of a coupler falls significantly as

inductive reactance is added to its burden. It also falls, but at a reduced rate
Current transformers with air-gapped cores 81
as the burden resistance is increased. The output can, however, exceed the
ideal value (Is = - Np INs Ip) if the secondary circuit is made sufficiently capa-
citive. The phase error is (1t/2 - ~) (leading) and examination of the expression
for ~ given above shows that increase of burden resistance or capacitive
reactance causes the phase error to increase.
It can be seen that there are three transient terms in the expression for the
secondary current (equation 2.13). One of these is produced by the transient
component of the primary current whilst the others are produced by the
coupler and its burden.
The deviations between both the steady state and transient outputs from the
normal ideal value may be considerable and as shown they depend on the
coupler and burden parameters. When particular protective schemes, for
example differential equipments, are being supplied, satisfactory behaviour
should be obtainable, however, provided that the coupler and burden parame-
ters are matched. Clearly couplers should be as nearly identical as possible,
i.e. their mutual inductances should be the same, and the burdens must be
virtually equal.
2.4 CURRENT TRANSFORMERS WITH AIR-GAPPED CORES

Current transformers with air-gapped magnetic cores were introduced many
years ago to provide more linear behaviour than that obtainable from trans-
formers with homogeneous cores of highly permeable magnetic material.
Figure 2.33(a) shows the excitation characteristics of magnetic material and
air and also a combination of the two which should be obtained when using
an iron core with one or more air gaps as shown in Fig. 2.33(b).
The resultant characteristic is obviously more linear than that of the mag-
netic material and the remanent flux which may be left in the core is lower.
Clearly, however, complete linearity would not be obtained until the magnetic
path was completely in air or other material of unit relative permeability, i.e.
as in a linear coupler.
Air-gapped core designs occupy therefore an intermediate position between
conventional current transformers and linear couplers and a satisfactory com-
promise performance has to be achieved. The introduction of air gaps or
sections of low permeability in a core increases the excitation needed. While
this could be counteracted by using large cores with large cross-sectional
areas, this in turn increases the lengths of the secondary winding turns and
their resistances. To obtain performances comparable with those of conven-
tional transformers, air-gapped designs have to be large and relatively expens-
ive.
Such transformers have been produced and used to supply protective equip-
ment which has functioned satisfactorily. As with linear couplers, care must
be taken during manufacture to ensure that transformers which are to operate
together in groups are correctly matched, particular attention being taken to
A.!! '
."..."..~ ......
" ,.
~ .... :-: ~..... Combination
m.m.f.
(a)
(b)
Fig. 2.33 Air-gapped core and its excitation characteristic.
ensure that the gapped sections have the correct reluctances. The so-called
distributed air-gap transformer (abbreviated to DAG), which was developed
many years ago, has several gaps uniformly distributed around its core, this
being considered preferable to a core with only One relatively long gap.
2.5 NON-CONVENTIONAL CURRENT TRANSDUCERS

It is evident from the previous sections that the iron-cored current trans-
formers, used for feeding measuring and protective equipment in power sys-
tems, are not ideal. In addition, those which are used on very high voltage
systems in positions where they cannot be fitted over bushings in major items
Non-conventional current transducers 83
of equipment have the disadvantage of being costly, because of the insulation
needed on their primary windings. These factors have caused several investi-
gations to be initiated during the last ten to twenty years to produce alternative
current transducers. Workers have incorporated modem signal-processing
equipment and other electronic devices in protective schemes to eliminate the
need for them to be provided with inputs of several volt-amperes.
Work was done to eliminate the costly insulation needed on high voltage
primary windings by mounting lightly insulated conventional secondary
wound cores directly around the main system conductors. The secondary
outputs were processed to produce light signals conveyed to the relaying
equipment via optical fibres. The latter can convey signals over a wide fre-
quency band and being insulators they do not need bulky insulation around
them. They also have the advantage that they are not affected by electromag-
netic interference.
Blatt [11] described an arrangement in which a number of sensing coils were
placed at different points in a switchgear site. Simultaneous outputs from these
coils, monitored under microprocessor control, were analysed to provide data
on the individual circuit currents. An alternative device, called a dynamic
current transducer, developed by Mercure [12], used the current-carrying con-
ductor as the primary of a coaxial plastic-core Rogowski coil. The signal from
the secondary was transmitted to relaying equipment using an analog fibre-
optic link.
Current and voltage transducers of an entirely new type making use of
certain optical materials have been developed in recent years. When these
materials are placed in magnetic or electric fields the polarization of light
beams passing through them is affected by the strengths of the fields.
A current transducer of this type contains a piece of opto-magnetic material
which is mounted near the conductor to be monitored and the material is thus
in a magnetic field with a strength proportional to the conductor current. Light
from a light-emitting diode is linearly polarized and then transmitted through
the opto-magnetic material. On exiting from the material, the direction of
polarization is shifted by an angle proportional to the conductor current.
Measurement of this angle enables a signal proportional to the current to be
generated and this may then be transmitted, via an optical fibre or otherwise,
to supply continuous information about the current to measuring and/or pro-
tective equipment. More detailed information about such devices and their
performance is provided in references [13-16] at the end of this chapter.
The above devices, which are clearly more complex than conventional
current transformers, will not necessarily provide ideal performance at all
times. As an example, variable current errors could be introduced because the
behaviour of opto-magnetic materials is affected by ambient temperature vari-
ations. Nevertheless, these transducers appear to be suitable for use with
modem protective schemes.
To date they have not been installed as replacements for existing current
transformers because the latter are performing satisfactorily and the large
capital costs involved would be too great to be justified. It does seem probable,
however, that current transducers containing optical components will be used
in future installations where advanced protective schemes, requiring accurate
detailed information about the system currents, are to be used. Conventional
current transformers will nevertheless continue to be used for the majority of
protective applications and certainly this will be the case on systems operating
at the lower voltage levels.
2.6 SPECIFICATIONS AND TESTING

Current transformers used with measuring instruments or accurate metering
equipment must perform within given error bands during normal steady state
conditions. The permissible current and phase errors for given classes of
accuracy are specified in documents such as BS 3938. The classes of accuracy
quoted in BS 3938 are 0.1, 0.2, 0.5, 1,3,5, these being the respective percent-
age current errors permitted at rated current. The corresponding phase-error
limits are 5, 10, 30 and 60 minutes of angle for the first four classes, no
phase-error values being specified for classes 3 and 5. The durations of faults
and other abnormal conditions are usually so brief that their effects on perfor-
mance are not considered to be significant and consequently no accuracy
limits are quoted for very high current levels.
The situation is more complex for current transformers supplying protective
equipment because satisfactory transformation must be achieved during fault
conditions. For applications such as time-graded overcurrent protection a
measure of core saturation may be accepted in the current transformers during
transient conditions because the minimum operating times of these relays are
achieved at ten times rated current and errors at currents somewhat below this
level do not significantly affect the performance. For example, an error reduc-
ing the current by 10% at 10 times rated current increases the operating time
by only 7%. Current transformers of classes 5P and lOP, which are defined in
BS 3938, are therefore suitable in these situations. The P is included to denote
protective applications and the 5P and lOP classes permit the exciting current
to be 5% and 10% of the rated current respectively.
For other protective applications, such as high-speed differential schemes,
current transformers must function to acceptable standards up to the highest
current levels which may be encountered during fault conditions. BS 3938
includes a class X category for such transformers. It is stated in the specifica-
tion that the requirements are so numerous and dependent on the protective
gear involved that useful guidance cannot be provided. It is further stated that
the transformers should be so designed that balance is maintained within the
protective equipment supplied by them.
ESI Standard 35-15 [17] covers transformers for low voltage distribution
systems up to 33 kV and a supplement to IEC Publication 185 (1966) [18]
References 85
quotes additional requirements for current transformers associated with pro-

tection systems for which transient performance is significant.
Testing methods to be used by manufacturers, to ensure that their current
transformers comply with the permissible accuracy limits, are outlined in
specification documents such as BS 3938. However, for those transformers
which are to be used with complex protective schemes, such as those applied
to long high voltage transmission lines, conjunctive testing of the scheme and
prototype transformers is usually performed on the manufacturers' premises to
ensure that satisfactory performance will be obtained. Thereafter, testing must
be done on production transformers to ensure that their characteristics are
similar to those of the prototypes.
It is clear from the above that general guidance on the selection of current
transformers to be used with protective schemes cannot be provided, but more
specific guidance is provided, where possible, in the later chapters in which
particular types of protective schemes are considered. In addition, references
to works which may prove helpful are provided at the end of this chapter.
2.7 THE FUTURE
It is probable that conventional current transformers will, because of their

simplicity and reliability, continue to be used to supply many protective
equipments in the future and this will certainly be so for those applications
associated with relatively low voltage networks.
It does seem probable, however, that non-conventional current transducers,
such as those containing optical components, will be used in future installa-
tions where advanced protective schemes, requiring accurate and detailed
information about the system currents, are to be used. Extensive
development and field trials have taken place, sponsored by major
manufacturers, in the 380 kV German power system, by KEMA in the
Netherlands (132 kV), Electrabel in Belgium (170 kV), Hydro Quebec in
Canada (345 kV), EDF in France (420 kV), Georgia Power and BPA in the
USA (525 kV), and have given excellent results [19,20].
REFERENCES
1. Shadin, H. I (1963) The Making of the Electrical Age, Abelard-Schuman, New York.
2. Parson, R. H. (1940) Early Days of the Power Station Industry, Cambridge
University Press.
3. Andrews, L. (1898) The prevention of interruptions to electricity supply, J. Inst
Elec. Eng., 27, 487-523.
4. Clothier, H. W. (1902) The construction of high-tension central station switch-
gears, with a comparison of British and foreign methods, J. lEE, p. 1247.
5. Edgcumbe, K. and Punga, F. (1904) Direct reading measuring instruments for
switchboard use, lEE, 33, 620--66.
6. Young, A. P. (1910) The theory and design of current transformers, 1. Inst Elec.
Eng., 45, 670-678.
7. Wright, A. (1968) Current Transformers - their transient and steady state per-
formance, Chapman and Hall.
8. Gray, W. and Wright, A. (1953) Voltage transformers and current transformers
associated with switchgear, Proc. lEE, 100, 223.
9. Poljac, M. and Kolibas, N. (1988) Computation of current transformer transient
performance, Trans. IEEE, PD-3, 1816-1822.
10. Mathews, P. and Nellist, B. D. (1962) The design of air-cored toroids or linear
couplers, Proc. lEE, 109, (A), 229.
11. Blatt, D. W. E. (1989) Monitoring and fault protection of high voltage switchgear
by the REMMIT method, Fourth Int Conf on Developments in Power System
Protection, 11-13 April 1989, Edinburgh, UK, lEE Conf Publ., vol. 302, pp.
167-171.
12. Mercure, H. P. (1987) Development of a novel measuring device: the dynamic
current transducer, IEEE Trans. on Power Delivery, PWRD-2, pp. 1003-1007.
13. Erickson, D. C. (1980) The use of fibre optics for communications, measurement
and control within high voltage substations, lEE Trans. on Power Apparatus and
Systems, PAS-99, pp. 1057-1065.
14. Kanoi, M., Takahashi, G., Sato, T., Higaki, M., Mori, E. and Okumura, K. (1986)
Optical voltage and current measuring system for electric power systems, lEE
Trans. on Power Delivery, PWRD-l, pp. 91-97.
15. Nojima, K., Nishiwaki, S., Okubo, H. and Yanabu, S. (1987) Measurement of
surge current and voltage waveforms using optical-transmission techniques, Proc.
lEE 134, (C) 415-422.
16. Mitsui, T., Hosoe, K., Usami, H. and Miyamoto, S. (1987) Development of
fibre-optic voltage sensors and magnetic field sensors, IEEE Trans. on Power
Delivery, PWRD-2, pp. 87-93.
17. ESI Standard 35-15, April 1989, Protection and measurement transformers for
H.V. distribution systems up to 33 kV, 58 pages.
18. Supplement to IEC Publication 185 (1966), Current transformers for protection
systems for which transient performance is significant.
19. Schwartz, H. and Hudasch, M. (1994) Optical current transformers - successful
first field test in a 380 kV system, ABB Rev., 3, 12-18.
20. Chatrefou, D. (1997) Application of optical sensors in extra high voltage
substations, GEC Alsthom TandD Rev., 111997,17-24.
FURTHER READING
British Standard Specification for Current Transformers

BS 3938: 1973
IEC Publication
185 (1966), 'Current transformers '; IEEE Relaying Committee Report (1990)
Gapped core current transformer characteristics and performance IEEE Trans. on
Power Delivery, 5, (4), 1732-1740.
3
Voltage transformers
The term voltage transformer (commonly abbreviated to v.t.) is used when

referring to a low output, step-down transformer with two or more windings
per phase mounted over the limbs of a core of magnetic material.
Both single- and three-phase designs are produced. Clearly their primary
windings must be rated at the voltage levels of the power circuits to which
they are to be connected, whereas their secondary winding voltages are stand-
ardized in Britain at 63.5 V (phase) i.e. 110 V (line). In other countries, similar
but slightly different values have been adopted, for example 120 V (line) and
100 V (line) are standards in the USA and many other countries.
The burden of a voltage transformer may consist of several parallel-
connected items such as a voltmeter winding and the voltage windings of a
wattmeter and directional and distance-type protective schemes. The total
current to the burden does not normally exceed a few amperes, i.e. a burden
of a few hundred volt-amperes at the most.

It was stated in the previous chapter that the operating voltages of electrical
power systems rose rapidly during the last two decades of the nineteenth
century and that current and voltage transformers must have been introduced
to feed measuring instruments in that period.
The production of voltage transformers presumably presented few problems
because they would have been merely small versions of the power trans-
formers then being produced. Understandably, therefore, there are few direct
references to them in the literature produced around the tum of the century.
However, in 1898, in a paper by Andrews [1], also referred to in the previous
chapter, details were given of a discriminating cut-out which had been de-
veloped to detect reverse-current flow in machines. This device, which oper-
ated on the same principle as that used in later directional relays, needed both
current and voltage inputs and from Fig. 3.1 (Fig. 1 in Andrew's paper) it can
be seen that a voltage transformer was included in the circuit.
In the paper presented to the lEE in 1902 by Clothier [2], reference was
made to the German practice of feeding voltmeters from transformers. Later,

88 Voltage transformers
Fig. 3.1 Circuit incorporating a voltage transformer. (Reproduced from Andrews,

1898, Journal of the lEE, 27, with the permission of the lEE).
in 1904, a paper by Edgcumbe and Punga [3] contained a section dealing with
both current and voltage transformers. It is therefore clear that voltage trans-
formers were being used to supply measuring instruments and protective
devices before that date.
Designs basically similar to those produced originally have been used in
very large numbers up to the present time on circuits operating at voltages up
to about 66 kV (line). For higher voltage applications, transformers incorpor-
ating capacitor dividers, known as capacitor voltage transformers, are used.
The name capacitor voltage transformer was suggested by Wellings et al. [4]
in a paper presented to the lEE in 1936.
Information about the behaviour, construction and use of both electromag-
netic and capacitor voltage transformers is provided in the following sections
of this chapter.
3.2 ELECTROMAGNETIC VOLTAGE TRANSFORMERS

Three-phase voltage transformers usually have both their primary and second-
ary windings connected in star and therefore their phases are effectively
independent and their step-down ratios would ideally be the same as the ratios
of the turns in their primary and secondary windings, i.e. vs/vp = Ns/Np. The
Electromagnetic voltage transformers 89
ipNp
-Ns
4
I ~
Vp Np Vs + Burden
Fig. 3.2 Equivalent circuit of a voltage transfonner.
behaviour of a single-phase voltage transformer or any phase of a three-phase

transformer may thus be determined from the well-known equivalent circuit
shown in Fig. 3.2.
3.2.1 Behaviour during steady state conditions

It is clear from the equivalent circuit that ideal transformation cannot be
achieved in practice, because of the voltage drops which occur across the
primary and secondary windings, i.e. Ip Zp and Is Zs.
To minimize the errors for a given secondary current, the winding imped-
ances (Zp and Zs) and the exciting current (Ie) should be as low as possible.
Economic and consequential dimensional constraints limit the levels to which
the winding resistances and exciting currents may be reduced however and the
leakage inductances (Lp and Ls) are dependent on the spacing needed between
the windings for insulation.
The voltage error is defined as:
Voltage error = V
I
Kn Vsl - Vp
per unit
p
in which Kn is the desired or rated transformation ratio. From the phasor

diagram shown in Fig. 3.3 and assuming that Kn = Np/Ns, the following
approximate expression for voltage error can be derived:
Voltage error =
N
Is (Rs cos <ps + mLs sin <Ps) + Ip N S [Rp cos (<pp - a.) + mLp sin (<pp - a.)]
p
per unit
Fig. 3.3 Phasor diagram. Note: To increase clarity of the diagram, voltage drops, such
as Is Rs, are shown to be much greater than those which occur in practice.
The phase displacement error, shown in Fig. 3.3, is - a. rad (VS lags its ideal
position) and for the small values normally encountered in practice it may be
expressed as:
3.2.2 Behaviour during abnormal conditions

An extreme condition may arise on energizing a circuit to which a voltage
transformer is connected. Should the switching occur at a voltage zero, the
voltage impressed across the primary winding of the transformer would be:
vp = Vppk sin rot
The core-flux variation required to induce the necessary e.m.f. in the prim-
ary winding (Np turns), assuming the latter to have zero impedance, would be:
 =~
Np
Jepdt = ~p Jvp dt
V ppk
=-----.:T
ro,Yp
{cos rot - k}
in which the constant of integration (k) would be equal to unity if the initial
core flux was zero.
This flux would then reach a peak value of 2Vppk/roNp, i.e. twice the normal
steady state peak excursion. This condition would usually cause the core to
Fig. 3.4 Possible conditions caused by the incidence of a fault.
saturate and an accompanying exciting-current surge, with peak values up to

several times the normal rated current of the primary winding, would flow.
The related situation in current transformers was discussed in section 2.2.4. In
the case of a voltage transformer, however, the surge of exciting current causes
a relatively large voltage drop across the primary winding and, as a result, the
secondary winding voltage is both distorted and has an r.m.s. value below the
ideal value for several cycles. Such a situation could affect the behaviour of
protective equipment.
Clearly, abnormal voltages are present in power circuits during fault condi-
tions. Usually, however, the voltages are reduced below their normal levels at
such times and, in these circumstances, saturation of voltage transformer cores
will not tend to occur and accurate transformation should be achieved. It must
nevertheless be recognized that the phase of the sinusoidal component of the
voltage applied to a transformer may be changed by the presence of a fault as
shown in Fig. 3.4. As a result, the period between two voltage zeros may be
greater than normal and therefore a greater flux variation will be needed
during this period. This situation is, however, not likely to be as severe as that
considered above when a circuit is energized, and it can probably be disre-
garded.
3.2.3 Voltage transformer construction

Voltage transformers are basically very small versions of power transformers
and therefore only those factors and features which are special to voltage
transformers are discussed below.
Connections
As stated earlier, three-phase voltage transformers usually have star-connected
primary and secondary windings. They may also have tertiary windings con-
nected in star or broken-delta. Single-phase transformers are also produced
and in many cases they are used on three-phase systems with the above
connections. In some countries an alternative arrangement that is popular is
to use two such transformers with their primary windings connected between
pairs of phases so that they are each energized at a line voltage. Their second-
ary windings are connected in open delta and thus outputs proportional to the
three-line voltages are provided. The phase voltages cannot however be deter-
mined from these outputs unless it is known that they do not have zero-
sequence components.
Magnetic cores
Cores constructed from laminations of silicon-iron alloy are produced in both
single-phase and three-phase forms.
While three-phase, three-limb cores would allow satisfactory operation to
be obtained when the three-system phase voltages add to zero at every instant,
they would not provide acceptable transformation in the event of faults to earth
on the system. In these circumstances, the phase voltages do not add to zero,
i.e. a zero-sequence component is present, and in consequence the fluxes in
the three limbs do not add to zero. An extra flux path must therefore be
provided. Often, to provide symmetry, two extra limbs of small cross-sectional
area are added, the five-limb configuration commonly adopted being shown
in Fig. 3.5(a)
Single-phase transformers use either the well known, core- or shell-type
cores shown in Fig. 3.5(b).
Primary windings
Each primary winding is normally energized at the rated voltage of the circuit
to which it is connected, i.e.
Up = Vppk sin (rot + a)
Neglecting the voltage drop in the winding, there must therefore be an e.m.f.
induced in the winding of ep = - up and this must be produced by a core-flux
variation of:
cI> = - ~
Np
f ep dt = ~
Np
f up dt
Vppk
= - - - cos (rot + a)
roNp
The peak flux density should not exceed the saturation density of the core
(B sat ) and therefore:
Extra limbs
(a)
[[ID]
Shell-type
(b)
oCore-type
Fig. 3.5 Core configurations.
in which Ac is the cross-sectional area of the core.

In practice, transformers, depending on their duty, are required to be able
to operate without saturating at voltages above the rated value of the circuits
to which they are to be connected. Standards include a voltage factor (Vr)
which may range in value from 1.2 to 1.9.
Taking this into account:
Vr Vppk
(3.1)
B sat ~ roNpAc
in which Ac is the cross-sectional area of the core.
Considering, as an example, a transformer with a core having a saturation
flux density of 1.5 T and limbs of 25 cm2 cross-sectional area, it can be found
from equation (3.1) that its primary windings would require to have at least
55 O()() turns if it were for use on a 66 kV (line), 50 Hz system and to have a
voltage factor of 1.2. For an output of 200 VA/phase, the primary winding
current would be less than lOrnA. It will be appreciated that the production
of a winding with such a large number of turns of very fine wire is very
difficult and expensive especially when it is borne in mind that it has to be
insulated to withstand a high voltage. It is for this reason that electromagnetic
voltage transformers are not produced for use on the very high voltage systems
in use today.
Secondary windings
These windings present no difficulties because they have relatively few turns
of conductor capable of carrying a few amperes.
Fig. 3.6 Typical oil-filled voltage transformer. (Reproduced from Low oil volume
voltage-transformers, GEe High Voltage Switchgear Limited with the permission of
Instrument Transformers Limited).
Tertiary windings
These may be provided and connected in open- or broken-delta to give an
output under earth fault conditions which can be fed to protective equipment.
This output is proportional to the sum of the three zero-sequence voltages
present on the system.
Housing and insulation

Voltage transformers are often mounted in metal tanks, the cores and windings
being immersed in oil, in a similar way to power transformers, a typical
example being shown in Fig. 3.6. In recent years, however, resin-insulated
designs have been produced for use on lower voltage circuits.
3.3 CAPACITOR-VOLTAGE TRANSFORMERS

As stated earlier these transformers, which incorporate capacitor dividers,
were introduced for use on very high voltage circuits, when it was recognized
Capacitor-voltage transformers 95
that suitable electromagnetic transformers could not be produced for such
applications.
3.3.1 Capacitor dividers

Capacitor dividers have the attraction of being loss-free, i.e. they do not
consume power. When they are unloaded, i.e. iout = 0, the output voltage
(vout), as can be seen from Fig. 3.7(a), is given by:
C[
Vout= C[ + C2 vp (3.2)
Using Thevenim's Theorem, a divider supplying a burden may be repres-

ented by the circuit shown in Fig. 3.7(b) and this makes it clear that the output
or burden voltage (vout) is affected by the current in the burden (iout) because
of the voltage drop across the capacitor (C[ + C2 ). This drop, for a given ratio
of the capacitances C [ and C2 , reduces as both capacitances are increased in
the same proportion and therefore the capacitances should be as large as is
practicable.
It can be seen from equation (3.2) that a reduction of the capacitance C2 for
a fixed value of the capacitance C[ increases the output voltage (vout) and thus
for a given voltage-ampere output, the output current (iout) reduces. As a result
the percentage effect on the output voltage reduces.
vp
~C1
-
iout
C2
lv~ Burden
(a)
C1+C2
1
C1C
--~
+C2 p
t '" Burden
(B)
Fig. 3.7 Capacitor divider and its equivalent circuit.

Fig. 3.8 Circuit of a capacitor-voltage transfonner.
3.3.2 Circuit of a capacitor-voltage transformer

Because of the above considerations, capacitor-voltage transformers incorpor-
ate dividers in which the capacitor connected to the system (C 1) is made as
large as possible, usually 2000 pF, and the lower capacitor (C 2) has a value
which will provide a voltage (vout) which can be further transformed down by
a conventional voltage transformer. The r.m.s. value of the voltage (Vout)
which is termed the intermediate voltage, may be up to about 30 kV (phase).
To further reduce the errors caused by the presence of the effective capacit-
ance (C 1 + C2 ), shown in Fig. 3.7(b), an inductor (L) is connected in series
with the primary winding of the step-down transformer. This is shown in
Fig. 3.8.
3.3.3 Steady state behaviour

An equivalent circuit of a capacitor-voltage transformer is shown in Fig. 3.9.
The sum of the reactance of the inductor (L) and the leakage reactance of
the primary windings and the leakage reactance of the secondary windings of
Fig. 3.9 Equivalent circuit of a capacitor-voltage transfonner.

Capacitor-voltage transformers 97
the step-down transformer, referred to the primary level is made equal to the
sum of reactances of the capacitors C, and C2 at the normal power-supply
frequency, i.e. roIL + Lp + Ls (Np/Ns )2} = l/ro(C, + C2 ) The transformation er-
rors, which can be calculated for any given condition using the equivalent
circuit, are dependent on the system frequency and they should be determined
not only at the nominal frequency but at the lower and higher limiting values
likely to be encountered in service.
3.3.4 Behaviour during abnormal conditions

When a fault occurs on a power system, the voltages at different points on it
may fall suddenly and then they may suddenly increase when the fault is
cleared. Such a condition is illustrated in trace (i) of Fig. 3.1O(a) which was
Fault Fault
applied cleared
(i)
(ii) Fault A
(iii)
(a)
(i)
(ii) Fault B
(iii)
(b)
Fig. 3.10 Output voltage waveforms from voltage transformers under fault conditions.
(Reproduced from Gray and Wright, 1953, Proc.IEE, 100 with the permission of the lEE).
obtained experimentally using an unloaded resistor divider. The outputs of an
electromagnetic voltage transformer and a capacitor voltage transformer ob-
tained at the same time are shown in traces (ii) and (iii) respectively in Fig.
3.1O(a). In both cases the transformers were unloaded. A further set of traces
obtained with the rated burdens connected to the transformers are shown in
Fig. 3.1O(b). These results clearly indicate that the response of the electro-
magnetic transformer is close to the ideal, whereas the capacitor voltage
transformer generates undesirable damped oscillations when the sudden
changes of system voltage occur. These oscillations, which arise because of
the tuned nature of the transformer circuit, together with sustained oscillations
which can arise because of ferro-resonance, can affect the behaviour of pro-
tective equipment. To counteract these undesirable transient errors, Harder [5]
developed compensating circuitry which he patented in the United States in
1950. Subsequently, Hughes [6] studied the behaviour of distance-relaying
equipment supplied from basic capacitor-voltage transformers and from those
fitted with compensating equipment. All capacitor-voltage transformers are
regarded, however, as having a narrow bandwidth and an alternative developed
to provide a wide bandwidth was the cascade voltage transformer which was
made up of several individual electromagnetic transformers, the primary wind-
ings of which were connected in series. In this way, the primary voltage was
broken down into sections, each of the transformers having sufficiently low
voltages across them to enable satisfactory primary windings to be produced.
3.4 RECENT DEVELOPMENTS

The acceptance in recent years of the inclusion of electronic equipment in
protective schemes has opened up the possibility of using devices such as
amplifiers in voltage transducers. The incorporation of an amplifier to provide
the volt-amperes required by a burden enables a capacitor divider to be effec-
tively unloaded at all times and thus to provide a constant step-down ratio. In
these circumstances, the tuning inductor (L) used in a conventional capacitor
voltage transformer is not necessary. Such designs have been in service since
1975.
Stalewski and Weller [7] presented a paper to the lEE in 1979 in which they
described a capacitor-divider voltage sensor which incorporated both a pre-
amplifier and a main amplifier, the basic circuit being shown in Fig. 3.11.
Experience gained during high voltage system trials of this sensor extending
over the five years prior to the publication of the paper showed that it provided
faithful reproduction of the primary voltage during transient conditions. Dur-
ing this trial period, 27 system faults occurred in the vicinity of the substation
where the prototype unit was installed, including three faults actually on the
400 kV line to which it was connected. Very satisfactory performance was
obtained and further improvements were effected as a result of the information
obtained. Sensors of this type, which must of course have a reliable power
Specifications and testing 99
I H.V. Electromagnetic
step-down
C1
transformer
c2 Amplifier
~@B"m,"
Fig. 3.11 Capacitor-divider voltage sensor.
source to supply the amplifiers, should be capable of supplying outputs of the

accuracies needed by the latest protective schemes.
An alternative approach, using electro-optic materials, has been examined
in recent years. These materials, which are similar to those referred to earlier
in section 2.5, change the polarization of light passing through them when they
are placed in electric fields. A voltage transducer may therefore be produced
by placing a piece of electro-optic material near a conductor so that it is
subject to an electric field proportional to the potential of the conductor. As
in the current transducer, a polarized light beam is passed through the electro-
optic material and the changes in polarization, which are proportional to the
conductor potential, are measured to provide a signal which may be used to
supply measuring or protective equipment via optical fibres. It has been shown
that these transducers are capable of faithfully reproducing the primary volt-
ages applied to them under steady state and transient conditions. Further
details of this development are provided in publications by Kanoi et at [8],
Najima et at [9] and Mitsui et al. [10].
3.5 SPECIFICATIONS AND TESTING

British Standard 3941, published in 1975, specifies the requirements for both
electromagnetic- and capacitor-voltage transformers intended for use with
electrical measuring instruments and electrical protective equipment. It is
generally in line with the corresponding IEC Recommendations given in Pub-
lications 186, 186A and 358.
The accuracy classes for measurement purposes are 0.1, 0.2, 0.5, 1.0 and
3.0 and for protective applications they are 3P and 6P. The permissible steady
state voltage and phase displacement errors for each of the above classes are
given in the Table 3.1.
These limits must be complied with at any voltage between 80% and 120%
of the rated value and with burdens of between 25% and 100% of the rated
burden, at a power factor of 0.8 lagging.
Table 3.1
Accuracy class Voltage errror Phase displacement error
(%) (minutes)
0.1 O.l 5
0.2 0.2 flO
0.5 0.5 20
1.0 1.0 40
3.0 3.0 Not specified
3P 3.0 120
6P 6.0 240
In addition to having to comply with the above steady state limits, capacitor
voltage transformers also have to meet conditions associated with their tran-
sient response and effects produced by ferro-resonance. They are required, one
cycle after a short circuit is applied across their primary terminals, to have a
secondary output voltage of less than lO% of the peak value which existed
before the short circuit. Also with their secondary terminals on short circuit
and zero volt-ampere burden, ferro-resonance should not be sustained for more
than 2 seconds after the short circuit is removed, and when 120% of rated
voltage is applied the secondary voltage should revert to a value which does
not differ from its normal value by more than lO% within lO cycles after a
short-circuit across the secondary winding is removed. Methods of conducting
type tests, commissioning tests and routine tests are described in specification
documents.
3.6 THE FUTURE
Conventional voltage transformers will continue to be used to supply many

protective equipments in the future and certainly this will be so for applica-
tions associated with relatively low voltage power networks.
The needs of protective equipments, which are to be applied to transmission
lines operating at extremely high voltages, for very accurate and detailed
information about system voltages will almost certainly justify the use of
non-conventional transducers operating in ways such as those described in
section 3.4. A description of developments and trial tests in this area may
be found in [11].
REFERENCES
1. Andrews, L. (1898) The prevention of interruption to electricity supply, J. Inst.
Elec Eng., 27, 487-523.
2. Clothier, H. W. (1902) The construction of high-tension central station switch-
gears, with a comparison of British and foreign methods, J. lEE, p. 1247.
Further reading 101
3. Edgcumbe, K. and Punga, F. (1904) Direct reading measuring instruments for
switchboard use, lEE, 33, 620-66.
4. Wellings, J. G., Mortlock, 1. R. and Matthews, P. (1936) Capacitor voltage trans-
formers, J. lEE, 79, 577.
5. Harder, E. L. (1950) Transient compensation of potential-device burdens, US
Patent 2510631.
6. Hughes, M. A. (1974) Distance relay performance as affected by capacitor voltage
transformers, Proc. lEE, 121, 1557-1566.
7. Stalewski, A. and Weller, G. C. (1979), Novel capacitor-divider voltage sensors
for high voltage transmission systems, Proc. lEE, 126, 1186-1195.
8. Kanoi, M. Takahashi, G., Sato, T., Higaki, M., Mori, E. and Okumura, K. (1986)
Optical voltage and current measuring system for electric power systems, Trans.
IEEE, PWRD-1, 91-97.
9. Najima, K., Nishiwaki, S., Okuho, H., Yanabu, S. (1987) Measurement of surge
current and voltage waveforms using optical-transmission techniques, Proc. lEE,
134,415-422.
10. Mitsui, T., Hosoe, K., Usami, H., Myamoto, S. (1987), Development of fibre-optic
voltage sensors and magnetic field sensors, Trans. IEEE, PWRD2, 87-93.
11. Chatrefou, D. (1997) Application of optical sensors in extra high voltage
substations, GEe Alsthom T and D Rev., 111997, 17-24.
FURTHER READING
British Standard Specification for Voltage Transformers

BS 3941: 1975
lEe Publication
186 (1987), Voltage Transformers.
British Electricity Supply Standard

35-15 (1989) Protection and Measurement Transformers for HV Distribution Systems
up to 33 kV (58 pages).
4
Overcurrent and earth fault protection
Fuses are a very satisfactory form of protection for the lower voltage and
current sections of power networks because their operating time/current char-
acteristics are similar in form to the withstand time/current characteristics of
the circuits they protect. Their use enables circuits to be kept in service until
times when faulted or overloaded parts of a network must be disconnected to
ensure that healthy equipment does not suffer consequential damage.
Fuses are not capable, however, of interrupting large currents in high volt-
age circuits and they have the disadvantage that they cannot be tested regularly
after installation, a facility which is usually required in protective equipment
associated with major plant. As a result, relays with inverse operating
time/current characteristics, similar to those of fuses, were developed to pro-
vide overcurrent and earth fault protection. These relays, which are supplied
by current transformers, initiate the opening of circuit-breakers, when neces-
sary, by completing trip-coil circuits. They may therefore be used to protect
circuits operating at voltages up to the highest levels. Over the years such
relays have been installed around the world in very large numbers to provide
either the main protection of circuits or to give secondary or back-up protec-
tion to other more complex protective schemes.
The current and time settings of these relays are adjustable to allow them
to be applied in a similar way to fuses so that correct discrimination may be
achieved during fault or overload conditions. In some applications, however,
the use of time grading alone may not be sufficient to ensure correct operation
under all possible system conditions, and to improve protective performance
in such circumstances relays monitoring the direction of current flow, i.e. direc-
tional relays, are used in conjunction with overcurrent and earth fault relays.
Following the next section (4.1), which traces early relay developments,
details are provided of modern relays used in time-graded protective schemes
together with methods of applying them.

Protective relays were introduced quite widely into power systems during the
first two decades of this century and there are many references to them in the
literature of that period.

104 Overcurrent and earth fault protection
Reverse-current relays, i.e. directional relays, were quite common. A refer-
ence was made in Chapter 1 to a design described in a paper [1] presented to
the Institution of Electrical Engineers (lEE) by Andrews in 1898. An alterna-
tive design was described in the Electrical Review in Jllne 1904 [2], the device
then referred to being shown in Fig. 4.1.
An overcurrent inverse-time limit relay was designed in 1902 by C. E. L.
Brown, one of the founders of Brown Boveri Ltd. He took out patents on the
relay in several countries including Britain (Patent No 15500-03), these prob-
ably being the first world-wide patents on an a.c. relay. His original relay had
an aluminium disc which rotated in the horizontal plane, the drive being
provided by a shaded-pole electromagnet. It employed eddy-current braking
and was basically similar to modern designs. This relay is shown in Fig. 4.2.
Apparently it overran significantly if the current supplied to it was interrupted
before the disc had completed its travel. Its behaviour in this respect was
improved by modifications introduced in 1905. A relay of this improved type
was described in an article by Frey in the November 1924 issue of the Brown
Boveri Review [3]. It was claimed that experience with these relays over the
previous 20 years had overcome the prejudice which had existed against
somewhat indefinite time settings.
In spite of the introduction of induction relays several other types of cur-
rent-operated relays and associated time-delay devices were in use in the first
decade of this century, one being described in reference [2]. It was stated that
Fig. 4.1 An early reverse-current relay.

Fig. 4.2 An early over-current relay type A2 produced by Brown Boveri and Co Ltd.
(Reproduced with the permission of ABB Relays AG, Baden, Switzerland).
this relay had a solenoid plunger fitted with a miniature diving bell with two
superposed compartments, the lower one being bottomless. Small holes were
pierced in the top and middle partitions. The upper chamber dipped into oil
and the lower one into mercury. This mechanism delayed the lifting of the
plunger.
Relays with both definite time lags and with inverse time/current charac-
teristics were referred to in an article entitled 'Protective features of high-
tension system' [4] which appeared in Electrical World in 1909. This article
showed time/current characteristics which could be obtained using relays
which incorporated bellows to provide the time delays, one such relay being
shown in Fig. 4.3.
The use of eddy-current braking was referred to by Murphy during the
discussion on a paper on protective equipment presented to the lEE by Wed-
more [5] in 1915.
Reference was made to induction-type relays by Edgcumbe in a paper [6]
presented to the lEE in 1920. He stated that such relays are to be preferred to
fuse-shunted solenoids and went on to say that a fuse is 'at best' a capricious
piece of apparatus.
Fig. 4.3 Bellows-type overload relay.
Subsequent relay developments and their applications are described in the

following sections.
4.2 RELAY CONNECTIONS AND OPERATION

The relaying arrangements needed to detect earth faults and overcurrent con-
ditions, which may occur on three-phase networks are considered below.
4.2.1 The detection of earth faults

Because of the impedance which may be present in the ground-return path of
a circuit, earth fault currents may be quite low. It is nevertheless desirable that
they be detected and cleared by the opening of the appropriate circuit-brea-
kers. Earth fault relays should therefore have settings which will enable fault
currents well below the rated currents of the circuits they protect to be de-
tected.
Relay connections and operation 107
It is well known [7] that a single-phase, earth fault current may be repres-
ented by the set of symmetrical-component phasors shown in Fig. 4.4 and that
its instantaneous value may be expressed as:
It is therefore clear that an earth fault on any phase of a circuit may be

detected by feeding a single relay with a current proportional to the sum of
the three phase currents. This arrangement is illustrated in Fig. 4.5(a). Because
a relay connected in this way is not affected by normal balanced currents, it
may have a low current setting corresponding to 20% or less of the rated
current of the protected circuit.
4.2.2 The detection of over currents

As with fuses, relays provided to detect overcurrents must have minimum
operating currents above those corresponding to the rated currents of the
circuits they protect.
When an earth fault relay is provided to protect a circuit, the associated
overcurrent relays only need to operate in the event of three-phase faults or
interphase faults (e.g. phase a to phase b). This may clearly be achieved by
installing two overcurrent relays supplied with currents proportional to the
'.
-+
1 1
10= 3" (fa + Ib + Ie) = 3" la
If = la = 3/0
Zero
sequence
12e
Positive Negative _ _ _ Total fault
sequence sequence current
If = 11a + !2a + lOa
Fig. 4.4 Symmetrical components for an earth fault on phase 'a'.

s W f
Earth-fault relay
(a)
s~f
Earth-fault
relay
(b)
Fig. 4.5 Current transformers and relay connections.
currents in two of the phases, say a and c. This arrangement, which is shown
in Fig. 4.S(b), is commonly used although other arrangements are sometimes
employed.
4.2.3 Electro-mechanical relays

All the early relays were electro-mechanical, movement of armatures or discs
being used to operate contacts, which in tum caused the tripping of circuit-
breakers. Several types were produced and used in large quantities. Where
necessary, designs were modified to improve their performances and simplify
their production and this has enabled them to meet protective requirements up
to the present time. In recent years, however, the use of electronic control and
protective equipment has been increasingly accepted in power systems, and
manufacturers now produce electronic relays which perform similarly to their
electro-mechanical equivalents.
Because protective equipment practice was established when only electro-
mechanical relays were available, the following sections are devoted to the
behaviour of these relays and then, in section 4.3, equivalent modem elec-
tronic relays are described.
Three different basic types of relays have been produced to provide over-
current and earth fault protection, namely those which operate instantaneously
when the currents in them exceed set values, those which operate after a
definite or fixed time delay, and those which have inverse time/current char-
acteristics.
Instantaneous relays
These relays are relatively simple, employing, for example, a hinged armature
which is attracted to the pole of an electromagnet when the current in the
operating coil exceeds a certain value. The armature is coupled, via a linkage,
to the contacts. The attractive force on the armature is dependent on the current
and therefore the time taken for the closure of the air gap does vary with
current, but it is so short, a few milliseconds, that it is regarded as instant-
aneous. The operating coil is often tapped, connections being brought out to
a plug board so that a range of current settings is available. As an alternative,
the air gaps may be varied by moving the armature back stop to provide the
desired current setting.
Typical relays of this type are shown in Fig. 4.6(a-c).
n=:i 1
-
(a) (b) (c)
Fig. 4.6 Attracted-armature relays. (a) D.C. relay; (b) Shaded loop modification to
pole of relay (a) for a.c. operation; (c) Solenoid relay. (Reproduced from Protective
Relays - Application Guide, 3rd edn, GEC Measurements, 1987 with the permission
of GEC Alsthom Protection and Control Ltd).
Relays with definite time-lag operation
Various forms ofrelay have been produced for this purpose. They clearly must
monitor the current supplied by their associated current transformers and
trigger a timing system whenever the current exceeds the set value. A relay
of the type described in the previous section could be used, closure of its
contacts causing the energization of a timing device which would in turn
initiate the opening of an appropriate circuit-breaker after a fixed (definite)
time.
In such an arrangement the timing device should instantly return to its initial
position if the current-sensing relay resets as a result of the circuit current
falling to an acceptable level before operation is completed.
To enable such relays to be applied to systems, both the current and time
settings must be adjustable.
Timing has been achieved in various ways including dashpots, mechanical
escapements and eddy-current braked discs.
Relays with inverse time/current characteristics (IDMT)

There are several ways in which inverse time/current characteristics could be
produced and indeed, as stated earlier, relays with bellows, dashpots and
thermal devices have been used. Induction-type relays based on watt-hour
meters or with shaded-pole electromagnets were introduced by several manu-
facturers in the early years of this century. They were soon in use in vast
numbers around the world and both national and international time/current
standards were based on their performance.
Movements based on those of watt-hour meters have two electromagnets of
the form shown in Fig. 4.7. The main electromagnet (1) has two windings, one
of which is fed from a current transformer in the protected circuit whilst the
other acts as a secondary which feeds the winding on the second electromagnet
Electromagnet 1
Electromagnet 2 Permanent
L.=:======::...J magnet
Fig. 4.7 IDMT relay movement with two electromagnets.

(2). The first electromagnet provides a magnetic flux (CPI) and because of it an
e.m.f. is induced in its second winding, the phase of this e.mJ. being displaced
from that of the current in the first winding. As a result, the current driven
through the winding on the second electromagnet is not of the same phase as
that of the current supplied to the first electromagnet and therefore the phase
of the flux (Cjl2) produced by the second electromagnet differs from that of the
flux <Pl'
The two fluxes both cut the aluminium or copper disc, which is mounted in
jewelled bearings, and thus a torque, dependent on the input current, is pro-
duced on it.
The disc is restrained against a movable back stop by either a hair spring
or, in designs in which the disc rotates in the vertical plane, by a weight
mounted on an arm driven by the disc. The disc therefore only rotates when
the input current exceeds a certain value, its speed of rotation being dependent
on the input current and the eddy-current braking produced by a permanent
magnet. As the disc rotates it drives moving contacts, either directly or via
gearing, until the contacts close to energize the trip coil of the associated
circuit-breaker.
Because of the air gaps associated with the electromagnets, the two fluxes
(<PI and Cjl2) cutting the disc tend to increase linearly with the input current
until a level is reached at which saturation of the core materials begins. Up to
this current level the driving torque on the disc increases in proportion to the
square of the input current but at higher currents the torque tends to become
constant. The opposing torque provided by the hair spring or weight varies
with the distance travelled by the disc, having had an initial value which
prevents movement below the minimum operating current. The eddy-current
braking, however, is dependent on the speed of the disc. The behaviour of
such relays is therefore quite complex and their time/current characteristics
were expressed either in graphical form or by quoting the operating times at
a number of current levels, as shown in Table 4.1.
The characteristic shown in Table 4.1, which was referred to as 3/10, i.e.
the operating time is 3 s at 10 times rated current, was widely used. Other
characteristics, such as 40/20 and 30/5, were also produced.
Relays with these time/current relationships were described as 'inverse
definite minimum-time relays' (abbreviated to IDMT relays). The definite
minimum operating time obtained at currents above a certain level is an
essential feature of these relays as adequate discriminating time margins could
not otherwise be obtained between relays applied to adjacent sections of
networks during short-circuit conditions.
Table 4.1
Multiple of current setting 1.3 2 5 10 20 or more
time (s) 10 4.3 3.0 2.2
To allow flexibility in the use of these relays, tappings were provided on
the main input windings, either a current value or percentage being assigned
to each tapping. Overcurrent relays for use with current transformers having
secondary windings rated at 5 A were provided with tappings marked either
50% to 200% in 25% steps or 2.5 A to 10 A in 1.25 A steps. Earth fault relay
tappings were marked 20% to 80% in 20% steps (1, 2, 3, 4 A) or 10% to 40%
in 10% steps (0.5, 1, 1.5, 2 A). In addition, to allow time gradings to be
achieved, the operating times of a relay could be adjusted using a control knob
labelled 'time multiplier'. Rotation of this knob moved the back stop of the
disc and varied the travel needed to cause the contacts to close. Calibration
was provided in decimal form up to a maximum of 1.0 (full travel). The times
quoted in Table 4.1 were for a time multiplier setting of 1.0 and therefore the
times quoted would be halved if a time multiplier setting of 0.5 was used.
These practices are still in use today.
As an illustration, an overcurrent relay for use with a current transformer
having a ratio of 400/5 could be used to protect a circuit, required to carry a
maximum current of 300 A initially, but suitable for eventual uprating to 500
A. The relay would therefore be given a 75% current setting, corresponding
to 300 A, on installation, but this would be changed to 125% (500 A) later
when the load developed. These settings would be acceptable because opera-
tion would not occur below about 1.3 times the settings, i.e. 390 A initially
and 650 A later.
To enable time grading to be achieved, a time multiplier setting of 0.5 might
be used. In these circumstances, and with a 75% current setting in the above
application, operation would occur after 1.5 s for a fault current of 3000 A,
i.e. 10 times the current setting.
Relays with the same ranges of current and time settings were also produced
for use with current transformers having secondary windings rated at 1 A.
The alternative design of IDMT induction relays has a single shaded-pole
electromagnet of the basic form shown in Fig. 4.8. It has the same time/current
characteristic and time and current setting controls as the two-magnet design
Fig. 4.8 IDMT relay movement with a shaded-pole electromagnet.

Fig. 4.9 Modem IDMT relay. (Reproduced from Protective Relays - Application
Guide, 3rd edn, GEC Measurements, 1987 with the permission of GEC Alsthom
Protection and Control Ltd).
described above. Both types of relay have been widely used up to the present
day and they are interchangeable.
Since the early induction relays were produced, attempts have been made
to modify their designs, one aim being to increase the driving power obtained
for a given volt-ampere input. Such work has not been very successful how-
ever, because drives with greater efficiency have not allowed the earlier
time/current characteristics, on which standards were based, to be obtained.
Some improvements have however been made to simplify production, a mod-
ern element being shown in Fig. 4.9.
A factor of significance associated with induction-type relays is that their
discs and other moving parts acquire kinetic energy during operation and in
consequence they do not come to rest or reset instantly in the event of their
input currents reducing to rated or lower before contact closure occurs. A relay
could therefore overshoot and cause undesired tripping of its circuit-breaker
when an adjacent relay operates correctly to initiate the removal of a fault. To
ensure that this situation will not arise, allowance must be made for this effect
when the time-grading intervals between relays are being selected.
4.2.4 Directional relays

It was recognized at the beginning of this century that protective devices
whose behaviour depended only on circuit currents would not provide the
necessary discrimination during some operating conditions. Polarized relays,
i.e. relays with directional properties, based on wattmeter movements were
therefore developed. These incorporated a disc or vane and had two electro-
magnets, one having a winding fed from a current transfonner and the other
having a winding supplied by a voltage transfonner. Under steady state sinu-
soidal conditions, the flux (<Pc) provided by the electromagnet fed with the
circuit current would ideally be in phase with the circuit current whereas the
flux (<Pv) in the other electromagnet would ideally be in quadrature with the
voltage applied to its winding. For given values of voltage and current, maxi-
mum torque would be exerted on the disc in one direction when the input
voltage and current were in phase and maximum torque would be exerted in
the opposite direction when the voltage and current were in antiphase.
The torque of such relays is therefore proportional to power or VI cos <p, as
would be expected of a wattmeter movement. A relay of this type could thus
be made to close its contacts when the phase displacement between the voltage
and current was less than 1t/2 radians, whereas for phase displacements be-
tween 1t/2 and 1t radians the disc or vane would restrain against its back stop.
A relay with such a characteristic might not discriminate satisfactorily,
however, because fault currents usually lag their respective voltages by quite
a large angle. As an example, if a relay associated with phase 'a' were supplied
with the 'a' phase voltage and current, these two quantities could be almost
1t/2 radians apart during an earth fault on that phase. As a result the torques
on the disc would be very small for faults on either side of the relaying
position. This difficulty can be overcome by modifying relay characteristics
so that they produce maximum torque when the currents fed to them lag their
respective voltages by an angle similar to that which is likely to be present
during system faults, say 1t/3 radians. This behaviour, illustrated in Fig. 4.10,
can be obtained by incorporating a phase-shift network between the voltage
,,
,
Vw Max
" t?rque
' ....Ime
Fig. 4.10 Voltages associated with a directional relay . (VA is the voltage provided by
the VT and Vw is the voltage applied to the relay winding).
terminals of a relay and its voltage winding. Clearly, maximum torque can be
produced in this way at any desired angle between the input voltage and current.
A further important factor which must be recognized is that the torque
produced on a relay disc is proportional to the magnitude of the input voltage.
As a consequence, very small torques are likely to be produced for faults near
a relaying position, the voltage of a faulted phase being very depressed under
such conditions.
This situation would occur for faults in either direction from the relaying
position and correct discrimination might not be achieved. To minimize the
zone in which a relay may not correctly detect faults, directional relays cap-
able of operating satisfactorily at voltage levels down to 3% or less of the
normal level have been produced and indeed such requirements are included
in national and international standards. A practice which can improve perform-
ance in this respect is to connect the current and voltage windings of individual
directional relays to different phases. As an example, a relay may have its
current winding energized from phase 'a' but its voltage winding supplied with
a voltage proportional to that between phases 'b' and 'c'. With such connec-
tions, the relay voltage would not be depressed in the event of an earth fault
on phase' a'. Account must clearly be taken of such connections when setting
the phase angle between the voltage and current at which a relay is to produce
maximum torque.
Modern induction-type directional relay elements are generally similar to
those produced early this century, an example being shown in Fig. 4.11.
Fig. 4.11 A modem directional element. (Reproduced from Overcurrent and Earth
Fault IDMTL Relays. 1989, with the permission of ReyroUe Protection).
4.2.5 Thermal inverse time/current relays
These relays, which have been produced for many years, operate as a result
of the heating of elements. In this sense they are similar to fuselinks. Each
relay contains an element which is fed from a current transformer in the circuit
being protected, the input power therefore being proportional to the square of
the current. The heat produced is transmitted to a bimetal strip, fixed at one
end. The strip bends as it heats and after a certain deflection has been produced
it causes contacts to close and thus energizes the trip coil of its associated
circuit-breaker. The time taken for operation to occur at a particular current
level can be adjusted by varying the deflection needed to cause contact
closure.
It will be appreciated that the bimetal strip deflects whenever the heater
carries current and therefore the deflection associated with the rated current
of the circuit being protected must not cause contact closure.
Because of this factor, the operating time of a thermal relay, unlike that of
an induction-type relay, depends not only on the overcurrent flowing but also
on the current which was flowing earlier. In some applications, this behaviour
is desirable because it matches the behaviour of the protected circuits, namely
that the periods for which they can tolerate overcurrents is dependent on the
currents they have been carrying earlier.
The deflections of the bimetal strips in these relays are affected by the
ambient temperature and this could affect the time/current behaviour obtained.
To eliminate this effect, many designs incorporate pairs of bimetal strips, only
Load indicating scale and pointer
\ Adjustable overload contact

Overload setting
/ )Cale and pointer
.---::~~
Heater
Fig. 4.12 A modern three-phase thermal relay movement. (Reproduced from Power
System Protection, the Electricity Council, 1981 with the permission of the lEE).
Electronic relays 117
one of the pair having a current-fed heater adjacent to it, the differential
movement between the strips being used to control the operating time.
As with other relays, it is necessary that a range of current settings should
be available. This can be achieved by including a tapped interposing current
transformer to feed the heater.
A typical modern thermal relay is shown in Fig. 4.12 in which the bimetal
strips are of spiral form.
A limitation of relays of this class is that they do not reset instantly when
the currents supplied to them fall to the rated value or less after an overcurrent
has been flowing. They are not therefore very suitable for inclusion in time-
grading schemes unless they are associated only with load circuits. They are,
however, particularly suitable for the protection of small motors because they
can be set so that they do not operate during current surges caused by starting,
but nevertheless they will operate when faults or sustained overcurrents occur.
4.3 ELECTRONIC RELAYS

Manufacturers have attempted to produce static relays to replace electro-
mechanical relays for many years. Although electronic devices such as ther-
mionic valves were readily available from the 1930s, they were not considered
reliable enough to be used in protective equipment. Alternatives such as
transductors were introduced in the 1950s, a paper on this development being
presented to the lEE by Edgeley and Hamilton in 1952 [8]. Subsequently
relays incorporating silicon planar transistors were developed.
The later rapid progress in electronic equipment including the introduction
of reliable integrated circuits, logic gates and microprocessors made it possible
to produce equivalents of overcurrent and earth fault relays and also direc-
tional relays using both analogue and digital processing.
Because of the inherent flexibility of the circuitry involved, it is now
possible to obtain the performance required in any application very readily,
for example, any desired time/current characteristic can be produced. This is
a feature which is not available on electro-mechanical relays. In addition,
weaknesses of the latter relays, such as slow resetting and overshooting are
not present in electronic versions. It must be recognized, however, that elec-
tronic relays must have reliable power supplies and they must be made im-
mune to electrical interference which may be produced by power equipment
in their vicinity.
At present manufacturers are producing both electro-mechanical relays,
which remain popular because of their simplicity and the high level of perform-
ance provided by them over many years, and also a range of equivalent
electronic relays.
It is not possible in this work to give details of the circuitry used in modern
electronic relays and indeed it is likely to change quite quickly in the future.
Because of this, only a few examples, based on block diagrams are given below.
4.3.1 Basic electronic processing
Modem solid-state relays function by using digital circuitry to process incom-
ing signals derived from current and voltage transducers. Usually the input
signals are sampled at fixed intervals to produce a succession of discrete
pulses of the form shown in Fig. 4.13(a,b). The magnitude of each pulse is
stored as a binary number with a fixed number of bits. As an example, if an
8-bit coding system is used, then 28 (i.e. 256) signal levels can be stored.
Clearly the greater the number of bits used, the greater is the resolution and
accuracy achieved.
The sampling process removes some of the high-frequency content of an
input signal and it is generally necessary to sample at a rate or frequency of
at least twice that of the highest frequency of interest.
The circuit block which samples and codes the input signals is called
an analogue to digital converter (abbreviated to AID). The digital signals
produced by such a converter can then be processed to extract relaying inform-
ation, by performing a logical operation based on an algorithmic implementa-
tion of the relaying principle. The logic used for this purpose may be discrete,
i.e. a combination of electronic gates and operational amplifiers, or a suitable
microprocessor. Which of these alternatives is adopted depends on factors
such as the complexity of the relaying principle, the speed of operation and
the cost.
Discrete logic can provide rapid operation and it can be configured to deal
with several tasks in parallel. It lacks flexibility, however, because its circuits
(J)
"0
.
Ci
E

(a)
(J)
"0 I
- , ,_ ...... ,-
",- 1",
E I
Ci I
E
"

I
nT
(b)
Fig. 4.13 Sampling of a continuous time-varying signal.

must be changed to implement different tasks. In contrast, microprocessors are
able to perform complex sequential tasks but are not so suitable for dealing
with parallel tasks. This limits their processing speeds, but continuing techno-
logical developments are providing significant improvements. Microproces-
sors possess the advantage of flexibility, in that they can be used for a variety
of tasks without circuit alterations because their logic functions are im-
plemented in software.
Relays employing both discrete logic and microprocessors have been pro-
duced in recent years and a few examples are described in the following
sections.
4.3.2 Current-operated relay

A microprocessor-based, current-operated relay [9] manufactured by GEC-
Alsthom is shown in Fig. 4.14. The output of the main current transformer is
fed to a step-down interposing transformer in the relay. Its output after full-
wave rectification is fed to a circuit consisting of a set of shunt-connected
resistors which may be switched by the user to obtain the desired current
setting. The voltage thus obtained is applied to an AID converter, the output
of which is supplied to a microprocessor based on the Motorola 6805 chip.
This relay allows users to select a current setting at which instantaneous
operation will be initiated and also a current setting and associated inverse
time/current characteristic, seven different curve shapes being available.
Configurations which can provide three-phase overcurrent and earth fault
protection are available and directional features can be obtained using the
relay described in the following section.
4.3.3 Directional relay

A directional relay employing discrete logic [10], produced by GEC-Alsthom,
is shown in Fig. 4.15(a).
A voltage signal provided by the main voltage transducer is supplied to an
interposing voltage transformer in the relay. The output from this transformer
is phase shifted by an angle selected by setting a switch on the relay panel.
After amplification, the phase-shifted signal is converted into square pulses
thus clearly identifying the zero-crossing points.
The current signal obtained from the main current transducer is treated
similarly except that a phase change is not introduced.
The two signals produced by the above processing are then fed to a com-
parator containing one 'or' and two 'and' gates, an integrator and level detec-
tor. The behaviour of this comparator is illustrated for three conditions in Fig.
4.15(b).
When the inputs to the squarers are in phase with each other then the inputs
to each of the' and' gates are always different, i.e. 1 and 0 or 0 and 1, as shown
~
~Q)
OVJ
. - Q)
a. "O~
E .s
a.
ii5
:s .s
.s::
t:
as
Q)
II)
gj
()
L...--=---------L-1I'
Fig.4.14 Microprocessor-based, current-operated relay. (Reproduced from MCGG 22,

GEC Measurements with the permission of GEC Alsthom Protection and Control Ltd).
SO
o..l!!
-c::
:::lo
00
,
- "-'-'-'~'-'-i
I
I
I
I
I
I
~
1111 I
I
I
I
I
I
3l;::
S'-
.r::..r::.
c..<Il
a: r::T-+-----'
~ p.=;T-+--9-------i
L..--oa:::J>---------L--~--l--_lI
zrll'~'
~,_,_,.J
!
c::, al
~'-'-'-'-'
I
,~
al (,,)
Fig. 4.15(a) GEC Alsthom directional relay METI II, (Reproduced from MITI II,
GEC Measurements with the permission ofGEC Alsthom Protection and Control Ltd),
JLf JLf JL.J

v JLf UL JLJ
io
JLf JLf JLf
Vo
JLf UL JLJ
&OV JLf JLJl
&OV --'L ~
orOV JUUl
INT
(i)
/ (ii)
~
(iii)
(b)
Peak value
of intergrated
o~te!J!... _______
-1t +1t Phase

(c) angle 4>
Fig. 4.15(b), (c)
in Fig. 4.1S(b)(i). As a result zero inputs are always applied to the 'or' gate
and a zero signal is provided to the integrator.
When the inputs to the squarers are in antiphase with each other then the
inputs to the 'and' gates will alternate between 1,1 and 0,0 as shown in Fig.
4.1S(b)(ii). As a result the 'or' gate inputs are either 0,1 or 1,0 and a continu-
ous output signal is fed to the integrator which then provides a ramp output.
For the intermediate condition, shown in Fig. 4.1S(b)(iii), in which the
squarer outputs are displaced by n/2 rad, each 'and' gate produces an output
for only a quarter of each period (1t/2 rad) and as a result the 'or' gate produces
an output (1) for half of each period.
It will be appreciated that the 'or' gate produces an output (I) for a fraction
of 9/21t of each period in which 9 is the phase displacement either lagging or
leading between the signals supplied to the squarers and that the output from
the integrator rises during a cycle to a peak value proportional to the value of
the phase displacement 9. This behaviour is illustrated in Fig. 4.15(c).
The angular width of the operating zone of the relay can be controlled by
adjusting the setting of the level detector. As an example, if a setting of
0.5 Vmal< were used, relay operation would occur for values of 9 in the ranges
1t/2 ~ 1t and - 1t/2 ~ - 1t or with a setting of 0.25 Vmal<, operation would occur
T-------------T---------L1
+-r-----------~r_------~
A+-t----r------+-lh------- L3
N L...L.------<I-f 1operation
Forward
direction
~t -- ~~~ll~~?TIhtr~IlIf.RIJ
@ 1
-l1t---9Ji--[jll-l--I.
52
riWw------~-
rrrr'I rrrr'I
/ ~
6 16 20
-
17 19 11 15 18 10 14 8
-- - -----
4 12
I
I
I
+1-< _"\....J"-----,
51
(ji-<-
<p-<-
+ 1-< _"\....J"-----,
52
Fig. 4.16 Block diagram for the directional-overcurrent relays SPAS lKI 13 and
SPAS lK5 J3 manufactured by ABB. (Reproduced fromABB Buyer's Guide 1989-90,
with the permission of ABB Relays AG, Baden, Switzerland).
for values of e in the ranges 'It/4 ~ 'It and -'It/4 ~ -'It. These operating zones
are shown in Fig. 4.15(c).
The operating zones can then be set in the desired positions by selecting the
appropriate phase shift in the voltage circuit and a shift of 'It radians can
be obtained by reversing the connections from either the current or voltage
transducers.
Whenever circuit conditions cause the phase displacements between the
voltage and current to fall within the operating zone of the relay, the level
detector produces an output signal which may be used directly to open a
circuit-breaker or it may be used in logic circuitry which also takes account
of overcurrent or earth fault signals.
A block diagram of an alternative design produced by Asea Brown Boveri
[11] is shown in Fig. 4.16.
The output of the main voltage transducer is supplied to a threshold detector
in this relay. It is then phase-shifted before being applied to the input of a
band-pass filter, the output of which is then squared before being fed to a
comparator. The second input to the comparator is obtained after similar
processing of the signal provided by the main current transducer. As in the
GEC-Alsthom relay described above, an output signal is provided when the
phase difference between the initial voltage and current signals is within
desired limits. Again the output signal may be used in conjunction with signals
obtained from current-operated relay elements.
Further information on microprocessor-based relays, which it has not been
possible to provide in this work, may be obtained from references [12, 13] and
manufacturers' literature.
4.4 APPLICATIONS OF OVERCURRENT, EARTH FAULT AND

DIRECTIONAL RELAYS
In most applications, several individual relays or elements must be associated
with each circuit-breaker to enable the necessary discrimination to be achieved
during fault conditions and several elements may, for the convenience of
users, be mounted in a common case or housing. As an example, one earth
fault IDMT and two overcurrent IDMT elements together with associated
directional elements are commonly mounted in one case. In a similar way in
the following treatment, a single box is shown adjacent to each circuit-breaker
in circuit diagrams to indicate a relay consisting of a group of associated relay
elements.
The various types of relays described in sections 4.2 and 4.3 enable dis-
crimination to be achieved during fault conditions on many networks in the
following ways:
(a) by monitoring only current levels, i.e. current grading;
(b) by introducing time-delayed operation, i.e. time grading;
Applications 125
(c) by combining time-delayed operation with sensing of the direction of
current flow.
Applications where these methods may be used are examined in the follow-
ing sections.
4.4.1 Current grading

Current grading can be used to obtain correct discrimination in circuits where
there will be a large difference in the ratio of fault current to rated current in
sections of a network. Such a situation can arise when there is a unit of
relatively high impedance between two sections. This is illustrated in the
network shown in Fig. 4.17. The current in a load circuit in the event of a fault
on it (at point F I) would be limited relative to its rated current because of the
impedance of the transformer, T. The value of the current fed to the primary
winding of the transformer for this condition would therefore be much lower
than the current which would flow in the event of a primary-circuit fault at
point F2.
To illustrate this, the currents which would flow in the main feeder cable of
the network shown in Fig. 4.17 during three-phase and phase-to-phase faults
at point FI and F2 would be as follows:
Es
Three-phase fault at point F2 I=-=20pu
ZSI
Three-phase fault at point FI 1= Es =4pu

ZSI +Ztl
Phase-to-phase fault at point F2 24.75pu
Phase-to-phase fault at point FI 1= ...J3 Es = 3.69pu

ZSI + Zt l + ZS2 + Zt2
Es= 1pu ~ y
Fig. 4.17 Power network. Zsl =0.05 pu. Zs2 =0.02 pu. ZtI =0.2 pu. Z. =0.2 pu.
The above values are based on the rating of the main feeder. If each of the
loads had the same VA rating, the rating of the main feeder would be four
times that of each of the load circuits and therefore, in the event of faults at
point F l , the per-unit value of the current in the load circuit would be four
times that for the feeder cable, i.e. for three-phase and phase-to-phase faults
at point Flo the per-unit currents in the faulted cable would be 16 pu and 14.76
pu respectively.
In the above circumstances instantaneous overcurrent relays could be used
in the load circuits with current settings of say 150% or 1.5 pu to protect
against faults and sustained overloads. To obtain correct discrimination how-
ever and give rapid clearance for large faults, instantaneous relays set at 500%
or 5 pu could be used on the main feeder circuit. In addition, relays with fixed
or inverse time/current delays might be included to clear sustained overloads
from the main feeder circuit.
With a delta-star step-down transformer as shown in Fig. 4.17, zero-
sequence currents would not flow in the main feeder circuit as a result of earth
faults on the load circuits. Because of this, instantaneous earth fault relays
with low current settings could be used on the main feeder and load circuits
without loss of discrimination.
It will be realized that current grading can only be relied upon in a
limited number of applications but, because of their simplicity, instanta-
neous relays with suitable current settings should be used whenever possible.
Certainly they can often be used on the load circuits of radially-fed networks,
time grading or other methods being employed on the circuits nearer the
source.
4.4.2 Time grading using relays with definite operating times

In networks where there are several sections connected in series without
significant impedances at their junctions and where the source impedance is
much greater than the impedances of the sections, there will be little difference
between the levels of the currents which will flow for faults in different
positions on the network. In these circumstances, current grading would clear-
ly not enable satisfactory performance to be obtained.
Correct discrimination could, however, be obtained during fault conditions
by using relays set to operate after different time delays. The timing differen-
ces between the relays associated with adjacent sections could clearly be made
sufficient to allow the appropriate circuit-breaker to open and clear the fault
on its section before the relay associated with the adjacent section nearer to
the source could initiate the opening of its circuit-breaker. In practice, grading
intervals of the order of 0.5 s are adequate, this being the interval used in the
scheme shown in Fig. 4.18(a), for the protection of a network containing
several short lines in series. Because discrimination is dependent only on time
grading, the individual overcurrent relays can be set to any desired levels
Applications 127
Relay s
t o.5
s
Relay 1
Operating
time Relay 2 to.5s
Relay 3
--
(a)
Es
Source
(b)
Operating
time
(c)
Fig. 4.18 Discrimination achieved by time-graded relays.
above the rated currents of their circuits. Similarly earth fault relays may be
set independently at levels of the desired sensitivity.
It must be recognized that relays used near the sources in networks with
several sections connected in series may have unacceptably high operating
times for faults in the sections they protect because such faults may be of
current levels which can only be allowed to persist for short periods. It is
evident therefore that although it is very easy to select the operating times
needed to ensure that correct discrimination will be achieved when relays with
definite time lags are used, they have the weakness that their use must be
restricted to networks with relatively few series-connected sections.
4.4.3 Time grading using relays with inverse time/current characteristics

In networks such as that shown in Fig. 4.18(b), where the impedance (Zs)
between a source of e.mJ. (Es) and the input end of a protected section is small
compared with the impedance (Zps) of the protected section, there will be a
significant variation in the current levels of faults with their positions on the
protected section. For the extreme cases of faults at the near and remote ends
of a section, the currents would be:
and
respectively. Taking as an example, a situation where Zps =5Zs, the current

(In) for a near-end fault would be six times that for a fault at the remote end
(If). In these circumstances, and if relays with truly inverse time/current char-
acteristics were used (i.e. l/t DC I or It = constant), the operating time for a
near-end fault would be one sixth of that for a remote fault. The behaviour
which would be obtained with such relays applied to a network with three-
lines in series, allowing 0.5 s discriminating margins, is illustrated in Fig.
4.18(c). Clearly this behaviour is more desirable than that which would be
obtained using relays with definite time delays because the large fault currents
would be interrupted more quickly.
The energy let-through during a fault may be taken to be proportional to
12 t and on this basis the energy let-through in the above example would be
six times greater for near-end faults than for those at the remote ends of lines.
This situation might be regarded as unacceptable and it could clearly be
avoided by using relays in which operation occurs at a constant 12 value (i.e.
t DC 1/12). Relays with time characteristics of this form would also enable better
discrimination to be achieved in applications where the protected sections have
relatively low impedances and where, as a result, the currents during faults are
not greatly affected by the fault position. In these circumstances, the greater
sensitivity of such relays to such current differences would be beneficial.
For networks with sources of low volt-ampere output or in which some
circuits are remote from their sources, the effective values of source imped-
ance may be high and the behaviour obtained when using relays with inverse
time/current characteristics of any form might not be significantly better than
that which relays with definite time delays could provide. In such conditions,
the latter relays could be used on sections remote from the source while relays
with inverse time/current characteristics could be used in the remainder of the
network.
It will be appreciated from the above that there is not a single time/current
characteristic which would be ideal for all applications. Certainly the most
commonly used IDMT relays have the characteristic which the earliest induc-
tion relays provided, namely the one specified in section 4.2.3 (Table 4.1).
This is now designated as the standard normal inverse characteristic in BS142.
Manufacturers have had to continue to produce relays with this characteristic
to replace existing relays or for use on extensions to networks where they may
be essential to provide correct discrimination with the relays already in use.
Relays with other characteristics are, however, now available, manufac-
turers in the USA having introduced two, one designated 'very inverse' and
the other 'extremely inverse', the latter providing approximately the constant
Applications 129
]2 t performance, referred to earlier. These relays, like those with the standard
inverse characteristic, do not operate below certain minimum current levels
and they also have fixed or definite minimum operating times above certain
current levels. The relative shapes of the above characteristics can be seen
from Fig. 4.19, in which each of the curves is for a time multiplier setting of
unity. In practice, performance may be improved in some applications by
having some relays with one characteristic and the remainder with another.
It will be appreciated from the above that it is more difficult to choose the
necessary settings for relays with inverse time/current characteristics than it
is when relays with definite-time delays are to be used. To ensure that correct
100 I \
I \
I \
\ \
\ \ \
\ \ \
\ \ \
\ \ \
\ \
\ \
\ \
\
\
\
10
Definite 8 seconds
Q) Longtime standy
E earth fault
"" t= 120
/-1
Definite 2 seconds
Standard inverse
t=~
/0.02_1
1.0
Very inverse
t= 13.5
/ -1
Extremely inverse
t= -...!!lL
/2 -1
0.1 '------'--'----'---L.....L....J....J...I....I.._ _...l---1_I-...L-l.....LJ...LJ

10 100
/ (x Is)
Current (multiple of setting)
Fig. 4.19 Time/current characteristic (Is = current setting). (Reproduced from MeGG 22,
discrimination will be obtained under all conditions on a particular network a
study of its behaviour should be done first to determine the currents which
will flow in its various sections, in the event of faults at several different points
on it. Because the relay operating times are relatively long, steady state
calculations using well established techniques such as symmetrical compo-
nents may be used.
As a guide, a possible procedure for determining the settings of overcurrent
relays, with standard inverse time/current characteristics, needed to protect the
network shown in Fig. 4.20 is outlined below.
Because the system is radial and currents cannot be fed back towards the
source, the relays protecting the loads would operate most rapidly. The char-
acteristics of the individual loads must be taken into account, however, when
determining the performances required of the relays associated with them.
Should loads contain large motors, for example, then relays with time-delayed
operation may be needed to ensure that maloperation will not be caused by
the surge currents which may flow when motors are started. In these circum-
stances thermal relays might be used. For other loads, instantaneous relays
may be satisfactory and their use would be advantageous because the operating
times of relays at points nearer the source would be lowered. In this example,
it is assumed that instantaneous relays could be used to protect all the loads
and that each relay would be set to operate at a current somewhat above the
rated value of its load.
When determining the settings of the other relays on the network, the
following factors must be taken into account:
1. That upstream relays may overrun after a fault is cleared by a downstream
circuit-breaker.
2. That tolerances are allowed on the time/current characteristics of IDMT
relays in the various standard specifications and in consequence an up-
stream relay may operate slightly faster than would be expected from its
basic characteristic whilst a downstream relay may operate more slowly.
3. That a finite time is needed for a circuit-breaker clearing a fault to open
and effect interruption.
4. That the current transformers feeding the relays may have errors within
the limits allowed in national and international standards.
To allow for these factors it is usual to have a discriminating time margin
between adjacent relays of about 0.5 s. This time interval may be reduced if
an assessment of the minimum time grading interval t' is made. This may be
done using the formula [14]
f
, =[2ER100
+ ECT] f + fCB + fo + fs, seconds
where ER is the percentage relay timing error,

Applications 131
Cij
~ ::l
0 0"
al CD
a. "0
al CD
0
E
~ E
::l
UlUl
> al
~*tD .r::.r::!!?
C\J I'- -.r:: a. a. al
>ci>~aaUl
-">5EEE~
- ~~
~::: OOe:
~ ~OOOal
_ C\J
-" -
C\J C\J C\J "0
- CD
~
(; ~ ~ a. 0
C\J
~ C') ~ C') .~ ..t
~g~C\J- fl e:
0
CD -" - e:
ex: -.5 X CD-"C')~LO(X)Q)
~,.... . rn ::J ~
e
Q)
...J L ::l~C')CDQ)CDO"
~~~:5:5:5~Q) a..
i!? > E
* *
~
0>
E CD
e:
>.
en
~
al
a.
"0
e: lii
al ~
~ 0
0
CD
> a..
:;::
c..i
~
z
00
0
a.
CD
iii
Fig. 4.20 The protection of a radially-connected network.
ECT is the current transformer ratio error,

t is the nominal operating time of the relay nearer to the fault,
tCB is the circuit-breaker interrupting time,
to is the relay overshoot time

and ts is a suitable safety margin.
For simplicity, in the example that follows a discriminating margin of 0.5 s
has been used, i.e. t' =0.5 s. In this example, therefore, relays Rl.R2 and R3
must not operate in less than 0.5 s for faults beyond the circuit-breakers in the
load circuits, i.e. for a fault at a point such as Fl'
To ensure that this requirement will be met the maximum current flowing
in line 1 for a fault at point Fl must be determined. It can be seen that this
condition would occur if the circuit-breakers CB2, CB3, CB4, CB6, CB7 and
CB9 were open and if the source impedance Zs had its lowest possible value.
With the parameters shown, the current flowing in line 1 in the event of a
three-phase fault at point Fl would be 3.9 kA. Line 1 is rated at 200 A and
relay Rl is fed from current transformers of ratio 200/1. Relay Rl could thus
be given a 100% setting as it would not operate at line currents of less than
about 260 A (i.e. 1.3 x 200 A). With this setting, the above fault current of 3.9
kA would represent a multiple of 19.5 (i.e. 3900/200) for which relay Rl would
operate in 2.29 s with a time multiplier setting of 1.0. To provide operation in
0.5 s a time multiplier setting of 0.22 would be needed.
In the event of a phase to phase fault at point Fl the maximum possible
current in line 1 would be 3.38 kA which would represent a multiple of 16.9
(i.e. 3380/200) to relay 1 giving an operating time of 2.4 s for a time multiplier
setting of 1.0. For this condition a time multiplier setting of 0.21 could be used
to obtain operation in 0.5 s. Similar calculations should be done for other types
of faults at point Fh e.g. two phase to earth and single phase to earth and
clearly the highest of the time multiplier settings determined would have to
be used to ensure correct discrimination.
It is improbable that the network would ever be operated with so many of
its circuits disconnected or unloaded and hence the currents flowing in line 1
for faults at point Fl would never reach quite such high values as those
determined above.
Although, in most cases, the presence of loads is likely to lead to only small
reductions in the currents flowing in lines during fault conditions, the above
calculations could nevertheless, if desired, be repeated to determine whether
significantly lower time multiplier settings could be used, as this would speed
up the operation of relays near the source. For the network, represented in Fig.
4.20, the current in line 1 would reduce from the above value of 3.9 kA to 3.5
kA, if all the loads were connected, each of them being assumed to have the
impedance corresponding to its rated load and a lagging power factor of 0.6.
In this case, relay Rl could have a time multiplier setting of 0.21 instead of
the value of 0.22 determined earlier. This is an insignificant reduction but in
other situations larger reductions might be possible.
The settings of relays R2 and R3 could clearly be determined in the above
manner.
Relay R5 protecting line 5, rated at 75 A and fed by current transformers of
ratio 100/1 could be given a current setting of 75% or 0.75 A, thus allowing
line overcurrents up to about 100 A to flow. This relay must have operating
Applications 133
times greater than those of relays Rio R2 and R3 by 0.5 s or more for faults on
lines 1,2 and 3. To ensure that this will be so, the operating times of relays
Rio R2 , R3 for faults at positions Fz, F3 and F4 must be determined for a range
of conditions. In this case the maximum fault current possible at point F2 on
line 1 should be determined, i.e. using the lowest value of source impedance
(Z8) likely to be present in service, a fault impedance of zero and with circuit-
breakers CB4, CB6, CB7 and CB9 open.
Using the parameters shown in Fig. 4.20, a fault current of 5.8 kA would
flow under these conditions in the event of a three-phase fault. This
would cause the greatest possible current multiple of 29 in relay R J and this
would cause it to operate in 0.44 s. The corresponding current for a phase to
phase fault would be 5.0 kA, i.e. a current multiple in relay RJ of 25. Operation
in this case would occur after 0.46 s. For these two conditions the current
multiples in relay Rs would be 9.7 (i.e. 727/75) and 8.4 (i.e. 629/75) which
would lead to operation in 3.0 and 3.2 s respectively using a time multiplier
setting of 1.0. To allow a discriminating time margin of 0.5 between relays
RJ and Rs the latter relay would require multiplier settings of 0.31 and 0.30
for the above two faults. Clearly other types of faults should be studied
similarly and the highest value of time multiplier setting would have to be
used.
This procedure could be repeated for conditions producing lower fault cur-
rents at point F2. For example, the highest possible source impedance likely
to be encountered could be used, all the circuit-breakers could be closed and
then fault impedances could be introduced to obtain fault currents at point
F2 down to the minimum level needed to operate relay Rio i.e. about 260 A.
Unless lines 2 and 3 and their loads are very similar to line 1 and its load, the
above procedure should be repeated for faults at points F; and F~. For each
condition the time multiplier setting of relay Rs needed to provide a 0.5 s
discriminating time margin should be determined and the highest value calcu-
lated would then be the one required.
Similar studies could then be performed to determine the settings of all the
other relays.
Repeating the studies assuming relays with other characteristics, e.g. very
inverse, would enable the arrangement which would provide the optimum
performance to be determined. Similar studies could also be used to determine
the settings of earth fault IDMT relays.
Although the above procedure appears to be lengthy, it can be performed
very quickly using modern fault computing programs.
Because the shortest relay operating times are obtained when large fault
currents are flowing, the time multiplier settings needed to provide the neces-
sary discriminating intervals are generally determined by the maximum fault-
current conditions and the amount of computation may be reduced if this is
assumed. Should it not be possible to undertake the above studies, then simpler
approximate methods may be used.
As an example, for the network shown in Fig. 4.20, the current fed to line
1 in the event of a fault at point Fz could be taken to be m times its rated or
full load current, i.e. Ir = mIre. Should lines 2 and 3 each have the same rating
as line 1 and assuming that they could be supplying full-load current in the
presence of the above fault, and further assuming that the three line currents
were of the same phase, then the current in line 5 would be equal to
(m + 2)Iu. In these circumstances, the rating of line 5 would be three times
that of each of the lines it supplies, i.e. 3Ire. As a result, relay Rs would be
operating at a current multiple of (m + 2)/3 when relay Rl was operating at a
multiple of m.
The value of m could vary from 1.9, the minimum level which would cause
relay Rs to operate up to the maximum value possible on the protected circuit
(20 in the earlier example).
A curve of the required operating times of relay 5 against either its own
current multiple or the value of m could thus be derived, such a curve being
shown in Fig. 4.21. From it the required time multiplier setting of relay Rs
could be obtained. The value needed for the relationship in Fig. 4.21 would
be 0.29. Again it is the maximum fault current level which governs the time
multiplier setting which must be used.
It should be noted that the approximations made in this latter method tend
to increase the discriminating margins slightly and they would not therefore
introduce the risk of incorrect tripping of circuit-breakers.
In some applications, high-set, instantaneous elements are used in conjunc-
tion with IDMT overcurrent elements to provide rapid operation in the event
of very high current faults. Such relays are usually used near sources where
current grading can be achieved, only IDMT relays being employed on the
other downstream circuits.
5
Q)
~ 4
C)
c:
~ 3
CD
8"2
Current multiple (m)
Fig. 4.21 Time/current characteristic of relays Rl and Rs (Fig. 4.20).

Directional and current-operated relays 135
4.5 THE APPLICATION OF DIRECTIONAL AND
CURRENT-OPERATED RELAYS
Satisfactory discrimination cannot be achieved for all the fault conditions
which can occur on highly-interconnected networks when using only overcur-
rent and earth fault relays. This can be seen from a study of the very simple
network containing a pair of parallel-connected lines, which is shown in
Fig. 4.22.
Because of the symmetry of the network, relays R\ and R3 would both have
the same time and current settings. Relays R2 and ~ would also have the same
settings. In the event of a short circuit on line A near circuit-breaker CB2, all
four relays (Rl-R4) would carry similar currents. Because relays R2 and ~
would have lower time settings than relays R\ and R3 which are nearer to the
source, relays R2 and ~ would operate together opening their respective
circuit-breakers CB2 and CB4. Relay R\ would then operate later opening
circuit-breaker CB 1 to clear the fault. As a result both lines would have been
disconnected, causing a complete loss of supply to the load and the advantage
of having duplicate lines would have been lost.
This incorrect behaviour could clearly be avoided by providing directional
elements in addition to the other elements at the ends of lines A and B remote
from the source. The directional relays would have to be set to operate only
when currents were being fed back towards the source, an abnormal condition
during which the current 12 would lead the voltage (V2) at the position of relay
R2 by an angle somewhat greater than nl2 radians. Normal currents flowing
to the load would appear to the relay to be displaced by an angle over nl2
radians from that during the fault. A suitable operating characteristic for the
directional relays would thus be of the form shown in Fig. 4.23. In practice,
a directional element would have to be provided with each of the overcurrent
and earth fault elements fitted near circuit-breakers CB2 and CB4.
It is usual to arrange that directional elements only allow their associated
overcurrent or earth fault elements to operate when faults occur on the appro-
Fig. 4.22 The protection of a net~ork_con.!llining_ two similar lines connected in

parallel. For h < < IF it follows that h "" h "" 14 "" - /z.
Relay operating zone

~--------------~~
Fig. 4.23 Conditions associated with relay R2 (Fig. 4.22) when a fault is present on
Ene A and during healthy conditions.
/zh = typical current during healthy conditions
2 =current when line A is faulty
Ir = fault current
V2 =voltage applied to relay.
priate side of the relaying position, i.e. on line A or B in the above example.
In applications in which IDMT induction-type elements are being used, the
directional element contacts are connected in the circuits of the windings on
the second electromagnets or in the shading-winding circuits. In this way the
disc of an induction element will not rotate unless the current is in the direction
for which operation should occur.
If the alternative of energizing both a directional element and its associated
IDMT element at the same time were adopted and the contacts of the two
elements were connected in series in the trip-coil circuit, incorrect tripping
could result. This might happen, for example, if an overcurrent element com-
pleted its operation while its associated directional element was restraining
during a fault in the reverse direction. If after clearance of the fault current
flowed in the opposite direction before the overcurrent element began to reset,
maloperation would occur.
Directional elements should clearly be suitable for operation over the same
current ranges as their associated elements and, as stated earlier in section
4.2.4, they should be capable of operating correctly even when the voltages
applied to them are very depressed. This feature would be essential in the
unlikely event of a three-phase fault near the relaying position when all
the voltages available would be low. For other types of faults, however, the
situation may be imporoved by obtaining the polarizing voltages for elements
from a phase or phases different to that from which their currents are supplied.
Clearly there are several possible connections available and these were con-
sidered in detail by Sonnemann in a paper entitled 'A study of directional
element connections for phase relays' [15].
Directional and current-operated relays 137
As an example, one connection is considered below. As stated earlier in
section 4.2.4, an induction-type directional element may be set to provide
maximum operating torque at any desired phase angle between the current and
voltage supplied to it, by introducing the necessary phase shift in the circuit
of the voltage winding. The term maximum torque angle (MTA) was intro-
duced, this angle being defined as the angle by which the current must lag the
voltage fed to an element to produce maximum torque, say 30 or 1t/6 radians.
A connection which is often used because it tends to ensure an acceptable
voltage level during faults is to connect the current windings in one phase (say
phase a) and the voltage windings between the other two phases (say b to c).
This is termed a 90 or 1t/2 connection because the voltage Vbc is normally in
quadrature with the 'a' phase voltage. In these circumstances when using an
element with an MTA of - 1t/6 radians, the maximum torque would occur
when the element current (/a) lagged its phase voltage (Va) by 1t/3, a condition
likely to be present under fault conditions. This arrangement, which is illus-
trated in Fig. 4.24, is described as a (90-30) or (1t/2-1t/6) characteristic.
Although static or electronic relays clearly do not have maximum torque
angles, they nevertheless do have operating zones corresponding to those of
electro-mechanical relays and they may be specified in the above manner.
The necessary time and current settings of the overcurrent and earth fault
elements to be used on a given network could be determined using the methods
described earlier in section 4.4.3. In addition the optimum settings of direc-
tional relays could readily be found.
Directional overcurrent and earth fault relays are widely used to protect ring
mains and therefore this application is considered below to show how satis-
factory discrimination can be achieved.
A ring-main circuit with a single source and five substations is shown in
Fig. 4.25.
The double-headed arrows shown adjacent to the relays Rio R2 and R5
Fig. 4.24 Condition which provides maximum torque in a directional element with an
MT A of - n/6 rad.
138 Overcurrent and earth/ault protection
--7SlOS-- --
0.15 0.55
1. Q9s
~ ~ ~
--
I I I
2.15
X X---.I.........-X
CB3 CB4
Source ~
--7SlOSf
I
-- --
CLX 0.15 0.55
~B1 Ms
1.
~ I
~I
~I
X X X
CBS CB6 CB7
Fig. 4.25 The protection of a ring-main circuit.
indicate that these relays would operate for current flows in either direction,
i.e. these relays do not incorporate directional elements. This is acceptable
because these relays have higher minimum operating times than the others,
the values being 1.7 sand 2.1 s, i.e. the times shown adjacent to the arrows.
In this connection it will be appreciated that current would not be fed back through
circuit-breakers CB2 and CBS in the event of a fault on or near the source.
The other relays (R4-R7) each have directional elements so that operation
is obtained in short times for current flows in the directions indicated by the
single-headed arrows and longer times for current flows in either direction.
As an example, relay R3 would operate in a minimum time of 0.1 s for currents
flowing towards the source (right to left) and a minimum time of 1.3 s for
current flowing in either direction. To achieve this, two current-operated
relays and one directional relay are needed.
It will be seen that a discriminating margin of 0.4 s has been allowed
between the relays at adjacent substations.
For the short circuit shown between the circuit-breakers CB3 and CB4,
correct discrimination would be obtained. Circuit breaker CB4 would open
after a minimum time of 0.5 s followed by circuit-breaker CB3 opening after
at least a time of 1.3 s to clear the fault. The remaining circuits would remain
in service and all loads would continue to be supplied. Similar conditions
would obtain for faults in other positions.
It must be appreciated, however, that completely satisfactory behaviour
cannot be obtained in highly-interconnected networks fed from several sources
because the directions of current flow and the current levels at the individual
relaying positions do not enable the positions of faults to be correctly deter-
mined.
Current and voltage transformers 139
4.6 CURRENT AND VOLTAGE TRANSFORMERS
It is clearly essential that the accuracies of transformers supplying relays
should be high enough to ensure that satisfactory performance will always be
obtained. The requirements, which depend on the types of relays being used,
are considered below.
4.6.1 Electro-mechanical relays with a fixed current setting

These relays may operate either instantaneously or after definite time delays.
To ensure correct operation it is necessary that the current transformers sup-
plying them have small errors at the operating current levels. At higher cur-
rents, the errors are unimportant because they will not cause the secondary
currents to fall below the operating levels. In addition, transient errors caused
by saturation of cores during the first few cycles of fault current will not cause
maloperation and therefore only steady state conditions need be considered.
As an example, a relay with a 2 A setting (200%) fed from a current
transformer of ratio 200/1 would operate satisfactorily if the transformer cur-
rent error for a secondary current of 2 A was 2% or less. Assuming a secondary
circuit of impedance 5Q, i.e. 5 VA at rated current, a current error of not
more than 2% would have to be achieved at a secondary winding e.m.f. of
lOV.
This would certainly be achieved by a current transformer rated at 15 VA
with an accuracy classification of lOP5 as defined in BS 3938 : 1973. This
classification corresponds to a current error of 3% at rated current (1 A) for
a 15 VA burden, i.e. 15 V and a composite error (exciting current) of 10% or
0.1 A at five times rated current (5 A).
4.6.2 Electro-mechanical IDMT overcurrent relays

To ensure that satisfactory grading will be obtained over the whole operating
characteristic of an IDMT relay it is necessary that the current transformer
supplying it will maintain high accuracy over the range from the minimum
current level needed for relay operation up to the current at which minimum
operating time is achieved, i.e. 1.3-20 times the relay current setting.
These relays impose a burden of about 3 V A at rated current, i.e. the voltage
drop across a 1 A relay operating with a plug setting of 100% is 3 V. In
addition, allowance must be made for the voltage drops in the current trans-
former secondary winding and the interconnecting leads. Assuming, as an
example, that such a relay is supplied by a current transformer of ratio 200/1
and that the total voltage drop at rated current is 10 V, then core saturation
should not occur in steady state at a secondary winding e.m.f. of 200 V. This
performance would be provided by a transformer rated at 15 VA with an
accuracy classification of 10P20 as defined in BS 3938: 1973.
4.6.3 Electro-mechanical directional relays
The operation of directional relays is not dependent on the magnitudes of the
voltages and currents supplied to them but only on the phase differences
between these quantities. It is nevertheless necessary that the wavefonns of
these inputs should not be so distorted during fault conditions by saturation
effects in the cores of the transfonners which feed them, that the fundamental
components become significantly displaced in phase. While such displace-
ments are not likely to be produced in voltage transfonners because the
voltages applied to them tend to be depressed when faults are present, they
could occur in current transfonners when large system currents' are flowing.
To avoid such situations current transfonners capable of supplying their con-
nected burdens at the highest possible fault current levels without core sat-
uration occurring in steady state should be used.
The total VA burdens on current transformers including those of the sec-
ondary windings, the connecting leads and both the current-operated and
directional relays should be detennined at the rated-current level (e.g. 15 VA).
Because high transformation accuracy is not needed Class lOP transformers
are adequate and indeed this is the recommendation in BS 3938 : 1973. In most
cases, the transfonners needed to ensure correct operation of IDMT relays
should prove satisfactory when directional elements are also to be used, i.e.
say 15 VA Class lOP20.
4.6.4 Electronic relays

Transducers supplying modern electronic relays must provide outputs which
will enable correct operation to be obtained under all conditions, for example
a current transducer supplying a relay with an IDMT characteristic must
respond correctly at currents up to 20 times the relay setting current.
Because electronic relays are provided with power supplies, however, they
present negligible burdens to their transducers and therefore the burdens are
limited to those presented by the output windings of the transducers and the
connections between these windings and the relays. This factor should gener-
ally reduce the difficulty of obtaining transducers with the required perfonn-
ance.
The situation can be further eased if current transducers are used which
produce voltage outputs proportional to the input currents as they would then
provide voltage signals and negligible currents to the relays.
4.7 STANDARD SPECIFICATIONS

BS 142: 1982 [16] is a very comprehensive document covering electrical
protection relays. It is based upon, and is in technical agreement with, the
appropriate parts of IEC Publication 255 Electrical Relays.
Standard specifications 141
BS 142 provides definitions of the various tenns used in connection with
electrical protection relays as well as specifying characteristics, constructions
and testing procedures.
The basic requirements for IDMT and directional relays are summarized
below.
4.7.1 IDMT relays

As stated earlier, the time/current characteristics of the early induction-type
relays were adopted as standards. They resulted from the construction of the
relays rather than from considerations of the withstand abilities of the equip-
ment being protected. They were not originally specified in mathematical
fonn. In recent times, however, the general time/current relationship over the
normal operating range has been expressed in the fonn:
k
t=(GJa
- -1
s
Gb
in which
t is the theoretical operating time,
G is the relay current,
Gb is the basic or setting value of the current and
k and a. are constants characterizing particular relays.
For the standard normal-inverse, the very-inverse and extremely-inverse
relay characteristics, which are shown in Fig. 4.19, the values of k and a. are
as given in Table 4.2.
Manufacturers are required to specify the minimum or threshold operating
current level which must not be greater than 1.3 times the basic or setting
value. They must also specify the tolerances or error bands within which their
relays will perfonn. This must be done by declaring an assigned error as either:
(a) a percentage of the theoretical time; or
(b) a percentage of the theoretical time or a fixed maximum time error (where
this may exceed the percentage value), whichever is the greater, eg 5% or
20 ms; or
(c) a fixed maximum time error.
Table 4.2
k a
Normal Inverse 0.14 0.02
Very Inverse 13.5 1
Extremely Inverse 80 2
Effective range
,,
cD
E
i=
,
,, Limiting error
...
<.-'l
--------
_:::_-_-_-_-_-_-_...;.::_~ Theoretical
curve
2 Gs 20 Gs
Lower limit Upper limit
Gs"'G r "'1.3G s
Characteristic quantity
Fig. 4.26 Declaration of limiting error by means of graphical presentation.
The assigned error can be declared either by:

(i) a theoretical time/current curve bounded by maximum and minimum cur-
ves as shown in Fig. 4.26; or
(ii) an assigned error at the maximum current level multiplied by the factors
shown in Table 4.3 at lower values of current in the operating range.
As an example, if the manufacturer of a normal-inverse relay declared an
assigned error of 5% or 0.15 s, whichever is the greater, then operation would
have to occur within the ranges shown in Table 4.4
4.7.2 Directional relays

BS 142, Section 4.2, 1984 specifies conditions with which directional relays
must comply. The main requirements are listed below.
1. The operating phase angle range of a relay must be within 2 of the range
declared by the manufacturer.
2. Relays must be capable of operating at voltages between 3% and 110% of
the rated value except for earth fault elements to be used on systems which
are not solidly earthed, for which the range must be extended to 3%-
190%.
3. A relay must not operate
(a) when rated voltage is applied and currents in the non-operating zone
of up to 20 times rated current are applied or switched off, and
References 143
Table 4.3
Assigned error as Value of the current

a multiple of the as a multiple of the
declared error setting value
2.5 2
1.5 5
1 10
20
Table 4.4
Multiple of Normal Permissible error Permissible
basic value operating time operating time
(s) (s)
2 10 2.5 x 5% = 1.25 s 8.75-11.25
5 4.3 1.5 x 5% = 0.32 s 3.98-4.62
10 3.0 5%=0.15s 2.85-3.15
or 0.15 s
20 2.2 5%=0.11 s 2.05-2.35
or 0.15 s
(b) when with zero voltage applied, currents up to 15 times the rated value
are applied or switched off.
4.8 THE FUTURE

It seems probable that electro-mechanical overcurrent and directional relays
will continue to be used for many more years because of their relative sim-
plicity and proven reliability.
Electronic relays are now being installed in large numbers in many systems
around the world. They have certain advantages over their electro-mechanical
equivalents. For example, they can reset immediately when faults are cleared,
a feature which enables reduced time-grading intervals to be used. Future
developments such as the closer integration of protection and control functions
will also favour the use of microprocessor-based relays.
REFERENCES
1. Andrews, L. (1898) The prevention of interruption to electricity supply, J. Inst.
Elec. Eng., 27, 487-523.
2. Andrews, L. (1904) Automatic protective devices for electrical circuits, Electrical
Review, 54 933-935 and 972-975.
3. Frey, H. E. (1924) Relays for the protection of distribution systems, The Brown
Boveri Review, 11, 235-242.
4. Schuchardt, R. F. (1909) Protective features of high-tension systems, Electrical
World, 53, 1539-1543.
5. Wedmore, E. B. (1915) Automatic protective switchgear for alternating current
systems, J. Inst. Elec. Eng., 53, 157-183.
6. Edgcumbe, K. (1920) The protection of alternating-current systems without the
use of special conductors, J. Inst. Elec. Eng., 58, 391-416.
7. Wagner, C. F. and Evans, R. D. (1933) Symmetrical Components, McGraw-Hill.
8. Edgeley, R. K. and Hamilton, F. L. (1952) The application of transductors as
relays in protective gear, Proc. lEE, 99, 297.
9. Overcurrent relay for phase and earth faults, Type MCGG, GEC Alsthom, Publ
R-6054E.
10. Directional relay, Type METI, GEC Alsthom, Publ R-6003H.
11. Single-phase directional overcurrent relay, Type SPAS lK1J3, Asea Brown
Boveri, Publ34 SPAS 3ENI.
12. Sachdev, M. S. (ed.) Microprocessor relays and protection systems, IEEE Tutorial
Course, 88EH0269-I -PWR.
13. Phadke, A. G. and Thorp, J. (1988) Computer relaying for power systems, RSP
and John Wiley.
14. Protective Relays - application guide (1987) (3rd edn), GEC Measurements,
England.
15. Sonneman, W. K. (1950) A study of directional element connections for phase
relays, Trans. AlEE, 69, (II).
16. British Standard Specification on Electrical Protection Relays, BS142: 1982.
FURTHER READING
British Standard BS142 Electrical Protection Relays

Part 0: 1982 General Introduction and list of parts.
Part I : 1982 Information and requirements for all protection relays.
Part 2 : 1982 Requirements for the principal families of protection relays.
Part 3 : 1990 Requirements for single input energizing quantity relays.
Part 4 : 1984 Requirements for multi-input energizing quantity relays.
Part 5 : Supplementary requirements.
5
Current-differential protective schemes
It will be clear from the previous chapter that overcurrent and earth fault
protective equipment employing time grading and directional detection cannot
provide correct discrimination on all power networks and in many instances
the clearance times for some faults would not be acceptable.
An alternative protective scheme which has been in use for many years to
protect individual sections of networks or pieces of equipment, such as alter-
nators, detects the difference between the current entering a section and that
leaving it. Such schemes are designated as 'current-differential' or 'Merz-
Price', the latter being a combination of the names of the two persons who
proposed the basic principle of the scheme, which is expressed in Kirchhoff's
first law, namely that the sum of the currents flowing to a node must equal
the sum of the currents leaving it.
One phase of a basic current-differential protective scheme is illustrated in
Fig. 5.1. Under normal conditions, the current entering the protected unit
would be equal to that leaving it at every instant as shown in Fig. 5.1(a). The
two current transformers, if ideal, would have equal secondary currents which
would circulate around the interconnecting conductors, which are often called
'pilot wires'.
In the event of a fault on the protected unit, the input current would no
longer be equal to the output current. For such a condition the input current
would clearly be equal to the sum of the output and fault currents as shown
in Fig. 5.1(b), i.e. iPA = iPB + ie. The ideal current transformers would drive
secondary currents (iPA Np/Ns and iPB Np/Ns), proportional to their primary
currents, and these could not circulate through only the interconnecting con-
ductors. An interconnection must therefore be provided as shown, and this
would carry the difference current, i.e. it Np/Ns. A relay included in the inter-
connection would therefore carry a current proportional to the fault current at
every instant and circuit-breaker operation could be initiated instantly for any
fault current above a certain minimum level.
Should a fault occur at a point beyond the protected unit, that is at a point
which is not between the current transformer primary windings, then each
current transformer would carry the same primary current (ipA = i pB )' This
situation is shown in Fig. 5.1(c). As a result, no current would flow in the

146 Current-differential protective schemes
Current ipA ipB Current

transformer - - transformer
~X---p{lro=t=ec=te=d=u=n::r\t--~X~
End A End B
(a)
-h I lif[ ~
~X---p-(r=ot=ec=t=ed=u=n::rit....!.~~X~
Relay
(b)
(c)
Fig. 5.1 Basic current-differential protective scheme. (a) healthy circuit; (b) internal
fault; (c) external fault.
relay under ideal conditions, relay operation would not occur and correct fault
discrimination would be achieved.
It will be clear that differential protective schemes provide the possibilities
of rapid fault clearance coupled with correct discrimination. It must be recog-
nized, however, that they only protect the circuit or zone between their current
transformers and do not provide a measure of back -up protection to other parts
of the network. They are therefore categorized as unit protective schemes.
Whilst the above principle is simple, several factors have to be taken into
account when applying current-differential schemes to actual sections of net-
works or particular items of equipment. These will be considered both in this
chapter and later chapters. Initially, however, some historical information is
provided in the following section.
In 1904, Charles H. Merz and Bernard Price obtained a British patent (No.
3896) entitled 'Improvements in the method and means for protecting appara-
tus on alternating current systems'. In the introduction of the patent specifi-
cation it was stated that the 'cut-outs', based on detecting overcurrents, which
were then in use had sometimes failed to act when they ought to have done
and acted when they ought not to have done so. It was proposed that one or
more electromagnetic devices or contact makers (relays) be connected by one
or more conductors, referred to as a pilot wire or wires. Current transformers
were to be used and so connected that no current would flow in the pilot wire
or wires under normal conditions, but would do so in the event of a fault on
the protected equipment.
It was stated that 'protective devices to act in the manner described can be
variously constructed and arranged and be used not only in connection with
high pressure (voltage) feeders, transformers or other apparatus, but also with
any part of a distributing system'.
Because of its great historical significance, the circuit proposed in the patent
specification, to protect a single-phase feeder, is shown in Fig. 5.2.
The proposed scheme required that the current transformers produce equal
secondary winding e.m.f.s during normal or external fault conditions and that
the connections be such that e.m.f.s opposed each other thus causing no
current to flow in the interconnecting conductors (pilot wires) and relay wind-
ings. Because of this, the term 'voltage balance' was introduced to differen-
tiate such schemes from those introduced later, in which the current
transformers produce equal secondary currents during normal conditions.
These currents flow in the interconnecting conductors and zero current flows
in a relay connected between the interconnecting conductors as explained in
the introduction to this chapter. The term 'circulating-current' is used in
connection with these schemes.
a d
d
c
Fig. 5.2 Protective scheme proposed by C. H. Merz and B. Price. a, feeder; b, genera-
tor; c, substation; d, primary winding of CT; e, secondary winding of CT; t, earth or
return conductor; g, pilot wire; h, relay windings; i, circuit breakers; k, t, movable and
fixed relay contacts; m, circuit; n, battery; 0, electromagnetic device with armature p.
The advantages of the scheme proposed by Merz and Price were soon
recognized and it was applied to many circuits. This is clear from an article
published in the Electrical Review of 28 August 1908 [1] under the title 'The
Merz-Price system of automatic protection for high-tension circuits', in which
it was stated that about 200 miles of high-tension network in the North of
England were then protected by the scheme. It was further stated that Merz-
Price protective gear had been in use for several years and that numerous faults
had been isolated without visible shock on the system and, better still, without
opening healthy sections even though the switches were closed repeatedly on
to already matured faults. It was also stated that the scheme was in use to
protect transformers on the same scale.
In spite of these early comments, which indicated that correct operation
would be obtained under all conditions, difficulties did arise in later years as
line lengths and current levels increased. Maloperations evidently occurred on
occasions when protected units were healthy, due to several factors. Among
these was the mismatching of current transformer outputs caused by them
operating at different core permeabilities during healthy conditions. In an
endeavour to overcome this particular problem current transformers having
multi air-gapped cores, capable of performing linearly at primary current
levels up to 10000 A, were developed and introduced by 1925.
The introduction and use of special cables containing compensated pilot
wires by 1925 indicates that incorrect operation must also have been caused
by the currents which can flow in the capacitance between the interconnecting
conductors of balanced-voltage schemes. During external faults these conduc-
tors may have quite high voltages between them and their capacity current is
fed through the operating windings of the relays. This behaviour is considered
in more detail in section 5.2.4.
Over the years various methods have been developed to ensure that correct
operation will be obtained under all conditions, examples being the inclusion
of biasing proportional to the current flowing through a protected unit and to
the harmonics in power-transformer exciting currents.
5.2 FACTORS AFFECTING CURRENT-DIFFERENTIAL SCHEMES

The behaviour of the circulating-current differential scheme, described in the
introduction to this chapter, during three different conditions, was based on
ideal current transformers. In practice, however, there are several factors,
including the errors of practical transformers, which could cause incorrect
operation and these are examined separately, on a single-phase basis, in the
following sections.
5.2.1 Current transformer errors

Current transformers must be used in differential schemes to enable the input
and output currents of protected units to be compared and also to isolate the
Factors affecting current-differential schemes 149
relays and interconnecting conductors from the high voltages associated with
power systems.
In practice, all current transformers have transformation errors and therefore
it is desirable that those at the input and output ends of a protected zone should
be identical to try to ensure that their errors during healthy and external fault
conditions will always be the same. In these circumstances, the secondary
currents of balancing pairs of transformers would be equal at all times and
zero current would flow in the relays except when internal faults were present.
It must be recognized, however, that this behaviour might not be obtained even
when using transformers which are physically identical, because mismatching
can occur as a result of earlier internal faults. During such faults the currents
(/pA and IpB) at the two ends of a protected zone could both have large
magnitudes and be almost in antiphase with each other, as shown in Fig. 5.3.
In these circumstances, the current transformers in which they flow would
produce large e.m.f.s (ESA and E sB ), also in antiphase with each other, and these
would drive large currents (ISA and I'B) in the interconnecting cables. The
difference of these currents, which would be very large, would flow through
the relay, causing it to operate correctly to initiate the opening of the appro-
priate circuit-breakers. As a result, large residual fluxes could be left in the
cores of the current transformers at the two ends of the protected unit after
clearance of the fault, as explained in section 2.1 of Chapter 2.
These residual fluxes, which would be of opposite polarities, could persist
and cause the two current transformers to operate at different points on their
excitation characteristics during a later external fault and, indeed, they could
cause the core of one of the transformers to saturate whilst the other did not.
Current
transformer lpA
s N.p f -
- - -
It=/A,-/B
.
tr
_ Current
transformer
IPB
-sNpf
ROO X----{I====r-....:..L...-X-uVllV.l.R,..-__
Protected unit
Rc!2 Rc!2
Fig. 5.3 Conditions in a scheme with a matched pair of current transformers when an
internal fault is present. R" secondary winding resistance; Rc resistance of each con-
ductor.
As a result, sufficient unbalance current could flow in the relay to cause it to
operate incorrectly.
To minimize these effects, it is desirable that current transformers with
relatively low errors and cores with low flux retentivities should be used.
5.2.2 Current transformer secondary ratings

During external fault conditions, the current transformers have to provide
e.m.f.s to drive their secondary currents not only through their own windings
but also through the cables that interconnect them. In applications where the
two ends of a protected unit may be a considerable distance apart, the resist-
ances of the interconnections may be quite high and the voltage drops along
them may be correspondingly large when large fault currents are flowing. In
such circumstances, it is desirable that current transformers with relatively low
secondary ratings, e.g. 1 A rather than 5 A, be used. Although the lower rated
transformers have higher secondary winding resistances they can produce
greater e.m.f.s for given flux changes and the effects of the resistances of the
connecting cables are reduced, as is illustrated in the example below.
Example
In an application where the maximum current which may be fed to a fault
external to the protected zone is 15 times the rated current of the circuit, the
connecting cables have a total resistance of 6 Q. When using current trans-
formers with 5 A secondary windings, each of 1Q resistance, a balancing pair
would have to provide a total e.m.f. of 600 V, i.e. 15 x 5(2 + 6) V. Current
transformers with secondary windings rated at 1 A and each of resistance
25 Q would have to provide a total e.m.f. of 840 V, i.e. 15 x 1(50 + 6) V.
Because of the greater number of secondary turns on the lower rated trans-
formers, the flux variations in them would only be 0.28 of those in the 5 A
transformers, i.e. 1/5 x 840/600. Indeed, cores of a cross-sectional area equal
to 0.28 of that needed in the 5 A transformers could be used.
It will be seen from the above example that the e.m.f.s which must be
produced by the transformers do tend to rise as their secondary winding ratings
are reduced and the windings costs also rise significantly. The possible reduc-
tions in core area do, however, become very small as ratings are lowered
further and, in practice, the optimum situation is likely to be achieved using
transformers rated at 0.5 A or 1 A.
5.2.3 Interconnecting cables (pilot wires)

It will be apparent from the previous section that the resistances of the inter-
connecting cables affect the magnitudes of the e.m.f.s which must be provided
by current transformers during fault conditions, the effect being very consid-
erable when secondary windings of high rating (e.g. 5 A) are used. Although
the effect is smaller when lower ratings (e.g. I A) are used, it is nevertheless
desirable that the cable resistances should not be high compared with the
secondary winding resistances of the current transformers. Clearly the cost of
long cables with a large cross-sectional area might be unacceptably high and
a compromise must be effected in such applications.
5.2.4 Symmetry of protective circuits

Figure 5.4(a) shows the conditions which would obtain in a protective circuit
when an external fault is present.
If the circuit was symmetrical, i.e. RCA = ReB, and the current transformers
were perfectly matched, the secondary currents iSA and iSB would be equal at
all times. The current transformers would provide equal secondary winding
e.m.f.s, i.e. esA = esB = iSA(Rs + RCA)' The voltage (vIJ across the relay would be
zero and no current would flow in it, i.e. iR = O.
The above ideal behaviour cannot be obtained, however, when protective
circuits are asymmetric. For such a circuit, in which the conductor resistances
Current Current
transformer transformer -=-
End A s
m
~p -i
X___-t::===:J
f
I I
4
___-X G~~ s f ] ~nd B
Protected unit
(a)
ESB ; 330 V ------------ I
0.15 0.40
(b)
Fig. 5.4 Conditions in a protective circuit when an external fault is present.

RCA and RCB are unequal the following relationships would hold if the relay had
zero impedance, i.e. if the voltage across it were zero at all times:
eSA = iSA(Rs + RCA)
(5.1)
and
The relay current (i.J is equal to the difference between the transformer
secondary winding currents, i.e. (isA - i sB ), this being equal to the difference
in the two exciting currents needed to supply the e.m.f.s eSA and eSB. In the
limiting and impracticable case of cores with infinite permeability, the excit-
ing currents would be zero and no relay current would flow. In practice, with
cores having a finite permeability, a relay current would flow, its magnitude
clearly increasing with the decrease ofthe permeability. In addition, the relay
current would be greater, the greater the difference in the resistances of the
two sections of the circuit.
It should also be noted that relay current would flow even when the
resistances of the two sections of the circuit were equal, if the cores of the
transformers had different permeabilities. Clearly, therefore, any lack of
balance in the circuit will cause relay current to flow when the protected
unit is healthy, the greatest values occurring in the event of large external
faults.
Numerical examples based on the circuit shown in Fig. 5.4(a) are given
below to illustrate the above behaviour.
1. When an external fault current of 3000 A flowed in the primary windings
of current transformers with a ratio of 200/1, each of them would have
secondary currents (IsA, I sB) of approximately 15 A. If the circuit resistan-
ces had the following values, Rs =2 n, RCA = lOn, RcB = 20 n, and the
relay had zero impedance then the current transformers would have to
provide the following e.m.f.s:
Current transformer A: ESA = 15(5 + 5 + 2) = 180 V
Current transformer B: ESB = 15(10 + 10 + 2) = 330 V
Assuming that the current transformers had the excitation characteristic
shown in Fig. 5.4(b), their exciting currents, referred to secondary levels,
would be 0.4 A and 0.15 A. In these circumstances the relay current IR
would be approximately 0.25 A.
2. For the above conditions, if the relay was connected between the mid
points of the conductors so that RCA =RCB = 15 n, then the current trans-
formers would each have to produce an e.m.f. of 15(7.5 + 7.5 + 2) =
255 V. If, however, the current transformers had the excitation charac-
teristics shown in Fig. 5.5 then the difference in the exciting currents and
thus the relay current would be approximately 0.25 A.
'es (A)
Fig. 5.5 Current transformer excitation characteristics.
Electro-mechanical relays require input energy to cause them to operate and

therefore their windings possess finite impedance. As a result, the effects
caused by asymmetry in their associated circuits must differ from that de-
scribed above, but, as would be expected, the differences are small when
so-called low-impedance relays, i.e. current-operated relays with low opera-
ting VAs, are used.
As is shown later in section 5.6, improved performance can be obtained in
some applications by using high-impedance relays, i.e. voltage-operated re-
lays. In practice, these relays have very low operating currents, but in the
limiting case, for which the impedance would be infinite, zero current would
be required and indeed the current flow in them would be zero at all times.
Should such a relay be used in a current-differential scheme the current-
transformer secondary currents and exciting currents would be constrained to
be equal under all healthy and external fault conditions. In these circumstan-
ces, and for the first unbalanced condition considered earlier in connection
with Fig. 5.4, each current transformer would produce an e.m.f. of 255 V to
drive a secondary current of 15 A around the loop of resistance 34 n. As a result,
a voltage of 75 V, i.e. 255 - 15(5 + 5 + 2), would be present across the relay.
Should the resistance be balanced by connecting the relay between the
mid-points of the conductors (17 n on each side of the relay) but the trans-
formers again have the excitation characteristics shown in Fig. 5.5, then, with
equal exciting currents, the sum of their e.mJ.s would have to be approximately
510 V. In this instance, current transformers A and B would provide secondary
e.m.f.s (ESA and E SB ) of 280 and 230 V respectively and the relay at the
mid-point would have a voltage of 25 V across it, i.e. 280 - 15 x 17 V.
It will be clear from the above that quite large voltages may be present
between interconnecting conductors, particularly when large currents are
Current Current
transformer transformer
A B
End A s m
--. Np f X-----[===:J----,v
I I 'A
s o~~ f ~ B
Protected unit
Fig. 5.6 Effect of capacitance between interconnecting conductors. iR = 0 when iCA

= i CB'
flowing to faults outside a protected zone. As a result, significant currents

(icA and i cB ) will flow through the capacitance present between the conductors
as shown in Fig. 5.6. In a completely symmetrical circuit this will not cause
imbalance and zero current will be present in the relay. In an asymmetric
circuit, however, in which the relay is not connected between the mid-points
of the interconnecting cables, the capacitive currents flowing on the two
sides of the relay (icA and i cB ) will not be equal and their difference will flow in
the relay.
From the above, it is clear that the circuits of circulating current protective
schemes should be symmetrical whatever types of relays are used. In particu-
lar, identical current transformers are desirable at both ends of a protected unit
and the relay should be connected between the mid-points of the interconnect-
ing cables. In practice, it is not always possible to comply with these require-
ments because different manufacturers may be supplying the equipment at the
two ends of a protected zone and, in consequence, space or other consider-
ations may preclude the use of similar transformers. Similarly it may be
impractical to connect relays at a point midway along the run of the intercon-
necting cables. In the latter case, ballast resistors may be added in one section
of a scheme to equalize the resistances in the two sections, and while this
reduces the possible unbalance currents it does increase the e.m.f.s which the
current transformers must be able to provide.
5.2.5 The setting of low-impedance relays

Current-differential schemes can in principle provide very low fault settings
because they do not have to grade with any other schemes. There should
therefore be no difficulty in obtaining operation for all interphase faults within
a protected zone and indeed even quite highly-resistive faults to earth should
be detectable.
In addition, very rapid or so-called instantaneous operation should be attain-
able.
It must be realized, however, that the minimum fault current at which
operation will occur is not determined solely by the relay settings. This can
be illustrated by considering the single-phase arrangement shown in Fig. 5.7
in which the fault current Ir is fed from end A of the circuit.
If an electro-mechanical relay which required a 2 V A input to operate at its
setting of 0.2 A were used, the voltage drop across it at this level would be 10
V. This voltage, minus the probably insignificant voltage drop in the cable
resistance, would appear across the secondary winding of current transformer
B requiring an exciting current less to induce the necessary e.m.f. (Ess) (ap-
proximately 10 V) in the winding. The sum of the relay current (IR) and the
exciting current less would flow through the resistances of the secondary
winding of current transformer A (Rs) and the interconnecting cables (Rc).
Current transformer A would thus have to provide an e.m.f. ESA of approxi-
mately 12.4 V, i.e. VR + 0.2(5 + 5 + 2). The minimum fault current for which
operation would occur would thus be 200(0.2 + lesA + less), i.e. C.t. ratio
x (IR + lesA + less) and this could be significantly greater than 0.2 pu or 20%
unless the transformers had cores of low reluctance, i.e. cores with a large
cross-sectional area and high permeability.
It is quite probable that full-load current (1 pu) could be flowing in the
protected circuit when a highly-resistive earth fault is present. In such circum-
stances the fault and load conditions would be superimposed and higher volt-
age drops, transformer e.m.f.s and core fluxes would be present than those
considered above. As a result, because of the non-linearity of the core-excita-
tion characteristics, the increases in the exciting currents caused by the fault
might exceed those above and require a higher minimum fault current for
operation. Clearly the above effects should be taken into account when selecting
Current Current
transformer - transformer
A
End A s Np f
~A = "
-
Tt.-r
f1l - "J;,B = 0 B
s Np fEnd B
~;";";'--'oDOID00'--X----I===JIr-..:.-.L..-X 0 DO
200/1 - Protected unit 200/1
Rs=2il
s N f,
liSA
~ ~--I--r---~--~~-r--~
ESA
Fig. 5.7 Conditions in a symmetrical protective circuit when an internal fault is

present.
relay settings. In some cases, the minimum relay-operating current would
have to be significantly lower than that given by the product of the current
transformer secondary rating and the minimum per-unit fault current level at
which operation is required; for example, if transformers with rated secondary
currents of 1 A are to be used and faults of 0.2 pu are to be detected, then the
relay must be set to operate at a current of less than 0.2 A.
5.3 THE USE OF BIASING FEATURES

It will be appreciated from the preceding section that although current-
differential schemes are basically very simple, great care must be taken in
applying them, to ensure that they will operate correctly under all the condi-
tions to which they may be subjected.
It is very desirable that they should be capable of detecting internal faults
which cause low currents to flow, say down to 20% of the rated current of the
protected unit. They must not operate, however, during external faults of the
maximum current level which may occur, say 20 times the rated current, i.e.
2000%. To achieve such performance from the basic scheme would be diffi-
cult, because it would require the total imbalance resulting from mismatching
of the current transformers and unequal circuit resistances to be less than 1%.
Biasing features, which effectively increase the currents needed to operate
relays when high currents are flowing, were therefore introduced many years
ago to reduce the degree of matching needed. Biasing may be effected on an
electro-mechanical relay by including one or more restraining windings in
addition to the operating winding. Current flow in the restraining windings
produces force or torque opposing relay operation whilst current flow in the
operating winding produces force or torque in the direction needed to cause
operation.
Because of the desirability of having symmetry in differential schemes, it
is common practice to use relays with two restraining windings, these being
connected as shown in Fig. 5.8. During healthy conditions or when external
faults are present, restraint is required and therefore the forces or torques
produced by the two restraining windings should be in the same direction
when the currents in them (/SA and I sB ) are in phase with each other.
It is clearly desirable that relay performance should not be greatly impaired
when internal faults occur and the limiting case tends to be that of a highly-
resistive earth fault on a unit which nevertheless continues to supply its rated
current. This situation is illustrated in Fig. 5.9. Assuming, for simplicity, that
the fault current (It) of 0.2 pu (40 A) is in phase with the normal current (/pB )
of 1 pu (200 A), then one restraining winding would carry 1.2 A and the other
1 A whilst the operating winding would carry 0.2 A. For operation to be
obtained under this condition, the restraining force or torque would have to
be less than the operating force or torque and indeed it would normally be
very much less.
The use of biasing features 157
Current Current
transformer transformer _
A
End A s Np f -
'PA
Protected unit
4 ~
~B~
s Np f End B
---...w..-XI - - -
A
200/1
Restraing windings I
s Nfl I _______________
s ~~rl~~----~-----0~r+--~
fI
I
I
I
I
I I
- - - - - - - - ______ 1
Relay
Fig. 5.8 A biased protective scheme.
In practice the amount of bias or restraint applied is nonnally expressed in

percentage tenns. As an example, with 2% bias applied, the minimum current
needed for operation would increase by 0.02 A if both restraining windings
were carrying a current of 1 A, hence a relay which operated at 0.178 A with
zero restraint would operate at 0.20 A if rated current was flowing through the
protected circuit on which a 0.2 pu fault was present. This perfonnance is
illustrated in Fig. 5.9.
Small amounts of bias are very effective when high currents are flowing to
external faults. As an example, in the case considered above, 2% bias would
raise the relay operating current froIl) 0.178 A to 0.58 A if a current of 20 pu
were flowing to an external fault, thus considerably increasing the allowable
amounts of imbalance.
Current
transformer
IpA = 1.2 pu
-t I, = 0.2 pu
Current
transformer
B
~B_ = 1.0 pu
End A A - Protected unit End B
~X-----C====J-~~X~
1- - - - "Reiay - - - - I
I I
I I
1- __________ I
Fig. 5.9 Conditions in a biased protective scheme when an internal fault is present
(minimum operating current of the relay without bias 0.178 A, bias 2%).
Clearly similar performance can be achieved using a relay with only one
restraining winding, but to obtain the same percentage bias its restraining
winding must have twice the number of turns of each of the two restraining
windings considered above and there is little advantage to offset the asym-
metry introduced into the scheme.
5.4 IMPLEMENTATION OF SCHEMES

The behaviour of current-differential protective schemes has been outlined
above on a single-phase basis to simplify the treatment. Normally, however, such
schemes are used to protect sections of three-phase networks and the arrange-
ments adopted in such applications are considered in the following sections.
5.4.1 Units of short physical length

When units such as rotating machines are to be protected, the current trans-
formers at the two ends of a protected zone are only a relatively short distance
apart and each of the phases could be covered by independent schemes. It is
normal, however, to have four interconnecting conductors, one of them for-
ming a common neutral return, as shown in Fig. 5.lD. Three relays, each of
them having the same current setting, are needed and they are virtually inde-
End A
a
-X'"Protected unitrX- End B
b X'" rX
-c X'" rX ,....-
s Np f s N. f s o~~ f
s o~~ f s o~~! --..uL- Current transfrmers
"" ~~f J'" .!~~f~
s 000 f r-'GGGT - GGG T - 000 i s 000 f s 000 f
Ns s Ns s Ns s Ns Ns Ns
Relays
r--------__
4~~~_~IlH
I
I
I
I
I
I J
;.., ;~
i1
I
I
I
I
________ J
Fig. 5.10 Differential protection of a three-phase unit of short physical length.

Implementation of schemes 159
pendent of each other. In the event of an internal single-phase to earth fault
only one of the relays will operate, whereas for interphase faults two or all
three relays will operate.
An advantage of this arrangement over one with three independent phases
is that balanced components of the phase currents, i.e. positive- and negative-
sequence components, do not flow in the fourth, or neutral, interconnecting
conductor and thus the e.m.f. to be provided by each of the current trans-
formers is reduced, there being a voltage drop on only one conductor instead
of two.
5.4.2 Units of considerable physical length
Implementation of the basic current-differential scheme to protect units, such

as distribution lines or cables, would often be too complex or expensive. As
examples, the cost of providing four relatively long, interconnecting conduc-
tors to enable comparisons to be effected on each of the phases could be
unacceptably high and the installation of relays at the mid-point of a unit could
be impractical. These factors and modifications to the basic scheme which
have been introduced to enable it to be used for these applications are con-
sidered below.
Reduction of the number of interconnecting conductors
It is clearly desirable that schemes should only require two rather than four
interconnecting conductors when the ends of a zone are a considerable dis-
tance apart. Independence of the phase comparisons cannot, however, be
retained using two conductors, because only one quantity can then be con-
veyed between the ends of a zone. This single quantity must be produced from
the three phase currents at one end of a protected zone for comparison with a
corresponding quantity produced at the other end. Such a quantity could be
obtained by feeding the output currents of the three transformers to a network
which would provide an output related to one or more of the sequence com-
ponents of the input currents. Because positive-sequence components are pres-
ent during all faults, signals derived from only this component could be used.
During single-phase to earth faults, however, which occur much more fre-
quently in practice than other types of faults, the positive-sequence component
is only one third of the magnitude of the actual fault current and therefore the
relays would have to be given very sensitive settings. As an example, to detect
a fault current of 20% of the rated current of a protected circuit, the relays
would have to be set to operate at 67 rnA if current transformers with second-
ary windings rated at I A were used. This would increase the difficulty of
meeting the requirement that relay operation should not occur when external
faults, including large interphase faults, are present. A similar situation would
arise if a signal related to the negative-sequence component of the currents
were used and clearly a signal related to the zero-sequence component alone
could not be used as it would not enable interphase faults clear of earth to be
detected. In practice, therefore, a signal produced from two or three of the
components must be used.
While such techniques have been used in some protective schemes, se-
quence networks tend to be somewhat complex because phase shifting is
necessary to c!etermin~ the 'a' phase positive- and negative-sequence current
components (/Ia and Iza) as can be seen from the following equations:
II a=1(fa + dib + dIe)

and
in which a is the operator 112n/3 and la, Ib and Ie are the phase currents.
In addition, the above equations are only valid when the currents are sinu-
soidal and therefore not applicable during the periods when transient compo-
nents are present.
Simpler alternatives, which are suitable for all current waveshapes, have
been widely used in current-differential schemes.
The simplest arrangement, which has been employed since the earliest years
of the century, is to sum the outputs of transformers having different ratios in
the three phases, say 300/1.25, 300/1 and 300/0.8 in phases 'a', 'b' and 'c'
respectively, i.e. 240/1, 300/1 and 375/1.
As shown in Fig. 5.11, this would give the following output signals for
primary currents of 300 A of each of the sequences.
Positive-sequence = (0.35 - jO.173)A or 0.391- 0.46rad A

output relative to the 'a' phase primary current
Negative-sequence = (0.35 + jO.173)A or 0.3910.46 rad A
output relative to the 'a' phase primary current
Zero-sequence = 3.05 A in phase with the primary currents
output
It is evident that this arrangement is much more sensitive to zero-sequence

currents than to those of the other sequences and as a result it gives larger
output signals for earth faults than for interphase faults, thus making it easier
to obtain the necessary stability during external faults. With transformers of
the ratios considered above, a signal of 0.39 A would be produced for a
balanced three-phase fault of 300 A, whereas a signal of 1.25 A would be
produced for an earth fault of 300 A on phase 'a'. It should be realized,
however, that the behaviour obtained is phase dependent, the outputs for earth
faults of 300 A on phases 'b' and 'c' in this instance being 1.0 A and 0.8 A
respectively.
21t13 11d = 300 A 21t13 12a= 1.25 A
2" ~-125A
12b=1A
- - - 039A
12a + 12~+
12c .
0.46 rad
21t13 11a = 300 A - ---
0.39 A
12c= 0.8 A
Zero - - . . 12a = 1.25 A

sequence -/2b=1A - - -
- /2c =0.8A 12a + 12b + 12c = 3.5 A
(b)
Fig. 5.11 Output signals obtained when using current transformers of different ratios.
Another alternative, which has been used widely for many years, is to feed
the outputs of current transformers, of the same ratio in each of the three
phases, to a summation transformer arranged as shown in Fig. 5.12. Assuming
...----X----L..__----'t-- - - - -
r+--I---X____ll...._ _---'t-- - - --
r+~-+-f-~-+--X____ll...._ __Jt-- - - - -
...... DDD T sOoo f s D~O tl-/sa
_I. -
s N. N.
Nab So
I.
liE
b ~
Noc
_/sc c~
Ncn Summation
~
f C
transformer
Fig. 5.12 Current transformers feeding a summation transformer.

ideal transformation, the output current of the summation transformer would
be given by:
is =--.L {(Nab + Nbc + NcrJisa + (Nbc + NcrJisb + NCR i2c }

Ns
It will be recognized that this behaviour is the same as that obtained from
three current transformers with different ratios. By making the turns NCR
greater than Nab and Nbc the sensitivity to earth faults may be made greater
than that to interphase faults. Tappings are usually provided on the primary
windings of summation transformers to enable the sensitivities to be varied to
suit particular applications.
The positions and connections of relays

As stated earlier, it is necessary that the circuits of current-differential
schemes should be as symmetrical as possible to assist in achieving balance
during external fault conditions. Ideally, therefore, relays should be connected
at the mid-points of units. In many cases this is impractical because the ends
of a unit will be at different substations and the provision of housing for relays
at some other position could be expensive. In any event, if relays were so
positioned extra cables would have to be installed to enable the relays to trip
the circuit-breakers at the two ends of the protected zone.
To overcome these difficulties, relays could be installed at each end of a
unit, using either ofthe arrangements shown in Fig. 5.13. With the connection
represented on a single-phase basis in Fig. 5.13(a), i.e. with the relays effec-
tively in parallel, each relay would carry current during external fault or
healthy conditions and clearly these currents would increase with either de-
crease in the resistance of the relays or increase in the length or resistance of
the interconnecting conductors. Numerical values are shown on Fig. 5.l3(a)
to illustrate this situation. The voltage across the relays of 150 V produced by
an external fault current of 15 pu is very high and unlikely to be acceptable.
The arrangement illustrated in Fig. 5.13(b), with the two relays connected
in series, is completely balanced and no current would tend to flow in the
relays during healthy conditions or when external faults were present. This
arrangement, which requires an extra interconnecting conductor per phase,
increases the burden imposed on the current transformers, because they have
to provide extra e.m.f.s to drive current through two relays in series. It has,
nevertheless, been used quite widely. Again, biasing features are necessary to
prevent maloperation occurring because of asymmetries in the circuit.
5.4.3 Balanced-voltage schemes

The above schemes are described as 'circulating-current' schemes, because
currents flow around the loop formed by the interconnecting conductors dur-
Current (pA =4500 A

End A transformer -
---,~ ~,r-X----~Ic====rl----
. . ~ Protected unit
300/1
s N.
S
f.
I'SA = 15A
RJ2=5o.
L-1--r--t-----.---~--_r~~
t
V=o
RJ2 =50. RJ2 = 50.

(a)
Current
transformer (pA = 4500 A
End A A -X
---,~ N~
. ~
,r- ---';:I===Jl--~
Protected unit
300/1
S N.
s
f.
I'sA = 15 A
RJ2=5o.
L-.--r--t-----~--~--_r~~
V= 75 V V= 0
V= 0 R 2 =50. V= 75 V
(b)
Fig. 5.13 Alternative relay connections. (a) parallel; (b) series. Rc is the resistance of
each conductor.
ing healthy and external fault conditions. Such behaviour tends to require that
the e.m.f.s which must be produced by the current transformers become very
high for applications where the ends of protected zones are quite long distan-
ces apart, i.e. several miles, and therefore alternative schemes based on the
original proposal by Merz and Price and designated as 'balanced voltage' or
'opposed Voltage' are preferred for such applications.
A basic scheme of this class is shown in single-phase form in Fig 5.14(a),
from which it will be seen that the secondary windings of the current trans-
formers at the two ends of a unit are connected, via interconnecting conduc-
tors, in opposition to each other. In the event of a fault within the protected
zone, both the primary currents and the e.m.f.s induced in the secondary
windings of the transformers will be unequal and current will circulate through
the interconnecting conductors. To enable the circuit-breakers at the two ends
of the protected zone to be opened when this condition occurs, two relays,
with their operating windings connected in series with the interconnecting
conductors, are included. When the protected unit is healthy, however, and
current is flowing through it, the current transformers will each have the same
primary current and will produce e.m.f.s which oppose each other. Provided
that the transformers are identical and performing at the same points on their
excitation characteristics, they will produce equal instantaneous e.m.f.s, as
shown in Fig. 5.14(b), and current will not circulate through the secondary
windings and interconnecting conductors. It will be clear that the voltage
between the two interconnecting conductors will then be equal to the e.m.f.
provided by each current transformer and this will cause a capacitive current
to flow between the conductors, its magnitude increasing with the length of
the conductors. Half this current will be fed from each end of the circuit
through the operating windings of the relays and unlike the situation which
occurs with circulating-current schemes, the relay currents cannot be reduced
to zero by making the circuit symmetrical. -
As in the schemes described earlier, the relays may be set to operate for
internal faults with current levels below the rated value of the circuit. Again,
however, relay operation must not occur during external faults up to the
highest level possible on the protected network. To satisfy this requirement
Current . ..
transformer IpA = If + IpS
.-r'
41L IpS
Current
transformer
~~~
~~ fr-:x:------[I:::::JI--~--L-~~~~
Protected unit ~
~~f~
"'-mryA ;,. R'OlaYB .m-~
~ esA*esB
(a)
r
ipA ipS = ipA
rn-
E!2~..t N~ fr-:X:
~ _ ~
I
Protected unit
I ~~~~
N~ f~
esA=
~ -Nr-Y-A------------R..,Cr ~
(b)
ipA = 4+ ipS 1[ ~
4 ipB
~~~
N~ fr-:X: I 1-- EndB
Protected unit
~
'),Operating 1
Restraining~ 1winding 1
winding 1 ___ I 1
(c)
Fig. 5.14 The behaviour of balanced-voltage schemes.

End A Protected unit End B

X~ ~X------------~
a
r+--+--X~ ~X------....,
b
Restraining winding
transformer Operating winding
Fig. 5.15 Balanced-voltage scheme applied to a three-phase unit.
when low relay settings are used it is essential that the current transformers
be well matched. In addition, biasing must be employed, this being achieved
by connecting the restraining windings of each of the relays between the
interconnecting conductors at their respective ends of the protected zone. This
arrangement is shown in Fig. 5.14(c) and it will be realized that the voltages
across the restraining windings will be high when large currents are flowing
through protected units to faults outside their zones. The percentage of bias
applied must be sufficient to ensure that maloperation will not be caused by
imbalances in the protective circuit and current transformers and by the capa-
citive currents between the interconnecting conductors. In practice the per-
centage of bias may be varied using tappings on the restraining windings or
by varying the impedances of the restraining-winding circuits.
Balanced-voltage schemes may be implemented for three-phase applications
in similar ways to those used with the circulating-current schemes described
earlier. Summation-transformers are widely used, a typical arrangement which
only requires two interconnecting conductors being shown in Fig. 5.15.
5.4.4 Schemes to protect zones with more than two ends

In the previous sections, schemes suitable for protecting zones with only two
ends have been considered. In such applications, pairs of current transformers
which carry the same primary current under healthy or external fault condi-
tions are required to balance with each other, e.g. the two transformers asso-
ciated with phase a should ideally produce identical secondary currents. In
e
Current
trans~rmer End B
Current
transformer
.
J!.
x-------, F ---;+-
IpS
~
End A
. I A X Current
trans~rme ~
W X
Fig. 5.16 Circulating-current protective scheme applied to a teed feeder.
such circumstances, the e.m.f.s to be produced and the flux variations should
be the same in a balancing pair of transformers.
These conditions do not apply, however, when the outputs of more than two
transformers are required to balance. Such situations arise when teed feeders
or busbars are to be protected. Considering the teed feeder shown in Fig. 5.16,
it can be seen that the input current (ipA) must be equal to the sum of the output
currents (ipB and ipc) at every instant when the protected unit is healthy. Under
these conditions, the three current transformers would be producing different
e.m.f.s at any instant and therefore operating over different flux ranges. As a
result, zero current would only flow in the relay if the transformer cores were
infinitely permeable or if their excitation characteristics were linear and the
permeabilities of the cores were the same. Such conditions do not obtain in
practice and a degree of unbalance must always exist. Clearly the resulting
relay currents will be greater, the higher the currents flowing in the feeder.
The behaviour is, however, dependent on the ratios of the currents flowing in
the sections of such a feeder and in practice the current levels for different
fault and operating conditions should be determined when such schemes are
to be applied. Two examples are considered below to indicate the situations
which can arise.
A teed feeder fed from a single source

Should an external fault occur just beyond one of the output ends of a teed
feeder, say at point F at end B in Fig. 5.16, and there was only a source at end
A, then the voltages at the output ends (Vb and vc) would tend to be depressed
and the load currents at those ends (/pB and Ipc) would probably be well below
the rated values. Currents well above rated levels would flow, however, in the
primary windings of two of the current transformers (A and B) and these
currents would have similar magnitudes and phases. Tlfe primary current of
the third transfonner would be low and in such circumstances it is probable
that the degree of unbalance would be small.
A teed feeder fed from two or more sources

In a highly-interconnected network, it is possible that all the ends of a teed
feeder would be connected to sources. In such circumstances, large currents
could be fed into two ends of a feeder in the event of a fault just beyond the
third end. The current leaving the third end which would be the sum of
the other two currents could require an e.m.f. in the secondary winding of the
current transfonner at its end, necessitating very high flux density levels in its
core whilst the other transfonners would be operating at much lower flux
density levels. Under such a condition significant non-linearity in the core
excitation characteristics could lead to a significant unbalance current flowing
in the relay.
5.4.5 The protection of busbars

Similar but even more extreme conditions than those described in the previous
section may arise in schemes used to protect busbars. In these applications,
several sources may feed a fault, external to the protected zone, one such
situation being illustrated in Fig. 5.17. To avoid maloperation during this fault
condition, balance would have to be achieved between the outputs of five
current transformers, one of which would have a primary current much greater
than those flowing in the others.
In these circumstances and also in teed-feeder applications, biasing must be
employed to ensure that stability will be maintained when large currents are
Busbar
I
X X
I I
X X X
s
N, Current
P transformers
f
Fig. 5.17 Scheme applied to a busbar.

being fed to faults external to the protected zone. Many different combinations
of currents may be present during such faults, two being shown in Fig. 5.18.
For the condition shown in Fig. 5.18(a) circuits 1 to 4 carry equal input
currents (5 pu) to feed a fault fed by circuit 5. If bias were obtained by
summing the input currents in circuits 1 to 4 and the current fed out of circuit
5, i.e. bias was proportional to 11 + 12 + 13 + 14 -15, then a bias proportional to
40 pu would be obtained.
If the same biasing arrangement was used for the fault condition shown in
Fig. 5.18(b), in which a fault was fed by circuit 1, circuit 1 would provide a
bias of - 20 pu. Because of the direction of its current flow, circuits 2 to 4
would each provide a bias proportional to 5 pu and circuit 5 would provide
- 5 pu. As a result the magnitude of the bias would only be proportional to 10
pu, i.e. (-20+5+5+5-5).
I I I
F3 F3F~fF3~1~~
Busbar I I
Current ~
transformers ;:J
I I
I I
(a)
1/1 = -20 pu I I
3~ t~3 t~3 t~3 ~I'

Busbar I I
Xt Xt/ 2 =5PU Xt/ 3 =5PU Xt/ 4 =5PU Xt/ s =5PU
I I, I I I I
I I I I I
(b)
Busbar Operating
_--I--_--I--_-.. . .- . . . - . . . - . . . ---,lindin g
Relay
Current
5 pu
L-+-~4-~~~~~+~40 pu tI
(c)
Fig. 5.18 Current distribution during faults and a biasing arrangement.
Earth fault protective schemes 169
Two similar fault conditions would therefore produce very different levels
of bias. This situation is clearly undesirable and in practice the larger value of
40 pu could be produced for a fault on any of the circuits by rectifying each
of the circuit currents before summing them to obtain the biassing current.
This arrangement is shown in Fig. 5.18(c).
5.5 EARTH FAULT PROTECTIVE SCHEMES

It was shown in section 4.4 of the previous chapter that earth faults may be
detected by summing the currents flowing in the three phases of a circuit. In
many applications, however, it is necessary that tripping of appropriate cir-
cuit-breakers should only occur for earth faults within a certain zone. As an
example, the star-connected windings of power transformers are usually pro-
tected with sensitive earth fault protection. It is common to earth the star
points solidly or through resistors and in these cases four current transformers
connected as shown in Fig. 5.19 are employed. Such schemes, which have
been described as 'balanced earth fault' or 'restricted earth fault' protection,
should only operate for earth faults on the star-connected windings or the
connections between the current transformers. Their behaviour differs from
that of the schemes described earlier, in that balance must be maintained
between current transformers in the three phases and neutral when faults occur
outside the protected zone, rather than between two or more current trans-
formers in the same phase.
This feature of summing the phase currents in earth fault schemes is signi-
ficant. Under balanced conditions the phase currents have equal magnitUdes
and are displaced in phase from each other by 21t/3 rad. These currents, which
Line current
Power transformer
transformers
windings
s Np f ----I
a
-
b
ipc
c
+ipn
s Current
Relay ~p transformers
Fig. 5.19 Earth-fault protective scheme applied to the star-connected windings of a

power transformer.
are fed to the primary windings of the current transformers, sum to zero. The
excitation currents of the transformers will have fundamental components
similarly displaced from each other, but in addition there will be harmonic
components if the excitation characteristics are non-linear. Any third harmonic
components which are present will be in phase with each other and will thus
sum directly and flow in the relay. While this effect will normally be very
small at rated currents, it could be significant in the unlikely event of an
external three-phase short circuit.
As with other protective schemes, it is a requirement that relay operation
should not occur during healthy conditions or when faults external to the
protected zone are present and in addition it is desirable that the relay settings
are low enough to detect small fault currents. The latter feature is particularly
important when star-connected windings are to be protected as it determines
the percentage of the windings that can be protected. This can be seen from
Fig. 5.20, which shows a transformer winding with an earth fault on one phase.
The smaller the number of turns (Nc) between the star point and the fault
position the lower is the voltage available to drive the fault current (ft). If the
impedance of the winding is neglected, the fault current is given by:
NrVph
Ir=-- (5.2)
NpRe
in which Re is the resistance in the fault path and Np is the number of turns in
the phase winding.
The fault current (ft) is made up of ft. flowing through the faulted part of
the winding (Nr turns) and Ir2 flowing in the remainder of the winding
(Np - Nr turns), the latter current, which is needed to counterbalance the am-
pere-turns In Nr, returning via a neutral point associated with the source, i.e.
ft. Nr =Ir2 (Np - Nr) and ft =Ir. + Ir2
Fig. 5.20 Earth fault on a transformer winding.

Schemes employing high-impedance relays 171
The scheme would therefore sense the total fault current (h), h2 flowing in
one of the current transformers connected in the phases and IfI flowing in the
transformer in the neutral connection.
The simple expression for the fault current given in equation (5.2) clearly
shows that small currents will flow in the event of winding faults to earth near
the star point and therefore it is desirable that low fault settings be used.
Nevertheless it must be accepted that the windings cannot be completely
protected.
As an example, if a relay set to operate at 0.2 A were used and this protected
80% of the winding, then reducing the setting to 0.1 A would provide protec-
tion for faults on 90% of the turns. Such an increase in sensitivity would,
however, increase the difficulty of maintaining stability during external faults.
A further factor, to which reference has already been made, is that the
balancing current transformers may be carrying different currents during ex-
ternal faults and could thus be operating at different points on their excitation
characteristics. In such circumstances, the core of one of the current trans-
formers might be saturated for much of the time whilst the other transformer
cores might not saturate. As a result, large unbalance currents could flow in
the relay. Whilst such an extreme situation would probably not arise in bal-
anced earth fault protective schemes, it could occur in schemes associated with
bus bars because several circuits could then be feeding relatively low currents
to a circuit on which there is a fault.
In many cases, where the above conditions arise, the required stability
during external faults may not be obtainable in schemes employing low-
impedance relays which are set to provide high sensitivity to internal faults.
Much improved performance may be obtained in such situations by using
relays having high impedances, a practice introduced and proved over 40 years
ago.
5.6 SCHEMES EMPLOYING HIGH-IMPEDANCE RELAYS

The behaviour of these schemes is quite complex. The basic principles on
which they are based are simple, however, and these are illustrated in this
section, using the balanced-earth fault scheme shown in Fig. 5.21(a) as an
example.
To simplify the treatment, the following assumptions are made:
(a) that all four current transformers are identical;
(b) that the resistances of the connections to the secondary windings of the
transformers are the same;
(c) that the relay circuit has an infinite impedance.
The last assumption requires that the relay must operate when carrying zero
current and thus must be voltage operated, i.e. it must have a voltage setting
rather than a current setting.
Line current
transformers
s Np f
Neutral current
transformer
s
Np t3/0p
f
(a)
Li ne cu rrent
I = 41 N IN transformers
~esp sNpf
t
4= Ip Protected
- - - 1transformer
l _ _ _ _1
s Neutral
Np current
f transformer
(b)
Fig. 5.21 Conditions in a high-impedance protective scheme applied to a transformer

winding. (a) external fault; (b) internal fault.
With these protective schemes it is necessary that the voltage setting of the
relay be high enough to ensure that maloperation will not occur when faults
occur at points external to the protected zone. Clearly, the relay setting must
be determined by considering the most onerous condition that may be en-
countered in service. Although such a situation could not arise in practice, the
behaviour which would be obtained if high zero-sequence currents flowed as
Schemes employing high-impedance relays 173
a result of an external fault is considered below because this represents a
particularly onerous condition which may be readily examined and it is a
condition which is often used in proving tests in laboratories, because of its
simplicity.
For such a fault, the primary windings of the three current transformers in
the phases would each carry currents of the same magnitude and phase whilst
the primary winding of the neutral transformer would carry the sum of these
three currents as shown in Fig. 5.21(a). As a result, the neutral current trans-
former should, under ideal conditions, produce an e.m.f. three times that
produced in each of the other transformers. If it is further assumed that this
situation would in practice cause the core of the neutral current transformer
to be saturated throughout each half cycle, whilst the other transformers did
not saturate, the most unbalanced condition possible would exist. Under this
condition, the neutral current transformer would produce zero e.m.f. and
each of the transformers in the phases would produce the e.mJ. needed to
drive their secondary currents through their own individual circuits, i.e.
los Rs = lop Np Rs/Ns, and in addition to drive the sum of their currents through
the secondary circuit of the neutral transformer, i.e. 3Ios Rs =3Iop Np Rs/Ns.
It is clear from Fig. 5.21(a) that this latter voltage would also appear across
the relay circuit and that operation would not occur if the setting voltage
exceeded the above value. It will be appreciated that the limiting conditions
assumed above will not obtain in practice. For example, the neutral trans-
former will produce some e.m.f. and not only zero-sequence currents will
flow, and therefore a factor of safety would exist if the above relay settings
were used.
Having determined the minimum relay-setting voltage needed to maintain
stability, the current needed to cause operation during internal faults must be
determined to ensure that it is sufficiently sensitive. When high-impedance
relays are used the fault setting is basically dependent on the excitation char-
acteristics of the current transformers, as can be seen from Fig. 5.21(b).
When the setting voltage (Vs) is present across the relay it is also present
across the secondary windings of each of the current transformers, each of
which must then have an exciting current, referred to the secondary winding,
of Ie. In the event of an 'a' phase to earth fault, the current transformers in
phases 'b' and 'c' and in the neutral must each have secondary currents of Ie,
supplied by the 'a' phase current transformer, i.e. its secondary winding must
carry a current of 3Ie. Because it also needs an m.m.f. of Ie Ns, it will need a
primary current of 4/e Ns/Np to cause relay operation.
It is clear that satisfactory operation would thus be obtained if the proposed
transformers had an excitation characteristic where for Ie =Irmin Np/(4Ns) the
induced e.m.f. is equal to Vs, and the minimum fault current at which operation
is required is given by:
(5.3)
If the above setting is lower than that required, then the proposed current
transformers could be used but the relay setting voltage would have to be
increased so that higher exciting currents would flow. If, however, the setting
determined (IrmirJ is higher than that required then transformers with the lower
exciting currents needed would have to be used.
In practice electro-mechanical relays require a finite operating current (IR)
and therefore they cannot be of infinite impedance. As a result, the expression
for the minimum fault setting given in equation (5.1) must be modified to
include the current supplied to relay, as follows:
I4le +IR INs

Irmin= A (5.4)
Np
Whilst satisfactory performance can be achieved when the number of cur-
rent transformers is small, as in balanced earth fault schemes, it can be difficult
to obtain low fault settings when a relatively large number of transformers
form a group, as in busbar-zone schemes, because the fault setting is depend-
ent on the number of transformers (n), as can be seen from equation (5.5):
(5.5)
5.7 RELAYS USED IN CURRENT-DIFFERENTIAL SCHEMES

Various forms of electro-mechanical relays have been used over the years
since current-differential schemes were introduced.
In schemes of the basic form shown in Fig. 5.22, current circulates around
the interconnecting conductors during healthy conditions and when external
faults are present. Internal faults are detected by current imbalance and there-
Current
transformer ipA = ips Ips
End A A 7:---[1==::::Jlr--~X End B
~ Protected unit ~
Relay
Fig. 5.22 Conditions in a current-differential scheme incorporating a low-impedance

relay, during healthy conditions or when an external fault is present.
Relays used in current-differential schemes 175
Fig. 5.23 Hinged-armature relay.
fore ideally the relay should have zero impedance so that the current through
it tends to be proportional to the fault current.
Simple relays, with hinged armatures, of the form shown in Fig. 5.23 typi-
cally require 2 VA at the operating current and therefore a relay with a 0.2 A
setting would have an impedance of son, i.e. the operating voltage would be
lOY.
Whilst relays of this type are suitable for some applications, they cannot
present near zero or infinite impedances, i.e. they are not truly current or
voltage operated, and therefore over the years more complex relays capable
of operating with lower volt-ampere inputs have been produced. One such
relay, which is shown in Fig. 5.24, was designated as being 'rotary sensitive'
by its manufacturers, A. Reyrolle and Co Ltd. It is described in some detail
below to illustrate the constructional features which it incorporated to enable
the desired performance to be obtained. The armature, which was a vane
mounted on a vertical spindle, was able to rotate in its own plane. The spindle
had hardened pivots working in jewelled bearings, the lower bearing being
spring loaded to reduce the stress on the pivot. The field-system consisted of
two electromagnets in opposition so that when the operating windings were
energized, the resulting flux threaded the armature, causing it to rotate, against
the restraint of the control spring, into slots in the magnets. The ends of the
armature were so shaped that, as it moved, the driving torque increased more
rapidly than the restraining torque so ensuring positive operation at the mini-
mum operating current.
Torsion head _ _ _ _ _ __
Inertia discs -------r-.

Fixed contact -----rOlllllr
11I1~~~_ _ _ _ _ _ _ Moving contact
Flux path
~..,...... __ Armature
Coils
Fig. 5.24 Rotary-sensitive relay.
Rotation of the annature and of the spindle closed sensitive leaf contacts,
the contact travel being limited by two adjustable stops which, acting on both
sides of the annature, balanced the operating shock and reduced the stress on
the pivots when large currents relative to the operating value flowed in the
windings during internal fault conditions. To relieve the leaf contacts of the
full tripping-current duty, an auxiliary d.c. operated contactor was included.
Because of its construction, this relay required an input of only 30 m VA at
its minimum setting.
Biasing features were possible by including an extra electromagnet which
produced flux opposing rotation of the armature in the operating direction.
This electromagnet had either a single tapped winding to allow the percentage
of bias needed to be selected or a number of windings could be provided to
allow biasing from several circuits to be effected. The burden of the restraining
windings was lower than 0.5 VA.
Further reading 177
In addition weights could be placed on a pan fitted to the spindle to increase
the mass of the movement and thereby slow the relay operation slightly to
ensure that it would be less liable to maloperate when high surges of current
were experienced.
Relays of this type have been widely used in both balanced-voltage and
circulating-current schemes.
Simple hinged-armature relays are used in high-impedance schemes, such
as balanced earth fault protection. As stated earlier, such relays must operate
at low current levels and therefore have windings of many turns. They there-
fore have quite a large impedance. An impedor (usually resistive) is connected
in series with the winding to set the operating voltage level to that needed to
maintain stability during external faults.
Modern electronically-based relays capable of operating at either particular
current or voltage levels may also be used in current-differential schemes and
because such relays have external power sources they may be arranged to have
imput impedances of any desired values including zero and infinity.
5.8 APPLICATION OF CURRENT-DIFFERENTIAL SCHEMES

Because of their ability to discriminate between faults in a particular zone of
a power system and those external to it, without relying on time delays, these
schemes are widely applied.
The conditions they encounter in practice are dependent on the equipment
or unit they protect. As an example, high exciting current surges may flow in
power transformers for considerable periods after they are energized. In con-
sequence, schemes of varying forms are used, as indicated earlier in this
chapter, which has been confined to considerations of features and behaviour.
More detailed information on the schemes used for particular applications
is provided in the following chapters.
REFERENCE
1. Electrical Review, 28 August 1908, The Merz-Price system of automatic protection
for high tension circuits.
FURTHER READING
GEC Measurements, Protective Relays Application Guide (3rd edn), 1987, Chapter to.
Power System Protection, Vol 3, edited by The Electricity Council, 1981, (2nd
edn), Peter Peregrinus Ltd.
Royle, J. B. (1990) Differential relay concepts and applications, Canadian Electrical
Association - Transaction of Electrical and Operating Division, Vol 29, Paper
No.90-SP-157.
Further references are included in chapters dealing with applications.
6
The protection of transformers
It was stated in Chapter 2 that power transformers were first included in power
supply systems in the last decade of the nineteenth century to enable electricity
to be distributed from central stations at relatively high voltages to consumers
spread over quite wide areas. This practice, which proved to be economically
sound, has been continued around the world.
As the demand for electricity has grown, large generating stations have been
built, each containing a few large three-phase alternators of high volt-ampere
ratings, a typical value being 600 MV A. Because the output windings of these
machines are fitted in slots in the steel stators, it is not practical to insulate
their conductors to withstand the voltage levels which are used to obtain high
efficiency transmission over long distances. Consequently, alternators are now
wound to generate voltages of the order of 25 kV (line) and these are stepped
up to transmission levels, such as 400 kV (line), by a transformer directly
connected to each alternator, the combination forming a generator-trans-
former unit. Clearly the volt-ampere rating of a transformer must be the same
as that of its associated alternator.
A so-called unit transformer is also connected to each alternator to step
down the machine voltage and provide supplies to the auxiliary plant associ-
ated with the driving turbine and boiler or hydro plant. The electrical layout
of a typical generator-transformer unit is thus of the form shown in Fig. 6.1.
In recent years relatively small gas-turbine driven alternators and machines
in combined heat and power plants providing relatively low voltage outputs
have been commissioned and again they are connected to large high voltage
networks or loads via transformers.
Transformers of various sizes and ratings are provided to step down voltages
from transmission to distribution levels, e.g. 400 kVj132 kV and further trans-
formers are used to provide the voltage levels used by consumers (e.g. 415 V
(line) for domestic supplies).
Many other transformers, both three-phase and single-phase, are used within
power systems.
The behaviour of transformers and methods of protecting them are examined
in this chapter after the following section in which historical information is
provided.

180 The protection of transformers
Alternator Main transformer ~B b

)---..--...;0 ~X us ar
Circuit
breaker
~transformer
X
---L..- Auxiliary service
Fig. 6.1 Alternator-transfonner unit.

As stated earlier, power transformers were being quite widely used in electri-
cal power systems in the last decade of the nineteenth century. Initially, fuses
were the only form of protection available. Satisfactory circuit-breakers were
produced shortly afterwards, enabling protective relays to be applied to trans-
formers. Induction-type relays with inverse time/current characteristics were
being manufactured by Brown Boveri, as stated in Chapter 4 and C. H. Merz
and B. Price patented the current-differential scheme named after them, in
1904. They stated that the scheme could be applied to both transmission and
distribution lines and also transformers. IDMT relays and differential schemes
have been applied to relatively large transformers since that time and a later
very significant development was the introduction of harmonic biasing in 1938
to ensure that differential schemes would not maloperate during the exciting-
current surges which may flow when transformers are energized.
Earlier, L. M. Buchholz had developed relays which detect the production
of gases and oil vapour and also oil surges produced by large arcs resulting
from major breakdowns of insulation. These relays, which were named after
their inventor, have been fitted to most large oil-filled transformers since that
time and they are still being applied.
6.2 THE CONSTRUCTION AND BEHAVIOUR OF TRANSFORMERS

As stated earlier, many transformers, ranging from small single-phase units to
extremely large, three-phase units associated with alternators and transmission
lines, are in use in power-supply networks around the world. They all need to
be protected, the complexity of the equipment used for this purpose being
related to the size and importance of the individual transformers. In each
application, the protective equipment should ensure that a transformer is not
allowed to remain in service when abnormal conditions exist, either within it
or external to it, which may cause it to suffer consequential damage, but it
should not be taken out of service unnecessarily. As an example, a fault on
one of the circuits supplied by a transformer should be cleared by opening the
Construction and behaviour of transformers 181
appropriate load circuit-breaker, rather than by disconnecting the transformer.
To ensure, when selecting protective equipment for particular applications,
that the above aims will be achieved, it is necessary that the construction of
the transformers and their behaviour during both normal and abnormal condi-
tions, be studied. In the following sections these factors are examined in a
general manner.
6.2.1 Construction
All transformers whether they be required to step-up or step-down voltage
levels, operate on the principle enunciated by Michael Faraday, namely that
an e.m.f. is induced in a winding when the magnetic flux linking with it
changes, i.e. the well-known relationship:
e=-N dq>/dt (6.1)
By having two windings with different numbers of turns (Np and Ns), both
linking with the same changing flux (q, two different e.m.f.s may be pro-
duced, i.e.
ep = - Np dq>/dt and e s = - Ns dq>/dt.
Physical dimensions
In practice the windings are usually mounted concentrically around a limb of
a core made up of laminations of a magnetic material. Such materials may not
be operated satisfactorily at flux densities above about 1.7-1.8 T. As a result,
cores of large cross-sectional areas and windings with large numbers of turns
are needed in transformers which are to be connected to high voltage trans-
mission networks. This is illustrated below:
Equation (6.1) may be rewritten as e=-NAcdB/dt in which Ac is the
cross-sectional area of a core and B is the flux density in it. For sinusoidal
conditions the flux density must be of the form:
B = Bpk sin (2nft)
allowing an e.mJ. e =- 2nfNAc Bpk cos (2nft) to be produced.
The r.m.s. value of this e.m.f. is:
E = ...J2nfNAc Bpk V
A transformer to operate on a three phase, 400 kV, 50 Hz network must

therefore have a NAc product for each of its star-connected high voltage
windings of
NA = 400x 103
C ...J3 x ...J2 x 50n x Bpk
i.e. NAc = 611.5 if Bpk is to be limited to 1.7 T.
To convey the significance of this value, a transfonner with a core of 0.5 m
square cross-section, i.e. Ac =0.25 m2, would require a high voltage winding
of 2446 turns. Should the transfonner be rated at 600 MV A, the high voltage
winding would need to have a conductor capable of carrying a current of 866
A. Clearly such a winding together with its insulation would be very large, as
also would be the lower voltage winding associated with it. As a result, the
magnetic core would not only be of large area but the flux path would be quite
long.
Winding connections and core configurations

Small single-phase transformers usually have two or more windings, i.e. pri-
mary, secondary, tertiary, etc., mounted concentrically around one of the limbs
of their magnetic cores, these being of either the core type or shell type which
are shown in Fig. 6.2(a) and (b).
Three-phase transfonners usually have three-limb cores of the fonn shown
in Fig. 6.2(c), windings being mounted over each of the limbs. This is a very
economical arrangement as it reduces the volume of magnetic material to the
lowest possible level. It will be appreciated, however, that some magnetic flux
must exist in non-iron paths if the sum of the three-limb fluxes in such a core
is not zero. This condition will not arise when the three-phase voltages are
balanced but it can be present during faults which cause zero-sequence com-
ponents to be produced.
As pointed out above, transformers of high ratings are very large and in
some cases three-phase units of the very highest ratings would be impractic-
ably large. Their weights and bulk would be too great to allow them to be
Phase b Phase c
/ /
(a) (b)
/ /
~, ~~ ~,....I J: J:...I
~~ ~~
Phase a
(c)
Fig. 6.2 Transfonner constructions. (a) core-type, (b) shell-type and (c) three-phase
transfonner with a three-limb core.
transported from the manufacturer's premises to the sites where they are to be
installed. In such cases, three separate single-phase transformers are produced
and installed.
The windings of three-phase transformers may be connected either in star
or delta, the arrangements adopted in particular applications being governed
by the conditions which will exist in the network in which they are to be used.
It is normal practice for windings which are to operate at very high voltages
to be connected in star. The star points are usually solidly earthed and, as a
result, the voltages to earth at points on the windings increase linearly with
their distances from the star point. This enables the insulation to be graded along
the windings and minimizes the total amount of insulating material needed.
In many applications, lower voltage windings are connected in delta. This
practice is, for example, almost invariably adopted on the transformers directly
connected to the alternators in power stations. The output voltages of the
machines are typically of the order of 25 kV (line) and the machine star points
are earthed. It will be appreciated that no points are at earth potential on
delta-connected windings, the two ends of each phase being at full phase
voltage to earth and the centre points at half that value. This increases the
amount of insulation needed above that required on a star-connected winding
but at relatively low voltage levels the resulting extra cost is not great.
An advantage obtained by having delta-connected low voltage windings on
the step-up transformers is that any third or higher-order triplen harmonic
voltages produced by the alternators are not transmitted to the high voltage
network. A further advantage is that the zero-sequence current components
which will flow in the star-connected, high voltage windings in the event of
an earth fault on the transmission network will be counterbalanced by a
zero-sequence current circulating around the delta-connected, low voltage
windings and no current will flow in the neutral connection of the alternator.
These advantages are considered to be great enough to justify the extra insu-
lation costs referred to above.
Whilst the majority of single-phase and three-phase power transformers are
of the two-winding type, there are applications where the use of other types
may be advantageous. As examples, three-phase auto transformers are used to
interconnect high voltage transmission networks because they are cheaper than
equivalent two-winding transformers and in some cases three-phase, three-
winding transformers connected in star-delta-star are used.
Insulation and cooling

Many transformers and certainly all those of high ratings are mounted in tanks
or casings and the cores and windings are immersed in oil. The ingress of
moisture is prevented to ensure that the insulants will not be adversely af-
fected. The oil not only forms part of the overall insulation but also conducts
heat away from the cores and windings.
Tapping 3
RS3
Tapping 2
LS2
Tapping 1
LS1 ~~o--------1--------~
Centre-tapped inductor
Fig. 6.3 Tap-changing arrangement. Note: The situation shown occurs during the
change from tapping 1 to tapping 2.
Whilst modern transformers operate at very high efficiencies, the actual

power losses which occur in the winding conductors and the magnetic core
materials may nevertheless be very large. As an example, the power loss in a
transformer rated at 600 MVA with an efficiency of 99.6% is about 2 MW.
To dissipate such amounts of power without temperatures being raised to
unacceptable levels, radiators are fitted and equipment is provided to circulate
the oil.
Tap-changing equipment
Many transformers used in power networks are provided with tap-changing

equipment to enable their ratios to be changed over a range, typically 15%
of the nominal ratio, so that the voltages supplied to consumers may be
maintained within the statutory limits. To enable reasonably fine control to be
achieved sufficient tappings are provided on the windings to provide changes
of ratio in steps of 1.5%.
Supplies to loads must be maintained during the switching operations
needed to effect the ratio changes. One method of achieving this is to include
a centre-tapped inductor which is brought into circuit, as shown in Fig. 6.3,
during transition periods when connections are made to two adjacent tappings.
The tappings may be provided on either the low voltage or high voltage
windings, it being common practice to use the latter alternative because of the
lower currents to be carried by the switching equipment.
6.2.2 Operation during normal and external fault conditions

The voltage and current relationships which would be provided at every
instant by an ideal, single-phase, two-winding transformer are:
. Np
and ls =--. lp
Ns
Because the windings of practical transformers possess resistance and in-
ductive reactance, the latter, as a result of fluxes which do not link both
windings, and because there are power losses in cores and also that m.m.f.s
are needed to set up the necessary core fluxes, the above ideal transformations
are not achieved. The actual behaviour that will be obtained during any
particular condition can be determined by using the well-known equivalent
circuit shown in Fig. 6.4(a). The corresponding phasor diagram for steady state
sinusoidal conditions is shown in Fig. 6.4(b).
In steady state, the exciting currents (Ie) are small relative to the rated
currents, this being particularly the case with very large transformers; for
example, values of 1%, i.e. Ie = 0.01 pu, are typical. The equivalent circuit
components Re and Xm therefore have high values.
The series resistors (Rp and Rs) have very low values in highly efficient
transformers but the series inductive reactances (Xp and Xs) are usually signi-
ficant, particularly in large transformers, being made sufficiently large to limit
the currents which will flow during external faults to acceptable levels. It is
(a) Referred to primary level
\ ill (rad/s)
Fig. 6.4 Equivalent circuit and the steady-state phasor diagram of a two-winding
transformer.
~ i~X, + i7,X. '" 2iV. =0.2 P"
Fig. 6.5 Conditions when the output is at unity power factor. Note: The per-unit values
of the phasors are shown.
common for each of the reactances to have a value up to 0.1 pu (i.e.

Xp =Xs =0.1 pu) in which event the maximum possible current fed to an
external fault would not exceed 5 pu.
When the above conditions apply, the current transformation normally ap-
proaches the ideal value, i.e. is ,0, - ip Np/Ns but the voltage transformation is
usually more removed from the ideal value at any instant. For example, during
full-load conditions, a 0.2 pu voltage drop will be present across the inductive
reactances. When operation is at unity power factor this drop primarily causes
a phase displacement between the primary voltage and the reversed secondary
winding voltage as shown in !::ig. 6.5, i.e. Vp/Vs ,0, Np/Ns, the phase displace-
ment being approximately 21 Ip 1Xp radians, Ip and Xp being in pu values. A
phase displacement of about 0.2 rad would clearly be present with the reac-
tances values quoted above.
When operation is at relatively low power factors the voltage transformation
ratio may be significantly different to the turns ratio.
As stated in section 6.2.1, cores have to be worked over a wide flux-density
range in steady state, e.g. between 1.7 T, to minimize their sizes. As there
is considerable non-linearity in excitation characteristics over wide ranges,
steady state excitation currents are not pure sinusoids and contain quite large
odd-harmonic components.
The above comments have been concerned with normal steady state behav-
iour. There are a number of other conditions, however, which can occur and
which must be taken into account when considering the suitability of protec-
tive schemes for particular applications. They are considered separately in the
following sections.
Excitation-current surges
It has been stated above that many transformers are so designed that symme-
trical flux swings between levels approaching the saturation flux densities of
the core materials are present during steady state operation, to enable their
costs and sizes to be minimized. It must therefore be accepted that saturation
densities will be reached in the cores of such transformers during some normal
and abnormal operating conditions. Particularly severe saturation may occur
in periods after a transformer is connected to a source of supply, the degree
of saturation being dependent on the instant in the supply voltage cycle at
which the transformer is energized.
The flux-density variation (B) which would be required in the core of an
ideal loss-free transformer after energization is given by:
B = -1 - It Vp
V ppk
dt=--- [cosa- cos (rot + a)] + Bres (6.2)
NpAc 0 roNpAc
in which the supply voltage is vp = V ppk sin (rot + a)

Np = primary winding turns
Ac = cross-sectional area of the core
Bres = Residual flux density at time t =0
If the initial flux density (Bres) was zero and switching occurred at a voltage
zero, the peak flux density would be twice that of the peak value of the
sinusoidal component and indeed a multiple of three times could be ap-
proached if significant residual flux was present at the instant of energization.
The associated exciting current (ie), which would be of the form shown in
Fig. 6.6(a), would reach peak values of several times the rated current of the
transformer and it would continue to do so indefinitely. This current contains
a fundamental component, a quite large unidirectional component and also
harmonic components, the latter decreasing the higher their order. The passage
of the unidirectional current component through the resistance which must be
present in the primary windings of practical transformers causes a unidirec-
tional voltage drop and a corresponding component of opposite polarity in the
e.mJ. which must be induced in the winding. This causes the unidirectional
component in the flux-density variation to reduce from an initial value of
(
Vppk 1
roNp Ac + Bres
and eventually become zero. It must be recognized, however, that the rate of
reduction of this component is very low in large transformers with very high
efficiencies and consequently the durations of their exciting-current surges are
long. Figure 6.6(b) shows a typical exciting-current waveform.
The above treatment, which was based for simplicity on a single-phase
transformer, shows that the greatest current surges are experienced when
transformers are energized at an instant when the voltage is zero. The same
situation arises with three-phase transformers but with these there are six
voltage zeros per cycle, i.e. two on each phase, and therefore it is highly likely
that saturation associated with one of the phases will occur. A further factor
which must be recognized is that whilst the fluxes of the three phases will
2<\1pk ('.ie .,
._"'
-----r~/. ., /./-.i,
/ I .,
/ .I \
\.
\<\1
\
,,
I . '
\ ,
\
'-. "
(a)
(b)
Fig. 6.6 Conditions when a transformer is energized.
always sum to zero and may therefore circulate in the magnetic material of
the cores of three-limb transformers, the permeabilities of the limbs and yokes
will differ considerably at any given instant. The m.m.f.s provided by the
phase windings must clearly allow the required fluxes to be set up, but it must
be appreciated that the phases are not independent of each other, and as a result
the windings mounted on the limbs with higher permeabilities will contribute
m.m.f.s to supplement that provided by the winding on a deeply-saturated
limb.
Current relationships during healthy conditions
The steady state current relationships associated with healthy single-phase

transformers are very simple, the magnitudes being given by:
The phase relationships are dependent on the winding connections and

conventions, the primary and secondary winding currents being almost in
phase or antiphase with each other, depending on the polarities adopted.
The same relationships apply to the currents in the windings mounted on
the individual limbs of three-phase transformers, i.e. the primary and secondary
winding currents have magnitudes almost proportional to the ratios of the turns
in the windings in which they flow and they are almost in phase or antiphase
depending on the conventions adopted. The currents in the windings are the
'phase' and 'line' currents when windings are star-connected whereas in delta-
connected windings they are the 'phase' currents, the line currents being the
differences of pairs of phase currents.
In the case of a transformer with delta-connected low voltage windings and
star-connected high voltage windings, the ratio of the magnitudes of the line
currents (hp and hs) under steady state balanced conditions is given by:
in which Np and Ns are the numbers of turns in the low voltage and high voltage
windings respectively. In addition, pairs of low voltage and high voltage line
currents are displaced in phase from each other by 5rc/6 or -5rc/6 rad depending
on the connections of the windings. This situation is illustrated in Fig. 6.7(a,b).
Clearly the general relationships which apply under all conditions, except
when core saturation occurs, for the connections and conventions shown in
Fig. 6.7, are:
(6.3)
The comparisons made in differential protective schemes utilizing line cur-

rents should take the above factors into account and should be based on
equations (6.3). It is evident that balance must be maintained, during healthy
conditions, between three groups of currents, each group containing three
currents; for example, balance must be obtained between i pa, i sa, isb.
It is clearly vital that balance be maintained when faults occur on the
networks connected to transformers.
A common condition which may arise is that of a single-phase to earth fault
on the circuit fed by a star-connected winding of a transformer. If the other
winding is delta-connected, the resulting current distribution would be as
shown in Fig. 6.8(a). It can be seen that equal and opposite currents would
flow in two of the line conductors connected to the delta-connected windings,
a condition normally associated with an interphase fault. During this fault,
balance would have to be maintained in a differential protective scheme
separately between the fault current (isc) and two of the input line currents
(ipb and i pc ), i.e.
. Ns . . Ns .
Ipb AN Ise and Ipe A - Np Ise
p
In the event of an interphase fault on the circuit connected to the star-con-
nected winding of a transfonner with a delta-connected primary winding, the
input line currents would be in the ratio 2: - 1 :- 1, as shown in Fig. 6.8(b).
In these circumstances, balance would have to be maintained in a differential
scheme separately between each input line current and one or two secondary
currents, i.e .
. A Ns . 2Ns
Ipa - N (/sa - Isb) = N If;
P P
Ns
s
-+ -+
ipa isa
Ns
s
-+ -+
ipb isb
S
Ns
-+ -+
ipc isc
(a)
IscNr/Ns
\00 (rad/s)
-I+--r-,k----- Isa
(b)
Fig. 6.7 Conditions in a delta-star connected transformer and the steady-state phasor
diagram.
ipb = -ifNsINp -
s Np f
-iSbNs1Np = 0
s Ns
-
isa = 0
ipc = -
-ifNJNp
s Ns
- -
(a)
s Ns f

isa
ipa = 24 NsINp
tif
-
ipb = - 4Nsl Np
s Ns f

- iSb
-
ipc= -4NJNp
(b)
--2., o~so f
-
isc= 0
Fig. 6.8 Conditions when faults are present on the secondary side of a delta-star
connected transformer.
and
6.2.3 Behaviour during internal fault conditions

The majority of internal faults are associated with the windings. These may
result from a number of causes such as overloading for long periods or over-
heating of cores. The fault currents which may flow during winding faults are
dependent on the winding connections and earthing arrangements. These fac-
tors are considered separately below.
Earth faults on windings

In all cases, the current which will flow when an earth fault occurs will be
dependent on the voltage to earth at the fault position and the impedance in
the fault path.
-
Np
0 'TIT' 0
Ns Ip
~'
I
n 1;
OVs
Re
- -
Fig. 6.9 Conditions when an earth fault is present on a winding.
When a single-phase transformer is earthed through a relatively high resist-

ance at one end of one of its windings as shown in Fig. 6.9 then the current
which would flow in the event of an earth fault on that winding would be
almost proportional to the distance along the winding from the earthed end,
the transformer winding impedance being negligible relative to the earthing
resistance. The actual current would be given by:
. VsNe
le~--
NsRe
The current in the other winding to cancel the m.mJ. produced by the fault
current would be:
Ne v.NI
ip = Np . Ir.a. NpN. Re
Clearly the current in the healthy winding is more affected by the position
of a fault than that in the faulted winding. As an example, if an earthing
resistor was fitted to permit full-load current (1 pu) to flow for an earth fault
at the end of the winding remote from earth, then a fault current of 0.2 pu
would flow for a fault 20% along the winding (NelNs =0.2) but a current of
only 0.04 pu would flow in the healthy winding.
Obviously much higher currents flow in the event of earth faults on wind-
ings which are solidly earthed. Whilst the resistance of a section of a winding
through which a fault current flows is proportional to the turns affected (Ne),
the leakage reactance does not vary in such a simple manner, tending to vary
with the square of the number of turns affected and also the path through
which the leakage flux passes. The leakage reactance normally exceeds the
resistance and thus has the greater effect on the fault current. When necessary,
information on the variations of fault current and current in the healthy wind-
ing with fault position should be obtained from the transformer manufacturer
so that the suitability of protective schemes may be determined.
Very similar conditions obtain when earth faults occur on earthed star-
connected windings of three-phase transformers. As stated in section 6.2.2
(page 189), equal and opposite currents flow in two of the line connections if
the other winding is connected in delta.
Delta-connected windings which are fed from earthed sources or which feed
earthed loads have no points on them which are at earth potential and therefore
currents will flow in any earth faults which occur on them. Faults at the ends
of the phase windings will normally cause large currents to flow to earth, the
actual levels being dependent on the impedance in the earthing connections
and that of the source or load. The voltage to earth at the mid-point of a phase
winding is normally half the phase voltage and again quite large currents
should flow in the event of a fault in such a position. This fault level, which
is the lowest that may be experienced, during an earth fault on a delta-
connected winding may be calculated in any given case, using the appropriate
system parameters. Little difficulty should be experienced in setting protective
equipment to operate at levels which will detect these faults.
Interphase faults on windings

Interphase faults not involving earth clearly cannot occur on large three-phase
transformers formed of three separate units mounted in their own housings.
They may, of course, occur in three-phase transformers in which the windings
are mounted on a single core, but in practice they rarely do so.
Because the voltages which exist between phases are high, the currents
which will flow in the event of an interphase fault will usually tend to be large.
If, however, a fault did occur between two points on different phases of a
star-connected winding, each of them being near the star point, the voltage
difference normally present between the points could be low but nevertheless
quite a high fault current would flow because the leakage reactances of the
faulted parts of the windings would be low. In these circumstances, the m.m.f.s
produced by the fault current (ir) would be cancelled by currents flowing in
the faulted phase windings (ipa, ipb), as shown in Fig. 6.10. These currents
-
-
lpa
-
Ipb
0000000
Ipc
ipaNp - 4Nt .0. 0

ipbNp + i,Nt A 0
Fig. 6.10 Conditions when an interphase fault is present on a star-connected winding.

would approach zero as the fault positions approached the star point.
Similar conditions would clearly apply if interphase faults occurred on
delta-connected windings and again the currents in the connections to the
winding would be small if a short circuit occurred between two phases at
positions near the point of connection of the phases affected.
lnterturn faults
Steep-fronted voltage surges propagated along transmission lines to the termi-
nals of transformers may, as explained later in section 6.2.4 (page 197), over-
stress the insulation of the end turns of windings and cause interturn faults.
The low leakage reactances of such groups of faulted turns may permit large
currents to be driven through them by the power frequency e.m.f.s induced in
them. As in the cases considered earlier, the m.m.f. produced in the faulted
turns of a winding must be cancelled by one resulting from a current flowing
through the whole winding. Once again, this current will usually be small, its
magnitude decreasing as the number of faulted turns reduces.
Although the input currents to windings may be small when interturn faults
are present, the current flowing in the fault path will nevertheless be high, as
stated above, and as a result local heating will occur. If the condition were
allowed to persist, further damage to the winding insulation would occur and
a fault to earth would be established. This in turn could cause the core to be
damaged. Clearly, therefore, it is desirable that the presence of interturn faults
should be detected soon after their inception to keep the resultant damage and
the associated repair costs to a minimum.
lnterwinding faults
The windings of small single-phase transformers are usually concentrically
positioned or they may be mounted in several sections adjacent to each other
along the axial length of a limb of the core. Adequate insulation is provided
between the individual windings and therefore faults rarely occur between
them. When they do so, however, quite large currents may flow, because of
the large voltages which may exist between the fault points on the windings.
Interwinding faults may also occur between the windings on one limb of a
three-phase transformer. Very high voltages may be present between points
on the affected windings and therefore large currents could flow in such
circumstances. The likelihood of faults of this type is very low, however,
because there is always much insulating material between the windings, this
including spaces through which oil flows to cool the windings.
6.2.4 Causes of internal electrical faults

There are a number of operating conditions which can be permitted to continue
for limited periods but which will cause damage that will lead to breakdown
and electrical faults if they persist for long periods. These conditions, which
should ideally be detected before their effects become serious, are considered
below. There are also conditions which will cause faults to develop very
quickly.
Overloading
Transformers may be operated at currents above their rated values for limited
periods. The power losses in the winding conductors are proportional to the
square of the currents in them and therefore overcurrents cause the tempera-
tures of the conductors to rise to cause extra energy to be dissipated from them
to their surroundings. The rate of rise of temperature tends to be very low,
however, this being particularly so in large transformers and, as a result, low
levels of overloading may be permitted to continue for fairly long periods
without unacceptably high temperatures being reached. The actual periods
allowable are dependent on the designs of the transformers and the operating
conditions prior to overloading. Clearly if a transformer has been lightly
loaded for a long period before it is overloaded then the temperatures of the
parts within it will all be well below the limiting values when overloading
commences and it may then be allowed to continue for a long period before
the temperatures which can be produced under full-load conditions are
reached. If, however, a transformer has been operating at its rated output level
for several hours before it is overloaded then this new condition may only be
allowed to continue for a limited period.
In practice, the permissible operating temperatures and their durations are
dependent on the insulating materials which are used, because their lives are
shortened if they are heated to temperatures above certain levels and indeed
the deterioration is dependent on the periods of operation at high temperatures.
It will be appreciated, therefore, that regular overloading for long periods
will eventually lead to breakdowns within insulation, even at normal voltage
levels and that faults between windings, between turns or to earth may occur
as a result.
Guidance on permissible overloading should be obtained from transformer
manufacturers if it is not included in the original specification documents.
Overvoltage operation
When a transformer is operated in steady state at its rated frequency the flux
density in its core is proportional to the e.m.f.s which must be induced in its
windings and therefore to the voltages at its terminals.
Overvoltage operation of a transformer, because of the greater flux density
swings associated with it, causes increased eddy-current and hysteresis power
losses in the core material and therefore in tum it causes the temperatures of
the core and other parts to reach above normal levels. In addition, the winding
currents may also be increased by the overvoltages and certainly in large
transformers, which operate over large flux ranges during normal conditions,
the exciting currents will be relatively high and distorted because of the
measure of saturation which will be caused. In addition, flux may be diverted
away from the saturated sections of cores into iron paths in which the flux
densities are normally very low. For example, core clamping-bolts, around
which insulation is provided to ensure that the core laminations are isolated
from each other, have low flux densities in them under normal conditions
because of the high reluctance presented by the insulation. When core satura-
tion is present, however, the flux densities in the bolts may be much higher
leading to large power losses and heating in the bolts. If this condition was
allowed to persist, the bolt insulation could be damaged to such an extent that
currents might flow between the core laminations during subsequent normal
operation causing further overheating which could damage the windings.
Operation at frequencies above and below the nominal value

At a given operating voltage, the peak flux densities in steady state are
inversely proportional to the supply frequency and therefore, at below-normal
frequencies, some saturation of the cores will occur in transformers which
operate over large flux density ranges under normal conditions.
Generally the eddy-current losses in cores are not greatly affected by fre-
quency variations because these losses are proportional to the product of the
squares of the peak flux density and the frequency, i.e. (Bid 2 ), and they thus
tend to be constant for a given voltage as the frequency varies. Hysteresis
losses are proportional to the product of the area of the hysteresis loop and
the frequency. In practice, these losses tend to reduce as the supply frequency
falls because the areas of the loops do not tend to increase significantly at
densities above the saturation level. At higher frequencies, the increase due to
the increased frequency tends to be offset by the reduced areas of the loops at
the lower-peak flux densities which are reached at the rated voltage level.
Nevertheless, the saturation caused by low-frequency operation can cause over-
heating of components such as core bolts, as described in the previous section.
In general the ratio of the per-unit values of the operating voltage to fre-
quency should be maintained near unity to ensure that damage due to over-
heating will not occur.
Faults on systems connected to transformers

Short circuits on a network fed by a transformer can cause large currents to
flow in it, resulting in relatively rapid heating of its windings. To ensure that
the magnitudes of such currents are limited, transformers are designed to have
particular leakage reactances ranging up to about 0.2 pu. Should this be the
only impedance in the fault, then clearly the maximum current during a short
circuit would be at least 5 pu. In general the higher the rating of a trans fonner,
the greater is its per-unit leakage reactance. In addition, transformers are
designed to carry their maximum fault currents for specified periods without
being damaged by overheating. Relationships between these withstand periods
and the fault current levels and leakage reactances are included in documents
such as BS 171: 1936 [1]. As an example, a transfonner with a leakage
reactance of 0.05 pu would limit the level of external fault currents to 20 pu
and these could be withstood for 3 seconds. If necessary, such data should be
obtained from manufacturers.
High fault currents, in addition to causing overheating, can cause large
mechanical forces between winding conductors and stresses in other compo-
nents. These forces and stresses reach their highest levels when the currents
have their highest instantaneous values. Because these conditions occur during
the first cycle of fault current when maximum asymmetry may be present, and
cannot be prevented, transformers must be so designed that they will not be
damaged by them.
A further factor which must be recognized is that steep-fronted voltage
waves may be propagated along overhead lines as a result of lightning strikes
near them, or to them or because of flashovers of insulators. Should these
waves reach a transformer and be applied across one of its windings, they will
distribute non-linearly across the winding in a manner dependent on the inter-
turn and turn to earth capacitances it possesses. In practice, a very high
percentage of the voltage would appear across a few of the turns near the high
voltage end of the winding. Because of this, it is the usual practice to apply
extra insulation to these turns, but even so the arrival of steep-fronted waves
could cause interturn breakdowns. Whilst normal protective schemes would
detect such breakdowns they could not prevent them. As a further measure,
devices such as arcing horns and voltage limiters, which are outside the scope
of this book, are fitted to overhead lines, and although these will operate when
steep-fronted waves reach them, they nevertheless take a finite time to do so
and it must be accepted that the initial sections of such waves may be applied
across the windings of transformers.
Mechanical failures
Many transfonners are mounted in oil-filled tanks, the oil being needed to
maintain the insulating materials in good condition and also to conduct heat
from the windings. In large transfonners the oil usually circulates through
air-cooled radiator pipes and in many cases the circulation rate is raised to the
necessary level by pumps.
Clearly a reduction of the effectiveness of the cooling caused by the block-
age of pipes by sludge or by a pump failure would not affect the transfonner
perfonnance immediately but it would lead eventually to unacceptable heating
and damage to insulation.
Similar unacceptable conditions would arise if oil leaked, for example, from
a faulty drain valve.
6.3 THE APPLICATION OF PROTECTIVE SCHEMES AND DEVICES

TO POWER TRANSFORMERS WITH TWO OR MORE WINDINGS
PER PHASE
The protective equipment used with transformers is dependent on their import-
ance and ratings. Single-phase and three-phase transformers used at voltage
levels up to 33 kV and with ratings up to 5 MV A are often protected by
fuselinks which must in all cases be capable of carrying the maximum excit-
ing-current surges which may occur without operating. Guidance on such
applications was provided earlier in section 1.9.8 (page 00).
The other forms of protective equipment which are used with transformers
are examined in the following sections.
6.3.1 Inverse time/current relays

IDMT overcurrent and earth fault relays are used as the main form of protec-
tion on both the primary and secondary sides of relatively small power trans-
formers. As with fuses, the current and time settings of these relays must be
such that they will not operate when the maximum exciting current surges
flow. They must also be such that they will enable correct discrimination with
other protective equipment on associated networks to be achieved.
Relays of this type are also used to provide back-up protection for large
transformers.
6.3.2 Current-differential schemes

As stated in earlier sections, current-differential schemes have been used to
protect transformers since the early years of this century. They are now applied
to all large transformers.
The basic principle of current-differential protection is that the current
entering a circuit is equal to that leaving it under healthy conditions. To
implement this principle when protecting a single-phase two-winding trans-
former, it would be necessary to compare the currents entering and leaving
both the primary and secondary windings separately using four current trans-
formers and two relays.
To protect a three-phase, two-winding transformer in this way would require
the use of twelve current transformers and six relays, and should one of the
windings be connected in delta, its current transformers would have to be
positioned as shown in Fig. 6.11. With this arrangement, faults on the connec-
tions between the phases would not be detected and nor would faults on the
connections between the windings and the external circuit.
Protective schemes and devices 199
Current
transformer
Np sNppf
Fig. 6.11 The protection of individual windings.
To reduce the number of current transformers and relays required and also
provide more complete protection, a modified form of differential protection
is used in transformer applications, the principle employed being of m.m.f. or
ampere-turn balance. The basic scheme and factors which must be taken into
account when applying it to a single-phase transformer are examined in detail
below and then its application to three-phase transformers is considered in
later sections.
Basic scheme for use with single-phase two-winding transformers

Two current transformers must be provided, their primary windings being
connected in series with the connections to the high and low voltage wind-
ings of the transformer being protected, as shown in Fig. 6.12. Because the
currents in the power transformer windings will be different at all times, unless
its ratio is unity, the primary ratings of the current transformers must also be
different, as must their transformation ratios to enable balance to be main-
tained between their secondary currents during healthy conditions.
If the power transformer were ideal, it would require no exciting m.m.f. and
therefore at every instant the sum of the winding ampere turns would be zero,
i.e. ip Npp + is Nps = O. In this case the ratio of the current transformer ratios
would have to be the same as that of the power transfonner being protected,
i.e.
Npp Nsp
~:-N =Np:Ns or
"ps ss
In practice, this arrangement is adopted with power transformers even though

they are not ideal.
It was pointed out in Chapter 5 that current will flow in the relays of
current-differential schemes during nonnal conditions or when external faults
are present if the protective circuit is asymmetric. The modified scheme de-
scribed above will be similarly affected and it must be recognized that it will
also be affected by the factors considered below.
Mismatching of current transformers Because the current transfonners on

the two sides of a protected transformer must have different ratios, they cannot
be of identical designs. Quite often they must have different numbers of turns
in their secondary windings and their core dimensions cannot be the same
because of the numbers of turns to be accommodated on them and because of
the difference in the amounts of insulation needed on the primary windings.
As a consequence, the exciting m.m.f.s needed at each instant by the two
current transformers will usually be different, thus causing their trans-
fonnation errors to be unequal. The resulting differences in the secondary
currents will cause imbalance current to flow in the relay.
Mismatching caused by the power transformer steady state exciting current

Power transfonners are not ideal and do require exciting m.m.f.s to set up the
necessary fluxes in their cores. The basis of the modified differential scheme,
namely that the primary and secondary winding m.m.f.s sum to zero, i.e.
ip Np + is Ns = 0, does not therefore apply in practice. As a result, an imbalance
current (iR ), proportional to the power transfonner exciting current, will flow
in the relay at all times, i.e.
iR = ie Npp/Nps
In steady state, the exciting-currents of power transfonners are quite low,
this being particularly so for transfonners with high ratings, values of 0.02 pu
or less being usual. Using this value as an example, it can be seen that a relay
current of 20 rnA could flow in a relay employed in a scheme in which the
current transformers had secondary current ratings of 1 A.
This imbalance, which will be present even when a protected transfonner is
lightly loaded, cannot be counteracted by percentage-bias windings on the
relays.
Mismatching caused by power transformer exciting current surges Although

the magnitudes of the exciting currents of power transfonners are usually very
small during healthy steady state conditions, they may rise to values many
times the rated currents for significant periods after transformers are energ-
ized, as explained in section 6.2.2 (page 186). These currents, which are present
only in one winding of a transformer, cause large currents to flow in the
operating windings of the relays used in current-differential schemes and
would cause operation unless effective biasing arrangements were provided.
Their effect cannot be overcome by applying a relatively small percentage bias
and therefore a large amount of bias obtained from one or more quantities
which are only present in exciting currents is necessary.
Mismatching caused by tap-changing Tap-changing is effected by operating

switches to select a particular number of turns in either the primary or second-
ary winding of a transformer so that the desired output voltage is obtained and,
as stated in section 6.2.1 (page 184), the turns ratio of a transformer may thus
be varied in steps over a range.
To retain balance in current-differential schemes which compare the wind-
ing m.m.f.s, the current transformers associated with the tapped windings
should also have tapped windings and their ratios should be changed appro-
priately when tap-changes are effected. If this practice were to be adopted it
would be necessary to ensure that the secondary windings of the current
transformers would never become open circuited and also that the sections of
their windings being switched would not become short circuited. In addition,
the changes would have to be effected at the same instants as those on the
transformers being protected.
Because of the complexity and cost of such switching and the general
feeling that switches should not be included in current transformer circuits, it
is the practice to use current transformers with fixed ratios which will balance
when the transformers being protected are operating at the mid-points of their
tapping ranges. As a result, the flow of imbalance currents in the operating
windings of the relays must be accepted at other times. These may clearly
reach a percentage value equal to half the total tapping range of a protected
transformer.
Implementation of the scheme
The basic scheme, as indicated above, employs two current transformers with
their primary windings in series with the primary and secondary windings of
the protected transformer as shown in Fig. 6.12. The current transformer ratios
must be related as follows:
They must be so connected that the currents in their secondary windings

circulate around the loop they fonn with the interconnecting conductors when
the protected transfonner is healthy. The operating winding of the relay should
be connected between the mid-points of the interconnecting conductors to
form a symmetrical circuit.
Relays Rough balance protective schemes are applied to relatively small

transformers, where rapid operation in the event of faults is not essential.
These schemes, which balance m.mJ.s, incorporate standard IDMT relays
having current settings of 20-80%. Their time settings should be such that
they will not be operated by the exciting current surges which may flow into
the transformers protected by them and they should also grade satisfactorily
with other IDMT relays connected to their associated networks.
Schemes applied to large transformers must operate rapidly when faults
occur within their protected zones and therefore they must detect relatively
low fault currents and must not operate when their protected transformers are
healthy. The settings and biasing arrangements needed to provide the neces-
sary performance are considered below.
Settings Relays must have a minimum operating current which will ensure
that faults on most of each of the windings will be detected. Clearly the
percentages of the windings protected in the event of earth faults is dependent
on the method of earthing. When solid earthing is employed very high per-
centages are protected by a relay with a setting of 20%, i.e. 0.2 A when current
transformers with 1 A secondary ratings are used. It was shown, however, in
section 6.2.3 (page 191), that should an earthing resistor be used, which would
allow rated current to flow for an earth fault at the high voltage end of a
winding, then a fault at a point 20% along the winding from its earthed end
would cause a current of only 0.04 pu to flow in the other (healthy) winding.
As a result a relay setting of 4% (0.04 pu) would be required to ensure that
faults on 80% of the earthed winding would be detected.
Percentage-bias windings As stated in the previous chapter (section 5.3,

page 156), small percentages of bias are very effective in preventing incorrect
relay operations occurring as a result of mismatching when faults are present
at points outside a protected zone. Such biasing will certainly ensure that
maloperations do not occur because of mismatching resulting from current
transformer errors, protective-circuit asymmetry and the normal exciting cur-
rent of a protected transformer.
It is also advantageous to have two bias windings, as shown in Fig. 6.12, to
maintain symmetry.
Higher bias percentages are necessary when a transformer with tap-chang-
ing facilities is to be protected because significant imbalances will occur when
operation is taking place at the upper or lower ends of the tapping range. As
an example, if a transformer is operating at full load on a + 15% tapping then
Biassing windings
- -
I-------~
I
IS
I s i
Current ip I
s Current
transformer Npp I Nsp transformer
f I f
L _______ J
I
Relay
Fig. 6.12 Protective scheme based on m.mJ. or ampere-tum balance.
a current of 0.15 pu will flow in the relay, i.e. 0.15 A when current trans-
formers with secondary ratings of 1 A are used. In such circumstances, a 20%
relay setting (0.2 A) would not provide a significant margin to allow for other
possible mismatching.
Whilst higher relay settings could be used, this would have the serious
disadvantage of increasing the amount of the windings on which earth faults
would not be detected. The alternative of using higher amounts of percentage
bias is preferable.
Harmonic biasing As explained in section 6.2.2, an exciting current with

peak values of many times the rated current of a transformer may be present
during significant periods after it has been energized and, as a result, large
currents will flow during such periods in the operating winding of a relay used
in a scheme of the form shown in Fig. 6.12.
It was realized many years ago that this effect could not be counteracted by
using percentage-bias windings and methods of dealing with the situation were
sought.
One possible solution was to open the relay tripping circuit for a short period
each time a protected transformer was energized. During these periods reliance
would have had to be placed on other protective equipment operating should
a transformer be faulty. This was not considered to be an acceptable practice
and therefore novel biasing methods were examined.
In 1938, Kennedy and Hayward published a paper [2] in which they quoted
values, obtained experimentally, of the unidirectional, fundamental and har-
monic components present in typical internal fault currents and exciting cur-
rent surges. Because the values of the unidirectional, second, third and fifth
harmonic components in the exciting current surges were much higher than
those present during internal fault conditions, they developed a relay which
contained two filter circuits. One of these allowed only the fundamental
component of any imbalance current to flow in the operating winding of the
relay whilst the second filter was tuned to block the fundametnal component
of the current but permit all the other components to be fed, after rectification,
to a bias winding. The relay was tested and it was shown to perform satisfac-
torily.
Subsequently the effectiveness of the individual components in biasing
relays was studied further.
It was recognized that large slowly decaying unidirectional components are
present in exciting current surges but it was found that these components could
not be reproduced faithfully by current transformers because of the large
unidirectional changes of core flux required by them. It was realized that the
secondary winding currents of current transformers would fall to zero and
become of the opposite polarity after relatively short times when their primary
windings carried decaying unidirectional currents and that bias derived from
unidirectional components could thus reach very low values and be ineffec-
tive, whilst quite large currents resulting from exciting current surges could
still be flowing in the operating windings of relays. It was apparent, therefore,
that acceptable performance would not be obtained from schemes employing
only bias derived from the unidirectional components of currents.
It was also realized that only a limited amount of bias obtained from these
components could be applied because of the quite large unidirectional com-
ponents which can be present initially in the currents fed to internal faults on
power transformers.
It was shown in other studies that the large second harmonic components
present in exciting current surges could be reproduced faithfully by current
transformers and that these current components could provide the necessary
bias to ensure that relay operation would not occur as a result of such surges.
It was found, however, that significant second harmonic components can be
present in the secondary winding currents of current transformers if their cores
saturate when high currents are flowing as a result of internal faults on power
transformers. It was then clear that either the degree of second harmonic
biasing applied in a scheme should be limited to ensure that relay operation
would not be prevented under such conditions or that separate unbiased high-
set relays should be provided to clear these faults.
The possibility of obtaining bias from the relatively large third and fifth
harmonic components present in exciting current surges was also considered.
Whilst such components are not present in the currents which flow when
internal faults are present on power transformers, they can nevertheless be
present in the secondary winding currents of current transformers when their
cores are driven into saturation for parts of half cycles. It was recognized
therefore that any bias obtained from these components would have to be limited.
As a result of these studies, several schemes employing relatively simple
analogue filters were developed to provide biasing for current-differential
schemes which were produced about 1950. In one arrangement filters were
used to extract the second harmonic components from the imbalance currents
flowing in the protective circuit, the extracted currents being fed through bias
winding on the relays which had the number of turns needed to ensure satis-
factory performance. Because the filters did not have narrow enough pass
bands, small proportions of the unidirectional and third and fifth harmonic
components of the exciting currents also flowed through the bias windings of
the relays. The extraction or diversion of some of the imbalance current to the
bias windings reduced the current flowing in the operating windings of the
relays, thus further helping to reduce the possibility of incorrect operation
being caused by exciting-current surges.
In an alternative arrangement [3], simple LC series-tuned circuits were used
to permit only the fundamental component of the imbalance currents to flow
in the operating windings of the relays. The other components were fed to the
bias winding. In this scheme the relays were set so that they would not operate
when their bias winding current exceeded 15% of the current in the operating
windings. The minimum operating current of the relays, i.e. when there was
zero bias, was 15% of the secondary rating of the current transformers and the
minimum operating time was about 40 ms.
When high-current internal faults occur, the cores of the current trans-
formers may saturate, causing their secondary winding currents to contain
second and third harmonic components, as stated earlier. These components
together with the unidirectional components present in the fault currents may
provide sufficient bias in schemes of the above type to delay relay operation
until the transients fall to low levels. Unbiased relays were therefore often
connected in series with the operating windings of the main relays, the extra
relays having current settings above the levels produced by exciting current
surges. These relays then provided rapid clearance of high current internal
faults.
Recent developments Current-differential protective schemes of the type de-

scribed above, with both percentage and harmonic bias, have been widely
applied to power transformers since 1950. Developments in electronic equip-
ment in recent years have enabled relays with both non-linear percentage
biasing and an improved method of biasing associated with exciting current
surges, to be introduced.
In older schemes, percentage biasing was obtained by having bias windings,
with fixed numbers, of turns connected so that the current circulating in a
current-differential scheme flowed through them. As a result, the bias was
directly proportional to the current and a fixed percentage bias was thereby
provided. There are, however, advantages in having non-linear or variable
percentage biasing. At currents up to rated level a relatively low percentage
bias, say 20%, is desirable to maintain high sensitivity for internal faults and
it is adequate to counter the small mismatches which occur at low current
levels and also any imbalance caused by tap changing. During external faults
when currents well above the rated level are flowing, saturation may occur in
one or more of the current transformer cores and large imbalance currents may
flow as a result. To counter such situations a high percentage bias slope, of
say 80%, is desirable. Relays [4,5] with such characteristics, e.g. 20% bias up
to rated current and a bias slope of 80% above rated current, are now produced.
Because of the possibility of the clearance of high-current internal faults
being delayed by harmonic biasing, which was referred to earlier, an alterna-
tive method of inhibiting relay operation during exciting current surges has
been introduced recently. It is based on detecting the significant periods in
each cycle during which an exciting surge current is almost zero. This feature,
which can be seen in Fig. 6.6(b), is very pronounced and as it is not present
in the currents associated with internal faults it can be used to inhibit relay
operation. In practice, the imbalance currents in current-differential schemes
are monitored because they tend to be proportional to the exciting currents
when switching surges are present.
The current transformers feeding schemes using this technique should be so
designed that they will not saturate when internal faults occur with current
magnitudes up to the levels of exciting current surges. This is necessary to
ensure that their secondary currents, when such internal faults are present, will
not be near zero for significant periods as a result of core saturation. In
addition, internal faults which cause very high currents to flow should be
detected separately by employing techniques which will provide rapid opera-
tion and do not involve bias.
A relay which incorporates the above features is described in more detail in
reference [4].
Application to three-phase transformers with two windings per phase

When current-differential schemes are applied to three-phase transformers
account must be taken of the connections of their windings. Six current trans-
formers must be used and their secondary windings must be so connected that
significant currents do not flow in the three relays unless internal faults are
present.
As an example, if a transformer has delta- and star-connected windings
arranged as shown earlier in Fig. 6.7, and for which the winding-current
relationships under healthy conditions are as expressed in equation (6.3), then
the secondary windings of the current transformers on the low voltage side
must be connected in star and those on the high voltage side must be connected
in delta.
The ratio of the sets of currents transformers on the two sides must be
different satisfying the condition given below:
Npp= _
_ Np 0Nsp
-
Current Power Current

transformers transformer . transformers
s Npp f
~Ol...
~ s Np f s Ns f~ s Nsp f
000 a
000
a
~No f f~
-isaNslNp
Nt.ns/sa
r-'oooT
S
ps ss
*npipa
b
N
s of~ f
~
~
Is Np f
00_0 -
s Ns isb
r--ooo f - s o~sf f b
sOoo f
Nps
-is-;;!;Np - 000
s Nss
T
t nSisb
* npipb N ipc N
s off f - s OODp - f
N isc
S f- ~SfJ-o c
I
~ooo s
c
~ ~rps f -isJ!jNp
5 Nsso f t
+ npipc
i=O+Y Y yRelays
Fig. 6.13 Protective scheme applied to a delta-star connected power transformer.

To balance np ipa = ns(isb - isa), but ipa = Z; (isb - isa) :. ns/np = Ns/Np
. Nsp Ns Npp
l.e.-=--
Nss Np Nps
Note: The above assumes that the power transformer and the current transformers are
ideal.
and the interconnections between the two groups of current transformers must
be as shown in Fig. 6.13.
The relays should be set as sensitively as possible whilst ensuring that they
will not operate during healthy conditions including exciting-current surges or
when faults are present at points external to the protected zone. To enable this
objective to be achieved, both percentage and harmonic or other biasing must
be employed.
Operation of one or more of the relays should be obtained for all types of
faults on the delta-connected winding, for interwinding faults, and for inter-
phase and earth faults on a high percentage of the star-connected windings.
A factor which must be recognized when a transformer has a delta-
connected winding is that zero-sequence currents and triplen harmonic
currents may circulate around it. In these circumstances third harmonic com-
ponents in exciting current surges may be confined to a delta-connected wind-
ing and not be present in the current transformers included in the input and
output line connections to a star-delta connected transformer. As a result, no
third harmonic components can be derived for biasing purposes.
Application to transformers with three or more windings per phase

The arrangements described above may clearly be extended to cover both
single-phase transformers with three or more windings and three-phase trans-
formers with three or more windings per phase.
Considering, as an example, a three-phase transformer with three windings
per phase, nine current transformers are necessary, and to enable balance to
be achieved during healthy conditions there will be groups of current trans-
formers with three different ratios.
In such applications, mismatching similar to that considered earlier will be
present, indeed it may be greater, and again imbalance currents will flow
during exciting current surges. Satisfactory behaviour may nevertheless be
obtained by adopting suitable relay settings and employing adequate percent-
age and harmonic or other biasing.
It has only been possible to consider the most commonly used connections
of three-phase transformers in the preceding sections but it will be appreciated
that current-differential schemes can be applied successfully to all trans-
formers. Basically, the currents in each of the windings must be monitored
using current transformers and their ratios and secondary winding intercon-
nections must be such that relay operation will not occur when the protected
transformer is healthy.
6.3.3 Restricted earth fault protection

It will be appreciated from the previous section that extremely low relay
settings may not be used in overall current-differential schemes of the type
described because of the mismatching which may occur and because of the
need to ensure that operation will not be caused by exciting current surges.
It will also be realized that the percentage of a star-connected winding on
which earth faults will not be detected by such schemes may be quite high
when relays with relatively high current settings are employed. This is par-
ticularly so when significant impedance is present in the neutral connection
to limit the levels of earth fault currents.
The above behaviour is not acceptable in many cases and therefore restricted
earth fault protection is applied to star-connected windings. This form of
protection, which was described in section 5.5, also belongs to the current-
differential group. As stated earlier, it employs four current transformers, each
of the same ratio, one having its primary winding in series with the neutral
connection, the others being connected in the connections to the three phases.
When the protected winding is healthy, or in the event of interphase faults not
involving earth, the secondary currents should circulate and balance at all
times, the current in the operating winding of the shunt-connected relay being
zero. When an earth fault occurs, however, balance is not achieved and a
current proportional to the fault current would flow in the relay winding if it
had zero impedance.
In practice, some mismatching occurs, possibly due to residual fluxes being
present in the cores of the current transformers at the instant when a fault
occurs. To prevent relays with low-impedance windings operating incorrectly
in these circumstances, percentage bias is usually necessary.
The alternative arrangement, shown in Fig. 6.14, in which the relay circuit
has a high impedance enables improved performance to be obtained. The basic
behaviour of this form of protection was explained in the previous chapter
(section 5.6, page 171). In principle the relay is set to operate at a voltage higher
than those which can be produced across it during external faults as a result
of one or more of the current transformers saturating whilst the others remain
unsaturated. A detailed theoretical and experimental study of this protective
scheme, made by Wright [6], confirmed that satisfactory performance can be
obtained under all conditions, provided that appropriate relay settings are used
and that the current transformers have the required characteristics.
Power Current
transformer
windings
r-:---...---I S Ns
Relay
Fig. 6.14 Protective scheme incorporating a high-impedance relay.

This particular scheme has been widely applied since 1950 and it is still
used successfully today. It is usual to arrange for operation of the relay to
occur when fault currents of 20% or more of the rating of the protected
winding are flowing from it to earth. Clearly this scheme, which effectively
compares the current flowing into the phase windings with that flowing to the
neutral, is not affected by the exciting currents in the windings and the con-
ditions needed to ensure that relay operation will not occur at current levels
up to the maximum which can be reached during faults can readily be met.
6.3.4 Combined differential and restricted earth fault protection

The application of separate current-differential and restricted earth fault pro-
tective schemes to a three-phase, two-winding, power transformer requires
that ten current transformers be provided, six for the current-differential
scheme and four for the earth fault protection. In addition, further current
transformers will be necessary for other purposes such as metering, back-up
protection and unit-type protective schemes covering adjacent zones.
In some cases, therefore, because of the difficulties of accommodating large
numbers of current transformers within switchgear or other power equipment,
it is considered to be advantageous to reduce the number of transformers
needed by feeding current-differential schemes and restricted earth fault
Current Power transformers Current

transformers .-_--..;.w;.;,;in.,.dings transformers
s f s f
s f s
s f
- Summation
m.m.f. balance transformer ~ Restricted
relay
1-------, t.c=]J earth-fault
rela
Fig. 6.15 Combined m.mJ. balance and restricted earth fault protective scheme ap-
plied to a delta-star connected power transformer. Note: For simplicity only one m.m.f.
balance relay is shown.
schemes from the same line current transformers. As a result, the number of
current transformers may be reduced from ten to seven. One arrangement
which is commonly adopted to achieve this reduction when protecting a three-
phase, delta/star connected power transformer is shown in Fig. 6.15, from
which it will be seen that an auxiliary summation current transformer is
included to supply the operating winding of the earth fault relay. Clearly the
summation transformer and the other current transformers must be so designed
that the necessary current balances will be achieved in both protective schemes
unless faults are present within the protected zone.
6.3.5 Earth fault protection of delta-connected windings

Should earth faults occur on delta-connected windings which are connected
to networks on which there are earthing transformers or on which there are
earthed neutral points, then quite large currents will normally flow because all
points on the winding will be at voltages of at least half the phase value, as
stated earlier in section 6.2.3 (page 193). Such faults should normally be
detected by current-differential protective schemes, but nevertheless separate
earth fault protection is often applied to delta-connected windings. This pro-
tection is basically the same as the restricted earth fault protection described
in section 6.3.3 (page 208) except that only three current transformers are used,
their primary windings being connected in the line connections to the winding,
as shown in Fig. 6.16. Either high-impedance or percentage-biased low-
impedance relays may be used with fault setting of 20% or less. These will
operate for earth faults on the protected winding because the three line currents
will not sum to zero for these conditions. They will not operate for faults on
the associated network, however, because delta-connected windings cannot
provide zero-sequence currents and the other current components add to zero
at all times, i.e. the line currents must always sum to zero unless there are
faults on the windings.
Current Power
transformers
windings
ao-----~~~-----4r_--------r,n~---------,
Fig. 6.16 Earth-fault protective scheme applied to a delta-connected winding.

6.3.6 Tank earth protection
When an earth fault occurs on a winding of a transformer protected by a
differential protective scheme in which comparisons are made of the m.m.f.s
associated with the windings, a current will flow in the operating winding of
one of the relays. This current, as shown earlier in section 6.2.3, will not be
solely dependent on the magnitude of the fault current, but also on the fault
position. As a result, faults on a significant part of a winding may not be
detected even when a relatively sensitive relay setting is used. This situation
will certainly apply if a large earthing impedor is present in the circuit.
Restricted earth fault protection, as stated in section 6.3.2 (page 208), can
detect faults on star-connected windings which cause currents above the set-
ting value to flow to earth. Because of the need to ensure that such schemes
will not maloperate when earth faults occur outside their zones, relay settings
of the order of 20% are normally used, i.e. 0.2 A when the current transformers
have secondary windings rated at 1 A. In those applications where impedance
is included in transformer neutral connections to limit earth fault currents to
the rated current (1 pU), operation of relays with 20% settings would not occur
for faults on the 20% of the winding adjacent to the neutral point.
The amount of winding protected can be increased in such applications if
the transformer tank can be lightly insulated from earth and a current trans-
former can be installed with its primary winding connected between the tank
and earth. A current-operated relay connected to the secondary winding of the
current transformer can be given a very low setting, say 5%, because current
can only flow in it when winding faults to the tank occur. With such a setting,
95% of the windings would be protected if an earthing impedor limiting fault
currents to 1 pu were present.
6.3.7 Overfluxing protection

As explained in section 6.2.4, it is desirable that the ratio of winding voltage
to frequency should not exceed a particular value at which the required flux
swings are approaching saturation levels. In some applications, therefore, a
relay is provided to monitor this ratio and provide an alarm when the set value
is exceeded.
One relay [7], which has been used for this purpose for some time, is
supplied by the output (V) of a voltage transformer connected to a winding of
the power transformer being monitored. The relay contains a resistor (R) of
high ohmic value connected in series with a capacitor (e) of relatively low
reactance. The input voltage (V) is applied across this combination and a
current proportional to the voltage flows through the capacitor, i.e. I = VIR.
The resulting voltage Vc across the capacitor is proportional to Vlf i.e.
Ve =lime = Vl2nfeR. A voltage-operated device connected across the capa-
citor is set to operate when the ratio exceeds a critical level.
More recently, a relay [8] has been introduced which monitors the imbal-
ance current in current-differential protective schemes. This current has a large
component caused by the exciting current of the protected transformer, and
when the VI! ratio rises core saturation occurs and the exciting current has
high peaks and periods when it is very low. This behaviour, which is detected
using modern processing techniques, is a direct indication of overfluxing.
6.3.8 Protection against overheating

The temperatures within a power transformer at any instant are not dependent
solely on the currents then flowing in its windings. Certainly in some instan-
ces, because of inadequate oil circulation resulting from faulty pumps or
blockages in ducts or pipes, local or general overheating can occur even when
transformers are lightly loaded. It will therefore be clear that reliance cannot
be placed on using overcurrent relays to detect unacceptable overheating, and
that devices which monitor temperatures directly are preferable.
Over the years, temperature sensors have been positioned in the oil and
within the windings of power transformers. In the past, thermostats or bulbs
containing volatile liquids which operated remote pressure-operated devices
to which they were connected by small diameter tubes were used. In recent
years, however, it has become common practice to install a thermal-sensing
element in a small compartment positioned near the top of a transformer tank
where the oil tends to be hottest. A small heater, which is also installed in the
compartment, is fed from a current transformer connected in series with one
of the phase windings. A local oil temperature comparable with that present
in the main windings is thereby produced in the compartment. Heat-sensitive
resistors are now used as sensors and these form parts of resistance bridges
which produce imbalance output signals that initiate either alarms or the
opening of the appropriate circuit-breakers when unacceptable overheating
occurs. A relay type TTT was produced for this purpose by GEC Measure-
ments [9].
Because the life of insulation is dependent on both the temperatures at which
it operates and the periods spent at each temperature, relays, designated tem-
perature-time integrators, have been developed to measure the periods for
which transformers operate within a number of temperature bands; for
example, the number of hours for which a transformer operates at a tempera-
ture within the band lOO-llOoC may be recorded. From such information an
estimate of the degree of ageing of the insulation in the transformer may be
made. Further information on these relays is available in reference [lO].
6.3.9 Buchholz relays

Protective schemes which monitor circuit currents or voltages can operate
rapidly and open the appropriate circuit-breakers when certain conditions
exist. As an example, a differential scheme will operate when the current
imbalance is at or above a particular level, a situation which only arises when
a significant fault current is flowing in the protected equipment. In many cases
serious damage will have been caused before the faults are cleared in this way
and expensive repairs will be needed.
It is clearly desirable therefore that, where possible, other methods be em-
ployed to detect potentially serious conditions when they begin so that action
may be taken before extensive damage is caused. For many years this objective
has been achieved by fitting Buchholz relays in the pipes between the main
tanks and the conservators of oil-filled transformers. A typical relay and
mounting arrangement is shown in Fig. 6.17.
Gases such as hydrogen and carbon monoxide are produced when decom-
position of the oil in a transformer occurs as a result of localized heating or
arcing associated with winding faults. Gases produced in this way rise from
the transformer and pass up the sloping pipe, shown in Fig. 6.17(a), towards
the conservator. They pass into the Buchholz relay, which is normally filled
with oil, and rise and become trapped in the top of the casing so displacing
oil. As a result, a pivoted float or bucket, depending on the relay design, falls
and operates a mercury switch or closes contacts. This situation is shown in
Fig. 6.17(b). An alarm signal can thereby be provided for a number of condi-
tions of low urgency, including the following:
Conservator
Transformer
._ m~'nn
Gas detector (a)
OO""ru
- --
UI~utij-f-
Trip contacts
- --
---------
Surge detector
Normal condition Alarm condition
(b)
Fig. 6.17 Buchholz protection.

(a) Hot spots in a core caused by short circuiting of the laminations.
(b) Breakdown of the insulation on core clamping-bolts.
(c) Winding faults causing only low currents to flow through the current
transformers.
In addition, an alarm would be given if the oil level fell below that of the
Buchholz relay as a result of a small leakage from the main tank.
Buchholz relays are provided with valves or cocks to allow samples of gas
collected in them to be analysed. This is a valuable feature because it allows
the cause of gas generation to be determined and may assist in determining
the remedial action which is needed. As examples, the presence of H2 and
C2H2 indicates arcing in oil between constructional parts and the presence of
H2, C2~' CO2 and C3H6 indicates a hot spot in a winding.
Major winding faults, either to earth or between phases or windings, invol-
ving severe arcing, cause the rapid production of large volumes of gas and oil
vapour which cannot escape. They therefore produce a steep build up of
pressure and displace oil, setting up a rapid flow towards the conservator. Such
flows cause either a vane or a second bucket or float in a Buchholz relay to
move and close contacts which initiate the tripping of the appropriate circuit-
breakers. Whilst such relays do not operate as rapidly as those in current-dif-
ferential schemes, they nevertheless operate much more quickly than IDMT
relays, and they therefore provide a very satisfactory back-up feature.
The physical size and the settings of a Buchholz relay are clearly dependent
on the transformer to which it is to be fitted. Firstly, it must be suitable for
mounting in the pipe between the main tank and the conservator of the trans-
former. Secondly, it must be set to provide an alarm when a particular volume
of gas has been collected and thirdly it must initiate circuit-breaker tripping
when the oil velocity exceeds a particular value. These values tend to increase
with the ratings of transformers, typical values being shown in Table 6.1.
A number of factors have to be taken into account when selecting the
settings of Buchholz relays, in particular the oil-velocity setting must be high
enough to ensure that relay operation will not occur as a result of pressure
surges which may be produced when oil-circulating pumps are started. This
and other factors are considered in more detail in reference [11].
It is now European practice to provide Buchholz protection on all trans-
formers fitted with conservators and suitable relays are either recommended
Table 6.1
Transformer rating Pipe diameter Alarm Volume of gas Trip min. oil velocity
MVA (in) (em3) (em/s)
up to 1 110 70-130
1-10 2 200 25-140
>10 3 250 90-160
Fig. 6.18 The protective equipment applied to a large transformer.
by transformer manufacturers or installed by them during manufacture or

erection.
6.3.10 The protection of large three-phase transformers

Several of the protective schemes and relays described in the preceding sec-
tions must be applied to each large transformer to ensure that all the unaccept-
able conditions which may arise in it will be correctly detected. A typical
arrangement, which will provide adequate cover, is shown in Fig. 6.18.
6.4 THE PROTECTION OF EARTHING TRANSFORMERS

Step-down transformers used to feed distribution networks from higher volt-
age transmission circuits usually have star-connected primary windings and
delta-connected secondary windings because the amount of insulation on the
high voltage windings is thereby minimized. Clearly the delta-connected, low
voltage windings cannot be directly earthed and therefore a separate earthing
transformer is often provided to ensure that the distribution circuits are not
floating free from earth.
Three-phase earthing transformers have six windings, each of the same
number of turns, mounted in pairs on the limbs of the core and they are
connected to the delta-connected windings of power transformers as shown in
Fig. 6.19(a). During normal balanced conditions, the winding voltages are as
shown in Fig. 6.19(b) and the fluxes in the earthing transformer limbs are each
of the same magnitude but displaced from each other by 2n/3 radians. Only
the small currents needed to provide the m.m.f.s flow in the earthing-trans-
former windings.
The protection of earthing transformers 217
Power transformer
(a)
fo fo fo
s S S
Ne Ne Ne
f f f
Earthing transformer
s s s
Ne
f
t.
/0
Ne
f
t.
/0
Ne
f
t. hio
/0
VeB
-
(b) (c)
Fig. 6.19 Earthing transformer connections and behaviour.
When an earth fault occurs on the low voltage circuit, positive- and nega-
tive-sequence current components are fed by the power transformer and the
fault current, which is equal to the sum of the zero-sequence currents, flows
into the neutral ofthe earthing transformers as shown in Fig. 6.19(c). Because
of the distributions of the currents in the windings, no resultant m.m.f.s are
produced by the currents and therefore no zero-sequence fluxes or e.m.f.s are
produced. As a result, the earthing transformer effectively presents zero reac-
tance to the earth fault current.
Current
transformers
"___ la_= 3/0 s Nsp f
, 0UUU~-------oa
s Nss f
Ib=O I NSPfl
,
-~~------~-----------~----~.sJUUU~.----_Ob
~----~----r----~~=~O---rI~~11 oc
3d
10'
-
r.:::;.;:::..-_-"="='1
... ...~
N N N N Io'
10' = ~ 10 and ~ = _s .--.P.
Nss Nss Np Nps
-
Auxiliary
Io'
transformers - 10~l/o~1
2 /0'
-
Fig. 6.20 (a) Arrangements of current-differential protective schemes when an earth-
ing transformer is present. To simplify the diagrams. only one phase of the current-
differential relays is shown.
Auto transformers and their protection 219
An earthing transformer, when it is mounted near a power transformer, may
be included in the zone of a current-differential protective scheme. When this
is done, the sum of the three zero-sequence current components flowing to an
earth fault on the distribution network through the line current transformers
will return up the neutral connection of the earthing transformer. To ensure
that these components will not cause imbalance an extra current transformer
must be included in the neutral connection of the earthing transformer. This
transformer may have the same ratio as the line current transformers on the
delta-connected side of the power transformer, in which case its secondary
winding must feed three auxiliary current transformers with a step-down ratio
of 3 to 1 to provide the appropriate zero-sequence currents to balance with
those in secondary windings of the line current transformers. This arrangement
is shown in Fig. 6.20(a).
An alternative arrangement, which may be adopted when the current-dif-
ferential protection applied to a transformer is not combined with restricted
earth fault protection, filters out the zero-sequence components present in the
secondary windings of the line current transformers on the delta-connected
side of a power transformer by including a three-phase, star-delta-star-con-
nected interposing transformer, as shown in Fig. 6.20(b).
Should an earthing transformer be mounted at some distance from a power
transformer or if it is felt desirable to exclude it from the zone of the current-
differential scheme protecting a power transformer, it must be provided with
its own protection. Because of its relatively low rating, the use of a separate
current-differential scheme is not justifiable. A relatively simple scheme
which is used in these applications employs three current transformers whose
primary windings are connected in the lines between the network and the
earthing transformer, as shown in Fig. 6.21. The secondary windings, being
connected in delta, allow the zero-sequence currents which are present when
there are earth faults on the network to circulate around them and do not cause
currents to flow in the relays during these conditions. The relays may therefore
be given relatively low current settings, which enables operation to be ob-
tained for most of the faults which may occur within the earthing transformer.
IDMT overcurrent relays are usually employed in these applications because
rapid operation is not necessary unless fault current levels are very high.
6.5 AUTO TRANSFORMERS AND THEIR PROTECTION

In many networks it is necessary to provide isolation between circuits opera-
ting at different voltage levels. As an example, transmission lines operating at
voltages such as 400 kV are isolated from alternators, producing voltages of
about 25 kV, by transformers with separate primary (Iv) and secondary (hv)
windings.
In some situations, however, where the ratio of the voltage levels of circuits
is not high and where isolation is not necessary, it is economically advant-
Current Power transformer Current
transformers ;--- transformers
s Nppf Nsp
~., s;..ULf oa
~. , I~;..UL ob
o I I I 000 I -~ L.......rorr' I I"";..UL oc

Ne
s
N
Ne
Interposing rl ~ rl
,,,,",,,,,me, ~
Current differential N
relay
r..::..:-_-. - -=..::. 1
'" " r" (b)
Fig. 6.20 (b)

Auto transformers and their protection 221
------T-----------------------------a
----~~--~---------------------b
------+-----~----_r----------------c
Current N.
transformers ~ Relays
Earthing transformer
s~-+--~~--s~~
Ne Ne
f f
Fig. 6.21 The protection of an earthing transformer.
ageous to use auto transfonners. This can be seen by comparing the ideal auto
and two-winding transfonners, shown in Fig. 6.22, which both provide a
step-down ratio of 2: 1. The auto transfonner has a total of Np turns of
conductor capable of carrying a current of I p, whereas the two-winding trans-
former has a primary winding of Np turns which must carry a current of Ip and
also a secondary winding of Np /2 turns which must carry a current of 2Ip. The
volume of conductor needed for the auto transformer in this case would be
half that needed for the two-winding transfonner and the power loss in the
auto transformer would also be half that in the two-winding transfonner. The
advantages of the auto transformer clearly increase as the required transfonner
ratio reduces towards unity but decrease as it rises above two.
(a) (b)
Fig. 6.22 Auto- and two-winding transformers.

For the above reasons, auto transformers are used whenever possible, a
common application being the interconnection of transmission networks
which operate at different voltages. As an example, a considerable number of
three-phase auto transformers are in use to link networks operating at 275 kV
and 132 kV in Britain. Such transformers are always star-connected. The
neutral points are invariably earthed either solidly or through earthing impe-
dors.
The forms of protection applied to auto transformers are considered below.

Auto transformers could be protected by current-differential schemes of the
type considered earlier in section 6.3.2 (page 198), in which the input and
output m.mJ.s are compared. Considerable imbalances occur in these
schemes, which require two current transformers per phase. In addition, as
stated earlier, these schemes are unable to detect earth faults on a significant
number of turns near the earthed ends of windings, this number representing
a considerable percentage of the total number of turns in a winding when
earthing resistors are present.
A superior alternative, which is normally used on auto transformers, is a
current-differential scheme which directly sums the currents entering a wind-
ing. Three current transformers, each of the same ratio, must be provided per
phase, the primary winding of one being in series with the neutral connection
and the others in series with the high and low voltage connections. Their
secondary windings and the operating winding of the relay must be intercon-
nected as shown in Fig. 6.23.
abc
s
Current N.
transformers PP
f
Auto N. ~----~~~~-+~C
transformer P f N. s
windings PP
Fig. 6.23 The protection of an auto-transformer.

The future 223
Unbalance is not produced in this scheme by exciting current surges, and
because the interconnections tend to be short other imbalances are usually
small. Either percentage-biased low-impedance or unbiased high-impedance
relays may be employed and they may be set to operate at low per-unit values,
0.2 or less, i.e. 0.2 A when current transformers with secondary windings rated
at 1 A are used. With such settings, earth faults on the windings which cause
currents of 0.2 pu or more to flow will be detected as also will interphase faults
of the same magnitudes. In the latter case, two or more relays will operate
depending on the type of fault.
It must be recognized that this scheme will not operate in the event of
interturn faults and preferably some other form of protection should be pro-
vided to detect them. If this is not possible, then it must be accepted that these
faults will persist until they cause breakdowns to earth or between phases.
6.5.2 Tank earth protective equipment

As stated in section 6.3.6 (page 212), the presence of earth faults on windings
can be detected by lightly insulating a transformer tank from earth and con-
necting the primary winding of a current transformer in the connection be-
tween the tank and earth. The secondary winding of the current transformer
may then supply the operating winding of a relay. Because only currents
flowing from the windings to the tank can flow in the primary winding of the
current transformer and no current comparisons are involved, the relay can be
set to operate at very low current levels without increasing the possibility of
maloperation. The application of this form of protection to auto transformers
allows the numbers of turns near the earthed ends of windings on which earth
faults cannot be detected to be reduced below the levels obtainable with
current-differential schemes. The reduction can be particularly great when
earthing impedors are present in the neutral connections to earth and the use
of tank earth protection in such applications should be considered.
6.5.3 Other protective equipment

Because the physical constructions of auto and multi-winding transformers are
similar, the methods of detecting overfluxing and overheating which were
described in sections 6.3.7 and 6.3.8 respectively may be applied to auto
transformers. In addition, IDMT relays are applied to provide back-up protec-
tion and Buchholz relays, which were described in section 6.3.9, are applied
to large auto transformers which are oil-filled and have conservators.
6.6 THE FUTURE

The basic principles of transformer protection are well established. Changes
which are envisaged in the future relate to the use of digital techniques for the
extraction of relaying parameters. In the case of transformers, it is likely that
restraint (or bias) signals, which hitherto have been obtained using analogue
filtering techniques, will in the future be obtained by digital filtering. To
illustrate this development, if the third and fifth harmonic of the current is
required to form a restraint quantity, these could be obtained by sampling the
current at regular time intervals T, and processing these sample values to
obtain the desired harmonic components. There are a number of algorithms
suitable for this purpose, e.g. those based on the discrete Fourier Transform.
The sampling frequency must exceed twice the highest frequency of interest
as described in section 4.3.1 (page 118). Hence, if the 5th harmonic of a 50 Hz
signal is required the sampling frequency must be at least 500 Hz, or 10
samples per period. Samples from all the current transformers or transducers
in a differential protective scheme could be processed by a single unit using
multiplexing techniques. An advantage of such methods over conventional
techniques is that it is easier to combine restraint signals from a number of
transducers or from different harmonic components to form a more secure
restraint signal. A description of these techniques may be found in references
[12-14] and in section 14.3.2 of this book.
Relays are now available of a fully numerical design with all the advan-
tages that this brings in terms of communications, sharing of data and
enhanced functionality. Examples of relays of this type which incorporate
all transformer protection and alarm functions may be found in [15,16].
Such relays include biased differential protection, restricted earth fault
protection and a host of other functions in a single unit.
A further development, which may affect transformer protection, is the
wider availability of relaying signals in an integrated protection and control
environment. Current practice relies on current transducers for transformer
protection. Voltage transducers are too expensive to be justified in terms of
transformer protection alone. In an integrated environment, however, voltage
signals may be available and shared by control, monitoring and protective
systems. In such cases, relaying algorithms which provide restraint signals
derived from voltage transducers may be more widely used [12, 14]. The
topic of integrated protection is discussed in more detail in section 15.5 of
this book.
REFERENCES
1. BS171 Power Transformers, Parts 1 to 5: 1978, British Standards Institution
(equivalent to IEC 76: 1976).
2. Kennedy, L. F. and Hayward, C. D. (1938) Harmonic restrained relays for dif-
ferential protection, Trans. AlEE, 57, 262-271.
3. Hayward, C. D. (1941) Prolonged inrush current with parallel transformers affect
differential relays, Trans. AlEE, 60, 1096-1101.
Further reading 225
4. Type MBCH biased differential protection for transformers, GEC Measurements,

Publication R-6070B.
5. Type DT92 three-phased biased differential relay, Asea Brown Boveri, Publica-
tion CH-ES 65-60, Vol. 1, Buyers Guide 1989-90.
6. Wright, A. (1954) The performance of current transformers and relay circuits
employed in the earth fault protection of power-systems. MSc Thesis, University
of Durham.
7. Type GTT overfluxing relay, GEC Measurements, Publication R-5159C.
8. Type RATUB 2 V/Hz overexcitation relay for transformers, Asea Brown Boveri,
Publication B03 -50 11.
9. Winding temperature relay and indicator, Type TIT, GEC Measurements, Pub-
lication R-5074G.
10. Protective Relays Application Guide, GEC Measurements, (3rd edn), 1987,
p.282.
11. Ibid., pp. 94 and 296.
12. Phadke, A. G. and Thorp, J. S. (1988) Computer Relaying for Power Systems.
RSP-J Wiley.
13. Rahman, M. A. and Jeyasurya, B. (April 1988) A state of the art review of
transformer protection algorithms, IEEE Trans. on Power Delivery, 3, (2), 534-
544.
14. Microprocessor relays and protection systems, IEEE Tutorial Course, 88EH0269-
I-PWR, 1987.
15. Reyrolle Protection (1996) Duobias-M numeric transformer protection,
Publication REY3/9506-069/1-1.
16. GEC Alsthom T and D (1996) Differential protection for transformers and
generators, Type KBCH, Publication R6530F.
FURTHER READING
IEEE Trans. on Power Delivery, PD3, 525-533.
GEC Measurements, Transformer protection, Publication R-4025.
Murty, Y. V. V. S. and Smolinski, W. J. (1988) Design and implementation of a digital
differential relay for a three-phase power transformer based on Kalman filtering
theory.
7
The protection of rotating machines
A very large number of electrical machines of a wide range of types and

ratings are used in power systems around the world. The vast majority of them
have a rotating member, i.e. a rotor and a stator, and both members usually
have windings associated with them.
The largest machines are the three-phase alternators used in generating
stations. In the UK, many stations have machines which provide 50 Hz volt-
ages of 22 kV (line) and power outputs of 500 MW. Their star-connected
output windings, which are mounted in the stator, are thus capable of carrying
currents of 15 kA. Their rotor windings, which provide the m.m.f.s needed to
set up the magnetic fields, carry direct currents up to 4.24 kA, these being
supplied by exciters rated at 2.5 MW. The field windings of these machines
are supplied in turn by pilot exciters. The alternators and their exciters are
driven by steam turbines supplied by boilers. Associated with such plants are
motors which drive pumps, fans and other items including coal pulverizers.
Even larger alternators are now in use. As an example those installed in the
Itaipu hydro-electric power station on the Parana River which flows between
Brazil and Paraguay are rated at 700 MW. This station contains 18 of these
machines, each of which weighs 2700 tonnes.
In an article [1) published in the lEE Power Engineering Journal in 1990,
Creck described machines rated at 985 MW, 1158 MVA which were being
produced at that time by GEC Turbine Generators for the Daya Bay nuclear
power station in the People's Republic of China. These machines are designed
to produce voltages of 26 kV (line) at a frequency of 50 Hz, the full-load
current of the output windings being 25.7 kA. Their stator and rotor cores are
to be hydrogen cooled and water flowing in a closed circuit will cool the stator
windings.
It will be appreciated that power units, such as those referred to above, are
very complex and it is necessary that their performances are completely moni-
tored at all times to ensure that continuity of supply is maintained and that
any faults or breakdowns which may occur do not cause unnecessary conse-
quential damage.
Machines with ratings ranging down from the very high levels quoted above
to levels of only a few watts or less are in service and many are produced

228 The protection of rotating machines
annually for widely ranging applications. Both three-phase and single-phase
a.c. synchronous and induction machines are produced as well as motors
suitable for operation from sources of direct voltage or rectifiers. In addition,
in recent years, reluctance motors [2] operating from controlled-rectifier banks
have been introduced to provide efficient variable-speed drives.
Over the years several forms of protective equipment have been developed
to enable the whole range of rotating machines to be protected at cost levels
which are acceptable. Clearly cheap devices must be used with small ma-
chines, whereas quite costly equipment is justifiable when very large and
important plant is to be protected.
In all applications, it is desirable that the most appropriate protective equip-
ment is used and to enable this to be achieved it is necessary that adequate
information about the machine to be protected is available when protective
schemes or devices are to be selected. As an example, if fuses or relays
with inverse time/current characteristics are being considered for use with a
particular motor, then it is essential that the magnitude and duration of the
current surges which will flow in it during starting periods are known.
A significant factor which should be recognized at this point is that although
machines which are to be used as motors may be physically the same as others
which are to be used as generators, their protection must take into account the
use to which they are to be put.
Motors, clearly, are loads and therefore only their operating characteristics
need to be considered when selecting the protective equipment to be used with
them. Should time-graded schemes be used, then the protection on the circuits
supplying the motors must be set so that the necessary co-ordination will be
achieved under all conditions, i.e. the devices on the supply circuits should
operate more slowly than those on the motors.
Generators are, of course, sources and therefore the settings of their protect-
ive equipment must be such that operation will not occur in the event of faults
on the circuits being supplied by them and, in addition, the generators should
not suffer consequential damage.
It will be appreciated that it is not possible in this work to provide detailed
guidance on the protective arrangements which are most suitable for each of
the many types and sizes of machines which may be encountered in practice.
In this chapter, therefore, the protective schemes and devices currently avail-
able are examined and a number of applications of them are considered, after
some details are given of the early relays which were introduced to protect
machines.

When motors, generators and alternators were being produced in the final
decades of the last century for installation in power systems the only electrical
protective devices available were fuses. Whilst they were suitable for use with
some small motors, it was apparent that they could not provide adequate
protection for generators and alternators. To have ensured that faults on an
output winding connected to a bus bar would be cleared, fuses would have had
to have been included at each end of the winding and their operating times
would have had to be high enough to ensure that they would not operate in
the event of faults on the load circuits. They were particularly unsuitable for
use with three-phase machines because the operation of a single fuse in the
event of a phase to earth fault on a machine would have caused it to carry
unacceptably high negative- and zero-sequence currents. In addition the oper-
ation of fuses in the three-phase circuits could not have initiated the opening
of field-winding circuits. .
H. W. Clothier stated in a paper entitled 'The construction of high-tension
central station switchgears with a comparison of British and foreign methods' ,
which was presented to the lEE in 1902, that for large a.c. generators it is
feasible to do without fuses or other automatic devices.
This view was also expressed by W. B. Woodhouse during the discussion
on a paper by H. L. Riseley entitled 'Some notes on continental power house
equipment', which was presented to the lEE in 1903.
In spite of these views, there were those who felt that protective relays
should be used and certainly reverse-current relays were referred to in a paper
by C. H. Merz and W. McLellan entitled 'Power station design' which was
presented to the lEE in 1904. These relays were directional relays and their
purpose was to detect current fed from busbars into faulted machines. The
authors stated that these relays would not detect short circuits within machines
because the voltages would be too low for the relays to function correctly. The
authors nevertheless felt that automatic protective devices were required. In
the discussion on this paper, L. Andrews stated that he had demonstrated a
reverse-current device in 1898 but recognized that it still was not entirely
satisfactory. Reverse-current relays, developed by Brown Boveri, were
referred to in a paper entitled 'Protection devices for H. T. electrical systems'
which was presented to the British Association in 1903. Five years later C. C.
Garrard referred to the use of reverse-current relays with 0.5 s time lags to
allow for transient effects in a paper presented to the lEE entitled 'Apparatus
and relays for a.c. circuits'. It is clear, therefore, that most machines were
operating without effective protective equipment up to this time.
In 1910 K. Faye-Hansen and G. Harlow presented a paper entitled 'Merz-
Price protective gear and other discriminative apparatus for alternating-current
circuits' to the lEE. They advocated the use of circulating-current protective
schemes on alternators and showed arrangements suitable for delta- and star-
connected machines. Even at that time several speakers in the ensuing discus-
sion expressed the view that such schemes were unreliable and not suitable
for application to alternators. One speaker, A. E. McKenzie, stated that two
4000 kW, three-phase machines in Manchester had then been protected
by Merz-Price schemes for two years and that similar schemes were being
applied to new 6000 kW machines. He also expressed the view that devices
should be provided to detect loss of excitation. Eventually it became the
standard practice to apply current-differential protective schemes to the main
windings of all large machines and other protective devices were applied to
detect the other unacceptable conditions which are examined in the following
sections of this chapter.
7.2 PROTECTIVE DEVICES AND SCHEMES

The performance required of the protective equipment to be used with a
particular rotating machine is dependent on its type, size and characteristics,
and therefore a variety of schemes and devices, including fuses and thermal
relays, have been developed to cater for the wide range of machines produced
for use in power systems.
The types of protective equipment suitable for machine applications are
considered briefly in the following sections.
7.2.1 Fuses
The operation of fuselinks results from the heating of their elements to their
melting temperatures and they therefore have operating time/current charac-
teristics similar in form to the withstand time/current characteristics of ma-
chines. Cartridge fuselinks are therefore an inherently suitable form of
protection for small and medium-sized machines. As a result, many are pro-
duced annually for this purpose and a vast number must now be in use around
the world.
For some applications, standard cartridge fuselinks are suitable, indeed
domestic fuses with current ratings of 13 A or less are used to protect motors
fitted in the many appliances present in homes and other buildings.
For other applications, special motor-protection fuselinks are produced with
voltage and current ratings up to 11 kV and 1000 A respectively. These fuse-
links, which have special elements to enable them to cope with the current
surges which will flow through them when the motors they are protecting are
started, are described in section 1.9.8 (page 29).
7.2.2 Thermal relays

Thermal relays usually incorporate bimetal elements which deflect by amounts
dependent on the currents flowing in heating elements mounted adjacent to
them. The deflections tend to be proportional to the square of the current
flowing and above a certain current level closure of associated contacts occurs.
As a result, an inverse operating time/current characteristic is obtained above
a minimum operating-current level. Such relays therefore behave similarly to
fuses and they may be used in conjunction with contactors or circuit-breakers
Protective devices and schemes 231
to protect rotating machines. A relay of this type is described in section 4.2.5
(page 116).
These relays, which are produced in single-phase and three-phase forms,
may have their heating elements connected directly in series with the windings
of the machines they protect or they may be supplied from current trans-
formers. They may therefore be used to protect machines of a wide range of
power outputs and operating voltages.
7.2.3 Thermal devices

The above items are mounted externally to the machines they protect and their
settings must be co-ordinated. In some applications, however, there are ad-
vantages in mounting protective devices within machines during their manu-
facture; for example, a temperature-sensing device may be sited in the stator
core of a machine or near to or within a winding. Such a device, which is not
directly and solely affected by the machine currents, would operate and dis-
connect its machine if overheating occurred because the cooling was not
adequate. It would also operate if the overheating was caused by overloading.
Similarly overheating of a short-time rated starting winding, caused by a
slow build-up of motor speed, could be detected by a temperature sensor
mounted near the winding.
Devices of the above types are clearly selected by the manufacturers of
machines who are fully aware of the setting levels needed because they have
all the necessary design details of their products.
Bimetallic devices have been used as temperature sensors but more recently
thermistor probes have been introduced for this duty. The resistances of these
probes vary with temperature and thus the limiting temperature may be de-
tected by an electronic circuit which measures the probe resistance and an
output relay can thus be operated to disconnect a machine from its supply.
Clearly such a device must have its own in-built power supply.
7.2.4 Instantaneous electromagnetic relays

The devices considered above allow overcurrents or abnormal conditions to
continue until almost the point where the protected machine will be damaged.
This is desirable as it allows machines to be kept in service as long as possible.
In the event of a fault on a machine, such as a breakdown of the insulation on
a winding, it is desirable that the machine be disconnected immediately so that
further damage is not caused and thus the cost of repair and the period out of
service are both minimized. This can be achieved by using instantaneous
electromagnetic relays with operating currents above the levels which may be
encountered under healthy conditions including starting currents, but below
those produced by short circuits;
Current-differential schemes, as explained in Chapter 5, are able to detect and
operate instantaneously at fault current levels well below the rated current of
the unit protected by them, say 0.2 pu or less. This feature is very important
in applications associated with very large machines. This can be appreciated
by considering the machines being installed in Daya Bay Power Station in
China. It was pointed out in the introduction to this chapter that the rated stator
current of these machines is 25.7 kA. Clearly currents of these levels could
not be allowed to flow in fault paths because the consequential damage would
be unacceptable and very sensitive protection settings in per-unit or percentage
terms are essential.
7.2.6 Current-balance schemes

Current-differential schemes are not capable of detecting intertum faults in a
machine winding because such faults do not cause differences in the currents
at the two ends of a winding. Unacceptably high currents may nevertheless
flow in the fault paths. In machines with high current ratings, each winding
may have two or more parallel paths in it, i.e. a complete winding may be
made up of two or more separate windings connected in parallel. In such cases,
each of the parallel paths should normally carry almost the same current at
every instant. If an intertum fault developed on one of the paths however, the
parallel-connected windings would share the total current unequally and the
presence of the fault could be detected by comparing pairs of winding currents.
As an example, if two windings were connected in parallel, and current
transformers were connected in series with each of them, a fault could be
detected using the current-balance arrangement shown in Fig. 7.I(a). Altema-
Current Current
Interturn Interturn
transformers i i transformers
sNpf p~B fault Np ip~B fault
11 11
CT Protected \
winding
s Np f
I
I Current balance
L.-____........- . J I relay
____ J
(b)
Fig. 7.1 Current balance arrangements. (a) iR * 0 when ipA * ipB. (b) Relay does not
*
balance when ipA ipB.
Protective devices and schemes 233
tively, a current-balance relay with two windings energized from the two
current transformers as shown in Fig. 7.I(b) would be unbalanced as a result
of an interturn fault.
If a winding had three parallel paths (1, 2 and 3) then two current-balance
relays would be needed to compare the currents in, say, paths 1 and 2 and 1
and 3 respectively.
7.2.7 Phase-unbalance relays
It is obviously desirable that three-phase machines should operate under bal-

anced conditions and even relatively small amounts of unbalance may cause
operating efficiencies to be significantly reduced.
A negative-sequence component in the supply voltages of a motor may
cause large losses in its rotor and it is therefore usually essential that a
three-phase machine should not operate with an open-circuited phase.
Several types of relays are available to detect unbalance in voltages or
currents. Clearly voltage- or current-fed sequence networks may be used to
feed outputs to operate electromagnetic relays when unacceptable levels of
unbalance are present.
7.2.8 Voltage-operated relays
Attracted-armature or other types of relays capable of being set to operate at

particular voltage levels may be used to detect when supply voltages to motors
or outputs from generators are outside acceptable levels.
7.2.9 Control equipment
Many motors must be provided with control equipment to ensure that the
correct starting sequences are used. As examples, the main windings of three-
phase motors may be connected in star during starting periods and then con-
nected in delta when the speeds are approaching the running value or resistors
may be connected in either the stator or rotor winding circuits during starting.
This equipment, which is often supplied by the manufacturers of machines, is
usually housed in a cabinet or case which is mounted near to the machine it
controls. In these situations protective equipment, such as overcurrent and
field-failure relays, is often also housed with the control equipment.
7.2.10 Applications of protective schemes to machines
It will be clear from the above that there is a range of relays and schemes
available for protecting machines and whilst only one device, such as a fuse,
may provide sufficient cover for a small machine, a combination of devices
and/or schemes is usually needed to protect a large machine. Guidance on the
practices currently adopted is provided in the following sections.
7.3 THE PROTECTION OF MOTORS

As stated earlier, motors represent loads in power systems and therefore their
protective equipments should be selected taking account of the withstand
abilities of their supply networks and the behaviours of the motors themselves.
A number of applications are considered below.
7.3.1 The protection of small motors

There must be many millions of relatively small, single-phase a.c. motors
incorporating commutators and with their armature and field windings con-
nected in series, i.e. series-commutator machines, in service in domestic ap-
pliances, such as vacuum cleaners and washing machines. Their power ratings
are of the order of a few hundred watts and so they draw currents of 3 A or
less when operating from the voltages normally available in homes and offices
(i.e. 240 V).
These motors are switched directly to their supply voltages and draw surges
of current during starting, but because of the relatively low inertias of their
rotors, the starting periods tend to be very short. Standard cartridge-type
domestic fuselinks, rated at 13 A or less, will not operate during these surges
and they are therefore suitable for protecting such motors. They will operate
if a fault occurs in a machine but relatively slow clearance may be obtained
if the fault current is not greatly in excess of the fuselink rating. Further
damage to a motor, which may result from slow clearance, will probably be
unimportant because it will not be likely to greatly increase the cost of the
repair above that which would have been necessary in any event, and even if
complete replacement of the motor is necessary it is not likely that the cost
will be unacceptably high.
The fuses would certainly ensure that cables would not be damaged by faults
on the motors supplied by them. It must be recognized, however, that the fuses
will not operate at low-overcurrent levels, but this is usually acceptable as
such conditions do not often occur with motors used in domestic appliances
because their mechanical loads tend to be limited.
Some applicances, such as refrigerators, incorporate single-phase induction
motors. These usually have a main winding which is continuously rated and
a short-time rated starting winding. These motors are protected by domestic-
type cartridge fuselinks, rated at 13 A in the UK, and these ensure that the
supply circuits will not be damaged by faults within a motor. They will not,
however, protect the starting winding if for any reason the accelerating period
is unacceptably long. To cater for this condition, some manufacturers incorp-
orate thermal devices containing bimetal strips to open the starting circuit
when necessary.
The protection of motors 235
7.3.2 The protection of large induction motors
Large induction motors invariably have three-phase stator windings suitable

for operation at voltage levels up to 11 kV (line). Power ratings up to 10 MW
are available. Such machines may suffer damage from a range of causes and
these are considered separately below.
Stator-winding insulation
All insulating materials deteriorate with age and the speed of this process
increases with the temperature of the insulation. As a result prolonged opera-
tion at high temperature is undesirable. In the case of a motor, heat is generated
as a result of the power loss in the stator windings. Clearly the flow of high
currents for short periods produces relatively small amounts of energy and
only small temperature rises because of the mass of the conductors being
heated. Such conditions, which can occur during starting, can therefore be
accepted. On the other hand, currents above the rated level flowing for long
periods because of overloading can cause the insulation to reach high tempera-
tures which will be maintained until the overloading ceases. Such conditions,
especially if repeated regularly, will significantly damage the insulation,
which could then break down.
The temperature levels at which insulation may be operated safely are
dependent on the insulants being used. The motor industry has established
standards, those associated with induction motors being given in NEMA
MGI-1987 [3].
It will be appreciated that the withstand time/current curve for any motor is
inverse and protective equipment must ensure that the limits are not exceeded.
As stated earlier, unacceptably high-temperature levels may be reached in
the stator of a machine due to defects in the ventilating system, such as
blocked ducts or a faulty fan, and these could arise even though the current
levels were at rated values or less. It is therefore desirable that stator windings
be protected by three-phase overcurrent relays and thermal devices.
Guidance on suitable relay settings is provided in manufacturers' literature
and standards such as ANSI/NFPA 70-1987 [4]. Depending on the design and
duty of a motor, its relay should initiate tripping at currents in the range
115-125% of the rated value of the motor. These relays may have either
thermal elements or induction-type movements.
Relatively small machines operating at voltage levels such as 415 V (line)
and currents up to 20 A may incorporate thermal protectors containing bimetal
elements. These are capable of directly interrupting currents somewhat above
the rated values of their motors when overheating occurs. Larger machines
incorporate temperature detectors embedded in their stators. These detectors,
forms of which were described in section 7.2.3, initiate the opening of their
motor circuit-breakers when necessary.
Faults on stator windings
Conditions which may affect the insulation of the stator windings of induction
motors have been considered and protective arrangements, which may be used
to try to ensure that consequential damage will not result from them, have been
examined. Nevertheless, breakdowns of the insulation of stator windings,
although not common, are experienced and these can be quite serious in
machines operating at relatively high voltages. It is desirable that they should
be cleared quickly and for this reason current-differential protective schemes
are applied to large machines.
As explained in section 5.2.5 (page 154), these schemes may be set to detect
fault currents of magnitudes much lower than the rated current of the protected
equipment. They do not operate if faults occur external to their zone or when
high currents are flowing as a result of overloading and they must therefore
be used in conjunction with overcurrent protective relays.
The application of current-differential schemes to three-phase motors is
clearly affected by the connections of the stator windings, i.e. star or delta.
Star-connected windings may be protected using the arrangement shown in
Fig. 7.2(a). All six current transformers, which will have the same ratios,
should ideally be produced by the same manufacturer and have identical
characteristics. The three transformers on the supply side of the machine
should be mounted on the incoming connections to the switchgear so that the
motor supply cables and the stator windings are in the protected zone and the
other three transformers should be mounted in the machine as near as possible
to the neutral or star-point connection.
The protection of delta-connected windings is not so simple and in practice
several arrangements are used. Six current transformers may be connected as
shown in Fig. 7.2(b) and (c). With the arrangement shown in Fig. 7.2(b) the
six transformers have the same ratios and should ideally be identical. Each
phase winding of the motor is protected by its group of two transformers and
relay, but no protection is provided for the supply cables to the motor or the
connections between the phase windings of the stator.
With the arrangement shown in Fig. 7.2(c), three current transformers are
mounted in the supply circuit to the motor on the incoming side of the switch-
gear. Their secondary windings are connected in star. The other three trans-
formers are connected in the phases of the stator windings with their secondary
windings connected in delta. All the current transformers have the same step-
down ratio and the secondary output current of each of the transformers in the
supply circuits balances with the difference of the output currents of two of
the transformers mounted in the motor under healthy conditions, e.g.
lR2 = Np/Ns(Ja - lab + lea). This arrangement has the advantage that the protected
zone includes the supply cables and switchgear, but the achievement of the
necessary balance is more difficult than with the scheme shown in Fig. 7.2(b)
because the current transformers in each balancing group will normally be
Bias winding~ -1 Operating winding
Current Current
.... ....
transformers Motor transfDmlers
s ~f s ~ f ~
a
I
'N. f
\
I s'N; f
,
s \
- ~
I \
I
s~ f
,,
S f
I
s N. f I
\
5 Ns f
I
S..~f ~
\ I f
Bias ,~Clt / / '"
~ Windi~~__ ~
operati '
winding-1- __ etay
------
t
~--~]
,,,-~ ,
~ __ __ 1
(a) (b)
Current
transformers
Np f
N. f
N" f
.r-"---!--""-iJLi."i"ob
(e)
Current
.N.,,,,
f
,,
I
_ _ _ _ _1
Motor { Retay
windings
(d)
Fig. 7.2 The application of current-differential schemes to three-phase motor stator

windings. (a) Star-connected winding; (b) and (c) delta-connected winding; (d) ar-
rangement suitable for both star- and delta-connected windings.
operating at different points on their excitation characteristics at each instant.
An alternative arrangement, which can be used for both star- and delta-
connected machines, in which the currents at the two ends of each phase
winding are compared directly is shown in Fig. 7.2(d). Both ends of each
winding form the primary windings of the three current transformers. The
secondary windings of each of these transformers are connected to a relay
which will operate in the event of a winding fault.
Rotor windings
Rotors with squirrel-cage windings are very robust and it is not necessary to
provide special equipment to protect them. Wound rotors, however, may have
quite high e.m.f.s induced in them at times, those present at the beginnings of
starting operations being particularly large due to the high relative speed
between the rotating magnetic field and the stationary or slowly-moving rotor.
During such periods, a flashover to earth or at the slip rings could occur. In
this event the voltage of the star point of the rotor winding would be displaced
from that of the ground and this could be detected by connecting a star/open-
delta transformer to the rotor winding, as shown in Fig. 7.3. The output of this
transformer may be used to operate either a voltage-sensitive or a current-
sensitive relay to disconnect the machine.
The sensitivity of this arrangement reduces as the speed of the protected
motor increases because of the consequent reduction in the rotor e.m.f.s.
Operation is not attainable therefore at speeds near to the synchronous value.
This behaviour is usually considered to be acceptable, because flashovers are
less likely when the induced e.m.f.s are low. Should a serious rotor-flashover
occur, however, on a motor running near synchronous speed, reliance has to
be placed on the protection associated with the stator windings.
Stator Rotor
:::3 t--'TIO'-~+-f
windings windings
Stan-delta
transformer
s s
Relay
Fig. 7.3 The detection of rotor flashovers.
Failure to start
An induction motor may fail to start for several reasons including those below:
1. A mechanical fault such as a seized bearing.
2. Excessive load torque.
3. A low supply voltage.
4. An open circuit in a winding.
When a motor is started, the motor is stationary initially and its windings
are cut by the rotating magnetic field set up by the stator. A large e.m.f. of
the same frequency as that of the voltage supplied to the stator is induced in
the rotor, the action being the same as that in a transformer, i.e.
Ns Ns
Es=-Ep=-- Vp
Np Np
in which Np and Ns are the turns in the stator and rotor windings respectively,
Ep and Es are the induced e.mJ.s in the stator and rotor windings and Vp is the
supply voltage. Relatively large currents (/p and Is) flow in the stator and rotor
windings, their relationship being given by:
Np
Is =- Ns Ip
The actual current values in any given application are dependent on the
arrangements used to start the motor, such as star connection of the stator
during starting and delta connection for running, or the progressive removal
of resistance included in the rotor circuit during accelerating periods. In all
cases, the initial current levels are significantly above the rated current, multiples
of three to seven being common. If a motor does not accelerate from standstill
for any reason, these above normal currents will persist and cause unaccept-
able heating. This situation is also aggravated by the fact that the cooling
effect produced by the rotation of the rotor is not provided.
It will be evident that these stalled conditions can be detected by a relay
which operates when the initial motor starting current persists for more than
a certain time. This time must be great enough to allow normal starting to take
place but it must, in addition, be short enough to ensure that conductors and
insulation do not suffer consequential damage. Alternatively, or in addition,
the condition may be detected by thermal devices set in the stator and/or in
the proximity of the rotor and these can initiate the disconnection of the
machine before unacceptable temperature rises are produced. In practice, the
stator of a machine may reach its limiting temperature before its rotor, during
locked-rotor conditions. This is designated stator limiting and thermal devices
should clearly be embedded in the stator of such a machine.
It is necessary that the time for which a motor can remain connected to its
supply with its rotor locked be determined from its manufacturers, to enable
the necessary time delays associated with overcurrent and thermal devices to
be selected. In cases when the permissible locked-rotor time is shorter than
the normal acceleration time, shaft speed sensors may be used to detect this
condition.
Unbalanced operation
Unbalanced operation may occur in several ways. The supply voltages may
be unbalanced and in these circumstances they will contain significant nega-
tive- and zero-sequence components. An extreme condition could occur if one
phase of the supply became open circuited, as shown in Fig. 7.4.
The application of negative-sequence voltages to a motor causes a magnetic
field to be set up rotating in the opposite direction to that set up by the
positive-sequence voltages. As is well known, such a field induces very large
e.m.f.s in the rotor windings of a machine rotating in its normal direction and
consequently large negative-sequence currents flow in both the stator and rotor
windings. In addition, e.mJ.s of nearly double the supply frequency are in-
duced in the magnetic material of the rotor. These conditions can cause quite
rapid heating of machines even though the total input currents are not very
high and, because of this, the operating times of overcurrent relays may be
too long to afford the necessary protection.
i /,...-,
a o~
~__
a"":;"_'-"TI~\T"_oo/c
(000 ..
I
\
4, I
\
b - \
ic \
co--~-~\~~o-~--~
" .... _ / I
Stator
windings
Fig. 7.4 Operation with one phase on open circuit. ia =0, ib =- ie, Iil.1 = Iilb I =
liblNf.
Similar unbalanced effects can be produced by faults such as short-circuited
turns in a winding. In such cases, negative-sequence currents flow because of
physical unbalance and will do so even though the supply voltages to the
machine are balanced.
It is evident that the above conditions cannot all be detected by fitting relays
which detect the negative-sequence components in the supply voltages to
machines and therefore relays which monitor the stator-winding currents and
detect the negative-sequence components in them should be used. Because the
damage which results from unbalanced operation is caused by overheating,
the period for which unbalance may be allowed to persist is dependent on the
magnitude of the negative-sequence current in the windings, the relationship
clearly being inverse. To ensure that relays with suitable time/current charac-
teristics are chosen for particular applications, it is necessary that the appro-
priate withstand-abilities of motors are obtained from the manufacturers.
Overvoltage and undervoltage operation

Standard specifications require that motors be designed to operate satisfactor-
ily at their rated outputs over a range of voltages. As an example, NEMA
MGI-1987 quotes a limit of 10% ofrated voltage. Clearly variations within
such a range will not overstress the insulation of a machine but they do affect
the heating which occurs because different levels of current will flow when
the rated output is being provided.
Outside the specified operating voltage limits a machine could be damaged
if the condition persisted for a significant period. If desired, voltage-sensitive
relays could be provided to monitor the supply to a motor and disconnect it if
necessary. Because excessive heating is caused, however, by the flow of
above-normal currents, the overcurrent relays provided to detect other condi-
tions should also give adequate protection when abnormal voltages exist.
The application of a very high surge-voltage to a motor may cause severe
damage to the insulation of its windings. It must be recognized that overvolt-
age relays are not able to provide protection against such an event, because
they could not operate and cause a motor to be disconnected from its supply
before damage had been caused. It is therefore desirable that suitable surge-
suppression devices or voltage limiters be fitted to the supply network if such
conditions may arise.
Under- or over-frequency operation

The frequencies of the voltages supplied by major networks are closely con-
trolled and the deviations are small, typically less than 1% of the nominal
value. In more localized networks, however, greater variations may occur and
therefore motors are usually designed so that they will operate satisfactorily
from supplies with frequencies in the range of the nominal 5%. Such
variations will affect the speed of an induction motor and therefore its power
output when its load requires a fixed torque. Corresponding changes will occur
in the currents supplied to the motor and the heating within it.
As with the overvoltage condition considered above, the overcurrent relays
should detect this abnormal behaviour and ensure that the motor is not allowed
to operate in a way which may cause it to deteriorate. Although the system
frequency could readily be monitored and motors could be disconnected when
values outside the normal limits were detected, this is not normally considered
necessary.
Co-ordination of the protective equipment applied to induction motors

It will be appreciated that relatively small machines are normally only pro-
vided with simple and cheap protective equipment whereas large machines
operating from relatively high voltages, such as 11 kV, are protected by several
devices and schemes including one operating on the current-differential principle.
Several of the protective devices have a single setting and duty. For
example, a temperature sensor, which will probably be installed by the motor
manufacturer, must operate at a level below that at which damage will occur.
Similarly a current-differential scheme should be set to operate at a level well
below the rated current of the windings being protected, the only constraint
being that it should not be so sensitive that it may operate incorrectly, say as
a result of a starting-current surge. Voltage- and frequency-sensitive relays
have settings which may be selected quite simply.
Overcurrent relays with inverse time/current characteristics have to be set,
however, to perform satisfactorily under several conditions. Firstly they must
not operate when a motor is carrying its rated current and indeed must allow
it to run when slightly overloaded. They should have a minimum operating
current corresponding to the maximum current which the motor being pro-
tected can carry continuously without being damaged, a typical value being
1.2 times the rated current of the motor. At higher current levels, the relays
should operate in times slightly below those for which the motor can carry
them. To ensure that this requirement will be met withstand time/current
curves, of the form shown in Fig. 7.5, should be obtained from motor manu-
facturers if they are not already available. A further requirement is that the
relays should operate in a time less than that for which the motor protected
by them can withstand the locked-rotor condition. It is not possible to directly
check that this latter requirement will be met by relays with a particular
time/current characteristic because the starting currents of motors are not
constant, but reduce from an initial high value in a manner dictated by the
starting arrangements. In practice, therefore, it is usual to assume that the
initial starting current will continue throughout the whole acceleration period
and to check that this would not cause relay operation. This clearly ensures
that a factor of safety will exist.
I
I
IMotor
I starting
characteristic
Locked rotor
current
Fig. 7.5 Withstand time/current curves used in the coordination of protective equipment.
A further requirement is that the relays must not operate during normal
starting periods. This latter requirement must always be achievable because
motors must be so designed that they will not be damaged as a result of
carrying their starting currents for the periods associated with load torques up
to the rated values.
Thermal relays are particularly suitable for protecting motors because their
operation is caused by heating and therefore their operating time/current char-
acteristics tend to be similar in form to the withstand time/current charac-
teristics of motors. When using such relays, it can be advantageous to omit
compensation for ambient temperature variations because the times for which
motors can withstand particular overcurrents are affected by the temperature
of their surroundings, as also are the operating times of uncompensated relays.
It can also be advantageous to omit instantaneous-reset features as relay
operation is then dependent on the current levels over a significant period and
because of this a motor could be prevented from making several normal starts
in a short period, a condition which cannot normally be allowed because of
the progressively higher temperatures it would produce in a motor. Clearly, a
relay which reset instantaneously would not provide the necessary protection
in these circumstances because each motor start would be sensed inde-
pendently and allowed to take place.
A typical arrangement of the protective equipment applied to an induction
motor is shown in Fig. 7.6.
In the past such arrangements were implemented by providing separate
relays to operate for each of the various forms of protection being applied, i.e.
inverse-time overcurrent, instantaneous overcurrent, etc. Whilst this practice
is still adopted today, electronic relays, which process information and are
capable of performing several functions simultaneously, are being produced
[5]. As a result, although the same forms of protection are being applied to
motors, the number of separate relay units is reducing.
xes
VT
Undervoltage relay 27
Iinverse overcurrent 1511
Unbalanced current 46
Fig. 7.6 Schematic arrangement of induction motor protective equipment. (System

device function numbers according to IEEE C37.2-1987.)
Further information about relaying equipment and other protective arrange-

ments are provided in references [6-9].
7.3.3 The protection of synchronous motors

The stators of three-phase synchronous motors are similar to those of induction
motors and they have the same limitations. Clearly the same current-differen-
tial schemes and thermal overcurrent relays may be used to protect them.
The rotors of the two types of motor are, however, dissimilar, the rotors of
synchronous machines having a winding fed with direct durrent.
Some of the conditions considered in section 7.3.2, which can cause induc-
tion motors to suffer consequential damage, can also adversely affect syn-
chronous motors. There are, however, some conditions which are solely
associated with synchronous machines and these are examined below.
Loss of synchronism
Non-synchronous running, which is clearly unacceptable, may be caused by
the factors considered below.
Mechanical overloading of a machine When a machine is running synchron-

ously on no load, its excitation must be such that the e.m.f.s induced in each
of the phases are of almost the same magnitudes and phases as those of the
supply. Slight differences must exist however to drive small currents through
the phase impedances of the stator windings and there must be a small power
input equal to the electrical and mechanical losses. This condition is illustrated
in Fig. 7.7(a). When load is applied to the motor shaft, the output power
.....-
::-V.E.,Z
E
~J'f.X
,
-
I ,E
-
- - --
,, /
/
'/
(a)
(b)
Fig. 7.7 Operation of a synchronous motor.
temporarily exceeds the input power and the rotor decelerates for a period. As
a result, the phases of the rotor e.mJ.s fall further behind those of the supply
voltages, causing the input currents and power to rise until equilibrium is
established and synchronous running is re-established. Such a condition is
shown in Fig. 7.7 (b). At a particular load, which causes the phases of the stator
e.m.f.s to be about n/2 radians behind those of the supply voltages, the limit
of synchronous running is reached and any further increase in load will cause
the motor to fall out of step. In these circumstances, slip-frequency e.m.f.s are
induced in the rotor winding and the deceleration of the rotor increases. The
stator-winding currents vary as shown in Fig. 7.8 and when the rotor reaches
standstill they are very high, the conditions being similar to those in a stalled
induction motor. As the conditions during deceleration cannot be allowed to
persist, it is necessary that they be detected so that the motor can be discon-
nected from its supply.
Faults on the supply network Faults occurring on the supply networks to

which motors are connected probably cause most of the incidents of loss of
synchronism. During faults, the supply voltages are often depressed, and under
such conditions the load torque at which a motor may begin to decelerate may
fall below its rated value. In such circumstances, loss of synchronism may
occur unless rapid fault clearance is effected.
2 r
1 I-
o IV V~ v
-1 I-
I I I I
-2~----~~----~~----~~----~
0.0 0.1 0.2 0.3 0.4
Time (s)
Fig. 7.8 Current variation during asynchronous operation.
Loss of excitation Another common cause of loss of synchronism is under-

excitation of a machine. This may arise because of the opening or short-
circuiting of its field-winding circuit. It is clearly necessary that such
conditions be detected and that the stator windings be disconnected from the
supply. In addition, the appropriate action should be implemented to protect
the field circuit. For example, if a short-circuit or an intertum fault is present
in a field winding it should be disconnected from the supply and connected to
its discharge resistor.
Starting
Synchronous motors, as shown above, can only run satisfactorily at synchron-
ous speed and arrangements must be provided to enable them to be started.
One arrangement employs a small starting motor, often called a pony motor,
which is coupled to the synchronous machine. During starts the synchronous
motor is not connected to its supply and provides no driving power. The pony
motor therefore provides the torque to accelerate the combination up to syn-
chronous speed. When this speed is reached and the stator-winding e.mJ.s of
the synchronous motor are close enough in magnitude and phase to the volt-
ages of the supply sytem, it is connected to its supply. It is subsequently
controlled to provide the required output. The stator-winding currents of the
synchronous motor should not be abnormal during this process and should not
affect the protective equipment. The pony motor must of course be provided
with its own protective equipment.
An alternative starting method which is often adopted is to run up a motor
in the induction-motor mode, special rotor windings being provided for the
purpose, or alternatively the normal rotor windings may be short-circuited.
When the speed has almost reached the synchronous value either the extra
windings are open-circuited and the normal rotor winding is energized with
direct current or the short-circuit is removed from the normal rotor winding
and then it is fed with direct current. In either case, the motor will pull into
step and run synchronously. During such starts, a large current surge flows
initially as the motor accelerates from standstill, this surge clearly being of
the form associated with normal induction motors and then a second smaller
current surge flows as the machine is pulled into the synchronous mode.
In recent years, developments in the field of power electronics have enabled
power-supply units with variable frequency outputs to be produced. Such
units, with relatively low power outputs, are now being used to supply some
synchronous motors during starting periods. The frequency output of a unit is
increased gradually in a controlled way until the motor supplied by it is
running at its synchronous speed and thereafter the motor is disconnected from
the supply unit and connected to the main power-supply network.
In such cases, no high current surges should flow in the motors during
starting periods.
Co-ordination of the protective equipment applied to synchronous motors

As with induction motors, the larger the rating of a synchronous motor, the
more justified are relatively complex and expensive protective arrangements.
Several of the available relays, devices and schemes, described in section
7.3.2, are applied to both types of machine.
Current-differential schemes with low fault-settings may be applied to the
stator windings of large machines to detect short-circuits and faults to ground.
Overcurrent relays with appropriate time/current characteristics are commonly
used to ensure that damage is not caused by prolonged operation under over-
load conditions. Thermal devices are incorporated in the stators of machines
to directly detect overheating and sensors to detect a failure to rotate from
standstill may be used.
In addition, protective equipment to detect asynchronous running and field-
circuit faults is essential on large machines. As stated earlier, asynchronous
operation causes the phase angle between the stator-winding currents and
voltages to vary to levels outside the normal operating range and indeed it
changes continuously as the rotor speed falls. All three phases vary similarly
and therefore a single relay, fed with the voltage and current of one phase,
may be made to operate when the power factor or phase angle is in a range
not encountered under synchronous conditions, say when the power factor is
negative, i.e. when the phase angle is in the range 1t/2 to 31t/2 rad. Further
information is provided on this topic in section 7.4.2 (page 261).
d.c. supply
Rotor winding
Slip rings
Current
transformer
s Np f
c;J Relay
Fig. 7.9 Method of detecting asynchronous operation.
Alternatively, or in addition, the field-winding current can be monitored to

detect the alternating-current components which are present when machines
are running asynchronously. This is done by including a current transformer
in the field circuit, its secondary winding being connected to a sensitive
current-operated relay as shown in Fig. 7.9. It must be recognized that one
cycle of alternating current flows in the field winding for a slip of one pole-
pair pitch, i.e. for one revolution of slip in a two-pole machine. The frequency
of the alternating current is therefore very low when non-synchronous running
begins and the current transformer must be able to reproduce these slowly
changing or low-frequency currents so that relay operation will occur quickly
after synchronous operation commences.
It is vital that adequate current should flow in the field winding of a motor
to maintain synchronous operation and therefore either the supply voltage to
the field circuit or the current in it should be monitored by a relay which can
disconnect the stator windings from their supply if the excitation becomes
inadequate. Current-sensing relays, which must be fed from a shunt in the field
circuit, have the advantage that they will operate in the event of an open circuit
in the motor field circuit as well as for a loss of the supply voltage.
Protection should also be provided to detect faults associated with the rotor
windings of relatively large synchronous motors. Rotor circuits are not norm-
ally earthed and therefore a single fault to ground does not affect the operation
or cause further damage. The condition may be allowed to persist but it is
desirable that an indication be provided to the operator so that action may be
taken as soon as the motor may be conveniently disconnected. One method
which is employed is shown in Fig. 7.10. It employs two lamps connected in
series between the supply connections. Under healthy conditions, both lamps
have the same voltages across them and they should therefore be equally
bright. Should a fault occur on the field winding at any point other than the
+o-----~------------_,
Power Synchronous-motor
supply field winding
Fig. 7.10 The detection of earth faults on a field winding.
centre, unbalance will occur causing one lamp to be brighter than the other,
because of the unequal voltages across them. As an alternative, the lamps may
be replaced by two equal resistors and a voltmeter may be connected from
their junction to earth.
Should it be considered desirable to disconnect a motor in the event of a
single earth fault on its rotor winding then the second arrangement described
above may be used, a voltage-operated relay being included rather than a
voltmeter. This scheme has the slight disadvantage, already referred to, that
operation cannot be obtained for faults near the mid-point of a winding.
Alternative schemes have been developed to eliminate this limitation. One
simple arrangement, shown in Fig. 7.11, includes a varistor in series with one
of a pair of divider resistors. Again there is a point on the field winding at
which an earth fault would not cause the relay to operate. Should the supply
voltage subsequently change, however, the fault would then be detected, be-
cause the varistor resistance would have altered and the fault would not then
be at the point on the rotor winding at which operation would not now be
obtained. These and other possible arrangements are discussed later in section
+o-----~------------_,
Power Synchronous-motor
supply field winding
Fig. 7.11 The detection of earth faults on a field winding.

7.4.2 (page 261). If a single earth fault is allowed to persist on a rotor winding
and then a second fault occurs, a short-circuit condition will exist. Such a
condition could cause a high current to flow in the rotor circuit and this should
be detected by the protective equipment associated with the exciter. This
equipment should then not only cause the opening of the rotor circuit but also
that of the stator as otherwise asynchronous running would ensue. It is also
probable that the protective relaying associated with the stator circuit will
operate when two faults are present on the rotor winding because the excita-
tion is likely to be significantly reduced causing high currents in the stator
windings.
Whilst temperature sensors can be fitted into the stators of machines and
their outputs can readily be connected to monitoring equipment, it is not
practicable to detect overheating of rotors in this way because extra slip rings
would have to be provided to enable the necessary connection to be made.
Consequently rotor temperatures were not monitored in the past. Recently,
however, a relatively simple method, based on the resistances of rotor wind-
ings, has been developed to detect rotor overheating. Clearly the resistance of
a winding is dependent on its surroundings and the current it carries. It will
therefore rise above the value it would have when carrying a given current, if
the adjacent rotor material is at an above-normal temperature. Relay operation
can therefore be initiated either when the resistance of a winding exceeds a
particular value or when it exceeds a value dependent on the current in it.
The protective equipment associated with the stator windings of synchron-
ous motors is generally similar to that applied to induction motors, as stated
earlier. Details of this were provided in section 7.3.2.
An extra feature which is required with synchronous motors is an out-of-step
relay to detect loss of synchronism. Such relays compare the phase of one of
the stator currents with a supply voltage, an example being shown in Fig.
7.12(a).
Voltage Current Loss of field

transformer
U
transformer
a
d
b
c
Power
supply .....-_ _-, Field
Field winding
pplication
relay
(b)
(a)
Fig. 7.12 The detection of asynchronous operation.

A typical arrangement of the equipment used to control and protect the field
windings of large synchronous motors is shown in Fig. 7.12(b). Included in
this arrangement is a field application relay, the duty of which is to initiate
the closing of the field circuit at the appropriate instant towards the end of
starting periods. This relay monitors the slip-frequency voltage across the field
discharge resistor and when its frequency is low enough and its instantaneous
value is appropriate the main field circuit is closed, after which the motor is
pulled into synchronism.
7.3.4 The protection of d.c. motors

Direct-current motors have stationary magnetic fields produced by field wind-
ings mounted within their stators. Armature windings mounted in the rotors
are fed via commutators. They may, depending on their duties, have their
armature windings connected in series with their field windings (series-
connected) or the windings may be connected in parallel (shunt-connected).
In addition, shunt-connected machines may have extra field windings con-
nected either in series with the armature windings or in series with the supply,
to provide compounding.
Normal operation
Alternating e.m.f.s are induced in the conductors of an armature winding as
they rotate in the constant magnetic field set up by m.mJ.s provided by the
field windings. As a result of the commutator action, a direct e.m.f. is effect-
ively present between the brushes, its magnitude being dependent on the
current in the field windings.
This so-called back e.m.f. is almost equal to the applied voltage in a well-
designed machine. As a result, the speed of a shunt-connected machine with
a fixed field current remains almost constant for all normal loads. Lowering
of the field current causes such a machine to run at a higher speed, so that the
necessary back e.m.f. can be produced.
The field currents of series-connected machines cannot be separately ad-
justed and they are governed by the mechanical load applied to them, because
the greater the load, the greater the motor current. Increase of load therefore
increases the field and for a constant e.mJ. to be produced the machine speed
must fall.
Compounding enables intermediate load speed characteristics to be obtained.
More detailed information on the normal behaviour of these machines may
be obtained from reference [10].
Causes of failures in d.c. motors

Breakdowns and failures may occur from a range of causes and these are
considered separately below.
Overspeeding Operation at very high speeds can cause very high stresses in
the rotating parts of a machine and it may therefore result in mechanical
failures. It is clear from section 7.3.4 (page 251) that this undesirable condition
will arise if the mechanical load on a series-connected machine is reduced to
a very low level. In the case of a shunt-connected machine, overspeeding will
occur if the current in its field winding is reduced below a certain level.
Overloading Mechanical overloading of all d.c. motors causes them to draw

currents above their rated values and this results in the temperatures of their
armature windings and any series-connected field windings rising. Should
such conditions be allowed to persist for long periods then temperature levels
will be reached which will cause the insulating material around conductors to
deteriorate.
The permissible periods for which particular degrees of overloading may be
permitted depend on the designs of the machines and the operating conditions
prior to the overloading.
The extreme condition of stalling, during which very high currents flow,
must not be allowed to persist.
Undervoltage operation A motor operating at its rated load will run at re-
duced speed and draw currents above the rated value if the supply voltage is
below its normal level. Such a condition clearly may not be allowed to persist.
Overheating Overheating may not only occur as a result of overloading but

also because of inadequate cooling due to factors such as blocked ventilating
ducts. Clearly mechanical defects of these types do not affect the electrical
behaviour initially, and input currents do not rise, but if the conditions are
allowed to persist insulating materials could then be damaged.
Commutator faults Severe sparking may occur due to unsatisfactory commuta-

tion and this can lead to arcs forming between the positive and negative
brushes. In these circumstances the armature is short-circuited and very high
currents will be drawn from the supply. Should the condition be allowed to
persist, severe physical damage would be caused, particularly to the commutator.
Unsatisfactory starting Manually-operated starters are used with many rela-

tively small machines and automatic starters may be used with large machines.
Should the resistance be reduced too rapidly during starts, excessive currents
will flow leading to possible physical damage.
The application of protective equipment

Relatively small machines may be protected by fuselinks in the main supply
connection. As stated in section 1.9.8 (page 28), special fuselinks capable of
withstanding starting-current surges are produced for this application. Gener-
ally, separate fuselinks are not included in the field-winding circuits of shunt-
connected machines.
Overcurrent and inadequate field-current protection is provided on many
machines, it being commonly incorporated within the starting equipment.
Overcurrent relays or devices are usually set to operate if the input current
rises above 1.3 times its rated value and, if necessary, operation may be
delayed to ensure that it will not occur during starting-current surges. Relays
or devices to detect inadequate field-current conditions are arranged to dis-
connect machines from their supplies.
Relays are often provided to monitor supply voltages and initiate the dis-
connection of machines should the voltages fall below acceptable levels.
Speed-measuring devices may also be provided to detect both overspeeding
and operation at unacceptably low speeds, including stalled conditions. Opera-
tion of these devices also causes their machines to be disconnected.
It is essential that circuits should be so arranged that starting equipment will
be reset whenever machines are disconnected from their supplies, so that the
correct sequence will be followed when machines are restarted.
Direct-current motors fed from rectifiers

As a result of the development of power-electronic equipment, electronically-
controlled rectifiers are being increasingly used to supply direct-current mo-
tors. Rectifier/motor units in a range of ratings are produced and it is necessary
that each unit be adequately protected. A typical arrangement of the protective
equipment applied to such units is shown in Fig. 7.13.
Overcurrent and overvoltage protective devices are provided within rectifier
housings. A fuse may be included in the supply circuit to a unit or alternatively
a circuit-breaker which can be tripped by an overcurrent relay may be pro-
vided. Fault clearance must be effected rapidly enough to ensure that the
D.C. circuit
1----1 breaker
o-ix~
X
Three-phase ~ ~Converterl
supply - IXI ____ 1 1
[ x~
~ r--~~~~I 1
L_
A.C. circuit ~ ____ 1
breaker or
fuses
Fig. 7.13 The protection of a motor fed from a converter.

let-through energy (I 2 t) does not exceed that which the equipment can with-
stand, typical maximum operating times being in the range 5-100 ms. Induc-
tors are often included in units to limit fault currents to acceptable levels.
Signals from current sensors and overcurrent devices are fed to the elec-
tronic equipment in the rectifiers to control their output currents.
Should it be possible for a drive to operate in a regenerative mode at times,
a d.c. circuit-breaker must be provided which can operate in the event of a
short-circuit occurring in the rectifier.
7.3.5 The protection of variable-speed drives

In recent years, both induction motors and reluctance motors fed from elec-
tronically-controlled supply units have been used to provide variable-speed
drives. Because of the diversity of the various arrangements which have been
developed it is not possible to consider the protection of them in a general
way in this work.
Clearly, in all cases, relays and other devices must be provided to detect
abnormal conditions in the motors, such as overheating and faults within
windings. Protective equipment for the supply units, however, may vary great-
ly, being dependent on their individual designs. To provide an illustration, the
protection of the supply unit shown in Fig. 7.14 is considered below.
During normal operation, the incoming three-phase supply is rectified and
fed to the inverter to produce a variable-frequency alternating supply to the
motor. A large inductive reactor (L) is included between the rectifier and the
inverter to ensure that the current fed to the inverter (Ide) cannot change
rapidly.
During faults in the inverter or the motor, the current can be brought to zero
quickly by adjusting the firing pulses to the rectifier and by opening the
circuit-breaker.
D.C. link
Three-Phase:X
supply
x
~X--+--+-"'"
Circuit
breake
sensor
Fig. 7.14 The protection of a supply unit of a variable-speed drive.

The protection of alternators 255
If the drive operates in the regenerative mode at times or a fault occurs in
the rectifier, the current fed back by the motor can be limited by adjusting the
firing pulses supplied to the thyristors in the inverter.
Current sensors may be included in the connections from the a.c. supply, in
the link between the rectifier and the inverter or in the motor circuit, but for
those applications where regeneration may occur satisfactory performance will
only be obtained using sensors in the last two positions.
More detailed information about particular electronic-drive systems and
their protection is provided in reference [11].
7.4 THE PROTECTION OF ALTERNATORS

Alternators are basically similar to synchronous motors but they have much
more other plant directly associated with them and they are produced with
power ratings ranging up to much higher levels. All alternators are driven by
prime movers, which at the lower power levels may be reciprocating diesel or
petrol engines. At the high power ratings, however, they are invariably driven
by water or steam turbines. In the past it was common to install a group of
boilers which together supplied a number of turbines, this arrangement being
referred to as the range system. In most installations today, however, complete
generating units are produced consisting of the water supplies or steam-raising
plant, the turbine and alternator. In addition, as stated in the introduction to
this chapter, it is the practice in the large generating stations to have a power
transformer directly connected to each alternator to step up its output voltage
to that of the transmission system. There are many auxiliaries, such as fans
and pumps, associated with each generating unit and the supplies needed by
them are usually supplied by a second transformer which is also connected
directly to the alternator, this transformer being referred to as a unit trans-
former. A typical arrangement is shown in Fig. 6.1 (page 180).
It will be appreciated that the items of plant in a generating unit are inter-
dependent and therefore they must all be monitored to provide a co-ordinated
protective scheme which will operate appropriately when any abnormality
occurs. Important factors which must be taken into account when selecting
protective equipment are examined in the following section.
7.4.1 Alternator construction and behaviour

Alternator construction
Alternators, except those of very low ratings, are invariably three-phase. The
main output windings of a machine are mounted in slots in the stator core and
the field windings, which are fed with direct current supplied by an exciter,
are mounted in the rotor slots. The exciter may in tum be fed by a pilot exciter,
the main machine rotor and the exciters being mounted on the common shaft
driven by the turbine. To assist in maintaining synchronous running, damper
bars or windings are also fitted in the rotor.
The rated stator currents of large attenuators are very high; for example, the
rated current of the 500 MW, 22 kV machines installed in several generating
stations in Britain is 22 kA. The stator windings of such machines, which may
be made up of two or more windings in parallel, are invariably connected in
star.
Earthing of stator windings

The neutral points of alternators are usually earthed to ensure that excessive
voltages are not likely to be present between their stator windings and the
surrounding core. Should a fault nevertheless occur between a winding and
the core, and a large current flow as a result, then severe arcing could melt
the core locally and cause laminations to become welded together. The con-
ductors affected might be replaced at a relatively low cost but the necessary
repairs to the core, to ensure that excessive heating would not occur in it
during subsequent normal operation, could be extremely costly.
To avoid this situation it is usual to include impedance in the neutral
connection to earth to limit the magnitudes of the earth fault currents which
may flow. In some installations, earthing resistors, which will limit earth fault
currents to the rated current of the machine, are used. On very large machines,
such as those referred to earlier, however, the rated currents are very high and
earth faults of these magnitudes would certainly cause extensive damage to a
stator core. It is the standard practice in such cases to include earthing resistors
which limit earth fault currents to maximum levels, 200 A being a commonly-
used value. In such instances even this value has been considered to be
excessive and higher values of earthing resistance have been used with a few
machines.
It will be obvious that limiting earth fault currents to very low values in the
above way requires the use of resistors of high ohmic value in the neutral
a b c
AlternatorI ..-
,
\
windings " ..- /
Earthing
transformer
Fig. 7.15 Method of earthing an alternator.

connections, say about 4000 n to limit currents to 5 A and such resistors may
not be very robust. As an alternative, earthing transformers of 5-100 kVA
rating are used in some installations, the connections being as shown in Fig.
7.15. In the extreme case of an earth fault on one of the alternator output
terminals, its phase voltage would be present across the primary winding of
the earthing transformer. Under this condition its secondary winding output
voltage would be typically in the range 100-500 V and this is applied across
a resistor. The fault current can thus be limited to a low value using a robust
resistor of relatively low ohmic value, i.e. reduced from the value needed with
direct connection by the square of the earthing-transformer ratio. Further
information on this method of earthing is provided in reference [12].
Interphase and interturnfaults on stator windings

In the previous section, methods used to limit the magnitudes of earth faults
on stator windings were considered. Such techniques, however, would quite
obviously have no effect on interphase and interturn faults which could cause
large currents to flow.
Because of the physical distribution of the phase windings around the stator
cores of machines, interphase faults are unlikely and can only occur on the
end connections of stator coils and in slots which contain the sides of two coils
belonging to different phases. Intertum faults are also quite rare. Should any
of these faults occur, however, it is likely that the resulting damage will cause
the faults to develop and quickly become earth faults.
Faults on rotor windings

The field circuits of alternators, like those of synchronous motors, are norm-
ally unearthed and therefore a single earth fault on a rotor winding has no
effect on the performance of the machine. It may be allowed to persist and
indeed some machines have been allowed to continue operating in this condi-
tion for lengthy periods.
However, it is clearly desirable that an indication should be given of the
presence of such a fault, so that remedial work may be undertaken at the
earliest convenient opportunity.
It must be realized, when a decision is taken to allow a large alternator to
remain in service with an earth fault on its rotor winding, that considerable
consequential damage could occur if a second earth fault then developed on
the winding. In such a situation a parallel path would be created in which a
large current could flow and this could damage both the conductors and the
rotor core material. In addition, the magnetic field then produced by the rotor
might no longer be sinusoidally distributed around the periphery and it might
also be asymmetric, the flux densities at corresponding points on the poles
being different. This could have serious consequences because the attractive
forces existing between the poles and the stator core could be unequal and
vibrations could be set up during each revolution of the rotor. This behaviour
could cause severe physical damage, particularly to components such as bear-
ings.
Intertum faults rarely occur on rotor windings, but should they do so it is
very probable that they will develop rapidly and become earth faults.
Inadequate excitation
The output of an alternator connected to busbars is dependent on its excitation
and the power supplied by its prime mover. For a given power input the power
output per phase must be almost constant, i.e. I cos q> must be almost constant
and therefore the current locus may be taken to be a straight line vertical to
the phase voltage (V), as shown in Fig. 7.16. With relatively weak excitation,
the phase e.m.f. (E) is lower than the terminal voltage (V) and as a result the
machine operates at a leading power factor. As the excitation is varied, the
locus of the e.m.f. must be the straight line shown in Fig. 7.16. It is evident
that synchronous running will not be maintained if the excitation is reduced
below that required to produce the limiting e.m.f. (Ed at point A on the locus.
Such situations should not normally occur, but they could arise because of
faults in exciters or excitation circuits.
Asynchronous operation
Insufficient excitation considered above can lead to a machine running at a
constant speed, above the synchronous value as an induction generator, the
zC)1v
-E
Locus of current tip I
/ -
Locus of e.m.f. E
/
I X COS!P1 =const.
Fig. 7.16 The operation of an alternator. (Note: The resistances (R) of the windings
of alternators are very much lower than the inductive reactances (X) and they have
been neglected in the figure. In practice the locus of E slopes slightly from the vertical.
Two operating conditions are shown, designated by the suffixes 1 and 2.
main field then being produced by the reactive components of the stator-wind-
ing currents. In this mode, e.mJ.s are induced in the rotor core, and in the
damper bars and the slot wedges, causing significant heating. Such operation
cannot be allowed to continue indefinitely, the permissible periods being of
the order of a few minutes. The precise withstand times, which are dependent
on the machine design, should be obtained if necessary from the machine
manufacturer.
An alternator with a healthy excitation system may also fall out of step and
operate asynchronously because of faults on the network being supplied by it.
For example, an interphase fault on a transmission line may cause the terminal
voltages of a machine to be depressed and the power transferred to the loads
may then be severely affected. As a result of the e.m.f. phasors moving relative
to the terminal voltages, the magnitudes and phases of the stator-winding
currents may vary widely and unacceptably. The machine may not return to
synchronous operation when the fault has been cleared, in which event it must
be disconnected and then reconnected after completion of the synchronizing
procedures.
Unbalanced loading
In the case of motors, unbalanced operation only normally occurs if the supply
voltages are unbalanced, whereas with alternators it is normally caused by
unbalanced loading of the network supplied by the machine or by faults.
An alternator, like a synchronous motor, overheats quite rapidly if it carries
even relatively low currents of negative sequence, because such currents set
up a magnetic field rotating in the opposite direction to that of the rotor. They
therefore induce significant e.m.f.s at double the operating frequency in both
the rotor core and winding.
The amount of negative-sequence current which a machine can carry de-
pends on its construction and factors such as the cooling methods it employs.
Because unbalanced loading may continue for long periods, machines are
assigned continuous negative-sequence ratings which are expressed as per-
centages of their continuous ratings. These values, which should be obtained
from machine manufacturers when selecting protective schemes, are typically
in the range from 10% to 15% for turbine-driven alternators with cylindrical
rotors.
During fault conditions, large negative-sequence currents may be provided
by an alternator. As examples, an interphase fault current, say phase a to phase
b, has a negative-sequence component with a magnitude of 57.7% of that of
the fault current and a single-phase to earth fault has a negative-sequence
component equal to one third of the fault current. The magnitudes of these
component currents, which can be quite large, may be calculated using the
sequence impedances of machines which are to be protected.
Type Continuous
Type and Curve Ilt
of
cooling medium
lit rating
number value
machine % F.L.C.
Turbo Direct
alternator hydrogen 1 10 7
30 Ib I in 2
Turbo Conventional
30 Ib I in 2
Turbo Conventional
151b/in2
Turbo Conventional
alternator air or hydro~en 4 15 20
0.51b I in
Typical
salient pole Conventional 5 40 60
machine air
200
100
80
60
40
30
U)
1J
c:
20
0
0
CD
~
CD 10
E 8
i=
6
4
3
2
1.0 5
0.8
0.6
4
0.4
3
0.3
2
0.2
(\J ('") q- (0 coo o ~ 0 0 00

c:i c:i c:i c:i c:i"': N C'i..t <ci <Xi c:i
Negative phase sequence current

(multiples of full load current)
Fig. 7.17 Typical negative sequence current withstand levels of alternators with dif-
ferent forms of cooling. (Reproduced from Protective Relays - Application Guide, 3rd
edn, GEC Measurements, 1987 with the permission of GEC Alsthom Protection and
Control Ltd).
Because fault durations are short, it may be assumed that little heat is lost
by a machine while they are present. In addition, it is the heating caused
by these currents which may cause damage and therefore it is the input
energy which must be limited. As a result, the time integral of the square of
the negative-sequence current must be kept below a particular value, i.e.
Is2 t ~ constant. A typical withstand time/current characteristic for a hydrogen-
cooled machine is shown in Fig. 7.17.
Mechanical equipment
Alternators are electro-mechanical structures subject to mechanical stresses.

Large machines are reliant on gas and/or water cooling systems which must
perform correctly, and action must be taken quickly if they become defective
in any way. Similarly prime movers, such as turbines, and other plant which
form part of an integrated generating unit are individually complex and may
not operate satisfactorily or safely because of a range of defects within them-
selves or their auxiliary equipment. Consequently, a large number of monitor-
ing or protective devices are incorporated in such plant during manufacture
and erection and they must function in conjunction with the protective equip-
ment associated with the alternators.
7.4.2 The application of protective equipment to alternators

Small single-phase alternators, such as those fitted in motor-cars, may be
protected by fuses fitted in the field and main winding circuits. The fuse in
the main winding circuit will operate for overloads or faults to earth on the
main winding whilst the fuse in the field circuit would operate for faults in- its
circuit. Machines of this type do not operate synchronously and loss of excita-
tion merely causes loss of output voltage and therefore a reduction, possibly
to zero, of the output current. Thermal devices may also be embedded in such
machines to disconnect them in the event of overheating.
With larger machines and certainly for those which are to run in synchron-
ism on a network and have three-phase outputs, fuses do not represent a
satisfactory form of protection. The operation of one of the output fuselinks
would produce unacceptable unbalanced operation and operation of a fuselink
in the excitation circuit would result in asynchronous running. In these
applications co-ordinated protected equipment should be provided to cover the
whole unit, including the alternator, driving machine and excitation circuit.
The cost of such units is usually very considerable and therefore the use of
several forms of protection is justifiable and, of course, very complex schemes
are not only justifiable but essential on very large machines such as those
installed in major generating stations. The various forms of protective equip-
ment available for application to alternators are considered below.
Current-differential schemes
In some installations, alternators are connected directly to busbars feeding
distribution circuits, a typical arrangement being shown in Fig. 7.18(a). In
such cases, six current transformers, which should ideally be identical, should
be installed, three in the connections between the windings and the neutral
point and the other three in the output connections. Biased relays may be used,
as shown in Fig. 7.18, to enable the sensitive settings, needed to detect small
Current Current
transformers Alternator transformers
N. windings N. Busbars
.-...!.OO~ f 000
s oo~ f
rg 0N.s0 f ~
S Ns f
F- - N.
!. oo~ f
s 0N.0 f
s
000
N.
s oo~ f
sO/./s f
-
N. N.
!. oo~ f 000
s oo~ f
Bias
s 0N.0 f 0.00 f
s windi~g~_ -:, s Ns
000 ,
Operating' 000
,Relays
winding--L-
L J
, 000
,
,,
000 :1
L __ J
, r~ ,
------
000 000 ,
I
L __ iO
~ __ J
(a)
"iii
E 20
~
CD Trip
;:-
'6i
CDt:::
~::J 10
,::10 Non-trip
c:
~
CD
a. o 2 4 6
Per unit through current
(b)
Fig. 7.18 The protection of an alternator supplying distribution circuits.

earth fault currents to be obtained. The relays, which may be of the induction
pattern, usually have a relatively low bias, say 5%, i.e. as shown in Fig.
7.18(b).
Alternatively, high-impedance relays may be used, their operating voltages
being set high enough to ensure that operation would not occur even if the
core of one of a pair of the balancing current transformers was continuously
saturated whilst the other remained unsaturated. This condition, which. was
examined in section 5.6 (page 171), becomes more difficult to satisfy as the
resistances of the interconnecting conductors increase. In applications associ-
ated with machines, however, the interconnections are usually quite short and
acceptable settings can readily be obtained. As mentioned earlier, it is usually
necessary to connect a non-linear resistor in parallel across the operating
windings and the series resistors of these relays to limit the voltages which
might otherwise be present across them during internal faults.
These schemes do not detect interturn faults but they will certainly operate
in the unlikely event of interphase faults on a stator winding because there is
usually a very low impedance in the paths of such faults. As stated earlier,
earth fault currents are limited by resistors included in the neutral connections.
These usually have high ohmic values and therefore the current magnitude
during an earth fault on a winding will vary almost linearly with the voltage
at the fault position and thus its distance from the neutral point. It must be
accepted therefore that faults near the neutral point will not be detected. As
an example, if the current which will flow when an earth fault is present at
the output terminal of a phase winding is limited to 200 A by the earthing
resistor and the relays are set to operate for currents of 20 A in their current
transformer primary windings, then the 10% of the winding adjacent to the
neutral point would not be protected. This limitation has not been considered
to be serious in the past but recently schemes capable of protecting all of the
stator windings have been investigated. One type detects earth faults by
examining the harmonic voltages produced by alternators whilst another is
based on the injection of additional relaying signals [13-15].
Earth fault protection

Earth fault protection may be applied to supplement that provided by a cur-
rent-differential scheme. This can be provided by mounting a low-ratio current
transformer over the neutral connection of a machine and feeding its output
to a relay, as shown in Fig. 7.19. Because this arrangement does not involve
balancing of currents, the relay may be given a very sensitive setting.
If this form of protection is applied to an alternator which is directly con-
nected to its loads, operation will be obtained for faults at any point on the
network and discrimination must then be achieved with other protective equip-
ments by introducing an adequate time-delay before the alternator is discon-
nected or shut down.
Alternator stator
windings
p----o~~-----------oa
~--~o------------ob
Current ~--~~---------------oc
transformer
~ Earth-fault
~relay
Fig. 7.19 Sensitive earth fault protection.
In applications associated with alternator-transformers, the earth fault relay

may, however, initiate immediate action when the transformer windings con-
nected to the alternator are delta-connected. In this, the usual arrangement,
relay operation will only occur for earth faults associated with the alternator
and the primary windings of the transformers.
Some machines have only four bushings, rather than six, three being used
for the output connections from the phases of the stator winding and one for
the neutral connection, the neutral point being within the casing. In such cases,
four current transformers interconnected as shown in Fig. 7.20 may be in-
stalled to provide earth fault protection. High-impedance or biased relays are
used in this application and again it must be accepted that faults near the
neutral point will not cause operation but nevertheless with sensitive settings
most of each phase winding can be covered. Clearly this protective arrange-
ment will not detect interphase or interturn faults, but this may not be regarded
as unacceptable because such faults occur so infrequently.
s Np
r---~~------~~----oa
Alternator
stator
r-;---..,..._-I S Ns
Relay
Fig. 7.20 Earth fault protection of the stator windings.

Many alternators, particularly those with high output ratings, operate in
conjunction with transfonners. As stated in the introduction to this chapter, it
is the practice in large generating stations to directly connect each alternator
to a step-up transfonner which then feeds busbars operating at the transmission
voltage level, for example 400 kV. The transfonners almost invariably have
their low voltage windings connected in delta and their high voltage windings
in star with the neutral point solidly earthed. It is also usual to connect a
second transfonner with a lower rating to the output connections of each
alternator. This transfonner steps down the voltage to a level suitable for the
supply of the other plant associated with the alternator. Figure 6.1 (page 180)
shows the electrical layout of an alternator and its two associated transfonners
which is common in large power stations. The step-up and step-down transfonners
are usually referred to as the 'main' and 'unit' transfonners respectively.
Because the three items, the alternator and the two transfonners are inter-
dependent, circuit-breakers are not installed in the connections between them
and all three are usually included in the zone of the current-differential scheme
which is invariably installed to protect them. The scheme therefore operates
from nine current transformers, as shown in Fig. 7.21, and balance has to be
Main transformer
Current Alternator Output Current
transformers 22 kV(line) 220 kV(line) transformers
s NPPf SOOOA Np SOOA N. s NSPf
r-~~~--~~~~----------------~~-' ~-T~~~~----~a
r-~======::;::===::::;I""' Unit transformer

Output 11 kV(line)
100A
Fig. 7.21 The protection of an alternator and its associated transfonners (the relay for
one phase only is shown).
To obtain balance Npp = Np Nsp = Nup Nusp
Nps Ns Nss Nus N uss '
effected in each phase, during healthy conditions, between the outputs of three
transformers of different designs and ratios. For the rated values in the
example shown in Fig. 7.21, the current transformers associated with the
alternator, main and unit transformers could have ratios of 5000/5, 500/5/...J3
and 10 000/5/...J3 respectively and it will be noted that they must be connected
in star and delta configurations to allow for the connections of main and unit
transformers. This situation is clearly similar to that examined in section 6.3.2
(page 198) of the previous chapter except that in the case of power transformer
protection there are normally only two sets of balancing current transformers.
As in those applications, either biased or high-impedance relays may be used
in schemes protecting alternator-transformer units and, because of the import-
ance of these units, it is vital that their protective schemes should perform
correctly at all times.
It is essential, first, that the settings of the relays should be such that they
will not operate in the event of faults on the networks connected to the
secondary (output) windings of the main and unit transformers. To enable this
to be achieved, fault calculations should be done to determine the most ex-
treme conditions which might occur. It should be recognized that the current
transformers in the balancing groups will normally all be operating at different
points on their excitation characteristics at each instant and therefore their
transformation errors will not balance out. If high-impedance relays are to be
used, their settings should be determined assuming that the core of one trans-
former will be continually saturated during external faults while the others are
producing their ideal secondary currents.
Having selected the relay settings, the percentages of the various windings
on which earth faults will not cause relay operation should be determined, in
a similar way to that described above, to ascertain that they are sufficiently
low to be acceptable.
Similar checks should be made if biased relays are to be used, to determine
that the optimum percentage bias is selected.
In the previous chapter it was pointed out that current-differential schemes
should incorporate features such as harmonic-biasing, to prevent them oper-
ating during the exciting current surges which may occur when a power
transformer is initially energized. Exciting current surges are less likely to
occur, however, when alternators and transformers are directly connected
together because during run-up the voltage applied to the primary windings
of a transformer increases as the exciting current of the alternator increases,
i.e. there is no sudden application of a high voltage. When a unit is sub-
sequently synchronized to the busbars, the transformer secondary voltages are
made almost equal in magnitude and phase to those of the busbars and there-
fore, when the circuit is closed, any associated current surge will normally be
very small.
It is, however, possible that significant exciting current surges may occur
as a result of faults on the network connected to a transformer. For example,
should a short-circuit occur on the network fed by a transformer at a point
Transformer
Y~X~x_
Duration of fault
011( ~
o----+---~------------------------
-ci>pk
Fig. 7.22 Conditions caused by a fault.
close to it, the secondary voltage could collapse to zero. If this happened at
an instant when the voltage had been small, the transformer core flux would
have been near its maximum value. If the fault was subsequently cleared at a
time when the voltage would have been at its maximum value, i.e. a current
zero in an inductive circuit, then a further flux variation could be required
which would take the core into saturation, thus causing an exciting current
surge. This condition is illustrated in Fig. 7.22. Because of the possibility of
such events occurring it is desirable that the current-differential protective
scheme should have the same features as those used to protect large power
transformers which are not associated directly with alternators.
A further limitation of overall current-differential protective schemes is that
they may not detect faults in the unit transformers. Such transformers have
relatively low ratings and therefore have impedances which restrict the cur-
rents which can be fed to faults on their windings. Because of this, unit
transformers are often protected by their own schemes.
Schemes to detect interturn faults

As with large synchronous motors, it is common for alternators with high
volt-ampere ratings to have stator windings with two or more paths in parallel.
In such applications it is desirable that a current-balance scheme be included
as part of the overall protective equipment. This scheme, which is illustrated
in Fig. 7.1, operates when the total current in a winding is not being carried
equally by the parallel paths. It can therefore detect faults, such as intertum
faults, to which current-differential schemes are insensitive.
/ - "Generator stator winding

~~~~-------- __~--~a
\
~~TP_Ir---~--+---Ob
I
I
Voltage
s~s transformers
Ns Ns Ns
f f f
Zero-sequence voltage
operated relay
Fig. 7.23 Method of detecting intertum faults in a winding.
Alternators with single-path stator windings clearly cannot be protected in

the above manner and, in the past, equipment was not provided to detect
interturn faults on such machines. This practice was considered acceptable
because these faults occur very infrequently and many of them develop quick-
ly into faults involving earth. Serious damage can, however, result from an
uncleared interturn fault and it is certainly desirable that they should be
detected and cleared quickly. A scheme which is now widely used to do this
determines the zero-sequence component present in the voltages across the
three phases of a machine. This component will be present when an interturn
fault is present, but it is normally insignificant during healthy conditions. The
scheme may therefore be implemented by connecting the primary windings of
three voltage transformers on a five-limb three-phase transformer across the
phases of a machine as shown in Fig. 7.23. The zero-sequence voltage is then
derived by connecting the three secondary windings in series, i.e. in open
delta.
It must be recognized that a zero-sequence component may be present in
the voltages of a machine when earth faults are present on the network con-
nected to it and therefore the voltage-operated relay used to monitor the output
in the above scheme should be arranged to operate after a time delay suffi-
ciently long to enable faults on the network to be cleared before the machine
is disconnected.
Schemes to detect unbalanced operation

Alternators, as stated earlier in section 7.4.1 (page 255), are not capable of
operating continuously under unbalanced conditions which cause their cur-
rents to contain significant negative-sequence components and therefore all
large machines should be provided with negative-sequence relays which have
inverse time/current characteristics coordinated with their withstand time/cur-
rent capabilities.
Schemes to detect asynchronous operation caused by low excitation
Asynchronous running of an alternator may be caused by its excitation falling
below critical levels which are dependent on the power it is supplying. A
machine may, however, continue to run synchronously at a low leading power
factor, even with very low excitation, if it is lightly loaded, i.e. if its prime
mover is supplying only a small amount of power. This can be seen from Fig.
7.16 in which the value of the limiting e.m.f. (Ed is low if I cos cp is low.
A voltage-operated relay may be connected across a shunt in the excitation
circuit to monitor the current flowing in it. The relay would operate in the
event of a complete loss of excitation, due to an open circuit or failure of the
exciter. It must, however, have a low setting, say 5% of the continuous rating
of the circuit, to ensure that its machine will not be disconnected at a time
when it could continue to run synchronously. This is not completely satisfac-
tory, however, as synchronism might be lost at higher field-current levels if
the power output was greater.
A superior method which is now widely used is to monitor the impedance
presented at the output terminals of an alternator. As explained earlier in
section 7.4.1 (page 255), the phase e.m.f.s produced by an alternator rotate
relative to its output voltages when it runs asynchronously, i.e. at both super-
and sub-synchronous speeds. As a result, the current varies during each slip
cycle.
This behaviour may be illustrated by considering the simple circuit shown
in single-phase form in Fig. 7.24(a) in which E and V are the e.m.f.s provided
by the asynchronous machine and the synchronous source respectively, Xa is
the inductive reactance of the machine and Zs is the reactance of the synchron-
ous source and the connecting circuit to the machine terminal at point A. To
simplify the treatment the alternator resistance is neglected. The current I and
the terminal voltage Va are given by:
1= E- V and Va = E _ . j Xa (E - V)
j~+~ J~+~
The apparent input impedance (Zap), seen at terminal A, is:
- -V '-1- EZs + j V Xa
Zap- a -
E-V
With zero excitation the machine e.m.f (E) would be zero, in which event
the apparent impedance (Zap) would be:
Zap=-jXa
With weak excitation, the magnitude of the e.m.f. E would be less than that
of the voltage V. As an example, if E = 0.5 V, i.e. the e.m.f. has a magnitude
half that of the voltage and they are of the same phase. In this case the apparent
impedance (Zap) would be given by:
Zap = - (2j Xa + Zs) n
jX
I
I
I
I
R
E/V= 1
(a) (b)
Fig. 7.24 Apparent impedance (Zap) at point A.
Half a slip cycle later, when E = - 0.5 V, the value of Zap would be:
Z=_),- -.2_X_a_-_Z_s Q
3
In practice the impedance Zs will be smaller than 2Xa and therefore the above
impedance will always appear to be capacitive when the e.m.f. E is less than
the voltage V. It can be shown that the loci of the apparent impedances as the
e.m.f. E rotates relative to the voltage V will always be circular, examples
being shown in Fig. 7 .24(b).
Clearly, therefore, a relay energized with a current and voltage proportional
to those at the stator terminals of a machine, and set to operate when the
apparent impedance is abnormal, may be used to detect asynchronous opera-
tion caused by low excitation. It will be clear that only one relay, associated
with one phase of a machine, is needed for this purpose because the behaviour
of each of the phases is the same for this condition.
Relays with circular operating zones in the impedance plane, the zone
centres being displaced from the origin as shown in Fig. 7.25, are commonly
used to detect asynchronous operation. These relays which are described as
mho type, are considered in some detail in section 11.3.2. To enable them to
jX
Radius 0.25 ~
Fig. 7.25 Operating zone of a relay set to detect asynchronous operation.

be correctly set for this application, the apparent impedance loci for protected
machines during asynchronous and both healthy and other abnormal condi-
tions should be determined to ensure that the relays will discriminate correctly.
It will generally be found that satisfactory performance will be provided by a
relay with a mho characteristic centred at a point R =0 and
X = - (0.75 Xd' + 0.25 Xd) and with a radius of 0.25 Xd, in which Xd' and Xd are
respectively the direct axis transient and synchronous reactances of the pro-
tected machine.
Asynchronous operation caused by low excitation may be allowed to persist
for periods ranging from a few seconds to several minutes, the permissible
times being lower the greater the rating of a machine. In attended stations it
is therefore quite common for a relay, which has detected that a machine is
running asynchronously, to give an alarm so that the operators may assess the
situation. They could firstly open the excitation circuit to enable the machine
to run in the induction-generator mode and then they could reduce the power
output of the prime mover and thus possibly re-establish synchronous running.
If this procedure did not prove successful then the machine could be discon-
nected from the network and re-synchronized after checking the excitation
circuit. If the network was not heavily loaded, and it was thought preferable,
the machine could be disconnected on receipt of the alarm.
In unattended stations, the relay could initiate the opening of the excitation
circuit and cause the power output of the prime mover to be reduced to attempt
to re-establish synchronous running and, should this not occur, then the ma-
chine could be disconnected from the network.
Schemes to detect pole slipping

Pole slipping may occur as a result of a disturbance or fault on the network
connected to a machine. In such cases, the speed of the machine rotor changes
slightly for a short period during which high currents will flow and large
mismatches will exist between the input and output powers if full excitation
is maintained. As a result, the mechanical stresses and vibrations produced
will be much higher than those present when asynchronous operation occurs
because of a loss of excitation. If the disturbance or fault is removed quickly,
synchronous running may be re-established when the rotor has moved for-
wards or backwards by two pole pitches, i.e. one revolution in a two-pole
machine, and no action need be taken. If this does not occur, however, the
condition cannot be allowed to continue because of the damage which will
result.
Pole slipping, because of its similarity to asynchronous running, may also
be detected by monitoring the apparent impedance at the output terminals of
a machine. Again a relay with an appropriate characteristic in the impedance
plane must be selected and it must be different from that chosen to detect
asynchronous running produced by low excitation, the impedances for this
condition being inductive rather than capacitive. Relays with mho-type char-
jX
R
Fig. 7.26 Relay characteristic used to detect pole-slipping.
acteristics may prove satisfactory but it may be found necessary, after examin-
ing the various impedance loci, to employ relays with quadrilateral
characteristics of the form shown in Fig. 7.26.
In attended stations, the relay could initiate the opening of the field circuit-
breaker to enable the machine to establish asynchronous running and thus
remove the damaging stresses. The power input to the prime mover could then
be reduced to a level where the machine might resume synchronous operation.
If this did not occur, the field circuit-breaker could be reclosed to allow a low
exciting current to flow. This should normally cause the machine to become
synchronous. The alternative is to arrange that relay operation will initiate the
opening of the main circuit-breaker to disconnect the machine from the net-
work, after which it could be resynchronized.
In unattended stations, the latter practice must usually be adopted, unless
the control systems are highly automated and capable of initiating the proced-
ures outlined above.
Overloading
Alternators in attended stations are not likely to be overloaded. Firstly the
power they can provide is limited to the maximum rating of their prime movers
and the operators will ensure that their excitations do not allow them to
provide excessive VAr outputs. It is not usual therefore to provide protective
equipment to detect the overloading of such machines.
Alternators in unattended stations may, however, supply excessive currents
at times due to failures of control equipment, such as voltage regulators, and
overload relays with suitable operating-time characteristics are often fitted to
such machines.
Overcurrent protection
Although the synchronous positive-sequence impedances of alternators are
usually high, values of 1 pu or more being normal on large machines, the
sub-transient and transient values are much lower, as also are the negative and
zero-sequence impedances. The currents which will flow in them to faults on
the networks they supply can therefore be quite high initially. They will,
however, decay to the normal rated values or less if a fault persists, the actual
rate of decay, in a given case, being dependent on the decrement characteristic
of the machine and the behaviour of its voltage regulator.
IDMT relays may be used to provide a back-up feature to the protective
equipment on a network. The time and current settings of such relays must be
selected after studying the characteristics of the machines to which they are
to be applied. When these relays are to be used in a network fed by a single
machine, the current transformers feeding the relays should be in the neutral
connections of the machine so that operation will occur for faults, both internal
and external to it. In those situations where several machines operate in
parallel, the current transformers should be in the output connections of the
machines, operation for both external and internal faults then being obtained
because of the currents that will be fed into a faulted alternator from the other
machines.
The detection of earth faults on field circuits

The field windings of alternators are normally unearthed and therefore the
incidence of a single earth fault will not cause current to flow. It is desirable,
however, that an indication of the condition should be given so that remedial
work may be done at the earliest convenient opportunity. Several methods are
used in practice to detect this condition. A simple scheme incorporates a
potentiometer connected across the field winding. The centre point of the
potentiometer may be connected to earth through a high resistance and a
sensitive relay. This scheme, which is also applied to synchronous motors, has
the disadvantage that it is not sensitive to faults near the centre of the field
winding. An alternative arrangement, described in section 7.3.2 (page 235), in
which a varistor is introduced does ensure that such faults will be detected
when the excitation changes significantly. Such a change may not occur for a
long period on large base-load alternators and therefore another arrangement
which may be adopted in attended stations is to include a switch, as shown in
Fig. 7.27, so that its operation will cause the relay to be connected to a
Alternator field
winding
Fig. 7.27 Scheme to detect earth faults on field windings.

~ Alternator field
d.c.
't--~
Alternator field
winding dI
supply winding
supply
~c
d.c. ~
Injection
transformer
[a.c. source
Voltage
sensitive
-=- reJay
(a) (b)
Fig. 7.28 Schemes for detecting earth faults on field windings.
different point on the potentiometer which would then enable a fault near the
centre of the field winding to be detected. This check would be made manually
every few hours.
More complex arrangements involving auxiliary supplies are used on large
motors and alternators, two being shown in Fig. 7.28 (a) and (b). In each,
current will flow through the relay in the event of an earth fault on the field-
winding circuit and an immediate indication of its presence can be provided.
Other monitoring and protective schemes

The protective schemes applied to alternators have been considered in the
preceding sections. Associated with them must be devices to detect other
abnormal conditions which may arise in the prime movers and auxiliary plant.
As an illustration, conditions such as low steam pressure, loss of vacuum and
loss of boiler water should be detected when driving power is provided by
stream turbines. In addition, the supply of lubricating oil and the performance
of governors should be monitored.
When transformers form units with alternators they are included in the zones
of the overall current-differential schemes, as stated earlier. In addition, how-
ever, the transformers should have earth fault protection to cover their second-
ary windings, Buchholz relays and devices such as temperature sensors.
Overall protective arrangements

For any particular application, the appropriate protective schemes should be
provided to ensure that all unacceptable behaviour which occurs in the pro-
tected plant will be detected. Each of the protective schemes or devices should,
when necessary, provide an indication of the abnormality it has detected and
also initiate the appropriate actions to ensure that any damage which might
have occurred is limited to the lowest possible level. As an example, the
operation of a current-differential scheme should cause the appropriate circuit-
U
Field
46
current protection.
Ne ative sequence
Fig. 7.29 Schematic of alternator-transformer unit protection.
breakers to be opened to ensure that the unit is disconnected from the network
and also that its excitation is removed. In addition, the steam or water supply
to the driving turbine should be cut off or reduced by operating the appropriate
control valves.
An arrangement suitable for the complete protection of a large alternator-
transformer unit is illustrated in Fig. 7.29. More examples may be found in
references [6, 7, 16, 17]. A typical tripping arrangement for an alternator-
transformer unit is shown in Fig. 7.30.
Electrical trip of oil servo system
Boiler
firing
LVCB
~~~~~~~~n~t-----f~:-t~~ ___
Low power
interlock
+----+-1 Field
circuit
breaker
HVCB
Fig. 7.30 Typical tripping arrangements for alternator-transformer unit. (Reproduced

from Protective Relays - Application Guide, 3rd edn, GEC Measurements, 1987 with
the permission of GEC Alsthom Protection and Control Ltd).
7.5 THE FUTURE

It is probable that small machines will continue to be protected by fuses and
relatively simple thermal devices and relays.
Schemes based on the principles presently in use, such as current compar-
ison and the detection of negative-sequence currents, will almost certainly be
applied to large machines well into the future. Increasingly they will be
implemented by employing electronic data acquisition and processing tech-
The future 277
niques which will ensure that the appropriate actions are initiated when ab-
normal conditions occur. Details of modern communication and digital
signal processing techniques may be found in Chapters 13 and 14 of this
book.
Examples of manufacturers' products which employ numerical relays
and digital signal processing techniques of the type described in Chapter
14 are available. Typical features of these relays are that they are
multifunctional and have comprehensive self-monitoring and communi-
cations facilities. In [IS] a digital integrated generator protection relay,
LGPG111, is described which incorporates 14 separate protection
functions. These are:
Generator differential (for protection against phase-to-phase and three

phase stator faults)
Stator earth fault (covering 95% of stator winding)
Neutral displacement
Sensitive directional earth fault
Voltage dependent overcurrent (for backup protection)
Reverse power (to detect loss of prime mover)
Negative phase sequence (to detect unbalanced conditions)
Field failure (to detect loss of excitation)
Under lover voltage
Underlover frequency
Voltage balance (to detect VT fuse failure)
Timer hold facility (for faster clearance of recurrent faults)
Integrated timer facility
Frequency tracking (to adjust internal sampling)
The heart of the relay is a 16-bit microprocessor (Intel SOCI86XL)

which controls all other modules. Communications are either two RS232
ports using the IEC870 protocol or one RS485 port using the K-Bus
protocol. In addition to the microcomputer and serial communications
modules, the relay also has a power supply unit, two output modules, an
analogue and status input module, communications isolation module and
a user interface.
Another example of a microprocessor based relay using digital signal
processing techniques is the Reygen-30 [19]. It incorporates 15 protective
relaying functions and it is primarily used to protect high-impedance
earthed unit generators.
Numerical generator protection is also available from the REG216 relay
described in [20,21]. Its main application areas are in generator and
generator transformer unit protection. Its modularity and multifunction-
ality allow adaptation to the desired protection scheme. It has been applied
to generators with ratings from 80 to 812 MV A.
REFERENCES
1. Creek, F. R. L. (1990) Power Eng. Journal.
2. Ray, W. F., Lawrenson, P. J., Davis, R. M. et al. (1986) High performance
switched reluctance brushless drive, IEEE Trans. on Industry Applications, IA-
22, pp 722-30.
3. NEMA MGI-1987, Motors and Generators.
4. ANSI/NFPA 70-1987, National electrical code.
5. Type MCHN02 Motor protection relays, in GEC Measurements Modular Protect-
ive Relays, 1, 155-64.
6. Protective Relays Application Guide, Third Edition, 1987, GEC Measurements.
7. Motor Protective Relays in Protective Relays, Catalog R. 1989, pp 14/1-14/18,
published by Siemens Actiengesellschaft.
8. Generator and Motor Protection, Section 4 in HV Protection and Protection
Systems, Buyer's Guide 1989-1990, Vol I, Asea Brown Boveri Relays.
9. IEEE Guide for AC motor protection, ANSI/lEEE C37.96-1988.
10. Say, M. G. and Taylor, E. O. (1980) Direct Current Machines, Pitman.
11. Finney, D. (1988) Variable frequency AC motor drive systems, Peter Peregrinus.
12. Protective relays application guide, 3rd edition, 1987, GEC Measurements, p. 301.
13. Pope, J. W. (1984) A comparison of 100% stator ground fault protection schemes
for generator stator protection, IEEE Trans. PAS-I03, pp 832-40.
14. Marttila, R. J. (1986) Design principle of a new generator stator ground relay for
100% coverage of the stator winding, IEEE Trans, PWRD-l, pp 41-51.
15. Type RAGEA 100% generator stator ground-fault relay in Asea Brown Boveri
Buyer's Guide B03-4012E.
16. Generator Protection: Application Guide, Asea Relays AG03-4005 Dec 1986.
17. IEEE Guide for AC Generator Protection, IEEE C37.102-1987.
18. GEC Alsthom T and D (1996) Digital integrated generator protection relay,
Publication R4J06F.
19. Reyrolle Protection (1996) Reygen-30 generator protection, Publication
REY3/9606-11O/1-1.
20. ABB (1992) Type REG216 generator protection, Publication IMDB02005-EN.
21. liar, M. and Stranne, G. (1993) Numerical protection systems for generators
and generator transformer units, ABB Rev., 1/93, 27-38.
FURTHER READING
IEEE Relaying Committee Report (1988) Survey of experience with generator protec-
tion and prospects for improvements using digital computers, IEEE Trans. on
Power Delivery, 3, 1511-22.
GEe Measurements, Industrial generator protection application guide, Publ R-
4016B.
IEEE Tutorial on the protection of synchronous generators (1995), Publication 95
TP 102.
8
The protection of busbars
Busbars are vital parts of power networks because they link incoming circuits
connected to sources, to outgoing circuits which feed loads. In the event of a
fault on a section of busbar all the incoming circuits connected to it must be
opened to clear the fault. In practice, because of the amount of interconnection
of circuits and the possibility of back feeds from load circuits, all the circuits
connected to a faulted section of bus bar are disconnected. Such disconnection
clearly causes considerable disruption and the greater the operating voltage
and current levels of a busbar, the greater will be the loss of supply resulting
from a fault. It is therefore necessary that busbars should be so designed and
constructed that the incidence of faults occurring on them is reduced to a very
low level and it is also essential that the protective schemes applied to busbars
are highly discriminative so that they will not wrongly cause a busbar, or a
section of it which they are protecting, to be disconnected when faults occur
on circuits external to it. In addition, because of the severe and extensive
damage which may result from bus bar faults it is necessary that they should
be detected by protective schemes and then cleared very rapidly.
The following section provides some historical background and then infor-
mation on busbar layouts and constructions and protective arrangements is
provided in the later sections of this chapter.

During the early decades of this century both the ratings of generating plants
and the degrees of interconnection between networks were relatively low and
therefore the fault levels, in terms of volt-amperes, were fairly limited. It was
also felt that the designs of switchgear were such that the likelihood of faults
occurring on or adjacent to bus bars was very small. As a result of these factors
busbar-protective schemes based on the then well-known Merz-Price prin-
ciple were not applied, it being thought that they were not sufficiently reliable
and that they might maloperate and cause unnecessary disconnection of bus-
bars. Reliance was usually placed on the definite or inverse-time overcurrent
and earth fault relays covering the circuits connected to the bus bars. This
practice was not wholly satisfactory because of the delays involved in clearing

280 The protection of busbars
busbar faults and the fact that such relaying systems cannot always discrimi-
nate correctly.
In some installations, therefore, the above arrangements were supplemented
by applying frame-leakage protection. This simple form of protection, which
was available before 1915, was also applied to other power equipment, such
as power transformers. Its principle, which was outlined in section 6.3.6 (page
212), is to lightly insulate the casing or enclosure of a piece of equipment from
earth and to then connect the primary winding of a current transformer be-
tween earth and the enclosure. A relay connected to the secondary winding of
the current transformer can be set to operate at a low current, say 20% or less
of the secondary rating of the current transformer, and thereby internal faults
to the enclosure can be detected. As faults external to the protected equipment
would not cause currents to flow to earth through the casing, these could not
cause relay operation and therefore discrimination could be obtained without
the need for time delays. It will be appreciated, however, that this scheme
cannot detect interphase faults which do not cause breakdowns to the enclosure.
In spite of the steady growth in generator outputs and networks in the first
decades of this century it was not until about 1935 that the necessity for
introducing and applying more advanced busbar-protective schemes was
accepted. Two basic forms were developed, namely interlock and current-
differential.
The interlock schemes basically employed directional relays associated with
each of the ~ircuits connected to the sections of bus bars being protected. In
the event of an external fault one of these relays would detect current leaving
the busbar and it would inhibit the tripping of the circuit-breakers. Such
schemes clearly required quite large numbers of relays and both current and
voltage transformers had to be provided to supply their windings. In addition,
starting relays, which operated when currents were above normal levels, were
included so that the tripping of circuit-breakers could only be initiated when
faults were present. In this way the possibility of incorrect operation was
reduced.
Current-differential schemes, which operated on the Merz-Price principle,
were introduced and biased relays of various forms were developed. By 1950
devices such as transductors were introduced to reduce the complexity of the
electro-mechanical relays required, it being felt that this would increase the
reliability of the schemes. A scheme of this form, designated 'Monobias' [1],
was produced by A. Reyrolle and Co Ltd.
Because of the initial reluctance of electricity supply authorities to install
bus bar protection, which was caused by their fear that healthy busbars might
be disconnected incorrectly, the practice of applying two independent schemes
to major busbars was adopted and this has continued up to the present time.
As an example, A. Reyrolle and Co Ltd produced Dualock protection which
had an interlock scheme and a current-differential scheme and, of course, both
had to operate before the tripping of circuit-breakers could be initiated.
Bushars 281
Because of the rarity of busbar faults, protective schemes are seldom re-
quired to operate and this leads to the possibility that defects might have arisen
of which operating staff could be unaware. For this reason, manufacturers have
always incorporated equipment into their schemes to enable operators to per-
form tests regularly and in many cases equipment is provided to initiate tests
automatically.
8.2 BUSBARS
Busbars, because they interconnect several circuits, are associated with cir-
cuit-breakers and therefore they are mounted within switchgear units produced
for systems operating up to medium voltage levels. When circuits operating
at very high transmission-voltage levels were introduced, the above practice
was not possible because of the spacings needed between the conductors, and
in these applications the busbars were normally mounted in air, connections
then being taken to the circuit-breakers. This practice is still used extensively
but recently fully-enclosed switchgear containing gaseous insulants has been
produced for use at very high voltage levels and the busbars are housed within
these units, as described later.
Constructional features associated with these arrangements are examined in
the following section and then busbar configurations are considered.
8.2.1 The construction of enclosed switchgear units

In switchgear produced at the beginning of this century, conductors, supported
on insulators, were housed in air-filled compartments, usually at the tops of
the units and they ran along to interconnect the units and thus formed busbars.
The voltage levels were low and small spacings between the phases were
acceptable. The amounts of generation and the load levels were small and
therefore the damage which could result from a busbar fault was not great and
the consequences of loss of supply were not serious.
As demand increased, larger machines were installed and operating voltages
increased. Because of the increased consequences of busbar faults, switchgear
was produced in which the bus bars were mounted in separate metal chambers
filled with bituminous compound. The spacings between adjacent busbars
were great enough to ensure that the breakdown voltages between the phases
were higher than those on the circuits connected to them.
Because faults on the conductors connecting the busbars to the circuit-
breakers would have the same consequences as faults on the busbars them-
selves, these connections, which were within metal enclosures, were also
insulated to high levels and surrounded by bituminous compound to ensure
that their breakdown voltages were as high as those of the busbars. In addition,
the circuit-breakers had not only to be capable of interrupting the fault currents
which could flow in the event of faults on the bus bars or on the circuits
Fig. 8.1 800 kY GIS substation produced by ABB, in operation since February 1988
in the 800 kY system of ESCOM in South Africa. (Reproduced from ABB Review 6/91,
with the permission of ABB Relays AG, Baden, Switzerland).
connected to them, but also to be so insulated that the likelihood of faults

occurring within them was reduced to a minimum.
Busbars 283
The above principle has been adhered to in all the forms of enclosed switch-
gear produced subsequently.
Air, oil, synthetic resins and gases have been and still are used as the main
insulants in different types of enclosed switchgear. In the past such switchgear
was only suitable for use in the lower voltage sections of networks, the
open-type switchgear layouts described below being used in sections operating
at transmission-voltage levels. In recent years however, very compact fully-
enclosed switchgear which uses the gas sulphur hexafluoride (SF6 ) as the main
insulant has been developed and designs are available for use at all voltages
and current levels. An exterior view of an installation operating at 800 kV is
shown in Fig. 8.1.
8.2.2 Open-type switching sites

Busbars which operate at high voltages may be made of either uninsulated
flexible cable or rigid conductors. They are either suspended in air from string
insulators or supported on post-type insulators. In all cases, the insulators are
longer than those used on the rest of the associated networks and the spacings
between the phases and to earth are great enough to ensure that the breakdown
voltages are very high.
The busbars are linked to air-blast, small-oil volume, or bulk oil circuit-
breakers, which are insulated to high levels by connectors with large spacings
between them and to earth.
Because of the sizes of these sites they are usually outdoors and allowance
must be made for the effects of the weather and possible pollution on the
withstand-voltage levels of the insulation. A further important factor is that
exposed conductors could be struck by lightning, thus initiating busbar faults.
To eliminate this possibility, a grid of earthed conductors is usually provided
above the bus bars.
In a number of installations, for example at the Ratcliffe-on-Soar Generating
Station of Powergen in Britain, it was felt that outdoor switching sites were
too vulnerable to climate conditions and therefore the costs of suitable build-
ings to house the busbars and switchgear were considered to be justified and
they were provided. Indoor switching sites are also used near coasts to avoid
them being polluted by salt.
8.2.3 Further methods to reduce the incidence of bus bar faults

Whilst direct lightning strikes to busbars can be prevented, extremely high
voltages can nevertheless be produced between phases or between phases and
earth on transmission lines as a result of lightning strikes on or near them. If
such a situation arises, voltage waves are set up, travelling in both directions
from the point where the disturbance has occurred, and should they reach
busbars to which the line is connected they could cause a breakdown of the
busbar insulation.
To try to eliminate this possibility and also to prevent damage to other
vulnerable equipment, such as power transformers, an earthed conductor is
provided above the phase conductors of each transmission line to reduce the
likelihood of high voltages being produced on the phase conductors by light-
ning discharges and, in addition, horn gaps are fitted across each of the string
insulators to try to prevent any high voltage surges which do occur from
reaching the ends of the line.
As a further measure, two earthed conductors are often run in parallel over
the end sections of transmission lines to try to ensure that high voltage surges
will not be induced, by lightning, on the phase conductors at points near
terminal equipment or busbars, because there are few opportunities for such
surges to be attenuated.
In addition to these measures, surge absorbers, i.e. voltage limiters, are often
fitted near the terminals of overhead lines to prevent excessively high voltages
reaching busbars or other equipment.
8.3 SECTIONALIZATION
Because of the practices described above, the number of faults which occur
on busbars and the connections to them represent only a tiny fraction of the
total number of faults which occur on power systems in given periods. Never-
theless it must be accepted that some busbar faults will occur and therefore it
is the general practice to sectionalize busbars to reduce the number of circuits
which must be opened to clear a fault and at the same time enable a significant
part of the network to continue in service.
The simplest and cheapest arrangement is to include one bus-section circuit-
breaker to divide a busbar into two sections as shown in Fig. 8.2(a). With this
arrangement it is usual to operate with the bus-section circuit-breaker closed
and should a fault occur on one of the sections all the circuit-breakers associ-
ated with the faulted section and also the bus-section circuit-breaker must be
opened to clear the fault. For this arrangement to be effective each of the
sections should contain about equal numbers of incoming and outgoing cir-
cuits and where possible important loads should be fed by two circuits, one
connected to each of the busbar sections.
Where it is thought to be desirable, busbars are provided with two or more
bus-section circuit-breakers, thus creating three or more sections. These prac-
tices, of course, increase the initial cost of an installation but they reduce the
amount of disconnection which will occur in the event of busbar faults.
Busbar faults can cause severe damage which may necessitate extensive
repairs and, as a consequence, consumers connected to a faulted section of a
bus bar could be left without supplies for a considerable time. To alleviate such
situations, it is usual to have duplicate busbars in important installations, a
typical arrangement b~ing shown in Fig. 8.2(b). The primary purpose of this
practice is to allow circuits, which become disconnected because of a fault on
Sectionalization 285
----1--.. 1--
.
Bus-section
Section 1 circuit breaker Section 2
III II II ~ II
x---I---I--~I- Phase a
Circuit
breakers XXX XXX XXX
11
XXX XXX XXX
Ii: ::::
III III III III III III
Circuit 1 Circuit 2 Circuit 3 Circuit 4 Circuit 5 Circuit 6
(a)
Bus-coupler
circuit breaker
Phase a
Reserve
Phase b
busbar
I I Phase c
X~X XXX
X - I Phase a
Main I X Phase b
busbar
X Phase c
.b Bus-section
circuit breaker
,b
ff
Selection
switches
Circuit
X X X
f fff
X X X
III III
breakers
Circuit 1 Circuit 2
(b)
Fig. 8.2 Busbar arrangements.
a busbar section, to be transferred to the other busbar. This requires that

selector switches, not capable of interrupting fault currents, be provided for
each circuit to enable the necessary transfers to be effected.
If the second set of busbars is used solely as a reserve and is not normally
energized, the transfer procedures may be quite lengthy because supply and
load circuits cannot be transferred simultaneously. It is therefore usual, but
more costly, to include one or more bus-coupler circuit-breakers, which are
normally closed, so that both busbars are energized, thus enabling transfers to
be effected relatively quickly in the event of a fault on any busbar. Two
three-phase bus-coupler circuit-breakers are shown in Fig. 8.2(b).
In some installations bus-section circuit-breakers are included in both sets
of bus bars thus providing symmetry and complete duplication. This arrange-
ment makes it possible to isolate individual sections of busbars for short
periods during which maintenance work may be done; for example, the insu-
lators in outdoor installations may be cleaned and inspected.
It will be evident that the clearance of a fault on a section of busbar will
require that all the circuit-breakers associated with incoming and outgoing
circuits connected to the section be opened, as must also bus-section and
bus-coupler circuit-breakers associated with the section.
8.4 FAULTS ON OR NEAR BUSBARS

It is essential when considering the designs of busbars and the protective
equipment which will be applied to them that the levels of the currents which
may flow when faults are present be known. Behaviour during both internal
and external fault conditions is examined below.
8.4.1 Internal faults

Busbars are unlike transformers and machines in that the currents which will
flow in the events of faults on them are not dependent on their positions
because at any instant, on any phase, the voltage is the same at all points.
Fault levels are dependent, however, on the number of circuits connected at
any time, which can feed currents to a fault. Clearly if all the circuit-breakers
in circuits connected either directly or indirectly to sources are closed at the
time of a short-circuit then the total fault current will be at the maximum
possible level. Should some circuit-breakers be open, however, then lower
fault currents will flow and, of course, similar effects will be produced if there
is significant impedance in a fault path. In addition, earth fault currents will
be affected by the earthing arrangements on the various circuits.
Clearly, the range of fault currents which may be experienced in a particular
application should be determined using one of the well-established techniques
to ensure that satisfactory performance will be obtained from the proposed
In general, fault levels are normally well above the rated currents of indi-
vidual circuits and very sensitive fault settings are not usually needed.
Whilst all types of faults, i.e. phase to earth and interphase faults, could
possibly occur on open-type outdoor busbar installations and in many other
enclosed switchgear arrangements, there are some switchgear designs in which
the phases are separated by earthed barriers and of course in such cases only
phase to earth faults are possible.
8.4.2 External faults

In the event of a fault on a circuit at a point near a circuit-breaker, as shown
in Fig. 8.3, the fault current would be the same as that for a busbar fault and,
if all the circuits were connected, the maximum possible fault level would be
reached. If the faulted circuit was feeding only a load, its circuit-breaker would
have to clear the total fault current and would have to have the appropriate
Positioning of current transformers 287
Bus-coupler Bus-section
Reserve circuit breaker circuit breaker
busbar
Main X X
busbar X
Selection
switches
Circuit
breakers
X X X X X X
I I ~ Fault
I I I
Fig. 8.3 Fault on an outgoing circuit. Circuit shown in single line form for simplicity.
breaking capacity. If, however, the faulted circuit was connected to a source,
its circuit-breaker would have to clear a current somewhat less than the total
fault current. In practice, however, it is usual, in the interests of stand-
ardization, for all the circuit-breakers associated with a set of busbars to have
the same breaking capacity, namely that needed to clear the maximum-
possible fault current.
It will be appreciated that current-differential protective schemes associated
with busbars have to be set sensitively enough to detect internal faults and yet
not operate in the event of external faults involving currents up to the highest
levels possible.
8.5 POSITIONING OF CURRENT TRANSFORMERS AND

INCORRECTLY PROTECTED ZONES
As pointed out in section 8.4.2, large currents flow in the event of faults on
circuits near the circuit-breakers which connect them to bus bars. Such faults
should be detected by the protective equipments associated with the circuits
on which they occur and only the circuit-breakers in a faulted circuit should
be opened. In practice, however, there must be connections between circuit-
breaker contacts and current transformers and these connections are protected
incorrectly as explained below.
Busbar-protection current transformers are invariably mounted on the out-
going sides of the circuit-breakers and should a fault occur on the connections
between the circuit-breaker contacts and the current transformers feeding a
current-differential protective scheme covering a section of the busbars, as
shown in Fig. 8.4, all the circuit-breakers associated with the section would
be opened incorrectly. This possibility cannot be eliminated, but to reduce it to
a minimum the connections between the current transformers and the circuit-
breaker contacts should be both highly insulated and short.
Reserve
bus bar
Main
busbar
Selection
switches
Circuit X
breakers s~ s
Current f [
transformers
Fig. 8.4 Fault within a protected zone.
This topic was examined in more detail in section 2.2.7 (page 71), as was
the positioning of the current transformers feeding the protective schemes
associated with the outgoing circuits.
8.6 PROTECTIVE ARRANGEMENTS FOR BUSBARS

Simple busbars used in direct-current distribution networks operating at rela-
tively low voltages may be protected by fuses. Correct discrimination in the
event of faults either on the busbars or the circuits connected to them can be
achieved by selecting fuselinks with the appropriate current ratings and oper-
ating times, when there are few circuits connected to sources and the other
circuits, supplying loads, cannot feed current back into the busbars. Fuses may
also be used to protect simple busbars used in single-phase distribution net-
works operating at relatively low voltages and, again, satisfactory operation
should be obtainable. Fuses are not suitable for application to three-phase
busbars, however, because complete circuits would not be cleared in the event
of single-phase faults and, as a result, equipment such as motors could be
supplied with imbalanced voltages.
IDMT relays are used to protect the busbars of some single-phase and
three-phase distribution networks. In these applications, the current and time
settings of the relays should be selected in the manner described in section 4.4
(page 124) and satisfactory grading should be achievable.
With these arrangements, fault clearance times may be quite long, a situ-
ation which cannot be accepted on major busbars where the fault levels may
be very high and the consequential damage could be very great. It will be
clear, therefore, that high-speed, current-differential schemes should be used
in these applications and this has been the practice for many years.
Protective arrangements for busbars 289
Depending on the busbar layouts and the designs of the switchgear, several
different current-differential schemes are used and these are now considered.
8.6.1 Application to simple single-phase, unsectionalized busbars

Such busbars probably would not be protected by current-differential schemes,
but if they were, then a current transformer would be included in each of the
circuits on the outgoing side of the circuit-breakers as shown in Fig. 8.S(a).
The current transformers, which would each have the same ratio, would
have their secondary windings connected in parallel and the sum of their
output currents would flow in the operating winding of the single relay.
The protected zone would, as shown in Fig. 8.S(a), include the connections
on the busbar sides of the current transformers, the circuit-breakers and the
busbar.
Unless there was a fault present within the protected zone, the primary
currents of the transformers would sum to zero at every instant, i.e.
Circuit r - - - - - -Protected zone- - - - - -IUne

breakers I
I
I
XI tiA XI t
is XI t
ic XI t
iD
I
X t
iE
:buSbar
~/r
.
Current IS 5 5 5 5 5 5 5 5 51
transformers L I Relay
each of the
same ratio
Loads
.....- ....- -.....- -......- ....- Neutral
(a)
Fig. 8.5 Current flows for different conditions.

iA + iB + ic + iD + iE = O. Ideally, the secondary winding currents should also
sum to zero causing the current in the operating winding of the relay to be
zero. In practice, because of the non-linearity of the excitation characteristics
of the transformers and the fact that their primary currents would probably all
be different, the secondary currents would not sum to zero and some current
would flow in the relay operating winding.
A particularly onerous condition would be a fault just outside the protected
zone, at a point such as P in Fig. 8.5(b). In this event, the fault current (it)
could have a very large magnitude as would the primary current iA The other
currents (iB-iE) would probably all have different magnitudes but all of them
would be lower than that of the current i A In these circumstances quite
considerable mismatching could occur and a significant current (ir) could flow
in the operating winding of the relay. To assist in minimizing this current, the
current transformers in a scheme should all be of the same design. There are
circumstances, however, where this may not be practicable, because the rated
currents of the individual switchgear units may vary considerably and it may
not be possible to accommodate current transformers of the same dimensions
in each of the circuits.
Should a fault occur within the protected zone, at a point such as Q in Fig.
8.5(c), the sum of the primary currents of the transformers would be equal to
the fault current, i.e. iA + iB + ic + iD + iE = it. This would cause a current, de-
pendent on the fault current, to flow in the relay and its setting would have to
be such that the necessary sensitivity could be obtained. Clearly biasing or
other features would be necessary to ensure that operation would not occur
when external faults up to the maximum possible current levels were present.
These factors and the types of relays available for this duty are considered
later in section 8.7.
8.6.2 Application to simple three-phase unsectionalized busbars

These busbars are protected by including a current transformer in each of the
phases on the outgoing side of each of the circuit-breakers. Again the current
transformers must be of the same ratio and ideally they should all be of the
same design.
When the design of the switchgear is such that both earth faults and inter-
phase faults clear of earth may occur within the zone to be protected, the
secondary windings of the current transformers must be connected in parallel
in three separate phase groups as shown in Fig. 8.6(a) and each group must
be connected to a relay.
Each phase group would operate independently and its behaviour would be
the same as that considered in section 8.6.1. Again appropriate relay settings
and biasing arrangements are needed to provide the necessary sensitivity to
internal faults and yet ensure that relay operation will not occur when external
faults are present.
ii, I I , , I I I
I - - - - - - - - - - - - - - Protected zone - - - - - - - - - - - Bu;b~r;1
I_ X X X X X X X X X X X X X X Xbreake~s
_ _ _ _ _ _
r
_
,
_
I II _
I Circuit i
___ J
Circuit 1 Circuit 2 Circuit 3 Circuit 4 Circuit 5

(a)
I,I
Busbars
o
00-..:::0
~
Q)
E
III
~
Q)
I , I
X X
III
X X X X
r , I I I
X X X X X breakers
, Circuit
.E~
-co-
(J) (J)
~~
e?.c
:s~ I
OQ)
Circuit 1 Circuit 2 Circuit 3 Circuit 4 Circuit 5
(b)
Fig. 8.6 Protective-circuit arrangements.
It will be evident that only one relay will operate in the event of a single
phase to earth fault but two or all three relays may operate when interphase
faults occur. In each case, however, relay operation will initiate opening of
the circuit-breakers associated with the faulty section which will thus be
completely isolated.
Should the switchgear be so designed that there are earthed barriers between
the phases, thus eliminating the possibility of interphase faults clear of earth,
then the secondary windings of all the current transformers could be connected
in parallel and connected to a single relay as shown in Fig. 8.6(b). The
protected zone would again include the connections on the busbar side of the
current transformers, the circuit-breakers and the busbars.
With this arrangement the number of current transformers operating in
parallel is three times the number of circuits connected to the bus bars. It will
be appreciated that greater imbalance is likely to be caused by mismatching
when the number of balancing current transformers is increased. In addition,
obtaining a given fault setting may require the operating current of a relay to
be reduced because of the increased total of the exciting currents of the
transformers. Such schemes may therefore require greater biasing to ensure
that correct discrimination will be achieved. In some circumstances it may be
preferable to segregate the current transformers into phase groups and use
three relays, i.e. use the arrangement shown in Fig. 8.6(a).
8.6.3 Application to complex three-phase sectionalized busbars

As explained earlier in section 8.3, important busbars are sectionalized, so that
internal faults will not cause all the circuits linked to a busbar to be discon-
nected. In addition, such installations usually have duplicate busbars, a main
set of bus bars to which the circuits are normally connected and a reserve set
of busbars to which the circuits may be connected when a fault has occurred
on the main busbars or when maintenance work has to be done on them. In
such installations, busbar section and coupler circuit-breakers are needed and
selector switches, incapable of clearing load and fault currents, must be pro-
vided to enable individual circuits to be connected to either the main or reserve
busbars. A typical arrangement is shown in Fig. 8.7.
Clearly a current-differential scheme providing a single protective zone,
covering all the busbars in an installation, would not be satisfactory as a single
fault anywhere on the busbars would cause the disconnection of all the cir-
cuits. In practice, separate and independent schemes must be used to cover
each of the sections of busbar. For the arrangement shown in Fig. 8.7, there
.......................................................................
: Zone 3 Bus-section :
Reserve
: circuit breaker :
busbar
Bus-coupler I
circuit-breakers I
Main busbar ~1"""+--"'9'""-+-~-t ........
I
Selector I
switches I
: ,
Circuit-breakers I X X X ,. X X X
'-r---1----t-----
h--
Zone 1 i - I+-'
-hZone2 h
Fig. 8.7 Protected zones.

would be schemes covering three zones, i.e. the two sections of the main
bus bar and the reserve bus bar.
As in other applications, all the current transformers would have the same
ratio and those associated with the incoming and outgoing circuits would be
mounted on the circuit sides of the circuit-breakers in positions as close to the
contacts as possible to minimize the incorrectly protected zones. In addition,
current transformers could be mounted on each side of the busbar section and
busbar coupler circuit-breakers. Ideally all the current transformers should be
of the same design to assist in achieving balance when the busbars are healthy.
Should the switchgear and busbars be of designs where both earth faults and
interphase faults could occur, then the phases would be protected separately
and for the arrangement shown in Fig. 8.7 the current transformers would be
connected to form nine groups each with its own relay, i.e. three zones.
Because each of the incoming and outgoing circuits could be connected
either to a section of the main busbar or the reserve busbar, the selector
switches must have auxiliary switches which will connect the secondary wind-
ings of the current transformers into the appropriate zone groups and thus
ensure that correct discrimination will be obtained at all times. An important
point which should be noted is that the selector switches are not capable of
breaking current and they are therefore only operated when their associated
circuit-breakers are open. There is therefore no possibility of the protective
schemes being imbalanced when any of the current transformers are trans-
ferred from one group to another because they will not be carrying current
when the auxiliary switches operate. The detailed current transformer circuitry
is shown in Fig. 8.8.
It is evident that the numbers of current transformers in the groups associ-
ated with the zones will vary and at any time they might be quite different,
the number in the phase groups of one zone being low while there may be a
large number in the groups of another zone. As an example, if none of the
circuits were connected to the reserve busbar, then there would only be the
two bus-coupler current transformers in its phase groups whereas there would
be large numbers of current transformers in the groups associated with the
sections of the main busbar.
For given relay current settings the minimum fault currents needed to cause
operation may be considerably affected by the numbers of current transformers
in the zone groups; for example, if high-impedance relays are used, the mini-
mum fault current for operation is approximately proportional to the number
of current transformers in a group. A further factor which must be recognized
is that the amount of mismatching and therefore the imbalance during external
fault conditions may be affected by the numbers of transformers in groups.
Studies should therefore be done to determine that the proposed relay settings
and biasing will provide acceptable performance under all operating condi-
tions. This topic is considered further in section 8.7.
The inclusion of auxiliary switches in the secondary circuits of current
Reserve busbar
Reserve busbar
s Zone 3
~~L Bus-coupler
X ciraJ~-breaker
Main busbar
~1: III I I II III I II :~;3

Fig. 8.8 Current-transfonner connections. Note: To simplify the diagram it is shown
single-phase and only the connections for zones I and 3 are shown. All the current
transfonners have the same ratio.
transformers has always been accepted with some reluctance and this is par-
ticularly so when they are feeding busbar protective schemes. The switches
could be eliminated if the zones were completed by including current trans-
formers in both the connections between each selector switch and the busbars,
as shown in Fig. 8.9. This arrangement would increase the total number of
current transformers required and also make it necessary to provide further
protective equipment and current transformers to enable faults on the selector
switches, the circuit-breakers and the connections between them to be de-
tected. This extra complexity has never been considered to be justified, be-
cause of the high performance record of the current schemes incorporating
auxiliary switches.
Whilst it is the common practice to have main and reserve busbars and to
only have circuits connected to the reserve bus bar after a fault has been present
on a section of the main busbar or when maintenance work is being done, it
can be argued that it is better to have circuits connected to each of the busbars
so that fewer circuits will be disconnected in the event of any busbar fault. If
this procedure is adopted it does not affect the protective arrangements in any
way because they must be such that satisfactory performance will be obtained
with all the possible connections.
As stated earlier, some switchgear units have the phase connections com-
pletely segregated by incorporating earthed barriers between them. In such
cases, all the current transformers associated with a protected zone may be
connected in parallel and only one relay is provided per zone. This clearly
reduces the number of relays needed but it may affect the sensitivity of a
scheme and increase the imbalances during healthy and external fault condi-
tions as stated earlier in section 8.6.2.
Reserve
bucbarl
Bus-coupler
circuit-breakers x Bus-section
x
circuit-breaker
Main busbar ~--~~--~~~---rX~----~--~~r----+
Current
transformers s
each 01 the f [ ] s [ ] s [ ] ; ; [ ] s [ ] s [ ] s
same ratio
Selector
switches
t t t t t t
Fig. 8.9 Current-transformer connections without auxiliary switches_
8.6.4 Interconnections in current-differential schemes applied to
busbars
Switchgear produced for relatively low voltage and current operation tends to
be compact and the interconnections between the current transformers associ-
ated with current-differential protective schemes are usually quite short and
therefore of low resistances. In addition, the primary current ratings of the
current transformers are not high and therefore low secondary current ratings,
e.g. 1 A, can readily be provided. As a result, the voltage drops on the con-
nections, even during external faults involving high currents, are not great and
the imbalances caused by mismatching and current transformer errors will be
limited.
On large outdoor switching sites, however, the connections between the
current transformers may be quite lengthy and the current ratings are high, e.g.
current transformers with primary windings rated at 2000 A are used in large
British generating stations. In such applications current transformers with 5 A
secondary ratings would be used if possible to limit the number of secondary
winding turns, but the voltage drops that would be produced in busbar protective
schemes with such transformers might not be acceptable and the extra expense
of producing 1 A secondary windings is often considered to be justified.
8.7 RELAYS USED IN CURRENT-DIFFERENTIAL SCHEMES

As explained in Chapter 5, both high- and low-impedance relays may be used
in current-differential schemes and their use in busbar protective schemes is
considered below.
8.7.1 High-impedance relays

High-impedance relays have proved very satisfactory in several different ap-
plications over many years, a particular example being in the restricted earth
fault protection of power transformers. These relays have the advantage that
the minimum operating voltage, at which they must be set to ensure that they
will not operate during external faults, can be determined easily. It is shown
in section 5.6 (page 171) that this is the voltage which would be produced across
the relay during the largest possible external fault, if the core of the current
transformer carrying the total current was continuously saturated whilst the
others in the group performed ideally. This voltage is not usually very high in
other applications but it could be in bus bar protective schemes, because the
current levels in the event of faults near busbars may be great and the resist-
ances of the secondary windings of high-ratio current transformers are high.
Whilst a relay could be set to operate at any voltage level, this voltage would
have to be produced at the fault-setting level and each current transformer in
a balancing group would require the necessary exciting current (Ie) in its
Relays in current-differential schemes 297
primary windings, i.e. the fault-setting current would be approximately nleA
in which n is the number of current transformers in a balancing group. Clearly
in applications to busbar sections to which a large number of circuits are
connected, the numbers of current transformers in balancing groups may be
high and therefore fault settings will be high unless the exciting currents (Ie),
needed to induce the relay operating voltage, are low. In practice exciting
currents can be made low by using current transformers with cores of large
cross-sectional area, but obviously there is a practical limit and it must be
recognized that an increase in core dimensions is accompanied by an increase
in secondary winding resistance which would increase the relay setting voltage
needed. This could be counteracted by using conductors of larger cross-
sectional area for the secondary windings which would cause a further in-
crease in the cost and size of the current transformers.
There are usually several circuits on a busbar which are connected to sour-
ces. As they are connected in parallel their effective impedance is relatively
low. In addition, in the case of very high voltage installations there may be
several solidly-earthed neutral points. As a consequence, fault currents on such
busbars will usually be high and therefore, unlike the situations which can
arise on other equipment such as power transformers, very sensitive fault
settings are not necessary.
It will be clear that fault studies must be done when protection is to be
applied to a bus bar installation to determine the levels of the internal and
external fault currents which may flow. The behaviour of the proposed scheme
must then be examined to determine the relay voltage settings which must be
used to ensure that operation will not occur when external faults are present.
Thereafter, the minimum internal fault current which will cause .operation
must be determined and the acceptability of this value must be considered.
8.7.2 Low-impedance relays

It will be apparent from the preceding section that satisfactory performance
may not always be obtainable from busbar protective schemes which include
high-impedance relays and indeed many schemes produced in the last 40 years
have used biased low-impedance relays.
Again it is desirable that the possible causes of imbalance be reduced to a
minimum by using both interconnecting conductors of relatively large cross-
sectional area and therefore of low resistance and current transformers with
low secondary current ratings, e.g. 1 A. The current transformers should be of
the same design and well matched.
A particular fault-current setting can be provided by setting the relays in the
zones to the appropriate values; for example, in a scheme using current trans-
formers of ratio 1000/1 and relays set to operate at 0.2 A, operation would be
obtained at a little over 200 A, the exciting currents of the transformers being
low at these current and voltage levels.
Relays with such low settings and without any form of biasing would almost
certainly maloperate when large currents were flowing to faults outside their
protected zones. Biasing must therefore be provided and it must clearly be
effective under all possible conditions.
It is evident that the greatest current will flow in the circuit on which a fault
is present and bias provided by this current would be effective. Clearly,
however, no advance indication is given of the circuit on which a fault will
occur and therefore bias must be derived from all the incoming and outgoing
circuits of a protected zone. It would not be practicable to provide relays with
a bias winding for each circuit of an extensive busbar zone and therefore the
individual currents must be summed to produce an input to, preferably, a
single bias winding.
A direct summation of the instantaneous currents in the individual circuits
would be zero at all times when external faults were present and therefore a
satisfactory bias input could not be derived in this way. In practice as was
shown in section 5.4.5 (page 167), no form of direct summation of the alter-
nating currents can provide satisfactory bias inputs which are independent of
the locations of external faults, i.e. the circuits on which they occur.
A satisfactory method which has been in use for over 40 years is to rectify
the secondary currents of the current transformers in the individual circuits of
a protected zone and then sum them. In this way a bias input, proportional to
the external fault current but independent of the circuit on which it is present,
is obtained. This bias input is also produced when internal faults occur and
therefore a fixed percentage bias is effective for all types of faults and this
clearly prohibits the use of biasing of 100% or more.
A typical protective scheme of the above form was shown in Fig. 5.18.
A major difficulty, which arises when applying schemes of this type, is to
determine the amount of biasing which must be used to ensure that operation
will be prevented from occurring when external faults occur. It will be ap-
preciated that the imbalances which could cause such incorrect behaviour can
be caused by several factors, all or some of which may be present on any
occasion. Some factors such as the resistances of the interconnections between
the current transformers in a balancing group and their relay could be deter-
mined and their effects could be assessed. Other factors such as the possible
residual fluxes in current transformer cores and various non-linearities make
it impossible to make relatively simple assessments of behaviour. Accurate
allowances for the effects of the saturation of current transformer cores are
also difficult to make.
This situation, together with the vital need to ensure that busbar protective
schemes will perform correctly, led to the introduction of the conjunctive
testing of schemes on the premises of manufacturers. In these tests the actual
current transformers to be used on a site were tested with the protective
scheme to be used with them. High test currents, provided at low voltages by
laboratory machines, were supplied to the primary windings of the current
transformers to simulate the conditions which could arise in service and in this
way the suitability of the scheme and its settings was ascertained.
In more recent years, the development of simulation techniques and accurate
mathematical modelling have enabled theoretical studies of schemes to be
undertaken. It must be recognized, however, that there are no simple methods
of assessing the performance which will be obtained from a particular protec-
tive scheme incorporating biased low-impedance relays. It must be added,
nevertheless, that many such schemes have been installed and that they have
performed excellently over many years.
8.7.3 Low-impedance relays with compensation for

current-transformer saturation
It will be clear from the two preceding sections that it can be difficult to apply
current-differential protective schemes to busbars. When high-impedance re-
lays are used very large current transformers may be needed so that the
exciting currents required to provide the e.m.f.s needed to operate the relays
are not so high that acceptable fault settings cannot be obtained. Alternatively,
when low-impedance relays are used lengthy conjunctive testing may be
needed to assess the suitability of protective schemes for particular applica-
tions and again large current transformers may be needed.
Because of the introduction of modem electronic equipment and acceptance
of its use in important protective schemes, other features can now be provided
in busbar protective schemes to enable the necessary performance to be ob-
tained without the need for very large current transformers.
It has always been apparent that the imbalances caused during external fault
conditions by mismatching of the current transformers and the resistances of
their secondary-circuit resistances are small compared with those which can
result if the core of one of the current transformers saturates whilst those of
the others in its balancing group do not. It will be evident that the current
transformer in the circuit on which the fault is present will be the one in which
saturation will occur because it will have the greatest primary current and loss
of secondary output from it will cause very significant imbalance.
It was therefore realized that current-differential schemes could incorporate
equipment which would monitor the current transformer outputs and provide
signals to inhibit relay operation during periods when the core of any current
transformer is saturated.
This technique has been developed and used in a scheme [2] produced by
GEe Measurements of Stafford, England. The basic feature on which the
technique is based is that the secondary current of a current transformer with
a resisitive burden collapses to zero when saturation occurs and it remains at
zero until the time when the next zero crossing would have occurred. There-
after normal behaviour is obtained until saturation occurs again and the pro-
cess is then repeated. This behaviour, which was discussed earlier in section
2.3.1 (page 78), would not be obtained if the secondary circuit was highly
inductive, but this is not the case in current-differential schemes in which the
resistances of the current transformer secondary windings and the intercon-
nections form the burdens.
The detection of the sudden collapse of the secondary current of a trans-
former during either a positive or negative excursion could be detected in
several ways. The method developed by GEe Measurements, referred to
above, is shown in simplified form in Fig. 8.10(a). The primary winding of an
auxiliary current transformer is connected in the secondary circuit of each
main current transformer. The output of the auxiliary transformer is full-wave
rectified and the resultant current passes through a resistor (R), connected in
Main current
transformer (a)
Output when the core

is saturated
(b)
D D D D (c)
Fig. 8.10 Detection of current transformer saturation. (Reproduced from MBCZ 10,
parallel with a rectifier-capacitor circuit. A voltage (up) proportional to the
rectified current is produced across resistor R as shown in Fig. 8.1O(b). A
voltage (uQ) derived from the rectifier capacitor circuit is provided at point Q
in the circuit. This voltage is approximately equal to half the voltage (up)
during the earlier part of each half cycle as shown in Fig. 8.1 O(b). The voltages
Up and uQ are applied to the inputs of a comparator which provides an output
when the voltage uQ exceeds the voltage Up and this output is used to close a
switch which short circuits the interconnecting conductors and thus the oper-
ating winding of the relay.
It will be appreciated that inhibiting pulses are produced at each zero
crossing of the current waveform but these will be of short duration when
saturation of the transformer core does not occur. However, when significant
core saturation occurs longer inhibit pulses will be obtained as shown in Fig.
8.1O(c). Clearly this will ensure that operation will not be caused by the effects
of core saturation when faults are present outside a protected zone. Similar
action would occur should a current transformer saturate during a fault within
a protected zone, and in these circumstances inhibit signals would also be
generated. However, a large current would have to be flowing to cause the
saturation of a core and therefore the current levels during the period when
there were no inhibit signals would be great enough to ensure that relay
operation would be obtained.
Further details of this particular scheme and other modem bus bar protective
schemes can be obtained from reference [2].
An alternative scheme, which will not be affected by the saturation of
current transformer cores, is currently being studied. In this scheme, the
currents in the circuits of a protected zone would be sampled many times
during each half cycle. Each group of samples would be summed algebraically
and clearly under healthy conditions the sums should always be zero. If the
core of a transformer saturated, however, its secondary current would be
affected and during such periods the groups of samples would not sum to zero.
This is somewhat similar to the scheme described above but in this case relay
operation would not be initiated if more than a certain percentage of the sums
was zero. Clearly it is not likely that the sums would be zero if a fault was
present within the protected zone.
8.7.4 The duplication of current-differential protective schemes

The numbers of faults which occur on busbars are very low because of the
levels of insulation associated with busbars and the spacing between adjacent
phase conductors and to earth and because of other arrangements such as the
installation of voltage limiters at the incoming ends of overhead lines.
Current-differential schemes applied to the zones of busbars are therefore
required to operate very infrequently but they must nevertheless balance
and not initiate the tripping of circuit-breakers for a wide range of healthy
conditions which may involve the individual circuits carrying different load
currents and this process must continue through all of every day. They must
also cope with more onerous conditions when faults occur on the circuits
connected to their busbars, at points outside the zones covered by the busbar
protective equipment and again balance of the current-differential schemes
must be maintained. Because the numbers of faults of this type will always
exceed the small number of busbar faults it is imperative that steps be taken
to reduce the possibility of zones of busbars being disconnected unnecessarily
because of the maloperation of a busbar protective scheme.
In practice this is achieved by applying two separate and independent pro-
tective schemes to cover each section of a set of busbars and it is so arranged
that both schemes must operate before the tripping of the appropriate circuit-
breakers may be initiated.
Clearly, complete duplication of all the current transformers and the con-
nections and relaying equipment would be very costly on major busbar instal-
lations in which there are main and reserve busbars and several bus-section
and coupler circuit-breakers. In such applications, therefore, a compromise
solution is usually accepted in which current transformers and relaying equip-
ment are provided to protect the various zones as described earlier. Extra
current transformers and relaying equipment are then provided to form an
overall zone covering the complete busbar installation. With such an arrange-
ment, which is shown in Fig. 8.11, current transformers must be included in
all the circuits connected to the complete installation but they are not needed
adjacent to the busbar section and busbar coupler circuit-breakers. In addition,
extra auxiliary switches do not need to be provided on the busbar-selector
switches in each of the circuits, because the current transformers do not have
to be switched into different zones when the positions of selector switches are
changed.
Should a busbar installation be so constructed that both earth faults and
interphase faults may occur then, as before, the extra current transformers
must be divided to form three separate balancing groups, i.e. one associated
with each of the phases, and three relays must be provided. In those installa-
tions, however, where only faults to earth are possible then, again, all the extra
current transformers can be connected to form one overall group and, of
course, only one extra relay is needed.
It will be recognized that this compromise arrangement has the slight dis-
advantage that the extra scheme will operate should a fault occur anywhere
on a busbar installation and therefore if a fault occurred in a zone and not only
the relays associated with that zone operated but those in another zone also
operated wrongly, then two zones of the busbars would be disconnected
instead of one. Another possibility is that the relays associated with a faulted
zone might not operate whilst those associated with a healthy zone might do
so incorrectly. In these circumstances the fault would not be cleared, a result
which would have occurred even if a duplicate scheme had not been fitted,
Manual and automatic testing 303
Protected zone
Reserve ~-----------------------------l
busbar I
I I
Bus-coupler I I
circuit-breakers:
I
x Bus-section
x I
I
I
I circuit-breaker I
Main ~1~~-+~--~4---X--~-r--~~~1-~-
bus bar
Selection
switches
Circuit
breakers
Current
transformers
Fig. 8.11 Overall protective zone.
but, in addition, a healthy zone would have been disconnected unnecessarily.

The possibilities of such happenings are considered to be so small that the
advantages of the compromise arrangement justify its usage.
8.8 MANUAL AND AUTOMATIC TESTING

As stated earlier, switchgear installations are so designed and constructed that
faults occur only rarely on their bus bars, but nevertheless, when they do so,
large fault currents will flow and considerable damage may result unless they
are cleared quickly. It was to limit such damage that busbar protection was
developed and applied in spite of fears that it might operate incorrectly on
occasions and that large and unnecessary disruptions of supply could thus be
caused.
To reduce the risks of circuit-breakers being opened incorrectly or failing
to operate on the infrequent occasions when it is necessary, manually-operated
or automatically-operated testing equipment has been produced for use with
busbar protective schemes. Such equipment enables checks to be made by
operators or automatically at regular intervals. During such checks it is necess-
ary that the protective equipments be inhibited from opening circuit-breakers
and thus for short periods the protection is ineffective. During the checks,
signals are injected to simulate both internal and external faults and the
behaviour of the protective equipment is monitored to determine whether it is
functioning correctly.
The inclusion of checking equipment does increase the complexity of a
protective scheme and there are those who feel that its reliability is thereby
reduced. As a consequence some supply authorities do not allow testing equip-
ment to be incorporated in busbar protective schemes.
Clearly the desirability of including such equipment should be considered
when schemes are to be installed and it must be recognized that the advantages
and disadvantages of particular arrangements must be studied.
8.9 THE FUTURE
Current-differential protective schemes applied to busbar installations have

performed extremely well over many years. Their basic principle, namely
Kirchhoff's law, that the algebraic sum of the currents entering a healthy
network must be zero at all times, is so simple that it must continue to be used
for such applications. These schemes have the advantages that they can be set
to operate rapidly when low-current internal faults occur and that they can
discriminate readily between faults internal and external to their protected
zones.
When conventional electro-mechanical relays are used in these schemes,
quite large current transformers and connecting cables may be required to
ensure that unacceptable imbalances will not occur when high fault currents
flow to points outside the protected zones. Electronic relays have already been
produced which enable smaller current transformers to be used and it seems
certain that further developments will use digitally-processed signals obtained
from relatively small current transducers to determine whether the circuit
currents sum to zero or not. It has been stressed earlier that busbars perform
such a vital role in the operation of a power system that their protective
equipment must be extremely reliable. Because of this, it will be necessary
that electronic relaying be produced to very high standards. Very secure power
supplies and even duplicate supplies will be necessary and built-in checking
features must be incorporated in all the equipment. In addition, the present
practice of applying duplicate sets of protective equipment to busbars will
almost certainly continue.
An area where the introduction of microprocessor technology is expected to
lead to further developments, is to provide busbar protection as part of an
integrated protection system. In such schemes signals needed for busbar pro-
tection are shared with other protective and control functions. Further details
may be found in references [3], [4], [5] and [6] and also in section 15.5 of
this book.
An example of a numerical bus bar protection relay is the REB 500
[7,8]. The flexibility of this system is such as to allow either a centralized
or a decentralized installation using fibre optic links for communication.
The relay incorporates two algorithms, a stabilized differential current
algorithm and a directional phase current comparison algorithm.
Additional features are also provided such as breaker-failure protection,
event and disturbance recording. A module may be added which allows
The future 305
interfacing to substation automation. The relay can work as part of the
PYRAMID substation automation described in section 15.5 of this book.
A numerical bus bar protection relay employing digital processing from
input sampling through to analogue-to-digital conversion and circuit
breaker tripping is also described in [9].
REFERENCES
1. Leybum, H. and Lackey, C. H. W. (1952) The protection of electric power systems

- a critical review of present-day practice and recent progress, Proc. lEE, 99, (II),
47-66.
2. Low impedance busbar protection MBCZlO, Publication R-4026A, GEC Measure-
ments.
3. Cory, B. J. and Moont, J. F. (1970) Application of digital computers to busbar
protection, lEE Conference on the Application of Computers to Power System
Protection and Metering, Boumemouth, England, May 70, pp 201-209.
4. Udren, E. A. (1985) An integrated microprocessor based system for relaying and
control of substations-design features and testing program, 12th Annual Western
Protective Relaying Conference, Spokane, Washington, October 24.
5. Udren, E. A. Protection function, in IEEE Tutorial course on Microprocessor
Relays and Protection Systems Course Text 88 EH0269-1-PWR, pp. 43-45.
6. Phadke, A. G. and Thorp, 1. S. (1988) Computer relaying for power systems, RSP,
pp. 182-186.
7. ABB (1995) REB 500 Numerical busbar and breaker-failure protection,
Publication 1MRB520002-Ben.
8. Ilar, M., Reimann, B. and Brunner, D. (1997) REB500 - Decentralised
numerical busbar protection with extended functionality, ABB Rev., 5, 24-32.
9. Siemens, A. G. (1992) 7SS5 Numerical busbar protection, Publication
R2.I2.I992.
FURTHER READING
Guide for protective relay applications to power buses, ANSI/IEEE Standard C37.97-
1979.
9
The protection of overhead lines and
cables by current-differential schemes
Current-differential protective schemes, as stated earlier, have the great ad-

vantages that they can discriminate between faults within and external to their
protected zones and can operate very rapidly when required to do so. Their
basic principle, namely Kirchhoff's first law, that the algebraic sum of the
currents flowing into a healthy circuit is zero, is very simple.
The behaviour of these schemes was considered in a general manner in
Chapter 5 and their application to transformers, rotating machines and bus bars
was examined in Chapters 6, 7 and 8. They have been widely applied to
relatively short cables and overhead transmission and distribution lines, which
are often referred to as feeders, since the beginning of this century and are
still being installed today.
Their behaviour and limitations are examined in detail in this chapter after
the following section in which some historical information is provided.

As stated in Chapter 5, C. H. Merz and B. Price submitted an application for
a British patent on a current-differential protective scheme in February 1904.
This was granted and numbered 3896. It included arrangements to cover both
feeder circuits and transformers and the proposed scheme employed the
balanced-voltage principle, the current transformers being so connected that
their outputs opposed each other during healthy and external fault conditions.
As a result, the secondary winding currents were zero under these conditions
and the secondary winding e.m.f.s were of equal magnitudes and opposite
polarities. One end of each secondary winding was earthed and hence only
one interconnecting conductor was needed to protect a single-phase circuit.
The inventors referred to the interconnecting conductors as pilot wires and
this term, which has remained in use, will be used in this chapter. During
internal faults a current flowed in the pilot wire and also in the relay windings,
which were connected in series with the pilot wire, at each end of a protected
line.

308 Protection by current-differential schemes
Merz and Price submitted two further patent applications in 1904 and were
granted patents numbered 11364 and 15796. These covered improvements to
the initial scheme. One proposal enabled the number of pilot wires needed in
three-phase applications to be reduced. The use of auxiliary transformers
which would allow several currents to be fed to separate windings mounted
on a common magnetic core was also proposed. An output could be obtained
from a further winding. These were forerunners of summation and core-
balance transformers and indeed Merz and Price were granted a patent, num-
bered 28186 in 1908, on the latter devices.
Schemes based on the above patents were applied to lines and cables in the
first decade of this century. In an article [1] published in the Electrical Review
in 1908, it was stated that about 200 miles of high-tension network were
protected in the North of England at that time by the Merz-Price system. It
was further stated that the scheme had not just been perfected but had passed
the perfected development stage for some 18 months or more. Most import-
antly, numerous faults of the most severe description had been isolated without
visible shock on the system, and, better still, without opening healthy sections.
It was further stated that the relays were of the simplest design. They
consisted of an ordinary electromagnet with a very light armature having, for
the sake of rapidity in action, a small air-gap. They clearly did not possess
any biasing windings.
It was presumably found, as the lengths of the lines and cables being
protected increased, that incorrect operations of Merz-Price schemes were
occurring because of the capacitance currents which were flowing between
the pilot wires and earth during external faults. This situation is illustrated in
Fig. 9.1(a), from which it can be seen that half the capacitance current to earth
flows in each of the relays and clearly this current is proportional both to the
length of the protected line and the current flowing in it.
To overcome this situation, special compensated pilot wires, covered by a
British patent, were introduced. These cables incorporated metallic sheaths
which surrounded the interconnecting conductors, there being a break in each
sheath at the mid-point of a cable run. The protective arrangement with such
cables was as shown in Fig. 9.1(b), from which it can be seen that the
capacitance currents between a pilot wire and its sheath do not return through
the relays.
As line operating voltages and fault current levels increased further, more
complex current-differential schemes were produced. One of these, which was
designated as 'split-pilot' protection, was described in a paper [2] presented
at the International Conference on Large Electrical High-Tension Systems,
which was held in Paris in 1931. This scheme, which incorporated summation
transformers and had three pilot wires, was connected as shown in Fig. 9.2.
The use of the parallel-connected pair of pilot wires ensured that imbalance
currents did not flow in the relays during healthy conditions and the inclusion
of the diverter reactors reduced the currents which flowed in the relays be-
Current Current
X E2f] J[
transformer A transformer B
x-
IpA IpB
End A Protected circuit End B
~ ----~~~~~---
Relay
operating
-
iCA = icl2
Pilot wire -
iCB = ic/2
-
Fault
Relay
operating
winding
winding
(a)
Current Current
__x
transformer A IpA IpB transformer B
X-
______~p~r~ot~e~ct~e~d~C~irC_u~it~_-
J[
End A EndB
~ E2f]
Pilot wire Fault
t - - - - -.....--E) Cable sheath O - -..... -_-----i

iCB Relay
Relay
operating operating
winding (b) - winding
Fig. 9.1 Operation when an external fault is present. C is the pilot wire capacitance
to earth in (a), and Cs is the pilot wire capacitance to the sheath in (b).
cause of mismatching. This scheme was used on three-phase lines operating

at voltages up to 220 kV (line).
A range of special relays was developed for use in Merz-Price schemes.
One, which was described as an 'anti-surge' relay, was covered by British
patent No. 235605. This relay had two movements, one being electromagnetic
and the other purely mechanical. The two movements were mechanically
coupled together. Because of the construction, the effective restraining torque
on the electromagnetic movement varied with the current in its operating
winding, being large when a high impulse current flowed as a result of an
external fault and low when a comparatively low current flowed.
Protection schemes which were developed subsequently are described in the
later sections of this chapter.
r-----------CD ~ ----------~
Fig. 9.2 Split-pilot protective scheme. (Reproduced from Leason and Leybum, 1931,
CIGRE Conference, Paper /06 with the permission of CIGRE.)
(1) Common pilot No.1. (2) Split-pilot No.2. (3) Split-pilot No.3. (5) Split-pilot transformer.
(6) Relay. (1) Mid-point trillling connection. (8) Small tuning condenser. (9) Current transformers.
(10 Air-gap summation transformer. (11) Diverter reactor.
9.2 CABLES AND OVERHEAD TRANSMISSION AND

DISTRIBUTION LINES
Cables are much more expensive per unit length than overhead lines and
therefore they tend to be used primarily for distribution circuits in built-up
areas where overhead lines would be considered to be unacceptable. They are
usually relatively short, typically 20 km or less. They are also used in a few
extra high voltage transmission circuits; for example, in areas of great scenic
beauty where overhead lines would be too obtrusive. Again they are usually
relatively short, transmission being continued by overhead lines when areas
of less beauty are reached.
Because cables incorporate dielectric materials with breakdown voltages
well above their operating levels and they are buried in the ground, the
incidence of faults on them tends to be low. They may, however, be damaged
as a result of subsidence or when excavations are taking place.
Overhead lines are engineered to high standards but it must be accepted that
breakdowns will occur when extreme weather conditions are experienced. The
Cables and overhead transmission lines 311
conductors are under tension and the stresses may increase greatly when thick
ice coatings build up on them and, as a result, fractures may occur. Conductors
may also swing violently during stormy conditions and again breakages may
result. In addition, flashovers between conductors or to ground or across string
insulators may occur during lightning storms. These factors, coupled with the
great lengths of many transmission lines, make it inevitable that significant
numbers of faults will be experienced on them. Whilst many of these faults
will be between a phase and ground, it must be accepted that interphase faults
will also be experienced and therefore protective equipments should be able
to detect all types of faults.
Whilst large currents flow in the event of many faults on cables and over-
head lines, there may be occasions when quite low currents will flow. As an
example, a broken conductor may make poor contact with rocky ground,
causing little fault current to flow. It is therefore desirable that protective
schemes should have sensitive fault settings.
A factor which should be recognized is that arcs will be present in most of
the faults which occur on overhead lines. It is therefore desirable that faults
be cleared very rapidly so that the resultant burning of vital parts is minimized
and so that the production of ionized gases is limited. In many cases, if rapid
fault clearance has been effected, it will be possible to reclose a line after a
dead time of say 0.2 s, during which period the ionized gases will have
dispersed, and normal operation will then be resumed. This practice, which is
termed 'auto reclosing', is common on overhead lines.
The presence of capacitance between pilot wires was referred to in section
9.1 and, of course, there is also capacitance between the conductors of over-
head lines and cables. Clearly their capacitances are dependent on the effective
diameters of the conductors and the spacings between them and also on their
lengths. The presence of capacitance currents causes the input and output
currents of the individual phase conductors of the lines and cables to be
unequal and therefore exact balance cannot be expected to occur in current-
differential protective schemes applied to them.
In practice, however, the capacitance currents of relatively short lines or
cables, say up to 20 miles in length, are very low relative to the rated currents.
As an example, the capacitance current per phase of a 20-mile long, three-
phase, 132 kV overhead line rated at 440 A is about 7 A, i.e. only 2%, a level
well below the lowest fault settings likely to be used. During limiting condi-
tions when a short circuit is present just outside a protected zone, the voltage
at the faulted end of the circuit will be almost zero and therefore the capacit-
ance current will be roughly half its normal value and extremely small relative
to the fault current.
It is usual, therefore, when considering the behaviour of protective schemes
to be applied to relatively short lines or cables to neglect the capacitance
currents and assume that the input and output currents are equal and this
practice will be assumed in the following sections.
9.3 THE APPLICATION OF CURRENT-DIFFERENTIAL
PROTECTIVE SCHEMES
As explained in Chapter 5, these schemes may be implemented in two basic
forms, i.e. circulating current and balanced voltage. Their application to over-
head lines and cables was considered in Chapter 5 but more detailed informa-
tion is provided below.
9.3.1 Comparison arrangements

All current-differential schemes must compare the currents flowing at the two
ends of their protected zones. In applications to equipment such as three-phase
transformers and rotating machines, the individual phase currents are com-
pared using four interconnecting conductors, as was indicated in the earlier
chapters. This practice is acceptable because the lengths of the interconnecting
conductors are relatively short and their cost is therefore low.
In applications to overhead lines and cables, however, the interconnecting
connections may be quite long, say 15 miles, and in such cases the cost of
providing four pilot wires could be very significant. It is therefore the practice
in these applications to perform a single comparison between quantities
derived from the three-phase currents at each of the ends of protected zones
so that only two pilot wires are needed.
Two possible methods of producing single quantities from sets of three-
phase currents for comparison purposes are examined below.
Sequence networks
Sequence networks of various forms may be used to determine the positive-
and negative-sequence components of sets of three-phase currents and the
zero-sequence component may be derived by simply summing the three-phase
currents.
Zero-sequence and negative-sequence components are not present in the
currents associated with balanced three-phase faults and therefore neither of
these components could be used on their own or together for comparison
purposes. Positive-sequence components are present however in the currents
associated with all types of faults. They represent only one third of the mag-
nitude of the current which flows in the event of a single phase to earth fault
and therefore it would be difficult to provide sensitive earth fault settings if
only this component was used. In practice, therefore, comparison quantities
obtained from at least two of the sequence outputs are needed to provide
acceptable performance.
A factor which should be appreciated is that the magnitudes of the output
signals derived from two sequences are dependent on the phases on which
faults are present and this would clearly cause the sensitivity to such faults to
Current-differential protective schemes 313
'a' phase
fault
rlIb
If
/1C 11a ="3 +
)b ' ':-'-
12c
3 +
~/",=;
-lOb
-lac
'b' phase
fault
'c' phase
fault
Fig. 9.3 Output current derived from positive- and zero-sequence currents in the ratio
1 to 4, for single phase faults on phases 'a', 'b' and 'c'.
vary. As an example, if a signal were derived from the 'a' phase positive-
sequence current and the zero-sequence current, in the ratio of I to 4, then the
ratios of the magnitudes of the outputs for single phase to earth faults of a
given magnitude on phases a, band c would be 5, ...)13, ...)13, respectively as
shown in Fig. 9.3.
A comprehensive study of the possible quantities which can be used for
relaying purposes was undertaken by Adamson and Talkan [3].
Variations of sensitivity can be accepted, but the networks needed are quite
complex when signals are derived from symmetrical components because
phase shifting of two of the input currents must be performed to determine the
positive and/or negative sequence components.
This situation has been accepted and such networks have been used in
differential protective schemes in the United States. In Britain, however, the
use of summation transformers or summation techniques has been preferred
and they are still used in differential schemes.
Summation transformers
A summation transformer may have three input windings supplied with the
currents from the secondary windings of the current transformers associated
with the three phases of a protected system or network. These input windings,
which each have a different number of turns, would be mounted together with
an output winding on a magnetic core.
The usual, and more economical, arrangement of summation transformers,
which may be used when the current transformer secondary windings are
connected in star, has a single, tapped input winding, and one or more output
windings. Such a transformer is shown in Fig. 9.4. It will be seen that the
common or neutral connection of the secondary windings of the current trans-
formers is connected to one of the tappings (x, y, z) on the primary winding
of the summation transformer. The other ends of the secondary windings of
the current transformers are connected to the terminals a, band c of the
primary winding of the summation transformer. Because of the asymmetry of
the arrangement, different sensitivities are obtained for the various current
combinations which may be experienced during power-system faults.
The input m.m.f.s to a summation transformer are proportional to the num-
ber of turns of its primary winding through which the input currents flow.
When phase to phase faults of a given magnitude (I) are present, the input
m.mJ.s are as follows:
[Nab for a fault between phases 'a' and 'b'.
[(Nab + Nbc) for a fault between phases 'a' and 'c'.
[Nbc for a fault between phases 'b' and 'c'.
It is usual for the turns Nab and Nbc to be the same, in which case the
sensitivity to faults between phases a and c will be twice as great as it is for
the other interphase faults.
For a three-phase balanced fault of magnitude I, i.e. I/a I = lIb I = I/c I = I,
the input m.m.f. is I/aNab -/cNbc I. For a transformer in which Nab = Nbc this
m.m.f. would have a magnitude of -,J3INab , i.e. the sensitivity to three-phase
faults would be --/3 times as great as that to faults between phases a and b and
band c.
The sensitivities to phase to earth faults, which are normally higher than
Current transformer Summation
secondary windings transformer
E seCOndary
E
S winding
rertiary
I winding
zo-:,....,........J
Primary winding
Fig. 9.4 A summation transformer.

those for interphase faults because Ncx is greater than Nab and Nbc also vary
with the phase on which a fault occurs being greatest for faults on phase a and
lowest for faults on phase c. The desired sensitivities may be selected by
connecting the star point of the secondary windings of the current transformers
to the appropriate tapping (x, y, z) on the primary winding of the summation
transformer.
Whilst it is the usual practice, as stated above, to use summation trans-
formers in which the turns Nab and Nbc are the same, there are applications
where it could be advantageous to make them different.
As an example, if a line is supplying a transformer which has delta-
connected windings on the line side and star-connected windings on its remote
side, then the phase currents on the line side will be in the ratio - 2 : 1 : 1 in
the event of an interphase fault on the remote side of the transformer. This
situation was discussed earlier in Chapter 6. It will be seen that should it be
the 'b' phase current which has double the magnitude of the others and
because it will be in anti-phase with the others there would be zero m.m.f.
applied to summation transformers in which the turns Nab and Nbc were equal.
If this occurred in a current-differential scheme applied to a line the quantities
to be compared would both be zero and relay operation would, quite correctly,
not occur, the fault being outside the protected zone. If, however, a current
transformer core at one end of the scheme saturated, imbalance would occur
and currents could flow in the operating windings of the relays at each end of
the scheme. No biasing current would be available however at the end of the
scheme at which the current transformers were behaving correctly because
there would be no output from the associated summation transformer and
incorrect operation of the relay at this end could occur.
It will be found, therefore, in some cases, that the primary winding turns
Nab and Nbc on summation transformers are not the same and as a result the
sensitivities to each of the types of interphase faults are different.
9.3.2 Circulating-current schemes

The basic form of a single-phase circulating current scheme is shown in Fig.
9.5(a). To achieve the ideal situation in which no current will flow in the relay
unless there is a fault within the protected zone, i.e. between the current
transformers, the following conditions must be met:
(a) the circuit must be physically symmetrical,
(b) the current transformers must be physically identical,
and
(c) the current transformers must produce identical secondary currents at each
instant when they have the same currents in their primary windings.
To meet the first condition the resistances of the pilot wires on each side of
the relay must be the same, i.e. Ral + Ra2 = Rbi + Rb2. If this were achieved by
Current Current
End A
~
A X--...;..;.;;.;.;=;.;;..;;=::.:.:..._-X
transformer 'PA
Protected circuit
Ips transformer
~
B End B
= =
Pilot wire
(a)
Pilot wire
Pilot wire
Pilot wire Relay
(b)
X
End A
IpA
X
Protected circuit
~ --...;..;.;;.;.;=;.;;..;;=::.:.:..._- I:,R~~;J
Ips
End B
s Ns f Pilot wire
~ ~,-~~--~~~~--~~--~~
'sA
Relay Rop
Pilot wire
(c)
Fig. 9.5 Circulating-current schemes.
connecting the relay between the mid points of the interconnecting cables, so
that Ral =Ra2 =RbI =Rb2, it would be remote from both ends of the protected
zone and in a position where it might not be readily accommodated. In addition
cables would have to be run to enable the relay to initiate the tripping of the
circuit-breakers. The alternative arrangement, shown in Fig. 9.5(b), in which
two relays are connected in series with each other and in which Ral = Rb2 and
RbI = Ra2 = 0, would provide both the necessary balance and position the relays
near the circuit-breakers. It has the disadvantage, however, that an extra
interconnecting conductor is needed.
An alternative arrangement in which a measure of imbalance is accepted is
shown in Fig. 9.5(c). The relays, which are mounted at the ends of the pro-
tected zone, have their operating windings connected across the interconnect-
ing conductors at points where the voltages are not zero during healthy or
external fault conditions. The application of restraint to the relays by biasing
windings carrying the current which circulates during such conditions can
ensure that the relays will not operate. The necessary perfonnance may never-
theless be obtained when faults occur within the protected zone.
In general, circulating-current schemes have not been widely applied to
overhead lines or cables in the past, the balanced-voltage schemes described
in the following section having been preferred. Some relatively modern
schemes, however, do operate on the circulating-current principle.
9.3.3 Balanced-voltage schemes

In these schemes the outputs of the current transfonners at the two ends of a
protected zone are effectively opposed to each other and therefore under ideal
conditions no current flows in the interconnecting cables when the protected
circuit is healthy. This was the arrangement patented by Merz and Price in
1904. When a fault occurs within the protected zone, balance is not achieved
and a current circulates in the interconnecting cables. This may be detected
by connecting the operating windings of the relays in series with the intercon-
necting cables at each end of the protected zone. These relays are then ideally
situated to initiate the tripping of the circuit-breakers.
As stated in section 9.1, such schemes were widely applied to transmission
and distribution lines and cables in the first decades of this century. As the
current and voltage levels rose, the need for biasing features became evident
and more advanced schemes were developed, two of these being' Solkor' and
'Translay' which were produced in Britain by A. Reyrolle and Co Ltd and
Metropolitan-Vickers Electrical Co Ltd respectively. These schemes are de-
scribed in some detail below to illustrate features needed to provide both the
necessary fault settings and discrimination between faults within and external
to their protected zones.
So/kor protective scheme

The circuit of the original Solkor scheme [4] which was applied to three-phase
transmission or distribution circuits is shown in Fig. 9.6, from which it will
be seen that the current transfonners, which were connected in star, all had
the same ratio. Ideally they should all have been of the same design and been
well matched. Because of the various features within the protected scheme,
current transfonners with solid cores could be used rather than the large
transformers with cores having distributed air gaps (DAGs) needed by earlier
schemes, such as split-pilot protection, which was referred to earlier in section
9.1. In fact, the name 'Solkor' was derived from reference to the 'solid cores'
of the transformers used with it.
To enable comparison to be achieved using only two interconnecting con-
ductors, summation transfonners were included.
CT
Red End A E ~
_ L - Feeder - Yellow ..or- ~ a
:.. ! ~ J Blue...!.. -!2 b
t A Arrows represent flow of current --..f.; ;.l,2 c
i i I . I i iS r:- i TC
t
C-Tuning-condenser
CT-Current-transformers
OC-Operating-coil
R~elay
R~estraining-coil
L
RT~elay-transformer
RI
S-Summation-transformer
I SB-SoIkor box
I TC-Trip-coil
I
TSI
I
I
I ~I ________________
<D I ~I
I ____ ~
Fig. 9.6 Connections of Solkor plain feeder protective scheme. (Reproduced from
Solkor protective schemes, Pamphlet 80613-54, with the permission of Reyrolle Pro-
tection.)
During healthy conditions or when faults were present external to the pro-
tected zone, balance had to be maintained between the e.m.f.s induced in the
secondary windings of the summation transformers. To ensure that this would
be achieved, the summation transformers had to be well matched. For condi-
tions such as earth faults of high current levels, large m.m.f.s were applied to
the summation transformers. To prevent the r.m.s. values of the secondary
winding e.m.f.s of the summation transformers being very high in these cir-
cumstances and very large voltages thus being present between the pilot wires,
the summation transformers were so designed that their cores saturated at
these levels. As a result, the higher secondary e.mJ.s were not sinusoidal but
contained harmonic components and a limited fundamental component. The
relationship of r.m.s. voltage to current was as shown in Fig. 9.7.
Ideally, no current should have circulated in the interconnecting conductors
during the above conditions but in practice, because of inevitable mismatch-
ing, a current containing fundamental and harmonic components did flow. In
addition, the presence of a large balancing e.m.f. caused significant current to
flow in the capacitance between the interconnecting conductors and half of
this current was fed from each end of the scheme. Clearly the magnitudes of
these capacitance currents were proportional to the length of the protected
zone, i.e. the length of the pilot wires. A further factor which had to be
recognized was that the presence of harmonic components in the balancing
e.m.f.s caused corresponding components to be present in the capacitance
current and indeed, because of the inverse relationship of the reactance asso-
ciated with the capacitance between the pilot wires to frequency, the resulting
harmonic components of the capacitance current increased with their order.
To cope with the above situation, the scheme incorporated relay trans-
formers at the ends of the balancing loop and capacitors were connected across
the primary windings of these transformers to form tuned circuits (RT, C in
200 Voltages for white and blue phases

are lower than those for red phase
2 4 6 8 10 12 14 16 18 20
Straight-through earth-fault current in red phase
(times full-load current) with current-transformers connected to Y1 tap
Fig.9.7 Relation between pilot voltage and primary current. (Reproduced from Solkor
protective schemes, Pamphlet 806/3-54, with the permission of Reyrolle Protection.)
Fig. 9.6). Because of this feature large proportions of the harmonic compo-
nents of the imbalance currents flowed through the parallel-connected capa-
citors and as a result the currents fed to the operating windings of the relays
were predominantly produced by the limited fundamental components of the
summation transformer e.m.f.s.
It nevertheless had to be accepted that currents would flow in the operating
windings of the relays when faults were present outside a protected zone and,
to counteract this effect, the relays were provided with biasing windings
connected to tertiary windings on the summation transformers. The amount of
restraint was dependent on the currents flowing in the primary windings of the
summation transformers, but it was not a fixed percentage bias because of the
core saturation which occurred at high current levels.
When a fault was present in the protected zone, the e.m.f.s produced by the
summation transformers did not balance and a current circulated in the pilot
wires. As an example, in the event of an earth fault on one of the phases,
currents would have flowed to it from each end. E.m.f.s were induced in the
secondary windings of the summation transformers and these e.m.f.s caused a
current to circulate in the pilot wires. As a result, currents flowed in the
operating windings of the relays. Clearly these currents and those in the pilot
wires were dependent on the impedance of the path around which the current
circulated and therefore the fault currents needed to cause operation increased
with the length of circuit to be protected and in practice there was an upper
limit to the impedance which could be present. It was usual to use conductors
with seven strands of 0.029 inch diameter copper and these had a resistance
of about 5 n per 1000 yards. They enabled satisfactory operation to be ob-
tained in applications in which the length of the protected zone did not exceed
15 miles. Whilst the impedance of longer pilot wires could be kept down to
the necessary levels by using cables with conductors of greater cross-sectional
areas, their costs tended to be unacceptably high and, in addition, the effects
of the high capacitances between such conductors would have made it neces-
sary to use very high percentage biasing to prevent incorrect operation.
To enable the necessary sensitivity to internal faults to be obtained in Solkor
schemes, rotary-sensitive relays of the form described earlier in section 5.7
(page 174) were employed. These relays could be set to operate with currents
in the range 35-80 rnA in their operating windings, the required value being
obtained by adjusting the travel and thus the initial air gaps between their
armatures and electromagnets. Restraining windings, supplied from the ter-
tiary windings of the summation transformers, enabled the percentage biasing
needed to ensure that correct discrimination would be obtained up to the
maximum fault-current levels likely to be encountered. To further assist in this
process, weights could be placed on pans mounted on the relay spindles. These
increased the inertias of the movements and thereby slightly increased the
operating times and reduced the possibility of operation due to transient im-
balances.
It will be clear from the above that the fault settings obtainable with this
scheme were dependent on both the minimum current at which the relay would
operate and the interconnecting conductors. As an example, if the scheme were
applied to an overhead line of length 6 miles, with interconnecting conductors
of seven strands of 0.029 inch conductor, and the relays were set to operate
at 55 rnA and if the least sensitive tappings on the summation transformers
(i.e. Ncx = minimum value) were used, the fault settings were as shown in
Table 9.1.
Figures 9.8 and 9.9 show the effects of the lengths of the pilot wires and
the operating currents of the relays on the fault settings.
Because the summation transformers included in Solkor schemes saturated
when their input currents were high, the demands on the main current trans-
formers were limited.
To ensure that acceptable balance would be obtained in the event of high-
current faults outside a protected zone, it was nevertheless necessary that the
current transformers installed at the two ends of a scheme were able to provide
adequate secondary winding e.m.f.s. A large number of the original Solkor
schemes and variants of it were installed around the world and many are still
in use and performing satisfactorily.
Translay protective scheme

This scheme [5] also employed the biased balanced-voltage principle, its
general arrangement being as shown in Fig. 9.1O(a).
Q) 1.15
::J
~
'iii 1.10
c:
'E
g 1.05
(/)
Q)
~ 1.00
C)
c:
E
3l 0.95
~
14
o 40 200 240
Pilot-loop resistance (ohms)
Fig. 9.8 Variation of fault setting with pilot length. (Reproduced from Solkor protect-
ive schemes. Pamphlet 806/3-54. with the permission of Reyrolle Protection.)
1.3
Q)
:::J
a;
> 1.2
a;
c
E
0
c 1.1
f/)
Q)
E
:e 1.0
Cl
c
S
Q)
'{l
'5ctI
u.
0.840
Relay-setting (milliamperes)
Fig. 9.9 Variation of fault setting with relay setting. (Reproduced from Solkor protec-
tive schemes, Pamphlet 806/3-54, with the permission of Reyrolle Protection.)
It will be seen that the secondary windings of the current transformers at

each end of a protected zone were again connected in star, They were all of
the same ratio and should ideally have been of the same design and well
matched.
At each end, the secondary windings of the current transformers were con-
nected to a summation winding mounted on the centre limb of an 'E' -shaped
electromagnet. As a result, the effect was the same as that produced by the
summation transformers incorporated in the Solkor scheme described above,
namely that the m.m.f.s applied were not solely dependent on the current
levels but also on the phases in which they flowed. This situation was again
accepted because it enabled a single comparison to be made to determine
Table 9.1
Type of fault Fault setting (percentage of
current-transformer
primary rating)
'a' phase earth fault 40%
'b' phase earth fault 48%
'c' phase earth fault 60%
'a-b' phase-fault 240%
'b-c' phase-fault 240%
'a-c' phase-fault 120%
Three-phase fault 140%
whether a fault was within a protected zone or not and, most importantly, it
reduced the number of pilot wires needed to a minimum, namely two.
In the Translay scheme, however, induction-type relays were employed and
the secondary windings on their main E-shaped electromagnets and the wind-
ings on the lower C-shaped electromagnets were connected in series via the
pilot wires, as shown in Fig. 9.1O(a).
During healthy conditions or when faults were present external to the pro-
tected zone, the m.m.f.s applied to the main electromagnets of the two relays
by the currents in the summation windings, would ideally have been of the
same magnitude at each instant, as would the e.m.f.s (e sr ) induced in the
secondary windings. Because the e.m.f.s were connected in opposition, no
current should have flowed in the pilot wires or the windings of the lower
electromagnets of the relays and therefore no torque should have been pro-
duced on the relay discs.
In practice, significant currents could circulate in the comparison circuit
because of mismatching when external faults involving high currents were
present. To ensure that these did not cause the relays to operate, the relays
were biased by copper shading loops fitted to the main E-shaped electromag-
nets. These provided restraining torques proportional to the square of the flux
magnitudes in the main magnets. In addition, permanent magnets were fitted
to provide eddy-current damping of the discs and this feature further improved
the performance of the relays.
Again, capacitance currents (ic) flowed between the interconnecting conduc-
tors during healthy conditions and they would be quite high when external
faults involving high currents were present. These currents flowed in both the
Current Current
End A transformers transformers End B
--------~~~~~~a
--------__~~=r--b
----------~~~~--c
Summation
winding
Secondary
Pilot
winding ~ri--4::::::jr---' wires L.-----i--1---.;::::t--.....
Bias loop
Fig. 9.10 (a) Translay differential protective scheme. (Reproduced from Protective
Relays - Application Guide, 3rd edn, GEC Measurements, 1987 with the permission
of GEC Alsthom Protection and Control Ltd.)
secondary windings of the main electromagnets and the windings on the lower
electromagnets as shown in Fig. 9.1 O(b). In the absence of resistance in the
comparison circuit, the capacitance currents would have lead the e.mJ.s (e sr )
by nl2 rad and they would thus have been in phase with the fluxes in the centre
limbs of the main electromagnets of the relays. In these circumstances, the
fluxes set up by the lower electromagnets would have been in phase with those
in the main electromagnets and the resultant torques produced on the discs
would have been zero. The presence of significant resistance in the pilot wires
affected the above conditions because it caused the capacitance currents to
lead the e.m.f.s (e sr ) by less than nl2 rad and caused the fluxes set up by the
relay electromagnets to be out of phase with each other, the displacement
increasing with increase of the resistance of the pilot wires. This situation was
counteracted by the shading loops fitted to the main electromagnets, which
displaced the phase of the pole-tip fluxes from that of the main fluxes in the
centre limbs of the main electromagnets. The phase adjustment so produced
increased the capacitance current which could be allowed and so enabled the
scheme to be applied to relatively long lines and cables.
In the event of a fault occurring within the protected zone in which currents
were fed to the fault from each end of the line or cable, as shown in Fig.
9.lO(c), fluxes would have been set up in the centre limbs of the main elec-
tromagnets of both relays and a current would have circulated in the compari-
son circuit and in the windings of the lower electromagnets of both relays.
Provided that the faults currents at both ends of the protected zone were above
the fault-setting levels, both relays would have operated and initiated the
opening of their associated circuit-breakers. If, however, a fault occurred
End A End B
Pilot wire Secondary windings on
h upper electromagnets
Windings on lower
tiJ2 Pilot tiJ2 electromagnets
capacitance
Pilot wire
(b)
Circuit Circuit
breaker breaker
-x-
End A X Protected line End B
Fault
]4
-=-
iIB
(c)
Fig. 9.10 (b), (c)

within the protected zone but current was fed to it from only one end of the
line or cable, then flux would only have been produced in the centre limb of
the main electromagnet of the relay at the end from which the fault was fed.
Current would nevertheless have circulated around the comparison circuit, but
only the relay at the end from which the fault was fed would have operated.
This situation, for which the current irB in Fig. 9.1O(c) would have been zero,
might not have been acceptable, in which case arrangements would have been
made to inter-trip the second circuit-breaker whenever one of them was op-
ened by the protective equipment.
This scheme has been widely applied to distribution circuits. In the majority
of applications the following settings were found to be satisfactory:
Least sensitive earth fault 40% of rating
Least sensitive interphase fault 90% of rating
Three-phase fault 52% of rating
Relays with more sensitive earth fault settings were available for applica-
tions in which earth fault currents could have been very low and phase-fault
settings above the rated-current level could be selected in some instances to
ensure that relay operation would not be caused by full-load currents if a fault
occurred on the pilot wires.
The e.m.f.s provided by the secondary windings on the main electromagnets
of the Translay relays were not directly proportional to the currents in the
summation windings even though there were significant air gaps in the mag-
netic circuits. Saturation of the cores, which occurred at the higher current
levels, was allowable because of the close matching achieved during manu-
facture. As a result, the r.m.s. voltages applied between the pilot wires did not
exceed about 180 V. As with the Solkor scheme, the voltage waveform did
become distorted and high peak values could be produced when external faults
of high current levels were present. To cope with such situations the relay
windings were insulated to withstand test voltages of 5 kV.
It will be appreciated that the currents which flowed around the comparison
circuit when external faults occurred were affected by the resistance of the
interconnecting conductors and could be quite low when the protected line or
cable was long. Because the relay torque was dependent on the fluxes pro-
duced by both its electromagnets, operation could be achieved with limited
current in the winding of the lower electromagnet, but then higher currents
were needed in the summation winding, i.e. the fault settings increased with
the resistance of the pilot wires. In practice it was found that the scheme could
operate satisfactorily with resistances up to 1000 Q in the comparison circuit.
Solkor Rf protective scheme

The original Solkor and Translay schemes have been described in detail in the
preceding sections because they incorporated the features which are necessary
"'l \;J
~.
~
\C
;... ~
.... Current Current
transformers Circuit Circuit transformers
(3
~ ~
Np breakers breakers Np
Vl c End A s 000 f X Protected line X s 000 End Ao ""~.
~ ::s
Q , N, f f
:;:c
I ,-,,; ~
...., ~ ~ ;::
sHof X X ~ ""
a
~
~
::s
o s Ns f 'i'
~. >:l.
C'l Np i
X X ', S Ho f ~
~
::r ~
Pilot wires I IIs --- f ::s
......
3!1> ,(Rpw is resistance of loop) Ns [
PA '... AA. /I AAn .. ' P~
..,
s g.
~
~
Summation
transformer f
when lines are to be protected. They did, however, operate relatively slowly
when internal faults occurred and the relays used in them were quite complex.
As a consequence several other schemes, which employ different techniques,
have been introduced. One of these is the Solkor Rf scheme [6] produced by
NEI Reyrolle, which incorporates simple polarized relays and several recti-
fiers. The basic arrangement of this scheme, which employs the circulating-
current principle, is shown in Fig. 9.11(a).
The resistors R are of greater resistance than half the total resistance of the
two pilot wires (Rpw). When an external fault is present the output windings
of the summation transformers produce equal e.mJ.s which assist each other
and thereby cause a current to circulate around the loop formed by the pilot
wires, causing the voltage distributions during the alternate half cycles to be
as shown in Fig. 9. 11 (b). Because of the arrangement of the rectifiers the
voltages across the relays are always of the opposite polarites to those needed
to cause current to flow in them. In effect the relays are negatively biased, the
bias being dependent on the amount by which the resistance RA exceeds that
of the pilot wires (Rpw).
When faults are present on a protected line the above conditions do not
obtain and the relays at both ends of the scheme operate when the fault-setting
level is exceeded if current is being fed into only one end of the line. In the
event of a current being fed from each end of a line to a fault, operation occurs
at levels below the setting values.
Typical fault settings are given below for the various types of fault:
Phase a to earth 25 %
Phase b to earth 32%
Phase c to earth 42%
V,elayA = VMA - VCA

V,elayB = VNB - VCB
MA
ipw positive
PB
PB
V 1
c" \.. "oga'"
'----------'("L
NB "'B
(b)
Fig. 9. 11 (b)
~0''''J'j f..;.)
~ ::3 1Ji N
;l ::i 00
~
t/}
'"
;:...
ers ~
(JQ "0 N ~~m ~~m
3 a ..., transformers Circuit Circuit transformers C:l
. < .... ~
3 0: Np breakers breakers Np
~ ('1) t/} oEnd A s 000 f X Protected line X s ou fEnd 8 0 a .....,c:;.
0;- ;:s
"0'<
o ~
e v; ..,;:::
o I'\.JW..,' ~000'1 ob
&."Slo X - - - - - - - - - - - - -X

3 ~
("l
.........-'Orr' ;:s
? ~. s Ns f s Ns f ...~
:-,">0
~
..., til Np f I X " Np f i
~
\: ("l 0---,IS __
000_~Ir-r- ---+I-II~~ Jll..; 1 0 c ...
X ~
3 g 11..-rnDf ;:s
'" 3 s Ns f s Ns
<:;!'t [
., Z
0
o ..,""
~
::I ff ~
5-..., Auxiliary ~
o \: ~
transformers ""
~ 3til TA
><
:=:~ Pilot
~"" wire
Q 0 Summation I~ Summation
... ::l transformer f R f transformer
....
~
::l 5- Pilot
0
'"O'~ wire
\:
3
0 ~
.... ;.
'" ....
"0'<
....
0 ::r
!!.
0.",
o ,
Phase a to b 125%
Phase b to c 125%
Phase c to a 62%
Further information may be obtained from reference [6].
Translay S protective scheme (Relay type MBC1)

The original Solkor and Translay schemes described above did not operate
particularly rapidly because of the inertias and travels of their relay move-
ments and therefore schemes incorporating electronic equipment were intro-
duced when it was felt that power-system engineers would consider them to
be sufficiently reliable.
The Translay S protective scheme, which is produced by GEC Measure-
ments, operates on the circulating-current principle. Comparisons are made
over pilot wires and discrimination is achieved using phase comparators of the
type described earlier in section 4.3.3 (page 119).
The basic circuit of the scheme is shown in Fig.9.12, from which it will be
seen that summation transformers are fed by the current transformers at each
end of a protected line. When faults are present at points external to the line,
the secondary windings of the summation transformers provide equal e.m.f.s
which cause current to circulate around the pilot wires and through the oper-
ating and restraining windings of the auxiliary transformers (TA). The phases
of the output voltages from the auxiliary transformers are compared with those
of the voltages produced by the tertiary windings of the summation trans-
formers and the phase comparators are so set that operation will not occur
during such conditions.
When faults are present on a protected line, the currents which circulate
around the pilot-wire loop tend to be lower and the phases of the voltage
outputs from the auxiliary transformers with respect to those provided by the
tertiary windings of the summation transformers are such that tripping of the
circuit-breakers will be initiated by the phase comparators.
The operating times of this scheme are dependent on the fault-current levels
and other factors, the minimum value being about 25 ms.
Further details of this scheme, including the current transformer require-
ments and allowable pilot-wire resistances are provided in reference [7].
9.3.4 Pilot wires
It will be clear from the foregoing that the pilot wires form crucial parts of
current-differential protective schemes.
In many cases special cables are provided for this duty and they may include
additional conductors for other purposes. The individual conductors must be
insulated to withstand the voltages which may be applied between them or to
earth. As shown earlier, the r.m.s. voltages between pilot wires in balanced-
voltage schemes may be limited to relatively low levels because of saturation
occurring in the cores of summation transformers but nevertheless high peak
levels may be reached when high currents are flowing in a protected circuit.
It is therefore recognized that the insulation on the conductors and any equip-
ment directly connected to them, such as summation transformers or relay
windings, should be capable of withstanding voltages of standard levels such
as 5 kV or 15 kV. It will be appreciated that devices such as voltage limiters
or diverters may not be connected between the pilot wires used with some
protective schemes because they would cause high currents to flow between
the conductors and could thus affect the operation of the protective scheme
when a high current was flowing to a fault outside the protected zone.
Whilst overhead air-insulated conductors could be used, they are particular-
ly vulnerable to damage because of their exposure to atmospheric conditions.
They are therefore not normally used for this purpose, buried cables being
preferred in spite of their extra cost.
As an alternative, it is sometimes more economical to rent circuits from a
telephone company. When this is done the circuits must be dedicated, i.e.
connected continuously and solely to the protective scheme, because of the
element of unreliability which would be introduced if circuit switching were
permitted. Special contracts are necessary with clauses guaranteeing that no-
tice will be given of any work that is to be done on the circuits by the telephone
company. The maximum voltages which may be applied to the circuits and
the currents which may flow in them are normally specified by the telephone
company, typical values being 130 V (peak) and 60 mA (r.m.s.) and these may
only be present during fault conditions.
Circuits available for rental have conductors of relatively small cross-
sectional areas, their weights typically being 20 lb per mile and 40 lb per mile.
Their resistances are therefore higher than those of the conductors which are
specially provided for current-differential schemes.
Whilst the schemes described earlier may be capable of operating satisfac-
torily with relatively high resistances in their comparison circuits, they may
not able to comply with the limiting voltage and current conditions set by
telephone authorities. Such protective schemes are not generally suitable for
operation with rented circuits and therefore special or modified schemes which
may use them have been developed, one being described in detail in the next
section.
9.3.5 Protective schemes which use rented telephone circuits

In addition to the limitations imposed by telephone companies on the voltages
which may be applied between the conductors of rented circuits and the
currents which may flow in them, the following conditions may also have to
be complied with:
1. The conductors of the rented conductors may have to be isolated from
most parts of a protective scheme and other circuits and earth by insulation
able to withstand test voltages, typically of 15 kV.
2. Spark-gaps may have to be provided between the secondary circuits of
current transformers and earth to ensure that high voltages, which could
cause breakdowns or that could lead to excessive voltages being present
on the rented conductors, will not be produced.
The Solkor Rf and Translay S schemes both satisfy the various conditions
outlined above. Special transformers are available to isolate the telephone-
circuit conductors and adequate high voltage insulation is included in the relay
circuits and the summation transformers.
Equipment is always provided with the schemes to monitor the telephone-
circuit conductors.
9.3.6 Monitoring of pilot wires

In spite of the precautions which are taken, faults may nevertheless occur on
pilot wires. As examples, underground cables may occasionally be damaged
by subsidence or when excavation work is being done.
The effects of faults on pilot wires depend on their nature and the type of
protective scheme with which they are associated.
An extreme condition which could arise is an open-circuiting of a compari-
son loop because of a break in one of the pilot wires. If the associated
protective scheme operated on the voltage-balance principle such a break
would not cause relay operation should a fault occur outside the protected zone
but it would prevent internal faults from being detected because current would
not be able to flow in the operating windings of the relays. On the other hand,
a break in a comparison circuit associated with a protective scheme operating
on the circulating-current principle would interrupt one of the balancing cur-
rents and therefore relay operation could occur in the event of a fault outside
the protected zone or because of the load current in the protected circuit if the
fault settings were below the full-load level.
A second extreme condition would be a short-circuit between a pair of pilot
wires. In this event and should the associated protective scheme employ the
voltage-balance principle, then a current would flow from each end of the link
and both relays would operate if a fault was present outside the protected zone.
Relay operation could again be caused by the flow of load current in the
protected circuit if low fault settings were being used.
Should the pilot wires of a circulating-current scheme be short circuited,
then current could be diverted away from the operating windings of the relays
when a fault was present within the protected zone and, as a result, opening
of the associated circuit-breakers might not be initiated.
In addition to the extreme conditions considered above, other faults may
arise on pilot wires. In the case of circuits rented from telephone companies,
occasions have arisen where circuits have been replaced by others with dif-
ferent resistances and, as shown earlier, this could affect factors such as fault
settings and the maximum current levels at which correct discrimination will
be achieved when faults occur outside protected zones.
Experience has shown that specially installed cables, buried in the ground,
are very reliable and therefore the likelihood of the above conditions arising
are so small that monitoring of the cables may not be considered justified or
necessary. Some authorities who are not willing to accept the relatively small
risks however, do, arrange for supervisory equipment to be provided. Whilst
the performance record of circuits rented from telephone companies is good
because of the conditions specified in the contracts, such circuits are never-
theless considered to be somewhat weak links in protective schemes and
therefore it is the usual practice to include supervisory equipment when they
are used.
It will be clear from these considerations that supervisory equipment should
be able to detect the following conditions:
(a) an open-circuit in the comparison loop of a protective scheme;
(b) a short-circuit between the pilot wires;
(c) a change in the resistances of the pilot wires.
For many years, monitoring has been achieved by injecting small direct
currents into comparison circuits, as shown in Fig. 9.13. In this example, a
balanced-voltage protective scheme is being supervised. It is common practice
to include a relay (R 1) in the circuit at the opposite end of the protected zone
to that at which the injected current is supplied from an auxiliary supply. The
relay Rl is continuously energized by the injected direct current and in addition
_X__~p~r~o.:::te;::.ct.:::e~d..::.lin:.::e;:....
~A ~B
....;E;.;.n~d..;.A;.....s..9.o.iJ~fl..-f _ _~
Current Current
differential I~l
__ J differential
relay relay
I
Pilot wire I
R3
d.c. supply
Fig. 9.13 Pilot wire supervision scheme.

a small power-frequency current may flow in it because of mismatching in the
protective circuit and capacitance currents associated with the pilot wires.
In the event of an open circuit in the comparison circuit the direct current
in the relay RI would fall to zero, causing it to reset and thereby initiate an
alarm.
If a short-circuit occurred between the pilot wires, the direct current in the
relay RI would again fall to zero and an alarm could be provided. If, however,
significant resistance was present in a fault between the conductors, i.e. the R
shown dotted in Fig. 9.13, the current in relay RI would reduce because of the
resistance present in the pilot wires, but it would not fall to zero and therefore
only faults in which the resistance R was below a certain value would be
detected. Better performance is obtained by including an extra relay R20 shown
in Fig. 9.13, in the supply connection. The direct current in the operating
winding of this relay would always rise if a fault occurred between the con-
ductors and the condition would be detected by relay R 2
Should the resistance of the comparison circuit rise the direct currents in
both relays would fall. This condition, which could arise if a rented telephone
circuit was replaced by another with smaller conductors andlor a greater route
length, could be detected by the resetting of relay R I Should the reverse
situation arise, namely the resistance of the comparison circuit fell significant-
ly, the injected direct current would thus increase causing relay R2 to operate.
It will be appreciated that the inclusion of the two relays would enable the
presence of unacceptable conditions in a comparison circuit to be detected. It
must be recognized, however, that the currents flowing in the relays would be
affected if a fault occurred within a protected zone because alternating cur-
rents would then circulate in the comparison circuit. To ensure that alarm
signals will not be provided in these circumstances, time delays are normally
included so that the supervision relays must indicate that a comparison circuit
is unhealthy for a period considerably longer than the times for which internal
faults may exist before an alarm signal is initiated.
An additional relay (R3 in Fig. 9.13) is often provided to detect the loss of
the injection voltage. Supervisory arrangements of the above types have been
applied successfully to many current-differential protective schemes. In recent
years, schemes, based on the same principles, using different techniques have
been developed and applied [8, 9].
9.3.7 Current-differential schemes incorporating optical-fibre

links
It will be clear from the preceding sections that the limitations of current-
differential schemes do not arise from the basic principle of comparison of the
currents entering and leaving a power circuit but because of factors such as
the resistances and capacitances of the interconnecting conductors over which
comparisons are effected.
The introduction of equipment to enable communications to be effected over
optical fibres caused groups of engineers concerned with protective systems
to investigate the possibilities of developing current-differential schemes in which
comparisons would be effected over optical-fibre links. Such schemes have
now been produced and applied to feeders. Two schemes are described below.
Scheme produced by Asea Brown Boved

This scheme, which is shown in block form in Fig. 9.14, has two main units
at each end of a protected line or cable, namely a differential relay type DL91F
and a fibre-optic system FOX6. In three-phase applications, three conventional
iron-cored current transformers are needed at each end of the protected zone.
The manufacturers recommend that these transformers should be designed for
the maximum symmetrical short-circuit current expected in the protected cir-
cuit, without regard for the time constant of the network. They state that
current transformers of class 5P20, P N = 30 VA are usually adequate. Each of
the relays (DL9l-F) contains a summation transformer so that single phase
comparisons may be made. An output (eAT) provided by a summation trans-
former at one end of a protected zone is fed as an input Tx to an interface
(N3AL) within the fibre-optic system unit (FOX6) and the digital signal
produced is then transmitted over the optical fibre link to an interface (N3AL)
in the unit (FOX6) at the other end of the zone. An analogue output (eBR) is
produced for comparison with the signal eBT, provided by the local DL91-F
relay. A similar comparison between the e.m.f.s eAR, and eAT is effected in the
End A EndS
Current Circuit
N. transformers breaker
~X ______~Pr~o~te~ct~e~d~lin~e~______ X
~~~~---X-----------------
~~~*-----X------------------
Optical
fibre line
N3AL
FOX 6
/
Fig. 9.14 Scheme incorporating relay DL91F. (Reproduced from ABB Buyer's Guide
1989-90, with the permission of ABB Relays AG, Baden, Switzerland.)
opposite direction. As with protective schemes using interconnecting conduc-
tors, the opening of the appropriate circuit-breakers is initiated by the relays
when the e.m.f.s being compared are sufficiently different.
The fibre-optic system units (FOX6) contain monitoring circuits to ensure
that the digital signals are transmitted without error and to provide blocking
signals to the relays in the event of false operation. In addition, both the relays
and the fibre-optic system units are able to detect an interruption of the
optical-fibre link and they will initiate the operation of alarms about 3 seconds
after such an occurrence.
It is recommended that multi-mode graded optical fibres 50/125 ~m stand-
ardized by lEe for the wavelength 900 nm be used with this scheme and these
should allow satisfactory performance to be obtained on circuits up to 8 km
long. By using a variant of the fibre-optic system unit, which is designated
FOX6L operating at 1300 nm, circuits up to 16 km long may be protected
satisfactorily.
The above scheme, which may be applied to circuits operating at frequen-
cies of 50 Hz or 60 Hz and which may be supplied from current transformers
with secondary winding ratings of 1 A or 5 A, operates in 25 ms or less for
faults of twice the setting values.
Further detailed information about this scheme is provided in reference [10].
Other schemes which are now available are described in references [11], [12]
and [13].
LFCB scheme produced by GEC Measurements
Digital processing is used in this scheme in which the three-phase currents at

each end of a protected line are sampled. Internal free running clocks at the
ends control the sampling but they are not directly synchronized and this leads
to phase differences of up to one half of the sampling period.
The data samples represent the instantaneous values of the phase currents
which may contain d.c. offset, harmonic and high-frequency components in
addition to the fundamental. The data are filtered and pre-processed to a form
which enables the magnitudes of differential and bias currents to be deter-
mined. The one-cycle window Fourier method, which is used because of its
I
stable transient characteristics, can be expressed as:
j
Is=~ ~] sinnwLlt in
N-]
in which
N = number of samples per cycle
ro = fundamental angular frequence (rad/s)
at = sampling time (s)
in = instantaneous value of signal at time of sample n
Is = Fourier sine integral of signal i
Ie = Fourier cosine integral of signal i
If the fundamental component of signal i is given by I sin (rot + 9) then
Is=Icos9 and Ie =1 sin9
From this information the instantaneous value of each phase current can be
determined at any instant.
Data messages are sent between the ends of a protected line at regular
intervals. As an example, a message will be sent about the phase a current
(I A ) at end A of a line at a particular instant to end B of the line. The
propagation time is known for the particular communication channel and
therefore the current at end B (lB) at the instant when the message was sent
from end A can be determined. Differential and bias currents (ldiff and Ibias)
can then be determined as:
IIdiffl = IIA+IBI
Ihiasl =~lIIAI + IIBI)

A per-unit biased characteristic is available, for which the tripping criterion is:
IIdiffl > kl Ihias I + lSI
in which kl is the pu bias setting and lSI is the minimum differential-current
setting. .
An alternative bias characteristic is available to improve the stability of the
scheme when high currents are flowing to external faults. In this case the
tripping criterion is as before at low current levels but at the higher currents
it is:
in which IS2 is the threshold at which the increased bias becomes effective and
k2 is the increased pu bias setting. This characteristic is shown in Fig. 9.15.
Typical settings are:
lSI = O.2pu
kl =O.3pu
IS2 = 2.0pu
k2 = 1.0pu
Application to multi-ended circuits 337
I
'S1 pu bias k1
I
'S2
I I
Bias current 'bias
Fig. 9.15 Bias characteristic.
This scheme can readily incorporate extra features, such as monitoring,

event recording, fault recording and inter-tripping. It may be operated over a
dedicated link, fibre optics being embedded within the power system earth
conductors run along the tops of the transmission-line towers or it may be
multiplexed in a communications system link. It may also be operated over
other communication links such as microwave or over shielded twisted pair
cables.
The scheme does not incorporate summation transformers and direct com-
parisons are made of each of the phase currents. As a result a low and constant
minimum fault setting is provided for each type of fault.
Further information about this scheme may be obtained from reference [14]
and details of other schemes are provided in references [15] and [16].
9.4 THE APPLICATION OF CURRENT-DIFFERENTIAL SCHEMES

TO MULTI-ENDED CIRCUITS
There are situations where it is economically advantageous to feed two or
more loads from one source or set of busbars and certainly teed feeders are
quite common in distribution networks. The protection of such circuits pres-
ents special problems [17].
Current-differential schemes are suitable for application to multi-ended cir-
cuits as was made clear in the previous chapter, in which the protection of
busbars was considered. They have also been applied to teed feeders for many
years and schemes such as Solkor and Translay have been widely used. In
such applications, three or more currents must be summed or compared per
phase to determine whether a protected unit is healthy or not. As an example,
when the teed feeder shown in Fig. 9.16 is healthy, the currents fed into
iAa iBa
~ X--.-:;..---<r------ X~
-X X-
-X X-
XXX
I II
EndC
Fig. 9.16 A multi-ended (teed) circuit.
each phase must sum to zero at each instant, i.e. iAa + iRa + ica =0 or
iAa = - (iBa + ica). The magnitude of one of these currents must clearly be
different to those of the others and indeed one current could be twice as great
as each of the others. Because of this, linearity must be maintained in the
components of a protective scheme up to the highest current levels which may
be encountered during faults external to its zone. Saturable summation trans-
formers, such as those used in the schemes described earlier, may not be used
and air-gapped cores must be used to provide linear voltage-current charac-
teristics. Other factors for which provision must be made in protecting multi-
ended circuits are referred to in the next section.
Basic scheme in which metallic pilot wires are used

A basic voltage-balance scheme suitable for application to teed circuits is
shown in Fig. 9.17. To enable balance to be achieved unless a fault is present
on the protected circuit, summation transformers with air-gapped magnetic
cores must be used as stated above. Their secondary winding e.mJ.s (eA, eB
and ec) must be proportional to the input m.m.f.s produced by their primary
windings. The secondary windings of the summation transformers must be
connected in series by two pilot wires (1 and 2) between each of the ends of
the protected zone in such a way that the e.m.f.s (eA' eB and ec) will sum,
ideally, to zero unless an internal fault is present. In these circumstances, zero
current should flow in the relay transformers but in practice small currents
will flow because of mismatching. In the event of a fault occurring within the
protected zone the e.m.f.s do not sum to zero and a current proportional to that
in the fault circulates in the pilot wires and relay transformers. The same current
will flow in the operating windings of all three relays, tripping of the three
circuit-breakers being initiated if the fault current is above the setting level.
Application to multi-ended circuits 339
As stated above, mismatching causes currents to circulate and flow in the
relay transformers when the protected circuit is healthy, and these currents
might reach levels which could cause relay operation when large currents are
flowing to faults external to the protected zone. To ensure that such maloper-
ation will not occur, the relays have windings which provide an adequate
percentage bias. It will be appreciated, however, that satisfactory performance
would not necessarily be obtained if the restraining winding of each relay was
connected only to a tertiary winding on the summation transformer at its own
end of the circuit because little current might flow in that end of the circuit
when high currents were flowing through the other two ends to an external
fault. The technique, employed also in busbar protective schemes and de-
scribed in section 8.7 (page 296), of biasing with a quantity derived from the
currents at all the ends of the protected circuit must therefore be employed.
Current Current
End A translormers Circuit Circuit translormers End B
breakers breakers s Np I
~R~g I
X X""""" RRR oa
s Ns I s Ns I
s Np I s Np I
X X b
s Ns I
s Np I
X X c
I:~
s Ns I
Current s
N
translormers sl
ST
I
R, R,
Fig. 9.17 Teed-feeder protective scheme. Rop, relay operating winding; Rr, relay re-
straining winding; ST, summation transformer (all have the same number of turns).
To achieve this the e.m.f.s provided by tertiary windings on the summation
transformers may be rectified and connected in series via two interconnecting
conductors (2 and 3). As a result a direct current will circulate and flow in the
restraining windings of each of the relays. Clearly, the restraint will then be
proportional to the sum of the magnitudes of the currents flowing at the ends
of the protected zone and it will not be dependent on their directions.
Other devices such as diverter reactors may be included in schemes to assist
further in obtaining the required discrimination between internal and external
faults. Such reactors have two windings on a magnetic core, one of the wind-
ings being connected in the restraining circuit and the other across the opera-
ting winding of one of the relays. When external faults of high-current levels
are present, the cores of the reactors saturate because of the high direct
currents flowing in them and consequently the impedances of the windings
connected across the relay operating windings decrease. As a result, increasing
percentages of the imbalanced currents flowing in the relay operating wind-
ings tend to be shunted through the diverter reactors as the current levels of
external faults increase.
The above techniques were used in the early Solkor and Translay schemes.
LFCB scheme produced by GEC Measurements

This scheme [14], which was described earlier in section 9.3.7, is suitable for
application to teed feeders. In such applications the three-phase currents at
each of the three ends of a line, i.e. nine currents, must be sampled and data
must be sent between the ends via either metallic conductors or optical fibres.
A differential current (Idiff) and a bias current (hias) are produced at each of
the ends (A, B, C), their values being given by:
IIdiffl = IIA + IB +Ic I
and
and as before the tripping cirterion is
IIdiffl >k,lhiasl + Is,

or a characteristic of the form shown in Fig. 9.15 is available.
Further information on the protection of teed feeders is provided in refer-
ences [18] and [19].
9.5 THE APPLICATION OF CURRENT-DIFFERENTIAL SCHEMES

TO LINES AND CABLES TERMINATED WITH TRANSFORMERS
These circuits, which are usually referred to as feeder transformers, and in
which there is no circuit-breaker between the feeder and transformer, are
commonly used in distribution networks. Whilst separate current-differential
Application to lines terminated with transformers 341
protective schemes could be applied to the feeder and the transformer in such
a circuit, it is clearly economically advantageous to employ a single overall
scheme, thus reducing the number of current transformers and the amount of
relaying equipment needed. It might appear that differential schemes suitable
for transformers would be suitable for those applications and that it would
only be necessary to position the current transformers on the primary side of
the transformer at the input end of the feeder. Overall schemes must, however,
be capable of coping not only with the special factors associated with trans-
formers, such as their exciting currents and the need to use current trans-
formers of different ratios, but also with the inclusion of long pilot wires or
communication links in the comparison circuits.
A further variant of the basic voltage-balance protective scheme has been
widely applied to feeder transformers. Figure 9.l8 shows the arrangement
needed to protect a circuit in which the power-transformer windings are delta-
connected on the side connected to the feeder and star-connected on the other
side, the neutral point of the latter winding being earthed. To achieve balance
during healthy conditions, the current transformers on the star-connected side
of the power transformer must be connected in delta and those at the other end
of the feeder must be connected in star, as was explained earlier in section
6.3.2 (page 198). In addition, the secondary current ratings of the delta-
connected current transformers must be 1/-13 times that of corresponding rating
of the star-connected transformers, i.e. either 2.89 A and 5 A or 0.578 A and
1 A respectively. Whilst a summation transformer with a primary winding
having three sections of turns and four terminals could be connected to the
star-connected current transformers, such a transformer could not be energized
from a delta-connected set of current transformers from which only three
connections may be made and therefore a summation transformer with only
two sections and three terminals, as shown in Fig. 9.18, must be used.
As explained earlier in chapter 6, zero-sequence current components cannot
be fed into a delta-connected winding and therefore, if an earth fault occurs
on the star-connected output winding of a transformer, currents of equal
magnitude and opposite phase will flow in two of the conductors feeding the
delta-connected winding. Should such a condition occur with the arrangement
shown in Fig. 9.17, the fault could appear to the protection to be an interphase
fault for which the fault setting could be unacceptably high, say 240% or more.
For this reason, the summation transformers at both ends have only two
sections with three terminals as shown, and earth fault relays are included at
each end of the protected zone. The earth fault relay in the neutral connection
of the current transformers at the input end of the feeder will operate for earth
faults on the feeder or the delta-connected windings of the transformer. The
second earth fault relay, which must detect faults on the star-connected wind-
ing, is fed from an output winding on a core-balance transformer which has
four input windings supplied by the current transformers in the phases and
neutral of the star-connected side of the power transformer. A factor which
el
.0 <: I... ,
FJ r J)-
(.)
!!! _dill-
E
.... Q) -
~ J=~ <:0
.. ~<:~
~<:~ l~ ~ll':.
0-
t:::: o~fI1
-
VI
::l1i)<' ~' ~.
o c:
g VI VI VI VI VI
J
,-,; = <:0
- ~VI
~
~ <:0
~= !'VI
... -m
I-LVl OOO 000- orr;:
EJ
Q)
fJ)~
(.)
c: Q) :;'J til
VI _
J!!E
X X co ...
.00
cbUic: cx:!l- ~
...
o co
o~
-~ -~ <II
;! ~ ~
ll<:~
<:0.1 <:<11
i"
VI~
.~ .~
VI
VlO0 0-
....
.Q
....
.Q
11
'----......----.roIiD'-------tI' a:: a::

<:0-
VI VI VI
Q)
c:
~
Fig. 9.18 Feeder-transformer protective scheme. Rop. differential relay operating

winding; Rr differential relay restraining winding; Re. earth fault relay winding; ST.
summation transformer (both have the same number of turns).
References 343
must be recognized with this arrangement is that any earth fault which occurs
within the protected zone will only operate one of the earth fault relays. It is
therefore necessary that intertripping arrangements be provided so that the
operation of either of the relays will cause the circuit-breakers at both ends of
the protected zone to be opened. This may be achieved independently of the
protective scheme or the comparison circuit may be imbalanced by injecting
a signal into it.
Other forms of current-differential schemes operating on current-balance
and voltage-balance principles are available for application to feeder trans-
formers, examples being Translay 'S' type MBCI [18] and the LFCB scheme,
both of which are produced by GEC Measurements.
9.6 THE FUTURE

Because of the simplicity of the principles involved in current-differential
schemes and the ability of such schemes to discriminate between faults within
and external to their protected zones and also to provide sensitive fault settings
and operate rapidly, it is probable that they will continue to be applied to
short and medium length overhead lines and cables and feeder transformer
circuits for many years.
Undoubtedly relay equipment will increasingly incorporate electronic cir-
cuitry and digital, rather than analogue, techniques. Whilst direct comparisons
will continue to be effected using pilot wires because of their simplicity and
reliability, increasing use of optical-fibre links is to be expected. Comparisons
are already being effected over long distances by high-capacity, digital com-
munication systems [17] owned by private and public companies and it is to
be expected that increasing use of these techniques will occur. These modern
methods of implementing current and voltage comparisons have the advantage
that errors, and therefore imbalances, are not produced when faults are present
external to protected zones because of the capacitances between metallic
interconnecting conductors. In addition, non-linearities caused by the use of
conventional current and summation transformers may be eliminated. As a
result, it is probable that the lengths of circuits protected by current-differen-
tial schemes will increase from the present levels and in addition they are
likely to be used in countries where supply authorities have been reluctant to
apply schemes dependent on metallic conductor links because of the poss-
ibility of faults occurring on them.
REFERENCES
1. The 'Merz-Price' system of automatic protection for high-tension circuits, Elec-
trical Review, August 28, 1908.
2. Leeson, B. H. and Leyburn, H. (1931) The principles of feeder protection and
their application to three modern systems, CIGRE Conference, Paris 18-27, June
paper 106,pp. 6-33.
3. Adamson, C. and Talkan, E. A. (1960) Selection of relaying quantities for dif-
ferential feeder protection, Proc. lEE, 107A, 37-47.
4. Solkor Protective Systems. Pamphlet 806, A Reyrolle and Co Ltd, 1954.
5. Protective Relays - Application Guide (1987) (3rd edn) GEC Measurements,
Chapter 10.
6. High speed pilot-wire feeder protection Solkor-R and Solkor-Rf, Pamphlet pub-
lished by NEI Reyrolle Ltd, 1990. Also' Digital feeder protection - Solkor M',
Data Sheet Solkor - M, 6/91.
7. Translay 'S' differential feeder and transformer feeder protection - Type MCBI,
GEC Measurements, publication R6011.
8. Pilot supervision equipment for Solkor R and Solkor Rf pilot-wire feeder protec-
tion, Data Sheet RIRf SUP (1985), NEI Reyrolle Ltd.
9. Supervision of AC pilot circuits - Relay type MRTP, R-6026, GEC Measure-
ments.
10. Differential line protection with fibre optics-type LD91-F (1987), Asea Brown
Boveri, Publication CH-ES 63-54.10.
11. Sun, S. c., Ray, R. E. (1978) A current differential relay system using fibre optic
communications, IEEE Transactions on Power Apparatus and Systems, PAS-I02,
410-419, February.
12. Takagi, T., Yamakoshi, Y., Kudo, H., Miki, Y., Tanaka, M., and Mikoshiba K.
(1980) Development of an intrastation optical-fibre data transmission system for
electric power systems, Trans. IEEE, PAS-99, 318-327, Jan.-/Feb.
13. Sugiyama, T., Kano, T., Hatata, M. and Azuma, S. (1984) Development of a PCM
current differential relaying system using fibre-optic data transmission, IEEE
Transactions on Power Apparatus and Systems, PAS-I03, 152-159, January.
14. LFCB digital current differential relay, Publications R-4054B and R-4028A, GEC
Measurements, England.
15. Kitagawa, M., Andow, F., Yamaura, M. and Okita, Y. (1978) Newly developed
FM current-differential carrier relaying system and its field experience, Trans.
lEE, PAS-97, 2272-2281.
16. Akimoto, Y., Matsuda, T., Matsuzawa, K., Yamaura, M., Kondow, R. and Mat-
sushima, T. (1981) Microprocessor based digital relay application in TEPCO,
Trans. IEEE, PAS-IOO, 2390-2398.
17. AlEE Committee Report (1961) Protection of multi-terminal and tapped lines,
Trans. AlEE, PAS-SO, pp 55-66.
18. Aggarwal, R. K. and Johns, A. J. (1986) The development of a new high speed
3-terminalline protection scheme, Trans. IEEE, PWRD-l, 125-134.
19. Aggarwal, R. K., Hussein, A. H. and Redfern, M. A. Design and testing of a new
microprocessor-based current differential relay for EHV Teed feeders, 91 WM
/65-1 PWRD.
10
Interlock and phase-comparison
schemes for the protection of
overhead transmission lines
INTRODUCTION
Protective schemes based on Kirchhoff's first law, namely those which deter-
mine the location of faults by comparing the instantaneous currents entering
and leaving protected zones, were considered in Chapters 5-9. Their applica-
tion to overhead lines and cables was described in Chapter 9, in which it was
shown that there are factors which limit the lengths of circuits which may be
protected when the comparisons are effected over interconnecting conductors
(pilot wires). These limitations arise because of the need to convey continuous
infonnation with fairly high accuracy between the ends of protected circuits.
To eliminate this need, alternative schemes were introduced in which only
relatively simple signals need to be sent between the ends of protected zones,
and because significant attenuation of these signals may be allowed it is
possible to apply such schemes to very long circuits.
Interlock schemes which employ directional relays sited at each end of a
protected zone initiate the opening of the circuit-breakers at each end of a line
if both sets of relays indicate that currents are flowing into both its ends.
Clearly, information must be conveyed between the ends of protected circuits
but it is only of a yes-no nature, e.g. are the directional relays detecting current
flow into the circuit or not? The magnitudes of the received signals in such
schemes do not matter provided they are above particular threshold levels.
Phase-comparison schemes are somewhat related to Merz-Price schemes in
that they compare the currents at the two ends of a protected circuit, but they
do so by taking into account only the phases of the currents. Basically a signal
is sent whilst the current at an end has a particular polarity, say positive, and
this is compared with a similar signal at the other end. When the circuit is
healthy the two signals will coincide because the currents at the two ends will
be of similar phases whereas the phases will be considerably different if a fault
is present within the protected circuit and in this event the signals will be
displaced from each other. Clearly in these circumstances the opening of the
circuit-breakers at the ends of the line must be initiated.

346 Interlock and phase-comparison schemes
Whilst the principles involved in both the above forms of protection are
simple, features have to be incorporated in practical schemes to both ensure
that they will not operate incorrectly under certain system conditions and also
minimize the amount of equipment required. As an example, it is desirable
that only a single signal, rather than three, be transmitted between the ends of
a three-phase line and therefore the most suitable signal derived from the
three-phase currents should be used.
Practical schemes are considered in detail later in this chapter, after the
following sections in which historical information is provided and the behavi-
our of long lines is examined.

During the first decades of this century, the transmission of electricity was
effected at relatively low voltages over fairly short distances and the amount
of interconnection of networks was small. At that time most of the overhead
lines and cables were protected by current-differential schemes based on the
Merz-Price principles considered in the previous chapter. Various detailed
improvements enabled these schemes to provide the fault settings and degrees
of discrimination needed for the applications which arose.
The situation changed quite significantly in Britain, however, around 1930,
with the advent of the 132 kV interconnection and transmission (grid) net-
work. Much longer lines than those which had been used before were con-
structed and the fault-current levels were significantly greater than those
previously encountered. Current-differential schemes, because of their limita-
tions, which were examined in the previous chapter, were not suitable for some
of the new circuits. Other schemes were therefore developed at that time,
including a group which employed simple interlocking signals to obtain the
necessary discrimination between internal faults on protected lines and other
faults or healthy conditions. These schemes were applied not only to long lines
but to shorter lines where simple signals could be transmitted over telephone
circuits or specially provided pilot wires. In addition, schemes were produced
in which high-frequency signals were transmitted via the conductors of the
lines being protected and, of course, such schemes contained electronic equip-
ment.
Interlock schemes discriminate by determining the directions of current
flows in lines, i.e. by determining the phase relationships of the line currents
and voltages. They therefore require inputs derived from both current and
voltage transformers. These latter items are costly and therefore, when devel-
opments in electronic equipment made it possible, schemes were developed
in which the phases of the currents at the two ends of a protected line were
compared. Two such phase-comparison schemes, namely Telephase and Con-
traphase, were introduced around 1950 by A. Reyrolle & Co Ltd and Metro-
politan-Vickers Electrical Co Ltd respectively. In both these schemes, which
Construction and behaviour of transmission lines 347
are considered in detail later in this chapter, the necessary comparisons were
effected by transmitting high-frequency signals over the lines being protected.
10.2 THE CONSTRUCTION AND BEHAVIOUR OF

TRANSMISSION LINES
Because of the great cost of high voltage underground cables, the relatively
long-distance transmission of electricity is effected over bare conductors sup-
ported by string insulations which are suspended from towers. In many cases,
two three-phase circuits are carried on the same set of towers, one circuit being
supported on each side of a tower, as shown in Fig. 1O.1(a). In other cases
towers may only support a single three-phase circuit, in which case the con-
ductors may be mounted in several configurations, the two most commonly
used arrangements being shown in Fig. 1O.1(b) and (c).
In all cases, conductors are run along the tops of the towers to which they
are directly connected. In many installations, only a single earthed conductor
is provided except near the ends of lines where it is common to run two, spaced
apart in a horizontal configuration. In some cases, however, two conductors
are used throughout the length of a line. These conductors are primarily to
screen the phase conductors in the event of lightning storms in the vicinity of
a line and it is well known that earthed conductors shield a zone between the
vertical beneath them and a certain angle each side of that vertical. Two
horizontally spaced earth conductors shield a larger zone than a single con-
ductor and, as stated above, they are usually installed at the ends of lines to
reduce the likelihood of direct lightning strikes close to vulnerable terminal
equipment.
It must be recognized that such shielding may not be fully effective and that
a strike may nevertheless occur to a phase conductor, thus raising its potential
and then causing flashovers across supporting insulators as the surge propa-
gates in each direction. Such events do not occur often and breakdowns result
Earth conductor

a b
i
c
(a) (b) (c)
Fig. 10.1 Overhead line arrangements.

more frequently from lightning strikes to the overhead earth conductors. When
such a strike occurs, the potential of the earth conductor is raised and surges
propagate along it in both directions. When each of these surges reaches a
tower a large current flows down the tower to earth and the voltage at the top
of the tower is raised above earth because of the impedance of the tower to
the surge. This can cause the potentials between the phase conductors and the
top ends of their supporting insulators to be so high that flashovers occur
across one or more of the insulators. Thereafter an arc or arcs may be main-
tained by the normal power-frequency voltages.
It will be clear from the above that both phase to earth and interphase faults
can occur as a result of lightning. In addition, faults can occur due to mech-
anical failures in spite of the high standards to which overhead lines are
designed and produced. Conductors do occasionally break due to overstress-
ing, a condition which may arise when thick coatings of ice are present on
them. Severe arcing may occur at a break in a conductor and the associated
ionization could cause flashover to the other phase conductors. In addition,
one or both parts of the broken conductor may fall onto the ground or on to
other conductors, causing earth and/or interphase faults.
It will be clear from the above that faults must be expected on long trans-
mission lines and indeed the numbers of faults are likely to increase with line
length because of the increased amount of exposed conductor.
As was stated in the previous chapter, there is capacitance between the
phases and to earth of three-phase overhead lines and there is also capacitance
between the conductors of parallel circuits which are carried on the same
towers. As a consequence, quite significant capacity currents flow between
the phases and to earth on high voltage transmission lines and these increase
with line length. Quite clearly the instantaneous currents entering individual
phase conductors are not equal to those which leave at the other ends, and in
steady state both the magnitudes and phases of the currents at the two ends of
a line are different. Because of the series impedances associated with the
conductors, the instantaneous voltages at the two ends of each phase conductor
must also be different and, again, in steady state the pairs of voltages must be
different in phase and/or magnitude.
The well-known steady state relationships between the quantities at the ends
A and B of a long, healthy single-phase line are:
VA =VB cosh yl + IHZo sinh yl (10.1)
- - V
IA =IB cosh yl + _B sinh yl (10.2)
Zo
in which
VA and VB are the r.m.s. voltages at ends A and B respectively
IA is the current flowing into end A
IB is the current leaving end B
Features of interlock protective schemes 349
1 is the line length
'Y the line propagation constant
andZo the line surge (or characteristic) impedance.
If an interlock protective scheme incorporating directional relays were to be
applied to the above single-phase line the magnitudes of the voltages and
currents would not be important. The angular zones over which the relays were
set to operate would, however, have to be such that both relays would never
indicate that currents were entering both ends of the line unless there was a
fault on it. To ensure that this condition would be satisfied, the displac~ments
Eetwee~ the voltages and currents at each end of the line, i.e. between VA and
IA and VB and IB' would have to be examined for healthy and external fault
conditions.
Should a phase-comparison scheme be applied to a single-phase lin~ then
!.he only quantities of significance would be the phases of the currents hand
IB and clearly their phase displacements, when the line was healthy, should
never be in the range in which tripping of the circuit-breakers would be
initiated. In addition, the phase displacements possible during all internal
faults should ideally cause operation.
In practice, long overhead lines are invariably three-phase and their protec-
tive schemes must be set so that the principles outlined above will obtain under
all conditions. The behaviour of overhead lines is considered in more detail
in sections 10.3.1 and 10.5.1, in which details of particular protective schemes
are provided.
10.3 FEATURES OF INTERLOCK PROTECTIVE SCHEMES

As indicated earlier, relays or detectors are installed at each end of a protected
line to determine the directions of current flow in the conductors and, when
necessary, signals are sent between the line ends to ensure that the appropriate
actions are taken.
In the following sections the relaying and signalling arrangements are con-
sidered.
10.3.1 Relaying arrangements

The construction and behaviour of electro-mechanical directional relays was
considered in some detail in section 4.2.4 (page 113). These relays must be
energized by both voltages and currents and basically they operate when the
phase displacements between the voltages and currents supplied to them are
within certain ranges. As an example, a relay may operate when the voltage
and current supplied to it are displaced from each other by 1t/2 rad or less and
it will then restrain for other phase displacements, i.e. those greater than
1t/2 rad.
Induction-type electro-mechanical relays tend to operate over an angular
range of about n rad, the actual values for particular relays being somewhat
affected by the magnitudes of the voltages and currents supplied to them.
Relays which use more complex circuitry and electronic processing can pro-
vide operating ranges of any desired value and these may not be affected by
the magnitudes of the voltages and currents.
All types of relays may have their operating ranges positioned asymmetri-
cally; for example, a relay could be set to operate for currents with displace-
ments between n/3 rad leading and 2n/3 rad lagging the voltages applied to
it.
The directions of currents associated with three-phase overhead lines may
be detected in several ways and these are considered below.
Basic detection arrangements

To detect each of the types of fault which could occur on or external to a
three-phase line, six directional relays could be provided at each of its ends,
i.e. three to detect phase to earth faults and three to detect interphase faults.
It was shown earlier in section 4.5 (page 135) that earth fault directional
relays supplied with voltages and currents from the same phase, e.g. Va and
la, are unable to discriminate correctly when faults occur near the end of a line
at which they are sited, because the voltages supplied to them may then be too
low. Clearly phase-fault relays are also unable to discriminate correctly for
faults near to them if they are supplied with voltages from the phases they are
monitoring, i.e. if a voltage proportional to that between phases a and b of a
line was supplied to the relay required to detect faults between these phases
it might operate incorrectly for close-up faults.
In practice, therefore, the voltage supplied to each relay is obtained from a
phase or phases not associated with the faults which it is to detect. As an
example, the 'a' phase earth fault relay could be energized by a voltage
proportional to that between the phases band c of the line being protected.
Whilst the cost and complexity associated with the use of six directional
relays at each end of a long overhead line will not usually be prohibitive, a
reduction in the number of relays required in a scheme is clearly desirable
provided that performance is not impaired. Such a reduction could be achieved
by using a single relay to detect the directions of all earth faults. The relay
would be energized by the sum of the outputs of the current transformers in
the three phases and the sum of the phase voltages, this latter quantity being
obtained from either open-delta-connected tertiary windings on the main volt-
age transformers or from an extra auxiliary transformer. Such an arrangement
is shown in Fig. 10.2(a) and the relay quantities obtained thereby during
internal and external fault conditions are shown in Fig 1O.2(b). It will be clear
that a directional relay could be set to discriminate between the two condi-
tions.
Current
transformers
o-r
a F2 N.
--+ F1
--l.i;~ f
b-=-
o--X
X
r"'O~.OT
s Ns N.
Si~~ f
Ib = 0
~
111
-
r-o~.o'j
s Ns N. le= 0
8---X s ~~~ f ~
r; 0;/s 'j s t ts sJt

Nv p : ' Va ~. Vb ~. Ve
f f f Voltage
s s s transformer
N~m
r----I Nvs
f f f
fW
~ s s s
directional relay: ,..r,.~_
1 Auxiliary
transformer
(a)
Ve
~
'
V, = k( Va+ Vb+ Ve) _ Va
- 4
Vb Vb
Earth fau It at F1 Earth fault at F2
(b) (c)
Fig. 10.2 The behaviour when faults are present.
As an alternative a single relay could be used to detennine the directions of

both a phase to earth fault and an interphase fault. As an example, a relay
could be supplied with inputs proportional to the 'c' phase voltage and the 'a'
phase current of a line. By setting the relay to operate in a suitable angular
zone it could be made, as shown later, to detect the directions of faults to earth
on phase a and faults between phases a and b of a protected line. It will be
clear that this practice would enable schemes to be produced with only three
relays at each end.
Several other methods could be used to reduce the number of relays required
in a scheme. A single relay could be used at each end of a protected circuit if
current transfonners of different ratios were employed or, alternatively, sum-
mation transfonners connected to current transfonners, each of the same ratio,
4,
Vb
Va
la
VV
O.aVe
O.9Vb
a.8~'~ 0.9,
(a) (b)
Fig. 10.3 Voltage and current summation for internal fault.
could be used. In each case a current would be provided to energize the relay
and a corresponding voltage would have to be provided.
The relay current (ir) would be equal to the sum of different fractions of the
currents in the phases of a protected line, i.e.
(10.3)
The outputs of the voltage transformers would also be summated to provide

a voltage (Vr) to energize the relay. The same proportions ofthe phase voltages
would be summed, i.e.
(10.4)
In the event of a three-phase fault on a protected line the voltages at the

ends of the line would be depressed and the currents could be well above
the rated value for the line. The phases of the currents would lag those of the
respective voltages by angles of cp, as shown in Fig. lO.3(a) and the outputs
from the summation processes, based on equations (10.3) and (10.4), using
values of kl = I, k2 =0.9 and k3 =0.8 would be as shown in Fig. 10.3(b). It will
be seen that the output current (lr) would lag the relay voltage (Vr) by the
angle cp, i.e. by the displacement between the phase voltages and their associ-
ated currents.
In the event of a three-phase fault external to a line, at end A as shown in
Fig. lO.4(a), the phase voltages could be depressed and the currents la, Ib and
Ie would be of the opposite polarities to those above, as can be seen from Fig.
lO.4(b). The summated current output (Ir) would then lag the output voltage
(Vr ) by (1t + cp) rad. As a result a single directional relay could be set to restrain
for this condition but operate for the internal fault condition considered above.
Similar behaviour would be obtained for other types of faults. As an
example, the conditions which would obtain during internal and external faults
to earth on the 'a' phase are shown in Fig. 10.5. In both cases the voltage of
X-----a
X b
X-----c
-::~ End A
(a)
Ij,
~v O~8'
la
Va O.9Vb
O.BVe
la
'r
(b)
Fig. 10.4 Voltage and current summation for external fault.
the 'a' phase at end A could be depressed and the currents for the two faults
would be approximately in antiphase with each other. A directional relay could
clearly be set to differentiate between the two conditions.
Summations of the above type produce outputs from the positive-, negative-
and zero-sequence components of the voltages and currents present during
fault conditions. The outputs caused by zero-sequence components of a given
magnitude are much greater than those produced by the other sequences, as
shown in Fig. 10.6. The actual outputs are dependent on the ratios kh k2 and
k3 and thus the turns in the primary winding sections when summation trans-
External fault
Fig. 10.5 Conditions when single-phase to earth faults are present.

)-v.. Positive sequence
O .2C: Q 8
O.9V2b
Vr2
V2a
Negative sequence
- - -... Voa
- - -... VOb
- - -... Vac VOa O.9VOb O.8Vac
Zero sequence
Fig. 10.6 Sequence outputs of a summation transformer.
formers are used. Whatever ratios are used, however, the sensitivities to
positive- and negative-sequence inputs are equal to each other.
Clearly, the same performance could be obtained from sequence networks
connected to current and voltage transformers of the same ratio in each phase.
In addition, however, the outputs obtained from quantities of a given magni-
tude of each of the sequences could be different from each other and, of
course, zero output could be obtained from one or more of the sequences if
desired. Outputs could be obtained for all types of faults from positive-
sequence networks but the outputs provided by such networks during phase to
earth faults are only one third of those obtained for three-phase faults of the
same current or voltage magnitudes. The resulting low sensitivity to earth
faults is normally unacceptable and therefore it is usual to employ networks
which provide relatively low outputs when their inputs are of positive-
sequence but higher outputs when the inputs are of zero sequence, e.g.
Ir = kIll + kola, ko being larger than k l.
Whatever methods are used and however many directional relays are used,
the angular operating and restraining zones of the relays must be such that
they will discriminate between faults within protected zones and those external
to them.
Relay settings
On very high voltage transmission networks, the neutral points are usually
earthed solidly and in these cases earth fault currents lag their phase voltages
by angles (q approaching n/2 rad. A typical condition during a short circuit
to earth on phase a of a line is shown in Fig. 10.7. The voltage of the faulted
phase (Va) could be low but the voltage between phases b and c (Vbc) would
be unaffected.
In a protective scheme with three earth fault and three phase-fault induc-
tion-type relays at each end, the earth fault relays could be set to provide
maximum torque for short-circuits to earth, e.g. maximum torque would be
provided when la lagged Va by q> rad. As these relays would have angular
operating zones approaching n rad, boundary conditions would exist near the
line PQ in Fig. 10.7. Should the 'a' phase relay be supplied with a voltage
proportional to that between the healthy phases (Vbc), it would have to be set
to operate whenever the fault current (/a) lead the voltage Vbc by an angle less
then (n - q rad or lagged by an angle less than q> rad.
The phases of the currents relative to the input voltages of a line on which
an interphase short-circuit is present are dependent on the line impedance
between its input and the fault. As a result, fault currents lag the corresponding
voltages by angles (q approaching nl2 rad, i.e. for a fault between phases a
and b, the current (/a) will lag the voltage Vab by almost nl2 rad and the current
h will be in antiphase with the current la as shown in Fig. 1O.8(a) and (b).
Again the voltages between faulted phases could be very depressed and there-
fore the relays are usually energized with the voltage of the unaffected phase.
As an example, the relay which is to detect faults between phases a and b
could be supplied with a current proportional to that in the 'a' phase of the
line (/a) and a voltage proportional to that of the 'c' phase of the line (Vc). In
these circumstances the operating angular zone of the relay would be posi-
tioned as shown in Fig. 1O.8(c).
It will be appreciated that all the voltages may be depressed in the event of
a symmetrical three-phase short-circuit near the end of a line and, in these
circumstances, the above cross-polarizing measures will clearly not be effec-
Vbc
Fig. 10.7 Conditions when a single-phase to earth fault is present.
Circuit
breaker la= If
ao----X __~~~~L~in~e~~,_--------
I~f If +[Short-circuit
bo----X--~------~-------
co----X-----------------------
(a)
(b) (c)
Fig. 10.8 Conditions when an interphase fault is present.
tive and relays might not operate. This situation might be thought acceptable
in some applications because of the unlikelihood of such faults. If this is not
so, however, some form of memory action, such as tuned circuits, could be
included to ensure that the voltages supplied to the relays will not collapse
suddenly when faults occur, but decay slowly, thus enabling the relays to
operate if necessary.
It was stated in the previous section that the number of directional relays
needed at each end of a scheme could be reduced from six to three by so setting
each of the relays that it will determine the directions of both a phase to earth
fault and an interphase fault. As an example, a relay could be supplied with
inputs proportional to the 'c' phase voltage and the 'a' phase current of a line.
In the event of a short-circuit to earth on the 'a' phase of the line, the fault
/.)F--Jc:..::...::~ Va
/'
/'
/'
/'
/'
/' /' lab
/'~/'
Vb la
Operating
zone
Fig. 10.9 Conditions when a single-phase to earth and an interphase fault are present.
current (fa) would lag the 'a' phase voltage by an angle approaching nl2 rad,
as shown in Fig. 10.9. Should a fault occur between phases a and b, the fault
current (lab) would lag the line voltage (Vab) by a similar angle. The angle
between the 'a' phase currents for the two faults would be of the order of
nl6 rad and the operating zone of the relay could be set to operate for both of
them, as shown in Fig. 10.9.
When a single relay fed with summated quantities is used at each end of a
scheme, suitable angular operating zones can be readily provided.
Relay sensitivity
In the previous section, the phase displacements between the currents and
voltages associated with transmission lines during short-circuit conditions
were considered and it was clear that the displacements when faults occur
within a protected zone are very different to those encountered when faults
are present on or beyond the bus bar near the relaying position. As a result, the
angular operating zones of directional relays can be set so that correct dis-
crimination will always be achieved for such conditions.
It must be recognized, however, that directional relays may not operate in
ways which will enable internal faults of limited current magnitudes to be
detected. To illustrate this in a simple manner, the behaviour obtained in the
event of a resistive fault to earth on a single-phase line is examined below.
To further simplify the treatment it is assumed that the fault is at the
mid-point of the line, as shown in Fig. 1O.1O(a) and that the line conductor
has no resistance. For the short-circuit condition (Rf = 0), the currents hand
IB would flow. The current IA would lag the voltage VA by nl2 rad, whereas
the current IB would lead the voltage VB by n!2 rad, as shown in Fig. 10.1 O(b)
and the currents would be given by:
The directional relays would be set so that the presence of an internal fault
would be detected.
In the extreme case of a fault with a resistance approaching infinity
(Rf ~ 00), the currents IA and IB would be almost the same, i.e.
V -V
I~I~-' A B
A B } 2x
This condition is shown in Fig. 10.1 O(c). It is very different from that shown
in Fig. lO.lO(b) and the relays would not indicate an internal fault.
Conditions as the fault resistances increase from zero are shown in Fig.
1O.l0(d), from which it can be seen that the phase displacement at end B would
increase from nl2 rad as the fault resistance rose from zero to a peak value of
about 2.8 rad. Thereafter the phase displacement would fall with increase of
fx -
Busbar
fA
Line
jx jx
.~
-X
Busbar
A B
(a)
(c)
~B
2
o~----~~============================~
,;---------------------------- R
,
I
'~A
-1
-2 (d)
Fig. 10.10 Conditions when resistive faults are present-on a single-phase line.
(a), (b), (c), (d)
- - - - - - - - - - - - - - - - - 5pu
or--------------------------------------------.
,
~---------------------------- R
I
,'~A
-1 '
-2
Fig. 10.10 (e)
fault resistance, the change for resistances between 1 pu and infinity being
small.
At end A, however, the displacement decreases continuously from a value
of about 1t/2 rad. when the fault resistance is zero.
Similar results would occur for faults at a point quarter the way along a line,
as is shown in Fig. 1O.10(e) and it is therefore clear that directional relays
cannot be set to detect internal faults with resistances above certain levels, i.e.
internal faults of current levels below particular values would not be detected.
It will be appreciated from the preceding sections that several factors have
to be considered when the operating zones of directional relays are being
selected for particular applications. Because of their experience, manufac-
turers should be able to supply interlock schemes which incorporate relays
with settings which will enable given lines to be adequately protected. If
desired, however, the suitability of the proposed operating zones of relays
could be determined by using either the three-phase long-line equations or
approximate methods to calculate the phase relationships between the various
voltages and currents which could be present during healthy conditions and
also those possible during both internal and external fault conditions.
10.3.2 Interlocking signals
If an interlock scheme associated with a transmission line has only one direc-
tional relay at each of its ends, then each relay should detect that current is
flowing into its end when a fault is present within the protected zone. In these
circumstances, opening of the circuit-breakers should be initiated.
When the line and the network of which it forms a part is healthy or when
faults are present outside the zone covered by the line protective scheme, one
relay should detect current flowing into the line, whereas the other relay
should detect current leaving the line. For these conditions the line circuit-
breakers should not be opened.
The above performance could be obtained by using one of the following
signalling arrangements.
Permissive signalling
In this method, also referred to as the transfer trip method, a signal is sent
when a directional relay detects current flowing into its end of a line. On
receipt of this signal at the other end of the circuit, tripping of the circuit-
breaker at that end is initiated if its directional relay also detects current
flowing into the line.
With this method, signals have to be sent from each end of a protected line
when a fault occurs on it so that tripping of the circuit-breakers at each of the
ends is initiated.
A disadvantage of this method is that the non-arrival of the signals would
prevent an interlock scheme from initiating the opening of the circuit-breakers
when internal faults were present. In these circumstances, reliance would then
have to be placed on the operation of other protective equipment or an alter-
native method of permitting tripping, such as that described later in section
10.4.2, must be used.
Because current normally flows into one end of a healthy line a signal would
be transmitted continuously unless the directional relays were prevented from
operating at currents up to levels somewhat above the rated values. It is clearly
desirable, however, that signals should only be sent when faults are present
and therefore current-sensitive starting relays must normally be provided. The
settings of these relays are considered later in section 10.3.3 ('Permissive
signalling' ).
Blocking signalling
In this method, signals are sent to prevent or block the tripping of circuit-
breakers when necessary. These signals are transmitted when a directional
relay detects that current is leaving its end of a protected circuit. Clearly this
method has the advantage that the non-arrival of a signal cannot prevent circuit-
breakers being opened when internal faults are present, but it could cause a
circuit-breaker at the end of a healthy line into which current is flowing, to be
allowed to open incorrectly.
Under healthy conditions, current must be flowing out of one end of a
protected line and a blocking signal could be sent out continuously, and indeed
such a signal would be required to prevent the directional relay at the other
end of the line from initiating the opening of its associated circuit-breaker.
This is clearly undesirable because any interruption of a blocking signal could
cause the opening of a circuit-breaker. For this and other reasons which are
considered later, current-operated starting relays must be included in interlock
schemes which use this type of signalling.
These considerations have been based on schemes with a single directional

relay at each end of a protected zone. Most schemes do, however, have several
relays at each end. In these cases permissive signals must be sent when any
directional relay detects current flowing into its end, whereas, when blocking
signals are employed, a signal must only be sent when all the directional relays
at an end detect currents leaving the protected zone.
10.3.3 Starting relays

It was made clear in the previous section that the signals sent between the ends
of lines to either permit or block the tripping of circuit-breakers should not be
sent continuously but only when faults are present. Extra relays, which are
usually referred to as starting relays, must therefore be provided to detect
abnormal conditions, namely the presence of both phase to earth and inter-
phase faults on or external to a protected line. Interphase faults can be detected
by including simple overcurrent relay with settings above the full load level
of the line. These need only be provided in two of the phases, say 'a' and 'c',
at each end of a line. Because the current flowing in the event of an earth fault
may be restricted, it is necessary that earth fault relays should have sensitive
settings, say 0.2 pu, and a single relay carrying the sum of the outputs of the
current transformers in the three phases of the line should be included at each
end of a scheme.
Should a scheme incorporate summation transformers then they could feed
starting relays which would operate above certain output levels.
Because the arrangements required with permissive and blocking signalling
are different, they are considered separately below.
Permissive signalling
The operation of one or more of the starting relays at an end of an interlock
scheme should permit an interlocking signal to be sent from that end if a fault
is detected in the appropriate direction.
With this arrangement, the opening of the circuit-breakers at both ends of a
line will be initiated when currents above certain levels are fed into both its
ends. Should the current infeed to one end of a line be below the level needed
to operate one or more of the starting relays, then neither of the circuit-break-
ers would be opened even though the current at one end of the line could be
very high. Because of this limitation, this type of signalling should only be
applied to lines in networks which are so interconnected that large currents
will flow into both ends of a faulted line or some alternative provision should
be made for such situations.
Blocking signalling
When this method of signalling is used it is imperative that the directional
relays at the end of a line to which a signal is being sent should not be allowed
to initiate the tripping of their associated circuit-breaker at a current level
Circuit Circuit
Current breaker breaker
transformer
~E_n_d_A____~____~~J-____
A
X__
~p~r~ot~e~ct~ed~lin~e___ X______
B End B
-1
Stage 1
Starting relay (SR1)
Voltage Stage 2
transformer Starting relay (SR2)
breaker A
~-
Trip circuit
oil
BS1 DR2
-
SR2
0
JC DR1
-0---
SR1
- 0---
Send signal
to end B
Fig. 10.11 Blocking signalling. Contact BSI opens when a blocking signal is received
from end B. Contact DRI closes when current flows from the line to the busbar at end
A. Contact DR2 closes when current flows into the line from the busbar at end A.
Contact SRI closes when fault currents are above the Stage I level. Contact SR2 closes
when fault currents are above the Stage 2 level.
below that at which the signal may be sent. To meet this requirement and
ensure that a margin of safety will exist, relays, termed Stage 1, should be
provided to allow signals to be sent and a second set of starting relays, with
higher current settings than the Stage 1 relays, must be provided at each end
of a line. One or more of these latter relays, which are termed Stage 2 starting
relays, must operate at an end of a line before the circuit-breaker at that end
may be tripped.
A typical arrangement of the circuitry needed at each end of a line to ensure
correct operation is shown in Fig. 10.11. It will be seen that the receipt of a
blocking signal must cause the trip circuit to be opened, i.e. contact BS 1 must
open. It is essential that the trip circuit should not be completed before the
blocking signal has had time to arrive and to achieve this an adequate time
delay must occur before the Stage 2 relay allows tripping to be initiated. This
requirement prevents such schemes from operating very rapidly in the event
of internal faults including those of high current magnitudes.
This arrangement does, however, have the advantage that should a fault on
a line be fed from only one end, then it will be cleared by opening the
circuit-breaker at that end.
10.3.4 Signalling channels

Because of the simple nature of the information which must be transferred
between the ends of interlock schemes, a wide range of signalling channels
may be employed. These are briefly considered below.
Interconnecting conductors (pilot wires)

Signals may clearly be sent over pilot wires specially provided for the purpose.
This practice may be thought desirable when major lines are to be protected
and in such cases a relay or electronic circuitry may be used to detect the
arrival of signals and then tripping of the circuit-breaker at the receiving end
will be either allowed or prevented depending on the type of signalling being
employed.
When less important lines are to be protected, signalling may be effected
over conductors used for telephony. In such cases, operation of starting relays
must cause the conductors to be available for the transmission of interlocking
signals if necessary. This requires special arrangements with the telephone
company so that transmission is always available within an acceptable time
delay. An analysis of pilot-relaying performance requirements can be found
in reference [1].
The conductors of protected lines

To avoid the expense of providing special interconnecting conductors for
major long lines, signals may be transmitted over the conductors of the lines
Tuned circuits to block
hf signals (line traps)
_ x : _....-...w.,-_.....__...;.p.;,;ro;.;,te;,;c;.;,;te;;,;d;.;l;;.;,;ine a
---X----0Wv---~----------b
U
-x:--~0Wv-~~--~------c
Coupling
capacitors
H
Transmit/receive signals
Fig. 10.12 Coupling equipment for PLC signalling.
being protected by interlock schemes. This method of transmission is referred

to as Power Line Carrier, often abbreviated to PLC.
Relatively high-frequency (typically 30-300 kHz) signals are transmitted.
These must be produced by electronic circuits including oscillators and am-
plifiers and they must be fed to and from two of the phase conductors of a line
via coupling equipments turned to the signal frequencies. A typical arrange-
ment is illustrated in Fig. 10.12.
The power supplies to the transmitting and receiving equipment used in
interlock schemes must be secure and must not be affected by faults on or near
the lines being protected.
When a scheme depends on blocking signals to obtain stability, correct
operation will be obtained when faults occur on circuits connected to the line
it protects, i.e. external faults, because the signals will be transmitted satisfac-
torily over the line. In the event of faults occurring on the protected line,
blocking signals are not required and therefore the presence of the faults in
the signalling channel is of no consequence.
When a scheme employs permissive signalling, however, signals must be
received satisfactorily when faults are present on the line being protected. This
requirement may not be met under all conditions and therefore pennissive
signalling over protected lines is not usually acceptable.
A discussion of the transmission behaviour of power lines at PLC frequen-
cies can be found in reference [21.
Optical-fibre links
Optical-fibre links are now employed for the transmission of interlocking
signals, and because such links are separate from the conductors of lines being
Interlock protective schemes 365
protected, they are not directly affected by system faults. Both permissive and
blocking signals can therefore be sent over these links. Again, reliable power
supplies must be provided for the equipment associated with this method of
signalling. Information about an interlock scheme employing optical links is
provided later in section 10.4.2. A review of protective relaying using fibre-
optic communications may be found in reference [3].
Microwave links
Line-of-sight communication links may be established using microwave sig-
nals at frequencies ranging from 1 to 10 GHz. Such links can rarely be justified
for protective purposes alone and thus when they are used they perform a
number of protection, monitoring and control functions.
A summary of the various protection signalling methods is given in refer-
ence [4].
10.4 INTERLOCK PROTECTIVE SCHEMES

A variety of interlock protective schemes have been applied to overhead lines
over the years. One of the first, which was produced by A. Reyrolle and Co
Ltd, was referred to earlier in section 10.1. It was designated as 'interlock
protection' and details of it are provided below and then information about a
recently developed scheme is provided.
10.4.1 Reyrolle interlock protective scheme

This scheme was described in papers [5, 6] presented at the International
Conference on Large Electric High-Tension Systems held in Paris in June
1931. The diagram which was included to illustrate the scheme in the second
of these papers is reproduced as Fig. 10.13.
It will be seen that the scheme, which is shown in single-phase form for
simplicity, incorporated directional relay elements (1), starting relays (2) and
interlocking relays (4). The interlocking relays (4) opened the tripping circuits
when they operated and therefore the scheme employed blocking signals
which were sent over pilot wires normally used for telephony. To meet the
requirement that time must be allowed for blocking signals to arrive before
tripping can be initiated, which was referred to in section 10.3.3 (page 361),
the tripping elements (7) operated after a time delay of 0.3 s.
10.4.2 Modern schemes

Modem interlock schemes incorporating directional relays are based on re-
laying quantities derived from incremental voltage and current signals. In-
cremental signals Av and Ai are obtained after removal of the prefault
[Jl
81 18
< ~-f-~
Fig. 10.13 Diagram illustrating the 'Interlock' protective scheme. (1) Three-phase
directional stabilizing element. (2) Three-phase over-current element. (3) Trip circuit.
(4) Interlock relay. (5) Telephone. (6) Pilot. (7) Tripping element with time lag.
(Reproduced from Leeson and Leyburn, 1931, CIGRE Conference Paper 106 with the
permission of CIGRE.)
components and their relative polarity indicates the direction of propagation

of the disturbance caused by the fault. Comparison of this information ob-
tained at the two ends of the protected line can be used to discriminate between
internal and external faults. Practical implementations of these ideas are de-
scribed in references [7, 8, 9] and the schemes are examined in more detail in
Chapter 12.
Phase-comparison protective schemes 367
A directional earth fault comparison relay 75N21, produced by Siemens, is
supplied with the sum of the outputs of the three line current transformers and
either a voltage proportional to the sum of the three line voltages or a polar-
izing current obtained from the station transformer neutral. The relay has a
two-stage (1 and 2) zero-sequence current detector. The first, more sensitive
stage initiates the directional measurement whilst the second stage allows
circuit-breaker tripping to be initiated. Current transformers with secondary
windings rated at 1 A or 5 A may be used and Stage 2 settings of 0.1 to 0.85 pu
in 0.05 pu steps are available, i.e 0.1-0.85 A or 0.5-4.25 A. Angular operating
zones of n rad, i.e. nl2 rad, are provided, and operating times are 25 ms or
more depending on the fault conditions.
Permissive signals are employed with this scheme and a special feature is
provided so that should a low current be fed into one end of a line to an internal
fault then tripping of the other end nevertheless will be initiated, if the fault
current is high enough, after the receipt of an echo of the signal sent out by
that end.
This scheme, which provides very sensitive earth fault settings, has been
developed to detect high-resistance faults on transmission lines and it is in-
tended to be used with other schemes which will detect the high-current
interphase faults.
Clearly, both the above schemes must be fed from current transformers
capable of faithfully reproducing the waveforms of their primary currents.
10.5 FEATURES OF PHASE-COMPARISON PROTECTIVE SCHEMES
The principle involved in comparing the phases of the currents at the two ends
of a line is related to that of comparing the instantaneous values of the current,
as in Merz-Price type protective schemes. It is also related to the directional-
interlock principle because the phase of the current at an end of a line changes
significantly when the direction of the current flow reverses to feed a fault on
a line.
Phase-comparison schemes have an advantage over current-differential
schemes in that only information about the phases of the currents must be
transmitted between the ends of lines and therefore attenuation of signals is
acceptable. They also have an advantage over directional-interlock signals in
that they do not require information about the line voltages and therefore
voltage transformers are not needed to supply them.
As stated earlier in section 10.1, phase-comparison schemes in which high-
frequency signals were transmitted along the conductors of protected lines
were introduced about 1950, by which time suitable electronic equipment was
available.
Several factors which affect phase-comparison schemes are considered in
the following sections.
10.5.1 The phase displacements of line currents
If there was no capacitance or leakage resistance between the conductors of a
healthy transmission line, then the currents at the two ends of each of its
phases would be the same at every instant and therefore they would be in phase
with each other.
If either a phase to earth or an interphase short circuit was present on such
a line this condition would not hold and the phase currents would be displaced
from each other by considerable angles. As an example, the currents for a short
circuit to earth at the mid-point of one phase of a line would be almost in
antiphase with each other.
The phase-displacements between the currents in the event of highly-resistive
faults could however be quite small as was shown earlier in section 10.3.1 and
Fig. 10.10.
In practice, of course, significant capacitance is present between the phase
conductors of long lines and therefore, although the leakage resistances are
very high, the instantaneous currents in each phase at the two ends of lines
are not the same. As an example, the capacitive current per phase of a British
400 kV line, 100 miles long, is about 140 A and the resultant phase displace-
ment between the currents at the two ends when operating at rated current is
approximately 0.036 rad.
It will be clear from the above that phase comparison schemes must be so
arranged that they will not initiate tripping of the circuit-breakers when the
phase displacements are less than a certain set value but that circuit-breaker
opening will be initiated when displacements exceed the set value.
The phase-displacement settings provided by particular manufacturers are
quoted later in section 10.6 and experience has shown that these provide
satisfactory discrimination and also enable internal faults above certain current
levels to be detected. If it is felt to be necessary the phase displacements which
may occur in particular applications may be calculated using either the full
three-phase long-line equations [(10.1) and (10.2)] or simpler approximate
equations and in this way the suitability of proposed angular settings may be
determined.
10.5.2 The production of comparison signals
Comparison of the phases of each of the currents at the ends of a line would
require three separate signals to be sent in each direction, and whilst this could
be done it is clearly preferable that only a single signal should be transmitted,
when necessary, from each end.
Summation processes using current transformers of different ratios or sum-
mation transformers fed from current transformers of the same ratios, as
described earlier in section 10.3.1, could be used to provide a single output at
each end of a line. It has, however, been the usual practice to use symmetri-
cal-component sequence networks for this purpose. The sensitivities to the
separate components of the networks used in practical schemes are quoted in
section 10.6.
10.5.3 The comparison process

During healthy conditions the currents at the opposite ends of each of the
phases of a line are almost in phase with each other, whereas they are displaced
by large angles when internal faults are present. These conditions are illus-
trated in Fig. lO.l4(a). If the current transformers are connected in the reverse
direction at one end relative to that at the other or the outputs from the
sequence networks are taken in reverse directions, then these outputs will be
almost in antiphase with each other when a line is healthy, as shown in Fig.
10.14(b), and, of course, the outputs will be almost in phase with each other
when internal faults are present.
As a result, if a signal is produced at end A of a line whenever the sequence-
network output at that end is positive and a signal is generated at the other
end B when the output there is positive then the receipt at end A of this latter
signal from end B together with the locally derived signal at end A will
provide a continuous signal when the line is healthy and this can be used to
prevent tripping of the circuit-breaker at end A. Clearly, a similar process
would occur at the second end of the line (B). These conditions are illustrated
in Fig. 10.14(c).
During internal faults, the two signals overlap, as shown in Fig. 1O.14(d),
and there are significant periods when there is no resultant signal and this
condition can be detected and used to initiate the opening of the circuit-brea-
kers.
In practice, the phase currents at the two ends of long healthy lines are not
exactly in phase and therefore there will be a short break each cycle in the
signals at each end with the arrangement described.
In addition, if the outputs of the sequence networks have to reach a particu-
lar positive level before signals are generated and transmitted, then the dura-
tions of each of the high frequency signals will be less than the half period of
the power system quantities, as shown in Fig. 10.15. As a result there will be
short breaks each half cycle in the signals fed to the receivers at each end of
a scheme and these breaks will be longer the lower the outputs of the sequence
networks, i.e. the effect will be more pronounced the lower the current flowing
in the line.
To allow for the above effects, the detection process must be so arranged
that circuit-breaker tripping is only initiated when the breaks exceed a certain
period, say a period corresponding to 1t/12 rad of the power-frequency cycle.
It will be clear that the above results may be produced in other ways. As an
example, the current transformers and sequence network output-windings
could be connected in the same directions and signals could be derived at one
Current
at end
A
rot rot
(a)
Output
at end
A
rot rot
(b)
Signal produced Signal obtained Signal produced Signal obtained

at end Infrom end B at end B end A
\
(c)
Signal produced Signal produced

at end A and at end Band
obtained from obtained from
end B end A
\
rot
(d)
Fig. 10.14 Conditions when faults are present on a line and external to it.
End A End B
Output Output
at end at end
A B
Level above which

_signal is produced
wt wt
Signal produced Signal obtained Signal produced Signal obtained
/
at end A from end B at end B end A
\ \
Fig. 10.15 Conditions obtained on a healthy line.
end when the network output was positive whilst signals would be sent from
the other end when the corresponding output was negative.
10.5.4 Comparison signals

These schemes were developed for application to long lines and they were
designed so that the signals could be transmitted along the conductors of two
of the phases. Initially high frequency signals in the frequency range 80--500
kHz were used.
A signal must be sent from each end of a line to the other end and two
frequencies sufficiently far apart must be used in a scheme so that filters of
reasonable bandwidths may be provided to pass the signals being transmitted
and received at each end without interference occurring. As an example, the
signals could be at frequencies of 140 kHz and 148 kHz.
During period of transmission, i.e. when the output of a sequence network
is positive, the transmitted signal is of constant amplitude.
10.5.5 Signalling equipment

It will be clear that an oscillator and amplifier must be provided at each end
of a line to produce the outgoing signals and a receiver is needed to feed
incoming signals to the comparison circuits. Coupling equipments to enable
signals to be both injected into and received from a protected line are required
as also are the filter circuits referred to above.
These items have varied in form over the years and details of the equipments
provided by some manufacturers are provided in the later sections of this
chapter.
It will be appreciated that alternative methods of signalling may be used and
certainly phase-comparison schemes could operate using pilot wires or optical
fibres to link the ends of lines although such links would often be very costly.
VHF radio links could also be used.
Whatever equipment is employed, reliable power supplies must be provided
to ensure that signals will be transmitted whenever they are required to ensure
that correct operation of the protective scheme will be obtained.
10.5.6 Starting relays

As with interlock protective schemes, it is undesirable that signals should be
sent continuously and therefore starting relays are necessary.
To ensure that the opening of a circuit-breaker will not be initiated because
a signal has not been received from the other end of a healthy protected line,
Stage 1 and Stage 2 relays must be provided at each end of a scheme. These
relays, which are fed from the sequence networks, are current-operated, the
Stage 2 relays having the higher settings.
Operation of the Stage 1 relays permits the comparison signals to be sent
and the operation of the Stage 2 relays allows the circuit-breakers to be tripped
if necessary. It is also desirable that the Stage 1 relays should operate slightly
faster than the Stage 2 relays.
10.5.7 Current transformers

Because phase-comparison schemes operate when there are significant breaks
in the signals sent to their receivers it is essential that such breaks should not
be produced as a result of the distortion of the output waveforms of current
transformers, because of. saturation of their cores when large currents are
flowing due to faults on circuits connected to them.
Current transformers must therefore be so designed that they will not satu-
rate during the transient conditions which may occur on the circuits in which
they are to be used. Protective equipment manufacturers specify the minimum
knee-point voltages of the current transformers which may be used with their
schemes and examples are quoted in section 10.6.
10.6 PHASE COMPARISON SCHEMES

Details of a number of schemes are provided in the following sections.
Phase comparison schemes 373
10.6.1 Telephase protective schemes
This scheme was developed jointly by A. Reyrolle and Co Ltd, and GEC
(Telephones) Ltd, prototype equipments [10] being produced in 1945. After
these were operated on an experimental basis, a final production version was
applied to one of the transmission lines between Loch Sloy Hydro-Electric
Station and Windyhill Switching Station, north of Glasgow in 1950. This line
was about 40 miles in length.
The equipment, which was supplied by current transformers with secondary
windings rated at 1 A, incorporated sequence networks. The connections were
such that their outputs were in antiphase with each other when the phase
currents at the ends of the protected line were in phase with each other.
The sequence networks produced voltage outputs given by:
VOU! = k (It. + 8/2.)
in which ft. and lz. are the 'a' phase positive- and negative-sequence currents
respectively and k is a constant.
The general arrangement of the equipment was as shown in Fig. 10. 16(a).
When the output of a network exceeded the level needed to operate its asso-
ciated Stage I relay, the high frequency signal produced in the oscillator was
modulated by the output signal of the sequence network, and a high frequency
signal of constant amplitude was injected into the line conductors, via coup-
ling equipment, during those periods when the output of the sequence network
was positive. Because of the tuned-line traps, inserted in series at each end of
the line, the injected signal travelled to the other end of the line, where after
passing through the line-coupling equipment it was fed to a receiver. Each
receiver not only received signals from the other end of the line but also the
high frequency signals produced at its own end.
As a result, signals would have been received continuously by each receiver,
as shown in Fig. 10.16(b) if the currents in the two ends of the line were in
phase. When the line currents were not in phase, however, the signals over-
lapped as shown in Fig. 1O.16(c) and there were periods during each power-
frequency cycle when the receivers did not receive signals. When the duration
of a period in which no signal was received exceeded I/lth of a cycle of power
frequency, i.e. rc/6 rad, an output relay operated. This relay then initiated the
opening of the associated circuit-breaker provided that the local Stage 2 relay
supplied by the sequence network had operated.
The phase-displacements between the currents at the ends of a healthy line
tend to be considerably less than rc/6 rad. As an example, the phase displace-
ment on a three-phase, 400 kV 3800 A line of 100 miles in length is 0.036 rad.
Should a fault occur on a circuit connected to such a line, the voltage at the
end near the fault would be depressed below normal, causing reductions in the
capacitive current and phase displacement.
Experience has shown that the above angular range of rc/6 rad in which
DF-Directional-filter. F-Wave-filtec. R-Output-relay. St-Starting-relay.
I SrStarting-relay. T3-Transformer. Ts-Transformec. TFt-Summation-transformer.
I TFrSummation-transbmer. TR-Tripping relay. V t-Oscil1ator. Vz-Modulator.
VrAmplifier. V4-Amp1ifier. V4-Mixer. V"Amplifier.
C:
o rJ)
UCi>
<I) rJ)
'- c:
.!: <I)
."0
u.c:
I'o
0
u:
_<I)
!!?
I-
~ -~
~ E
!]ltt,ttll------------------I
Fig. 10.16 (a) Telephase protective scheme, (b) basic operating principle when cur-
rents are in phase, (c) and out of phase. (Reproduced from The Reyrolle Review, No
142, Jan-Mar 1950, with the permission of Reyrolle Protection.)
External fault Internal fault
End A End B End A End B
Primary f\f\ f\f\ Primary

4JV current VV VV current N
4JV Secondary f\f\ ~ Secondary -4JV
current current
VV
______ Transmitted Transmitted
. .
signal
.--.-T.:Ir:....-----------_II H I signal 1
H
1r-=--=--=--=--=--=--='
I Rectified 6 1 Rectified 6 1
--r--.-..,.t~ received .;-L...,-_----.~ .--r--r--.-~t received r-'-r-"T--'-
o U signal 0 o 0 signal 0
f\ l\ Locally l\ l\ I
l'\ l'
,\
Locally 1\ l\
v, \I ' derived ' \ I 'V derived V' \I
V voltage V V\} voltage V
Tripping Tripping
valve on valve nn
output output
(b) (c)
Fig. 10.16 (b), (c)

operation does not occur, does usually provide an adequate factor of safety
and assures correct discrimination. On very long lines, however, a larger range
may be needed and schemes are arranged so that ranges up to 1tf3 rad may
be selected.
The various electronic units in the early Telephase schemes, such as the
oscillators and amplifiers, which were mounted on racks, employed ther-
mionic valves. These had limited lives and therefore the circuits were moni-
tored to detect failures and in addition provision was made to enable operators
to check the condition of the equipments regularly. Equipment could be pro-
vided to initiate checks automatically at regular intervals.
It will be appreciated that the schemes needed secure power supplies and
standby generating equipment was provided for this purpose.
The fault settings available on these schemes, which were controlled by the
sequence networks and the Stage 2 starting relays, were as shown in Table
10.1.
It will be seen that the settings were dependent on the phases on which faults
occurred and the setting for three-phase faults was above the rated current
level so that the comparison process was not initiated by normal load currents.
This high three-phase setting was considered unacceptable for some applica-
tions and later versions of the scheme incorporated sequence networks which
provided separate voltage outputs (VI and V2) proportional to the positive- and
negative-sequence inputs supplied to them. The separate outputs were fed to
modulation amplifiers which produced outputs proportional to (kv2 and VI) and
these outputs were used to modulate the high frequency carrier signals which
were injected into the protected lines.
With this arrangement four level detectors were provided at each end of a
scheme, i.e. a pair of low-set devices and a pair of high-set devices. These
replaced the Stage 1 and Stage 2 relays used in the original Telephase scheme.
The low-set detectors again allowed the comparison process to be initiated and
the high-set devices allowed tripping to take place when faults occurred within
the protected zone. One low-set and one high-set detector at an end were fed
with the output of the local negative-sequence network and because this is
Table 10.1
Type of fault Setting (percentage of current
transformer secondary rating)
Phase 'a' to earth 75
Phase 'b' to earth 90
Phase 'c' to earth 90
Phase 'a' and 'b' 45
Phase 'b' and 'c' 60
Phases 'c' and 'a' 45
Three-phase 225
zero under normal healthy conditions, the detectors could be given settings
well below the value corresponding to the rated current of the protected line.
The other low-set and high-set detectors were fed with the output of the
positive-sequence network via impulse networks which only provided outputs
when the line currents changed suddenly. This second set of detectors was not
therefore sensitive to the passage of normal load currents and could thus be
set to detect relatively small three-phase fault currents, unlike the earlier
scheme.
Various forms of Telephase have been applied to many transmission lines
over the years and they have performed well, operating for internal faults in
times of about 20 ms.
The latest form of Telephase protection which is designated T3 [11], is
produced by Reyrolle Protection, NEI Electronics Ltd. It incorporates the
above impulse-detection arrangements and is arranged as shown in Fig. 10.17.
As a result it can provide the range of fault settings shown in Table 10.2.
Because the scheme utilizes modem electronic equipment, several facilities
such as automatic checking procedures are incorporated. A further feature of
this scheme is that the angular zone in which operation will not occur, i.e. the
stabilizing zone is not fixed as in the early schemes, but increases as the line
current decreases. It varies between limits of about 'It/6 rad to 'It/3 rad. It
thus takes account of the fact that the phase displacement caused by the
capacitive current of a line decreases as the current fed through it increases.
The operating time of this scheme is somewhat dependent on the levels of
fault currents and also on the transient components present and as can be seen
from Fig. 10.18 it may vary between about 18 ms and 40 ms.
To ensure that incorrect operation will not occur because of saturation of
the cores of current transformers feeding Telephase T3 schemes, it is recom-
mended that the secondary knee-point voltage of the transformers should
satisfy the following conditions:
X
Vk ;;;;. 1.5 -lr (A + B + C)
R
in which
1.5 is a stability factor,
Table 10.2
Type of fault Fault setting (percentage of current
transformer secondary rating)
Balanced Three-phase 30 40 50 60 70
(impulse)
Unbalanced Phase to earth 36 48 60 72 84
Phase to phase 20 27 34 41 48
"''''''l
- IJQ
-" I-/-X 1f'l'Pt I - 50 HzlPLC intertace
50 Hz circurts PLC circurts
.....,,"
t-- .... t
(J? I
~~ I
I:> t::C I
'"~ g -
g~
~I:>CICl~
.... ...
~. 3
;:, 0
."...,
~ -l
~ ~
" 0
5 -g.
;:, I
-..,0'"
:->4'
"""0
~g
~ g.
::." ::s
::>"-l
~~
~'Stl
3~
~.
_. 8-
c::
o (")
= 0
.....
o..., Q.
za
~3
~;;:
'!i~
50 Hz circurts ---+----1
:~
~~
80
70
CIl Operating-time measured from the fault

"0
c: 60
0
application to operation of the trip relay
0
(J)
~ 50
e
.!: 40
(J)
~
C, 30
c:
~
(J)
a. 20
0
Minimum current transient
10
0
0 5 10 15 20 25 30
(a) Multiples of current-setting
Fig. 10.18 Overall operating time/current characteristic (a), and typical operate
time/point-on-wave initiation b-e fault at three times fault-setting (b).
X/R = Inductive reactance/resistance ratio of the power-system impedance
corresponding to the maximum through-fault conditions,
If is the maximum through-fault in the secondary winding of the current
transformers (A),
A is the resistance of the current-transformer secondary windings (0),
B is the resistance of the connection between each current transformer and
the sequence network (0),
C is the impedance of the sequence network (0), which has maximum
values of 0.1 0 and 0.2 0 respectively when current transformers with
1 A and 5 A secondary current ratings are used. The value of C is
normally very small compared with A + B.
10.6.2 Contraphase protective schemes

The original Contraphase scheme was produced by Metropolitan Vickers Elec-
trical Company Ltd, at about the same time as Telephase and both schemes
worked on the same principles and performed in similar ways.
GEC Measurements now produce two modern versions of Contraphase
which are designated PIO and P40 [12]. The scheme PlO, which has an output
signal power of 10 Wand can operate over links with attenuations up to 36
dB, is suitable for use on transmission lines up to 200 miles in length. The
scheme P40 has an output power of 40 Wand it can operate over links with
40 dB attenuation. It may therefore be applied to lines longer than 200 miles.
These schemes are so connected that the transmitted signals are in antiphase
with each other when the currents at the two ends of a protected line are of
the same phase.
As in the Telephase scheme, low- and high-set starters are provided which
operate when the negative-sequence outputs of the sequence networks exceed
pre-set levels and low- and high-set impulse starters are also provided which
operate when there are sudden increases above set levels in the positive- and
negative-sequence outputs of the sequence networks. As a result the schemes
have a high sensitivity to all types of faults, the settings available being as
follows:
Positive-sequence impulse currents 30-60% of rated current in
5% steps
Negative-sequence impulse currents 5-20% of rated current in
2.5% steps
Negative-sequence non-impulse currents eight times the impulse value
selected.
To ensure that current comparisons may be maintained when faults are
present, it is arranged that the impulse starters have dwell times during which
they remain operated after being initiated, provided that fault current is still
flowing. The high-set dwell time is 0.5 s and the low-set dwell time is 0.6 s.
Auto-reclosing 381
In addition, the low-set starters have time delays which maintain the trans-
mission of carrier signals for lOOms after faults are cleared.
The carrier signals which are transmitted are modulated by an output derived
from the positive- and negative-sequence networks, i.e. k\ /\a + K2 /2a in which
k\ =- 1 and k2 = 16.
The stability angle, i.e. the phase displacement over which operation will
not occur, is nominally about 1t/6 rad but, as with Telephase T3, this angle
increases at low-current levels for which the effects of the capacitive currents
in a line are greater.
This scheme, which has an operating time between 10 and 40 ms for fault
currents of five times the setting levels, incorporates in-built testing features
and secure power supplies.
To ensure that Contraphase schemes will operate satisfactorily, the manu-
facturers recommend that current transformers with knee-point voltages (Vk)
of the following levels be used:
in which
X/R is the primary system reactance/resistance ratio,
It is the maximum through-fault current (secondary A),
ReT is the resistance of the secondary winding (0),
Rw is the resistance of leads between the current transformers and the
protective equipment (0).
10.6.3 Other schemes

Schemes with similar characteristics to those described above are produced
by other manufacturers, an example being protection relay 7SD32 [l3] which
is available from Siemens AG.
10.7 AUTO-RECLOSING
Many of the faults which occur on overhead lines are caused by flashovers
between phase conductors or between one or more of the phase conductors
and earthed metal or the ground. After the initial breakdown quite large
currents flow in the arcs which are formed and considerable ionization is
produced.
Such conditions cannot be allowed to persist and they must therefore be
detected by protective equipment which will initiate the qpening of the cir-
cuit-breakers associated with the faulted line.
Clearly the arc or arcs will extinguish after the line is de-energized and the
ionized products will then disperse. As a result, the line could be fe-energized
after a certain period, which is termed the dead time.
It will be appreciated, however, that the disconnection of a line connected
between sources affects power transfers and it could lead to asynchronous
operation, i.e. instability. It is therefore necessary that dead times be so chosen
that the line circuit-breakers will be reclosed before instability will occur. In
most cases a dead time of about 0.2 s proves sufficiently long to allow arc
products to disperse and yet enables stability to be maintained.
The above process, which is referred to as auto-reclosing, is now widely
used on both transmission and distribution lines. It is implemented in several
different ways. In each case, operation of the protective equipment associated
with a faulted line initiates the opening of the circuit-breakers which then
automatically reclose after the pre-set dead time. Should the isolation of the
line clear the fault then no further action is initiated and the line is left in
service. If, however, the fault is not cleared at the end of the dead time, either
because it is a solid fault, i.e. direct contact between conductors or to ground,
or because the products of arcing have not dispersed sufficiently, then the
protective equipment detects the presence of the fault when the line is reclosed
and initiates the opening of the circuit-breakers again. Because it is clearly
undesirable to repeatedly reclose a major circuit on to a fault, schemes are
arranged so that either only one reclosure is possible, i.e. if a fault has not
been cleared when the circuit-breakers reclose initially, then when they are
opened for the second time a further reclosure will not occur. Two or more
reclosures are allowed, however, on less important circuits such as distribution
lines. To differentiate between the arrangements described above, they are
referred to as single shot reclosing and multi-shot reclosing.
It will be appreciated that the disconnection of a major transmission line
significantly affects other parts of the network of which it forms a part and,
as stated above, it can lead to asynchronous operation after a relatively short
period of time. Clearly the effect of opening only one phase of a line is less
than that caused by opening all three phases because power can be transferred
on the two unopened phases and therefore instability might not be caused or
it will not occur as quickly as it would with three-phase opening. It is therefore
advantageous, when single-phase to earth faults occur on a line, to open only
the faulted phase and then to use one of the reclosure methods described above
and this is the usual practice when suitable circuit-breakers are provided, i.e.
those in which the phases may be operated independently.
To enable this arrangement to be implemented, it is clearly necessary that
the protective schemes should be able to determine the types of faults which
may occur on a transmission line, e.g. interphase and single-phase to earth,
and the phases on which they are present so that the appropriate circuit-breaker
operation and reclosure procedures may be initiated.
Faults involving all three phases are relatively rare and when they do occur
it is usually because of errors, such as the failure to remove earths, applied
while maintenance work has been done, before a line is re-energized. Reclo-
sure in such circumstances is undesirable and, as a result, some users prefer
References 383
to inhibit automatic reclosing when three-phase faults are detected by the
In some cases, when voltage transformers are present on a network in
positions which enable the voltages on both sides of line circuit-breakers to
be measured, check-synchronizing equipment is provided to ensure that auto-
matic reclosures do not take place unless the phase displacements across
circuit-breakers are within acceptable limits. It will be appreciated that when
a line is disconnected at both ends to clear a fault, the voltages on the line
sides of both its circuit-breakers will be zero and therefore the phase displace-
ments at neither end can be determined. It is thus necessary that the circuit-
breaker at one end be allowed to close after the necessary dead time and that
check-synchronizing equipment at the other end should determine whether
conditions are suitable for the second circuit-breaker to be reclosed. Further
details of problems associated with auto-reclosing may be found in reference [14].
10.8 THE FUTURE
It will be evident from the preceding sections that both interlock and phase-
comparison schemes have been used successfully to protect transmission lines
for many years and modem versions of these schemes are being implemented
using the latest electronic circuitry and equipment. It therefore seems likely
that they will continue to be applied in the foreseeable future and although their
hardware and that of the communication links will continue to be up-dated,
the basic principles and features will not be changed.
REFERENCES
1. Pilot relaying performance analysis - IEEE Committee Report, (1990) IEEE
Trans, PWRD-S, 85-102.
2. Eggimann, F., Senn, W. and Morf, K. (1977) The transmission characteristics of
high voltage lines at carrier frequencies, Brown Boveri Review, 8, 449-459.
3. Fibre optic channels for protective relaying, report prepared by the Fibre Optics
Relay Channels Working Group for the IEEE Power System Relaying Committee
(1989): lEE Trans, on Power Delivery, 4, (1),165-176.
4. Power System Protection. Vol 1 Principles and Components (1981), (ed. Elec-
tricity Council), Peter Peregrinus 2nd edn, Chapter 7.
5. Clothier, H. W. (1931) Overhead line and feeder protection, CIGRE Paris, 1931,
Paper No. 88.
6. Leeson, B. H. and Leybum, H. (1931) The principles of feeder protection and
their application to three modem systems, CIGRE Paris, 1931, Paper No. 106.
7. Chamia, M. and Liberman, S. (1978) Ultra high speed relay for EHV/UHV
transmission lines - development design and application, Trans. lEE, PAS-97,
2104-2112.
8. Johns, A. T. (1980) New Ultra-high-speed directional comparison technique for
the protection of ehv transmission lines, Proc. C, lEE, 127, 228-229.
9. Vitins, M. (1981) A fundamental concept for high speed relaying, Trans. lEE,
PAS-lOO, 163-168.
10. Lackey, C. H. (1950) A review of British practice in the protection of electric
power systems, The Reyrolle Review, No 142, pp 1-19.
11. Telephase T3, PLC Phase-Comparison Protection, Pamphlet T3.4186, (12 pages),
NEI Reyrolle Ltd.
12. Type Contraphase PlO/p40, Phase Comparison Carrier Protection, Publication
R-5247B (11 pages), GEC Measurements.
13. Phase Comparison Protection Relay 7SD32, Protective Relays, Siemens Catalog
R.1989, pp 6/25-6/33.
14. Automatic reclosing of transmission lines - an IEEE Power Systems Relaying
Committee Report (1984), Trans. lEE, PAS-103, 234-245.
11
Distance-type protective schemes for
overhead lines and cables
In Chapters 7-10 various schemes which are used to protect overload lines
and cables have been considered. All of these schemes are of the unit type in
that they should operate whenever faults above certain levels of current occur
on the protected units, i.e. between the current transformers mounted at each
of the ends of the units, and they should not operate when external faults are
present or during any healthy conditions. In each case this desired performance
is obtained by making use of information about the conditions at both ends of
a protected unit at any time. As examples, the instantaneous input and output
currents of lines are compared in current-comparison (Merz-Price) schemes,
the phases of currents are compared in 'phase-comparison' schemes and the
directions of currents are taken into account in interlock schemes. To effect
the necessary comparisons, signals must be sent between the ends of protected
units which may, of course, be quite long distances apart, and either pilot wires
or optical fibres or other high frequency electronic communication equipment
may be required for this purpose.
An alternative method of protecting overhead lines and cables, without the
need for comparisons between the quantities at both ends of the protected
units, is based on measuring the input impedances of lines and cables and to
do so clearly requires that information about both input currents and voltages
be available. It will be clear, therefore, that such schemes must be fed from
both current and voltage transformers mounted near the ends of protected
circuits.
It is shown later that it is not possible to measure impedance with such
accuracy that discrimination can be achieved between faults a short distance
from the end of a long line and those just beyond its end, and therefore these
schemes are not of the unit type and hence features, such as time delays, must
be incorporated to enable correct discrimination to be obtained between inter-
nal and external faults.
The input impedance of a short-circuited line or cable varies from zero for
a fault at its input end to a finite value for a fault at its remote end, the actual
impedance value increasing with the distance to the fault; therefore schemes

386 Distance-type protective schemes
based on such measurements are referred to either as 'distance- or distance-
measuring protection' or alternatively as 'impedance-measuring protection'.
These schemes have been applied successfully to many circuits over the
years and they are still being produced today. Detailed information of various
schemes and their behaviour is given later in this chapter. In the following
section, however, details of the original impedance-measuring schemes and
other historical information are provided.

About 1920 some difficulties were being experienced in obtaining correct
discrimination when existing relays were applied to power lines. This caused
P. Ackerman to examine the possibility of using the input impedance to a line
to obtain discrimination. He subsequently produced a relay with a beam,
pivoted at its centre, which had armatures at each of its ends, these being
attracted to electromagnets. One of the electromagnets carried a current pro-
portional to the input voltage of the protected circuit and the other carried a
current proportional to the input current to the circuit. Under normal condi-
tions the net torque on the beam tilted it towards the voltage-energized electro-
magnet but when the ratio of the input voltage to current of the circuit fell
below a particular value, the beam tilted in the opposite direction, causing the
operation of contacts. This relay, which was described in a paper [1] published
in the J oumal of the Engineering Institute of Canada in December 1922, was
the forerunner of the beam relays which were produced in large numbers in
the following decade.
Shortly afterwards Crichton described a different design of impedance-
measuring relay in a paper [2] published in the Transactions of the lEE in
April 1923. This relay had an induction-type element in which the disc was
driven by an electromagnet energized by a current proportional to the current
in the protected circuit. When the disc rotated it wound up a helical spring,
one end of which was connected to an armature attracted by an electromagnet
energized with a current proportional to the input voltage of the protected
circuit. This relay, which was similar to that described later in section 11.3.1,
operated in a time which was approximately proportional to the input imped-
ance of the protected circuit.
Development work continued and it was recognized that impedance relays
could not discriminate between faults on a protected circuit and those on other
circuits connected to the same busbars, i.e. between forward and reverse faults.
The need to use impedance relays in conjunction with directional relays was
therefore appreciated and certainly suitable directional elements with watt-
metric-type movements were available when McLaughlin and Erickson of
Westinghouse Electrical and Manufacturing Co had a paper published in the
Transactions of the lEE in 1928 [3].
At this time it was realized that the behaviour of impedance relays was
The behaviour of overhead lines 387
affected by the resistances of fault arcs and ground-return paths and relays
which were only affected by the input reactances of protected circuits had been
produced.
The publication of a paper [4] by George in March 1931 in the Transactions
of the lEE indicates that the use of distance protection was increasing. It was
stated in the paper that both impedance- and reactance-measuring relays pro-
duced by the Brown Boveri, General Electric and Westinghouse companies
were in use on circuits of the Tennessee Electric Power Company in the USA.
11.2 THE BEHAVIOUR OF OVERHEAD LINES

The conductors of overhead lines possess resistance and inductance and in
addition there are capacitance and leakage conductance between them. These
parameters are distributed throughout the lengths of the conductors, R, L, C
and G usually being used to represent the values per metre length.
The effects of capacitance and leakage conductance on the behaviour of
relatively short lines operating at power frequencies are very small and they
will be neglected in the following section in which the conditions which may
be encountered on such lines are examined. To further simplify the treatment
a single-phase line will be considered.
11.2.1 The input impedance of a short single-phase line

It will be evident from Fig. 11.1(a) that the input impedance (Zin) to end A of
the line P during steady state conditions if a short-circuit was present at a point
a distance x along it would be given by:
Zin = ~A =(R + jroL) x Q

fA
This impedance clearly increases linearly with the distance to the fault (x)
until it reaches its maximum value for a fault at the remote end of the line,
i.e. Zin = (R + jroL) lQ. This variation is shown in Fig. 11.1(b). On most lines,
the ratio of the inductive reactance to resistance, (roL/R), is high and therefore
the phase angle <p of the input impedance is quite high. As an example the
roL/R ratio of the 400 kV transmission lines in Britain is 16 and the corre-
sponding value of <p is 1.508 rad.
If the line P supplied a similar line Q on which a short-circuit occurred at
a point a distance XQ along it, then the input impedance to line P would be
given by:
Clearly therefore, an input impedance (Zin) in the range between the points 0
and M on Fig. 11.1 (b) would indicate a fault on line P and a value above the
Line P(R and L per End B

unit length)
1-=- ~-t-
Fault v.
B
x--
LineQ
lin x
I"
(a)
I
I
I
TI
0 RI
Resistance
(b)
Fig. 11.1 Input impedance of a faulted line.
value at point M would indicate that line P was healthy. This is one of the
simple principles on which distance- or impedance-measuring schemes are
based.
It will be appreciated that discrimination must also be achieved between
short-circuits on a protected line and normal healthy operating conditions.
During the latter conditions a protected line (P) may be supplying a range of
loads connected to the busbar at its remote end (B). Should these loads be
resistive then the input impedance (40) to line P would be given by:
40= (R+ jroL) L+RL Q
in which RL is the effective resistance of the loads.

The locus of the input impedance would thus be on the line MN in Fig.
11.1 (b). Considering also the limiting cases of highly-inductive and capacitive
loads, the measured impedances would lie along the lines MS and MT respec-
tively.
Impedances with values in the zone to the right of the line ST in Fig. 11.1 (b)
could therefore be encountered during normal healthy conditions. In practice,
however, the load impedance has a minimum magnitude related to the total
rated current (lr) of the loads, i.e. VB/lr, and therefore the input impedances
(Zin) to line P would not have values within a semicircle such as that shown
shaded in Fig. 11.1 (b).
From the above considerations it is clear that the input impedances which
may be present when short circuits occur on a protected line (P) are different
to those which may be present at other times and therefore satisfactory dis-
crimination is achievable. A further factor which must be taken into account,
however, is that significant resistance may be present in a fault path. The
input-impedance (Zin) to a protected line when such a fault occurs at a point
a distance x along it will be given by:
Zin = (R + jroL)x+Rf
in which Rf is the fault resistance.
The range of input impedances which may occur when faults with resistan-
ces up to a certain value (Rfmax) occur on a protected line will thus be within
the shaded parallelogram shown on Fig. 11.2(a).
It will be seen from Figs. 11.1(b) and 11.2(a) that correct discrimination
could be achieved for faults with resistances up to a particular value at which
the part of the parallelogram to the right of the vertical through the point RL
in Fig. 11.2(a) lies in the shaded semicircle in Fig. 11.1(b), i.e. when the
vertical height OA in Fig. 11.2(a) is greater than the height OU in Fig. 11.1 (b).
This condition is shown in Fig. 11.2(b).
Should a protected line form part of a highly interconnected network, then
current may be fed through it in the reverse direction to normal when faults
occur on the circuits connected to its input end. This situation is illustrated in
Q)
0
c
C\I
ti
C\I
~
Q)
>
t5::l
-0
.E A
U
o RI
Resistance 0
Resistance
(a) (b)
Fig. 11.2 Range of input impedances for resistive faults.

--x-tx-
End A _ End B
T{-LX--.;.~-.-;;L;;;.;.;..ine.;.....p
-=-
Fault
vJ I
Z
VB
In
(a)
Ql Ql
> 0
+=~
0_
::J 0
"0111
E~
Resistance
(b)
Fig. 11.3 Apparent input impedance of a line in a highly-interconnected network.
Fig. 11.3. In the~ circumstances the fault current (IF) would usua.!!y lag the
busbar voltage (VA) and the apparent input impedance of line P (Zin) would
then have both a resistance and inductive resistance which are negative, an
example being shown in Fig. 11.3(b). Should the fault be on busbar A then
the apparent impedance would be zero.
It will be clear from the above that faults on a protected line could be
detected and that correct discrimination could be achieved in this simple case
by providing an impedance-measuring relay which would operate for imped-
ances in a section of the Z plane, such as that shown in Fig. 11.2(b).
11.2.2 The input impedance of a long single-phase line

When long lines are being considered it is necessary to take into account the
distributed nature of the impedance parameters R, L, C and G. The last com-
ponent (G) usually has a very low value which is difficult to assign. It has
little effect on the line behaviour and is therefore usually neglected.
In the general case, however, the voltage - current relationships are ex-
pressed by the well-known long-line equations below.
Vx = VA cosh yx - IA Zo sinh yx (1Ll)

- - V
Ix = IA cosh yx - _A sinh yx (11.2)
Zo
in which Vx a!!.d fx ar~ the quantities at a point a distance x from the input end
of a line and VA and fA are the input-end voltage and current. The quantities 'Y,
Zo are given by the expressions
Y= ~(R + jroL) (G + jroC)
and
_",lR+jroL
Zo- "G+jroc
where R, L are the series resistance and inductance per unit length and
G, C the shunt conductance and capacitance per unit length.
It can be seen from equation (11.1) that the input impedance to a line with
a short circuit at a distance x along it, when Vx would be zero, would be:
~o= ~A =~ tanh yx n
fA
The form of the variation of 40 with the distance x is shown in Fig. 11.4(a),
from which it can be seen that the input impedance becomes very large at a
6
kn2
t
Zin 0
-2
-4 (a)
500
400
300
0
t 200
lin
100 200 ........ 400- 0

1m........ ~
-100 ........ ....
....
-200 (b)
Fig. 11.4 (a) Typical input impedance of a line with a short circuit fault at a distance
x. (b) Variation of input impedance with fault resistance.
particular distance which is dependent on frequency. For the frequencies used
in power systems, however, these distances are great, for examples 1500 km
for 50 Hz operation and 1250 km for 60 Hz operation.
In practice, therefore, the positions of short circuits on all the lines in use
today could be determined by impedance measurement and the non-linearity
of the input impedance variation with fault position does not pose problems.
Should resistl!.!lce (Rr) be present in the path of a fault on a long line then
the value of Vx/fx would be Rf an~ it can be shown from equations (11.1) and
(11.2) that the input impedance (Zin) would then be given by:
Zin= ~A =Rr cosh yx+ Zo sinh 'Ix

fA cosh 'Ix + ~r sinh 'Ix
Zo
The variation of the value of the input impedance with fault resistance is
not as simple as that which was shown in the previous section, in which
relatively short lines were considered. The effect is similar, however, as can
be seen from Fig. 11.4(b), which shows the variation which would be obtained
for a fault at a point 300 km along a single-phase line with parameters related
to those of the 400 kV, three-phase lines in Britain. The same variation would
clearly be obtained if current was supplied to a resistive-load connected to the
end of a 300 km-Iong line with the same parameters but in this case the ohmic
value of the load resistance would be relatively high.
Clearly the effect produced by the presence of resistance in faults close to
the input end of a line is the same as that obtained on short lines.
The above considerations demonstrate that correct discrimination could be
achieved between faults on protected lines and other conditions by detecting
input impedances within certain areas in the Z plane.
11.2.3 The input impedances of a three-phase line

All major transmission and distribution lines are three-phase and therefore
several different types of faults may occur either on them or on other associ-
ated circuits, namely single-phase to earth, phase-to-phase, phase-to-phase to
earth and, more rarely, three-phase. Clearly the positions of all these types of
faults cannot be determined by a single impedance measurement and therefore
a protective scheme must be energized with voltages and currents proportional
to those associated with the phases of the line being protected. As examples,
faults between phase a and earth would be detected by monitoring a current
and voltage proportional to the 'a' phase current and the phase a to earth
voltage whilst faults between phases a and b would be detected by monitoring
a current proportional to either the current in the 'a' phase or the 'b' phase
and a voltage proportional to the voltage between phases a and b.
It will be evident from the above that distance-protection schemes to be used
Impedance measurement 393
with three-phase lines require that a large number of impedance measurements
be performed.
11.3 IMPEDANCE MEASUREMENT

It must be accepted that impedance-measuring relays or other devices cannot
be produced with the accuracy needed to enable them to discriminate between
faults on an overhead line and all faults elsewhere. As an example, a relay
could not be expected to differentiate between a fault one metre from the
remote end of a line 100 kIn long and a fault on the busbars to which the
remote end is connected as this would represent a difference of only 0.001 %
of the setting.
As a consequence, impedance-measuring schemes do not inherently provide
unit protection and extra features must be incorporated in them to enable correct
discrimination to be achieved in service. Before considering these features,
however, methods available for measuring impedance are examined below.
11.3.1 Relays dependent on the magnitude of impedance

Two different types of electro-mechanical relays were used in early protective
schemes. One of them operated in a time related to the magnitude of the ratio
of the voltage to the current applied to it whilst the other operated rapidly
whenever the magnitude of the ratio fell below a set level. These two basic
forms are considered below.
Relays which operated in times related to impedance

The earliest impedance-measuring relays were intended to employ time grad-
ing similar to that of the IDMT relays in use at the time and they were based
on the induction elements then available. A particular design is illustrated in
Fig. 11.5(a), from which it can be seen that a current proportional to that in
the circuit being protected was fed to the winding on the driving magnet of
the induction element. When the current exceeded a set value the disc rotated
at a speed dependent on the current and wound up the spiral spring attached
to it. The other end of the spiral spring was attached to the armature of an
electromagnet which was energized with a voltage proportional to that of the
protected line. The greater the level of this voltage, the greater was the spiral
spring torque needed to move the armature and operate the contacts. This
required the disc, which was braked by a permanent magnet, to travel a greater
distance. As a result, the operating time increased with increase of voltage and
decreased with increase of current and therefore it increased with the ratio of
voltage to current, i.e. impedance. Linearity was not achieved however, the
actual characteristics being of the form shown in Fig. 11.5(b), which also
shows how discrimination was obtained between circuits in series.
Armature
L_~===- To trip circuit
(a)
(b)
Fig. 11.5 Principle and application of time distance relay.
Relays which operated rapidly when impedances were below set levels
The above relays did not provide rapid clearance of faults on protected lines
and the operating times clearly increased with the distance of a fault from the
relaying position. This was not always acceptable because high fault currents
could persist for quite significant periods and therefore relays were produced
which operated rapidly for all faults which caused the magnitudes of the input
impedances to protected circuits to be below appropriate set levels. In early
distance-protection schemes, relays with a pair of electromagnets above which
was a balanced beam were used. The basic form of such relays is shown in
Fig. 11.6(a).
The beam, because of its inertia, could not respond to the power-frequency
variations of the fields set up by the two electromagnets. Under normal con-
ditions the voltage applied to the winding on the left electromagnet was high
and a current flowed in the winding producing a large m.m.f. (/yNI ) and
therefore a large flux <PI' The current in the protected line at such times would
be relatively low and therefore both the m.m.f. (IN2 ) in the right electromagnet
and the flux <pz would be low. During these conditions the beam was restrained
by the backstop.
"-
\
\
\
\
\ Operation I R
\
\ within "
\ circle I
(a) " "- .... ..- /

/
(b)
x
I
,f
\
\
\
Beam relay \ " /
"- .... ..- /
(c)
Fig. 11.6 Balanced-beam relay.
During periods when faults were present on the line, the voltage (V) would
be depressed and the current (I) would be above normal and as a result the
magnitude of the flux CP2 would exceed that of the flux CPh causing the beam to
tilt and close the contacts. These relays were therefore amplitude comparators.
Because of the air gaps in the magnetic circuits, almost linear behaviour was
obtained and operation occurred when the ratio of the magnitude of the m.m.f.
IN2 to that of m.m.f. Iv!V.J !.xceeded a set value. As a result operation occurred
when the magnitude of I IV exceeded a set value. This provided an impedance
setting, independent of phase, below which operation occurred.
These relays therefore had a circular impedance characteristic centred on
the origin of the Z plane, as shown in Fig. 11.6(b) and not offset characteristics
of the forms shown earlier in Fig. 11.2. They did not therefore have directional
properties and had to be used in conjunction with separate directional elements
so that the combination operated only in semicircular zones of the form shown
in Fig. 11.6(c).
11.3.2 Relay characteristics

Relays with circular operating zones centred on the origin of the Z plane are
known as plain-impedance relays. As was shown above, this characteristic is
obtained by comparing the magnitudes of the applied voltage and current and
operation is obtained when the ratio of the two quantities falls below a set
value. Over the years a number of methods have been developed to effect such
comparisons and the devices used for this purpose have been designated
'amplitude-comparators' .
Because relays with plain-impedance characteristics could not be used on
their own in distance protection schemes, relays with other characteristics
were produced. One of the first of these was the 'mho' characteristic, which
was first obtained using beam relays as described below.
Mho characteristic obtained using an amplitude comparator

As shown in Fig. 11.7(a), the restraining electromagnet had two windings. One
of them, which had NI turns, was supplied via an impedor to make the total
impedance of the circuit have a set value (ZI l!J.
This is referred to as the
replica impedance. The voltage applied to the circuit (k l V ~ was propor-
tional to that of the protected circuit, i.e. V~. The second winding on this
electromagnet had N2 turns which carried a current (k2/), I being the current
in the protected circuit. The current k21 also flowed in the single winding of
N3 turns on the operating electromagnet.
For a relay in which operation would occur when the m.m.f. applied to the
operating electromagnet exceeded that applied to the restraining electromag-
net the following relationship would exist:
k lIN
2 3
I ~ I klNI V~ - k2 IN2 I
ZIl1...
( 11.3)
;;;;
BackstoR::=i::::::::J:!::;==:::::;r:==::r-~g C,zrtact x
-
/
I \
/ \
I \
~I
, b
\
--,--~\
,
\ I
\ /
\ I
" a // R
kNZ ..... - - - - /
Restraining Operating oa = -LL!. cos ~
electromagnet electromagnet k,N,
k2N2Z, . R
ob =--Sln",
k,N,
k2NaZ,
r =--
k,N,
(a) (b)
Fig. 11.7 Mho characteristic.

The limiting condition, expressed in complex form, is thus:
Ik2IN31 = I kl ZlVNl cos(a-p)-k2IN2+jklZlVNl sin(a-p) I

Substituting R and X, the input resistance and input inductive-reactance of the
protected circuit respectively, for the quantities V cos all and V sin a/I in the
above expression leads to the following equation:
(11.4)
This is the circle of radius
and centre
shown in Fig. l1.7(b).

It can be seen from the above expressions that the circle can be made to
pass through the origin by making the turns of the two current windings (N2
and N3) the same. Relays with this characteristic are described as 'mho' relays
whereas the term 'offset mho' is used for other relays with circular charac-
teristics in the Z plane. It is worthy of note that the term 'mho' was adopted
because the circular characteristic in the Z plane becomes linear when it is
represented in the Y (admittance, i.e. mho) plane.
It can also be seen from Fig. 11.7(b) that the position of the circle with
respect to the axes is dependent on the phase angle of the impedance of the
voltage circuit, i.e. ~, and this can be adjusted to provide the characteristics
needed by the circuits being protected.
Although this characteristic was produced using beam relays it can clearly
be produced by other forms of relays including those which employ electronic
circuitry, in which the amplitudes of the quantities on the two sides of equation
(11.3) are compared. Mho characteristics can also be produced by relays
incorporating digital circuitry and in such cases direct amplitude comparison
is not necessary.
Ohm characteristic obtained using an amplitude comparator

Beam relays were also used to produce the so-called 'ohm' characteristic,
which is linear in the Z (impedance, i.e. ohm) plane, by having a winding of
Nl turns on each electromagnet, these being connected in series and fed
through an impedor to make their total impedance have a value of ZI ~ as
before. This circuit was energized by a voltage (k l V~ proportional to that
of the protected circuit. Each electromagnet also had a winding which carried
a current (k 2 /), I being the current in the protected circuit. The windings on
the restraining and operating electromagnets had N2 and N3 turns respectively
and the current flowed through them in opposite directions.
Operation thus occurred when the following condition applied:
V~+k IN I ~ I klNI
I klNIZI LIL Zd~
2
V~_k IN I
3 2 2
(11.5)
By again substituting R = V cos oJI and X = V sin 0./1, the following express-
ion is obtained:
R ZI k2 (N2 - N3)
X = - R cot I-' + ---'---=--'-----=----"..:... (11.6)
2 kl NI sin~
The resulting operating characteristic of these relays is therefore of the
general form shown in Fig. 11.8(a).
It will be clear that the characteristic would pass through the origin, as
shown in Fig. 11.8(b), if the current windings had the same number of turns,
i.e. N2 =N3, and such a relay would operate for currents varying in phase over
7t rad relative to the applied voltage, i.e. it would behave as a directional relay.
It will also be clear that a horizontal ohm characteristic could be produced
by making the phase angle ~ equal to 7t/2 rad. A relay with such a charac-
teristic could be set to operate when the inductive reactance component of the
input impedance of a line fell below a particular value and it would therefore
not be affected by the presence of resistance in a fault path. Relays with this
characteristic, which were described as reactance relays, were produced using
induction type elements, and were employed to detect earth faults in early
protective schemes.
x x
R R
OA = Z1 k2(N2 - N3)
2k1N1 sin ~
~>N3
(a) (b)
Fig. 11.8 Ohm characteristic.
The above characteristics can be produced by other forms of relays in which
the amplitudes of the quantities on the two sides of equation (11.5) are compared.
Characteristics obtained using phase comparators

Phase comparators, as their name implies, are relays or devices which are
supplied with two quantities, for example a voltage and a current, and they
operate whenever the phase displacement between the quantities is in a par-
ticular range. As an example, a device could be set to operate when the two
quantities are within rr./2 rad of each other. A directional relay is thus one form
of phase comparator.
The characteristics considered on page 396 may be obtained using a phase
comparator supplied with two quantities produced by summing and taking the
differences of voltages and currents proportional to those associated with a
protected line.
As an example, if the following quantities are produced:
and
and operation occurs when the phase angle between the quantities is rr.!2 rad
or larger, either lagging or leading, then the limiting condition would occur
when:
Ivl =W
lIT ki
=constant
i.e. such a relay would have a plain-impedance characteristic.
As a further example, a 'mho' characteristic can be produced by a relay
which operates when the following two quantities are displaced in phase by
rr./2 rad or less:
- --
SI=kr/+kvV and S2=~/-kvV
The resulting circle has a radius of

~+kr
21 kv I
and a centre at
~-kr e ~-kr . e
-1-1
2kv cos '-I-I
2kv sm
in which -9 is the phase angle of 1<... This is an offset-mho characteristic unless

kr is zero, in which case the circle passes through the origin of the Z plane.
Ohm relays and directional relays can also be produced using phase com-
parators.
Quadrilateral and other characteristics
To reduce the possibility of incorrect operation of impedance-measuring re-
lays it is desirable that the areas of the zones in the Z plane in which they will
operate are limited to those needed to detect faults on the circuits being
protected. It was shown earlier in section 11.2.1 that faults up to a certain
resistance can be detected using the characteristic shown in Fig. 11.2, i.e. a
parallelogram.
Such characteristics cannot be produced simply by a relay employing either
an amplitude or phase comparator. As these relays can, however, have linear
characteristics, i.e. an ohm characteristic, then four of them with suitable
characteristics could be used to produce an overall quadrilateral characteristic.
This would, however, be expensive.
Over the years combinations of relays have been used to produce desired
characteristics. As stated earlier, plain-impedance relays have been used with
directional (ohm) relays to obtain the semicircular characteristic shown in Fig.
11.6(c). Similarly, an ohm relay could be used with a mho relay to produce
the operating characteristic shown in Fig. 11.9.
Impedance relays have been produced in several different forms, e.g. beam
relays, induction relays and rectifier comparators [5]. The range of charac-
teristics which could be produced by such relays, was limited but the intro-
duction of electronic digital processing in recent times has made it possible
for any desired characteristic to be provided including quadrilaterals and those
with lenticular shapes with a range of aspect ratios.
11.3.3 The effects of system transients

The operating characteristics considered in section 11.3.2 were all presented
in the complex Z plane which is related to sinusoidal steady state conditions.
In practice, however, relays must operate relatively quickly after the incidence
Fig. 11.9 A combined mho and ohm characteristic.

of faults, say in 20 ms or less, and steady state conditions will not usually have
been obtained in such times and indeed r.m.s. values determined over fractions
of a power frequency cycle are not those used in phasor diagrams. In addition,
quite significant unidirectional transient components may be present in the
voltages and currents at the ends of faulted lines.
To ensure that large errors will not occur in the assessments of the distances
to faults, relays must be so designed that the effects of the above factors will
be either eliminated or minimized.
As an illustration of the way in which this can be done, the behaviour of a
'mho' element incorporating a beam relay as an amplitude comparator is
examined below.
Behaviour obtained during transient conditions resulting from/aults on short

lines
To simplify the treatment, the application of the relay to a relatively short,
single-phase line is considered. The capacitive current between the conductors
of such a line would be very low and therefore to detect short circuits up to a
distance C along a line, which has a resistance R per unit length and an
inductive reactance (roL) per unit length, the diameter of the mho charac-
teristic would be made equal to C I(R +jroL) I in steady state.
It was shown on page 397 that the diameter of the mho circle is given by:
and therefore to provide the necessary setting the magnitude of the impedance
of the voltage circuit of the relay must be:
(11. 7)
If the impedance ZI is made up of a resistance
R1=k1N1CR
2k2 N3
in series with an inductance
then the transient and sinusoidal components of the current in the voltage
winding of the relay will match those in the current windings at all instants.
As an example, should the voltage at the input to a line remain sinusoidal
during a short circuit condition, i.e. v = Vpk sin (rot + a) in which t is measured
from the time of fault occurrence, then the instantaneous line current would
be given by:
i= Vpk ] {sin(oot+O;-~)-Sin(o;-~)exp(-!it)}
f(R 2 + 002L2) Z L
in which tan ~ = ooL/R

The current (iv) in the relay voltage winding would be given by:
iv = 2k2 N3 Vpk ] {Sin (oot + 0; - ~) - sin (0; - ~) exp (_!i t)}

N](R2+oo2L2)z L
and the quantities on both sides of equation (11.3) would be equal at all
instants.
As a result the m.mJ.s applied to the two electromagnets would be the same.
The impedance Z] given in equation (11.7) is thus a replica of that in the line
and its use would ensure that in practice only the small errors caused by
neglecting the capacitive current of the line would occur during transient
conditions. In addition the behaviour would not be affected by the relay
operating time and operation would not therefore need to be delayed to allow
r.m.s. values to be determined.
It will be clear that errors will occur, however, in the assessments of the
distances to short circuits if the phase angle of the impedance Z] is not the
same as that of the line impedance.
It can be shown that accurate assessments of the distances to short circuits
can be obtained with relays having the other characteristics considered earlier
and this is so when either amplitude or phase comparators are used.
It must be recognized that significant errors may be made in determining
the input impedances of lines on which resistive faults are present because
under such conditions the ratio of the inductance to resistance in the replica
impedor (Z]) will not be the same as that in the line input impedance. In these
circumstances, the waveforms of the currents in the voltage circuits of the
comparators will not be the same as those in the current circuits when transient
components are present and certainly those in the current circuits will decay
more rapidly than those in the voltage circuits. As a result, the values being
compared will not be in a constant ratio and the phase differences will not be
constant.
Behaviour during transient conditions which result from faults on long lines
The capacitive current between the conductors of long lines tends to be small
relative to the normal or fault currents in lines, but nevertheless the input
impedances when short circuits occur on lines do not increase linearly with
the distances to faults as shown in section 11.2.2. In addition the phase angles
of the impedances are affected by the distances to faults.
Basic schemes 403
Fig. 11.10 T-section line model connected to a relay.
Clearly, a relay could be set to detect when the input impedance of a line
fell below a certain value during steady state conditions, but the technique of
matching the impedance in the voltage circuit of a relay to that of the protected
line does not enable completely accurate assessment of the positions of faults
to be achieved. This is basically because the distributed nature ofthe impedors
in a long line cannot be matched by a single impedor.
The behaviour of a line would be approximately mirrored by making allow-
ance for its capacitance by connecting a replica capacitor in the voltage circuit
of a relay as shown in Fig. 11.10. In this way a T-section modelling of the
line up to the distance at which short circuits should be detected would be
produced. This would provide improved accuracy of measurement but it would
not be completely accurate because the current (Ir) in the relay winding would
not contain the components caused by the travelling waves which are present
under fault conditions.
In practice a degree of error in the assessment of the distances to short
circuits or long lines must be accepted [6] and errors will again be present
when resistive faults occur.
11.4 BASIC SCHEMES

It is desirable that faults on major transmission and distribution lines should
be cleared rapidly and therefore relays which operate in times proportional to
the distances of faults along lines are not acceptable and relays which operate
in short times independent of the fault positions are always used. Such relays
cannot be set so that they will detect faults anywhere on a protected line and
yet not operate for faults elsewhere on a network. They must therefore be set
to operate when faults occur on most of the length of a line, a percentage of
80% being usual. Such a setting provides a factor of safety to ensure that
operation for faults beyond the end of a line will not be caused by measuring
errors. In addition to this setting, which is designated Stage 1 or Zone 1 reach,
another setting which can detect faults both on the protected line and on
circuits beyond the remote end of the line must be provided and operations
within this setting, which is designated Stage 2 or Zone 2 reach, must initiate
circuit-breaker tripping after a short time delay (t2)'
To ensure the isolation of faults on a protected line, similar sets of relays
must be provided at each of its ends, the arrangements being as shown in
single-phase form in Fig. 11.11. It will be clear that such a scheme would
ensure that faults in the centre 60% of the line, i.e. 0.2t' ~ O.se would cause
the opening of the circuit-breakers at both ends to be initiated after a time tl
say 20 ms. Faults in the 20% end sections would cause the opening of the
nearer and more remote circuit-breakers to be initiated after times of tl and
t2 respectively. To ensure that circuit-breaker opening will not be initiated in
the event of faults on the other circuits connected to the bus bars at the line
ends, the operating time t2 must be greater than that of the protective equip-
ments associated with the other circuits.
In some applications, a further setting (Stage 3 or Zone 3) is included, this
being greater than the Stage 2 setting, and in this way faults which are more
remote than those in the zones covered by the Stage 2 setting can be detected.
Stage 3 operation occurs after a longer time delay than that associated with
the Stage 2 setting and therefore it only initiates the opening of the circuit-
breakers if faults on connected circuits are not cleared by their own protective
schemes. This setting thus provides a form of back-up protection.
As stated above, a fault near one end of a protected line would not be
completely cleared until after the operation of the Stage 2 relays at the remote
Dst
EndA D
--
st D 1
-
1
EndB
_ _ xl,v-st2----s-t2SJ=rX-
_A Protected line A
st2, End A
------------- -r----
"-"-l"-'~t2, 'End '6-"
Cl
c: i
i
.~
~
o
t1
i
st1, End AL.
st1, End B
0.2/ 0.8/
Fig. 11.11 Application of relays with Zone 1 and Zone 2 settings.

Basic schemes 405
end, i.e. after a time t 2 In some applications, such relatively slow clearance
of faults might be considered unacceptable. In these circumstances, the trip-
ping of the remote circuit-breaker can be accelerated by sending inter-tripping
signals between the line ends as described later in section 11.12. The resultant
scheme is then clearly a form of interlock protection incorporating impedance-
measuring relays.
11.4.1 Application to three-phase lines

To ensure that the various types of faults which may occur on three-phase
networks will be correctly detected, separate relays could be used on each of
the phases to determine whether faults to earth are within the protected zones,
i.e. Stage 1, 2 and possibly 3. Separate relays could also be provided to detect
interphase faults. Clearly one or more of the above relays would operate in
the unlikely event of a three-phase fault.
It will be seen from the above that 18 impedance-measuring elements would
be needed at each line end in such an arrangement to provide three impedance
zones and certainly when electro-mechanical relays were used the costs were
significant and the burdens placed on the current and voltage transformers
supplying the relays were high. This latter factor in turn made for greater
expense because of the need to use larger cores and/or windings, particularly
in the current transformers, to ensure that they would perform satisfactorily
under all the conditions which could be encountered.
At the other extreme, schemes having only a single impedance-measuring
element may be produced by including simple elements to sense the types of
fault present on lines and using these to provide the necessary quantities to
the impedance-measuring element. As an example, simple overcurrent starting
elements may be used to detect interphase faults and in the event of a fault
between, say, phases a and b, the appropriate overcurrent elements would
operate and connect an input voltage proportional to that between the 'a' and
'b' phases of the line and an input current proportional to either the 'a' or 'b'
phase line current to the impedance-measuring element. Should this element
not operate within a certain time (tl) because the fault was not within its Stage
1 reach, then its setting would be increased to that of Stage 2 and this p{oce-
dure could, if desired, be repeated to extend the reach to that of Stage 3 after
a further period.
Over the years a variety of schemes ranging between these two extremes
have been produced. Clearly the advantage obtained by including starting
elements is the reduction in the number of impedance-measuring elements
needed. The disadvantage of these schemes is the element of unreliability
introduced by the inclusion of switches in the circuits supplying the measuring
relays and certainly some authorities have considered the risk to be unaccept-
able. Nevertheless such schemes, which are designated as switched-distance
schemes, have been used in many applications and have given good service.
11.4.2 The detection of faults close to the input ends of lines
Various relay characteristics which may be produced in the Z plane were
considered in section 11.3 and it was shown in some cases that the charac-
teristics could pass through the origin. Such behaviour implies a limiting
condition at zero impedance and thus zero voltage.
In the case of a semi-circular characteristic, shown in Fig. 11.6(c), being
produced by combining a directional element with a plain-impedance element,
it will be evident that the direction of a current cannot be determined when
the reference voltage being used is zero. Such a relay could not therefore be
expected to discriminate between faults close to the current and voltage trans-
formers on a protected line and those on the adjacent busbars or near the ends
of other circuits connected to the busbars. This situation was considered earlier
in section 4.5 (page 135) when the behaviour of directional relays was exam-
ined, and it was then shown that satisfactory operation could be obtained by
energizing relays with voltages derived from phases other than those from
which the currents were obtained, e.g. a directional relay associated with phase
a to earth faults could be supplied with a current proportional to that flowing
in phase a of the protected circuit and a voltage proportional to that between
phases b and c. This technique could also be used successfully with the relay
considered above. To correctly assess the input impedance during an earth
fault on phase a of a line, the plain-impedance element would be supplied with
a voltage and a current proportional to those of phase a whilst the directional
element would be supplied with quantities proportional to the phase a current
and the 'b' to 'c' phase voltage of the line.
The technique cannot, however, be applied directly to relays with 'mho'
characteristics which have only two inputs. In such applications if a current
from one phase and the voltage between the other two phases were used then
the limiting conditions would not be those of the required circle. To produce
an acceptable compromise, three inputs must be provided as in the previous
example, e.g., the 'a' phase voltage (va) and current (ia) and the voltage
between the other phases (Vbc).
The relay would combine the 'mho' and 'ohm' characteristics and to do so,
in the case of an amplitude comparator, an equation containing the terms of
both equations (11.3) and (11.5) would define the conditions needed for
operation, i.e.
(11.8)
Such a relay does not have a single operating characteristic in the Z-plane
because it is controlled by four variable terms, i.e. Valla, Vab/la, 'Yand a., unlike
a mho relay which is ,only controlled by two of these terms, namely Valla
and 'Y.
Basic schemes 407
It can be seen, however, that in the event of a short circuit on the 'a' phase
near the current and voltage transformers, for which the voltage Va would be
zero the equation (11.8) would reduce to that of an ohm relay, i.e. equation
(11.5), and it would pass through the origin of the Z plane. It would thus act
as a directional relay and the angle of the linear characteristic could be set at
the required angle in the Z plane by choosing an impedor Z3 with the necessary
phase angle.
For short circuits on phase a of a line near the point beyond which operation
is not to occur, the voltage Va would be quite high and provided that the
magnitude of the term k3N3/Z3 is very much less than that of the term
k\NtfZh the effect produced by the polarizing voltage Vbc would be small
compared with that produced by the phase a voltage (Va). As a result the
behaviour would approach that of an mho relay. Such relays are referred to as
being cross-polarized and clearly the same performance can be obtained by
providing the necessary inputs to other forms of amplitude or phase comparators.
It will be evident that neither self- or cross-polarized relays would operate
correctly in the unlikely, but nevertheless possible, event of a three-phase
short circuit on a line at a point near the current and voltage transformers
because under such a condition all the voltages would collapse to near zero levels.
In the past, so-called memory circuits were connected at the outputs of the
voltage transformers. These were oscillatory circuits which would continue to
provide power-frequency voltages that would decay in amplitude slowly
should the power-system voltages collapse. These voltages which were related
to the pre-fault system voltages were then used to provide polarization to the
various relays.
More recently, with the introduction of modern electronic equipment, volt-
ages have been continuously sampled and stored for short periods and these
have been used to provide a measure of polarization in the event of system faults.
It will be appreciated that the basic circular mho characteristic is distorted
by the use of cross-polarization and clearly the degree of distortion increases
as the amount of cross-polarization is increased. This feature is utilized on
occasions and the amount of cross-polarizing is so set that the circle is dis-
torted to increase the value of the intercept on the R axis of the Z plane
diagram whilst retaining the cut-off impedance value at the characteristic
angle ~. In this way, relay operation for relatively high-resistance faults at
points along a protected line can be obtained.
To differentiate between the types of mho characteristics, the term 'self-
polarized' mho is now used for relays with a single voltage input, the term
'cross-polarized' being used for those with two voltage inputs.
11.4.3 The settings of impedance-measuring relays

The setting of relays used in three-phase applications must take account of the
current distributions which can exist during the various types of faults which
may occur on a line. Those which arise during phase to earth faults and
interphase faults are considered separately below.
Phase to earth/aults
If a line had a completely symmetrical conductor configuration and it was
supplied from an earthed source, as shown in Fig. 11.12(a), and there were no
other earths on the system, then the conditions during a short circuit between
phase a and earth could be determined from the sequence network shown in
Fig. 11.12(b). This is demonstrated in Appendix 3. The fault current would be
given by:
- - - - 3 Via
la =Iia + lu + loa =----==-----==---,=-
X (ZI + Z2 + Zo)
and therefore the apparent input impedance to phase a (Ze) would be given by:
Z = ~Ia =x ZI +Zz+Zo (11.9)

e la 3
in which Zo, ZI and Z2 are the sequence impedances up to the fault position.
Each of the earth fault impedance-measuring relays could be set to operate
at impedances up to the value given by the above expression for a fault at the
desired cut-off or reach, say 80% of the line length.
In practice, lines are not symmetrically spaced in triangular configurations
and therefore the impedances per unit length of the phases differ and measur-
ing errors would be present if the above setting was used on all phases. The
-
- Line impedances/unit length
- r-- are 2'1' ~ and Zo V1a
la
--"+-
x_ a
X
-
x b
~ X---+--c ~a=T2a=foa
It =1a
(a) (b)
Fig. 11.12 A phase-to-earth fault and the associated sequence network.

Basic schemes 409
imbalance is usually small, however, and the resulting errors tend to be ac-
ceptable. _
It will be realized that the current (/a) would flow in the faulted phase during
the above condition and it would all return through the earth. In multiply-
earthed systems, however, different current distributions can occur and currents
can flow in the healthy phases. This can be seen by considering the circuit
shown in Fig. 11.13(a) in which a line is connected between two earthed
sources. The sequence network would then be connected as shown in Fig.
11.13(b) to determine the conditions which would obtain during a short circuit
to earth_on phase a of the line. It will be clear that the positive.:sequence
current I Ala wou~ not be_the same as the zero-sequence current I Aoa if the
source voltages VAla and VBla were not of the same magnitudes and phases,
and under these conditions currents would_flow in phases band c of the line.
As a resul~ the ~pl!!ent input impedance (ZAe) to phase a at end A of the line,
given by ZAe =VAa/I Aa , would not be equal to the value given by equation
(11.9), i.e.
A relay set to operate up to this latter value would thus assess the distance
to the fault incorrectly and it could either under-reach or over-reach.
Sound-phase compensation To avoid the above situation, allowance for the

currents in the healthy phases can be made as explained below.
It can be seen from Fig. 11.13(b) that the input voltage to phase a and end
A of the line is given by:
- - --
VAa = VAal =X {lAlaZI + I A2a Z 2 + I Aoa Zo}
-
For a line Zl =Z2, therefore,
VAa = X {Zl (/Ala + I A2a ) + Zo I Aoa }
- - - -
=X {ZMAla + IA2a + I Aoa ) + I Aoa (Zo - Zl)}
The distance to the fault (x) on phase a is thus given by:
A method based on the above equation was employed in Ratio-Balance

protective schemes [6] produced by A. ReyroUe and Co Ltd, a sound-phase
compensation transformer of the form shown in Fig. 11. 13(c) being included
for the purpose. It will be seen that the phase output currents ofthe transformer
were of the form:
in which
End A Line impedances/unit length End B

VAa VAb VAc are~, ~ and Zo VBa VBb VBc
I
~~- X -
X
X
-
-
- -
VA1a (a)
XZ2 ~a xZo ~oa
(b)
(I-x)Zo -leoa
- -
101!!4 sNout f 'out a
""--./YY'>t"'--- - J "........" ' - - _ ..J'"YlnI"",-" ~ ~~~~~~~~ ~
- -
101!!4. sNout flout b
n.../YY'lr...... U--r>nC'lr"'- L...L.r'X'X'"J(~ ~ ~
-
lo~
L.-rntY'l...n~
N3
I ~~1 I
(c) (d)
Fig. 11.13 Current flow during earth faults in multiply-earthed systems.

Basic schemes 411
W.!!ilst ac~ount should strictly be taken of the phase angles of the impedan-
ces ZI and Zo, this form of compensation does enable earth fault relays to
perform with acceptable accuracy because the line resistances are much lower
than the reactances and the assessments of the positions of faults are not
therefore significantly affected by the distributions of the currents in the
phases. _ _
To give an output current of I out = X I A (ZI + Z2 + Zo) when there are no
currents in the healthy phases the turns ratio N31NI must be unity.
Residual compensati~n _ The ~quation above relating the voltage VA. to the
three-phase currents I., Ib and Ie may also be expressed in the form:
An alternative form of compensation, known as residual compensation, c~

thus .!?e provided and if the phase displacements between the impedances ZI
and Zo are neglected then the compensation can be obtained using a trans-
former connected as shown in Fig. 11.13(d). The turns ratio must be
N2 INI i
= I Zo ZI l
I
Interphase faults
In the event of a short circuit between two phases of a line, as shown in Fig.
11.14(a), the sequence network would be connected as shown in Fig. 11.14(b).
The magnitude of the fault current would be given by:
lId = -..51 VI. I

x IZI +z21
and the magnitude of the apparent input impedance (Zbe) given by
IVbel
IIbl
would be:
~ ~=b
7 \j :
Line impedances/unit length
are 21 and 22
(a)
Fig. 11.14 An interphase fault and associated sequence network.
In practice, the positive and negative sequence impedances of lines are equal
z,
and therefore IZbc I = 2x I I.
During a three-phase fault the fault currents are given by:
VIa
h=---=-
xZ,
The magnitude of the apparent input impedance Zbc in these circumstances
would be:
It can be seen from this that phase-fault relays would operate for three-phase
faults at distances up to 2/...[3 times those for which they would operate for
phase to phase faults. As a result, if Stage 1 phase-fault relays were set to have
a reach of 80% of the length of a protected line in the event of a three-phase
fault, then they would only detect phase to phase faults on 69% of the line
length.
Basic schemes 413
This situation may be avoided by supplying each of the phase-fault relays
with a line voltage and the difference of two phase currents. Considering again
a short circui!....between phases b and c, the magnitude of the apparent input
impedance ( IZbe I) would be determined from the expression:
12bel == IVbel
lIb-lei
For this fault
- - -
Ib==h=-/e
During a three-phase fault, the fault currents are given by:
I == VIa. Ib== VIa 1_ 2n I == VIa 1_ 4n

a 21 ' 21 3' e 21 3
and
For this condition the magnitude of the apparent input impedance (I 2be I)
would be:
As ZI = (ZI + Z2)/2, the relays would operate for all interphase faults up to the
same distance from the relays and therefore this method of measurement is
now used in most schemes.
The effects of the physical asymmetry of lines

The settings considered in the previous sections have been based on complete
line symmetry, whereas in practice conductors are run in asymmetrical con-
figurations and therefore the self and mutual impedances associated with the
phases are not the same. Whilst conductors could be transposed to reduce the
effects of their physical asymmetry, this practice is now seldom adopted and
certainly it would be difficult to implement on major lines because of the large
spacings which would have to be maintained at the points where transpositions
were made.
Clearly errors will be present in the assessments of fault positions if the
impedance-measuring relays in each of the phases of a scheme applied to an
untransposed line are given the same settings and, of course, these could be
reduced by employing different settings.
In a study [6] on the behaviour of an un transposed line it was shown that
the apparent impedances seen by relays during a three-phase fault at a point
500 miles from the relaying position were affected by the zero-sequence
source impedance, variations up to 10% being possible. In addition, the im-
pedances seen on the individual phases could differ by about 20%.
In general, the above effects are insignificant when schemes are being
applied to relatively short lines but they should be taken into account when
the protection of very long lines is being considered.
The effects of mutual coupling between lines

When overhead lines run close to each other, there is mutual inductance and
capacitance between them and therefore they do not operate independently of
each other. The effects of coupling can be quite significant on pairs of lines
which are carried on double-circuit towers because of the close proximities of
the various conductors.
The mutual inductances between pairs of conductors in the two lines are not
all the same and therefore e.m.f.s are produced in the conductors of one line
by the currents in the conductors of the other line. Although the effect is
greatest when zero-sequence currents are flowing in a line there is nevertheless
a small effect when the line currents are balanced. As a result, the apparent
input impedances of a line on which a fault of any type is present are affected
by load currents or currents being fed to the fault by a parallel healthy line
and clearly the operation of impedance-measuring relays can be affected. This
may be seen by considering the situation which would arise in the event of a
__~~L~in~e_P_________ x
?O...,
-
~
e
~ Ip
X
~X To Ip+/e
Source A
_-:;.,--=L:::,in::::;e.;:Q:...-_.....;-;....._X
_ Is Source B
End A .. x Jjault End B
is = self impedance of lines/unit length

Zm = mutual impedance of lines/unit length
VA = loZsx+ IpZmx
VA = 7p [z,,(2/-x)-2Zm(I-x))
+ To Zm x + le(Zs-Zm)(/-x)
Fig. 11.15 Effects of mutual coupling on line input impedance.

Basic schemes 415
fault on one of a pair of single-phase lines connected in parallel, as shown in
Fig. 11.15. _
For this condition, the lines would feed the currents Ip IQ and 18 to the short
circuit a dist~ce x fr~m end A along line Q and the apparent input impedances
to the lines, ZioP and ZinQ, at end A would be:
-
ZinQ = I: ( I: )
-V -
= Zs +
-1-
Zm x (11.1 0)
and
- if - - IQ- 1 - -
ZioP = _A =Zs (U -x) - 2Zm(t' -x) +=- ZmX+ _8 (Zs - Zm)(t' -x) (11.11)
~ ~ ~
It will be seen that the input impedance of the faulted line (ZinQ) js_not
directly proportional to the distance to the fault (x) because the ratio Ip/IQ is
affected by this distance. It is clear, however, that the correct distance to the
fault could nevertheless be determined from equation (11.10) if the ratio of
the two currents was known and in practice this could be achieved by feeding
the relays with the currents from both lines~
The input impedance to the he~lth}'. lin~ (ZioP) is dependent on the distance
to the fault (x) and also the ratio IQ/ (/p + 18). This ratio varies in practice with
the operating conditions of the sources and as the current input at end B (/8)
cannot readily be provided as an extra input to the relays at end A, the distance
to the fault (x) cannot be determined by the relay on the healthy line when it
is only compensated with the current at end A of the_other line (IQ).
When an uncompensated relay, i.e. one in which Ip is taken to be zero in
equation (11.10), is present on a line on which a fault occurs, it will clearly
not determine fault positions correctly and the errors will increase with in-
crease in the value of the current (/p) in the healthy line. These errors will
cause the distance to the fault to appear greater than the actual value and
therefore the reach at any given setting will be less than the nominal cut-off
value unless the current in the healthy line (/p) is zero. The relay would
therefore never over-reach.
When a line is protected by a relay which is compensated only with the input
current from the other line, i.e. IQ in equation (11.11) then errors will be
present in the determinations of the distances to faults as stated earlier, but in
this case operation can occur during faults beyond the nominal cut-off position
and indeed Stage 1 operation could occur for faults beyond the remote end of
the protected line. Such over-reach is not acceptable and it is therefore much
more serious than the under-reach, referred to above, which can occur with
uncompensated relays.
The behaviour of both uncompensated and compensated relays applied to
three-phase, parallel-connected lines was examined in detail by Davison and
Wright [7] in 1963 and it was concluded that the normal arrangements used
on single-circuit lines should be used on double-circuit lines and that compen-
sation for the mutual effects should not be provided. The study showed that
over-reach could occur with certain faults but that maloperation would not
result whatever first-stage setting was adopted.
In the past, some schemes produced by manufactures in the USA did incor-
porate compensation dependent on the currents in parallel lines, but today
uncompensated relays are usually employed.
The micromho relay [8] produced by GEe does, however, include provision
for compensation derived from the zero-sequence current in the parallel cir-
cuit. It is occasionally applied to long double circuit lines but care must be
taken when choosing the settings.
11.4.4 The operating times of relays

Electro-mechanical relays, because of the inertias of their movements and the
displacements needed to close their contacts, have finite operating times and
these vary with the operating forces applied to them.
If a beam relay is considered, it will be clear that the torques applied to it
will be equal at the limits of its operating zone, i.e. the nett operating torque
will be zero on the limiting circle of a mho characteristic. As a result, the
operating times will be quite significant when faults occur on a line near the
current and voltage transformers and also when faults occur near the point
beyond which operation does not occur. When faults occur at intermediate
points, however, the operating torques will exceed the restraining torques by
significant amounts, and therefore more rapid operation will occur.
A further factor which must be taken into accoun~ when considering relay
operating times is the effective source impedance (Zs) at the input end of a
protected line. It can be seen from Fig. 11.1~(a), which shows a single-phase
line fed from another circuit of impedance Zs, that the current fed to a short
circuit a distance x along the line would be:
1=_ Vs_
Zx+Zs
The voltage (Vin) at the input end of the line would be
- -- Zx -
Vin =1 Zx ==--=-' Vs (11.12)
Zx+Zs
The impedance seen by an impedance-measuring relay would be given by:
and it would thus assess the fault position correctly but both the voltage and
current would decrease with increase in the effective source impedance. As a
Basic schemes 417
Vs ~
6< z.>-x~'" x'_~_X_._ _x I

t-- I IShort circuit ---t
Line impedance to fault = Z.
Total impedance of line = Zt.
(a)
100
Source impedance angle = 88 lag
Ui'
-0
c:
90 Line impedance angle = 70 lag
0
(J
80 Relay characteristic angle = 70 lag
til
70
~
60
I
Q) 50
E
." 40
Ol
c: 30
~ 20
~ 10
0
00 10 20 30 40 50 60 70 80 90 100
Fault position (% relay setting)
(b)
100
~============~==~~=--- ........ ,
" Relay boundary
Ci
c: 80 \characteristic
-=.,., \
,
\
~ I
~ 60
,,
~ I
c:
,g 40 Operating time curves ......4~~=====}_)
iii I
8. I
I
'3ca I
LL 20 I
I
I
/
/
0
0.1 0.2 0.3 0.5 2 3 4 5 10 20 30 4050 100
Source to line impedance ratio :~
(C)
Fig. 11.16 Effects of source impedance. (a) Network configuration. (b) Typical oper-
ating characteristics for Zone 1 phase-fault mho relay. (c) Isochronic time charac-
teristics for mho relay (three-phase faults).
105 Phase-earth faults
01
c:
=:CD 100
!Il ~
950
01
c:
=:CD
!Il 105 Phase-phase faults
CD
c:
0 100
N
:>!! 950
~
.s:::.
l;l
10't
~ Three phase and three
ph...-,.,,,, "'"Us
CD
0
c:
nI
"0
CD
a.
100~
.E 95 I I I I I I
o 20 40 60 80100120
Voltage
Fig. 11.16 (d) Typical reach to relay voltage relationship. (Reproduced from Protec-
tive Relays - Application Guide, 3rd edn, GEC Measurements, 1987 with the per-
mission of GEC Alsthom Protection and Control Ltd.)
consequence, the relay operating time would increase with source impedance.
It is therefore customary to present information on relay operating times either
~n graphs of the time variation with distance to a fault for several ratios of
ZS/ZL as shown in Fig. 11.16(b) or in the contour form shown in Fig. 11.16(c).
In addition to affecting the relay operating times, the reductions of the
voltages and currents applied to relays, caused by the presence of source
impedance, can also affect the distances up to which they can detect faults,
i.e. the cut-off or limiting distances may vary. Because account must be taken
of such variations, curves showing cut-off distance to relay voltage relation-
ships, of the type shown in Fig. 11.l6(d), are produced.
It will be appreciated from equation (11.12) that the effect of a given source
impeclance increases the shorter the length of a protected line, because the
ratio Zs/ZL increases and particular account must be taken of this factor when
impedance-measuring relays are to be applied to very short lines.
11.4.5 Relay performance standards

It will be evident from the preceding sections that an impedance relay cannot
be set to operate in the event of faults occurring at points up to a fixed distance
along a protected line and it must be accepted that the cut-off distance or reach
will be affected by the system operating conditions. To enable relays to be
applied it is necessary that their accuracies be known and therefore British
Standard BS 3950, 1965 was produced to define their performance require-
ments. In this Standard the following two quantities are defined:
Asynchronous system operation 419
1. System Impedance Ratio (SIR), which is the ratio of source impedance to
relay setting impedance, expressed at the same impedance level, either
primary or secondary.
As an example, should the source impedance (Z8) at primary level be
40U and a relay be set to detect faults on a line, with an impedance of
0.40/km, at distances up to 200 km, i.e. a setting (Zd of SOU at primary
level, then the system impedance ratio would be:
IZ8105
SIR= IZLI = . or 50%
2. Characteristic Impedance Ratio (CIR) is the maximum value of system

impedance ratio up to which a relay will operate at its prescribed accuracy.
This ratio enables the voltage range over which a relay must be able to
operate with the prescribed accuracy to be determined.
It will be clear from equation (11.12) in the previous section that at the
cut-off distance or reach,
1 1
Vin = 1 + Z8/ZX Vs = 1 + SIR . Vs
If therefore, a relay were required to operate with a CIR value of 24, then
it would have to maintain its prescribed accuracy of say 5% at voltages
(Vin) down to Vs/25, i.e. if the rated secondary voltage provided by the
voltage transformers was 110 V then the relay would have to maintain its
accuracy at input voltages down to 4.4 V.
Because of the significant changes in relaying in recent years, such as the
use of digital processing, the above standard was thought to be inappropriate
and it has therefore been withdrawn. If necessary, manufacturers should now
be approached for details of the performances of their schemes to ensure that
they will be suitable for application to particular circuits.
11.5 CONDITIONS WHEN A HEALTHY LINE IS AFFECTED BY

ASYNCHRONOUS SYSTEM OPERATION
In the earlier sections, the input impedance values of a protected line which
may be present when faults occur have been considered and various relay
characteristics which are available have been outlined.
As with all protective schemes, it is necessary that incorrect operations
should not occur and certainly the opening of line circuit-breakers should not
be initiated when a protected line is healthy. By setting relays so that they will
only operate when the input impedances of a protected line are below particu-
lar levels, the requirement that faults on circuits connected to it will not cause
incorrect operation will be met.
A condition which could, however, cause incorrect operation is the asyn-
chronous operation of the system of which a protected line forms a part, i.e.
system instability. The behaviour during such a condition is considered below
and a single-phase treatment is again employed in the interest of simplicity.
For the circuit shown in Fig. 11.17(a) in which the line is considered to be
short and in which the voltages VA and VB have slightly different frequencies,
the phase angle between them would vary slowly over 21t rad. At an instant
(t) when the phase angle between the voltages was e, as shown in Fig.
11.17(b), the line current I would be given by:
1= (VA - VB cos e) + jVB sine
R + jroL
The circular loci of the current I for several ratios of the magnitudes of VA
to VB for values of e over 21t rad are shown in Fig. 11.17(c) in which the
circular loci of the apparent input impedances at end A, which is given by
VAI I, are also shown.
End B
--.
I X-t-x----0
VB __
(a) VB
(b)
(c)
Fig. 11.17 Conditions during asynchronous operation. (a) Network configuration. (b)
Locus of VA - VB. (c) Loci of I and Zin for various values of k= IVA/vB I
Asynchronous system operation 421
It will be seen that the magnitudes of the impedances vary between very
large and very small values when the voltages at the two ends are almost equal
and when the frequency differences are small, the apparent input impedance
changes slowly. In these circumstances, tripping of the line circuit-breakers
would be initiated when the apparent impedance came within the Z plane
operating zone of an impedance relay.
Now although two sources which are operating asynchronously should be
disconnected from each other, it would not be satisfactory if the whole net-
work was split into several sections because of the operation of several dis-
tance-protection schemes and ideally the separation should be effected by
opening circuit-breakers near the mid-point between the sources. It is therefore
necessary that the instability should be detected and the appropriate actions
then be taken.
A method which has been used for many years is to include in a distance-
protection scheme an extra relay element (A) with a characteristic within
which the operating zones of the other measuring elements all lie. As an
example an offset-mho characteristic could be used, as shown in Fig. 11.18.
It will be appreciated that system instability affects all the phases in a similar
manner and that its presence can be detected by a single element supplied for
example by the 'a' phase voltage and current. In the event of instability the
apparent input impedance to a line changes quite slowly initially around its
particular circular locus, as shown in Fig. 11.18. As a result the instability
detector element A would operate a significant time before any of the other
impedance-measuring elements. Subsequently, the measuring element with
the greatest reach, either Stage 3 in a three-zone scheme or Stage 2 in a
two-zone scheme, would operate. This behaviour, which would clearly indicate
the onset of instability, would be used to prevent the impedance-measuring
..... ..... Locus of Z during

.. power swing
Fig. 11.18 Principle of the instability detector element A.

relays from initiating the opening of their associated circuit-breaker. A signal
could also be provided to cause the opening of one or more circuit-breakers
to split the network at the most suitable points.
When faults occur within the zones protected by distance-protection
schemes, the input impedances of the lines change suddenly and in these
circumstances the instability detector A could operate, but it would do so at
the same time as the appropriate impedance-measuring relays; in these cir-
cumstances the latter relays would be allowed to initiate the opening of their
associated circuit-breaker.
Should faults occur which would not cause the detector A to operate, e.g. a
fault on a phase different from that to which the detector is connected, then
operation of one or more of the impedance-measuring relays would again be
allowed to initiate the opening of their associated circuit-breaker.
Summarizing, the opening of circuit-breakers as a result of the operation of
impedance-measuring relays would only be inhibited if the instability detector
relay A operated more than a certain time interval before the other relays.
This method of detecting instability is quite widely used. There are, how-
ever, some conditions when the impedance trajectory could occasionally enter
the tripping zone of the impedance relays quite quickly, causing them to
operate incorrectly.
An alternative method, based on the rate of change of voltage, is now used
in the LU91 relay [9] produced by Asea Brown Boveri.
11.6 SCHEMES APPLIED IN THE PAST

As indicated in the previous sections, early distance-protection schemes
employed quite large numbers of separate relay elements to provide two- or
three-stage cover to overhead lines.
The schemes produced by different manufacturers operated on the principles
outlined earlier but they differed in the types of relays used to determine fault
directions and line-input impedances. To illustrate the methods used, details
are provided below of the Ratio Balance type XZA scheme [10] which was
produced by A. Reyrolle and Co Ltd around the middle of the century.
This scheme incorporated directional relay elements. Three were provided
to detect phase to phase and three-phase faults and these controlled the phase-
fault distance-measuring elements, i.e. when particular directional elements
operated they caused the appropriate measuring elements to be supplied with
input voltages and currents. A single directional element, which was included
to detect earth faults, determined fault directions by comparing the phase of
the sum of the three main current-transformer outputs (3/0 ) with the phase of
the current in the system neutral.
It was claimed that the use of directional control had several advantages,
one being that the delay which occurred in energizing the measuring elements,
after the incidence of a fault, typically 20-40 ms, greatly reduced the measur-
Present-day schemes 423
ing errors which could be caused by the transient components in the system
currents and voltages.
As a further feature, three earth fault phase selectors were included, each of
which was fed with the sum of the output currents from the main current
transformer and the output of one phase of a sound-phase compensation trans-
former, the latter being connected as described earlier in section 11.4.3 (page
407). Operation of both the earth fault directional relay and an earth fault
phase selector caused the appropriate measuring element to be energized.
Three-beam type relays with plain-impedance characteristics were provided
to deal with phase to phase and three-phase faults and three induction-type
relays with reactance (ohm) characteristics were included to deal with earth
faults. The latter elements were used because of their insensitivity to resistance
in the fault path.
Two timing relays were included and these were energized when any of the
directional relays operated. The measuring relays were set to operate for faults
within the Stage 1 reach, but their reaches were extended when the first timing
relay operated, i.e. Stage 2, and later, if the second timing relay operated, the
reaches were further extended to Stage 3.
Several contactors were necessary to enable the various operations de-
scribed above, for example the extension of the reaches, to be achieved.
The phase-fault directional relays were set so that they would not operate
at current levels below 120% or 150% of the rated current of the line. The
measuring elements were therefore not energized during normal healthy con-
ditions. The ohmic settings of the measuring elements could be selected to
enable different lengths of lines to be protected.
It will be appreciated that a complete scheme contained a considerable
amount of equipment including at least eighteen electro-mechanical relays as
shown in [10]. Nevertheless, this scheme and similar schemes produced by
other manufacturers were applied to many lines and their performance records
were very good.
11.7 PRESENT-DAY SCHEMES

Schemes based on modern electronic equipment which utilize digital-signal
processing are now produced by the manufacturers of protective equipment.
These schemes are extremely versatile and can include features which could
not readily be provided in the earlier schemes which employed electro-mech-
anical elements.
Two schemes which are currently available are described in some detail below.
11.7.1 Optimho static-distance protection schemes

These schemes, designated LFZP, which are produced by GEC Measurements
and described in detail in their literature, [11], are suitable for the whole range
of overhead lines from those employed as main transmission links to those used
in distribution circuits. To cover this range several different models are available.
All the models have a full set of measuring elements for each main zone of
protection to avoid the need to rely on phase selection hardware or software
to provide the appropriate voltage and current inputs to a single measuring
element. As a result 18 measuring units are provided in schemes with three
distance zones.
Each measuring element uses a micro-controller to produce a software
equivalent of the hardware phase comparators which were used in the past.
To enable schemes to be suitable for the various applications which may arise,
the measuring elements may be arranged to have a wide range of charac-
teristics in the Z plane, including mho, quadrilateral and lenticular of variable
aspect ratio by selecting from the various forms of polarizing which are
available, including self-polarizing and a range of degrees of cross-polarizing.
Partial synchronous polarizing, derived from the pre-fault system voltages,
is also applied to the measuring elements associated with phase to phase faults
to ensure that they will operate in the event of three-phase faults near relaying
positions during which all the voltages provided to a scheme by the transdu-
cers will be very low. The pre-fault voltages, used to provide the necessary
polarization, are stored for 16 cycles. Because of this feature, operation of the
measuring elements can be obtained for all types of faults with applied volt-
ages down to zero and with currents down to very low levels.
The characteristic angles of the measuring elements are adjustable and the
impedance settings can be adjusted over a wide range enabling lines of all
types and lengths to be protected. As an example, impedance settings, referred
to the voltage- and current-transformer secondary rated values (110 V and 1
A) of 0.2 to 2500, are available with accuracies of 5%.
Because these schemes derive power from auxiliary supply units they im-
pose very low burdens on the voltage and current transformers, typical values
being 0.1 VA per phase on the voltage transformers at rated voltage and 0.08
VA per phase on current transformers with 1 A secondally windings.
The operating times for faults in Zone 1, which vary with the types of
characteristics of the measuring elements and the power system frequency (50
or 60 Hz), are between 14 and 23 ms.
The schemes can detect power swings associated with system instability and
tripping is inhibited under these conditions. Several other features can be
included, such as, (i) single-pole tripping of the line circuit-breakers in the
event of single phase to earth faults, (ii) fault location and (iii) the recording
of fault data.
11.7.2 Distance protection relay 7SL32 [12]

This relay, which is produced by Siemens, incorporates six starting elements,
which detect the various types of faults which can occur and a memory feature
Duplicate protection of transmission lines 425
x x
Operate
R I R
x I
(a) (b)
(c)
Fig. 11.19 Production of a polygonally shaped characteristic. (a) Starting element

(directional). (b) Measuring element (distance). (c) Tripping characteristic.
is provided to store pre-fault data so that close-up faults can be correctly

detected. The starting elements, which are cross-polarized, have directional
characteristics of the form shown in Fig. 11.19(a) and those which detect phase
to phase faults only operate at current levels above the rated values.
There are also six measuring elements which have offset characteristics of
the form shown in Fig. 11. 19(b) and these elements are energized when one
or more of the starting elements operate. As a result, polygonally-shaped
operating zones are produced in the Z plane, a typical zone being shown in
Fig. 11.19(c). Should a fault be present in the Zone 1 reach, tripping will be
initiated in a minimum period of 14 ms. If this does not occur then the reach
is extended to cover Zone 2 after a fixed delay and then it can be further
extended to produce a three zone scheme after a further interval.
This scheme can also detect power swings and can, if required, initiate
single-pole opening of the circuit-breakers. It imposes the low burdens of 0.1
VA per phase and 0.5 VA per phase respectively on the current and voltage
transformers.
11.8 THE DUPLICATE PROTECTION OF TRANSMISSION LINES

Transmission lines form vital parts of networks and therefore, as with busbar
protection, it is common practice to apply two protective schemes to each
major line.
In Britain many of the 400 kV transmission lines are provided with both
phase-comparison and distance schemes. This practice is adopted because
phase comparison schemes can detect faults with high resistances but they
require signalling equipment which may at some time be defective. In such an
event, the distance scheme would nevertheless operate to clear the fault.
Clearly, in such applications, the two schemes must be able to initiate the
tripping of the circuit-breakers independently.
In a few applications two protective schemes are applied to lines and they
are so connected that both schemes must operate before the opening of
circuit-breakers can be initiated. This practice; similar to that adopted with
busbars, ensures that a major line will not be opened because of the mal-
operation of one scheme. Again, a given line is provided with schemes oper-
ating on different principles because two identical schemes could both
maloperate for a particular condition.
11.9 THE PROTECTION OF FEEDER TRANSFORMERS

Feeder transformers of several forms are used in networks because they reduce
the number of circuit-breakers needed and thus allow financial savings to be
made. Two typical arrangements are shown in Fig. 11.20.
These circuits may be protected in various ways. Overcurrent or unit-type
schemes are employed and in many cases distance-type schemes are used to
protect the feeder. It will be appreciated, however, that the transformers will
have their own protective schemes and devices such as Buchholz and thermal
relays and therefore intertripping equipment must be provided to open circuit-
breakers at the remote ends of feeders.
Because transformers possess a significant series impedance which is effec-
tively lumped at the end of the feeder, a distance protective scheme at the
opposite end of the feeder (end A in Fig. 11.20) may be provided with a single,
high-speed zone covering the whole length of the feeder and reaching part way
into the transformer without any risk of over-reaching to faults on the low
voltage side of the transformer.
Such distance schemes could provide cover for both phase to earth and
interphase faults but in many cases instantaneous earth fault protection is
applied at the input end of the feeder (end A) and the distance schemes are
then only required to detect interphase faults.
In interconnected networks where power could flow in either direction in a
feeder transformer, then a distance-type scheme could also be provided at the
End A HV LV End B End A HV LV End B
r-X X-t t-x Xl

- -
(a) (b)
Fig. 11.20 Typical arrangements of feeder transformers.
The setting of distance type schemes 427
transformer to cover the feeder, the necessary current transformers being
mounted within the power transformer. Such a scheme would then be set in
the normal way.
11.10 THE PROTECTION OF CABLES

Underground cables, which are used extensively in the lower voltage sections
of networks, behave in similar ways to overhead lines and they may be
protected by distance-type schemes. The ratios of the series inductive reactan-
ces to resistances of cables are lower than those of overhead lines because of
the reduced spacings of the phase conductors and therefore the characteristic
angles of mho and other relays must be set at the appropriate values in such
applications.
The phase displacements between the positive- and zero-sequence imped-
ances of cables may be greater than those of overhead lines and therefore when
sound-phase or residual forms of compensation are to be used the phase
displacements should be taken into account and the transformers described
earlier in section 11.4.3 may not be suitable. When allowance for the phase
displacement is made the compensation is described as vectorial compensa-
tion.
11.11 THE SETTING OF DISTANCE TYPE SCHEMES

Modern schemes provide ranges of ohmic values at which the various measur-
ing elements may be set to operate. Continuous adjustment is usually possible
and the calibrations are quoted in terms of positive-sequence impedance
referred to the secondary level, i.e. the secondary levels of the voltage and
current transformers. A typical range is 0.25-32Q for a scheme operating from
current and voltage transformers with secondary ratings of 1 A and 110 V
(line) respectively. In the case of mho elements the above values are the
diameters of the characteristic circle, whereas for reactance relays they are the
values of reactance below which operation will occur. In addition the charac-
teristic angles of elements may also be set to correspond with those of pro-
tected line, i.e. tan -1 roL/R.
To determine the required settings of elements, it is necessary to calculate
the desired reaches along line sections in primary ohms, i.e. for a Zone 1 reach
e,
of SO% of a line of length, the value would be o.se [R 2+ ro2L2]~ in which
Rand L are the series resistance and inductance per unit length of the line.
The required element settings would then be given by:
Zs= CT rat~oxO.8f [R 2+ro2L2 ]t

VT ratIO
Calculations should be performed using symmetrical component or other
techniques to determine the minimum voltages which may be applied to
measuring elements during faults of all types at the Zone 1 reach to ensure
that the necessary accuracy will be obtained during such conditions. Examples
of such calculations are provided in reference [13].
11.12 ACCELERATED CLEARANCE OF FAULTS NEAR THE ENDS

OF LINES
There are clearly sections at each end of a line where a fault will be in the
Zone I reach of one of the distance-type schemes and in the Zone 2 reach of
the scheme at the other end. Such a situation, which is illustrated in Fig. 11.21,
would cause the circuit-breaker at end B to open rapidly but the circuit-breaker
at end A would open after a time delay. This behaviour, as stated earlier in
section 11.4, may be unacceptable, this certainly being so when major lines
are being protected, and means of accelerating the clearance of such faults
must be provided. Several arrangements which are available for this purpose
are considered below.
Transfer tripping
Considering the situation shown in Fig. 11.21, operation of a Zone 1 element
at end B could cause a signal to be sent to end A to initiate the tripping of the
circuit-breaker at that end directly. This arrangement, which is referred to as
a 'transfer trip under-reaching scheme' requires high security signalling equip-
ment and coding of signals to avoid maloperation.
An alternative scheme may be used in which the tripping of the circuit-
breaker at end A is only initiated on receipt of a signal from end B if at least
a Zone 2 or 3 element at end A has detected a fault. This scheme is termed a
'permissive intertrip under-reaching scheme'.
A further alternative which may be used is to detect faults within the Zone 2
reach instantly and send intertripping signals to the other end. Should a Zone 2
element or an extra directional element at the second end have also detected
the fault then tripping of the circuit-breaker would be permitted and this
process would clearly occur at both ends. These schemes are known as
'permissive intertripping over-reaching' or 'directional-comparison' schemes.
It will be clear that Zone 1 elements should not be needed when these inter-
End A Zone 1 reach of End B
_xix_En-rdA_pro_tec_tio_n---t-- ~X-
I. ~FA~~
Zone 1 reach of
End B protection
Fig. 11.21 Network configuration used to illustrate tripping arrangements.
The protection of teed lines 429
tripping arrangements are used but such schemes would be very dependent on
the signalling channel and equipment. To provide greater security Zone 1
elements are therefore often provided.
Acceleration schemes
A further alternative is to arrange for a signal to be sent when a Zone 1 relay
operates at one end of a line. Receipt of this signal at the other end then
extends the reach of the elements at that end from Zone 1 to Zone 2 immedi-
ately. This scheme, which is termed an 'acceleration scheme', does not operate
quite so rapidly as the permissive intertrip schemes because time is required
for the measuring elements to operate after their range has been extended, but
it is not prone to maloperation.
Blocking schemes
When these schemes are used, reverse-looking directional relays are provided
at each end of a protected line to detect faults on circuits connected to it, i.e.
external faults. Should these relays operate at one end (B) of the line then
blocking signals are sent to the other end of the line (A) to ensure that the
circuit-breaker at that end will not trip. In the absence of such a signal, tripping
of the circuit-breaker at end A is initiated by the Zone 3, or Zone 2 elements
in two-zone schemes, at end A when they detect faults.
It will be appreciated that time must be allowed for the operation of the
reverse-looking directional relays and the transmission of the blocking signal
before tripping may be initiated by the Zone 2 or 3 relays.
It will be apparent that the schemes described above all have advantages
and disadvantages. A factor which must be recog9ized when schemes are used
in which signals have to be sent when faults are present on a protected line is
that should the line be used as the signalling channel, then the signals may be
attenuated significantly at the fault position.
More detailed information is provided on these schemes in reference [13].
11.13 THE PROTECTION OF TEED LINES

Teed lines with three terminals, which are commonly used in lower voltage
networks, can be protected by distance-type schemes at each of the terminals,
as shown in Fig. 11.22.
Because each of the sections of such lines will normally be of different
lengths it is not possible to set the Zone 1 reaches of the three schemes to the
same values. This can be seen from Fig. 11.22 in which the section of line
from the tee point to busbar B is much shorter than that to busbar C. In these
circumstances, Zone 1 elements at end A would be set to reach to a point A'
near busbar B. These elements would leave a considerable section of the line
to busbar C in Zone 2.
A further factor which must be recognized is that the input impedance to a
-x-Melxt":"I!-'X-~-Fa-UI-t {=tx-
' ~
EodA - EndC
-"'f~!I---- End B
XI
Fig. 11.22 Network containing a teed line.
section of line will not be directly proportional to the distance of a fault from
the relaying position. This can also be seen from Fig. 11.22 by considering a
short circuit on the section of line to busbar B.
If current could be fed to the fault from busbars A and C, i.e. IA and Ie, then
the voltage at bus bar A would be:
The apparent input impedance of line A would thus be:
ZinA =VA =(R + jroL) {fA + X + Ie x}

IA h x
It can be seen that the magnitude of the apparent input impedance would
increase with increase in the infeed from busbar C. As a result, the length of
the section of line to busbar B on which faults would be detected by the
protection on line A would decrease with increase of the current fed by bus bar
C, but it would never fall to zero.
Such conditions would cause delayed tripping of one or more of the circuit-
breakers for faults in significant sections of the lines because they would be
within the Zone 2 reaches. In some applications this performance may be
acceptable, but where this is not so, one of the forms of inter-tripping described
in the previous section may be provided to accelerate the clearance of faults.
11.14 THE PROTECTION OF SERIES COMPENSATED LINES

The power which can be supplied from the remote end (B) of a line is
depe~ent on the magnitudes ~nd phases of the voltages at both the re.!.D0te
end (VB) and at the near end (VA) and the series impedance of the line (ZAB)'
For simplicity, the behaviour of a theoretical loss-free, single-phase line
with no shunt admittance is considered below. The current (l) in such a line
would be given by:
1= VA - VB = VA - VB
ZAB j XAB
Protection of series compensated lines 431
. VA . VB
=-j-+j-
XAB XAB
If VB is taken as the reference phasor, then the power output of the line
would be:
p=Re(-j VB VA)
XAB
This clearly has a value of zero when the voltage VA is in phase with the
voltage VB and it has a maximum positive value when the voltage at end A
leads that at end B by n/2 rad. In addition, the maximum power is inversely
proportional to the inductive reactance of the line.
Similar behaviour occurs on actual three-phase lines and transfer powers
would be unacceptably low on many very long lines because of their high
inductive reactances and the fact that operation cannot be permitted near the
limiting value because of the possibility of instability. To improve the behavi-
our, it is now common practice to connect capacitors in long lines to effec-
tively reduce the overall reactance between their ends and thus increase their
power transfer capabilities.
The ratios of the capacitive reactances to the effective series inductive-
reactances of lines used in installations vary, but a typical figure is 0.5. Such
degrees of compensation on high voltage lines require that very large voltages
be present across the capacitors when their lines are carrying their rated
currents and therefore unacceptably high voltages would be produced across
the capacitors if they carried high fault currents. It was therefore the practice
to limit the voltages across capacitors by connecting arc gaps across them but
metal-oxide varistors (MOYs), which have been produced in recent years, are
now usually employed. These latter devices, which have resistances which
decrease with increase of currents, have the advantage that they are self-ex-
tinguishing when normal current conditions are re-established and they pro-
vide significant damping to the power systems in which they are installed.
It will be appreciated that the capacitors remain in circuit for short periods
after the incidence of faults and are then effectively short-circuited. As a
result, protective schemes are presented with two different sets of conditions
during the period when a fault is present on a line.
Considering the single-phase line shown in Fig. 11.23(a), in which a capa-
citor C is installed at the mid-point to provide 50% compensation, i.e.
l/mC = 0.5.(mL, the steady state input impedance at end A (Zin) for a short
circuit at a point on the line at a distance x from end A would vary as shown
in Fig. 11.23(b) when the capacitor was in circuit. The variation would,
however, be as shown in Fig. 11.23(c) when the capacitor was short circuited.
In practice, the situation is further complicated by the fact that the voltages
and currents would contain transient components initiated both by the fault
and by the short-circuiting of the capacitor.
In the above consideration, the capacitor was taken to be at the mid-point
of the line. Whilst such arrangements are used, it is common practice to install
Busbar A Busbar B
112
~R. L per metre II~12
c X
2in (a)
2in
tl~~W
o 112
(b)
-x
1 0 112
(c)
-
x
Fig. 11.23 Conditions in a line with series compensation at mid-point.
capacitors at both ends of a line near the circuit-breakers. Total compensation

is typically 70% and such practice has advantages. Firstly a site near the
mid-point of a line is not needed and the capacitors can be more readily
inspected. A further important factor is that the voltage transformers can be
connected on the line sides of the capacitors as shown in Fig. 11.24(a). With
this arrangement, the voltage drops across the capacitors do not affect the
behaviour of impedance-measuring relays when faults occur on the line al-
though they will do so in the event of faults external to the line, the input
impedance presented to the relay at end A of the line shown in Fig. ll.24(a)
would vary with the distance to a fault on the line or beyond its remote end
as shown in Fig l1.24(b) when the remote capacitor is in circuit. The input
impedance (Z) would, however, vary as shown in Fig. l1.24(c) when the
remote capacitor is short circuited.
Clearly, first and second zone settings which will ensure correct operation
for all faults can be selected more readily than is the case when a single
--
X} End A
X""JJ-L"-{
I-FI~:::.-n_ _ _,.1
Line (R and L per metre)
y-tX-
.--.
End B
-X rm+
Current
[ Voltage
X-
transformer transformer
'I"''''IVJ
(a)
2in~
- - - - - - - - - - --
'IR+~LI
0"'1"''''1 ~
.z:n
r--------/
o (b) 1 7 o 1 ~
Totaf compensation = 70% (c)
Fig. 11.24 Conditions in a line with series compensation at the two ends.
Protection of series compensated lines 433
capacitor is connected at some point in a line and satisfactory operation can
certainly be achieved when accelerating or transfer-tripping features are in-
cluded. It must be recognized, however, that two sets of transients will be
present whenever a capacitor is short circuited. Further details may be found
in references [14, 15].
It will be clear from these comments that many factors must be considered
when distance-type protective schemes are to be applied to series-compen-
sated lines. To ensure that satisfactory performance will be obtained it is
desirable that studies be performed in which systems are accurately modelled
on computer-based test benches, capable of dynamically testing proposed
schemes.
Newbold and Taylor described studies performed in the above way in a
paper presented at an lEE Conference in 1989 [16]. The results of a study of
the behaviour of the series-compensated lines of the West Coast, USA were
provided and it was shown that distance-type protective schemes could be set
so that they would provide satisfactory performance under all conditions.
11.15 THE FUTURE

Distance-type protective schemes have been used successfully for over 50
years to protect overhead lines and cables and it seems probable that they will
continue to be used well into the future.
The principles of distance protection have remained essentially the same
over the years but the introduction of microprocessors into protective-relay
design has introduced new methods of deriving the necessary relaying quan-
tities. The outputs of current and voltage transformers or transducers can be
passed through an analogue-to-digital (ND) converter and thus transformed
into a sampled digitized form. Further processing can be employed to imple-
ment sophisticated relaying functions [17]. Relaying can thus be regarded as
a parameter estimation task whereby samples of the voltage and current signals
are processed to obtain quantities such as the r.m.s. value of post-fault voltage
and current or the impedance to the fault. Much greater flexibility thus exists
in implementing any desired characteristic such as a quadrilateral tripping
characteristic of a distance relay.
There are two general approaches to estimating relaying parameters by
using digital signal processing techniques.
An estimate can be made based on a model of the waveform, e.g. voltage
or current. For conditions encountered during faults it may be postulated that
a transducer signal consists of the fundamental, second harmonic, and an
exponentially decaying term. Consecutive samples of the signal can then
processed to obtain estimates of the various terms. For example, the samples
may be processed by an algorithm which implements a Discrete Fourier Trans-
form of the signal and can thus provide the amplitude and phase of the
fundamental term.
To illustrate this procedure a voltage signal v(t) of period 2n/ro is con-
sidered. If this signal is sampled at time intervals T to obtain N samples in any
one period (NT = 21t/ro), then the equivalent discrete signal v (nT) is obtained
consisting of a succession of samples v (0), v (T), v (2T) ....
The Discrete Fourier Transform (DFT) of this signal may be obtained by
implementing the following algorithm:
L
N-I
V (kilo) = v (nT) e- JOkno",T

o. 21t
where ilo = -
n=O NT
The fundamental-frequency component corresponds to k= 1:
n=O
A similar procedure may be followed for the current signal and an imped-
ance estimate is then obtained from:
This value is then compared with the required tripping characteristic in the
impedance plane and the appropriate action is taken. The number of samples
that should be processed before a secure decision is reached is a compromise
between speed of operation and immunity to noise. Sophisticated methods are
available which can make allowances for the presence of noise in the sampled
waveforms so that a rapid estimate is made with the minimum error. An
example of this technique is Kalman filtering. This is an example of an
adaptive method where after each sample an update of the expected error is
made so that estimates improve rapidly [18].
An alternative approach to parameter estimation based on a model wave-
form, is one based on a model of the system. It may, for example, be assumed
that the system may be presented by a series R-L combination. Then the
voltage v (t) and the current i (t) seen by the relay are related by [19].
di(t)
v(t) = Ri(t) + Ldt
Integrating this equation from tl to t2 gives

l
f 2v(t)dt=R fl2 i(t)dt+L [i(t )-i(td] 2 (11.13)
1\ 1\
These integrals may be evaluated using the trapezoidal rule, i.e.
f
I
l+M
v (t)dt =.1 [v (I + At) + v (t)] At
2
where At is the time interval between consecutive samples.
The future 435
Implementing equation (11.13) between time intervals kilt to (k + 1) Ilt and
(k + 1) Ilt to (k + 2) Ilt and solving for Rand L gives:
R= (Vk+1 - vk)(ik+2 - ik+l) - (Vk+2 - Vk+l)(ik +2 - ik)
(ik+ 1+ ik)(ik+2 - ik+ I) - (ik+2 + ik+ I) + (ik+ 1- ik)
L = Ilt [(ik+ 1 + i k )(vk+2 + Vk+ I) - (ik+2 + ik+ I)(Vk+ 1 + Vk)]

2 (ik+l+ik)(ik+2-ik+I)-(ik+2+ik+I)(ik+l-ik)
where Vk. ik is the value of the voltage and current sample respectively at time
k . Ilt. The estimates Rand L can then be used to implement any desired
tripping characteristic. The advantage of this formulation is that transient
terms are fully accounted for in the model and do not constitute an error.
However, it is obvious that the model does not take account of capacitive
effects, and hence of travelling-wave phenomena. Another aspect of practical
significance is that the algorithm described above is based only on three
samples and may therefore be unacceptably sensitive to signal noise. Improve-
ments can be made where the processing window of three samples is made
longer so that better noise immunity is obtained [20, 21]. Further details of a
number of relaying algorithms may be found in references [22,23] and in
Chapter 14 of this book.
Fully algorithmic relay designs based on the principles outlined above
are becoming increasingly available from manufacturers. The fully digital
design allows greater flexibility in algorithm optimization and updating,
comprehensive interrogation of the relay from a remote terminal, more
sophisticated interlocking arrangements with other relays (e.g. on double-
circuit lines), closer coordination with power system control functions via
SCADA systems and sophisticated self-test facilities.
A digital relay incorporating two different protection algorithms, a
high-speed and a conventional one, is described in [24,25]. The high-speed
algorithms use superimposed quantities and enable very fast response
times 2 ms) to be achieved. The conventional algorithms are used when
the power system is not reliable for a sufficient period of time prior to a
fault. Both algorithms are on standby at the same time. The relay
architecture follows closely the principles described in Chapter 14 of this
book with anti-aliasing filters, multiplexer, sample-and-hold and analog-
to-digital converter. Further filtering and the derivation of relaying
quantities are performed by the first digital signal processor and the
protection algorithms by the second. A standard 16-bit microprocessor
implements trip logic and other housekeeping functions.
For the purposes of illustration the LFZR high-speed numerical distance
relay, produced by the same manufacturer, is described here [26]. This is
a fully numerical relay, with operating speed typically less than one cycle,
five independent zones of protection, mho characteristics for phase and
earth faults, selectable quadrilateral characteristics for earth faults and a
comprehensive range of other protection and communication facilities. A
IRIG-B Serial pori B
'K'Bus
Communication module
Back ~ane
Analogue
Power board
Buffer supply
board 16biAOC
72 sample~cycle
Host processor
board
\\)Iage & currenl

transformers
Va Vb Ve
r=:=Q] la Ib Ie
- -
r=:=Q] In Im/lpol
J-e
inputs
r=:=Q] Ves
r=:=Q]
r=:=Q]
Indicator LEDs
~~'lJ!J} Parallel p-inter
Fig. 11.25 Electrical structure of type LFZR high speed numerical distance relay.
(Reproduced with the permission of GEe Alsthom T and D.)
schematic diagram of the architecture of this relay is shown in Fig. 11.25.

The protection processor board is an essential part of the relay capable of
accepting and processing sampled values from the analogue input board.
A fast digital signal processor capable of 25 million instructions per second
(MIPS) implements this function and another 25 MHz scheme logic
processor evaluates information from the digital signal processor and the
relay optically isolated inputs to operate the relay's output elements. The
host processor board has two processors, one for ASCII RS232
communications and the other for communications through the K-Bus and
RS232 using the IEC870 protocol. A description of communication
interfaces and protocols used in protection may be found in Chapter 13 of
this book. Typical of all numerical relays and also of this particular
example are the enhanced functionality and self-check facilities. A host of
ancillary functions are available such as measurements, fault!event!
disturbance recording, and fault location. Features of the relay are
accessed through the front panel interface shown in Fig. 11.26. A menu-
driven system can be navigated using the arrow keys to select particular
sections and items. The soft keys under the liquid crystal display (LCD)
have different functions depending on the position in the menu. The menu
may also be accessed and changes made remotely through a PC connected
to the relay through its communications ports.
Reyrolle Protection manufacture the Ohmega Numeric Distance
Protection Relay [27]. It is of modular design with self-monitoring and RS-
The future 437
Fig. 11.26 The front panel of type LFZR relay. (Reproduced with the permission
of GEe Alsthom T and D.)
232 serial communications facilities. A range of characteristics, e.g. mho,

quadrilateral, is available for up to five zones of protection. The standard
protection schemes are available, i.e. stepped distance, permissive/inter-
tripping/ accelerated undereach, blocking/unblocking/permissive overeach.
Another example of a numeric distance protection relay is the 7SA5l3
[28]. Computer-based signal processing removes the influence of inrush
currents, d.c. components, harmonics and high-frequencies. Errors due to
currents in the load or in a parallel line can be compensated. A parallel
calculation of six impedance loops is implemented to give a high degree of
sensitivity and selectivity. The relay implements all normal protection
functions expected of transmission line protection. Standard serial
interfaces permit the integration of the relay into substation installations.
A numerical line distance protection relay is described in [29]. Its basic
architecture consists of standard isolation and filtering, an analogue-to-
digital converter, ten digital signal processors, a 32-bit micro-controller,
communications interfaces, and outputs. It implements two algorithms
operating in parallel. One algorithm is a distance scheme and the other is
an adaptive hybrid algorithm for high-speed operation. The idea of hybrid
algorithms with adaptive features is explored further in [30]. Adaptive
relaying is also discussed in section 15.4 of this book.
REFERENCES
I. Ackerman, P. (1922) Journal of the Engineering Institute of Canada, Dec. 1922 (also
New radial relay protection, Electrical World, 81, March 17, 1923, pp. 619- 623).
2. Crichton, L. N. (1923) The distance relay for automatically sectionalising electrical
networks, Trans. lEE, XLII, 527-537.
3. McLaughlin, H. A. and Erickson, E. O. (1928) The impedance relay development and
application, Trans. lEE, July 1928, 47,776-784.
4. George, E. E. (1931) Operating experience with reactance type distance relays, Trans.
lEE, SO, 288-293.
5. Ellis, N. S. (1982) Distance protection of feeders. In Power System Protection Manual,
Reyrolle Protection, (ed. D. Robertson), Oriel Press, pp. 149-179.
438 Distance-type protection schemes
6. Wright, A. (1961) Limitations of distance-type protective equipment when' applied to
long extremely high voltage power lines, Proc. lEE, (C), 108, 271-280.
7. Davison, E. B. and Wright, A. (1963) Some factors affecting the accuracy of distance-
type protective equipment under earth fault conditions, Proc. lEE, 110, 1678-1688.
8. Micromho-Static distance protection relay, GEC Measurements, Pub/. R5406.
9. Power Swing Blocking Relay, Type LU91, Asea Brown Boveri, (1987), Publ. CH-ES-
66.51.11.
10. Reyrolle Ratio Balance Distance Protection Type XZA Pamphlet 1044 (1951).
11. Type LFZP Optimho Static distance protection relay, GEC Measurements, Publ. R-4056 C.
12. High-speed distance protection relay 7SL32, Siemens Protective Relays, Catalog R-
1989, pp. 5/21-5/26.
13. Distance protection schemes, In Protective Relays Application Guide, GEC Measure-
ments (3rd edn), 1987, Chapter 12.
14. Mathews, C. A. and Wilkinson, S. B. (1980) Series compensated line protection with a
directional comparison relay scheme, 2nd lEE Int. Con! on Developments in Power
System Protection, lEE Conf. Publ. No. 185, pp. 215-220.
15. EI-Kaleb, M. M. and Cheetham, W. J. (1980) Problems in the protection of series
compensated lines, 2nd lEE Int. Con! on Development in Power System Protection, lEE
Conf. Publ. No. 185.
16. Newbold, A. and Taylor, I. A. (1989) Series compensated line protection: system
modelling and relay testing, 4th lEE Int. Con! on Developments in Power System
Protection, lEE Conf. Publ. No. 302, pp. 182-186.
17. Mann, B. J. and Morrison, I. F. (1971) Digital calculation of impedance for transmission
line protection, Trans. lEE, PAS-90, 270-279.
18. Girgis, A. A. and Brown, R. G. (1981) Application of Kalman filtering in computer
relaying, Trans. IEEE, PAS-lOO, 3387-3397.
19. McInnes, A. D. and Morrison, L. F. (1970) Real time calculation of resistance and
reactance for transmission line protection by digital computer, Elec. Eng. Trans., IE,
Australia, EE7, (1), 16-23.
20. Ranjibar, A. M. and Cory, B. J. (1975) Algorithms for distance protection, Developments
in Power System Protection, lEE Conf. Publ. 125, London, March 1975, pp. 276-283.
21. Sanderson, J. V. H. and Wright, A. (1974) Protection scheme for series-compensated
transmission lines, Proc. IEEE, 121, 1377-1384.
22. Microprocessor relays and protection systems, lEE Tutorial Course, (Course coordinator
Ms Sachdev), 88EH0269-I-PWR.
23. Phadke, A. G. and Thorpe, J. S. (1988) Computer Relayingfor Power Systems, RSP.
24. Holweck, C. (1990) Design and performance of PXLN digital protection devices for
EHV /UHV power systems, GEC Alsthom Technical Review, No.2, pp. 23-24.
25. GEC Alsthom Tand D (1992) Digital Distance Protection PXLN, Publ. N16869A-02/92.
26. GEC Alsthom T and D (1995) Type LFZR High-Speed Numerical Distance Relays,
Publ. R-4117C.
27. Reyrolle Protection (1997) Ohmega Numeric Distance Protection, Publ. REY3/9703-
166/1-2.
28. Siemens, A. G. (1996) 7SA5I3 Line Protection Relay, Publ. LSA 2.1.12, Apr. 1996.
29. ABB (1995) REL 531 Line Distance Protection, Publ. IMRK 506 004-BEN.
30. Saha, M. M., Wikstrom, K. and Lindahl, S. (1997) A new approach to fast distance
protection with adaptive features, Proc. Int. Con! On Developments in Power System
Protection, 25-27 March, University of Nottingham, pp. 32-35.
FURTHER READING
Analysis of distance protection, V. Cook, RSP 1986, 188 pages.
Protection aspects of multi-terminal lines, lEE Power Eng. Soc. Publ, IEEE Report No
79, TH 0056-2-PWR, 1979, pp. 1-17.
Determination of time settings in a distance protection scheme based on statistical
methods, J. De Haas, ClGRE-report 31-05.
12
Ultra-high-speed schemes for the
protection of long transmission lines
INTRODUCTION
It has been made clear in earlier chapters that unit-type protective schemes
which compare the quantities at the two ends of transmission lines can provide
the necessary discrimination between faults on the circuits they protect and
both healthy conditions and faults on other circuits. Such schemes cannot
however operate extremely rapidly because time must be allowed for the
interchange of signals between the ends of the protected lines.
These factors do not arise when impedance- or distance-type protective
schemes are used and the tripping of circuit-breakers may be initiated for
faults within the Zone 1 reaches without delays being introduced during which
information is obtained about conditions at the remote ends of the protected
lines. To achieve the required measurement accuracies, however, a significant
portion of a power-frequency cycle must be monitored after fault inception
and therefore circuit-breaker tripping cannot usually be initiated in times much
less than 20 ms on systems operating at a frequency of 50 Hz.
In many applications the performance provided by these schemes is accept-
able but there has long been a desire to obtain more rapid clearance of faults
on long major transmission lines. Such performance would reduce the damage
which can be caused by high currents flowing in healthy equipment and in
addition it would reduce the possibility of asynchronous operation being in-
itiated as a result of faults on transmission lines. It will be appreciated that the
incidence of a fault on a line will reduce the power transmitted by it and should
it be connected between two sources then the machines at one end will supply
less power whilst those at the other end will supply more. As a result one set
of machines will accelerate whilst the others will decelerate and their angular
displacements will increase. If this situation were allowed to persist, asyn-
chronous operation would occur and, as stated above, the likelihood of such
behaviour is reduced if the time needed to clear faults is reduced.
In recent years the possibilities of producing protective schemes using mod-
ern electronic-processing equipment to analyse the travelling waves present

440 Ultra-high-speed schemes
on long transmission lines during fault conditions have been examined exten-
sively, because such schemes should enable the most rapid operation possible
to be obtained. After the next section, which provides historical material, the
principles and operation of schemes which may be used extensively in the
future are examined.

It has long been known that a sudden change in the voltage at a point on a
transmission line, resulting from a fault or short-circuit, causes travelling
waves of both voltage and current to propagate away from the point in both
directions.
When studying such conditions it is usual to assume that the steady state
conditions prior to a fault or disturbance continue and that transient travelling
waves will also be present during the fault. The situation is usually simulated
by considering a source of e.m.f. to be connected suddenly at the fault point and
it is this e.m.f. which causes the transient currents and voltages to be present.
Because major transmission lines have relatively low series resistances and
low shunt conductances, the travelling-wave components which do not involve
the earth have a velocity near that of light, i.e. 3 x lOS mls and thus waves
caused by a fault on a line reach its ends very quickly. For example, the first
current and voltage waves resulting from a fault at a point 300 km from the
end of a line would arrive at that end 1 ms after the incidence of the fault. It
will be appreciated that this would be the first indication at the end of the line
of the presence of the fault and if information could be quickly obtained from
these waves by protective equipment then the most rapid fault clearance
possible could be effected.
Analyses performed over 30 years ago indicated the possibilities of produc-
ing protection schemes which analysed travelling waves but the equipment
available at that time was not capable of performing the necessary signal
processing in real time. Subsequent developments in electronic equipment did,
however, provide the opportunity for the development of travelling-wave
schemes and in recent years much research effort has been devoted to exam-
ining the various detection processes which may be employed.
In the later 1970s a considerable number of papers were published on
methods of determining the directions of faults from the first travelling waves
reaching the ends of long lines, it being intended to use this information in
directional-comparison schemes. Two important papers were published by
Chamia and Liberman in the lEE Transactions in 1978 [1] and Johns in the
Proceedings of the lEE [2] in 1980 respectively. Because directional-compari-
son schemes have the disadvantage that signals must be sent between the ends
of lines to ensure correct discrimination, other workers sought to obtain further
information from the travelling waves so that schemes could be produced in
which fault positions could be determined from the quantities at a single end
Travelling waves 441
of a line. Papers published by Crossley and McLaren [3] in 1983 and by
Christopoulos et al. [4, 5] in the Proceedings of the lEE in the years 1988 and
1989, describe such schemes. It is made clear [4] that the necessary informa-
tion cannot be obtained from the first travelling current and voltage waves to
arrive at the end of a line after a fault occurs on it and that information must
also be extracted from the next wave which arrives as a result of reflections
at the line end and the fault.
Detailed considerations of the behaviour of travelling waves and protective
schemes based on them are provided in the remainder of this chapter.
12.2 TRAVELLING WAVES

As stated earlier, travelling waves propagate in both directions away from any
point on a transmission line at which the voltage is suddenly changed as a
result of a disturbance such as a fault.
The voltage waves progressively change the voltage at other points on the
line and they are accompanied by current waves which change the charges on
the line capacitances.
12.2.1 Initial waves on d.c. lines when short-circuits occur

The above behaviour can be most simply illustrated by considering an ideal,
i.e. loss-free, two-conductor line operating in a d.c. system. Referring to Fig.
12.1(a), the voltage between the conductors would be the same at all points
vJo.c. Characteristic
impedance
Short
circuit
vt~] Load
supply
~-------~~--~--------~
(a)
"=4 I-v
(b)
"1
F
1'=-~20
(c)
Fig. 12.1 Voltage and current waves on a d.c. line following a short-circuit.
under healthy conditions. Each of the capacitors per unit length (C) would be
charged to the system voltage (V) and there would be no voltage drops across
the series inductors per unit length (L).
In the event of a short-circuit between the conductors of a line at a point F,
travelling voltage waves of magnitude V and negative polarity would be
propagated from point F at a velocity of ...Jl/LC mis, as shown in Fig. 12.1(b).
The accompanying current waves would have a magnitude of V/Zo, where the
characteristic impedance Zo is equal to ...JLlC Q. The polarities of the travelling
waves of voltage and current which would reach both ends of the line would
all be the same, i.e. negative in the above example.
Should two such lines be connected in series as shown in Fig. 12.2(a) and
a short-circuit occurred on one of them, say line P, then the above waves
would again be propagated. At both ends of line P the arriving voltage and
current waves would again have the same polarities, i.e. negative. As these
waves propagated into line Q, the voltage wave would retain its negative
polarity but a device detecting the current flowing to the bus bar from line Q
would regard the current wave as positive. These conditions are illustrated in
Fig. 12.2(b). It will therefore be clear that the arrival of voltage and current
waves of the same polarity at the end of a line would indicate a fault on the
line or beyond its remote end whereas the arrival of waves with opposite
polarities would indicate a fault in the opposite direction, i.e. on line P in the
above example.
- --
Busbars
--
Positive Positive
current current Line Q
Line P
Positive Short
current circuit
(a)
"1 I-v
F"
rr"
(b)
"1 I -1HZ,
(c)
Fig. 12.2 Voltage and current waveforms for conventional positive current reference
directions.
It must be recognized that the travelling waves exist in addition to the
pre-fault voltage and current and therefore to produce a directional detector
the pre-fault, i.e. steady state quantities, must be subtracted from the actual
fault currents and voltages to obtain the travelling-wave components.
An important feature of this technique is that it enables correct operation to
be obtained under all conditions, unlike other types of directional relays which
may not operate when voltages collapse to very low levels during faults.
12.2.2 Initial waves on ideal single-phase a.c. lines when

short-circuits occur
As stated earlier, the extra voltages and currents which are superposed on the
d.c. lines considered above, in the event of a short-circuit between the con-
ductors, may be determined by connecting a source at the fault position with
an e.m.f. equal to but of opposite polarity to that present at the instant of fault
occurrence, i.e. - V.
The same technique may be used when the behaviour of a.c. lines is being
considered, but in such cases the source connected at the position of the
short-circuit must provide an e.m.f. equal and opposite to that which would
have been present if the fault had not occurred. This situation is shown in Fig.
12.3(a), from which it can be seen that the source e.m.f. would be given by
e =- Vpk sin (Olt + a) if the voltage at the fault position F would have been
v = Vpk sin (Olt + a) had the fault not occurred. In each case the time t is
measured from the instant of fault occurrence.
~X
I
Characteristic
impedance Zo
Short circuit
te = Vpksin(rot + a)
Xl
(a)
"=c==c=:F" I-v~
F
(b)
"1 :=c=F" I-~ (c)
Fig. 12.3 Voltage and current waves on an a.c. line followin~ a short-circuit.
As a result, travelling waves of voltage and current would propagate in both
directions from the fault position. Again these waves would travel at the
velocity of light on ideal lines and they would not be attenuated as they
travelled. Unlike the waves on d.c. lines, however, their amplitudes at given
positions would vary with time after the wavefront had passed the various
positions but the amplitudes at the wavefronts would be constant. This behavi-
our is illustrated in Fig. 12.3(b) for a condition in which a short-circuit has
taken place at an instant when the voltage at the fault position was positive
and of its peak value, i.e. v = Vpk.
The accompanying current would again be given by i = e/Zo and it would
thus have the same waveform as the voltage. As in the case of d.c. lines, the
polarities of the travelling waves of voltage and current which would arrive
at the ends of the faulted line would be the same but, as before, they would
be of opposite polarities in the event of faults on other lines. The directions
of faults could therefore be determined as before.
A factor which must be recognized, however, on an a.c. line is that the initial
magnitudes of the travelling waves of voltage and current may be very small
or even zero if faults occur at or near instants when the voltage at the fault
position is zero. Clearly the directions of faults could not then be determined
immediately a fault occurred.
12.2.3 Initial waves on ideal lines when resistive faults occur

The initial value of the travelling voltage waves which would propagate away
from a fault position in the event of a fault with a resistance of Rr n between
the conductors would be given by:
Zo
VI =- Zo+2Rr Vr (12.1)
in which Zo is the characteristic impedance of the line and Vr is the voltage at

the fault position immediately prior to the incidence of the fault. In the case of a
d.c. line Vr would be equal to the line voltage V and for an a.c. line it would be
Vpk sin ex, ex being the angle after the voltage zero at which the fault occurred.
The initial values of the accompanying current waves, taking the directions
of the waves as the references, would be given by:
VI I
II=Zo = Zo+2RrVr (12.2)
It will be clear from equations (12.1) and (12.2) that the amplitudes of the
travelling waves reduce with increase of the resistance in the fault path.
12.2.4 The effects of line resistance and leakage conductance

In the sections 12.2.1-12.2.3, the behaviour which would be obtained if faults
occurred between the conductors of ideal d.c. or single-phase a.c. lines were
considered. The conductors of practical lines possess resistance (R), however,
and leakage conductance (G) is present between them. The latter quantity (G)
which is usually very low and variable, occurs mainly because of deposits and
moisture on the surfaces of the insulators. The presence of resistance and
leakage conductance causes the shapes of the fronts of travelling waves to be
modified and the amplitudes of the waves attenuate as they travel. These
effects tend, however, to be small and this is particularly so when faults occur
between the conductors of major three-phase transmission lines which employ
several conductors in parallel per phase, i.e. bundle conductors. In such cases,
as shown earlier, the directions of interphase faults relative to the ends of lines
can be readily detected from the polarities of the initial travelling waves which
arrive at the ends of lines after the incidence of such faults.
It must be recognized, however, that in the event of faults which cause
currents to flow in the earth, the velocities of the ground components of the
travelling waves and also their wavefronts may be affected significantly, the
differences being dependent on the resistivities of the ground-return paths.
This factor, which must be taken into account, is considered later under section
12.2.6 (page 452).
12.2.5 Later travelling waves resulting from reflections

It will be clear from the preceding sections that the directions of faults relative
to the ends of lines can be determined from the polarities of the initial voltage
and current waves which reach those ends after faults occur.
The ratio of the amplitudes of the initial incoming voltage and current waves
at the end of a line is always equal to the characteristic impedance (Zo) of the
line, which is known, and therefore this parameter does not give any extra
information about any fault which is present.
In the case of a d.c. line, the voltages at all points along its length will be
almost equal to the voltages at its ends during healthy conditions and should
a fault occur at a point on it, the approximate value of the pre-fault voltage at
that point (Vf) would be known. Measurement of the magnitude of the first
travelling voltage waves to arrive at the line ends (Vd would therefore enable
the fault resistance (Rf) to be assessed from equation (12.1). No information
about the position of the fault, other than its direction could, however, be
obtained from the initial waves.
In the case of an a.c. line the situation is basically similar. During healthy
conditions there may, however, be significant phase displacements between
the voltages at various points along the line and therefore the instantaneous
voltage at the ends of the line may be considerably different to those at other
points. Should the instantaneous voltage at an end, at the time when a fault
occurs, be used in equation (12.1) as the value of Vf then the assessment of
the fault resistance could be quite inaccurate.
Further information, which could be used in protective schemes to enable
Busbar A Busbar B
x
-x Line P characteristic x-
~____~im~e~da~n~ce~z~__ X
x-
(a)
(b) (c)
Fig. 12.4 A Bewley lattice diagram for different fault positions.

satisfactory discrimination to be obtained without recourse to signalling be-
tween the ends of lines, could be derived from the later travelling waves which
are caused by reflections. This is illustrated by making use of the well-known
Bewley [4] lattice diagram. To simplify the treatment an ideal single-phase
line P connected between busbars A and B as shown in Fig. 12.4(a) is now
considered.
The initial values of travelling waves produced by faults on a protected line

In the event of a fault with a resistance Rf at a point a distance x from busbar
A, a voltage wave would travel towards busbar A. The initial values of this
wave and the accompanying current wave would, as stated earlier in equation
(12.1), be given by:
20
VI =- 2o+2Rf Vf
1
and II = - 20 + 2Rf . Vf
As there are invariably several connections to a set of busbars, any waves,
such as those above, would initiate reflected waves (Vrl and I rl ) when they
arrived at the discontinuity presented by the busbar. This situation is shown
in Fig. 12.4(b).
It is well known that the voltage and current reflection coefficients for the
waves arriving at bus bar A would be:
kVA = [R A - Zo]/[R A + 20] (12.3)
kiA =- [RA - 2o]/[RA + Zo] =- kVI (12.4 )
In both these equations, RA is the input resistance presented at busbar A to the
incident waves arriving from line P. For the purpose of illustrating the prin-
ciple of the technique RA is assumed resistive.
Protective voltage and current transducers on line P near busbar A would
not detect the incident and reflected waves separately because their time
separation would be very short. They would therefore provide outputs propor-
tional to the sums of the initial values of the two sets of waves, i.e.
(12.5)
and
(12.6)
The ratio of these equations is

(12.7)
By determining the value of RA from the above equation and knowing the
characteristic impedance of the line (Zo), the reflection coefficients, kVA and
kiA, could be determined from equations (12.3) and (12.4) and then the initial
values of the incident at reflected voltages and currents (VI. Vrh II. I rl ) could
be found from equations (12.5) and (12.6).
The first wave reflected from busbar A (Vrh Irl ) would return along the line
P towards busbar B until it arrived at the fault position where further reflected
waves (Vr2, I r2 ) and transmitted waves (Vll, Ill) would be initiated as indicated
in the lattice diagram shown in Fig. 12.4(b).
The voltage reflection coefficient kvf at the fault position would be given
by:
[Zo Rr/(Zo + Rf)] - Zo

(12.8)
kvf= [ZoRr/(Zo+R f ) +Zo
The reflected voltage wave Vr2 which would return to busbar A would
therefore have an initial value given by:
Vr2 =kvf Vrl
This wave would in turn reflect from busbar A to initiate a further reflected
voltage wave with an initial amplitude Vr3 given by:
The increment applied at this time to the voltage transducer on line P at its
end connected to busbar A would be equal to:
VS2 =Vrz + Vr3 =kvf (1 + kvA ) Vrl (12.9)
Having already determined the initial value of the reflected voltage wave
Vrl , the value of the coefficient kvf could be found from equation (12.9) and
this would then enable the fault resistance (Rf) to be determined from equation
( 12.8).
As shown in Fig. 12.4(b), the voltage increment VS2 would occur at a time
2tl seconds after the increment VSI produced by the first travelling wave (VI),
indicating that the distance x of the fault from busbar A would be given by:
x =uti metres (12.10)
in which u is the propagation velocity of the travelling waves which is known

for any particular line.
It will be seen that a considerable amount of information could be obtained
by measuring the increments VSh lSI and VS2 and the time interval between the
voltage increments VSI and Vs2 The results would, however, only be correct
provided that the wave reflected from bus bar B and transmitted through the
fault to produce the increment VS3 in Fig. 12.4(b), arrived after the increment
VS2 ' This condition would be met if a fault occurred within half the line length
from busbar A, i.e. x < fp/2.
For faults on the other half of the line, the increment VS3 caused by the
reflection from busbar B would, however, occur before the increment VS2
caused by the reflection of the wave Vrl as shown in Fig. 12.4(c). In these
circumstances, if the increment VS3 were used in equations (12.1) to (12.7) to
determine the fault resistance (Rf) and the distance to the fault (x), the values
obtained would be incorrect.
Nevertheless, it will be clear from the above that the positions of faults could
be determined using equation (12.10) provided that the increments VS2 and
I s2 , produced by waves travelling between the faults and the busbar A, could
be identified.
Several methods of identifying the waves VS2 and IS2 have been suggested
and schemes based on such techniques have been proposed. These are exam-
ined later in section 12.3.2.
It will be appreciated that whilst this procedure would enable the positions
of faults to be determined, it has the same limitation as existing distance type
protective schemes, namely that it could not be set to operate immediately in
the event of all faults on a protected line. Provision would have to be made
for inevitable errors and it would be necessary to have settings covering two
or more zones.
Travelling waves caused by faults external to a protected line

It will be clear that travelling waves may be initiated at many points on an
interconnected network when faults or other disturbances occur, and it is
essential that any protective scheme which analyses these waves should be
able to discriminate correctly.
It has been shown earlier that faults on circuits connected to busbar A in
Fig. 12.4(a), other than those on line P, would cause travelling voltage and
current waves of opposite polarities to be sensed by a scheme associated with
line P, whereas faults on line P would set up waves of the same polarity.
This situation would not obtain, however, for faults on the network which
would cause travelling waves to be propagated from busbar B to busbar A.
Other checks, such as those described later in section 12.3.2, could be made
to ensure correct discrimination between such faults and those on the protected
line.
The voltage and current variations at the end of a line when a fault occurs on it
The effects of reflections on the initial values of travelling waves were
considered under section 12.2.5. As stated in section 12.2.3, however, the
transient voltages and currents at points on a line vary with time after faults
occur on it.
When a fault occurs on an a.c. line, therefore, the transient voltages and
currents at its end may have step changes at intervals and sinusoidal variations
between the steps. For simplicity the behaviour obtained when a short circuit
occurs on a loss-free line connected to an infinite source is examined below,
the circuit being shown in Fig. 12.5(a).
As stated earlier, a travelling voltage wave (VI) with an initial value of
- Vpk sin a. would propagate towards busbar A, the voltage at the fault position
at the instant of fault occurrence being Vpk sin a.. A travelling current wave
(il) with an initial value of - Vpk sin alZo would accompany the voltage wave.
The waves would reach busbar A at a time tl =x/u, after the incidence of the
o-xp
fault, U being the velocity of light.
Busbar A x Busbar B
X~x-
1-x-
LineP
Infinite Short
source ]ircuit
(a)
(b)
i,+i" i,+i"t
t, I t,
t
O~I
-2Vpk -2VPk : - -
-
Zo (c)
Zo
i t, 3t, 5t, 7t,
0
(d)
Fig. 12.5 Current variations at the end of a line when a fault occurs on it.
Because of the presence of the infinite source, no change in the level of the
voltage of the busbar would occur and therefore a reflected voltage wave
(Vrl) with an initial value of Vpk sin a would be propagated along the line
towards the fault, as shown in Fig. 12.5(b). The accompanying current wave
(irl) would have an initial value of - Vpk sin a/Zo, taking current flow into the
bus bar from the line to be positive. The current transformer in the line near
busbar A would therefore carry a current increment with an initial value of
h + Irl which would be equal to - 2 Vpk sin a/Zoo
These conditions are illustrated in the Bewley Lattice diagram shown in Fig.
12.5(b). The current in the primary winding of the current transformer would
then vary sinusoidally, i.e. i l + i rl = - 2 Vpk/ZO sin (rot + a). For faults occurring
at instants of voltage zero and at peak voltage, i.e. a =0 and 1t/2, the currents
would vary as shown in Fig. 12.5(c). The waves Vrl and i rl would travel to the
fault where they would be reflected, the initial values again being the same as
those of the arriving waves, the polarity of the voltage reversing but that of
the current remaining the same. The waves Vr2 and ir2 would then travel to
busbar A, causing further reflections, i.e. V r3 and i r3 The increment in the
current on this occasion would again be - 2Vpk sin a/Zoo This process would
continue indefinitely on a loss-free circuit and currents of the form shown in
Fig. 12.5(d) would result.
It will be seen from Fig. 12.5(d) that the polarities ofthe current increments
would always be the same as that of the initial voltage wave arriving at busbar A,
i.e. VI> and that the current would have a finite value after a short time even though
a fault occurred at an instant when the voltage at the fault position was zero.
It will also be seen that currents would reach high values very rapidly if
faults occurred near the end of a line, i.e. when x and therefore tl would have
small values. Clearly the build up of current would be slow when faults
occurred at more remote positions.
In practice, the travelling waves would attentuate as they travelled because
of the line losses and the successive steps would reduce and steady state
sinusoidal conditions would eventually be achieved. At this time, the current
levels would be controlled by the line series impedance and again the current
magnitude would decrease with increase of the distance to a fault, i.e. high
currents would flow for close-up faults and lower currents would flow when
more remote faults were present.
It will be appreciated that practical busbars are not infinite and that their
voltages are affected by the arrival of travelling waves initiated by faults. As
a result, the reflected currents do not have the same magnitudes as those of
the incoming waves and therefore currents steps of 2Vpk sin a/Zo are not
produced. Nevertheless quite rapid build-ups of current are produced when
short-circuits occur on a line at points near the busbar and in this connection
it is worthy of note that the time interval between the current steps would only
be 0.2 ms for a fault at a distance of 30 km from the busbars, i.e. tl = 0.1 ms
for propagation near the velocity of light.
12.2.6 Travelling waves on three-phase lines
In the preceding sections, the propagation of travelling waves has been con-
sidered on d.c. and single-phase bases for simplicity. In such situations a line
has a single characteristic impedance (Zo) and velocity of propagation (u) and
under ideal loss-free conditions these parameters are given by:
and u= -V LC1 mls
in which Land C are the inductance and capacitance per unit length of the
line.
The value of Zo is dependent on the diameter of the conductors and the
spacing between them. The inductance increases with increase in the spacing
and decrease in the conductor diameter, whereas such changes have the oppo-
site effects on the capacitance. As a result, the characteristic impedance be-
haves similarly to the inductance, i.e. it increases with increase in the spacing
and decrease in the conductor diameter. The velocity (u) is not, however,
affected by the line spacing and the conductor diameters, and under ideal
conditions the velocity of propagation is the same as that of light.
The behaviour of three-phase lines is, however, much more complex. Firstly
it must be recognized that some waves propagate only in the line conductors,
i.e. say the phase a and b conductors in the event of an interphase fault between
phases a and b. Such waves are said to propagate in an aerial mode. In the
event of a single-phase to earth fault, however, the current waves flow in a
single line conductor, say the phase a conductor, and return in both the earth
conductor or conductors which run above the line being supported on the tops
of the towers, and also in the ground. These waves are said to propagate in
the ground mode.
Aerial-mode propagation
Because the conductors of major transmission lines have very low resistances
per unit length and the leakage conductances between the phase conductors
are also very low, aerial-mode waves travel at almost the velocity of light and
therefore any errors caused by using the velocity of light to assess the positions
of faults in the manner described in the previous section would be very small.
The characteristic impedances presented to the waves set up when the
various types of interphase faults occur will not all be the same because of the
asymmetric positioning of the phase conductors and the fact that conductors
are not transposed on major lines because of the practical difficulties that
would be entailed in so doing.
Although the phase conductors of a line are each the same, the interphase
spacings vary with the configuration employed. As an example, for a horizon-
tal arrangement with equal spacings between the centre and outer conductors,
the spacing between the outer phases, say 'a' and 'c', is twice that between
the other pairs of phases, 'a' and 'b' and 'b' and 'c'. In this case the charac-
teristic impedance for propagation on the outer phases (Zoac) would be greater
than those for the other phases, (Zoabo Zobc), which would both have the same
values. Because the inductances and capacitances per unit length are propor-
tional to the logarithms of the spacings and their inverses respectively, the
characteristic impedances are also proportional to the logarithms of the spa-
cings. They therefore do not vary as greatly as the spacings. For given lines,
they can be readily determined and the appropriate values could be used when
the particular interphase faults present have been identified from the initial
travelling waves.
Ground-mode propagation
As stated above, voltage and current waves are propagated in parallel along
the overhead earth conductors and in the ground when phase to earth faults
occur on a line. Because the earth conductors are effectively connected to
ground at each tower both sets of waves must propagate at the same velocity.
Clearly, in the absence of the ground, the waves would travel in the faulted
phase conductor and the overhead earth conductor and they would thus be of
the aerial mode. In the absence of earth conductors, the waves would have to
travel in a phase conductor and the ground. The current in the ground tends
to follow the route of the line but it is not confined to a particular cross-
sectional area and it is affected by the resistivity of the ground. In these
circumstances, the inductance per unit length of the conductor-ground loop
cannot be calculated simply from basic principles.
To assist in studying telephone and telegraph circuits in which signals were
transmitted via a single conductor and returned through the ground, W. Carson
produced a model circuit [6] in which the ground was represented by an image
conductor, of the same diameter as the aerial conductor, at a depth in the
ground given by:
dg=660~ m
in which p is the average resistivity of the ground and f is the system fre-
quency. For typical resistivities usually encountered, i.e. 10000m, the depth
of the image conductor in power applications, i.e. at frequencies of 50-60 Hz,
is 933 metres.
Using this model, the inductance per metre (L) of a conductor above ground
is given approximately by:
lio dg
L=-ln- Him
1t d
in which d is the effective diameter of the overhead conductor and dg is an
effective height.
This modelling does not enable the capacitance between a conductor and
ground to be determined satisfactorily, and for this purpose it is common to
assume that the surface of the ground is an equipotential, i.e. at zero volts.
The capacitance is then determined assuming an image conductor with a
charge equal and opposite to that on the overhead conductor, the depth of the
image conductor below the ground surface being equal to the height of the
aerial conductor above the ground (h).
Using this technique, the capacitance to ground (C) of a conductor per metre
is given by:
REo
C = In 4h/d F/m
in which d is the effective diameter of the aerial conductor and h is its height
above ground.
Because the depth of the image conductor (dg ) in the model used to deter-
mine the inductance is greater in practice than the depth of the image conduc-
tor in the capacitance model (h), the velocity of propagation of travelling
waves which flow in the ground tends to be significantly lower than the
velocity of light. As an example, for the line with the configuration shown
in Fig. 12.6, the velocity of propagation for waves travelling in a phase con-
ductor and returning through the ground and the overhead earth conductor
is 1.88Hf mis, i.e. about two thirds of the speed of light. It will be appreci-
ated that such velocities are so far removed from the velocity of light that
fault positions calculated using the latter value would be too inaccurate
and therefore the actual values for particular lines must be calculated and
used.
The values of inductance and capacitance determined in this way must also
be used to calculate the ground-mode characteristic impedance.
1015'1015'1
1 111 11:
: -: I4+j
- --1- t- + -$ -I--- Earth wires
11 ~~ ~ ~ ~ Phase conductors
1 1 1
15 1 1 1
1 1 1
~'77;;0'77.mm77;~ Ground
Fig. 12.6 Typical line configuration. Dimensions in metres, steel earth wires 9.5 mm
diameter, phase conductors 76 mm effective diameter, resistance = 25 mO/km.
Modal signals suitable for protective schemes
The general equation relating the steady state voltage drops and currents in
three-phase circuits is of the form shown below:
The use of this equation involves considerable computation because the Z

matrix contains nine non-zero elements, i.e. three self impedances and six
mutual impedances.
For a completely symmetrical circuit in which the three self impedances
CZaa, Zbb, Zee) are equal and the six mutual impedances are equal CZab. Zae,
Zba, Zbc, Zea, Zeb) the above equation can be transformed to diagonalize the
impedance matrix so that it only contains three non-zero elements.
A very well-known transformation, dealt with in Appendix 3.2, is that
associated with symmetrical components, i.e. positive-sequence, negative-
sequence and zero-sequence. For steady state sinusoidal conditions this equa-
tion is of the form:
[ ~:l =
Vo
[Zl1 Z22
Zoo
] . [~:]
10
in which Vh V2, Yo, 11'/2'/0, ZII, Z22 and Zoo are the positive-, negative- and
zero-sequence voltages, currents and impedances respectively.
A great advantage of this particular equation is that there are no mutual
effects between the sequences and as a result the three rows are independent
of each other.
The relationships between the positive- and negative-sequence voltages and
currents and the actual phase quantities are complex, as shown in Appendix
3. As an example
in which la, h and Ie are the phase currents and
a=!C-l+
2
JV 3)=11 21t
3
rad
It will be appreciated that instantaneous voltages and currents cannot

be transformed in the above manner. There are, however, a number of trans-
formation matrices containing only real numbers which may be used. A trans-
formation that has been widely employed is:
V~ = [l~
[Va] t - _l]~. [va]
0
1 Vb
Vo 1 liVe
The same transformation applies to the currents and the following relationship
between the voltages and currents associated with travelling waves is pro-
duced.
(12.11)
It will be clear that three independent propagation-modes of waves could

be used in studies of circuits containing symmetrical overhead lines. Two of
these modes, i.e. a and ~, would be associated with the phase conductors and
they are thus referred to as aerial modes. The current associated with the other
mode, i.e. 0, which is the same as the zero-sequence mode, must at least flow
partly in the ground and it is therefore referred to as the ground mode. It will
be clear that a single characteristic impedance and propagation velocity could
be determined for each mode.
It must be recognized, however, that the conductor configurations of major
overhead lines are always asymmetric and it is not the practice to transpose
the conductors at regular intervals along the lengths of line. As a result, the
self and mutual inductances and capacitances per unit length of lines are not
the same for each of the phases. The differences are not great, however,
because the various quantities are dependent on the logarithms of the spacings.
If the actual values were used to produce the elements in the Z matrix of
equation (12.11), all of them would have non-zero values. Those on the
principal diagonal, i.e. Zan, Z~ and Zoo, would, however, have larger values
than the others and limited errors would be produced in determining the
voltages and currents, if the elements not on the diagonal of the Z matrix were
regarded as zero.
It will be appreciated that the same approximation is usually made when
symmetrical-component techniques are used, i.e. it is assumed that the sequen-
ces are independent of one another. Various other transformations are
now available which allow the errors caused by neglecting the physical asym-
metry of overhead lines to be reduced. Some are considered in detail in
Appendix 3.3.
12.3 PROTECTIVE SCHEMES WHICH DETECT

TRAVELLING WAVES
As indicated in the preceding sections, it is possible to determine the direction

of a fault from the first travelling current and voltage waves which arrive at
Protective schemes which detect travelling waves 457
the end of a line and directional-comparison schemes using this technique
have been produced. Two of them are considered in the following section.
12.3.1 Directional-comparison schemes

During the 1970s, the possibilities of producing protective schemes based on
travelling waves were being studied with the object of obtaining very rapid
detection and clearance of faults. Such relays, which were designated as
UHSR, i.e. ultra-high-speed relays, were later defined to include schemes
which would initiate the opening of circuit-breakers in one quarter of one
cycle in 60 Hz systems, i.e. in 4.17 ms. Some of the schemes which have been
produced are described below.
RALDA Scheme produced by ASEA

The Bonneville Power Administration in the United States launched a deve-
lopment programme and awarded a contract to ASEA in Sweden, in July 1974,
to develop a UHSR system using the travelling-wave approach. The contract
specified an operating time of 4 ms or less for close-up faults and 8 ms or less
for all other faults.
As a result, after optimizing the relay performance using digital and ana-
logue simulation techniques, a scheme given the designation Ralda was pro-
duced. The scheme was described in detail in a paper [1] by Chamia and
Liberman which was presented at the IEEE PES Winter Meeting in January
and February 1978.
The scheme, represented in block diagram form, is shown in Fig. 12.7. The
analogue interfaces were designed to accept signals from either conventional
or other types of current and voltage transducers and to provide both isolation
and surge immunity.
As stated earlier in section 12.2, it is necessary that the steady state voltages
and currents which would have been present in the absence of a fault be
subtracted from the voltages and currents which are present during faults so
that the transient superimposed qualities are obtained for processing. In the
Ralda scheme, this was achieved by including steady state frequency suppress-
ors which had transfer functions which ensured that the initial changes caused
by faults were transmitted without significant time delays. Amplifiers were
included to provide signals of the levels needed for subsequent processing.
These signals were fed to directional detectors I and D and with the conven-
tions adopted a tripping direction signal (TD) was produced by detector D if
the polarities of the voltage and current signals present after the incidence of
a fault were different. This corresponded with the voltage and current waves
approaching the end of the line being of the same polarity, i.e. the condition
for a fault on the protected line, as shown in section 12.2.2. In the event of
the polarities of the signals being the same a blocking directional signal (B D )
was produced.
IT
End B
Transmission line
)
ror'
~
,..u..L.L..--=-L..I...l:~ - ,
Information link
R R
x
Tr Tr
R= Receiver
Tr = Transmitter
Tripping and Tripping and

signalling signalling
Fig. 12.7 Block diagram of the RALDA scheme. (Reproduced from Chania and Liber-
man, 1978, IEEE Trans PAS-97, with the permission of the IEEE.)
To ensure that the above tripping and blocking signal outputs would not be
produced by minor disturbances on a system, for example, by a sudden loss of
one of the output voltages of the voltage transducers, the directional detectors
D were so arranged that the magnitudes of the transient current and voltage
inputs should both be above certain set levels before they provided output
signals.
Clearly this process had to be performed separately for the quantities
obtained from each of the three phases of a protected line and the outputs from
the directional detectors D were fed to logic circuits. These determined the
sequence of the To and Bo signals produced after the incidence of a fault to
establish whether the first signal was To or Bo. This was done to avoid errors
which could otherwise have arisen if the signals changed because of the arrival
of later reflected waves.
The production of the TD and BD signals energized the communication link
between the ends and either permissive or blocking signals could be sent and
clearly tripping of the circuit-breakers at both ends of the protected line was
initiated if TD signals were present at both ends of the line or it was prevented
if a BD signal was present.
Because of the time delay introduced by the communication link, the time
between the incidence of a fault and the initiation of tripping of the circuit-
breakers was about 8 ms.
To enable the operating times for close-up faults to be shorter, the direc-
tional detectors I were included. These detectors only operated if the magni-
tude of the transient current was above a level which could only be present
for faults within a short distance along the protected line. For such a condition,
the current increases rapidly, as explained under section 12.2.5 (page 445),
because many waves travel between the fault and the end of a line in a very
short time. Operation of a directional detector I produced a TI signal which
directly initiated the tripping of the circuit-breaker at its end of the line without
energizing the communication link. This mode of operation was referred to as
the independent mode and this led to the designations directional detector I
and signal TI Its use enabled tripping of the circuit-breaker to be initiated in
4 ms or less.
The other mode of operation was referred to as the dependent mode and this
led to the terms directional detector D and signals TD and BD , used above.
It will be appreciated that the directional detectors I had to be set so that
they would only operate in the event of faults on a section of protected line,
i.e. they had to under-reach, whereas the directional detectors D had to operate
for all faults on the protected line and in consequence they had to over-reach.
It was necessary therefore, when applying the Ralda scheme, to do extensive
fault studies to ensure that the directional detectors I and D were set at levels
which would ensure correct operation.
The scheme was so arranged that the types of fault present were determined,
e.g. phase a to earth, phase b to c, three phase, and the logic circuits then
initiated the appropriate circuit-breaker operation, i.e. single-pole or three-
pole opening.
The communication equipment incorporated a single-tone transmitter-re-
ceiver which operated at a carrier frequency of 72 kHz I kHz, its general
arrangement being as shown in Fig. 12.8. This was an expensive arrangement
with four channel spaces which was suitable for use in schemes for application
to short lines where rapid fault clearance was needed. For longer lines where
8-10 ms relaying times would be acceptable in the event of faults near the
remote ends, an audio-tone communication equipment operating on a single
voice channel was adequate.
During the later stages of testing the prototype equipments, part of the
Bonneville Power Administration's 500 kV system was modelled on an ana-
logue transient-system simulator. System voltages and currents for a large
tii
> .... CD ....
.- CD
CD_ .!!l CD
O~
~:E Z-
"ON
CI a:
til..>:: N
U'<tI
(/)N..>::
0 .... ,...
uj ~+I o 0 <0
CD
u...>:: ~ T I
tii W8P lndlnO
uj .~
o
...J
CD
U.o Jall!J 1::I1\~1::I
a: ~
E
e
u.

o
...J
~
~
Fig. 12.8 Block diagram of the communication equipment used with the RALDA
scheme. (Reproduced from Chania and Liberman, 1978, IEEE Trans PAS-97, with the
permission of the IEEE.)
number of conditions were recorded and these were later used to provide
inputs to the Ralda relays.
Subsequently two Ralda terminals were installed on the 500 kV John Day-
Grizzly No. I line of the Bonneville Power Administration in April 1976. The
setting procedures and one year's operational experience together with the
performance of the relays during staged fault tests were summarized in a paper
[7] presented by Yee and Esztergalyos at the IEEE PES Winter meeting held
in January and February 1978. In the conclusion of the paper it was stated that
the collected field data, combined with laboratory test results, prove that the
UHSR system is directional and that it has consistent high-speed performance
from 2-6 ms. It was further stated that the UHSR system reliability in the
forward direction and in the reverse blocking mode is 100%.
Since that time many series-compensated lines in the United States have been
protected by Ralda schemes and satisfactory behaviour has been obtained.
Scheme developed by the University of Bath and GEe Measurements

This scheme [2, 8] processed the transient currents and voltages produced at
the ends of lines by faults and discrimination was obtained using carrier-block-
ing type communication equipment. Although it was basically similar to the
Ralda scheme, produced by ASEA, the processing techniques were different
and it did not detect the directions of faults relative to the ends of lines by
directly comparing the polarities of the incoming travelling waves of voltage
and current initiated by faults.
It will be clear that if a fault occurs beyond end A of an ideal line as shown
in Fig. 12.9(a), i.e. a reverse direction fault, then the resultant change of
voltage at the fault position would cause travelling waves of voltage (VI) and
End A End B
Line P (characteristic
X___i;;,;m.:.!:p..:.ed~a:;.n..:.ce:.....::lz911i-)_ _ _X
-V1
-i1
-1 h
Fault - Reference direction
for current
(a)
End A End B
x- _
Reference direction
for current
V,1 LineP V1-
X
- i,1 i1 - -=- Fault
(b)
Fig. 12.9 Voltage and current waves for fault at ends A and B.
current (il) to be propagated into the protected line P. Taking the current flow
from busbar A to busbar B to be positive, the transient current would be given
by:
. VI
11=-
Zo
This condition would persist until the waves had travelled to busbar Band
then returned to busbar A, the elapsed time being:
Signals 51 and 52 are given by:
and
with values of 51 = 0 and 52 = 2vI throughout this period.

In the event of a fault on the protected line or beyond its remote end, a
travelling voltage wave (VI) would be propagated towards busbar A as shown
in Fig. 12.9(b). The accompanying current wave would be given by:
. VI
11=-
Zo
with the above reference direction for the current.
Because several circuits would be connected to busbar A it would present
an impedance RA to the waves which would not be equal to the characteristic
impedance of the line (Zo). As a result reflections would occur at the busbar
and travelling waves Vrl and i rl would be propagated back along the line
towards busbar B.
As shown under section 12.2.5 (page 445), the incoming and reflected waves
would sum at busbar A to produce voltages and currents given by:
2RA VI
Z
VSI=VI+Vrl=-R
A + 0
and
. .. 2VI
lSI =II + IrJ =- -A RZ
+ 0
The voltage and current transducers near busbar A would not detect the initial
incoming waves but would give outputs proportional to the above quantities
(VSI and i sl )'
Under these conditions the signals SI and S2 would have the values:
and
Comparison of these values with those obtained during reverse faults shows
that discrimination can be obtained because the magnitude of S1 is less than
that of S2 for a reverse fault but it is greater for faults on a protected line or
beyond it.
Clearly both signals would have zero magnitudes initially for both external
and internal fault conditions which occurred at instants of voltage zero but the
signals S1 and S2 would grow to significant levels for a fault on the protected
line before further reflected waves arrived, i.e. after a time of 1 lu s.
This detection process was used in the scheme developed at the University
of Bath and it was implemented using the a, ~, 0 modal quantities referred to
under section 12.2.6(page 452).
A block schematic diagram of the equipment required at one end of a line
is shown in Fig. 12.1O(a). The whole arrangement was initially analogue based
and the outputs of the voltage and current transducers were fed to mixing
circuits to produce the a, ~ and 0 components of both sets of quantities.
The asymmetry of the line configuration was neglected and a single char-
acteristic impedance was used for each mode of propagation, i.e. Zaa, Z~~,
Zoo and in fact these impedances were assumed to be purely resistive in the
surge-replica circuits. These circuits were fed with the three current compo-
nents and the quantities Zaa ia, ZI3I3 il3 and Zoo io were produced for subtraction
from and addition to the components of the voltages to obtain quantities such
as Va - Zaa ia and Va + Zaa. ia. These quantities contained steady state terms as
well as extra transient terms caused by faults and therefore the transient
components were extracted in the superimposed component circuits. This was
achieved by feeding differential amplifiers shown in Fig. 12.10(b), with the
V - Zi and V + Zi signals. A time delay of one period of the power system
frequency. i.e. T= l/f. was introduced in the connection to the negative input
of each of the six differential amplifiers so that the output would be the
difference of the input and its value one cycle earlier. This clearly provided
the transient components only for the first period after the incidence of a fault
and this was considered satisfactory because the fault direction would have
been determined within this time. The signals produced by pairs of these
circuits. e.g. Sal and Sa2 were fed to detectors which determined the fault
directions indicated by each of the modes and either trip or block outputs were
obtained. These outputs were fed to 'OR' gates so that a single trip or block
signal was provided by the equipment at an end of a line. In the event of a
fault being detected on the line or beyond its remote end, tripping of the local
circuit-breaker was initiated unless a blocking signal was obtained from the
communication link.
To evaluate the performance of the scheme the conditions which would
obtain on the network shown in Fig. 12.11 were determined using a digital
simulation technique described in reference [2]. Real-time signals proportional
to the system voltages and currents calculated for a wide range of different
conditions were supplied to a prototype relay in which the replica surge
d1+ 1
ystem R X r.r.
R :-.~ .....
<>
~ lCi[--""":
,Jl. -=- .ff-
...--------':;:- :;:-
m
,.g:
-co
.ff-
Current- Voltage
mixing mixing
Similar arrangement
-
circuits circuits
-
to end R
~
N
,.g: M
,.g:
Surge
replica
--
circuits
;::. (\j
MII:
oc II:
'N 'M
ri rF rF
--
N C\i
jf ,.g:
t..:,-:J ~ ~
To similar arrangement r:r:.0 r:t To similar arrangement
for mode-1 signals ;::::. for mode-3 signals
N N
Trip outputs from .ff-.ff- Block outputs from

modes 1 and 3 modes 1 and 3
channels
~
I I
I I
Sequence
detector
(mode-2)
Trip output Block output
- }Similar
Tripping , ,- arrangement
signal ,-'-
'< , ~---I to end R
Trip relay
Fig. 12.10 Block diagram of the Bath/GEe scheme, (Reproduced from Johns, 1980,
lEE Proc C, 127 with the permission of the lEE,)
impedances were set to be proportional to the real component of the actual
characteristic impedances, the values of which were ZWJ. =Z~~ =266 nand
Zoo =585 n.
Time
\
VR2 (t - T) - R02 iR2 (t - T)

VR2(t) - R 02 iR2 (t)
51R2
Fig 12.10 (b)
Fig. 12.11 Schematic diagram of network configuration. (Reproduced from Johns,

1980, lEE Proc C, 127 with the permission of the lEE.)
5 5
.s end
(i) (i)
4 ___ _-- end .s 4
E
Ol 3 --;><'::'---)R
~~ __
+:
Ol
E 3 _--;_..-::.':::._----_.
__-----~A
~ ... fIII'" ........
C>
~~~
c S
.~ 2 >S '0. 2
a. a.
~ a-earth ~ b-c
o~~--~~--~--~-- 0
o 20 40 60 80 100 0 20 40 60 80 100
Distance from R to point of Distance from R to point of
fault, % of line length fault, % of line length
5
(i)
.sOl
4
end
E 3
+:
c
C> _---------./R
'0. 2
a.
c "-...S
I-
b-c-earth
O~~--~~~--~--
o 20 40 60 80 100
Distance from R to point of
fault, % of line length
Fig. 12.12 Variation of relay operating times with fault position. (Reproduced from
Johns, 1980, lEE Proc C, 127 with the permission of the lEE.)
The scheme operated satisfactorily under all conditions and operating times
under 4 ms were obtained, the actual times for a range of conditions being
shown in Fig. 12.12.
Unlike impedance-measuring schemes, directional-comparison schemes are
not directly affected by the resistances present in fault paths, but nevertheless
the magnitudes of the travelling voltage and current waves set up by faults
reduce with increase in the fault resistance. Tests were done on this scheme
to determine the limiting fault resistances at which correct operation could be
obtained. Understandably, the values were dependent on the position of the
fault on the line and also the point in the cycle at which it occurred.
It was found that the limiting values were in the range 40-600 nand
therefore correct operation would always be obtained in the event of faults
with resistances below 40 n, a level considerably above that which could be
accepted by a distance-measuring scheme. A high-speed digital directional-
comparison relay using a signalling channel based on the above principles is
described in reference [9]. Typical operating times of half a power frequency
cycle plus signalling delays are possible with this relay. In addition to normal
relaying functions, the relay features continuous self-monitoring and diagnos-
tic facilities, and can be interrogated from a remote location via an RS232
serial communication link.
Scheme developed by Asea Brown Boveri

This scheme was described in paper [to] by Vitins which was presented at the
IEEE PES Meeting held in February 1980. It employed a correlation technique
which had been described in a paper [11] presented at the IEEE PES Meeting
held in July 1977.
The algorithm proposed by Vitins determined the trajectories followed by
the transient voltages and currents caused by faults (~v and ~i) in the ~v,
R~i plane. This technique is a variation of the directional-sensing processes
described earlier, in which the polarities of the voltage and current increments
caused by faults are compared. In such schemes, the polarities of the first
increments of voltage and current produced at a busbar in the event of a fault
either on a protected line or beyond it, i.e. in the forward direction, are the
same whereas they are different when faults occur in the reverse direction, i.e.
for a fault in the forward direction, ~v and ~i are either both positive or both
negative. Such quantities when plotted in the ~v, R~i plane must therefore
lie in the first and third quadrants for faults in the forward direction and in the
second and fourth quadrants for faults in the reverse direction. They move
when the first waves arrive at the end of the line and take up positions
dependent on the angle in the voltage cycle at which the fault has occurred.
As subsequent increments are produced by later waves new positions are
taken up and the trajectories follow a clockwise path when faults are in the
forward direction and an anticlockwise path when faults are in the reverse
direction.
By setting boundaries in the ~v, R~i plane of the form shown in Fig. 12.13,
discrimination between forward and reverse faults can be obtained by deter-
mining whether the boundaries in the second or fourth quadrants are crossed
before the boundaries in the first or third quadrants. In addition, the fault
direction can also be determined by detecting the direction of the trajectory.
This has the advantage that discrimination is possible at times after the arrival
of the initial waves.
The above principle is employed in the relay, designated LR91, produced
by Asea Brown Boveri [12]. This scheme, like that described under section
12.3.1 (page 457), incorporates current detectors so that should current levels
be so high during a fault that it could not be beyond the end of the line
protected, then tripping of the circuit-breaker would be initiated immediately
if the directional detector indicated a forward fault. In these circumstances
operation of the circuit can be initiated in less than 5 ms. In the event of more
remote faults then the directional detection process is used and tripping of the
circuit-breakers is only initiated if the equipments at both ends indicate that
the faults are in their forward directions.
Fig. 12.13 Operational characteristics of the ABB scheme.
The burdens imposed on the voltage and current transducers are very low.
Schemes of this type have been installed since about 1985 and they are now
in use in several countries including China where they have been applied to
several lines.
12.3.2 Schemes which determine the positions of faults

It was indicated in section 12.2.5 (page 445) that the distance of a fault from
a relaying position may be determined by measuring the time interval (2t l )
between the arrival at the end of a line of the first travelling wave caused by
a fault and the next wave which results from reflections at the busbar and the
fault, i.e. the time between the voltage increments VSI and VS2 shown in Fig.
12.4(b).
Because many travelling waves may be present on lines during fault condi-
tions, the increment VS2 may not always be the next increment after VSI and
therefore, to ensure that fault positions will be determined correctly, it is
necessary that protective schemes operating on this basis should identify the
increment VS2 '
Extensive studies have been conducted on two schemes incorporating dif-
ferent methods of wave identification and these are described in some detail
in the following sections.
Scheme developed at the University of Cambridge

This scheme, which was described in a paper [3] by Crossley and McLaren,
presented at the IEEE PES Meeting in January and February 1983, employed
similar techniques to those described earlier (section 12.3.1 (page 457)).
The signals SI and S2 given by
and S2= USI + i SI Zo
_I
i
... 500
G"" -'000'- -X--- -Xf~laYing pOint .

L' Q BUSbar30
X- Ilne---X- -'lflf1f" - ...
35 GVA X Line P X -=- F2 10 GVA
f::\
0-'000'-- X------x
FI
1 -
f::\
x----x --'lflf1f"-O
10 GVA Busbar A Busbar B 5 GVA
I.. _I
500
(a)
41.510
I Earth wire
31.23
1 C
1
1
1
1
____ '- ________ 1 b
20.87
1 .1.
E 1 1
1 1
:>: 1 1
1 1
1
1
12.03 --------~ a
1
1
1
1
1
1
1
1
~LI-6.~7-8----8~.~-5-----9.-9-8-----
X,m
(b)
Fig. 12.14 Schematic diagram of network configuration. (Reproduced from Christo-

poulos, Thomas and Wright, 1988, lEE Proc C, 135 with the permission of the lEE.)
were obtained from the aerial-modal quantities and the directions of faults
were determined from the first travelling waves of voltage and current which
arrive at the relaying point after the incidence of faults. Thereafter the cross-
correlation technique described in the above paper was used to identify waves
which have been reflected at the busbars, returned to the fault and then
travelled back to the busbar.
The assumption was made that the largest peak in the cross-correlation
function was caused by the second incident wave (VS2 and I s2 ). The positions
of faults were then determined by timing the interval between the first and
second incident waves.
To assess the performance of the scheme, the voltage and current variations
which would be obtained on a single-circuit, three-phase line with two infeeds
were computed for a wide range of conditions and an off-line program was
then used to simulate the relaying processes.
The results obtained indicated that satisfactory performance should be ob-
tainable. In further studies which were done to evaluate a number of ultra-
high-speed relay algorithms [13] it was found, however, that the necessary
discrimination would not be obtained under some conditions. Enhancements
to the original algorithm have been suggested in references [14] and [15].
Scheme developed at the University of Nottingham
This scheme was described in considerable detail in papers published by the

Institution of Electrical Engineers in 1988 and 1989 [4, 5]. For simplicity the
scheme is considered below on a single-phase basis.
It was shown earlier (section 12.2.5 (page 445)), that a great deal of infor-
mation can be extracted from the first travelling waves of voltage and current
which reach the end of a line after the incidence of a fault on it and the later
waves of voltage and current caused after reflections at the busbar and the
fault.
On arrival of the first waves of voltage and current, with initial values of
VI and 110 reflections occur at busbar A and the increments VSI and lSI are
produced, their values being as given in equations (12.5) and (12.6). Their
ratio (Vst!/ sl ), as shown in equation (12.7), is the surge impedance presented
by the bus bar A (R A). In practice, because RA may not be entirely resistive, the
reflected wave from the relaying point may be obtained directly from the
expression Vr = (V + IZ)/2, without the need to calculate the reflection coe-
fficient at the busbar which is, in general, time-dependent. This is explained
more fully in reference [5].
As shown in section 12.2.5, the increment VS2 caused by reflection of the
wave Vrt at the fault would have an initial value given by:
VS2 = kvf (l + k vl ) Vrl
in which kvf is the voltage reflection coefficient at the fault. The magnitude of
this coefficient can thus be found and because it is given by equation (12.8),
i.e.
[Zo Rr/(Zo - Rf)] - Zo

kvf = [Zo Rr/(Zo + Rf)] + Zo
the value of the fault resistance (Rr) can be determined.

In addition, and most importantly, the time interval between the voltage
increments VSI and V S2 enables the distance x to the fault to be determined from
equation (12.10).
A check must be made, however, to ensure that the second increment which
occurs at the busbar has been caused by the reflected voltage Vrl being re-
flected at the fault.
Referring to Fig. 12.4(b), it can be seen that the fault must have occurred
at a time tl seconds before the arrival of the first travelling wave at the busbar,
i.e. half the time between the increments VSI and V s2 A knowledge of the
voltage and current conditions at the bus bar A end of the line, prior to the fault,
would enable the voltage (Vf) at a point distance x along the line at the end of
fault incidence to be determined using the well-known long-line equations.
Substitution of this voltage in equation (12.1) would allow the value of the
fault resistance to be calculated again and obviously this value should corre-
spond closely with that derived earlier.
In the event of a fault near the remote end of a line, an increment VS3 would
arrive before the increment VS2 as shown in Fig. 12.4(c) and should the second
value of the fault resistance be determined using the time elapsed between the
increments VSI and VS3 then the value obtained would not agree with the earlier
value. In these circumstances, the process would be repeated later when the
increment VS2 arrived and the necessary agreement would be obtained.
It was proposed that the scheme should be implemented by processing
digital samples of the sums of the two aerial-mode quantities and the ground-
mode quantities associated with three-phase lines, i.e.
(12.12)
It was further proposed that the values of fault resistances would be deter-
mined using the time intervals between the increments present in the aerial-
mode phase voltages because the velocity of propagation associated with the
aerial mode is constant and near that of light, whereas the propagation velocity
of the ground mode is dependent on the resistivity of the ground, which may
vary along the length of a line and it may also be affected by climatic condi-
tions.
To evaluate the perfonnance of the proposed scheme its behaviour when
applied to transmission lines connected in various power networks was
studied. The various voltages and currents at the end of a line at which the
scheme was assumed to be mounted were determined for a variety of fault
conditions, using a frequency-domain, power-system simulation program pro-
vided by GEe Measurements. These quantities, which were sampled at 25 kHz
in all numerical computations, were processed as described above.
Studies were based on the network shown in Fig. 12.14(a) in which the lines
were assumed to be single circuit and untransposed with the conductor con-
figuration shown in Fig. 12.14(b). The protective scheme being studied was
taken to be associated with line P and to be mounted near bus bar A. Faults at
various points on the network were simulated and the responses of the scheme
were determined. In each case, the sampled voltages and currents were added
to the corresponding values of the previous half cycle and in this way the
presence of a disturbance was indicated by a set of non-zero values and these
and the subsequent values were the transient quantities produced by the dis-
turbance. The third set of non-zero quantities were taken to represent the
magnitudes of the initial voltage and current increments (VSh lSI) at the re-
laying point.
A set of aerial-mode transients caused by a symmetrical, three-phase fault
on a line P (of length 200 km) at a point 140 km from busbar A are shown in
Fig. 12.15. These were obtained assuming fault resistances of 600 between
each of the phases, the fault occurring at an instant of voltage maximum on
phase a at the fault position. Because the fault was beyond the mid-point of
the line, the initial waves and reflections which travelled between the fault
and bU1\bar A arrived at the times tl and t3, whereas the waves which went
from the fault to busbar B before returning to busbar 1 at time t 2
The two sets of estimates obtained of the fault resistances between the
phases using the voltage and current increments at times tl and t2 were as
follows:
Phase 'a': 760 and - 568 0
Phase 'b': 00 and 4590
Phase 'c': -600 and -5160
This lack of agreement indicated that the increments were not those referred
to earlier as VSI and Vs2 The estimates obtained using the increments at times
tl and t3 were:
Phase a: 670 and 480

Phase b: 670 and 560
Phase c: 680 and 630
This agreement indicated that the increments were VSI and VS2 and the time
300
21
I" .. I
200
100
>
""'-
cD
0
~
0
>
-100
-200
-300
2.0 21
I.. ..I
1.5 t1 ~t3
1.0 a
0.5
>
""'-
E 0
~
::l
() -0.5
._........ _.-.
-1.0 '0_--.-0"",_".,
"-'-.
-1.5
-2.0
Fig. 12.15 Aerial-mode transients caused by a symmetrical three-phase internal fault

140 km from the relaying point (fault resistance 60 Q/phase, fault at phase a positive
voltage maximum). (Reproduced from Christopoulos, Thomas and Wright, 1988, lEE
Proc C, 135 with the permission of the lEE.)
interval between tl and t3 corresponded with a fault 146 km from the relaying
position, i.e. an error of 6 km.
In this connection it should be noted that the use of a 25 kHz sampling rate
introduces an inherent 40lls uncertainty into the timing estimates and an
associated 12 km uncertainty into the distance estimates, this being the dis-
tance travelled in one sampled interval at the aerial-mode velocity.
Satisfactory estimates were also obtained for phase to phase and most single
phase to earth faults. In a few cases, however, it was not possible to distinguish
between earth faults on line P and line Q when using aerial-mode quantities
but it was found that satisfactory discrimination could be achieved by incorp-
orating a further check feature based on the ground-mode waves.
Later studies based on circuits containing double-circuit lines also gave
satisfactory results.
Having established that the scheme was soundly based, it was felt that
features should be included to ensure that the necessary information would
always be extracted from the very complex waves which can be present on
protected networks and therefore the use of cross-correlation processing was
studied. As a result, an algorithm was developed which enabled correct
behaviour to be obtained for a wide range of test conditions. Details of this
algorithm, the performance obtained when using it and the influence of factors
such as window length are provided in reference [5]. Implementation of the
basic algorithms in real time using hardware based on a digital signal proces-
sor (DSP) confirms that the required processing can be done within a time
period not exceeding 2 ms and therefore a relay based upon this principle can
have a very fast operating time. To date, however, this scheme has not been
applied to an actual line.
12.4 THE APPLICATION OF ULTRA-HIGH-SPEED RELAYS TO

SERIES-COMPENSATED LINES
As stated earlier, in section 11.14 (page 431), it is now common practice to
install capacitors in transmission lines so that their effective series impedances
in steady state are reduced. In this way the phase displacements between the
voltages at the opposite ends of lines are reduced for a given power transfer
and as a result the rated power transfer over a line may be increased without
jeopardizing the stability of the associated network.
To be effective the voltages present across such capacitors at the rated
currents of their associated lines must be quite high. The voltages which would
appear across them if they carried currents above the rated values during fault
conditions would rise to unacceptably high levels and therefore metal oxide
varistors are connected across them to ensure that they will be effectively short
circuited at such times.
It will be clear therefore that the performance of protective schemes associ-
ated with series-compensated lines should not be affected by the presence of
the capacitors or their effective removal.
Because the voltages across capacitors cannot be changed suddenly, the
initial values of travelling waves are not affected by the capacitors included
in series-compensated lines. This can be seen by considering the situation
Ultra-high-speed relays 475
End A End B
-X1X----~cII-I
-X -_-:::-v-
x-
xtX-
Vc- -.
I
(a)
It~

Distance
(b)
Fig. 12.16 Conditions in a series compensated line.
illustrated in Fig. 12.16(a). When the wave of voltage magnitude v reaches

the capacitor C, the current i = v/Zo will begin to flow through the capacitor
producing a voltage Vc across it, Vc being given by:
The initial magnitudes of the waves travelling from the capacitor towards
busbar A would therefore be v and i, but at any point behind the wavefronts
the voltages and currents would reduce with time as the voltage across the
capacitor increased. The form of these variations is shown in Fig. 12.16(b).
In practice the effects produced by the voltage across capacitors during the
periods in which ultra-high-speed relays are operating is small. As an example,
if a single-phase 100 kV, 50 Hz line of length 150 km and inductive reactance
of 0.3 n /km was provided with 50% series compensation, the capacitive
reactance required would be 22.5 n, i.e. 150 x 0.3 x 0.5 n. The capacitance
would therefore be 141.51lF.
In the event of a short-circuit at an instant of peak voltage a travelling
voltage wave of initial magnitude 141.4 kV would be propagated and for a
line with a characteristic impedance of 250 n the accompanying current wave
would have an initial magnitude of 565 A. This current flowing through the
capacitor would cause the voltage across it to rise at a rate of 4 kV/ms, i.e.
2.83% of the initial voltage per millisecond. The time taken for a wave to
travel the line length would only be 0.5 ms and therefore the presence of the
capacitor would only reduce the voltage surge magnitude by 1.4% in this time.
Clearly this would have no effect on schemes which detect the directions of
faults from the polarities of the first travelling waves arriving at the relaying
positions. Even in schemes which determine fault positions from later travel-
ling waves, the effects would not be significant; for example, the changes in
the magnitudes of waves which have traversed the line length three times
would be less than 5%.
As stated in section 12.3.1 (page 457), Ralda directional-comparison
schemes have been applied to series-compensated lines and to date they have
performed satisfactorily.
The suitability of the scheme described in section 12.3.2 (page 468) for
application to series-compensated lines has been investigated in some detail.
This scheme, as explained earlier in section 12.3.2, discriminates by compar-
ing estimates of the resistances present in faults, one of the estimates being
obtained from data stored of the conditions prior to the incidence of a fault
and the other from post-fault conditions.
Because the changes in voltage across a capacitor in a short period after
fault incidence are small, its presence may be ignored in obtaining the second
estimate of fault resistance. During this calculation the time taken for waves
to travel between a fault and the end of a line is determined, and should this
indicate that the fault is between the capacitor and the relaying point, then the
first estimate of resistance obtained from the pre-fault data need not take
account of the presence of the capacitor. If, however, a fault appears to be
beyond the capacitor then the first estimate must be obtained from a three-sec-
tion model of the complete line because there may be quite large voltages
across a capacitor during healthy conditions. The capacitor forms the centre
section of the model and the two end sections represent the lines on each side
of the capacitor.
Apart from the above feature, the scheme operates in the same manner as
that described earlier for use with uncompensated lines.
Clearly the scheme could not operate satisfactorily if the voltage limiter
connected across a compensating capacitor operated within the time interval
required for the signals to be processed, but extensive investigations have
shown that this situation will not occur. It has also been confirmed that
accidental operation of voltage limiters would not generate travelling waves
of sufficient magnitude to significantly affect the operation of the scheme.
Operation of voltage limiters could occur, however, in the event of faults
external to protected lines when the series capacitors are sited near one of the
line ends. Various studies have shown, however, that these would not cause
the scheme to maloperate.
Further information about the above points and other details are provided
in a paper [16] which was presented at the IEEE Meeting held in July 1991.
12.5 THE FUTURE

It is clear that protective schemes based on processing information contained
in the travelling waves set up by faults can potentially offer very high-speed
operation. Such schemes are particularly useful on long important lines, where
a short relaying time is desirable. A number of directional-comparison
References 477
schemes based on travelling wave, or superimposed quantities are already
commercially available [9, 12] and more such schemes are likely to be intro-
duced. The need for communication between the two line ends introduces
inherent delays and additional costs.
Distance schemes where the position of the fault is determined are being
developed. In order to exploit the inherent speed advantages of such schemes,
it is necessary to employ voltage and current transducers of sufficient band-
width to allow accurate detection of the fast-rising waveforms associated with
travelling waves. Although such transducers are available, most existing in-
stallations use voltage and current transformers which are designed with 50
or 60 Hz signals in mind. Noise immunity and the response of travelling-wave
algorithms when faults occur near a voltage zero are additional factors which
suggest that the introduction of such schemes will in practice be slow. The
most likely initial application of such techniques will be as part of a combined
scheme where, for example, a conventional distance protection scheme is
supplemented by an ultra-high-speed algorithm based on travelling waves, to
obtain accelerated performance for faults at the remote end of the line.
REFERENCES
1. Chamia, M, and Liberman, S. (1978) Ultra high speed relay for EHV/UHV
transmission lines-development design and application, Trans. lEE, PAS-97, 2104-
2112.
2. Johns, A. T. (1980) New ultra-high-speed directional comparison technique for
the protection of ehv transmission lines, Proc. lEE, 127 (C), 228-239.
3. Crossley, P. A. and McLaren, P. G. (1983) Distance protection based on travelling
waves, Trans. lEE, PAS-I02, 2971-2983.
4. Christopoulos, C., Thomas, D. W. P. and Wright, A. (1988): Scheme, based on
travelling waves, for the protection of major transmission lines, Proc. lEE, 135
(C),63-73.
5. Christopoulos, C., Thomas, D. W. P. and Wright, A. (1989) Signal processing and
discriminating techniques incorporated in a protective scheme based on travelling
waves, Proc. lEE, 136, (C), 279-288.
6. Wagner, C. F. and Evans, R. D. (1933) Symmetrical Components, McGraw-Hill.
7. Yee, M. T. and Esztergalyos, J. (1978) Ultra high speed relay for ehv/uhv trans-
mission lines-installation staged fault test and operational experience, Trans. lEE,
PAS-97,1814-1825.
8. Johns, A. T. and Walker, E. P. (1988) Co-operative research into the engineering
and design of a new directional comparison scheme, Proc. lEE, 135 (C), 334-
368.
9. Type LFDC digital directional comparison protection relay, Publication R-4078,
GEC Measurements.
10. Vitins, M. (1981) A fundamental concept for high speed relaying, Trans. lEE,
PAS-I00, 163-168.
11. Vitins, M. (1978) A correlation method for transmission line protection, Trans.
lEE, PAS-97, 1607-1617.
12. Type LR91 ultra-high-speed directional relay, Publication CH-ES 63-85.11, Asea
Brown Boveri.
13. Cabeza-Resendez, L. Z., Greenwood, A. N. and Lauber, T. S. (1985): Evaluation
of ultra-high-speed relay algorithms, EPRI Report EL-3996.
14. Rajendra, S. and McLaren, P. G. (1985) Travelling-wave technique applied to the
protection of teed circuits: principle of travelling-wave technique, Trans. lEE,
PAS-I04,3544-3550.
IS. Shehab-Eldin, E. H. and McLaren, P. G. (1988) Travelling wave distance protec-
tion - problem areas and solutions, lEE Transaction on Power Delivery, 3, 894-
902.
16. Thomas, D. W. P. and Christopoulos, C. (1991) Ultra-high speed protection of
series compensated lines, Paper presented at the IEE/pES summer meeting, San
Diego, California, Paper 91 SM 359-0 PWRD.
FURTHER READING
McLaren, P. G., Travelling wave and ultra high speed (UHS) relays Chapter 6 in lEE
Tutorial Course-Microprocessor relays and protection systems (ed M. S. Sach-
dev), 88 EH 0269-I-PWR.
Bollen, M. H. J. and Jacobs, G. A. P. (1988) Extensive testing of an algorithm for
travelling-wave-based directional detection and phase-selection by using TWON-
FIL and EMTP, Eindhoven University of Technology, EUT Report 88-E-206.
Bollen, M. H. J. and Jacobs, G. A. P. (1989) Implementation of an algorithm for
travelling-wave-based directional detection, Eindhoven University of Techno-
logy, EUT Report 89-E-214.
Johns, A. T., Martin, M. A., Barker, A., Walker, E. P. and Crossley, P. A. (1986) A
new approach to E.H.V. direction comparison protection using digital signal
processing techniques, Trans. lEE, PWRD-l, 24-34.
Crossley, P. A., Elson, S. F., Rose, S. J. and Williams, A. (1989) The design of a
directional comparison protection for ehv transmission lines, Fourth International
Conference on Developments in Power System Protection, lEE Con! Publ. 302,
pp. lSI-ISS.
13
Digital communications
INTRODUCTION AND HISTORICAL BACKGROUND

Many protection arrangements require some means of communication.
Examples are, distance schemes with some form of intertripping and phase
comparison in differential current protection schemes. For a long time,
communications were based on lines leased from the Post, Telephone and
Telecommunications (PTT) authorities, Power-Line Carrier (PLC) systems,
or microwave channels. Advances in information technology in recent years
have meant that it is now possible to employ digital communication
networks of high capacity for protection control as well as data and voice
channels. In fact, technological changes mean that the differences between
voice, data and video communications are blurred as are the differences
between data communications and data processing in single- and multi-
processor systems. There is now a plethora of information about the power
system, potentially of importance to protection, which can now share data
and communications with the control and management functions of a
network. The rate at which this sharing and interdependence will increase
and what final level it will reach is difficult to foresee. Undoubtedly, long
term reliability of protection schemes relying on communications has
depended on the communication system and therefore the introduction of
new communications technologies deeper into protection procedures has to
be examined carefully so that system security and dependability is
maintained at high levels. It is now essential that the protection engineer is
aware of digital communications issues so that he or she can influence the
development of the communications network, take advantage of the
possibilities that it offers and maintain the high integrity of the protection
relays designed according to the principles presented in previous chapters.
The purpose of this chapter is to introduce the basic principles of digital
communications and their use in protection.
13.1 BASIC COMMUNICATION PRINCIPLES

The purpose of a communication system is to convey information (data).
Before this can be done, data has to be encoded in some way to produce

480 Digital communications
electrical signals which can then be transmitted and processed [1]. Some
data is inherently analogue (i.e. has continuous values) such as voice,
current obtained from CTs, etc. In other cases data may be digital in
nature, e.g. representing the status of a circuit breaker (open or closed). A
communication system can only propagate electrical signals hence data,
whether analogue or digital, has to be converted to analogue or digital
signals. In the former case the signal varies continuously and in the latter it
consists of pulses of typically two levels representing values (bits) '1' or
'0' (binary system).
Text is an example of data which must be converted to bits (ones and
zeros) before transmission. The most widely accepted system for this is the
American Standard Code for Information Interchange (ASCII) which uses
a sequence of 7 bits to represent a range of useful characters. As an
example, character 'b' is represented by the sequence 1100010. In an actual
transmission further bits may be added to indicate start/finish, to im-
plement error checking etc.
A continuous signal may be converted into a sequence of pulses by
sampling it at regular intervals (discrete signal) and each pulse can then be
represented by a sequence of 8 bits. This means that each pulse amplitude
may be represented by one of 28 = 256 levels and that a 'quantization
error' of half a level may be present.
RZ
volts
+v
- r-
NRZ
volts
+v
- -
II
I
-v
loE ~
I
Signal penod T
Fig. 13.1 Schematic of two waveforms, return to zero (RZ) and non-return to zero
(NRZ).
Basic communication principles 481
13.1.1 Basic theorems
A binary sequence representation may be unipolar (voltage levels +V and
ovolts) or bipolar (voltage levels +V and -v volts) as shown in Fig. 13.1
for the binary sequence 1010.... In the former case the code is return to
zero (RZ) and in the latter non-return-to-zero (NRZ). Fourier has shown
that any periodic signal V(t) of period T may be decomposed into an
infinite series of sinusoidal components of frequencies which are multiples
of the fundamental frequency fo = 1/ T.
00 00
V(t) = llo + Ldncos(nroot) + Lb.sin(nroot) (13.1)

n=) n=)
where n is an integer
roo = 2nfo = 2n/T
1
ao = T
JT V(t)dt
0
2
an = T JT V(t)cos(nroot}dt
0
b. = -2 JT V(t) sin(nroot) dt
T 0
Assuming that the mark-space ratio (relative duration of the two

voltage levels) in Fig 13.1 is equal to one then the Fourier series
representation of the two signals is
Vet) ="2V + -;- II}

2V {cos(root) - 3"cos(3root) +scos(Sroot) ... RZ-code (13.2)
4V{ cos(root} - 3cos(3root)

Vet) = -;- II }
+ 5co s(Sro ot) ... NRZ-code (13.3)
It should be noted that in neither case there are even harmonics and that
there is a d.c. component for the RZ code. The amplitude of high-order
harmonics gets increasingly smaller. It therefore follows that if this signal
is passed through a communication system of a limited bandwidth some of
the higher order harmonics, and therefore some of the energy of the
original signal, will be lost. This loss of high frequency spectral content
means that the edges indicating transitions of the signal between two levels
may not be sharp enough. This may make it difficult to interpret correctly
the data at the receiving end. In addition to this limitation the signal may
be attenuated, delayed or corrupted by noise during its transmission [1-3].
An important parameter is the signal-to-noise ratio (SNR) at the
receiving end defined as
SNR = 10 IOg(!) , in dB (13.4)
where S is the power in the received signal and N is the power associated
with noise. For the signal in Fig 13.1, the signal rate (lIn, the bit rate
(2/n and the data rate (which may be smaller than the bit rate when
additional bits are transmitted for control purposes) cannot be increased
indefinitely for a given channel bandwidth.
In 1924 Nyquist obtained a fundamental limit to the maximum data rate
for a noiseless channel which is
maximum data rate = 2(BW) log2 M bits per second (bps) (13.5)
where (BW) is the bandwidth in Hz and M is the number of discrete
voltage levels. This formula is obtained by observing that a symbol which
can be in M independent states carries information equal to log2 M and
that a signal of bandwidth (BW) requires at least 2(BW) samples for each
reconstruction (sampling theorem). Hence the information carried is the
product of these two expressions as shown in equation (13.5).
As an example, assuming a voice channel (BW) = 3 kHz, and binary
signals M = 2 gives
Max data rate = 6000 bps
Assume now that the bit rate is B and that to send a character 8 bits
are used. Then the time required is 81 B seconds, hence the fundamental
frequency is BI8 Hz. If B is chosen to be 19200 bps then the frequency
of the fundamental is 2.4 kHz. Hence a voice channel (BW = 3 kHz) will
just manage to propagate the fundamental frequency component. In the
presence of noise the maximum number of states that can be identified
is M = (I + SINi/ 2 hence in this case the maximum data rate is given
by the Shannon-Hartley law
maximum data rate (bps) = (BW)log2(1 + SIN) (13.6)
Assume as an example a voice channel with (BW) = 3 kHz, SNR = 25 dB,
hence SIN = 316. Then the maximum data rate is 3000xlog2317=
24925 bps.
Similarly, assume that a digitized signal is to be transmitted through
an analogue channel BW = 4 kHz and that an 8-bit code is used to
represent each sample. This represents 28 = 256 possible levels and hence
a quantization error of half a level. The signal-to-noise ratio is at best
256/1 or in power terms approximately 2562.
Substituting in (13.6)
Bit rate = 40001og(1 + 2562) = 64 kbitsl s.
This result may also be derived directly from the sampling theorem.
Since a 4 kHz signal requires 2 x 4000 = 8000 samples and since each
Basic communication principles 483
sample is coded by 8 bits, the required rate is 8 x 8000 = 64 kbitsl s. In
practice this is an upper limit and due to complex noise statistics much
lower bit rates should be expected [1,4]. Another quantity frequently
used in communication system is the baud rate. This is the number of
signal line transitions per second [2]. If for example there are 300 signal
transitions per second and 4 bits per signal then the bit rate is
300 x 4 = 1200 bps.
13.1.2 Basic Structure of Transmission Networks

Current trends in the installation of new communication networks and in
the refurbishment of old is towards a digital technology. Several advantages
come with digital technology including higher noise immunity, better
transmission quality, greater flexibility and ease of routing, sophisticated
error correction, monitoring and network management, and reduced costs
[4,5].
In a complex network there is a whole range of communication tasks
to be addressed including:
(i) signal generation, synchronization and interfacing
(ii) detection and correction of errors and recovery from faults in the
networks
(iii) in shared networks correct addressing and routing of messages
(iv) agreed message structure (formatting protocols)
(v) system management, protection and utilization.
It is beyond the scope of this chapter to cover in detail all these aspects.
The reader is referred to texts [1-3] for details. It is however evident from
the previous list that great care has to be taken to structure network
messages and procedures in order to achieve maximum reliability and
economy.
Networks may be structured in several ways. It is normally impractical
to provide direct communication paths between each station in a complex
system. Normal practice is to share resources, that is the same com-
munication path is shared by several stations using multiplexing techniques
and indeed more than one communication path may be available between
two stations. A typical topology is shown in Fig. 13.2. This is described as
a switched network, where paths may be selected (switched) to establish
communication between different networks. Switched networks are in tum
distinguished into circuit-switched and packet-switched. In the former case
a definite path between two stations is established prior to transmission,
while in the latter, the message is broken down into packets which are
stored and transmitted from each node - no definite path needs to be
established in advance.
An alternative to a switched network is a broadcast network where all
Fig. 13.2 Typical switched network (A to D are stations and 1 to 4 are nodes).
stations receive a broadcast message. Examples are networks using

terrestrial radio, satellites, a local bus or a local ring network.
Messages may be structured and transmitted in several ways. A
character consisting of 8 bits (a byte) can be transmitted in parallel or
serially as shown in Fig. 13.3. Parallel communication is fast but expensive
over long distances. Serial communication is slower but cheaper and the
only realistic choice for transmission over long distances. Depending on
requirements, there may be several modes of communication. Simplex
communication refers to data transmitted in one direction only. Half-
duplex refers to communication both ways but not simultaneously. Duplex
communication refers to simultaneous communication in both directions.
Finally, two transmission modes are possible - asynchronous and
synchronous transmission - depending on whether the transmitter and
receiver clocks are synchronized or not. In asynchronous transmission the
receiver synchronizes at the start of each byte of data. In synchronous
transmission the receiver attempts to keep in synchronism with the stream
of bits it receives for a whole block of bytes. In the case of asynchronous
transmission start and stop bits are necessary for each byte of information
to give the receiver the chance to re-synchronize. This results in a simple
Fig. 13.3 Parallel (a) and serial (b) transmission.

Transmission media and interfaces 485
and cheap system but extra bits are required per channel as a synchroniz-
ation overhead. In the case of synchronous transmission there are no start
and stop bits. Synchronization of the two clocks is achieved either by a
dedicated clock synchronization line between receiver and transmitter, or
by embedding synchronization information in the data bit stream. For
large blocks of data synchronous transmission is more efficient.
Asynchronous transmission requires a 20% overhead in extra bits while
extra control information in synchronous transmission using a common
scheme (HDLC) requires 48 bits. Thus for a 1000 bit message the overhead
is less than 5% [1]. More details regarding the structure of messages to
achieve orderly and error-free data transmission are given in section
13.3.5.
13.2 TRANSMISSION MEDIA AND INTERFACES

The selection of a transmission medium and of the appropriate interface
affect the rate at which signals may be transmitted. In this section the most
common arrangements are examined.
13.2.1 Transmission media

Signals may be transmitted either by guiding along wires or by radiation
in free space. In the former case the nature of the guiding medium is
paramount in determining data rates while in the latter it is the nature of
the signal that is important.
Guided transmission
The simplest arrangement is that of two parallel wires forming a
transmission line. As the bit rate and therefore the frequency increases,
current flows near the wire surface (skin effect) thus increasing resistance.
At high frequencies there are also losses due to radiation. As a result,
bandwidth is limited to a few kHz and the maximum transmission distance
to a few tens of metres. Moreover, this transmission line is highly
susceptible to crosstalk and coupling to incident radiated fields. This
causes difficulties with electromagnetic compatibility (EMC) in the very
harsh electromagnetic environment of substations [6-8]. The flat ribbon
cable is an example of this type of transmission medium consisting of
several parallel lines.
A much better alternative is the use of a twisted pair of wires preferably
covered with copper tube or brass as an electromagnetic shield. The wire
twists minimize crosstalk and coupling to radiated fields. The use of shields
with low transfer impedance and suitable earthing arrangements can
increase immunity to and minimize emission of electromagnetic
interference. A further enhancement is the use of balanced transmission.
In normal unbalanced transmission, the signal is applied to one wire and
the reference to the other. However, the two wires are not in every way
identical in their proximity and coupling to adjacent circuits, grounds, etc.
Therefore, the degree of coupling to each wire due to crosstalk and
external fields differs. Assuming as an illustration that the induced
voltages, with respect to some reference, on each wire A and B, are VA and
Va respectively, then the interference voltage seen by the load connected
across wires A and B is VA - ViI, Only if the two wires are balanced with
respect to adjacent circuits is it likely that VA ~ Va and therefore that
interference is minimized. Control of interference and therefore of signal
integrity on transmission lines demands the restriction of EMI current
components both differential- and common-mode [6J. Twisted-pair wire
transmission has typically a bandwidth of 250 kHz, a maximum data rate
of 4 Mbps, with repeater stations to regenerate the signal every few km.
Screened, balanced twisted pair wire is used for interconnecting local
computer networks and teleprotection and telecommunication equipment.
Further details of this and other transmission media may be found in [1 J.
An alternative suitable for achieving higher data rates is the coaxial
cable. This has high immunity to EMI and is used extensively in telephone
and television long distance transmission and in local area networks. The
maximum bandwidth is approximately 400 MHz and the maximum data
rate 500 Mbps. Amplifiers or repeaters are required every few km. The
characteristic or surge impedance of coaxial cable is low, hence careful
interfacing to circuits is required.
Another alternative as a transmission medium is the optical fibre cable
[9J. It consists of a core of thin glass or plastic fibre surrounded by
cladding of a lower refractive index with an outer jacket of plastic for
mechanical and environmental protection. The fibre transmits a light beam
which is constrained, guided, by total refraction inside the core. It is the
medium of choice for long haul and high-speed transmission. Plastic fibres
are cheap but have much higher attenuation, i.e. 150dB/km, than glass
fibre which offers attenuation typically 100 times lower. Typical maximum
data rates are 2 Gbps and the bandwidth is 2 GHz. It is clear that optical
fibre cable has significant advantages in higher bandwidth and longer
distances between repeaters (tens of lan). It is also smaller, lighter and it is
immune to electromagnetic interference. It is however necessary to have
equipment converting electrical to optical signals and vice versa, thus
increasing costs and the risk of interference. Fibre optic cables are used
exclusively for long haul and/or high-speed transmissions. For protection
applications fibre optic cables may be installed as the core of a composite
overhead earthwire, wrapped around phase or earth conductors, as a self
supporting cable on existing towers, or in underground ducts and
conduits.
Transmission media and interfaces 487
Free-space transmission
Transmission in unguided media may be distinguished into microwave

(terrestrial or satellite) and radio. In the former case transmission is at
frequencies above approximately 2 GHz and in the latter below 1 GHz. At
high frequencies propagation can be highly directional thus making it
possible to establish point-to-point communications for long distance data
transmission or for transmission between nearby sites. At radio
frequencies, propagation is less directive compared to microwaves, thus
antenna design and alignment are less critical. In both cases high data rates
may be achieved but attention must be paid to electromagnetic interference
from natural sources and other users of the electromagnetic spectrum.
13.2.2 Interfacesfor protection and communication

Effective communication between different types of equipment referred to
as Data Terminal Equipment (DTE) either through dedicated private
circuits or through the Post, Telephone and Telecommunication (P1T)
authority requires that appropriate standards are available which specify
mechanical, electrical, functional and procedural details of data trans-
mission. As an illustration, mechanical details will include the type of plugs
and sockets, number and configuration of pins, while electrical details will
include voltage levels etc. Similarly, specification of the function of each
circuit and the procedures for transmitting data is necessary. In order to
facilitate the widest possible use of data transmission systems, various
organizations have sought to specify standards. The most important
organizations are the International Standards Organisation (ISO) and the
International Telecommunication Union (lTU). A complete survey of
interfaces and standards is beyond the scope of this book. However,
enough details will be given here to reach an understanding of the issues
and main standards involved.
In a protection application there will be many interfaces. A typical
configuration is shown in Fig. 13.4 [4]. Of particular relevance to data
communications is the interface between protection equipment and
telecontrol equipment and between teleprotection equipment and the
telecommunications network. The other interfaces shown in Fig. 13.4 are
normally an integral part of protection designed and supplied by the
supplier of the protection.
The interface between protection and telecontrol is normally via a slow-
speed serial port or possibly a local area network (LAN). Connection be-
tween teleprotection and the telecommunication network is normally via a
dedicated circuit or via a network through which a circuit is established each
time (a virtual connection) with specified limits for delays. In all cases
standards are necessary to allow interconnection. In the remainder of this
('rf\
"-
--- "
---------------,
Protection Teleprotection Telecom

~--~ ~--~
equipment equipment system
Other
L..-------l~ protection
equipment
--------------------
Telecontrol
equipment
Fig. 13.4 Schematic of protection and associated interfaces.
section the basic mechanical/electrical standards are described relating to

basic circuit interconnections (physical layer standards). Other specifications
required for orderly communications are presented in section 13.3.
The EIA-232D interface

This is an interface commonly used for connecting to public switched
telephone networks. It is suited to normal low bit-rate telephone circuits.
The original specifications, known as the RS-232 interface, appeared in
1962 and the current improved version dates from the mid eighties. It
provides for a 25 pin connector and binary '1' is represented by negative
voltage 3 volts or higher. Binary '0' is represented by a positive voltage
higher than 3 volts (typical values 12V). Transmission over distances of
the order of tens of metres is possible. Dedicated circuits deal with data
transmission, timing and control functions. Multicore or ribbon cables are
used as the transmission medium. A signal ground and shield are provided
and full-duplex operation is possible. Although the high voltage levels used
provide some immunity to noise, the presence of a single wire ground
reference causes noise and crosstalk problems. This interface defined by
the Electrical Industrial Association is equivalent to the V.24 interface
defined by the lTD. The electrical specification is defined in V.28. Further
details may be found in [1].
Communication systems and protocols 489
The EIA-530 interface
This interface (the equivalent lTV specification is V.35) differs from EIA-
232D mainly in the introduction of balanced transmission. This increases
noise immunity and makes the interface more suitable for longer
transmission lengths and faster bit rates. It represents the evolution of
specifications through the RS-449, which was based on a 37-pin connector
and proved unpopular, to electrical standards based on either the RS-432-A
for unbalanced transmission or the RS-422-A for balanced transmission. In
the latter case bit rates in excess of 100 kbps over 1 Ion long transmission
paths are possible or several Mbps over distances of the order of tens of
metres. This interface can be used with a 25-pin connector.
Other interfaces
There are a number of other interfaces to connect to the wide variety of
networks available. For more details the reader is referred to the
publication already cited. To summarize, specifications with prefix V refer
to connections to the existing Public Switched Telephone Network
(PSTN). Typical examples are specifications V.2 and X.33. Where public
networks specifically designed for data transmission (Public Switched Data
Network - PSDN) are used, specifications with the prefix X apply. A
typical example is the X.25 recommendation. For the emerging digital
networks (Integrated Services Digital Network - ISDN) specifications
with prefix I must be used. Other examples are time division multi-
plexed networks (prefix G), and Local Area or Wide Area Networks
(LAN/WAN) where IEEE 802.* and ISO 8802.* specifications apply.
13.3 COMMUNICATION SYSTEMS AND PROTOCOLS

In this section modulation, multiplexing, the handling of errors and the
overall structure of messages for orderly transmission are discussed.
13.3.1 The encoding of data

Data which may be analogue or digital may be encoded into a digital
signal which fits with the properties of the transmission medium. On
reception it is decoded to recover the original signal. Alternatively,
analogue or digital data may be encoded on a carrier signal by a process
described as modulation and transmitted in analogue form. The carrier
frequency is chosen to best exploit the properties of the medium. The
original signal is recovered at the receiver by a process described as
demodulation. It is therefore possible to envisage the following encoding
possibilities [1]:
(i) Digital data -+ digital signals
The simplest method to transmit digital signals is to use different

voltage levels for binary '1' and binary '0', e.g. apply a voltage to
transmit '1' and no voltage to transmit '0' (on-off keying). This is
described as a non-return to zero level (NRZ-L) code. An example is
the code used for the RS-232 interface shown in Fig. 13.5. Most of the
signal energy in this code is contained between d.c. and half the bit rate.
Disadvantages of the code are the presence of a d.c. component (hence
transformer isolation cannot be used to minimize noise) and the lack of
a synchronization capability (since a long sequence of ones or zeros
gives a constant voltage over a long period). For these reasons this
code is used for short distance communications only. An alternative
arrangement which deals with the synchronization problem is bipolar
alternative mark inversion (bipolar-AMI) encoding. Here binary '0' is
represented by no line voltage and binary '1' by a negative or positive
pulse so that binary '1' pulses alternate in polarity. This encoding is
shown in Fig. 13.5. It is clear that a long sequence of zeros may still
=u
o 1 o o o 1 1 1 Binary sequence
- + - - + - - - - + - - - ov NRZ-L
Bipolar-AMI
l nnn nnr-
u uuuuu
OV Manchester
Fig. 13.5 Examples of encoding techniques.

cause synchronization problems. Yet another alternative is Manchester
coding where there is a transition in the middle of each bit period
so that a low-to-high transition represents binary '1' and a high-to-
Iowa binary '0' as shown in Fig. 13.5. The transition in the middle
of the bit period represents the data and also offers a means of
synchronization.
The G703 64 kbps interface specified by the ITU uses four intervals
to encode one binary bit. Binary '1' is described by the sequence 11 00
and binary '0' by the sequence 1010, giving for the same binary sequence
used in Fig. 13.4 the sequence shown in Fig. 13.6(a). In the next step
the polarity of each alternate encoded bit is reversed to give the
sequence shown in Fig. 13.6(b) [10]. In this step d.c. levels are removed
and therefore the interface may be isolated by transformers to deal with
noise and electromagnetic compatibility problems. A further feature is
added as shown in Fig. 13.6(c) whereby a violation of the alternate
inversion rule is introduced every eighth bit. This violation can be
detected and used for synchronization. For a 64 kbps interface this
represents an 8 kHz synchronization signal embedded into the data
stream.
There are many other encoding schemes, details of which may be found
in the references cited.
o 1 o o o 1 1 1
(a)
(b)
(c)
t
Violation
t
Violation
Fig. 13.6 Encoding scheme for the G703 interface ('0' = 1010, '1' = 1100).
(ii) Analogue data ~ digital signal
This is essentially a process of sampling an analogue waveform. A codec
(coder-decoder) is a device that converts analogue data to digital signals.
A common technique for doing this is pulse code modulation (pcm)
whereby the data are sampled at regular intervals to produce pulses of
amplitude proportional to the original data. These amplitudes are then
quantized using an n-bit integer. The resulting binary sequence can be
coded for transmission using any of the techniques described earlier. An
alternative to pcm is delta-modulation.
(iii) Digital data ~ analogue signals

The advantage of connecting into analogue signals lies in the better use
of the propagation properties of the medium and the sharing of the same
medium by several users simultaneously. A typical application is the public
telephone network designed to carry voice signals with a bandwidth of
about 4 kHz. Digital data can only be transmitted through this network
after conversion into analogue signals using a modulator-demodulator
(modem). Common techniques for converting are:
1. Amplitude-shift keying (ASK) where a single carrier is used to represent
binary '1' and '0' by varying the amplitude as shown in Fig. 13.7(a).
2. Frequency-shift keying (FSK) where two different carrier frequencies
are used to transmit binaries '1' and '0' (Fig. 13.7b). Full-duplex
operation may be achieved over a voice channel by using carrier
frequencies centred around say 1170 Hz and 2129 Hz for each trans-
mission direction.
3. Phase shift keying (PSK) where the phase of the carrier is shifted by
180 degrees to represent binary '1' and remains unchanged to represent
binary '0' as shown in Fig. 13.7(c).
Other modulation schemes which transmit more than one bit for each
transmitted state have been developed, e.g. quadrature amplitude modula-
tion (QAM) where there are 16 levels per element and hence 4-bit symbols
may be transmitted [2].
(iv) Analogue data ~ analogue signal

In some cases analogue data may be transmitted without any modification
in what is described as baseband communication. However, in many
applications higher frequencies are desirable to improve transmission.
Also, by suitable modulation and frequency-division multiplexing, the
same medium may be shared simultaneously by several users. Common
modulation techniques are amplitude modulation (AM), frequency (FM)
and phase modulation (PM).
o 1 1 o 1
ASK
(a)
AAAAAA
VVVVWij
FSK
(b) AAAAAAAAAAAAA.
\[VVVVVV\{V VVVij t
PSK
AA AMAAA AA.
Vool) \[V \[ool)
(c)
Fig. 13.7 Modulation techniques for analogue signals.
13.3.2 Multiplexing
Multiplexing is a technique which allows the sharing of the communication
system by more than one source. This means that the communication
system is utilized more efficiently. There are two approaches to
mUltiplexing.
First, in cases where the communication system bandwidth exceeds the
required bandwidth of signals originating from different sources, it is
possible to use frequency-division multiplexing (FDM). This is depicted in
Fig. 13.8 and is a popular approach, e.g. TV and radio. Each source
modulates a different carrier (carrier frequencies ft,h, ... ,fN) and at the
de-multiplexer the different carrier frequencies are separated to obtain N
different signals originated from the N different sources.
A second approach is time-division multiplexing (TDM) whereby
samples from each source are taken in rotation and transmitted as shown
schematically in Fig. 13.9. Consider as an example the case of a voice
channel which is sampled 8000 times per second. This means that a sample
(8 bits) is taken every 1/8000 = 125 ~s. In the case of several such channels,
each channel may be assigned a separate time slot. If 32 channels need to
be multiplexed then the bit rate must be set at
(8000 samples/s) x (8 bits/sample) x (32 channels) = 2.048 Mbps
Source 1 - Channel 1 (frequency (1)

- Output
Cii
Cii x
2 x
Q)
Ci
2
'2 Q)
Ci
2
+'
"S
~ E
:::!! Q)
n
-
n
'n -
0 n
Fig. 13.8 Frequency-division mUltiplexing (FDM).
Source 1 Output 1
Cii
2 ~ g1ii t---- 2
Ci
+' "S
"S E
:::!! Q)
n o n
Fig. 13.9 Time-division multiplexing (TDM).
Thirty-two consecutive samples, one from each channel, constitute one

frame 125 ~s in duration.
Both FDM and TDM can be used with either analogue or digital
signals.
13.3.3 Error detection

Errors, error detection and therefore the reliability of a teleprotection system
is a complex subject. The reader is referred for more details to specialist work
(1, 4, 11, 12]. In this section the basic issues will be considered and guidance
to likely performance will be offered. An error occurs when a received bit
differs from the bit that was sent. Noise, and especially impulsive noise,
means that there will always be errors. Long term measurements can
establish a mean error ratio. If, for instance, an average of one in every
thousand bits is in error then the bit error rate (BER) is equal to 10-3
Consider the case of a frame consisting of N bits. The probability that the bit
is not in error is 1 - (BER) and hence the probability that the complete frame
is received with no bit errors is (1- (BERN. If the frame consists of 16 bits
and BER = 10-4 (a typical value for a public telephone line connected to a
modem) then the probability of this frame arriving without errors is
(1 - 10-4 )16 = 0.9984. Hence the probability that a frame arrives with one or
more errors is 1 - 0.9984 = 4.6 x 10-3 This error performance, in a com-
munication system which tra~mits 16-bit frames continuously over 24 hours
at a rate of 300 bps (i.e. 24 x 3600/(16/300) = 1.62 x 106 frames per day),
would imply that, on average, 1.62 x 106 x 4.6 X 10-3 = 7452 frames with
errors would be received per day. This performance may not be acceptable
and therefore error detection and correction technique need to be used.
In a teleprotection application it is appropriate to consider the
probability Puc of an unwanted command. If an n-bit message is
transmitted where the 'normal' message and the 'trip' message differ by x
bits then Puc is simply the probability of x bits being in error and all other
bits (n - x) being correct, i.e.
Puc = (BERY(1 - BERr-x
This probability may be plotted for the different values of the BER and
different combinations of n and x [II]. These plots show that there are
particular values of BER when the probability of an unwanted command
is maximum. Increasing the length of the message n and also the number
of different bits x between 'normal' and 'trip' messages reduces the
maximum value of Puc. Similarly, re-transmitting the trip message r times
reduces the probability of an unwanted command from the value given
above to Yoc. CIGRE guidelines for 64kbits/s power utility circuits, based
on the CCITT G.821 recommendation, state a BER < 10-6 for 98.5% of
time and BER < 10-3 for 99.97% of time.
The strict requirements for security of transmission in many applications
demand some form of error checking or correction to be implemented. A
common approach is the introduction of a parity bit. Taking the ASCII
character set as an example which is a 7-bit code, an 8th bit is added so
that the 8-bit word has either an even number of '1' (even parity,
commonly used in asynchronous transmission) or an odd number of , l'
(odd parity, commonly used in synchronous transmission).
If a parity violation is detected the receiver may reject the message or
request re-transmission. This scheme can cope with a single bit reversal but
it fails when two bits are corrupted. This is an example of a vertical
redundancy check (VRC). It is also possible to introduce additional parity
bits at the end of a block of characters (fixing the parity of the first, second
etc., bit for a group of characters) in what is known as a longitudinal
redundancy check (LRC).
A more complex but powerful approach is to use the cyclic redundancy
check (CRC). Here a k-bit long message is augmented by a frame check
sequence (FCS) consisting of n bits so that the k + n number is exactly
divisible by a predetermined number. The receiver performs this division
and if no remainder is detected then the transmission is assumed to be
error-free. Further details and formulae for the probability of not detecting
error bits in commonly used protocols, such as the HOLC with CRC error
detection, may be found in [13].
13.3.4 Open Systems Interconnection (OSI) model
Every complex task can be best tackled by breaking it down to smaller

more manageable parts. Although this process may sometimes appear over
pedantic it nevertheless has many advantages, especially when the task
involves communications in a complex network of computers with a
variety of users and software. In order to make it possible that the
maximum number of users with the widest possible variety of software can
communicate, a basic architecture and a set of standards or protocols are
necessary. These must be widely accepted so that manufacturers and users
see distinct advantages in designing hardware and software complying with
them. This widely accepted architecture referred to as the Open Systems
Interconnection (OSI) model is described in this section. Protocols are the
subject of the next section. OSI divides communication tasks into seven
vertical layers as shown in Fig. 13.10. Each layer provides services to the
layer above it and relies on the layer below it to implement more primitive
functions which remain transparent to it. The purpose of this architecture
r'
Sending process Receiving process
..
--------------
Application Interfaces to the user Application
layer layer
---------------
Presentation Coding to the same format Presentation
layer layer
--------------
Session Establishes orderly dialogue Session
layer layer
--------------
Transport Makes sure message Transport
layer reaches destination layer
--------------
Network Routes packets to Network
layer destination layer
---------------
Data link Starts, or ends, frame Data link
layer layer
---------------
PhYSical Transports raw bits across Physical
layer network layer
--------------
Physical transmission path
Fig. 13.10 The OSI data transmission model.

and its associated protocols is that each layer in the sending end can
establish a relationship with its associated peer layer at the receiving end.
No direct communication takes place between peer layers. Instead all
information is pushed up or down so that the direct link across
transmitter-receiver is through the physical layer.
The actual functionality of each layer depends on the set of
standards (and there are several) relevant to it. However, in general
terms, the purpose of these layers can be appreciated from the following
example:
Consider the task of sending a book by train from Paris to Berlin. This
task may be broken down into seven simpler subtasks. These are,
Go to appropriate railway station in Paris to talk to the relevant official.
This is the Application Layer where the interface with the user takes
place.
Package the book in accordance with railway parcel standards. This is
the Presentation Layer where the particular contents, book in this case,
are no longer of relevance. In an electrical application, data at this level
are coded to a common format.
Establish that there are trains between Paris and Berlin and that the
recipient of the book will collect it. This is the Session Layer where
connection is established between transmitter and receiver.
Make sure that there is a means of knowing whether the book has been
received. This is the Transport Layer which ensures that data has been
received correctly.
Make sure that the book gets on the right train, or trains if several
changes are involved. This is the Network Layer responsible for routing
data to its destination. The task of the Transport Layer is to establish
whether the book was received in Berlin and it is not concerned whether
trains were routed through Brussels or Frankfurt. This detail is the task
of the Network Layer.
Make sure that the package is clearly marked with address details to
distinguish it from all other packages. This is the Data Link Layer
providing error free orderly transmission of data to their destination
through the next lower layer.
Make sure that there are people who know how to operate the railway
(track, trains, signalling) so that trains reach their destination. This is
the Physical Layer which in the electrical application is concerned with
the transmission of bits over the physical medium.
In many protection applications, where a simple dedicated Local Area
Network (LAN) is used, it is not necessary to distinguish all seven layers
since some choices have already been made and are adhered to. A simpler
model then applies based on the Application, Data Link and Physical
Layers.
Start of Frame End of
check
Information
nbits
8 bits Frame header 8 bits
Fig. 13.11 The HDLC frame format
13.3.5 Message structure and protocols

In order for messages to be correctly received and interpreted by different
users a set of conventions is necessary. A protocol is such a set of
conventions so that communicating processes can exchange information
reliably and efficiently. Appropriate protocols apply to each OSI layer so
that each layer can communicate with its peers.
An example of a data link protocol is the High-Level Data Link Control
(HDLC) protocol used in many current differential protection schemes.
The structure of this protocol is shown in Fig. 13.11. Flags delimit the
frame both at its start and finish. The sequence 01111 110 is used as a flag.
Since there is no guarantee that a set of data may not arise with the same
bit sequence and thus incorrectly be interpreted as a flag, the transmitter
always adds an extra '0' after five consecutive '}' except for the flag (bit
stuffing). At the receiver if a '0' is detected after five consecutive '}' it is
deleted. If the sixth and seventh bits are '10' this is interpreted as a flag or
if '11' is detected an abort operation is indicated. The address field makes
sure that the message goes to the right recipient. The control field is used
for handshaking (nature of message, frame length). Error checking is based
on the CRC approach described in the last section. The information
transmitted can consist of any amount of data.
13.3.6 Electromagnetic compatibility

It has to be accepted that all protection equipment, including associated
communications, operate in an electromagnetically hostile environment.
Wideband communications are particularly vulnerable to electromagnetic
interference. Installation and use of communications equipment should
therefore be subject to strict guidelines so that electromagnetic compati-
bility is ensured. This dictates the use of balanced transmission, high
quality screened cable (with low transfer impedance) and good quality
earthing at both ends of the cable, if possible. Electrical isolation must also
be used whenever possible. Particular attention should be paid to interfaces
between electrical and optical transmission paths to avoid coupling and
References 499
propagation of EM!. Guidelines (e.g. lEC 801-4, 255-4 and CCITT 1-
431) have been formulated with the public telephone system in mind. In
power plants and substations a harsher electromagnetic environment is
prevalent which is described in some detail in [14]. A more general
introduction to EMI and mitigation techniques may be found in [6].
13.4 THE FUTURE

Undoubtedly, wideband digital communications are bound to play an
increasingly important part in protection. What is less certain is the rate at
which dependency on communications will increase and the degree to
which protection will be integrated with other power system functions.
Arguments are focused on dependability and security of communications
systems when measured against the stringent requirements of protection. It
is technically possible and may be commercially profitable to use dedicated
fibre optic circuits for protection. However, sharing communication
circuits with other users has many commercial advantages which in the
end may be decisive in making the final choice. The question to be asked
then is what is the long term security of these arrangements. The argument
is essentially one of efficient use of hardware/ software, balanced against
the high levels of reliability demanded of protection circuits. Another
argument relates to the degree of shared functionality between protection
and other power system functions such as control and supervision. Many
novel ideas in protection such as adaptive relaying (which requires
extensive information about the power system which most probably will be
supplied over general communication networks) and the use of Global
Positioning Satellites (GPS) to effect precise timing of events and
functions, make the main protection function dependent on system and
functions which lie outside the normally understood boundaries of pro-
tection. Technological changes and the informed judgements reached on
these seemingly conflicting requirements will influence the degree of
interdependence between protection and communications in future
systems.
REFERENCES
1. Stallings, W. (1994) Fourth edn, Data and Computer Communications,
Macmillan.
2. Halsall, F. (1994) Fourth edn, Data Communications, Computer Networks and
Open Systems, Addison-Wesley, NY.
3. Tanenbaum, A. S. (1996) Third edn, Computer Networks, Prentice-Hall, NJ.
4. CIGRE Report (1991) Application of Wideband Communication Circuits
to Protection - Prospects and Benefits, Working Group 05 of Study
Committee 34.
5. Aggarwal, R. and Moore, P. (1993) Digital communication for protection,
Part 1 - General principles, Power Engng J., pp. 281-287.
6. Christopoulos, C. (1995) Principles and Techniques of Electromagnetic
Compatibility, CRC Press, Boca Raton.
7. Christopoulos, C. (1992) Electromagnetic compatibility - Part I (General
principles), lEE Power Engng J., 6, 89-94.
8. Christopoulos, C. (1992) Electromagnetic compatibility - Part II (Design
principles), lEE Power Engng J., 6, 239-247.
9. Aggarwal, R. and Moore, P. (1994) Digital communication for protection, Part
3 - Fibre optics, Power Engng 1.,241-246.
10. lEE Tutorial on Modern Techniques for Power Systems Protection, University
of Nottingham, 24 March 1997.
11. Hall, I. J. (1993) Performance assessment of digital teleprotection systems,
Proc. 5th Int. Conf. on Developments in Power System Protection, lEE Con!
Publication 368. York, 30 Mar.-l Apr., 1993, 1-4.
12. Hall, I. J. and Potts, S. (1996) Experience in the use of digital communication
links for protection, CIGRE Report 35/34-03.
13. Selga, J. and Rivera, J. (1981) HDLC Reliability and the FRBS Method to
Improve it, Seventh Data Comms. Symp. Proc., Mexico City, Oct. 27-29,
260-267.
14. Guide on EMC in Power Plants and Substations, prepared by ClGRE Group
36.04, April 1997.
FURTHER READING
CIGRE WG 34.03, (1996), Communication requirements in terms of data flow
within substations, Final Report 96-12-01.
14
Digital signal processing (DSP) and
protection
INTRODUCTION AND HISTORICAL BACKGROUND

The development of digital computers in the 1960s led to investigations
to establish the possibility of using them to implement some or all of the
protection functions in a power system. In essence, developments in
technology have gradually permitted the migration of the well established
fundamental protection algorithms described in previous chapters to new
hardware and software environments. The volume and speed of processing
that can be implemented in modern hardware is increasing all the time.
While the fundamental principles remain the same, as already mentioned,
new ways of extracting relaying information from transducer signals are
now possible. The first comprehensive survey of the application of
computers to protection was published in 1969 [1]. Within three years, the
application of these principles to distance protection was described in two
publications [2,3]. The first developments of digital relaying algorithms
and therefore the beginning of numerical relays were described in [4,5]. As
regards hardware, developments were very rapid with basic discrete digital
circuits being replaced by general purpose microprocessors and in the early
1990s by special Digital Signal Processors (DSPs). Further information to
trace the development of microprocessor based protection may be found in
[6,7]. These developments continue to keep pace with very rapid
technological change. The purpose of this chapter is to present the
fundamentals of microprocessor based environments which involve a tight
connection between hardware and software.
14.1 IMPACT OF DSP ON PROTECTION

Many of the new protection relays are microprocessor based and are
generally referred to as numerical relays. This means that signals from
transducers are sampled at fixed time intervals, digitally encoded, and
processed by equipment which resembles a computer to derive relaying

502 Digital signal processing ( DSP) and protection
information, e.g. phase and rms value of currents. The processing function
may be an elaborate one involving a substantial amount of computation.
In a sophisticated modern relay processing is done by a combination of a
general purpose microprocessor and a digital signal processor (DSP). The
former deals mainly with protection logic, user interface and supervisory
tasks, while the latter is designed to execute fast complex algorithms such
as Fourier transform on a time sequence of data. Undoubtedly, the
increasing use of numerical relays owes much to economies that can be
made compared to more conventional designs, but it is worth indicating
here the technological benefits numerical relays and the associated DSP
techniques can bring to protection:
Converting transducer signals to binary bits means that very sophis-
ticated and complex processing can be done and thus highly sophisticated
relaying functions may be implemented. It is also straightforward to
implement on the same piece of hardware several different protection
functions thus giving a multifunctional product.
Since the relaying function is implemented in software, it can be changed
at any time leaving the hardware platform largely unaffected. This offers
great versatility and flexibility during the lifetime of the product.
Self-check facilities may be incorporated in software routines of a
numerical relay, thus reducing the need for regular maintenance, and
offer instant indications of developing problems needing attention.
Numerical relays can be easily interfaced to communication systems
thus allowing remote interrogation of the relay to report faults, tripping
statistics, alter settings, etc.
Physically smaller relays are obtained with greater functionality when
using microprocessor based technology compared to other more con-
ventional designs.
Numerical relays offer the possibility of a closer integration of pro-
tection, control, and monitoring functions in a power system. This
potentially can reduce costs and increase functionality.
Although these are substantial advantages it is worth pointing out some
of the potential problems which must be addressed as new technology
makes greater inroads into power systems. Modern microprocessors
employ fast clocks and therefore pulses with steep edges. This means that
potentially substantial levels of electromagnetic interference (EMI) are
generated which may affect adversely the operation of adjacent circuits.
Moreover, since such systems are necessarily wideband they are more
susceptible to interference from external electromagnetic radiation. The
design of all microprocessor-based relays, associated communications and
all other connections should be done to a standard set to achieve high
levels of electromagnetic compatibility (EMC) [8]. Many of the projected
benefits of numerical relays are in the area of closer integration with other
Hardware and software environments 503
power system function such as control and monitoring. A certain amount
of coordination appears inevitable but there are concerns that protection,
because of its critical importance to the health of a power system, should
remain an autonomous function and should not share facilities with other
less critical functions. The question of closer integration, which is
addressed again in Chapter 15, warrants careful consideration and can
only be resolved after extensive operational experience is available to reach
informed judgements based on the philosophy and practices of the
different utilities.
14.2 HARDWARE AND SOFTWARE ENVIRONMENTS

A microprocessor based system is a form of intelligent controller which
can be used to implement computing operations for the purpose of
controlling processes, acquiring and evaluating data. Its basic construction
is such as to permit its use on many applications. This last feature is its
greatest advantage. The structure of a typical general purpose micro-
processor is shown in Fig. 14.1. Depending on the particular design most
Data bus (typically 8 bits)
U
Data bus interface
Arithmetic
Instruction
logic unit
decoder
(ALU)
Register 0----< +5V
0----< GND
Timing
0----< Clock
=>
Timing and
control
Address
logic
and buffers
Control inputs
and outputs
Address bus
(typically 16 bits)
Fig. 14.1 Schematic diagram of a microprocessor.

504 Digital signal processing ( DSP) and protection
or all functions shown may be implemented on a single chip. A
microprocessor thus constructed forms the central processing unit (CPU),
and when it is supplemented by additional memory, printers, terminals and
other peripherals forms a powerful computational system, as shown in
Fig. 14.2. There are many different types of microprocessors such as
general purpose (e.g. Z80) or optimized for digital signal processing
applications (e.g. TMS320). The design aim in such a system is to produce
a program that the microprocessor system will execute. As an illustration,
the application may demand that the processor obtains a signal from one
of its input ports, compares it with a value stored at a particular location
in its memory and depending on the result of this comparison places an
appropriate signal at an output port (e.g. makes the output high if the
input is larger than the value stored in memory and low otherwise). This is
an example of an action initiated by the program, i.e. it is internal to the
computer system. Other options may involve responding to external
events. In all cases the hardware (physical circuit layout) is the same. What
changes is the controlling program (the software). This is the key feature
of microprocessor-based instrumentation as it allows standardized
hardware (hence low cost), the introduction of new protection functions
without hardware modifications (hence flexibility and low cost) and the
option of introducing more sophisticated relaying schemes (hence technical
benefits). The different parts of a microprocessor based system com-
municate with each other through a collection of signals which we describe
as busses. Data is transferred on the data bus which in its simplest form
may consists of 8 lines (one for each bit). In order to be able to identify the
device or memory location for which the data is intended it is necessary
D/A Digital
NO I/O
Data/address bus
Disc, Printers, Keyboard,

RAM ROM
tape plotters VDU
CPU
Fig. 14.2 Schematic of a microprocessor with associated peripherals.

to have a collection of bits (say 16) specifying a particular address for the
data. We thus need a l6-line address bus. Several other lines are required
for signals which specify particular actions to be taken, e.g. to interrupt
the normal functioning of a system to accept urgent data from an external
device.
Interfacing to external devices may be in parallel or in serial fashion
and custom-made chips are available to facilitate input! output operations,
i.e. buffers to hold data in a stable condition, hardware to select port
address and ascertain whether peripherals are busy or not. There are
different types of memory. Random Access Memory (RAM) can be read
and written to. At least some Read Only Memory (ROM) is also available
containing information which cannot be changed and it is used for an
orderly start of the computer following power on.
The CPU implements the following functions:
Places addresses on the address bus
Decodes instructions
Fetches data required by instructions
Acknowledges and deals with interrupts
Provides control and timing signals for all devices
To increase the flexibility and speed of its operation the CPU has its
own internal storage locations called 'registers' (e.g. program counter,
stack pointer, accumulator etc). Some processors may also have their own
'cache' memory which can be accessed very quickly.
As an illustration, the pin diagram of a very basic processor the Z80 is
given in Fig. 14.3 [9]. The power supply to this chip is through the +5V
and GND pins. Timing and synchronization is derived from the clock
(CLK). Pins BUSREQ and BUSACK control access to the bus. The
overbar indicates that these lines are active when the signal level is low.
The next group of signals control the CPU. The RESET signal initializes
the CPU. The NMI stands for a non-maskable interrupt, i.e. a high-
priority interrupt which is recognized at the end of the current instruction.
The INT is an interrupt generated by external devices and it is of a lower
priority compared to NMI. The WAIT signal indicates to the CPU that an
external device is not ready to receive data. The HALT signal indicates
that the CPU is awaiting an interrupt. The system control signals are refresh
memory (RFSH), write to memory or external device (WR), read from
memory or external device (RD), request input/output (IORQ), request
access to memory (MREQ) and indicate that the current machine cycle is
the one fetching an operation code (Ml). Signals AO to A15 are used to
specify addresses and signals DO to D7 contain data. Other processors will
have different signals, but this example indicates a typical arrangement
for a simple CPU. All instructions executed by such a processor are merely
a series of very basic operations such as, memory read or write, input!
506 Digital signal processing (nSP) and protection
M1 Ao
A1
MREQ A2
System IORQ A3
control
RO A4
WR A5
A6
RFSH A7 Address
As bus
-
HALT A9
A10
--
WAIT A11
CPU Z80CPU A12
control
INT A13
-
NMI A14
A15
RESET
0 .-
CPUbUS{_ BUSREQ 1 .-
control .-
--
BUSACK 2
.-
-
3 Oata
ClK 4 .- bus
+5V 5 .-
GNO 6
.-
7
.-
Fig. 14.3 A basic general purpose microprocessor pin diagram.
output device read or write and interrupt acknowledge. For this particular
processor each of these basic operations can take three to six clock periods
to complete or even longer if an external device is slow. A typical sequence
of events during an 'instruction OP code fetch' is shown in Fig. 14.4. This
figure shows the timing diagram for an MI machine cycle consisting of four
clocks periods. The contents of the program counter (PC), which is one of
the internal registers and contains the start of memory location where the
next OP code is stored, are placed on the address bus at the beginning of the
MI cycle. Half a cycle later (to give time for address signals to stabilize)
the MREQ goes active (low) to enable the dynamic memory. Simul-
taneously, the RD signal goes active to indicate that the memory read data
should be placed on the CPU data bus. This is done with the rising edge
of T3 clock pulse which also turns off the RD and MREQ signals. During
T3 and T4 cycles the contents of the dynamic memories are refreshed.
M1 cycle
t1 t2 ~ t4 t5
.J~ ~ t---- ~ 't----

-II PC I Refresh ADDR I
MREQ
-0-
\ n I L-
RD -1"\ I
------ ------ - - - - - -
WAIT
=======1 \. -----
-,
------ ------
,
MI I \_----
IiN'
~J
RFSH \
Fig. 14.4 Timing diagram for the instruction 'op code fetch' for the Z80
microprocessor.
The WAIT signal if active can be used to add delays (extra clock periods)
between T2 and T3. Virtually all operations involve a sequence of events
such as shown in Fig. 14.4 where appropriate control signals are used to
effect an orderly transfer of data between specific memory locations or
devices.
The CPU understands its own language (the 'machine code' which is a
long sequence of ones and zeros). Since it is difficult for humans to
program at this level they write 'source' programs in a high-level language.
These then are translated by another program into machine code readily
recognizable by the computer system. If the source program is written
using a symbolic 'assembly language' the program that translates the
source code is called an 'assembler'. If the source program is written using
high-level languages such as 'pascal', 'fortran' or 'C', the program doing
the translating is called 'compiler'. Normally, for time-critical real-time
applications, it is best to generate source programs in assembly code as this
gives higher efficiency and therefore speed. As an illustration, the
508 Digital signal processing (DSP) and protection
operations done by a computer system based on the Z80 processor
following the assembly language command shown below will be
described:
LD A, (0200H)
This command means: load the contents of the memory location 0200
(given in hexadecimal) into register A (known as the accumulator). This
command may appear in the program as a line:
0102H LD A, (0200H)
Since binary number representation can get very long other number
systems are used to simplify reading. A popular one is the hexadecimal
number system (base 16) where binary numbers are represented in groups
of 4 digits. The letter H indicates that such a code is used, hence
0120H <=> 0000 0001 0000 0010
0200H <=> 00000010 0000 0000
3AH <=> 0011 1010
Data memory

Registers
A I 0000 0100 0000 0100 0200H
II I
PC 0000 0001 0000 0010 or 0102H ...

~"~"I, A~"$
Program memory

0011 1010 or 3AH 0102H
0000 0000 0103H
0000 0010 0104H
~

Fig. 14.5 Operations involved when executing command LDA, (0200H).
Figure 14.5 depicts the relevant registers and the contents of the data
and program memories. The program counter register contains the
location of the next line of the program to be executed, in this case 0102H.
This location contains 3AH which stands for 'load register A' (LD A) in
hexadecimal. The next two locations in program memory (0103H, 0104H)
contain the location (0200H) where the data to be loaded into register A is
located. In this case this location contains binary data 0000 0100 which is
then loaded into register A. The next program memory location (0105 H)
is loaded into PC ready for executing the next command. To execute this
instruction the following operations are required:
Fetch the op code: 4 cycles
Read low byte of address: 3 cycles
Read high byte of address: 3 cycles
Read contents of address indicated into the accumulator: 3 cycles
In total 13 cycles are required and if the clock frequency is say 5 MHz then
the time required is (! x 10-6 ) 13 = 2.6 ~s. This calculation indicates that
delays are inherent in processing data in a microprocessor system and that
a strict and orderly timing of all operations has to be maintained.
A typical numerical relay configuration is shown in Fig. 14.6. It consists
EJ
1- - - - - - - - - - - - -I r Ti-;;'i;;-g -;';olsign~s - - - - - - - - - - - - - - - - - --I
I ~--r-~~~~ I
I Isolation I
I 1 I
~i Micro- Man/machine
:I Va~ processor interface ROM/RAM
Vb~ JJ JJ JJ
Vc~ Bus
la~ ~ ~ ~ ~~ ~ ~ ~: ,---''''''---,
I
I Digital Serial
Ib~ ADC
I
I
I/O ports
I
Ic~
I L.,--------...J
Anti-aliasing Samp~ I- - - - - - 11 PROCESSING ~

low-pass and hold
filter I Relay Relay I
II _______
sensor trip JI
IDATA ACQUISITION ~ IACTUATION.

Fig. 14.6 Schematic diagram of a typical numerical relay.
(K-1)R
Fig. 14.7 Basic two-pole low-pass active filter.
of three major parts, namely, data acquisition, processing and actuation.

Each of these parts is described below. The data acquisition section is
concerned with the sampling of the analogue waveforms obtained from the
VTs and CTs or other similar transducers. Isolation or other measurement
transformers and non-linear devices may be used to scale signals and to
protect the digital circuits from electromagnetic interference originating
from the primary part of the power system. The process of sampling was
described briefly in sections 4.3.1 and 13.1.1. According to the sampling
theorem the sampling frequency I. must be chosen to be at least twice as
large as the highest frequency of interest. To avoid the phenomenon called
aliasing (section 14.3.1) a simple low pass filter is incorporated with a
cut-ofT frequency set at half the sampling frequency 1./2. A simple RC
circuit will suffice in most cases but a more elaborate design such as that
shown in Fig. 14.7, which is based on a non-inverting amplifier of gain K,
may be configured to obtain a Butterworth filter response. Basic filter
design tables may be found in specialist text and also in [10].
The next component in the data acquisition section is the sample
and hold circuit. The need for this circuit is due to the fact that the
sampling circuit (the analogue-to-digital converter) requires some time
to effect sampling (typically a few ~s) and also the multiplexer presents
each channel to the AID converter in rotation. During this delay the
sampled signal cannot be guaranteed to remain unchanged except if
captured and stored intact for a short period. The sample and hold
circuit ensures that all channels are sampled at the same instant in time
and that the samples are kept unchanged for the time required for the
multiplexer and AID converter to complete conversion. A typical circuit
is shown in Fig. 14.8. The input signal is applied to ICI which is
configured as a voltage follower with a low output impedance so that
capacitor C may be charged quickly to the sampled voltage. This is done
0----1+
Sample n
hOld...JL
1 FET
Fig. 14.8 A sample and hold circuit.
during the 'sample' phase of the FET switch. The switch is then opened
and remains in the 'hold' phase until the next sample. During this period
the sample voltage remains stable across the capacitor C (except from
a slow leakage of charge through the input impedance of IC2 which is
however chosen very high).
The multiplexer is a multi-way switch which allows the selection of one
of several inputs as specified by digital signals from the microprocessor. It
removes the need for using a dedicated AID converter for each channel
which would be expensive but it also slows down the rate at which each
channel may be sampled.
At the heart of the data acquisition unit is the analogue-to-digital
converter. There are several techniques employed, the most common being
the successive approximation converter (SAC). In an SAC the input is
compared with various output codes starting with all bits set to '0'. Then
the most significant bit is set to '1' and if the output is smaller than the
input, is left at '1', otherwise is set to '0'. The process is then repeated for
the next most significant bit and so on. For an n-bit SAC n times the
setting time is required to complete conversion. An end of conversion
output from the SAC signals the completion of the conversion process.
Output is in the form of n bits in parallel or in a serial form. The number
of bits n varies, typical values being 8, 12 and 16. The AID converter is
connected to the microprocessor bus and operated under the control of
signals from the microprocessor.
The second part of typical numerical relay is the processing section. This
is based on a microprocessor and associated peripherals. In the case of
more complex relays more than one processor may be present to deal with
a multitude of tasks. In complex high-performance relays the second
processor is normally a specialized digital signal processor such as the
TMS320 produced by Texas Instruments. This type of processor is
optimized for DSP tasks. While in a general purpose processor multi-
plications are done in software (i.e. by implementing a multiplication
algorithm), in a DSP processor there is a dedicated hardware multiplier.
As a result, a multiplication which in a general purpose processor takes
several instruction cycles, in a DSP processor can be done in one
instruction cycle. This and other similar features make the DSP processor
particularly suitable for time-critical real-time protection tasks.
Finally, the actuation section transmits trip signals from the micro-
processor to the trip coil of the circuit breaker. A sensor at the
circuit breaker (CB) is used to ascertain its status so that this in-
formation is available at the relay. Some form of electrical isolation is
necessary to control electrical noise and interference between the CB and
the relay.
It is important to realize that in such a relay the signals from the VTs
and CTs are first converted into samples each represented by binary
numbers and are processed by the microprocessor system to extract
relaying information. For example, under instruction from the CPU a
sample of the voltage of phase 'a' may be placed in the accumulator then
compared with the value stored in memory and an action taken depending
on the result of this comparison. A whole sequence of such operations
may be programmed to effect a particular protection scheme. This allows
complicated algorithms to be implemented (in contrast to electromechani-
cal relays where comparisons and associated logic must be simple) and also
changed in a future date without any changes in hardware (in contrast to
solid-state relays where the protection function is embedded in the circuit
structure - in the hardware - and cannot be changed without circuit
modification). A new algorithm can be loaded into the memory of the
relay through one of its serial ports or from a remote location through
the communication system if one is available and if the relay is interfaced
to it. These features give numerical microprocessor-based relays their
advantages.
In a microprocessor-based relay a certain amount of self-checking is
possible. There are many techniques for doing checks. As an illustration of
the techniques used the following are listed:
Add all data in the PROM memory and compare with the original fixed
value. If different then probably the PROM is corrupted. An alarm is
then raised to alert the operator to a potential problem.
Data is written in each memory location and then it is read back. If a
difference is detected an alarm is raised.
Check the in/ out circuits by taking data from output port, placing them
on the input port and comparing.
The integrity of software can be checked by an independent 'watchdog
timer' which raises an alarm if a certain operation takes too long to
complete (software locked in an everlasting loop). This checking is done
on-line during time intervals when the CPU is not engaged in processing.
Principles of digital signal processing (nsP) 513
These techniques allow for maintenance when .needed and avoid serious
faults with the relay which remained dormant and only became evident
when the relay failed to operate correctly.
14.3 PRINCIPLES OF DIGITAL SIGNAL PROCESSING (DSP)

It is not possible here to give a complete survey of all techniques and
processes used in digital signal processing. DSP finds application to a very
wide range of problems including communications, imaging, instrumen-
tation and power system protection, the subject of this book. Therefore,
emphasis will be placed on particular techniques which are relevant to
protection. For a more complete picture the reader is referred to specialist
texts [11]. The material that follows is organized into two subsections, the
first dealing with general DSP techniques and the second with particular
algorithms applied to protection problems.
14.3.1 General nsp principles

Central to the digital processing of signals is the sampling process whereby
a continuous signal is rendered discrete. This process was described in
section 4.3.1 and in Chapter 13. Once a discrete signal is obtained it can be
converted to a binary number and processed in a variety of ways to extract
relaying or other useful information. The sampling process and subsequent
treatment of digital signals are described below.
Sampling
Let us consider a continuous time-varying signal with a spectrum in
the frequency domain shown in Fig. 14.9(a). It is assumed that all the
energy in the signal is at frequencies less than f.. The sampling theorem
dictates that this signal should be sampled at a frequency I. ~ 'lfc to
recover information up to frequency f.. If I. = 21. then the spectrum of
the sampled signal is as shown in Fig. 14.9(b). Using a low-pass filter
with a cut-ofT frequency I. will recover the spectrum of the original
signal. If a lower sampling frequency is used if. < 21.) the situation is as
shown in Fig. 14.9(c) and the original signal cannot be recovered by
filtering (due to overlap between curves). In many practical problems
there are high-frequency signal components which are of no interest.
However, following sampling, they may corrupt the original signal by
overlapping with the desired part of the spectrum as shown in Fig.
14.9(c). This phenomenon is called aliasing. To avoid these problems,
after the sampling frequency I. is set, it is normal practice to introduce,
prior to sampling, an anti-aliasing low-pass filter with a cut-ofT
frequency equal to 1./2.
Fig. 14.9 Spectrum of continuous signal (a), spectrum of sampled version when
!, = 21. (b), and spectrum of sampled version when!, < 21. (c).
Digital filters
Filtering is a process whereby a particular property may be added to a

signal. This may, for example, be that it only contains frequencies within a
specified band (bandpass filter) or below a specified frequency (lowpass
filter). Filters may be described either in the time or frequency domains as
shown schematically in Fig. 14.10. In the time-domain filters are
characterized by their impulse response h(t), i.e. their output to an
impulsive input (delta function c5(t. In the frequency domain, filters are
(a) 0 , Filter 0
o(t) h(t)
(b) 0 ..
Vin(w)
Filter I ' 0
Vout(w) = H (w) Vin(w)
Fig. 14.10 Schematic of filter response in the time- (a) and frequency-domains
(b).
Principles of digital signal processing (nsp ) 515
described by their transfer function H(ru). Whichever domain is used to
describe the filter it gives complete information on its response as the
transfer function H(ru) may be obtained from the impulse response h(t) and
vice versa. H(ru) and h(t) are Fourier transform pairs thus each may be
calculated when the other is known.
Digital filters operate on sampled signals in contrast to analogue
filters which operate on continuous signals. In order to illustrate the
operation of digital filters and their relationship to analogue filters an
example is given here of a simple RC lowpass analogue filter and its
digital equivalent.
The analogue filter is shown in Fig. 14.11(a). Its response to a unit step
voltage
Vin(t) = 1, for t 2:: 0
IS v.,ut(t) = 1 - exp(-tjRC), for t 2:: 0
Let as now sample this input with a period T to produce a simple version
Vin(nT) which is defined at multiples n of the sampling period T. It will be
shown that the arrangement in Fig. 14.1 1(b) is a digital filter implementing
a function similar to that of the filter in Fig. 14.11(a) but on the discrete
signal.
In the digital filter, triangles indicate a multiplier by the factor indicated,
circles are adders and square boxes signify a delay of one sampling period
T. Inspection of the diagram in Fig. 14.1l(b) indicates that input and
output are related by
v.,ut(nT) = bVin(nT) + av.,ut(nT - T) (14.1)
Output
Vout(nT)
Fig. 14.11 Schematic of an analogue (a) and its equivalent digital (b) filter.
Table 14.1 Comparison of analogue and digital filter responses
Time [s] 0.2 0.4 0.6 0.8 2
Analogue filter response 0.18 0.33 0.45 0.55 0.63 0.86
Digital filter response 0.227 0.37 0.49 0.59 0.66 0.89
Let us now look at this output when the input is a sampled unit step
voltage.
n=O v.,ut(OT) = bVin(OT) + aO = b
n=1 = b + ab = b(1 + a)
v.,ut(l T) = b Vin (1 T) + a v.,ut(O)
n=2 v.,ut(2T) = bVin (2T) + av.,ut(T) = b(1 + a + a2 )
In general
v.,ut(nT) = b(1 - an+1)/(l - a)
If we choose a = exp( - T / RC) and b = T / RC we can then compare the
response of the analogue and digital filters. For the case when RC = 1 s
and T = 0.05s the output is shown in Table 14.1.
Thus the two filters implement similar functions. The sampling period
must be short enough for high accuracy. The arrangements shown in
Fig. 14.11(b) may be regarded as a blueprint for a hardware implemen-
tation of the digital filter. Alternatively, the various blocks may be
regarded as indicating arithmetic operations on digital signals which may
be implemented in software in any microprocessor based system. It is clear
from this that a software implementation of the digital filter requires that
sufficient time is available between samples for the various arithmetic
operations to be completed and that fast dedicated processors may be
necessary in practical applications. In the filter shown in Fig. 14.11(b) and
the associated equation (14.1) the output at anyone time depends on inputs
and also a previous output. These filters are described as recursive.
Another of their features is that the response to a single impulse is an
infinite train of output pulses. They are thus described as infinite impulse
response (IIR) filters. It is possible to design non-recursive digital filters
where the current output does not depend on previous outputs. An
example of such an implementation is shown in Fig. 14.12. Its response to
an impulse is a finite length train of pulses and it is thus referred to as a
finite impulse response (FIR) filter. Its output is given by
v.,ut(nT) = h(O)Vin(nT) + h(l)Vin(nT - T) + ...
+h(N - l)Vin(nT - (N - I)T) (14.2)
An important parameter in the choice and design of filters is the time
Principles of digital signal processing (DSP) 517
Fig. 14.12 Example of an FIR filter with N weights.
taken for a signal to pass through the filter (the group delay). For fast
operating times the group delay must be short. In IIR filters it is not easy
to specify group delays during design. In contrast in FIR filters the group
delay never exceeds NT where N is the number of weights and T is the
sampling period. Thus FIR filters are much more commonly used in
protection applications. Another interesting property of digital filters is
the possibility of changing coefficients and weights to respond to changing
conditions in the power system. This results in adaptive filters which are
increasingly considered for protection application.
A full understanding of digital filtering requires more elaborate mathe-
matical tools such as the Z-transform and is outside the scope of this book.
The reader is referred to specialist texts [l2].
Discrete Fourier transform (DFT)
The Fourier transform establishes a relationship between a continuous

signal described in the time- and frequency-domains. A discrete Fourier
transform (DFT) establishes a similar relationship for discrete signals.
Starting with N samples in the time-domain (signal duration NT)
V(O). VeT). V(2T) ... VN - I)T)
The OFT can be used to describe the signal by N complex numbers in
the frequency-domain
L V(nT)e-jk/lOo
N-l
V(kQ o) = T (14.3)
"=0
where the signal in the frequency domain is calculated at multiples of
Qo = 21t/(NT) radians and is repeated every 21t/T radians.
The index k determines at what frequency the OFT evaluates the signal.
In many protection applications only the 50 Hz component is required
and hence equation (14.3) is used to evaluate only this component. An
inverse OFT may also be calculated when required.
When all frequency components are required and provided the number
of samples is a power of 2 the fast Fourier transform (FFT) may be used.
The FFT is computationally very efficient especially for a large number of
samples.
Other signal processing techniques

There are several other processes which may be applied to signals to
extract useful information. Only a brief, selective, description will be given
here to alert the reader to other potentially useful possibilities.
In many applications it is desirable to have a measure of the similarity
between two signals. Such a measure is the cross-correlation between
signals l'1, and l'; defined as
R 12 ('r) = roo l'1(t)l';(t - r)dt (14.4)
This expression compares one signal with a shifted version (by 1) of the
other. Two identical signals which are however shifted in time by To will
have maximum R 12 (r) when r = To. In this way the time delay between
the two signals may be accurately determined. Such techniques are used
in connection with ultra-high-speed protection schemes, described in
Chapter 12.
Much of the noise observed in measurement and control circuits is
random in nature and special signal processing techniques are available
to characterize random processes and the response of circuits to them [13].
A technique which has been considered for application in protection is
Kalman filtering [14,15]. This is a form of adaptive filtering whereby
parameter estimates are improved as each measurement data point
becomes available. It appears to have advantages when initial estimates
are available and/ or noise does not remain constant during the
measurements.
Many protection algorithms require the estimation of a parameter (e.g.
nT
Fig. 14.13 Voltage measurements as function of time and associated errors.
rms value of voltage) from a number of samples in the presence of noise.

The problem is similar to fitting a curve to a set of measurements obtained
over time as shown in Fig. 14.13. At time nT the best fit to the signal
V(nT) is related to its estimate v"T and error by
V(nT) = V(nT) + e(nT) (14.5)

In order to obtain the best fit a useful strategy is to minimize the sum
of the squares of all errors for the available N measurement points, i.e.
N T
",A
E =" '2 2 i2
L...J E(nT) = L...J[V(nT) - (Co + C1(nT) + C2 (nT) + ... + Ci(nT]
(14.6)
where polynomial fit to the curve of i + 1 coefficients C has been assumed.
In order to minimize E it is demanded that its derivative with respect to
the i + 1 unknown coefficients is zero, i.e.
aE aE aE
aco = aCt = ... = aCi = 0 (14.7)
The system of i + 1 simultaneous equations may be solved to find the

unknown coefficients. This is known as the least squares method of
obtaining a polynomial fit to the measurement data.
14.3.2 DSP algorithms in protection

There are two general approaches to estimating relaying parameters by
using DSP techniques.
In the first approach relaying information is extracted based on a model
of the observed waveforms, e.g. voltage and current (model of the signal
algorithms). In the second approach relaying information is extracted from
a model of the physical system (model of the system algorithms). Examples
of both approaches for the case of distance protection schemes were given
in section 11.15.
In this section further details are given of typical algorithms and
reference is made to more specialist material [16, 17].
The model of the signal based algorithms is illustrated by three examples.
First consider the application of the OFT in estimating the fundamental
frequency phasor of a signal. Starting from a signal
V(t) = .J2v cos(c.ot + cp) (14.8)
or its sampled version where N samples are taken with a sampling period
T such that NT = 21C/ro
V(nT) = .J2 V cose: + cp) (14.9)
then from equation (14.3) the fundamental component (k = 1) is

+
VI = L: V(nT)e-jkn~
N-I
n=O
2 T
NT
= .J2vL: e
N-I
n=O
j(~+Ip)
2
e
-j(~Ip)
e-jn1f
where the cosine in (14.9) has been replaced by the exponential represen-
tation.
Recognizing that
L 1 = N, and that the geometric series L

N-I N-I
e-j ' " sums to zero
n=O n=O
gives
v,1- -..[iN
--
V e-jlp
2
hence the required phasor is
. 2V,
Ve llp = __
2 N-I
I = __
..[iN
L
V(nT) e-J~
..[iN n=0
= ~N~[V(nT)Cos(n~) -jV(nT)Sin(n~)] (14.10)
The samples V(nT) of the signal can be used in (14.10) to estimate the
fundamental phasor. This procedure may be repeated for other quantities
(e.g. fundamental current phasor) and thus the impedance to the fault may
be obtained to form the basis of a distance protection scheme.
The OFT method requires the sampling of a complete power frequency
period. A typical value of the nwnber of the samples per power frequency
period may be set at 72. If a faster determination of parameters is required
than other approaches are possible [19]. Consider as an example the
estimation of the peak current Ipk in a waveform
i(nT) = I pk sin.9 (14.11)
where .9 = co(nT), T is the sampling period and therefore qJ = coT is the
phase angle between consecutive samples; then,
~i(nT) = in + l)T) - in - I)T)
(14.12)
d.9 2 sin qJ
clearly
di
d.9 = Ipk cos.9 (14.13)
The required peak value is obtained from

I~k = (lpk sin .9)2 + (Ipk cos.9i (14.14)
where the first term is estimated directly from (14.11) and the second from
(14.12). This calculation requires only three samples (at n - 1, n, n + 1)
and hence is very fast. In order to avoid the sensitivity of this algorithm to
noise it is necessary to filter the signal using a bandpass filter prior to
processing. Alternatively, more samples may be used to increase con-
fidence in parameter estimation. These principles may be extended to
estimate more complex quantities. Following a fault the voltage may be
estimated as a function of the current and its rate of change
V(t) = Ai(t) + Bdi(t)/dt (14.15)
where A and B are coefficients to be determined. Using these samples, the
current and its rate of change can be determined giving one equation
(containing the two unknowns A and B) relating the voltage at time (nT)
and current measurements at (n - l)T, nT and (n + I)T. Using another
three samples results in another equation. Solving the system of two
equations allows the estimation of A and B and hence of the quantity
V / I = A + jB. This information, obtained over six samples, may be used
to implement a distance protection scheme.
A second example involves a case where the model of the current signal
involves an exponentially decaying term and five harmonics. This may be a
typical model of a transformer magnetizing current. In such cases a least
squares technique may be used to obtain the fundamental component
[17,18].
The third example involves estimation of the frequency. Asswning the
model of the signal to be V(t) = Vpk sin(cot + qJ), expanding the trigo-
nometric function and replacing sin(cot) and cos(cot) terms by the first three
terms of their Taylor expansion, an expression is obtained for V(t) at
t = t) in terms of six unknown coefficients depending on Vpk and the
deviation !:J.ro from the nominal frequency roo [20]. If several samples are
taken the method of least squares may be used to provide good estimates
of the frequency derivation !:J.ro. Alternatively, correlation techniques may
be used to obtain frequency estimates when the model of the signal
contains an exponentially decaying term and several high-frequency
components as shown in [21].
The model of the system based algorithms were illustrated by an example
from distance protection based on a differential equation describing the
system (section 11.15). A further example is presented here from
transformer protection [22,16]. The essence of this algorithm is the
reconstruction of the current-flux linkage relationship to provide a
restraint signal. If ip is the transformer primary current, Lp is the leakage
inductance and A. the flux linkage, then the model of the system is
V ( ) = L dip(t)
dA.(t) (14.16)
pt Pdt+dt
Integrating this expression using the trapezoidal rule as explained in
section 11.15 gives
(14.17)
where the subscript n indicates the value of the sample at time n T.

The flux can be obtained from (14.17) and the magnetizing current from
i . n = ip.n - i. n where i. n is the secondary current at time nT. For an internal
fault the A. - i. curve is shown by the broken line in Fig 14.14. Under
healthy conditions and depending on the value of the residual flux the solid
...... ......
...... ... ... ... "'I;ernal fault
Fig. 14.14 Flux-i:urrent relationship for a transformer under various conditions.

References 523
curves are obtained. A restraining signal may be obtained if the slope of
the A. - i. is used. Low slope at the initial part of these curves indicates a
fault as shown in Fig. 14.14.
14.4 THE FUTURE

Digital signal processing techniques have made steady inroads in recent
years into the way in which relaying information is extracted from
transducer signals. There is a great variety of algorithms which can be
devised, some of them involving substantial amounts of real time
processing. Two factors drive developments in this area. Firstly,
increasingly faster and cheaper processors are becoming available and
general purpose hardware appears capable of implementing relatively
complex protection schemes. Secondly, algorithm and software
developments in other engineering disciplines such as computer science,
image/ signal processing and communications provide alternative ways of
processing information. Both these factors point to the increasing adoption
of DSP techniques in protection. Adaptive techniques where on-line
modifications of protection algorithm characteristics may be done to
provide an optimum response will become more and more common as will
the coordination of protective schemes with other power system functions
such as control and monitoring. These developments will go hand-in-hand
with an increase in the security and reliability of hardware and software
both as regards malfunctions due to interference and conflicts between
different levels of functionality.
REFERENCES
1. Rockfeller, G. D. (1969) Fault protection with a digital computer, IEEE Trans.
PAS-88, No.4, pp. 438-464.
2. Gilcrest, G. B., Rockfeller, G. D. and Udren, E. A. (1972) High-speed distance
relaying using a digital computer, Part I: System description, IEEE Trans.
PAS-9J, pp. 1235-1243.
3. Rockfeller, G. D. and Udren, E. A. (1972) High-speed distance relaying using
a digital computer, Part II: Test results, IEEE Trans. PAS-9J, pp. 1244-1258.
4. Mann, B. I. and Morrison, I. F. (1971) Digital calculation of impedance for
transmission line protection, IEEE Trans. PAS-90, pp. 270-279.
5. Mann, B. I. and Morrison, I. F. (1971) Relaying a three-phase transmission
line with a digital computer, IEEE Trans. PAS-90, pp. 742-750.
6. IEEE Power Engineering Education Committee (1986) Microprocessor Relays
with Protection Systems, IEEE Tutorial Course 88EH0260-J-PWR.
7. IEEE Power Engineering Education Committee (1997) Advancements in
Microprocessor Based Protection and Communication, IEEE Tutorial Course
97TPl20-0.
8. Christopoulos, C. (1995) Principles and Techniques of Electromagnetic
Compatibility, CRC Press, Boca Raton.
9. Leventhal, L. A. (1979) Z80 Assembly Language Programming, Osborne/
McGraw-Hill.
10. Horowitz, P., Hill, W. (1980) The Art of Electronics, Cambridge University
Press.
11. Rabiner, L. R. and Gold, B. (1975) Theory and Application of Digital Signal
Processing, Prentice-Hall, NJ.
12. Poularikas, A. D. and Seely, S. (1988) Elements of Signals and Systems,
PWS-Kent.
13. Phadke, A. G. and Thorp, J. S. (1988) Computer Relaying for Power Systems,
Research Studies Press, Taunton.
14. Girgis, A. A. and Brown, R. G. (1981) Application of Kalman filtering in
computer relaying, IEEE Trans. PAS-100, pp. 3387-3397.
15. Girgis, A. A. (1982) A new Kalman filtering based digital distance relay, IEEE
Trans. PAS-1OJ, pp. 3471-3480.
16. Johns, A. T. and Salman, S. K. (1995) Digital Protection for Power Systems,
Peter Peregrinus.
17. Sachdev, M. S. (ed.) (1997) Advancements in microprocessor based protection
and communication, IEEE Tutorial Course, 97TP120-0, Chapter 5 on
Algorithms.
18. Sachdev, M. S. and Baribeau, M. A. (1979) A new algorithm for digital
impedance relays, IEEE Trans. PAS-98, pp. 2232-2240.
19. Keeling, D. A. and Pickering, S. D. A. (1997) High speed numerical techniques
for transmission line protection, Proc. 6th Int. Con! On Developments in Power
System Protection, 29-27 March 1997, University of Nottingham, UK, lEE
Publication No. 434, pp. 14-17.
20. Sachdev, M. S. and Giray, M. M. (1985) A least error square technique for
determining power frequency system, IEEE Trans. PAS-104, pp. 437-444.
21. Fromm, W., Halinka, A. and Winkler, W. (1997) Accurate measurement of
wide range power system frequency changes for generator protection, Proc. 6th
Int. Con! On Developments in Power System Protection, 27-29 March 1997,
University of Nottingham, UK, lEE Publication No. 434, pp. 53-57.
22. Phadke, A. G. and Thorp, J. S. (1983) A new computer relay flux-restraining
current differential relay for power transformer protection, IEEE Trans. PAS-
102, pp. 3624-3629.
15
New relaying concepts
INTRODUCTION
The fundamental ideas in protection have developed over many years and
there is now a settled understanding and a maturity of approach to power
system protection. This however does not mean that new ideas cannot find
application in protection work. In recent years, advances in information
technology (IT) have given the impetus to protection engineers to re-
examine the way protection is practised and to seek to learn and benefit
from the new technologies. This process is slow as protection engineers
need to be reassured that the traditionally high reliability of relaying
systems is maintained and that any benefits that new technologies bring
are not accompanied by new and unexpected shortcomings. Therefore,
some of the ideas presented in this chapter have not as yet found
widespread practical application. They are however in the forefront of
considerations as the new generations of relays are designed initially at the
conceptual level and eventually through to prototyping and manufacture.
In the last two chapters two of the major areas where new technologies
have made an impact on the protection field were described. Digital
communications and advanced digital signal processing techniques have
been the prerequisite for considering new relaying concepts. The new
information technologies can make an impact in one of two ways. Firstly,
well established algorithms may be implemented digitally using techniques
described in Chapter 14. This may offer cost and/ or speed advantages over
conventional approaches. In addition, the enhanced processing ability
and flexibility offered when signals are in digital form means that more
complex and sophisticated algorithms may be implemented which were
either too cumbersome or simply impossible using older techniques. An
example of this is the ultra-high-speed schemes described in Chapter 12.
The implementation of old and new algorithms in digital form in the so
called numerical relays is not however the only way that the new
technologies impact on protection. A second possibility now arises which
is fundamentally different to the algorithmic processes typical of

526 New relaying concepts
traditional relaying schemes. Artificial intelligence (AI) techniques attempt
to embody in software the way humans deal with problem solving. In
solving problems humans employ tacit knowledge (knowledge which is
implied but not explicitly expressed) and they do not rely on explicit rules
or algorithms. Some problems in protection, monitoring and control may
be tackled in this way using AI techniques such as expert systems, fuzzy
logic and artificial neural networks. Other developments which are
possible with better communications within the power system open the
possibility of adaptive relaying and a closer integration of protection with
other power system functions. All these topics are briefly reviewed in this
chapter to give protection engineers a basic grasp of technological
developments in this area and permit an assessment of these emerging
technologies for application in power system protection.
15.1 EXPERT SYSTEMS

Expert systems (ES) are computer programs (software) which possess
expertise, which emulates human thinking, in a narrow application area. It
is important to grasp in what wayan ES differs from a conventional
program. In a conventional program large amounts of data are processed
using precisely defined algorithms. Such programs are good at number
crunching and involve strong analytical, deterministic processes. They are
not very good at handling problems which cannot be precisely and
analytically formulated.
In an ES, instead of 'algorithms' 'heuristics' are employed. The term
heuristics refers to rules which are based on good judgement such as is
characteristic of an expert in this field. These heuristic rules are applied
not on data but on knowledge structured in some symbolic way. This
knowledge may consist of facts, objects and their attributes etc. Using
heuristic rules applied on knowledge stored in symbolic form it is possible
to generate new knowledge much like a human thinker may be able to do.
In conventional algorithmic programming, knowledge is represented
procedurally. The program is a description of the processing sequence [1].
The programming found in an ES is described as declarative and it consists
of facts and rules. Before describing in more detail expert systems it is
important to point out that a human thinker possesses far higher capacity
for pattern recognition and can home in on profitable solution strategies
with astonishing efficiency and speed. The breadth of human reasoning, its
scope for imaginative and intuitive thinking far exceeds anything that an
ES can do. However, human numerical (arithmetic) capabilities are limited
compared to those of an ES.
A typical expert system consists of four parts as shown in Fig. 15.1.
The knowledge base contains the expertise of the ES. The rules or
relationships are stored here and it is this part which makes the ES
Expert systems 527
Inference
engine
Knowledge
Database
database
User
interface
Fig. 15.1 Schematic of the building blocks of a typical expert system.
different from a conventional program. These rules are extracted from an

expert in the particular application covered by the ES (the domain expert)
by a knowledge engineer (someone capable of structuring the expertise in
a symbolic form appropriate for the ES). The database contains
information about the particular application. As an example, it may
contain information about the configuration of parts of the power system
which can be used together with the knowledge base to draw conclusions.
The inference engine contains general problem solving knowledge capable
of using the available expertise of the ES. This part of the ES ensures that
all relevant facts are supplied by the user and all rules which may be
applicable are identified. The appropriate rule is then selected and
implemented. This may lead to an action or the implementation of another
rule etc. Inference strategies employ either forward reasoning (i.e. data
driven, suitable for planning applications) or backward reasoning (i.e. goal
driven, suited to diagnostic applications) [2].
Let us consider the rule
if A then B
where A is called the antecedent and B the consequent of the rule. An
equivalent rule is
Bif A
In forward reasoning the first form of the rule is used and the input data
are compared with the antecedent of all rules in the ES and those which
match are considered further. In backward reasoning the second form of
the rule is used but here the consequents of all rules in the ES are examined
to find those which match the goal or objective. Those rules which match
are considered further.
Finally, the interface which typically consists of keyboard and terminal,
is the means by which the user interacts with the ES.
Crucial to the effectiveness of an ES is the knowledge base which is the
result of the interaction between the domain expert and the knowledge
engineer. This knowledge must be structured and represented in a way
suitable for manipulation by the ES. There are several ways in which
knowledge may be represented including semantic networks, production
rules, frames and predicate calculus [1, 3]. Predicate calculus is described
here in some detail as it forms the basis for many of the methods of
structuring and manipulating knowledge. Formal logic is used in mathe-
matics and philosophy to arrange words into statements, manipulate them,
and establish whether they are 'true' or 'false'. A logical proposition is a
statement which may be true or false. A predicate is something (a
property) that is affirmed or denied of a subject in a logical proposition.
As an example, in the statement 'snow is white', 'white' is the predicate.
Predicate calculus describes knowledge in the forms of propositions. It
encompasses semantic rules (rules which relate to the meaning of an
expression, dealing with the relation between symbols and the objects they
refer to) and processing rules (grammar) which can generate new valid
logical expressions by manipulating existing valid logical expressions.
Propositions may be combined using logical connectives such as 'A' which
stands for 'and' and '--+' which stands for 'implies'. Hence the expression
A A B is true if A is true and B is true, otherwise it is false. A simple
example adapted from [1] illustrates predicate calculus. The example has
four statements. The first two which are facts and the last two are rules.
1. Alexander and Katherine are husband and wife.
This statement in predicate calculus form appears as:
husband_and_wife (Alexander ,Katherine)
This simply means that the relationship 'husband and wife' exists between
Alexander and Katherine.
2. Katherine lives in Athens
lives-in (Katherine,Athens)
Meaning that the relation 'lives in' exists between Katherine and Athens.
3. If two people A and B are husband and wife, they are married
husband_and_wife (A,B) --+ married (A,B)
4. If two people C and D are married and D lives in city E then C also
lives in city E.
married(C,D) A lives-in(D,E) --+ lives-in(C,E)
Using these four statements the assertion that 'Alexander lives in Athens'
may be proved thus generating new knowledge. This assertion in predicate
form
Fuzzy logic 529
liveLin(Alexander ,Athens)
matches the conclusion part of rule 4 if C is associated with Alexander
and E with Athens.
We now check the two conditions of rule 4 for validity. Condition
'married(C,D)' is not supported by a fact but it is the conclusion to rule 3
provided the condition of rule 3 is verified. This condition is
husband_and_wife(A,B)
From statement 1 this is true provided that A is associated with Alexander
and B is associated with Katherine.
It follows therefore that the conclusion to rule 3
married(Alexander ,Katherine)
is true. The first condition of rule 4 is thus true.
Let us now look at the second condition of rule 4
liveLin(D,E)
Since D is already associated with B i.e. Katherine, E is associated with
Athens, this proves the conclusion of rule 4 that
liveLin(Alexander ,Athens).
In a complex case of many facts and rules similar checks are made and
if contradictions are not found the conclusion is accepted as true,
otherwise it is considered false. In power system applications facts and
rules relate to the state of circuit breakers, operation of main and backup
protection etc. Several computer languages may be used to develop ES but
the most suitable are PROLOG and LISP.
ES have found most applications in control monitoring and load
restoration applications. A survey of a number of applications may be
found in [4]. Another survey more specific to protection applications may
be found in [2]. System restoration studies are referred to in [5,6].
Application to alarm processing is described in [7]. It appears that most ES
applications are found in event diagnosis, protection settings and
coordination, fault identification and load restoration. Applications of ES
in connection with fuzzy logic are described in the next section.
15.2 FUZZY LOGIC

Engineers are accustomed to deal with data and make decisions on an all-
or-nothing basis [8]. This is reinforced by the binary logic used extensively
in signal processing where two values are accepted '0' or 'I'. However,
human thought processes rarely follow this crisp logic. Instead, vagueness
and grey areas are more characteristic of human thinking until sufficient
options and alternatives have been considered and a definite decision is
reached. Our language abounds with words such as 'not much', 'too bad',
'sort of' etc. One can recognize in these words some of the fuzziness
associated with our perceptions. When we say that someone is 'good
looking' we are not making a judgement which can be described as '1' (in
this case) or '0'. This person is a member of the set of 'good looking
people' with a membership value which ranges between 0 and 1. Thus our
observation results in a fuzzy statement regarding the looks of this
particular person. A similar situation exists in engineering applications. A
current flowing on a line may be 'normal' or 'very high' etc. The 'normal'
current lies in a range and in the middle of this range we assign a fuzziness
or membership value of '1' and at the extremities of the range a value of
'0'. This is shown schematically in Fig. 15.2. A particular current value is a
member of the normal set of currents if it lies within 1\ and 12 and its
degree of membership of the set of normal currents is given by the
membership value ranging between zero and one. Similarly, a current in
excess of 13 belongs to the set of very-high currents with a membership
value depending on its exact magnitude. Fuzzy theory is the branch of
mathematics which studies logical relationships within the context of fuzzy
variables. The mathematical details are outside the scope of this book but
the interested reader may consult specialists texts [9]. In order to help the
reader understand how protection problems may be formulated for
solution using fuzzy logic and show the manner in which results are
obtained, a particular example adapted from [10] is presented in some
detail.
The specific example refers to transmission line protection. The first step
Membership
value
Normal
current Very high current
Current
Fig. 15.2 Typical membership functions for fuzzy variables (solid line). The broken
lines show a possible conventional (crisp) membership function.
Fuzzy logic 531
is the identification of suitable variables which for the particular example
are the current I, voltage V and the corresponding incremental values 11 V
andM.
The second step is the conversion of these variables into fuzzy variables
- a process referred to as fuzzification. This means that the variables are
expressed in linguistic terms which are essentially qualitative and vague.
For the voltage and current five ranges are defined:
Very large value (VL)
Large value (LV)
Normal value (NV)
Small value (SV)
Very small value (VS)
These ranges are shown schematically in Fig. 15.3(a) where at the centre
of each range the membership value is one.
Similarly six ranges are defined to fuzzify I1V and 111:
Membership
value
I or V
Membership
value
LN MN SN 1 SP MP LP
M or ~V
Fig. 15.3 Schematic of the membership functions for the fuzzy variables V, J(a),
and AV, M(b).
large positive value (LP)
medium positive value (MP)
small positive value (SP)
small negative value (SN)
medium negative value (MN)
large negative value (LN)
These ranges and their membership values are shown schematically in
Fig. 15.3(b).
The third step involves the fonnulation of a fuzzy matrix, based on
knowledge about the process, which relates the fuzzy inputs mentioned
above to a set of possible fuzzy outputs. These may, for example, be:
absolute trip (AT)
possible trip (PT)
possible no trip (PN)
no trip (NT)
This involves a substantial amount of work since a 5 x 5 x 6 x 6 = 900
input combinations must be related to the four fuzzy outputs. The type of
rules that may be available to do this are
IF, I is VL, III is LP, Vis VS and AVis LN, THEN the output is AT.
The inference procedure is used to obtain all the required elements of
the fuzzy matrix. The final step is to convert the fuzzy outputs into definite
(crisp) actions - a process described as defuzzification. Various approaches
are available to do this, for example the output with the highest
membership value may be chosen to represent the crisp output.
A block diagram of a fuzzy relay is shown in Fig. 15.4 [9,10]. A survey
reported in [4] indicates that the majority of fuzzy theory applications are
in real-time control and fault diagnosis in many cases in conjunction with
expert systems and artificial neural networks. Examples from fault
diagnosis are reported in [11] and from transfonner protection in [12].
Protected
plant
Crisp output for Sensor
r
plant control data
Defuzzification
H Inference
engine
Fuzzification

Rule base
+
Knowledge
base
Fig. 15.4 Schematic of a typical fuzzy relay.

Artificial neural networks 533
15.3 ARTIFICIAL NEURAL NETWORKS
It was mentioned before in connection with expert systems that brain-like
activity is fundamentally different to normal algorithmic computation.
Although it is not possible to describe fully and imitate completely brain-
like activity, it is nevertheless profitable to examine how humans approach
problem solving and to seek to learn from the manner in which the brain
functions. To this end, a brief description of a biological neuron (the basic
unit of the brain) and its connections to other neurons is presented here.
An artificial neuron is then described and further developed into an
artificial neural network (ANN) which can be used for problem solving.
A biological neuron is shown schematically in Fig. 15.5. It consists of
the nucleus (the 'soma' from the Greek word for body) and a series of tree-
like structures (the 'dendrites' from the Greek word for trees) which are
receiving inputs into the neuron. The stimuli from all the dendrites are
combined to generate an output which is transmitted via the axon to the
dendrites of other neurons. Thus communication between neurons is
through axon-dendrite junctions described as synaptic junctions. It is
believed that the human brain contains 1011 neurons and approximately
10 15 synaptic junctions [8]. Axons can be very long (1 metre in the spinal
cord) and make many complicated connections but each neuron connects
only with a small fraction of the total number of neurons [13]. It appears
Dendrites
Axon from
another
Axon neurone
Fig. 15.5 Schematic of a biological neuron.

that the nervous system has evolved slowly over a very long period to
adapt to the environment in a stable and predictable way. Neurons in the
brain die all the time and new connections are made throughout a human's
life. It is believed that intelligence is based on the rich pattern of inter-
connections inside the brain. The importance of the human nervous system
is confirmed by the fact that although it makes up no more than 2% of
the total body weight it consumes approximately 25% of the body's total
energy requirements [13]. The vast number of neurons and their inter-
connections explain the outstanding capacity of the human brain for
creative, imaginative thinking and its rich pattern of emotions. Only in
number crunching can a conventional computer outperform the human
brain. Currently available artificial neural networks are far inferior to the
human brain. If the number of interconnects and the number of
connections per second are taken as a performance measure, then ANNs
have surpassed worms in performance and are now challenging the
housefly [8]. This fact is at the same time depressing and exhilarating as it
confirms our limitations compared to nature and the scope to design more
powerful ANNs in the future.
Let us now focus on the topology of a typical ANN shown in Fig. 15.6.
This replicates closely the topology of the biological neuron shown in
Fig. 15.5. The inputs XI> X 2 , X3 (which can be as many as required)
represent input from the dendrites. Input Xo is set to one for convenience
to represent, effectively, a bias term. In a biological neuron connections
Fig. 15.6 Schematic of an artificial neuron.

can be broken and new ones established as the brain 'learns' but this is
impractical in an artificial neuron. Instead weights W;, ~, J.f3 etc., are used
to simulate the strength of each particular connection. In effect 'learning'
is the process of determining the correct value for each of the weights.
Clearly, the weights may be changed as new evidence becomes available
and this gives artificial neurons great flexibility and adaptability. After the
inputs are multiplied by the weights the signals are combined to produce
n
U= L:Jt;x j (15.1)
;=.
where n is the total number of inputs. The sum U in equation (15.1) may
be positive (stimulus) or negative (inhibition). The quantity U is called the
activation of the neuron and must be processed further to determine the
final output. The normal approach is to make the output y a non-linear
function of the sum U
y =f(U) (15.2)
This is necessary, since U is the sum of many inputs multiplied by weights
and hence is potentially unbounded in magnitude. The non-linear function
f is chosen to restrict the output within specified limits. Several choices
are available as shown in Fig. 15.7. A hard limiting function is shown in
(a), a linear function in (b) and the popular sigmoid function defined by
feU) = 1/(1 + exp( -kU (15.3)
is shown in Fig. 15.7(c) for two values of parameter k.
An ANN consists of the interconnection of a number of such neurons
in several topologies. Training is the selection of the correct value of
weights to 'learn' a range of tasks. An example is presented here, adapted
from [14], showing in a simple way how a single neuron may be employed
as a pattern classifier.
Let us assume that we have a neuron with two inputs X., X 2 a bias Xo
and corresponding weights W;, Wz and Wo. Then
U = WoXo + WjX. + WzX 2
The hard limiting function shown in Fig. 15.7(a) is chosen as follows
F(U) = {+I if U;;:'O

-I ifU<O
Hence, the output y = feU) is mapped as shown in Fig. 15.8. This
neuron maps all possible combinations of input (X., X 2 ) to only two
possible outputs i.e. y = +I (light area) and y = -I (dark area).
A practical neural network will consist of many interconnected neurons
leading to a generalization of the basic idea presented in Fig. 15.8.
ANN architectures may be divided into three classes, namely, feed-
f(u)
......................,-----
(a)
u
f(u)
..............................., . - - - -
(b)
f(u)
(c)
u
Fig. 15.7 Schematic of non-linear (limiting) functions.
Fig. 15.8 An example of simple pattern classification.

forward networks, feedback networks and self-organizing maps [15]. In
general ANNs may be classified according to the way in which they learn
(learning phase - supervised or unsupervised) and according to the way
they process new data (recall phase - feedforward or feedback).
In supervised learning, the teacher presents an input to the network,
obtains its output and compares with the desired output. The difference is
then used to alter the weights. In unsupervised learning only inputs are
provided. In the former case cross-correlation between the error signal and
the input is involved, while in the latter, the input is cross-correlated with
the output.
Crucial in supervised learning networks is the algorithm used for weight
adjustment. Let us examine a simple case illustrating how this can be done.
Consider the case of a neuron with bias and two inputs where for a
particular input test pattern the desired output is Yo. The actual output
obtained from the neuron is, say,
Yo = Wo + wd(x 1) + wd(x2)
Hence the error is
(15.4)
The training objective is to minimize the mean squared error for a
succession of input test patterns (delta rule). We are looking for the
minimum point (the bottom of the hill) on the surface describing the error
as a function of the weights. To find our way to the bottom we work out
the gradient (slope) and move in the opposite direction. The gradient for
weight ~ is
(15.5)
where the chain rule has been applied. From (15.4) the derivative of E with
respect to ~ is
hence
(15.6)
Moving in the opposite direction to the gradient (to reach minimum) and
adjusting the weight by an amount proportional to the gradient
(15.7)
should get us closer to required weight. The parameter '1 is known as a
learning rate and is in the range zero to one. It is chosen smaller with each
search. Similar procedures may be used for the other weights to make
adjustments for a particular input test pattern. This process is repeated for
all the input test patterns so that learning by the network follows the
average gradient.
The unsupervised learning process is more difficult to explain. The
reader is referred to specialist texts [13].
Some of the common types for ANNs are the multilayered feedforward
with error back propagation learning shown in Fig 15.9 and the self-
organizing map shown in Fig. 15.10. The former example has the learning
strategy similar to the one described above. The network shown in
Fig. 15.10 is an example of unsupervised learning network and is known as
the Kohonen layer or map. Each input is connected to all the units forming
a two dimensional layer (only one input is connected for clarity). This self
organizing network can map multi-dimensional data into categories. If
two sets of input data are similar then they are mapped in units close to
each other on the Kohonen layer. During the learning stage, the weight of
the most active unit and its neighbours are modified to move closer to
the current input. Thus if patterns exist in the input data, these will be
mapped in distinct areas in the layer. This allows categorization of the
input data without the need to know what is the desired 'output'.
Networks like the one shown in Fig. 15.9 with supervised error back
propagation may display slow convergence or convergence to local
Sa. '"a.
:;
c: :;
o
Input Hidden Output

layer layer layer
Fig. 15.9 Schematic of a three-layer feedforward ANN.

Adaptive relaying 539
Fig. 15.10 The Kohonen layer architecture. The most active unit for a test pattern
and its neighbours are shown.
minima. Self-organizing maps require substantial amounts of input data

for training. Such data can sometimes be obtained by numerical
simulation. A discussion of ANNs in connection with protection may be
found in [16]. Their advantages are perceived to be their capacity for
implementing adaptive behaviour, their ability to generalize and handle
data outside the original training set, their inherent parallelism and
handling of non-linearities. On the negative side, it is sometimes difficult
to provide sufficient data for training and there is no clear procedure for
choosing the best architecture, number of units, etc. Development tools for
ANN s are surveyed in [17]. In the area of power system protection the
use of ANN is described in [18] for autoreclosing applications, in [19] for
phase selection, in [20] for series compensated lines, in [21] for transmission
line protection and in [22] for transformer protection. Other applications
are reported in [23,24]. Hybrid approaches have been described, whereby
an expert system is used to identify a faulted section based on waveform
recognition performed by an ANN [6]. A survey of ANN applications in
power systems is reported in [4].
15.4 ADAPTIVE RELAYING

The introduction of numerical protection relays and better communi-
cations between relays and other power system functions has opened the
possibility of protection which is capable of automatically adjusting
settings, selecting the most suitable algorithm, compensating for errors
etc., so that the optimum protection function is implemented for the
prevailing power system conditions. To some extent all relays are adaptive.
An overcurrent relay has an operating time which gets smaller as the fault
current increases. In a traditional relay of this type the adaptation cannot
be complete, as the operating time depends on the current and time
multiplier settings which remain fixed. If these settings could be changed
automatically to reflect the prevailing conditions then faster operation
could be achieved [14]. These principles if applied widely will result in an
improved protection performance.
Lists of application areas in protection where there is scope for adaptive
relays are given in [14,25]. Application areas include:
Selection of the most suitable algorithm,
Change settings according to system conditions,
Transformer differential protection,
Compensating for errors,
Compensating for mutual coupling,
Protection of multi-terminal lines,
Back-up protection,
Detection and control of power swings.
A comprehensive discussion of these and other areas where adaptive
protection is of value may be found in [14,25-28].
A precondition for the introduction of adaptive techniques is the
availability of data from different parts of the system and the capacity to
process and evaluate data for relaying purposes. The availability
requirement depends on the existence of a fast reliable communication
system as described in Chapter 13. Processing can be done provided that
numerical relays are available. The basic operations required to extract
information from data including filtering, parameter estimation etc., were
described in Chapter 14. The approach used will vary in each application
but artificial intelligence techniques such as expert systems and artificial
neural networks described in previous sections may be particularly
advantageous. Some examples are presented here to illustrate the potential
benefits of adaptive relays.
In case of failure of a relay, it may be possible to change the back-up
protection so that tripping time is reduced. This can be done automatically
by intelligence incorporated within the substation. A prototype relay
permitting the instantaneous operation of a back-up relay when the main
relay has failed is described in [29].
An example of an adaptive scheme for a distance protection which takes
account of fault resistance and the remote-end in feed on reach accuracy is
reported in [30]. The adaptive setting of zone 3 distance protection is described
in [31]. Adaptive autoreclosure techniques using ANNs are explored in [32].
An adaptive protection scheme for a city-wide distribution network is
presented in [33]. The scheme described in this paper operates at three
levels: central computer, substation computer and relay. System
configuration, fault level and load changes are assessed by the substation
computer and communicated to the central computer. Relays monitor
voltage, current and state of switches and communicate this data to the
substation computer. The central computer then determines the most
appropriate settings for the prevailing system conditions and communicates
Integrated protection 541
them to the relays via the substation computer. The overcurrent relays
are based on the TMS 320C25 digital signal processor and implement all
standard functions including directional features. The voltage and current
phasors are obtained using a least error square filtering technique.
Dedicated software is available to assess network topology, estimate
current and voltages at branches and nodes of the network, calculate fault
current levels, and calculate revised relay settings. The coordinated relay
settings are obtained by a linear programming technique based on the
Simplex method [34]. Communication software to allow information
exchange between relays, substation computer and central computer are
also installed. Updates of settings are done at hourly intervals with default
setting determined locally in case of communication system failure. The
prototype implementation described in [33] uses seven personal computers
(PCs). Five of these are used to host the relay card which is based on the
TMS 320 DSP. One PC is used as a substation computer and one as the
central computer. To exchange information between relays and the
substation computer a local area network (LAN) is used based on the
token ring format (IEEE802.5). Physical connection is through shielded
twisted pair cable. Communication with the central computer is through
an RS-232 modem connection. System studies done using this prototype
have shown that relay times are reduced and selectivity is improved.
Many utilities are considering the benefits that adaptive protection can
bring and operating experience is gathered from which a quantitative
measure of these benefits may emerge. Inherent in adaptive protection is
the need for reliable standardized communications. All adaptive schemes
involve larger areas of the system (as several relays may be affected) and
distributed intelligence than for traditional stand-alone relays. This
inevitably increases complexity and must be carefully managed to maintain
high reliability and ease of maintenance. Testing of adaptive relays must
consider the implications of the greater complexity of these schemes. Many
ideas for adaptive schemes involve a closer integration between the
protection, measurement and control functions. The options for in-
tegration are explored further in the next section.
15.5 INTEGRATED PROTECTION

In a modern power system a large amount of data is available at anyone
time. The information contained in the great diversity of data may be
classified according to various criteria:
functional purpose
reaction time
priority in terms of security of the power system
volume of data
In tenns offunctional purpose the following functions may be identified.
(i) Power system protection (e.g. main, backup protection etc.). This
function requires fast reaction time (in the order of msec) and has
very high security implications.
(ii) Abnonnal condition monitoring (e.g. load and frequency control,
load shedding etc.). Reaction time varies between msec and seconds
and security implications are high.
(iii) Automatic control (e.g. autoreclosing). Reaction time is in the msec
and second ranges and security implications are high.
(iv) Supervisory control (e.g. synchronizing, alarm processing). Reaction
times in msec and second range and high to medium security impli-
cations.
(v) Metering (e.g. tariffs, measurements). Reaction times range widely
and security implications are medium.
(vi) Event recording and monitoring (e.g. fault location, disturbance
recording). Reaction time is minutes and security implications are
low to medium.
(vii) Data analysis (e.g. analysis of long-tenn perfonnance data).
Reaction time is days or months and security implications are low.
(viii) Engineering (e.g. maintenance). Reaction times are generally long
and security implications are medium.
It is difficult to give the full picture of all relevant power system
Fig. 15.11 Traditional weakly coupled arrangement.

Power system
Fig. 15.12 A closely coupled arrangement.
functions and reflect correctly the complexity and importance of some of

the tasks. This list should be seen as an attempt to illustrate the diversity
and amount of data that is available. In terms of the volume of data
transmitted at anyone time there are also wide variations depending on
the particular function involved. For example, data volumes for fault
recorder data may be classed as medium in size, whilst transmission of
visual information involves very high volumes.
Traditionally, protection and control functions were kept essentially
apart with limited interaction as shown in Fig. 15.11. The arrows show the
information flows. However, it is clear from this diagram that protection,
local control and central (remote) control operate practically indepen-
dently and share only important operational data. The different functions
are weakly coupled. Modern technology with numerical relays of very high
functionality, sensors, and advanced communications, has raised the
question of whether a more closely coupled arrangement is possible as
shown in Fig. 15.12. Some of the functionality associated with protection,
control, and monitoring functions, is now shared and the distinctions
between the three functions are blurred. Numerical relays possess very
high processing capabilities and in a single unit it is now possible to
incorporate up to twenty distinct protection functions. Control functions
could also be incorporated as already done at the medium voltage levels. A
single unit may contain main and backup protection, autoreclosing, fault
locator, synchronizing etc. Practising power system engineers therefore
ask the question as to how far this closer coupling can and should go and
what the implications are for reliability and availability. Technological
developments with high processing rates, parallelism of tasks and
improvements in the quality and standardization of communications,
suggest that a closer coupling is possible. Protection functions will then be
able to share data available from the Supervisory Control and Data
Acquisition System (SCADA) and the Energy Management System
(EMS). SCADA is responsible for retrieving and transmitting data from
substations to the central control computer and EMS processes these data.
A high degree of integration already exists between SCADA and EMS.
References in technical literature on closely coupled protection, control
and monitoring sometimes use the term 'coordinated protection' or,
alternatively, 'integrated protection'. Coordinated protection refers to a
system where protection operation is essentially autonomous, with its own
data acquisition, processing and communications. Interaction with the
other functions is only for non-critical tasks. In contrast, integrated
protection refers to the situation where protection shares processing, data
acquisition and communications with other functions. These two
approaches are shown schematically in Figs 15.13 and 15.14. The dividing
line is not however very clear and confusion often arises.
The perceived advantages of coordinated/integrated over segregated
protection, monitoring, and control are [16,35-37]:
reduced commissioning costs based on pre-tested modular units
lower hardware and installation costs due to the use of standard units
and less cabling
savings in space required for installation
high availability of data, better self-monitoring facilities
Primary plant
Substation
....----L.L...---, } bay
L...----,,~---l
U
SCADA
Fig. 15.13 An example of a coordinated architecture.

Primary plant
II II II
Sensors! Sensors! Sensors!
actuators actuators actuators
II /I II Substation
bay
Protection! Protection! Protection!
monitoring! monitoring! monitoring!
n n
control control control
n
n
Communication bus
SCADA
Fig. 15.14 An example of an integrated architecture.
ability of reprogramming, to cope with changes in technology, system

configuration, and expansion.
Factors which may be seen as difficulties or disadvantages are:
redundancy and duplication are required to increase reliability;
common-mode failures must be avoided
high self-test provisions are necessary to maintain the highest possible
availability of the most critical functions such as protection
highly standardized communication interfaces to allow for interchange-
ability and expansion
testing and maintenance of software has to be thought out carefully to
reflect the changed nature and the closer coupling between functions
electromagnetic compatibility measures must be designed in and main-
tained throughout the lifetime of the system
The degree of integration and the speed at which closer integration
proceeds will depend on the balance between perceived advantages and
disadvantages and an evaluation of the practical experience obtained from
such systems.
Some examples of manufacturers' products which adopt, to some extent,
the philosophy of tighter coupling between protection, instrumentation
and control are available.
The K Range of multifunctional relays provides comprehensive protection
facilities, control and data gathering. It includes functions such as broken
conductor detection, thermal overload alarm and trip, underfrequency
protection etc. Serial communications and interfacing to SCADA are also
provided [38]. Another example of a coordinated protection and substation
control system is the LSA675 [39]. It has a coordinated system archi-
tecture, high processing capabilities, an optical communications network
with serial interfaces and covers a comprehensive range of protection and
control functions. ABB markets a total coordinated protection and control
concept known as PYRAMID [40]. This system has comprehensive
protection and control functions, fibre optic communications, self-
supervision and diagnostics.
15.6 THE GLOBAL POSITIONING SYSTEM (GPS)

The wider availability in recent years of highly accurate synchronized time
measurements has opened new possibilities for protection. The measure-
ment of phasors at different parts of the power system requires a common
time reference. This reference can be provided by synchronization pulses
transmitted along fiber optic links or by the satellites of the Global
Positioning System (GPS) [41]. The wide coverage and accuracy of GPS
make its use the preferred option. The GPS system consists of 24 satellites
on six orbits at a height of 10898 miles. Each satellite has an unobstructed
view of 42% of the globe. Signals are transmitted at two frequencies
(1227.6 and 1575.42 MHz). The GPS master clock is kept within lOOns of
Universal Coordinated Time (UTC). Messages received from a GPS
satellite contain information about the location of the satellite and a date
and time stamp.
A GPS receiver receives time information about the position of the
satellite (x j Yj. Zj), and the GPS system clock. The differences in time of
arrival from the four satellites and the receiver clock can be calculated.
Four equations may be formulated with four unknowns the receiver's
location and its time difference with respect to the GPS master clock.
(Xj - xi + (yj - y)2 + (Zj - Z)2 = (D j - ci (15.8)
where (x. Y. z) is the position of the receiver, Dj = (Atj)c is the distance
between receiver and satellite. C is a clock bias error due to lack of
synchronization between the satellite and receiver clocks.
The non-linear equation (15.8) may be solved to obtain the position of the
receiver and the time difference between the receiver's clock and the GPS
clock. For a fixed receiver, once its position has been thus determined, only
one satellite signal is necessary to give it accurate time information with
respect to UTC. The timing accuracy available for commercial applications
is better than 1 ~s. Signals intended for military use and not available for
commercial applications give even better accuracy.
Protection relays may thus exchange information with the appropriate
time stamp so that synchronized phasor measurements may be made at
points separated by large distances [14,16]. An example of the timing
accuracy required for differential protection is given in [16,42]. If the
Genetic algorithms and simulated annealing 547
timing error is At then the phase error is
A8 = 2At2n! (15.9)
The difference in measured current between the two points is
Ipk sin(rot + A8) - Ipk sin(rot - A8) = (2Ipk sin A8) cos(rot)
The expression in brackets is the error in current due to timing errors i.e.
M/lpk = 2 sin A8 (15.10)
For At = 1 JlS and! = 60Hz, it follows that M/lpk = 7.5 X 10-4 This error
is far lower than is normally required (4%). Even at a timing accuracy of
20 JlS the current measurement error will be within the above normal limit.
With four satellites visible at anyone location (and only one strictly
required) the probability of losing all four is considered very low. Even
under these circumstances receiver clock accuracy will take several hours
to deteriorate beyond 20 JlS. The long term future of the GPS system
appears to be secure, hence its use for power system protection is actively
investigated.
15.7 GENETIC ALGORITHMS AND SIMULATED ANNEALING

Many problems in power systems involve a process of searching for an
optimum solution. For this type of problem a heuristic search may be
implemented by techniques based on genetic algorithms (GA) and
simulated annealing (SA) [43].
Genetic algorithms are based on ideas borrowed from genetic evolution
in biological systems [44], and simulated annealing on the cooling of metals
from the liquid to the solid state [45]. Some details of the GAs are given
below.
The ideas behind a genetic algorithm may be described simply as
follows:
Using a random generation process genes are produced. A 'gene' is a
binary string representing an encoding of a parameter. A sequence of
genes produces a 'chromosome'.
The merit of each chromosome is then evaluated using an appropriate
cost function and the various chromosomes are arranged in order from
the least fit to the most fit.
Unfit chromosomes are discarded and the ones that survive are used as
genetic material to produce new offspring. This process is done so that
an equal number of new, better chromosomes is produced to replace
those discarded, thus leaving the total number of chromosomes
constant. This is a procedure akin to natural selection occurring in
nature.
The process of reproduction from the most promising chromosomes
may involve a simple crossover as shown below:
Parent 1
0101 / 1010 Child 1
'\. /
gene1/gene2 010111111
Parent 2
0000 / 1111 Child 2
'\. /
gene1/gene2 000011010
Another feature which may be introduced during reproduction is a
random mutation. One percent of bits in a chromosome may be changed
to guard against the possibility of the optimization process converging
to a local minimum .
The reproductive procedure is repeated for a fixed number of times or
when an acceptable solution is reached. Knowing when to stop is not a
simple matter and the computational effort to cover the problem space
and to evaluate solutions can be very considerable.
Examples of the application of GAs to generator protection have been
reported in [46]. GA techniques may be used for overcurrent relay co-
ordination. Further details of GA and SA techniques may be found in
[4,16,43,47,48].
15.8 FLEXIBLE AC TRANSMISSION SYSTEMS (FACTS)

Conventional power systems are characterized by the free flow of energy
determined by the characteristics of the network. This situation is illus-
trated in Fig. 15.15(a). Once the phase angle, the line voltages at ends A and
B and the line impedance have been set, the power flow between A and B
is fixed. Any change in power flow is only possible by slow mechanical
controls on the drive plant. For the configuration shown in Fig. 15.15(b) if
one of the two parallel lines is lost, due to a fault, there is very little that
can be done quickly to control power flow on the remaining healthy line.
This has serious stability implications which in turn affect protection
arrangements. In order to cope with such situations it is advisable to keep
the loading of the link below its full capacity and/ or construct additional
transmission capacity. These difficulties are a consequence of the fact that
power flow in each of the two parallel lines cannot be set independently -
the power system is inflexible. One way of controlling power flow is to
introduce a d.c. link as shown in Fig. 15.15(c). The disadvantage of this is
Flexible A C transmission systems (FACTS) 549
(a)
(b)
(c)
~-----iJu
HVDC link
(d)
~r---k?
Facts device
Fig. 15.15 Inflexible and flexible power systems.
that a completely new design for lines and sending/receiving stations is

necessary. An exciting alternative was described in [49] and is known as a
Flexible AC Transmission System (FACTS). The essence of this arrange-
ment is that power flows through several semiconductor switching devices
(thyristor, GTO, etc.) and thus is electronically controlled. This is shown
schematically in Fig. 15.15(d). A typical FACTS device may be a shunt
connection of variable inductance and capacitance, or a variable series
capacitance. An example of a thyristor-controlled series capacitor is shown
in Fig. 15.16. FACTS permit high-speed control of the power system and
more precise routing of energy transfer across the network. DC links and
the introduction of FACTS will have substantial impact on protection
arrangements as conventional protection schemes must be modified to
coordinate with the switching of the semiconductor devices and to
Fig. 15.16 Thyristor-controlled series capacitor.

recognize the different protection requirements. Further details may be
found in [50-52].
15.9 EMBEDDED GENERAnON

The development in the late 50s of the HV transmission network has led
to a power system configuration where very large generators supply
electrical energy which flows in most cases from high to low voltage levels.
In recent years concern about the difficulty of erecting new HV lines in
populated areas, recognition of the high efficiency of combined heat and
power (CHP) schemes, and commercial/legislative pressures have opened
the possibility of local generation connected at lower voltage levels. These
small embedded generators are not at present directly controlled by central
grid control centres. The issues which need to be addressed as regards
embedded generation are [53]:
power flow and losses
power quality
changes in fault levels
transient stability
protection
earthing.
Protection of embedded generation schemes presents several difficulties.
Generator protection must be capable of detecting internal generator faults
by monitoring current flow from the network into the generator. Also,
protection must be included to ensure that the embedded generator does
not feed into a faulted or islanded system. Normally, protection is
provided by the electricity utility at the interface to the embedded scheme
to provide backup against the generator feeding into a faulty network, and
to clear faults internal to the embedded scheme. Particular protection
difficulties are the lack of sufficient fault current flowing from the
generator to trip overcurrent relays following a fault on the high voltage
side of the delta-star transformer connected to the generator, and the
detection of islanded operation. Further details may be found in the
reference cited above.
15.10 THE FUTURE

Several relatively new ideas and technologies for protecting and controlling
power networks were presented. The extent to which these will impact on the
operation of power systems is difficult to predict. However, it seems certain
that a greater degree of control and a closer integration of protection and
control functions will take place as utilities and protection engineers adapt
to technological opportunities, commercial and environmental pressures.
References 551
REFERENCES
1. Nebendahl, D. (ed.) (1988) Expert System: Introduction to the Technology and
Applications, Siemens AG and John Wiley.
2. IEEE Working Group DlO, Power System Relaying Committee (1994)
Potential applications of expert systems to power system protection, IEEE
Trans. On Power Delivery, 9, No.2, 720-728.
3. Lo, K. I. and Nashid, I. (1993) Expert systems and their application to power
systems, Power Engng J. (in three parts): Part 1: Components and methods of
knowledge representation, Feb. 1993, 41-45; Part 2: Search methods and
languages, June 1993, 141-144; Part 3: Examples of application, Oct. 1993,
209-213.
4. Dabbaghchi, I., Christie, R. D., Rosenwald, G. W. and Liu, C. C. (1997) AI
application areas in power systems, IEEE Expert Intelligent Systems and their
Applications, 12, No.1, 58-66.
5. Kimura, T., Nishimatsu, S., Ueki, Y. and Fukuyama, Y. (1991) Development
of an expert system for estimating fault reaction in control centers based
on protective system simulation, Proc. lASTED Con!, Tainan, March 1991,
91 SM 356-6 PWRD.
6. Fukuyama, Y. and Ueki, Y. (1991) Development of an expert system for
analysing faults in power system using artificial neural network based
waveform recognition, IEEE Int. Symp. On Circuits and Systems, Singapore,
June 1991, 1137-1140.
7. Young, D. J., Lo, K. L., McDonald, J. R., Howard, R. and Rye, J. (1992)
Development of a practical expert system for alarm processing, lEE Proc., C,
139, No.5, 437-447.
8. Chester, M. (1993) Neural Networks - A Tutorial, Prentice Hall, NJ.
9. Ross, T. 1. (1995) Fuzzy Logic and Engineering Applications, McGraw-Hill.
10. Song, Y. H., Aggarwal, R. K. and Johns, A. T. (1993) Fuzzy logic in the
development of new protection equipment for power systems, 28th Universities
Power Engineering Conference Proceedings, Staffordshire University, 21-23
Sept., 1993, Vol. 1,97-100.
11. Monsef, H., Ranjbar, A. M. and Jabid, S. (1997) Fuzzy rule-based expert
system for power system fault diagnosis, IEEE Proc. Gener. Transm. Distrib.,
144, No.2, March 1997, 186--192.
12. Wiszniewski, A. and Kasztenny, B. (1993) Fuzzy set approach to transformer
differential relay, Fifth Int. Con! in Development in Power System Protection,
lEE Conf. Publication 368, University of York, 30 March - 1 April, 1993,
169-172.
13. Anderson, J. A. (1995) An Introduction to Neural Networks, MIT Press.
14. Sachdev, M. S. (coordinaror) (1997) Advancements in Microprocessor Based
Protection and Communication, IEEE Tutorial Course, 97 TP120-0.
15. Kohonen, T. (1990) The self-organising map, Proc. IEEE, 78, No.9, 1464-
1480.
16. lEE (various authors) (1997) Tutorial on Modern Techniquesfor Power System
Protection, University of Nottingham, 24 March 1997.
17. Hammesstrom, D. (1993) Working with neural networks, IEEE Spectrum, July
1993, 46--53.
18. Aggarwal, R. K., Johns, A. T., Song, Y. H., Dunn, R. W. and Fitton, D. S.
(1994) Neural network based adaptive single pole autoreclosure technique for
EHV transmission systems, lEE Proc. Gener. Transm. Distrib., 141, No.2,
155-160.
19. Bo, Z., Aggarwal, R. K., Johns, A. T., Li, H. Y. and Song, Y. H. (1997) A
new approach to phase selection using fault generated high-frequency noise
and neural networks, IEEE Trans. On Power Delivery, January 1997, No. 12,
106-115.
20. Song, Y. H., Johns, A. T. and Xuan, Q. Y. (1996) Artificial neural-network-
based protection scheme for controllable series-compensated EHV trans-
mission lines, IEEE Proc. Gener. Transm. Distrib., 143, No.6, 535-540.
21. Sidhu, T. S., Singh, H. and Sachdev, M. S. (1995) Design implementation and
testing of an artificial neural network-based fault direction discriminator for
protecting transmission lines, IEEE Trans. On Power Delivery, 10, No.2,
697-706.
22. Perez, L. G., Flechsiz, A. J., Meador, J. L. and Obradovic, A. (1994) Training
an artificial neural network to discriminate between magnetizing inrush and
internal faults, IEEE Trans. On Power Delivery, 9, No.1, 434-441.
23. Sultan, A. F., Swift, G. W. and Fedirchuk, D. J. (1992) Detection of high
impedance faults using a multi-layer perceptron, IEEE Trans. On Power
Delivery, 7, No.4, 1871-1877.
24. Dalstein, T. (1995) Neural network approach to fault classification for high-
speed protective relaying, IEEE Trans. On Power Delivery, 10, No.2, 1002-1011.
25. Phadke, A. G. (1995) Adaptive protections and control, ClGRE Colloquium
SC 34, Stockholm, 11-17 June 1995, paper 34-212.
26. Thorp, J. S. (ed.) (1993) Feasibility of adaptive protection and control, IEEE
Trans. On Power Delivery, 8, No.3, 975-983.
27. Horowitz, S. H., Phadke, A. G. and Thorp, J. S. (1988) Adaptive transmission
system relaying, IEEE Trans. On Power Delivery, 3, No.4, 1436-1445.
28. Rockfeller, G. D., Wagner, C. L., Linders, J. R., Hicks, K. L. and Rizy, D.
T. (1988) Adaptive transmission relaying concepts for improved performance,
IEEE Trans. On Power Delivery, 3, No.4, 1446-1458.
29. Eisman, J., Gomez, G., Torres, J. and Cobelo, F. (1995) ClGRE Colloquium
SC 34, Stockholm, 11-17 June 1995, paper 34-207.
30. Xia, Y. Q., Li, K. K. and David, A. K. (1994) Adaptive relay setting for
stand-alone digital distance protection, IEEE Trans. On Power Delivery, 9,
No. 1,480-491.
31. Stedall, B., Moore, P., Johns, A. T., Goody, J. and Burt, M. (1995) An
investigation into the use of adaptive setting techniques for improved distance
back-up protection, IEEE, 95 SM 435-8 PWRD.
32. Fitton, D. S., Dunn, R. W., Aggarwal, R. K., Johns, A. T. and Bennett,
A. (1995) Design and an implementation of an adaptive single-pole auto-
reclosure technique for transmission lines using artificial neural networks,
IEEE, 95 SM 432-5 PWRD.
33. Sachdev, M. S., Sidhu, T. S., Chattopadhyay, B., Ta1ukdar, B., McDonald,
G. and Chan, R. (1995) Design and evaluation of an adaptive protection
system for a distribution network, ClGRE Colloquium, SC 34, Stockholm, 11-
17 June 1995, paper 34-202.
References 553
34. Chattopadhyay, B., Sachdev, M. S. and Sidhu, T. S. (1995) An on-line relay
coordination algorithm for adaptive protection using linear programming
technique, IEEE, 95 WM 035-6 PWRD.
35. Booth, C., McDonald, J. R., Laycock, W. J. and Bennett, A. (1995)
Distributed intelligence within power systems: the future for substation control
and management systems, CIGRE Symp. on Integrated Control and Com-
munications Systems, August 1995, Helsinki.
36. Mitchell, B. C. (1996) Advanced features of the integrated control and
protection system used in Ontario Hydro substations, CIGRE 1996 Meeting,
August 28-30, paper 34-101.
37. Electricity Training Association (eds) (1995) Power System Protection, Vol. 4:
Digital protection and signalling, published by the lEE.
38. GEC Alsthom Protection and Control (1996) Publication R6500A.
39. SIEMENS AG, Protection and Substation Control Systems (1996) Publication
E50001-U321-A92-X-7600, 6/46-6/73.
40. ABB Network Control and Protection (1992) Publication IMDB00004-EN.
41. Kaplan, E. D. (1996) Understanding GPS - Principles and Applications, Artech
House.
42. Southern, E. P., Li, H. Y., Grossley, P. A., Potts, S. and Weller, G. C. (1997)
GPS synchronised current differential protection, Dev. In Power System
Protection Proc., lEE Con! Publication 434,25-27 March 1997, University of
Nottingham, 342-345.
43. Davis, L. (ed.) (1987) Genetic Algorithms and Simulated Annealing, Morgan
Kaufmann Pub!.
44. Holland, J. H. (1992) Genetic Algorithms, Scientific American, 66--72.
45. Kirkpatrick, S., Gelatt, C. D. and Vecchi, M. P. (1983) Optimisation by
simulated annealing, Science, 220, 671-680.
46. Sutoh, T., Suzuki, H. and Nagai, N. (1994) Large scale generator maintenance
scheduling using simulated evolution, Proc. ISAP, 94, 567-573.
47. Goldberg, D. E. (1989) Genetic Algorithms, Addison-Wesley, NY.
48. Haupt, R. L. (1995) An Introduction to genetic algorithms for electro-
magnetics, IEEE Antennas and Propagation Magazine, 37, 7-15.
49. Hingorani, N. G. (1988) Power electronics in electric utilities: role of power
electronics in future power systems, Proc. IEEE, 76, 481-482.
50. lEE Colloquium (1994) Flexible AC Transmission Systems (FACTS) - The
key to increased utilisation of a power system, London, 12 Jan. 1994, Digest
No. 1994/005.
51. Moore, P. and Ashmole, P. (1995) Flexible AC transmission systems, Power
Engng J., Part 1, 9, 282-286; Part 2 Dec. 1996, 273-278; Part 3 Aug. 1997,
177-183.
52. Andersson. F. and Juhlin, L.-E. (1992) Power system protection in HVDC
environments, ABB Rev., 3/92, 27-32.
53. Jenkins, N. (1995) Embedded generation, Part 1, Power Engng J., 9, 154-150,
and Part 2, October 1996, 10, 233-239.
Appendix A
The testing and application of
power-system protective equipment
The capital costs of power-system equipment are high and this is particularly
so for major items such as large turbines, alternators, transformers and over-
head lines and cables. They represent investments which provide subsequent
income derived from the payments made for the electrical energy supplied to
consumers. It is therefore vital that the maximum continuity of supply be
maintained to consumers and that power-system equipments should not suffer
unavoidable consequential damage when faults occur within networks. Similar
considerations also apply to smaller installations such as those in manufactur-
ing plants.
To satisfy these requirements, protective equipments, such as those de-
scribed in the earlier chapters, must be provided and their not insignificant
costs may be considered to be insurance premiums, because they do not
directly increase the income from consumers. Having paid the premiums,
supply authorities and others understandably expect protective equipments to
operate correctly under all conditions. To achieve this goal it is necessary that
they and the suppliers of protective equipment ensure that the necessary
studies are conducted so that suitable protective schemes are applied and that
the appropriate fault settings are chosen for each application. In addition,
adequate development, production and site tests must be performed.
The procedures and tests which are undertaken to ensure that the various
types of protective equipments now available will meet these requirements are
examined in the following sections.
A.I PERFORMANCE REQUIREMENTS

It will be evident that a knowledge of the performances required of protective
equipments is needed both when new schemes are to be developed or when
existing schemes or items are to be applied. Such information may be derived
from studies of the behaviour of power systems or from the manufacturers of
items of equipment and it will affect the selection of the types of protective
equipment needed for different applications. As examples, fuselinks may be
556 Appendix A
used to protect relatively low voltage loads or components whereas distance-
or impedance-measuring schemes may be applied to major transmission lines
and in each case the information required to apply the protection will be
different, as will the testing which must be done.
A.2 TESTING OF PROTECTIVE DEVICES AND EQUIPMENT

The testing requirements for the various forms of protection are considered
below.
A.2.1 Fuselinks
Fuselinks should ideally operate at currents above their minimum-fusing
levels in times slightly shorter than those for which the items or circuits being
protected by them can withstand the currents without being damaged. Clearly,
therefore, the withstand time/current characteristics of items to be protected
must be known and at very high current levels the [2 t, i.e.
values which may be withstood must also be known.

Over the years ranges of fuselinks have been produced with known and
published characteristics and in many cases fuse links with characteristics
suitable for particular applications can be selected from those already avail-
able.
The testing of new designs of fuselinks

Unlike other items of protective equipment, an individual fuselink cannot be
fully tested to determine its time/current characteristic because it is destroyed
when it operates for the first time. Consequently, when a new fuselink is being
developed, a large number of prototypes are produced, care being taken to
ensure that they are all as similar as possible. The actual tests done depend on
the particular fuse links being examined, but in general their dimensions are
checked to ensure that they are within the tolerances that will be used in the
subsequent volume production. The electrical resistance of each fuselink is
measured in an ambient temperature, typically in the range 20-25C and
normally, in the case of low voltage fuselinks, the power dissipated by each
fuselink at its rated current when it is mounted in a standard test rig is
measured. The prototype fuse links which meet the necessary conditions are
then used in the following tests.
The determination of conventional fusing current Conventional fusing cur-

Testing ofprotective devices and equipment 557
rent is strictly the current which is needed to cause a fuselink to operate after
an infinite time and its determination is clearly impractical. British and Inter-
national Standards therefore define the conventional current as that current
which will cause operation in a given period, say 4 hours, but even the
determination of this value would be very time consuming and therefore to
simplify the situation, lEe specifications require. that all of a specified number
of fuselinks mounted in a standard type-test rig should operate in less than the
conventional time, for example 4 hours, and not operate in the conventional
time when carrying the conventional non-fusing current. The conventional
time, which depends on the current rating, is specified in the various standards.
For industrial fuses the conventional fusing current is 1.6 x (rated current) and
the conventional non-fusing current is 1.25 x (rated current).
The determination of breaking-current capacity Fuselinks must be capable

of operating in service under the most onerous conditions which can arise.
Their breaking capacities must therefore be determined from tests done under
specified conditions in single-phase inductive circuits of low power factor
(typically 0.2 for low voltage fuselinks), arranged as shown in Fig. A.1.
Equipment must be included to enable test circuits to be closed at any desired
points in the voltage cycle so that conditions of varying severity may be
produced. The fuselinks are mounted in standard rigs and tests are performed
to determine not only their breaking capacities but parameters such as [2 t let
through, arc voltages and cut-off currents and when the latter are being found,
the circuits must be switched so that arcing commences just prior to an instant
when the voltage is at its peak value. During testing the appropriate source
voltages specified in the various standards must be provided, for example lO%
more than the rated voltage of low voltage fuselinks, and the voltages must
be maintained for at least 30 s and in the case of high voltage fuselinks they
must be maintained for 60 s.
Type tests must be done on fuselinks over the range of currents between
their maximum and minimum breaking capacities and at the lower-current
Variable impedance
Make
CB switch
--rr-....~---:;;;ro...--X - - - - 0 -
Voltage Fuse under

source test
Fig. A.I Fuse test circuit.

558 Appendix A
levels the power factors of the test circuits are usually higher than those used
at the maximum. This takes account of the situations which arise in service
and IEC specifications usually require tests at high current levels to be done
at power factors at or below 0.2, whereas tests at lower currents may be done
at power factors in the range 0.3 to 0.5.
The maximum breaking-capacity tests are usually conducted in short-circuit
testing stations used also for testing switchgear. Such stations have high MV A
outputs which enable high currents to be obtained at the required voltage
levels. At lower-current levels for which the operating times of fuselinks may
be of the order of an hour or more it would be uneconomic to use a major test
plant and two-part test methods, which are allowed in IEC specifications, may
be used. In such cases, a fuselink is supplied initially from a low voltage
source and then when the element is nearing its melting temperature, the test
circuit is switched to obtain the current from a short-circuit alternator which
is excited to provide the required recovery voltage.
Time/current characteristics Specific time/current characteristic curves have

not been standardized or quoted in standard specifications because this could
stifle developments and prevent the introduction of new fuselinks with differ-
ent characteristics. The present trend is therefore to specify a number of points
which form gates through which the actual characteristics of all manufacturers'
fuselinks must pass. As an alternative, some of the IEC recommendations on
low voltage fuses specify zones in which all characteristics must lie. The zones
have been chosen so that all fuselinks of a given rating would operate in a
shorter time at any current than any fuselink of twice the rating. This method
of specification is likely to be discontinued and in future the gating method
will probably be adopted as standard.
Compliance with the gates or zones is checked during the breaking-capacity
tests, described above, by supplying prototype fuse links with the necessary
currents and noting the corresponding operating times.
The determination of overload-withstand capability Some fuse links must be

capable of carrying specified overload currents for particular periods without
their subsequent performance being affected. Such a capability is necessary
for example when they are to be used to protect motors where current-surges
of particular magnitudes and durations will be experienced during starting and
run up.
Various tests are specified in the standards covering different types of
fuselinks. As an example, miniature fuse links must be subjected to a 100
cycles, during each of which a current of 1.2 times rated current must flow
for 1 hour followed by 15 minutes of zero current, and then finally 1.5 times
rated current must flow for one hour. This test must be done on three fuselinks
and at the ends of the tests the voltage drops across the fuselinks must not
have changed by more than 10%.
Manufacture, quality assurance and inspection
The standards set for the construction of prototype fuses in both materials and
manufacturing processes become the standards which must be maintained
when bulk production commences. This is essential to ensure that the produc-
tion fuses will have the performance characteristics indicated by the type tests
and because electrical tests cannot be performed on the bulk output.
Detailed specifications and procedures must be prepared for all stages of
manufacture from the purchase of materials and components to the final
inspection and testing. Many physical checks must be made to ensure that
components are of dimensions within the acceptable limits. In the case of high
voltage fuse links, radiographs of completed units are taken to check that the
elements are not broken, twisted or damaged in any other way and that they
are correctly positioned within their bodies.
As a final check, the resistances of finished fuselinks are usually measured
with a digital instrument with an error of 0.5% or less. After correcting for
ambient temperature the measured values must comply with the original de-
sign limits.
Service checks
In general fuses do not require any regular maintenance but it is desirable that
those in important circuits be inspected physically at regular intervals to check
that the housing and fuse links are in good condition. In addition, fuse link
resistances can be measured and compared with the original values.
Further information on testing is provided in reference [1] and in National
and International Standard Specifications.
A.2.2 Current and voltage transformers

It will be clear from Chapter 2 and the later chapters, which dealt with
protective schemes and relays, that separate current transformers are used to
supply the various equipments in power systems. As an example, a housing
within or adjacent to a circuit-breaker may contain several magnetic cores and
secondary windings associated with a common primary winding. One of
the secondary windings may supply accurate metering equipment, whilst
other secondary windings may separately supply IDMT relays and a current-
differential protective scheme.
To enable appropriate current transformers to be supplied the burdens which
will be imposed on them by the equipment to be connected to them must be
known and in addition the behaviour required of them over the current range
which they will encounter must be known.
In many cases transformers conforming to particular classes designated in
National and International Standards will be found to be suitable; for example,
560 Appendix A
transformers with a ratio of 500/1 and of Class 5P may be chosen to supply
IDMT relays in a particular installation. In these circumstances, the trans-
formers, on completion, are routinely tested to check that they meet the
requirements laid down for their class in the relevant standards. Various limits
are quoted and bridges and other equipment including sources capable of
driving the necessary primary currents, are used.
When the more complex protective schemes, such as those based on current
balancing, are being developed matched sets of current transformers are pro-
duced for use in conjunctive tests which are performed in the laboratories of
manufacturers. In these tests, where the current conditions which could occur
in service are produced, the suitability of the current transformers is deter-
mined and subsequent production transformers are checked to ensure that they
have very similar characteristics. Further information on conjunctive testing
is provided later in section A.2.4.
Whereas separate current transformers are used to supply different items of
measuring and protective equipment, individual voltage transformers supply
several burdens connected in parallel and clearly they must have the necessary
volt-ampere outputs. In addition they must provide output voltages within the
accuracy ranges required. Each transformer is tested on completion to verify
that it complies with the limits specified in the appropriate Standard Specifica-
tion using equipment such as bridges.
In many applications of both voltage and current transformers it is essential
that the polarities of their outputs should be correct. As examples, current
transformers feeding current-differential protective schemes must be so con-
nected that operation will not occur during healthy or external fault conditions
and current and voltage transformers must be so connected that directional
relays supplied by them will operate and restrain correctly. It is therefore
necessary that site tests be conducted during commissioning to ensure that the
interconnections between transformers and their burdens are correct.
Detailed information on the testing of current and voltage transformers is
provided in references [2-4].
A.2.3 The testing of relays

Very large numbers of IDMT and other relays are produced annually for use
in power systems. Many of them have adjustable settings to enable them to be
applied to any network. As an example, IDMT relays are provided with ranges
of current and time settings.
National and International Standard Specifications have been produced in
which the operating characteristics of many relays are defined and acceptable
performance limits are stated.
Both prototype and production relays are tested by their manufacturers to
confirm that their performances comply with the requirements. Because the
current and voltage inputs to relays are usually relatively low, typically no
more that 100 V and 100 A, there is little difficulty in providing adequate
testing facilities.
It is clearly desirable that relays be checked both when they are put into
service and at regular intervals thereafter. In the case of electro-mechanical
relays, it is necessary that their performances be checked electrically by en-
ergizing them from suitable test sets and in addition they should be inspected
physically. The electrical performances of electronic relays should also be
checked by energizing them with the appropriate signals obtained from test
sets.
Further information on this topic can be obtained from reference [5].
A.2.4 The testing of protective schemes

Protective schemes are usually quite complex, and because of their important
roles in power systems their performances must be thoroughly examined
during their development and subsequent production. Testing is then necessary
after schemes are installed and at regular intervals thereafter.
It will be clear that the procedures which must be adopted are dependent on
the operating principles and methods of implementation of individual schemes
and these cannot be covered in detail in this book. The basic methods and
testing equipments which are used are, however, described in the following
sections.
Development testing
Because protective schemes are required to operate when the conditions on
power systems are abnormal it is essential that their performances be deter-
mined during such conditions. Manufacturers have therefore to set up testing
facilities which will enable schemes to be subjected to the full range of
conditions which they may encounter in service. Over the years, facilities have
been developed and extended to meet the changing needs.
The testing of current-differential and phase comparison schemes As stated

in Chapter 4, current-differential schemes, as proposed by C. H. Merz and B.
Price, were produced and installed on networks in the early years of this
century and such schemes have been continually improved and are still being
installed. Their performance, as explained earlier, is dependent on the relaying
arrangements, the communication links between the ends of the protected units
and the current transformers. It therefore became the practice to perform
development tests, called conjunctive tests, on prototype schemes fed from
current transformers the same as those which were subsequently to be used in
service.
Because fault settings are relatively low they can be checked readily by
supplying the appropriate currents to the current transformers. To check that
562 Appendix A
stability will be maintained under external fault conditions, it is required that
the maximum currents which may flow in service be supplied to the current
transformers and thai these should contain the transient components with the
highest time constants which may be encountered. This latter factor is of great
importance as it affects the core fluxes present in current transformers, as
explained in Chapter 2, and determines the possible degrees of saturation
which may occur.
Initially, manufacturers installed motor-driven-three-phase alternator sets of
relatively low volt-ampere outputs which supplied step-down transformers
which could supply high currents at low voltages to the test circuits. Impedors
with high X/R ratios were available for connection in series with the primary
windings of the current transformers and in this way the possible service
conditions could be produced. A circuit-breaker was included in the test circuit
to enable tests to be initiated and terminated. At that time it was not possible
to control the point on the voltage wave at which a circuit-breaker would close
and therefore it was the practice to do many stability tests so that the most
onerous conditions were likely to have been produced.
These testing arrangements are still used today but features such as the
control of the point on the wave at which the circuit closes are now standard
and modem recording facilities and instruments are incorporated. A modern
medium-current test plant [5] is shown in single line form in Fig. A.2. It can
be seen that it incorporates a 500 kV A supply transformer with 11 kV primary
windings and secondary windings which can provide line voltages of 880,660
and 440 V. The inductive reactors can be set to give wide ranges of X/R ratios
and current magnitudes, the maximum current being 510 A per phase. This
plant feeds relays from built-in current transformers with ratios of 10/1 A and
Test current transformers

Group B
-x
Supply transformer
500kVA
11,OOO/880-660-440V
Fig. A.2 Schematic diagram of a medium-current test plant. (Reproduced from Pro-
tective Relays - Application Guide, 3rd edn, GEC Measurements, 1987 with the
permission of GEC Alsthom Protection and Control Ltd.)
Supply transformer
500kVA
11 ,OOO/880-660-440V
Thyristor
-x
II~ c
II _ _ _ _ _ _ Pomt-on-wave
.
control unit
-------_ ...
Fig. A.3 Circuit to achieve a high X/R ratio in current transfonner primary current.
(Reproduced from Protective Relays - Application Guide, 3rd edn, GEC Measure-
ments, 1987 with the pennission of GEC Alsthom Protection and Control Ltd.)
20/1 A and the effective knee-point voltages of the transformers can be varied
between 160 V and 2000 V by connecting one or more of them in series.
The nominal X/R ratio of the reactors is 30, but this value is reduced in
practice by the resistances of the connections, the windings of the current
transformers and the relays. It will be appreciated that current transformers
with the cores and secondary windings to be used in service, but with the
appropriate multi-turn primary windings, could be fed by test plants such as
that described above to enable completely conjunctive tests to be performed.
The increase in power-system fault levels and X/R ratios in recent years has
created conditions which cannot be simulated economically with these types
of equipment. Because of this Cavero et al [6] commenced investigations at
the University of Nottingham into alternative methods of generating the re-
quired test currents. This led to development of test plants with the basic
arrangement shown in single-phase form in Fig. A.3. In this plant, the transient
component of the test current is produced by discharging the capacitor C
through the resistor R2 and the primary winding of the current transformer
which supplies the protective scheme or relay. The desired transient current
is obtained by previously charging the capacitor to the appropriate voltage and
by setting the resistance (R 2 ) to the necessary value. The steady state compo-
nent of the current is supplied from a transformer, the desired magnitude being
obtained by choosing the appropriate voltage for the secondary winding of the
transformer and adjusting the resistance of resistor RI . This current also flows
through the primary winding of the current transformer. Current flow is in-
itiated at the appropriate instant in the cycle of the voltage provided by the
secondary winding of the transformer by firing the triac and thyristor.
564 Appendix A
The technique has been implemented on a three-phase basis and such plants
can be used in the development of protective schemes, and again the desired
currents can be fed to either current transformers similar to those to be used
in service or to built-in current transformers interconnected as described ear-
lier to achieve desired knee-point output voltages.
The most recent development has been the introduction of programmable
power-system simulators which can be widely used to test all types of protect-
ive equipment. These are described in some detail in the next section.
The testing a/impedance-measuring schemes Unlike the schemes considered

in the previous section, impedance-measuring, or distance protection, schemes
are supplied with both voltages and currents and in service these are obtained
from voltage and current transformers connected near the ends of the lines
being protected.
In the past, the power-system currents and voltages which would be present
at the end of a given transmission line or cable during faults at different
positions on it or beyond it could have been calculated for the range of possible
operating conditions. It would, however, have been difficult to then feed the
appropriate quantities to the voltage and current transformers connected to a
protective scheme during conjunctive testing. Because of this, testing equip-
ments containing so-called 'artificial lines' were produced. Such an equipment
is fed from a three-phase voltage source and it contains variable impedors
which can be set to represent the effective source impedance at the busbars at
the end of the line to which the protective equipment was considered to be
connected and also to represent the line and fault impedances and that of the
load at the remote end of the line. Prior to a test, the impedances have to be
set so that they represent particular network conditions. These have to take
account of the source MV A, and voltage, the ratio of source and line imped-
ances (Zs/Zd and other factors such as the X/R ratios of sections of the
network.
These equipments are supplied from three-phase sources at voltage levels
of 415 V or 660 V (line) at a frequency of 50 Hz (in Britain) and it is usual to
be able to obtain an alternative supply from a variable-frequency machine.
The maximum currents available are of the order of 120 A (r.m.s.) and these
are fed to current transformers with multi-tum primary windings with cores
and secondary windings similar to those which are to be used in service with
the protective scheme being tested.
All types of faults can be simulated and they can be applied in different
positions; for example, on the protected line or beyond it. Point-on-wave
control is invariably provided so that faults may be applied at any desired
instant in the voltage cycle and adjustable resistors are included so that the
presence of fault-arc resistance can be simulated.
As stated in section 11.14 (page 431), programmable power-system simula-
tors have been developed to test distance-measuring schemes including those
intended for application to series-compensated transmission lines. In addition,
they can be used to test other protective schemes including those in which
current comparisons are effected or those in which discrimination is achieved
by examining travelling waves.
Simulators may be produced in several ways but basically they contain a
computer programmed to calculate the voltages and currents which will be
present under a wide range of healthy and fault conditions on a network to
which a particular protective scheme is to be applied. Several computer codes
are available for this task, a particular example being the Electromagnetic
Transient Program (EMTP) [7]. The digital data are converted into analogue
quantities which are applied to analogue models of the voltage and current
transducers and their outputs are in tum applied to the actual protective
scheme. Its behaviour is monitored and its suitability is thereby determined.
Research institutions [8], and manufacturers have constructed such test
equipments. As an example, a block diagram of a simulator used by GEe
Measurements is shown in Fig. A.4.
The testing of travelling-wave schemes These schemes are implemented in a

similar way to modem distance-type protective schemes and because of their
complexity their behaviour should be determined using programmable power-
system simulators.
In addition to the tests described above, which determine the electrical
performances of relays, additional tests are necessary to ensure that relays are
able to withstand a range of environmental conditions. Protective relays must
be capable of withstanding vibrations [9], impact and extremes of temperature
and humidity. Another class of important tests, particularly for digital relays,
is those related to electromagnetic compatibility (EMC). These include high-
frequency bursts, impulses, fast transients, static discharges and radiated
interference tests [10].
The production testing of schemes

Tests on protective schemes before they are dispatched to site may be con-
ducted using equipments such as those described in the preceding sections but
only a limited number of checks should be necessary at this stage. Alternat-
ively, tests conducted using simpler equipment may be adequate to ascertain
that the schemes are operating correctly.
Commissioning and subsequent routine testing

Tests are performed before schemes are commissioned using secondary injec-
tion and other equipment and at this stage checks should be made to ensure
that the correct settings have been selected and that voltage and current
transducers have been connected correctly.
5-~~
~ 0 (JQ
9
'g "'tl >
3F;;" ~ ~
~ t:C
~. ~ 0' IA
g ~. ~ r-'""-
g,~e:
otIl ~-~ '"1
EJI
Keyboard
d
Tape
unit 18
~
Digital to
analogue t- Linear
interpolation t-
CT
model
~-
(")"'3 [ J circuits_
~~o -
~"""
::r"l::s ~
~
1 f. l
t-V - Equipment
t-~ under
~ ['Sl -- IA test
4' ::l'.~ --.,
~ g ~ I GDU I I/O Linear
g C) 9 Digital to CT
r- Mini-computer r-- sub- r- interpolation r- model
g. E. ~ analogue
~
~ -
::I~g: system circuits t-~
~. ~ 1
::I ....
0. ~ Keyboard
(") ~
o
'8
!:>. ~
~
I J ~
Ie Contact
- J
::I ;:s '"1 r status
::: .. ~
o 0,< monitor r-
-tIl'"
r(")S
p. ,.. 9
1 r
Disc Key to abbreviations: Signalling
'-'"'~" '" ,-
unit CT = Current transformer channel
~ g-. I CVT = Capacitor voltage tr~nsformer simulation
=
~ -
~ r,nll = r,r",nhi"" tii"nl",v omot
9 0
EJ1
~ :-<
ia,-... [ Keyboard
J" :;c J Communications .. (When required) .. To second PPSS
~~
000 link to second
-.10.
_. (>
PPSS
~ g
9-0.
Testing relays and protective schemes 567
Routine checks should be perfonned at regular intervals after schemes have
been commissioned to ensure that they are capable of operating correctly, the
nature of the checks being dependent on the way in which the schemes are
implemented. An indication of the on-site testing procedures adopted by a
major utility may be found in reference [11].
A.3 AUTOMATIC TESTING OF RELAYS AND PROTECTIVE

SCHEMES
As indicated above, it is necessary that relays and protective schemes be tested
at regular intervals after they are installed and commissioned. In many cases
this is done by perfonning individual tests using secondary-injection equip-
ment and by making physical checks. This was certainly the practice in the
past. Equipments are now available, however, to automate the process and
they may be used with electro-mechanical relays, such as IDMT elements or
modern static relays and schemes.
These equipments provide testing voltage and current wavefonns repres-
enting power-system quantities which are mathematically synthesized. Digi-
tal-to-analogue converters and power amplifiers are incorporated to produce
the levels of voltage and current required to cause the relays under test to
operate. The relay outputs are monitored and quantities such as operating
times are recorded.
The use of such equipments enables tests to be conducted quickly even on
quite complex schemes, such as distance protection. Further infonnation on
such equipment is provided in reference [12] .
. It was the practice in the past for manufacturers to provide some built-in
test features in relatively complex schemes. As an example, the early phase-
comparison protective schemes applied to transmission lines incorporated
switches which enabled high-frequency signals to be sent between the line
ends to check that correct operation would be obtained during both internal-
and external fault conditions. These switches could be operated manually or
automatically at regular intervals, say every 6 hours.
This practice has continued and many modern systems incorporate facilities
for regular testing. Comprehensive self-test facilities can be easily incorpor-
ated in digital relays where hardware and software are designed to allow
continuous testing of the functioning of various parts of the relay, including
the accuracy of computation and memory integrity. Self-checking is per-
formed as a background task, without interfering with the operation of the
relay, and it is claimed that identification of malfunctions is possible in most
cases. When malfunctioning is detected an alarm is raised to alert the operator
to the problem.
568 Appendix A
REFERENCES
1. Wright, A. and Newbery, P. G. (1982) Electric Fuses, Peter Peregrinus Ltd.
2. BS 3938 : 1973 British Standard Specification for Current Transformers.
3. ANSI/lEE C57.13.1-1981, lEE Guide for Field Testing of Relaying Current
Transformers.
4. BS 3941 : 1975 Specification for Voltage Transformers.
5. GEC Measurements, Protective Relays Application Guide (3rd edn), 1987, Chap-
ter 23.
6. Paull, C. J., Wright, A. and Cavero, L. P. (1976) Programmable testing equipment
for power-system protective equipment, Proc. lEE, 123, 343-349.
7. Electromagnetic Transients Program-Reference Manual (EMTP Theory book),
Prepared by H. W. Domme1 for the Bonneville Power Administration, August
1986.
8. Redfern, M. A., Aggarwal, R. K. and Husseini, A. H. (1990) A laboratory power
system simulator for evaluating the performance of modern protective relays, 25th
Universities Power Eng Conf, Aberdeen 12-14 Sept 1990, pp 167-170.
9. BS 142: 1989, Electrical Protection Relays, Section 1.5, Vibration, shock, bump
and seismic testing.
10. IEC 255-22: Electrical disturbance tests for measuring relays, Parts 1 to 4.
11. Laycock, W. J. (1989) On site testing of protection and automatic switching
equipment in 275 and 400 kV substations, Proc. 4th Int Conf on Development in
Power System Protection, Edinburgh 11-13 April 1989, lEE Conf. Publ. No 302,
pp 27-29.
12. Webb, A. C. and Webb, M. (1988) Automated testing of power system protection
relays, Power Eng J., November, pp 291-296.
Appendix B
Percentage and per-unit quantities
Whilst the perfonnances of power-system networks can be determined using

the actual ohmic impedances of the components within the networks, the
various transfonnations caused by the presence of power transformers have to
be taken into account. These transfonnations could be eliminated in a model
network used to represent a particular power system if one voltage level was
chosen as the reference and all the impedances were converted to the equival-
ent values associated with this voltage level. A similar effect is obtained
somewhat more simply by using percentage or per-unit values in calculations
associated with power systems.
A given percentage value for an impedor means that the voltage drop across
it when it is carrying the rated current of the section of the circuit in which it
is connected is the percentage of the rated voltage of that section of the circuit.
As an example, if a resistor of 20 Q is connected in a circuit for which the
rated voltage and current are 1000 V and 10 A respectively, then the resistor
has a percentage value of 20%, i.e. it drops 200 V when carrying 10 A in a
circuit rated at 1000 V. It may alternatively be given a per-unit value of 0.2,
Le. 200/1000.
Voltages and currents are also assigned percentage or per-unit values based
on the rated voltages or currents of circuits. As an example, if the voltage
between two points in a circuit rated at 100 V is 50 V, it is referred to as 50%
or 0.5 per unit. Similarly, if a current of 20 A flows in a circuit rated at 50 A
then the current is either 40% or 0.4 per unit. The tenn per-unit is usually
abbreviated to pu, e.g. I = 0.4 pu.
It is well known that impedances can be referred across transfonners, Le.
an impedor in the secondary circuit of a transfonner may be represented by
an equivalent impedor in the primary circuit. This is illustrated in Fig. B.l (a)
in which an ideal tran~former with Np and Ns primary and secondary winding
turns has an impedor Zs in its secondary circuit which carries a current Is. The
impedor has a per-unit value of lis Zs IIV.. VS being the rated voltage of the
secondary circuit. In the equivalent circuit, shown in Fig. B.l (b), the impedor
Zs is replaced by an impedor Zs' in the primary circuit. The current in this
ci!.cuit _is Ip = -Is Ns/Np and the drop across the impedor Zs' is Ip Z; =
-Is Ns Z; /Np. The rated voltage of the primary circuit is Vp = - Vs Np/Ns and
570 Appendix B
Ideal
s
Np
f
II
(a)
Fig. B.l Equivalent circuit and parameters of a transformer.
therefore the per-unit value of the equivalent impedor (Z.') is given by:
I/pzs'l = lIs I 1-'1 [Ns)2

Vp Vs Zs Np pu
This has the same per-unit value as the actual secondary impedor,
IIs Zs I/Vs when
i.e. the value used when impedance is transferred across a transformer.

It will also be appreciated that if a particular per-unit voltage is applied to
the primary winding of an ideal transformer then the secondary output voltage
has the same per-unit value. As an example, if the primary voltage has a value
of 0.5 pu then the secondary voltage has a value of 0.5 pu. Similar relation-
ships exist for the currents.
Figure B.2(a) shows a simple circuit containing a transformer and Fig.
B.2(b) shows the equivalent circuit in which per-unit values are used.
The technique is also applied to three-phase circuits where further simplific-
ations can be made. For example, if the phase voltages of a circuit are each
of a particular per-unit value then the line voltages also have the same per-unit
values.
Actual voltages, currents and impedances associated with circuits during
steady state conditions are usually expressed in complex or polar form, for
Appendix B 571
example Jp could be (50 - j 50)A or 50'l'2I- n /4 A and Zs could be
20(1 +/-/3) 0 or 40 In/3 O. If these quantities were associated with a circuit
in which one per-unit current and impedance were 100 A and 2000 respec-
tively then they would be expressed as:
Ip =(0.5 - j 0.5) pu and Zs=O.1 (1 + j.g) pu
An important factor which must be taken into account when percentage or
per-unit quantities are to be used is the volt-ampere ratings of the component
parts of the circuit to be studied. This will be evident from the following
example, which is based on a star-connected three-phase circuit in which there
is an impedor Z in one of the phases.
The full load or rated current (lr) of each phase of the circuit is given by:
l-~
r - --j3vL
in which VA is the rated volt-amperes of the three-phase circuit and VL is the

rated line (phase-to-phase) voltage.
The voltage drop across the impedor Z when carrying the rated current of
the circuit would be:
z" = 0.2 1m3 Q

s
II ~s
f
(a)
-
Zp = 0.05 L1tI3 pu ~p=0.1 1.zr!3pu ~=0.1L1tI3pu Zs= 0.05L1tI3 pu
Ip (pu)
1 Vp
(pu)
(b)
Fig. D.2 Transformer (a) and its equivalent circuit in per-unit quantities (b).
Primary rated voltage and current: 500 V, 20 A. Primary winding impedance:
Zip = 2.511t/3 n. Exciting current at 500 V: 0.21-1t/3 A. Secondary rated voltage
and current: 200 V, 50 A. Secondary winding impedance: Zts = OAI1t/3 n.
572 Appendix B
The per-unit impedance of Z would therefore be given by:
Zpu=IpZ= VAZ =VA. Z (B.l)

Vph .~ VL vl
. . . 3 VL .,f3
in which Vph is the rated phase voltage of the circuit.

It will be clear that the per-unit value of an impedor Z in a single-phase
circuit is given by:
(B.2)
in which V is the rated voltage of the circuit.

When determining the behaviour of circuits which contain sections or items
with different VA ratings it is necessary to use a common or reference volt-
ampere base throughout. This procedure is used in the following example
which should also illustrate the other points referred to above.
Example For simplicity, the single-phase circuit shown in Fig. B.3(a) is

examined. A transformer with a step-down ratio of 2 : 1 is included ~d this
~eds three load circuits with different current ratings. The impedors 4p and
Zts represent the impedances of the primary and secondary windings of the
II
(a)
ZL1 =O.48L1tIS pu
ZL2 =O.72L1tIS pu
~s = 0 .24L 1tIS pu
ZL3 = O.9SL 1tIS pu
--------"-'-----1'03
Za= 20L1tI4 pu
(b)
Fig. B.3 Example of a network (a) and its per-unit equivalent (b).
Appendix B 573
transformer and Ze is its ex~ting impedance. The source, which provides an
e.m.f. Es, has an impedance Zs anj the connections b~tween the source and the
transformer have an impedance Zc. The impedance Zp shown in Fig. B.3(a) is
therefore given by:
- --
Zp = (Zs + Zc + 4p)O
The outgoing circuits have impedances of ZLl, ZL2 and ZLJ as shown.
The rated voltages of the primary and secondary circuits are 1000 V and
500 V respectively and the rated currents are as follows:
Rated current of load circuit 1 (ILl) = 50 A
Rated current of load circuit 2 (lu) = 40 A
Rated current of load circuit 3 (lLJ) = 30 A
Rated current of primary circuit = (50 + 40 + 30)/2 = 60 A.
The impedances in the circuit are:
- -
ZLl=2In/6 O;ZL2=3In/6 0
- -
ZLJ=4In/6 O;Zp=5In/3 0
- -
Ze = 1000/3In/4 0; 4s = Iln/6 0
The rated VA associated with the primary circuit is 1000 x 60 = 60 kV A and

this is used as the reference base below.
Voltage drop across Zp at 1000 V 60 k VA = 60 x 51n/3
- 300
:. Zp = 10001n/3 pu = 0.3In/3 pu
Alternatively
zp= VA
V2
Z = 60 x 1000. 51n/3 =0.3 In/3 u
1000 2 P
For lo~d circuit 1 the rated current = 50 A and at this current the voltage
drop on ZLl = 100 In/6 V.
- 100
:. ZLI = 500 In/6 pu at 500 x 50 = 25 kV A
With reference to the 60 kVA base,

- 60
ZLl = 25 0.2In/6 = 0.48In/6 pu
Alternatively,
- VA -
TVP"
ZLl =
60x 103
ZLl = 5002 x 21n/6 = 0.48In/6 pu
574 Appendix B
Similarly,
- 60x 103
ZL2 = 5002 X 31 n/6 =0.72 In/6 pu
- 60x 103
ZL3 = 5002 X 41n/6 = 0.96 In/6 pu
-Z = 60 X 103 1000 I ,"/1114 = 20 I ,"/1114

e 10002 X 3 ~ ~ pu
- 60x 103
and Zts = 5002 x Iln/6 =0.24 In/6 pu
The equivalent circuit, using the 60 kVA base, is shown in Fig. B.3(b).
This circuit could be used more simply than the one shown in Fig. B.3(a)
to determine the behaviour over widely ranging conditions.
As an example, should load terminals 2 and 3 be open-circuited and the load
terminal 1 be short circuited to earth when the source e.m.f. (Es) was 1000 V,
i.e. 1 pu, and the small effect of the exciting impedance (Ze) was neglected,
then the short-circuit current (lsc) would be given by:
- 1 1
~= = ~
Zp + Zts + ZLl 0.3In/3 + 0.72ln/6
= 1.0181 - 0.676 pu
This is based on a 60 kVA base, i.e. 60 A in the primary circuit and 120 A
on the secondary side of the transformer.
The actual current in the primary circuit would therefore be 61.1 A and the
current in the short circuit would be 122.2 A.
Appendix C
Transformations of three-phase quantities
The steady-state voltage/current relationships in three-phase circuits with

three phase conductors and a neutral conductor are of the general form:
(C.I)
The current in the neutral conductor (/n) is given by:

- - - -
In=-(/a+/b+/c)
There are basically three degrees of freedom and each phase voltage is
dependent on the three phase currents. It will be clear that both the amount of
computation involved in determining the behaviour of power systems and the
processing in protective schemes can be reduced if equation (C.I) is trans-
formed so that the elements in the Z matrix, other than those in the principal
diagonal, all have a value of zero. In these circumstances, three independent
equations are provided by the rows of the matrix equation.
It will be evident that the voltage, current and impedance matrices in equa-
tion (C.l) must all be so transformed that the two sides of the resulting
equation are equal. The basic technique when dealing with steady-state r.m.s.
quantities is outlined in the following section.
C.I VOLTAGE, CURRENT AND IMPEDANCE TRANSFORMATION

Using a matrix [C], the relationship between the original and transformed
current matrices may be expressed as:
[I] = [C][i'] (C.2)
in which [I] and [I'] are the original and transformed matrices respectively.
To ensure that the volt-amperes are not changed the original and trans-
formed ~~tities must be ~e same and because voltage-amperes are givenJ>y
[VA] = [Vt1[/*], in which [VtJ is the transpose of the voltage matrix and [/*]
contains elements which are the complex conjugates of the elements in the
576 Appendix C
current matrix, the following relationship must be maintained:
(C.3)
in which [V{] is the transpose of the transformed voltage matrix.
The relationship in equation (C.2) may be expressed as:
a*] = [Co][/'o]
and thus equation (C.3) may be expressed as:
[Vt']a'*] = [Vt1[C*][/'*]
and therefore:
This equation may be rearranged to give:
(C.4)
Equation (C.I) may be expressed as:

[V] = [Z][I]
and therefore equation (C.4) may be written in the form:

[V'] = [C;][Z][/] = [C;][Z][C][/'J
= [Z'][I']
It is clear that the transformed impedance matrix [Z'] is given by:
[Z'] = [C;][Z][C] (C.S)
When the matrix [C] is square and has an inverse [C] - 1 then the following
equations may all be used.
[I] = [C][I'] a'] = [cr 1[/]

[V] = [C;] -I [V'] [V'] = [C;][V]
[Z] = [C;] -I[Z'][C]-I [Z'] = [C;][Z][C]
When the matrix [C] is also orthogonal then
[C][C;] = [C;][C] = I
i.e. and
Using these relationships, the following equations are obtained:
[I] = [C][I']
Transformation into symmetrical components 577
(V] = [C][V'] [V'] = [C;][V]
[Z] = [C][Z'][C;] [Z'] =[C;][Z][C]

It should be noted that [C~] can usually be determined more rapidly than
[Cr l
C.2 TRANSFORMATION INTO SYMMETRICAL COMPONENTS

Symmetrical components are a special form of transformation in which steady
state three-phase voltage and current phasors are replaced by three sets of
phasors designated as being of positive-, negative- ant! z~ro-se9.uence.
As an example, an unbalanced set of current phasors (la, Ib and Ie) is replaced
by the nine current phasors shown in Fig. C.l, the relationships being that:
- -
la = Iia + 12a + loa
- - - -
h = lib + 12b + lOb
- - - -
Ie = lie + 12e + lOe (C.7)
Again there are only three degrees of freedom and therefore the relationship
can be expressed in the form:
[~:]=[~ ~ lj [f:] (C.8)
in which the operator a = - 0.5 + j ...[3/2 and a 2 = - 0.5 - j...[3/2 or a = II 21t/3

Tad and a2 = 11 41t/3 rad.
The transformation matrix [C] is thus:
(C.9)
n
The matrix [C;] is therefore:
IC:l=[: ~
It must be recognized, as stated above, that when sequence quantities are
used, the three phase voltages are replaced by nine sequence voltage and
currents.
The volt-amperes associated with the actual phase voltages and currents are
given by:
578 Appendix C
and the voltage-amperes associated with the sequence quantities are given by:
VA= 3~'\.I7a + 3V2aI~ + 3VoaI~ = 3 [V(][I'*]
With this transformation therefore:
and as a result:
3 [Vt'] =[Vtl[C*]
or [V'] = 1/3 [C;UV] (C.IO)
The full transposed voltage matrix [V'] is therefore:
(C.lt)
In a circuit which is physically symmetrical, the elem~nts ~n t!!e principal

diagonal of the impedancE matrix, i.e. the impedances Zaa, Zbl Zee are each
equal to self impedance (Zs) and the o~her elements, which are all the same,
are equal to the mutual impedance (Zm). The actual impedance matrix is
therefore given by:
~
[2YJ3~'
'2C
Fig. C.I Transformation into symmetrical components

Ilia I = I/lb I = I/le I
1/2a I = 1/2b I = 1/2c I
Iloa I = IlOb I = 1/0e I
Other transformations 579
The sequence-impedance matrix is:
[2'] = 1/3 [C;][2][C]
[1
n~ n[~
Zm 1
a
tl
=1/3 1
1 1
Zs
Zm tl
a
1]
i.e
[Z.-~
Z.+~2~]
0
= 0 Zs-Zm (C.I2)
0 0
Hence ZI = ~ = Zs - Zm and Zo = Zs + 2Zm.

It will be clear that there are no mutual impedances in the transfonned
matrix and the sequences are thus independent of each other when circuits are
physically symmetrical, i.e.
and VOa =ZoIOa
In practice, three-phase transfonners and ~ver!!ea<!. transmission and dis-

tribution lines are physically asymme.!.ric, i.e. ~aa, Zbb, Zcc are not all equal nor
are the mutual impedances such as Zab and Zac. In these circumstances, the
transfonned sequence impedance matrix does not only have non-zero elements
on the principal diagonal and the sequences are not therefore independent of
each other.
Methods of detennining the behaviour of three-phase power systems during
unbalanced conditions, using symmetrical component techniques, are de-
scribed later in Appendix D.
C.3 OTHER TRANSFORMATIONS

In many situations, such as the treatment of transient signals, the sequence
transfonnations described in the previous section cannot be used. It is however
useful in such cases to be able to decouple phases and the approach outlined
below may be used. _
The phase voltages [V(P)] and currents [/(P)] are related by the following
expression:
.E:.. [I (p)] = - [y(P)][V(p)]

dx
where [Z(P)] and [Y(P)] are the system impedance and admittance matrices
580 Appendix C
respectively. In general they contain components which are frequency depend-
ent and have real and reactive parts.
These equations may be combined to obtain:
where
T...ransformation matrices [S]and [Q] are sought so that component voltages

[V(el] and currents [/(e l ] can be obtained.
[V(Pl] = [SHV(e l ]
[I(Pl] = (QHI(e l ]
Expressing the phase quantities in terms of the component quantities gives:
The three component quantities are decoupled provided the matrix [1 2 ] is

diagonal.
Matrices [S] and (Q] are selected to diagonalize [1 2 ].
For lines in which the conductors are symmetrically positioned, diagonaliz-
ation is straight forward and the transformation matrices are:
[S] = [Q] =[ 11 1
0 - 2
1]
1 -1 1
This is known as Clarke's or (a.,~, 0) transformation.

In practice the conductors of lines are not arranged symmetrically. Diago-
nalization is again possible but the transformation matrix varies with fre-
quency. As an example, for a 400 kV, 50 Hz line in which the conductors a.!.e
in a vertical configuration and for a ground resistivity of 100 Urn the [S]
matrix is
0.66 -0.553+jO.l -0.27- j O.08j

[ 0.58- j.0.006 0.23- jO.ll 0.82~
0.48-} 0.02 0.787 - 0.47+ } 0.13
In transient calculations two approximations are necessary. Firstly, an [S]

matrix at a particular frequency is selecte~ typically at 1 kHz. Secondly, the
imaginary part of the components of the [S] matrix is neglected. For the 400
kV line considered above, the approximate matrix at 1 kHz is:
References 581
0.68 -0.56 -0.2J

[S] == [ 0.57 0.36 0.79
0.46 0.75 -0.59
The component or modal voltages and currents are then related by the expres-
sion
where
_
[Z(c) =[542.4
- 3.2
1.53
293.2 -0.7 0.1~
-0.2 0.1 246.6
It will be seen that because an approximate S matrix was used the off-
diagonal elements are small but not zero and hence complete decoupling has
not been achieved. In practice the contribution from the off-diagonal elements
can be neglected.
Further details of transformation matrices may be found in references [1-3].
REFERENCES
l. Wilson Long, R. and Gelopulos, D. (1982) Component transformations - eigen-
value analysis succinctly defines their relationships, Trans, lEE PAS-101, 4055-
4060.
2. Wasley, R. G. and Selvavinayagamoorthy, S. (1974) Approximate frequency re-
sponse values for transmission-line transient analysis, Proc. lEE, 121, 281-286.
3. Wedepohl, L. M. and Mohqwed S. E. T. (1969) Mu1ticonductor transmission lines
- theory of natural modes and Fourier integral applied to transient analysis Proc.
lEE 116, (C), 1553-1563.
Appendix D
The determination of power-system
behaviour using symmetrical components
As indicated in Section C.2 of Appendix C, the sequences are only inde-

pendent of each other when circuits are physically symmetrical. Although
items of plant such as alternators are completely symmetrical, this is not
the case for overhead lines or three-phase transformers but, nevertheless, the
effects of asymmetry are usually neglected to simplify the calculations of the
behaviour which will occur in power systems during both normal and abnor-
mal conditions. This practice is acceptable because it is not normally neces-
sary to determine fault-current levels with great accuracy.
In all networks, current flows are affected by the various impedances which
are present and therefore the impedances presented to each of the current
sequences must be known. In the following section the sequence impedances
of items of equipment used in power systems are considered.
D.l SEQUENCE IMPEDANCES

Because of the differences in the constructions and modes of operation of
rotating machines and other static equipment, their effective impedances are
examined separately below.
D.I.I Three-phase synchronous machines

These machines are always physically symmetrical in that they contain three
a.c. windings, each of the same number of turns, distributed similarly around
their peripheries. A set of steady-state positive-sequence currents in these
windings sets up a magnetic field of constant magnitude which rotates relative
to the windings at synchronous speed (ns) in the same direction as the main
magnetic field set up by the field winding. An extra positive-sequence set of
e.mJ.s is thereby induced in the windings, their magnitudes being proportional
to the winding currents. This effect is normally allowed for by assigning a
reactance to each winding, i.e. a positive-sequence reactance, which together
with the winding leakage reactance is termed the positive-sequence synchron-
584 Appendix D
Rotor
Stator 'a' phase winding Motion of field
\ Motion of field produced by
produced by stator windings
stator windings
'c' phase
winding
Field winding
(a) (b)
Fig. D.I Field set up by stator currents rotates at n, as shown for positive (a) and
negative (b) sequence currents.
ous reactance (Xd. The three windings each present the same resistance (R 1 )
to the currents and as a re~ult a machine can be assigned a positive-sequence
synchronous impedance (ZI). In practice this impedance is made relatively
high to limit the currents which may flow in the events of faults, but, of course,
the resistance is kept to a minimum to ensure that high efficiencies are ob-
tained. As a result XI R I
It will be appreciated that the magnetic field set up by positive-sequence
currents does not cut the field circuit of a machine. This is illustrated in Fig.
D.l(a).
A set of steady-state negative-sequence currents in the windings of a ma-
chine sets up a magnetic field which rotates relative to the windings at syn-
chronous speed (ns) in the opposite direction to that produced by the
positive-sequence currents. This field, shown in Fig. D.l(b), cuts the three-
phase windings and induces e.m.f.s in them but it is of reverse or negative
sequence. This field rotates at twice synchronous speed relative to the field
windings of the machine and the core material associated with them, causing
hysteresis_effects and eddy-currents. As a result, the negative~equence im-
pedance (Z2) is not equal to the positive-sequence impedance (ZI).
A set of zero-sequence currents in the windings of a machine sets up three
time-varying magnetic fields which are stationary with respect to the wind-
ings. They induce equal e.m.f.s of the same phase in each of the windings, i.e.
a set of zero-sequence e.m.f.s with a magnitude dependent on the zero-
sequence current. These zero-sequence fields move at synchronous speed
relative to the field windings and the core material causing powerJosses within
them. A machine has therefore a zero-sequence impedance (Zo) , which is
different from the other two sequence impedances.
Sequence impedances 585
For a large turbo-alternator, typical values for the steady-state positive-
sequence reactance (synchronous reactance) range between 2 and 2.5 pu.
However, immediately after the incidence of a fault, the machine reactance is
much lower, typically 0.1-0.2 pu (subtransient) reactance. The subtransient
phase lasts for approximately one or two cycles and it is followed by the
transient phase when typical reactance values range from 0.15-0.28 pu. Steady
state is reached after typically four or five cycles. Typical values for negative-
and zero-sequence reactances range from 0.13-0.25 pu and 0.05-0.13 pu
respectively.
D.1.2 Transformers
When positive-sequence currents flow in the primary and secondary windings
of a transformer, magnetic fields are set up in leakage paths and positive-
sequence e.mJ.s are induced in both sets of windings. This effect may be
allowed for by assigning a positive-sequence leakage reactance to the primary
and secondary windings (X lp and XIS)' The currents flow through the resist-
ances of the _windins..s and therefore the windings possess positive-sequence
impedances Zip and ZIS'
Similar effects are produced by negative-sequence currents and the imped-
ances which may be assigned have .,!he s.!IDe val~es a! those associated with
the positive-sequence currents, i.e. ~p =Zip and ~ =ZIS'
The passage of zero-sequence currents also causes magnetic fields to be set
up in leakage paths, but because these fluxes are in phase with each other they
may not follow the same paths as those set up bI the ot~r sequence currents
and as a result the zero-sequence impedances, Zop and Zos may not have the
same values as the other sequence impedances.
Typical per-unit positive- and negative-sequence impedances for trans-
formers rated at 10 MVA and above range between 0.05 and 0.2, depending
on their insulation levels. construction and winding connections.
The zero-sequence impedance depends on the type of the core used and
winding connections as described in section D.2.2. In general it has a value
which is smaller than the other sequence impedances as explained in more
detail in reference [1].
D.1.3 Overhead lines and cables

A similar situation to that with transformers exists with overhead lines and
cables. namely that if their physical asymmetry is neglected then positive-.
negative- and zero-sequence impedances may be assigned to them, the posi-
tive- and negative-sequence values being the same whereas the zero-sequence
value is different. In the case of lines and cables, this occurs because the
positive- and negative-sequence currents flow only in the phase conductors
but the zero-sequence currents flow in the phase conductors and return in the
586 Appendix D
earth or neutral conductor and ground. This causes not only the reactances to
differ but also the resistances of the paths taken by the currents. As an
example, the positive and negative sequence impedances for a 4QO ky line in
the British system, consisting of bundles of four c~nductors, is ZI =Z2 - 0.02
+ jO.3n/km, and the zero-sequence impedance is Zo =0.1 + jO.8 n/km.
D.1.4 The etTects of physical asymmetry

Because physical asymmetry is present in parts of power networks, the mutual
impedances that are present cause interactions between the sequences. These
can clearly be taken into account but they are generally neglected in the
interests of simplifying the determination of the behaviour of networks during
both normal and fault conditions and in the following sections the techniques
which are usually employed are described.
D.2 SEQUENCE NETWORKS

An actual three-phase network is represented by three separate sequence net-
works which have to be interconnected appropriately to represent particular
conditions.
As an example, the network shown in Fig. D.2(a), which is represented in
single-line form for simplicity, would be replaced by the three networks shown
in Fig. D.2(b).
Alternators produce only positive-sequence e.m.f.s, i.e. EAI and E8h and
therefore no e.mJ.s are present in the negative- and zero-sequence networks.
Other important factors which must be taken into account are considered below.
D.2.1 Earthing of neutral points

In Fig. D.2(a), the star points of the generators were shown to be solidly
earthed and therefore they were shown to be connected directly to the zero-
voltage conductor in each of the sequence networks.
In practice alternators are normally earthed through resistors to limit the
currents which will flow in the event of faults to ell!!h.lt will Ee appreciated
that the sum of the three positive-sequence currents (Iah hi and lei) is zero and
that no current flows through an earthing resistor as a result of the flow of
these currents and no voltage drop is caused across such resistors. They are
therefore not included in the positive-sequence network. Similar consider-
ations apply when negative-sequence currents flow. When zero-sequence cur-
rents flow, however, the current in an earthing resistor (Rg) is 3/0 because the
three-sequence currents are of the same phase and the voltage drop across the
resistor is 3/0Rg. To allow for this situation, a resistor of 3Rg must be included
in the zero-sequence network to allow for the fact that the current flow is only
10 This representation is shown in Fig. D.3.
Sequence networks 587
EA
-+-
0-X~
~
ZSA
Line Z;
X-8 -
EB ZSB
-
IA
~
-
(a)
V1 =0
EA1
-+-
-EB1 Positive
sequence
~ ~ _ I
L-2:J
Negative
~l---~~ sequence
-+-
~
G
~=o
-+-
~l---~~ ~:eoc,
7;..0
(b)
Fig. D.2 A network (a) and its three phase-sequence networks (b).
r---------o Va = 0
Fig. D.3 Zero-sequence network for an alternator earthed through a resistor Rg

588 Appendix D
D.2.2 Transformer connections
The three-phase windings of power transformers may be connected in star or
delta and indeed it is common practice to have the high voltage windings
connected in star because this reduces the cost of insulation, as stated in
Chapter 6, and the low voltage windings are often connected in delta. There
are also situations where transformers have three sets of three-phase windings,
a common arrangement then being to have the two higher sets of voltage
windings connected in star and the other set of windings connected in delta.
Clearly positive- and negative-sequence currents can flow in both star- and
delta-connected windings through the sequence impedances of the primary
and secondary windings. It should be noted that the exciting impedances are
usually regarded as infinite in the interests of simplicity, i.e. the exciting
currents are assumed to be zero, and therefore ideal transformation is assumed,
namely IpNp+ isNs = O. A transformer may therefore be represented in both the
positive- and negative-~quenc~ netwo!.ks by t~o percentage or per-unit im-
pedances in series, i.e. Zip and ZIS and Z2p and Z2p as shown in Fig. D.4(a).
Clearly zero-sequence currents cannot flow from a circuit into a delta-con-
nected set of transformer windings because there is no path by which they may
leave. A delta-connected set of windings therefore presents an infinite imped-
ance to incoming zero-sequence currents. A set of zero-sequence currents can
however circulate in a delta-connected winding and such currents would en-
~ ~
Positive-sequence Negative-sequence
(a)
Primary Secondary
(b)
(c)
Fig. D.4 Sequence networks of a transformer.(The quantities shown are per-unit.)

Sequence nt?tworks 589
(a)
0---0
(b)
Fig. D.S A star-delta connected transformer (a) and its zero-sequence network (b).
counter the zero-sequence leakage impedance of the winding (Zo). In the

unlikely event of a transformer having two delta-connected windings as shown
in Fig. D.4(b), the zero-sequence network would be as shown in Fig. D.4(c).
Zero-sequence currents can clearly flow into a star-connected set of wind-
i~gs provided that the star-point is earthed, either directly or via an impedor
(Zg). As stated !bove, earthing impedors must be represented in zero-sequence
networks by 3Zg because the actual impedor carries the three zero-sequence
currents.
When a transformer is connected in star-delta, zero-sequence currents cir-
culate around the delta-connected windings when zero-sequence currents flow
in the star-connected windings. They do not, however, flow into the lines
connected to the delta-connected windings. Because of this, the impedance
presented to the zero-sequen~e curreE-ts is that of the leakage impedances of
the two sets of windings, i.e. Zop and Zos. This situation is shown in Fig. D.5(a)
and (b).
If both windings of a two-winding transformer are connected in star, then
zero-sequence currents will flow in each of the windings if their star points
are both earthed and the magnitudes of the currents in those circumstances
will be dependent on the leakage impedances of the windings and the imped-
ances in the ground paths. This situation is illustrated in Fig. D.6. If the star
point of only one winding is connected to earth then the zero-sequence current
in it will be limited by the zero-sequence exciting impedance of the trans-
former (ZOe). This situation is shown in Fig. D.7.
590 Appendix D
o o
(b)
Fig. D.6 A star-star connected transformer with both star points earthed, and its
zero-sequence network.(All quantities are per-unit.)
(a)
(b)
Fig. D.7 A star-star connected transformer with only one star point earthed (a), and
its zero-sequence network (b).
Sequence networks 591
SW1
SW2
Fig. D.S General form of zero-sequence network for two-winding transformers.(All

quantities are per-unit.)
When dealing with two-winding transformers their zero-sequence circuits

can be readily determined from Fig. D.S, in which each switch 1 is closed if
its associated winding is star connected and each switch 2 is closed if its
associated winding is delta connected.
Similar circuits may be used for three-winding transformers, an example
based on a star-delta-star transformer being shown in Fig. D.9.
ZaP
All quantities are shown in per unit
Fig. D.9 A three-winding transformer (a) and its zero-sequence network (b).
592 Appendix D
D.3 THE INTERCONNECTION OF SEQUENCE NETWORKS
To enable the actual operating conditions of power systems to be determined,
the appropriate sequence networks must be interconnected in particular ways.
These are described below for a range of possible conditions.
D.3.1 Balanced three-phase normal or fault conditions

Under conditions associated with either normal operation or when three-phase
faults are present, the phase voltages and currents are balanced. As a result
they have no negative- or zero-sequence components and the behaviour may
be determined from the positive-sequence network, an example being shown
in Fig. D.lO. It will be clear that the phase currents and voltages at points in
the network can be readily determined using complex quantities.
D.3.2 Conditions when a phase-to-phase fault is present

When an interphase fault is present, the current leaving one of the faulted
phases at the fault position is equal to the current entering the other faulted
phase. This situation is illustra.!.ed in Fig. D.ll(a) for a ~ult between phas~s
'a' and 'b':... The iault current h is equal to ~e current la and equal to - lb.
As a result la = - h, and of course the current Ie is zero. These currents, which
are shown in Fig. D.II (b), may be transformed into the positive- and negative-
sequence currents shown in Fig. D.ll(c). It will be seen that the positive- and
negative-sequence currents of the healthy phase (lIe, 12c) are in antiphase and
have equal magnitudes given by:
-
EgA - - - - EgB
~ Z1tA 11A 11B Z1tB -
0-i~X Z11A ]4 z X -iK)

1IB
Z1gA
- Z1gB
(a)
-
EgA 4t
- E9B
All quantities are shown in per unit

(b)
Fig. 0.10 A network (a) and its sequence network following a balanced fault (b). The
excitation impedance Zoe is neglected.
The interconnection of sequence networks 593
1_[ 1-1-[ 1_IIal_11fl

Ic - -...J3 ...J3
2c
or the complex form:
(D.I)
The positive- and negative-sequence networks representing an actual circuit

must be interconnected so that the healthy-phase sequence currents at the fault
position are of equal magnitude but of opposite polarities. This is illustrated
in the example shown in Fig. D.ll(d) which is based on the circuit shown in
Fig. D.lO(a). It will be appreciated that the actual fault current can be deter-
mined using equation (D.I) and the actual phase and voltage currents at other
points in the network can be found by transforming the sequence quantities at
Phases
----.--------------0-",ba
-c
Ta=t;+L-~~ !1c=O
(a)
~ .."'f---_....--.. Ta = t;
(b) Ic=O
Positive-sequence Negative-sequence
(c)
~IB
(d)
Fig. D.ll Conditions and sequence network for an interphase fault. (All quantities are
per-unit.)
594 Appendix D
the corresponding points. A numerical example is included later in section
DA.
D.3.3 Conditions when a single phase to earth fault is present

When a single phase to earth fault is present, a current leaves the faulted
conductor and flows in earthed conductors and/~r the ground. For the fa~lt on
phase a sho~n in ~g. D.12(a), the fault current h is equal to the current la and
the currents Ib and Ie are zero. These currents, which are shown in Fig. D.12(b),
may be transformed into the positive-, negative- and zero-sequence currents
shown in Fig. D.12(c). It will be seen that the three sequence-currents associ-
Phases
----~--------------oa
--~--~--------~b
----+--+--~~-----oc
~~=ol+fc=O
(a) (b)
- -
~
/1c
) ",".li3, +
/1b /2c
(c)
(d)
Fig. D.12 Conditions and sequence network for a phase-to-earth fault.

Numerical example 595
ated with phase a are in phase with each other and have the same magnitude,
i.e.
- - - -
lla =l2a =loa =h/3 (D.2)
The three sequence networks representing a circuit must therefore be con-
nected in series, as in the example shown in Fig. D.12(d). Again the actual
currents at any point in a network can be determined and the fault current can
be calculated using equation (0.2).
A numerical example is provided in section 4.4 and this indicates how
factors such as transformer connections, methods of earthing and the presence
of impedance in fault paths may be taken into account.
D.3.4 Impedances used in sequence networks

It will be clear that steady-state conditions are not established in a power
system immediately after a disturbance, such as the incidence of a fault,
occurs, and if it is necessary, it is now possible with the aid of modem
computing equipment to determine the instantaneous values of the various
currents and voltages which will result from a particular disturbance.
In many cases, however, such detailed information is not necessary. As an
example, when choosing the fault settings of protective equipment, a know-
ledge of the minimum currents likely to flow during fault conditions may be
adequate. When considering the stability of a protective scheme during exter-
nal fault conditions, however, the maximum currents which may flow at such
times must be known.
The higher currents which flow immediately after fault incidence are deter-
mined by using the sub-transient values of the sequence impedances of the
sources which are present in the circuit being considered. The lower currents
which are present shortly after a fault occurs may be determined by using the
transient values of the sequence impedances and the eventual steady state
conditions can be determined by using the synchronous impedances. These
various values were referred to earlier in section D.I.
0.4 NUMERICAL EXAMPLE

A small network is shown schematically in Fig. D.13a. The parameters of each
item of plant referred to the same base and neglecting the resistances are given
below,
Generator GA ZAI =jO.3 pu, ZA2 =jO.2 pu, ZAO =jO.05 pu
Generator GB ZBI = jO.25 pu, ZB2 =jO.15 pu, ZBO =jO.03 pu
Transformer TA
596 Appendix D
(a)
Nl
Positive
Fl ~1 = jO.15
N2 Negative
F2
No Zero
Fo ~o=jO.4
(b)
El=1PU~
Zo=jO.17
Positive Negative Zero

(c) (d)
(e)
Fig. D.13 Network (a) and sequence networks for a range of faults.
Reference 597
Transformer TB ZtBI =ZtB2 =Zmo =jO.1 pu
Line Zll =ZI2 =jO.15 pu, ZIO =jO.4 pu
It is further assumed that the two generators produce voltages of the same
phases and of magnitude 1 p.u. The currents for faults near the secondary
terminals of transformer TA are determined below.
The positive-, negative- and zero-sequence networks are shown in Fig.
D.13b. These networks may be further simplified as shown in Fig. D.13c.
D.4.1 Short-circuit between phases band c and earth

For this fault the three sequence networks are combined as shown in Fig.
D.13d. The sequence currents are then
11 = V = 1 =-j3.15pu
Z Z2Z0 0230.18xO.17
I+ Z2 + Z0 J. +JO.18+0.17
- -
Similarly, 12 =j1.53 pu and 10 =j1.62 pu.
The phase currents at the fault may be found by substituting the above
values and are:
- - -
la = II + 12 + 10 = 0
Ib =/0 + crll + al2 = - 4.05+ j2.43pu
Ie =10 + all + a2/2 =4.05 + j2.43 pu
D.4.2 Short-circuit between phase a and earth

For this fault the three sequence networks are connected as shown in Fig.
- - - 1 .
D.13(e). Hence II =h =10 =j(0.23 + 0.18 + 0.17) - J 1.72 pu
The phase currents may be found by substituting the above values and are
la = 3/0 =- j5.17pu
h=/e=O
REFERENCE
1. Roeper, R. (1985) Short-Circuit Currents in Three-Phase Systems, (2nd edn),
Siemens and A. G. and J. Wiley and Sons.
598
CONCLUDING REMARKS
It was stated in the Preface to the first edition that all aspects of the
protection of power systems cannot now be covered in a single book. It is
hoped, however, that most of the important areas have been dealt with in
this volume and that the information provided will enable readers to cope
with any situations which may arise. Recent advances in information
technology and communications permit high levels of signal processing
and data availability across the entire range of power system functions
such as protection, monitoring and control. Innovative ways of extracting
relaying information from data gathered from the power system based on
artificial intelligence techniques are now being explored. Ways of co-
ordinating and integrating more closely protection with other power
system functions are being sought. All these developments must go hand in
hand with high levels of reliability traditionally expected from power
system protection relays.
It is not reasonable to expect from any single individual to have a
complete mastery of all the techniques which impact on protection today.
Hence, the first sentence appearing in these concluding remarks is more
true today than when it was first written five years ago. The second edition
of the book seeks to give the reader a sound grounding in protection and
maintain a balance between two essentials - grasp of fundamentals and
appreciation of technological opportunities.
Index
Acceleration schemes 429 Artificial intelligence 526-57

Adaptive relaying 539-41 Artificial neural networks 533-9
Aerial mode 452,456,471 delta rule 537
Alternators feedback networks 537
asynchronous operation 258, 269 feedforward networks 537
Buchholtz relays 275 kohonen layer 538
complete protection 277 self-organizing maps 537-8
construction 255 Asymmetry effects 579
current differential schemes 262 Asynchronous operation
earth-fault protection 262 alternators 258, 269
earth faults 274 synchronous motors 248
excitation 258 Auto-recIosing 311, 381
fuses 262 Auto-transformers 219
harmonic biasing 267
100% winding protection 263 Back-up protection 404
IDMT relays 273 Balanced-voltage schemes 147, 162,
interturn faults 267 307
loss of excitation 269 Bewley lattice 447
main transformer 265 Bias, transformer protection 202, 203,
mechanical equipment 261 205
mho relay 271 Bias winding 308
monitoring 275 Blocking schemes 429
negative-sequence components 269 Blocking signals 459
negative-sequence currents 259 Buchholtz relays, alternators 275
operation 258 Burden, voltage transformers 87
overcurrent protection 273 Busbars
overloading 273 bus-coupler circuit breakers 285
pole slipping 272 bus-section circuit breakers 284
rotor-winding faults 257 CT positioning 287
stator-winding faults 256, 257 enclosed 281
time/current characteristic 261 gas insulated 283
tripping arrangements 277 open type 283
unbalanced loading 259 reserve 285
unbalanced operation 269 sectionalization 284
unit transformer 265 sectionalized three-phase 292
zero-sequence components 268 unsectionalized single-phase 289
Arcing period 9, 10 unsectionalized three-phase 290
600 Index
Busbar protection 279 Current differential protection
cr saturation 299 balanced-voltage principle 307
current differential schemes 167,280, bias characteristic 337
288,296,301 bias winding 308
external faults 286 delta/star transformers 206
frame leakage 280 effects of capacitive current 319
fuses 288 lines and cables 307
IDMT relays 288 multi-ended circuits 337
incorrectly protected zones 287, optical fibres 333, 335
293 pilot wires 329,331
interlock schemes 280 sequence networks 312
internal faults 286 split pilot 308
numerical relay 304 summation transformers 308, 313, 322
relays high-impedance 296 telephone circuits 330
relays low-impedance 297, 299 voltage-balance schemes 341
testing 303 zero-sequence currents 341
Current differential schemes 145, 307,
Cables 310 312,333,343
Cables, protection 28, 427 alternators 262
Cables, sequence impedances 585 balanced-voltage 147, 162,312,317
Capacitive current 368 bias 168
Capacitor divider 95 busbar protection 167, 280, 288, 301
Capacitors circulating-current 147, 315
protection with fuses 33 compensated pilot wires 148
series 430 CT core permeability 152
Capacitor voltage transformers 96 CT errors 148
Carrier, power line 364 CT ratings 150
Circuit breakers earth-fault protection 169
bus-coupler 285 faults on star-connected transformer
bus-section 284 windings 170
Circulating-current schemes 147 high-impedance relays 171
Clarke's transformation 580 imbalance 152
Commissioning 567 load current 155
Commutator faults, dc motor 252 long physical length units 158
Comparators mismatching 148, 149, 156
amplitude 395, 396 patent 147
phase 399 pilot wires 307
Compensation relays 174
residual 411 rotating machines 232
sound-phase 409 sequence components 159
vectorial 427 short physical length units 158
Contraphase 380 summation transformers 161
Coordination of protection teed feeders 165
induction motors 242 testing 561
synchronous motors 247 Current ratings, current transformers 48
Cross-polarization 407 Current transducers
CT, see Current transformers electronic 83
Current differential principle 145 fibre-optic 83
Index 601
Current transfonners analogue data -+ analogue signals
air-gapped cores 81, 148 492,493
bar primary winding 76 Digital relay 435-7
construction 71 Digital signal processing and
core flux 51 protection 501-23
core saturation 59, 68 algorithms 519-23
current ratings 48 digital filters 514-17
equivalent circuit 42 discrete Fourier transfonn 517
errors, 46, 148 sampling 513
exciting current 43 Directional comparison 457
insulation 75 Directional electronic relays 119
leakage reactance 73 Directional relays 114
linear couplers 78 Discrete Fourier Transfonn 434
magnetic core 47,68, 73 Discrimination and coordination, fuses
non-sinusoidal conditions 49 26
open-circuit behaviour 68 Distance protection 385
positioning 74 compensation 415
post-type 75 implementation 423-5
remanent core flux 65 instability 420
saturation 299 operating time 416
short-circuit forces 76 reach 415
specifications 84 source impedance effects 416,419
testing 84 switched schemes 405
turns 48 teed lines 429
turns compensation 46 three-phase lines 405
windings 74 Drives, variable speed 254
Current-transfonner core penneability
152 Earth-fault detection 106
Current-transfonner positioning, Earth-fault protection
busbars 287 alternators 263
Current-transfonner rating, current current differential schemes 169
differential schemes 150 restricted 208
Cut-off characteristics, fuse 25 Earthing, neutral point 510
CVT, see Capacitor-voltage Earthing transfonners 216
transfonners Electromagnet shaded-pole 104
Embedded generation 550
Dead time 381 EMTP 565
DC motors, see Motors, direct current Energy product 23, 556
Differential schemes, see Current Equivalent circuit, current transfonner
differential schemes 42
Digital communications 479-500 Error detection 494-5
adaptive relaying 499 cyclic redundancy check 495
Global Positioning Satellite 499, frame check sequence 495
546 longitudinal redundancy check 495
Digital protocols 489 vertical redundancy check 495
encoding of data 489 Errors
digital data -+ digital signals current transfonners 46
490-1 voltage transfonners 89, 90
602 Index
Excitation, alternators 258 fusing current 556
Excitation loss fusing factor 10, 23
alternators 269 high-voltage 6, 7, 8
synchronous machines 246 low-voltage 6, 7, 8
Exciting current miniature 6, 7
current transformers 43 minimum fusing current 10, 13
transformers 186,200,202 in motors 29
Expert systems 526-9 non-conventional fusing current 9
heuristics 526 operating frequency 25
inference strategies 527 overload withstand capability 558
predicate calculus 528 in power transformers 30
pre-arcing period 9
Faults prospective current 10
accelerated clearance 429 quality assurance 559
close-up 406 rated current 10, 23
interphase 411 rating 16
phase-to-earth 408 rotating machines 230
resistive 389 testing 13, 24, 557
three-phase 412 time/current characteristic 6, 9, 22,
Feeders 307 24,556,557
teed, differential schemes 165 virtual time 24
Feeder transformer protection 426 in voltage transformers 32
Fibre-optic current transducers 83 Fusing current
Flexible AC transmission systems conventional 9
543-50 non-conventional 9
Fuse construction Fusing factor 10,23
cartridge 6, 7, 14 Fuzzy logic 529-32
expulsion 6, 8, 18 defuzzification 531
liquid 6, 8, 21 fuzzification 531
semi-enclosed 6, 8, 18
Fuses Genetic algorithms 547-8
alternator 261 Grading, current 125
application requirements 21 Ground mode 452, 456, 471, 474
arcing period 9, 10
behaviour 9 Hardware and software environments
breaking current capacity 556 503-13
in busbars 288 busses 504
in cables 28 random access memory 505
in capacitors 33 read only memory 505
cartridge 7, 15, 16 relay configuration 509-13
conventional fusing current 9 sample and hold circuit 510--11
cut-off characteristic 25 Harmonic biasing 267
in dc motors 252
discrimination and coordination 26 IDMT relays 110
elements I, 7, 10, 12, 14,29 in alternators 273
energy product 23 in busbars 288
fillers 8,15, 16 electronic 118
full-range 23 in transformers 198
Index 603
Image conductor 453 Low-impedance relays 154
Imbalance in current differential
schemes 152 Magnetic core, current transformers
Impedance 47,68,73
measurement 393 Main transformer 265
mutual 578 Memory circuits 407
referred 569 Merz-Price schemes 147, 346, 385
replica 396, 402 Mho relay
self 578 in alternators 271
sequence 583 characteristic 396
m sequence networks 595 during transients 401
synchronous machine 583 Modal components 581
Impedance-measuring schemes, testing Model
564 of system 435
Impedance protection, see Distance of waveform 434
protection Modes 455
Impedance-relay, settings 407 aeriaI452,456,471
Instability detector 421 ground 452,454,456,471,474
Integrated protection 541-6 Motors, direct current
Energy Management System 544 commutator faults 252
Supervisory Control and Data fault conditions 251
Acquisition System 544 and fuses 252
Interlock schemes 385 normal operation 251
busbar protection 280 overheating 252
lines and cables 345, 349, 365 overloading 252
Interlocking signals 360 overspeeding 252
Inter-tripping 405 rectifier-fed 253
Interturn faults, alternators 267 starting 252
Inverse definite minimum-time relays undervoltage operation 252
(IDMT) 111 Motors, induction
coordination of protection 242
Kalman filtering 435 faults on stator 236
over/under frequency operation 241
Leakage reactance, current over/under voltage operation 241
transformers 73 stalling 239
Lenticular characteristic 400 starting 239
Lightning strikes 194, 197 thermal relays 243
Line traps 364 time/ current curve 235
Linear couplers 78 unbalanced operation 240
Lines Motors, protection using fuses 29
artificial 564 Motors, small 234
overhead 310 Motors, synchronous
Links asynchronous operation 248
microwave 365 loss of excitation 246
optical fibre 364, 372 loss of synchronism 244
radio 372 overloading 244
Loss of synchronism, synchronous pony motor 246
motors 244 protection coordination 247
604 Index
rotor faults 248 signals 371
rotor temperature sensors 250 testing 485
slip frequency 245 Phase displacement 368
starting 246 Phase-to-phase faults, sequence
Multiplexing 493-4 networks 592, 593
Multi-ended circuits 337 Pilot wires 307, 363
Multi-ended circuits, distance compensated 148
protection 429 metallic 338
split 308
Negative-sequence component, PLC, see Carrier, power line
alternators 269 Polarizing voltage 407
Negative-sequence impedance, Pole slipping, alternators 272
synchronous machine 584 Polygonally-shaped characteristic 425
Noise 236 Post-type current transformers 75
Power transfer, maximum 431
Ohm relay 397 Pre-arcing period, fuses 9
Open System Interconnection Model Prospective current, fuses 10
496-9 Protection, duplicate 425
electromagnetic compatibility 498-9 Pu, see Per-unit quantities
message protocols 498
message structure 498 Quadrilateral characteristic 400
Operating frequency, fuses 25
Optical fibres 333, 335, 364, 372 Reactance relays 398
Overcurrent detection 107 Reactance, synchronous 583
Overcurrent protection Reclosing, auto- 311, 381
alternators 273 Reflection coefficient 447
application 124 Relays
current grading 125 algorithmic 436
relay connections 106 balanced beam 394
ring circuit 137 beam 386
specification 140 cross-polarized 355, 407
VT requirements 139 directional 114, 345, 350, 357, 395
Overhead lines, sequence impedances eddy current braking 104, 105
585 electromechanical 108
Overloading electronic directional 119
alternators 273 electronic processing 118
dc motors 252 high-impedance 171,296
synchronous motors 244 IDMT 110
Overspeeding, dc motors 252 IDMT electronic 118
induction 104
Parameter estimation 434 instantaneous 109
Percentage quantities, see Per-unit inverse-time 104
quantities low-impedance 154, 297-9
Permissive intertrip 428 mho 396
Permissive signals 459 offset mho 397
Per-unit quantities 569 ohm 397
Phase-comparison schemes 372,385 overrun 104, 113, 117
lines and cables 345, 367 plain-impedance 395, 399
Index 605
reach 403, 409 Simulated annealing 547-8
reactance 398 Simulator, programmable power
reverse current 104 system 564
self-polarized 407 Slip frequency 245
sensitivity 357 Solkor scheme 317,325
settings 354 Stalling, induction motors 239
starting 361, 372 Starting
testing 560--1 dc motors 252
Remanent flux, current transformers 65 elements 405
Resistivity, ground 453, 471 induction motors 239
Rotating machines 227 synchronous motors 246
control equipment 233 Sub-transient reactance 585
current-balance schemes 232 Summation transformers 161
current differential schemes 232 Superimposed quantities 457, 477
fuses 230 Symmetrical components 573, 583
phase unbalance relays 233 Synchronous generators, see
ratings 227 Alternators
thermal devices 231 Synchronous machines
thermal relays 230 negative-sequence impedance 584
Rotor temperature sensors 250 zero-sequence impedance 584
Saturation, current transformer cores Tank earth protection 212,223

59,68 Teed circuits, see Multi-ended circuits
Self-polarized relays 407 Teed feeders, current differential
Sequence components 455 schemes 165
Sequence impedances 455 Telephase 373
cables 585 Testing
overhead lines 585 automatic 567
transformers 585 bus bar protection 303
Sequence networks 312,354,381,586 CTand VT 559
balanced conditions 592 current differential schemes 561
interconnection 592 current transformer 84
phase-to-earth faults 593 impedance-measuring schemes 564
phase-to-phase faults 593 phase-comparison schemes 561
transformer 588 protective schemes 561
Series-compensated lines 474 relays 561
Settings voltage transformers 100
distance schemes 427 Thermal devices, rotating machines
impedance relay 407 231
stage 1, 2 and 3 403, 404 Thermal relays
time multiplier 130 induction motors 243
Signalling 360 rotating machines 230
blocking 361, 363 Time / current characteristic
channels 363 alternator 261
permissive 361, 363 fuse 6, 9, 22, 24
Signals induction motors 235
blocking 459 Time multiplier 130
permissive 459 Transducers, voltage and current 140
606 Index
Transfer trip 360, 428 summation 161, 308, 313, 322, 351
Transfonnation tank earth protection 212, 223
(a,~,O) 580 tap changing 184, 201, 202
Clarke's 580 three-winding 591
Transfonned quantities 575 Transients 400
Transfonners long lines 402
auto- 219 Translay scheme 321, 329
connections 588 Transmission lines 345, 347, 387
cooling 183 asymmetry 413, 456
core losses 196 capacitance 348
current 372 characteristic impedance 349, 391
current differential protection in equations 348, 390
delta/ star 206 earthed conductors 347
current differentiai protection, three input impedance 387, 390
or more windings 208 mutual-coupling effects 414
delta-connected windings 211 propagation constant 349,391
earthing 216 series compensated 430, 474
exciting current surges 186, 200, surge impedance 349
202 three-phase input impedance 392
external faults 196 Transmission media 485-7
faults 184, 191 free-space transmission 487
feeder 340 guided transmission 485-6
healthy conditions 188 interfaces for protection and
and IDMT relays 198 communication 487
internal faults 194 EIA-232D 488
iron losses 195 EIA-530 489
and lightning strikes 194, 197 other 489
mechanical failures 197 Transmission networks 483-5
operation at frequencies other than Traps, line 364
nominal 196 Travelling-wave schemes 440, 456
overloading 195 non-unit 468
overvoltage operation 195 Trip, transfer 360
protection 179 Turns compensation, current
protection and CTs 200 transfonners 46
protection using digital techniques
223 Ultra-high-speed protection 439
protection using fuses 30 Unbalanced loading, alternator 259
protection using hannonic bias 203 Unbalanced operation, alternator 269
protection using non-linear Unit transfonner 265
percentage bias 205
protection using percentage bias 202 Variable-speed drives 254
protection of single-phase two- Voltage transfonners
winding 199 abnonnal conditions 90
relay settings 202 burden 87
relays used for protection 202 capacitor (CVT) 96
restricted earth-fault protection 208 capacitor divider 95
sequence impedances 585 connections 91
sequence networks 588 construction 91
Index 607
cores 87, 92 faults, external 449
with electronic equipment 98 faults, internal 447
errors 89, 90 reflections 445
housing and insulation 94 resistive faults 444
with opto-electronic materials 99 three-phase lines 452
oscillatory outputs 98 voltage and current 442, 443, 447
protection using fuses 32 Wires, pilot 329
specification 99
testing 100 Zero-sequence components, alternators
windings 87, 91, 93 268
VT, see Voltage transformers Zero-sequence currents 341
Zero-sequence detection 107
Waves, travelling 439, 440, 441, 449 Zero-sequence impedance,
effects of ground return 445 synchronous machines 583
effects of losses 444 Zone settings 403, 404

(C. Christopoulos, A. Wright (Auth.) ) Electrical PDF

Uploaded by

Copyright:

Available Formats

(C. Christopoulos, A. Wright (Auth.) ) Electrical PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(C. Christopoulos, A. Wright (Auth.) ) Electrical PDF

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the book?

What topics are covered in the book?

Electrical Power System

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Printed on acid-free paper

All Rights Reserved.

Preface to the Second Edition xvii

Preface to the First Edition xviii

4 Overcurrent and earth fault protection 103

5 Current-differential protective schemes 145

6 The protection of transformers 179

7 The protection of rotating machines 227

8 The protection of busbars 279

9 The protection of overhead lines and cables by

10 Interlock and phase-comparison schemes for the protection

11 Distance-type protective schemes for overhead lines

12 Ultra-high-speed schemes for the protection of long

13 Digital communications 479

14 Digital signal processing (DSP) and protection 501

15 New relaying concepts 525

Appendix A The testing and application of power-system

Appendix B Percentage and per-unit quantities 569

Appendix D The determination of power-system behaviour

The death of Professor Arthur Wright in the summer of 1996 deprived

1.1 HISTORICAL BACKGROUND

C. Christopoulos et al., Electrical Power System Protection

During this century, relay-based protective schemes have been produced.

1.2 BASIC REQUIREMENTS

1.3 FUSE TYPES AND CONSTRUCTIONS

~~!,,->01I11""_ _.....:...._ _ _ _ _ Fuselink

Granular quartz _ _ _ __+_'~

1.3.2 Semi-enclosed fuses

1.3.3 Expulsion fuses

1.3.4 Liquid fuses

Asbestos liner _ _-+_-++-

"----~- ____ Fuse carrier

1.4.1 The pre-arcing period

1.4.2 The arcing period

Fig. 1.8 Equivalent circuit.

of providing current limiting, which assists in reducing damage to the circuits

1.4.3 Determination of fuselink performance

Fig. 1.10 Schematic diagram of test circuit.

1.5 THE CONSTRUCTION OF CARTRIDGE FUSES

As indicated earlier in section 1.3, cartridge fuses are produced in a number

1.5.1 Fuse elements

(a) low specific heat;

1.5.2 Fuselink bodies

1.5.3 Filling material

1.5.4 Mountings and ratings

1.6 SEMI-ENCLOSED FUSES

1.7 EXPULSION FUSES

Tin or copper element

~~i~~~~~~i:iJcl~opper button head rivet

Strain tape Copper sleeve

...g._ _ _ _ Arc-extinguishing liquid

-!-_ _ _ _ Glass tube

1.9 THE APPLICATION OF FUSES

i~~~~i:iJcl~opper button head rivet