Xinzhou Dong

AC/DC Hybrid
Large-Scale
Power Grid
System
Protection
AC/DC Hybrid Large-Scale Power Grid System
Protection
Xinzhou Dong

AC/DC Hybrid Large-Scale

Power Grid System
Protection
Xinzhou Dong
Department of Electrical Engineering
Tsinghua University
Beijing, Beijing, China

ISBN 978-981-19-6485-5 ISBN 978-981-19-6486-2 (eBook)

https://doi.org/10.1007/978-981-19-6486-2

The translation was done with the help of artificial intelligence (machine translation by the service DeepL.
com). A subsequent human revision was done primarily in terms of content.
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Foreword by Zhou Xiaoxin

A major power outage is a disaster that modern society cannot afford; therefore, it is
the basic responsibility of the relay protection and safety and stability control system
of a power system to prevent the occurrence of such power outages. Chain failures
and incorrect actions of relay protection and safety stability control systems during
fault handling often become the main drivers for the occurrence and development of
major blackouts.
China has built a giant hybrid AC-DC power grid with the largest installed capacity
and power generation capacity, highest voltage level, largest consumption of new
energy generation and highest proportion of DC transmission worldwide. Ensuring
the safety of this hybrid grid is a matter of national security. However, we are facing
a very serious security situation. The grid of East China, with Shanghai as its core,
and the grid of South China, with Guangzhou as its core, have formed a grid pattern
of ultrahigh voltage and multiple DC feeds. AC grid failure often leads to multiple
simultaneous DC phase change failures. Therefore, an investigation of the safety
technology of AC-DC hybrid power grids is critical.
This book analyzes the action behavior of traditional relay protection during chain
failures in AC-DC hybrid grids from the perspective of relay protection for the
purpose of protecting the safety of large grids. The book’s findings clearly point
out that the accidental overload after fault removal and the incorrect action of relay
protection during oscillation are the main problems of traditional relay protection.
With the emergence of hybrid AC-DC grids, the failure of DC system phase changes
due to voltage reduction during AC system failure becomes almost certain, and the
long-term phase change failure will inevitably cause DC lockout, which will trigger
more serious chain reactions.
In response to these problems, this book presents the principles and techniques
of distance protection immune to overload and distance protection immune to oscil-
lation, control techniques to prevent long-term phase change failures of converters
and techniques for emergency tidal control of DC system participation to avoid large
power surpluses in the AC grid, to which the DC system is connected. These tech-
niques have an important role in ensuring that the power balance of the AC grid is
maintained during faults and fault handling and in maintaining the integrity of the

v
vi Foreword by Zhou Xiaoxin

grid topology, as well as allotting valuable time for the subsequent functioning of a
safe and stable control system.
This book is innovative; the concepts of non-tripping relay protection and phase
change failure prevention and control are concrete; and the understanding of inter-
locking faults and interlocking faults leading to large outages is comprehensive and
objective. The theoretical and technical measures described in the book are complete,
self-contained and have an important role in improving the safety prevention and
control system of the power grid. China’s relay protection has excellent performance
in correct action to remove faults, and the correct action rate of relay protection
ranks second worldwide. China is a global leader in both theoretical research and
technical application. The theoretical research and practical application of techniques
for interlocking faults and major outages are equally valuable for not only correctly
reflecting faults and quickly removing faulty equipment but also avoiding negative
consequences, further enriching existing knowledge of power system safety. There is
therefore good reason to consider this book another important contribution of relay
protection practitioners to power systems in general and to power system safety in
particular.
The author of this book has been engaged in research on relay protection tech-
nology of power systems and has made many achievements in research on traveling
wave protection and system protection. The contents of this book are the result of
his long-term research on the problems of interlocking faults and large outages.
The publication of this book will contribute to the integration of relay protection
technology and safety and stability control systems in a guiding role. The book is
an important reference for readers, especially senior students and postgraduates, as
well as engineers and technicians engaged in relay protection and safety and stability
control technology of power systems.

Beijing, China Zhou Xiaoxin

May 2022
Foreword by Sun Guanghui

Ensuring safe and stable operation is the primary task of a power system. After
decades of unremitting efforts of power workers, China’s power system, with its safe
and stable operation, has made world-renowned achievements, surviving nearly three
decades without large-scale power outages, to ensure national economic development
and social stability. As the first line of defense of relay protection and the second and
third lines of defense of safety, automatic devices have great merit.
Various components and equipment of a power system often experience short-
circuit faults, equipment tripping and other accidents. Relay protection for a single
component fault can always function through the circuit breaker to remove the faulty
components. The power system generally meets the N-1 setup standards and restores
normal operation for multiple components impacted by successive faults tripping.
To implement the predetermined control strategy, the stability control system can
also maintain the stability of the power system operation. However, with the rapid
development of the power system, the demand for west–east power transmission,
north–south mutual aid and large-scale, new energy construction and consumption
in the western region, China has built a hybrid AC-DC power grid with ultrahigh
voltage transmission as the main source. Due to the strong, direct and weak AC power
grid, the electromagnetic ring network still exists and exhibits other characteristics,
rendering the power grid structure more complex. DC lockout may cause serious
overload of the AC lines in parallel with it, and short circuiting of the AC system
will almost always cause the nearby drop point of the DC transmission phase change
failure. AC and DC mutual influence and coupling are triggered by the power system
to increase the probability of a chain reaction, when the relay protection set for a
single component fault encounters a challenging situation, which may even appear
to be an incorrect action. For example, distance three sections will not be able to
effectively escape the tidal shift caused by line accident overload, causing the same
section of multiple lines to be consecutively tripped (on “8.14” in 2004, the USA
experienced an increase in blackouts; on “7.1” in 2006, our accident occurred).
Distance protection may incorrectly function in the event of out-of-step oscillations
in the system (distance protection was incorrectly disconnected from the grid during

vii
viii Foreword by Sun Guanghui

the Indian blackout that occurred on two consecutive days, “7.30” and “7.31”, in
2012).
China’s relay protection workers contributed many years of dedicated research,
practice and accumulated experience in preventing line overload caused by the
distance three false operation, which has produced a very effective technology. In the
1960s, the distance protection of the oscillation blocking concept, which indicated
that during power system oscillation, protection should be reliable non-action via the
dedicated step loss unlisting device in the predetermined location to unlist the grid.
To address the problem of AC side short circuits caused by DC side phase changes,
numerous measures have also been taken in the area of phase change failure on the DC
side due to short circuits on the AC side. Previous studies on the prevention of large
grid chain accidents did not comprehensively address this issue. In summary, the
majority of relay protection workers and related engineering and technical personnel
would benefit further in-depth research and application of relevant technology, which
is urgently needed.
This book was born out of the need to investigate power grids from the perspec-
tive of relay protection. The authors provide an in-depth analysis of the differences
between short-circuit faults and accidental overloads as well as screening methods;
an in-depth study of the differences in the characteristics of short-circuit faults and
out-of-step oscillations; an introduction to a series of distance protection princi-
ples and configuration schemes immune to overloads; and a proposal of distance
protection principles and implementation methods immune to oscillations. For the
problem of DC converter phase change failure inherent in the formed AC-DC hybrid
grid in China, methods to prevent continuous phase change failure and simultaneous
phase change failure leading to DC blocking are introduced, current DC fast control
techniques are introduced, and techniques to prevent protection misoperation under
overload are proposed. These techniques are oriented to power system safety from
the perspective of relay protection and are mechanically different from traditional
relay protection and the principle of safety and stability control system action. The
value of these techniques is to prevent relay protection from not only incorrectly
operating during chain accidents and expanding the scope of an accident but also
causing secondary disasters and to provide a reliable guarantee for the correct opera-
tion of the safety and stability control system and the adoption of the intended control
strategy.
This book is novel in intention and unique in perspective, with profound analysis
and distinct mechanisms for problems and specific and feasible methods for system-
atically solving problems, and has an important role in improving the safety and
stability systems of power systems and upgrading the level of safety and stability
operation. This book enriches relay protection theory and technology, clarifies the
boundary between relay protection and safety and stability control technology and
has positive significance in promoting the development of the three lines of defense
technology.
The author of this book has been engaged in long-term research on fault analysis
and relay protection technology of power systems. This book comprises the culmi-
nation of this research by the author and his team concerning the safety problems of
Foreword by Sun Guanghui ix

AC-DC hybrid power grids. I believe that the publication of this book will contribute
to the construction and safe and stable operation of AC-DC hybrid power grids both
in China and worldwide and to the construction of a new power system that can
adapt to the increasing proportion of new energy sources. The book has important
reference value for engineers and technicians engaged in the research and application
of relay protection and safety and stability control technologies for power systems.

Nanjing, China Sun Guanghui

June 2022
Preface

Faults are inevitable. In response to faults, researchers have proposed a series of

relay protection theories and methods based on in-depth fault analysis and devel-
oped numerous relay protection technologies, which have effectively ensured the
safety of power grids worldwide. Relay protection has become one of the most
important branches of power systems and thus an expansive industrial field. With the
promotion and application of DC transmission technology, China has built a giant
AC-DC, hybrid power grid, with the world’s highest voltage level and largest global
consumption of new energy generation. A common form of fault in the hybrid AC-
DC grid is the chain fault, especially the cross-propagation of simple faults between
AC networks and DC networks.
Unlike simple faults, interlocking faults have a large impact range and involve
many electrical devices. It is necessary to pay special attention to the special type
of “interlocking faults” in AC-DC power grids and to explore special techniques to
address them, which entails the system protection explored in this book.
System protection is different from relay protection in that it aims to guarantee
system safety under the premise of ensuring the safety of electrical equipment. The
main manifestation of system protection is to maintain the integrity of grid topology
and to ensure the balance of power generation and power consumption, thus reducing
the risk of frequency instability, power angle instability and voltage instability,
interrupting the chain of chain fault occurrence and development and effectively
preventing the occurrence of large-scale, malignant power outage accidents.
The main elements of system protection include distance protection that is
immune to accidental overloads; distance protection that is immune to oscillations;
converter control techniques to prevent long-term or continuous phase change fail-
ures; DC participation in emergency tidal control techniques for sharing accidental
transfer overloads caused by converter blocking; and adaptive overload protection
for overhead transmission lines.
System protection is dynamic in nature: it is not set up for a static section but
is a protection technology that is set up around the event that will occur after an
event. System protection is active in nature: it is a measure taken after the previous
event to prevent the next chain event. System protection also involves non-tripping

xi
xii Preface

protection: it neither removes the faulty equipment nor cuts the machine to cut the
load. Thus, system protection is distinguished from the safety and stability control
system to avoid repeated settings that may cause logical confusion and undesirable
consequences. Customarily, system protection falls between the first safety line of
relay protection and the second safety line of emergency control of a power system
and comprises 1.5 safety lines of defense of the power grid.
Dong Xinzhou participated in writing all chapters of this book; Chen Binshu
participated in Chap. 2 Immunity to Distance Protection for Overload and Chap. 6
Adaptive Overload Protection for Overhead Transmission Lines; Wang Hao partic-
ipated in writing and revising Chap. 3 Distance Protection Immune to Oscillation;
Wang Bin and Wang Haonan participated in writing and revising Chap. 4 on phase
change failure prevention and control; and Chen Zhong participated in writing and
revising Chap. 5 on DC participation in emergency current control.
This book presents the results of the National Key R&D Program “Large-scale
AC-DC Hybrid Grid Operation Control and Protection”, which was supported by
the National Natural Science Foundation of China’s major international cooperation
project “Local Information-based System Protection Research”. Chapters 6 and 2
are based on the postdoctoral report of Ding Lei and dissertations by Liu Kun, Ph.D.,
and Cao Runbin, Ph.D. Chapter 3 is supported by the research of Mr. Shengshi Zhu,
Dr. Liu Cui and Dr. Hao Wang. Chapter 4 is supported by the postdoctoral reports
of Soharb and Liuming Jing. Li Bai participated in the editing and proofreading of
the whole book.

Beijing, China Xinzhou Dong

December 2021
Contents

1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Analysis of Typical Chain Failures in Domestic
and International Power Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 2012 Southern Grid AC Failure Triggers Phase
Change Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Power Outage in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Interlocking Faults in AC-DC Hybrid Grids . . . . . . . . . . . . . . . . . . . . 21
1.2.1 Current Status and Development Trend of Domestic
and International Power Grids . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.2.2 Key Features of the AC-DC Hybrid Grid . . . . . . . . . . . . . . . . 22
1.2.3 Interlocking Faults in AC-DC Hybrid Grids . . . . . . . . . . . . . . 23
1.3 System Protection Against Interlocking Faults in Mixed
AC-DC Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.3.1 Defined Functions and Components of System
Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.3.2 Our Three Lines of Defense . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.3.3 Special Protection Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2 Distance Protection Against Overload . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.1 Analysis of the Action Behavior of the Distance III Section
Under Accidental Overload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.1 Triggering Events for Accidental Overload . . . . . . . . . . . . . . 38
2.1.2 Action Behavior of Distance III Segments During
Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2 Network Analysis of Accidental Overload . . . . . . . . . . . . . . . . . . . . . 46
2.2.1 Station Domain Accidental Overload Versus
Nonstation Domain Accidental Overload . . . . . . . . . . . . . . . . 46
2.2.2 Network Analysis of Nonstation Domain Accidental
Overload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

xiii
xiv Contents

2.3 Analysis of the Conditions for the Operation of the Distance

III Section Under Accidental Overload . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.1 Operating Condition 1: The Line is Heavily Loaded,
and the Equivalent System Power Angle is Stable
at Both Ends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.2 Operating Condition 2: The Line is Heavily Loaded,
and the Equivalent System Voltage at Both Ends is
Stable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.3.3 Adjustment Conditions: Protection Installed on Long
Lines and Large Fixed Values . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4 Identification Methods for Accidental Overload . . . . . . . . . . . . . . . . . 60
2.4.1 Division of the Incident Overload Action Domain
and Protection Action Domain . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.2 Identification of Dynamic Processes of Accidental
Overload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.5 Distance III Accidental Overload Blocking Scheme . . . . . . . . . . . . . 73
2.5.1 Description of the Lockout Program . . . . . . . . . . . . . . . . . . . . 73
2.5.2 Lockout Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.6 Solving for the Davinan Equivalent Impedance . . . . . . . . . . . . . . . . . 80
2.7 Adaptive Adjustment of Distance Protection Based on Shared
Information in the Station Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
2.7.1 Factors Affecting Distance III Segment Adjustment
and Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2.7.2 Adaptive Tuning Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.7.3 Simulation Analysis of Overload Blocking
Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.7.4 Simulation Analysis of Intra-Zone Faults and Complex
Fault Opening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
2.7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3 Immunityin Distance Protection of Oscillations . . . . . . . . . . . . . . . . . . . 109
3.1 Multiphase Compensated Distance Relays . . . . . . . . . . . . . . . . . . . . . 110
3.2 Multiphase Compensated Distance Relay Performance
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.2.1 Oscillation Without Fault Condition . . . . . . . . . . . . . . . . . . . . 111
3.2.2 Fault Conditions Without Oscillation . . . . . . . . . . . . . . . . . . . 111
3.2.3 Oscillation Accompanied by Fault Conditions . . . . . . . . . . . . 113
3.2.4 Effect of Transition Resistors on Multiphase
Compensated Distance Relays . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.3 Distance Protection from Power System Oscillations . . . . . . . . . . . . 121
3.3.1 Improved Multiphase Compensated Distance
Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.3.2 Distance Protection Based on Information from Both
Ends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Contents xv

3.4 EMTP Simulation Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

3.4.1 Simulation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3.4.2 Power System Oscillations Without Fault Conditions . . . . . . 133
3.4.3 Power System Oscillation with a Single-Phase
Ground Fault in the Zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
3.4.4 Power System Oscillation with Out-Of-Area
Single-Phase Ground Fault Conditions . . . . . . . . . . . . . . . . . . 134
3.4.5 Transition Resistance Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4 Commutation Failure Prevention and Control . . . . . . . . . . . . . . . . . . . . 141
4.1 Analysis of the DC Commutation Failure Mechanism . . . . . . . . . . . . 142
4.1.1 First Commutation Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.1.2 Continuous Commutation Failure . . . . . . . . . . . . . . . . . . . . . . 146
4.1.3 Multifed DC Commutation Failure . . . . . . . . . . . . . . . . . . . . . 158
4.2 Early Warning Measures for Commutation Failure in a Hybrid
AC/DC Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
4.2.1 Early Warning Measures for the First Commutation
Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
4.2.2 Early Warning Measures for Continuous Commutation
Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
4.2.3 Early Warning Measures for Multifeeder DC
Commutation Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid . . . . 184
4.3.1 Suppression Measures for the First Commutation
Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
4.3.2 Suppression Measures Against Successive
Commutation Failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
4.3.3 Suppression Measures for Multifeed DC Successive
Commutation Failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
5 DC Participation in Emergency Tidal Control . . . . . . . . . . . . . . . . . . . . . 207
5.1 Study of Multidimensional Coupling Mechanism
of an AC-DC Hybrid Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
5.1.1 Methodology for Evaluating the Degree
of Commutation Bus Voltage Interactions in Hybrid
DC Networks with Different Control Methods . . . . . . . . . . . . 209
5.1.2 Calculation of the Maximum Delivered Power
of HVDC Based on Equivalent Impedance . . . . . . . . . . . . . . . 213
5.1.3 Short-Circuit Ratio and Operational Evaluation
Method for Multifeed-In Operation Based
on Equivalent Impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
5.1.4 Multifeeder System Tuner Capacity Calculation
Method Based on Power Support Requirements . . . . . . . . . . 222
xvi Contents

5.2 Multi-indicator Static Security Domain for AC-DC Hybrid

Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
5.2.1 Definition and Model of a Multimetric Static Safety
Domain for AC-DC Hybrid Grids . . . . . . . . . . . . . . . . . . . . . . 224
5.2.2 A Method for Inscribing the Full-Dimensional Static
Safety Domain of Hybrid AC-DC Grids Considering
Different Control Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
5.2.3 Methodology for Inscribing Low-Dimensional Focal
Variable Safety Sections (Profiles) in the Static Safety
Domain of AC-DC Hybrid Grids . . . . . . . . . . . . . . . . . . . . . . . 234
5.2.4 Static Safety Domain Inscription Method for AC-DC
Hybrid Grids Containing Controllable Series
Capacitor Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
5.2.5 Methods for Inscribing the Decoupled Security
Domain of an AC-DC Hybrid Grid . . . . . . . . . . . . . . . . . . . . . 238
5.2.6 Evolutionary Characteristics and Impact Analysis
of the Static Safety Domain of Hybrid AC-DC Grids . . . . . . 241
5.3 Coordinated Control Objectives and Control Methods When
Multiple DC Systems Are Involved in the Rapid Control
of Tidal Currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
5.3.1 AC-DC Static Safety Domain Under Meter and Time
Characteristic DC Active Adjustment Method . . . . . . . . . . . . 254
5.3.2 Safety Correction Strategy for Hybrid AC-DC Grids
Based on Safety Distance Sensitivity . . . . . . . . . . . . . . . . . . . . 258
5.3.3 Optimal Scheduling Based on Safety-Corrected
Control in the Static Safety Domain of Hybrid AC-DC
Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
5.3.4 Preventive Correction Coordination Control Based
on Static Safety Domain for Hybrid AC-DC Grids . . . . . . . . 271
5.3.5 Fast Tidal Control Based on Decoupled Security
Domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
6 Adaptive Overload Protection for Overhead Transmission Lines . . . . 285
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
6.2 Line Emergency Current-Carrying Capacity Analysis . . . . . . . . . . . . 286
6.2.1 Mechanical Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
6.2.2 Arc Drape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
6.2.3 Fittings and Various Types of Connectors . . . . . . . . . . . . . . . . 289
6.2.4 Summary of Emergency Current-Carrying Capacity . . . . . . . 289
6.3 Line Adaptive Overload Protection Action Time . . . . . . . . . . . . . . . . 290
6.3.1 Line Temperature Calculation and Action Time
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
6.3.2 Prediction Based on the Echo State Network Method . . . . . . 294
Contents xvii

6.4 Component Scheme for Line Adaptive Overload Protection . . . . . . . 298

6.4.1 Rectification Scheme and Action Logic . . . . . . . . . . . . . . . . . 298
6.4.2 Algorithm Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
6.4.3 Applications and Calculations . . . . . . . . . . . . . . . . . . . . . . . . . 303
6.4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

Appendix: Transient Temperature Calculations . . . . . . . . . . . . . . . . . . . . . . 309

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Chapter 1
Overview

1.1 Analysis of Typical Chain Failures in Domestic

and International Power Grids

1.1.1 2012 Southern Grid AC Failure Triggers Phase Change

Failure [1]

1. Overview of failures

At 16:07 on August 11, 2012, the 500 kV Zeng (City) Sui (Guangzhou) B line tripped
on the phase C fault and was successfully reclosed. During the fault, five inverter-side
converter stations (Zhaoqing converter station, Baoan converter station, Goose City
converter station, Beijiao converter station and Suidong converter station) of the ±
500 kV DC transmission line of the Southern Power Grid experienced phase change
failure. Among them, the Chu Sui (Yunnan Chuxiong–Guangzhou Suidong) DC pole
I double valve group and pole II high end valve group each experienced one phase
change failure; the pole II low end valve group encountered three phase change
failures; and Xingan (Guizhou Xingren–Shenzhen Baoan) DC double pole, high
Zhao (Guizhou Gao Pp–Guangzhou Zhaoqing) DC double pole, Tian Guang DC
(Tianshengqiao, Guizhou—Beijiao converter station, Guangzhou) and Jiangcheng
(Jiangling, Hubei–Goose City, Guangdong) DC double poles each experienced one
phase change failure, as shown in Fig. 1.1 [2].
After the AC fault was removed, the power and bus voltages of all the major
sections of the 5-back DC and west–east transmission were restored to the prefault
operation level, and the system resumed smooth operation.
2. Failure history
Before the fault, the 500 kV main network was fully wired, and the system frequency
and voltage operated at normal levels. The margin of each feeder section met the
requirements. The Yunnan AC feeder channel and Guangdong AC entrance were
operating at the average voltage limit. The main network feeder section power and
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 1
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_1
2 1 Overview

Fig. 1.1 2012 Southern power grid wiring diagram (DC shown in blue)

limit values are shown in Table 1.1. Five DCs, with the exception of Xing’an DC
(rated power of 3000 MW) with 2800 MW power operation, were operating at full
load. The AC voltages of the inverter side converter stations were operating at normal
levels. The operating power and inverter side voltages of each DC are shown in Table
1.2.
On August 11, 2012, at 16:07, the 500 kV Zeng Sui B line tripped with a phase C
fault and was successfully reclosed. During the fault, phase change failure occurred
on all five DC inverter sides of the Southern Power Grid. The power in the west–
east AC Yunnan section and each major AC section in Guangdong during the phase
change failure, is shown in Table 1.3. After the fault, the tide of both the outgoing

Table 1.1 Power and limit values for each feeder section before the fault, MW
Categories Guangdong AC entrance Guizhou AC exit Yunnan AC exit
Limit value 8000 3500 4000
Power 7430 1650 3300
Margin 570 1150 700

Table 1.2 Power and inverter-side voltage for each DC operation before the fault
Categories Chusui DC Gaozhao DC Xing’an DC Tianguang DC Jiangcheng DC
Bipolar 5000 300 2800 1800 2800
power(MW)
Inverter-side AC 531 533 536 229 534
voltage(kV)
1.1 Analysis of Typical Chain Failures in Domestic … 3

Table 1.3 Power of each major AC section in Yunnan and Guangdong MW

Categories Guangdong AC Guizhou 500 kV AC Yunnan 500 kV AC exit
entrance exit
Limit value 8000 3500 4000
Prefailure power 7430 1650 3300
Postfailure power 9114 2124 4619
Margin −1114 1376 −619

500 kV section in Yunnan and the 500 kV AC inlet section in Guangdong significantly
increased and crossed the limit for a short period, with a maximum crossing limit of
619 MW in Yunnan and a maximum crossing limit of 1114 MW in both sections.
During the oscillation, the time of crossing in Yunnan was less than 2 s at the outgoing
section and less than 3 s at the Guangdong AC inlet.
During the DC phase change failure, the voltages of all major stations in the west–
east transmission channel were reduced to different degrees, as shown in Table 1.4,
among which the transient voltage drop at the 500 kV Xianling Shan substation was
the largest, falling to 456 kV, but recovering to 516 kV after 60 ms. The transient
voltage drop at all other points exhibited even shorter values. After the transient
process, the voltage at each point can be quickly recovered, and the steady-state
voltage can be restored to the prefault voltage level.
3. Causes of failure
An inherent characteristic of conventional DC transmission systems is that they are
prone to phase change failure when a grid fault causes a sudden drop in grid voltage.
To meet its large-capacity needs, long-distance transmission, multiple DCs are
intensively fed to the PRD region of Guangdong. The electrical distance between
two DC inverter stations is short, and in the event of a fault in the adjacent AC
system, the AC voltage of each DC inverter station significantly drops, resulting in
the simultaneous phase change failure of multiple DCs.
During the DC phase change failure, the DC power significantly drops and is
transferred to the AC line in parallel with it, causing large fluctuations in system
currents and voltages.
4. Conclusions and insights

Table 1.4 Voltage changes at selected plant stations before and after DC phase change failure kV
Categories Chongzuo Luoping Hezhou Xianling Shan
Prefault voltage 523 535 537 525
Transient voltage 501 499 489 456
Transient voltage dip 22 36 58 69
Steady-state voltage 523 535 537 525
Steady-state voltage dip 0 0 0 0
4 1 Overview

On August 11, a single-phase fault occurred on the Zeng Sui B line with an AC fault
excision time of 60 ms, resulting in a phase change failure on all five DCs of the
Southern Power Grid. The following conclusions and insights are noted:
(1) Multiple DC phase change failures occurred in this fault. After the AC fault was
normally removed, each DC power was quickly restored to the prefault power
level, and the system maintained stable operation.
(2) For multiple DC dropout areas, attention needs to be given to the problem
of simultaneous multiple DC phase change failure. The risk of multiple DC
phase change failure should be reduced during the planning stage by reasonably
controlling the scale of DC transmission and by optimizing DC dropout points.
Safety and stability analysis of multiple DC phase change failure problems
should be carried out during the operation stage, and necessary operational
measures should be taken.

1.1.2 Power Outage in Brazil

1. Overview of the Brazilian grid blackout

The transmission pattern of Brazil’s power grid is similar to that of China’s power
grid, with energy and load distributed in the reverse direction [3]. The overall trans-
mission pattern shows a long-distance, high-capacity “north–south, west–east” form.
The Brazilian power grid can be divided into four major grids: the northwestern grid,
northern grid, northeastern grid and southern grid. The southern grid can be divided
into three parts: the central region, western region and southern and southeastern
region. The northwestern grid is an independent grid, while the northern, northeastern
and southern grids are coupled to form a large-scale, regional, coupled grid.
On March 21, 2018, at 15:48 (local time in Brazil), a large power outage occurred
in Brazil. The incident originated from the overcurrent protection operation of a
segmental circuit breaker of the AC system, which caused the shutdown of the
bipolar blocking of the first phase of Mount Meliá DC in Brazil. The shutdown
caused an instantaneous power loss of 3.73 million kW and eventually led to the
successive uncoupling of the northern and northeastern Brazilian grids to form an
isolated network. Nine states located in southern Brazil were also affected to varying
degrees. The regional coupled transmission pattern in normal operation before the
fault is shown in Fig. 1.2.
The outage originated at the Xingu converter station of the Miryama hydroelectric
transmission system in northern Brazil. The Miryama hydropower plant transmits
power to the southern grid through the Miryama I ± 800 kV DC line. The northern
grid is connected to the northeastern grid, and the northeastern grid is connected to
the southeastern grid by a 500 kV AC line, which collectively form a large hybrid
AC-DC grid.
Before the accident, a total of seven units of the Miryama hydroelectric power plant
were operational, and the Miryama Phase I DC line was operating at a normal level,
1.1 Analysis of Typical Chain Failures in Domestic … 5

Fig. 1.2 Power generation,

load demand and power
flows in the Brazilian
regional grids

delivering 3.73 million kW of power to customers. The electrical wiring diagram

of the Hingu converter station, whose AC system was still under construction, is
shown in Fig. 1.3. To meet the power output of the Miryama hydroelectric power
plant and to put the Miryama Phase I DC project into operation at an earlier stage,
the Brazilian Electricity Regulatory Commission (ANEEL), in consultation with the
project company, proposed a construction transition period of single bus operation
at Hinku station, which was divided into 3 phases [4, 5].
2. Incident history
The course of events in a major power outage is described as follows [3,4]:
(1) Tripped circuit breaker at Hinku converter station, DC double pole blocking
Prior to the fault, Miryama DC Phase I wasIn the first phase, Bus B was operated
without a circuit breaker to ensure commissioning and operation of the double pole;
in the second phase, Bus A was operated with a circuit breaker, and Bus B was
out of service; and in the third phase, Bus B was operated with a circuit breaker.
The incident occurred during the second phase of the transition period. Since circuit

Fig. 1.3 Electrical wiring

diagram of Shingo Station
[8]
6 1 Overview

breakers were not utilized in the first stage, the relevant departments did not set
the rectification value of the sectional circuit breaker protection in time after the
conversion to the second stage, and the rectification value of the sectional circuit
breaker 9522 overcurrent protection was the factory-set value of 4000 A [9].
Conducting a full power test, and the line load current reached 4400 A, exceeding
the overcurrent protection preset threshold (4000 A). At 15:48:03.245 (0 ms after the
fault), segment circuit breaker 9522 tripped due to overcurrent protection, resulting
in a loss of voltage at bus A and a bipolar blocking of the Miryama DC Phase I line.
(2) The stabilization device did not operate in a timely manner; the DC tide shifted
to AC; and the northern and northeastern grids were consecutively unlisted.
After a DC lockout occurs, if the generating station unit is removed in a timely
manner, the output power at the sending end can be quickly reduced to prevent large-
scale transfer of DC transmission power to the AC line. However, because the stability
device (SEP) did not consider the extreme case of simultaneous loss of voltage at
both AC buses (reflected in this accident as a loss of voltage with single bus operation
during the transition period), the stability device (SEP) decided that the signal was
invalid after the DC blocking sent a signal to the stability device to remove six units
and failed to act in time to cut the units. The seven units of the Beautiful Mountain
Hydropower Station continued to operate with the same output power, so the DC
current was transferred to the AC line. As shown in Fig. 1.2, after the blocking of
Mirishan Phase I DC line ➀ between the northern grid and the southern grid, all 3.73
million kW of power carried on it was transferred to the parallel AC corridor ➁ ➂
➃.
After the tide shifts to AC, the system oscillates and overloads. Various AC lines
consecutively tripped. Approximately 738–1134 ms after the fault occurred, the
loss-of-step protection and distance protection on AC line ➁ between the north and
the south tripped in many places, and the northern grid was disconnected from the
southern grid. Subsequently, the out-of-step and distance protection devices on AC
line ➂ between the north and the northeast operated and disconnected the lines,
causing the northern grid to form an isolated network. At this point, the northern
grid is severely overpowered and completely disintegrates after a certain period. The
distance protection failure at two locations on line ➃ between the northeast and the
south disconnected the grid and isolated the northeast.
(3) Disintegration of the northern and northeastern grids and restoration of stability
in the southern grid
The northern grid was unlisted with excess power; the frequency rose to a maximum
of 71.69 Hz; some lines tripped due to overvoltage; the system oscillated after the
high-frequency cutter action removed some units; and the northern grid disintegrated
85 s after the fault occurred, losing approximately 93% of the load.
After the power shortage in the northeastern grid was lifted, the frequency was
restored to near the rated value after five rounds of low-frequency load shedding
action. However, after approximately 10 s of stable system operation, the grid
1.1 Analysis of Typical Chain Failures in Domestic … 7

frequency dropped again due to the tripping of two units of the Paulo Afonso hydro-
electric power station and certain thermal power station units due to inappropriate
action of their own protection, which eventually caused the disintegration of the
northeastern grid and the loss of approximately 99% of the load.
The system returned to stable operation after the power shortage and frequency
reduction after the southern grid was unlisted, with the frequency reaching a low of
58.44 Hz, and after a total of 3.67 million kW of load was removed after the first
round of low-frequency load shedding action.
3. In-depth analysis of the causes of power outages in Brazil
The original cause of the blackout was the blocking of the Mei Shan Phase I DC
line, which was caused by the tripping of the sectional circuit breaker. However, DC
blocking alone would not have caused the massive blackout. The root cause of the
blackout was the subsequent chain failures attributed to DC blocking. An analysis of
the chain failures caused by the DC lockout in this accident is presented as follows:
(1) DC blocking causes tidal shift
DC transmission systems are generally required to operate under some kind of power
command, i.e., the power of the DC transmission system remains essentially constant
during normal operation. After a DC or AC line fault, the DC converter station will
be blocked out of operation to protect the safety of the DC converter station and the
line.
For this incident, since only one ± 800 kV DC line, Mirador I, is currently in
operation in the Brazilian grid, the entire 3.73 million kW of power carried on the
Mirador I DC line was transferred to the parallel AC line after the DC lockout, i.e.,
triggering a massive tidal shift.
For the AC-DC hybrid grid, where multiple DC lines exist, due to the nature of
the DC transmission system control system, the power transmitted by the other DC
lines is basically unchanged, and all the power originally transmitted on the blocked
DC lines will also be transferred to the parallel AC lines, making the AC lines carry
a much larger load than they would in normal operation.
(2) Tidal shift triggers chain tripping
When all the power transmitted on a DC transmission line is transferred to a parallel
AC line, an overload phenomenon may occur on the AC line and is often accompanied
by system oscillations. The overload and oscillation phenomena triggered by tidal
current transfer are analyzed here.
(1) Tidal shift triggers line overload
With the existing protection configuration, the relay protection devices on the AC line
cannot distinguish between fault overload and tidal shift overload, and a false-action
chain trip will occur, eventually leading to the unlisting of the sending and receiving
ends in the network structure. Compared with the small probability event of chain
tripping triggered by accidental overload in a conventional AC system, the possibility
of chain tripping triggered by severe overload on the AC line due to DC blocking is
8 1 Overview

Fig. 1.4 Impedance relay

directional circle
characteristics

extremely high in a hybrid AC-DC network. A simple description is provided here.

The directional characteristic impedance circle shown in Fig. 1.4 is selected as an
example.
In the figure, Z set is the distance protection section III adjustment value and ϕ L is
the load impedance angle. The actual power delivered by the AC line is P, the rated
voltage is U N , and the measured impedance amplitude is
√ 2
3U N cos ϕ L
ZM = (1.1)
P
As shown in Fig. 1.4, the minimum measured impedance value that will not cause
the action of the distance protection section III is

Z M min = Z set cos(ϕset − ϕ L ) (1.2)

When meeting

Z M < Z M min (1.3)

The protection operates when the condition of the,

Substituting (1.1) and (1.2) into (1.3), we obtain,
√
3U N2 cos ϕ L
< Z set cos(ϕset − ϕ L ) (1.4)
P
Collation (1.4) has
√ 2
3U N
P1 = cos ϕ L < P (1.5)
Z set cos(ϕset − ϕ L )

From (1.5), distance protection III operates when the AC delivered power P is
greater than P1 . The value of Z set can be expressed as
1.1 Analysis of Typical Chain Failures in Domestic … 9

K r el
Z set = Z L min (1.6)
K ss K r e cos(ϕset − ϕ L )

where K r el is the reliability factor, K ss is the motor self-start factor, K r e is the

protection return factor, and Z L min is the load impedance at minimum load. Take the
typical values of the three coefficients and substitute (1.6) into (1.5) with
√
3U N2
P1 = 2.07 cos ϕ L < P (1.7)
Z L min
√
where 3U N2 cos ϕ L /Z L min is the maximum transmitted power during normal
operation, denoted as Pmax ; then, (1.7) can be written as

P1 = 2.07Pmax < P (1.8)

The above derivation shows that when the transmission power of the AC line is
greater than 2.07 times the maximum transmission power, distance protection III of
the AC line is highly likely to operate due to overload.
(2) Tidal shift triggers system oscillations
Tidal shifting can also cause the power delivered by the AC transmission line to
exceed the static stability limit, which triggers system oscillations. When the system
oscillates, the amplitude and phase of voltage, current and power at various points
periodically change, causing the current and impedance relays of the protection
devices to misactivate [6].
The directional characteristic impedance circle shown in Fig. 1.5 is selected as an
example.
The figure at Z K is the measured impedance of the relay. When the system oscil-
lates, the measured impedance of the relay will move along the vertical bisector of
the total line impedance OO . Let the intersection of the relay action characteristic
with line OO be O and O . Then the measured impedance of the relay lies within the
action characteristic circle between these two intersection points, i.e., when the end

Fig. 1.5 Effect of

oscillations on relays
10 1 Overview

point of the measured impedance lies on the intersection point, the distance protec-
tion will be affected by the oscillation and produce a false operation. Tidal shifting
causes the end point of the measuring impedance Z K to move from right to left along
OO into the action characteristic circle, causing the protection to misactivate.
In addition, the effect of system oscillations on the distance protection is also
related to the protection installation location. The closer the installation point of the
protection is to the center of the oscillation, the greater the oscillation is affected; if
the center of the oscillation is outside the protection range or in the opposite direction
of the protection, the distance protection is not affected by the oscillation, and a false
operation occurs [7].
Our relay protection devices achieve good oscillation blocking and fault
unblocking during oscillation [12]. However, the distance protection of the Brazilian
grid does not have an effective oscillation blocking function. For this incident, the
Miryama Phase I DC line (1) originally delivered approximately 3.73 million kW,
the parallel AC line (2) delivered approximately 0.77 million kW, and the AC line
(3) delivered approximately 2.88 million kW. After the tidal shift, the AC line took
on a load that far exceeded its normal delivered power, and the system oscillated.
Under the dual action of overload and oscillation, the AC line triggers a protection
misoperation chain trip.
(3) AC interlocking trips impact the grid at the receiving end
Tripping of AC transmission lines can cause shocks to the grid at the sending and
receiving ends, leading to oscillations in the system at the sending and receiving ends
and great fluctuations in frequency, most likely generating frequency crossing limits.
For the sending system, the DC blocking interrupts the delivered power, the
sending system has excess active power, and the frequency rises. If the system’s
own primary frequency regulation can ensure that the system frequency is within
the rated range, no subsequent action is needed. If the primary frequency regula-
tion cannot ensure that the frequency is within the rated range, it is necessary to
cut the machine to reduce the power. In this incident, with the northern grid as the
sending system, the system frequency rose rapidly after the unlisting, a large oscil-
lation occurred after the high-frequency cutting action removed some units, and the
grid was completely shut down 85 s after the fault occurred.
For the receiving system, the DC blocking interrupts the delivered power, the
receiving system has an active power deficit and the frequency is reduced. If the
system frequency has been reduced outside the rated range, to prevent system
frequency collapse, the automatic under frequency load shedding device in the system
(Under frequency Load Shedding, UFLS) will automatically operate to remove part
of the load in a graded manner (in rounds), according to the high or low frequency,
so that the frequency can be restored to the rated range as soon as possible. After the
system at the sending and receiving ends is regulated and stabilized, if the protection
in the system unreasonably operates when it only focuses on its own protection object,
it may cause the system to collapse. In this incident, the frequency was reduced after
the southern grid was unlisted but was restored to the rated range after the partial load
was removed by low-frequency load shedding. In contrast, after the northeastern grid
1.1 Analysis of Typical Chain Failures in Domestic … 11

was unlisted, the low-frequency load shedding action was restored to stability within
a short period after five rounds, but the low-frequency load shedding strategy was
not adequately configured, and the unit protection in the system mistakenly acted to
cut the machine, leading to the final collapse.
In summary, the cause of the blackout in Mirador, Brazil, lies in the three compo-
nents of the AC-DC interlocking fault: DC blocking, tidal shifting and interlocking
tripping. The loss of voltage on the rectifier side bus of the northern grid triggered DC
blocking, while the failure of the stabilization device to cut the machine caused the
tidal shift. To correctly identify overloads and oscillation blocking, the failure of the
relay protection device caused the AC line chain to trip. The chain reaction among
the various links led to the eventual system collapse and major outage. In addition,
after the system was unlisted, the low-frequency load shedding configuration was not
in place and became the “last straw” that crushed the temporarily stable northeastern
grid.
4. Simulation studies
To verify the chain fault process in the AC-DC hybrid grid, a simple AC-DC hybrid
grid model, as shown in Fig. 1.5, is built in PSCAD and simulated accordingly.
As shown in Fig. 1.6, the system consists of three grids, each with a bus voltage
level of 500 kV, a DC transmission line voltage level of ± 800 kV, and a system
frequency of 50 Hz. Grid 1 is connected to grid 2 via DC transmission and AC
transmission Lines 8–9. The DC transmission power is 3000 MW. Grid 1 is connected
to grid 3 via AC transmission Lines 1–11. Grid 2 is connected to the grid via AC
transmission Lines 10–15. The tidal state and nature of AC and DC transmission of
the three grids in the simulation platform are similar to the operation of the Brazilian
grid, and grid cutting and load cutting are not considered at this time.
Figure 1.7 shows the power variation of AC Lines 8–9 and Lines 1–11 after DC
blocking. Figures 1.8, 1.9 and 1.10 show the frequency fluctuation of Grids 1–3 after
DC blocking. Figure 1.11 shows the voltage variation of Bus 11 of Grid 3 after DC
blocking.
Figure 1.7 shows that during normal operation, AC transmission Lines 1–11
transmit approximately 2900 MW, Lines 8–9 transmit approximately 900 MW and
Lines 10–15 transmit approximately 750 MW.
At 5 s, the DC line connecting grid 1 to grid 2 is blocked, and the transmitted
power of AC Lines 8–9, which operate in parallel with the DC line, rapidly increases.
The transmitted power from grid 2 to grid 3 simultaneously decreases (Lines 10–15),
and the transmitted power from grid 1 to grid 3 accordingly increases (Lines 1–11)
due to DC line blocking. Grid 1 and Grid 3 have excess power and increases in
frequency, while Grid 2 decreases in frequency due to a power deficit.
At 5.5 s, Lines 8–9 trips due to overload distance protection in section III. All
channels for direct power transmission from grid 1 to grid 2 are disconnected; grid 1
transmits power to grid 2 via grid 3; Lines 1–11 power further increases; Lines 10–15
power appears to backfeed and significantly increases; and grid 3 supplies power to
grid 2. Grid 1 and Grid 3 have excess power and increase in frequency; Grid 2 loses
power input from Grid 1 and decreases in frequency.
12 1 Overview

Fig. 1.6 Simple AC–DC

hybrid model

Fig. 1.7 Variation in power

transmitted by AC lines
1.1 Analysis of Typical Chain Failures in Domestic … 13

Fig. 1.8 Grid 1 frequency

variation

Fig. 1.9 Grid 2 frequency

variation

Fig. 1.10 Grid 3 frequency

variation
14 1 Overview

Fig. 1.11 Grid 3 bus11

voltage variation at bus 3

At 6 s, Lines 10–15 trips due to overload distance protection section III, and Grid
1 is declassified as an isolated network operation, whose frequency continues to rise
when internal generation within the grid is greater than the load. Grid 2 has a power
shortage, and its frequency decreases. At this point, the tidal direction resumes for
Grid 2 to supply Grid 3, and Grid 3 rises in frequency after a sudden drop.
At 6.3 s, Lines 1–11 trip due to overload and oscillation, Grid 2 continues to
drop in frequency due to power shortage without cutting load, Grid 3 operates alone,
reactive power cannot be supported and the system collapses. The voltage variation
in Bus 11 on Grid 3 bus is shown in Fig. 1.10.
The simulation results characterize the occurrence of chain faults in the hybrid
AC-DC system. DC blocking, tidal shifting and chain tripping, as the three major
links of the hybrid AC-DC grid, are progressively and sequentially affected. The
DC lockout is caused by a single fault on the inverter or rectifier side of the DC
transmission line. However, due to the chain effect, the whole system is affected,
which causes a large-scale tidal shift of the overall grid, the AC transmission lines
between local grids are chain tripped due to overload and oscillation, and the local
grids are subjected to frequency shock, which eventually leads to the total collapse
of the large grid.
5. Analysis and insights into chain failures from the Brazilian blackout
Blackout is usually caused by chance events; however, behind these chance events,
there are often many fundamental flaws [6], The blackout in Brazil was essentially a
chain of failures caused by the control system characteristics of the DC transmission
system and the unreasonable actions of the protection and stabilization devices in
the context of the “strong direct and weak transmission” of the Brazilian grid.
(1) “Strong direct and weak AC” hybrid AC-DC grid structure
The development of AC-DC hybrid grids roughly undergoes three stages: the small-
capacity DC-AC hybrid stage (strong AC and weak straight), large capacity DC-AC
hybrid stage (strong straight and weak AC), and comparable capacity DC-AC hybrid
stage (AC and straight strong and weak) [8].
1.1 Analysis of Typical Chain Failures in Domestic … 15

In the “strong direct and weak AC” stage, the transmission capacity of a single
DC is much larger than that of a parallel single AC, and the number of interoperable
DC transmission channels is also greater than that of AC transmission channels. As
the transmission power of DC lines is greater than that of AC lines, once a single
DC or several DC lines are blocked, a large-scale tidal shift is imminent. The AC
grid cannot withstand large power fluctuations, and the distance protection cannot
correctly distinguish faults and overloads, resulting in incorrect actions and unlisting
of the system, which produces a series of chain reactions that cause the final blackout
incident.
(2) Unreasonable coordination of relay protection and safety devices in the context
of AC-DC hybrid connections
Despite the expanded scale of UHV DC transmission and the rapid development of
hybrid AC-DC grids at this stage, the collaborative methods of relay protection and
safety devices has not been updated.
Traditional relay protection focuses on the protection object itself and does not
consider the whole grid system; such a mode of work has a negative side on the
operation of the AC-DC hybrid grid. In this case, the transmission channel parallel to
the DC transmission system should be kept running for as long as possible to provide
more time for the stabilization device to operate. However, the line protection often
operates under overload or oscillation, causing another shock to the system at the
sending and receiving ends and aggravating the impact of the accident.
In addition, this renewed frequency drop in the Northeast Brazilian grid after
stabilization was also triggered by the generator’s own protection acting again after
the successful regulation of the stabilization device.
The “strong direct and weak AC”, AC-DC hybrid grid structure and unreasonable
operation of protection and stabilization devices jointly cause chain failures.
The Great Indian Blackout
Two major blackouts occurred in India on 30 July and 31 July local time. Both
blackouts were caused by overload trips on the same 400 kV transmission line, which
triggered a chain of trips that resulted in widespread blackouts. The 30 July incident
involved a major blackout of the northern grid with a power deficit of approximately
36,000 MW, covering nine states in northern India, including the capital city of
New Delhi. The July 31 incident involved a major blackout of the northern, eastern
and northeastern grids, which affected 20 states and reduced the power supply by
approximately 48,000 MW, affecting more than 670 million people and creating the
most widespread disaster in history [9,10,11].
1. Basic information on the Indian electricity grid
The Indian grid consists of the national grid (consisting of the interregional grid and
five regional grids in the North, West, South, East and Northeast across states) and
29 state grids, which are subordinate to the central government. The regional grids
have 400 kV as the main grid and are interconnected by 765 kV (actually operating
at 400 kV at reduced voltage), 400 kV, 220 kV AC and ± 500 kV DC lines. The four
16 1 Overview

Fig. 1.12 Schematic of the

regional grid in India

regional grids in the north, west, east and northeast are synchronously interconnected
using a hybrid AC/DC connection to form the NEW grid and are asynchronously
interconnected with the southern grid via DC, as shown in Fig. 1.12 [9].
With a total installed power generation capacity of approximately 200 GW, India
has the fifth largest power generation capacity in the world; however, the per capita
electricity consumption is grossly inadequate, and power restrictions are frequent
everywhere [10].
2. “7.30” power outage
India has a tropical monsoon climate. Prior to the power outage, the delayed monsoon
triggered drought and heat, resulting in a sharp increase in electricity load in the north
due to agricultural irrigation and air conditioning use. Additionally, the delayed
monsoon meant that hydropower generation was lower than usual.
Before the “7.30” accident, the northern grid generated 32,636 MW and demanded
38,322 MW, leaving a power gap of 5,686 MW (14.8%), and the system operated
at 49.68 Hz. The northern and western grids are connected by two 400 kV liaison
lines, and the northern and eastern grids are connected by six 400 kV liaisons and
one 220 kV AC liaison lines. The regional intergrid liaison line operated by the NEW
grid prior to the incident is shown in Fig. 1.13 [1,9]
The accident unfolded as follows [9]:
(1) At 02:33:11.907, the 400 kV contact line Bina-Gwalior I of the northern and
western grid was tripped due to overload, causing the distance protection section
III to operate. The natural power of the Bina-Gwalior I line is 691 MW (uncom-
pensated). Before the accident, the line was delivering 1450 MW to the northern
1.1 Analysis of Typical Chain Failures in Domestic … 17

Fig. 1.13 NEW regional grid contact line before the “7.30” outage

grid and was in an overload condition, and the voltage on the Bina side had
dropped to 374 kV.
With the disconnection of the Bina-Gwalior I line, only one AC link remains
between the western grid and the northern grid: the Zerda (400 kV)-Bhinmal
(400 kV)-Bhinmal (220 kV)-Sanchore (220 kV) and Dhaurimanna (220 kV)
lines, with the tidal flow shifted from the western-northern section to this
liaison line. The power angle difference between the northern grid and the
western-eastern-northeastern grid increases, and subsequently, the system starts
to oscillate.
(2) At 02:33:13.438, the 220 kV Bhinmal-Sanchore link between the northern
section and the western section was tripped due to system oscillations resulting
in distance protection Section I. Subsequently, another 220 kV Bhinmal- Dhau-
rimanna link was also removed by distance protection Section I due to the
same oscillations. At this point, the northern grid lost all AC links to the
western grid. After the disconnection of the western-northern grid link, the
cross-sectional currents were transferred and sent to the northern grid through
the western-eastern-northern path, creating a massive tidal shift.
(3) At 02:33:13.927, an important channel for the west-east-north tidal shift-400 kV
Jamshedpur-Rourkela double return line located within the eastern grid tripped
due to overload, resulting in phase action of distance protection section III.
At this point, although the northern grid is still connected to the eastern grid,
the generator speed in the network drops, and the power angle difference with
the western-eastern-northeastern grid further increases, subsequently leading to
power angle instability (loss of synchronization).
(4) 02:33:15.400–02:33:15.542, six 400 kV contact lines between the northern grid
and the eastern grid (Gorakhpur-Muzaffarpur double, Balia-Biharsharif double
and Patna-Balia double) were tripped due to system oscillations, resulting in
distance protection. The center of oscillation is located in the northeast grid
section.
18 1 Overview

The entire 400 kV AC link between the northern grid and the eastern grid was
removed. The Sahupuri load, which was part of the northern grid, was integrated
into the eastern grid through the 220 kV Pasauli-Sahupuri line. The northern
grid was unlinked from the western-eastern-northeastern grid.
(5) The unlisted northern grid experienced a power deficit of 5800 MW and a
sudden drop in frequency. Due to inadequate load shedding by emergency
control measures (low-frequency load shedding and df/dt slip shedding), the
northern grid collapsed, and only a few areas of Badarpur and NAPS were
left to maintain islanding. The western-eastern-northeastern grid experienced a
5800 MW power surplus, and the frequency rose to 50.92 Hz, which stabilized
at 50.6 Hz after removal of 3340 MW units by a special protection system.
3. “7.31” power outage
After the restoration of the “7.30” outage, the power demand in the northern grid
remained tight. Before the “7.31” incident, the northern grid generated 29,884 MW
and demanded 33,945 MW, with a power gap of 4,061 MW (12.0%), and the system
operated at 49.84 Hz. The northern and western grids are connected by one 400 kV
liaison line and two 220 kV liaison lines, and the northern and eastern grids are
connected by one 765 kV AC liaison line, nine 400 kV AC liaison lines and one
220 kV AC liaison line. The western and eastern grids are connected through six
400 kV liaison lines and three 220 kV liaison lines. The regional intergrid links
operating on the NEW grid prior to the incident are shown in Fig. 1.14 [9, 10].
The accident unfolded as follows [9]:
(1) At 13:00:13, the 400 kV liaison line Bina-Gwalior I of the northern and western
grids tripped due to overload causing the distance protection section III to
operate. Subsequently, the 220 kV Bina-Gwalior double circuit was discon-
nected, and the Gwalior area was disconnected from the western grid. Thus,
the northern grid was unlinked from the western grid, the cross-sectional tide

Fig. 1.14 NEW regional grid contact line before the “7.31” outage
1.1 Analysis of Typical Chain Failures in Domestic … 19

was sent to the northern grid through the western-eastern-northern path, and the
northern grid was destabilized from the western grid at the angle of merit.
(2) At 13:00:13.600, the 400 kV Jamshedpur-Rourkela I line, an important tide-
shifting channel within the eastern grid, tripped due to overload resulting in
protection action. Before the trip, the line voltage was approximately 362 kV,
the line current was 1.98 kA and the apparent power was approximately 1241
MVA.
The system begins to oscillate.
(3) At 13:00:17.948–13:00:20:017, multiple 400 kV lines within the eastern grid
tripped due to protection action from oscillations. The center of the system oscil-
lation is located within the eastern grid (near the western-eastern grid section).
Areas such as Ranchi and Rourkela, which were part of the eastern grid, were
incorporated into the western grid, and the eastern grid was unlinked from the
western grid.
After unlisting, the frequency of the western grid rises to 51.4 Hz, and the
final frequency of the western grid is stabilized at 51 Hz by cutting measures
and boosting the DC power sent to the southern grid. A power deficit of approxi-
mately 3000 MW occurs in the northeastern-northeastern grid, and the frequency
drops to 48.12 Hz due to insufficient load cutting and system power angle
oscillations caused by unit tripping.
(4) Subsequently, more than 50 lines tripped within the northern and eastern grids
and on the northeastern grid liaison due to distance section III protection, over-
voltage protection, and out-of-step protection operations, unlinking the northern
grid from the eastern-northeastern grid.
The northern grid and eastern-northeastern grid collapsed in most areas, with the
exception of a few areas of isolated operation, causing another major power outage.
4. Implications of the Indian Blackout for Relay Protection
Inappropriate relay protection was the trigger for the Indian blackouts of July 30
and July 31, which directly caused chain tripping and contributed to the development
of the blackouts. The two blackouts revealed obvious deficiencies in the Indian relay
protection distance protection (especially distance section III): overload tripping and
lack of oscillation blocking.
Both outages were caused by the tripping of the distance protection section III of
the 400 kV Bina-Gwalior I line on the same western-northern transmission section.
The natural power of the Bina-Gwalior I line was 691 MW (uncompensated), and
before the “7.30” outage, the line delivered 1450 MW to the northern grid and was
heavily loaded, with a voltage drop of 374 kV on the Bina side. Before the “7.31”
incident, the line carried 1254 MVA of apparent power, and the voltage on the Bina
side was only 362 kV. Due to the “unreasonable” principle and “inappropriate”
adjustment value, the distance protection section III inappropriately operated, causing
tripping and then triggering a chain fault.
Distance protection in India, North America and Western Europe does not have
oscillation blocking. The two Indian blackouts and numerous foreign blackout, were
20 1 Overview

not in the grid oscillation, allowing the line protection to jump off the line and rapid
splitting of the grid, resulting in a large area blackout. Compared with the domestic
“7.1” Central China blackout and “7.30” “7.31” India blackout, it is not difficult
to identify the oscillation blocking function to prevent widespread blackouts from
having a valuable role.
In addition, these two major blackouts in India and several other major blackouts
worldwide have exposed deep flaws in existing relay protection guidelines.
(1) Existing relay protection is oriented toward individual electrical devices and
does not have a global perspective to protect the safety of the power system
Large outages are mostly caused by interlocking faults [6]. A load shift will occur after
electrical equipment is removed, which will easily cause overload of other nonfaulty
electrical equipment in the transmission section. If the backup protection of nonfaulty
electrical equipment (especially the remote backup protection, such as distance III) is
tripped, it will trigger a new round of tidal shift, which will spark a chain of overload
faults and eventually a large area of network paralysis and large-scale power outage.
In the above process, the protection action complies with the existing relay protec-
tion guidelines and protects the safety of electrical equipment but causes system shut-
down and extensive power outages. This action contradicts the concept of protecting
the safe operation of the power system because existing relay protection guidelines
are aimed at protecting the safety of electrical equipment not that of the power system.
(2) Fast-acting relay protection cannot be matched with relatively slow safety and
stability control
The safety and stability control response time ranges from a few hundred millisec-
onds to a few seconds or even longer and cannot be matched with fast-acting relay
protection. There is a blind spot between the two: after an overload of nonfaulty
electrical equipment (abnormal operation), the backup protection acts quickly to
remove it, while the safety stabilization control is its subsequent action. After a load
transfer due to an accident, the removal of the overloaded line or electrical equipment
can only aggravate the overload of other safe electrical equipment, which impedes
system safety.
(3) Existing relay protection lacks effective countermeasures for overload,
including oscillation conditions
A study of power system blackout accidents based on complex network theory
revealed that the grid has self-organizing characteristics and that the load is similar
to sand in a sandpile model. Grid expansion enables a reduction in the system oper-
ating margin and an increased risk of large blackout accidents. Past experience also
confirms that transmission line overloads often cause major outages and are the root
cause of accident development.
Prior to the “30 July” blackout in India, four states in the northern grid experienced
a sharp spike in load, and on 31 July, when the eastern and northern regional grids were
reconnected, four states in the northern grid were overloaded beyond the dispatch
schedule [9]. Under normal circumstances, overloads can be eliminated through
1.2 Interlocking Faults in AC-DC Hybrid Grids 21

dispatch and stability control systems. However, major outages never occur under
normal circumstances, and in the event of a failure or oversight of dispatch and
stability control systems, as in the US-Canada and Italy outages on 8.14 and 9.28,
lines can experience prolonged overloads. Negligence, overloading the line for an
extended period, which causes wire heating, and an increase in arc sag [11, 12],
which aggravates the probability of short circuits, also increase the safety risk of
the system. China’s distance protection has an oscillation blocking function (many
foreign manufacturers of relays do not have this function), so in the face of system
oscillation, the probability of protection misoperation sharply increases. Even in
China, with the deepening of AC-DC coupling, the unlocking conditions of distance
protection oscillation blocking are very easy to meet, rendering oscillation blocking
virtually useless.
In addition, the existing stability control system may have a “strategy meter
mismatch” in the face of complex and unpredictable blackout situations, which may
prevent effective measures or delay the best time to act, leading to further expansion
of the accident. In response to the above problems, the existing relay protection can
only passively wait or fail to implement effective countermeasures.

1.2 Interlocking Faults in AC-DC Hybrid Grids

1.2.1 Current Status and Development Trend of Domestic

and International Power Grids

In 1882, Edison built a DC power system due to the limitations of the transmission
distance, which was replaced by the AC transmission system designed by Westing-
house in 1886. Because the transmission distance can be excessive, the AC trans-
mission system quickly became the mainstream transmission system worldwide. In
1954, ABB built the world’s first commercial, high-voltage DC (HVDC) transmis-
sion lines. Since the 1980s, large-scale, renewable energy generation has employed
inverters to connect to the AC grid. Two DC elements were added to the traditional
AC grid, but the main body of the grid is still the AC grid.
In recent years, the rapid development of DC transmission technology and renew-
able energy generation technology represented by wind power and photovoltaics has
changed the traditional AC grid pattern. DC transmission has become the preferred
method for long-distance power transmission. In Europe, due to the reconstruction
of energy markets, increasing environmental awareness, security of energy supply
and many other issues, high-voltage DC transmission technology has been utilized
as an effective way to solve these problems and is gradually replacing high-voltage
AC transmission technology [13].
In India, DC transmission has become a viable solution for enhancing the trans-
mission capacity of the power system and for eliminating existing development
bottlenecks, with 11 DC projects commissioned and under construction in China
22 1 Overview

alone as of 2014, including five bipolar DC transmission lines, one unipolar DC line,
and five back-to-back transmission lines [14].
In China, due to the inverse distribution of energy and load centers, the pattern
of west–east power transmission and north–south supply is objectively formed.
Considering the large transmission capacity and long transmission distance, ultrahigh
voltage DC has undergone rapid development in China, with 29 DC transmission
lines accounting for 36% of the transmission power between two regional grids. In
2020, a 500 kV flexible DC network will be put into operation. To solve the current
problem of “strong direct and weak AC”, the State Grid Corporation has adopted a
new transmission mode of UHV DC layered access to the AC grid [15, 16].
Renewable energy generation technologies, represented by wind power and photo-
voltaic, are becoming increasingly mature, and the installed capacity and power
generation capacity of renewable energy are rapidly growing. By the end of 2019,
the total installed capacity of renewable energy in Europe was 573 million kW, of
which wind power and photovoltaic power generation accounted for 58.29%. The
total installed capacity of renewable energy in the United States was 265 million
kW, of which wind power and photovoltaic power generation accounted for 62.05%.
The total installed capacity of renewable energy in China was 759 million kW, of
which wind power and photovoltaic power generation accounted for 54.78%. The
total renewable energy generation capacity worldwide was 2.04 trillion kWh. The
high proportion of wind turbines and PV plants connected to the AC grid by inverters
further deepens the depth and breadth of AC-DC hybrid grid coupling [17].
In summary, China has built a giant hybrid AC-DC power grid with the largest
scale, highest voltage level, highest percentage of DC transmission, and largest
amount of renewable energy generation consumption worldwide. The world’s power
grids are developing toward hybrid AC-DC connections.
The AC-DC hybrid grid is a new form that emerged in the history of grid devel-
opment, and the deepening degree of AC-DC coupling, especially the interaction
between AC systems and DC systems, has yielded new characteristics of faults in
AC-DC hybrid grids. For current-source LCC DC transmission, the converter inher-
ently suffers from a phase change failure problem [18]. Continuous and repeated
phase change failures can cause DC blocking [19], so the hybrid grid failure shows
“chain” characteristics, chain failure if not prevented, and the direct consequence
is the occurrence of large-scale blackouts. Therefore, an analysis if the interlocking
faults in AC-DC hybrid grids and the proposal of targeted system protection theory
and technology has theoretical and practical significance.

1.2.2 Key Features of the AC-DC Hybrid Grid

The hybrid AC-DC grid is assuming a new form in the history of power development
with the following main characteristics [20–21]:
1.2 Interlocking Faults in AC-DC Hybrid Grids 23

1. Controllability
The use of inverters to connect renewable energy generation to the grid is a direct
manifestation of the controllability of the hybrid AC-DC grid. The controllability
of DC transmission provides new options for operational control of the hybrid AC-
DC grid, but the current DC transmission system does not fully participate in its
tidal and stability control, and too many control variables create uncertainty in fault
prevention.
2. Vulnerability
DC system converter components have low overcurrent capability, are unable to
withstand fault shocks, and exhibit grid vulnerability, as exemplified by the poor
high- and low-voltage ride-through capability of early wind turbines, leading to their
large-scale off-grid.
3. Close AC and DC influence
As the degree of interconnection of AC-DC hybrid grids gradually increases, the
interaction between AC and DC becomes increasingly significant. Disturbances on
the AC side affect the normal operation of the DC side, and likewise, disturbances
on the DC side affect the AC side.

1.2.3 Interlocking Faults in AC-DC Hybrid Grids

Early cascading failures (cascading failures) is a term derived from network commu-
nication technology and refers to a chain failure in which a unit component fails. For
relay protection, interlocking faults can be considered developing faults (developing
faults) [22–24], transforming faults (transferring faults) [25], and cascading failures
(cascading failures, some people in China translate this term as chain failures) [26],
etc.
Cascading failures in the AC grid (cascading failures) have been defined in national
and international studies [27–31].
More representative of these studies, [27] defines a chain fault in an AC system
as a sequence of events in which multiple independent components fail in sequence
caused by an initial disturbance or a series of disturbances.
AC grid interlocking faults have many commonalities [32–33]: (a) they are acci-
dental; (b) they are random; (c) they involve multiple components or subsystems;
and (d) they are very destructive. Hidden faults [34–37] and interlocking trips are the
main manifestations of AC grid interlocking faults, which have been fully explained
in the previous analysis.
1. Interlocking faults in mixed AC-DC grids
A cascading fault in a hybrid AC-DC grid (cascading fault) is defined as a fault in an
AC-DC hybrid grid in which the failure of one device causes the failure or shutdown
24 1 Overview

of other devices or systems, especially the cross propagation of simple faults between
two AC-DC grids. The trigger of a chain fault in an AC-DC hybrid grid is a single
simple fault, and the core mechanism is the interaction of the AC-DC system. As
the converter is the link connecting the AC-DC system, its fault and failure will
propagate and fuel the fault chain process.
The definition of interlocking faults indicates that there are numerous scenarios
and forms of interlocking faults in an actual grid. A typical chain fault scenario is
described as follows: short circuit fault in the AC system at the receiving end causes
a voltage drop; low voltage leads to phase change failure of the converter; long-term
or multiple phase change failures cause DC blocking; after single DC blocking, the
tide will be transferred to the parallel running AC line; the AC line cannot withstand
the overload and chain trips; the frequency of the grid at the sending end rises due
to excess power; and the grid at the receiving end causes a frequency drop. Notably,
the abovementioned chain fault link is also the main fault form or performance of
the AC-DC hybrid power grid. The respective description is provided as follows:
(1) Phase change failure
A fault in the AC grid at the receiving end of the DC transmission system, such as a
single-phase ground fault or a three-phase fault, may cause a voltage reduction at the
receiving AC bus, resulting in a phase change failure of the converter, during which
the DC line transmitted power will fluctuate. Figure 1.15 illustrates the mechanism
of phase change failure.
Taking the process of phase change from valve 1 to valve 3 as an example, the
phase change bridge arm satisfies the relation

di 3 di 1
Uab = L c − Lc (1.9)
dt dt

Fig. 1.15 Circuit diagram of the six-pulse commutation bridge phase change process
1.2 Interlocking Faults in AC-DC Hybrid Grids 25

Considering the large value of impedance of the DC-side flat wave reactor, the
DC current Id is considered constant and satisfies i 1 + i 3 = Id . Then, Eq. (1.9) can
be transformed as,

d(Id − i 1 ) di 1 di 1
Uab = L c − Lc = −2L c (1.10)
dt dt dt
Integration of Eq. (1.10) yields.

t1
Uab
dt = Id (1.11)
−2L c
t0

Because Valve 1 and Valve 3 share a common anode, the lower voltage valve will
conduct during the phase change. The AC side line voltage is expressed as
√
Uab = 2U L sin(ωt) (1.12)

Rectifying Eqs. (1.9)–(1.12) yields

√
2  
X c Id = U L cos(γ + ϕ) − cos(β) (1.13)
2
where γ is the arc quenching angle and ϕ is the phase shift of the phase change
voltage. Transforming Eq. (1.13), γ can be expressed as
√
2Id X c
γ = arccos( + cos β) − ϕ (1.14)
UL

If γ is less than a certain value (e.g., 10°), a phase change failure is considered to
have occurred. Additionally, if the inverter stations of multiple DC lines are in close
electrical proximity to each other, a simultaneous phase change failure may occur on
the DC lines in the case of an AC fault on the receiving grid. If the AC fault is more
severe, it may eventually cause a continuous simultaneous phase change failure on
all multiple DC lines, which may have serious consequences.
(2) DC blocking
The above analysis clearly shows that if the AC fault at the receiving end of the grid
is more serious, the DC will experience continuous phase change failure. If a long-
term phase change failure or more than 2 consecutive phase change failures occur,
the DC will be blocked [19]. When DC lockout occurs, for the sending system, the
instantaneous surplus of large amounts of active and reactive power may cause gener-
ator power angle instability, transient high frequency and overvoltage phenomena.
For the receiving system, the instantaneous generation of active power deficit and
reactive power surplus will cause frequency drop and voltage fluctuation, which may
26 1 Overview

create stability problems in severe cases. During DC fault removal and recovery, the
converter station needs to absorb a large amount of reactive power in the sending and
receiving systems, which may easily cause voltage collapse and other problems.
(3) Tide shifting
Since the current DC system does not fully participate in the system’s tidal current
and stability control, its own transmission of active power has constant value. In AC-
DC hybrid grids, DC transmission lines often form a transmission cross -section with
several AC transmission lines to jointly link the sending and receiving grids. Once
the DC is blocked due to an accident, as the power transmitted by other DC lines is
set, the power transmitted by the DC line can only be carried by the AC contact line
located in the same transmission section, and the tide can only be transferred to the
AC line connected in parallel with it. This scenario may therefore cause overloading
of the AC lines operating in parallel.
(4) Chain tripping
DC systems transmit much more power than AC transmission lines. For example, a
single AC overhead line at 500 kV transmits approximately 1000–1500 MW, while
a ± 500 kV DC line transmits approximately 1500–3000 MW [38].
Therefore, the tidal shift caused by DC blocking will make the AC transmission
line seriously overloaded, causing the AC transmission line backup distance protec-
tion to operate, further aggravating the remaining AC transmission line overload. In
addition, power shortages or excess will cause system oscillation, and the incorrect
action of distance protection during oscillation will also aggravate the occurrence
and development of chain failures, which in extreme cases may lead to large area
malignant blackout accidents.
The above analysis clearly reveals that in the AC-DC hybrid grid, the mutual
coupling characteristics of the AC and DC grids determine that the concurrent action
of any one or more links may lead to other subsequent chain failures. In a narrow
sense, phase change failure is related to low voltage, and low voltage caused by any
reason may trigger phase change failure. In this sense, the chain fault in the AC-DC
hybrid grid is a frequent fault type and becomes the new normal fault.
2. Characteristics of interlocking faults in mixed AC-DC grids
As a new normal of faults in AC-DC hybrid grids, interlocking faults have the
following characteristics:
(1) Almost certain events
In the event of a severe AC fault, the occurrence of a chain fault is certain. In the best
case, phase change failures and transient tidal shifts will also occur.
(2) Not caused by a hidden fault
Hidden faults generally refer to faults caused by improperly set equipment param-
eters, which are not within the scope of the AC-DC hybrid grid interlocking faults
investigated in this book.
1.2 Interlocking Faults in AC-DC Hybrid Grids 27

(3) Not equivalent to chain tripping

Chain tripping is a traditional AC system chain event. The AC-DC hybrid grid chain
fault defined in this paper is a series of faults or equipment outages triggered by a
single fault, including not only AC side chain tripping but also DC system phase
change failure and DC single- and double-pole blocking.
(4) Potential for major power outages
When a chain fault occurs, the tide will fluctuate greatly between the AC grids
operating in parallel columns, and if the system works in the mode of AC (power
side, sending grid) direct (DC transmission system) AC (load center, receiving grid),
there will be a large power imbalance in both the sending grid and receiving grid,
and any incorrect action of relay protection and safety control devices in this process
will trigger a large-scale power outage. During the process, any incorrect action of
relay protection and safety control devices can lead to a massive blackout.
As a new type of fault, chain faults necessarily involve the analysis and research
methods of chain faults. In terms of theoretical analysis, all chain faults can be
considered complex faults, and mature theories and methods for complex faults
exist [39]. In terms of simulation calculation, existing simulation software such as
EMTP also has such capabilities. Unlike the familiar complex faults, chain faults
sequentially occur in the time domain and are therefore secondary to complex faults.
3. Risk areas for interlocking faults in hybrid AC-DC grids
Interlocking faults occurring in AC-DC hybrid grids are determined by a combination
of AC power supply characteristics, receiver system strength, converter characteris-
tics and network topology. Conventional fault analysis methods based on network
transformation and Davinan’s equivalence principle cannot be applied to hybrid grids.
To highlight the problem under study, a brief qualitative description of the causal
factors of interlocking faults, possible interlocking processes and risk factors leading
to interlocking faults are given below.
Figure 1.16 illustrates an AC-DC parallel transmission scenario. The power trans-
mission at the sending and receiving ends is shared by the DC line and shunt AC
line; the DC line of the same voltage level transmits much more power than the AC
line; and the DC power transmission is performed under a set control law. In general,
the rectifier-side and inverter-side control systems operate in constant current (CC)
control and constant arc quenching angle (CEA) control modes, respectively. Area
1 in the figure is the fault area of the AC field at the sending end, where 112Region
2 is the DC field fault region, including the DC line fault (F3) and converter station
fault (F6). Region 3 is the receiver AC field fault region, including the receiver AC
grid fault (F5) and parallel AC line fault near AC bus B2 (F4).
In the case of a fault in the sending AC grid (F1), the receiving inverter system
switches to the constant current control mode, while the sending rectifier is forced
to operate in the minimum trigger angle mode. Therefore, a fault in the sending AC
system has minimal effect on the DC line and the receiving AC grid. A fault in the
parallel AC line (F2) does not pose a significant safety threat to the DC line and
receiving AC grid.
28 1 Overview

Fig. 1.16 Regional analysis

of risk for hybrid AC-DC
grids

In the case of a fault in the receiving AC system (F5), the voltage of the receiving
AC bus B2 significantly drops and will cause the inverter to fail the phase change.
The fault will trigger low-voltage current limiting control (VDCOL) in the HVDC
control system to protect the commutation valve. The VDCOL will gradually increase
the DC current according to its characteristics, which helps the inverter station to
recover after a phase change failure. However, a severe inverter-side AC fault may
cause a long phase change failure or continuous phase change failure at the inverter
station. In this case, if the fault is not cleared within 400 ms [26], the converter is
forced to block, causing the HVDC line to drop out of operation, which has serious
consequences. Similarly, a fault at the end of the parallel AC line (F4) can also lead
to a phase change failure or even DC blocking.
When a fault occurs in a DC line or converter station (F3 and F6), the DC system
is forced out of operation and is blocked, at which point the tide carried by the DC
line is transferred to the AC line operating in parallel. Due to the large difference
in transmission power capacity, the AC line may be severely overloaded and cause
protection to operate, thus inducing a chain reaction.
The above analysis shows that the DC field (region 2) and the receiving AC field
(region 3) are high risk areas for initiating chain failures, while a fault in the sending
AC field (region 1) is a low risk area for chain failures.

1.3 System Protection Against Interlocking Faults in Mixed

AC-DC Grids

Based on the analysis in Sects. 1.1 and 1.2, many major outages are caused by
interlocking faults, and an interlocking fault in mixed AC-DC grids is a serious and
frequent type of fault. This fault certainly needs a protection technique, which must
also have an active and positive role in countering the interlocking faults to block the
occurrence and development of the fault and to minimize the scope of its impact and
outage. Such a protection technique is the system protection described in this book.
1.3 System Protection Against Interlocking Faults … 29

1.3.1 Defined Functions and Components of System

Protection

System protection is defined as the protection technology that guarantees the safety
of a power system. Thus, various relay protection technologies that help to quickly
remove faults and maintain the safe operation of the remaining power system, safety
and stability control technologies, cutter and load cutting technologies, unlisting
technologies, etc., all which belong to the system protection category. When a fault
occurs in the power system, the relay protection technology that quickly detects and
removes a fault already has its own clear definition and professional characteristics.
Therefore, system protection is the integration of all technologies other than over
relay protection that can continue safe and stable operation after a large grid distur-
bance. Notably, grid disturbance means that a fault or other large disturbance occurs
in the grid, after which the adopted All technical means are part of system protection.
However, such a concept is too broad, which renders the research and imple-
mentation of system protection nonspecific, does not easily capture the essence of
the problem, and cannot objectively reflect the actual progress of system protection
technology. The related domestic and foreign concepts include the “Three Lines of
Defense” in China, the “System Protection Scheme (SPS)” defined by Cigre [40], and
the “Special Protection System (SPS)” in the United States, Canada and other North
American countries. “Special Protection System (SPS)” [41], etc. will be explained
later in the book. The definition and technical scope of system protection in this book
are given below.
1. Functions and definitions of system protection [42, 43]
System protection is a grid safety-oriented protection technology for stopping faults
that occur after one fault has been removed or with a high probability of occurrence;
it is a specialized, anti-chain fault technical measure. Precisely, system protection
is a collection of protection and control techniques for interlocking faults in mixed
AC-DC grids:
(1) Prevention of prolonged or continuous phase change failures of converters.
(2) Mitigation of transfer overloads following DC lockout.
(3) Distance protection techniques to prevent chain tripping of the AC grid.
Figure 1.17 illustrates the system protection block diagram.
The purpose of system protection, as with all safety protection control techniques,
is to safeguard electrical equipment and the stability of the remaining system.
2. Main technical features of system protection
System protection is set up for chain failures, which are sequential recurring fail-
ures in the time domain. The corresponding system protection should be dynami-
cally configured layer by layer so that the system protection has the following basic
characteristics:
30 1 Overview

Fig. 1.17 System protection block diagram

(a) Has a dynamic character, not for a specific fault section, but for successive faults
following one fault.
(b) Has a proactive character, not passively waiting for the next failure to occur, but
starting to prevent the next failure from occurring based on the current failure.
(c) Blocks the development of chain failure.
Combined with the aforementioned chain fault scenarios and without loss of gener-
ality, chain fault prevention measures mainly include (a) protection control to prevent
DC phase change failure; (b) control techniques to prevent DC blocking; (c) protec-
tion control to prevent DC current transfer; and d) protection control to prevent AC
chain tripping.
System protection differs from relay protection in that relay protection targets
faults in a stable operating power system, which quickly acts and removes the faulty
equipment. System protection also differs from safety and stability control tech-
niques, such as cutting machines and loads, in that it does not remove any generators
or loads. In this sense, it is a protection technology that does not require circuit
breaker action to trip.

1.3.2 Our Three Lines of Defense [44]

The three lines of defense have a vital role in ensuring the safe and stable operation
of power systems and comprise the generic term for all relay protection, safety and
stability control devices and many related technologies. The three lines of defense
are set up to enable the power system to withstand disturbances, and similar to other
countries worldwide, China developed its own safety and stability guidelines.
1.3 System Protection Against Interlocking Faults … 31

1. Safety and stability standards for Chinese power systems to withstand large
disturbances
The Guidelines for the Safety and Stability of Electric Power Systems stipulate that
the safety and stability criteria for our power systems to withstand large disturbances
are divided into the following three levels:
Level 1 criterion: Maintain stable operation and normal power supply to the grid
[single fault (fault with high probability of occurrence)];
Level 2 criterion: Maintain stable operation but allow for partial loss of load
[single critical fault (fault with low probability of occurrence)];
Level 3 criteria: When the system cannot be maintained in stable operation, system
collapse must be prevented, and load loss must be minimized [multiple critical
failures (failures with low probability of occurrence)].
2. Three lines of defense
For the three levels of security and stability, three technical lines of defense are in
place to ensure the safe and stable operation of the power system in the event of any
major disturbance.
The first line of defense: fast and reliable relay protection, effective preventive
control measures to ensure stable grid operation and normal grid supply in the event
of a common single fault;
The second line of defense: the use of stabilization control devices and emergency
control measures, such as cutting machines and loads, to ensure that the grid can
continue to stably operate in the event of a serious fault with a low probability;
The third line of defense: installing out-of-step unlisting, frequency and voltage
emergency control devices, which are relied upon to prevent the expansion of acci-
dents and to prevent large-scale blackouts when the grid encounters multiple serious
accidents with low probability of stabilization damage.
In recent years, with the rapid development of DC transmission, new energy
generation, GPS and communication technology, the shape of China’s power grid
has profoundly changed, and to meet the changing needs of power grid security and
stability, Shu Yinbiao and Chen Guoping of the State Grid Corporation proposed
the concept of “system protection” based on the three lines of defense. To meet the
changing, the concept and specific requirements of “system protection” were further
given by Chen Guoping and Li Mingjie [44,45].
There is no doubt that the three lines of defense or the “system protection” concept
and technology system proposed by the State Grid have ensured that China’s power
grid rarely suffers from nationwide large-scale power outages and will continue to
ensure the safe and stable operation of China’s AC-DC hybrid power grid in the
future.
2.2 General Components of the National Grid “System Protection” [45]
System protection is still based on the traditional three lines of defense of the AC
power grid, and by consolidating the first line of defense, strengthening the second
line of defense and expanding the third line of defense, the contents and measures
32 1 Overview

Fig. 1.18 Overall system protection components

of the original three lines of defense are expanded to form a new comprehensive
defense system for the UHV AC/DC power grid. The relationship between system
protection and the traditional three lines of defense is shown in Fig. 1.18.
(1) Consolidation of the first line of defense

The purpose is to reduce the severity of the fault from the source of the fault to
suppress the fault to transfer the disturbance shock to the grid. When a component
fault occurs in the AC system, it will cause a disturbance shock to the AC system,
which will easily induce a phase change failure or blocking in the DC system and then
may cause a further chain reaction due to DC power fluctuation, AC grid tide transfer
and other AC-DC interactions. Therefore, it is necessary to take measures such as
fast fault removal, adaptive reclosing and station protection to suppress the impact of
component faults on the system and to cut off or inhibit the source of subsequent chain
reactions at the first stage of component faults that disturb the AC system. When the
initial fault is a DC device fault (e.g., due to the reliability of the DC body equipment
or improper control and protection adjustment), a series of chain reactions may also
occur due to AC-DC interactions. To curb this phenomenon, it is also necessary
to adopt new technical means similar to AC system relay protection and reclosing
to remove or suppress DC faults at the first stage of the disturbance impact of DC
faults on the AC system. In addition, it is necessary to change the concept of design,
protection and testing of network-related equipment only from the equipment itself,
to strengthen the degree of safety of the connected system to improve the operational
reliability of network-related equipment and network performance, and to reduce the
probability of equipment failure.
(2) Strengthening the second line of defense
When a serious fault occurs that has a large impact on the safe and stable operation
of the system, it will cooperate with various types of control resources and new
control means for a wide range of multivoltage sources and networks to achieve
active emergency control based on event triggering or combined with response-driven
control to interrupt the system chain reaction and prevent system destabilization.
1.3 System Protection Against Interlocking Faults … 33

(3) Expanding the third line of defense

The content includes (i) expanding the types of control resources (e.g., pumped
storage units, DC transmission systems, etc.) to include more control devices in local
decentralized control based on electrical quantity overrun detection; (ii) combining
fault events and response information to implement additional control based on
response information when the control quantity is insufficient in the emergency
control mode based on event triggering, etc.
The concept and content of “system protection” of the national grid comprises
the three lines of defense in the new grid form, which is derived from practice and
has practical research background and broad application prospects. Comparing the
three lines of defense with the system protection proposed in this book, it is easy
to determined that the three lines of defense are broad, while the system protection
proposed in this book is specific, which is a line of defense between the first line of
defense of relay protection and the second line of defense of emergency control and
load cutting, which can be considered 1.5 lines of defense.

1.3.3 Special Protection Systems [48]

Cigre’s System Protection SPS concept and the North American Special Protection
System SPS are similar in that they are a series of technical measures to maintain
safe and stable system operation after a grid disturbance. The following discussion
focuses on the North American SPS.
1. SPS concept
A Special Protection System (SPS) is a class of protection schemes that is designed to
maintain system stability [27]. In some systems, studies of system dynamic behavior
have shown that large disturbances occurring on critical lines or equipment can lead
to drastic, even catastrophic, consequences. These disturbances can occur in systems
interconnected by heavily loaded, long-distance or weak contact lines. Once an inci-
dent occurs, the system may split in an unpredictable manner into several islands of
very unbalanced load and generation. Such islands cannot maintain operation and
eventually cause a system-wide blackout. One way to prevent such a catastrophic
outcome is to unbundle the system into several islands with more balanced genera-
tion and load in the manner envisioned, which would increase the likelihood of the
islands surviving.
Another situation that requires special treatment is when a disturbance occurs in
one utility due to its inherent problems or size but creates serious consequences for
the transmission system of a neighboring utility. In the North American Intercon-
nected Power System, the interconnection among systems is carefully designed and
controlled to prevent this situation. This design requirement ensures that no system
designed by either utility can create serious problems for the adjacent utility. This
requirement stipulates that measures be taken during the design process to ensure that
34 1 Overview

the introduction of new equipment does not cause operational difficulties for adja-
cent utilities, even in the event of a failure or other unanticipated disturbance event.
An example of a type 2 disturbance event is the bipolar blocking of a large HVDC
converter station. Since the HVDC system interconnects multiple power companies
on the AC side and carries a large amount of load prior to the outage, a converter
station failure will force the liaison lines between two power companies to absorb
the sudden change in tide, causing a large number of line trips.
Stability studies have shown that all of the above instances are so severe that
control measures must be taken in advance. These controls then form a dynamic
safety barrier that quickly responds to maintain stable system operation in the event
of an anticipated accident.
2. SPS characteristics
The above control strategies are known by many different names in addition to SPS
[43], for example, special stability control systems, dynamic safety control systems,
emergency incident handling schemes, recovery control schemes, adaptive protection
schemes, corrective control schemes, and enhanced safety schemes. Depending on
the problem caused by the disturbance, these schemes provide different types of
control measures. Some schemes jump-start generators; some schemes purposefully
disconnect lines; and other schemes unlock the system at predetermined locations.
The common features of these schemes are presented as follows:
(1) All schemes are dynamic safety control systems designed for control system
stability. If these control schemes are not employed, the system will have serious
consequences.
(2) In contrast to online real-time control, both schemes are designed offline. The
reason is that the power system responds so quickly that there is no time to
implement the following common sequential control system logic: observe in
real time, determine the extent of the disturbance, decide what action should be
taken, and take the needed action.
(3) Most control schemes can be selected depending on the actual needs of the
system, that is, under certain operating conditions, special control logic is not
needed, at which point the SPS should be withdrawn.
(4) All control scheme measures are designed to restore the dynamic behavior of
the system or to take predetermined control measures when an event will have
serious consequences.
As outlined above, the SPS and our safety and stability control system have similar
functions, which prevent the grid from being disturbed to avoid chain reactions and
to minimize the scope of outages.
3. Examples of special protection systems
The following is an example of an actual SPS through the Intermountain Power
Project (IPP) of the Western U.S. Power Grid. [46] The IPP consists of the Inter-
mountain Power Station (IGS) in Utah, the Intermountain Converter Station (ICS),
the Adelanto Converter Station (ACS) in Southern California, and a 784 km ± 500 kV
1.3 System Protection Against Interlocking Faults … 35

Fig. 1.19 IPP geographic

wiring diagram [45]

HVDC transmission line connecting the ICS to the ACS. The geographical wiring
diagram of the IPP is shown in Fig. 1.19. In addition to the abovedescribed equip-
ment, the IPP has transmission lines connecting to the Utah, Nevada, and Southern
California grids. The IGS has two 827 MW generators, and the HVDC system is
rated to transmit 1600 MW, so the entire power plant can be delivered to southern
California or have a portion to supply the local grid.
Studies of the design process have shown [47] that under certain load conditions,
unipolar or bipolar blocking of HVDC will immediately force power plant output to
the Utah and Nevada grids, which is likely to cause severe voltage reductions and line
tripping or load shedding on these grids. Accordingly, in this case, the SPS would
remove the IGS units.
The SPS only needs to be operational under heavy load operating conditions,
which requires monitoring load levels along several key transmission corridors in
the western U.S. grid. Of these, the Pacific Interconnection Corridor, which connects
Southern California to the Pacific Northwest, and the transmission corridor from
Arizona to Southern California are particularly important because they also deliver
power to Southern California. Based on the load levels in these corridors, the Los
Angeles Department of Water and Power’s Energy Control Center (ECC) determines
whether to establish an SPS.
The SPS itself consists of the ICS, ACS, and volume measurements of the
converter stations on both sides of the ICS and ACS. The following input triggers are
defined therein: fault type, fault location, voltage level, 1 AC line trip, 2 AC line trips,
single pole blocking, and double pole blocking. A total of 17 of these input triggers
are distributed on both ends of the HVDC system. The status of these input triggers is
36 1 Overview

transmitted via a microwave or fiber optic cable to the computer logic unit located at
the power plant. The combination of these input triggers to the power plant identifies
13 single and 6 double (or multiple) disturbances, referred to as emergency state
triggers. The trigger logic was designed on the basis of hundreds of stability anal-
yses and is very complex [35]. Ultimately, these emergency state triggers form five
“super triggers” through “or” logic, which are used to cut the machine or to take other
measures to prevent the propagation of disturbances to adjacent power systems. The
system is designed to be scalable, allowing for the integration of additional control
measures should the need arise in the future. This integration may include line trip-
ping, application of dynamic brakes, or other appropriate measures. The entire SPS
is fully redundant so that if one set is taken out of operation and maintenance, the
other set can still perform its full function. The SPS does not require that the trip
commands of the two systems be identical.
A comparing of SPS with our three lines of defense and the system protection
proposed in this book indicates that SPS has similarities with our three lines of
defense, especially in how to ensure system power balance after simple fault removal
to maintain an acceptable system frequency and voltage, which can be considered
the second and third line of defense in our three lines of defense. One very important
reason for defining the new “system protection” in this book is that phase change fail-
ures occur in converters in AC-DC hybrid grids, and chain failures become inevitable.
In addition, a very important purpose is to promote the integration of traditional relay
protection and safety and stability control systems to achieve a four-thirds effect. The
effect is to pave the way for the subsequent safety and stability control system (second
and third line of defense) or to facilitate the role of SPS in preventing a fault from
developing into a serious stability damage accident in a timely and proactive manner.
Chapter 2
Distance Protection Against Overload

The action of distance protection in the case of accidental overload is an important

cause of chain tripping and poses a serious threat to the safety of the grid. The solution
to this problem consists of three parts: identification of accidental overloads; reli-
able protection blocking without endangering equipment safety; and reliable protec-
tion action in the case of endangering equipment safety. Identification of accidental
overloads and reliable blocking need to be addressed while ensuring the safety of
equipment. Of these, the effective identification of accidental overloads is the first
step.
Complex dynamic processes and variable operation modes hinder the analysis
and identification of accidental overloads. Some studies [72] consider the dynamic
process of accidental overload and classify it into asymmetric and symmetric cate-
gories; however, the proposed asymmetric identification criterion has defects, which
may cause the protection to be rejected under a single phase via a large resis-
tance ground fault and a reduction in the protection sensitivity. The proposed
scheme in [118–119] may not be able to mitigate accidental overloads after
multiple branch disconnections and asymmetric accidental overloads. The interphase
relay constraints proposed in [126] may be misjudged during accidental overload
dynamics. The proposed scheme in [68, 120, 121] may be misjudged during asym-
metric accidental overloads and may not be able to fully mitigate accidental overloads
under different operation modes. The problem of accidental overload identification
has not been completely solved by domestic and international research.
To address this problem, it is necessary to conduct an in-depth analysis of why
and when accidental overload triggers distance protection. This chapter provides a
comprehensive analysis of the action behavior of distance section III under accidental
overload. First, in the time domain, the action of protection during the dynamic
process of accidental overload is analyzed. Second, in the spatial dimension, network
analysis is carried out according to different types of accidental overload. Third, the
conditions of protection beyond accidental overload are analyzed. Last, the concept
of accidental overload identification and blocking combining multiple disturbance
time information, station domain switch information and station domain electrical

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 37
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_2
38 2 Distance Protection Against Overload

information is proposed, thus paving the way for obtaining a thorough solution to
the problem of accidental overload identification.

2.1 Analysis of the Action Behavior of the Distance III

Section Under Accidental Overload

2.1.1 Triggering Events for Accidental Overload

Incidental overload is caused by the tripping of one or more operating lines due to
a fault or no-fault condition, the process of which is a shift in tide, and causes an
overload on the operating lines of the parallel transmission cross-section. A parallel
transmission cross-section is a group of interconnected lines that have the same load
area or power area [139]. Two types of components are involved. To simplify the
narrative, lines tripped due to faults or no faults are subsequently denoted by Line I,
and lines overloaded due to tidal shifts after an accident are denoted by Line J. Line
I faults and subsequent perturbations, such as switch movements, are the initiating
events of accidental overloads and are classified in terms of fault type, location and
dynamic processes as follows:
(1) Line I fault types can be classified as hidden faults (no-fault trips), single-phase
faults, two-phase ground faults, two-phase interphase faults and three-phase
faults.
(2) If the Line I fault is located in the Line J distance III section of the fixed value of
the zone, then the Line I fault is reunited in the fault phase, and Line J protection
starts, that is, outside the zone fault; otherwise, protection does not start.
(3) Line I failure and subsequent switch action and other disturbances are accidental
overload triggering events, including four main stages: ➀ fault, ➁ first selective
tripping, ➂ reclosing, and ➃ three-phase permanent jump. If the line successfully
recloses, the tidal current transfer will be terminated, and distance III section
exit action is not likely. Here, only the Line I reclosing failure in the three-phase
permanent trip case is considered.

2.1.2 Action Behavior of Distance III Segments During

Dynamics

Because the backup protection export action is based on the time period of continuous
judgment, taking into account the Line J protection action, the main distance III
section protection is continuous in the action zone. For Line I, different fault types,
locations, and Line J distances III start are continuous in the action area of the
situation analysis (Table 2.1). Z is the distance III ground relay, and Z is the
distance III phase relay.
2.1 Analysis of the Action Behavior of the Distance … 39

Table 2.1 Start-up and continuous operation of various types of accidental overload protection
Fault of line I Distance protection III of line J
Type In the area of distance Startup stages Continuous in action
protection III of line J? zone stage
Latent fault – ➁ ➃ or ➃ Z & Z : ➃
Phase A ground fault No ➁ ➃ or ➃ Z & Z : ➃
Yes ➀➂➃ ZA : ➂ ➃, other: stage ➃
Yes ➀➁➂➃ ZA : stage ➀–➃, other:
stage ➃
Phase BC ground fault No ➁ ➃ or ➃ Z & Z : stage ➃
Yes ➀➁➂➃ ZBC & ZB & ZC : stage
➀–➃, other: stage ➃
BC interphase fault No ➁ ➃ or ➃ Z & Z : stage ➃
Yes ➀➁➂➃ ZBC : 阶段 ➀–➃, other:
stage ➃
Three-phase fault No ➁ ➃ or ➃ Z & Z : stage ➃
Yes ➀➁➂➃ Z & Z : stage ➀–➃

Due to hidden faults on Line I, i.e., no fault trips, the Line J distance III section
protection will not enter the action zone during Line I faults and reclosures and needs
to be continuously maintained in the action zone after stage ➃, i.e., Line I three-phase
permanent trips.
When a permanent, single-phase earth fault occurs on Line I, if the fault is not
within the Line J distance III section rating zone, the Line J distance III section will
only continuously fall into the action zone from phase ➃. If the fault is within the
Line J distance section III rating zone, the Line J distance section III earth relay
phase A ZA will fall into the action zone during phases 1 and 3. If ZA starts during
Line I non-full-phase operation (phase A selective jump), i.e., during the asymmetric
tidal shift caused by phase ➁, the continuous start timing of ZA starts from phase ➀,
increasing the possibility of exit action.
When a permanent, single-phase earth fault occurs on Line I, if the fault is not
within the Line J distance III fixed value zone, Line J distance III continuously falls
into the action zone only from stage ➃. Conversely, the fault-related phase relay is
continuously located in the action zone from stage ➀, and the other phase relays
continuously start from stage ➃.
The current study of accidental overload identification based on wide-area infor-
mation [118–121] and the study of accidental overload identification based on in-situ
information [97,126] do not consider the case when the Line I fault falls within the
Line J distance section III fixed value zone. However, only the symmetrical tidal
shift after phase ➃, i.e., Line I reclosing failure three-phase tripping, occurs. The
phase-to-phase, cosine voltage constraints mentioned in the literature [126] and the
voltage plane-based constraints of the literature [68] will open the Line J distance
III protection, causing an unreasonable trip of Line J if the Line I phase fault falls
40 2 Distance Protection Against Overload

within the Line J distance III fixed value area. On the other hand, the above method
does not take into account the asymmetric tidal shift overload process mentioned in
the literature [137], which may also lead to misclassification during the trip open,
single-phase, non-full-phase operation of Line I.
The literature [138] proposes criteria based on literature [137] for symmetrical
and asymmetrical tidal shifts. The symmetrical tidal shift identification criteria and
logic flow can block accidental overloads caused by two-phase or three-phase faults
of Line I falling into the fixed value area of the distance III section of Line J. However,
the proposed criterion for asymmetric tidal shift overload identifies a single-phase
grounded via transition resistance as an asymmetric tidal shift and blocks it, which
will reduce the sensitivity of the distance III section under a single-phase fault.
Therefore, the identification of accidental overloads has not been completely
solved. The influence of out-of-zone fault phases ➀ and ➂ and out-of-zone non-full-
phase operation (single-phase postfault phase ➁), especially continuously enabling
the protection in the action zone, on the accidental overload identification cannot be
disregarded, as illustrated in the following simulation in PSCAD/EMTDC:
1. Impact of out-of-area faults on accidental overload identification
Based on the 3-machine, 9-node model, combined with the research content of this
chapter, the transformer node is removed, and receiver node 5 is added to build a
500 kV, 7-node, ring network simulation system in PSCAD/EMTDC, as shown in
Fig. 2.1. To analyze the protection action behavior triggered by accidental over-
load, the rectified EM , EN , EP , and EQ potential angles are 55°, −60°, 10°, and 15°,
respectively, and the system equivalent impedance is shown in 0.
The simulation system takes into account reactive power compensation, and the
loads carried by the nodes are S4 = 1000 + j50MVA, S5 = 2000 + j150MVA, S6 =

EN Bus5

Bus1 Bus4 Bus2

R
L14-1 L45 L25-1
EM EP
L14-2 L25-2
EN
L17 L26
S4 S5

L37 L36
Bus7 Bus6

S7 S6
Bus3

EQ

Fig. 2.1 500 kV, 7-node simulation system

2.1 Analysis of the Action Behavior of the Distance … 41

Table 2.2 500 kV, 7-node simulation system parameters

System parameter Sending end Receiving end Sending end Sending end
system M system N system P system Q
Positive (negative) 10∠87◦ 30∠87◦ 20∠87◦ 50∠87◦
sequence
impedance ()
Zero sequence 8∠85◦ 40∠85◦ 10∠85◦ 100∠85◦
impedance ()

Table 2.3 500 kV line

Line parameters Positive sequence Zero sequence
parameters
R (/km) 0.027 0.1984
XL /km) 0.2701 1.131
XC (M·km) 0.25879 0.62414

1000 + j100MVA, and S7 = 800 + j100MVA. The active power flows are represented
by the arrows in the figure. The length of l25 is 120 km; the length of each remaining
line is 200 km; and typical 500 kV parameters are chosen, as shown in Tables 2.2
and 2.3 [186].
The Line L14-1 distance section III phase-to-phase relay action criterion is

U̇
90◦ < arg < 270◦ (2.1)
U̇ − I˙ Z set

where U̇ is the protection measured phase voltage, I˙ is the protection measured
phase current and Zset is the protection fixed value.
In conventional identification of phase-to-phase faults and accidental overloads,
cosine voltage limiting conditions, which open the protection once satisfied, are
usually established [90]:

U cos ϕ < mUNN (2.2)

where ϕ FF is the angle between the phase voltage and the phase current at the
protection installation, U NN is the rated phase voltage, and m can be set to 0.5.
N-3 faults occur in the system, causing L14-2 , L25-1, and L25-2 to sequentially trip
open; the sequence of events is shown in Table 2.4. The protection and reclosing time
in the simulation is set as follows [68]: the main protection trips the outlet 20 ms after
the fault occurs, and the breaker opening time is set to 40 ms, so the interval between
fault occurrence and excision is 60 ms; considering the latent supply current in non-
full-phase operation, the single-phase reclosing is 0.7 s. According to the People’s
Republic of China Power Industry Standard DL/T 599-2007, “220–750 kV Grid
42 2 Distance Protection Against Overload

Table 2.4 Sequence of accidental overload events

Time (s) Event
0.1 BC interphase fault occurs at 20% from m side of L14-2
0.16 Three-phase tripping of main protection action of L14-2
0.6 BC phase-to-phase grounding fault at 50% from p side of L25-1
0.66 Three-phase tripping of main protection action of L25-1
0.86 Reclosing within the fault of L14-2
0.92 Three-phase acceleration tripping of L14-2
1.1 BC phase-to-phase grounding fault occurs at 100% from p side of L25-2
1.16 Three-phase tripping of main protection action of L25-2
1.36 Reclosing within the fault of L25-1
1.42 Three-phase acceleration tripping of L25-1
1.86 Reclosing within the fault of L25-2
1.92 Three-phase tripping of main protection action of L25-2

relay protection device operating calibration procedures” [90], with an end-of-the-

line ground fault sensitivity of 3.0, grounding relay section III is adjusted to 162.9
card 84.3 , and the section III action time is set to 1.5 s.
The L14-1 Line M Side phase ratio results for distance segment III are shown in
Fig. 4.2, with the shaded portion of the figure showing the distance relay action zone
(i.e., distance relay phase ratio results between 90° and 270°). When the incident
proceeds to N-3, i.e., after a phase-to-phase fault on L25-2 , the BC phase-to-phase
distance relay on Line M side III of L14-1 enters and remains in the action zone,
while phase B grounded section III enters and remains in the action zone (phase
C grounded section III had exited the action zone when L25-1 was reclosed). If no
measures are taken to effectively eliminate line overload, the BC interphase distance
relay enters the action zone for 1 s, and the distance section III exits and removes
Line L14-1 (Fig. 2.2).
The overload limits are currently proposed for the interphase relays. However,
during a BC phase fault on L25-2 , as shown in Fig. 2.3, the BC interphase cosine
voltage drop measured by the relay at the head of the L14 line is 0.28 p.u., satisfying
the interphase cosine voltage condition. The interphase relay will open, generating an
unreasonable action. Note that 1.0 s after the L14-1 interphase III section starts, if the
emergency control strategy is not effectively implemented to mitigate the overload
situation, the L14-1 line will jump open in three phases, further expanding the tidal
shift.

2. Impact of out-of-zone non-full-phase operation on accidental overload identifi-

cation

Figure 2.4 shows a 500 kV, double-ended power transmission system with the M-
side system at the sending end, a 60˚ power angle over the N-side system at the
receiving end and the N-side system with a load of 2500 + j1000 MVA. The system
2.1 Analysis of the Action Behavior of the Distance … 43

Fig. 2.2 Specific phase of L14-1 first distance section III under accidental overload

Fig. 2.3 BC phase-to-phase cosine voltage at the first end of L14-1 under accidental overload

parameters are shown in Table 2.5. The length of Lines L1-L6 is set to 200 km, and
the line parameters follow the data shown.
The Line L1 distance section III grounding relay action criterion is:

U̇
90◦ < arg < 270◦ (2.3)
U̇ − ( I˙ + K I˙0 )Z set
44 2 Distance Protection Against Overload

Fig. 2.4 500 kV dual power R1 L1

supply simulation system R5 L5
diagram EM R2 L2 EN
R3 L3
R6 L6
R4 L4

Table 2.5 Parameters for the 500 kV dual power system

System parameter Sending end system M Sending end system N
Positive (negative) sequence impedance () 10∠87◦ 20∠87◦
◦
Zero sequence impedance () 5∠85 10∠85◦

where U̇ is the measured phase voltage of the protection; I˙ and I˙0 are the measured
phase current and zero sequence current, respectively, of the protection; and K is the
zero sequence compensation factor.
A system N-3 fault occurs, causing L1 -L3 to sequentially trip open, as simulated
in the sequence of accidental overload events shown in Table 2.6
The results of the L4 Line side M phase ratio of the distance III section grounding
relay, after the L1 ~ L3 successive faults, are shown in Fig. 2.5. The shaded part of the
figure shows the distance relay action zone (i.e., the distance relay phase ratio result
falls between 90° and 270°). Figure 2.5 shows that when the fault develops to N-3,
the L4 line grounding distance relay section III phase A enters the action zone, and
after L3 is removed for the second time, the distance relay section III between phase
CA and phase AB also remains continuous in the action zone. If no control measures

Table 2.6 Sequence of accidental overload events

Time (s) Events
0.1 Phase A fault occurs at 80% from side m of L1 (through 20  transition resistance)
0.16 Main protection action jumps out of phase A of L1 , and non full phase operation
0.6 Phase A fault occurs 20% from side m of L2 (through 20  transition resistance)
0.66 Main protection action jumps out of phase A of L2 , and non full phase operation
0.86 Reclosing phase A at fault of L1
0.92 Three-phase acceleration tripping of L1
1.1 Phase a fault occurs at 50% from side m of L3 (through 20  transition resistance)
1.16 Main protection action jumps out of phase A of L3 , and non full phase operation
1.36 Reclosing phase A at fault of L2
1.42 Three phase acceleration tripping of L2
1.86 Reclosing phase a at fault of L3
1.92 Three-phase acceleration tripping of L3
2.1 Analysis of the Action Behavior of the Distance … 45

Fig. 2.5 Specific phase of L4 first distance section III under accidental overload

are taken to effectively eliminate the line overload, after 1.5 s of grounding section
III initiation, i.e., 0.68 s after L3 is removed, the L4 grounding distance section III
phase A exit is operated to protect the excised line.
The identification of single-phase faults and accidental overloads is usually distin-
guished by the magnitude of the negative sequence, zero-sequence currents or current
asymmetry. The negative sequence current |Ia2|, zero sequence current |I0| and current
asymmetry (|Ia2|+|Ia2|)/|Ia1| in this process are shown in Fig. 2.6.
During the non-full-phase operation of L2 and L3, the negative sequence current
at the first end of L4 reaches a maximum near the rated current. The zero-sequence
current is smaller due to the high zero-sequence impedance of the line. Addition-
ally, the current asymmetry is small because of the large magnitude of the positive
sequence current under accidental overload, and its maximum value is less than
0.3. After the protection starts, if the negative sequence current threshold is set to
distinguish between fault overload and accidental overload, i.e., if the protection is
opened when |Ia2| is greater than the threshold value [125], it may cause unreasonable
protection operation during out-of-zone, non-full-phase operation.
The previous analysis revealed that there are multiple disturbances in the whole
process of accidental overload and that the existing accidental overload identification
46 2 Distance Protection Against Overload

Fig. 2.6 Negative sequence and zero sequence currents and current asymmetry at the first end of
L4 under accidental overload

judgment does not sufficiently consider the influence of out-of-zone faults and out-
of-zone non-full-phase operations. Thus, a thorough identification is difficult, and
there is the risk of export action. Therefore, to avoid unreasonable action of protection
in dynamic processes, the identification of accidental overload should be extended
from the judgment in the time section to the dynamic judgment in the time domain.
Local-based information acquisition is necessary.

2.2 Network Analysis of Accidental Overload

2.2.1 Station Domain Accidental Overload Versus Nonstation

Domain Accidental Overload

Before performing the network analysis, a distinction is made between station

domain accidental overload and nonstation domain accidental overload. During
station domain accidental overload, if Line I and Line J are delivered at the same
substation and Line I is removed, the tidal shift causes an accidental overload on
2.2 Network Analysis of Accidental Overload 47

Line J, as noted in the case presented in the previous section. Conversely, for the case
where faulty Line I and overloaded Line J are not in the same substation, nonstation
domain accidental overload occurs, as noted in the case of the L25-1 , L25-2 fault and
trip causing L14-1 overload in the previously discussed case.
For station domain overloads, a typical situation is that one or more of the
double/multiple return lines are opened, causing overloads on the remaining parallel
lines. The dynamic identification of accidental overloads can be achieved if the elec-
trical and switching information of all parallel lines is available. Therefore, only
nonstation domain accidental overloads need to be analyzed in the network.

2.2.2 Network Analysis of Nonstation Domain Accidental

Overload

As discussed in the previous section, the accidental overload caused by different

faults contains three main states: out-of-zone faults, out-of-zone three-phase open
and out-of-zone non-full-phase operation. The three states are separately compared
with the in-zone faults.
1. Out-of-zone failure versus in-zone failure
The root cause of the protection override problem caused by out-of-zone faults lies in
the fixed value adjustment, which is essentially a contradiction between the selectivity
of backup protection and sensitivity of backup protection. The root cause cannot
be distinguished by fault network analysis, but the probability of out-of-zone fault
override can be reduced by dynamic judgment blocking in the time domain and by
reasonable adjustment of the fixed value.
2. Out-of-zone three-phase open versus in-zone fault
The opening of three phases outside the zone will cause a symmetrical tidal shift,
and if the distance protection is activated in this case, it will be either the three-phase
ground relay or all three interphase relays that are activated, which is the same as
noted in the case of a three-phase fault within the zone. A single consideration of
this phase has been more comprehensively analyzed and is easier to distinguish, e.g.,
based on the interphase, arc resistance characteristics of the three-phase fault that
can be distinguished and will not be analyzed in this chapter.
3. Out-of-zone non-full-phase operation versus two/three phase faults
Given the phase-to-phase arc resistance characteristics of two/three phase faults,
which are easier to distinguish from non-full-phase operating conditions outside a
single zone, they are not analyzed here.
48 2 Distance Protection Against Overload

Fig. 2.7 Simulation of a k

branch circuit after a short i j
circuit fault trip
YFβ yik y jk YFγ
k
i j
m n
YFα

4. Out-of-zone non-full-phase operation versus single-phase faults

1) Out-of-zone non-all-phase operation

This state is caused by two events: a single-phase fault and a single-phase trip open.
The effect of a single-phase fault on the network after a single-phase jump-off is
analyzed.
There is a new node m and n on each side of the line switch, and the line switch
is simulated with a section of branch circuit, as shown in Fig. 2.7.
Y Fα is the equivalent derivative of the k-point ground fault branch, and Y Fβ and
Y Fγ are the equivalent branch derivatives of the 2 line switches. For this equivalent
circuit, the nodal derivative matrix, including k, m and n, is created [136]:

i j k m n
⎡ ⎤
i Y Fβ 0 0 −Y Fβ 0
j⎢⎢ 0 Y Fγ 0 0 −Y Fγ ⎥
⎥ (2.4)
⎢ ⎥
k⎢ 0 0 y −yik −y jk ⎥
⎢ ⎥
m ⎣ −Y Fβ 0 −yik −yβ 0 ⎦
n 0 −Y Fγ −y jk 0 yγ

where y = yik + yjk + YFα , yβ = yik + YFβ and yγ = yjk + YFγ .

Elimination of the added nodes k, m, n gives the derivative matrix with respect to
nodes i and j.
⎡  ⎤−1 ⎡ ⎤
y −yik −yjk 0 0
Y Fβ 0 0 −Y Fβ 0 ⎣ −yik yβ
Yf = − 0 ⎦ ⎣ −Y Fβ 0 ⎦
0 YFβ 0 0 −Y Fβ
−yjk 0 yγ 0 −Y Fγ
(2.5)

where Y Fα , Y Fβ, and Y Fγ are the conductors of the faulty branch. Taking the form of
0, 1, and 2 sequential components, we have
2.2 Network Analysis of Accidental Overload 49
⎡ ⎤ ⎡ ⎤
111 2 −1 −1
yf ⎣ y
1 1 1 ⎦, Y Fβ = Y Fγ = ⎣ −1 2 −1 ⎦
o
YFα = (2.6)
3 3
111 −1 −1 2

where yf is the k-point ground conductance and yo is the line switch conductance.
The following section presents an analysis of the effect of a single-phase ground
fault on a non-all-phase operating network. Let the switch conductance yo be set
to 10–6 ; in the case of a nonfault, direct single hop, yf is set to 10–3 , and in the
case of single hopping after metallic single-phase grounding, yf is set to 102 (i.e.,
the transition resistance is 0.01 ). Substituting into Eq. (2.5), the difference in the
sequence component conductance impedance calculated for the two cases has an
order of magnitude of 10–4 , so the network analysis can be performed only for the
non-full-phase operation condition after single-phase tripping.
Figure 2.8 shows a dual power system, where N is the receiving end, M and P are
the sending ends and r is the distance III section protection installed at the head of
Line J. The diagonal area is the protected section of protection R. (Both this section
of the line and the next section of the line in the far backup area are long and of
equal length. In the figure, Z Lj indicates the total impedance of the two sections of
the line.)
The boundary condition at the disconnection after the tripping of phase A of Line
I is

I˙k1 + I˙k2 + I˙k0 = 0

(2.7)
U̇k1 − I˙k1 Z Li = U̇k2 − I˙k2 Z Li = U̇k0 − I˙k0 Z Li

where Z Li is the value of the inter-break impedance, i.e., Line I impedance. The
composite sequence network for two-phase operation after phase A trips is shown in
Fig. 2.9, and the bus NP can be considered a node between
2) Single-phase fault
The boundary conditions at the short-circuit point after a single-phase fault on Line
J are shown in Eq. (2.8), and its composite sequence network is shown in Fig. 2.10.

I˙f1 = I˙f2 = I˙f0

(2.8)
U̇f1 + U̇f2 + U̇f0 = ( I˙f1 + I˙f2 + I˙f0 )Rf

Fig. 2.8 Dual power system network

50 2 Distance Protection Against Overload

Fig. 2.9 Composite Z 00

sequence network for
non-full-phase operation Z Li0
after phase a tripping I k1 Z11 Z Li1 I k0 U k0

I k2
[0] U k1 Z Li2
U PN
Z 22
U k2

Z ∑ (1) Rf I f1 Z ∑ (2) Rf I f2 Z ∑ (0) Rf I f0

U f[ 0 ] U f1 U f2 U f0

Fig. 2.10 Composite sequence network for phase A short circuit

3) Comparative results of network analysis

The fault network analysis based on the fault component method is separately carried
out for single-phase tripping outside the zone and a single-phase short circuit inside
the zone, and the fault components at the fault and at the protection installation are
shown in Table 2.7.
where U̇fA[0] is the prefault voltage at the short-circuit point, whose amplitude is
similar to that of Ė M , and the phase lags behind β; C j1 and C j0 are the positive
(negative) sequence and zero sequence current fault distribution factors of Line
J; C m1 and C m0 are the positive (negative) and zero sequence voltage fault distri-
bution factors of bus M; Z  (1) and Z  (0) are the equivalent positive (negative)
sequence impedance and equivalent zero sequence impedance, respectively, viewing
the system from the short-circuit point; and Z 11 and Z 00 are the equivalent positive
(negative) sequence impedance and equivalent zero sequence impedance, respec-
tively, viewing the system from the line break. The values are Z 11 = Z Li1 + Z P1 +
(Z M1 + Z N1 )Z N1 /(Z M1 + Z N1 + Z Lj1 ) and Z 00 = Z Li0 + Z P0 + (Z M0 + Z N0 )Z N0 /(Z M0
+ Z N0 + Z Lj0 ).
A comparison of the fault component felt at the Line J protection installation after
a single-phase fault on Line J with a single-phase trip on Line I reveals the following
findings:
(1) Since the fault component after single-phase tripping is affected by the system
operation mode and network structure and the fault component of single-phase
grounding is affected by the fault operation mode and fault boundary conditions
Table 2.7 Fault components at the fault and at the protection installation
Sequence current at fault (broken wire) Fault component at Line j protection

−U̇fA[0]
I˙f1 = I˙f2 = I˙f0 =
2Z  (1) +Z  (0) +3Rf
Line j I˙j1 = I˙j2 =Cj1 I˙f1 , I˙j0 =Cj0 I˙f0
Phase a fault − Ė M ejβ
≈  U̇m1 = U̇m2 = Cm1 I˙f1 Z m1
2Z (1) +Z  (0) +3Rf
2.2 Network Analysis of Accidental Overload

U̇m0 = Cm0 I˙f0 Z m0

1 Z 00
I˙k1 = −(U̇P − U̇N )( − 2 )
Z Li1 Z 11 + 2Z 11 Z 00
(U̇P − U̇N )Z 00 I˙j1 = Cj1  I˙k1 , I˙j2 = Cj1  I˙k2
Line I I˙k2 = − 2
Nonfull line operation after phase a tripping Z 11 + 2Z 11 Z 00 I˙j0 = Cj0  I˙k0
(U̇P − U̇N )
I˙k0 = − U̇m1 =U̇m2 =Cm1  I˙k1 Z m1
Z 11 + 2Z 00
U̇m0 =Cm0  Ik0 Z m0
51
52 2 Distance Protection Against Overload

(fault location and transition resistance), it is difficult to distinguish them from the
magnitude.
In the 500 kV, dual power supply simulation system shown in Fig. 2.4, two types
of operations are performed. The system impedance parameters in the large opera-
tion mode are shown in Table 2.5. The system impedance parameters in the small
operation mode are set to ZM1 = 200∠87° , ZM0 = 100∠85° , ZN1 = 100∠87°
, and ZN0 = 50∠85° . The occurrence of a single-phase ground fault at different
locations in this section and the lower section of the line via the transition resistance
of 1  and 300  single-phase ground faults is simulated, and zero-sequence current
and negative-sequence current amplitudes are obtained, as shown in Table 2.8.
When the transition resistance is small, the fault current component of single-
phase grounding is more obvious. According to a comparison of the negative
sequence and zero sequence currents in the dynamic process of accidental over-
load shown in Fig. 2.6, the maximum value of the negative sequence current in
the dynamic process of accidental overload is greater than the negative sequence
current of single-phase grounding via a 300  resistor in the large operation mode
and the value of the negative sequence current of single-phase grounding via a 1 
resistor at the end of this line in the small operation mode. The case of zero-sequence
current is similar. Therefore, it is difficult to distinguish the accidental overload and
single-phase fault triggered by single-phase tripping only from the magnitude of
the fault component, especially for single-phase faults via large transition resistance
grounding or even small resistance single-phase grounding faults in small operation
mode.
(2) It is difficult to identify single-phase tripping and single-phase grounding faults
from the magnitude ratio and phase angle relationship between two fault compo-
nents at the protection installation due to the system operation mode and network
structure.
As shown in Table 2.7, at a Line J single-phase fault, its positive sequence, negative
sequence, and zero-sequence fault components are equal. At the protection installa-
tion, the positive sequence fault components and negative sequence components have
equal amplitudes and phases and zero sequence components and negative sequence
components in the amplitude and phase deviation. The amplitude ratio of nega-
tive sequence current to zero sequence current at the protection installation depends
mainly on the negative sequence (positive sequence) distribution coefficient and
zero sequence distribution coefficient. This distribution coefficient is related to the
fault location the system operation mode. Therefore, it is difficult to determine the
threshold value for differentiating the ratio of negative sequence component to zero
sequence component under asymmetric accidental overload and single-phase faults
due to the operation mode and grid structure.
Similarly, the phase angle relationship among the fault sequence components is
affected by the mode of operation and grid structure. Since the positive sequence
impedance angle of the transmission grid is similar to the zero sequence impedance
angle (both are approximately 80°), it is difficult to distinguish the fault sequence
components from the single-phase fault case when they do not deviate much in phase.
Table 2.8 Single-phase ground fault negative sequence and zero-sequence currents for different operating modes, transition resistances, and fault locations
Fault location (km) Large operation mode Small operation mode
Fault resistance 1  Fault resistance 300  Fault resistance 1  Fault resistance 300 
2.2 Network Analysis of Accidental Overload

|Ia2 | (p.u) |I0 | (p.u) |Ia2 | (p.u) |I0 | (p.u) |Ia2 | (p.u) |I0 | (p.u) |Ia2 | (p.u) |I0 | (p.u)
100 1.86 1.87 0.29 0.29 0.80 0.88 0.23 0.25
200 1.06 0.97 0.21 0.19 0.64 0.58 0.19 0.17
300 0.81 0.62 0.14 0.11 0.58 0.38 0.15 0.10
400 0.92 0.34 0.08 0.03 0.67 0.17 0.12 0.03
53
54 2 Distance Protection Against Overload

Although the additional network in accidental overload is different from the single-
phase fault network, the fault component magnitude, phase angle, and ratio felt by
Line J are highly random due to the uncertainty of the network structure and operation
mode and the small difference in the positive zero sequence impedance angle of the
transmission network. The fault characteristics of single-phase faults greatly vary
due to the transition resistance and fault location and may also hold under accidental
overload.
Therefore, it is difficult to take into account the selectivity and sensitivity prob-
lems in the identification of asymmetrical tide transfer and single-phase faults in
protected areas in accident overload by extracting certain types of fault character-
istics as distinguishing accident overload identification criteria from fault network
analysis. Traditional distance protection based on information about this line cannot
completely solve the identification problem of accidental overload. It is urgent to
introduce new information and processing methods.

2.3 Analysis of the Conditions for the Operation

of the Distance III Section Under Accidental Overload

Due to the complex and variable operation of the system, the asymmetric components
of its dynamic process are also difficult to distinguish from single-phase faults by
network analysis. Unlike faults, accidental overload under certain conditions can
only lead to the action of distance protection section III. Therefore, before carrying
out accidental overload identification and blocking, it is necessary to clarify the
necessary conditions for the protection action caused by accidental overload. The
previous analysis indicates that the action of distance section III under accidental
overload is related to the protection setting and system operation mode. The necessary
conditions for the action of distance section III are analyzed here.

2.3.1 Operating Condition 1: The Line is Heavily Loaded,

and the Equivalent System Power Angle is Stable
at Both Ends

The accidental overload triggering distance III section starts to occur mainly at the
active sending end, i.e., the active power direction flows from the bus to the line
[25]. The power angle characteristic Eq. (2.9) shows that the equivalent system
output active power at the sending end is mainly influenced by the difference in the
system power angle between the two ends δ, system impedance Z M and Z N and line
impedance Z L .
2.3 Analysis of the Conditions for the Operation of the Distance … 55

Fig. 2.11 Accidental post-overload phase ratio results with changing power angle

EM EN
P= sinδ (2.9)
ZM + ZN + ZL

An analysis of the power angle difference before the disturbance of the system
shown in Fig. 2.4 is performed. When the power angle difference increases from 0°
to 90°, after the N-3 disturbance, L14-1 is the first distance from section III. The phase
results are shown in Fig. 2.11. When the power angle difference is greater than 55°,
the grounding relay section III A phase will only enter the action zone after the tidal
shift is stabilized. Therefore, only if the line itself has a large active transmission can
accidental overload “snowball” and cause protection to start.
Accidental overload occurs in the case of system transient instability out-of-step
oscillation because the oscillation line voltage current and measured impedance show
periodic changes, as shown in Fig. 2.12. The currently observed power system out-of-
step oscillation period does not exceed 1.5 s. The continuous time in the action zone
is less than the protection action time and does not exit tripping. Thus, in the case
of transient system instability, no distance III section exit action due to accidental
overload will occur. Note that the distance I and II segments may be misoperated in
oscillation and still requires reasonable oscillation blocking.
The power angle condition of the protection action caused by accidental overload
without out-of-step oscillation is further investigated. As an example, the power
characteristics and power angle variation curves of the full process of accidental

Fig. 2.12 Distance III section-specific phase under system oscillation and accidental overload
56 2 Distance Protection Against Overload

Fig. 2.13 Power characteristic curve and power angle variation curve

overload triggered by a permanent single-phase ground fault outside the zone are
shown in Fig. 2.13.
Based on the extended equal area method, it is known that the generator speed
decreases after the first selected jump of the line. However, the speed is still greater
than the synchronous speed at this time, so the work angle will continue to increase
from δ C1 . If δ C1 ≥ 90°, then the active power will decrease as the work angle increases,
which is not consistent with the increase in active power in accidental overload.
Therefore, the accidental overload that can cause the action of the distance III section
must satisfy the condition of δ C1 < 90° at the moment when the out-of-area fault is
removed.

2.3.2 Operating Condition 2: The Line is Heavily Loaded,

and the Equivalent System Voltage at Both Ends is
Stable

The line transmitted power is not only related to the power angle stability but also
bounded by the voltage stability. As demonstrated by the PV curve in Fig. 2.14,
the voltage destabilization point coincides with the point at which the line trans-
mitted power reaches its maximum value [170]. Since the line active power increases
under accidental overload, the current line transmission power must be less than the
maximum transmission power, i.e., the accidental overload must be in a voltage stable
state.
The substation and system at both ends are Davinan equivalent, as shown in
Fig. 2.15a. When the complex of load impedance is equal to the conjugate complex
of transmission impedance, i.e., |Z L |=|Z thev |, the load obtains the maximum power,
which is at the critical point of voltage instability [60], as shown in Z L2 in Fig. 2.15b.
When the impedance modulus ratio |Z L |/|Z thev |>1, the node voltage is stable, as
shown in Z L1 . When |Z L | ||Z |thev < 1, the node voltage is destabilized, as shown in
2.3 Analysis of the Conditions for the Operation of the Distance … 57

Fig. 2.14 PV curve V 1

2
3

a) Davinan equivalence for b) Voltage stability versus

substation and system impedance modulus ratio

Fig. 2.15 Schematic of the Davinan equivalent system

Z L3 . Thus, the voltage stability can be characterized by the impedance modulus ratio
|Z L |/|Z thev | when an accidental overload occurs with |Z L |/|Z thev | > 1.
In the case of voltage instability, the voltage reduction is accompanied by current
reduction, which generally does not affect the protection action. However, if proper
control measures are not taken, the protection may unreasonably operate under severe
voltage instability or even voltage collapse. [172–175] has investigated how to avoid
the unreasonable operation of protection under voltage instability. This chapter will
not address this topic.

2.3.3 Adjustment Conditions: Protection Installed on Long

Lines and Large Fixed Values

The adjustment of distance section III is mainly affected by the notion that the
protection is installed on long lines. The longer the line is, the larger the distance
protection section III rectification value, the smaller the relay load capacity, and
the easier it is for the protection to start under overload. Unlike the line carrying
capacity, the relay load capacity is limited by the action characteristics of the distance
protection under overload, as shown in Fig. 2.16 for the impedance relay, for example.
58 2 Distance Protection Against Overload

Fig. 2.16 Schematic of jX

impedance relay load
capacity
Z set

ZL
ϕZ
ϕL
R

Where Z set is the distance relay segment III constant, ϕ set is the line positive sequence
impedance angle, and ϕ L is the load power factor angle. At voltage U, the relay load
quantity, i.e., the minimum load S min that triggers the start, can be obtained from the
following equation [8]:

U2
Smin = (2.10)
Z set cos(ϕset − ϕL )

Taking a 500 kV line as an example, the line positive sequence impedance angle
is 85°, and the load impedance angle is set to 30° at 0.85 U N [63].. When the Z set
magnitude increases from 50  to 300  (1  corresponds to an approximately
1.2 km line section III rectification value without considering the crosstalk case), a
start is triggered. The minimum load S min to be moved is shown in Fig. 2.17.
When the Z set amplitude values are 50  (approximately 60 km line section III
rectified value), 100  (approximately 120 km line section III rectified value) and
200  (approximately 240 km line section III rectified value), a start is triggered. The
minimum loads to be moved S min are 7200 MVA, 3566 MVA and 1774 MVA. The
line transmission capacity of 7200 MVA has exceeded the 500 kV line transmission
capacity and transient limit and is unlikely to occur, while 3566 MVA and 1774 MVA
are possible under large operation mode.

Fig. 2.17 M relay load

quantity versus rectified
impedance magnitude
2.3 Analysis of the Conditions for the Operation of the Distance … 59

Therefore, the longer the protection line is, the higher the probability of protection
activation under overload, and the protection action study under accidental overload
only needs to be carried out for long lines. In addition, the longer the line is, the
less obvious the fault characteristics of the backup protection zone, especially the
far backup protection zone, and the more difficult it is to identify.
In summary, on the premise of measuring the Davinan equivalent impedance, it
is possible to classify the protection actions and thus achieve the identification of
accidental overloads and single-phase faults. However, it is difficult to obtain the
Davinan equivalent impedance under the traditional framework of single interval
information of protection, and the sharing of electrical information in the substation
area is essential for solving this problem.

2.3.4 Summary

In this chapter, the dynamic action process of the distance III section under accidental
overload is analyzed in the time domain; the necessary network analysis of the
accidental overload is carried out; and in particular, the necessary conditions for
the distance III section to be exceeded under accidental overload are analyzed. The
following conclusions are obtained:
(1) Accidental overload contains a complex dynamic process, and the existing solu-
tion method is still defective. To avoid the unreasonable action of the protec-
tion in the dynamic process, the identification problem of accidental overload
should be extended from the judgment on a single time section to the dynamic
continuous judgment in the time domain. The use of multiple disturbance time
information is feasible and necessary.
(2) The essential reason for the distance III section to exceed under accidental
overload is the contradiction between protection selectivity and sensitivity. For
station domain accidental overload, blocking can be directly achieved by station
domain information. However, solving the identification of nonstation domain
accidental overload is difficult by traditional fault network analysis.
(3) Whether distance section III is exceeded under accidental overload is determined
by a combination of the protection adjustment conditions and system operating
conditions.

(1) The line is heavily loaded; the equivalent system power angle at both ends is
stable; and the node voltage is stable.
(2) Protection is installed on long lines with large fixed values.

The scope of the study can be narrowed based on the protection rectification
value. Based on the system operation mode, it can be predicted whether accidental
overload may trigger protection activation. Based on the station domain information
for the back side of the line, i.e., the system side, Davinan equivalence is performed
60 2 Distance Protection Against Overload

to judge the system operation mode, which is a feasible method for solving the
accidental overload identification.
4) In IEC61850-based intelligent substations, more complete identification and
blocking of accidental overloads can be achieved based on multiple distur-
bance time information, station domain switch information and station domain
electrical information.

2.4 Identification Methods for Accidental Overload

2.4.1 Division of the Incident Overload Action Domain

and Protection Action Domain

The system operating conditions and protection rectification conditions for protec-
tion action under accidental overload are presented in Sect. 2.3 as a new idea to
solve the identification problem. In this section, the Davinan equivalent impedance
is chosen to characterize the system operation mode, and the relationship between the
amplitude and phase angle of the Davinan equivalent impedance and the measured
impedance and the relationship with the protection rectification impedance are further
investigated.
1. Magnitude of the measured impedance versus the Davinan equivalent impedance
In the three-node system shown in Fig. 2.18, the sending nodes M and P jointly deliver
power to the receiving end N. R is the distance III segment protection installed at the
head of Line J. The diagonal area is the far protected segment of protection R. The
total impedance of this segment and the lower segment of the line is represented by
ZLj.
Line J is the ground distance protection measurement impedance Z RG .

U̇ Ė M − U̇thev.
Z R = = (2.11)
˙I + 3K0 I˙0 I˙ + 3K0 I˙0

where Ė M is the M-side system phase potential, U̇thev. is the system equivalent
impedance phase voltage drop, K0 is the line zero sequence compensation factor, and

Fig. 2.18 Three-node system network

2.4 Identification Methods for Accidental Overload 61

Fig. 2.19 Schematic of the jX

measured impedance and
compensated Davinan
equivalent impedance Z R1
Z R2
Z R3

R
Z thev

I˙ and I˙0 are the phase sequence current and zero sequence current, respectively, at
the protection installation.
Zthev1 and Zthev0 are the Davinan positive-sequence equivalent impedance and
zero-sequence equivalent impedance, which are equivalent to ZM1 and ZM0 for Line
J. Let Kt0 = (Zthev0 −Zthev1 )/3Zthev1 , and define the compensated Davinan equivalent
impedance as

 U̇thev. I˙0 (3K0 − 3Kt0 )

Z thev = = [1 − ]Z thev1 (2.12)
˙IA + 3K0 I˙0 I˙A + 3K0 I˙0

Due to the small magnitude of the zero-sequence current under accidental over-
load, the difference between Z’thev and Z R does not differ much from the Z thev1
amplitude. After compensation, Z’thev and Z R are in the same impedance plane, as
shown in Fig. 2.19.
Similarly, the measured impedance Z R at the interphase distance protection is

U̇ Ė M − U̇thev. Ė M

Z R = = = − Z thev1 (2.13)
˙I I˙ I˙

In the phase-to-phase distance protection impedance plane, Z thev1 and Z R also
satisfy the relationship shown in Fig. 2.19.
According to the previous analysis, only the node voltage stability under acci-
dental overload may protect the load impedance more than the Davinan equivalent
impedance. From the distance protection impedance plane, there should be |Z  R
|/|Z’thev |, and the |Z  R |/|Z thev1 | ratio is greater than the threshold value OTH , where
Z  R is the parallel value of the measured impedance to ground for all outgoing lines.
Z  R is the parallel value of the measured impedance between two phases for all
outgoing lines.

2. Phase angle relationship between measured impedance and Davinan equivalent

impedance

Define the ground distance protection electromotive force measurement impedance

ZE as
62 2 Distance Protection Against Overload

Ė M 
Z E = = Z R + Z thev (2.14)
˙I + 3K0 I˙0

In the system in Fig. 2.1, when Line I is open in three phases causing a three-phase
tidal shift, the phase angle of Z E is given by Ė M , I˙M is determined by the power
angle characteristic Ė M , and the angle Ė N is less than 90°, i.e., Ė M and I˙M The
maximum angle is 45° + (90° − ϕ L1 ).
When considering the single-phase tidal current transfer caused by a single-phase
trip, the Z E substate term is determined by the load current before the disturbance
of Line J I˙j [0]
and the current transferred from Line I is I˙ ji1 + I˙ ji2 + (3K0 + 1) I˙
ji0 . Disregarding the difference in electric potential magnitude, we have

⎧ π−δMN
⎪
⎪ ˙j
[0] Ė M − Ė N 2 Ė M sin(δMN /2)ej( 2 )
⎪
⎪ I = +Z +Z
= +Z +Z
⎪
⎪
Z M1 Lj1 N1 Z M1 Lj1 N1
⎪
⎪ I˙ji1 = (Z M1 +Z Lj1 +Z Z P1 Z 00 ˙iA
I [0]
⎨ P1 )(Z 11 +2Z 00 )
I˙ji2 = (Z M1 +Z Lj1 +Z Z P1 Z 00
)(Z +Z
I˙[0]
) iA (2.15)
⎪
⎪ P1 11 11 Z 00
⎪
⎪ I˙ji0 = (Z M0 +Z Lj0 +Z Z 11 Z P0
I˙[0]
⎪
⎪ P0 )(Z 11 +2Z 00 ) iA
⎪
⎪ π−δPN
⎩ I˙[0] = Ė P − Ė N
= 2 Ė P sin(δPN /2)ej( 2 )
i Z P1 +Z Li1 +Z N1 Z P1 +Z Li1 +Z N1

where Z11 = Z Li1 + Z P1 + (Z M1 + Z N1 )Z N1 /(Z M1 + Z N1 + Z Lj1) and Z 00 = Z Li0 +

Z P0 + (Z M0 + Z N0 )Z N0 /(Z M0 + Z N0 + Z Lj0 ). δ MN and δ PN represent Ė M with Ė N
and Ė P with Ė N , respectively. There are δ MN < 90° and δ PN < 90° during the fault
phase outside the accidental overload zone, i.e., the pretidal shift. When δ MN = 90°,
the Ė M angle with the maximum angle I˙M is 45° + (90° − ϕ L1 ), where ϕ L1 is the
line positive sequence impedance angle. Similarly, the Ė N with a negative I˙N angle
is also the maximum angle of 45° + (90° − ϕ L1 ). Because the transmission grid
equivalent system and the difference between the line positive sequence impedance
angle and the zero sequence impedance angle are small, the current transferred to
Line J and the predisturbance Line I load current angle are small. Additionally, since
Line I and Line J have a common receiving end node, the angle ZE has a maximum
value of 45° + (90° − ϕ L1 ) when both Line I and Line J send ends are operating in
the static stability limit.
To define the phase distance protection electromotive force measurement
impedance ZE, refer to the following equation. As in the previous analysis, the
maximum phase angle of ZE is 45° + (90° − ϕ L1 ).

Ė M
Z E = = Z R + Z thev1 (2.16)
I

Compensating for the line impedance angle, there is

 ◦
arg(Z E ) = arg(Z E eϕL1 −90 ) < 45◦
 ◦ (2.17)
arg(Z E ) = arg(Z E eϕL1 −90 ) < 45◦
2.4 Identification Methods for Accidental Overload 63

Fig. 2.20 Schematic of the jX

rectified impedance,
measured impedance, and
compensated Davinan
Zset
equivalent impedance ZE
ZE

90
Z thev_1
L1
ZR _1
Z thev_2
Emax
L1
ZR _2

3. Magnitude of the rectified impedance versus the Davinan equivalent impedance

In the case of impedance relays, for example, the protective start is subject to the
following specific amplitude conditions:
     
 Z set 2 >  Z set 2 − Z R  (2.18)

Combined with the previous analysis, the accidental overload situation can only
satisfy Eq. (2.18) if Eqs. (2.14) and (2.17) are satisfied. zE , z’E , ZR , Z thev and
the phase relationship of the protection rectification impedance Z set are shown in
Fig. 2.20.
The rectified impedance Z set is large to enable distance section III to start in the
case of accidental overload [9]; thus, the |Z’thev |/|Z set | ratio is set to the value OMAX .
The following conditions are set for the calculations; the results are shown in Table
2.9.
1) Z set = 1∠85° p.u.
2) Z E : moving along the ray of maximum impedance angle, i.e., ϕ Emax = 50° (ϕ L1
takes 85°).
3) Z’thev : values taken in the region of the circle in the direction below Z E , where
the phase angle changes from 81° to 89°.
4) Z R : According to Z’thev , determine Z R that satisfies Eq. (2.17), as shown in
Fig. 2.20. Correspondingly, Z‘ thev_1 ↔ Z R_1 , Z‘ thev_2 ↔ Z R_2 , and OTH is set
to 2.
Table 2.9 shows that the phase angle Z’thev has a minimal effect on OMAX , and
for the M relay, OMAX can be set to 0.3. Therefore, the rectification condition for

Table 2.9 Different Davinan equivalent impedance angles for the |Z thev |/|Zset |extreme value

arg(Z thev ) 81° 82° 83° 84° 85° 86° 87° 88° 89°
OMAX 0.24 0.23 0.22 0.21 0.21 0.21 0.21 0.20 0.20
64 2 Distance Protection Against Overload

accidental overload triggering distance III starting can be quantified as

  
Grounded: Z thev  |Z set | < OMAX
 (2.19)
Interphase:|Z thev1 | |Z set | < OMAX

Without satisfying Eq. (2.19), it is not possible to enter the distance III section
action zone, even if an accidental overload occurs. Equation (2.19) also qualifies the
relationship between the equivalent impedance of the sending end system and the
rectified impedance, which can be utilized for further identification.

4. Delineation of protection start-up domains

Based on the above analysis, the Overload Start Index (OSI) is proposed for distance
protection in section III. Only if the OSI index falls into the accidental overload start
domain can the current distance protection section III start be caused by accidental
overload and can the accidental overload caused by distance protection section III
start be excluded for the fault start domain. For the grounding and phase-to-phase
distance protection III segments, the OSI conditions for the accidental overload
starting domain are shown as follows:
⎧      
⎨ O S I1 =  Z thev / Z  < O M AX
 set 
Phase to ground: O S I2 =  Z  R / Z thev  < OTH
⎩  
O S I3 = imag(Z E )/real(Z E )<1
⎧
⎨ O S I1 = |Z thevl |/|Z  set | < OMAX
Phase to phase : O S I2 =  Z  Rφφ /|Z thevl | < OTH (2.20)
⎩  
O S I3 = imag(Z E )/real(Z E )<1

Based on the description of the accidental overload process in this chapter and in
conjunction with the initiation of relay protection, this section proposes identification
and blocking schemes for nonstation domain accidental overloads caused by three
types of events: out-of-zone faults, out-of-zone single-phase trips and out-of-zone
three-phase openings.

2.4.2 Identification of Dynamic Processes of Accidental

Overload

1. Identification of accidental overloads during the out-of-area fault phase

The distance backup protection is divided into protected zones by rectifying the fixed
values. If the fixed value is not properly set, it makes the out-of-zone fault fall into the
fixed value area, which is difficult to distinguish from the principle. To essentially
solve the problem of false start of out-of-area fault protection, it is necessary to
reasonably adjust the protection setting value.
2.4 Identification Methods for Accidental Overload 65

Table 2.10 Maximum excision time for distal and proximal faults on system lines by voltage level
220 kV 330 kV 500 kV 750 kV
Near end maximum time (ms) 120 82 92 55.25
Maximum remote time (ms) 100 80 88 54

Due to the correct action of the main protection, out-of-zone faults will be removed
in a short time, as shown in Table 2.10 [69]. Additionally, because the backup protec-
tion has a certain action delay, the misjudgment caused by such faults can be avoided
in terms of time. Taking the 500 kV system as an example, according to Table 2.10,
we can choose 100–150 ms as avoidance of the out-of-zone fault time.

2. Identification of single-phase tidal shifts

Based on OSI conditions, the accidental overload starting domain is defined in terms
of system operating conditions and rectification conditions. However, influenced by
conditions such as single-phase fault transition resistance, there is a mixed region of
single-phase faults and single-phase tidal shifts in the accidental overload starting
domain. The single-phase ground fault has a high probability of occurrence, and the
single-phase tidal current transfer after the fault phase selection jump may cause
other line distance III actions in the parallel section, whose identification problem is
a difficult and key point. The following section will further distinguish single-phase
faults from single-phase tidal shifts on the basis of the OSI.

1) Ratio of zero sequence current amplitude to negative sequence current amplitude

Under a single-phase open circuit that does not constitute an earth circuit, the zero-
sequence current is usually less than the negative-sequence current, and the zero-
sequence current and negative-sequence current are equal in magnitude at a single-
phase earth fault. However, limited by the system operation mode and network struc-
ture, it is difficult to define by network analysis alone. According to the conditions of
the previous analysis, further study of Line J protection at the A-phase tidal shift and
A-phase fault reveals that the ratio of zero sequence current magnitude to negative
sequence current magnitude I t0 /I t2 and I f0 /I f2 is expressed as
⎧  2 
 Z Z Z Z + Z +Z 
⎨ It0
=  Z 112 Z P0 Z N0 ( Z M1 + Z Lj1 +Z N1 ) 
It2
 00 P1 N1 ( M0 Lj0 N0 )  (2.21)
⎩  (1−k)Z +Z ]( Z M1 + Z Lj1 +Z N1 
)
If0
=  [(1−k)Z Lj0 +Z N0 
N1 ]( Z M0 + Z Lj0 +Z N0 )
 
If2 [ Lj1

where k is the proportion of the fault location in the line, Z’N1 = (Z Li1 + Z P1 )Z N1 /(Z M1
+ Z N1 + Z Lj1 ) and Z‘N0 = (Z Li0 + Z P0 )Z N0 /(Z M0 + Z N0 + Z Lj0 ). From Eq. (2.21),
if the system equivalent impedance tends to 0, I f0 /I f2 tends to 1 under a single-phase
fault, and I t0 /I t2 under a single-phase tidal shift depends on the ratio of line zero-
sequence impedance to the positive-sequence impedance [13], which is generally
0.2–0.3.
66 2 Distance Protection Against Overload

When the system impedance cannot be disregarded, according to the accident

overload protection action conditions, the system equivalent impedance assumes a
value to set reasonable constraints. After the overload Line J, the power P2 must
be greater than the distance III section into the action area of the apparent power to
make the distance III section start, and the line open section impedance increases. P2
must be less than the total power transferred before the transfer P1 , and the power
factor is set to 0.9.
 0.9×(0.85U N )2
P2 = Z set cos(ϕset −ϕload )
< P1 
2
E2 (2.22)
P1  < (Z M1 + Z Lj1 )//(ZNP1 + Z Li1 )+Z N1 < Z M1 //Z P1N +Z N1
E

where Z M1 //Z P1 denotes the parallel impedance of Z M1 and Z P1 . The load power
factor angle ϕ load is set to 30°, ϕ set is set to 85°, and E N is set to 1.1 times U N , from
which the following relationship can be derived

Z M1 //Z P1 + Z N1 cos(ϕset − ϕL )E N2
< <1 (2.23)
Z set 0.9 × (0.85UN )2

If the voltage at the receiving end is destabilized, the third line of defense is
employed for low-voltage load shedding to relieve overloads. Therefore, the main
consideration when calculating the protection action is the voltage stability at the
receiving end, with [12]

(Z M1 +Z Lj1 )//(Z P1 + Z Li1 ) < Z N1 (2.24)

From Eqs. (2.23) and (2.24), the range of Z P1 and Z N1 can be determined. Consid-
ering the line impedance and rectified impedance, the constraints under accidental
overload are shown in Table 2.11.
In the case of a single fault, the values of Z M1 , Z N1 and Z P1 are set from 0.05 to
1.0; the fault location is set within the range of 0.1 to 1.0 times the total length of
this section and the next section of the line; other parameters are set with reference
to Table 2.11.

Table 2.11 System and line constraints under incidental overload

Parameter Constraint condition
Line impedance Z Lj1 : 1 Z Li1 : 0.1–1 Z Lj0 : 3–5 Z Li0 : 0.3–5
Setting impedance Zset :0.8–1.5Z Lj1 (current and lower level line impedance)
System equivalent impedance Z M1 : 0.05–0.45 Z N1 and Z P1 : Refer to
Formulas (2.23) and (2.24)
Z M0 : 0.025–0.9 Z N0 and Z P0 : 0.5–3 times
positive sequence value
2.4 Identification Methods for Accidental Overload 67

Table 2.12 Range of the negative sequence current amplitude to the zero sequence current
amplitude ratio
Negative sequence current to zero sequence Minimum value Maximum Average value
current ratio
I t0 /I t2 0.025 0.43 0.095
I f0 /I f2 0.39 1.08 0.91
I t0 /I t2 (Unconstrained) 0.013 0.66 0.12

The ranges of the I t0 /I t2 and I f0 /I f2 ratios derived from the above analysis are
compared with the range of I t0 /I t2 without the constraints obtained from the previous
analysis in Table 2.12.
As shown in Table 2.12, without OSI-related constraints, the mixing zone between
the I t2 /I t0 and I f0 /I f2 range of values reaches 0.27. After considering the constraints,
the mixing zone is reduced to 0.04. Ensuring reliable operation under a distance III
segment fault, let OD = 0.45.

(1) Protection installation at I 0 /I 2 < OD for a single-phase tidal shift.

(2) Protection installation I 0 /I 2 ≥ OD : for single-phase faults and partial single-
phase tidal shifts, further identification is needed.

2) Dynamic identification based on protection action conversion

The dynamic analysis indicates that out-of-zone non-full-phase operation will be

converted to three-phase permanent tripping after a reclosing failure. Using this
multidisturbance time domain information, the identification of the accidental over-
load will be converted from the protection single-phase start domain to the protec-
tion three-phase start domain, as shown in Fig. 2.21. Under the three-phase starting
domain, the accidental overload is more distinguishable from the fault. The dynamic
identification timing is considered here.
Given the high correct rate of transmission network protection action and reclosing
action, the faulty element is reliably removed from the most severe case, considering
the time of overload line protection in the single-phase starting domain: the maximum
time of the first fault removal is set to 0.09 s [60]; the single-phase reclosing cycle
(including breaker closing time) is set to a maximum of 1.5 s; and the maximum time
of accelerated opening after reclosing is set to 0.2 s. The single-phase asymmetric

Fig. 2.21 Protection start

domain transition
68 2 Distance Protection Against Overload

component is presented. if multiple disturbances are considered, the single start time
of protection under accidental overload will be longer. If multiple disturbances are
considered, the single start time of protection under accidental overload will be
longer.
The criteria in the previous section have been able to distinguish between single-
phase ground faults and single-phase tidal shifts via small resistances, which need to
be further identified by this subsection as single-phase faults and single-phase tidal
shifts via large resistances. Since single-phase faults via large resistance grounding
do not have a significant impact on system stability, an appropriate delay in line
removal is allowed. Therefore, this chapter appropriately sacrifices rapidity to ensure
the selectivity and sensitivity of the protection and introduces a dynamic identification
method with additional delay: a 1.0–2.0 s delay is added to the 1.5 s action time of
the grounding distance III section to avoid the single-phase tidal shift overload in the
most unfavorable situation and further identification in the protection three-phase
starting domain.

3) Out-of-zone single-phase trip identification based on the zero sequence negative

sequence voltage ratio

When a single-phase ground fault occurs outside the zone, if the transition resis-
tance is large, it may result in insignificant impedance angle changes that cannot be
identified by the impedance angle discriminator module. A method of the negative
sequence zero sequence voltage ratio is proposed to identify the single-phase trip-
ping process. The method is independent of the transition resistance and can reliably
identify out-of-zone single-phase tripping after protection activation.
Schematic of line relationships (Figs. 2.22 and 2.23).
L PM is the nonfaulty line and the protection installation line under study. L MN is
the far backup line for the protection and the fault line under study.

1) In the case of a single-phase short circuit, the line diagram is shown as follows:
. .
In the figure, IAf is the fault current of phase A at the fault point, IBf is the fault
.
current of phase B at the fault point, ICf is the fault current of phase C at the fault
point, and Rf is the fault resistance.
Boundary conditions.
. . . .
I = I, I = I = 0 (2.25)
Af f Bf Cf

Fig. 2.22 System schematic P M N

LPM
LMN
2.4 Identification Methods for Accidental Overload 69

Fig. 2.23 Schematic of a

single-phase short circuit line

. . .
I Af I Bf I Cf
Rf

Converting to the boundary conditions for positive and negative zero-order

networks yields

If
I1 = I2 = I0 = (2.26)
3
The negative order equivalent network is (Figs. 2.24 and 2.25)
where Z M is the system equivalent negative sequence impedance to the left of bus
M, Z L is the line negative sequence impedance to the left of the faulty line fault, Z R
is the line negative sequence impedance to the right of the faulty line, and Z N is the
system equivalent negative sequence impedance to the right of bus N.
where UM2 is the voltage measurement of bus M at the left end of the faulty line, and
the negative sequence voltage value of the opposite bus M can be calculated from

Fig. 2.24 Negative UM 2

sequence equivalent circuit

ZM ZL ZR ZN
I2

Fig. 2.25 Zero-sequence UM 0

equivalent circuit

ZM 0 ZL 0 ZR 0 ZN 0
I0
70 2 Distance Protection Against Overload

the single end voltage and current measurement, as well as from the line parameters
in the event of a remote backup line fault, i.e., UM2 .
Theoretical calculation. The negative sequence voltage is given by

Z M (Z R + Z N ) Z M (Z R + Z N ) If
UM2 = − I2 = − (2.27)
ZM + ZL + ZR + ZN ZM + ZL + ZR + ZN 3

Definition

Z all = Z M + Z L + Z R + Z N
Z Lall = Z M + Z L
Z Rall = Z N + Z R (2.28)

These equations can be simplified to

Z M Z Rall If
UM2 = − (2.29)
Z all 3

The zero-order network is similar to.

Theoretical calculation. The zero-sequence voltage is given by

Z M0 (Z R0 + Z N0 ) If
UM0 = − (2.30)
Z M0 + Z L0 + Z R0 + Z N0 3

Definition

Z all0 = Z M0 + Z L0 + Z R0 + Z N0
Z Lall0 = Z M0 + Z L0
Z Rall0 = Z N0 + Z R0 (2.31)

These equations can be simplified to

Z M0 Z Rall0 If
UM0 = − (2.32)
Z all0 3

The ratio of negative sequence voltage amplitude to zero sequence voltage

amplitude is
   
 UM2   Z M Z Rall Z all0 
 =  (2.33)
U   Z Z 
M0 M0 Rall0 Z all

Since the PM line and P bus backside system impedance ratios do not vary with
single-phase tripping and single-phase short circuits,
2.4 Identification Methods for Accidental Overload 71

A I AM 0 I AN 0

B UBM 0 UBN 0

C UCM 0 UCN 0

Fig. 2.26 Schematic of a single-phase tripping line

   
 UP2   
  = C  Z M Z Rall Z all0  (2.34)
U  Z Z 
P0 M0 Rall0 Z all

2) When a single phase is tripped, the line diagram is shown as follows (Fig. 2.26):
Boundary conditions [107, 108]:
. . . .
U = U =U =U =0 (2.35)
BM CM BN CN

Converting to boundary conditions for positive and negative zero-order networks

yields:

U1M = U2M = U0M , U1N = U2N = U0N (2.36)

The negative order equivalent network is (Fig. 2.27)

Theoretical calculation. The negative sequence voltage is given by

Z M (U2M + U2N )
UM2 = (2.37)
ZM + ZL + ZR + ZN

This expression can be simplified to

Z M (U2M + U2N )
UM2 = (2.38)
Z all

UM 2
U 2M U2N

ZM ZL ZR ZN

Fig. 2.27 Negative sequence equivalent circuit

72 2 Distance Protection Against Overload

UM 0
U0M U0N

ZM 0 ZL 0 ZR 0 ZN 0

Fig. 2.28 Zero-sequence equivalent circuit

The zero-order network is (Fig. 2.28)

Theoretical calculation. The zero-sequence voltage is given by

Z M0 (U0M + U0N )
UM0 = (2.39)
Z M0 + Z L0 + Z R0 + Z N0

This expression can be simplified to

Z M0 (U0M + U0N )
UM0 = (2.40)
Z all0

The ratio of negative sequence voltage to zero sequence voltage is

   
 UM2   Z M Z all0 
 =  (2.41)
U   Z Z 
M0 M0 all

Similar to the single-phase short circuit analysis, the p bus voltage relationship is
expressed as
   
 UP2   
  = C  Z M Z all0  (2.42)
U  Z Z 
P0 M0 all

From a comparison of the negative sequence zero sequence voltage ratio in the
two cases, we determine that the ratio of negative sequence voltage magnitude to zero
sequence voltage magnitude in the case of a single-phase short circuit is smaller than
that in the case of single-phase tripping. We also discover that the magnitude ratio
size does not change with the transition resistance value of the fault. The ratio of the
two amplitude ratios is |Z Rall /Z Rall0 . The ratio size depends on the ratio of negative
sequence impedance to zero sequence impedance for the right-hand impedance. The
method is not affected by the transition resistance, and the ratio hardly changes
when the fault resistance changes, allowing reliable identification of out-of-zone
single-phase tripping.
2.5 Distance III Accidental Overload Blocking Scheme 73

Fig. 2.29 Interphase and phase-to-ground fault measurement impedance

Judgment.

(1) Device start-up.

(2) The results are consistent.
(3) If the criterion of Eq. (2.43) is satisfied, a single-phase trip is considered.
   
 UP2   
  > 1.05 UP2  (2.43)
U  U 
P0 2 P0 1

3. Identification of three-phase tidal shifts

Based on the arc resistance characteristics of the phase-to-phase and phase-to-ground
faults, the phase-to-phase measured impedance distribution is obtained, as shown in
the diagonal section of Fig. 2.22. For the phase distance III section, the three-phase
tidal current transfer is directly distinguished from the phase-to-phase and phase-to-
ground faults by the OSI3 < 1 condition. For the ground distance III section, when
all three phases are at start-up, a three-phase tidal shift can be identified as long as
the OSI is satisfied (Fig. 2.29).

4. Full process identification method for accidental overload

The logic for identifying single-phase tidal shifts with three-phase currents under
accidental overload is shown in Fig. 2.30.

2.5 Distance III Accidental Overload Blocking Scheme

2.5.1 Description of the Lockout Program

1. Prerequisites for accidental overload blocking

1) Relay section III has at least one phase started, and the main protection is not
active.
74 2 Distance Protection Against Overload

Fig. 2.30 Identification

logic for
single-phase/three-phase
tidal shift

The start here requires both the overall start of the device (including phase current
surge, negative sequence current, zero sequence current, switching variables and
remote trip start) and the entry of a phase into the action zone of relay section III.
The existing main protection can export trip information within 10–20 ms of the
protection start, and based on the GOOSE switch status message transmitted by the
network, the backup protection can be quickly informed of the main protection action
and restored after the circuit breaker is opened.
(2) The line is in a three-phase operation before starting
When the line is not fully operational in the asymmetric state, it will produce a
negative sequence, zero sequence fault component if the accident overload is difficult
to distinguish. In general, the time of the non-full-phase operation is less than the
action time of grounding distance relay section III. To avoid unreasonable action,
one can wait until the end of the non-full-phase operation before making a judgment.
For the non-full-phase operation caused by line break, the longitudinal protection or
zero sequence current backup protection will normally jump open the three phases.
In this case, the implementation of the accidental overload blocking scheme is no
longer meaningful.
2. Specific requirements of the incidental overload lockout program
(1) For grounding distance relay section III, when an earth fault occurs (including
single-phase grounding, two-phase grounding, and three-phase grounding),
open the overload blocking, make a protection judgment, and cooperate with
the main protection to remove the faulty line. For accidental overload and
voltage collapse, block the protection to ensure that it does not go beyond the
non-misoperation.
(2) For phase distance relay section III, when a phase fault occurs (including two-
phase, two-phase ground, and three-phase fault), open the overload blocking,
make a protection judgment, and cooperate with the main protection to remove
the faulty line. For accidental overload and voltage collapse, block the protection
to ensure that it does not go beyond the non-misoperation.
2.5 Distance III Accidental Overload Blocking Scheme 75

Table 2.13 Accidental

Single-phase Phase-to-phase
overload identification
grounding in the area grounding in the
methods
area
Single-phase Delay avoidance + Protection
grounding outside reasonable setting criterion
the area
Phase-to-phase Protection criterion Delay avoidance
grounding outside + reasonable
the area setting
Single-phase power OSI Protection
flow transfer criterion
Three-phase power Protection criterion Phase-to-phase
flow transfer cosine voltage
criterion

According to Sect. 4, the method of identifying the various phases of accidental

overload with different types of in-zone faults is collated and shown in Table 2.13.

3. Accidental overload blocking scheme for distance protection section III

Phase distance protection section III can be blocked by the phase cosine voltage
condition, and the out-of-zone fault phase under the dynamic process is considered
for delayed judgment. Blocking of ground distance protection section III involves
the conversion of the protection starting phase, so the protection starting situation
needs to be fully considered, especially for the analysis of various types of compound
faults.
(1) Complex fault conditions containing interphase or phase-to-ground within the
zone can be identified based on the phase-to-phase cosine voltage constraints.
The voltage cosine criterion can well distinguish between intra-zone faults and
overloads from a voltage plane perspective. However, the assumption of this
analysis method: equal magnitude of system potential on both sides is approx-
imately valid for in-zone faults and slow overloads [91–92]. In the case of out-
of-zone faults and accidental overloads, there is no guarantee that the system
potential will remain stable near the rated voltage in the short term due to faults
or severe disturbances in the receiving system. Out-of-zone faults and over-
loads are difficult to distinguish in terms of information at a given moment. A
relatively minor out-of-zone fault appears to the system to be similar to a large
load that is being connected at a nonload point. To distinguish between out-
of-zone faults and accidental overloads, the protection needs to use dynamic
process analysis to identify accidental overloads and dynamically adjust the
phase-to-phase cosine voltage criterion to improve the reliability of the distance
protection.

By identifying accidental overload by the aforementioned method, the fixed value

of Eq. (2.2) is modified. If no accidental overload is detected, then the original
76 2 Distance Protection Against Overload

overload criterion is kept unchanged, U cos ϕ < mUNN , and m is set to 0.5.
If accidental overload is detected, then the system may be in an accidental overload
condition, and the assumption condition of the original voltage cosine value criterion
is not valid. The criterion should be further set to U cos ϕ < mUNN , and m is set
to 0.35 to ensure that the system in relatively minor out-of-zone faults and overload
conditions blocking protection will not operate to maintain power transfer and open
protection to correctly trip the line in the case of system destabilization and severe
faults.

(2) In the case of compound faults that contain only single-phase faults in the zone,
the blocking scheme is proposed based on the number of phases starting in
section III of the ground distance relay, considering all types of compound
faults for their analysis.

(1) Multiple fault situations containing only single-phase faults in the zone (no two-
phase or three-phase faults in the zone) are single-phase faults in one phase at
different fault points in the zone, single-phase faults in two phases at different
fault points in the zone, single-phase faults in three phases at different fault
points in the zone, and single-phase faults in the zone + faults outside the zone.
The above situations may occur simultaneously or in succession, and the relay
may start one, two or even three phases.
(2) For multiple, same-phase, individual single-phase ground faults in the zone,
only one phase relay is started. Due to the superposition of three same-phase
faults, the zero-sequence current and negative-sequence current are large and
can be opened by the current asymmetry degree.
(3) Three separate, simultaneous, single-phase faults exit at different locations
in the zone. According to the principle of superposition, the additional power
supply for three different categories of faults has a partly offsetting role, which
normally makes it difficult to cause a three-phase relay to start. Only when the
distance between different fault points is short and the transition resistance is
small does the situation at this time tend to be a three-phase short circuit, which
can trigger a three-phase start. This situation is characterized by a small inter-
phase cosine voltage and small current asymmetry, which satisfies the interphase
cosine voltage limitation condition.
(4) Protection of two phases starting under single-phase faults of different phases at
multiple fault points in the zone are corrected for reliable starting in accordance
with the interphase cosine voltage limit.
(5) When multiple fault points inside and outside the zone have different single-
phase faults, only one phase start can be treated as a single-phase fault depending
on the mode of operation, as subsequently discussed.
(6) If a single-phase fault within the zone is superimposed on a single-phase fault
or single-phase trip outside the same phase, it will increase the zero sequence
current and other fault characteristics, which can better ensure that protection
starts. If a single-phase fault within the zone is superimposed on a non-co-phase
fault or non-full-phase operation outside the zone, the protection will start two
2.5 Distance III Accidental Overload Blocking Scheme 77

phases. In this case, the interphase cosine start condition may not be satisfied.
In this case, the Imax phase should be chosen to keep the start and revert to
the other one-phase. If the single-phase fault in the zone is superimposed on the
two-phase fault outside the zone, the protection will start three phases. In this
case, if the interphase cosine voltage limitation condition is not satisfied, it will
be restored. However, due to the short duration of the two-phase fault outside
the zone, the protection will be converted to a single-phase start after a short
period of time.

4. Blocking criterion for ground distance protection Section III

Since the distance protection section III is a remote backup protection with a three-
phase permanent trip, no phase selection is involved. As a result of the above findings
and in conjunction with the start-up of the grounded distance protection III, the
following criteria can be proposed.

(1) If all three phases are activated, a three-phase fault or multiple repetition fault is
opened by means of a phase-to-phase cosine voltage limiting condition, which is
not met by a three-phase reversion of the relay to reliably block the symmetrical
accidental overload.
(2) If there are two phases starting, the cosine voltage between the same phase to
identify the phase fault or multirepeat fault does not meet this condition. Next,
select the Imax phase into a single-phase starting judgment and revert to the
other..
(3) If there is only a single-phase start, further classification is required to open
according to the OOI indicator.

Open according to the different protection starts, refer to Table 2.14, where ZG
*1 means that one phase is connected to ground distance protection section III starts,
and so on. According to the superposition theorem, the additional power supply for
single-phase faults of different phases acts as a partial offset and can only trigger a
two-phase or three-phase Z start if the distance between two different fault points
is close and the transition resistance G is small. This case does not satisfy the OSI
condition [55], which can open the protection.
In addition, to avoid the operation of the distance protection section III under
voltage collapse, the voltage dip speed can be identified [161] when the three phases
at start-up open protection, satisfying the following equation.

|dU/dU dt|max >VTH (2.44)

78 2 Distance Protection Against Overload

Table 2.14 Open conditions for compound faults and different protection start-ups
Location and type of complex fault Protection start Open conditions
Multiple single-phase faults in the area ZG *1 OSI
Multiple single-phase faults of different phases ZG *1 OSI
in the area ZG *2 Choose I max phase + OSI
ZG *3 Cosine voltage
Single-phase fault in the area + in-phase fault or ZG *1 OSI
trip outside the area
Single-phase grounding in the area + grounding ZG *1 OSI
or tripping of other phases outside the area ZG *2 Choose I max phase + OSI
Single-phase grounding in the area + ZG *1 OSI
phase-to-phase fault outside the area ZG *2 Choose I max phase + OSI
ZG *3 Delay avoidance + OSI

2.5.2 Lockout Logic

Based on the previous analysis, overload blocking is proposed for distance segment
III. The blocking logic for grounding relay segment III and phase-to-phase relay
segment III is shown in Figs. 2.31 and 2.32, respectively.

1. Distance III accidental overload blocking process

Compared with the grounding relay, the phase-to-phase relay blocking process is
relatively simple. An example of the grounding relay blocking process is provided
here. The flow of the nonstation-domain accidental overload blocking scheme for

Fig. 2.31 Grounding relay

section III accidental
overload blocking logic
2.5 Distance III Accidental Overload Blocking Scheme 79

Fig. 2.32 Interphase relay section III accident overload blocking logic

the grounding distance relay, when the line is under full-phase operation, is shown
in Fig. 2.33.
The blocking procedure for grounding relays to respond to accidental overloads
in nonstation areas consists of five main components.

1) First-level sufficient condition judgment and preopening

If the current asymmetry degree condition or the (phase or interphase) cosine voltage
condition is met, open the protection. Otherwise, the relay is preopened, and the
number of starting phases is judged. If three phases are started and sufficient condi-
tions are not met, the protection is restored and blocked. If two phases are started
and sufficient conditions are not met, the phase is selected to enter the single-phase
starting judgment, and the other is restored to one phase. If a single phase starts

Fig. 2.33 Grounding relay

nonstation domain accidental
overload blocking process
(full phase operation)
80 2 Distance Protection Against Overload

and does not meet the sufficient conditions, then based on the OSI indicators for
classification, open or revert to the relay.
2) Davinan equivalent impedance solution
The Davinan equivalent impedance is solved from the fault component data of the 2
circumferential waves after start-up.
3) OSI classification judgment
Under the distance protection, a single phase starts to meet the OSI indicators, may
be a fault, and can open the protection. Otherwise, continue to judge the power
direction. If the power direction is negative, revert to protection, and vice versa,
revert to delayed dynamic open.
4) Protection of openness
As soon as the protection is opened, the timing starts, and each phase protection is
judged. If the phase ratio result is still in the action zone, the protection is opened
at the action time. If the phase ratio result exits the action zone, the protection is
reverted, and latching is performed. For the case in which the single-phase start
does not meet the OSI index, and the protection is opened again after the Davinan
equivalent impedance is calculated, the Davinan equivalent impedance calculation
time needs to be considered. Thus, the protection timing is increased by 40 ms in
this case in the scheme of Fig. 2.33.
5) Dynamic opening with additional delay
After dynamic opening, latching judgment is always performed. After the asymmetric
tidal shift is transformed into a symmetric tidal shift, the protection three-phase start
can be reverted to protection by the phase-to-phase cosine voltage limiting condition
to ensure the selectivity and reliability of protection. When the transition resistance
in the high resistance ground fault becomes small, the fault component increases, the
protection can be opened by other judgments, retaining the timing in the dynamic
opening. When the action time in the grounding III section arrives, it can be exported
and tripped. When the dynamic opening time is greater than the additional delay
time, the exit trips.
The above process pertains to the protection start in the case of full-phase of
the line. For the protection start in non-full-phase operation, its blocking process is
shown in Fig. 2.34.

2.6 Solving for the Davinan Equivalent Impedance

Presently, there are many methods to solve the Davinan equivalent parameters, which
can be generally classified as the parameter identification method based on the obser-
vation equation [181, 182] and adding various auxiliary space, curve and filtering
2.6 Solving for the Davinan Equivalent Impedance 81

Fig. 2.34 Nonstation

domain accidental overload
blocking process
(non-full-phase operation
state)

methods to solve the parameter drift and other problems [183, 184]. All of the above
methods are aimed at real-time, online identification of Davinan parameters for
voltage static stability, which need to cope with factors such as system operation
mode, fault disturbance and load dynamic changes and have unknown time-varying
characteristics and poor discriminability.
Only the Davinan equivalent impedance modulus under asymmetric faults |Z thev
| is needed in the nonstation domain accidental overload identification, which can be
considered constant for a short time after the fault |Z thev |. Thus, the backup protection
with action delay can be used to derive the Davinan equivalent parameters from the
data before the protection is operated after the fault. As shown in Fig. 2.35, the fault-
attached network is shown on the right and the left to provide an enlarged diagram
of the electrical wiring of the substation for the Davinan equivalent.
where the bus fault component voltage and
 the
 incoming fault component current
satisfy the following equation. When 3U̇2  > 1.5V (Secondary), the negative

M M
Z thev I2

U2

Fig. 2.35 Schematic of the Davinan equivalent impedance fault component solution
82 2 Distance Protection Against Overload

L3
Ep EN
L4
R
L1
EM
Eq
L2

Fig. 2.36 500 kV 4-node simulation model

sequence voltage and current are extracted. If the negative sequence voltage is too
small, the positive sequence variation is utilized instead.

or (2.45)

When an asymmetric fault occurs on the line, the main protection action removal
time is approximately 40–90 ms, and the backup protection action removal time is
approximately 1.5 s. When the line occurs in non-full-phase operation, the general
duration is 500–700 ms. Therefore, regardless of whether the protection is started
after the fault or after a single-phase trip, there is a certain time to extract the fault
component, and the shortest time is the protection 2 cycles after starting (Fig. 2.36).
The 500 kV four-node system in Fig. 2.36 is selected as an example to investigate
the Davinan equivalent impedance measurement problem under nonstation domain
accidental overload and fault conditions. Note that m and N are the sending ends of
the system with power angles of 0◦ and 10◦ , respectively, P and Q are the receiving
ends with power angles of −60° and −30°respectively, and the P side carries a load
of 3000 + j800 MVA. The system impedance is shown in Table 2.15.
The lengths of Lines L1 and L2 and Lines L3 and L4 are 200 km and 50 km,
respectively, and the line parameters follow the typical parameters in Table 2.3.
During non-full-phase operation after a single-phase fault in L3, a single-phase
fault occurs in L4, which is removed after 40 ms, and the L1 first-end protection
starts during the L4 fault. Its post-start fault component is calculated from |Z thev |,
as shown in Fig. 2.37. The protection starts at the 0 ms moment, and the data from

Table 2.15 Equivalent impedance table for the 500 kV 4-node system
System Delivery end Delivery end Recipient system P Recipient system Q
parameters system M system N
Positive 10∠87◦ 10∠87◦ 20∠87◦ 50∠87◦
(negative)
sequence
impedance ()
Zero-sequence 5∠85◦ 10∠85◦ 10∠85◦ 100∠85◦
impedance ()
2.6 Solving for the Davinan Equivalent Impedance 83

Fig. 2.37 Calculated

Davinan equivalent
impedance under accidental
overload

Fig. 2.38 Calculated

Davinan equivalent
impedance under
single-phase short circuit via
150  resistor ground at 50%

20 to 40 ms after the start are taken for calculation, as shown in the left shaded
area in the figure, and averaged. The negative-sequence component is obtained from
|Z thev | = 10.09 with an error of 0.9%. The positive sequence mutation quantity is
obtained at |Z thev | = 10.07 with an error of 0.7%. If the protection is started after
the single phase of Line L4 is removed, based on the fault components during the
non-full phase period, as shown in the shaded area on the right, the negative sequence
components are obtained at |Z thev | = 9.999, and the positive sequence mutations are
obtained at |Z thev | = 9.996 with errors of 0.01% and 0.04%, respectively (Fig. 2.38).
If a phase A ground fault via 150  resistance occurs at 50% of the L1 line, the
protection is removed after 40 ms of start-up, and the calculation obtained from the
fault component |Z thev | is shown in Fig. 2.36. Taking the data from 20 to 40 ms after
starting for calculation, as shown in the shaded area in the figure, and taking the
average value, the negative sequence component is determined to be |Z thev | = 9.97,
and the error is 0.3%. The positive sequence mutation is determined to be |Z thev | =
9.98, and the error is 0.2%.
Continued testing of the L1 line is conducted at 50% for metallic A-phase ground
faults, 90% of the line for metallic A-phase grounding and ground faults via 50 
resistance. The calculated values of the Davinan equivalent impedance are shown in
Table 2.16. Its accuracy is not affected by transition resistance, and the fault distance
is less.
The method does not have to consider parameter drift and other problems and is
not affected by fault location or transition resistance, has high accuracy and is easy
to implement.
84 2 Distance Protection Against Overload

Table 2.16 Calculated values of Davinan impedance

Line 50% fault 0  Line 50% fault Line 90% fault 0  Line 90% fault 50
150  
|Zthev | 9.97 9.97 9.94 9.10
(negative
sequence)
|Zthev | 9.97 9.98 9.95 9.13
(positive
sequence
mutation)

2.7 Adaptive Adjustment of Distance Protection Based

on Shared Information in the Station Domain
  
The overload override indicator O O I3 =  Z thev /|Z set | defines the relationship
between the protection adjustment value and the system operation mode and the
protection override action under accidental overload. Under a certain system oper-
ation mode, reducing the protection adjustment value can reduce the possibility
of protection overrun and thus improve the selectivity of protection. However, the
distance protection section III, which is overrun under accidental overload, needs to
provide the next level of line for backup protection. Reducing the setting value will
shorten the protection range and weaken the action performance under the transition
resistance, reducing the sensitivity of the protection. Therefore, the essential reason
for the distance protection section III to be exceeded under accidental overload is the
contradiction between the selectivity of the protection adjustment and the sensitivity
of the protection adjustment.
The correct rate of action of the protection is directly related to the setting value.
The rectification principle of the distance protection III section is more complex, and
the general sensitivity requirement is 1.8 to 3.0, with a lower line distance II. The
distance protection section III is compatible with the distance section or distance III
section of the lower line, and its calibration is more difficult in the ring network.
In Table 2.17, the rectification is performed from 2.0 to 3.0 by sensitivity and
simulated according to Table 2.5 Order of events and model (system Davinan posi-
tive sequence equivalent impedance of 10 ). As the system impedance is rela-
tively small, the change in the rectification value has a minimal effect on OOI1 .
However, the impact on the protection start-up results is larger. As the sensitivity
decreases, the maximum specific phase results of the protection in the out-of-zone
fault arg(U̇p /U̇op )max .F and the maximum specific phase results under tidal shift
arg(U̇p /U̇op )max .T decrease, and the maximum time t max to stay in the action zone
continuously decreases. For backup protection, the lower the sensitivity is, the lower
the probability of exceeding under accidental overload.
Table 2.17 Operational performance of backup protection with different sensitivity requirements
 
Sensitivity O O I3 max arg(U̇p U̇op )max.F /° arg(U̇p U̇op )max.T /° tmax /s
2.0 0.09 106.2 85.8 0.06
2.2 0.08 111.7 90.1 0.78
2.5 0.07 117.9 95.3 1.19
2.8 0.06 122.3 99.4 1.22
3.0 0.06 124.6 101.6 1.57
2.7 Adaptive Adjustment of Distance Protection Based on Shared …
85
86 2 Distance Protection Against Overload

2.7.1 Factors Affecting Distance III Segment Adjustment

and Performance

1. Minimum overload impedance avoidance

One of the basic principles of distance relay section III is to avoid the minimum over-
load impedance. The minimum overload impedance is usually adjusted for maximum
operating power and possible accidental overloads. However, the actual accidental
overload may be difficult to accurately predict, and the value is difficult to match with
other protections. Further analysis of overloads has been conducted in the previous
section and will not be repeated here.
2. Transition resistors
Transition resistances are usually present when short-circuit faults occur on transmis-
sion lines, especially for single-phase ground faults with a wide range of values. By
considering the transition resistance, an additional component will be added to the
measured impedance of the distance relay Z R . Taking the M and N double-ended
system in Fig. 2.8, it is not difficult to introduce the additional component

I˙M + I˙N (Z N Ė M + Z M Ė N )Rf

ZR = Rf = (2.46)
I˙M Z N Ė M + ( Ė M − Ė N )Rf

where Rf is the transition resistance, I˙M is the M-side current, I˙N is the N-side current,
Z M is the equivalent impedance from the fault point to the M-side electric potential,
Z N is the equivalent impedance from the fault point to the N-side electric potential,
Ė M is the M-side power supply electric potential, and Ė N is the N-side power supply
electric potential.
When I˙M leads I˙N , Z R is capacitive; conversely, Z R is inductive. Since the
accidental overload occurs mainly at the sending end, i.e., the M side ahead of I˙M
and I˙N , the effect of the transition resistance on the sending side protection is the
main analysis. As shown in Fig. 2.39, the measured impedance of the feeder side
attachment Z R may cause an overrange action of the protection or a rejection.
Neither of these situations is desirable.
To overcome the effect of transition resistance on distance relays, relay protection
workers have performed much research [186, 187], which has improved the ability of
distance protection to recognize transition resistance and increased the risk of relay
protection overrunning under asymmetrical accidental overloads.
3. Impact of branch lines
In a transmission grid, adjacent transmission lines are usually separated by a busbar
from a power plant or substation to which power lines, load or parallel lines, and ring
and parallel lines are connected to form branch lines.
Figure 2.40a illustrates a network with powered branch lines. When a fault occurs
at point K on line NP, for a distance relay mounted on the M side of the MN line,
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 87

Fig. 2.39 Effect of jX

transition resistor on
protected area at the sending
ΔZ R
end
Z set

ZR

ΔZ R′

Z R′
R

the current flows to the fault point but not through the protective device, creating a
helper current, and the measured impedance of the M side relay is

I˙K z 1lK + I˙MN z 1lMN

ZR = = z 1lMN + K b z 1lK (2.47)
I˙MN

where z1 is the positive sequence

 impedance per unit length of the line, K b is the
branching factor, K b = 1+ I˙QN I˙MN > 1, which is referred to as the boosting factor,
and the QN line is referred to as the boosting line.

Fig. 2.40 Network with branch lines

88 2 Distance Protection Against Overload

In the network shown in Fig. 2.40b, in the event of a fault  at point K, I˙MN does
not flow into the faulty line, the branch factor K b = 1 − I˙K1 I˙MN < 1, which is
referred to as the draw-out factor, and the other return of the NP double-back line is
the draw-out branch.
The protection timing needs to take into account the boosting factor and draw-off
factor, with

Z set.III.MN = K K z 1lMN + K K K b Z set.II.NP (2.48)

where K K is the reliability coefficient, which generally ranges from 0.7 to 0.8. When
K b is the incremental coefficient, Z set.III.MN has a larger value. However, in actual
operation, this value is affected by the system operation mode and fault disturbance,
as shown in Fig. 2.1. When L14-2 and L24 are out of operation, the actual incre-
mental coefficient decreases. However, because the protection does not modify the
adjustment value, it may cause the protection to be exceeded.
4. Other factors
In addition, the input or withdrawal of FACTS devices, such as (controllable) series
compensation capacitors, shunt reactors, series reactors, and dynamic performance
adjustments, also affect the performance of the protection [188]. However, the effect
on the action characteristics and the adjustment of the distance III section is relatively
small. For example, for the distance III section, when a fault reaches the action exit
time, the series capacitor is bypassed and withdrawn, which has no effect on its action
[202]. From the point of view of the rectification fit, even if the line is equipped with
a series complement, it should be calculated as if the series complement capacitor is
withdrawn.

2.7.2 Adaptive Tuning Scheme

1. Adaptive offset properties based on Davinan equivalent impedance

The following adaptive scheme is proposed for the difficult problem of dodging over-
loads and the adjustment of anti-transition resistance, combined with the accidental
overload exceedance indicators analyzed in the previous section.
(1) Appropriately increase the sensitivity of the protection to the transition resis-
tance in system operating mode, where the accidental overload will not be
exceeded.
(2) In the operation mode of the system where accidental overload may overtake,
the focus is ensuring selective action of the protection, i.e., as far as possible
without false activation.
An adaptive offset feature is introduced: the real-time feed-side system impedance,
i.e., the Davinan equivalent impedance is measured based on shared information in
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 89

Fig. 2.41 Adaptive offset jX

circle characteristic Z set

Z′set

−kZ thev.1
R
−kZ thev.2

the station domain, and the impedance round relay offset is thus adjusted. The addi-
tion of the offset characteristic improves the sensitivity of the relay to the transition
resistance when the protection adjustment range can be appropriately reduced to
avoid out-of-zone overruns to ensure selectivity. The action criterion
    
U̇ + k Z thev İ  Z set − k Z thev   k Z thev − Z set
 
◦
90 < <270
◦
⇔  İ − U̇ < İ (2.49)

U̇ − Z set I  2   2 

The action characteristics are shown in Fig. 2.41.

Figure 2.41 shows that with a small impedance at the sending end of the system
Z thev.1 , i.e., with the possibility of accidental overload overrun (O O I3 < 0.3 ),
the overload and transition resistance overlap zones do not change much after the
adaptive offset adjustment −k Z thev.1 , maintaining the ability of the protection to
recognize faults within the home line. In addition, the possibility of out-of-zone fault

overrun is reduced by the reduction of the rectification value from Z set to Z set . Note
that the task of distance relay segment III is to reflect faults at the end of the line and to
protect the next segment of the line as far as possible. Therefore, the slight reduction
in the protection range of the lower line does not affect the protection performance.
In the case in which the impedance at the system sending end Z thev.2 is large, i.e.,
the probability of overrunning of the accidental overload protection is low (O O I3 ≥
0.3 ), the offset −k Z thev.2 is large, which improves the ability to cope with transition
resistance and reduces the out-of-zone fault overrun. Simultaneously, measures need
to be taken to avoid overexpansion of the protection range and changing the protection
action under accidental overload from the exceedance rate.
The key to the scheme is the selection of the adaptive coefficients. For this purpose,
six schemes are proposed, and simulation of the protection action characteristics
90 2 Distance Protection Against Overload

under fault action and accidental overload in the zone with different Davinan equiv-
alent impedances is carried out, considering factors such as fault distance, transition
resistance, system work angle, Davinan equivalent impedance, and load conditions.
Statistics on the fault action rate and accidental overload protection beyond rate for
this section of the line are presented in Table 2.18.
The new rectification scheme should exhibit an increase in the fault action rate
of this section of the line, i.e., sensitivity, and a decrease in the overload exceedance
rate, i.e., an increase in selectivity. For option 4, the rectification impedance is set
to a sensitivity of 2.5, that is, 2.5 times the impedance value of the single-phase,
metallic earth fault at the end of the line. The impedance circle criterion is set to
0.3 times the Davinan equivalent impedance offset in the fault action to enhance and
avoid load override of the combined ability of the optimal, selected as the adaptive
offset characteristic adjustment factor.
2. Fast adjustment of parallel line branching coefficients based on GOOSE
information
To eliminate the effect of branching lines, the branching factor needs to be deter-
mined in real time, and the protection settings need to be adjusted. Parallel lines are
special cases of helper lines, which when disconnected can cause protection over-
runs. The problem can be quickly solved in the substation: the GOOSE trip message
is employed to obtain information about the switch movement and to readjust the
settings, which enables a more accurate adjustment of the protection setting, a short-
term adjustment of the single-phase setting of the nonfaulty line and avoidance of
the single-phase tidal shift when a single-phase selected trip occurs on the faulted
line. After the adjustment, K b < K b , and the protection setting is reduced, which is
more conducive to avoiding accidental overload. When reclosing is successful, the
original setting is restored; when reclosing is unsuccessful, the same correction is
made to the nonfaulted phase setting.

Z set.III.MN = K K z 1lMN + K K K b Z set.II.NP (2.50)

For line cutoffs in nonsubstations, the relay protection can be adjusted by wide-area
protection or other control systems for timing, which is useful for symmetrical tidal
shifting after a three-phase permanent jump. For a single phase, short-term setting
adjustment after an open break may be too late to implement. In practice, the station
domain setting adjustment should be combined with the wide domain to complement
the strengths and weaknesses.

2.7.3 Simulation Analysis of Overload Blocking Performance

This chapter focuses on the latching scheme of protection beyond accidental over-
load and is more concerned with the action behavior of protection under the dynamic
process of tide transfer compared with the tide transfer path in the grid, which can be
configured to equate different large system grid structures by configuring parameters,
Table 2.18 Comparison of adaptive offset schemes
Setting  sensitivity
Z set k Z thev = 10 Z thev = 50n Z thev = 100
Change rate of Load override Change rate of fault action Load override Change rate of fault action Load override
fault action change rate change rate change rate
Original 3 0 – – – – – –
scheme
Option 1 2.8 0.1 0 0 9% 0 13% 0
Option 2 2.8 0.2 5% 0 23% 0 28% 0
Option 3 2.5 0.2 −3% −12% 6% 0 19% 0
Option 4 2.5 0.3 2% −5% 23% 0 31% 0
Option 5 2.2 0.3 −17% −33% 6% 0 19% 0
2.7 Adaptive Adjustment of Distance Protection Based on Shared …

Option 6 2.2 0.4 −11% −33% 13% 0 34% 0

91
92 2 Distance Protection Against Overload

such as small system operation mode (including system equivalent impedance, load,
etc.), substation electrical wiring, and line length. A global perspective is replaced
with a more detailed analysis of electrical quantity changes and the dynamic process
of protection actions. Since the grounding distance section III explored in this chapter
does not show unreasonable actions under voltage collapse and out-of-step oscilla-
tions, considering the electromechanical transient characteristics for the study in this
chapter is not important and may limit the scale of the tidal shift and hinder simulation
of the protection beyond under accidental overload.

1. Simulation analysis of accidental overload identification

Unlike conventional faults, accidental overload is affected by the system operation
mode, and the difficulty in its identification mainly lies in the difficulty of ensuring
the validity of the proposed criterion under various operation modes. For this reason,
this chapter simulates the accidental overload calculation under different operation
modes on the basis of the 500 kV 7-node ring network model shown in Fig. 2.1 for
comprehensive verification.
Consider the nine modes of operation shown in Table 2.19 [11]. The system power
angle positive sequence impedance angle is set to 87°; the zero sequence impedance
angle is set to 85°; and the system zero sequence impedance magnitudes are set to
0.5 times and 3.0 times, respectively, the positive sequence impedance magnitudes.
The potential angles of EM , EP , and EQ are adjusted to 0°, 5°, and −10°, respectively,
and the potential angle of EN varies from −30° to −70°. After considering reactive
power compensation, the loads at each node are S 4 = 1000 + j50MVA, S 5 = 2000
+ j150MVA, S 6 = 1000 + j100MVA, and S 7 = 800 + j100MVA, and the active and
reactive loads are adjusted to 100%, 120% and 150%, respectively. The voltage at
each node is adjusted at 500 and 525 kV.
The line unit positive (negative) sequence parameters are Z1 = 0.027 + j0.2701
/km, C1 = 0.0123F/km, zero sequence parameters Z0 = 0.1984 + j1.131 /km,
and C0 = 0.0051F/km. The lengths of l14 and L45 are 200 km, and the remaining line
lengths are adjusted by 50, 100, 150 and 200 km in turn.

Table 2.19 System power parameters for different modes of operation

Operation mode Z thev1.M / Z thev1.N / Z thev1.P / Z thev1.Q /
Normal 54∠87° 54∠87° 27∠87° 54∠87°
Minimum 10.8∠87° 10.8∠87° 5.4∠87° 10.8∠87°
Maximum 108∠87° 108∠87° 54∠87° 108∠87°
Special 1 36∠87° 5.4∠87° 36∠87° 27∠87°
Special 2 108∠87° 27∠87° 54∠87° 27∠87°
Special 3 8.1∠87° 108∠87° 4.05∠87° 108∠87°
Special 4 8.1∠87° 54∠87° 8.1∠87° 54∠87°
Special 5 27∠87° 36∠87° 13.5∠87° 54∠87°
Special 6 54∠87° 10.8∠87° 27∠87° 10.8∠87°
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 93

Simulations are performed in the order of the accidental overload events shown
in Table 2.4. The phases 0.66–0.86 s and 1.16–1.36 s are taken as single-phase tidal
shift samples for a total of 31,104 groups. As a comparison group, A-phase faults
are simulated at 50% and 100% of the L14-1 line and 50% and 100% of the L45-1 line,
with the transition resistance increasing from 0  by 50  to 300  for a total of
3024 groups. The single-phase tidal shift and identification under single-phase faults
are shown in Fig. 2.42.
In Fig. 2.42, “▲” indicates the single-phase tidal shift that caused the start of the
Distance III section; “◯” indicates the single-phase fault that caused the start of the
Distance III section.
Further statistics on the single-phase tidal shift causing distance III section start-
up versus meeting the identification condition under a single-phase fault are shown
in Table 2.20.
As shown in Fig. 2.42 and Table 2.20, the single-phase OSI ~OSI14 conditions of
Eq. (2.20) are satisfied for all OSI indicators under tidal shifts. In addition, 8.99% of
single-phase faults simultaneously satisfy the conditions of OSI1 ~OSI3, and after
further identification by OSI4, the single-phase faults in the calculation case are

(a) |Z'thev |/|Zset |-|ZRΦ |/|Z ' thev||Z

(b) arg(Z'EΦ)-I0/I2

Fig. 2.42 Identification results of single-phase tidal shifts with single-phase faults
94 2 Distance Protection Against Overload

Table 2.20 Statistics on single-phase tidal shift and single-phase faults meeting identification
conditions
Meet the conditions OSI1 (%) OSI1~OSI3 (%) OSI1~OSI3 & I0/I2 (%)
Single phase power flow transfer 100 100 100
Single phase fault 25.93 8.99 0

completely removed from the accidental overload action domain. Single-phase tidal
shifts that cannot be fully identified in special cases can be dynamically identified by
protection start transition information with delay to ensure the reliability of distance
protection section III.
Interphase and phase-to-phase ground faults are verified with three-phase tidal
shifts. The phase 2–2.1 s is taken as a sample of a three-phase tidal shift with 15,552
groups. For the comparison group, BC interphase, BC phase-to-ground and ABC
three-phase ground faults are simulated at 50% and 100% of the L14-1 line and 50%
and 100% of the L45-1 line, and the transition resistance is taken as 10  for a total
of 2700 groups. The identification of a three-phase tidal shift with phase-to-phase
and phase-to-ground faults is shown in Fig. 2.43. The “▲” in the figure indicates
the distance III segment that causes the three-phase tidal shift situation to start; “◯”
indicates the three-phase fault that causes the distance III section to start.
Figure 2.43 shows that the three-phase tidal shift can be completely distinguished
from phase-to-phase faults and phase-to-ground faults by OSI2 and OSI.3
2. Simulation analysis of full process blocking of accidental overload

1) Effect of out-of-zone fault conditions on overload blocking

Whether the protection is exceeded under an out-of-zone fault is an important ques-

tion regarding overload blocking and is related to 2 main conditions: fault location
and fault type.

Fig. 2.43 Identification results of the three-phase tidal shift with phase-to-phase/phase-to-ground
faults
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 95

(1) For faults that can trigger overruns, the location is relatively close to the end of
the protection and far backup zone; if they are far apart, they will not trigger
overruns.
(2) For fault types, out-of-zone, three-phase faults can only trigger symmetrical
accidental overloads, which are easier to distinguish, and the asymmetrical
component of out-of-zone, two-phase faults exists for a short time (less than
0.1 s), which can be blocked by the 100 ms continuous phase-to-phase voltage
cosine condition restriction. This restriction can block such cases from over-
riding the protection. The difficulty of identification remains the asymmetric
accidental overload triggered by multiple single-phase faults. For single-phase
short circuits, the transition resistance also needs to be considered. If the tran-
sition resistance is large, an out-of-zone fault will also not trigger protection
overruns.

Since the exit action of section III is based on a continuous judgment of the
protection criterion within a certain time, it will greatly reduce the threat of accidental
overload to the backup protection section III if the protection is not exceeded during
an out-of-zone fault, especially a single-phase fault.

2) Effect of multiple disturbances on overload blocking

For the system shown in Fig. 2.1, the fault disturbance sequence is set as shown in
Table 2.21. The entire process protection start-up and sufficient condition criterion
are shown in Fig. 2.44
After a fault occurs at L35-1 , the L14-1 first ground relay section III A phase is
activated, and the protection reverts after the single phase of the faulty line trips

Table 2.21 Sequence of accidental overload events

Time (s) Events
0.1 Phase A fault occurs 100% from L25-1 to the P side (grounded through 10 
transition resistance)
0.16 Main protection of L25-1 action trips phase A, nonfull phase operation
0.2 Phase A fault occurs at 50% of L25-2 from the P side (grounded through 20 
transition resistance)
0.26 Main protection of L25-2 trips phase A and operates in incomplete phase
0.96 L25-1 reclosing phase A at fault
1.00 Phase A fault occurs at 80% of L14-2 from side M (grounded through 50 
transition resistance)
1.02 Three-phase tripping of L25-1 acceleration
1.06 Main protection of L14-2 trips phase A and operates in incomplete phase L25-2
reclosing phase A at fault
1.12 Three-phase tripping of L25-2 acceleration
1.86 L14-2 reclosing phase A at fault
1.92 Three-phase tripping of L14-2 acceleration
96 2 Distance Protection Against Overload

Fig. 2.44 Protection criterion and start-up under multiple single-phase accidental overloads

open. Subsequently, a fault occurs at L25-2 and removes the single phase, the L14
line carries an accidental overload of the L25 double return Line A phase, and the
ground relay section III A phase enters the action zone. The maximum value of
current asymmetry is 0.384, and the minimum value of the A-phase voltage cosine
component is 0.744 p.u. during the protection into the action zone, which does not
satisfy the fault opening condition.
After the protection starts, it starts to calculate the Davinan equivalent impedance,
and the positive sequence incremental data can only be used for the 2nd-3rd week
wave after the start. As long as there is a negative sequence quantity, the negative
sequence data can be continuously applied. After 100 ms of blocking, if the main
protection does not operate, the OOI indicators O O I1 , O O I2 are calculated to satisfy
the accidental overload override condition, and the negative sequence power direction
is −98° (positive direction), entering the accidental overload delay processing, as
shown in Fig. 2.45.
During the protection start-up, the L14-2 line had an A-phase fault selected to
jump A-phase, and the L35 double-loop line overlapped the fault and permanently
tripped three phases. During the disturbance, the Davinan equivalent impedance
values calculated based on negative sequence quantities slightly fluctuate but do
not affect the overall judgment. Line L14-1 takes on two line symmetrical accidental
overloads and one line single-phase asymmetrical accidental overload, and the degree
of overload overtaking further increases. There is a decrease in O O I1 and a slight
increase in O O I2 , maintaining its credibility during the dynamics.
If the general grounding distance III section delay time in the protection starts
1.5 s after approximately 1.78 s of outlet action, at this time, the protection is still
in the single-phase start state and cannot be restored, which will cause unreasonable
protection action. Therefore, in such cases, the additional delay is added for the III
section action time. After L24 three-phase trips, the protection is started in three
phases, and the protection is restored because the interphase cosine voltage limit
condition is not satisfied, thus avoiding beyond the false operation. Due to the dual
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 97

Fig. 2.45 Protection criterion and start-up under multiple single-phase accidental overloads

configuration of China’s transmission network protection and high reliability, the

fault under the main protection rejection rarely occurs (2012 line protection correct
action rate of up to 99.95%, and incorrect action 7 times are false [177]), and the
probability of high-resistance ground fault in the large operation mode is not high.
This condition does not have much impact on the stability of the system. The appro-
priate delay to ensure reliable non-misoperation of the grid safe operation outweighs
the disadvantages.
If the protection is adjusted to the adaptive offset characteristic after the starting
element starts, the ratio phase of the ground distance section III is determined as
shown in Fig. 2.46. Compared with Fig. 2.44, the protection stays in the action zone
for a much reduced time, reducing the probability of tripping the protection exit.

Fig. 2.46 Protected start with adaptive offset characteristics

98 2 Distance Protection Against Overload

Fig. 2.47 Protection starts with dual adjustment of the adaptive offset characteristic and branching
factor

The adaptive setting of the protection is modified according to the GOOSE tripping
information, and the action characteristics of the protection are further optimized, as
shown in Figs. 2.46 and 2.47. Such a short time start does not pose a threat to the
unreasonable action of section III, which is a more desirable mode.

3. Effect of protection feeder impedance on overload blocking

The effect of the Davinan equivalent impedance on protection beyond accidental

overload is analyzed in the previous section, and the OOI index is presented. When
the Davinan equivalent impedance amplitude of the protection feeder is changed from
10 to 60  (only the positive sequence equivalent impedance amplitude is shown
in the table; the positive sequence impedance amplitude is 87°; the zero sequence
impedance is taken as 0.5 times the positive sequence impedance amplitude; and the
zero sequence impedance angle is 85°), other parameters remain unchanged. The
OOI index and starting of the protection under the disturbance sequence listed in
Table 2.6 are shown in Table 2.22.
where arg(U̇p / U̇op )max is the maximum value of the protection ratio phase result
during the tidal shift (excluding the out-of-zone fault phase) and t op is the maximum
time that the protection continuously stays in the action zone. As the impedance at
the sending end increases, the result of the protection ratio phase decreases, the time
spent in the action zone decreases, and the danger of unreasonable action decreases.
When the impedance amplitude at the sending end is 50 , the conditions of
O O I1 < 2, O O I2 > 1 and O O I3 > 0.3 are simultaneously satisfied, and according
to the analysis in this chapter, opening the protection in this case does not cause
accidental overload under transcendence. The data in the table also confirm this
determination, with Davinan equivalent impedances of 50  and 60 , where tmax
is 0.06 and 0.05 s, posing no threat to the protection of Section III. This table also
shows that the protection did not start in the asymmetric tidal shift from a sending
impedance of 40 . The OOI indicator not only accurately reflects the situation but
also leaves a margin.
Table 2.22 OOI indicators and protection start-ups for changes in impedance at the sending end
|ZM1 | O O I1 min O O I2 max O O I3 max Uφ min arg(U̇p / U̇op )max tmax
10 7.25 0.74 0.06 0.86 101.6 1.57
20 3.94 0.80 0.12 0.83 96.7 1.22
30 2.84 0.87 0.18 0.81 92.3 0.93
40 2.29 0.94 0.24 0.80 88.4 0.08
50 1.96 1.03 0.30 0.79 84.9 0.06
60 1.74 1.13 0.35 0.77 81.6 0.05
2.7 Adaptive Adjustment of Distance Protection Based on Shared …
99
100 2 Distance Protection Against Overload

4. Effect of the main protection rejection on overload blocking for out-of-area lines

When a fault occurs outside zone Line I, the main protection refuses to operate and
will trigger the Line J distance III section start. If the final fault of Line I backup
protection II Section is removed, Line J protection reverses. If the fault is removed
by the Line I backup protection III section action, the Line J protection will be unable
to cooperate in time, and misoperation may occur. Of course, the probability of such
a situation is extremely low. In addition, the severity of a long line backup protection
III section sensing out-of-area fault is higher. If in the III section action time has not
been removed, it may affect the system stability, and its severity will greatly exceed
the line accident overload unreasonable action.

2.7.4 Simulation Analysis of Intra-Zone Faults and Complex

Fault Opening

Because the application of cosine voltage limiting conditions to open interphase

faults is well established, this chapter does not verify this type of fault again, only for
all types of single-phase ground faults, especially single-phase grounded complex
faults.

1. Single-phase ground fault in the zone

Simulations are performed for different operating modes, fault locations and transi-
tion resistances for ground faults in the zone. For the large operation mode, refer to
the system parameters in Section I. For the small operation mode, the load is reduced
to 50%, and the system equivalent impedance is adjusted to ZM1 = 50∠87°, ZM0
= 40∠85° , ZN1 = 80∠87° , ZN0 = 50∠85° , ZP1 = 100∠87° , and ZP0
= 200∠85° . The simulations are used to derive the maximum transition resis-
tance at each operation mode and fault point to satisfy the power asymmetry The
maximum transition resistance to meet the power asymmetry condition at each oper-
ation mode and fault point, the maximum transition resistance of section III of the
ground protection into the action zone, the minimum value of the OOI indicator into
the single-phase and the minimum and maximum values of OOI indicators for open
judgment are shown in Table 2.23.
Since the positive sequence current is larger in the large operation mode, the
current asymmetry may be smaller for the same fault condition (same fault location
and transition resistance) than in the small operation mode. The value is also influ-
enced by the system impedance at the opposite end. In general, the absolute values
of negative and zero sequence currents are greater in the large operation mode than
in the small operation mode. Compared with a small operation mode, a large oper-
ation mode distance grounding III section can resist greater transition resistance.
A large operation mode into a single-phase starting judgment after the minimum
O O I1 min > 2, in the nonaccident overload starting domain, can be directly opened.
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 101

Table 2.23 Protection actions under single-phase faults

Fault Large operation mode Small operation mode
location Current Starting O O I1min Current Starting O O I1max
(%) asymmetry Max Rf asymmetry Max Rf
opening Max opening Max
Rf Rf
50 120 190 6.73 150 140 1.65
100 80 150 7.88 90 110 1.99
150 30 90 9.49 40 60 2.26
200 10 20 15.13 0 10 2.55

A small operation mode in the grounding via large resistance O O I1 max < 2. In the
accident overload beyond the domain, needs to be additional delayed dynamic open.

2. Start-up of three-phase protection for developmental faults in the zone

In the small mode, phase A via a 10  resistive ground fault occurs at 80% of this
section of the line at 0.1 s, phase B via a 10  resistive ground fault occurs at 20%
of the lower level line at 0.2 s, and phase C via a 50  resistive fault occurs at 50%
of this section of the line at 0.3 s. The interphase cosine voltage, current asymmetry,
OOI and grounded section III are shown in Fig. 2.48.
In the event of a phase A fault, only the phase A relay starts, and the current
asymmetry is large enough to open the protection by the current asymmetry criterion.
Subsequently, phase A remains in the action zone, and the protection exit is tripped
after 1.5 s. If the first and second faults simultaneously occur, i.e., directly for the 0.2–
0.3 s protection start and electrical quantity situation, the protection can be opened
by the BC interphase cosine voltage. If the triple faults simultaneously occur, i.e.,
for the situation after 0.3 s, the protection can be opened by the three-phase cosine
voltage.

3. Start-up of two-phase protection for developmental faults in the zone

Multiple cases of two-phase relay start under single-phase faults in the simulation
area.
In the large operation mode, phase A occurs at 80% of this section of the line at
0.1 s via a 10  resistive earth fault, and phase A relay enters the action zone; phase
B occurs at 20% of the lower line at 0.2 s via a 100  resistive earth fault, and phase
B relay does not start; phase C occurs at 50% of this section of the line at 0.3 s via a
50  resistive earth fault, phase C relay enters the action zone and phase A continues
to hold, as shown in Fig. 2.49.
After the start of phase A, the protection is opened due to O O I2 > 1 Therefore,
phase A is always maintained in the action zone, and the protection outlet is tripped
after 1.5 s. If the first and second faults simultaneously occur, only phase A starts and
is affected by the phase B fault; the measured impedance decreases, and O O I2 > 1
can still directly open the protection outlet. If the triple fault simultaneously occurs,
102 2 Distance Protection Against Overload

Fig. 2.48 Three-phase protection under triple single-phase fault start open result of operation

i.e., for the case after 0.3 s, phase A and phase C start, phase C current suddenly
changes the amount of phase A, and phase C will open. At this time O O I1 is 4.86
is greater than the threshold, and the protection outlet opens after 1.5 s.

4. Zone Developmental fault single-phase protected start-up situation

Alternating single-phase relays under multiple single-phase faults in the zone start.
The simulation is performed for the case of action. In the small operation mode, at
0.1 s, phase A is grounded at 80% of this line via 100  resistance, and the phase A
relay enters the action zone; at 0.2 s, phase B is grounded at 20% of the lower line
via 40  resistance, phase B enters the action zone, and phase A resumes; at 0.3 s,
phase C is grounded at 50% of this line via 50  resistance, phase C enters the action
zone, and phase B resumes, as shown in Fig. 2.50.
The starting phase keeps changing due to the superposition of the fault compo-
nents, so it can only be opened after the last phase has been started by a
delay.
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 103

Fig. 2.49 Two-phase protection under triple single-phase fault start open result of operation

Fig. 2.50 Alternating single-phase protection under triple single-phase faults start open result of
moving
104 2 Distance Protection Against Overload

A comparison of the three-phase protection start, two-phase protection start and

single-phase alternate start under triple single-phase faults reveals that because of the
120° phase angle difference between its fault components between single-phase faults
of different phases, it affects the identification of the original fault by the protection.
Only when the fault characteristics of single-phase faults of three different phases are
obvious, is it possible to start three-phase, at which time the opening by the current
asymmetry can be achieved. If two or more single-phase faults of different phases
lead to single-phase or two-phase protection start, often requiring single-phase open
discriminations and delayed tripping in some cases, but will not refuse to operate.
All the above are special cases discussed under the assumption that the main
protection and near backup protection do not operate. Normally, the main protection
can reliably trip the fault, and it is extremely difficult for the above cases to occur.
However, for the reliability of backup protection, such verification is essential.

5. Incidental overload switching failure

If a two-phase or three-phase fault occurs in the zone under asymmetric accidental

overload (phase A fault outside the zone and trips phase A), it may be safely opened
without being affected by accidental overload. If the same name phase fault occurs
at the end of the line, with different transition resistances, the OOI indicators, criteria
and protection ratio phase results are shown in Table 2.24.
When the transition resistance is 0, the protection can be opened by U  cosϕ min ;
when the transition resistance is 150 , the protection ratio phase result is still 114.7,
compared with Table 2.8 The large operation mode of the end of this line can only be
opened to 150 , and the fault characteristics of a single-phase fault are more obvious
due to the superposition of the same phase component of accidental overload. It is
possible that the current system mode makes the section III exit action time slightly
longer, but does not reduce the ability of the protection to identify under faults.
Table 2.25 shows the start of the B-phase grounding relay when a B-phase fault
occurs in the zone under accidental overload of the line in phase A. The effect of the A-
phase accidental overload on the B-phase relay is small. Simultaneously, the phase
angle difference between the B-phase fault current and the A-phase tidal transfer
current makes the A-phase protection revert, avoiding the unreasonable action of the
protection under single-phase accidental overload. The dynamic process of protection
starting is shown in Fig. 2.51.
It is verified by simulation that the accidental overload blocking scheme proposed
in this chapter can reliably block unreasonable protection actions under accidental
overload by identifying the system state and time domain dynamics and can effec-
tively open under all types of faults. The proposed adaptive rectification scheme has
a good suppression effect on both out-of-zone fault override and tidal shift over-
ride during accidental overload and maintains and improves the sensitivity of the
protection end of this line, which is simple and feasible and has good practicality.
Table 2.24 (Phase A) single-phase fault of phase A in the zone under accidental overload
Fault resistance OOI1min OOI2min arg(U̇p / U̇op )max m max Uφ cos ϕmin
0 5.95 183.8 174.2 0.43 0.45
50 5.86 1.68 131.5 0.27 0.71
100 6.22 1.20 119.8 0.22 0.78
150 6.45 1.04 114.7 0.19 0.80
2.7 Adaptive Adjustment of Distance Protection Based on Shared …
105
 106

Table 2.25 Simultaneous single-phase faults in phase B within the zone under phase a tidal shift

Fault resistance Phase A arg(U̇p / U̇op )max Phase B arg(U̇p U̇op )max m max Uφ min
0 515 173.3 0.41 0.43
50 1.19 117.5 0.17 0.76
100 0.84 104.2 0.12 0.82
150 0.73 98.5 0.11 0.84
2 Distance Protection Against Overload
2.7 Adaptive Adjustment of Distance Protection Based on Shared … 107

Fig. 2.51 Protection start-up for a phase B fault under phase A tidal shift

2.7.5 Summary

In this chapter, identification, blocking and adaptive tuning methods are proposed
for accidental overload, a further adaptive tuning scheme is proposed. The proposed
scheme is verified by simulation. The following conclusions are obtained:

(1) Based on the system operation and protection rectification conditions of distance
section III operation under accidental overload, the OSI index based on station
domain information, including the relationship among rectification impedance,
measured impedance and Davinan equivalent impedance, which can effectively
divide the accidental overload protection starting domain, is proposed.
(2) The identification of accidental overload needs to consider its dynamic process,
including three stages of out-of-zone fault, single-phase tidal shift and symmet-
rical tidal shift, which can be separately identified by station-domain electrical
information and time-domain dynamic information.
(3) Accidental overload can be identified and blocked based on station domain
switching information and protection start information This chapter proposes a
distance III section protection accidental overload adaptive blocking scheme.
(4) The essential problem of protection beyond accidental overload is analyzed, and
the adaptive adjustment method based on station domain information is further
proposed to ensure the selectivity and to take into account the sensitivity of the
distance III section protection, which reduces the risk of protection action under
accidental overload.
(5) It is verified by simulation that the identification method proposed in this chapter
can thoroughly identify single-phase and three-phase tidal current transfer iden-
tification problems under different system operation modes, can reliably block
unreasonable actions of protection under accidental overload during the whole
process of accidental overload, and can open effectively under various types of
faults.
(6) The identification, blocking and adjustment method of accidental overload
based on station domain information proposed in this chapter is simple and
feasible and solves the problem of unreasonable action of the distance III section
under accidental overload more thoroughly, which is important for preventing
chain tripping and ensuring safe operation of the power grid.
Chapter 3
Immunityin Distance Protection
of Oscillations

The phenomenon of loss of synchronization in a power system or power plant oper-

ating in parallel is referred to as oscillation. When a power system oscillates, the
angle δ between the equivalent electric potentials on both sides periodically varies
within the range of 0°–360°, which causes the voltage and line current at various
points in the system to experience periodic variations. Correspondingly, the measure-
ment impedance of the distance protection also exhibits the same periodic varia-
tion. During oscillation, the measured impedance may enter the action range of the
distance protection with a change in δ and cause protection misoperation. There-
fore, it is often necessary to include an oscillation blocking element to prevent the
protection from malfunctioning when oscillation occurs in the system. The oscilla-
tion blocking method of short-time open protection, in which timing begins after the
disturbance starts and blocks the protection after a period of time delay (generally set
to 160 ms), is widely employed in China. This oscillation blocking method is simple
to implement but fails to truly distinguish between oscillation and short-circuit faults
and only closes the protection from opening again before oscillation may occur;
thus, this oscillation blocking method is unable to mitigate short-circuit faults that
occur during oscillation [182]. A complete oscillation blocking element should meet
the following requirements: reliable differentiation between oscillation and short-
circuit faults; i.e., the protection is not affected by system faults, and the protection
is reliably blocked when the system oscillates without faults; short-circuit faults are
quickly identified in system oscillations and the protection is opened in a timely
manner [205]. To achieve the requirements of oscillation blocking, scholars world-
wide have carried out extensive research. Common oscillation blocking methods
include the oscillation blocking method based on a comparison of three sequence
current amplitudes, the oscillation blocking method based on the measurement of the
rate of change in the impedance, and the oscillation blocking method based on the
voltage amplitude of the center of oscillation [7, 206–208]. However, the rectified
values obtained by these methods often cannot be used to completely distinguish
between oscillation and short-circuit faults and require a certain delay for matching.
This situation also affects the response speed of the protection of the fault, and the

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 109
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_3
110 3 Immunityin Distance Protection of Oscillations

risk of false opening of the protection in oscillation still exists, leading to protection
misoperation. If it is possible to propose immunity in oscillation distance protection,
which no longer relies on oscillation blocking elements, the above problems can
be effectively avoided. The multiphase compensated distance protection is oscilla-
tion independent in principle, which provides a certain degree of immunity to the
proposed oscillation-immune distance protection. Thus, oscillation-immune distance
protection provides a certain theoretical basis for this study.

3.1 Multiphase Compensated Distance Relays

Multiphase compensated distance relays were widely employed in the era of elec-
tromagnetic protection [209–213]; one of their important advantages was their reli-
able inactivity in the absence of faults in a purely oscillatory system. At that time,
multiphase compensated distance relays were proposed to solve the problem of
a single relay reflecting multiphase faults. Thus, they were mostly composed by
comparing phases between different phase compensated voltages or by comparing
phases between different phase-to-phase compensated voltages, i.e., comparing
phases among U̇ A , U̇ B and U̇C or comparing phases among U̇ AB  
, U̇ BC and U̇C A .
In this paper, we focus on the characteristics of multiphase compensated relays that
do not respond to oscillations without faults for in-depth analysis and explore the
feasibility of the composition of distance relays that are not affected by oscillations.
Thus, we choose to analyze the composition with phase compensated voltages and
phase-to-phase compensated voltages.
The multiphase compensated distance relay (hereafter referred to as the P-
element) consists of three comparison elements: PA-BC , PB-CA and PC-AB . The action
conditions of these three compared elements are [205]:
⎧
⎪ U̇ 
⎪
⎪ P : 360◦ > arg U̇  A > 180◦
⎨ A−BC BC
U̇ 
PB−C A : 360◦ > arg U̇ B > 180◦ (3.1)
⎪
⎪ CA
⎪ 
⎩ PC−AB : 360◦ > arg U̇C > 180◦
U̇ AB

The above three criteria are applied to identify different phase faults and are
output with “or” logic. In Eq. (3.1), U̇ A , U̇ B and U̇C represent the phase A, B and C
 
compensated voltages, respectively; U̇ AB , U̇ BC and U̇C A represent the interphase AB,
BC and CA compensated voltages, respectively. The phase compensated voltages and
interphase compensated voltages are defined as follows:
⎧
⎨ U̇ϕ = U̇ϕ − ( I˙ϕ + 3k I˙0 )Z set
⎪ 

k = Z 03Z−Z 1 (3.2)
⎪
⎩ ϕ = A,1B, C
3.2 Multiphase Compensated Distance Relay Performance Analysis 111

 
U̇ϕϕ = U̇ϕϕ − I˙ϕϕ Z set
(3.3)
ϕϕ = AB, BC, C A

Z 0 and Z 1 are the zero-sequence impedance and positive-sequence impedance,

respectively, of the line that is being protected, and Z set is the rectified impedance.

3.2 Multiphase Compensated Distance Relay Performance

Analysis

The purpose of selecting a multiphase compensated distance relay as a core element

for distance protection is to take advantage of its excellent characteristics, which are
unaffected by power system oscillations. Therefore, it is necessary to analyze these
operation characteristics under various oscillation and fault conditions. Although
research on this subject exists [214–216], the analysis results are typically represented
as phase diagrams, which does not facilitate numerical calculations. This paper gives
results that can be numerically analyzed from the point of view of analyzing the
equations of action of a multiphase compensated distance relay.

3.2.1 Oscillation Without Fault Condition

When a system simply oscillates without fault, the compensated voltage of each
phase is actually the voltage at the rectification point of each phase, its phase
angle synchronously changes, and the relative angle between two compensated volt-
ages remains unchanged, as demonstrated by (3.2) and (3.3). Thus, the multiphase
compensated distance relay will not operate.

3.2.2 Fault Conditions Without Oscillation

As discussed in the literature [217], from the point of view of the compensated voltage
vector diagram, a multiphase compensated distance relay can reliably operate in the
absence of oscillations in the fault condition. In this paper, the problem is analyzed
from the point of view of the action equation. Selecting a single-phase ground fault
occurring in phase A as an example, the performance of the multiphase compensated
distance relay operation under the fault condition but without an oscillation system
is analyzed. A schematic of the system is shown in Fig. 3.1.
In Fig. 3.1, point R is the location of the protection; point F is the fault point;
point Y is the end of the protection range; Ė M and Ė N are the power potentials on
both sides of the system; U̇ M and U̇ N are the voltages on both sides of the line being
112 3 Immunityin Distance Protection of Oscillations

Fig. 3.1 Schematic of the

forward fault system

protected; I˙M is the current measured at the protection installation; I˙f is the fault
branch current; and Z M , Z L , Z N , Z f , Z set , and Z Σ are the line impedances of each
section shown in the figure.
When a single-phase fault occurs in the system, it may be useful to set Ė M = Ee jθ
and Ė N = AEe j(θ−δ) . Next, the PA-BC action equation is expressed as follows:

U̇ A j (Z f − Z set ) Z
arg 
= arg √ × (3.4)
U̇ BC 3(Z f + Z M ) (Z  − Z M − Z set ) + Ae− jδ (Z M + Z set )

When there are no oscillations in the system, assume that.

(1) The impedance angles for all parts of the system are equal.
(2) The positive and negative sequence impedances of the system are equal.
(3) The fault is metallic in nature.
Additionally, setting δ = 10° and A = 1, the equation of action becomes

U̇ A j (Z f − Z set ) 1
arg = arg √ × (3.5)

U̇ BC 3(Z f + Z M ) (1 − m) + me− j10◦

In Eq. (3.5), m is a real number defined as in Eq. (3.6).

Z M + Z set
m= <1 (3.6)
Z
 
When an internal fault occurs, Z f < Z set , and the action equation, arg U̇ A /U̇ BC

≈
270◦ , is in the most sensitive region of the action. When an external fault occurs,
Z f > Z set , the action equation, arg U̇ A /U̇ BC

≈ 90◦ , is in the most sensitive region
of inactivity.
Since the multiphase compensated distance relay is phase-specific between the
compensated voltages of each phase, when a symmetrical fault occurs, the phase of
each compensated voltage simultaneously changes, and the relative angle between the
compensated phase voltages remains unchanged. Thus, the multiphase compensated
distance relay will not operate, that is, the relay does not reflect symmetrical faults.
3.2 Multiphase Compensated Distance Relay Performance Analysis 113

3.2.3 Oscillation Accompanied by Fault Conditions

1. Oscillation accompanied by intrazone failure

The oscillation schematically accompanies the fault system within the zone in
Fig. 3.1. Consistent with the previous assumptions, power system oscillations imply
that δ continuously varies during system operation from 0° to 360°.
When a ground fault occurs in phase A of the system, the characteristics of compo-
nent PA-BC , which reflect the faulty phase, and those of components PB-CA and PC-AB ,
which reflect the nonfaulty phase, differ. Thus, they are separately analyzed here.
For the comparison element PA-BC , which reflects the faulty phase, the corre-
sponding compensated voltage is determined, and the action criterion is obtained as
follows [218]:

U̇  −j −j
arg U̇  A = arg (1−m)+Ake − jδ = arg 1+Be− jδ
BC (3.7)
B= Am
1−m
>0

In Eq. (3.7), A is the voltage amplitude ratio at both ends of the system, which
usually falls within the following range: 90.9% ≤ A ≤ 110%. The effect of oscillation
on PA-BC can be obtained by taking A = 1 and plotting the curve of the calculated
result of PA-BC with δ for different values of parameter B. This curve is shown in
Fig. 3.2. The shaded part of the figure shows the action interval of the relay:
When the system oscillates, δ varies between 0° and 360°. Figure 3.2 shows that
in most cases, PA-BC is still able to reliably act in the case of a fault. Only when B
> 1 is there a possibility of internal fault rejection near δ = 180°, and the rejection
range is related to the value of B. Setting A = 1, solving inequality (3.8) yields m >
0.5.

Fig. 3.2 Curves of PA-BC

calculations with δ for 360
changes in parameter B

270
Results of PA-BC

180
B=0.7
B=0.9
90
B=1.1
B=1.3
0
0 45 90 135 180 225 270 315 360
δ
114 3 Immunityin Distance Protection of Oscillations

Am
B= >1 (3.8)
1−m

According to the definition of m in Eq. (3.6), m > 0.5 means that the distance from
the end of the protection range to the power supply on the protection installation side
is greater than the distance to the power supply on the opposite side or that the end
of the protection range is outside the center of the oscillation. In this case, there is
a possibility of rejection in the zone when the oscillation develops to the extent that
the power supply at both ends is swinging away at an angle near 180°. In the case
of power system oscillation, δ periodically varies in the range of 0°–360°, and for
the instantaneous action I-section, the I-section protection can still correctly operate
when the oscillation angle decreases. In the case of power system oscillation, the
tacho-activity of the protection can be appropriately relaxed so that the rejection of
the I-section protection is acceptable.
However, for the time-delayed operation of the section II and section III protection,
if the correct duration of operation during the oscillation cycle is less than II, the relay
will be rejected in the case of oscillation and single-phase ground fault in the zone.
The analysis of Eq. (3.8) clearly indicates that the greater the extent that the distance
from the end of the protection range to the local power supply is greater than the
distance to the opposite power supply, the shorter the correct action time of PA-BC .
The correct action angle range of PA-BC can be determined. When B in Eq. (3.7) takes
different values from small to large, as shown in Table 3.1. δ the angle of PA-BC from
incorrect action to correct action is defined as the positive action initial angle; the
angle of PA-BC from correct action to incorrect action is defined as the positive action
end angle; and the angle range between them is defined as the positive action angle
range.
As shown in Table 3.1, the correct action angle range of PA-BC decreases as B
increases. When the value of B is increased to a certain extent, the correct action
range only slightly changes. Therefore, it is considered that the minimum correct
action range of PA-BC is approximately 185° when B is taken as 20. Literature [219]
pointed out that the oscillation period usually ranges from 0.2 to 2 s. Based on the
minimum oscillation period of 0.2 s, the minimum correct action time of PA-BC is
0.1 s. This action time does not meet the requirements of segment II and III delay
action. Therefore, if a multiphase compensated distance relay is used to form the
distance protection of sections II and III, the protection may refuse to operate when
an internal single-phase fault occurs during the power swing whose oscillation period

Table 3.1 Correct action

B Correct action Correct action end Correct action
range for P A-BC
initial angle angle angle range
5 258 102 204
10 265 96 191
15 266 94 188
20 267 92 185
3.2 Multiphase Compensated Distance Relay Performance Analysis 115

is small and oscillation center is in the protection range. The solution to this problem
is to add counters to the timers of distance protection II and III sections. If it is
detected that the protection repeatedly enters and exits the action zones for a certain
number of times within a certain period, then a tripping signal is given.
For element PB-CA , which reflects the nonfaulted phase, the action equation is
obtained after calculating the corresponding compensated voltage, as shown in
Eq. (3.9).
Equation (3.9) clearly shows that the range of action of PB-CA is related to the
parameters T, A, and k. The physical significance of each of these three parameters
is explained as follows:
T: relative electrical distance from the fault point to the end of the protective range
and to the power supply on the installation side of the protection.
A: ratio of the opposite side power amplitude to the protection installation side
power amplitude.
m: Electrical distance from the end of the protection range to the power supply
on the installation side of the protection.
⎧ ◦ ◦
⎨ argU̇ B (1−m)e− j120 +m Ae− j (δ+120 )
= (1−m)e
U̇C A j120◦ +m Ae− j (δ−120◦ ) +T
−Z (3.9)
⎩T= Z set f
>0
Z f +Z M

To investigate the effect of each parameter on the calculated results of PB-CA , the
curves of the calculated results of PB-CA with the three parameters are separately
plotted.
According to the definition of m and A, A = 1 and m = 0.8 are reasonably set.
The variation range of T is set from 0.3 to 0.9, and the calculated results of PB-CA for
different values of parameter T are plotted with δ, as shown in Fig. 3.3.
Figure 3.3 shows that element P, which reflects the nonfaulted phaseB-CA , will
misfire near δ = 270° when parameter T is large, that is, in the case of a large

Fig. 3.3 Curves of PB-CA

calculations with δ for 360
changes in parameter T T=0.3
T=0.5
270
T=0.7
Results of PB-CA

T=0.9
180

90

0
0 45 90 135 180 225 270 315 360
δ
116 3 Immunityin Distance Protection of Oscillations

Fig. 3.4 Curves of PB-CA

calculations with δ for 360
changes in parameter m m=0.3
m=0.5
270 m=0.7

Results of PB-CA
m=0.9
180

90

0
0 45 90 135 180 225 270 315 360
δ

power outlet, fault PB-CA has the possibility of false operation in the case of power
system oscillation. In all other cases (including the case of a simple fault in the
system without oscillations), the element PB-CA , which reflects the nonfaulted phase,
is reliably inoperative.
Next, T = 0.7 and A = 1 are set. According to the definition of m, the range of m
is set from 0.1 to 0.7, and the calculated results of relay PB-CA for different values of
parameter m are plotted with δ, as shown in Fig. 3.4.
As shown in Fig. 3.4, when m is set near 0.5, element P, which reacts to the
nonfaulted phaseB-CA , will falsely activate when the power system is oscillating and
an internal fault occurs. When the end of the protection range (i.e., rectification point)
is near the center of oscillation, PB-CA will falsely activate when the power swing
angle on both sides of the system falls within a certain range, and the closer it is to the
center of oscillation, the greater the range of false activation. When the rectification
point is far from the center of oscillation, PB-CA will reliably operate at all oscillation
angles.
Considering the large voltage dips in extreme cases and for the completeness of
the analysis, T = 0.7, m = 0.8 and A varies from 0.5 to 1.1. The calculated results
of the relay PB-CA for different values of parameter A are plotted with δ, as shown in
Fig. 3.5.
According to Fig. 3.5, when the value of A is too small, PB-CA will be faulty
when the power system is oscillating and there is a fault in the zone. A small value
of A means that the protection is installed on the supply side and that the supply
electromotive force on the supply side is much larger than the supply electromotive
force on the receiving side. In the case shown, where T = 0.7, m = 0.8 and A < 0.7,
a false alarm occurs. Unless a voltage collapse occurs, a voltage drop on one side to
70% of the voltage on the other side is almost impossible in a power system.
PC-AB has a similar conclusion to PB-CA , with the exception that the oscillation
angle at which the malfunction occurs is different.
3.2 Multiphase Compensated Distance Relay Performance Analysis 117

Fig. 3.5 Curves of PB-CA

calculations with δ for 360
changes in parameter A A=0.5
A=0.7
270
A=0.9

Results of PB-CA
A=1.1
180

90

0
0 45 90 135 180 225 270 315 360
δ

A comprehensive analysis of this section shows that the comparison elements

that reflect the nonfaulted phase in the case of a single-phase earth fault in most
oscillations have correct action results. False operation is only possible when the
following three conditions are met and when they achieve a certain degree of fit:
(1) A major power outlet failure.
(2) The end of the protective range is in the vicinity of the center of oscillation.
(3) The protection is installed on the supply side, and the supply electromotive force
on the supply side is much greater than the supply electromotive force on the
receiving side.
There is a similar conclusion for a two-phase ground fault in the system. The
analysis process is omitted, and the conclusion is directly written as follows:
If the distance from the end of the protection range to the local power supply
is greater than the distance to the opposite power supply when a zone fault occurs,
the comparator PA-BC , which reflects the faulty phase, may reject the zone when
the oscillation develops to an angle of approximately 180° between the two power
supplies. Due to the BC-phase fault, the other two comparators PB-CA and PC-AB can
also reflect the fault. These two elements also operate when the power supply angle
on both sides of the protection installation is less than 90° ahead of the opposite
power supply angle; otherwise, they do not operate.
2. Oscillation accompanied by positive external faults
Similar to the previous analysis, only the schematic and conclusions of the system in
the case of a single-phase ground fault are given here. The schematic of the system in
the case of a positive external fault is shown in Fig. 3.6, the symbols in the diagram
are defined as shown in Fig. 3.1.
In the event of power system oscillations and positive external faults, the three
elements of the multiphase compensated distance relay have the following properties:
(1) The case where the end of the protection range is within the center of oscillation
or where the angle of oscillation δ < 135° or δ > 225°, the components that react
118 3 Immunityin Distance Protection of Oscillations

Fig. 3.6 Schematic of the

forward external fault system

to the fault phase are reliably inoperative, and the more precise correct range
of operation is related to the relative position of the end of the protective range
and the center of oscillation.
(2) When the end of the protective range is outside the center of oscillation and the
oscillation develops to an angle of 135° < δ < 225° between the two ends of
the power swing, the element that reflects the faulty phase may misfire, and the
more precise misfire range is related to the relative position of the end of the
protective range and the center of oscillation.
(3) The end of the protection range is not near the center of oscillation or the angle
of oscillation δ < 90° or δ > 225°. The components reacting to the nonfaulted
phase are reliably inoperative, and the more precise correct range of operation
is related to the relative position of the end of the protective range and the center
of oscillation.
(4) When a large power outlet fault occurs and the end of the protection range is
near the center of the oscillation, there is a risk that the element reacting to the
nonfaulted phase will misfire at 90° < δ < 225°.
3. Oscillation with reverse out-of-area fault
Here, only the schematic and conclusions of the system in the case of a single-phase
ground fault are given. In the event of a reverse external fault in the system, the
schematic is shown in Fig. 3.7, and the symbols in the diagram are defined as shown
in Fig. 3.1.
In the event of power system oscillations and a reverse external fault, the three
elements of the multiphase compensated distance relay have the following properties:
(1) At the end of the protection range outside the center of oscillation or at an
angle of oscillation δ < 135° or δ > 225°, the components reacting to the faulty
phase are reliably inoperative, and the more precise correct range of operation

Fig. 3.7 Schematic of the

reverse external fault system
3.2 Multiphase Compensated Distance Relay Performance Analysis 119

is related to the relative position of the end of the protective range and the center
of oscillation.
(2) When the end of the protective range is within the center of oscillation and the
oscillation develops to an angle of 135° < δ < 225° between the two ends of
the power swing, the element that reflects the faulty phase may misfire, and the
more precise misfire range is related to the relative position of the end of the
protective range and the center of oscillation.
(3) At the end of the protection range on the inside of the oscillation center or at an
oscillation angle δ < 135° or δ > 270°, the elements that react to the nonfaulted
phase are reliably inoperative, and the more precise correct range of operation
is related to the relative position of the end of the protective range and the center
of oscillation.
(4) When the end of the protection range is inside the oscillation center, the opposite
side power supply is a large power supply or a long-distance reverse fault occurs,
and the power supply swing angle at both ends is near 135° < δ < 270°, there is
a possibility of a reverse, out-of-area fault, misoperation situation.

3.2.4 Effect of Transition Resistors on Multiphase

Compensated Distance Relays

The analysis in the previous section omitted the effect of the transition resistance on
the multiphase compensated distance relay. This section will analyze the operation
with the transition resistor case.
Using the phase A fault as an example, the three-phase compensated voltage form
can be expressed as Eqs. (3.10) to (3.12). In the case of the phase A ground fault, the
action equation of element PA-BC is expressed as Eq. (3.13).

U̇ A = (1 − m) Ė M A + m Ė N A
  
m(1 − t) 2Z  + Z 0 (1 − t) Ė M A + t Ė N A
−   (3.10)
3Rg + t(1 − t) 2Z  + Z 0
 
U̇ B = α 2 (1 − m) Ė M A + m Ė N A
  
m(1 − t) Z 0 − Z  (1 − t) Ė M A + t Ė N A
−   (3.11)
3Rg + t(1 − t) 2Z  + Z 0
 
U̇C = α (1 − m) Ė M A + m Ė N A
  
m(1 − t) Z 0 − Z  (1 − t) Ė M A + t Ė N A
−   (3.12)
3Rg + t(1 − t) 2Z  + Z 0
120 3 Immunityin Distance Protection of Oscillations
⎧ U̇  m(1−t)(2Z  +Z 
⎪ =√j + √
0
)((1−t) Ė M A +t Ė N A )
⎪
⎨ U̇ BC

A
3 j 3[3Rg +t(1−t)(2Z  +Z  0
)][(1−m) Ė M A +m Ė N A ]
Z M +Z f Z 0M +Z 0f (3.13)
⎪ t = =
⎪
⎩
Z 0
Z
m = Z MZ+Z
set

The effect of the transition resistance on Eq. (3.13) is reflected in the second term.
In the second term, Ė M A and Ė N A are present in the numerator and denominator in
the same form. Therefore, the actual values can be disregarded and taken as Ė M A = 1
◦
and Ė N A = 1∠10 . The relative sizes of Z 0 and Rg and the values of the parameters
m and t have a great influence on the calculation results of Eq. (3.13). The values
of these parameters are closely related to the actual system parameters. Therefore,
an example is given below to illustrate the effect of the transition resistance on the
multiphase compensated distance relay.
The structure of the simulation system is shown in Fig. 3.1, where the positive
sequence parameters are Z M = 1.8 + j2.8, Z N = 1.8 + j2.8 and Z L = 3.6 +
j5.6; the zero sequence parameters are Z 0M = 16 + j7.47, Z 0N = 16 + j7.47
and Z L0 = 32 + j14.94. The length of the protected line is 200 km, and the positive
sequence rectification value Z set = 25.6 + j11.95, which is equivalent to 160 km.
The above parameters are substituted into Eq. (3.13), and the transition resistance
Rg is changed from 0  to 400  for calculation. With different values of transition
resistance selected, the farthest fault distance that can cause PA-BC to act (referred to
as its action boundary in this paper) is plotted, as shown in Fig. 3.8.
As shown in Fig. 3.8, the protection range (i.e., action boundary) sharply decreases
with increasing transition resistance. When the transition resistance is 0 , the protec-
tion range is 160 km, which is exactly the rectification distance. When the transition
resistance increases to just 50 , the relay completely loses its protection range.
Thus, the transition resistance has a great influence on this relay.

Fig. 3.8 Action boundary of 200

PA-BC for different values
of Rg
Operation zone/ km

150

100

50 Rg=50Ω

0
0 100 200 300 400
Rg /Ω
3.3 Distance Protection from Power System Oscillations 121

3.3 Distance Protection from Power System Oscillations

3.3.1 Improved Multiphase Compensated Distance Protection

1. Multiphase compensated distance relays with resistance to transition resistance

As shown by Eqs. (3.11) and (3.12), since the second terms of U̇ B and U̇  C are equal,
they cancel out in the calculation of U̇  BC . Thus, the transition resistor Rg has no
effect on U̇  BC , and its effect on PA-BC mainly affects the real part of the denominator
of the second term of U̇  A . The second term of U̇  A is separately presented and defined
as RES, as shown in Eq. (3.14). Next, Rg influences the result of PA-BC by affecting
the real part of the denominator of RES. If this effect can be removed, the effect of
the transition resistance on P can be eliminated A-BC . When Rg = 0, RES = -m/t, i.e.,
the effect of the transition resistance can be eliminated if -m/t can be obtained [220].
  
m(1 − t) 2Z  + Z 0 (1 − t) Ė M A + t Ė N A
RES = −   (3.14)
3Rg + t(1 − t) 2Z  + Z 0

If the effect of Rg on the real part of the RES denominator is to be eliminated, the
following two problems need to be solved:
(1) How can RES be obtained under the condition that only three-phase voltage and
three-phase current can be collected at the protection installation?
(2) After obtaining RES, how can the effect of Rg on its denominator part be
accurately eliminated?
To extract RES, determine a measurable or calculable electrical quantity that has a
form similar to that of Eq. (3.14). Additionally, this electrical quantity should be a
quantity near the end of the calibration range (i.e., near the compensated point) to
ensure that the multiphase compensated distance relay retains its good characteristics
and remains unaffected by oscillations.
Equation (3.15) gives the analytical expression for the zero sequence compensated
voltage U̇  0 . A comparison of Eqs. (3.14) and (3.15) shows that U̇  0 is extremely
similar to the RES form. They differ only by a factor related to Z 0 and Z  . This
coefficient can be obtained from the parameter k defined in Eq. (3.2) and is necessary
for distance protection. RES is expressed using parameter k with the zero sequence
compensated voltage U̇  0 , as shown in Eq. (3.16).

U̇  A + U̇  B + U̇  C
U̇  0 =
3
m(1 − t)Z 0  
= −   Ė M A (1 − t) + Ė N A t (3.15)
3Rg + t(1 − t) 2Z + Z 0
3k + 3 
RES = U̇ 0 (3.16)
3k + 1
122 3 Immunityin Distance Protection of Oscillations

Equation (3.14) shows that the numerator of the RES contains the term of the
power supply potential at both ends, which is a rotating term that varies at an operating
frequency and should be removed if the effect of Rg on the RES is to be removed. This
term in the RES is (1 − t) Ė M A + t Ė N A , which is the voltage at the fault point before
the fault occurred and cannot be calculated or measured. (1 − m) Ė M A + m Ė N A (i.e.,
U̇ A − R E S) for the rectification point voltage, considering the normal operation of
the system when the line voltage difference between two points is not large [239],
can be applied instead of (1 − t) Ė M A + t Ė N A to eliminate the rotating term caused
by the electric potential of the power supply at both ends. The part obtained after the
rotating term is eliminated is defined as COM, as shown in Eq. (3.17).

m(1−t)Z e jϕ
COM = RES
U̇  A −R E S
= − 3Rg +t(1−t)Z e jϕ (3.17)
Z e =2Z  + Z 0
jϕ

When the assumptions of Eq. (3.18) are available, take the inverse of COM in
Eq. (3.17) to obtain a complex number expressed as shown in Eq. (3.19). If A in
Eq. (3.19) can be eliminated by calculation, the effect of the transition resistance on
the relay can be eliminated.

3R
A = − m(1−t)Z
g

(3.18)
B = −mt

1
= (A cos ϕ + B) − j Asinϕ (3.19)
COM
The effect of Rg can be eliminated by transforming the COM shape, as shown in
Eq. (3.20).

1
C O M =     (3.20)
r eal 1
COM
+ imag C O1 M ctanϕ

Ultimately, the improved compensated voltage is shown in Eq. (3.21), and the
improved equation of action for PA-BC is shown in Eq. (3.22).
  
U̇  AComp = C O M  +1 U̇ A − R E S (3.21)

U̇  AComp
PA−BC : 360◦ > arg > 180◦ (3.22)
U̇  BC

For phases B and C, the compensated voltage can be similarly improved, but the
expressions of COM for phase B and C are different from that for phase A, and are
written as
3.3 Distance Protection from Power System Oscillations 123

C O MB = RES
α 2 (U̇  A −R E S )
(3.23)
C O MC = RES
α (U̇  A −R E S )

Notably, in some cases, the presence of U̇  0 = 0 leads to COM = 0 and thus to

infinity in the inverse calculation. Thus, the presence of U̇  0 should be included as a
criterion in practice. Although the improved method in this paper uses the protection
rectification point voltage instead of the fault point voltage, this approach does not
destroy the good characteristics of the multiphase compensated distance relay that
“the power system oscillates but does not reliably operate in the absence of faults”.
When the system is simply oscillating, no zero sequence voltage is generated, and
the action equation is not modified. The improved characteristics in this case are
identical to the original characteristics.
The above approximation was made using (1 − m) Ė M A + m Ė N A instead of
(1 − t) Ė M A + t Ė N A , and in the actual system, these two values slightly deviate.
The error introduced by this deviation to the improved multiphase compensated
distance relay is analyzed below. Assumptions.

(1 − t) Ė M A + t Ė N A
e jθ = (3.24)
(1 − m) Ė M A + m Ė N A

Pattern (3.20) becomes

sin ϕ
C O M = 3Rg
(3.25)
m(1−t)Z
sin θ − t
m
sin(ϕ − θ )

In the above equation, ϕ is the system impedance angle; θ is the phase angle
difference between the fault point voltage and the rectification point voltage when
there is no fault in the system. θ decreases with increasing fault distance when the
protection is installed on the delivery side; θ increases with increasing fault distance
when the protection is installed on the receiving side.
Comparing observation Eqs. (3.21), (3.22) and (3.25) shows that the larger the
denominator of COM’ is, the more the protection tends to act and that the smaller the
denominator of C O M  is, the less the protection tends to act. Since the sine function
is an increasing function between −90° and 90°, the smaller ϕ-θ is, the more likely
the protection is to act, and the larger ϕ-θ is, the less likely the protection is to act.
When the protection is installed on the delivery side, θ decreases with increasing
fault distance, and the protection is more inclined to actuate during internal faults and
to inactivate during external faults. The error in the voltage approximation renders
the improved multiphase compensated distance relay more effective.
When the protection is installed on the receiving side, θ increases with increasing
fault distance, and the protection is more inclined to inactivate during internal faults
and to actuate during external faults. The error in the voltage approximation renders
the improved multiphase compensated distance relay less effective. However, this
124 3 Immunityin Distance Protection of Oscillations

result does not mean that the method in this paper cannot be applied to the receiving
side; it only means that the receiving side is not as effective as the sending side.
The line impedance angle ϕ is employed in the modified multiphase compensated
distance relay. Although the line impedance angle is readily available in the field,
the error introduced into the modified multiphase compensated distance relay when
the value of the line impedance angle ϕ deviates from the actual value (α) is given
for completeness of the analysis.
When the line impedance angle is taken to be ϕ + δ, Eq. (3.25) becomes

sin(ϕ+α)
C O M = 3Rg
(3.26)
m(1−t)Z
sin(θ −α) − t
m
sin(ϕ − θ +α)

As analyzed in the previous section, the smaller ϕ-θ + α is, the easier the protec-
tion is to act, and the larger ϕ-θ + α is, the less likely the protection is to act. Relative
to the actual value of the line impedance angle, if the estimated value is larger, the
protection tends to refuse to move within the zone; if the estimated value is smaller,
the protection tends to misfire outside the zone. Therefore, the actual application
should be as accurate as possible to estimate the line impedance angle under the
premise of taking a slightly larger estimated value. Additionally, the larger ϕ is, the
smaller the effect of α on ϕ. Therefore, the relay proposed in this paper is suitable
for lines with relatively small line resistance and long lines, that is, it is applicable
to high voltage lines.
Applying the improved multiphase compensated distance relay to the arith-
metic example in Sect. 3.2 gives the action boundary of the improved multiphase
compensated distance relay, as shown in Fig. 3.9.
The solid line in the figure shows the operation boundary of the multiphase
compensated distance relay for the case of accurate estimation of the line impedance
angle. When accurately estimating the line impedance angle case, the improved
multiphase compensated distance relay proposed in this paper is completely inde-
pendent of the transition resistance, and it is easy to accurately estimate the fault

Fig. 3.9 Action boundary of 200

the improved multiphase
compensated distance relay
Operation zone /km

150

100
Proposed relay α= 0°
50 Proposed relay α= 5°
Traditional relay
0
0 100 200 300 400
Rg /Ω
3.3 Distance Protection from Power System Oscillations 125

impedance angle. Even if there is an error in the estimation of the line impedance
angle, as shown in the dashed line in the figure, its performance is far better than that
of the conventional multiphase compensated distance relay (dotted line).
2. Out-of-area false-action blocking elements
As analyzed in Sect. 3.2, if a forward or reverse external fault occurs when the system
is oscillating, the multiphase compensated distance relay will misfire. To avoid this
situation, an external false-action blocking element is needed.
For the case of misoperation outside the forward zone, the blocking element Q is
set. The action criterion for the Q element is also set.
  
Q ϕ : Z Y I˙ϕ +k I˙0 > U̇ϕ , ϕ= A, B, C
(3.27)
Q ϕϕ : Z Y I˙ϕϕ > U̇ϕϕ , ϕϕ= AB, BC, C A

Z Y is 2–2.5 times the rectified impedance Z set of the P element I. U̇ϕ and U̇ϕϕ are
the protection installation point voltages A-BC . For example, if QA operates and QBC
does not operate, it is considered an A-phase fault; if QBC operates and QA does not
operate, it is considered a BC-phase fault; if QA and QBC simultaneously operate, it
is considered an oscillation or three-phase fault. If only either QA or QBC is active
and the other is inactive does the calculation result of PA-BC open. Additionally, the
Q element assists the P element in distinguishing between single-phase faults and
two-phase faults and acts as a phase selector.
To solve the problem of oscillation and false activation of faults outside the reverse
zone, it is sufficient to simply establish the directional elements. A zero-sequence
directional element D0 is selected for single-phase faults, and a negative-sequence
directional element D2 is selected for two-phase faults. The action criteria for D0
and D2 are

U̇s
360◦ > arg > 180◦ (3.28)
I˙s

In the above equation, U̇s and I˙s represent the zero sequence (negative sequence)
voltage and current, respectively, of the protection measurement point.
Figures 3.10 and 3.11 give the composition of the components that reflects single-
phase faults and those that reflect phase-to-phase faults (the components reacting to
phase A faults and BC faults are applied as examples).

Fig. 3.10 Component PA-BC

composition logic diagram
for reaction to a single-phase QA
& Phase A fault
ground fault QBC
D0
126 3 Immunityin Distance Protection of Oscillations

Fig. 3.11 Component PA-BC

composition logic diagram
for reaction to an interphase QBC
& Phase BC fault
fault QA
D2

3. Symmetrical fault detection elements

Multiphase compensated distance relays can only reflect asymmetric faults but cannot
reflect not three-phase faults. Although three-phase faults are rare, this paper gives a
protection composition scheme to detect three-phase faults for the sake of protection
integrity.
Since the three-phase current and three-phase voltage are symmetrical when a
three-phase fault occurs, any phase-to-phase current and phase-to-phase voltage can
be utilized for measurement. To distinguish a three-phase fault from a single-phase
fault and two-phase fault, a single-phase criterion and two-phase criterion are concur-
rently applied to determine them. In this paper, phase A and BC interphases are
selected as criteria.
⎧
⎨ M A : 270◦ > arg U̇ A|o| > 90◦
U̇ A
(3.29)
⎩ M BC : 270◦ > arg U̇ BC|o| > 90◦
U̇ 
BC

In formula (3.29), U̇  A and U̇  BC are the compensated voltages of phase A and

phase BC, respectively, at the current moment; U̇ A|o| and U̇ BC|o| are the voltages
of phase A and phase BC, respectively, at the protection measurement point 40 ms
before the current moment; and the logical relationship between the two criteria is
expressed by the “with” logic.
To prevent the abovementioned criterion from being misactivated when an oscil-
lation occurs in the system without a fault, the “with” logic is formed by using the
phase-to-phase frequency variation distance element ΔM and the abovementioned
criterion. ΔM’s criterion is.

U̇ BC|o|
M : 270◦ > arg 
> 90◦ (3.30)
U̇ BC


In the above equation, U̇ BC is the compensated voltage of the BC phase at the

current moment, and U̇ BC|o| is the compensated voltage of the BC phase before 40 ms
at the current moment. If the system is in normal operation, the current compensated
 
voltage U̇ BC between the BC phase and the compensated voltage U̇ BC|o| after 40 ms
should be equal, the ΔM criterion calculation result is 0°, and ΔM will not operate.
When the system is in oscillation without faults, considering the long period of
 
oscillation, usually on the order of seconds, the difference between U̇ BC and U̇ BC|o|
is only 40 ms, which does not produce a particularly large phase angle difference,
3.3 Distance Protection from Power System Oscillations 127

Fig. 3.12 Logic diagram of MA

the composition of the M M
MBC &
component
ΔM

and ΔM does not act. The composition of element M that reflects a three-phase fault
is shown in Fig. 3.12.

4. Overall relay operation logic

The analysis of the above sections is combined to give the block diagram of the whole
group action logic of the distance relay that is both unaffected by power system oscil-
lations and unaffected by transition resistance proposed in this paper, as shown in
Fig. 3.13. All asymmetric faults are detected by the modified multiphase compen-
sated distance relay (P) with a forward external misoperation blocking element (Q)
and reverse external misoperation blocking element (D), and three-phase faults are
detected by a symmetric fault detection element (M).

Fig. 3.13 Block diagram of PA-BC

distance relay logic
QA & Phase A fault

QBC
& Phase BC fault

PB-CA
QB & Phase B fault

QCA
& Phase CA fault

PC-AB
QC & Phase C fault

QAB
& Phase AB fault
D0
D2
M Phase ABC fault
128 3 Immunityin Distance Protection of Oscillations

3.3.2 Distance Protection Based on Information from Both

Ends

Both the multiphase compensated distance protection and the proposed improved
multiphase compensated distance protection distinguish between internal faults and
external faults by comparing the magnitude of 1 with that of m/t. Although the
improved multiphase compensated distance protection improves the transition resis-
tance, it still has certain shortcomings: in the derivation process, to simplify the
calculation, the strong assumptions that the impedance angle is constant in all parts
of the system and that the positive sequence network structure of the system is iden-
tical to the negative. In the process of calculating m/t, it is necessary to pairwise
estimate the equivalent impedance of the end system. The deviation in the strong
assumption of the system from the actual system and the deviation in the end-system
equivalent impedance estimation may affect the performance of the protection in the
practical application of the derived m/t and the desired m/t with large deviations.
This subsection proposes a distance protection method that is not affected by oscil-
lations and that has a strong resistance to transition resistance. This method follows
the idea of multiphase compensated distance protection by comparing the magnitude
of 1 with that of m/t to distinguish between internal faults and out-zone faults.
1. Basic principles
The two-terminal system shown in Fig. 3.1 is still accordingly illustrated. In the subse-
quent derivation, only the positive sequence parameters of the system are assumed
to be equivalent to the negative sequence parameters, and the parameters m 1 , m 0 , t1
and t0 are defined as
⎧
⎪
⎪ m 1 = Z M1Z+Z set1
⎪
⎪
⎪
⎨ t1 = Z M1Z+Z f 1
(3.31)
⎪
⎪ m 0 = Z M0Z+Z set0

⎪
0
⎪
⎪
⎩ t0 = Z M0 +Z f0
Z0

The subscripts “1” and “0” denote the positive sequence values and zero sequence
values, respectively, of the corresponding impedances in Fig. 3.1. The expressions for
the three-phase compensated voltage under a single-phase ground fault, two-phase
short-circuit fault and two-phase short-circuit ground fault are derived.
When a single-phase ground fault occurs, taking a phase A ground fault as an
example, the expression for the three-phase compensated voltage derived from the
current voltage on the M side after the fault occurs is
3.3 Distance Protection from Power System Oscillations 129
⎧  
⎪ 2m 1 (1−t1 )Z +m 0 (1−t0 )Z 0
⎪
⎪ 
U̇ AM = Ė m − 3R +2t (1−t )Z +t (1−t )Z 0 Ė t
⎪
⎪
⎨  1 
g 1 0 0

m 1 (1−t1 )Z −m 0 (1−t0 )Z 0

U̇ B M = α Ė m + 3R +2t (1−t )Z +t (1−t )Z 0 Ė t
2 (3.32)
⎪
⎪  g 
⎪
1 1 0 0
⎪
⎪ m 1 (1−t1 )Z −m 0 (1−t0 )Z 0
⎩ U̇  = α Ė + Ė t
CM m 3R +2t (1−t )Z +t (1−t )Z 0
g 1 1 0 0

where Ė m = (1 − m) Ė M A + m Ė N A and Ė t = (1 − t) Ė M A + t Ė N A . The zero

compensated voltage corresponding to the M side can be obtained from Eq. (3.32)
as

m 0 (1 − t0 )Z 0
U̇  0M = − Ė t (3.33)
3Rg + 2t1 (1 − t1 )Z + t0 (1 − t0 )Z 0

If Z L − Z set is considered the rectified value of the protection on the N side of

the line, the expression for the three-phase compensated voltage obtained with the
current voltage on the N side is
⎧  
⎪ 2t1 (1−m 1 )Z +t0 (1−m 0 )Z 0
⎪
⎪ 
U̇ AN = Ė m − 3R +2t (1−t )Z +t (1−t )Z 0 Ė t
⎪
⎪
⎨  1 
g 1 0 0

t1 (1−m 1 )Z −t0 (1−m 0 )Z 0


U̇ B N = α Ė m + 3R +2t (1−t )Z +t (1−t )Z 0 Ė t
2 (3.34)
⎪
⎪  g 
⎪
1 1 0 0
⎪
⎪ t1 (1−m 1 )Z −t0 (1−m 0 )Z 0
⎩ U̇  = α Ė + Ė t
CN m 3R +2t (1−t )Z +t (1−t )Z 0
g 1 1 0 0

The zero compensated voltage corresponding to the N side can be obtained from
Eq. (3.34) as

t0 (1 − m 0 )Z 0
U̇  0N = − Ė t (3.35)
3Rg + 2t1 (1 − t1 )Z + t0 (1 − t0 )Z 0

With Eqs. (3.33) and (3.35), we can calculate

U̇  0M m 0 (1 − t0 ) Z M0 + Z set0 Z N 0 + Z L0 − Z f 0
= = · (3.36)

U̇ 0N t 0 (1 − m 0 ) Z M0 + Z f 0 Z N 0 + Z L0 − Z set0

An analysis of Eq. (3.36) shows that the magnitude of the result of Eq. (3.36)
is greater than 1 when an internal fault occurs and less than 1 when an external
fault occurs. Therefore, by detecting the magnitude of U̇  0M U̇  0N , it is possible
to distinguish between internal single-phase ground faults and external single-phase
ground faults. In addition, a comparison of Eq. (3.33) and Eq. (3.35) reveals that the
term containing the transition resistance is in the denominator of both expressions,
so the differentiation between internal faults and external faults using Eq. (3.36) is
independent of the transition resistance.
130 3 Immunityin Distance Protection of Oscillations

Similarly, when a two-phase short-circuit fault occurs, taking the BC-phase short-
circuit fault as an example, the corresponding negative sequence compensated voltage
on both sides of the system can be calculated as

1 m 1 (1 − t1 )Z 
U̇  2M = Ė t (3.37)
2 t1 (1 − t1 )Z  + R p
1 (1 − m 1 )t1 Z 
U̇  2N = Ė t (3.38)
2 t1 (1 − t1 )Z  + R p

where R p is the transition resistance between phases, and Ė t = (1 − t) Ė M A +t Ė N A .

With Eqs. (3.37) and (3.38), we can calculate.

U̇  2M m 1 (1 − t1 ) Z M1 + Z set1 Z N 1 + Z L1 − Z f 1
= = · (3.39)

U̇ 2N t1 (1 − m 1 ) Z M1 + Z f 1 Z N 1 + Z L1 − Z set1

When a two-phase, grounded short-circuit fault occurs, the corresponding zero-

sequence compensated voltages on both sides of the system can be obtained as
follows, taking the BC-phase grounded short-circuit fault as an example:

m 0 (1 − t0 )Z 0
U̇  0M = Ė t (3.40)
6Rg + 3R p + 2t0 (1 − t0 )Z + t1 (1 − t1 )Z 0

(1 − m 0 )t0 Z 0
U̇  0N = Ė t (3.41)
6Rg + 3R p + 2t0 (1 − t0 )Z + t1 (1 − t1 )Z 0

With Eqs. (3.40) and (3.41), we can calculate.

U̇  0M m 0 (1 − t0 ) Z M0 + Z set0 Z N 0 + Z L0 − Z f 0
= = · (3.42)

U̇ 0N t0 (1 − m 0 ) Z M0 + Z f 0 Z N 0 + Z L0 − Z set0

With Eqs. (3.36), (3.39), and (3.42), it is possible to distinguish between internal

faults and external faults by calculating the magnitude of U̇  0M /U̇  0N or U̇  2M U̇  2N
when an asymmetric fault occurs and is not affected by the transition resistance.
Considering the case of a fault in oscillation, the expressions for the three-phase
compensated voltages corresponding to the M and N sides in the case of an A-phase
ground fault, a BC-phase short-circuit fault, and a BC-phase short-circuit ground fault
in oscillation are similar to those in the case of a simple fault, with the exception that
the amplitudes and phases of Ė m and Ė t in the various equations periodically  vary
compared with the case of only a line fault. However, when using U̇  0M U̇  0N or
U̇  2M /U̇  2N to identify a fault, Ė m and Ė t are eliminated from the calculation process
and do not affect the final result. Therefore, the method of distinguishing internal
faults from external faults using U̇  0M /U̇  0N or U̇  2M /U̇  2N is also not affected by
system oscillations.
3.4 EMTP Simulation Experiments 131

Based on the above analysis, the differentiation between the internal fault criterion
and external fault criterion can be written as
 
U̇0M U̇2M
K0 = 
> 1||K 2 = 
>1 (3.43)
U̇0N U̇2N

If the magnitude of U̇  0M /U̇  0N or U̇  2M /U̇  2N satisfies Eq. (3.43), then an internal

fault is considered to have occurred; conversely, an external fault is considered to
have occurred. Notably, the criterion uses the negative sequence component and zero
sequence component, so when using Eq. (3.43), the magnitudes of U̇  0M , U̇  0N , U̇  2M
and U̇  2N need to be judged, i.e., Eq. (3.43) is used to identify the fault only when
their magnitudes reach a certain value. Of course, the protection is also unable to
identify three-phase short-circuit faults and still requires a module with three-phase
short-circuit fault identification.
2. Protection flow chart
The corresponding flow chart for this protection is shown in Fig. 3.14. After the
protection is activated, the calculation uses the current and voltage on this side to
calculate the corresponding negative and zero sequence compensated voltage and
receives the calculated sequence compensated voltage from the remote side. The
calculated negative sequence or zero sequence compensated voltage is compared
with the threshold value. If the voltage is less than the threshold value, then end
 the
process; if it is greater than the threshold value, then the amplitude of U̇  0M U̇  0N

or U̇  2M U̇  2N is calculated and compared with 1; if the amplitude is greater than 1,
the protection acts, and vice versa.

3.4 EMTP Simulation Experiments

3.4.1 Simulation System

To verify the correctness of the distance relay that is not affected by both power
system oscillations and transition resistance, as proposed in this paper, a simulation
model is built in PSCAD for simulation experiments. A schematic of the simulation
system is shown in Fig. 3.15.
In Fig. 3.15, point R is the protection installation point, point F is the fault point,
and point Y is the end of the I section protection; SM and SN are the power supply at
both ends of the system, and their electric potentials are Ė M and Ė N , respectively; M
and N are the busbars at both ends of the protected line; U̇ M and I˙M are the voltage
and current, respectively, measured by the protection; I˙f is the fault current; Rg is
the transition resistance; and ZM , ZL , ZN , Zf , Zset and Z are the impedance values
of the corresponding lines in the figure. The length of the protected line is 200 km,
and the rectified distance of Section I is 160 km.
132 3 Immunityin Distance Protection of Oscillations

Fig. 3.14 Protection flow Start

chart

Sampling

Calculate & 2 and & 0

Obtain & 2 and & 0 from

another side˛
No
Yes

&
0M 0
No
Yes
&
2M 2
No
Yes
& & 1
0M 0N
No & & 1
2M 2N
Yes
No
Yes

Relay acts

End

Fig. 3.15 Schematic of the

simulation system
3.4 EMTP Simulation Experiments 133

Table 3.2 Line impedance

Impedance Positive sequence () Zero sequence ()
values for each section
ZM 0.8 + j28 16 + j75
ZL 0.8 + j28 16 + j75
ZN 1.6 + j56 32 + j150
Zf 0.4 + j14 8 + j37.5
Z set 1.44 + j22.4 12.8 + j60

To simulate power system oscillations, the frequency of SM is set to 50.4 Hz, and
the frequency of SN is set to 49.6 Hz; thus, the oscillation frequency is 0.8 Hz. The
initial phase angle difference between Ė M and Ė N is 30°, and Ė M is ahead of Ė N .
The line impedance values for each segment are listed in Table 3.2.

3.4.2 Power System Oscillations Without Fault Conditions

The electric potential difference between the two ends of the line (reflecting the power
system oscillation) and the action of the three elements of the modified multiphase
compensated distance relay, for the case when a power system oscillation occurs
without a fault, are plotted in Fig. 3.16. The shaded area in the figure represents the
action region. The action curves of each element of the relay are labeled in the figure.
As shown in Fig. 3.16, in the case of a purely oscillating system with no faults,
all three elements are able to reliably inactivate regardless of the value of the
power system oscillation angle. This capability is the advantage of the multiphase
compensated distance relay in overcoming the effects of power system oscillations.

3.4.3 Power System Oscillation with a Single-Phase Ground

Fault in the Zone

Simulation experiments are carried out with the system in oscillation with a phase
A ground fault in the zone. The phase A metallic fault occurs at 0.5 s, and the fault
distance is 100 km, which is within the protection range.
Figure 3.17 plots the difference between the electric potentials on both sides of the
system (to reflect the power system oscillations) and the calculation of the relevant
components PA-BC , QA, and QBC for the modified multiphase compensated distance
protection to determine the grounding of phase A. Figure 3.18 gives the action of
the distance protection based on the information at both ends. The shaded area in the
figure is the action zone; the action curves of each element of the relay are marked
by the solid black line in the figure (below); and the responding element acts when
the solid black line enters the shaded area.
134 3 Immunityin Distance Protection of Oscillations

Fig. 3.16 Calculated results 1000

of the improved multiphase
compensated distance relay

U /V
for the case of a system with
0
pure oscillations and no
faults -1000
0 1 2 3
t /s
(a) EMA-ENA

360

PA-BC /deg
180
Action curve of relay
0
0 1 2 3
t /s
(b) Calculation result of P A-BC

360
PB-CA /deg

180
Action curve of relay
0
0 1 2 3
t (s)
(c) Calculation result of P B-CA

360
PC-AB /deg

180

Action curve of relay

0
0 1 2 3
t /s
(d) Calculation results of P C-AB

3.4.4 Power System Oscillation with Out-Of-Area

Single-Phase Ground Fault Conditions

Simulation experiments are conducted for a positive external A-phase ground fault
occurring in the system during oscillation. The A-phase, metallic, positive external
fault occurs at 0.5 s, and the fault distance is 210 km, which is outside the protection
range in Section I.
Figure 3.19 plots the difference between the electric potentials on both sides of
the system (to reflect the power system oscillations) and the calculated results of the
relevant components PA-BC , QA , and QBC for determining the grounding of phase A.
As shown in Fig. 3.19, when a fault occurs outside the positive zone of phase A,
element PA-BC undergoes a periodic misoperation when the oscillation angle reaches
3.4 EMTP Simulation Experiments 135

Fig. 3.17 Calculated results 1000

for each component in the
case of a phase A ground

U /V
0
fault within the oscillation

-1000
0 1 2 3
t /s
(a) EMA-ENA

360
PA-BC /deg
180

0
0 1 2 3
t /s
(b) Calculation result of PA-BC
1000
QA /kV

500

0
0 1 2 3
t /s
(c) Calculation result of QA
1500
QBC /kV

1000

500

0
0 1 2 3
t /s
(d) Calculation result of QBC

Fig. 3.18 Distance 4

protection action based on 3
information from both ends
K0

in the case of a ground fault 2

in phase A within the
1
oscillation

0.49 0.495 0.5 0.505 0.51 0.515 0.52

t/s
136 3 Immunityin Distance Protection of Oscillations

approximately 180°. The misoperation region is marked with a vertical, black, dashed
line in the figure, which is consistent with the analysis of the action performance of
the multiphase compensated distance relay in the case of power system oscillations
accompanied by a forward external fault in Sect. 3.2. The element QBC periodically
operates with power system oscillations, and its action region is marked in the figure,
which is the misactivation region of element PA-BC . Component QA is always active
during the entire duration of the fault. According to Fig. 3.13, the given action
logic, when the elements QA and QBC simultaneously act, is judged as a three-phase
fault or power system oscillation, and the protection does not act. The simulation
example shows that the external fault blocking element Q can effectively block the
misoperation of element P in the case of power system oscillation and a positive
external fault.

Fig. 3.19 Calculated results 1000

for each component in the
case of an external phase A
U /V

0
ground fault in oscillation

1- 000
0 1 2 3
t /s
(a) EMA-ENA

360
PA-BC /deg

180

0
0 1 2 3
t /s
(b) Calculation result of PA-BC
1000

500
QA /kV

0
0 1 2 3
t /s
(c) Calculation result of Q A
1500
QBC /kV

1000

500

0
0 1 2 3
t /s
(d) Calculation result of QBC
3.4 EMTP Simulation Experiments 137

Fig. 3.20 Distance 3

protection action based on
information from both ends 2

K0
in the case of an external
phase A ground fault in 1
oscillation

0.49 0.495 0.5 0.505 0.51 0.515 0.52

t /s

Figure 3.20 plots the action of the distance protection based on information from
both ends, and Fig. 3.20 shows that this protection is not affected by the oscillation
and reliably does not act when an external phase A ground fault occurs during the
oscillation.

3.4.5 Transition Resistance Test

The improved multiphase compensated distance relay resistance to transition resis-

tance is examined in the absence of oscillations in the system. The frequency of the
power supplies S M and S N on both sides of the system in Fig. 3.15 is set to 50 Hz,
and the transition resistance Rg is set to 400 . A single-phase ground fault occurs in
the system with a fault moment of 0.5 s and a fault distance of 100 km. The potential
difference between the power supplies at both sides of the system Ė M − Ė N and the
operation results for elements PA-BC of the conventional and improved multiphase
compensated distance relays are shown in Fig. 3.21a–c, respectively. Figure 3.22
gives the corresponding actions of the distance protection based on the information
at both ends.
Figure 3.21a shows that the system does not oscillate. Figure 3.21b shows that
the conventional multiphase compensated distance relay rejects in the case of 400
 transition resistance. Figures 3.21c and 3.22 show that the improved multiphase
compensated distance relay and the distance protection based on information from
both ends immediately operate on fault occurrence and remain active for the duration
of the fault with good resistance to transition resistance.
Considering that the ultimate goal of the improved multiphase compensated
distance relay is to achieve distance protection independent of power system oscilla-
tions, it is necessary to examine its resistance to transition resistance in the presence
of power system oscillations.
The frequency of S M is set to 50.4 Hz; the frequency of S N is set to 49.6 Hz; and
the transition resistance Rg is set to 100 . An A-phase single-phase ground fault
occurs in the system with a fault moment of 0.5 s and fault distance of 100 km. The
action results of the power potential difference between the two ends of the system
Ė M − Ė N , the conventional multiphase compensated distance relay and the element
PA-BC of the modified multiphase compensated distance relay are plotted in Fig. 3.23.
138 3 Immunityin Distance Protection of Oscillations

Fig. 3.21 Comparison of 500

conventional and modified
multiphase compensated
0

U /V
distance relays against
transition resistance
-500
0 1 2 3
t /s
(a) EMA-ENA
400

PA-BC /deg
200

0
0 1 2 3
t /s
(b) Result of P A-BC using traditional MCDR
400
PA-BC /deg

200

0
0 1 2 3
t /s
(c) Result of PA-BC using improved MCDR

Fig. 3.22 Distance 3

protection against transition
resistance based on 2
K0

information from both ends

1

0.49 0.495 0.5 0.505 0.51 0.515 0.52

t /s

Figure 3.23a shows that the system oscillates. Figure 3.23b shows that the conven-
tional multiphase compensated distance relay rejects in the presence of 100  tran-
sition resistance. The improved multiphase compensated distance relay illustrated in
Fig. 3.23c periodically operates with power system oscillations, that is, the improved
multiphase compensated distance relay is periodically rejected during the duration
of the fault. However, as pointed out in Sect. 3.2, for the analysis of the multiphase
compensated distance relay in the case of oscillation with an accompanying fault
in the zone, for the instantaneous-acting Segment I, its rejection in other oscillation
angle ranges is accepted as long as the protection can be actuated when the oscillation
3.4 EMTP Simulation Experiments 139

Fig. 3.23 Comparison of 1000

conventional and modified

U/V
multiphase compensated
distance relays in oscillation
0
against transition resistance
1- 000
0 1 2 3
t /s
(a) EMA-ENA
400

PA-BC /deg
200

0
0 1 2 3
t /s
(b) Result of PA-BC using traditional MCDR
400
PA-BC /deg

200

0
0 1 2 3
t /s
(c) Result of PA-BC using improved MCDR

angle reaches a certain range. For the time-delayed action of segment II and segment
III, the solution proposed in section II is still valid.
Figure 3.24 plots the corresponding actions of the distance protection based on
two-terminal information. A comparison of Figs. 3.23c and 3.24 shows that the
two-end information-based distance protection is less affected by oscillations and
transition resistances than the improved multiphase compensated distance protection,
which quickly enters the action zone after a fault and remains in the action zone for
the duration of the fault.
According to the above four sets of simulation results, the improved multiphase
compensated distance relay, the distance protection that is not affected by power

Fig. 3.24 Distance 3

protection against transition
resistance in oscillation 2
K0

based on information from

both ends 1

0.49 0.495 0.5 0.505 0.51 0.515 0.52

t /s
140 3 Immunityin Distance Protection of Oscillations

system oscillation and the distance protection based on information from both ends
composed of it can not only reliably inactivate in the case of pure system oscillation
without fault but also reliably operate in the case of internal fault in the system. The
simulation results also show that the blocking element can reliably block the multiple
phases in the case of power system oscillation and external fault. The distance relay is
compensated for misoperation and exhibits a strong resistance to transition resistance.

3.5 Summary

To present the immunity to the distance from oscillations, this chapter analyzes the
dynamic characteristics of multiphase compensated distance protection in the face
of oscillations and faults. The analysis results show that the multiphase compensated
distance protection does not react to oscillations but is still rejected out of zone
and misactivated out of zone when faults occur in system oscillations. In addition,
the mechanism of the influence of the transition resistance on the performance of
multiphase compensated distance protection is analyzed. On this basis, an improved
multiphase compensated distance protection and a distance protection based on two-
terminal information are proposed. The simulation results show that both of the
proposed protections have the following advantages.
(1) Reliable inactivity in the absence of mere oscillation of the system.
(2) Reliable operation in the absence of oscillation in the event of a simple fault in
the system.
(3) Reliable operation in the event of oscillation accompanied by internal faults.
(4) Reliable inactivity in the event of oscillations accompanied by external faults.
(5) Good resistance to transition resistance
Chapter 4
Commutation Failure Prevention
and Control

Faced with the problems of highly inverse source-load distributions and large-scale
renewable energy consumption, China is vigorously developing high-voltage, large-
capacity, long-distance transmission technologies [1–3]. Compared with a high-
voltage AC transmission system, a high-voltage DC transmission system has the
advantages of higher voltage, higher capacity and lower cost and can realize the
asynchronous interconnection of different AC grids, which is widely considered a
key link in building the energy internet [4]. In recent years, several conventional,
high-voltage, direct-current (line-commutated converter high-voltage direct current,
LCC-HVDC) transmission lines have been in operation or are under construction in
China [6]. AC-DC hybrid grids have become a new form of grid development [7].
Complex hybrid AC-DC grids, which are characterized by large scales, multivoltage
levels and deep AC-DC integration, have emerged [8].
With the increasing capacity of high-voltage DC and the development of a single
DC drop point into a dense drop point for multiple DC feeds, the characteristics
of the current hybrid AC-DC grid have profoundly changed compared with those
of the traditional AC grid. This change is summarized by the power engineering
and academic communities as the strong direct and weak AC characteristics of the
large-scale, UHV (Ultra High Voltage) AC-DC hybrid grid, where AC and DC are
interdependent [9]. This finding renders the propagation characteristics of simple
faults in the AC hybrid grid more complex, which can easily trigger a chain reaction
in the power system, cause chain failures and eventually lead to large-scale power
outages.
A cascading fault in a hybrid AC-DC grid (cascading fault) is defined as a fault
in an AC-DC hybrid grid, in which the failure of one device causes the failure or
shutdown of other devices or systems, especially the cross propagation of simple
faults between two AC-DC grids. The trigger of a chain fault in an AC-DC hybrid
grid is a single simple fault, and the core mechanism is the interaction of the AC-
DC system. As the converter is the link that connects the AC-DC system, its fault
and failure will have the role of propagating and boosting the fault chain process.
In the AC-DC hybrid grid, a fault in the AC line, causing commutation failure or

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 141
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_4
142 4 Commutation Failure Prevention and Control

even blocking of the DC system, is the primary stage of a wide range of interlocking
faults. Studies of traditional, high-voltage, DC-commutation failure suppression are
important prerequisites for ensuring the safe and stable operation of the AC-DC
hybrid power grid.
Based on the time sequence of commutation failure, commutation failure can be
divided into a first commutation failure and successive commutation failure. Based
on the number of DC drop points, commutation failure can be divided into single-fed
DC commutation failure and multifeed DC commutation failure. In this chapter, the
mechanisms of the first commutation failure and continuous-commutation failure
of a single-fed DC are analyzed, and the simultaneous commutation failure and
successive commutation failure of a multifeed DC system are introduced. Based
on the generation mechanism of commutation failure, early warning methods and
suppression measures for these four typical commutation failures are introduced.

4.1 Analysis of the DC Commutation Failure Mechanism

4.1.1 First Commutation Failure

When the phase change between two bridge arms is completed, if the valve that
has just exited conduction does not recover its blocking capacity within a certain
period when the reverse voltage is applied, or if the phase change process has not
been completed during the reverse voltage period, the valves that were phase changed
when the valve voltage changed to positive will phase change to the original valve that
was scheduled to exit conduction, which is referred to as the first commutation failure
[10]. The root cause of the first commutation failure is the physical characteristics of
the thyristor and phase change inductor. The thyristor lacks the ability to turn itself
off and requires the AC grid to provide a reverse voltage long enough for the thyristor
carriers to compound and restore the blocking capability. The magnetic chain that
crosses the inductor is nonleaped, resulting in a continuous inductor current (Fig. 4.1).

Fig. 4.1 Characteristic curve of a thyristor

4.1 Analysis of the DC Commutation Failure Mechanism 143

id Ld
+ VT4 VT6 VT2

Lc ua
ub
ud N
uc

VT1 VT3 VT5

Fig. 4.2 Schematic of the inverter of a conventional, high-voltage, DC transmission system

The six valves VT1-VT6 of a conventional HVDC transmission system inverter

are triggered in a sequential manner. Each valve has a conduction angle of 120°.
When the valves are on, the valve current has a magnitude of i d , the current flowing
in the upper row of valves is positive, and the current in the lower row of valves is
negative (referred to as the return current) (Figs. 4.2 and 4.3).
If an AC fault occurs after the triggering of valve VT3, valve VT1 cannot recover
its forward blocking capability. The AC fault interferes with the normal commutation
process so that the first commutation failure occurs as the current in valve VT3 again
decreases to zero and the current through valve VT1 increases. Figures 4.4 and 4.5
illustrate the normal commutation failure and first commutation failure, respectively,
during the valve currents.
In the inverter, valve VT2 conducts; the phase change process from valve VT1 to
valve VT3 is shown in Fig. 4.6. According to Kirchhoff’s voltage law, the following
differential equation can be written:

di 3 di 1 di 3 d(i d − i 3 ) √ Ul sin θ
Lr − Lr = Lr − Lr = ub − ua = 2 (4.1)
dt dt dt dt n
where L r is the equivalent phase-commutation bridge arm inductance, and i 1 , i 3 ,
and Ul are the valve currents of valves VT1 and VT3 and the amplitude of the line
voltage on the net side of the converter transformer, respectively. n is the ratio of the
converter transformer; i d is the DC current, and u a , u b , and u c are the three-phase
voltages on the valve side of the converter. Ideally, the DC current i d is a constant
value Id , and the commutator bus voltage is a three-phase symmetric sine wave. The
start moment of phase change t1 (t1 = α/ω) valve VT3 is turned off, and i 3 = 0.
The end moment of phase change t2 (t2 = π − γ /ω) valve VT1 is turned off, and
i 3 = Id The phase change process, in the form of simultaneous integration on both
sides, yields
144 4 Commutation Failure Prevention and Control

id Ld id Ld
VT2 + VT2

i2 ua i2 Lc
ua
ub
ud N ud ub
uc N
i1 uc
VT1 VT3
VT3
-
-

(a) ωt = -60° to 0 ° (b) ωt = 0° to 60 °

id Ld id Ld
+ VT4 + VT4

i4 Lc i4
ua ua
ud ub ud ub
N N
uc uc
i3
VT3 i5
VT5
-
-

(c) ωt = 60 ° to 120° (d) ωt = 120° to 180°

id Ld id Ld
+ VT6 + VT6

i6 i6
ua ua
ud ub ud ub
N N
uc uc
i1
i5
VT5 VT1

- -

(e) ωt = 180° to 240° (f) ωt = 240° to 300°

Fig. 4.3 Order of conduction of each valve group

Fig. 4.4 Normal phase change at valve current

4.1 Analysis of the DC Commutation Failure Mechanism 145

Fig. 4.5 First commutation failure at valve current

Fig. 4.6 Schematic of phase

change

t2
1
i 3 (t2 ) − i 3 (t1 ) = u ba (t)dt (4.2)
2L r
t1

When the system is stable, solve for the shut-off angle γ as follows:
√ 
2n Id X c
γ = arccos + cos β (4.3)
Ul

where X c denotes the equivalent phase commutation impedance and β (β = π − α)

denotes the overrunning trigger angle. According to Eq. (4.3), the commutation
bus voltage, DC current and trigger angle have a direct effect on the phase change
process. The bus voltage falls and distorts; the DC current rises; the phase change
impedance increases; and the overtrigger angle decreases, which reduces the actual
system shutdown angle γ The first commutation failure is triggered. After the first
commutation failure occurs, the DC current flowing through the commutation valve
is bound to increase, which may cause two commutation failures [11]. The first
commutation failure occurs, and the DC current through the commutation valve
must increase, which may cause two commutation failures.
146 4 Commutation Failure Prevention and Control

4.1.2 Continuous Commutation Failure

When the first commutation failure occurs in DC, commutation failure will occur
again during its recovery process, resulting in 2 or more fluctuations in DC power,
which threatens the safe and stable operation of the system; thus, it will be referred
to as continuous commutation failure [12]. In the engineering application, if the
commutation failure of the converter is continuously detected within 200 ms, the
pole control system and VBE system will be switched to the standby system from
the duty system to avoid DC shutdown due to the control system equipment failure,
which also makes the commutation failure occurring within 200 ms count as one
commutation failure. When a commutation failure occurs again after 200 ms, it is
considered a continuous commutation failure. Continuous commutation failures can
potentially cause DC blocking. Note that continuous commutation failure differs
from “two commutation failures”. Continuous commutation failure is a system level
concept that causes long DC power fluctuations and is related to the DC control
system and recovery characteristics [13].
In essence, continuous commutation failure occurs for the same reason as the first
commutation failure because the shutdown angle fails to meet the minimum shutdown
angle requirement, i.e., the bus voltage, DC current and trigger angle fail to meet the
requirements. After the first commutation failure, the AC/DC control and protection
system starts the adjustment action. Thus, the occurrence of continuous commutation
failure is necessarily related to the AC/DC control and protection system.
1. Phase change voltage time area theory
During phase change, the transfer of current requires a certain amount of time for the
voltage to act and build up the inductive magnetic chain. The phase change voltage
acts on the phase change branch, the cumulative effect in time, which is the phase
change voltage time area. The phase change voltage time area is divided into two
parts. The first part is the stacked arc area. The stacked arc area is the process by which
the inductor builds up the inductive magnetic chain. The second part is the turn-off
area, which is the process by which the thyristor goes free after the current crosses
zero and restores the blocking capability under the action of the reverse voltage. In
the case of a constant trigger angle, the sum of the stacked arc area and turn-off area
is certain. When the stacked arc area is too large so that the turn-off area is less than
the minimum required turn-off area, the thyristor cannot be completely turned off,
thus triggering commutation failure.
The inductor builds the magnetic chain in response to the AC voltage, which is
determined by the magnitude of the current. When the current increases, it causes an
increase in the inductor magnetic chain, which makes the AC voltage take longer to
build up the chain, i.e., it requires more stacked arc area. A decrease in the AC-phase
change voltage amplitude, a phase angle shift past zero, and waveform distortion
due to harmonics cause a decrease in the amount of arc stack area provided, while
an increase in DC or phase change reactance causes an increase in the amount of arc
stack area required.
4.1 Analysis of the DC Commutation Failure Mechanism 147

The phase change process starts at the trigger delay angle α and ends at the arc
extinguishing delay angle δ. The current transfer process corresponds to the phase
change overlap angle μ. At the end of the phase change process, the arc extinguishing
angle for the thyristor to recover the forward blocking capability is γ , and β is the
trigger override angle. The interrelationships of α, δ, μ, γ , and β
⎧
⎪
⎪ α + μ + γ = 180◦
⎨
β = 180◦ − α
(4.4)
⎪γ = β − μ
⎪
⎩
δ =α+μ

During a phase change, phases B and C of the phase change branch are connected
in parallel, and the voltages of the two branches are equal, satisfying the following
relationship:

di 2 di 6
Lc + UC = L c + UB (4.5)
dt dt
In the electrical angle from α to δ during the phase change, the DC current of
valve arm 6 decreases from i d to 0, and the DC current of valve arm 2 rises to i d . The
process of current transfer, i.e., the energy stored in the inductance of valve arm 6, is
transferred to the inductance of valve arm 2. As the inductor current increases from
zero to i d , the power supply to the inductor is converted to magnetic energy, which is
stored in the magnetic field generated by the inductor current. The inductor current
is continuous; in effect, the magnetic chain that crosses the inductor is nonleaping.
The energy stored in the inductor can be expressed by the equation

I d
1 2 ψ2
Wm = Li L di L = Li d = (4.6)
2 2L
0

To establish the checks and balances between the amplitude of the phase change
voltage and the duration of the phase change process, the integration of both sides
of Eqs. (4.5) within the phase change process yields
a+μ a+μ
ω ω √
di 2 di 6
Lc − Lc dt = 2U L sin(ωt)dt (4.7)
dt dt
a a
ω ω

Assuming no change in the DC current, Eq. (the left side of (4.7)) becomes
a+μ
ω
di 2 d(i d − i 2 )
2Sμ−need = Lc − Lc dt = 2L c i d (4.8)
dt dt
a
ω
148 4 Commutation Failure Prevention and Control

This result is the amount of phase change voltage area demanded. The right-hand
side of Eq. (4.7) is the amount of phase change voltage supplied.
a+μ
ω
2Sμ− pr o = U L sin(ωt)dt (4.9)
a
ω

A fault in the AC system on the inverter side will cause the phase change voltage
of the commutation valve to decrease, at which time (1) if the phase change voltage
decrease triggers commutation failure, a short circuit exists on the inverter side and
the DC current abruptly increases; (2) if the phase change voltage decrease does not
trigger commutation failure, the DC current will also increase due to the increased
voltage difference between the rectifier side and the inverter side because the constant
current control has not produced regulation at the beginning of the fault. Therefore,
when considering the phase change process under fault conditions, the DC current
i d change cannot be disregarded, and rearranging Eq. (4.8) yields
a+μ
ω
L c [i d (α) + i d (α + μ)] = U L sin(ωt)dt = 2S (4.10)
a
ω

Equation (4.10) shows that the commutation area is no longer constant if the
change in DC current is considered, and its value is related to the DC current at
the start and end of the commutation. Accordingly, define the phase change area
requirement as

1
Sneed2 = · L c [i d (α)n + i d ((α + μ)n )] (4.11)
2
Next, the lack of phase change area at this point is

Sins = Sneed2 − S pr o (4.12)

Under normal operating conditions, Sins = 0 holds, and no commutation failure

occurs. The phase change area requirement after a failure can be considered an
additional requirement superimposed on the normal requirement, which is reduced
by the phase change area availability. The phase change is not completed, and the
phase change process needs to be extended until it is satisfied and Sins = 0. Next, the
phase change voltage time area Sμ expands to the right, the turn-off area Sγ decreases
and the turn-off angle γ also decreases. The phase change is not completed.
Figure 4.7 the shaded area in Sγ can be calculated using Eq. (4.13). According to
the PN junction principle, when the thyristor is subjected to a large reverse voltage,
the deionization recovery time decreases, and the required turn-off time is shortened.
When the thyristor receives a small reverse voltage, the deionization recovery time
4.1 Analysis of the DC Commutation Failure Mechanism 149

UL UL

γ min
Sμ Sγ
S μ'
S 'γ

0 α μ γ ω0t 0 ω 0t
α' μ' γ'

Fig. 4.7 Commutator phase change voltage waveform graph

increases, and the required turn-off time is lengthened.

π
ω √
Sγ = 2U L sin(ωt)d(t) (4.13)
π−γ
ω

2. Mechanism of continuous commutation failure

According to Eq. (4.9), the commutation area is provided by

1
Sμ− pr o = U L cos(γ ) − cos(β) (4.14)
2
According to Eqs. (4.9), (4.11) and (4.12), the following equation can be obtained.

L c [Id (α)n + Id (α + μ)n ] = U L cos(γ ) − cos(β) (4.15)

Next, the angle of shutoff can be obtained as.

L c [Id (α)n + Id (α + μ)n ]

γ = arccos + cos β (4.16)
UL

Continuous commutation failure occurs when the external phase change condition
continuously or intermittently fails in the time domain. Considering that the minimum
turn-off angle required for the thyristor to recover its blocking capability is γmin ,
it is generally considered that the actual turn-off angle γ > γmin ; otherwise, the
commutation valve will experience commutation failure. The turn-off angle γ <
γmin . As a criterion for judging commutation failure, continuous commutation failure
means that after the first commutation failure of the DC transmission system, the
system does not reach the original or new equilibrium point, and then commutation
150 4 Commutation Failure Prevention and Control

Fig. 4.8 Turn-off angle

waveform plot when
continuous phase change
fails

failure occurs again, corresponding to a shutdown angle γ that is multiple times less
than γmin . As Fig. 4.8 shows, due to the control system, other electrical parameters
are also in a state of violent fluctuations.
According to Eq. (4.16), the main variables that determine the magnitude of γ
are the DC current, commutation voltage amplitude, overrunning trigger angle and
the commutation reactance. The effect of γ on the variables by taking the partial
derivative yields
⎧ ∂γ
⎪
⎪ ∂ Id
= −1 · UL CL
⎪ m
⎨ ∂γ = 1 · L C · [Id (α)n +Id (α+μ)n ]
∂U L m UL UL
∂γ (4.17)
⎪
⎪ = m1 · sin β
⎪ ∂β
⎩ ∂γ
∂ LC
= m1 · [Id (α)n +IUdL (α+μ)n ]

where

2
L c [Id (α)n + Id (α + μ)n ]
m= 1− + cos β (4.18)
UL

As shown in Eq. (4.17), in the fault situation, with a change in the phase change
voltage, the change in DC current will affect the size of the shutdown angle, and the
phase change voltage to the partial derivative of the shutdown angle is the exponential
change in the phase change voltage. The three conditions affect each corresponding
constraint; therefore, the occurrence of continuous commutation failure must be 3
electrical parameters that do not cooperate, a sudden change in a parameter, and the
situation in which other parameters cannot be adjusted in time and cause a commuta-
tion failure. Among them, the fault recovery type of continuous commutation failure
is a typical case of continuous commutation failure due to the improper cooperation
of the inverter side control strategy.
Conventional DC transmission systems use hierarchical control: from top to
bottom, the main control stage, pole control stage and valve bank control stage.
The fault recovery characteristics of the DC system are mainly related to the pole
control stage when the inverter-side AC system fails. The controller in the CIGRE
HVDC standard model is the pole control stage, and the specific structure is depicted
4.1 Analysis of the DC Commutation Failure Mechanism 151

Fig. 4.9 Block diagram of inverter-side control structure of DC system

in Fig. 4.9. The controller in the CIGRE HVDC standard model is the pole control
stage, as shown in the structure.
In Fig. 4.9, Ud−inv is the inverter-side dc voltage, Id−inv is the inverter-side DC
current, Id−or der is the DC command value passed down from the main control
stage, Idr −inv is the current command value passed down from the inverter side to
the rectifier side, and βinv−i and βinv−γ are the overtrigger angle values output by
the constant current control and constant off angle control, respectively. As shown
in Fig. 4.9, the inverter side is equipped with constant current control, constant off
angle control and current error control (CEC). During the operation of the DC system,
the output of the constant-off angle control and constant-current β angle control is
selected at any given time; the greater of the two is taken according to the operating
characteristics of the inverter.. The CEC can realize the smooth transition between
constant-off angle control and constant-current control.
Under normal operating conditions, the inverter operating state is regulated by
the closed-loop control of the fixed shutdown angle so that the shutdown angle γ
is kept at the rectified value. Once a fault occurs in the AC system on the inverter
side resulting in commutation failure, the inverter operating state will be switched
to constant current control, and under the action of low voltage current limiting
control, it will recover to the operating point of the fault state and then switch to
constant shutdown angle control. In this recovery process, there is mutual switching
of constant current control, current deviation control and constant shutdown angle
control, which is likely to trigger continuous commutation failure.
When a fault occurs in the AC system, the control stage in the DC transmis-
sion system Pole is able to quickly respond and allows the postfault DC current to
operate to a new stable operating point. This fault recovery process is affected by
the combined action of multiple controllers on the rectifier and inverter sides. To
more clearly portray the process, the process is divided into three stages based on
152 4 Commutation Failure Prevention and Control

Fig. 4.10 CIGRE HVDC standard test model steady-state operating curve

the state switching of the inverter-side controllers, and the role of the controllers
at different stages is separately analyzed. To obtain mechanistic-level conclusions,
without relying on simulation analysis, the CIGRE-HVDC standard test model is
selected as an example to analyze the trajectory of the operating point of the inverter-
side system, starting from its steady-state operating curve, to obtain the mechanism
of continuous commutation failure. The steady-state operation curve is shown in
Fig. 4.10, where Ud and Id are the standardized values of the DC voltage and current,
respectively.
The external characteristics of the rectifier side consist of a constant DC current
control with a low-voltage current limiting link and a constant αmin . The equation
of the external characteristics of the constant DC current control with a low-voltage
current limiting link can be expressed as function I = f (U ) as follows:
⎧
⎨ 0.55, U ≤ 0.4
I = 0.9U + 0.19 0.4 < U ≤ 0.9 (4.19)
⎩
1 0.9 < U

where U is the value of the starting voltage of the low-voltage current-limiting link
and I is the target value of the constant current control.
The equation for the external characteristic of the minimum fixed trigger angle
αmin control is

Udr = 1.3Er cos αrd min (4.20)

4.1 Analysis of the DC Commutation Failure Mechanism 153

where Udr is the rectifier-side DC voltage, Er is the RMS value of the rectifier-side,
commutation transformer, valve-side, no-load voltage, and X r is the rectifier-side
equivalent phase change reactance.
The external characteristics of the inverter side consist of three segments: constant
off angle, constant DC current, and current deviation control. The equation for the
external characteristics of the fixed-off angle control is:

Udi = 1.3E i cos γ − (3/π )X i Id (4.21)

where Udi is the inverter-side DC voltage, E i is the RMS value of the valve-side
no-load voltage of the inverter-side converter transformer and X i is the inverter-side
equivalent phase change reactance.
The external characteristic of constant current control is only one current margin
value smaller than that of rectification side. The external characteristic of constant
current control is only one current margin smaller than that of rectification Id which
is usually taken as 10% of the rated DC current.
When the DC system is operating in the rated state, its DC current is determined by
the constant current control on the rectifier side, the DC voltage is determined by the
constant shutdown angle control on the inverter side, and the system is operating at
point A. When a fault occurs in the inverter-side AC system and causes a commutation
failure to occur, the fault recovery process can be divided into three phases.
(1) Phase 1: Commutation failure occurs, and the operating point of the inverter-side
system is shifted. After the inverter-side AC system fault triggers commutation
failure, the inverter upper and lower bridge arm commutation valves are soon
short-circuited, resulting in the inverter-side DC voltage Ud−inv The inverter
side DC voltage greatly decreases, and the inverter side DC current Id−inv .
The inverter-side system operating point is shifted to point B. At this time,
the deviation from the inverter-side DC current Id−inv and the target point of
constant current control increases, current deviation control is activated, and
the inverter side is switched from constant shutdown angle control to constant
current control, at which time the control target point corresponding to operation
point B is in constant current control because of the Btarget point. Therefore,
operating point B on the inverter side is moved toward the control system by
the action of Btarget movement.
(2) Phase 2: Low-voltage current-limiting action and normal phase change are
resumed by the commutation valve. When operation point B is operated to
point C under the action of low-voltage current limiting link in constant current
control, the inverter-side DC current at this time Id−inv greatly decreases, the
commutation valve realizes normal phase change, and then the inverter side DC
voltage Ud−inv . The system control target point at this time has been run to
Ctarget point, so the system operation point C will continue under the action of
constant current control to Ctarget . The system operation point C will continue
to move under the action of constant current control. During this motion, the
inverter-side DC current Id−inv will continue to increase, which will cause the
154 4 Commutation Failure Prevention and Control

inverter-side DC voltage Ud−inv to increase, resulting in the Ctarget point moving

on the inverter-side low voltage current limit curve.
(3) Stage 3: The inverter-side system operating point coincides with the fixed current
control target point and enters current deviation control. The system operating
point C will gradually move toward its target point under the action of the control
system Ctarget until it meets. At this time, the inverter-side system operating
point is located at point D. The inverter-side DC current is Id−inv , and rectifier-
side target current is Id−or der . The difference between the DC on the inverter
side and the target current on the rectifier side is 0.1 pu, and the inverter enters
current deviation control and runs from point D to point F; that is, it is converted
from constant current control to constant shutdown angle control, and then point
F is the stable operation point jointly decided by the inverter side and the rectifier
side.
The low-voltage current-limiting control has a crucial role in stages 1 and 2. Equa-
tion (4.19) shows that the low-voltage current-limiting control U − I . The greater the
slope of the characteristic curve is, the faster the DC system recovers, the higher the
level of DC current recovery, and the greater the possibility of continuous commuta-
tion failure. Therefore, as long as the reasonable setting of the low-voltage current-
limiting parameters is ensured, the recovery speed and level of the DC current can
be ensured, and simultaneously, the failure of the fault-recovery continuous phase
change will not occur in phase 1 and phase 2. Therefore, it is known that the fault
recovery type continuous commutation failure occurs in stage 3, the current deviation
control link.
The main function of the current deviation control is that when the slope of the
characteristic curve of the inverter fixed-off angle control is greater than the slope
of the rectifier fixed min When the slope of the characteristic curve of the control is
greater than the slope of the characteristic curve of the rectifier, there is no stable
operating point between the fixed values of the current regulator at both ends, and
the DC current will oscillate back and forth between the 2 values [14, 15]. To avoid
this situation, the current deviation control is designed in the actual control system so
that the external characteristic of the inverter becomes a straight line with a positive
slope when the DC current is between the fixed value of the current on the inverter
side and the fixed value of the current on the rectifier side, i.e.,

γ = γr e f + K (Id0 − Id )/Id0 (4.22)

where γr e f is the inverter-side shutdown angle rectification value, Id0 is the rectifier-
side current constant, Id is the value of the inverter-side current, K is a constant and
a properly chosen K value, and the slope of the external characteristic of the current
deviation control can be positive. The current deviation control can simultaneously
realize the smooth switching between constant current control and constant off angle
control, and its external characteristic curve is shown in Fig. 4.11. The external
characteristic curve is shown in the figure.
4.1 Analysis of the DC Commutation Failure Mechanism 155

Fig. 4.11 Current deviation control steady-state operation curve

The dashed line in Fig. 4.11 shows that when the current margin between the
rectifier side and the inverter side is 0.1 pu, the larger the slope of the outer char-
acteristic of the current deviation control is, the longer the process of stage 3. As a
result, the inverter side may run on current deviation control instead of switching
smoothly from constant current control to constant off-angle control. Therefore, the
slope of the external characteristic curve of the current deviation control needs to be
kept at a low level, and the fixed trigger override angle β control can be employed.
The equation for the voltage-current characteristic is expressed as follows:

3X i
Udi = E i × cos β + × Id (4.23)
π
The above 2 implementation schemes for current deviation control are different
but functionally identical, and both of the characteristic curves maintain a positive
slope. It can be considered that the control objective of the current deviation control
process, regardless of the implementation, can be approximated as maintaining a
constant trigger override angle on the inverter side β. The inverter-side DC current
gradually increases in stage 3, and the increase is Id > 0.1 pu. It is obtained from
the theory of phase change area that the phase change area can be divided into the
phase change demand area Sneed and phase change supply area Ssupply , which are
expressed as follows:

Sneed = 2X C Id (4.24)
156 4 Commutation Failure Prevention and Control


α+μ 
π−β+μ
√
Ssupply = u n dω0 t = 2E sin(ωt)dω0 t (4.25)
α π−β

where X c is the equivalent phase commutation inductance, Un is the phase commu-

tation voltage, E is the rms value of the phase-commutation voltage, ω is the AC
system frequency and α is the trigger delay angle.
Expanding Eq. (4.25), we obtain
√
Ssupply = 2E[cos(β − μ) − cos β] (4.26)

According to (4.23), in phase 3, Id continues to increase, the phase change demand

area Sneed continuously increases, and to ensure successful phase change, the phase
change supply area Ssupply should also be continuously increased. However, since in
stage 3, its overrun trigger angle β should keep approaching its command value, to
judge the effect of the change in β on the Ssupply of the phase change, the effect of
Eq. (4.26) on the β partial derivative is obtained as

∂ Ssupply √
= 2E[sin β − sin(β − μ)] (4.27)
∂β

From Eq. (4.26), the overrunning trigger angle β and phase change angle μ can
be obtained in the range of values of∂ Ssupply /αβ > 0. Next, Ssupply decreases as β
decreases. The simulation analysis shows that in stage 3, β decreases and approaches
the target value with minimal range of variation. Therefore, it can be obtained that
in stage 3, the area of phase change demand Sneed continues to increase, while β
slightly decreases, which leads to a decrease in Ssupply . To maintain a successful
phase change, only a continuous increase in μ is necessary to ensure that Ssupply
continuously increases and is equal to theSneed phase. According to Eq. (4.26),
when β slightly decreases and μ gradually increases, then the angle of closure γ will
gradually decrease in stage 3 when γ is less than the inherent limit off angle γmin
when the fault recovery type continuous commutation failure will occur, which is
the root cause of the fault recovery type continuous commutation failure.
The above analysis clearly indicates that the occurrence of fault recovery type
continuous commutation failure is highly dependent on the severity of the fault,
which tends to occur in nonserious fault scenarios. During nonserious faults, the DC
system has a higher degree of DC current recovery after the first commutation failure,
resulting in the risk of continuous commutation failure when the inverter side enters
current deviation control.
The mechanism of continuous commutation failure can be further described with
a test system built on the CIGRE DC transmission standard model. A rated trigger
angle α N of 142° is selected; a minimum shutdown angle γmin of 7° is chosen; and
the other control parameters and fault settings of the model are identical to those of
the original standard system. A three-phase ground fault is set at the inverter-side
4.1 Analysis of the DC Commutation Failure Mechanism 157

commutation bus; the fault occurrence time is set to 1.5 s; the fault duration is set
to 0.5 s; and the fault is inductively grounded via 0.5H. The results are shown in
Fig. 4.12.
The above mechanism analysis reveals that the main reason for the continuous
commutation failure is the increase in the DC current during the current deviation

Fig. 4.12 Inverter-side control system response process after a fault

158 4 Commutation Failure Prevention and Control

control phase in stage 3, which leads to a continuous decrease in the turn-off angle.
Figure 4.12c shows that the actual value of the inverter side current at the moment
corresponding to point D has reached the inverter Side target value of constant current
control and that the system operating point is located at point D in Fig. 4.12d, i.e.,
the inverter-side system has entered current deviation control. In Fig. 4.12d, at the
moment corresponding to point F, the actual value of the inverter-side current has
reached the target value of the rectification side. The target value of the fixed current
control, the system operating point, is located at point F in Fig. 4.12d, i.e., the inverter-
side system is already in fixed-off angle control. Therefore, as shown in Fig. 4.12, the
shaded part of the above represents stage 3 in the mechanism analysis by Fig. 4.12c,
d. It can be found that the inverter-side DC current will continuously increase in stage
3, while by Fig. 4.12b, in stage 3, its inverter-side trigger angle α gradually increases.
However, the change is very small, approximately 10°, which leads to a decrease in
the trigger overrun angle β, i.e., the phase change margin decreases. Thus, only the
phase change angle μ can only ensure a successful phase change with increasing DC
current, which leads to a gradual decrease in the shutdown angle γ in stage 3. As
shown in Fig. 4.12a, the inverter-side shutdown angle continues to decrease in stage
3 until it is less than the critical shutdown angle γmin , thus leading to the occurrence
of successive commutation failures.

4.1.3 Multifed DC Commutation Failure

With the construction of the grid, multifeed LCC-HVDC transmission patterns with
multiple LCC-HVDC feeds concentrated into a particular area have emerged in large
numbers [16]. Although multifeeder LCC-HVDC increases the transmission capacity
and improves the flexibility of operation, the interaction between different LCC-
HVDC transmission systems and the AC-DC system becomes very complicated due
to the dense settlement of inverter stations [17]. If the electrical distance between each
DC drop point is small, the AC system at the receiving end is not strong enough,
and AC grid failure at the receiving end can easily cause commutation failure of
multiple LCC-HVDCs, which can even cause interruption of multiple LCC-HVDC
power transmission in severe cases, threatening the safe and stable operation of
the whole AC-DC system. In the multifeeder LCC-HVDC, the commutation failure
problem becomes more complex due to the interaction between AC systems and
DC systems. An AC fault may simultaneously trigger a commutation failure of
multiple DC systems, or it may cause a change in the system state after triggering a
commutation failure of one DC system and then cause commutation failures of other
DC systems, i.e., multifeeder DC commutation failures.
1. Simultaneous commutation failure
The structure of the actual system with a multifeeder DC system is shown in Fig. 4.13,
where the receiving grid feeds into the multifeeder DC transmission system, and the
drop points of each converter station are relatively dense and electrically connected.
4.1 Analysis of the DC Commutation Failure Mechanism 159

Fig. 4.13 Multifeeder DC

transmission system
architecture

Therefore, when a fault or voltage disturbance occurs near a converter station, it is

likely to cause voltage fluctuations in the nearby converter stations, resulting in a
concurrent commutation failure (CCF) of the multifeeder DC system, which will
cause a vast power shortage and threaten the stability of the receiving grid. The
mechanism of multifeeder DC simultaneous commutation failure is identical to that
of single-return DC lines. The key point is that when faults occur at different buses in
the inverter-side AC system, the voltage changes at other buses are different under the
influence of the DC system, i.e., there are boundary conditions for the occurrence of
multifeeder DC simultaneous commutation failure, which are related to the structural
parameters of the AC-DC system, as described below.
(1) Multifeed-in interaction factor
The concept of Multi-Infeed Interaction Factor (MIIF) was proposed by the Interna-
tional Large Grid Working Group [18] and was originally intended to characterize the
level of voltage interaction between two inverter-side converter buses in a multifeed
DC transmission system. MIIF is defined and calculated as follows: by artificially
connecting reactance elements in parallel so that the voltage at commutation bus node
i falls by approximately 1%, at which time the voltage change at the other inverter
station AC node j is observed, the ratio of the voltage change at the two nodes is
MIIFji , as expressed by Eq. (4.28).

Vj
MIIFji = (4.28)
Vi

where V i is the voltage drop value of inverter station commutation bus i and V j
is the voltage drop value of inverter station commutation bus j. When MIIF is equal
to 0, the two nodes are infinitely far apart; when MIIF is equal to 1, the two nodes are
the same node. Therefore, it is known that a larger MIIF value represents a stronger
interaction ability of the voltage between two nodes, while MIIF can quantitatively
portray the effect of the voltage drop of one commutator bus on the voltage of other
160 4 Commutation Failure Prevention and Control

commutator buses in a multifeed DC transmission system. When the MIIF value

between two nodes is greater than 0.6, it belongs to strong interaction, and when the
MIIF value between two nodes is less than 0.15, it belongs to weak interaction.
The definition is intuitive, concise, and has good adaptability for small-scale
systems. However, its greatest drawbacks are that the physical meaning is unclear,
that the results can only be derived from simulation analysis, which cannot theoret-
ically derive the influence of the network structure, DC system control mode and
other factors on the voltage interaction between two nodes and has large limitations
when facing large complex systems. For this reason, many scholars have performed
research to explore the physical interpretation of the voltage interaction relationship
of multifeeder DC transmission systems, are no longer limited to the MIIF and other
similar experimental indicators, and have derived the voltage interaction factor, tran-
sient voltage support indicator and other related parameters based on the structure
of the receiver grid. However, the above indicators only consider the influence of the
receiver grid structure on the voltage interaction relationship and do not consider the
DC system’s own power. The calculation results are not sufficiently accurate. Based
on the power characteristics of the DC system, the analytical calculation method
of the multifeed interaction factor can be obtained for different control modes. The
voltage interaction between the converter buses of the inverter station is not only
related to the structure and parameters of the AC system but also influenced by the
power characteristics of the DC system. The analytical expression of MIIF based on
the steady-state current of the receiver grid can more accurately quantify the voltage
interaction between the inverter side buses, and the physical meaning is clearer.
(2) Simultaneous commutation failure boundary condition
The purpose of the proposed multifeed-in interaction factor is to analyze the boundary
conditions for simultaneous phase commutation failure. In a multifeed DC transmis-
sion system, if a three-phase metallic short circuit occurs at the inverter-side commu-
tation bus i, the voltage at that node U Li = 0, resulting in the inevitable commutation
failure at the converter station feeding that node. The amount of voltage magnitude
change at the other converter buses on the inverter side is defined by the multifeed
interaction factor MIIF as
UL j N UL j N
U L j = MIIF ji U Li = MIIF ji U Li0 (4.29)
U Li N U Li N

where ULiN and ULjN represent the rated line voltages of the inverter-side AC buses
i and j, respectively.
From Eq. (4.29), the voltage of the inverter-side AC bus j after a fault U L j can be
obtained, as shown in Eq. (4.30).

UL j N
U L j = U L j0 − U L j = U L j0 − M I I F ji U Li0 (4.30)
U Li N
4.1 Analysis of the DC Commutation Failure Mechanism 161

where U Li0 and U Lj0 represent the rated line voltages of the inverter-side AC buses i
and j, respectively, before the fault.
When a fault occurs in the AC grid at the receiving end, it will lead to an increase
in the DC current at the adjacent converter station, resulting in a smaller phase change
margin. Thus, the change in the DC current after the fault should be fully considered
when analyzing the boundary conditions for simultaneous commutation failure. Let

Id be the DC current of DC system j after the fault, and let the shutdown angle after
the fault be γ  , which is obtained from Eq. (4.3).
√
 2U L j  

Id = cosγ − cosβ (4.31)
2n X L

To investigate the boundary conditions of commutation failure, the period of

interest in this subsection is from after the occurrence of the fault to before the
occurrence of the first commutation failure, which is referred to as “after the fault”
in the subsequent sections. The inverter side is controlled by a fixed shutdown angle,
where the shutdown angle γ is the minimum of the measured inverter-side angle in
the DC system. The angle γ measured by the DC system is taken as the minimum
value, so the control command given by the fixed shutdown angle β control command
given by the fixed shutdown angle control does not change until the first commutation
failure occurs. It can be assumed that the short period between the occurrence of the
fault and the first commutation failure β angle does not change. Therefore, Eqs. (4.3)
and (4.31) before and after the fault can be compared to obtain Eq. (4.32).

UL j I cosγ − cosβ
= d · (4.32)
U L j0 Id cosγ  − cosβ

Since the fault occurs on the inverter side, the transmitted power of the DC system
is constant for a short time for the nondirect faulty node j. Thus, Eq. (4.33) is obtained
as

Id Ud
=  (4.33)
Id Ud

From the converter characteristics, the relationship between the DC voltage and
the commutation voltage can be obtained as Eq. (4.34).
√
3 2U L j0
Ud = (cosγ + cosβ) (4.34)
2π
Combining the DC voltage in the stable case Ud and the DC voltage after the fault

Ud ratio based on Eq. (4.34), Eq. (4.35) is obtained as follows:

Ud U L j0 cosγ + cosβ
 = · (4.35)
Ud U L j cosγ  + cosβ
162 4 Commutation Failure Prevention and Control

From the above derivation, Eqs. (4.30), (4.32), (4.33), and (4.35) can be solved
in conjunction to obtain the shutdown angle after a fault in the receiving AC grid

inverter Station j γ j :
 
 cos2 γ − cos2 β
γ j = arccos  2 + cos β2
U L j /U L j0
⎛ ⎞

 cos2 γ − cos2 β
⎜ ⎟
= arccos⎝   2 + cos2 β ⎠ (4.36)
UL j N
U L j0 − M I I F ji U Li0 U Li N /U L j0

According to the above equation, the size of the shutdown angle of the inverter
station of the AC grid at the receiving end after the failure is related to not only
the voltage of the converter bus but also the operation mode of the converter station
before the failure occurred, i.e., the overtrigger angle and shutdown angle before
the failure, and influenced by the multifeed interaction factor between the adjacent
converter station and this converter station. Notably, the larger MIIFji is, the smaller

the shutdown angle of the converter station γ j , and the higher the possibility of
simultaneous commutation failure at converter Station j.

Referring to the minimum turn-off angle criterion, when γ j = γmin = 8◦ time,
the multifeed-in interaction factor corresponding to the minimum turn-off angle can
be calculated from the above equation, which will be defined as the simultaneous
commutation failure interaction factor. Therefore, from Eq. (4.36), when MIIFji >

CCFIFji , γ j will be less than the minimum shutdown angle (8° ). At this time, if a
three-phase metallic short circuit occurs in inverter station AC bus i, inverter Station
i will fail to commute due to phase change voltage landing. As indicated by the
above analysis, the shutdown angle of inverter Station j is less than the inherent limit
shutdown angle, which will also fail to commute. Therefore, the equation of CCFIFji
is calculated as follows:
 2
cos γ−cos2 β
1 − cos 2 γcos2 β
CCFIFji = U Li0 U L j N
min
(4.37)
U L j0 U Li N

2. Successive commutation failure

Due to the different parameters and operating states of each LCC-HVDC, a fault in
the AC system at the receiving end may trigger a commutation failure in one LCC-
HVDC, while its adjacent LCC-HVDC does not experience a commutation failure
[19]. After a commutation failure, significant changes in DC power, voltage, etc.,
occur, especially when the continuous activation of the LCC-HVDC control system
will further aggravate changes in electrical quantities. Therefore, the LCC-HVDC
where the commutation failure occurs is bound to affect the adjacent normal oper-
ating LCC-HVDC through close electromagnetic coupling, resulting in a secondary
4.1 Analysis of the DC Commutation Failure Mechanism 163

response of the normal operating LCC-HVDC, which may lead to a serious commu-
tation failure. This type of commutation failure is referred to as multifeeder DC
successive commutation failure. Compared with studies on simultaneous commuta-
tion failure, fewer studies address the mechanism and consequences of successive
commutation failure.
(1) Mechanisms of successive commutation failure
After a fault in the AC grid at the receiving end, the voltage of the ith return LCC-
HVDC commutation bus drops to U L f i . The reactive power emitted by the AC filter
decreases, while the control system causes the reactive power consumption of the
ith return LCC-HVDC to rise, and the change in its reactive power consumption is
shared by the receiving system and the adjacent LCC-HVDC. The reactive power

consumption of the inverter station of the ith LCC-HVDC after fault Q I i is satisfied.
  
Q I i = Q aci + Q f i + Q ex j (4.38)

where.Q aci is the reactive power absorbed by the inverter station from the receiving

grid after the fault; Q f i is the reactive power supplied by the filter of the ith LCC-

HVDC inverter station after the fault, which can be written as Q f i = B f i U L2 f i , where
B f i is the equivalent power of the inverter station filter and U L f i is the value of the
phase change voltage of the ith LCC-HVDC after the fault, and Q ex j is the reactive
power exchanged with the adjacent jth LCC-HVDC after the fault.

Q aci determines the voltage dip of the converter bus and the short-circuit capacity
of the equivalent system at the receiving end [20]. The amount of reactive power
exchanged between the ith LCC-HVDC and the adjacent jth LCC-HVDC after a
commutation failure of the ith LCC-HVDC can be expressed as
     U i Saci
Q ex j = Q I i − Q li − B f i U L2 f i − B f i U Li
2
− (4.39)
UL N i

where Q li denotes the reactive power of the ith LCC-HVDC inverter station before
the fault, U Li is the value of the phase change voltage of the ith return LCC-HVDC
before the fault, U i is the voltage dip of the ith return LCC-HVDC commutation
bus, Saci is the short-circuit capacity of the system at the receiving end of the ith
LCC-HVDC, and U L N i is the phase change voltage rating of the ith LCC-HVDC.
According to the calculation, the reactive power consumption of the inverter
station is always positively related to the converter bus voltage and DC current.
The reactive power rises with the increase of the firing angle [21]. After AC grid
failure, the decrease in the converter bus voltage causes a decrease in the reac-
tive power consumption of the inverter station. With the activation of the voltage-
dependent current order limit (VDCOL), the reduction in the DC current command
value causes the DC current to decrease and the reactive power consumption to
decrease. Therefore, the reactive power consumption of the inverter station increases
and then decreases when the DC current changes after a fault. The smaller the starting
voltage of the VDCOL and the later it is input after a fault, the higher the reactive
164 4 Commutation Failure Prevention and Control

power consumption of the inverter station. With the incoming VDCOL limiting the
DC current to a lower level, constant extinction angle control (CEAC) continues
to act, especially when the VDCOL action makes the DC current less than the
commanded value, and CEAC will cause a sudden increase in the overtrigger angle.
The overtrigger angle continuously rises under the CEAC action, causing an increase
in reactive power consumption at the inverter station. Under the action of the control
system, the reactive power consumption of the inverter station has a “2-up” trend.
Under steady-state operation, the reactive power provided by the AC filter can
meet the reactive power consumption requirement of the inverter station, and the
reactive power absorbed by the inverter station from the receiving system is basically
0. The input of the AC filter is generally on the order of seconds [22]. The input of
the control system of the inverter station is on the millisecond level. The “2 rise”
of reactive power consumption of the inverter station occurs within 50 ms, and the
reactive power compensation of the AC filter cannot track the change of reactive
power consumption of the inverter station in time. When the receiving grid is a weak
system, the reactive power supplied to the inverter station during the commutation
failure is limited [23]. The total amount of reactive power supplied by the adjacent
LCC-HVDC is similar to the reactive power consumption of the ith LCC-HVDC.
After grid failure, the “2 rise” of the reactive power consumption of the ith LCC-
HVDC inverter station under the weak grid at the receiving end leads to the same “2
rise” in the reactive power exchange of the adjacent LCC-HVDCs.
After a fault in the AC grid at the receiving end, the commutation bus voltages of
multiple LCC-HVDCs will dip. If the voltage dip triggers a commutation failure of
multiple LCC-HVDCs, the current experiences simultaneous commutation failure.
The distance of each LCC-HVDC inverter station from the failure point varies, and
the amount of the commutation bus voltage dip after the failure varies. An AC grid
fault may cause a commutation failure in only part of the LCC-HVDC. When an LCC-
HVDC phase change fails, its control system starts to trigger a “2 rise” in the reactive
power exchange of the adjacent LCC-HVDC, which causes the bus voltage of the
adjacent LCC-HVDC inverter station to drop again. The voltage drop triggers another
drop in the shutdown angle of the adjacent LCC-HVDC inverter station, i.e., a new
commutation failure may have occurred. This commutation failure is a secondary
commutation failure triggered by the commutation failure of the adjacent LCC-
HVDC, i.e., a successive commutation failure. The successive commutation failure
is a causally related secondary commutation failure among multiple LCC-HVDCs,
where the fault and controller response caused by the coupling of the adjacent LCC-
HVDC phase change voltage dip through the AC system is one of the main reasons
for the occurrence of the successive commutation failure.
(2) Analysis of a typical successive commutation failure process
Figure 4.14 shows the two-loop LCC-HVDC shutdown angle, exchanged reactive
power and DC current under an AC grid fault at the receiving end. The Multi-Infeed
Effective Short Circuit Ratio (MIESCR) of the two-loop LCC-HVDC receiver system
is denoted as MIESCR . A three-phase short circuit occurs at 30 km from the i-bus at 1 s,
4.1 Analysis of the DC Commutation Failure Mechanism 165

Fig. 4.14 Exchange of reactive power, DC current and shutdown angle between the two LCC-
HVDCs under an AC grid fault at the receiving

and the fault lasts for 0.5 s. The blue dashed line in the figure represents the i-bus LCC-
HVDC electrical quantity; the red solid line denotes the j-bus LCC-HVDC electrical
quantity. As shown in the figure, the ith return LCC-HVDC fails to change phase
at 1.012 s. At this point, the jth return LCC-HVDC off angle slightly decreases, but
no commutation failure occurs. Until 1.034 s later, the jth return LCC-HVDC turn-
off angle drops below the critical turn-off angle, and commutation failure occurs.
The commutation failure of the jth LCC-HVDC lags 22 ms behind the ith LCC-
HVDC, and the two commutation failures have a significant time delay and are not
simultaneous commutation failures caused by having a grid fault. As observed based
on the reactive exchange and DC current, the jth LCC-HVDC turn-off angle shows
3 significantly different change processes, which can be divided into 3 phases.
Phase 1: Commutation failure occurs in the ith LCC-HVDC, and the jth LCC-
HVDC shutdown angle slightly decreases. After a commutation failure of the ith
return LCC-HVDC triggered by a fault in the AC system at the receiving end, an
instantaneous drop in the turn-off angle causes a sharp rise in the DC current. Under
the action of the VDCOL, the ith return LCC-HVDC current then decreases, and
thus, the reactive power exchange continues and then slightly decreases.
Phase 2: Commutation failure occurs at the jth LCC-HVDC. CEAC causes the
override trigger angle to continuously rise, leading to another increase in reactive
power consumption at the ith inverter station and another increase in reactive power
exchange between the two LCC-HVDCs. The increase in reactive power exchange
triggers another dip in the jth-back converter bus voltage as well as the shutdown
angle, resulting in the commutation failure of the jth-back LCC-HVDC. The first dip
in the jth LCC-HVDC shutdown angle is similar to the moment of the ith shutdown
angle dip, which is triggered by an AC grid fault. The 2nd dip is accompanied by
a rise in reactive power exchange, which indicates that the commutation failure of
the jth LCC-HVDC is a commutation failure triggered by the commutation failure
of the ith LCC-HVDC.
166 4 Commutation Failure Prevention and Control

Phase 3: The system is restored, and the inverter station resumes normal phase
exchange. The jth return LCC-HVDC control system is put into action, and the
reactive power exchange is slightly increased.

4.2 Early Warning Measures for Commutation Failure

in a Hybrid AC/DC Grid

The detection methods for single-commutation failure can be divided into two types:
real and predictive. Among them, the real measurement type method actually detects
the shutdown angle by detecting the moment when the phase change voltage crosses
zero and the moment when the valve current crosses zero (γ ). When the detected
turn-off angle is less than the minimum value, a commutation failure has occurred.
The predictive commutation failure detection method, on the other hand, predicts
whether an AC fault triggers a commutation failure by detecting a voltage dip,
which can also be referred to as a commutation failure warning measure. Effec-
tive commutation failure warning facilitates the rapid input of commutation failure
suppression measures and has a vital role in preventing the generation and further
development of commutation failure. In this section, several early warning measures
for the first commutation failure, successive commutation failure and multifeeder
DC commutation failure are presented, distinguished by the type of commutation
failure.

4.2.1 Early Warning Measures for the First Commutation

Failure

1. Study on early warning of commutation failure based on phase change area

prediction
Based on the whole process of DC phase change, the mathematical analysis of the
phase change demand area and minimum supplied phase change area is derived
based on the theory of phase change voltage time area and utilized as the prediction
criterion for commutation failure. The voltage waveforms before and after the fault
are fitted by the three-point method to achieve fast warning of commutation failure.
The phase change voltage time area theory is described in more detail in Sect. 4.1.2
of this chapter.
(1) Prediction criterion for commutation failure based on phase change area
According to the phase commutation area theory, the two phase commutation area
indicators that are critical in determining the success of the phase commutation
process are defined as follows.Sneed is the phase change demand area, and Smin is
the minimum phase change area that can be supplied. The criterion for commutation
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 167

failure is as follows: when Smin > Sneed , the inverter will not experience commu-
tation failure; when Smin < Sneed , the inverter will experience commutation failure.
Therefore, the accuracy of the commutation failure prediction based on the phase
change area depends on Sneed and the accuracy of the Smin calculation.
As shown in Eq. (4.8), considering the variation the DC current, the area of phase
change demand Sneed is no longer constant. However, the method described in that
subsection is for the prediction of the first commutation failure occurring after the
fault, and since the DC current changes at a slower rate before the commutation
failure occurs and the flat wave reactor on the DC side reduces the rate of increase
of the DC current, the start and end moments of the phase change Id are taken as the
sampling moment t0 corresponding to the Id (ω0 t0 ), according to which the area of
phase change demand can be Sneed .

Sneed = 2XC Id (ω0 t0 ) (4.40)

Based on the phase change area theory, the minimum phase change area that can
be supplied Smin mainly depends on the postfault line voltage curve, assuming that
the postfault AC voltage frequency ω0 remains constant and is still sinusoidal. Thus,
the minimum line voltage curve after the fault can be defined as

u Lmin = U Lmin sin(ω0 t + ϕ1 ) (4.41)

With Eq. (4.8) and Eq. (4.41), the minimum phase change area that can be supplied
Smin is defined as

ω0 t2
ULmin
Smin = ULmin dω0 t = [cos(ω0 t1 + ϕ1 ) − cos(ω0 t2 + ϕ1 )] (4.42)
ω0
ω0 t1

As shown in Eqs. (4.40) and (4.42), the Smin and Sneed numerical solution process
needs to be performed to solve the phase change start moment t1 and termination
moment of phase change t2 , as shown in Fig. 4.15.
Where t0 is the sampling moment, α is the inverter-side trigger delay angle, θ0 is
the line voltage curve during normal system operation t0 and the voltage phase angle
at the moment.θ0 = ω0 t0 + α0 . The phase change onset time t1 can be obtained from
Fig. 4.15

t1 = t0 + t1
(4.43)
t1 = αiω−θ
0
0

The line voltage during normal operation before system failure was

u L = U L0 sin(ω0 t + ϕ0 ) (4.44)
168 4 Commutation Failure Prevention and Control

uL 0 Δφ

uL min
S2min

θ0 Δt1 t
γ min
Δt2
α
t0 t1 t2
Fig. 4.15 Minimum area that can be supplied by the system after a failure

Based on the physical characteristics of the phase change process, the sampling
moment t0 is the interval of (0, 90°) in the positive half-cycle of each phase change
line voltage, which can ensure the accurate prediction of the next phase change after
the fault. Substituting the sampling moment t0 into Eq. (4.44), we can solve θ0 as

θ0 = arcsin[U L0 sin(ω0 t0 + ϕ0 )] (4.45)

Phase change termination time t2 From Fig. 4.15, we have.


t2 = t0 + t2
(4.46)
t2 = π−γminω−θ
0
0− ϕ

where γmin is the minimum shutdown angle and ϕ is the line voltage commutation
offset angle, which can be obtained from the initial phase of the prefault line voltage
curve and postfault line voltage curve as follows:

ϕ = ϕ1 − ϕ0 (4.47)

From Eq. (4.47), the phase change onset time t1 and the phase change termination
time t2 can be obtained. With Eq. (4.40) and Eq. (4.42), we can obtain the phase
change demand area Sneed and the supply minimum phase change area Smin , that is,
by comparing Sneed and Smin , the magnitude of the phase change and the prediction
of commutation failure within a very short sampling time after a fault can be enabled.
The above analysis reveals that the method proposed in this subsection has a
significant impact on Sneed and Smin . The required parameters in the solution process
have basically been solved, and only the minimum line voltage curve after the fault
u L min is still unknown, so the minimum voltage curve will be solved in the following
section.
(2) Fast prediction of commutation failure based on voltage waveform fitting
The reliability and speed of the proposed commutation failure prediction method
depend mainly on the accuracy and speed of the fitted voltage curve, while the
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 169

accuracy of the voltage curve is mainly influenced by the accuracy of the system
sampling value, the calculation method of the fitted curve and the strength of the
receiver system. The stronger the receiver system is, the smaller the frequency devi-
ation of the receiver system is, the closer the voltage waveform is to sinusoidal, and
therefore the higher the accuracy of the commutation failure prediction.
The principle of the three-point method is to calculate the sine curve amplitude,
frequency, and phase angle from three sampling points on the sine curve. Assuming
that the frequency of the AC system remains unchanged after the fault, which is still
the working frequency of 50 Hz and the angular frequency of ω0 = 2π × 50 rad s
, the
voltage curve can be determined by determining the amplitude and phase angle of
the voltage curve through two sampling points. Let the postfault AC side line voltage
expression be

u = k1 sin ω0 t + k2 cos ω0 t = A sin(ω0 t + ϕ) (4.48)

The two sampling points of the AC bus line voltage on the inverter side are (t1 ,
u 1 ) and (t2 , and u 2 ). Next, we have

u 1 = k1 sin ω0 t1 + k2 cos ω0 t1
(4.49)
u 2 = k1 sin ω0 t2 + k2 cos ω0 t2

To make the expression of Eq. (4.49) more concise, construct the matrix
     
u1 k sin ω0 t1 cos ω0 t1
U= K= 1 =
u2 k2 sin ω0 t2 cos ω0 t2

Next, Eq. (4.49) can be expressed as

U= K (4.50)

The coefficient matrix K can be derived as follows:

−1
K= U (4.51)

The amplitude and phase of the voltage curve can be further derived from the
coefficient matrix of the voltage curve as follows:

A= k12 + k22 (4.52)

arctan −k 1
k2 ≥ 0
ϕ= k2
−k1 (4.53)
arctan k2 + π k2 < 0

The expression for the postfault I.F. voltage curve can be determined.
170 4 Commutation Failure Prevention and Control

The method based on the three-point method to fit the sinusoidal curve is a direct
solution method. The number of sampling points is equivalent to the number of
unknown quantities of the voltage curve; the number of sampling points required is
the least; the sampling time used is the shortest; and the calculation is the smallest
and has the advantage of rapidity. Through simulation analysis and theoretical calcu-
lation, the voltage curve fitted by the three-point method is basically consistent with
the trend of the system voltage change and can respond more quickly to the voltage
drop after the fault. The output voltage amplitude of the three-point method responds
to the voltage drop at a faster rate than the response speed of the measured RMS
voltage, which overcomes the requirement that the RMS voltage measurement cannot
meet the rapidity of control system regulation.
By predicting the postfault AC voltage waveform by the three-point method
and by sampling the real-time DC current values, it is possible to achieve a phase
change demand area for the LCC-HVDC-type DC transmission system Sneed and the
minimum phase change area that can be supplied Smin of the numerical analysis, thus
achieving fast prediction of commutation failure. The specific prediction scheme is
implemented as shown in Fig. 4.16.

Fig. 4.16 Schematic of

commutation failure
prediction
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 171

Fig. 4.17 Block diagram of

the control structure

The sampling point of the line voltage after the fault is obtained at the AC bus on
the inverter side, the line voltage magnitude and phase angle at the current sampling
moment are calculated by the three-point method, and the line voltage parameters
before the fault are fitted by the delay link. Parameters such as the trigger delay angle,
DC current and sampling time on the inverter side are input to the phase change area
calculation unit, whose output is the system phase change demand area Sneed and the
minimum area that can be supplied Smin When the difference value is greater than 0,
the inverter is judged to have imminent commutation failure, and the Commutation
Failure Prediction Signal (CFPS) output is 1. When the difference value is less than
or equal to 0, the inverter is judged not to experience commutation failure, and the
CFPS output is 0. The specific control structure is shown in Fig. 4.17.

2. Early warning measures for commutation failure based on Davinan equivalent

impedance

According to Davinan’s theorem, for a network, the Davinan equivalent impedance

of a port is the value of the impedance of the passive network as observed from that
port.
Figure 4.18 shows the fault impedance for Z F and the Davinan equivalent
impedance as observed from ports A and B under a short-circuit fault, where both the
upstream equivalent voltage source and downstream equivalent voltage source are
set to 0. In the figure, the upstream equivalent impedance and downstream equivalent
impedance are denoted as Z U and Z D , respectively, and x denotes the distance from
the fault point to port AB. The Davinan equivalent impedance as seen from port AB
is

Z t = Z U ||(x Z D + (Z F ||(1 − x)Z D )) (4.44)

Equation (4.44) shows that as the severity of the fault increases, Z t will decrease.
172 4 Commutation Failure Prevention and Control

Fig. 4.18 Fault impedance

of ZF Davinan equivalent
impedance from ports A and
B under a short-circuit fault
of

Fig. 4.19 Davinan

equivalent model

Based on the Davinan equivalent model shown in Fig. 4.19, the voltage at any
node of the grid can be expressed as

V = Et − Z t I (4.45)

Thus, the inverter AC bus voltage can be expressed as

Vib = E tib − Z tib Iib (4.46)

where Etib and Ztib denote the Davinan equivalent voltage source and Davinan
equivalent impedance, respectively, as viewed from the inverter AC bus. Figure 4.20
shows two sets of data measured by the same Phasor Measurement Unit (PMU)
installed on the inverter AC bus at two different moments during the fault (Vib , Iib ).
In the phase diagram of Etib , the magnitude should be identical in both measurements,
but due to fluctuations in the system frequency, the Etib of the phase is shifted.
The first measurement of Etib can be expressed as
2
E tib = VibI
2
+ IibI
2 2
Z tib + 2VibI IibI Z tib cos(θtib − φibI ) (4.47)

By extending cos(θtib − φibI ) = cos θtib cos φibI + sin θtib sin φibI , and rewriting
cos θtib = Rtib /Ztib , sin θtib = X tib /Z tib , cos φibI = PibI /(VibI IibI ), and sin φibI =
Q ibI /(VibI IibI ), one obtains
2
E tib = VibI
2
+ IibI
2 2
Z tib + 2PibI Rtib + 2Q ibI X tib (4.48)
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 173

Fig. 4.20 Phase diagram of

the measured values of the
two PMUs at different
moments of the AC bus on
the inverter side

where Rtib , X tib , PibI , and Q ibI denote the Davinan equivalent resistance, Davinan
equivalent reactance, active power and reactive power, respectively, at the AC bus of
the inverter.
The same calculation gives the second measurement of Etib
2
E tib = VibI
2
I + IibI I Z tib + 2PibI I Rtib + 2Q ibI I X tib
2 2
(4.49)

The subtraction of the two measurements gives

 2  2
2
VibI − VibI
2
I + IibI − IibI I Z tib + 2(PibI − PibI I )Rtib + 2(Q ibI − Q ibI I )X tib = 0
2

(4.50)

Z tib = 2
Rtib + X tib
2
can be rewritten to obtain

2 2
PibI − PibI I Q ibI − Q ibI I
Rtib + + X tib +
2
IibI − IibI
2
I
2
IibI − IibI
2
I
2 2
2
VibI − VibI
2
PibI − PibI I Q ibI − Q ibI I
= I
+ + (4.51)
2
IibI − IibI
2
I
2
IibI − IibI
2
I
2
IibI − IibI
2
I
174 4 Commutation Failure Prevention and Control

 2  2
V2ibI −V2ibII PibI −PibII QibI −QibII
This is a circle with radius r I,I I = IibI −IibII
2 2 + I −I
2 2 + IibI −IibII
2 2 , and
 ibI ibII
−PibI QibII −QibI
the center of the circle is Q I,I I = PIibII , I2 −I2 . The circle which is in the
ibI −IibII
2 2
ibI ibII
impedance plane, represents the Davinan equivalent impedance at the AC bus of
the inverter. The subtraction of two measurements at different moments cannot give
Z tib the exact value of the intersection of the three impedance circles. Therefore,
we need to make another measurement and use the same method to obtain three
impedance circles to give Z tib . Note that parameters such as fault location and type,
ground impedance, and inverter-side AC grid structure affect the measured voltage
and current values of the inverter AC bus, and therefore, are included in Z tib . From
this, it is possible to calculate Z tib whether a sudden change occurs to determine
whether a short-circuit fault has occurred in the inverter-side AC system and to
prepare for further prediction of the shutdown angle [34].

4.2.2 Early Warning Measures for Continuous Commutation

Failure

Before the AC fault is cleared, the HVDC system usually starts to enter the recovery
phase of the first commutation failure, and there is a risk of continuous commutation
failure, which will cause further power impact to the power system at the transmitter
and receiver side again. Therefore, by using the real-time collected operating param-
eters of the HVDC system to assess the risk of continuous commutation failure and to
correctly warn of the occurrence of continuous commutation failure, the active inter-
vention and suppression of continuous commutation failure can be realized through
appropriate advance emergency control means to ensure the safe and stable recovery
of the DC system and to further interrupt the occurrence and spread of chain failures
in the hybrid AC-DC power system. The mechanism of continuous commutation
failure is described in more detail in Sect. 4.1.2 of this chapter.
1. Early warning of commutation failure based on recovery characteristics measures
The occurrence of successive commutation failures during fault persistence is closely
related to the controller regulation and system response during recovery. In addition
to the commonly associated factors that induce the first commutation failure, factors
such as the output fluctuations of the current error controller and the periodic oscil-
lations of the inverter DC current under asymmetric faults need to be considered.
Therefore, the risk of successive commutation failures occurring can be sensed in
advance by monitoring the control system response and the state of the main circuit
electrical quantities.
When the 6-pulse converter is operated in “2–3” mode, the voltage time area
(VTA) demand for a certain phase change process can be expressed according to the
quasi-steady state analysis of the phase change process as follows:
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 175

V T Ad = ωL c Idiπ−γ + ωL c Idiπ−β (4.52)

where Idiπ−β and Idiπ−γ are the inverter DC currents at the start moment and end
moment, respectively, of the phase change; ω is the rated angular frequency of the
system; L c are the phase commutation reactances; and β and γ are the actual trigger
overshoot and shutdown angles, respectively. If the time axis is expressed in electrical
angles, the voltage–time area provided by the AC supply is the integral of the phase
change voltage with respect to time and can be expressed as


π−γ

V T As = Ucom (ωt)dωt (4.53)

π−β

where Ucom (ωt) is the base wave instantaneous value of the phase change voltage.
Usually, sufficient V T As is conducive to successful phase change, so the necessary
condition for successful phase change considering the change in electrical quantities
of the main circuit of the HVDC system is

 0
π−γ
√
 ! !
2ωL c Idr + ! I˙di ! < 2E min sin ωtdωt (4.54)
π−β+ϕ

In the above equation, the left-hand side and right-hand side represent the
maximum demand and supply, !respectively,
! of the commutation voltage time area;
Idr is the rectifier DC current; ! I˙di ! is the amplitude of the inverter-side oscillation
current under an asymmetric fault; γ0 is the electrical angle corresponding to the
minimum valve shutdown time; ϕ is the maximum phase shift of the valve-side
commutation voltage due to the AC fault; and Emin is the rms value of the minimum
phase change voltage. Thus, the normalized criterion for the continuous commutation
failure warning can be further introduced as
√  ! !
2ωL c Idr + ! I˙di !
cos(β E W S − ϕ) < cos γ0 − (4.55)
E min

where the left side reflects the controller boundary denoted by BC (β EWS ) and the
right side reflects the circuit boundary denoted by BE (γ 0 ). β E W S The controller
structure and regulation risk perspective is expressed as follows:

β E W S = βiC E A_K i − K PC E A · σiC EC (4.56)

where βiC E A_K i is the integral link output of the CEA controller, K PC E A is the scale
factor of the PI regulator, and σiC EC is the output of the current error controller. The
block diagram of the constructed early warning system is shown below. The core
of the early warning system is based on the normalized criterion of the continuous
176 4 Commutation Failure Prevention and Control

commutation failure warning expressed in Eq. 4.55, where γiY and γi D denote the
turn-off angles of the Y-bridge and D-bridge, respectively, on the inverter side; γimin
denotes the minimum value of the shutdown angle on the inverter side in one cycle;
γir e f is the reference value of the shutdown angle; Uth denotes the negative sequence
voltage reference value; and WFlagY, WFlagD and Wflag denote the determination
results output by the warning system (Fig. 4.21).

2. Early warning measures for commutation failure based on statistical learning

methods

The continuous commutation failure fault has a certain development process, with
the longest duration of several seconds, during which the AC-DC system is mutually
disturbed by strong coupling and the control system rapidly jumps. The involved
phase change mechanism is not as clear as the single-phase change process, with
complex characteristics such as AC-DC strong coupling and nonlinearity, and the
physical mechanism level is not analytic, so it is difficult to manually preset rules to
achieve early warning. To address this problem, the premise of continuous commuta-
tion failure warning based on statistical learning method is to consider the problem as

Fig. 4.21 Schematic of the early warning system for continuous commutation failure in HVDC
systems
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 177

a pattern recognition problem based on locally known measurement quantities, that

is, to take part of the measurement signal at the early stage of the fault as the recogni-
tion feature, to make an inference whether the continuous commutation failure will
occur throughout the fault dynamics based on the feature, and to achieve the purpose
of advance prediction and early warning.
The purpose of statistical learning is to obtain an optimal, strongly fit-capable
parametric mapping function based on the data level to mine the correlation mapping
relationship between the features of the data and their corresponding markers, which
can be theoretically nonlinear or even arbitrarily complex. In the framework of the
continuous commutation failure warning algorithm, the features correspond to some
of the measured signals at the early stage of the fault, and the markers correspond to
the values indicating whether the continuous commutation failure will subsequently
occur in the DC system based on the features. The optimal parametric mapping
functions of the features and markers for the continuous commutation failure are
obtained by training the data from the field operation fault data or from the offline
simulation. The statistical mapping based on statistical learning algorithms to solve
correlations can be freed from the mechanistic analysis of continuous commuta-
tion failure, which no longer requires analytical analysis for physical processes and
artificially gives judgment thresholds. However, this analysis focuses on obtaining
sufficient descriptive information to obtain a satisfactory warning algorithm and to
apply it in engineering practice by reasonably building a learner model.
Among statistical learning methods, integrated learning methods are generally
more generalizable than single methods and are good at solving small data learning
and learning challenges under sample imbalance. As a result, integrated learning is
widely employed in various fields, such as computer vision and image processing,
biology, medicine and engineering. Specifically, integrated learning methods refer
to learning methods that significantly improve the learning performance of weak
learners by constructing and combining the outputs of multiple weak learners to
accomplish a learning task using certain rules. Weak learners are often referred to
as learners with slightly better generalization performance than random guesses,
and in integration learning, weak learners, which are also known as base learners,
are an important part of the composition of homogeneous integration models. The
weak learners in integration learning can be any of the popular learning algorithms,
such as “neural networks”, “decision trees” and “support vector machines”. The
continuous failure warning problem is a binary problem of pattern recognition, i.e.,
the warning system outputs a “0–1” warning signal based on early measurement
characteristics, which represent that “continuous commutation failure will occur”
or that “continuous commutation failure will not occur”, respectively. Among the
integrated learning methods, the traditional boosting method is the best at solving
binary classification problems with a split linear decision plane, while the family
of boosting algorithms, Adaboost, is the most famous and can be derived using an
“additive model”. The output of the AdaBoost learner is a linear combination of the
outputs of several weak learners.
Assume that the initial training sample is m, that the sequence of samples D =
{(x1 , y1 ), (x2 , y2 ), . . . , (xm , ym )}, where yi ∈ {−1, +1}, that L is the base learning
178 4 Commutation Failure Prevention and Control

algorithm corresponding to the weak learner, that f (·) is the true function assumed
to exist, and that h i (·) is the first-round iteration i generated by the weak learner
corresponding to the discriminant function.
(1) Each training sample is assigned an initial weight, where Di j denotes the first
sample j in the first round of training i with sample weights, calculated as shown
in Eq. (4.57).

1
D1l = , l = 1, 2, . . . , m (4.57)
m

(2) for i = 1, . . . , T , , loop through steps (3)–(6).

(3) For the training set D with the training set weights Di , the training is performed
to obtain the first iteration i of the weak learner discriminant function.

h i (·) = L(D, Di ) (4.58)

(4) Calculate the residuals of this weak learner according to Eq. (4.59).

1 "
m
i = I (h i (x) = yi )Di j (4.59)
m j=1

(5) If the residual is greater than 0.5, the algorithm aborts; otherwise, for the first i
round of training, the resulting weak classifier is assigned the classifier weight
αi , which is calculated as shown in Eq. (4.60).

1 1 − εi
αi = ln (4.60)
2 εi

(6) Reassign the training sample weights according to Eq. (4.61).

Di+1 (x) = Di (x)e−αi f (x)h i (x) (4.61)

In this context, a weak classifier is a less accurate classifier, with a classification

accuracy generally ranging from 60 to 80%. The use of weak classifiers in Adaboost
simply provides a training direction and then enhances the training weights in this
direction, which is referred to as strong training. The final combination is the final
result. After one round of training T, Adaboost combines the outputs of the base
learners based on the accuracy of the voting weights with the outputs of the individual
base learners and passes a symbolic function as the final output, which is calculated
as shown in Eq. (4.62).
 
"
T
H (x) = sign αi h i (x) (4.62)
i=1
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 179

According to the previously described phase change mechanism, the occurrence

of commutation failure is directly related to the AC commutation voltage and DC
current during phase change. Continuous commutation failure is essentially a sequen-
tial occurrence of multiple single-commutation failures; thus, the same factors that
affect single-commutation failure also affect continuous commutation failure. The
AC voltage, DC voltage, and DC current are utilized to construct the character-
istic quantities, including AC voltage U̇a , U̇b , U̇c , inverter-side DC current Idr , and
inverter-side DC voltage vdr , from the eigenvectors of the Adaboost learner.
To achieve fast continuous commutation failure warning, the length of the feature
vector data used for warning should not be excessive; otherwise, the information used
for fault warning is redundant and increases the computational effort of the warning
algorithm and the training difficulty of the learner. Therefore, a window operation
for each feature quantity within the feature vector is needed; simultaneously, the data
window of the feature quantity cannot be too small for reliable warning; otherwise,
it may lead to insufficient information for warning, thus increasing the probability
of misclassification. To simplify the model to reduce the training difficulty of the
classifier, the standard youngest value is selected for all measured quantities.
To correspond to the data structure of a single-level decision tree, these windowed
and scalarized notation-valued eigenvolumes are joined in the first place as a column
vector as follows:
T
x ∗ = U̇a∗T , U̇b∗T , U̇c∗T , Idr
∗T ∗T
, vdr (4.63)

To integrate with field practice and to make the problem of adding windows to
the characteristic data concrete, the following assumptions are made:
(1) The DC warning system has the ability to determine the occurrence of the
first commutation failure and to mark the moment of its occurrence within a
negligible time cost.
(2) The early warning system has some recording capability for measurement data.
Based on the above assumptions, the data addition windowing operation and subse-
quent warning process based on the trained Adaboost learner are given, combined
with that shown in Fig. 4.22. The specific process is described as follows:

(1) After a large disturbance occurs in the inverter-side AC system, the warning
system instantaneously determines that the first commutation failure has
occurred and calibrates the moment of occurrence of the first commutation
failure to activate the fault warning system.
(2) The early warning system reads from the previous measurement data, records
the AC voltage of the half-period before the moment of occurrence of the first
commutation failure U̇a , U̇b , U̇c , the inverter-side DC current Idr , and inverter-
side DC voltage vdr of the recorded data, and continues to record the measure-
ments of the characteristic variables for half a perimeter wave to complete the
windowing operation of the measured data.
180 4 Commutation Failure Prevention and Control

. . .

Fig. 4.22 Early warning system data addition windowing method and warning process

(3) A scalarized feature vector is constructed according to the feature vector design
described in Eq. (4.63), input into a pretrained Adaboost learner for the mapping
calculation of Eq. (4.62) and for making an early warning of whether successive
commutation failures will occur after the first commutation failure based on the
output of the Adaboost learner.

The application scenario that was previously described is given in the hypothetical
presence of a trained Adaboost learner, while during the training process, to ensure
uniformity between the training process and the field engineering application, the
data addition windowing methods and labeling methods for the training and test
samples used during the classifier training process are kept consistent with those
mentioned above for the field run hypothesis [24].
3. Other early warning measures
Relatively few relevant studies on early warning measures for continuous commu-
tation failures exist, and other than early warning measures based on recovery char-
acteristics, no other prediction methods based on the characteristics of continuous
commutation failures have been solely proposed for continuous commutation fail-
ures. The nature of the occurrence of continuous commutation failures is multiple
consecutive occurrences of single-commutation failures. Therefore, early warning
measures for the first commutation failures can also be utilized to predict the occur-
rence of each commutation failure in a continuous commutation failure as well.
Successive commutation failures occur because the shutdown angle is repeatedly
less than the minimum value.

γ K < γmin (4.64)

4.2 Early Warning Measures for Commutation Failure in a Hybrid … 181

where K denotes the number of cycles.

In addition, Sect. 4.1.3 of this section analyzes the mechanism and characteristics
of successive commutation failures in multifeed DC and concludes that the absorption
of reactive power by a particular DC line during recovery is the cause of commutation
failures in its adjacent DC lines. This mechanism is also applicable in the single-fed
DC system: this DC line will absorb a large amount of reactive power during the
recovery period, and when the reactive power of the AC system is not sufficient
for supporting its consumption, it will trigger a secondary drop in the commutation
bus voltage, and a commutation failure may occur, i.e., a single-fed DC successive
commutation failure. Therefore, the risk of continuous commutation failure can also
be reflected from the other side by online, real-time evaluation of the dynamic reactive
power standby of the system [25].
The dynamic reactive power standby of a generator is usually defined as the
reactive power that can actually be added to the generator during transients under
a given fault. For continuous commutation failure requirements, the reactive power
generated by the generator before the AC fault disappears is not the focus of attention
and the reactive power generated by the generator after the fault is cleared has a major
role in suppressing the secondary dip in the commutation bus voltage. Therefore,
this reactive power is defined as the effective dynamic reactive power reserve for
continuous commutation failure.
 
Q D R R j = K qls j Q j max − Q j0 (4.65)

where Q j0 is the generator reactive power output before the fault and K qls j is the
reactive network loss correction factor to be considered at the converter bus for
equating the reactive power reserve to the converter bus.
The values of the effective dynamic reactive power reserve of the generator that
were previously defined differ for different operating conditions. As the DC delivered
power increases, more reactive power is required for DC power recovery after a fault,
and the corresponding generator generates more additional reactive power. As the
load power of the DC recipient grid increases, which can be equated to the reduction
of reactive power compensation capacity in the DC near zone, the secondary voltage
dip of the converter bus is larger and requires more reactive power from the generator.
The effect of the converter bus voltage dip on the effective dynamic reactive power
standby of the generator is shown in Fig. 4.23.

4.2.3 Early Warning Measures for Multifeeder DC

Commutation Failure

Few studies investigate early warning measures for multifeeder DC commutation

failure. The mechanism of multifeed DC simultaneous commutation failure is iden-
tical to that of the first commutation failure, and the electrical distance between each
182 4 Commutation Failure Prevention and Control

Fig. 4.23 Maximum effective dynamic reactive power standby

DC converter station is short. Thus, the risk of simultaneous commutation failure can
be effectively predicted at each DC converter station by using early warning measures
for the first commutation failure. Alternatively, the multifeeder interaction factor and
the simultaneous commutation failure boundary condition can be used to discrimi-
nate the adjacent DC directly after a commutation failure occurs in a certain return
DC, which has certain requirements on the communication between two returns.
For the prediction of successive commutation failures in multifeeder DCs, there
are no targeted measures from mechanism analysis, with the exception of the use
of early warning measures for the first commutation failure. Some related studies
have proposed a neural network-based, phase change voltage prediction method for
predicting the shutdown angle; however, this method is more dependent on the sample
dataset [26].
For a fault with a particular parameter, both the phase change voltage and the
arc extinguishing angle during the fault are uniquely determined. As shown in
Fig. 4.24, a fault process is divided into two phases according to time, with one set
of characteristics {x1 , x2 , . . . , xn } to characterize the phase change voltage of phase
I and another set of features {y1 , y2 , . . . , yn } to characterize the arc-extinguishing
angle of phase II. With sufficiently complete feature extraction, {x1 , x2 , . . . , xn } and
{y1 , y2 , . . . , yn } should correspond to each other. This correspondence is indepen-
dent of the fault parameters and can therefore be used for subsequent prediction of
the arc extinguishing angle.
The data drive is to obtain this correspondence from the samples; its effectiveness
depends on the completeness of the feature extraction and the number of training
samples. For commutation failure prediction, the actual available fault samples are
usually small, and the number of features that can be extracted from the phase change
voltage is quite limited due to the limited sampling accuracy. Therefore, data-driven
direct application to commutation failure prediction cannot better reflect the mapping
4.2 Early Warning Measures for Commutation Failure in a Hybrid … 183

Fig. 4.24 Correspondence between phase change voltage and arc extinguishing angle characteris-
tics during a fault

relationships among data but can serve as a complement and reinforcement to the
mapping relationships disregarded in the mechanism-driven approach. A schematic
of the data physical fusion approach to commutation failure prediction is shown in
Fig. 4.25.
The method of commutation failure prediction based on the physical fusion of
data can be divided into two parts: training of the error correction model and actual
commutation failure prediction. The training process of the error correction model
can be divided into the following steps:
(1) A mechanism-driven approach based on data collected during faults for each
electrical quantity based on historical fault samples is used to obtain predicted
values of the arc extinguishing angle.
(2) Input features associated with the arc-out angle are extracted from the histor-
ical fault samples used as training samples with the true arc-out angle and
mechanism-driven arc-out angle predictions.

Fig. 4.25 Schematic of the method for predicting commutation failure with physical fusion of data
184 4 Commutation Failure Prevention and Control

(3) The error correction model is obtained by selecting a suitable algorithm for
training.
In the actual commutation failure prediction process, a fixed time interval T is
chosen. The phase change voltage at the beginning of the fault is sampled, each time
for a one-week wave, and N sets of sampling samples are obtained. The mechanically
driven arc-out angle prediction and input characteristics, which are input into the error
correction model to obtain the corrected arc-out angle prediction, are obtained in the
same way.

4.3 Commutation Failure Suppression for the AC/DC

Hybrid Grid

When a commutation failure is detected or predicted, commutation failure can be

suppressed by increasing the phase change area provision, decreasing the phase
change area requirement, increasing the shutdown area provision, and connecting
a current limiting device in series [27]. When the phase change area provision is
reduced, the on-time of the voltage–time waveform can be increased by using the
advance trigger method, and the voltage amplitude of the voltage–time waveform
can be increased by using SVC, STATCOM and other reactive power compensation
equipment to increase the phase change area provision and to avoid the commutation
failure caused by the incomplete phase change process. The rise in DC current will
cause the demand of the phase change area to rise, and the DC current can be limited
by a low voltage current limiting link (VDCOL). Series current limiting devices can
limit the fault current and slow the voltage dip, thus increasing the phase change
area. It is also possible to increase the shutdown angle adjustment value to maintain
the minimum shutdown area required for phase change, thus avoiding commutation
failure due to thyristor failure. In this section, typical suppression strategies for single-
fed DC first commutation failure, continuous commutation failure, and multifeed DC
commutation failure are presented, distinguished by the type of commutation failure.

4.3.1 Suppression Measures for the First Commutation

Failure

1. Commutation failure suppression based on advance trigger control

The previous commutation failure prediction control is based on the zero sequence
voltage of the asymmetric ground fault and the magnitude of the Clark transform
static coordinate system voltage of the three-phase symmetric fault as the judgment
criterion of commutation failure, which itself is not the same judgment criterion of
commutation failure for different fault types The fault voltage of the same fault type
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 185

is influenced by the fault time and fault distance and cannot be completely unified,
which leads to the existing commutation failure. This situation contributes to the
shortcomings of the existing commutation failure prediction control, which is mainly
reflected in the excessive reliance on simulation and the lack of quantitative analysis at
the theoretical level, while the control is likely to overshoot the reduction of the trigger
angle, leading to an increase in reactive power consumption, which is not conducive
to the suppression of commutation failure [26, 28]. The main shortcomings are that
the prediction criterion is too reliant on simulation and lacks quantitative analysis.
However, the suppression measures based on advance trigger control can be inde-
pendent of the fault type, fault time and fault distance, which unifies the judgment
standard of commutation failure. Thus, on the theoretical basis of the fast prediction
method of commutation failure based on voltage waveform fitting, the corresponding
delayed trigger angle is calculated using the critical phase change area for successful
phase changes under different fault types and degrees with real-time sampling. Early
triggering can be realized to suppress the occurrence of commutation failure.
According to the phase change voltage time area theory, if commutation failure
does not occur, i.e., Sneed < Smin where γmin is generally taken as 7–10°, and the
subsection is taken as γmin =7°. When Sneed = Smin , the maximum delayed trigger
angle on the inverter side that satisfies the phase change success condition can be
obtained when αimax . From the commutation area, it is obtained that

π−γmin − ϕ π−γmin − ϕ

2X C Id = u L min dω0 t = U L min sin(ω0 t + ϕ1 )dω0 t (4.66)

αi max αi max

The maximum delayed trigger angle on the inverter side i αi max is expressed as

2X C Id
αi max = arccos + cos(π − γmin − ϕ) (4.67)
U L min

where X C is the inverter-side phase commutation reactance, Id are the DC currents,

and ULmin and φ are obtained by fitting the curve by the three-point method.
Equation (4.67) shows that the postfault inverter-side trigger angle can be quan-
tified by the phase change area and no longer relies on empirical values to give
a strong universality, which can meet the demand for dynamic adjustment of the
delayed trigger angle under different operating and fault conditions.
The commutation failure fast suppression method control module is triggered
by the commutation failure prediction signal (CFPS) output from the commuta-
tion failure prediction module and is compared with the trigger angles output from
the original control system, such as fixed shutdown angle control and fixed current
control. Their minimum value is used as the final delayed trigger angle on the inverter
side. The control structure of this suppression measure is shown in Fig. 4.26.
When the DC system is normally operating, the CFPS output from the commuta-
tion failure prediction module is always zero, and the prediction control signal PCS is
always zero, at which time the trigger angle of the inverter side αi is the trigger angle
186 4 Commutation Failure Prevention and Control

Fig. 4.26 Block diagram of trigger angle control based on phase change area

of the original control system output αi SY S and π is the minimum of the two. Since
the value range of the inverter-side trigger angle is π/2 ∼ π , the trigger angle of the
inverter side, in this case αi must be αi SY S , ensuring that the commutation failure
control strategy based on the phase change area does not affect the normal operation
of the DC system. When the CFPS output of the commutation failure prediction
module changes from 0 to 1, the prediction control signal PCS is 1, the trigger angle
of the inverter side αi is the trigger angle of the original control system output αi SY S ,
and the trigger angle obtained based on the phase change area αimax is the minimum
value of the two.
To test the effectiveness of the proposed method in this subsection, a test system
based on the CIGRE DC transmission standard model in PSCAD/EMTDC is built to
implement the proposed fast prediction and suppression measures for commutation
failure. The signal sampling frequency is 20 kHz; a minimum turn-off angle γmin
of 7° is chosen, and all other control parameters and fault settings of the model are
identical to the original standard system.
A single-phase ground fault, phase-to-phase fault and three-phase fault on the
inverter-side commutation bus are employed to count the occurrence of commutation
failure in the CIGRE standard test model and the predictive control model based on
the critical phase change area for different moments and degrees of fault conditions.
The single-phase ground fault is used as an example for the A-phase ground fault.
The phase-to-phase short-circuit fault is used as an example for the B- and C-phase
short-circuits. The statistical results are shown in Fig. 4.27.
According to the figure, this predictive control model reduces the value of ground
inductance for critical phase changeover failure under different types of AC side
faults, indicating that the trigger angle control strategy based on the phase changeover
area can effectively suppress the occurrence of phase changeover failure. As the
data shows, the highest reduction in critical inductance value is 0.2H for single-
phase faults, 0.12H for three-phase faults and 0.7H for phase-to-phase faults, and
the performance of the system against commutation failure has been significantly
improved.
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 187

(a) Results of commutation failure suppression under single-phase ground fault conditions

(b) Results of commutation failure suppression under three-phase fault conditions

(c) Results of commutation failure suppression under interphase fault conditions

Fig. 4.27 Suppression results of the commutation failure control strategy under different types of
faults
188 4 Commutation Failure Prevention and Control

To more intuitively observe the ability of the system to resist commutation failure,
the Commutation Failure Immunity Index (CFII) is introduced as an evaluation index
of the system’s ability to resist commutation failure [30]. The CFII is calculated as
shown in Eq. (4.68).
2
UacN
CFII = × 100 (4.68)
Z max × PdcN

where UacN is the rated voltage value of the inverter-side commutation bus. Zmax is the
phase commutation failure critical impedance. PdcN A larger CFII value indicates
that the smaller the fault critical impedance is, the more serious the fault causing
the commutation failure, the stronger the system’s ability to resist the commuta-
tion failure. The CFII is determined by the commutation failure critical inductance
provided in Fig. 4.27 and can calculate the commutation failure immunity factor
CFII for different fault types and fault moments, as Fig. 4.28 shows.
When different types of faults occur in the AC system, the ability of the quan-
tized trigger angle control model based on the area of commutation failure to resist
commutation failure is higher than that of the CIGRE standard model, especially
under single-phase grounding and phase-to-phase faults, and the resistance of the
system to commutation failure is significantly improved. Under three-phase faults,
the resistance improvement is relatively small because three-phase faults are gener-
ally more severe and the AC bus voltage drastically drops, which is not conducive
to the suppression of commutation failure.
2. Fault limiting-based commutation failure suppression
Series current limiting devices can compensate for the deficiencies caused by the
limited range and speed of trigger angle adjustment and help suppress commuta-
tion failures caused by severe faults. In addition to the AC side, current limiting
devices can be applied on DC lines to limit DC short circuit currents and to suppress
commutation failures caused by DC faults. Current limiting devices are devices with
resistive characteristics, such as resistors or superconducting current limiters.

(1) resistors

Figure 4.29 shows a schematic of a resistor-based commutation failure suppression

measure in an AC-DC hybrid grid [31]. As shown in the figure, the measure consists
of the following five main components:

1) A three-phase transformer of ratio a is connected in series with the receiving AC

grid.
2) Three-phase uncontrolled bridge rectifiers.
3) A is connected in parallel with a large resistor, which is used to limit the fault
current.
4) The CFPM Control System is utilized to determine the optimum resistance throw
ratio to increase the shutdown angle during a fault.
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 189

(a) CFII curve under a single-phase ground fault

(b) CFII curve under a three-phase fault

(c) CFII curve under phase-to-phase failure

Fig. 4.28 CFII curves for different fault types

190 4 Commutation Failure Prevention and Control

Fig. 4.29 Schematic of resistor-based, commutation failure suppression

5) A small inductor for limiting the rate of change of current. Due to the small
inductance value, the inductor can be designed as an empty core.

The core idea of this measure is to design a controllable resistor in series with the
faulty current using a switch that can be switched off and used to suppress commu-
tation failure. During normal system operation, the switch closes, the resistor is
bypassed, and the inductor can be considered shorted. Disregarding the voltage drop
across the power electronics and inductor, the voltage drop across the uncontrolled
bridge rectifier is almost 0. Therefore, the measure has almost no effect on the grid
during normal grid operation. When a fault occurs at point F, the controller detects
the occurrence of the fault by measuring the B1 bus voltage. During the fault, the
shutdown switch is operated under the control of the controller.
Selecting a six-pulse converter as an example and disregarding the phase change
overlap, the inverter AC side current can be expressed as

π √
1 6
I1acinv =√ Id cosωtdωt = Id (4.69)
2π π
−π

According to the relationship between the DC current and the AC current, with
Eq. 4.69, we obtain
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 191
√
12kV L L
I1acC F P M = (cosα + cosγ ) (4.70)
2πa X i

where k denotes a constant in the range 0 to 1. Considering the relationship between

the currents at the ends of the uncontrollable rectifier, Eq. 4.70 can be rewritten as
√ 
2aXi IdCFPM
γ = arccos − cosα (4.71)
kVLL

By controlling the shutoff switch, the controller can achieve, IdCFPM , which can
control the shut-off angle γ.
Figure 4.30 shows the control block diagram of the resistor-based commutation
failure suppression measure, which consists of 2 main parts: the fault detection part
and switchable gating signal generator. The fault detection section detects three-
phase, symmetrical and asymmetrical faults. The gate signal generator can control
the opening and closing of the switchable switch for the purpose of current control,
as shown in Fig. 4.31.

(2) Superconducting current limiter

The commercialization of high-temperature superconductors provides the basis

for the practical application of superconducting current limiters. According to
the working principle, superconducting current limiters can be divided into the
loss-of-superconductivity type and nonloss-of-superconductivity type. The loss-of-
superconductivity-type superconducting current limiters utilize the process of the
superconductor’s own transformation from the superconducting state to the normal
state. The superconducting current limiter works in the superconducting state during
normal operation of the power grid, which does not affect the normal operation
of the power grid. The superconducting current limiter loses its superconducting

Fig. 4.30 Block diagram of control of resistor-based commutation failure suppression measures
192 4 Commutation Failure Prevention and Control

Fig. 4.31 Turn-off switchable signal versus current

state during a short-circuit fault in the power grid and works in the normal state,
forming an impedance series into the circuit, thus realizing the function of limiting
the short-circuit current. Using the superconducting properties of superconductors,
the superconducting current limiter impedance is almost zero in normal operation,
and the impedance automatically increases when a fault occurs, quickly limiting the
short-circuit current to a level acceptable to the circuit breaker or system, and has
the characteristics of a fast response time. From the point of view of limiting the
fault current, the superconducting current limiter and resistor principle are equiv-
alent and rely on their own resistance to limit the current. The difference is that
the superconducting current limiter has the function of overcurrent automatic loss
of superconducting resistance. In the system, steady-state operation can be kept in
series in the line and will not affect the normal operation of the system but can also
assume part of the fault detection function (Fig. 4.32).
The resistance-type superconducting fault current limiter (RSFCL) is a device that
solely relies on the loss of superconducting resistance to limit current. When the grid

Fig. 4.32 Resistor versus

superconducting current
limiter
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 193

is in normal operation, the superconducting coil is in the superconducting state, and

the impact on the normal operation of the grid is negligible. When a fault suddenly
occurs, the superconducting coil is triggered by the fault current and changes from the
superconducting state to the normal state, generating a certain resistance and effec-
tively suppressing the short-circuit current. After the short-circuit fault is eliminated,
the superconducting coil automatically returns to the superconducting state.
Figure 4.33, VT1-VT6, VTB and VTF are thyristor groups. SupDC and SupBri are
thyristor-controlled, superconducting DC modules and thyristor-controlled, super-
conducting bridge arm modules, respectively, which contain superconducting current
limiting units. The circuit topologies of thyristor-controlled, superconducting DC
modules and thyristor-controlled, superconducting bridge arm modules are shown
in Fig. 4.34. The Y and D in the suffixes of all component names indicate the 2
different commutation bridges to which they belong; a, b, and c indicate the three
AC phases to which they belong; and IN and OUT indicate a parallel connection or
series connection, respectively, with the superconducting current limiting unit. zinv
and zfil indicate the equivalent impedance of the AC line on the inverter side and the
equivalent impedance of the filter on the inverter side, respectively. SupCoil is the
superconducting current limiting unit.
A total of two superconducting DC modules and six superconducting bridge arm
modules are required for a twelve-pulse converter, each containing one supercon-
ducting current limiting unit and two sets of thyristor valve banks. The structure
of the superconducting current limiting unit is shown in Fig. 4.35. Each supercon-
ducting current limiting unit is composed of multiple superconducting coils that are
connected in series and parallel to meet the requirements that the superconducting
current limiting unit does not lose its superconducting coil during steady-state oper-
ation of the system and that the resistance of the superconducting current limiting
unit is large enough when a system fault occurs. The superconducting DC module
has 2 ports; the left end shown in the figure is connected to the DC rectifier side, and
the right end shown in the figure is connected to the DC inverter side. The super-
conducting bridge arm module has 3 ports: the upper end and lower end shown are
connected to the upper bridge arm and lower bridge arm, respectively, and the middle
port is connected to the secondary winding of the converter transformer.
The superconducting current limiting unit can exhibit 2 states: a superconducting
state and a normal state. When the current that flows through the superconducting
current limiting unit is less than its critical current, the superconducting current
limiting unit is in the superconducting state, and the resistance is 0. When the current
flowing through the superconducting current limiting unit is greater than its critical
current, the superconducting coil loses its superconducting coil, and the supercon-
ducting current limiting unit is in the normal state, externally reflecting the resis-
tance characteristics. When the current flowing through the superconducting current
limiting unit is bypassed by the thyristor, the superconducting current limiting unit
gradually returns to the superconducting state in the liquid nitrogen environment.
194 4 Commutation Failure Prevention and Control

Fig. 4.33 Commutation failure suppression based on a superconducting current limiter

4.3.2 Suppression Measures Against Successive

Commutation Failures

1. Commutation failure suppression considering harmonics and phase change

voltage over zero offset

The strategy objective of the fixed shutdown angle controller in an HVDC trans-
mission system is to control the shutdown angle of the inverter to the rated value,
which is generally set to the minimum shutdown angle 18◦ . The control strategy for
a fixed shutdown area is to control the magnitude of the shutdown angle rectification
value such that the shutdown area is the minimum shutdown area. The flow chart of
the proposed algorithm is shown in the following figure. The sampled values of the
three-phase voltage of the converter bus are collected, and the zero sequence voltage
and the Clarke transformed, three-phase voltage amplitude are calculated. After the
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 195

Fig. 4.34 Thyristor-controlled superconducting current limiting module

Fig. 4.35 Superconducting current limiting unit

zero-sequence voltage is higher than the threshold or the three-phase voltage is lower
than the threshold, the commutation failure control module is activated. First, the
maximum phase angle offset is calculated, while the harmonic component calcula-
tion module starts to calculate the harmonic content, and the reference value of the
shutdown angle is calculated based on the maximum phase angle offset, harmonic
content, and degree of voltage amplitude drop. Since the current controller and fixed
shutdown angle controller use a take small strategy, when the output trigger angle
of the fixed shutdown angle controller is greater than the output trigger angle of
the current controller, the rectification value of the shutdown angle controller is
increased. The output of the current controller can be tracked to avoid sudden output
changes and to ensure that the shutdown angle controller trigger pulses are applied to
196 4 Commutation Failure Prevention and Control

the high voltage DC valve bank to control the DC transmission system and to avoid
continuous commutation failures [32].
The injected harmonics are mainly distributed in the range of 90–350 Hz when
the HVDC system rapidly changes. Therefore, the 2nd harmonic, which has a large
impact on the HVDC, and the (nk ± 1) harmonic are selected for calculation, where
n is the number of converter pulsations. The pulsating converter is then selected as
the 2nd harmonic, the 11th harmonic, and the 13th harmonic for calculation. Using
the second harmonic during HVDC operation as an example, we obtain

ϕ2 1 2U N
γ =π+ − arccos (1 − cos γmin 0 ) + cos ϕ2 (4.72)
2 2 U2

The amplitude and phase angle of the second harmonic can be calculated using
the sliding window iterative discrete Fourier algorithm as follows:

x2 (kτ ) = A2 cos(2ωkτ ) + B2 sin(2ωkτ ) (4.73)

−N +1
NCur"
2
A2 = x(iτ ) cos(2ωiτ ) (4.74)
N i=NCur

−N +1
NCur"
2
B2 = x(iτ ) sin(2ωiτ ) (4.75)
N i=NCur

where τ = T /N and NCur denotes the latest sampled data point.

Application of the proposed algorithm to the commutation failure control module
is depicted by the flow chart shown in Fig. 4.36. The zero sequence component is
used to detect asymmetric faults, and the α-β transformed modulus is used to detect
three-phase faults. When a fault is detected in the AC system, the commutation failure
control module is activated, and the shutdown angle is increased [33–35].
For a more intuitive representation, a test model was built based on the CIGRE
benchmark model, which was modified with the control system. The test model
is equipped with a constant current controller and constant voltage controller on
the rectifier side and a constant off angle controller, constant current controller and
constant voltage controller on the inverter side [36, 37]. In the case of a system
failure, the control system contains a low voltage current limiting link to allow the
system to recover [38]. In the event of a system failure, the control system contains
a low voltage current limiting link to allow the system to recover.
The simulation of the created DC voltage control module was performed for 0.9
p.u., 0.95 p.u., and 1.0 p.u. For the three states, the simulation results are shown in
Fig. 4.37.
The inverter-side AC bus contains three sets of filters to filter out the 11th harmonic
and 13th harmonic and to provide reactive power compensation.
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 197

Fig. 4.36 Flow chart of the

continuous commutation
failure suppression algorithm
considering harmonics and
phase change voltage over
zero offset

For the simulation with the 13th harmonic filter removed at 1.0 s, the inverter-
side AC bus voltage waveform is shown in Fig. 4.38. A significant voltage wave-
form distortion occurs after the filter removal moment. A more significant waveform
distortion occurs at the moment of filter removal and 150 ms after removal.
Continuous commutation failure occurred on the HVDC transmission line during
the filter throwing and cutting process. During filter removal, the first commuta-
tion failure occurred, and then the second commutation failure occurred at approxi-
mately 200 ms. The corresponding DC current and shutdown angle waveform plots
198 4 Commutation Failure Prevention and Control

Fig. 4.37 DC voltage simulation results

Fig. 4.38 Inverter side AC bus voltage

are shown in Fig. 4.39. The DC current rapidly rises when commutation failure
occurs and then falls and slowly recovers. The shutdown angle drops to zero, and
commutation failure occurs (Fig. 4.39).
The current of the commutation valve is shown in Fig. 4.40; note that two
commutation failures occurred.
Fourier decomposition of the commutation bus was performed to obtain the ampli-
tude of each harmonic, as shown in the figure below. The filter throw caused numerous
harmonics, with the fundamental voltage amplitude falling and the harmonic voltage
amplitude rising, triggering the first commutation failure (Fig. 4.41).
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 199

(a) DC current

(b) Angle of disconnection

Fig. 4.39 High voltage DC response without mitigation

Fig. 4.40 Valve current during commutation failure

The total harmonic distortion of the phase change voltage (total harmonics distor-
tion, THD) is shown in Fig. 4.42. The harmonic distortion rate rises when 2 commuta-
tion failures occur, and it can be judged that the harmonics caused the first commuta-
tion failure. The total harmonic distortion rate also increases more during the second
commutation failure.
The 2nd, 11th and 13th harmonic amplitudes are extracted. After the first commu-
tation failure, the 2nd harmonic amplitude rapidly rises, which can be judged as a
rise in DC current causing saturation of the converter transformer, which causes a
large number of harmonics leading to the second commutation failure (Fig. 4.43).
200 4 Commutation Failure Prevention and Control

Fig. 4.41 Amplitude of the fundamental wave and each harmonic

Fig. 4.42 Total harmonic distortion rate of the phase change voltage

Fig. 4.43 Amplitudes of the 2nd, 11th, and 13th harmonics

To suppress the harmonic-induced continuous commutation failure, the proposed

algorithm is modeled in PSCAD/EMTDC, the dynamic calculation of the off angle
controller rectification value is performed during filter bank throwing, and the simu-
lation results are shown in Fig. 4.44 When the filter bank is removed, the shutdown
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 201

Fig. 4.44 Commutation failure control module output

angle integer value rapidly increases to improve the shutdown margin of the subse-
quent valve bank. Subsequently, to track the fixed current controller, the shutdown
angle adjustment value continues to increase and then plateaus.
The response of the DC system is shown in Fig. 4.45. After applying the proposed
algorithm, the overshoot of the DC current is smaller, and the recovery is smoother.
While the turn-off angle drops to zero when the AC filter is removed and the first
commutation failure occurs, the turn-off angle rapidly increases and remains at
a larger value afterward, ensuring the phase change margin and thus effectively
avoiding the second commutation failure.
The response of the DC control system is shown in Fig. 4.46. After the commu-
tation failure occurs, the inverter-side control system switches from a constant-off
angle controller to a constant-current controller. The proposed algorithm tracks the
output of the current controller and acts on the shutdown angle rectification value,
and the HVDC control system switches again from the constant current controller to
the constant shutdown angle controller, ensuring that the commutation failure control
module takes the lead.

2. Commutation failure suppression based on AC protection optimization

The converter components of LCC-HVDC are unable to withstand fault shocks and
do not have their own shutdown capability. Simple faults in the AC system in the
near zone of the inverter-side converter station, if not cleared in time, can easily
induce chain accidents, such as simultaneous and subsequent commutation fail-
ures in multiple converter stations and single/bipolar blocking in the DC system.
Existing studies of commutation failure have given less consideration to the subse-
quent commutation failure caused by AC protection. After the occurrence of an AC
fault, according to the existing DC control system logic, DC blocking is easily trig-
gered. The DC system cannot be recovered yet after the AC protection has excised
the fault. In AC protection, directional elements, reclosing, dead zone protection and
failure protection may make HVDC evolve from the first commutation failure to a
continuous commutation failure. Therefore, from the perspective of AC protection
optimization, avoiding the occurrence of continuous commutation failure is a new
202 4 Commutation Failure Prevention and Control

(a) DC current

(b) Angle of disconnection

Fig. 4.45 High voltage DC response with the proposed algorithm

Fig. 4.46 High voltage DC control mode switching

idea for suppression of commutation failure in AC-DC hybrid networks. Increasing

AC protection quick-action, such as failure protection and rapid excision of high
resistance faults, can avoid commutation failure due to fault excision delay. Selecting
a high-resistance fault as an example, due to the small fault current, which affects
the sensitivity of the protection start element, the protection exit time can exceed
120 ms in the case of slow fault development, resulting in a long time reduction of
the phase change area, which can easily trigger continuous and subsequent commu-
tation failure and increase the risk of DC station lockout. In addition, the nonlinear
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 203

characteristics of the arc current will cause waveform distortion of the phase change
voltage; the phase change area is reduced, and the total harmonic distortion rate
exceeds 4%, which may cause DC system commutation failure.
High resistance fault protection of AC lines in the near zone of LCC converter
stations based on voltammetric characteristics is one of the measures to avoid commu-
tation failure due to high resistance fault excision delay [39]. The distortion char-
acteristics of the voltammetric characteristic curve are used to construct the high
resistance fault initiation element, which can effectively accelerate the initiation
speed. The algorithm flow is shown as follows:
Step 1: Check the three-phase voltage at the protection installation u a (t) u b (t),
u c (t) and zero sequence current i 0 (t). (a) The one-week waveform data of the
protection installation are normalized to the maximum value.
Step 2: Use the phase voltage and zero sequence current industrial frequency phase
angle difference to select the phase. If there exists a certain phase angle difference
absolute value ϕ in the range of 0 to 15° , then the phase is the faulty phase, and
proceed to step 3; otherwise, repeat step 1.
Step 3: Phase shift the zero sequence current so that its phase difference with the
fault phase voltage is minimized. Use the least squares method to segmentally linearly
fit the voltammetric characteristics of the fault phase voltage and zero sequence
current. The slope of the line near the fault phase voltage over zero point is measured
as k1 , and the slope of the line near the maximum point is calculated as k2 . Calcu-
late the fault phase voltage u ϕ (n) and zero-sequence current i 0 (n) The correlation
coefficients of Rc are shown as follows:
#N
k=1 i 0 (k)u ϕ (k)
R c = # #N (4.76)
N
k=1 i 0 (k) u 2ϕ (k)
2
k=1

Step 4: If 0 < k2 < 1, k1 > kset , and Rc > 0.966, then it is determined that a
suspected high resistance ground fault has occurred; where kset is the integer value,
take a real number greater than 1. In this section, use kset = 1.05. Considering
that the phase error between the fault phase voltage and the zero sequence current
at the measurement point generally does not exceed 10°, and then considering a
certain margin, take Rc The threshold value is 15°, corresponding to the value of
cos 15◦ = 0.966.
Step 5: Repeat steps 1 to 4. If the suspected high-resistance ground fault lasts
longer than one cycle, the fault is judged to be a high-resistance ground fault, and
the protection is activated.
On the basis of the high-resistance fault identified by the starting element, the total
area of the volt-ampere characteristic curve enclosed in the first and fourth quadrants
can be used to determine the direction of the fault, essentially characterizing the
dissipated energy of the arc. The calculation is presented as follows.
204 4 Commutation Failure Prevention and Control

N /4
"
S1 = sign(i 0 (k)) · u ϕ (k) · |i 0 (k + 1) − i 0 (k)| (4.77)
k=1

N /2
"
S2 = sign(i 0 (k)) · u ϕ (k) · |i 0 (k + 1) − i 0 (k)| (4.78)
k=N /4+1

S = sign(S1 + S2 ) · |S1 − S2 | (4.79)

where S1 is the area enclosed by the voltammetric characteristic curve during the
increase of the zero-sequence current from zero to the maximum value, S2 is the
area enclosed by the voltammetric characteristic curve as the zero-sequence current
changes from the maximum value in the opposite direction, i 0 (k) is the standard-
ized zero sequence current sequence, u ϕ (k) is the standardized fault phase voltage
sequence. Calculate the total area S. When calculating the total area, the area of the
part of the voltammetric curve in the first quadrant is positive, and the area of the
part in the fourth quadrant is negative.
According to Fig. 4.47a and c, the voltammetric characteristic curves of the posi-
tive direction high resistance fault are mainly distributed in quadrants one and three,
and the area located in the first quadrant is much larger than that in the fourth quad-
rant, with the total area S > 0 The voltammetric curves of a high-resistance fault
in the opposite direction are mainly distributed in the second and fourth quadrants,
and the area in the fourth quadrant is much larger than the area in the first quadrant,
with a total area of S < 0. Therefore, the directional element rectification value
Sset is set. If S > Sset , the positive direction is judged to be a high resistance fault;
conversely, if Sset > 0, it is judged to be a high resistance fault in the opposite
direction. The smaller the value is, the more sensitive the directional element; in the
case where the voltage and current data are standardized, considering that the total
area of the first quadrant is 1, and the phase voltage and zero sequence current are
almost in phase when the positive direction is determined to be a high resistance
fault, the voltammetric characteristic curve is thin and long, and the enclosed area
is very small. Thus, the integer value should be a value much smaller than 1, which
can be set to Sset = 0.01.

4.3.3 Suppression Measures for Multifeed DC Successive

Commutation Failures

Successive commutation failures are caused by the evolution of faults in multifeeder

DC transmission systems in severe fault scenarios, which lead to the following two
scenarios.
1) Voltage interaction effect between two converter buses in a multifeeder DC trans-
mission system: internationally, the multifeeder interaction factor MIIF is usually
4.3 Commutation Failure Suppression for the AC/DC Hybrid Grid 205

1 1 1

fault phase voltage /pu

fault phase voltage /pu

fault phase voltage /pu

0 0 0

-1 -1 -1
-1 0 1 -1 0 1 -1 0 1
zero sequence current /pu zero sequence current /pu zero sequence current /pu
(a) Positive fault phase (b) Positive nonfault phase (c) Reverse fault phase

Fig. 4.47 Voltammetric characteristic curves of fault and nonfault phases for high-resistance
ground faults

sampled to measure this effect. This subsection achieves the analytical calcula-
tion of MIIF through mathematical derivation and shows that the voltage fluctu-
ation of the faulted proximal converter station does have a certain effect on the
voltage of the adjacent converter station. Thus, if a commutation failure occurs
at the faulted proximal converter station, its converter bus fault voltage change
will directly affect the adjacent converter bus voltage.
2) Interaction between AC and DC systems during faults: The analysis of the fault
evolution mechanism in the previous section reveals that the fault characteristics
of the AC fault at the receiving end are largely influenced by the DC control
system and that if the DC system is under a conventional control strategy, there
is a negative interaction between the AC system and the DC system, i.e., the DC
control strategy triggers a secondary drop in the converter bus voltage.
In summary, the causative factor for the occurrence of successive commutation failure
is the secondary drop in the converter bus fault voltage caused by the interaction
between AC systems and DC systems, and the scenario of successive commutation
failure depends on the strength of the voltage interaction between two converter buses.
Therefore, the technical ideas regarding suppression of the successive commutation
failure mainly follows two directions: 1) weaken the negative interaction between
AC and DC systems during the fault process by improving the DC system control
strategy to reduce not only the secondary drop of the converter bus fault voltage but
also the possibility of successive commutation failure from the causative factors; 2)
destroy the scenario of successive commutation failure by reasonably planning the
DC drop point to weaken the voltage interaction between two converter buses. The
scenario of commutation failure occurs.
206 4 Commutation Failure Prevention and Control

4.4 Summary

In this chapter, the mechanism of DC first commutation failure and continuous

commutation failure, as well as multifeed DC simultaneous commutation failure
and successive commutation failure, are analyzed according to their fault charac-
teristics, which provide a theoretical basis for the study of DC commutation failure
warning methods and suppression measures.
Based on the generation mechanism of commutation failure, several early
warning measures and suppression measures for first commutation failure, contin-
uous commutation failure and multifeeder DC commutation failure are introduced
to distinguish the types of commutation failure. This research lays the foundation for
realizing early warning and protection of interlocking faults in AC-DC hybrid power
grids and for ensuring safe and stable operation of AC-DC hybrid power grids.
Chapter 5
DC Participation in Emergency Tidal
Control

With the extensive development and application of high-voltage direct current

(HVDC) transmission technology, China’s large-scale regional power grid with
multiple DC interconnection scenarios is increasing and has become the world’s
largest, highest-voltage-level, and most-complex-to-operate large-scale AC-DC
hybrid power grid. While high-voltage DC transmission brings many technical
advantages, the nonlinear discrete control process of the DC system also makes
the dynamics of the AC power grid change. Multiple types of constraints exist, the
coupling mechanism for the control and state variables is complex, and different
control objectives have the possibility of conflict. This makes difficulties in system
operation and control increase sharply. At the same time, the vulnerability of the
DC system to faults changes the scope of faults, which significantly increases the
complexity of power system operation and brings great challenges to grid scheduling
and safety control.
Some regions in the country have formed a structure of multiple-outer-zone DC
feeds into the same recipient AC grid, i.e., a multifeed DC system structure. In a
hybrid DC network containing line-commuted-converter (LCC) and voltage source
converter (VSC), the voltage support capability of the LCC recipient grid is an impor-
tant factor affecting the stability of the system. In actual operation, the interaction
between LCC-HVDC and VSC-HVDC is manifested in the reactive power support
of VSC to LCC. While the centralized feed-in of multiple DCs enriches the control
means of the system, it also brings challenges to the safe and stable operation of the
system. Considering the fast power adjustment characteristics of HVDC, it is neces-
sary to analyze the maximum delivered power of HVDC and the voltage support
capability of the receiving system under actual operation scenarios to provide a
reference for designing the system operation control strategy. The maximum deliv-
ered power of HVDCs in multifeeder systems is influenced by the operating state of
adjacent HVDCs, and it is important to shorten the maximum delivered power calcu-
lation time when developing online control strategies. Based on the HVDC external
characteristics equivalence method, the multifeeder system is equated to a dual-bus
system to obtain its maximum delivered power calculation model, and the number

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 207
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_5
208 5 DC Participation in Emergency Tidal Control

of variables and equations is greatly reduced. The original multifeeder short-circuit

ratio MISCR index evaluates the system strength based on the system grid and rated
parameters of the HVDC, which leads to inaccurate evaluation results when used for
actual system evaluation. The HVDC is equated to the grounding impedance while
ensuring that the external characteristics of the HVDC remain unchanged before and
after equating. The multifeeder system is equated to a dual-bus system through grid
transformation, and the multifeeder operating short-circuit ratio (MIOSCR) index is
proposed for evaluating the system operating condition and margin. When multiple
HVDCs adjust their delivered power simultaneously, they may be unable to adjust
their delivered power to the commanded value due to reactive power deficiency. The
access of dynamic reactive power compensation equipment is required to enhance
the transmission capacity. The relationship between the access capacity of the regu-
lator and the grid parameters under the same power support demand and the effect
of the power support demand on the access capacity of the regulator under the same
grid parameters are investigated.
The traditional calculation and analysis method of a power system often adopts
the “point-by-point method,” i.e., the system is verified and analyzed under different
operation scenarios to determine whether the system is currently in a safe state. This
results in a large amount of calculation and time-consuming system operation anal-
ysis, and the formulation of existing AC/DC coordination control strategies mostly
relies on simulation calculation. It is difficult to control and quantify the overall
safety state of the system in the adjustment process. The safety domain method real-
izes the overall evaluation of the system safety state. This determines whether the
system is safe based on the location of the operation point in the safety domain. In
addition, the scheduler can obtain the safety margin of the operation point in different
directions as an important basis for the next safety control measures. Therefore, the
safety domain theory provides a new idea for the safety assessment and control of
the current complex AC-DC hybrid grid.
Based on the safety domain theory, this part proposes a definition of the multi-
objective safety domain of LCC- and VSC-type AC-DC hybrid power grids and its
portrayal method and puts forward a safety judgment method and safety margin calcu-
lation method of the operation point based on the flexible controllability of the DC
system in a large AC-DC hybrid system. The definition of the static safety domain
section of a hybrid AC-DC power grid and its inscription method are proposed,
and the influence of different control methods on the static safety domain of a DC
system is studied, i.e., the evolution characteristics of the static safety domain. A
definition of the decoupling safety domain and its inscription method are proposed,
and cooperative static safety control among multiple converter stations is studied
by using the decoupling safety domain. A control safety domain with time char-
acteristics is proposed, and the control safety domain with time characteristics is
studied. Another control safety domain with time characteristics is proposed, and
the time-series coupling relationship between the active and reactive power of DC
transmission is studied to ensure the safety of the DC adjustment process. Based
on the static safety domain, the correction control strategy of the AC-DC hybrid
grid, the cooperative control strategy of preventive control and correction control
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 209

measures and the optimal dispatching strategy after correction control are studied.
The above research takes advantage of the geometric advantage of the static safety
domain and the ability to continuously characterize the system operation state to
provide a theoretical basis and application guidance for the operation and control of
the hybrid AC-DC grid.

5.1 Study of Multidimensional Coupling Mechanism

of an AC-DC Hybrid Grid

In a hybrid DC network containing LCC and VSC, the voltage support capability of
the grid at the receiving end of LCC is an important factor affecting system stability.
In actual operation, the interaction between LCC-HVDC and VSC-HVDC is mani-
fested in the reactive power support of VSC to LCC. For a multifeeder system, taking
into account the coupling relationship between Pd , I d and voltage and other electrical
quantities of each DC, the HVDC maximum delivered power calculation method is
proposed. On this basis, taking into account the operating constraints of the HVDC
maximum delivered power and critical phase change angle, the multifeeder operating
short-circuit ratio and multifeeder operating margin coefficient are proposed to eval-
uate the actual operating state of the system. To achieve the required multi-HVDC
power support capability, a method for calculating the fixed capacity of the regulator
under this control objective is proposed.

5.1.1 Methodology for Evaluating the Degree

of Commutation Bus Voltage Interactions in Hybrid DC
Networks with Different Control Methods

In a hybrid DC network containing LCC and VSC, the voltage support capability of
the grid at the LCC receiver is an important factor affecting the stability of the system.
The interaction laws are as follows: (i) from the perspective of LCC, the different
control methods of VSC make its support to LCC different, and (ii) for multifeeder
DC, the change of LCC interstation control method also affects its interaction. From
the above analysis, it is clear that the change in LCC and VSC control methods makes
their interaction change as well.
The tidal equation for the incremental form of the system before and after the
hybrid DC grid disturbance is
   ∗
 
P J Pθ J PU θ
= ∗ (5.1)
Q JQθ JQU U
210 5 DC Participation in Emergency Tidal Control

where the power delivered by the DC system is connected to the DC because of the
small phase angle dependence of the AC system J Pθ , the J Qθ remain unchanged and
are the submatrices corresponding to the original AC Jacobi matrix. J∗PU J∗QU is the
modified Jacobi matrix after adding DC; there are 4n variables and 2n equations in
Eq. (5.1), and two variables need to be determined in advance to solve. Assuming a
reactive disturbance at commutator bus i, the amount of change in injected active and
reactive power at the relevant nodes can be determined from the system parameters,
with a total of 2n − 1 variables assuming that the amount of change in voltage
magnitude at commutator bus i is known. Then, only 2n variables remain to be
solved, which can be found by the above equation. The amount of change in the
injected active and reactive power is related to the

P j = P jd
(5.2)
Q j = Q jd

The amount of change in active and reactive power at the nodes of the converter
station varies for different control methods of the converter station, and the combi-
nation of different control methods for multiterminal DC needs to be discussed
separately.
For the DC network with a mixed connection of LCC and VSC, assuming that the
rectifier station is LCC and the inverter station is LCC and VSC, the control methods
are as follows: (1) rectifier station fixed DC current, inverter station VSC fixed DC
voltage, and inverter station LCC fixed arc quenching angle; (2) rectifier station fixed
DC current, inverter station LCC fixed arc quenching angle, inverter station LCC,
and inverter station VSC fixed current; and (3) rectifier station fixed current, inverter
station LCC, and inverter station VSC with sag control. Now, the control method
adopted is fixed current in the rectifier and fixed arc quenching angle adopted in the
inverter. At this time, the characteristic equation of the inverter is
⎧
⎪
⎪ I = Iorder
⎪ d
⎪
⎪
⎪ = γorder
⎪
⎪
γ √
⎨ Udi = 3 2 Nb K i Ui
π
Udi = Udii cos γ − π3 Nb X ci Id (5.3)
⎪
⎪
⎪
⎪
⎪
⎪ Pdi = Udi Id
⎪
⎪
⎩ Q di = −Id U 2 − U 2
dii di

where I d and I order denote the DC current and its given value, respectively; γ and γ order
denote the arc extinguishing angle and its given value, respectively; U doi denotes the
no-load DC voltage; U di denotes the DC voltage; N b and K i denote the number
of six-pulse converter bridges and the converter transformer ratio of each converter
station, respectively; and Pdi and Qdi denote the active and reactive power absorbed
by the converter station, respectively.
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 211

Based on the Taylor expansion of the multivariate function, the DC power

expression in incremental form before and after the perturbation is obtained as
⎧ √
⎪
⎨ Pdi = 3 2
Nb K i cos γorder Iorder Ui
π
√ (5.4)
⎪
⎩ Q di = 3 2
Nb K i cos γorder Iorder √Udi −U doi
Ui
π Udoi −Udi
2 2

Assuming a reactive disturbance in commutation bus 2, calculate the voltage

impact on bus 1. The above equations are combined to give
⎡ ⎤ ⎡ ⎤
θ1 C1
⎢ θ2 ⎥ ⎢ C2 ⎥
⎢ ⎥ ⎢ ⎥
⎣ θ3 ⎦ = ⎣ C3 ⎦U2 (5.5)
θ4 C4

U1 = C3 U2 (5.6)

M I I I F12 = C3 (5.7)

[(J21 J23 − J31 J22 )J14 + (J31 J12 − J11 J32 )(J24 − A24 ) + (J11 J22 − J21 J12 )J34 ]
M I I F12
 −1
J11 J12 (A33 − J33 ) + J11 J32 J23 − J21 J12 (A33 − J33 ) + J21 J32 (A13 − J13 )
=
−J31 J12 J23 − J31 J22 (A13 − J13 )
(5.8)
⎧ √
⎪
⎨ A13 = √ Nb K 1 cos γorder 1 Iorderl
3 2
⎪ π
A24 = π Nb K 1 cos γorder Iorder √Udi −U
3 2 doi
(5.9)
⎪
⎪ √
2
Udoi −Udi2
⎩
A13 = 3 π 2 Nb K 2 cos γorder 2 Iorder 2

where Aij is determined by Eq. (5.4).

In a DC network containing VSC, the DC affected by voltage action may be LCC,
VSC or mixed LCC-VSC DC. The calculation of reactive-voltage external charac-
teristics of mixed LCC and VSC DC is given here, where STATCOM is considered
a special form of VSC.
For VSC fixed reactive power control mode, d Q V SC /dU = 0, the reactive-voltage
characteristics of the VSC converter are
d Qd PLCC.N d Q LCC
= (5.10)
dU (PLCC.N + PV SC.N )dU
212 5 DC Participation in Emergency Tidal Control

Table 5.1 Results of MIIF

Capacity U-I Simulation Analytical
calculations
characteristic results results
ratio
100Mvar 0.015 0.244 0.229
0.02 0.285 0.278
0.025 0.320 0.310
150Mvar 0.015 0.186 0.178
0.02 0.229 0.217
0.025 0.256 0.246
200Mvar 0.015 0.149 0.144
0.02 0.180 0.175
0.025 0.209 0.204

1. VSC Constant AC Voltage Control

At this point, d Q V SC /dU = +∞, and the reactive-voltage characteristic of the VSC
converter is also positive infinity.
2. STATCOM control
At this time.
d Q ST I
= U+I (5.11)
dU U
where k = U/I is the U-I characteristic of STATCOM.
Based on the reactive-extravoltage characteristics of the LCC and VSC hybrid
DC, the degree of voltage interaction can be introduced in other control methods in
the same way.
The simulation is based on the CIGRE standard test model consisting of a double
feed into the DC system. The DC system receiver is connected through the AC system.
One DC is controlled by a constant current and constant arc quenching angle. The
ESCR is 4. In the other DC bus with STATCOM, the capacity and U-I characteristic
ratio change based on the simulation method and the above analytical method to
calculate MIIF. The results are listed in Table 5.1. STATCOM can significantly reduce
the MIIF value, and the error of the results obtained by both methods is small,
similarly replacing STATCOM with VSC. When VSC is controlled with the same
control method, the effect is similar; when VSC is the other control method, the
method is feasible.
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 213

5.1.2 Calculation of the Maximum Delivered Power

of HVDC Based on Equivalent Impedance

After equating the HVDC to impedance, the single feed-in system is shown in Fig. 5.1.
In Fig. 5.1, τ denotes the converter transformer ratio; δ denotes the phase angle
of the DC receiver bus voltage; E and ϕ denote the equivalent voltage magnitude
and phase angle of the AC system at the receiver, respectively; Z and θ denote the
equivalent impedance and phase angle of the receiver system, respectively; Bc and
Qc denote the converter station reactive power and compensated reactive power,
respectively; Z d and θ d denote the equivalent impedance and phase angle of the DC
system, respectively; and Pac and Qac denote the active and reactive power transmitted
from the converter bus to the receiver AC system.
The HVDC injection of active power can be expressed as Eq. (5.12). In the equa-
tion, Zdc and θ dc are variables that are only related to μ-related variables, and the
remaining quantities in Eq. (5.12) are constants. Therefore, the above equation is an
expression related only to μ related expressions. It can be introduced that d Id /dμ > 0
holds in the normal operating range, while d Id /dμ > 0, and d Pd /d Id = 0 corre-
sponds to the point of maximum delivered power. Therefore, d Pd /dμ = 0 likewise
corresponds to the maximum delivered power point. By solving for d Pd /dμ = 0, the
corresponding μ is substituted into the above equation to find the maximum HVDC
transmission power. Compared with the traditional method, the calculation can be
effectively reduced, and the HVDC maximum delivered power can be obtained by

Fig. 5.1 Equivalent impedance model of HVDC

214 5 DC Participation in Emergency Tidal Control

solving the equation only once, which greatly reduces the calculation volume.

1+Z 2
· ZZdc · cos θdc
Pd = − Z
 2 (5.12)
1 + ZZdc + 2 ZZdc · sin θdc

After equating the HVDC to impedance, the multifeeder system can be equated
to each bus based on the network transformation, as shown in Fig. 5.1. For each
commutation bus there is
    
U̇ = A Eeq Ė eq (5.13)

Among others,
⎛⎡ ⎤ ⎡ ż eq11 ż
⎤⎞
1 0 · · · żeq1n
  ⎜⎢ . ⎥ ⎢
ż d1 dn
⎥⎟
A Eeq = ⎜ ⎣ . . ⎦ + ⎢ .. . . . .. ⎥⎟
⎝ ⎣ . . ⎦⎠
ż eqn11 ż
0 1 ż d1
· · · żeqnn
dn

⎡ ⎡
⎤ I˙ ⎤
ż eq11 · · · ż eq1n ac1
  ⎢ . . . ⎥ ⎢. ⎥
Ė eq = ⎣ .. . . .. ⎦⎢ ⎣ .. ⎦
⎥
ż eqn1 · · · ż eqnn I˙acn
   
For [AEeq ], the terms in Eq. are only related to μ, and U 2 = [U̇ ]T U̇ ∗ , so for
commutation bus i, Ui 2 = gi (μ1 , · · · , μn ) holds. All equations of the commutation
bus in the network can be expressed by the following equation:
⎧ 2
⎪ f 1 (μ1 ) = g1 (μ1 , · · · , μn )
⎪
⎪
⎪
⎪ .
⎪
⎪ .
⎨.
⎪
f i2 (μi ) = gi (μ1 , · · · , μn ) (5.14)
⎪
⎪
⎪
⎪ ..
⎪
⎪ .
⎪
⎪
⎩
f n (μn ) = gn (μ1 , · · · , μn )

The above equation has a total of n equations and n unknowns. The original model
contains 3n equations and 3n unknowns. Compared with this, the number of variables
and equations is greatly reduced, and thus the computational efficiency is improved.
For the single-feed-in system, varying the system impedance, the time required
for the calculation of this paper’s method and the original model is compared. As
shown in Table 5.2, compared with the original method, this paper method greatly
reduces the calculation time while ensuring the accuracy of the calculation results.
For multifeeder systems, the system grid parameters are changed with the adjacent
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 215

HVDC operating state based on this paper method and the traditional method to
calculate Pdmax , as shown in Table 5.3.
As seen from Table 5.4, the calculation results of the two methods are the same for
each scenario, and the proposed method in this paper greatly reduces the calculation
time.

Table 5.2 Comparison of calculated results for single-feeder systems with different SCRs P dmax
SCR The maximum power curve method The equivalent impedance method
Pdmax(p.u.) Time/s Pdmax (p.u.) Time/s
2.5 1.0087 2.2056 1.0087 0.0005
3 1.042 2.1666 1.042 0.0005
3.5 1.0857 2.1364 1.0857 0.0005
4 1.133 2.1188 1.133 0.0006
4.5 1.1808 2.1647 1.1808 0.0005
5 1.2278 2.1707 1.2278 0.0006

Table 5.3 Comparison of calculation results for multifeeder system Pdmax under different MISCRs
MISCR 1 The maximum power curve method The equivalent impedance method
Pdmax1 (p.u.) Time/s Pdmax1 (p.u.) Time/s
2.3333 1.0598 4.4156 1.0598 2.6455
2.4667 1.0774 4.3118 1.0774 2.5055
2.6 1.0957 4.3605 1.0957 3.0605
2.7333 1.1143 4.3644 1.1143 3.1322
2.8667 1.1331 4.3384 1.1331 2.8351
3 1.1521 4.3213 1.1521 2.8519
3.1333 1.1710 4.3154 1.1710 2.6690

Table 5.4 Comparison of calculation results for multifeeder systems with different operating states
of adjacent HVDC Pdmax
I d2 (p.u.) The maximum power curve method The equivalent impedance method
Pdmax1 (p.u.) Time/s Pdmax1 (p.u.) Time/s
0.8 1.2178 4.3071 1.2178 2.6531
0.85 1.2017 4.3310 1.2017 2.6300
0.9 1.1854 4.3384 1.1854 2.6532
0.95 1.1688 4.3135 1.1688 2.8475
1 1.1521 4.3213 1.1521 2.8672
1.05 1.1351 4.3000 1.1351 2.8957
1.1 1.1181 4.2905 1.1181 3.0644
216 5 DC Participation in Emergency Tidal Control

5.1.3 Short-Circuit Ratio and Operational Evaluation

Method for Multifeed-In Operation Based
on Equivalent Impedance

For a pair of ground conductors Yd , if the parallel bus voltage is U∠δ,U, then the
power injected into the bus by this pair of ground conductors is
∗
Sd = −U 2 Yd = −U 2 (G d − j Bd ) (5.15)

The joint vertical HVDC injection power and ground-directed injection power
expressions yield this ground-directed expression as follows:

G d = −C(cos 2γ − cos(2γ + 2μ)) (5.16)

Bd = Bc − C(2μ + sin 2γ − sin(2γ + 2μ)) (5.17)

Id
μ = ar c cos cos γ − −γ (5.18)
KU

Accordingly, it can also be expressed as an impedance to ground. The equiva-

lent impedance is actually the variable impedance associated with the HVDC oper-
ating state. Equations (5.16)–(5.18) above comprise the mathematical model of the
equivalent impedance.

1 1
Z d = Rd + j X d = = (5.19)
Yd G d + j Bd

For multifeeder systems, the voltage source nodes are eliminated by Davinan
equivalence. For the equivalent system, there are
 
 
     −Ui2  U2
 Z deqi  =  Z ∗ =   i 
deqi  • =  • 
(5.20)
 Sdeqi   Sdeqi 
 

The equivalent system is shown in Fig. 5.2.

Define the multifeeder operation short-circuit ratio indicator, as shown in the
following equation.
 
 Z deqi  Ui2
M I O SC Ri = =   (5.21)
|Z ii |  • 
|Z ii | Sdeqi 
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 217

Fig. 5.2 Multifeeder system after equivalence

Similarly, it can be assumed that the reactive power absorbed by each HVDC
inverter station is fully compensated by the shunt capacitor under the rated opera-
tion of the multifeeder system. The equivalent power injected by each HVDC into
converter bus i under rated operation can be obtained as follows.
    
 •  Z i j Pd j N
 Sdeqi N  = (5.22)
  |Z ii |

Ui2N U2
M I O SC Ri =   =   i N = M I SC Ri (5.23)
 •   Z i j  Pd j N
|Z ii | Sdeqi N 

Thus, MISCR is in fact a special case of the MIOSCR metric proposed in this
paper for the rated run case.
HVDC operation is limited by two factors: the critical phase change angle and the
maximum delivered power. In a strong system, as Id increases, the HVDC reaches
30° and then its maximum delivered power. The opposite follows for weak systems.
Therefore, the HVDC operating boundary depends on which of the two constrained
operating points is reached first, as shown in Fig. 5.3. The critical multifeeder oper-
ating short-circuit ratio is defined for each of the two operating points and is noted
as CMIOSCR_Pdmax and CMIOSCR_cri. Figure 5.3 shows the nominal values.
218 5 DC Participation in Emergency Tidal Control

Fig. 5.3 LCC-HVDC system operating boundary

As Id grows, the trend of MISCR and MIOSCR is shown in Fig. 5.4. Since MISCR
is related to the grid parameters only, it is a fixed value for a particular grid. MIOSCR
gradually decreases with increasing Id, which is consistent with the actual situation
in which the voltage support capacity of the system becomes weaker.
To evaluate the multifeeder system operating margin, the following multifeeder
system operating margin factor indicators are proposed.

Fig. 5.4 Comparison of MISCR and MIOSCR

5.1 Study of Multidimensional Coupling Mechanism of an AC-DC … 219

M I O SC Ri − max(C M I O SC R_Pd maxi , C M I O SC R_cri i )

M I O MCi =
max(C M I O SC R_Pd maxi , C M I O SC R_cri i )
(5.24)

Create a three feed-in DC system, fixing Id2 and Id3 to remain unchanged, and
obtain HVDC1 its Pd −Id curve for each scenario, as shown in Fig. 5.5. MIOSCR1
at Id1 = 1 for each scenario is shown in the figure. Obviously, the MIOSCR value
under each scenario corresponds to Idmax , while MISCR is a fixed value that cannot
reflect the actual voltage support capacity of the system.
For a three-feed-in system, Z1 = Z2 = Z3 = 1/2 and Z12 = Z13 = Z23 = 1/6 are
chosen, the adjacent HVDC operating state and shunt capacitor capacity are changed,
and Id1 = 1 and the MIOSCR values at the boundary points are calculated as follows
(Tables 5.5 and 5.6).
From the above results, it can be seen that MIOMC can accurately reflect the
multifeeder system operating margin compared with Id1cri . MIOMC reflects the
actual operating state as well as the operating range of the system in addition to the
system operating margin; therefore, it is more widely applicable.

Fig. 5.5 HVDC1 maximum power curve and variation in MISCR and MIOSCR for each scenario
 220

Table 5.5 MIOMC1 and CMIOSCR_Pdmax1 for each scenario

Pd2 Pd3 Bc1 Bc2 Bc3 MIOSCR1 (Id1 = 1) I d1max CMIOSCR_Pdmax1 MIOMC1 (%) Id I pd max
1 1 0.592 0.592 0.592 2 0.96 2.1632 −7.54 −0.04
1 0.9 0.592 0.592 0.592 2.3195 1.06 2.1565 7.56 0.06
1 0.8 0.592 0.592 0.592 2.5091 1.16 2.1348 17.53 0.16
0.9 0.9 0.592 0.592 0.592 2.5137 1.16 2.1456 17.16 0.16
0.9 0.8 0.592 0.592 0.592 2.6647 1.25 2.1479 24.06 0.25
1.05 1 0.592 0.592 0.592 1.8277 0.9 2.1935 −16.68 −0.1
1.1 1 0.592 0.592 0.592 1.8678 0.84 2.2198 −15.86 −0.16
1 1 0.65 0.592 0.592 2.1585 1 2.1585 0 0
1 1 0.7 0.592 0.592 2.2549 1.02 2.1981 2.58 0.02
1 1 0.65 0.62 0.592 2.2176 1.01 2.1877 1.37 0.01
1 1 0.65 0.62 0.62 2.2688 1.03 2.1845 3.86 0.03
0.9 0.9 0.65 0.592 0.592 2.5657 1.19 2.1582 18.88 0.19
0.9 0.9 0.7 0.592 0.592 2.6016 1.21 2.1816 19.25 0.21
1.1 1 0.65 0.65 0.592 1.8669 0.92 2.2274 −16.18 −0.08
1.1 1 0.65 0.65 0.62 1.945 0.94 2.2262 −12.63 −0.06
5 DC Participation in Emergency Tidal Control
Table 5.6 MIOMC1 and CMIOSCR_cri1 for each scenario
Pd2 Pd3 Bc1 Bc2 Bc3 I d _cri1 MIOSCR1 (Id1 = 1) CMIOSCR_cri1 MIOMC1 (%) Id I cri
1 1 0.592 0.592 0.592 1.37 3.7946 2.9826 27.22 0.37
1.1 1.1 0.592 0.592 0.592 1.32 3.4081 2.698 26.32 0.32
0.9 1.2 0.592 0.592 0.592 1.35 3.6037 2.8292 27.38 0.35
1.1 0.9 0.592 0.592 0.592 1.37 3.7794 2.9635 27.53 0.37
0.9 0.9 0.592 0.592 0.592 1.41 4.1465 3.2341 28.21 0.41
1 1.1 0.592 0.592 0.592 1.35 3.6103 2.8427 27.00 0.35
1.1 1.2 0.592 0.592 0.592 1.29 3.1915 2.5202 26.64 0.29
1.1 1.2 0.592 0.65 0.68 1.32 3.3362 2.6208 27.3 0.32
1.1 1.2 0.64 0.62 0.62 1.31 3.3038 2.6018 26.98 0.31
1.2 1 0.592 0.65 0.62 1.33 3.4657 2.7339 26.77 0.33
1.2 1 0.64 0.65 0.62 1.34 3.5142 2.7665 27.03 0.34
5.1 Study of Multidimensional Coupling Mechanism of an AC-DC …

1.2 1 0.592 0.68 0.592 1.34 3.4694 2.7173 27.68 0.34

1.2 1.2 0.63 0.65 0.65 1.28 3.1195 2.4322 28.26 0.28
221
222 5 DC Participation in Emergency Tidal Control

5.1.4 Multifeeder System Tuner Capacity Calculation

Method Based on Power Support Requirements

HVDCs require a large amount of reactive power to be absorbed during operation, so

their delivered power is strongly influenced by the bus voltage support capability. For
multifeeder systems, when multiple HVDCs adjust their delivered power simultane-
ously, the voltage and reactive power coupling between them and the AC system may
result in the HVDC not being able to adjust its delivered power to the target value.
The incoming shunt capacitors are usually close to full input, and their switching
speed is too slow to meet the real-time requirements. Therefore, dynamic reactive
power compensation equipment with faster regulation speed needs to be connected
to provide additional reactive power to enhance the HVDC delivered power value.
This paper analyzes the required regulator capacity under the target power demand
using a synchronous regulator as an example.
The operation of the governor camera is limited by the stator winding current. As
the DC current is continuously raised from the rated value, the synchronous regu-
lator will issue reactive power to stabilize the AC bus voltage. When the reactive
power output from the synchronous regulator reaches the rated capacity, the external
characteristics of the synchronous regulator behave as a controllable voltage source.
When the bus voltage can be maintained at the rated value, the DC power is contin-
uously increased with the DC current. When the reactive power output from the
synchronous regulator reaches the rated capacity, the external characteristics of the
synchronous regulator change to a current source, and the DC voltage continues to
drop because the bus voltage cannot be maintained. Thus, the DC power starts to
drop after reaching the maximum value.
When the system voltage support capacity is high, the HVDC delivered power
limit is determined by the operating point corresponding to μ = 30°. At the same
delivered power, the capacity of the regulator required to maintain the commutation
bus voltage of 1 is obviously larger than the capacity of the regulator operating in
current source mode. At the same time, a smaller μ under the same transmission
power means that the larger the U, the larger the capacity of the required input
regulator. Therefore, when each HVDC reaches its target delivered power value at
the same time, the scenario with μ of 30° corresponds to the smallest regulator access
capacity.
A two-feed-in system is used as an example to study the effect of grid parameters
and target delivered power values on the access capacity of the regulator. A two-
feed-in system model is developed as shown in Fig. 5.6. Set the target power Pd1set =
Pd2set = 1.2, Z2 = 1/3, Z12 = 1/6, and vary Z1 . Calculate the corresponding transfer
machine capacity. The results are shown in the following figure. In the figure, Qsc1
indicates the modulator access capacity.
As seen in the right panel of Fig. 5.6, the required regulator capacity increases
with increasing target power. In the calculation example MISCR1 = 3.5, when Pdset1
= 1.2, Qscmax1 = 0.22, which is too large for reactive power compensation. Therefore,
setting Pdset1 = 1.2 is not very reasonable for MISCR1 = 3.5. The HVDC target power
5.2 Multi-indicator Static Security Domain for AC-DC … 223

(a) Effect of grid parameters on access capacity (b) Effect of target power values on access
of regulator

Fig. 5.6 Effect of grid parameters and target power values on access capacity of regulator

should be reasonably designed, and the power support strategy should be developed
based on strengthening the current system grid, taking into account the system grid
and the capacity of the regulator.

5.2 Multi-indicator Static Security Domain for AC-DC

Hybrid Grids

China has formed the world’s highest-voltage-level and largest hybrid AC-DC grid,
and the safe operation of the AC-DC system is a major and urgent national need.
In the safety verification and monitoring of AC-DC systems, the traditional method
is often the “point-by-point method,” which is computationally intensive and time-
consuming. It is difficult to obtain feedback on the safety information of the system
from a global perspective. It is also difficult to determine the safety margin value
for the current operation state. Since the mathematical model of power systems can
be constructed by algebraic and differential equations and the set of mathematical
equations and geometric topology are interconnected, the study of power systems
can be transformed into the study of high-dimensional geometric problems.
On the basis of the mathematical model of the power system, taking into account
the constraints, we can obtain the “safety domain.” The current operating state corre-
sponds to an operating point in space. When the operating point is inside the safety
domain, the system can be judged to be safe and stable in the current operating state.
If the operating point has a large distance from the boundary of the safety domain,
then the system can be considered to have a large safety margin in the current oper-
ating state. Introducing the concept of a safety domain broadens the research ideas
of the power system discipline. Due to the special advantages that geometry has in
the research problem, the safety domain can provide a measure of the safety of the
system, which is beneficial to safety monitoring and safety control.
224 5 DC Participation in Emergency Tidal Control

According to the different research objects, the safety domain of the hybrid AC-DC
grid is the intersection of the static safety domain of the tidal operation constraint,
the small disturbance stability domain, the dynamic safety domain that takes into
account dynamic stability problems such as transient stability and voltage stability,
and the subsynchronous oscillation safety domain. Since the steady-state operation
control of AC-DC grids is the basis of various types of research and grid operation
regulation, the static safety domain is the focus of research in this part. The study
of the static safety domain needs to establish the relationship between the quasi-
steady-state model of the power system and the graphs in this coordinate system
on the basis of the high-dimensional right-angle coordinate system constructed in
Euclidean space, form the defined static safety domain, and complete the solution
and analysis of the static safety domain to reveal the safety problems existing in the
operation of the AC-DC hybrid grid and find the mechanism of the influence of DC
power transmission on the safe operation of the AC-DC grid.

5.2.1 Definition and Model of a Multimetric Static Safety

Domain for AC-DC Hybrid Grids

Since the static safety domain only takes into account the algebraic equations of
the power system, this part will first give the quasi-steady-state models of the AC
system, the LCC DC system and the VSC DC system. Then, a detailed analysis of
each operating constraint of the AC-DC system is presented as a basis to complete
the definition, geometric description and set of equations description of the static
safety domain of the AC-DC hybrid grid.
Let there be n + 1 nodes in the system and n l AC branches. Let node n + 1 be a
balance node, and classify the nodes into pure AC nodes and DC nodes. The nodes
connected to the primary side of the converter station are DC nodes, while pure AC
nodes are the nodes that are not connected to any converter station. Let there be a
total of n a pure AC nodes and n d DC nodes, where 1, 2, 3, . . . , n a is a pure AC node,
and n a + 1, n a + 2, . . . , n is a DC node.
For pure AC nodes, the set of tidal equations satisfies the following equation:

Pi = Ui U j (G i j cosθi j + Bi j sin θi j )
j∈i

Q i = Ui U j (G i j sinθi j − Bi j cos θi j )
j∈i

i = 1, 2, 3, . . . , n a (5.25)

where j ∈ i denotes node j (containing node i) connected to node i. It is possible for

j to be a pure AC node or a DC node, Ui is the voltage amplitude of the ith node, θi j
is the phase angle difference between nodes i and j, and G i j and Bi j are the real and
5.2 Multi-indicator Static Security Domain for AC-DC … 225

Fig. 5.7 Schematic diagram

of single line diagram for kth
LCC station

imaginary parts of the elements of the ith row and jth column of the node derivative
matrix, respectively.
For the n d DC nodes, writing both LCC DC transmission and VSC DC transmis-
sion into the same set of tidal equations and neglecting the commutation losses, we
have
 
Pi = Ui U j G i j cos θi j + Bi j sin θi j ± Udk Idk + Psk
j∈
 
Q i = Ui U j G i j sin θi j − Bi j cos θi j ± Udk Idk tan ϕk + Q sk
j∈

i = n a + 1, n a + 2, . . . , n (5.26)

where the subscript k denotes the kth converter station; Udk and Idk are the DC
voltage and DC current of the LCC DC converter station, respectively; ϕk denotes
the power factor angle of the AC side of the kth converter station; and Psk and Q sk
are the active power and reactive power transmitted to the AC bus by the VSC DC
converter station, respectively. Udk Idk , Udk Idk tanϕk correspond to the inverter and
rectifier stations, respectively.
A single line diagram schematic of the kth LCC station is shown in Fig. 5.7.
For the rectifier side of the LCC DC converter station, the DC voltage equation is
√
3 2 3
Udk = n k Ui cos αk − X ck Idk (5.27)
π π
where n k is the transformer ratio of the converter transformer, αk is the valve bridge
firing angle of the converter station, and X ck is the phase change reactance.
For the inverter side of the LCC DC converter station, the DC voltage equation
of Eq. (5.28) is satisfied.
√
3 2 3
Udk = n k Ui cos γk − X ck Idk (5.28)
π π
where γk is the arc extinguishing angle of the inverter-side converter station. In this
chapter, Ui is the AC bus line voltage.
The power factors of the AC side of the rectifier and inverter stations are
226 5 DC Participation in Emergency Tidal Control

Ud1k
cos ϕ1k = √ (5.29)
3 2
π
n k Ui
Ud2k
|cos ϕ2k |= √ (5.30)
3 2
π
n k Ui

where Ud1k and Ud2k denote the DC voltage on the rectifier side and the DC voltage
on the inverter side, respectively. ϕ1k and ϕ2k denote the power factor angle of the
rectifier side and the power factor angle of the inverter side, respectively (ϕ1k ∈
quadrant I, and ϕ2k ∈ quadrant II).
For a multiterminal DC system, the DC network equation of Eq. (5.31) is obtained
after eliminating the contact nodes in the DC network.

nd
Idk = ±Idk − gdk j Ud j = 0 (5.31)
j=1

where gdk j is the nodal conductance matrix of the DC network after eliminating the
contact node G d and the elements of the kth row and jth column. Ud j denotes the
DC voltage of the converter station connected to bus j, calledIdk . The first positive
and negative signs correspond to the rectifier and inverter stations, respectively. In
particular, in a two-terminal DC transmission system with

Rd Idk = Ud1k − Ud2k (5.32)

where Rd denotes the DC line resistance.

The single line diagram of the kth VSC flexible DC transmission converter station
is shown schematically in Fig. 5.8.
The active and reactive power transferred from the AC bus to the VSC station are

Psk = Ui Uck Yk sin(δk − βk ) + Ui2 Yk sin βk (5.33)

Q sk = −Ui Uck Yk cos δk + Ui2 Yk cos βk + Ui2 / X f k (5.34)

Fig. 5.8 Schematic diagram

of single line diagram for kth
VSC station
5.2 Multi-indicator Static Security Domain for AC-DC … 227

where Uck is the amplitude of the port voltage on the AC side of the converter station,
and δk is the phase angle difference between the AC bus and the port voltage of the
converter station (i.e., the phase shift angle).
Neglecting the losses in the converter, the active power injected into the converter
should be balanced with the DC power, i.e.,

Ui Uck Yk sin(δk + βk ) + Uck

2
Yk sin βk = Usk Is (5.35)

where Usk and Is denote the DC voltage and DC current of the VSC DC converter
station, respectively.
The relationship between converter station port voltage Uck and DC voltage Usk
is
μk Mk
Uck = √ Usk (5.36)
2

where μk is the DC voltage utilization, and Mk is the regulating system.

For the DC network, the DC network equation of Eq. (5.37) is shown below.

nd
Idk = ±Idk − gdk j Ud j = 0 (5.37)
j=1

The voltage magnitude of the ith node in the system Ui should satisfy the operating
constraints (including generator excitation voltage limits) of Eq. (5.38), and the
symbols corresponding to the superscripts max and min of the equation indicate the
upper and lower limits of the corresponding constraint limits, respectively.

Uimin ≤ Ui ≤ Uimax (5.38)

The phase angle difference between node i and node j θi j should satisfy
constraint (5.39), which is an approximation of the line current constraint.

j ≤ θi j ≤ θi j
θimin max
(5.39)

The active output of the ith generator other than the balancer Pi has an operating
constraint of

Pimin ≤ Pi ≤ Pimax (5.40)

The balance active output Pn+1 and reactive power output Q n+1 are constrained
to be
min
Pn+1 ≤ Pn+1 ≤ Pn+1
max
(5.41)
228 5 DC Participation in Emergency Tidal Control

n+1 ≤ Q n+1 ≤ Q n+1

Q min max
(5.42)

Pn+1 and Q n+1 The output power of the branch connected to the balancing node
and the load power of the balancing node are related and need to be found by tidal
calculations.
For LCC DC transmission, the ignition angle is too small with great insecurity, and
the ignition angle is generally required to be greater than 5°. If the effect of internal
reactance is not considered, then the ignition angle is equal to the power factor angle
of the AC bus on the rectifier side, which is limited by the reactive power, and the
power factor angle cannot be too large. Then, the ignition angle has a corresponding
upper limit.
The arc extinguishing angle must be large enough to avoid the failure of phase
change if the arc extinguishing time is too short. Usually, the arc extinguishing angle
is controlled at approximately 17°–21°.
Thus, the following constraints exist on the ignition angle α and arc extinguishing
angle γ .

α min ≤ α ≤ α max (5.43)

γ min ≤ γ ≤ γ max (5.44)

LCC DC transmission requires high reactive power support for the AC system.
When the reactive power injection is insufficient, the bus voltage drops. If the voltage
drops more and the reactive power compensation rate and capacity cannot keep up,
this will cause voltage instability, phase change failure and other problems. The
reactive power compensation capacity is partly provided by the filtering capacitor,
the reactive power compensation capacity of the rectifier station Q d1 and the reactive
power compensation capacity of the inverter station Q d2 . Their constraints (5.45) and
(5.46) are

d1 ≤ Q d1 ≤ Q d1
Q min max
(5.45)

d2 ≤ Q d2 ≤ Q d2
Q min max
(5.46)

To ensure the voltage stability of the system, it is generally required to operate

at a high SCR (i.e., SCR greater than 3). When the short-circuit capacity at the
commutation bus is constant, the DC delivered power Pd cannot be too large. The
low DC delivered power caused by the blocking of the converter station can lead
to problems such as power angle instability and frequency stability due to the wide
range of tidal shifts, so there is

Pdmin ≤ Pd ≤ Pdmax (5.47)

5.2 Multi-indicator Static Security Domain for AC-DC … 229

DC current Id and DC voltage Ud operating constraints (5.48) and (5.49) are as

follows:

Idmin ≤ Id ≤ Idmax (5.48)

Udmin ≤ Ud ≤ Udmax (5.49)

In addition, there is a constraint on the adjustment range of the converter

transformer ratio n, i.e.,

n min ≤ n ≤ n max (5.50)

Usually, when the ignition angle and arc extinguishing angle cross the limit for a
few seconds, tap control can be used to set it back to ensure that the trigger angle is
near the nominal value. The variable ratio quantity is a discrete quantity with a dead
zone. To simplify the analysis, this paper is equivalent to a continuous quantity.
The above constraints are established in the LCC DC transmission operation
under normal operating conditions. If the operation mode is abnormal mode and
fault mode, then the corresponding constraints must be changed and will not be
described in detail.
For VSC flexible DC transmission, active power PS and reactive power QS , the
DC voltage U S and DC current I S transmitted by the converter station to the AC bus
have the following constraints.

Psmin ≤ Ps ≤ Psmax (5.51)

Q min
s ≤ Q s ≤ Q max
s (5.52)

Usmin ≤ Us ≤ Usmax (5.53)

Ismin ≤ Is ≤ Ismax (5.54)

For VSC flexible DC transmission, the following constraints exist on the regu-
lating system M and the phase shift angle δ of the converter to ensure a suitable
operating condition for the converter.

M min ≤ M ≤ M max (5.55)

δ min ≤ δ ≤ δ max (5.56)

The static safety domain of the AC-DC hybrid grid can be defined as the set
consisting of each control quantity of the AC system, the active and reactive power
230 5 DC Participation in Emergency Tidal Control

of the load, and each control quantity of the DC system. This can be expressed as

= { (A, D)|s.t. f (X ) = 0, g(X ) ≤ 0} (5.57)

where A is each control and disturbance quantity associated with the AC system, and
D is each control quantity of the DC system. x is each electrical quantity of the full
system, f is the set of AC and DC current equations, and g is the operating constraint
that each electrical quantity of the full system should satisfy.
From the tidal equation, the static safety domain is a high-dimensional nonlinear
geometry enclosed by multiple high-dimensional surfaces in a Euclidean space, and
the static safety domain is a geometry composed of effective boundary surfaces.
Note that d1 and d2 are the control and state quantities of the DC system, respec-
tively. (P i , Ui ) are the active power output and end voltage magnitude excitation
settings of the ith generator, θ (P j , Q j ) is the active and reactive power of the jth
load, and θ is the nodal voltage phase angle. Then, the basic tidal equation of the
AC-DC system can be expressed as
⎧
⎪
⎪ d2 = f 1 (Ui , U j , θ, d1 )
⎪
⎪
⎪
⎪ Pi = f 2 (Ui , U j , θ, d1 , d2 )
⎪
⎪
⎪
⎪
⎨ = f 2 (Ui , U j , θ, d1 , f 1 (Ui , U j , θ, d1 ))
⎪
P j = f 3 (Ui , U j , θ, d1 , d2 ) (5.58)
⎪
⎪
⎪
⎪ = f 3 (Ui , U j , θ, d1 , f 1 (Ui , U j , θ, d1 ))
⎪
⎪
⎪
⎪
⎪ Q j = f 4 (Ui , U j , θ, d1 , d2 )
⎪
⎪
⎩
= f 4 (Ui , U j , θ, d1 , f 1 (Ui , U j , θ, d1 ))

where f 1 ∼ f 4 is the set of nonlinear equations. Combining d2 , U j and θ as dependent

variables, the above system of equations can be expressed as
⎧
⎨ d2 = g1 (Pi , Ui , P j , Q j , d1 )
U = g2 (Pi , Ui , P j , Q j , d1 ) (5.59)
⎩ j
θ = g3 (Pi , Ui , P j , Q j , d1 )

where g1 ∼ g3 is the system of equations to be solved. Transforming the phase angle

θ into the branch phase angle difference and accounting for d2 , U j and the bounded
upper and lower bounds on θ , the dependent variable in Eq. (5.59) should satisfy
⎧ min
⎪ d ≤ g1 (Pi , Ui , P j , Q j , d1 ) ≤ d2max
⎨ 2
U min ≤ g2 (Pi , Ui , P j , Q j , d1 ) ≤ U max (5.60)
⎪
⎩
j j

θ min
≤ A θ = A g3 (Pi , Ui , P j , Q j , d1 ) ≤ θ max
T T
5.2 Multi-indicator Static Security Domain for AC-DC … 231

Taking into account again the upper and lower limits of the constraints on the
control quantities and equilibrium nodes of the system, the set of constraints can be
converted into the set of inequalities in Eq. (5.57), where g(X ) ≤ 0.
Converting the inequality sign in Eq. (5.60) into an equality sign, a set of high-
dimensional surface equations is obtained, accounting for the linear hyperplane corre-
sponding to the linear constraints of the control quantities themselves and the equi-
librium nodes. These boundary surfaces form the corresponding static safety domain
of the AC-DC hybrid grid.

5.2.2 A Method for Inscribing the Full-Dimensional Static

Safety Domain of Hybrid AC-DC Grids Considering
Different Control Methods

In this chapter, a linear regression fit is used to approximate the boundary surface
expressions. Assuming that all LCC DCs in the system are controlled with constant
DC current on the rectifier side and constant DC voltage on the inverter side, the
fitted expression can be written as

n1
  n2
 
ε j PG j + χ j UG j + η j PL j + ξ j Q L j
j=1 j=1
n3
 
+ λ j Id j + ω j Uinv j + E i = PLi
max
(5.61)
j=1

where: n1 , n2 and n3 are the number of generators, loads and DC lines respectively,
ε j , χ j , η j , ξ j , λ j , ω j and E i are the coefficients to be fitted, PG j and UG j denote the
generator active output and end voltage amplitude respectively, PL j and Q L j denote
the load active and reactive power respectively, Id j and Uinv j denote the rectifier side
current and inverter side voltage respectively, and PLimax denotes the upper limit of
power transmission of line i.
For other types of DC control methods, simply change the DC dimension variable
in Eq. (5.61) and find its corresponding coefficient.
The linear goodness of fit can be expressed using R 2 . Let yi be the actual value of
the dependent variable of the fitted equation, ŷi be the actual value of the dependent
variable obtained from the fit, and y i be the mean of the dependent variable obtained
 2
from the fit. Noting the total sum of squares of deviations T SS = yi − y i , the
 2
regression sum of squares E SS = ŷi − y i , and the residual sum of squares
 2
RSS = yi − ŷi , the goodness of fit R 2 value can be expressed as Eq. (5.62)

E SS RSS
R2 = =1− (5.62)
T SS T SS
232 5 DC Participation in Emergency Tidal Control

In research on the static safety domain of power systems, it is usually considered

that the reactive power balance, network loss and voltage level of the system are
important factors affecting the accuracy of linear fitting, and the linear fitting method
is often difficult to apply to systems with low voltage levels, heavy reactive power
tides or high network losses. To further ensure the feasibility of the linear fitting
method, the system adopts the station self-compensation method to achieve the local
reactive power balance.
The IEEE14 node system is chosen as shown in Fig. 5.9. The basic operating
parameters of LCC DC transmission and VSC DC transmission are shown in
Tables 5.7 and 5.8, respectively.

Fig. 5.9 Modified IEEE14 AC/DC Hybrid Grid

Table 5.7 Basic operating

DC voltage 200 kV
parameters of LCC DC
system DC current 350A
Firing angle 21.79°
Extinction angle 23.61°
Commutation reactance 0.067H
DC resistance 28.54
Filter capacitance capacity 50MVar
Ratio of converter transformer 138 kV:166 kV
5.2 Multi-indicator Static Security Domain for AC-DC … 233

Table 5.8 Basic operating

DC voltage 200 kV
parameters of VSC DC
system DC current 266.7A
Reactive power at rectifier side −15.7MVar
Reactive power at inverter side 13.7MVar
Commutation reactance 11.42H
DC resistance 37.46
Rectifier side modulation degree 1.097
Phase shift angle of rectifier side 1.5616°
Inverter side modulation degree 0.997
Phase shift angle of inverter side 1.5947°

The fitting solution of the 71 effective boundary surface equations is completed.

In particular, the coefficients of the boundary surface equations corresponding to the
generator terminal voltage excitation limit, balancer active-reactive limit, DC current
limit and arc quenching angle limit can be obtained directly without fitting solutions
and are not found in this section.
The closer the goodness-of-fit correction value is to 1, the better the fit. From
Fig. 5.10, it can be seen that the goodness-of-fit correction values of the linear multiple
regression results are all above 0.96, indicating that the boundary surface of the
static safety domain of the hybrid AC-DC grid is approximately linear. There is
no significant difference in the goodness of fit of the effective boundary surface
corresponding to each type of operational constraint (Table 5.9).
fitting accuracy

Effective boundary surface number

Fig. 5.10 R2 values of the goodness of fit for each effective boundary of the linear fit
234 5 DC Participation in Emergency Tidal Control

Table 5.9 Limits of each operational constraint

Electrical parameters Upper limit Lower limit
Busbar voltage 1.15 0.95
line phase angle difference 20° −20°
Balancer active power output 5 0
Balancer reactive power output 2.5 −2.5
The DC voltage of LCC 1.8 1.2
The DC current of LCC 0.6 0.1
Firing angle 50° 5°
Extinction angle 35° 15°
Reactive power compensation capacity of LCC converter station 0.8 0
VSC DC voltage 1.6 1.0
VSC DC current 0.6 0.1
phase-shift angle 3° 0°
modulation degree 1.4 0
Reactive power setting value of VSC converter station 1 −1
Active power setting value of VSC converter station 1.1 0

5.2.3 Methodology for Inscribing Low-Dimensional Focal

Variable Safety Sections (Profiles) in the Static Safety
Domain of AC-DC Hybrid Grids

The solutions of the static safety domains in the previous section all extract safety
information in a high-dimensional space. To obtain a visual graph, a dimensionality
reduction process is needed. In this section, an accurate depiction of the static safety
domain boundary is given on a two-dimensional section (three-dimensional sections
are similar and will not be elaborated in this section). The section inscription is more
visual and facilitates safety control by dispatchers, but it is time-consuming and does
not give feedback on safety information in higher dimensional spaces. The depiction
of static safety domain boundaries on two-dimensional sections can be converted to
a nonlinear optimization problem. The LCC DC current and the arc extinguishing
angle are selected for the two-dimensional section as an example. In the optimization
solution, the constraints are still the tidal current equation constraint, equipment limit
inequality constraint, DC current control amount (which is gradually increased in a
certain interval by a given step), and objective function (the maximum and minimum
values of the arc-quenching angle). Finally, the DC current and its corresponding
maximum and minimum values of the arc extinguishing angle are connected point
by point to form the boundary of the static safety domain. That is, to solve the
optimization problem of Eq. (5.63).
5.2 Multi-indicator Static Security Domain for AC-DC … 235

h 1 = min γ
obj
h 2 = max γ
⎧
⎪
⎪ f (X ) = 0
⎪
⎪
⎨ g(X ) ≤ 0
s.t. (5.63)
⎪
⎪ (A, D) = (A0 , D0 ) (Id , γ ) ∈
/ D
⎪
⎪
⎩
Id = Id min + Id × k k = 0, 1, 2 . . .

where h 1 and h 2 are the objective functions. A and D are the respective AC gas and
DC control quantities in the static safety domain dimension, but D does not include
the selected section dimension variables (i.e., DC current Id and arc extinguishing
angle γ , the two control quantities). Ud A0 and D0 are the set values of each AC gas
quantity and DC control quantity in the static safety domain dimension, respectively,
i.e., the static safety domain is downscaled. Idmin is the left boundary of the DC
current corresponding to the horizontal coordinate of the 2D section, and Id is the
step size. (Id ,h 1 ) and (Id ,h 2 ) are the desired boundary points.
Near the left and right boundaries, the step size can be reduced to improve the
graphical accuracy. The later calculation example shows that due to the better linearity
of the boundary line, the step size can be increased appropriately at the rest to improve
the computational speed.
The algorithm is the same as before. Section 1 (active and reactive power section
of bus 5 load) and Section 2 (active and reactive power setpoint sections of VSC DC
rectifier station) are used as examples to complete the section depiction of the static
safety domain. The results are shown in Fig. 5.11. The red dots in the figure are the
current operating points.
After obtaining the required two-dimensional cross section of the static safety
domain, it is necessary to know the operating constraints corresponding to the
Bus 5 load reactive power

reactive power of VSC

References of

Bus 5 Load Active Power References of active power of VSC rectifier

(a) Section 1 (b) Section 2

Fig. 5.11 Two-dimensional cross-section of static safety domain depicted using point-by-point
method of boundary
236 5 DC Participation in Emergency Tidal Control

boundary line to determine the possible risk of overstepping the limits at the current
operating point (i.e., what operating constraints pose a greater threat to system
safety), to improve the corresponding equipment limits and increase the operating
margin, or to adjust the system control volume and translate the cross section until
the operating constraints corresponding to the boundary line are changed to avoid
difficult-to-improve operational constraints.

5.2.4 Static Safety Domain Inscription Method for AC-DC

Hybrid Grids Containing Controllable Series Capacitor
Converters

The analysis in this section will take the controlled series capacitor converter as an
example to visualize the static safety domain section cut and qualitatively determine
the coordinated control measures needed to operate the CSCC DC system in a way
that minimizes the reactive power demand on the AC system, reduces the probability
of phase change failure, reduces the stress on the valve voltage, reduces the control-
lable capacitor and investment cost of the commutation valve, improve the safety
margin of each operating constraint, etc. Since these requirements are often contra-
dictory, the scheduler or designer needs to find a "compromise" operating point for
coordinated control through qualitative judgment.
A simplified model schematic of the topology of a controlled series capacitor
converter station is shown in Fig. 5.12. Its topology is based on the conventional LCC
DC transmission topology with controlled capacitors connected in series between the
converter transformer and the valve bank. In Fig. 5.12, L is the phase commutation
reactance, Cf is the filter capacitor, U is the AC bus voltage, and Z is the impedance
of the AC system Davinan equivalent circuit.
The single-phase circuit structure of the controllable capacitor module C in
Fig. 5.12 is shown in Fig. 5.13. In Fig. 5.13, C is the capacitor, L is the inductor, and
VT1 and VT2 are thyristors. In practice, the module can be increased in capacity
by series connection. The controllable capacitance can be adjusted independently of
its equivalent impedance by control of the thyristor, which is used in this paper as a
dimensional variable in the static safety domain space.

Fig. 5.12 Controlled series capacitor converter station as simplified model schematic
5.2 Multi-indicator Static Security Domain for AC-DC … 237

Fig. 5.13 Single-phase

circuit structure of
controllable capacitor
module

The quasi-steady-state model of the CSCC DC system consists of the following

equations:

kU cos α Id
A+ √ + =0 (5.64)
2(K − 1)X L
2 2
kU cos(α + μ) Id
A cos(K μ) + B sin(K μ) + √ = (5.65)
2(K − 1)X L
2 2
1 2π Id X C kU K sin(α)
B= ( − U1 ) + √ (5.66)
2K X L 3 2(K 2 − 1)X L
kU K sin(α + μ)
−A sin(K μ) + B cos(K μ) − √
2(K 2 − 1)X L
(5.67)
1 2π Id X C
− ( − U2 ) = 0
2K X L 3

U1 + U2 = μI d X C (5.68)

√
3 2 cos α + cos(α + μ) 3μ
Ud = − kU − (1 − )(U1 − U2 ) (5.69)
π 2 4π

α + μ + γapp = π (5.70)

√ 2π
2kU sin(α + μ + γr eal ) + Id X C ( − γr eal ) − U2 = 0 (5.71)
3

Pd = Ud Id (5.72)

cos α + cos(α + μ)
cos ϕ + =0 (5.73)
2

XC 1 ω2
K = = 2 = 02 (5.74)
XL ω LC ω
238 5 DC Participation in Emergency Tidal Control

In the above equation, U is the average value of the AC bus line voltage, k is
the commutation transformer ratio, α is the trigger angle, μ is the phase change
angle, and X L and X C are the impedance values of the commutation reactance and
series capacitor, respectively. Id , Ud and Pd are the DC current, DC voltage and DC
delivered power, respectively, and U 1 and U 2 are the voltage variations (constant
when steady state is reached) on the capacitor in series with the exiting and putting-in
phase during phase commutation, respectively. γapp and γr eal are the apparent arc
quenching angle and the actual arc quenching angle, respectively. ϕ is the power factor
√
angle at the AC bus, K is the series capacitor compensation degree, ω0 = 1/ LC
is the resonant angular frequency, ω is the fundamental angular frequency, C and L
are the controllable capacitor and phase change inductor, respectively, and A and B
are intermediate variables.
The series capacitor makes the valve voltage peak increase and usually requires
that the maximum voltage withstood by the converter valve does not exceed 1.1 times
the maximum valve voltage in the LCC DC system and reduces the investment cost
of the converter valve as much as possible. Thus, the valve voltage peak must meet
the relevant limits, where UVmax is the maximum valve voltage peak allowed.

√ π π Id X C
2kU sin( + γapp ) + + U2 ≤ UV max (5.75)
3 3
√ π π Id X C
2kU sin( + γapp + μ) + − 2U1 ≤ UV max (5.76)
3 3
In this section, the method for inscribing static safety domain sections containing
CSCCs is described. The qualitative judgment method for coordinated control of the
cross section using CSCC is given later.

5.2.5 Methods for Inscribing the Decoupled Security Domain

of an AC-DC Hybrid Grid

When a multifed AC-DC hybrid grid is subject to large disturbances requiring emer-
gency power control of the DC subsystem, the proposed control strategy is often
the main factor limiting the fast response of the DC subsystem. For example, in a
multifed AC-DC hybrid grid, if the total power delivered by DC subsystem i and
DC subsystem j needs to be increased, then the current point-by-point method is a
nonlinear programming problem based on multiple constraints. This is often compu-
tationally intensive and takes a long time to compute, making it difficult to develop
the corresponding control strategy quickly.
Even for the double-fed AC-DC hybrid grid, the power of the DC subsystem Pdi
and Pd j are extracted as the two key control variables, and the static safety domain
sections of the DC power Pdi and Pd j in two-dimensional space are inscribed based
on the relevant theory in the previous section. The irregular geometry shown in the
5.2 Multi-indicator Static Security Domain for AC-DC … 239

black curve, when adjusting the size of Pdi and Pd j in the static safety domain section,
although the relative position of the current operating point and the boundary line of
the section can be visualized, it is still necessary to consider the coupling relationship
between them. ij Pobj = Pdi + Pd j = const If the coupling between Pdi and Pd j is not
taken into account in the process of adjusting the operating point A(Pdi,0 , Pd j,0 ), e.g.
if only Pd j or Pdi is adjusted according to the dotted line for fast power adjustment,
the operating points are migrated outside the static safety domain section, leaving
the system in an unsafe operating condition. If the sizes of Pdi and Pd j are adjusted in
concert so that the adjusted operating points lie on the line BC, the essence is again
transformed into solving a multi-constrained nonlinear programming problem.
A rectangle with sides parallel to each of the coordinate axes is found within the
nonregular geometry in Fig. 5.14. Solving the new run point based on the rectangle,
although the number of solutions is reduced (from the length of line BC to the length
of line DE) with some conservatism, the complex multiconstraint nonlinearized plan-
ning problem can be transformed into a four-constraint linear planning problem, as
shown in Eq. (5.77), thus increasing the computational speed and reducing the plan-
ning time. Additionally, decoupled control between two electrical quantities Pdi and
Pd j can be achieved within the rectangle.

obj Pobj = Pdi + Pd j

!
Pdi,min ≤ Pdi ≤ Pdi,max (5.77)
s.t.
Pd j,min ≤ Pd j ≤ Pd j,max

Based on the above analysis, the inscribed rectangle within the static safety domain
section with each side parallel to the coordinate axis is defined as the decoupled safety
domain. Thus, the decoupled safety domain is a subset of the static safety domain

Fig. 5.14 Schematic

diagram of run point
adjustment
240 5 DC Participation in Emergency Tidal Control

section. At the same time, since the static safety domain section is independent of
the operation state of the multifed AC-DC hybrid grid, when the network topology
and constraints of the system are determined, the static safety domain section is
uniquely determined. Then, the decoupling safety domain can be uniquely deter-
mined according to different solution objectives such as maximizing the number of
operating points in the decoupling safety domain.
When modulating the power of the DC subsystem, the focus will often be on
the transmission capacity of the DC subsystem, i.e., the maximum and minimum
values of the total power of the DC subsystem, and secondarily on the number of
safe operating points. Therefore, when determining the decoupling safety domain,
the decoupling safety domain should be such that it contains, first, the maximum and
minimum values of the total power, and second, the maximum number of operating
points within the decoupling safety domain, which can be characterized by the area
of the decoupling safety domain. Thus, the process for determining the decoupling
safety domain within the static safety domain section is shown in Fig. 5.15.
For Step I in Fig. 5.15, this paper uses the enumeration method to “search for the
inner rectangle containing point A with the largest area in the static safety domain

Fig. 5.15 Flowchart for

decoupled security domain
determination
5.2 Multi-indicator Static Security Domain for AC-DC … 241

Fig. 5.16 Schematic

diagram of coordinate points
during search

section 0 ”. Suppose the static safety domain section in two dimensions is shown
as the black closed curve in Fig. 5.16, and the total transmission power at the point
Pd1 +Pd2 A (Pd1,A , Pd2,A ) has the largest value, then there must be dPd2,A /dPd1,A ≤
0 at the point A. Therefore, the upper boundary of the decoupled safety domain
containing the point A with the largest area must start from the point A and extend
to the left in the direction parallel to. Pd1 .
When A1 (Pd1,A1 , Pd2,A1 ) is searched to the left, the search process is stopped
when the left boundary of the static safety domain section (Pd1 = Pd1,min ) or the
value of the upper boundary vertical coordinate is less than the vertical coordinate
of the point A(Pd1,A , Pd2,A ) is encountered. Each search calculates the area of the
current rectangle S0 and compares it with the current saved maximum area Smax ,
whose search flow diagram is shown in Fig. 5.17.

5.2.6 Evolutionary Characteristics and Impact Analysis

of the Static Safety Domain of Hybrid AC-DC Grids

Since the section inscription method has the characteristics of stronger intuitive
visibility and higher accuracy, and the section dimension selection is focused on in
the analysis, this section will study the evolution characteristics and influence of
various control measures on the static safety domain section and finally complete the
generalization of the influence law of DC transmission on the static safety domain
boundary evolution according to the analysis results.
First, the evolutionary characteristics and impact of different control methods of
LCC-type DC on the static safety domain are analyzed. The analysis of the impact
of DC transmission on the static safety domain in this section is all based on the left
zone transmission capacity (active output range of generators 1 and 2) (Fig. 5.18;
Tables 5.10 and 5.11).
242 5 DC Participation in Emergency Tidal Control

Fig. 5.17 Flowchart to search for decoupled security domain with point A and maximum area

Fig. 5.18 Two-area system with parallel DC and AC liaison lines

5.2 Multi-indicator Static Security Domain for AC-DC … 243

Table 5.10 LCC DC system

DC voltage 400 kV
operating parameters
DC current 500A
Firing angle 21.36°
Extinction angle 28.93°
Commutation reactance 0.067H
DC resistance 40.05
Filter capacitance capacity 125MVar
Ratio of converter transformer at both ends 230 kV:345 kV

Table 5.11 Upper and lower

Electrical parameters Upper limit Lower limit
limits of each operating
constraint for AC and LCC Node voltage 1.1 0.9
DC systems Balancer active power output 20 0
Balancer reactive power output 10 −10
Line phase angle difference 20° −20°
DC voltage of LCC 2.5 1.2
DC current of LCC 1.8 0.2
Firing angle 50° 5°
Extinction angle 35° 15°
Reactive power compensation 2 0
capacity

1. Impact of the introduction of LCC DC transmission on the static safety domain

The results of the static safety domain sections (for convenience of presentation, the
static safety domain abbreviation is used in the following of this chapter) of the active
output of generators 1 and 2 when the liaison line is pure AC, AC-DC parallel and
pure DC are shown in Fig. 5.19. The horizontal axis is the active output of generator
1 (PG1), and the vertical axis is the active output of generator 2 (PG2). The values
on both the horizontal and vertical axes in this section are the standardized values
(base capacity 100 MVA).
It can be seen that the active output of generators 1 and 2 has the least adjustable
range when the contact line is only LCC DC transmission and has the largest
adjustable range when AC-DC parallel is used and more power can be delivered
between the two regions. The static safety domain of the AC-DC hybrid grid is shifted
to the upper right, and the total area is slightly enlarged when LCC DC transmission
is introduced compared to the static safety domain of the AC-only system.
The screening method using the effective boundary of the two-dimensional section
can identify that the operational constraints governing the upper limit of the trans-
mission capacity of the contact line as a pure DC system are the upper limit of the
DC current and the lower limit of the ignition angle. The through-current capacity
of the DC line should be increased to improve the power transmission capacity. The
244 5 DC Participation in Emergency Tidal Control

Fig. 5.19 Comparison of

static safety domains for
pure AC systems and hybrid
AC-DC grids

operational constraints governing its lower limit are the lower limit of the DC current
and the lower limit of the reactive power compensation capacity.
The operating constraint governing the upper limit of the AC-DC parallel system
transmission capacity is the upper limit of the balancer reactive power output, while
the operating constraints governing the lower limit of the transmission capacity are
the upper limit of the phase angle difference of the line 11–10 (when PG1 is small)
and the upper limit of the balancer reactive power output (when PG2 is large). Thus,
the through-current capacity of the line 11–10 should be increased, multiple lines
should be built, the upper limit of the balancer reactive power output should be
increased, or the reactive power compensation capacity should be increased to give
generators 1 and 2 a greater adjustable range.

2. Impact of different control methods of LCC DC on the static safety domain

In a parallel AC-DC operation system, the basic control methods of the LCC DC
system under normal operation are compared. The rectifier station can be controlled
by constant current control or constant power control, and the inverter station can be
controlled by constant DC voltage control or constant arc quenching angle control,
for a total of four combinations.
From Fig. 5.20, it can be seen that the different basic control methods of LCC
DC have negligible effects on the static safety domain. The different basic control
methods are equivalent to giving approximately the same DC voltage and DC deliv-
ered power. If the remaining DC quantities have a large adjustable range and are
not involved in forming the effective boundary of the static safety domain, then their
effects on the static safety domain of the original AC system are essentially the same.

3. Impact of different modes of operation of LCC DC on the static safety domain

LCC DC transmission can be operated in unipolar or bipolar operation, bipolar

symmetrical or asymmetrical operation, earth return or metal return operation, power
5.2 Multi-indicator Static Security Domain for AC-DC … 245

(a) Static safety domain ratio (b) Local zoom-in

Fig. 5.20 Comparison of static safety domains with different basic control methods and their partial
enlargement

forward or reverse transmission, full voltage operation or reduced voltage operation,

etc. The operation of a DC transmission system is a combination of these. This
subsection will complete the analysis of different operation modes of LCC DC. VSC
DC is similar and will not be described in detail. Either earth return or metal return
operation acts on the line resistance and environment and will not be analyzed in this
section; power forward or reverse transmission can be found in Sect. 4 of this section
(Sect. 5.2.6) for section leveling analysis. When abnormal operating conditions occur
in system operation, the control mode of the DC system will switch. This explanation
is included in this section as a different operating mode in the analysis (only one of
the operating modes is analyzed for switching, the rest of the cases are analyzed
similarly and will not be analyzed in detail).
First, the impact of switching the operation mode of the system under abnormal
conditions on the static safety domain is analyzed. When the rectifier side AC voltage
is low or the inverter side AC voltage is too high, to avoid the insecurity of an ignition
angle that is too small, the rectifier converter station will take fixed ignition angle (5°)
control at this time, the inverter converter station will take fixed DC current (10%
smaller than the original value of the rectifier side) control, the operation mode is
recorded as operation mode 2, and the previous normal condition is operation mode
1. When the ignition angle is detected to drop to the minimum ignition angle limit,
the system changes from operation mode 1 to operation mode 2, and the static safety
domain is changed accordingly. A comparison of the static safety domains for the
two operation modes is shown in Fig. 5.21.
As seen in Fig. 5.21, the two upper boundaries overlap, the static safety domain of
the system under operation mode II is greatly reduced, and the operating constraint
corresponding to its lower boundary is converted to the lower limit of the arc extin-
guishing angle, thus ensuring that a sufficiently high DC voltage is maintained in
conjunction with the ignition angle. If the operation mode is other than normal and
fault mode, then the tidal equation is changed.
246 5 DC Participation in Emergency Tidal Control

Fig. 5.21 Comparison of

static security domains for
run mode 1 and run mode 2

The operating characteristics of DC transmission are closely related to the number

of valve bridges. When it is a single-pole double-bridge DC, the DC voltage, AC
current, and capacity are twice as high as when it is a single-pole single-bridge. A
similar equivalence can be made when it is a double-pole double-bridge.
Let the DC current limit be kept constant and the upper DC voltage limits for the
single and double bridges be set at 400 kV and 800 kV, respectively. The static safety
domains for the single-pole single bridge, single-pole double bridge and double-pole
double bridge (symmetrical operation) when only DC lines are used for the contact
line are shown in Fig. 5.22.

Fig. 5.22 Comparison of

static safety domains for DC
transmission under different
modes of operation
5.2 Multi-indicator Static Security Domain for AC-DC … 247

(a) Comparison of static security domains (b) Local zoom-in view

Fig. 5.23 Comparison of static safety domains of AC-DC parallel systems when delivering same
power during unipolar and bipolar operations

As the valve bridge increases, the DC voltage and the power delivered by the
system increase. However, the upper boundary does not move up parallel to the other
electrical quantities constrained by the AC-DC system but shows a certain arc.
The static safety domain of the system when the DC delivered power is kept
constant at 200 MW, and the system is operated with unipolar and bipolar symmetry
(each pole takes 100 MW with constant DC voltage). This is shown in Fig. 5.23.
It can be seen that there is little effect on the static safety domain when unipolar
DC or bipolar DC is used for transmitting the same power. This is because the
effective boundary of the static safety domain for parallel AC-DC transmission does
not include the DC current constraint, and the main factor affecting the difference
between the two originates from the magnitude of the AC side current in the converter
station.
If there is no restriction on the DC transported power, then the bipolar DC is able
to transmit more power than the unipolar DC in a system where the contact line is
pure DC because the upper boundary of its static safety domain corresponds to the
upper limit of the DC current.
Figure 5.24 shows a comparison of the static safety domain for unipolar and
bipolar operation when only LCC DC is used for the contact line.
For bipolar DC transmission, the two poles can be operated symmetrically or
asymmetrically. In an AC-DC parallel system, let the total DC transmission power
be the same for both at 200 MW. The static safety domain of the system and its local
enlargement when the two poles are operated symmetrically and asymmetrically
(150 and 50 MW of transmission power at the two poles, respectively) are shown in
Fig. 5.25.
Again, since the operational constraints corresponding to the effective boundary
of the static safety domain for AC-DC parallel transmission do not include DC current
constraints, the effect on the static safety domain for bipolar symmetric operation or
bipolar asymmetric operation is small.
248 5 DC Participation in Emergency Tidal Control

Fig. 5.24 Comparison of static safety domains for unipolar and bipolar operations

(a) Comparison of static security domains (b) Local zoom-in view

Fig. 5.25 Comparison of static safety domains for bipolar symmetric and bipolar asymmetric
operations

In a system with only LCC DC in the contact line, the results of the system static
safety domain carved for the DC system with full voltage operation and reduced
voltage operation (70% DC voltage) are shown in Fig. 5.26.
There is a significant reduction in the area of the static safety domain during
reduced voltage operation, and upon screening, the operating constraints corre-
sponding to the upper boundary of the static safety domain have switched to the upper
DC current limit and the lower bus3 voltage limit, and the operating constraints corre-
sponding to the lower boundary have switched to the upper arc-quenching angle limit
and the upper ignition angle limit. This is because to maintain a lower DC voltage,
the ignition angle and arc extinguishing angle need to operate at higher values while
the converter bus voltage is reduced. Therefore, to increase the operating range of the
5.2 Multi-indicator Static Security Domain for AC-DC … 249

Fig. 5.26 Comparison of static safety domains for full-voltage operation and reduced-voltage
operation

system during voltage reduction operation, the reactive power compensation capacity
of the converter stations at both ends needs to be increased.

4. Evolution of static safety domain boundaries when LCC DC is involved in

operational control

When the LCC DC is involved in the operational control of the AC-DC hybrid grid
(this section refers to adjusting the tidal distribution of the system), the adjustment
of the DC delivered power has an impact on the static safety domain boundary. In the
AC-DC parallel system, the static safety domain of the system is shown in the left
panel of Fig. 5.27 for the inverter station with constant arc quenching angle control
and the rectifier station with constant power control and the DC power given values
of 100, 150, 200, 250 and 300 MW.
As the given value of DC delivered power increases, the upper boundary of the
static safety domain gradually shifts upward, and the operating constraint corre-
sponding to this boundary is the upper limit of the reactive power output of the
balancer. When the DC delivered power is low, the operating constraint corresponding
to the lower boundary is the upper limit of the phase angle difference of line 11-10,
and increasing the DC delivered power will increase the burden on this line. When
the DC delivered power changes from 250 to 300 MW, the lower boundary shifts
significantly upward. At this time, as the operating point moves in space, the oper-
ating constraint corresponding to the lower boundary has switched to the upper limit
of DC current because when the arc quenching angle remains unchanged, the DC
current constraint is no longer satisfied by simply increasing the power delivery.
Therefore, when the LCC DC participates in tidal control, when its delivered power
increases to a certain value, the control quantities of the AC-DC system should be
250 5 DC Participation in Emergency Tidal Control

(a) Comparison of static security domains (b) Local zoom-in view

Fig. 5.27 Impact on system static safety domain boundary when LCC DC is involved in tidal
control and local zoom-in

coordinated and controlled in advance to ensure that the range of the static safety
domain is large enough.
Analysis of the evolutionary characteristics and effects of different control
methods of VSC-type DC on the static safety domain.
The LCC DC is replaced by a VSC light DC transmission line with a DC transmis-
sion power of 200 MW and DC voltage of 400 kV in parallel AC-DC. The rectifier
station is controlled by constant active and reactive power, and the inverter station is
controlled by constant DC voltage and constant reactive power.

1. Impact of the introduction of VSC DC transmission on the static safety domain

The static safety domain of the system when the liaison line is AC-only, AC-DC
parallel (with a certain DC transmitted power) and DC-only is shown in Fig. 5.28.
For comparison purposes, the black line in Fig. 5.27 shows the static safety domain
for the liaison line with only LCC DC. The green line and the lower boundary of the
black line in the figure largely overlap.
Similar to LCC DC, the contact line has the smallest controllable range of power
transmission between regions when only VSC DC transmission is used, while it has
the largest controllable range when AC DC parallel is used and more power can be
delivered between the two regions. After the introduction of VSC DC transmission,
the static safety domain of the AC DC hybrid grid is shifted to the upper right
compared to the static safety domain of the AC-only system, and the total area is
slightly enlarged. The static safety domains of the LCC DC and VSC DC static safety
domain lower boundaries basically overlap, and the upper boundaries are closer, but
the total area of the static safety domain of LCC DC is slightly larger.
By screening of the two-dimensional section boundary, it can be found that the
operational constraints governing the upper limit of the interregional power transfer
capability of the liaison line as a VSC DC system are the upper limit of DC current,
upper limit of DC voltage and upper limit of reactive power control of the rectifier
5.2 Multi-indicator Static Security Domain for AC-DC … 251

Fig. 5.28 Comparison of

static safety domains for
pure AC systems and hybrid
AC-DC grids

station. To improve the power transfer capability, the DC line through-current capa-
bility, device withstand voltage level, improved converter structure and upper limit
of the control parameters of the converter station should be increased. The operating
constraints that restrict the lower limit of power transfer capability are the lower limit
of DC current, lower limit of the phase shift angle and lower limit of active power
control of the converter station.
2. Evolution of static safety domain boundaries when the VSC DC is involved in
operational control
In an AC-DC parallel system where the converter station participates in the tidal
control of the AC-DC hybrid grid, the static safety domain and local enlargement
of the system are shown in Fig. 5.29 when the power setting values of the rectifier
station fixed power control are 100, 150, 200 and 250 MW. The rest of the control
quantities are kept constant.
As seen from Fig. 5.29, similar to LCC DC, the upper boundary of the static safety
domain corresponding to the upper limit of the reactive power output of the balancer
gradually shifts upward as the DC delivered power setting increases. Meanwhile, the
upper limit of the phase angle difference of lines 11-10 and the lower boundary of
the static safety domain corresponding to the upper limit of the reactive power output
of the balancer also gradually shift upward.
However, when the DC delivered power exceeds 263 MW, the static safety domain
section starts to shrink and disappears rapidly. It is tested that the operating point
no longer satisfies the upper limit of reactive power output of the balancer at this
time. This is geometrically interpreted such that the tangent plane where the high-
dimensional boundary plane corresponding to this operating constraint intersects
with the static safety domain of the three-dimensional section including the DC active
control volume is approximately parallel to the hyperplane where the DC power is
given as 263 MW. Therefore, when the DC participates in the operation control, to
avoid rapid reduction of the static safety domain section, the section inscription and
252 5 DC Participation in Emergency Tidal Control

(a) Comparison of static security domains (b) Local zoom-in view

Fig. 5.29 Impact on system static safety domain boundary when VSC DC is involved in tidal
control and local zoom-in

analysis should be carried out in advance, while the control quantities of the AC-DC
system should be coordinated. In particular, in scenarios where the volatility and
randomness of the load or intermittent energy are large, the DC power is changed to
the relevant disturbance quantity, and the evolutionary analysis of the static safety
domain section is completed in the same way to prepare the system for the worst
operation.
3. Impact of the simultaneous introduction of VSC DC and LCC DC on the static
safety domain
The parallel connection of both the VSC dc and LCC dc between buses 7 and 9
without the AC contact line and the parameters remains unchanged. Figure 5.30
shows a comparison of the static safety domain of the system after the simultaneous
introduction of both DC transmissions with pure AC, LCC DC only and VSC DC
only.
The parallel connection of both the VSC dc and LCC dc between buses 7 and
9 without the AC contact line and the parameters remains unchanged. Figure 5.30
shows a comparison of the static safety domain of the system after the simultaneous
introduction of both DC transmissions with pure AC, LCC DC only and VSC DC
only.
As seen from Fig. 5.30, the area of the static safety domain of the contact line
with parallel operation of LCC dc and VSC dc is larger than that of either dc alone
under the current dc system operation constraint setting. The static safety domain
of the hybrid AC-DC grid is shifted to the upper right compared to the static safety
domain of the pure AC system with a slight expansion of the total area after the simul-
taneous introduction of VSC and LCC dc transmission. The operating constraints
corresponding to the upper boundary of the static safety domain of the AC-DC hybrid
grid include the lower limit of the trigger angle and upper limit of the DC current
for the LCC DC and upper limit of the rectifier-side DC voltage, upper limit of the
DC current and lower limit of the rectifier-side reactive power control amount for
5.2 Multi-indicator Static Security Domain for AC-DC … 253

Fig. 5.30 Comparison of static safety domains for pure AC systems and hybrid AC-DC grids

the VSC DC. The corresponding operating constraints of its lower boundary include
the lower limits of DC current and reactive power compensation capacity for LCC
DC and the lower limits of the phase shift angle, DC current, and active power for
VSC DC. The static safety domain sections pass through the “intersection lines” of
the boundary surfaces corresponding to these constraints.
To analyze the “short link” in the tidal operation control of the LCC-VSC hybrid
DC system, one of the parallel VSC DC and LCC DC systems is separately involved
in the tidal operation control of the grid, at which time the static safety domain
boundary of the system evolves, as shown in Fig. 5.31.

(a) VSC DC fixed power operation (b) LCC DC fixed power operation
involved in operation control involved in operation control

Fig. 5.31 Evolution of boundary of static safety domain when one DC system is involved in tidal
operation control of system
254 5 DC Participation in Emergency Tidal Control

When VSC DC or LCC DC is involved in tidal regulation alone, as the DC

delivered power increases, VSC DC crosses the limit before LCC DC, and the static
safety domain disappears earlier. When the static safety domain disappears in the
left panel of Fig. 5.31, the ignition angle and DC current of LCC DC and DC current
of VSC DC will cross the limit at this time. Therefore, similar to the analysis above,
on the one hand, it is necessary to improve the line through-current capability and
maximize the design operating value of the DC voltage of the VSC DC. On the other
hand, since the VSC DC exists as a “short link” in the common current regulation of
the parallel DC system, it is necessary to expand the generator in conjunction with
the control volume of the LCC DC or AC system. On the other hand, since the VSC
DC is a “short link” in the common tidal regulation of the parallel DC system, the
control amount of the LCC DC or AC system is needed to expand the operating range
of generators 1 and 2 and the common tidal regulation of the parallel DC system to
give full play to the cocontrol potential of AC DC.
In addition, as seen from the right panel of Fig. 5.31, the static safety domain
gradually shrinks as the LCC DC delivered power increases and ceases to exist
when the LCC DC delivered power exceeds 323 MW. Throughout the process, the
operational constraints corresponding to the boundary of the static safety domain
also change, with the upper limit of the DC delivered power playing a role at first.
Constraints such as the LCC DC current, VSC DC voltage and VSC converter station
reactive power control value play a major role in the process after the static safety
domain section shrinks sharply and disappears. To further expand the operation range
of the LCC-VSC hybrid DC system in tidal common control, it is necessary to identify
the weak points that constrain the system operation at the current moment and make
improvements while moderately increasing the arc extinguishing angle to achieve
coordinated control.

5.3 Coordinated Control Objectives and Control Methods

When Multiple DC Systems Are Involved in the Rapid
Control of Tidal Currents

5.3.1 AC-DC Static Safety Domain Under Meter and Time

Characteristic DC Active Adjustment Method

In previous studies based on static safety domain control strategies, although the goal
is to adjust the system operating point to the inside of the static safety domain and
leave a certain safety margin, the time scale, control rate and response characteristics
of each control quantity are generally different from each other or even have a large
variability in the process of “pulling” to the inside of the safety domain. However,
the time scale, control rate, response characteristics, etc., of each control quantity
are generally different from each other and even have a large difference, so in the
process of “pulling” to the inside of the safety domain, the operating point may be
5.3 Coordinated Control Objectives and Control Methods … 255

shifted to the “outside” of the static safety domain. Thus, although the operating
point of the system corresponding to the given control quantity has satisfied the
safety constraints of the AC-DC hybrid grid, the actual system adjustment process
has already exceeded the limits, resulting in the “appearance” that the adjustment
command is valid.
In Fig. 5.32, when the dimension variable x1 needs to be increased from the
current value x A to x B , adjusting only x1 at this time will inevitably cause the run
point to cross the limit relative to the boundary 1, so the dimension variable x2 is
needed to participate in the coordinated control. If the response speed of x2 is slow,
i.e., then x1 is adjusted earlier than x2 , the operating point moves along curve 1, and
the operating point is shifted to the outside of the static safety domain during the
adjustment process. If the response speed of x2 is too fast, i.e., x1 cannot keep up
with the adjustment of x2 , then the operating point moves along curve 2, which is
also unsafe. Only the adjustment process of x1 matches the time characteristics of
x2 (as in curve 3) and can ensure the safety of the system in the whole process of
coordinated control. Only if the regulation process matches the time characteristics
of (as in curve 3) can the system be safe during the whole process of coordinated
control.
According to the definition and description of the control domain with time char-
acteristics given above, if the two-dimensional control domain exists at any moment
and remains continuous over the defined time range, then theoretically there are an
infinite number of high-dimensional “curves” from the “pull” operating point to the
target operating point in the control domain. There are infinitely many of them. In the
actual operation of the grid, there is always a corresponding adjustment strategy for
a specific operation control target, so this section gives a general method to calculate
the optimal adjustment curves for different targets in the control domain.
The time range defined by the control domain t0 ≤ t ≤ tn is equally spaced
over the time range containing the initial moment points m + 1. The first k(k =
0, 1, 2, . . . , m) moment point is noted as tk , and tm is the total time required for
the control quantity to complete control (tm < tn ). At the moment of tk , the system
electrical quantity, dimensional variables with known time response characteristics

Fig. 5.32 Schematic

diagram of run point
adjustment based on
different time response
devices
256 5 DC Participation in Emergency Tidal Control

and the control variables to be adjusted are Xk , Xak and Xck , respectively. xa0 and xam
are the values of the dimensional variables with known time response characteristics
at the initial and final moments, respectively. xc0 and xcm are the values of the control
variables to be adjusted at the initial and final moments, respectively. L is the set
objective function. Then, the solution of the optimal adjustment curve in the control
domain is transformed into the optimization problem of Eq. (5.78).

obj S = L(X0 , X1 , X2 , . . . , Xm )
⎧
⎪ O(Xk ) ≤ 0 k = 0, 1, 2, . . . m
⎪
⎪
⎪
⎨ h(Xk , tk ) = 0 k = 0, 1, 2, . . . m
a
s.t. (5.78)
⎪
⎪ Xa = xa0 , Xck = xc0 k = 0
k
⎪
⎪
⎩ k
Xa = xam , Xck = xcm k = m

For the safe operation of the AC-DC system, such as whether phase change failure
occurs in LCC-HVDC and voltage stability are closely related, and to guarantee a
small fluctuation of the voltage magnitude at each bus specified in the system when
the DC boosts the delivered power, especially the stable operation of the voltage
sensitive points, it is necessary to complete the solution of the optimal adjustment
curve in the control domain. The objective function L of Eq. (5.79) is given as
nb m
L = min (a(U j (tk ) − Uavr ( j)))b (5.79)
j=1 k=0

Assume there are n b such buses and m + 1 discrete points in the model. U j (tk )
is the voltage amplitude of the jth bus at moment tk , and Uavr ( j) is the average of
the voltage amplitude of the jth bus in the time range where the control quantity is
involved. a and b are coefficients greater than zero that characterize the importance
of that bus. To avoid sudden changes in the bus voltage magnitude during operation,
a and b can be adjusted to larger values. Equation (5.79) reflects the overall degree
of voltage fluctuation for all buses designated to participate in the examination with
high requirements for voltage magnitude fluctuations.
The control variable to be regulated in this part of the arithmetic analysis is selected
as the DC delivered power, with the objective of inscribing a two-dimensional control
domain. Since the LCC station has high requirements for reactive power support
during operation, it is more important to match the temporal characteristics of each
reactive power control device during the adjustment of the DC delivered active power.
The previous modified Kundur two-zone, four-machine calculation example is
used with an initial DC delivery power of 100 MW at parallel AC-DC. The system
is modeled in the static safety domain and linearly fitted to 82 effective boundary
surfaces. The operating parameters of the LCC DC system at the end of regulation
when the delivered power of the DC system needs to be increased from 100 to
440 MW are listed in Table 5.12.
5.3 Coordinated Control Objectives and Control Methods … 257

Table 5.12 Adjusted LCC

Electrical parameters Rectifier side Inverter side
DC system operating
parameters DC voltage 300 kV 289.8 kV
DC current 1466.6A 1466.6 A
Trigger/extinction angle 13.2° 16.8°
Commutation angle 21.6° 22.3°
Converter transformer ratio 1.03 1
Reactive power consumption 225.2Mvar 240.7Mvar

In this section, to simplify the analysis and graphical effect, the synchronous regu-
lator and STATCOM start to respond from moment 0 and use the first-order inertia
function to simplify the characterization of the time response of these two devices.
The STATCOM reaches its output target value at 30 ms, while the synchronous regu-
lator reaches its output target value at 750 ms. The mechanical capacitor throw and
converter ratio adjustment are both in seconds and are set to 1500 ms for the converter
transformer taps and 2500 ms for the mechanical capacitor throw of the rectifier and
inverter stations.
The following reactive coordination scenario is given for this adjustment
requirement.
Scenario: 150 Mvar is provided by the throw capacitors on both sides of the
rectifier and inverter, the rest of the reactive power is provided by the synchronous
regulator, and the STATCOM is inactive (Fig. 5.33).
When the objective function is selected as Eq. (5.79), the optimal adjustment
curve obtained from Eq. (5.79) (bus 5-11 is set as the voltage sensitive point in the
simulation) is shown in Fig. 5.34. The optimal adjustment curve can minimize the
voltage fluctuation of the sensitive bus of the system during the whole regulation
time and further ensure the safe operation of the AC-DC system.

Fig. 5.33 DC delivered

power control domain
Upper limit of DC transmission power
Inverter side active power/pu

Lower limit of DC transmission power

258 5 DC Participation in Emergency Tidal Control

Fig. 5.34 Scenario 3 DC

delivery power control
domain and optimal Upper limit of DC transmission power

Inverter side active power/pu

adjustment curve

optimal path

Adjust path 2
Lower limit of DC transmission power

In the current setting, the full process of DC transport power boosting and reactive
power control device cooperation takes at least 2500 ms. In the control domain of
Fig. 5.34, boosting the DC transport power does not result in operation constraint
crossing, and when the DC transport power is boosted along the control domain
boundary, the tide check and constraint screening reveals that at least one operation
constraint reaches the limit value at any moment. Thus, the calculation obtained in
this section of the control domain is reasonable.

5.3.2 Safety Correction Strategy for Hybrid AC-DC Grids

Based on Safety Distance Sensitivity

On the basis of the boundary surface fitting solution, the safety distance of the
operating boundary surface is solved. According to the set of tidal equations and
the linear fitting relationship, the linear fitting equation for each electrical quantity
in the system can be expressed as

X = C1 x1 + C2 x2 + · · · Cn xn + C0 (5.80)

where X is a vector of system-wide electrical quantities, x1 , x2 , . . . , xn is a variable

for each dimension in the safety domain, C1 , C2 , . . . , Cn is a vector of coefficients
for each dimension, and C0 is a vector of constants. Substituting the upper and lower
limits of the constraints for each electrical quantity into Eq. (5.80) translates into a
fitted boundary surface such that the boundary Hi can be expressed as

ci,1 x1 + ci,2 x2 + · · · ci,n xn + ci,0 − ei = 0 (5.81)

5.3 Coordinated Control Objectives and Control Methods … 259

where ci,0 , ci,1 , . . . , ci,n is the fitting coefficient of the corresponding boundary plane
Hi in the coefficient vector C0 , C1 , . . . , Cn , and ei is the corresponding constraint
limit. Then, the boundary distance di from the running point (x10 , x20 , . . . , xn0 ) to
the ith hyperplane in Euclidean space can be calculated by Eq. (5.82).

ci,1 x10 + ci,2 x20 + · · · ci,n xn0 + ci,0 − ei

di = (5.82)
2
ci,1 + ci,2
2
+ · · · ci,n
2

The distance formula of Eq. (5.82) cannot be used directly to characterize the
safety margin of the system because of the inconsistency of the magnitudes of the
electrical quantities. Therefore, the values of each electrical quantity need to be
normalized. Let the range of variation of the values of each dimensional variable be
transformed to 0-1, and let b j max and b j min be the upper and lower limits of the jth
dimensional variable. Then, the boundary surface (5.81) can be transformed into

ci,1 [(b1 max − b1 min )x1 + b1 min ] + ci,2 [(b2 max − b2 min )x2 + b2 min ]
+ · · · + ci,n [(bn max − bn min )xn + bn min ] + ci,0 − ei = 0 (5.83)
 
Notin ai, j = ci, j b j max − b j min , where j = 1, 2, . . . , n and ai,0 = ci,0 +
ci,1 b1 min + · · · ci,n bn min , the corrected boundary distance can be expressed as

ai,1 x10 + ai,2 x20 + · · · ai,n xn0 + ai,0 − ei

di = (5.84)
Ni 
 
Ni = ai,1 , ai,2 , . . . , ai,n (5.85)

where Ni is the normal vector of the hyperplane Hi .

The distance found by Eq. (5.84), di , has a positive or negative nature and char-
acterizes the relationship between the location of the operating point relative to the
hyperplane, i.e., whether the operating point is located inside or outside the security
domain, and is used to characterize the security of the system.
In the AC-DC hybrid grid, due to the strong coupling relationship between AC
and DC electrical quantities, adjusting the electrical quantities with higher sensitivity
regardless of the system security constraints is highly likely to make the important
AC current cross section cross the limit and cause the secondary phase change failure
of the DC system, exposing the system to greater security threats. Thus, the following
integrated degree concept is proposed from the perspective of the static safety domain
of the hybrid AC-DC grid.
From the above equation, the sensitivity of variable x j to vertical distance di from
the run point to hyperplane Hi is

∂di ai, j
Sdi = = (5.86)
∂x j Ni 
260 5 DC Participation in Emergency Tidal Control

Fig. 5.35 Schematic

diagram of safety distance
and its sensitivity in
two-dimensional projection

Referring to Eq. (5.86), the sensitivity of the safety distance of each dimensional
variable for any boundary can be obtained. This can clearly characterize the degree of
comprehensive influence of the change of each dimensional variable on the electrical
quantity characterized by the boundary surface. The traditional method is to correct
the safety of the boundary by this sensitivity, but this also means that it cannot take
care of the impact on other safety constraints. This often makes the safety correction
control less efficient and produces the “sawing” phenomenon. The safety distance
and its sensitivity model are further illustrated in the two-dimensional projection of
the safety domain shown in Fig. 5.35, where the horizontal coordinate is the DC
transmission current, and the vertical coordinate is the active output of generator 1.
When the run point is inside the security domain, di is positive, such as the
distances from run point B to boundary 3 and boundary 4 in Fig. 5.34 (d3B and d4B ).
When the run point is outside the security domain, di is negative, such as the distance
from run point A to boundary 1 (d1A ). Suppose that at this time, the run point is located
outside the domain at point A and is in an unsafe state. Now, to pull the run point
into the domain, a corresponding safety correction strategy needs to be developed.
The traditional sensitivity method is to move the run point along the normal vector
corresponding to boundary 4 according to the observed transgression limit, such as
boundary 4 transgression limit in Fig. 5.34, until the run point falls on that boundary
4. However, at this time, boundary 1 is still in the transgression state, so after that,
it is adjusted according to the normal vector of boundary 1. This is adjusted back
and forth between the 2 boundaries, which is less efficient. In this section, taking
advantage of the fact that the safety domain can be used for the overall evaluation of
−
→
the safety state of the system, the path AC can be obtained as a guiding vector for
the development of the safety correction strategy, which is defined as the minimum
value of the distance from the run point to each safety boundary.
In this section, the static safety domain theory is applied to the hybrid AC-DC
grid, and a safety correction strategy based on the sensitivity of the safety distance
is proposed. According to the expression of the safety domain boundary surface,
5.3 Coordinated Control Objectives and Control Methods … 261

it is known that the electrical quantities required to fit the boundary surface of the
safety domain are composed of the generator, DC channel and load. However, in
the safety correction, considering the different regulation capacity, regulation rate
and reliability of different electrical quantities, the corresponding weights shall be
assigned to different dimensional variables, i.e., the original static safety domain is
scaled and transformed.
The shortest Euclidean distance within 3 and after the transformed initial run
point Di solves the equation as
⎧
⎨ obj D = min  (x − x )
n
i j j0
j=1 (5.87)
⎩
s.t. g(X ) ≤ 0

Denote X n 1 as the point obtained after inverse telescoping transformation of the

adjustment endpoint
" " 3 to 1 , guiding the vector modulus to vc . Then,
found in
vc =  Vc  = " X n 1 − X 0 1 ". Based on the above analysis, the static sensitivity of
all adjustment means to the safe distance can be found. Let Sd j be the sensitivity of
the jth control quantity to the guiding vector VC , which is mathematically modeled
as follows.

VTc ej
Sd j = (5.88)
 Vc 

where ej is the unit vector [0, · · · , 0, 1, 0, · · · , 0]T , and 1 is the jth element of ej . Sd0 ,
the sensitivity vector, can be divided into three subvectors according to the positive
and negative values of sensitivity: Sd+ consists of all positive sensitivity subvectors,
and the corresponding adjustment quantity is recorded as A+ . Decreasing the size
of its value can improve the safety margin of the system.Sd− consists of all negative
sensitivity subvectors, and the corresponding adjustment quantity is recorded as
A− .A0 consists of sensitivity subvectors with all values of 0, and the corresponding
adjustment quantity is noted as A0 .
Based on the above study, the ability of each electrical quantity to enhance the
safety distance within the actual adjustment range can be found, and the formula
is shown in (5.89). The electrical quantities in the set A+ and A− are arranged in
descending order according to the magnitude of the absolute value of sensitivity.
When there is the same absolute value of sensitivity, then M j , which is obtained
according to Eq. (5.89), is arranged in descending order. Assume that the current
operating point of the system is outside the safety domain after the system is disturbed.
Let the set of adjustment quantities for the plus out force be A(r p
)
A(r )
p ∈ A
−
, and
(r )
 (r ) +

the set of adjustment quantities for the minus out force be Am Am ∈ A during the
rth round of adjustment A(r )
p . The adjustment increment for the jth control quantity in
+ −
x j and the adjustment decrement for the kth control quantity in A(r )
m is x k . The
adjustment quantities in the safety correction model are the decision variables, all of
which are injected in the static safety domain of the hybrid AC-DC grid, including
262 5 DC Participation in Emergency Tidal Control

the DC control quantity, generator active output, generator terminal voltage setting
and load.
⎧ +
⎨ Sd x j ∀x j ∈ A+
M j = 0  ∀x j ∈ A0 (5.89)
⎩  −
Sd x j ∀x j ∈ A−

It is important to note that the system must consider the external AC system
strength with a reasonable creep rate when performing DC power modulation to
ensure that the DC power conversion does not interfere with the connected external
AC grid. To minimize the total adjustment in the correction process, the following
linear programming model for the safe correction of the system is developed and
solved in this chapter using the simplex method in linprog under the MATLAB
compiled environment.
⎛ ⎞

min⎝ x +j + xk− ⎠
x j ∈x (∗)
p xk ∈∈(∗)
m
⎧ +
⎨ 0 < x j ≤ x j − x j
max 0
−
s.t. 0 < xk ≤ xk − xk
0 min (5.90)
⎩
Di (X + X) ≥ Dimin i = 1, 2, . . . , m

where X is the amount of electrical variation involved in this round of adjustment,

and the safety distance constraint ensures that the operating point is pulled into the
safety domain with a safety margin at the end of the adjustment.
To verify the effectiveness of the method in this section, the modified IEEE 39
node system is used as an example, and the simulation results are compared with the
traditional safety correction method.
Take the modified IEEE 39 node system as an example. Assume that in the initial
operating state, DC line 14-4 is taken out of operation due to a fault bipolar blocking,
at which point line 11-6 is overloaded with 110.26 MW, and the operating point is
outside the safety domain. The system is now required to be safety corrected, and
the operating margin dimin is taken to be 0.1.
The sensitivity of each electrical quantity to the guidance vector and its actual
regulation capability are obtained after calculation, as shown in Table 5.13. Since
the power deficit caused by the single-return DC blocking in this example can be
fully compensated by the two nonfaulty DCs as well as the parallel AC, the guidance
vector contains only the DC control quantities.
After the absolute value of the sensitivity size ranking, the DC current value and
arc extinguishing angle of the two nonfaulty DC lines are adjusted using DC lines
25-2 and 17-18 for power support and substituted into the safety correction model.
The model has a solution. The adjustment strategy obtained is as follows: the DC
current of DC line 25-2 is adjusted to 646 A, the arc extinguishing angle is adjusted to
17.2◦ , the DC current of DC line 17-18 is adjusted to 570 A, the arc quenching angle
5.3 Coordinated Control Objectives and Control Methods … 263

Table 5.13 Safety distance

Need to adjust Security distance Actual adjustment
sensitivity and actual
electrical quantities sensitivity ability
regulation capacity of each
control quantity under 25-2 dc current −0.5628 170 A
single-return DC blocking 25-2 extinction angle 0.5377 2.7°
17-18 dc current −0.5013 260 A
17-18 extinction 0.4751 3.1°
angle

is adjusted to 17.5°, and the voltage on the rectifier side of both lines is kept constant
at 500 kV, i.e., the transmission power of 25-2 is increased to 323 MW, and the
transmission power of 17–18 is increased to 285 MW. After this safety correction
strategy is implemented, the load factor of lines 11-6 is reduced from 1.2297 to
0.8917. The quantitative approach taken in this section can take into account the
static safety of the system and ensure a minimum amount of regulation.
From the above calculation example, it can be seen that when precise load cutting
is involved in the correction control, the static safety crossing limit can be eliminated
in a short time, and the regulation cost is low.

5.3.3 Optimal Scheduling Based on Safety-Corrected Control

in the Static Safety Domain of Hybrid AC-DC Grids

By calculating the safety distance sensitivity of each control quantity to the transmis-
sion limit of the critical section, the sensitive units and DCs with significant impact
on the critical section are identified. The safety subdomain projections of the high-
sensitivity control quantity on the optimized dispatch index, generation cost, voltage
deviation and critical section are inscribed with different system safety margins as
constraints to provide richer and more accurate operational information for grid
dispatch.
Under the safety constraint of heavy-duty AC line N − 1, two DC transmission
powers Pd1 and Pd2 are considered as examples, and Pd1 , Pd2 and the system opti-
mization scheduling target TS are used as the X, Y and Z axes, respectively, to form
a three-dimensional space. The scheduler can adjust the control target in real time
according to the system operation condition to complete the boundary inscription of
the safety domain three-dimensional projection.
To harmonize the safety and economy of hybrid AC-DC grid operation, the
minimum generation cost and minimum system voltage deviation are adopted as
the optimization objectives, taking into account the static safety constraint of heavy
load line N – 1.
The AC-DC hybrid system has a large capacity and high generation cost, so to
ensure the economic operation of the system in the medium and long term, this section
minimizes the generation cost FG as the objective function to efficiently distribute
264 5 DC Participation in Emergency Tidal Control

the active power of the generators in the system. Its expression is

NG
 
FG = αi PGi
2
+ βi PGi + γi (5.91)
i=1

where NG is the number of generators, PGi is the active output of the ith generator,
and αi , βi and γi are the respective generation cost factors of the generators.
After the DC participates in emergency power support, the DC cross-section
transmission power on the parallel AC-DC system tends to be higher, thus forming
a strong direct and weak AC grid structure, which is not conducive to the long-
term operation of the system. For multiple DC feeds into the receiving grid, the grid
voltage is affected significantly when allocating DC transmission active power due to
the constraints of the reactive power control strategy of the inverter station. Therefore,
it is necessary to optimally allocate the power of multiple DCs and generator reactive
power to improve the reactive power distribution of the multiple DC-fed grid, avoid
the risk of AC bus voltage overrun, and improve the voltage quality. Therefore, the
system voltage deviation Vde is used as the objective function, and its expression is
as follows:
N1 Ndc
 2  2
Vde = Ui − Uref,i + Udc, j − Uref,dc, j (5.92)
i=1 j=1

where Ui is the voltage amplitude of AC system node i, Uref,i is the reference voltage
of the AC system node, Udc, j is the DC voltage amplitude of converter station j, and
Uref,d c, j is the reference DC voltage of the converter station.
The generation cost and voltage deviation of the system are coordinated using the
ideal point method, which is expressed as
   
FC0 = w1 FG − FG,ide + w2 Vde − Vde,i de (5.93)

where FC0 is the co-optimization scheduling objective; w1 and w2 are the indicator
weights for the system generation cost and voltage deviation, respectively; and FG,ide
and Vde,i de are the ideal points for the single-objective optimization of the system
generation cost and voltage deviation, respectively.
Due to the inconsistency of the above two indicators, they must be normalized. Let
the variation range of the system generation cost and voltage deviation be transformed
into 0-1, and set the system generation cost and voltage deviation at FG,opt,max and
Vde,opt,max , respectively, in the process of objective optimization to find the maximum
value. Then, the modified co-optimal scheduling objective is
   
w1 FG − FG,opt w2 Vde − Vde,opt
FC = + (5.94)
FG,opt,max − FG,opt Vde,opt,max − Vde,opt
5.3 Coordinated Control Objectives and Control Methods … 265

Due to the large scale of the hybrid AC-DC grid, more sections need to be observed.
Since the shunt AC line tide has an important impact on the safe operation of the
hybrid AC-DC grid and faces the risk of bearing DC power shocks, to reduce the
observation pressure, this section focuses on the sensitive generating units that have
a greater impact on the shunt AC line tide or the changes in DC in the process of
optimal scheduling on the shunt AC line tide.
The sensitive unit or DC with a greater degree of influence on the parallel AC cross
section is selected according to the absolute value size of this safety distance sensi-
tivity. In the process of optimal scheduling of the AC-DC hybrid grid, the system oper-
ating point needs to be pulled inside the safety domain, and the scheduler can also set
a certain safety margin to further improve the safety of system operation according to
the actual grid operation requirements. That is, after optimal scheduling, the distance
from the operating point to the ith effective boundary should meet Di Di ≥ Dimin
(i = 1, 2, . . . m), where m is the number of effective boundary surfaces in the static
safety domain, and Dimin is the minimum safety margin to each effective boundary
surface after optimal scheduling. Dimin is the minimum safety distance margin to
each effective boundary surface that the system operating point  should satisfy. Let 
the operating point of the system after optimal scheduling be x1,d , x2,d , . . . , xn,d .
Then, according to Eq. (5.80), Di can be expressed as

ai,1 x1,d + ai,2 x2,d + · · · ai,n xn,d + ai,0 − ei

Di = (5.95)
Ni 

In this section, the relationship between the high-sensitivity control quantity and
the optimal dispatching target is obtained based on the three-dimensional projection
inscription of the N − 1 security domain in Sect. 5.2.3. In addition, considering the
important impact of the shunt AC line tide on the safe operation of the hybrid AC-DC
grid, which is exposed to the risk of DC power shock, the correspondence between
the high-sensitivity control quantity and the shunt AC tide is further inscribed during
the change in the high-sensitivity control quantity. The dispatchers can determine the
operation status of the grid in all aspects and create effective dispatching strategies
quickly and intuitively. The mathematical model for optimal scheduling of the system
taking into account the N − 1 constraint of the heavy load AC line is shown in
Eq. (5.96). The fmincon function is used in this chapter to solve this nonlinear
programming model in the MATLAB compiled environment.

min FC
f (X) = 0
obj
 ≤ 0
g(X)
f q Xq = 0
  (5.96)
gq X q ≤ 0
Di (X + X) ≥ Dimin
s.t.
xa = xa min + xa × k k = 0, 1, 2 . . .
xb = xb min + xb × k k = 0, 1, 2 . . .
266 5 DC Participation in Emergency Tidal Control

where xa and xb are the selected high-sensitivity units or nonfaulty DCs, xa min and
xb min are the initial values of the optimization calculation, xa and xb are the
steps of the optimization calculation, and X is the amount of change of electrical
quantities involved in the optimization scheduling process.
In this section, computational analysis is carried out with the modified IEEE 39-
node system and the simplified real grid with the same topology, rated transmission
power and long-term overload capacity of DC as in Sect. 5.3.2. The N − 1 fault
static safety domain of the counted and heavily loaded line is constructed to verify
the feasibility of the proposed optimal dispatching strategy.
When DC line 14-4 is bipolar blocked, the AC line active power transfer limit and
the AC-DC system operating constraints are the same as in Sect. 5.3.2. All converter
stations use the reactive power compensation method of throwing capacitors in groups
of 30 Mvar each, and the lines with initial load ratios exceeding 0.7 are listed as heavy
load lines, namely, 2-3, 10-11, 10-32, 16-19 and 22-35, for a total of 5. Considering
the actual operating conditions of the grid at this voltage level, transmission lines are
usually supplied using double-loop lines, so all heavy load lines N − 1 counted in
this section refer to single-loop line breaks. The generation cost coefficients of αi ,
βi and γi are 0.01, 0.3 and 0.2, respectively, where i = 1, 2…, 10.
First, the selection of sensitive units is carried out. The parallel AC section of a
large AC-DC hybrid grid is of great significance for interregional energy intercon-
nection, so based on the method of selecting sensitive units in the previous section,
the safety distance sensitivity value of each generator unit of the whole network
to line L 11−6 is analyzed. The detailed calculation results are shown in Table 5.14.
From the results in the table, it can be seen that G8 and G2 are the units with the
largest absolute value of safety distance sensitivity to the transmission limit of L 11−6 .
Therefore, the sensitive unit of L 11−6 is selected.
According to the inscription method of the three-dimensional projection of the
safety domain, the optimal objective function values in the current operation state
are searched in the range of the available output of the two sensitive motors, with
G2 and G8 as the X and Y axes, respectively, and to express the# optimal operation
interval more intuitively, the optimal scheduling index IOD = 1 FC is taken as the
Z axis, the minimum safety distance margin Dimin is taken as 0 and w1 = w2 = 1,
and the optimal scheduling safety subdomain considering the heavy load line N −
1, as shown in Fig. 5.36, is an inscribed projection in this 3D space. In contrast to
the optimal scheduling safety subdomain with Dimin as 0.1, the optimization process
is used to minimize the generation cost as the main optimization objective by taking

Table 5.14 Safety distance sensitivity of units in modified IEEE 39 node system
Unit number Sensitivity Unit number Sensitivity Unit number Sensitivity
G1 0.3493 G5 −0.1780 G9 −0.2445
G2 0.5415 G6 −0.1921 G10 0.2969
G3 -0.1572 G7 −0.1572
G4 -0.1801 G8 −0.3144
5.3 Coordinated Control Objectives and Control Methods … 267

w1 = 20 and w2 = 1 considering that the safety margin of 0.1 has effectively reduced
the voltage deviation of the system. At the same time, considering the abstract nature
of the co-optimization objective and the important influence of the weak AC cross
section on the safe and economic operation of the hybrid AC-DC grid, it is necessary
to monitor the generation cost, voltage deviation and parallel AC cross section of
the system in real time when optimizing the scheduling of the AC-DC system and
to avoid the area where the parallel AC cross section changes drastically. Therefore,
the three-dimensional projection comparison of two safety subdomains in different
spaces, as shown in Figs. 5.37 and 5.39, is given again with the transmission power
of the generation cost, voltage deviation and shunt AC cross section L 11−6 as the Z
axis.
Let the scheduling reference index be taken as IOD = 1.3, as shown in the green
shaded surface in Fig. 5.36. Then, the optimal scheduling subdomain above the
scheduling reference index is the optimal operation interval of the current system in
a saddle shape, and the optimal operation point is at IODmax
= 2.06 in the figure. At
this time PG2 = 585.2 MW, PG8 = 651.7 MW, and the reserve capacity left by G8
is smaller. The optimal operation interval is concentrated in the active output of G2
between 470 and 630 MW. When the active output of G2 is higher than 700 MW, the
optimal scheduling index of the system decreases significantly, and the active output
of G8 has a relatively small impact on the optimal scheduling index. Thus, adjusting
the operation point to the optimal operation interval can reasonably coordinate the
safety and economy of the system operation.
Because of the safety margin constraint, the higher the margin that is taken, the
narrower the adjustable range of the system control is for the optimal scheduling
subdomain. From Figs. 5.37 and 5.38, it can be seen that the overall system generation

Optimal operation point

Optimal schedule index

Optimal operation range

IOD=1.3

Fig. 5.36 Three-dimensional projection of sensitive motors with co-optimized scheduling objec-
tives
268 5 DC Participation in Emergency Tidal Control

Security margin is
0
Security margin is
0.1
Generation cost/dollars

Fig. 5.37 Three-dimensional projection of sensitive motor versus generation cost

cost with a safety margin of 0.1 at Dimin is higher than that with 0 at Dimin , but the
overall level of voltage deviation is relatively low, and the boundary pattern of each
safety domain is generally nonlinear. With a safety margin of 0.1, the system reduces
the adjustable range of electrical injection and increases the generation cost but also
improves the safety of system operation in all aspects, avoids the risk of critical safety
operation, and leaves a certain amount of spare capacity for the unit. Figure 5.37 also
allows real-time monitoring of the transmitted power at the parallel AC cross section
L 11−6 under different optimal dispatch measures. From Fig. 5.39, when the safety
margin is 0 and the operating point is in the optimal operating interval, the system
tends to make maximum use of the transmission capacity of the parallel AC section
L 11−6 to optimize the system operation. This leads to the load factor of the critical
section L 11−6 exceeding 0.95, which is on the verge of crossing the limit. Therefore,
if we want to improve the economy and safety of the system operation in the long
term, we can consider the reasonable expansion of the line L 11−6 .
After DC line 14-4 bipolar blocking, DC lines 2-25 and 17-18 become the only
two remaining DC transmission channels. Considering the important role of the DC
system in operation and control, the optimal objective function values in the current
operation state are searched in the range of the available power output of the two
DC transmission powers according to the inscription method of the safety domain
three-dimensional projection, with DC lines 2-25 and 17-18 as the X and Y axes,
respectively, and the optimal scheduling index as the Z axis. The safety subdomain
projection shown in Fig. 5.40 is drawn with the X and Y axes, and the optimal
scheduling index is the Z axis. The safety subdomain projection with different safety
margins is compared and analyzed with the generation cost, voltage deviation and
transmission power of the parallel AC section L 11−6 using the Z axis as a reference
to the calculation of the high-sensitivity unit.
5.3 Coordinated Control Objectives and Control Methods … 269

Security margin is
0

voltage deviation

Security margin is
0.1

Fig. 5.38 Three-dimensional projection of sensitive motor versus voltage deviation

11-6 AC section/MW

Security margin is 0.1

Optimal
operating range
Security margin is 0

Fig. 5.39 Comparison of 3D projection of sensitive motor and critical section

As shown in Fig. 5.40, the optimal scheduling subdomain above the scheduling
reference index IOD = 1.3 is the optimal operating interval of the current system, with
the difference that the optimal operating interval in the domain is more dispersed and
hump-shaped. The safety domain boundary also shows a strong hump because the
reactive power control strategy of the converter station leads to constant capacitor
throwing, which in turn makes the reactive power inject into the system change
continuously. Usually, in an AC-DC parallel system containing multiple DCs, after
DC blocking and to give priority to the safety of the system, the fast transfer band
effect of the nonfaulty DC is maximized to ensure the power balance at the sending
and receiving ends. Thus, the nonfaulty DC transfer power will usually operate at
270 5 DC Participation in Emergency Tidal Control

Fig. 5.40 Three-dimensional projection of DC and co-optimization scheduling objectives

full capacity or even over the quota, which creates a major hidden danger to the long-
term stable and economic operation of the system. From Fig. 5.40, it can be seen that
the current optimal operation interval of the system is concentrated in the following
areas: Pd (2 − 25) between 240 and 270 MW and Pd (17 − 18) between 240 and
260 MW, i.e., the area where the optimal operation point (IOD max
= 2.06) is located.
This verifies that the oversized DC transmission power cannot always guarantee the
long-term economic and safe operation of the system. When schedulers are facing
multiple optimal operating intervals, they should choose the one that occupies a
larger area, as this effectively avoids the risk caused by adjustment errors. If they
choose an optimal scheduling similar to operating interval 1 in Fig. 5.40, then there
is a high risk that it will not be adjusted to the established target value.
From the perspective of the safety domain, this section proposes an optimal
scheduling subdomain inscription method for hybrid AC-DC grids after emergency
control, considering that the system is in a critical safety state after emergency control.
This makes it difficult to meet the demand of long-term safety and economic opera-
tion of the system. In this way, not only can the accurate information of the current
system operation be obtained, but the dispatcher can also reasonably avoid the safety
risks in the process of optimal dispatching, optimize the system operation according
to local conditions, and enhance the safety and economy of the system operation.
5.3 Coordinated Control Objectives and Control Methods … 271

5.3.4 Preventive Correction Coordination Control Based

on Static Safety Domain for Hybrid AC-DC Grids

The process of preventive correction coordination control for hybrid AC-DC grids
based on the static safety domain is divided into 3 steps as follows.
(1) Expected accident set  into a preventive control subset p and the correction
control subset c The preventive control subset p the static safety of the system
guaranteed by preventive control and the corrective control subset c the static
safety of the system guaranteed by corrective control.
(2) Develop a preventive control measure x ap that relocates the initial operating
point of the system x a0 ∈/ ∩λi ∈ p (λi ) ∈ Rn relocated to x ap ∈ ∩λi ∈ p (λi ) ∈
n
R that is, the initial operating point x α0 to an operating point within the inter-
section of all the preconceived accident static safety domains of the preventive
control subset p x αp (2) Develop a preventive control measure that relocates the
initial operating point of the system to the operating point within the intersection
of the static safety domain of all anticipated incidents.
(3) For the correction control subset c for each anticipated incident in the λ j (λ j ∈
c ) select a corrective control measure x acj that relocates the operating point
after preventive control x ap ∈ / (λ j ) ∈ Rn to xacj ∈ (λ j ) ∈ Rn to form
the set of corrective control measures.x ac = {x ac1 , x ac2 , . . . , x acj }T . A
diagram of coordinated control is shown in Figs. 5.40 and 5.41.
After performing preventive controls, the system still needs to meet certain static
security margins in the normal network topology. For illustration purposes, this
section always treats the operational state in the normal network topology as a special
case in the preventive control set  p .
In economic terms, the solution of the coordinated control target operating point
requires consideration of the preventive control scheduling cost before the occurrence
of the expected accident, the corrective control scheduling expectation cost after the
occurrence of the expected accident, and the risk cost. Therefore, in this paper,

Fig. 5.41 Schematic diagram of preventive/corrective coordination control

272 5 DC Participation in Emergency Tidal Control

the following three main costs are considered: preventive control scheduling cost,
corrective control scheduling expected cost, and risk cost.
(1) Preventive control of movement costs

Preventive control pulls the current operating point of the system xα0 to the intersec-
tion of the static safety domain of the accident prevention set p . Thus, the scheduling
cost of preventive control can be expressed as

Cpd (xα0 , xαp ) = CT · |xαp − xα0 | (5.97)

where C = [c1 , c2 , . . . , cn ] is the unit adjustment cost of each dimension variable

in the static safety domain, xαp = [xαp 1
, xαp
2
, . . . , xαp
n
] is the operating point after
preventive control, and xα0 is the initial operating point.

(2) Corrective control scheduling expectation costs

The purpose of corrective control is to develop a corresponding corrective control

scheme for each envisioned incident xαc j λ j (λ j ∈ c ) in c and to pull the oper-
ating point xαp of the system after preventive control into the static safety domain
corresponding to the envisioned incident λ j , thus forming the set of corrective control
measures xαc . For example, for the envisioned incident λ j , the expected cost of
corrective control scheduling Ccd (xαp , xαc j ) is

Ccd (xαp , xαc j ) = P(λ j ) · CT · |xαc j − xαp | (5.98)

where P(λ j ) is the probability of occurrence of the expected event λ j , and xαc j =
[xαc
1
j , x αc j , · · · , x αc j ] is the operating point after corrective control is performed for
2 n

the expected event λ j .

This in turn leads to the expectation that the cost of corrective control scheduling
d
Cc(x,x) is

Ccd (xαp , xαc ) = Ccd (xαp , xαc j ) (5.99)

λ j ∈c

(3) Correction of the cost of controlling risk

In this paper, the risk of voltage overrun and tidal current overrun is quantified as a
cost indicator and quoted in the total cost of coordinated control to characterize the
degree of harm to the AC-DC hybrid grid from overrun before the corrective control
scheme takes effect. The mathematical model is

Cr ( f ) = (RV (λ j ) + R L (λ j )) (5.100)
λ j ∈c
5.3 Coordinated Control Objectives and Control Methods … 273

where RV (λ j ) is the voltage crossing risk due to the expected event λ j , and R L (λ j ) is
the tidal crossing risk due to expected event λ j . The two crossing risks are explained
separately below.
1) Risk of voltage overrun
Define the voltage crossing severity function G V (λ j , m) for PQ node m after the
occurrence of the envisaged accident λ j as shown in Eq. 5.101.
⎧ $ %ev
⎪
⎪ |2V − [V max
+ V min
)]| 1
⎪
⎪ ωm
m,λ j m m
−
⎪
⎪ 2[Vmmax − Vmmin ]
⎪
⎨
2
G V (λ j , m) = Vm,λ j > Vmmax || Vm,λ j < Vmmin (5.101)
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩ min
Vm ≤ Vm,λ j ≤ Vmmax

where Vm,λ j is the voltage value of node m when the expected accident λ j occurs;
Vmmax and Vmmin are the maximum and minimum values of node m voltage, respec-
tively; ωm is the node m importance factor; and ev is the voltage crossing severity
index factor. The higher its value, the easier to identify the severe expected accident.
The system voltage overrun severity indicator caused by the envisaged incident
λ j E V (λ j ) can be expressed as

E V (λ j ) = G v (λ j , m) (5.102)
m∈N B

where NB represents all PQ nodes in the AC-DC hybrid grid.

Therefore, the risk value Rv (λ j ) for voltage overruns in the event of a foreseen
incident λ j in the AC-DC hybrid grid is

RV (λ j ) = P(λ j )E V (λ j ) (5.103)

2) Risk of tide crossing limits

Define the tidal overload severity function G L (λ j , n) for AC line n at the time of the
expected accident λ j as
⎧  max eL
⎨ ω |Pn,λ j | − Pn
⎪
Pn,λ j > Pnmax
n
G L (λ j , n) = Pmax (n) (5.104)
⎪
⎩0 Pn,λ j < Pnmax

where Pn,λ j is the transmission power of line n when the expected accident λ j occurs,
Pnmax is the maximum value of transmission power of line n, ωn is the importance
factor of line n, and eL is the tidal transgression severity index factor. The larger its
value, the easier to identify a serious expected accident.
274 5 DC Participation in Emergency Tidal Control

Define the system tidal current overrun severity indicator E L (λ j ) caused by the
envisaged incident λ j as

E L (λ j ) = G L (λ j , n) (5.105)
n∈NL

where NL are all AC lines of the system.

Therefore, the system’s tidal current crossing risk value RL (λ j ) at the time of the
envisaged incident λ j is

RL (λ j ) = P(λ j )E L (λ j ) (5.106)

(4) Coordinated control of total costs

The objective of the optimal coordinated control scheme is to satisfy all the fore-
seen accident static safety requirements through preaccident preventive control and
postaccident corrective control while minimizing the total cost of coordinated control.
Thus, the mathematical optimization model of coordinated control can be expressed
as

min F(xα0 , xαp , xαc , f ) = C p (xα0 , xαp ) + Cc (xαp , xαc , f )

= C dp (xα0 , xαp ) + Ccd (xαp , xαc ) + Cr ( f )
s.t.
|xαcj − xαp | ≤ ki t
xαp ∈ ∩ (λi )
λi ∈ p

xαcj ∈ (λ j ) (∀λ j ∈ c )
Dλm ,n ≤ −Dssm
min
(∀λm ∈ , n ∈ Bup )
Dλm ,n ≥ min
Dssm (∀λm ∈ , n ∈ Bdown ) (5.107)

where Cp (xα0 , xαp ) is the cost of preventive control, which is equal to the cost of
preventive control Cpd (xα0 , xαp ) ; Cc (xαp , xαc , f ) is the corrective control expecta-
tion, which includes the corrective control scheduling expectation cost Ccd (xαp , xαc )
and the cost of risk Cr ( f ); and the rest of the variables have the same meaning as
before.
The first equation in the constraint condition represents the coupling between the
corrective control and preventive control, where ki is the regulation rate allowed
by the ith dimensional variable of the static safety domain; t is the dispatch
time allowed in the case of an accident in the corrective control; the second and
third equations represent the static safety constraints of the system expressed by the
static safety domain, where the third equation is the intersection of the static safety
domains of the set of prevented accidents p ; (λ j ) is the static safety domain
corresponding to the expected accident λ j ; the fourth and fifth equations represent
5.3 Coordinated Control Objectives and Control Methods … 275

the minimum static safety domain margin constraints; and dλm ,n is the operating point
to the expected accidentλm . The fourth and fifth equations represent the minimum
static safety margin constraint, which is the static safety distance from the opera-
tion point to the boundary surface, and Bn Bup and Bdown are the boundary surfaces
corresponding to the upper and lower limits of the constraint, respectively.
If the static safety requirements for all validly envisioned accidents are met by
preventive control, i.e., p = , then the cost of preventive control is the largest and
the expected cost of corrective control is the smallest. p To make the total cost of
coordinated control lower, we try to remove the most severe envisioned accident (the
envisioned accident with the smallest static safety margin after preventive control)
from p and merge it into c . At this point, the cost of preventive control will decrease
and the expected cost of corrective control will increase, and the rise and fall of the
total cost of coordinated control needs to be further observed.
Based on the above analysis, the outer layer optimization method of “constraint
relaxation” is designed in this paper. First, p contains all effective expected acci-
dents, i.e., p = , and calculates the total cost of coordination control. Then, the
accident with the smallest static safety margin in p is removed from p and incorpo-
rated into c , and the total cost of coordination control is reoptimized and calculated
in the inner layer to verify whether the total cost of coordinated control is verified
to be reduced. If it is no longer reduced, then the result of the previous step is the
optimal coordination control strategy; if it continues to be reduced, then the incident
with the smallest static safety margin is removed from and merged into p c , and
the inner optimization is continued. This process is repeated until the total cost of
coordinated control is no longer reduced.
The line and node importance factors ωm and ωk are both taken as 2500. The
severity index factors eV and e L for voltage overrun and tide overrun are both taken as
1.1. The PV active output and voltage adjustment rate are both taken as 2%/min. The
PQ node load cut volume and DC control variables are considered to be instantaneous.
The corrective control allowable dispatch time is taken as 5 min.
The optimal coordination control strategy is obtained after four iterations by
applying the coordination control optimization algorithm proposed in this section
to the effective set of expected accidents obtained by filtering based on the dominant
event principle . The optimal coordination control strategy is obtained after four
iterations. The changes in the preventive control cost, corrective control cost and total
coordination control cost during the iterations are shown in Fig. 5.42. The turnover
of the expected incidents in the preventive control subset p and corrective control
subset c are shown in Table 5.15.
During the 3rd iteration, the anticipated accident λ 27 dominated by the anticipated
accident λ 33 needs to be added to the iterative process of coordinated control in p
starting from the 3rd iteration since the anticipated accident λ 33 in p is removed.
The optimal coordinated control strategy, p contains 6 expected accidents
for p = { λ 0 , λ 8 , λ 12 , λ 18 , λ 27 , λ 36 } and c contains 3 expected accidents for
c = { λ 17 , λ 22 , λ 33 } . Since λ 17 is dominated by λ 18 , only 2 expected accidents
(λ 22 and λ 33 )are unsafe after preventive control under the optimal coordinated control
strategy, so corresponding corrective control measures need to be formulated for these
276 5 DC Participation in Emergency Tidal Control

Fig. 5.42 Changes in each Total cost of coordinated

cost during iterative process Total cost of prevention
Total Cost of Correction

Cost/million yuan Iterative process

Table 5.15 Variation in expected accidents in preventive control subset and corrective control
subset
Iterations Expected accidents in the expected Correction for expected accidents in the
control subset control subset
1 λ0 , λ8 , λ12 ,λ17 , λ18 , λ22 , λ33 , λ36 ∅
2 λ0 , λ8 , λ12 , λ17 , λ18 , λ33 , λ36 λ22
3 λ0 , λ8 , λ12 , λ17 , λ18 , λ27 , λ36 λ22 , λ33
4 λ0 , λ8 , λ12 , λ18 , λ27 , λ36 λ17 , λ22 , λ33
5 λ0 , λ8 , λ12 , λ27 , λ36 λ17 , λ18 , λ22 , λ33
6 λ0 , λ8 , λ27 , λ36 λ12 ,λ17 , λ18 , λ22 , λ33
7 λ0 , λ8 , λ36 λ12 ,λ17 , λ18 , λ22 , λ27 , λ33
8 λ0 , λ36 λ12 ,λ17 , λ18 , λ22 , λ27 , λ33 , λ36
9 λ0 λ8 , λ12 ,λ17 , λ18 , λ22 , λ27 , λ33 , λ36

2 expected accidents. The corresponding control strategy is executed when the acci-
dent λ 22 or λ 33 actually occurs so that it meets the static safety requirements. In the
case of the optimal coordination control strategy, the preventive control scheduling
strategy involves adjusting the control variables and their adjustment amounts, as
shown in Fig. 5.43. The electrical quantities of each conceivable accident in the
preventive control subset p = { λ 0 , λ 8 , λ 12 , λ 18 , λ 27 , λ 36 } that exist beyond the
limit and the changes in electrical quantities before and after the preventive control
are listed in Table 5.16. The control strategy corresponding to each conceivable
accident in the corrective control subset c is shown in Table 5.17.
To facilitate comparative analysis, Table 5.18 lists both the control cost compar-
ison results for pure preventive control and pure corrective control. In the optimal
coordinated control optimization solution process, the result of the first iteration is the
pure preventive control result. The result of the 9th iteration is the purely corrective
control result.
In this section, based on the static safety domain of an AC-DC hybrid grid, a
preventive-corrective coordinated control two-layer optimization algorithm with the
5.3 Coordinated Control Objectives and Control Methods … 277

preschedule preschedule

post-schedule post-schedule

Node 32 Node 35 Node 36 DC17-18 Node 34 DC17-18 DC17-18

active power active power active power constant dc Voltage constant voltage constant voltage

Fig. 5.43 Variation in amount of control involved in preventive control under optimal coordinated
control scheme

lowest coordinated control cost as the objective function is proposed. Compared

with existing studies, this paper eliminates invalid expected accidents based on the
dominant event principle to reduce the model size. Then, for a two-layer opti-
mization problem with dynamic constraints, an outer-layer optimization method
with relaxed constraints is proposed to effectively avoid the “dimensional disas-
ter” problem brought by the exhaustive method. In the inner layer optimization, the
static safety domain, the linear expression of the hyperplane, is used as the safety
constraint instead of the nonlinear tidal equation to improve the efficiency of the
optimization solution. Finally, the effectiveness and correctness of the optimization
method proposed in this section are verified by analyzing and calculating the modified
IEEE 39 AC-DC hybrid grid.

5.3.5 Fast Tidal Control Based on Decoupled Security

Domains

When the network topology and constraints of the system remain unchanged, the
decoupled security domain is uniquely determined so that the decoupled security
domain can be computed offline and applied online. The decoupled security domain
proposed in this chapter has the following two advantages.
(1) Realization of decoupling control between DC powers. Each DC subsystem
can ignore the mutual coupling relationship with other DC subsystems when
adjusting its own active power and can significantly increase the speed
of changing the system operating state when performing emergency power
modulation.
 278

Table 5.16 Changes in value of transgressive electric gas before and after preventive control
Anticipated accident Overlimit electric Upper limit/p.u Before scheduling/p.u Before scheduling dssd Post-scheduling/p.u Post-scheduling dssd
parameters
λ8 P6-11 upper limit 4.8 6.0498 0.0314 4.5556 −0.0050
λ12 P4-14 upper limit 5 6.0107 0.0253 4.5215 −0.0154
P10-13 upper limit 6 6.1740 0.0140 5.6720 −0.0323
P13-14 upper limit 6 6.3988 0.0308 5.8762 −0.0211
λ18 P6-11 upper limit 4.8 5.3728 0.0393 4.6877 −0.0718
P10-11 upper limit 6 6.5000 0.0385 5.9750 −0.0053
λ27 P23-24 upper limit 6 6.8558 0.0460 5.9502 −0.0053
λ36 P16-21 upper limit 6 6.7948 0.0432 5.8990 −0.0112
P21-22 upper limit 9 9.6051 0.0340 8.6959 −0.0422
5 DC Participation in Emergency Tidal Control
Table 5.17 Correction control scheme and change in values of transgressive electric gas before and after control
Anticipated Correction control scheme Changes of electrical quantities before and after correction control
accident Control Before Post-scheduling/p.u Overlimit Before Post-scheduling Postscheduling/p.u Post-scheduling
variable scheduling/p.u electric scheduling/p.u dssd dssd
quantity
λ22 Node 32 5.980 4.862 P6-11 upper 6.3986 0.1234 4.7668 −0.0050
active limit
power
output
Node 33 5.151 5.417
active
power
output
DC 1 0.930 0.865 P10-11 upper 6.1744 0.0140 4.6090 −0.2622
constant limit
voltage
DC 2 0.977 0.941
constant
voltage
5.3 Coordinated Control Objectives and Control Methods …

λ33 Node 32 5.980 7.877 P16-24 upper 6.3003 0.0165 2.7961 −0.1262
active limit
power
output
Node 35 6.050 5.975 P22-23 upper 65,000 0.0385 5.9750 −0.0050
active limit
power
output
(continued)
279
Table 5.17 (continued)
280

Anticipated Correction control scheme Changes of electrical quantities before and after correction control
accident Control Before Post-scheduling/p.u Overlimit Before Post-scheduling Postscheduling/p.u Post-scheduling
variable scheduling/p.u electric scheduling/p.u dssd dssd
quantity
Node 36 5.150 2.478 P23-24 upper 9.5847 0.2018 5.9555 −0.0052
active limit
power
output
5 DC Participation in Emergency Tidal Control
5.3 Coordinated Control Objectives and Control Methods … 281

Table 5.18 Comparison of control costs for different control options

Control Prevention and Expected cost of correction control Total
scheme control Expected cost of Risk costs/million cost/million
costs/million correction control yuan yuan
yuan scheduling/million
yuan
Single 43.2081 0 0 43.2081
prevention
control
Single 0 0.0047 48.8487 48.8534
correction
control
coordinated 7.5332 0.0023 11.1449 18.6804
control

(2) The nonlinear planning problem can be transformed into a linear planning
problem. If the power of multiple DC subsystems needs to be adjusted simul-
taneously so that their total power is a certain value, then the planning method
based on the decoupled safety domain can transform the system constraints into
linear constraints. This can improve the computational efficiency compared with
the point-by-point method.
For advantage (2), after an accident can be triggered by some signal, quickly match
the corresponding decoupling safety domain, combined with the corresponding
control objectives. Then, the corresponding optimal control strategy can be quickly
solved. For example, for a triple-fed AC-DC hybrid grid, when the rated compensa-
tion capacity of STATCOM is SV = 0.3 p.u. and the system is out of operation for
its own reasons at the rated operating state of HVDC3 , the total power of HVDC1
and HVDC2 needs to be urgently adjusted to Pd1 + Pd2 = 3.5 p.u., as shown by the
red line in 5.44, and solved to obtain as many power combination pairs as possible
(Fig. 5.44).
In the case of the point-by-point method planning solution, the DC power Pd1
and Pd2 of HVDC1 and HVDC2 are specified, and then the multifeeder system
nonlinear equation system model is solved to retain the solutions satisfying the system
inequality constraints and form the corresponding control policy table. If based on
the decoupled safety domain planning method, then the decoupled safety domain is
inscribed first, and then the planning problem with four linear constraints shown in
Eq. (5.108) is solved.

obj Pobj = Pdi + Pd j

!
Pdi,min ≤ Pdi ≤ Pdi,max (5.108)
s.t.
Pd j,min ≤ Pd j ≤ Pd j,max
282 5 DC Participation in Emergency Tidal Control

Fig. 5.44 Example of

decoupled security
domain-based power
planning

Table 5.19 Comparison of planning results

Planning method Procedure Essence Number of Solution Total time/s
solutions time/s
Security domain Describing Multiconstrained / 40.9542 40.9587
planning method decoupling nonlinear
based on security optimization
decoupling domain
Calculation Four Constraint 1405 0.0045
of feasible Linear
solution Programming
Point-by-point Calculation Multiconstrained 3765 82.3546 82.3546
planning method of feasible nonlinear
solution optimization

For the same computer, the planning time and results are shown in Table 5.19.
If the decoupled security domain is computed offline and applied online, then the
time to solve the runpoint based on the decoupled security domain is controlled
in milliseconds, which is basically negligible. Compared to Sect. 5.2.2 based on the
static security domain, it takes seconds to solve the corresponding control policy. This
fully demonstrates the advantages of the decoupled security domain-based planning
method in terms of computational volume and time.

5.4 Summary

The proposed improved maximum delivered power algorithm ensures the accuracy
of the calculation results while greatly improving the efficiency of the maximum
delivered power calculation. This can be applied to the design of online control
5.4 Summary 283

strategies. Compared with MISCR, MIOSCR can more accurately assess the actual
voltage support capability of the receiving system to the current HVDC and reflect
the operating margin of the system. The existing MISCR is actually a special case of
MIOSCR. The required regulator capacity under the same network frame increases
with the required power support demand. Under the same power support demand, the
required capacity of the regulator decreases as the system strength increases, so the
HVDC power demand should be reasonably arranged and the regulator reasonably
configured, taking into account the system strength.
The application of the safety domain approach to the hybrid AC-DC grid is
described in three aspects: the method of portraying the static safety domain of the
hybrid AC-DC grid, the evolution characteristics and influencing factors of the static
safety domain, and the safety control strategy of the hybrid AC-DC grid based on
the static safety domain. The coupling mechanisms between different types of DC
transmission systems, DC networks and AC systems in multiple dimensions such
as electrical quantities, control objectives and operational constraints were inves-
tigated, especially focusing on the interactive effects between active and reactive
power control under the fast tidal control of DC participation in AC systems. The
static safety domain of AC-DC hybrid power grids after the hybrid connection of
large LCC-type DC and VSC-type DC systems was studied. The evolution charac-
teristics of the multiobjective static safety domain when a DC system participates
in operation control were studied. The influence of the DC converter station control
mode and interstation interaction mode on the static stability characteristics of the
grid and multi-indicator static safety domain were studied. The coordinated control
objectives and control methods when multiple DC systems participate in fast tidal
control were studied.
Chapter 6
Adaptive Overload Protection
for Overhead Transmission Lines

6.1 Introduction

Relay protection is the first line of defense in defending the safety of the power
system, yet the unreasonable action of transmission line relays under overload has
been the direct cause of multiple chain trips. To defend against a major outage,
overloaded elements should not be removed or should at least be delayed, leaving
valuable time for relatively slow dispatch and stability control measures. However,
components operating under overload may damage their electrical performance and
affect normal operation. Relay protection is caught in a dilemma between tripping
or not tripping under overload, securing the system or securing the component.
In response to the above problems, [93] proposed line adaptive overload protec-
tion, according to the amount of line overload and overload bearing capacity, adaptive
adjustment of the protection action time. On this basis, this chapter further specifies
the action guidelines for relay protection under line overload: the line is overloaded
rather than faulty, and the protection does not trip if it does not cause damage to the
line performance and life. When the overload causes damage to the line performance
and life, the protection trips. By dynamically evaluating the safety of the line under
overload, the mode and time of action of the relay protection for overloaded lines is
determined: tripping, nontripping or delayed tripping, thus protecting system safety
and component safety.
As the current transmission system mostly uses overhead lines, the cable rate is
low, and overloads are not allowed during normal operation due to the high oper-
ating temperature requirements of cables [145]. Therefore, the line adaptive overload
protection proposed in this chapter focuses on overhead transmission lines to carry
out research.
In response to the problems with line overload protection in the introductory
chapter, three areas are examined in this chapter:
1) In response to the current line emergency current-carrying capacity setting being
conservative and single, the next step should be to fully exploit the transmission
line’s ability to bear overloads in the short term to ensure component safety
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 285
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2_6
286 6 Adaptive Overload Protection for Overhead Transmission Lines

under the premise of buying time for dispatch and emergency control as much
as possible. The emergency current-carrying capacity of the line is analyzed in
Sect. 6.2 of this chapter.
2) To address the problem of missing or slow updating of monitoring data along
the whole line (mainly conductor temperature), the substation station monitoring
data and monitoring data along the line are used to supplement each other, and
the real-time prediction data of the maximum temperature difference along the
line are used as a backup plan in case of missing data.
When the line is operated at overload, the distributed capacitance current along the
line has a smaller magnitude compared to the overload current, so it can be assumed
that the current is equal along the line, i.e., the Joule heat absorption is equal along the
line. The difference in conductor temperature along the line depends mainly on the
difference in meteorological factors such as wind speed and sunlight intensity. There-
fore, the difference between the temperature along the conductor and the measured
temperature at the substation station can be predicted from the meteorological data.
During normal communications, The difference between the maximum temperature
along the entire length of the conductor and the conductor temperature measured
inside the substation is called the maximum temperature difference along the line.
The maximum temperature difference along the line measured in real time is used
as a sample to establish a mathematical model reflecting its dynamic relationship.
When there is a communication problem that makes it impossible to obtain the moni-
toring data along the line, the maximum temperature of the conductor is predicted
by the historical data of the maximum temperature difference along the line and the
mathematical model. For the problem of slow updating of monitoring data along
the whole line, the current maximum conductor temperature can be projected based
on the maximum temperature difference along the line and the real-time measured
temperature at the substation.
(3) The problem of adaptive overload protection action time is the prediction of
conductor temperature, and the difficulty is to handle the parameter changes.
In this chapter, based on the calculation of the transient temperature rise, the
echo state network method is used to solve the problem of predicting the mete-
orological parameters and the maximum temperature difference along the line.
Cooperation with conventional backup protection is further considered to design
the adaptive accidental overload protection implementation scheme.

6.2 Line Emergency Current-Carrying Capacity Analysis

The emergency current-carrying capacity of the line is mainly limited by the thermal
stability of the conductor. Too high a temperature may lead to a reduction in the
mechanical strength of the conductor, an increase in line arc sag, a reduction in safety
spacing, and damage to the life and safety of the fixtures and various connectors.
Taking a 500-kV typical four split steel core aluminum strand as an example, using
6.2 Line Emergency Current-Carrying Capacity Analysis 287

the conductor mechanical strength, conductor arc sag, gold tool heat and grip, three
factors are analyzed.

6.2.1 Mechanical Strength

The slow annealing process of the conductor after high-temperature operation leads
to a loss of mechanical strength and accelerated aging of the conductor. In response to
this problem, a number of studies and tests have been carried out at home and abroad.
Research shows that the line runs below 200 °C and does not affect the mechanical
strength of the steel core in the wire, while for the outer shell of the aluminum wire,
its mechanical strength at approximately 95 °C begins to decrease [112]. China’s
Shanghai Cable Research Institute tests show that the steel-core aluminum stranded
wire (model LGJ-400/35) at 100 and 120 °C under an operation of 250 h has aluminum
line mechanical strength losses of 3.3% and 7.9%, respectively, while the steel core
at 120 °C under the operation of 1000 h shows no mechanical strength loss [146].
Another test shows that even if the strength of the steel-core aluminum strand is
slightly reduced when running at 90 °C, the pull-off force is restored after the high
temperature and will not cause practical problems.
The mechanical strength loss is the result of the combined effect of temperature
and time. The mechanical strength loss of the aluminum wire can be calculated from
the following equation [148]:

RSA % = (134 − 0.24TC )t − 0.63d (0.001TC −0.095)

1.6
(6.1)

where RS A % indicates the percentage of mechanical strength remaining in the

aluminum conductor, T C is the conductor temperature, d is the diameter of the single
aluminum strand (taken as 3.22 mm), and t is the duration of the conductor temper-
ature greater than 95 °C (in h), which is calculated as 100 when 135–0.24T C is less
than 100. The mechanical strength loss of the aluminum conductor was evaluated for
short runs at different temperatures, and the results are shown in Fig. 6.1 (converting
the time units to min).
As seen from Fig. 6.1, the conductor has almost zero damage to mechanical
strength within 60 min of operation at 120 °C, 60 min of operation at 150 °C causes
a 2% loss, and 60 min of operation at 200 °C causes a loss of up to 14.1% in
mechanical strength and should be removed at a rapid rate. Therefore, considering
mechanical strength alone, the conductor can be allowed to run at high temperature
for short periods (e.g., 60 min) when the line temperature is lower than 120 °C.
288 6 Adaptive Overload Protection for Overhead Transmission Lines

Fig. 6.1 Loss of mechanical strength of aluminum conductors during short-term high-temperature
operation

6.2.2 Arc Drape

With an increase in arc sag, the conductor-to-ground and crossover distances may
not meet the safety margin, increasing the probability of accidents such as tree flash
and discharge to ground. This endangers the safe operation of the line. Arc sag is
mainly caused by wire elongation. Wire in normal operation produces creep elon-
gation under the long-term role of stress (belonging to the elastic elongation); at the
same time, when the stress exceeds the wire elastic limit, there is irreversible plastic
elongation [149]. At high temperature, metal wire will also elongate (also known
as high-temperature creep), where the same temperature aluminum wire elongation
length is twice as long as the steel wire [150]. Wires are operated at high temperatures
to avoid plastic elongation and minimize creep elongation.
At a constant temperature, the creep elongation of the steel core aluminum strand
is proportional to the stress and duration. When the wire temperature is above 75 °C,
the unit creep elongation can be calculated according to the following equation [153]:

εC = 0.24(%RS)TC t 0.16 (6.2)

where εC is the unit creep elongation (in mm/km), %RS is the mechanical strength
residual rate (%), and t is the running time (in h). The creep elongation of the wire is
shown in Table 6.1 for 50% of the mechanical strength residual rate, run at different
temperatures for 1 h. The stall distance is taken as 400 m [146] between stalls.
In short-term high-temperature operation (within 1 h), creep elongation is small
and has little effect on the arc sag. Another test proved that at a 19% mechanical

Table 6.1 Creep elongation of conductors during 1-h high-temperature operation

Operation temperature 100 °C 120 °C 150 °C 200 °C
Creep elongation(m) 0.48 0.58 0.72 0.96
6.2 Line Emergency Current-Carrying Capacity Analysis 289

strength residual rate, the line short-circuit current reached 10 kA and lasted 1 s, and
the wire arc sag was almost unchanged [155].
Operational examples show that for a steel core aluminum strand (model LGJQ-
400) operating at 1200 A current for 3 min, the temperature rises to 127 °C, resulting
in a nonrecoverable plastic elongation of approximately 0.5 mm/m, causing an
increase in arc sag, endangering operational safety and causing a serious impact on
life [151]. Another study showed that steel-core aluminum stranded wire operating
at 120 °C does not produce plastic elongation of the wire stress [152]. Therefore, the
short-term operating temperature should not exceed 120 °C.

6.2.3 Fittings and Various Types of Connectors

Overhead transmission lines of the wire through the connection tube connection
and by the endurance wire clamps and other gold and insulator strings connected
to the endurance tower, gold and all kinds of connection head damage are difficult
to troubleshoot, and the maintenance costs are very high. The test shows that the
temperature of the metal tool is generally lower than the running temperature of the
wire at only 60–80%. Tests conducted in the Soviet Union showed that to prevent
oxidation damage, long-term operation of the gold and connecting head temperature
should not exceed 70 °C [156]. In addition to the heat resistance of the tooling, the
tooling grip (conductor, connecting tube, and composite crimping strength of the
tension-resistant clamps connected together) must also be considered. After the test,
the wire was continuously run for 48 h at 120 °C, and the grip force of the gold
tool was still more than 95% of the calculated breaking force to meet the operational
requirements [156]. The study shows that the gold and all kinds of connectors are
mainly affected by the long-term cumulative effects of high temperature. In the short-
time high temperature operation, the impact on its life and various performance is
small However, to ensure its performance, according to the temperature of the metal
tool for the wire running temperature of 70% of the calculation, the wire temperature
should not exceed 100 °C for an extended period of time.

6.2.4 Summary of Emergency Current-Carrying Capacity

The above experimental and calculated data are combined and discussed in three
categories.
(1) The transmission overhead line will not cause damage to the safety and perfor-
mance of the conductor when operated for a long period of time below the
maximum allowable temperature (70 or 80 °C) specified by the power grid.
(2) When the line temperature is higher than 120 °C, it can cause permanent damage
to the transmission overhead line. This situation should be avoided at all costs.
290 6 Adaptive Overload Protection for Overhead Transmission Lines

(3) Transmission overhead lines are higher than the maximum allowable temper-
ature for long-term operation and below 120 °C for short-term operation. The
mechanical strength of the conductor, arc sag and loss of fixtures can be ignored.
The damage caused by the overload operation of the conductor is the result of
the combined effect of temperature and time and should be calculated according
to the cumulative effect of heat.
Based on this analysis, adaptive overload protection thermal ratings are developed
for overhead transmission lines.

6.3 Line Adaptive Overload Protection Action Time

For system safety reasons, the emergency current-carrying capacity of the line was
studied in the previous section to exploit its potential current-carrying capacity as
much as possible under the premise of line safety. In the next step, the line temperature
is predicted, and the protection action time is adaptively adjusted for the thermal
inertia of the line temperature rise and the random changes in parameters to buy
as much time as possible for the dispatch and emergency control system to take
measures to eliminate the overload threat under the premise of component safety.
The determination of the adaptive overload protection action time is essentially
conductor temperature prediction, which, according to the previous analysis, is
divided into two parts: real-time temperature prediction based on conductor temper-
ature and ambient temperature monitoring data in the substation, especially for tran-
sient temperature rise processes under changing current or meteorological condi-
tions; and modeling and predicting the maximum temperature difference along the
line based on long-term online monitoring temperature data.

6.3.1 Line Temperature Calculation and Action Time

Analysis

1. Calculation of transient temperature rise of the line

The overhead line temperature is the combined result of the conductor’s own heat
production, heat absorption from and heat dissipation to the outside world. Among
them, Joule heat absorption PJ and insolation heat absorption PS are the main sources
of heat; convection heat dissipation PC and radiation heat dissipation PR are the main
parts of heat dissipation. The heat balance equation of the conductor is [158]

dTc
mCp = PJ + PS − PC − PR (6.3)
dt
6.3 Line Adaptive Overload Protection Action Time 291

where m is the wire mass (which can also be expressed by the density), and C p is
the specific heat capacity of the wire. When the total heat absorption and total heat
dissipation are equal, the wire is in thermal equilibrium, and the temperature remains
unchanged. When the heat absorption is greater than the heat dissipation, the line
temperature rises continuously until it reaches the heat absorption and heat dissipation
equilibrium. The temperature rise process is called the transient temperature rise.
To improve the efficiency of the solution, the transient heat balance equation
is usually approximated as a first-order differential equation with constant coeffi-
cients, and the general solution can be obtained as the approximate expression for
the transient temperature rise of the wire.

A A B(t−t0 )
Tc (t) = Ta (t) − + [Tc (t0 ) − Ta (t)+ ]e mCP (6.4)
B B
where t is the current moment, t 0 is the moment of initial calculation of the tran-
sient temperature rise, T c (t 0 ) is the initial temperature of the conductor, and T a (t)
is the ambient temperature at time t. Parameter A contains the Joule heat absorp-
tion temperature uncorrelated component and the insolation heat absorption compo-
nent; parameter B contains the Joule heat absorption temperature correlation compo-
nent, the convection heat dissipation component and the radiation heat dissipation
component. Each heat can be calculated according to the IEEE and CIGRE formulas
[159–160]. See Appendix A for details.
From Eq. (6.4), the calculated value of the maximum wire temperature T cmax.CAL
when the inverse of the wire temperature is 0 is

A
Tcmax.CAL = Ta (t) − (6.5)
B
When T cmax.CAL is greater than the overload protection temperature constant T c.TH ,
the time t TH for the temperature to rise from T c (t) to T c.TH can be further calculated
from Eq. (6.5). This gives

mCP Tc.TH − Ta (t)+A B
tTH = In  (6.6)
B Tc (t) − Ta (t)+A B

2. Temperature prediction and action time analysis considering parameter variations

The transient temperature rise considered in the previous section occurs when param-
eters A and B are constant. However, in practice, parameters A and B are time-
varying and dominated by changes in Joule absorption (current) and convective heat
dissipation (wind speed and wind angle).
Since the current is a known quantity accessible to the adaptive overload protec-
tion, if the current changes during the transient temperature rise, then the parameter
A’ is updated with the starting moment of the transient temperature rise calculation
t 1 and the initial temperature T c (t 1 ) to obtain the new temperature rise trajectory, as
shown in the following equation:
292 6 Adaptive Overload Protection for Overhead Transmission Lines
 
A A B1mC
(t−t1 )
Tc (t) = Ta (t1 ) − + Tc (t1 ) − Ta (t1 )+ e P (6.7)
B B

This is further analyzed by the following 2 arithmetic examples.

Example 1: The current rises from 400 to 1000 A at the 20-min moment and
continues to fall to 600 A at the 40-min moment. Take the wind speed data of a
wind farm in Jiangsu on December 1, 2013, for calculation. The wind angle is taken
uniformly between 0° to 90° and changes every 20 min, the ambient temperature T a
= 38 °C, and the light intensity Qs = 1000 W/m2. For Eq. (6.7) according to the
current parameters to calculate the wire temperature T c after 10 min (t + 10 min),
the results are shown in Fig. 6.2.
Example 2: The current rises uniformly at a temperature rise rate of 0.06 °C/min
at the bottom 20-min moment and finally stabilizes at 1000 °C. Meteorological
conditions are the same as above. The wire temperature T c (t + 10 min) after 10 min
is calculated from Eq. (6.7) based on the current parameters, and the result is shown
in Fig. 6.3.
From Figs. 6.2 and 6.3, it can be seen that
(1) When there is a sudden increase in current in Example 1, the fluctuation of
meteorological conditions such as wind speed has less effect on the temper-
ature prediction during the transient temperature rise of the wire to the new

Fig. 6.2 Calculation of wire temperature for Example 1

6.3 Line Adaptive Overload Protection Action Time 293

Fig. 6.3 Calculation of wire temperature for Example 2

temperature equilibrium point, and the temperature rise time prediction is more
accurate.
(2) The error and delay in the temperature prediction after the slow increase in
current in Example 2 also affect the prediction of the temperature rise time.
(3) When the current is reduced in Example 1, the wire temperature drops and runs
at 70 °C for a short period of time. According to the previous analysis, the impact
on the performance and life of the transmission overhead line is very low and
does not pose a security threat. However, according to the previous overload
protection action implementation plan, when the conductor temperature reached
70 °C, the line was removed, and full play was not given to the ability of the line
to operate under overload. For this reason, according to the previous analysis,
within a certain temperature range, the cumulative effect of temperature time
can be used as the basis for action.
(4) If the cumulative effect of temperature time is used as the basis for action, then
to determine the adaptive protection action time, the conductor temperature is
predicted. At this stage, the wire temperature is near the new equilibrium point,
wind speed and wind angle caused by the change in convective heat dissipation
on its temperature prediction. If still calculated according to formula (6.7) after
10 min of wire temperature T c (t + 10 min), its error is large. Figures 6.2 and
6.3 show the calculated value of the temperature after 140 min.
294 6 Adaptive Overload Protection for Overhead Transmission Lines

The temperature prediction error is caused by the time-varying parameters, domi-

nated by the convective heat dissipation component in parameter B and the Joule heat
absorption component in A. To reduce the error, the conductor temperature should be
predicted by the real-time prediction of A(t), B(t) instead of the constant B obtained
at the current moment.

6.3.2 Prediction Based on the Echo State Network Method

From the previous analysis, it is clear that the errors in the real-time prediction
of line temperatures in substations are mainly caused by time-varying current and
meteorological conditions. Another issue raised in Sect. 6.2.2 is the difficulty in
securing the line in emergency situations due to the slow update rate and possible
missing data of the transmission line online monitoring system. Since the currents
along the line are generally consistent, there are no differences in Joule absorption
along the line, and the differences in conductor temperatures monitored by the line
are mainly due to different meteorological conditions.
To accurately predict line temperature variations and determine reasonable
adaptive overload protection action times, a mathematical model reflecting the
dynamic relationship between time-varying current and meteorological parameters
and temperature should be established and predicted based on this model. This
problem belongs to the category of nonlinear time-series prediction. This chapter
will use the Echo State Network (ESN) to build a dynamic model and predict
the maximum temperature difference along the line as well as the time-varying
parameters in real time. This method overcomes the problems of complex training,
large amounts of computation and difficulty in determining the network structure of
the traditional neural network method, has high training efficiency and prediction
accuracy, and has greater online prediction advantages [161–165].
1. Mathematical model of the rebound state network
The mathematical model of the echo state network structure consists of three layers:
input, output and cryptosphere (geology). In hidden layer A, a large number of
sparsely connected neurons form the “reserve pool,” which maps the signal from the
input space to the high-dimensional state space to characterize its complex state and
forms a feedback structure connecting the output layer to the reserve pool with a
certain "memory" function. The typical structure is shown in Fig. 6.4 [161].
In the echo state network shown in Fig. 6.4, the input layer signal u(k) is connected
to the reserve pool via the input connection rights matrix W in ∈ RN×M . The internal
state of the reserve pool x(k) is internally connected via the internal connection rights
matrix W ∈ RN×N for processing the input signal and connected to the output layer
via the output connection rights matrix W out ∈ RL×(M + N + L) . The output signal y(k)
can also be fed back to the reserve pool via the feedback connection rights matrix
W fb ∈ RN×L .
6.3 Line Adaptive Overload Protection Action Time 295

Fig. 6.4 Structural diagram of echo state network

The update equation for the internal state of the reserve pool and the output
equation for the echo state network are [161]

x(k + 1) = f (W x(k) + W in u(k) + W fb y(k)) (6.8)

y(k + 1) = f out (W out [x(k + 1), u(k + 1), y(k)]) (6.9)

where f = [ f 1 , f 2 , . . . , f N ] is the internal neuron activation function, usually

chosen as a nonlinear action function such as a tanh or sigmoid function. f out =
[ f 1out , f 2out , . . . , f Nout ] is the output function, which is usually taken as a linear
constant function.
The connection power matrices W in , W and W fb are generated randomly during
the initialization phase of the network and remain constant thereafter. The output
connection right matrix W out is obtained by training the input and output signals of
the system and is the training object of the echo state network.
To further improve the prediction accuracy, a recursive least squares (RLS) method
is used to update the output weight matrix W out of the echo state network online.
This is referred to as the RLS-ESN method later on.
Define the RLS cost function, i.e., the cumulative squared error performance
function, as


k
E(k) = λk−i e(k)2 (6.10)
i=1

where e(k) is the kth point output error, i.e., ytarget (k) − y(k). λ is the forgetting factor,
which serves to attenuate the effect of historical data and fast-track data changes.
296 6 Adaptive Overload Protection for Overhead Transmission Lines

To ensure that E(k) of Eq. (6.10) is minimized, the following relationship needs
to be satisfied for Eqs. (6.11) to (6.13):

w(k + 1) = w(k) + e(k)g(k) (6.11)

P(k − 1)x(k)
g(k) = (6.12)
λ + x(k)T P(k − 1)x(k)

P(k) = λ−1 P(k − 1) − g(k)x T λ−1 P(k − 1) (6.13)

2. Echo state network training algorithm

The training algorithm for building an echo state network consists of two main parts:
network structure initialization and network training.
Network structure initialization consists of the following steps [166]:
1) Select the input and output sample signals: [u(1), y(1)],.[u(2), y(2)]……[u(t), y(t)].
u is the input signal, y is the output signal, and t is the sample time.
2) Generate the connection weight matrix W, W in , W fb : First, randomly generate
the internal connection weight matrix W 0 . Perform the normalization such that
W 0 = W 1 /λmax, where λmax is the W 0 spectral radius. Randomly generate the input
weights matrix W in and the feedback connection weights matrix W fb .
3) Scale and shift the input and output sample signals.
4) Let ESN output the weight matrix W out with column vector w(k) = 0 and state
matrix x(k) = 0 at moment k. P(0) = δ −1 I, I is the unit array, δ  1, 0 < λ < 1.
Network training consists of the following steps [166]:
1) Dynamic sampling forms the state collection matrix S and the desired output
matrix E. From moments 0 to t, the internal state x(k) is computed according to
Eq. (6.5) to form the state collection matrix S, where s(k) =[u(n); x(n); y(n −
1)]T ; and forms the desired output matrix E, E = [(f)out −1 y(t 0 ), (f out )−1 y(t 0+1 ),
where (f)out −1 y(t)]∈ R(t−t0 +1)×L .
2) Calculate the output weight matrix W out . From Eqs. (6.8) and (6.9), the following
relationship exists between the weight matrix W out , state collection matrix S and
desired output matrix E:

(W out )T = S −1 E (6.14)

Update W by RLS online learningout : update g(k) by Eq. (6.12), update the output
weight matrix w(k + 1) by Eq. (6.11), and update the inverse matrix P(k) by Eq. (6.13).
3. Specific applications
The specific steps of ESN-ΔT max illustrated by applying the RLS-ESN method to
predict the maximum temperature difference along the line are shown in the schematic
block diagram in Fig. 6.5.
The specific steps are as follows:
6.3 Line Adaptive Overload Protection Action Time 297

Fig. 6.5 RLS-ESN-based prediction of maximum temperature difference along line block diagram

1) Online sample training

To increase the training sample and improve the prediction accuracy, the temperature
difference ΔT max between the wire temperature station monitoring value T local and
the along-line monitoring maximum T max during normal operation is used as a sample
for the RLS-ESN online training on a smaller time scale with a uniform difference. At
moment k, the historical data ΔT max (k-N) ~ ΔT max (k) of the maximum temperature
difference along the line from moment k-N to moment k are used to predict the
maximum temperature difference along the line from moment k + 1 to moment k
+ M ΔT Emax (k + 1) ~ ΔT Emax (k + M). From moment k + 1 to moment k + M,
the predicted data errors are corrected by the actual measured data ΔT max (k + 1)
~ ΔT max (k + M) of the maximum temperature difference along the line and are
continuously trained to update the ESN network model.
2) Offline/online real-time forecasting
When data are missing along the line after k 2 hours due to communication, offline
prediction can be made based on historical data ΔT max (k)2 by the ESN method.
If the communication is open, then ΔT (k max2 ) and previous historical data can be
predicted online by RLS-ESN Tmax (k 2 + 1) ~ ΔT max (k 2 + M) based on T (k)
and previous historical data.
Real-time prediction of time-varying parameters using the rebound state network
method, whose inputs are current and historical data A(j), B(j) (j ≤ k). The output is
the predicted values AE (j), BE (j), and the procedure is the same as above. Based on
the predicted time-varying coefficients, the predicted local temperature data of the
substation can be obtained by iterative calculation by substituting Eq. (6.15) (where
298 6 Adaptive Overload Protection for Overhead Transmission Lines

Δt is the prediction step).

A(t) A(t) B(t)t

Tc (t+t) = Ta (t) − + [Tc (t) − Ta (t)+ ]e mCP (6.15)
B(t) B(t)

In this chapter, the prediction of time-varying parameters followed by the predic-

tion of local conductor temperatures is represented by ESN-T local . Combined with
the maximum temperature difference along the line predicted by ESN-ΔT maxl , the
predicted maximum line temperature is obtained, which is used to predict the adaptive
overload protection action time.

6.4 Component Scheme for Line Adaptive Overload

Protection

6.4.1 Rectification Scheme and Action Logic

From the previous analysis, it can be seen that the damage caused by the wire overload
operation is the result of the combined effect of temperature and time, and the effect
grows nonlinearly with increasing temperature. Comprehensive test data and the data
calculated above show that when the line is at 100 °C for short periods of operation,
the wire mechanical strength, arc sag and loss of fixtures can be ignored. When
the line temperature is too high, temperatures higher than 120 °C will cause greater
permanent damage to the line and should be quickly removed.
Therefore, the three concepts of alert temperature, emergency temperature and
limit temperature are proposed. The so-called alert temperature refers to the line that
can be allowed to run for a long time in the temperature and below. Attention should
be taken once the temperature exceeds the alert temperature, The alert temperature
can be the maximum allowable operating temperature of the existing regulations. The
so-called emergency temperature is reached when the wire temperature exceeds the
value; if this will cause damage to the line in the long run, then the situation is more
urgent in the delayed tripping phase. The so-called limit temperature is when the line
temperature exceeds the value and is on the mechanical strength of the conductor,
and other properties cause permanent damage and need to be removed immediately.
Through the previous analysis, the following scheme is created for adaptive over-
load protection logic and thermal calibration with a typical steel core of a 500-kV
overhead transmission line.
1) Alert temperature T w
Define the alert temperature T w as 60 °C. When the wire temperature keeps rising
and exceeds 60 °C, the alert state is entered. The adaptive overload protection predicts
the temperature change within 10 min. If the line temperature rises to the emergency
temperature within the predicted time, then an alert message is sent to the dispatch
6.4 Component Scheme for Line Adaptive Overload Protection 299

center (message sent via IEC 61,850 to IEC 61,970 communication interface [167])
informing about the remaining time to reach the emergency temperature.
2) Emergency temperature T E with emergency thermal setting C max
The emergency temperature T E is defined as the maximum allowable temperature
for long-term operation, which is commonly taken as 70 °C in China’s power grid
and 80 °C in East China’s power grid. This chapter takes 70 °C as an example to
illustrate that when the conductor temperature is higher than 70 °C, the adaptive
overload protection enters the emergency state.
According to the analysis in Sect. 2.3, the effect of overload on the performance
and life of the conductor is the result of the combined effect of temperature and time,
with a thermal accumulation effect. When the conductor temperature is 70 °C and
above, the temperature time product is calculated as C. When the real-time calculated
temperature time product C exceeds the line emergency thermal constant C max , the
adaptive overload protection outlet trips. The criteria are shown in the following
equation:

C= TC dt ≥ Cmax (6.16)

where the emergency thermal setting C max is determined by the emergency

temperature T H and the emergency support time t.H .

Cmax = TH × tH (6.17)

The transmission network commonly used steel-core aluminum stranded wire as

an example. In the previous analysis, the line runs at 100 °C for 60 min, and the wire
mechanical strength, arc sag and loss of performance of the fixture are low. Therefore,
the line emergency heat can be fixed as follows: T H = 100 °C, t H depending on the
degree of aging of the line and transmission corridor conditions (wire to ground
spacing, vegetation growth), which can be taken from 1 to 60 min within the value.
When the adaptive overload protection enters the emergency state, the transient
temperature rise is predicted by the echo state network method. The time when the
line reaches the emergency thermal setting is predicted as the tripping time of the
adaptive overload and reported to the dispatch center.
3) Limit temperature T L
Define the limit temperature T L as 120 °C. To prevent permanent damage to the line
due to temperature spikes, when the conductor temperature is greater than T L , it
should be quickly removed. In this case, it is necessary to consider the action with
the main protection and backup protection in the case of a short circuit.
The line short-circuit currents were 70 kA, 50 kA, 30 kA and 10 kA; the initial
temperature of the line before the fault was 70 °C; ambient temperature was 40 °C;
wind speed was 0.5 m/s; wind angle was 0°; the sunlight intensity was 2000 W/m2 ;
300 6 Adaptive Overload Protection for Overhead Transmission Lines

Table 6.2 Maximum line temperature at short-circuit current

Short circuit current 70 kA 50 kA 30 kA 10 kA
Trip at 0.1 s 70.19 °C 70.10 °C 70.03 °C 70.00 °C
Trip at 5 s 79.48 °C 74.81 °C 71.72 °C 70.19 °C

0.1 s elapsed after the fault; and the 5-s line was tripped and removed before the
highest temperature of the line occurred, as shown in Table 6.2.
Table 6.2 shows that even if a very serious short-circuit fault occurs on the line and
is removed only after 5 s, the line temperature does not reach the limit temperature T L .
There is no adaptive overload protection that trips the line in the case of a short-circuit
before the conventional protection (main protection and backup protection).
4) Other functions
Detection alarms for fire and other dangerous conditions along the line. When there
is a sudden increase in the maximum temperature difference between the line and
the station and the maximum temperature cannot correspond to the current current-
carrying and meteorological conditions, an abnormal alarm is issued and uploaded
to the dispatch center, and the adaptive overload protection outlet is blocked [167].
According to the previous section, the adaptive overload protection of the overhead
transmission line is independent of the line protection reflecting the fault and is
adjusted according to the thermal stability of the line. This is accomplished with
full consideration of the line emergency current-carrying capacity and the transient
temperature rise process in order to buy time for the implementation of system safety
and stability control measures as reasonably as possible. Its action logic is shown
in Fig. 6.6. When the maximum temperature along the line T Cmax is greater than
the limit temperature T L or the line is in an emergency state and meets the line
temperature time product C ≥ Cmax , then the protection trips. When the line is in
the alert state, an emergency state or abnormal temperature along the line is detected
and reported to the dispatch center.

Fig. 6.6 Line adaptive overload protection action logic

6.4 Component Scheme for Line Adaptive Overload Protection 301

6.4.2 Algorithm Flow

The adaptive overload protection algorithm can be divided into five modules:
temperature acquisition, online training, real-time prediction, action logic and
communication. The algorithm flow is shown in Fig. 6.7.
The temperature acquisition consists of two parts: one part comes from the temper-
ature and meteorological data collected by the online monitoring system along the
line, which is updated every 5 min (may be longer for nonemergency conditions);
the other part comes from the monitoring data in the substation station, which can be
updated in seconds. However, since temperature change is a relatively slow process,
taking into account the calculation volume and accuracy, the sampling period is set
as 10 s in this chapter.
The training of the echo state network is implemented with data during normal
operation, including ESN-ΔT max for predicting the maximum temperature difference
along the line and ESN-T local for predicting the temperature in real time based on
substation station data. The ESN is trained in real time with a large number of
samples online, and its prediction accuracy is higher than that of offline prediction.
However, to ensure the prediction ability of the protection in the initial state, it is
still necessary to reserve offline sample data for initial training. The data along the
line are calibrated to exclude the line temperature rise due to nonoverload such as
hill fires, and the abnormal alarms are uploaded to the dispatch center.
If the data pass the calibration, then the current line maximum temperature T Cmax
value transmitted along the line or projected from the station data is used to deter-
mine whether the limit temperature, emergency temperature or alert temperature is
reached. The well-trained rebound state network method is invoked to predict the line
temperature in real time, and the line is removed according to the action logic, or an
alert message is sent to the dispatch center to ensure the safety of the line operation.
The flow of the emergency handling subroutine is shown in Fig. 6.8. Since the line
temperature change has a cumulative temperature effect, it does not make the case that
the line temperature changes frequently between the alert temperature T w and emer-
gency temperature T E even if the line current changes frequently. However, the situ-
ation where the wire temperature changes up or down in the emergency temperature
is still considered, and the following emergency handling flow is designed.

(1) Adaptive overload protection when the line temperature is greater than the
emergency temperature T E . The adaptive overload protection can be set The
emergency status flag is 1.
(2) If the line temperature drops below the emergency temperature TE alert temper-
ature Tw above, adaptive overload protection remains in the emergency state,
and the temperature time product C is not cleared. If the line temperature is
again higher than TE, continue to calculate the temperature time product.
(3) If the line temperature drops to the alert temperature Tw and below, then the
protection lifts the emergency state; and if the temperature time product is still
less than Cmax, the temperature time product C is cleared.
302 6 Adaptive Overload Protection for Overhead Transmission Lines

Fig. 6.7 Line adaptive overload protection program flow

6.4 Component Scheme for Line Adaptive Overload Protection 303

Fig. 6.8 Emergency handling subroutine flow

(4) When the protection is in an emergency, once C > C max is satisfied, the line is
tripped.

6.4.3 Applications and Calculations

1. Forecast of maximum temperature differences along the route

The initial wire temperatures measured at the substation and at the hottest point along
the line are 40 °C and 42 °C, respectively. After the line is overloaded, the current
is taken uniformly between 800 and 1000 A and changes every 10 min, as shown in
Table 6.3. For other environmental parameters, it is assumed that the wind speed in
the substation obeys a uniform distribution, and the wind speed at the hottest place
along the line subtracts an additional value obeying a normal distribution on this
basis, weakening its convective heat dissipation and increasing the temperature.
In Table 6.3, U(a,b) denotes a uniform distribution with values taken between (a,b).
N(a,b2 ) denotes a normal distribution with mean a and variance b2 . The 10,000-min
substation conductor temperature data T local and conductor temperature data T max
304 6 Adaptive Overload Protection for Overhead Transmission Lines

Table 6.3 Temperature-related parameters at substation and hottest point along the line
Location Initial Wind speedvw (m/s) and wind angle Ambient Sunlight
temperature δ(°) temperature intensity
T a (°C)
Substation 40 °C vw ~ U(1,3),δ ~ U(0,90) T a ~ N(35,12 ) 1000 W/m2
Hottest 42 °C vw ~ [U(1,3)-N(1,0.12 )],δ ~ U(0,90) T a ~ [N(35,12 ) 1100 W/m2
point + U(1,3)]

at the hottest point along the line (with a sampling period of 5 min) were generated
under the above conditions (see Fig. 6.9).
Using the echo state network method to train the sample data for 8000 min, the
maximum temperature difference data between the substation conductor and highest
temperature along the line at moment k is input as u(k) = [ΔT max (k - 30 min)…,
ΔT max (k)]T . The output is y(k) = [ ΔT max (k + 1)…., ΔT max (k + 30 min)]T .
The reserve pool network connection power matrix scales as W in ∈ R100×1 , W ∈
R100×100 , W fb ∈ R100×1 and W out ∈ R1×102 . When the reserve pool spectral radius is
set to λmax = 0.8, the input and output scales are scaled to 0.1 and 0.3, respectively,
the shifts are 0 and -0.2, the forgetting factor λ is taken to be 0.9995 and δ is taken
to be 10–6. When the trained echo state network is utilized, iterative prediction is
performed on the remaining data. When the prediction duration is 30 min, i.e., the
data before 7970 min are used to predict the maximum temperature difference along
the line at 8000 min, and so on. The 30-min predicted values are compared with the
actual values in Fig. 6.10.
The maximum temperature difference along the line is predicted at different time
durations. The maximum, minimum and average root mean square errors are shown in
Table 6.4. The method has a high prediction accuracy and can meet the requirements
of adaptive overload protection.
Further finer predictions of conductor temperatures are based on substation
measured temperature data.
2. Temperature and action time prediction under current and wind speed variations
The RLS-ESN method is used for the real-time prediction of time-varying parameters
in the presence of current and wind speed variations.

Fig. 6.9 Substation conductor temperature and maximum temperature along the line
6.4 Component Scheme for Line Adaptive Overload Protection 305

Fig. 6.10 Comparison of predicted and actual 30-min values of maximum temperature difference
of conductors

Table 6.4 Root mean square error in predictions of maximum temperature differences along the
line at different time durations
Time duration 5 min 10 min 30 min
RMS error(%) 0.089% 1.13% 2.11%

The predicted input data are the values of parameter A at and before moment
k: u(k) = [A(k − 5 min)…, A(k)]T . To ensure the tracking effect on the current
variation, the RLS-ESN input samples are taken only for the first 5 min of the k
moment data. Output time-varying parameter prediction data: y(k) = [A(k + 1)…,
A(k + 10 min)]T . The RLS-ESN parameters are set as follows: the reserve pool
spectral radius is taken as λmax = 0.4; input scaling scale and displacement are
taken as 0.5 and 1.0, respectively; forgetting factor λ is taken as 0.9995; and error
limit δ is taken as 10–6.
The predicted input data are the value of parameter B at and before moment k:
u(k) = [B(k − 5 min)…, B(k)]T . The input sample is taken as the 30-min data before
moment k. The output time-varying parameter prediction data are y(k) = [B(k +
1 min), B(k + 5 min), and B(k + 10 min)]T. The RLS-ESN parameters are set as
follows: the reserve pool spectral radius is taken as λmax = 0.4; input scaling scale
and displacement are taken as 0.5 and 1.0, respectively; forgetting factor λ is taken
as 0.9995; and the error limit δ is taken as 10–6.
The predicted time-varying parameters A and B are predicted using the RLS-ESN
method for the two previous examples, and the predicted time-varying parameters
are substituted into Eq. (6.15), which in turn leads to the predicted values of the wire
temperature, as shown in Fig. 6.11.
Comparing the RLS-ESN method and the conventional method applied in
Example 1 and Example 2, the root mean square error of the parameter values and
wire temperature prediction after 10 min is calculated, and the results are shown in
Table 6.5.
From Table 6.5, it can be seen that the RLS-ESN method is effective in improving
the real-time prediction accuracy of real-time pairs of conductor temperatures.
Combined with the prediction of the maximum temperature difference along the
line, take the emergency thermal setting C max = 100 °C × 60 min. According to
306 6 Adaptive Overload Protection for Overhead Transmission Lines

a) Example 1

b) Example 2

Fig. 6.11 Comparison of real-time predictions of conductor temperature

Table 6.5 Comparison of 10-min prediction errors for time-varying parameters and temperatures
under different methods
RMS error Example 1 prediction values (10 min) Example 2 prediction values (10 min)
Parameter Parameter Wire Parameter Parameter Wire
A B temperature A B temperature
Conventional 22.9% 17.49% 2.85 °C 6.04% 24.92% 3.36 °C
method
Proposed 22.9% 12.42% 1.45 °C 0.10% 18.09% 1.17 °C
method

adaptive overload protection action logic, Case 1 does not trip, and the moment
when Case 2 reaches the emergency thermal setting, i.e., the actual tripping moment,
should be 552.8 min (with the initial moment of the current step-up as 0 min).
If within the predicted time [k, k + 10 min] the temperature time product reaches
the emergency thermal setting, i.e., the predicted value of adaptive overload trip time
t O (k) for the current moment k is noted, if the emergency thermal setting is not
6.4 Component Scheme for Line Adaptive Overload Protection 307

Table 6.6 Comparison of trip time prediction errors under different methods
Time of prediction Conventional method Proposed method
(min) Tripping time Prediction error Tripping time (min) Prediction error
(min)
480 not predicted – 563.3 10.5
500 560 min 7.2 min 556.7 3.9
520 555.3 min 2.5 min 553.7 0.9
540 553 min 0.2 min 552.7 − 0.1

reached within the predicted time and the predicted temperature T c (k + 10 min) is
greater than the emergency temperature T E , then the temperature of the conductor
after the moment k + 10 min is predicted by T c (k + 10 min) and the temperature
time product continues to be calculated to obtain the predicted value of adaptive
overload trip time t O (k). If T c (k + 10 min) is less than the emergency temperature
T E , then stop calculating t O (k).
The time t O (k) at which the line ambient temperature reaches the emergency
thermal fix is predicted by applying the RLS-ESN method and the conventional
method at different moments; see Table 6.6.
Since the conventional method cannot track the temperature change in time for
prediction, it may occur that T c (k + 10 min) is less than emergency temperature T E ,
resulting in a later prediction of the effective tripping moment. The prediction error
is approximately twice the prediction error of the method proposed in this chapter.
In summary, for current and wind speed changes, the proposed method in this
chapter can more accurately predict the conductor temperature, provide a more timely
and more accurate adaptive trip time of overload protection, and reliably ensure the
safe operation of transmission overhead lines.

6.4.4 Summary

This chapter addressed three problems of existing line adaptive protection. First,
the emergency current-carrying capacity of the line was analyzed, and the allowable
operating temperatures of the line at different time scales were discussed. Second,
to address the problems of temperature prediction errors due to time-varying param-
eters and slow updates or even missing monitoring data along the line, a tempera-
ture prediction method based on substation and monitoring data along the line was
proposed to achieve higher accuracy temperature prediction in real time through the
echo state network method. Finally, an line adaptive overload protection rectifica-
tion scheme, action logic and process were designed. The features of the adaptive
overload protection for overhead transmission lines proposed in this chapter are as
follows.
308 6 Adaptive Overload Protection for Overhead Transmission Lines

(1) The line overload situation protection action guidelines are proposed to protect
the line and system security. On the one hand, the protection line does not
cause permanent damage to the mechanical strength of the conductor, plastic
deformation, fixtures and connectors due to overload, thus guaranteeing the
safe operation of the line. On the other hand, there exists further tapping of the
emergency current-carrying capacity of the line and tripping as few times as
possible within the safety margin.
(2) Adaptation of the protection action time in the line thermal stability. Through
analysis of the emergency current-carrying capacity of the line, the proposed
emergency thermal rating with temperature time product instead of the
maximum allowable temperature or maximum current-carrying capacity as the
line thermal stability criteria, both a more realistic reflection of the tempera-
ture on the wire thermal cumulative effect, fully takes into account the transient
temperature rise process of the line with the advantages of static temperature
and dynamic monitoring.
(3) We performed real-time prediction of the maximum temperature difference
along the line with a large time step, completion of the spatial expansion of
temperature prediction, and prediction of time-varying parameters with a small
time step based on monitoring data in the substation, realizing the time exten-
sion of temperature prediction. Combined with the maximum error along the
line, this can overcome the difficulties of slow update of online temperature
monitoring data and data deficiency, realizing a more accurate real-time predic-
tion of the maximum temperature along the line. The reliability of the adaptive
overload protection action and the accuracy of the alarm time are guaranteed.
(4) The line adaptive overload protection proposed in this chapter requires only
station domain monitoring information with a small amount of monitoring
information along the line and is easy to implement.
Appendix
Transient Temperature Calculations

The overhead line temperature is the joint result of the wire’s own heat production,
heat absorption and heat dissipation to the outside world. Among them, Joule heat
absorption PJ and sunlight heat absorption PS are the main sources of heat; convection
heat dissipation PC and radiation heat dissipation PR are the main sources of heat
dissipation. The heat balance equation of the conductor is [158]

dTc
mCp = PJ + PS − PC − PR (A.1)
dt
where m is the mass of the wire (also expressed by density), and Cp is the specific
heat capacity of the wire. The equations for solving for the heat of each component
are given below.

Joule Heat Absorption

The Joule heat absorption P of wire J can be found by the following equation [167]:
 
PJ = Iac
2
Rac = kj Iac
2
Rdc20 1 + αj (Tc − 20) (A.2)

where Iac , Rac , and Rdc20 are the AC current, AC resistance and DC resistance
of the wire at 20 °C, respectively; kj is the correction factor to consider the skin
effect; for the wire containing ferrous materials, kj = 1; for the wire without ferrous
materials, kj = 1.0123; αj is the temperature resistance coefficient of the wire at
20° C, αj = 0.0036/°C; DC resistance Rdc20 = 0.07389/km at 20 °C; and Tc is the
wire temperature.

© The Editor(s) (if applicable) and The Author(s), under exclusive license 309
to Springer Nature Singapore Pte Ltd. 2023
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2
310 Appendix: Transient Temperature Calculations

Heat Absorption by Insolation

The insolation heat absorption Ps can be expressed by the following equation [167]:

PS = αs Q s D (A.3)

αs is the absorption rate of light by the conductor and ranges from 0.23 to 0.95
depending on the degree of weathering of the conductor (0.23 for new conductors,
0.5 for approximately 1 month of operation, and approximately 0.9 after one year),
D is the diameter of the conductor, and QS is the sunlight intensity in W/m2 .

Convection Heat Dissipation

Wire convection heat dissipation can be divided into natural convection (zero air
velocity) and forced convection (nonzero air velocity). The formula for forced
convection heat dissipation PC is [167]

PC = πλf (Tc − Ta )Nu (A.4)

where λf is the thermal conductivity of air, Ta is the ambient temperature around

the conductor, and Nu is the Nusser number, a dimensionless number indicating the
strength of convective heat transfer, which can be expressed as Nu = B1 (Re)n. The
Reynolds number Re is the ratio of fluid惯性 force to viscous force ratio, charac-
terizing the fluid flow, and is calculated as Re = Dρf vw/μf . ρf is the air density,
vw is the wind speed, μf is the air viscosity coefficient, and B1 and n depend on
the wire surface roughness Rf = d/[2(D − 2d)]. d denotes the outer strand diameter.
When Rf ≤ 0.05, B1 , and n take the values 0.178 and 0.633, respectively. When Rf >
0.05, B1, n takes the values 0.048, 0.8. λf , ρf , and μf are experimentally determined
coefficients. The specific parameters can be found in [167].
Considering the wind angle, correcting for the Nussle number, we have [167]

Nuδ = Nu [A1 + B2 (sin δ)m1 ] (A.5)

δ is the angle between the axial direction of the conductor and the wind direction.
When 0° < δ < 24°, A1 = 0.42, B2 = 0.68, and m1 = 1.08. When 24° < δ < 90°, A1
= 0.42, B2 = 0.58, and m1 = 0.9.
Under natural convection, the Nussle number is solved by

Nu = A2 (G r · Pr )m2 (A.6)

where the solutions for Gr and Pr are given in [167].

Appendix: Transient Temperature Calculations 311

Radiation Heat Dissipation

The radiation heat dissipation PR is calculated as [167]

PR = πDεσB [(Tc + 273)4 − (Ta + 273)4 ] (A.7)

where ε is the emissivity, which can be taken as 0.27 for new conductors and
0.95 for conductors operating for more than 1 year, and σB is the Stefan Boltzmann
constant (the constant in the blackbody radiation law), which has a value of 5.67 ×
10–8 W/(m2 K4 ).

Method of Calculation

To improve the efficiency of the solution, the transient heat balance equation can be
suitably simplified. The insolation heat absorption is only related to climatic factors
(insolation intensity), independent of the wire temperature, and is presented by Joule
heat absorption, convection heat dissipation, and radiation heat dissipation (Tc −
Ta ).

d(Tc − Ta )
mCP = A + B(Tc − Ta ) (A.8)
dt
where parameter A = (I2 β1 + αs Qs D/mCp), β1 = kj (αj (Ta − 20) + 1)Rdc20 ;
parameter B = (I2 β2 -πλf Nuδ -πDεσB Tθ mCp), β2 = kj αj Rdc20 , and Tθ = [(Tc
+ 273)2 + (Ta + 273)2](Tc + Ta + 546).
Parameter A contains the insolation heat absorption component αs Qs D and Joule
heat absorption component I2 β1. In the case of line overload, the insolation heat
absorption has a small value compared to Joule heat absorption and can be consid-
ered constant for a certain time. The value can be predicted from the local insolation
intensity curve and wire parameters. Furthermore, the Joule heat absorption compo-
nent can be found from the wire parameters and current values. Therefore, A is a
known quantity and can be considered a constant when the current is smooth.
The Joule heat absorption component I2 β2 and radiation heat dissipation compo-
nent πDεσB Tθ in parameter B can also be obtained from the wire parameters, current
values, and substation temperature data. The ambient temperature changes less during
the transient temperature rise and can be regarded as a constant. The convective heat
dissipation component πλ Nfuδ in parameter B can be derived from the current
temperature data and the approximated known quantities to obtain the current πλf
Nuδ component because the wind speed and wind direction parameters are difficult
to obtain and the measurement accuracy is not high.
To obtain the general solution of the (A.8) differential equation to predict the tran-
sient temperature rise, an analysis of the correlation between Tθ and Tc is performed,
and its approximate decomposition leads to
312 Appendix: Transient Temperature Calculations
 
Tθ ≈ πDεσB (Ta + 273) 3.92(Ta + 273)2 − 0.006(Tc − Ta ) (A.9)

Tθ with the (Tc − Ta ) component is an order of magnitude smaller than that without
the (Tc − Ta ) component, and the radiation heat dissipation usually accounts for only
10–40% of the total heat dissipation [61]. Therefore, Tθ can be neglected; that is, with
Tθ K2 as a constant, Formula (A.8) can be approximated as a constant coefficient
first-order differential equation and can be derived from the general solution of the
transient temperature rise of the wire expression.

A A B(t−t0 )
Tc (t) = Ta (t) − + [Tc (t0 ) − Ta (t)+ ]e mCP (A.10)
B B

Wire Parameters

In this paper, a typical 500-kV four-split steel-core aluminum strand, type LGJ-
400/35, is used as a case for calculation. m = 1349 kg/km, Cp = 1074 J/(kg-K), D =
26.82 mm, d = 3.22 mm, Rf = 0.079, B1 = 0.048, and n = 0.8 for the LGJ-400/35
conductor density.
Bibliography

1. National Technical Committee for Standardization of Power Grid Operation and Control.
Interpretation of the provisions of the guidelines for safety and stability of power systems and
study guide [M]. Beijing: China Electric Power Publishing House, 2020.
2. Zeng YG. Safe and stable operation practices of the Southern Power Grid [R]. 2012.
3. Lin Weifang, Sun Huadong, Tang Chung, et al. Analysis and insights of the "11-10" blackout
in Brazil [J]. Power System Automation, 2010, 34(07): 1-5.
4. State Grid Corporation Office in Brazil. Report on the situation of the 3.21 widespread power
outage in Brazil [R]. 2018.
5. Liu YY, Zuo J, Lopsided Hu, et al. Analysis of Brazil’s 3-21 blackout accident and its
inspiration for Hunan power grid [J]. Hunan Electric Power, 2019, 39(02): 25–29.
6. Dong XZ, Cao RB, Wang B, et al. Indian blackouts and the three functions of relay protection
[J]. Power System Protection and Control, 2013, 41(02): 19-25.
7. He JL, Li YL, Dong XZ, et al. Principles of power system relay protection (4th ed.) [M].
Beijing: China Electric Power Publishing House, 2010.
8. Zheng Chao, Ma Shiying, Shen Xuhui, et al. Definition, connotation and form of strong
straight and weak intersection and its countermeasures [J]. Power Grid Technology, 2017,
41(8): 2491–2498.
9. Report of the Enquiry Committee on Grid Disturbance in Northern Region on 30th July 2012
and in Northern, Eastern &North-Eastern Region on 31st July 2012[R]. 2012.
10. Tang, Y., Bu, G., Yi, J.. Analysis and insights of the "7.30" and "7.31" blackout accidents in
India [J]. Chinese Journal of Electrical Engineering, 2012, 32(25): 167–174.
11. U.S.-Canada Power System Outage Task Force. final report on the August 14th blackout in
the United States and Canada: causes and recommendations[R] . . 2004.
12. UCTE. Final report of the investigation committee on the 28 September 2003 blackout in
Italy [R]. 2004.
13. L Angelo, F Gianluca. Modeling and application of VSC-HVDC in the European transmission
system [J]. Int. J. Innov. Energy Syst. Power, 2010, 5(1): 8-16.
14. Kamalapur G D, Sheelavant V R, Hyderabad S, et al. HVDC Transmission in India [J]. IEEE
Potentials, 2014, 33(1): 22-27.
15. Shu Yinbiao, Liu Zehong, Yuan Jun, et al. An overview of the demonstration work of ultra-
high voltage transmission of the State Grid Corporation in 2005 [J]. Power Grid Technology,
2006, 30(5): 1-12.
16. Liu Z.Y., Qin X.F., Zhao L., et al. Research on the application of UHV DC hierarchical access
method in multi-feeder DC networks [J]. Chinese Journal of Electrical Engineering, 2013,
33(10): 1-7.

© The Editor(s) (if applicable) and The Author(s), under exclusive license 313
to Springer Nature Singapore Pte Ltd. 2023
X. Dong, AC/DC Hybrid Large-Scale Power Grid System Protection,
https://doi.org/10.1007/978-981-19-6486-2
314 Bibliography

17. National Development and Reform Commission, National Energy Administration. 2019
Annual National Renewable Energy Power Development Monitoring and Evaluation Report
[R]. Beijing: National Development and Reform Commission, National Energy Administra-
tion, 2020.
18. J. Tu, J. Zhang, G. Bu, et al. Analysis of the sending side system instability caused by multiple
HVDC commutation failure [J]. CSEE Journal of Power and Energy Systems, 2015, 1(4):
37-44.
19. Li M. J. Characterization and operation control of large-scale extra-high voltage AC-DC
hybrid grid [J]. Power Grid Technology, 2016, 40(04): 985-991.
20. Paulo F T, Jiuping P, Kailash S, et al. Case study of a multi-infeed HVDC system [C]// Power
System Technology and IEEE Power India Conference, 2008.
21. H. Clark, A. Edris, M. El-Gasseir, et al. Softening the blow of disturbances [J]. IEEE Power
Energy Mag, 2008, 6(1): 30–41.
22. Liang ZF, Kang SN, Sonam Gale, et al. Fault component extraction algorithm for develop-
mental faults [J]. Power System Automation, 2007, (06): 44–47.
23. Tang, N.. Analysis of the action behavior of a continuously developing single-phase ground
fault protection [J]. Power Grid Technology, 2006, (04): 103-104.
24. Sonam Galle, Xu Qingqiang, Li Xiaobin, et al. Developmental fault discrimination elements
for ultra-high voltage transmission lines [J]. Chinese Journal of Electrical Engineering, 2006,
26(04): 93–98.
25. Yang Q, Zheng T, Xiao S W, et al. A new criterion for transformer differential protection with
fast identification of transitive faults [J]. Power System Automation, 2007, 31(20): 61–64+107.
26. Xue Y. S., Xie Y. Y., Wen F. S., et al. A review on the study of successive faults in power
systems [J]. Power System Automation, 2013, 37(19): 1–9+40.
27. Baldick R, Chowdhury B, Dobson I, et al. Initial review of methods for cascading failure
analysis in electric power transmission systems[C]//2008 IEEE Power and Energy Society
General Meeting-Conversion and Delivery of Electrical Energy in the 21st Century. IEEE,
2008: 1–8.
28. Xiaohui Ye, Wuzhi Z, Xinli S, et al. Review on power system cascading failure theories
and studies[C]//2016 International Conference on Probabilistic Methods Applied to Power
Systems (PMAPS). IEEE, 2016: 1–6.
29. Sun Ke, Han Zhenxiang, Cao Yi. Review of complex grid interlocking fault models [J]. Power
Grid Technology, 2005, 29(13): 1-9.
30. Gan, D. Q., Hu, J. Y., Han, J. C.. Reflections on several international power outages in 2003
[J]. Power System Automation, 2004, 28(03):1–4+9.
31. M. Vaiman, K. Bell, Y. Chen, et al. Risk Assessment of Cascading Outages: Methodologies
and Challenges [J]. IEEE Transactions on Power Systems, 2012, 27(2): 631-641.
32. Lu Q., Mei S. W. Research on major scientific issues of disaster prevention, control and
economic operation of large power systems in China [J]. China Basic Science, 1999, (Z1):
61-67.
33. Fan Wenli, Liu Zhigang, Hu Ping, et al. Cascading failure model in power grids using the
complex network theory [J]. IET Generation, Transmission & Distribution, 2016, 10(15):
3940–3949.
34. Xue Shimin, Sun Wenpeng, Gao Feng, et al. Transmission system chain fault risk assessment
based on accurate hidden fault model [J]. Power Grid Technology, 2016, 40(4): 1012-1017.
35. Ding M, Zhu ZQ, Zhang JJ, et al. Protection of hidden faults and their impact on the devel-
opment of interlocking faults in power systems [J]. High Voltage Technology, 2016, 42(01):
256-265.
36. Fan W.L., Liu C.G.. Analysis of the impact of hidden faults on chain failures in small world
grids [J]. Power System Automation, 2013, 37(21): 23-28.
37. Yang M.Y., Tian H., Yao W.Y.. Chain fault analysis of power system based on relay protection
hidden faults [J]. Power System Protection and Control, 2010. 38(09): 1-5.
38. Zhao Wanjun. High voltage DC transmission engineering technology [M]. Beijing: China
Electric Power Press, 2004.
Bibliography 315

39. Wang Xifan. Power system calculations [M]. Beijing: Water Conservancy and Electric Power
Press, 1978.
40. P. M. Andeson. power system protection [M]. John Wiley & Sons Ltd, 1999.
41. Dong XZ, Ding L, Liu Kun, et al. Local information-based system protection [J]. Chinese
Journal of Electrical Engineering, 2010, 30(22): 7–13.
42. Dong Xinzhou. System protection theory and technology for defending against interlocking
faults in AC-DC hybrid grids [C]// Report of the Chinese Society of Electrical Engineering,
2018: 1–31.
43. SHU Yinbiao, CHEN Guoping, YU Zhao, et al. Characteristic analysis of UHV AC/DC
hybrid power grids and construction of power system protection [J]. CSEE Journal of Power
and Energy Systems, 2017, 3(4): 325-333.
44. Chen G. P., Li M. J., Xu T. Protection of ultra-high voltage AC-DC grid systems and its key
technologies [J]. Power System and its Automation, 2018, 42(22): 2-10.
45. Effects of Frequency and Voltage on Power System Load [R]. New York: IEEE PES Winter
Meeting, 1966.
46. P. M. Anderson, B. K. Lereverend. industry experience with special protection schemes [J].
IEEE Transactions on Power Systems, 1996, 11(3): 1166-1179.
47. China Electricity Council. 2014 National Electricity Supply and
Demand Situation Analysis and Forecast Report [EB/OL]. http://
www.cec.org.cn/guihuayutongji/gongxufenxi/dianligongxufenxi/2014-02-25/117272.html.
2014-2-25.
48. Lv Weiye. Development of China’s electric power industry and industrial restructuring [J].
China Electric Power, 2002, 35(1):1–7.
49. Liu Z.Y., Zhang Q.P.. Research on national grid development model [J]. Chinese Journal of
Electrical Engineering, 2013, 33(7): 1-10.
50. Zhang Yunzhou, Li Hui. A strategic discussion on the development of ultra-high voltage
power grid in China [J]. Chinese Journal of Electrical Engineering, 2009, (22): 1–7.
51. Shang Chun. Development and application of ultra-high voltage transmission technology in
southern power grid [J]. High Voltage Technology, 2006, 32(1): 35-37.
52. Yin Y. H. Study on the development planning of extra-high voltage power grids [J]. Power
Grid and Clean Energy, 2009, 25(10):1-3.
53. Ding Daoqi. In-depth study of complex grid dynamic behavior characteristics to construct
China’s ultra-high voltage grid safety guarantee [J]. China Electric Power, 2008, 41(8): 1-7.
54. He He Jingbo, Lun Tao, Chen Gang, et al. Analysis of tidal currents and voltage fluctuation
characteristics of extra-high voltage AC contact lines [J]. Power Grid Technology, 2012, 36(9):
56–60.
55. Power System Relaying Committee. Application of Overreaching Distance Relays [EB/OL].
http://www.pes-psrc.org/Reports/D4_Application_of_Overreaching_Distance%20_Relays.
pdf. 2010-3-28.
56. Cao Y, Guo J-B, Mei S-W, et al. A system complexity theory study for large grid security
assessment [M]. Beijing: Tsinghua University Press, 2010.
57. Zhao L, Liu FL, Chen T, et al. Safety and stability characteristics and defense countermeasures
of urban power grids at the receiving end on UHV transmission channels [J]. High Voltage
Technology, 2012, 38 (12): 3304-3309.
58. Yang Dong. Research on the optimization of UHV transmission grid structure and future grid
structure morphology [D]. Jinan: Shandong University, 2013.
59. State Council. Regulations on Emergency Response and Investigation of Electricity Safety
Accidents [S]. State Council Decree No. 599, 2011.
60. Technical Group of the Accident Investigation Team, Huazhong Power Grid. Technical report
on the investigation report of the "7.1" accident in Central China Power Grid [R]. 2007.
61. UCTE. Final Report on System Disturbance on 4 November 2006[R]. 2007.
62. CIGRE Task Force. Defence Plan Against Extreme Contingencies [R]. 2007.
63. APOSTOLOV A P. Distance relays operation during the August 2003 North American
Blackout and methods for improvement[C]//2005 IEEE Russia Power Tech, 2005.
316 Bibliography

64. Wang, Meiyi. Large grid accident analysis and technology application [M]. Beijing: China
Electric Power Press, 2008.
65. W. R. Lachs. Controlling Grid Integrity after Power System Emergencies [J]. IEEE Power
Engineering Review, 2002, 22(2): 61-62.
66. Ma R, Tao JN, Xu HM. Risk assessment of power system chain tripping based on tidal shift
factor [J]. Power System Automation, 2008, 32(12): 17-21.
67. W. R. Lachs. A New Horizon for System Protection Systems Schemes [J]. IEEE Power
Engineering Review, 2002, 22(11): 59.
68. Huanzhang Liu, Zexin Zhou. Strategies for line distance protection against accidental overload
[J]. Chinese Journal of Electrical Engineering, 2011, 31(25): 112-117.
69. Ho, D. Y.. Subsequent awareness of two major accidents in the western U.S. system 1996
(layered analysis) [J]. China Electric Power, 1998, 31(5): 37–40.
70. Xue Yusheng. A framework for spatio-temporally coordinated major outage defense (I) from
isolated defense to integrated defense [J]. Power System Automation, 2006, 30(1): 8-16.
71. A. G. Phadke, J. S. Thorp. Expose Hidden Failures to Prevent Cascading Outages [J]. IEEE
Computer Applications in Power, 1996, 9(3): 20-23.
72. Zhang HH, Su SH, Hu Y. Calculation of grounding distance protection rectification for 750kV
co-tower double-return line [J]. Power System Automation, 2009, 33(22): 102-105.
73. S. H. Horowitz, A. G. Phadke. third Zone Revisited [J]. IEEE Transactions on Power Delivery,
2006, 21(1): 23-29.
74. Yan Hongquan, Yan Xiaoding, Ren Zuyi. Microcomputer-type transformer overload inter-
connection device and its application [J]. Power System Automation, 2004, 28(16): 86–87.
75. Yang Fugang. Application of transformer overload linkage cutting device in Nanning power
grid [J]. Guangxi Electric Power, 2008, (6): 52-54.
76. Fan Shouzhong. Implementation of standby overload interconnection function [J]. Power
System Protection and Control, 2010, 38(5): 139-140.
77. Yuan Jixiu. Safety and stability control of power systems [M]. China Electric Power Press,
1996.
78. Xue Y. S., Wang D., Wen F. Shuan. A review on optimization and coordination of emergency
control and correction control [J]. Power System Automation, 2009, 33(12): 1–7.
79. Buldyrev S V, Parshani R, Paul G, et al. Catastrophic cascade of failures in interdependent
networks [J]. Nature, 2010, 464(7291): 1025-1028.
80. Yeh H.S., Gong D., Huang W.C., et al. Feasibility study and engineering implementation of
increasing allowable conductor temperature [J]. Power Construction, 2004, 25(9): 1-7.
81. Murray W. Davis. A new thermal rating approach: The Real Time Thermal Rating System
for Strategic Overhead Conductor Transmission Lines -- Part II: Steady State Thermal Rating
Program[J]. IEEE Transactions on Power Apparatus and Systems, 1977, 96(3): 810-825.
82. D. A. Douglass, D. C. Lawry, A. A. Edris, et al. Dynamic Thermal Ratings Realize Circuit
Load Limits [J]. IEEE Computer Applications in Power, 2000, 13(1): 38-44.
83. Xu Qingsong, Ji Hongxian, Hou Wei, et al. New technology for monitoring conductor temper-
ature to realize capacity increase of transmission lines [J]. Power Grid Technology, 2006,
30(S1): 171–176.
84. S. Maslennikov, E. Litvinov. Adaptive Emergency Transmission Rates in Power System and
Market Operation [J]. IEEE Transactions on Power Systems, 2009, 24(2): 923-929.
85. ISO New England. Procedures for Determining and Implementing
Transmission Facility Ratings in New England [EB/OL]. www.iso-
ne.com/rules_proceds/isone_plan/pp07/pp7_r3.pdf. 2007-2-27
86. Peng Xiangyang, Zhou Huamin. Feasibility study of short-time overload operation of overhead
transmission lines under emergency conditions [J]. Guangdong Electric Power, 2012, 25(6):
24-29.
87. H. A. Darwish, A. I. Taalab, H. Assal. A Novel Overcurrent Relay with Universal
Characteristics[C]//Atlanta, GA, 2001.
88. G. W. Heumann. Overload Relays and Circuit Breakers for Protecting Motorized Appliances
and Their Branch Circuits [J]. Electrical Engineering, 1953, 72(12): 1056-1060.
Bibliography 317

89. S. E. Zocholl, G. Benmouyal. on the Protection of Thermal Processes [J]. IEEE Transactions
on Power Delivery, 2005, 20(2): 1240-1246.
90. D. S. Baker. Generator Backup Overcurrent Protection [J]. IEEE Transactions on Industry
Applications, 1982, IA-18(6): 632–640.
91. G. Swift. Adaptive Transformer Thermal Overload Protection [J]. IEEE Power Engineering
Review, 2001, 21(8): 60.
92. Zhang Pinjia, Du Yi, T. G. Habetler. A Transfer-Function-Based Thermal Model Reduction
Study for Induction Machine Thermal Overload Protective Relays[J]. IEEE Transactions on
Industry Applications, 2010, 46(5): 1919–1926.
93. D. L. Ransom, R. Hamilton. Extending Motor Life With Updated Thermal Model Overload
Protection [J]. IEEE Transactions onIndustry Applications, 2013, 49(6): 2471-2477.
94. Liu Kun. Research on tide-shifting overload identification and protection measures for power
grids based on local information [D]. Beijing: Tsinghua University, 2011.
95. Zhou Zexin, Wang Xingguo, Du Dingxiang, et al. Protection and stability control coordination
strategy under overload condition [J]. Chinese Journal of Electrical Engineering, 2013, 33(28):
146–153.
96. Y. Serizawa, M. Myoujin, K. Kitamura, et al. Wide-area current differential backup protection
employing broadband communications and time transfer systems[J]. IEEE Transactions on
Power Delivery, 1998, 13(4): 1046-1052.
97. Ma J, Li JL, Ye DH, et al. A new principle of wide-area backup protection based on fault
matching degree [J]. Power System Automation, 2010, 34(20): 55-59.
98. He Zhiqin, Zhang Zhe, Chen Wei, et al. Wide-Area Backup Protection Algorithm Based on
Fault Component Voltage Distribution[J]. IEEE Transactions on Power Delivery, 2011, 26(4):
2752–2760.
99. P. Kundu, A. K. Pradhan. synchrophasor-assisted zone 3 operation [J]. IEEE Transactions on
Power Delivery, 2014, 29(2): 660-667.
100. P. V. Navalkar, S. A. Soman. Secure Remote Backup Protection of Transmission Lines Using
Synchrophasors [J]. IEEE Transactions on Power Delivery, 2011, 26(1): 87-96.
101. K. Kangvansaichol, P. A. Crossley. Multi-zone current differential protection for transmission
networks [C]// 2003 IEEE PES Transmission and Distribution Conference and Exposition
(IEEE Cat. No. 03CH37495), 2003.
102. Lv Y, Zhang B, Wu W Chuan. Wide-area relay protection algorithm based on augmented state
estimation [J]. Power System Automation, 2008, 32(12): 12–16.
103. Cong W, Pan Zengcun, Zhao J. Research on wide-area relay protection algorithm based on
the principle of longitudinal comparison [J]. Chinese Journal of Electrical Engineering, 2006,
26(21): 8–14.
104. Lin Xiangning, Li Zhengtian, Wu Kecheng, et al. Principles and Implementations of Hierar-
chical Region Defensive Systems of Power Grid[J]. IEEE Transactions on Power Delivery,
2009, 24(1): 30–37.
105. Wang Bin, Dong Xinzhou, Bo Zhiqian, et al. Negative-Sequence Pilot Protection With Appli-
cations in Open-Phase Transmission Lines[J]. IEEE Transactions on Power Delivery, 2010,
25(3): 1306–1313.
106. J. C. Tan, P. A. Crossley, D. Kirschen, et al. An Expert System for the Back-Up Protection of
a Transmission Network [J]. IEEE Transactions onPower Delivery, 2000, 15(2): 508-514.
107. Yang Wang, Xianggen Yin, Yijun Zhao, et al. Genetic algorithm-based intelligent protection
for regional power grids [J]. Power System Automation, 2008, 32 (17): 40-44.
108. Zhou Shu, Wang Xiaoru, Qian Qingquan. Bayesian network fault diagnosis method in power
system wide area backup protection [J]. Power System Automation, 2010, 34 (4): 44-48.
109. Zhou Liangcai, Zhang Baohui, Bo Zhiqian. Adaptive tripping strategy for wide area backup
protection system [J]. Power System Automation, 2011, 35(1): 55–60+65.
110. T. S. Sidhu, D. S. Baltazar, R. M. Palomino, et al. A New Approach for Calculating
Zone-2 Setting of Distance Relays And Its Use in an Adaptive Protection System[J]. IEEE
Transactions on Power Delivery, 2004, 19(1): 70-77.
318 Bibliography

111. J. Yang, W. Li, T. Chen, et al. Online Estimation and Application of Power Grid Impedance
Matrices Based on Synchronised Phasor Measurements [J]. IET Generation, Transmission &
Distribution, 2010, 4(9): 1052-1059.
112. Bi Zhaodong, Wang Ning, Xia Yanhui, et al. On-line calibration system for relay protection
calibration based on dynamic short-circuit current calculation [J]. Power System Automation,
2012, 36(7): 81–85.
113. Lv Y, Sun HB, Zhang BM, et al. Development of an intelligent early warning system for
online relay protection [J]. Power System Automation, 2006, 30 (4): 1-5.
114. E. Orduna, F. Garces, E. Handschin. Algorithmic-Knowledge-Based Adaptive Coordination
in Transmission Protection [J]. IEEE Transactions on Power Delivery, 2003, 18(1): 61-65.
115. Ma J, Ye D, Wang T, et al. Minimum breakpoint set calculation for complex ring networks
in multiple regions of power systems [J]. Chinese Journal of Electrical Engineering, 2011,
31(28): 104-111.
116. Lin Xianing, Xia Wenlong, Xiong Wei, et al. Research on adaptive regulation of distance
backup protection action characteristics independent of tidal shift [J]. Chinese Journal of
Electrical Engineering, 2011, 31 (S1): 83-87.
117. Xu Huiming. Research on wide area backup protection with identifiable tidal shift and its
control strategy [D]. Beijing: North China Electric Power University (Beijing), 2007.
118. Xu Y, Lv B, Lin X-T. Research and analysis of tidal shift identification methods [J]. Power
Grid Technology, 2013, 37(2): 411-416.
119. Lim Seong-Il, Liu Chen-Ching, Lee Seung-Jae, et al. Blocking of Zone 3 Relays to Prevent
Cascaded Events [J]. IEEE Transactions on Power Systems, 2008, 23(2): 747–754.
120. Ming Jin, Tarlochan S. Sidhu. Adaptive Load Encroachment Prevention Scheme for Distance
Protection [J]. Electric Power Systems Research, 2008, 78(10): 1693–1700.
121. Yang Wenhui. Research on critical line backup protection and emergency control strategy to
prevent chain tripping [D]. Beijing: North China Electric Power University, 2012.
122. Xu Y, Lv B, Wang Zengping. A tidal shift identification method based on wide area
measurement system [J]. Chinese Journal of Electrical Engineering, 2013, 33(28): 154–160.
123. Zhang T S, Luo Chenglian, Du Ling, et al. Analysis of the performance of quadrilateral
characteristic distance protection for load avoidance [J]. Relays, 2004, 32(1): 28–31.
124. D. Novosel, G. Bartok, G. Henneberg, et al. IEEE PSRC Report on Performance of Relaying
During Wide-Area Stressed Conditions [J]. IEEE Transactions on Power Delivery, 2010,
25(1): 3-16.
125. Gao X, Xu Guixian, Guo Dengfeng, et al. Method of load-limiting resistor rectification for
heavy-current lines in North China Power Grid [J]. Power System Automation, 2007, 31(5):
94–96.
126. Iii Edmund O. Schweitzer. Distance relay with load encroachment protection for use with
power transmission lines [P]. 1994.
127. Zhu Xiaotong, Zhao Qingqing, Li Yuanyuan, et al. A new method to prevent misoperation
of phase distance III section protection during overload [J]. Power System Protection and
Control, 2011, 39(9): 7–11.
128. A. Apostolov, B. Vandiver. Functional testing of IEC 61850 based IEDs and systems [C]//IEEE
PES Power Systems Conference and Exposition, 2004.
129. G. D. Rockefeller. Fault Protection with a Digital Computer [J]. IEEE Transactions on Power
Apparatus and Systems, 1969, PAS-88(4): 438–464.
130. Yi YH, Cao Y, Zhang JJ, et al. A new centralized IED based on IEC 61850 standard [J]. Power
System Automation, 2008, 32(12): 36-40.
131. Liu Dongchao, Wang Kaiyu, Hu Shaogang, et al. Centralized protection based on digital
substation [J]. Power Automation Equipment, 2012, 32(4): 117–121.
132. Liu Yiqing. Research on the principle and implementation technology of station domain
backup protection of intelligent substation [D]. Jinan: Shandong University, 2012.
133. Xiong Jian, Liu Chenxin, Deng Feng. Centralized protection measurement and control device
for intelligent substation [J]. Power System Automation, 2013, 37(12): 100-103.
Bibliography 319

134. Gao Houlei, Liu Yiqing, Su Jianjun, et al. Research on new station domain backup protection
for intelligent substations [J]. Power System Protection and Control, 2013, 41(2): 32–38.
135. Song Xuankun, Li Yingchao, Li Jun, et al. A new generation of intelligent substation
hierarchical protection system [J]. Power Construction, 2013, 34(7): 24–29.
136. Wang Yue. Research on hierarchical protection system based on intelligent substation [J].
North China Power Technology, 2013, (9): 26-30.
137. Dong XZ, Ding L. Research on digital integrated protection and control system structure
design scheme [J]. Power System Protection and Control, 2009, 37(1): 1-5.
138. Liu Kun, Dong Xinzhou, Wang Bin, et al. Analysis of the impact of fault dynamic processes
on tidal shift identification [J]. Power System Automation, 2011, 35(13): 31-36.
139. Liu Kun, Dong Xinzhou, Wang Bin, et al. Tidal shift identification based on local information
[J]. Power System Automation, 2011, 35(14): 80-86.
140. Bo Ren. Research on wide-area backup protection based on tidal shift identification [D].
Baoding: North China Electric Power University (Hebei), 2007.
141. Xu Chengbin, Sun Yimin. Digital substation process layer GOOSE communication scheme
[J]. Power System Automation, 2007, 31(19): 91–94.
142. Kanabar M G, Sidhu T S. Performance of IEC 61850-9-2 Process Bus and Corrective Measure
for Digital Relaying [J]. IEEE Transactions on Power Delivery, 2011, 26(2): 725-735.
143. Hu Zhongshan. Research on relay protection networking scheme based on station domain
information [D]. Beijing: North China Electric Power University, 2012.
144. Cao RunBin, Dong Xinzhou, Wang Bin, et al. Discussion of Protection and Cascading
Outages from the Viewpoint of Communication[C]//Beijing: 2011 International Conference
on Advanced Power System Automation and Protection, 2011.
145. Commission of the European Communities. Undergrounding of Electricity Lines in Europe
[R]. 2004.
146. V. T. Morgan. the Loss of Tensile Strength of Hard-Drawn Conductors by Annealing in
Servicec [J]. IEEE Transactions on Power Apparatus and Systems, 1979, PAS-98(3): 700–709.
147. Ye Hong-sheng. Discussion on the calculation of the load capacity of high-voltage transmis-
sion line conductors [J]. Power Construction, 2000, (12): 23–26.
148. Yeh H.S., Gong David, Huang W.C.. Research on increasing the allowable conductor temper-
ature to increase line transmission capacity and its application on 500 kV lines [J]. East China
Electric Power, 2006, 34(8): 43–46.
149. J. R. Harvey. Effect of Elevated Temperature Operation on the Strength of Aluminum
Conductors [J]. IEEE Transactions on Power Apparatus and Systems, 1972, PAS-91(5):
1769–1772.
150. Li Bozhi. Treatment of plastic creep elongation of overhead lines [J]. Power Construction,
2001, 22(6): 20–25.
151. J. R. Harvey. creep of Transmission Line Conductors [J]. IEEE Transactions on Power
Apparatus and Systems, 1969, PAS-88(4): 281–286.
152. Yuan Y. Y., Sun T. X. Factors affecting the allowable load capacity of steel-core aluminum
stranded wire [J]. Power Safety Technology, 2001, 3(5): 19-21.
153. PJM Interconnection. dms #590159. guide for Determination of Bare Overhead Transmission
Conductors [P]. 2010.
154. F. Massaro, L. Dusonchet. risk evaluation and creep in conventional conductors caused by high
temperature operation [C]//Padova: 2008 43rd International Universities Power Engineering
Conference, 2008.
155. Liao Tianquan. Research on improving transmission line transmission capacity [D]. Chengdu:
Xihua University, 2009.
156. F. Jakl, A. Jakl. Effect of Elevated Temperatures On Mechanical Properties of Overhead
Conductors under Steady State and Short-Circuit Conditions[J]. IEEE Transactions on Power
Delivery, 2000, 15(1): 242-246.
157. A. H. Bingham, F. C. Lambert, M. R. Monashkin, et al. An Accelerated Performance Test of
Electrical Connectors [J]. IEEE Transactions on Power Delivery, 1988, 3(2): 762-768.
320 Bibliography

158. Liu Changqing, Liu Shengchun, Chen Yongzhao, et al. Experimental study on increasing the
allowable temperature of wire heat generation [J]. Power Construction, 2003, 24(8): 24-26.
159. Cigre, Cigre TB 207-2002. thermal Behaviour of Overhead Conductors [S]. 2002.
160. L. Staszewski, W. Rebizant. The Differences Between IEEE and CIGRE Heat Balance
Concepts for Line Ampacity Considerations [C]//Wroclaw: 2010 Modern electric power
systems, 2010.
161. Zhang Hui. Research on thermal load capacity of transmission lines under operating conditions
[D]. Jinan: Shandong University, 2008.
162. Jaeger H. The "Echo State" Approach to Analyzing and Training Recurrent Neural Networks
[R]. 2001
163. Yin, Guotao. Research on the evaluation method of current-carrying capacity of transmission
lines based on temperature detection [D]. Chongqing: Chongqing University, 2011.
164. Feng Chen. Research on network traffic prediction algorithm based on ESN [D]. Beijing:
Beijing University of Posts and Telecommunications, 2013.
165. Fei Yang. Echo state network-based traffic flow prediction model and its related research [D].
Beijing: Beijing University of Posts and Telecommunications, 2012.
166. Ning, Meifeng. Research on wind speed and wind power short-term prediction methods [D].
Zhengzhou: Zhengzhou University, 2012.
167. O. Obst, X. R. Wang, M. Prokopenko. Using Echo State Networks for Anomaly Detection in
Underground Coal Mines [C]//St. Louis, MO: 2008 International Conference on Information
Processing in Sensor Networks (ipsn 2008), 2008.
168. Jaeger H. Tutorial on Training Recurrent Neural Networks, Covering BPTT, RTRL, EKF, and
the "Echo State Network" Approach [R]. 2002.
169. E. I. Koufakis, P. T. Tsarabaris, J. S. Katsanis, et al. A Wildfire Model for the Estimation of the
Temperature Rise of an Overhead Line Conductor[J]. IEEE Transactions on Power Delivery,
2010, 25(2): 1077-1082.
170. Hu, Liang. Research on object-oriented communication system for power dispatch automation
[D]. Chengdu: Southwest Jiaotong University, 2007.
171. Zhang, B. M.. Analysis of higher power networks [M]. Beijing: Tsinghua University Press,
2007. 335.
172. C. D. Vournas, A. Metsiou, M. Kotlida, et al. Comparison and Combination of Emergency
Control Methods for Voltage Stability [C]//Denver: IEEE Power Engineering Society General
Meeting, 2004.
173. National Electricity Dispatch and Communication Center. Relay protection training materials
of the State Grid Corporation (upper volume) [M]. Beijing: China Electric Power Publishing
House, 2009.
174. K. Vu, M. M. Begovic, D. Novosel, et al. Use of Local Measurements to Estimate Voltage-
Stability Margin [J]. IEEE Transactions on Power Systems, 1997, 14(3): 1029-1035.
175. Li, L.F., Liu, C.. Emergency situation analysis based on Davinan equivalence parameters [J].
Power Grid Technology, 2008, 32(21): 63-67.
176. M. Parniani, M. Vanouni. a Fast Local Index for Online Estimation of Closeness to Loadability
Limit [J]. IEEE Transactions on Power Systems, 2010, 25(1): 584-585.
177. M. Jonsson, J. Daalder. an Adaptive Scheme to Prevent Undesirable Distance Protection
Operation during Voltage Instability [J]. IEEE Power Engineering Review, 2002, 22(11): 61.
178. A. F. Bin Abidin, A. Mohamed. On the Use of Voltage Stability Index to Prevent Undesirable
Distance Relay Operation During Voltage Instability [C]/ /Prague, Czech Republic: 2010 9th
International Conference on Environment and Electrical Engineering, 2010.
179. Shen Gang, V. Ajjarapu. A Novel Algorithm Incorporating System Status to Prevent Undesir-
able Protection Operation during Voltage Instability [C ]//Las Cruces, NM: 2007 39th North
American Power Symposium, 2007.
180. Chen Deshu. Research on power system relay protection [M]. Wuhan: Huazhong University
of Science and Technology Press, 2011. 312.
181. China Electric Power Research Institute. 2012 State Grid relay protection equipment analysis
and assessment report [R]. 2013.
Bibliography 321

182. Zhu Sheng Shi. Principles and techniques of relay protection for high-voltage power grids
[M]. Beijing: China Electric Power Publishing House, 2005. 330.
183. Wu, A. P.. Determination of fault excision time for 500 kV lines in stability calculation [J].
Sichuan Electric Power Technology, 2008, 31(5): 7-8.
184. Tang Chung, Sun Huadong, Yi Jun, et al. A full differential based Davinan equivalent
parameter tracking algorithm [J]. Chinese Journal of Electrical Engineering, 2009, 29(13):
48–53.
185. Wang, M. N.. Research on the improvement algorithm and application of Davinan equivalence
based on measurement data [D]. Beijing: North China Electric Power University, 2012.
186. Liao G., Wang X. R.. Uncertainty model for identification of Davinan equivalent parameters
in power systems [J]. Chinese Journal of Electrical Engineering, 2008, 28(28): 74-79.
187. Li L.F., Yu J.L., Liu C.C.. Study on parameter drift problem of Davinan equivalence tracking
[J]. Chinese Journal of Electrical Engineering, 2005, 25(20): 1-5.
188. Li Juan, Liu Xiu Kuan, Cao Guochen, et al. A study on a node-oriented method for tracking
estimation of grid equivalent parameters [J]. Chinese Journal of Electrical Engineering, 2003,
23(3): 30–33.
189. Wang Qian. Research on a new tri-polarized volume distance relay [D]. Jinan: Shandong
University, 2006.
190. Li Yan, Chen Deshu, Yin Xiangen, et al. Research on a new adaptive m-ohm relay [J]. Chinese
Journal of Electrical Engineering, 2003, 23(1): 81–84.
191. M. M. Eissa. Ground Distance Relay Compensation Based on Fault Resistance Calculation
[J]. IEEE Transactions on Power Delivery, 2006, 21(4): 1830-1835.
192. Dong XZ, Su B, Bo ZQ, et al. Research on special problems of relay protection for extra-high
voltage transmission lines [J]. Power System Automation, 2004, 28(22): 19-22.
193. Zhang Y. Operation and calibration of line relay protection with series complementary
capacitors [J]. Southern Power Grid Technology, 2008„ 2(1): 75-79.
194. Dong XZ, Wang B, Cao RB, et al. A fast start-up and phase calculation method for integrated
protection of intelligent substation [P]. CN102761106A, 2012.
195. J. S. Thorp A. G. Phadke. Synchronized Phasor Measurements and Their Applications [M].
New York: Springer, 2008.
196. Li Xia Yang. Research on the implementation of cross-differential protection for 500kV
co-pole double-return line [D]. Chongqing: Chongqing University, 2008.
197. K. P. Brand, M. Ostertag, W. Wimmer. Safety Related, Distributed Functions in Substations
and the Standard IEC 61850 [C]//2003 IEEE Bologna Power Tech Conference Proceedings,
2003.
198. Huang Lei. Research and implementation of intelligent switchgear controller for intelligent
substation [D]. Nanjing: Nanjing University of Technology, 2013.
199. UCA International Users Group. Implementation Guideline for Digital Interface to Instrument
Transformers Using IEC 61850-9-2 [R]. 2006
200. Zhou Yiran. Research and hardware and software implementation of 35kV intelligent
substation merging unit [D]. Nanjing: Nanjing University of Technology, 2013.
201. Huanzhang Liu, Zexin Zhou, Delin Wang, et al. Principles of distance protection with the
ability to cope with overloads [J]. Power Grid Technology, 2014, 38(11): 2943-2947.
202. Zhou Zexin, Liu Huanzhang, Wang Delin, et al. A distance protection implementation scheme
with the ability to cope with overloads [J]. Power Grid Technology, 2014, 38(11): 2948-2954.
203. Yan X.W., Gao W., Zhang S.S., et al. A new method for locating single-phase disconnec-
tion zones in distribution networks based on phase current characteristics [J]. Electrical
Measurement and Instrumentation, 2019, 56(03): 76-81.
204. Chang Zhongxue, Song Guobing, Huang Wei, et al. A method for locating single-phase ground
fault zones in distribution networks based on the characteristics of phase voltage and current
abrupt changes [J]. Power Grid Technology, 2017, 41(07): 2363-2370.
205. Xu Zhengya. A new type of distance protection for transmission lines [M]. Beijing: China
Water Conservancy and Hydropower Press, 2002.
322 Bibliography

206. S. M. Hashemi, M. Sanaye-Pasand, Distance protection during asymmetrical power swings:

challenges and solutions [J]. IEEE Transactions on Power Delivery, 2018, 33(6): 2736-2745.
207. Z. D. Gao, G. B. Wang, A new power swing block in distance protection based on
a microcomputer-principle and performance analysis [C]// International Conference on
Advances in Power System Control, Operation and Management, APSCOM-91., 1991.
843–847.
208. B. Su, X. Z. Dong, Z. Q. Bo, et al. Fast detector of symmetrical fault during power swing for
distance relay [C]//IEEE Power Engineering Society General Meeting, 2005. 1836–1841.
209. Gao Houlei, Wen Feng, Wang Guangyan, et al. Numerical analysis of the operational
performance of two multi-phase compensated impedance relays [J]. Relays, 1995, (3): 3-9.
210. Huanzhang Liu. Analysis of δ-R characteristics of multi-phase compensated distance relays
[J]. Relays, 1990, (01): 23-31.
211. Li, X.. A new grounded multi-phase compensated impedance relay [J]. Relay, 1987,(1):21-26.
212. Yu, Ming-Yuan. Analysis and calculation of action characteristics of multi-phase compensated
ground distance relays [J]. Power System Automation, 1981, (6): 16-38.
213. Zhu, Sheng-Shi. Multi-phase compensated phase-to-phase distance relays [J]. Power System
Automation, 1978, (4): 1-26.
214. Qingkai Jiang. Analysis of a multi-phase compensated impedance relay unaffected by system
oscillations [D]. Beijing: North China Electric Power University, 2011.
215. Jiang Qingkai, Huang Shaofeng, Zhang Hongyan, et al. Analysis of a multi-phase compen-
sated distance relay [C]// Annual academic conference of Chinese higher education institutions
for power systems and their automation and annual meeting of the power system committee
of Chinese electrical engineering society, 2010.
216. Huang, Shao-Feng, Jiang, Qing-Kai. Analysis and countermeasures of the effect of transition
resistance on multi-phase compensation impedance elements [J]. Power Grid Technology,
2011, 35(8): 202-206.
217. Zhu Singshi, Cui Liu, Dong Xinzhou. Distance protection unaffected by power system
oscillations [J]. Chinese Journal of Electrical Engineering, 2014, 34(07): 1175-1182.
218. Cui Liu, Dong Xinzhou, Shi Shenxing, et al. Performance analysis of multi-phase compen-
sating distance relays under oscillation and accompanied by single-phase ground fault [J].
Chinese Journal of Electrical Engineering, 2013, 33(34): 214-222.
219. Liu Huanzhang, Zhou Zexin, Zhou Chunxia, et al. Improvement measures of relay protection
oscillation blocking [J]. Chinese Journal of Electrical Engineering, 2012, (19): 125-133.
220. Cui Liu, Dong Xinzhou. Multiphase compensated distance relay with transition resistance
resistance [J]. Chinese Journal of Electrical Engineering, 2014, 34(19): 3220-3225.
221. Liu Z.Y., Qin X.F., Zhao L., et al. Study on the application of UHV DC hierarchical access
method in multi-feeder DC networks[J]. Chinese Journal of Electrical Engineering, 2013,
33(10): 1-7.
222. Liu C, Zhang B, Hou Y, et al. An improved approach for AC-DC power flow calculation with
multi-infeed DC systems[J]. IEEE Transactions on Power Systems, 2011, 26(2): 862-869.
223. Rao H, Xu S, Zhao Y, et al. Research and application of multiple STATCOMs to improve
the stability of AC/DC power systems in China Southern Grid [J]. IET Generation,
Transmission & Distribution, 2016, 10(13): 3111-3118.
224. Shao Yao, Tang Chung, Guo Xiaojiang, et al. Analysis of phase change failure of multi-feeder
DC transmission systems in North and East China of the 2015 UHV planning grid[J]. Power
Grid Technology, 2011, 35(10): 9-15.
225. Li Shilin. Analysis of the current situation and development of smart grid technology [J].
Technology and Innovation, 2021(03): 28–29+32.
226. Han MX, Wen J, Xu YH. Principle and operation of high-voltage DC power transmission
[M]. Beijing: Mechanical Industry Press, 2020: 1.
227. Dong Xinzhou, Tang Chong, Bu Guangquan, Shen Shen, Song Guobing, Wang Zengping, Gan
Deqiang, Hou Junxian, Wang Bin, Zhao Bing, Shi Shenxing. Problems and challenges for safe
operation of large AC-DC hybrid power grids[J]. Chinese Journal of Electrical Engineering,
2019, 39(11): 3107–3119.
Bibliography 323

228. Yao Liangzhong, Wu Jing, Wang Zhibing, Li Yan, Lu Zongsang. Analysis of the future
development pattern of high-voltage DC power networks[J]. Chinese Journal of Electrical
Engineering, 2014, 34(34): 6007–6020.
229. Jing Liuming, Wang Bin, Dong Xinzhou, et al. Research on the suppression method of high-
voltage DC continuous phase change failure caused by AC filter throwing [C], The 34th
Annual Academic Conference of Chinese Higher Education Institution of Power System and
its Automation, , Taiyuan, 2018.
230. Li X.-Y., Chen S.-Y., Pang G.-H., et al. Optimization of phase change failure prevention
and automatic recovery capability for multiple DC feed-in systems in East China [J]. Power
System Automation, 2015, 39(06): 134-140.
231. DC Transmission Research Group, Power Generation Teaching and Research Group, Zhejiang
University. Direct current transmission [M]. Beijing: Water Resources and Electricity
Publishing House, 1985.
232. Jing, Liu-Ming, Wang, Bin, Dong, Xin-Zhou, Wang, Harbour, Yu, Bin, Xie, Min, Chen, Shi.
A review of research on continuous phase change failure in high-voltage DC transmission
systems [J]. Power Automation Equipment, 2019, 39(09): 116-123.
233. Liu Xiyang, Wang Zengping, Qiao Xin, Zheng Bowen. A review of research on classification
and suppression measures for phase change failure in AC-DC hybrid grids [J]. Smart Power,
2020, 48(06): 1–7+32.
234. Tao Yu. Analysis and application of DC transmission control and protection system [M].
Beijing: China Electric Power Press, 2015: 155–170.
235. Zhao Wanjun. High voltage DC transmission engineering technology [M]. 2 ed. Beijing:
China Electric Power Press, 2011: 124.
236. Lin Lingxue, Zhang Yao, Zhong Qing, Zong Xiuhong. A review of research on phase change
failure in multi-feeder DC transmission systems [J]. Power Grid Technology, 2006(17): 40-46.
237. Song JZ, Li YL, Zeng L, Zhang YK. A review of research on phase change failure in high-
voltage DC transmission systems [J]. Power System Automation, 2020, 44(22): 2-13.
238. Davies B, Williamson A, Gole A M, et al. Systems with multiple DC infeed [R]. Paris: CIGRE
Working Group B4.41, 2008.
239. Aik D L H, Andersson G. Analysis of voltage and power interactions in multi-infeed HVDC
systems[J]. IEEE Transactions on Power Delivery, 2013, 28(2):816-824.
240. Zhang W.C., Xiong Y.X., Li C.H., Yao W., Wen J.Y., Gao D.X.. Improved VDCOL-based
successive phase change failure suppression and coordinated recovery for multi-feeder DC
systems [J]. Power System Protection and Control, 2020, 48(13): 63-72.
241. Ouyang JX, Ye JJ, Zhang Z, Xiao Chao. Mechanisms and characteristics of successive phase
change failure in multi-feeder DC transmission systems under grid faults[J]. Power System
Automation, 2021, 45(20): 93–102.
242. Yin Chunya, Li Fengting, Song Xinfu, Wan Qi. A fast discrimination method for phase change
failure of multi-feeder DC systems[J]. Power Grid Technology,2019,43(10):3459–3465.
243. Zhang Zhengwei, Chen Dezhi, Bu Guangquan, Guo Jingyi, Xu Zhanke, Song Yunting, Ji Ping,
Li Lixin, Wang Qing. Research on safety and stability control measures for multi-DC-fed EHV
ring network [J]. Power System Protection and Control, 2019, 47(19): 46–53.
244. Zhang G. F., Li C. C., Wang B., et al. Adaboost-based warning method for continuous phase
change failure of high-voltage DC lines[J]. Power System Protection and Control, 2019,
47(19):9.
245. Jiang Yefeng, Bao Yanhong, Zhang Jinlong, Cui Zhanfei, Xu Wei, Ren Xiancheng, Chen
Yingjie. Dynamic reactive power standby evaluation to cope with DC continuous phase change
failure [J]. Power System Protection and Control, 2021, 49(19): 173–180.
246. Chao-Ming Zhang. Research on the mechanism of phase change failure and control strategy
of multi-feeder DC transmission system [D]. Southeast University, 2020.
247. Jing, Liu-Ming. Research on continuous phase change failure suppression method for high-
voltage DC transmission based on fixed-off area method. Postdoctoral Discharge Report,
Tsinghua University, 2019.
324 Bibliography

248. Wang, N., Huang, Y. L., Zhou, Q.. Analysis of response strategy and predictive control tech-
nology route for phase change failure in high voltage DC transmission [J]. Power System
Protection and Control, 2014, 42(21): 124-131.
249. Jovcic D, Pahalawaththa N, Zacahir M. Analytical modeling of HVDC-HVAC systems [J].
IEEE Transactions on Power Delivery, 1999, 14(2): 506-511.
250. Cao W.Y., Han M.X., Wen Q., et al. Analysis of the current status of research on the parallel
control technology of AC-DC distribution grid inverters[J]. Journal of Electrical Engineering
Technology, 2019, 34(20): 4226-4241.
251. Mirsaeidi S, Dong X, Said D M. A Fault Current Limiting Approach for Commutation Failure
Prevention in LCC-HVDC Transmission Systems [J]. IEEE Transactions on Power Delivery,
2019: 2018–2027.
252. Son H I, Kim H M. An algorithm for effective mitigation of commutation failure in high voltage
direct current systems [J]. IEEE Transactions on Power Delivery, 2016, 31(4): 1437-1446.
253. Guo C, Liu Y, Zhao C, et al. Power component fault detection method and improved current
order limiter control for commutation failure mitigation in HVDC [J]. IEEE Transactions on
Power Delivery, 2015, 30(3): 1585-1593.
254. Mirsaeidi S, Dong X, Tzelepis D, et al. A predictive control strategy for mitigation of commu-
tation failure in LCC-based HVDC systems [J]. IEEE Transactions on Power Electronics,
2019, 34(1): 160-172.
255. Wei Z, Yuan Y, Lei X, et al. Direct-current predictive control strategy for inhibiting commu-
tation failure in HVDC converter [J]. IEEE Transactions on Power Systems, 2014, 29(5):
2409-2417.
256. Kwon D, Kim Y J, Moon S I. Modeling and analysis of an LCC HVDC system using DC
voltage control to improve transient response and short-term power transfer capability[J].
IEEE Transactions on Power Delivery, 2018, 33(4): 1922-1933.
257. Zou G, Huang Q, Song S, et al. Novel transient-energy-based directional pilot protection
method for HVDC line [J]. Protection and Control of Modern Power Systems, 2017, 2(2):
159-168.
258. Rao YF, Zhang PF, Li C H, et al. Mechanism and evaluation method of the influence of
excitation inrush on phase change failure in high-voltage DC transmission systems[J]. Power
System Protection and Control, 2019, 47(13): 54-61.
259. Ren Xuan, Wang Bin, Yu Bin, Xie Min, Xie Hua, Huang Tao. High resistance ground fault
protection for AC lines in the near zone of LCC-HVDC inverter-side converter stations[J].
Power System Automation, 2021, 45(23): 162–169.