HVDC LCC Reference Book

DC Transmission Systems
LINE COMMUTATED
CONVERTERS
GEGridSolutions.com
DC Transmission Systems: Line Commutated Converters. © GE Vernova, 2023
ISBN: 978-0-9809331-2-3
Forth edition
All rights reserved. No part of this book may be reproducted or transmitted

in any form by means, electronic or mechanical, including
photocopying, recording, or by any information storage and retrieval
system, without permission in writing from the publisher.
GE VERNOVA
The Lord Nelson Building
Redhill Business Park
William Bagnall Drive
Stafford, ST16 1WS
www.GEVernova.com
DC TRANSMISSION SYSTEMS
LINE COMMUTATED
CONVERTERS
FOREWORD
The requirements for power networks are continuously evolving. Modern electricity generation,
transmission and distribution networks demand high security, high efficiency and good network stability.
There is also a shift towards harnessing renewable energy resources. High Voltage Direct Current (HVDC)
Transmission technology offers an effective solution to these aspects. HVDC is seen as an attractive option
due to its low transmission loss over long distances, its ability to inter-connect asynchronous systems
and its fast controllability. It has enabled countries like China, India, the USA and Brazil to harness remote
energy resources and convey the energy to large urban centres. It is increasingly being used in Europe and
other parts of the world for evacuating power from off-shore wind farms. In North America, it has permitted
the interconnection of asynchronous parts of continental networks, which could not have been achievable
with traditional ac transmission.
The use of HVDC transmission is growing rapidly. For today’s engineer to become proficient in this area
requires a knowledge of power engineering, power electronics and control systems. A resource that
consolidates these areas is not easy to come by. This book addresses this important need. An excellent
book on the subject was published about a decade ago by several of the authors of the current book.
However, HVDC is a rapidly growing field and the current book builds on the previous one and includes
significant new material as well as an updating of the old material. In particular there is a comprehensive
coverage of current and past HVDC projects fromGE Vernova (and all former entities including Alstom)
and a modernising of the material on control and protection. The material on valve testing now includes
GE Vernova’s latest VTF (valve test facility) in Stafford, UK. A very useful addition is the inclusion of modern
methods for the planning of HVDC projects.
Indeed, this book is probably the best single knowledge-resource for HVDC transmission and should prove
to be invaluable to the university student, the practicing engineer and the expert alike.
Ani GOLE
Professor Ani Gole is Distinguished Professor and NSERC Industrial Research Chair at the Universty of
Manitoba, Canada. He is an expert on the modelling and simulation of HVDC systems and Flexible ac
transmission systems (FACTS). He was one of the original design team members for the PSCAD/EMTDC
simulation program, which is widely used for simulating HVDC systems. Dr. Gole is a recipient of the IEEE
PES Nari Hingorani FACTS award. He is a Fellow of the IEEE and a Fellow of the Canadian Academy of
Engineering. He is a registered professional engineer in the Province of Manitoba, Canada.
4 | DC Transmission Systems: Line Commutated Converters

The HVDC market continues to grow. The growth is on two fronts, the interconnections between neighbouring
systems and the integration of renewables. It is clear that the two are linked together. The interconnections
between neighbouring systems also facilitate the integration of renewables. For example, if we consider wind
generation, if the generation is not available in a certain area, the fact that the system is interconnected allows
one to count on the renewable generation in another area.
This expansion in the HVDC market brings with it technical as well as human resources challenges. However,
it is clear that again both are linked. In order to solve the technical challenges, we need to have skilled and
experienced engineering resources. Such resources must have the tools as well as practical reference
material to tackle such challenges. Having such reference material enables engineers to convert challenges
to opportunities.
The first edition of this book was published almost ten years ago. The book was unique because it provided
the practicing engineer with practical reference material. It combined both the theory and the experience in
HVDC. Such information is important for the practicing engineers. This new edition has the same approach
to the subject with added information on some of the developments in the area.
Dr. Mohamed Rashwan

President
TransGrid Solutions Inc.
DC Transmission Systems: Line Commutated Converters | 5

HVDC Transmission into the Next Decade
HVDC transmission interconnections are a major part in modern power system worldwide development.
HVDC systems are being planned and implemented to bring renewable energy resources into major
metropolitan areas through ever more reduced and restricted geographical paths which present not only
environmental challenges to permitting, and siting, but are pressing the HVDC development community into
ever more innovative solutions including the substantial growth in capacity of Insulted Gate Bipolar Transistor
(IGBT) based Voltage Source Converters. However, HVDC projects based on conventional thyristor-based
HVDC technology is still the choice of the higher capacity links at longer distances and UHVDC voltage levels.
HVDC transmission systems based on thyristor-based technology still make up the majority of the technology
implemented in the fleet of existing converters, some of which are now requiring replacement and updating
of many of the critical subsystems up to and including complete replacement of major system converters.
Many parts of the world have no significant investment in HVDC technology, however the range of capacities
and technical improvements in converter design, including advancements in thyristor-based systems, have
made HVDC systems attractive to both the existing utility customers and new investor-backed projects. This
heightened interest has resulted in ever more innovative and demanding HVDC projects, project replacements,
and new interconnection opportunities even in areas without significant HVDC transmission systems.
Traditional development of HVDC systems was for long distance transmission up to 1500 km in existing
systems and some interconnections planned to 2000 km and potentially even longer under sea proposed
projects. Not only under sea configurations are installed and in planning but more recently within the USA
and other countries long terrestrial cable systems are being planned to make use of railway right of ways
(ROW) for the development of high capacity HVDC systems into congested urban areas. Multi-terminal HVDC
projects are being planned in several regions of the world with several intensive developments being planned
in Western Europe specifically to take advantage of renewable offshore WTG generation developments with
major exploitations of multiple terminals distributed in several regions within this region. Multi-Terminal HVDC
projects are also being considered for large multi-national interconnections in the Middle East and between
the Central European Regions. Within North America, there has been a renewed interest in Multi -Terminal
HVDC for the development of multiple large-scale WTG offshore developments.
None of these advances in the LCC or VSC applications would be possible without the better system modeling
tools and system simulation utilizing advanced real-time simulation tools now available from a number of
sources. These tools allow very complete initial equipment dimensioning and further within the system design
and testing allow actual simulated system conditions to stress and prove out todays more complex control
hierarchy’s and hardware arrangements.
Resources such as this extensive HVDC reference guide have proven invaluable to the global community
involved in the proliferation and advancement of HVDC technology.
Mark A. Reynolds, P.E.

Senior Project Engineer
Power Engineers, Inc.

PREFACE
The transmission of bulk electricity has revolutionized the way we live and work. From the early days of the
‘battle of the currents’ between Direct Current (DC) and Alternating Current (AC), our industry has been
fuelled by continuous innovation and challenged by ever increasing customer demands. Thomas Edison,
the founder of GE Vernova, was also the leading proponent of DC transmission in the late 19th Century,
and although the battle of the currents was initially won by AC, since the 1950s, High Voltage Direct
Current (HVDC) using Line Commutated Converters (LCC) gradually began to re-establish itself as a niche
application. HVDC interconnections allowed the delivery of bulk power in ever increasing voltage ratings
with reduced losses compared to AC connections and in the last 10 years, HVDC has made the transition
from a niche technology to an important part of mainstream electrical power transmission.
As thyristors (another GE Vernova innovation) started to become available at higher power ratings in
the 1960s, solid state technology began to replace the older mercury arc technology, bringing improved
reliability and higher power ratings to HVDC. GE Vernova and its ancestors have supplied both the
world’s first fully thyristor-based HVDC system (1972) and the last and largest mercury-arc-based system
(completed in 1977) as well as the world’s first multi-terminal HVDC system (1988) and many other
important firsts.
Today, power sector challenges have dramatically changed. Environmental concerns, the skyrocketing cost
of land for transmission lines and substations and the price of the energy itself are making energy managers
and engineers take another look at HVDC transmission. Wider adoption of renewable energy sources bring
new challenges of control and stability to AC grids. Innovations in Direct Current technology, the development
of ever more powerful semiconductor devices, improved mastery of insulating materials and the progress of
AC/DC converter transformers have allowed us to move into transmission ratings that our forefathers would
never have dreamed possible. Our technologies now allow us to truly create the energy highways of the future
with DC transmission voltages as high as 1100 kV. Renewable energies can be connected to the grid with
minimal losses. Bulk power can be transmitted longer distances.
GE Vernova and our ancestor companies have been innovating the HVDC world for over half a century.
Our experience and expertise in this highly complex field is the basis for this book. For future generations
to continue the development, they must understand the roots of history. This book serves not only as a
complete, detailed explanation of the High Voltage Direct Current world – for use by energy managers – but
can be considered an academic reference for students and researchers. The team of authors has covered
all aspects required to plan, specify and build a LCC-HVDC interconnection or transmission scheme. A later
volume will present a similar treatment of the newer Voltage Source Converter (VSC) HVDC technology,
which is starting to revolutionize the power transmission market once again. Special acknowledgement
and thanks must be given to contributors – customers, colleagues and academics - for their valuable input
on subjects where they are considered world authorities. We hope that all our readers will find answers to
their questions; that these answers will drive them to seek new challenges, to push the limits further and
to continue the innovation. It is now with great pleasure that I introduce you to our world: High Voltage
Direct Current.
Thomas Bjork
CTO, Grid Systems Integration
GE Vernova

AUTHORS
This book is the result of a collaborative effort by many technical experts within GE Vernova. This team of
dedicated professionals have shown impressive team work and the willingness to share their collective
experience and expertise.
Carl BARKER Alban JENKINS

Colin DAVIDSON David EBOCKAYUK
Jurek GOLD Bruno KAYIBABU
Neil KIRBY Mike LI
Norman MacLEOD Brian MATTHEWS
Ian McCONNACHIE Gaurav MENDIRATTA
José MONTEIRO Radnya MUKHEDKAR
Gearoid Oh’EIDHIN Shaukat SADULLAH
David STEVENSON Tim STOTT
Robert WHITEHOUSE Xueguang WU
Several industry experts also contributed their time and expertise to this book. Recognition and thanks are
given to the following authors for their contribution of specific subject matter presented in the book:
Georg BALOG (Nexans - retired)

Prof. José JARDINI, professor at the University of Sao Paulo, Brazil
David A. N. JACOBSON, Tony WEEKES, Narinder S. DHALIWAL, William McDERMID, Kelvin KENT,
(Manitoba Hydro, Winnipeg, Canada)
Mohamed RASHWAN (TransGrid Solutions, Winnipeg, Canada)
Daniel CASTAÑA, Mauro CROCE, (UTE)
The team wishes to thank the following people:

Roger CRITCHLEY and François GALLON for their time and tremendously valuable validation work.
Barbara VANTHIELEN, the HVDC Book’s project manager, for her dedication, patience and perseverence.
Carl BARKER, Colin DAVIDSON, Rafael BONCHANG, Neil KIRBY and Manuel RELVAS for the time
they allotted to review and edit this book.
The authors would like to thank their predecessors for the steps they made in understanding the technology
and developing it, which has made it possible for us to produce this book today. Special thanks must go to
John AINSWORTH, whose wide-ranging experience and fundamental understanding of all aspects of HVDC
and power engineering have helped to form the basis for many parts of this book.

GE VERNOVA’S UNO LAMM HIGH VOLTAGE DIRECT CURRENT AWARD RECIPIENTS
Uno Lamm High Voltage Direct Current Award

“Engineering – Accomplishment Through Imagination and Truth”
The Uno Lamm High Voltage Direct Current Award was established in 1980 by the Power Engineering Society
of the IEEE on the recommendation of the DC transmission Subcommittee. It provides a means for special
recognition of those outstanding engineers and scientists who have contributed to the advancement of high
voltage direct current (HVDC) technology.
The award is named for the man most responsible for the research and development that led to the first
practical application of an HVDC connection between AC systems. The keys to the solution of this problem
were the development of an electric valve which could be used in high capacity, high voltage converters, and
a fundamental system technology. This outstanding engineer and scientist was Dr. Uno Lamm, an IEEE Fellow
and the 1965 recipient of the Benjamin Lamme Medal.
Recipients of the Uno Lamm Award are an elite fraternity, and GE Vernova is proud to have had so many
within our ranks over the years that these achievements have been recognized. The contributions toward
the advancement of HVDC technology made by these GE Vernova engineers has been impactful and
meaningful, with many of their innovations still in practice today across the industry.
GE Vernova recognizes the significance that these pioneers have meant to our own HVDC legacy as well
as the mark they have made on the HVDC community at large. We are forever thankful to have worked
alongside many of them … to learn from them, to be mentored by them, and to count them as true friends.
GE Vernova Uno Lamm Award Recipients:

1983 – John D. Ainsworth
1989 – Aleksa Gavrilovic
1990 – Glenn D. Breuer
1993 – Donald M. Demarest
1994 – Thomas E. Calverley
2005 – Michael L. Woodhouse
2012 – Bjarne R. Andersen
2022 – Colin Davidson

TABLE OF CONTENTS
1
A BRIEF HISTORY OF
GE VERNOVA AND HVDC 14
1.1 GE VERNOVA GRID HVDC: PAST TO PRESENT............................................. 17
1.2 GE VERNOVA GRID INNOVATIONS..................................................................... 20
1.3 GE VERNOVA GRID HVDC EXPERIENCE.......................................................... 37
2
HVDC CONVERTER THEORY 66
2.1 DC TRANSMISSION APPLICATIONS.................................................................... 69
2.2 STEADY-STATE ANALYSIS OF CONVERTER BRIDGE OPERATION..... 74
2.3 REACTIVE POWER EXCHANGE BETWEEN HVDC
CONVERTER STATIONS AND AC SYSTEMS.................................................... 95
3
HVDC CONVERTER THEORY 110
3.1 HVDC CONVERTER STATION CONFIGURATIONS.................................... 113
3.2 HVDC STATION INSULATION CO-ORDINATION....................................... 127
3.3 TRANSIENT OVERVOLTAGE STUDIES............................................................. 134
3.4 CONVERTER STATION LAYOUTS....................................................................... 141
4
HARMONICS: CAUSES,
CALCULATIONS AND FILTERING 158
4.1 SOURCES, EFFECTS AND LIMITS OF HARMONIC DISTORTION................. 161
4.2 HARMONIC IMPEDANCE OF AC AND DC SYSTEMS...................................... 164
4.3 HVDC CONVERTER HARMONICS....................................................................... 173
4.4 CONVERTER DC HARMONICS............................................................................. 178
4.5 AC HARMONIC FILTERS......................................................................................... 182
4.6 DC HARMONIC FILTERS......................................................................................... 203
4.7 POWER LINE CARRIER FREQUENCY FILTERS.................................................. 213
4.8 RADIATED INTERFERENCE AND EMC................................................................ 221

5
AC/DC SYSTEM INTERACTIONS 230
5.1 E FFECT OF THE AC SYSTEM SHORT CIRCUIT LEVEL
ON AC/DC SYSTEM INTERACTION..................................................................... 233
5.2 P OWER TRANSMISSION INTO AN AC SYSTEM
FROM A HVDC CONVERTER................................................................................. 233
5.3 SYSTEM INERTIA...................................................................................................... 240
5.4 HVDC CONVERTER STATIC CHARACTERISTICS............................................. 242
5.5 EFFECTS OF AC SYSTEM VOLTAGE.................................................................... 248
5.6 MULTI-TERMINAL HVDC SCHEMES.................................................................... 251
5.7 TYPICAL TRANSMISSION SCHEME STATIC CHARACTERISTICS................. 254
5.8 BACK-TO-BACK HVDC SCHEMES – STATIC CHARACTERISTICS................ 258
5.9 OTHER CONTROL CHARACTERISTICS............................................................... 264
5.10 PROPOSED DEVELOPMENTS............................................................................... 266
5.11 STABLE OPERATION OF A HVDC CONVERTER IN AN AC SYSTEM........... 268
5.12 SYSTEM REPRESENTATION.................................................................................. 271
6
HVDC CONVERTER
STATION EQUIPMENT 276
6.1 THYRISTORS FOR HVDC.................................................................................... 279
6.2 THYRISTOR CONVERTERS................................................................................. 295
6.3 VALVE COOLING SYSTEM................................................................................. 335
6.4 HVDC TRANSFORMERS..................................................................................... 338
6.5 DC SMOOTHING REACTOR............................................................................... 352
6.6 MEASURING TRANSDUCER.............................................................................. 362
6.7 OVERVOLTAGE PROTECTION........................................................................... 368
6.8 CIRCUIT BREAKERS............................................................................................. 370
6.9 WALL BUSHINGS................................................................................................. 383
6.10 AUXILIARY POWER SUPPLIES.......................................................................... 384

7
CONTROL AND PROTECTION 390
7.1 HVDC SCHEME CONTROL................................................................................393
7.2 CONTROL OF A HVDC STATION.....................................................................396
7.3 CONVERTER STATION PROTECTION.............................................................408
7.4 PROTECTIVE ZONES..........................................................................................410
7.5 FAULT CASES AND CORRESPONDING PROTECTIONS..............................415
7.6 OVERVOLTAGE....................................................................................................415
7.7 TELECOMMUNICATIONS REQUIREMENTS...................................................416
7.8 CONVERTER TRANSFORMER FAULTS...........................................................416
7.9 DC POLE PROTECTION.....................................................................................416
7.10 HVDC CONTROL AND PROTECTION ARCHITECTURE...............................420
7.11 STRUCTURE OF A TYPICAL HVDC CONTROL SYSTEM..............................420
7.12 DUPLICATION OF CONTROL EQUIPMENT...................................................427
7.13 CONTROL FOR SERIES CONNECTED CONVERTERS..................................431
8
DC TRANSMISSION CIRCUITS 434
8.1 HVDC OVERHEAD TRANSMISSION LINES................................................... 461
8.2 HVDC SUBMARINE AND UNDERGROUND CABLES.................................. 474
8.3 EARTH / SEA ELECTRODES............................................................................. 474
9
HVDC SCHEME PERFORMANCE 484
9.1 CONVERTER STATION ELECTRICAL LOSSES............................................. 487
9.2 MODELING OF HVDC CONVERTER SCHEMES.......................................... 493
9.3 DYNAMIC PERFORMANCE STUDIES............................................................ 501
9.4 TRANSIENT AND DYNAMIC STABILITY INVESTIGATION......................... 512
9.5 C
ONTROLS DYNAMIC PERFORMANCE
ASSESSMENT INVESTIGATION...................................................................... 516
9.6 FUNDAMENTAL FREQUENCY TOV STUDY................................................. 518
9.7 TYPICAL STUDY RESULTS............................................................................... 520
9.8 RELIABILITY, AVAILABILITY AND MAINTAINABILITY................................ 523

10
PLANNING A HVDC
CONVERTER SCHEME 532
10.1 PLANNING A HVDC PROJECT.........................................................................535
10.2 OPERATIONAL EXPERIENCE............................................................................544
10.3 STUDIES ASSOCIATED WITH CONVERTER STATION DESIGN.................551
10.4 INFORMATION REQUIRED
FOR THE DESIGN OF A HVDC SCHEME........................................................557
APPENDIX 564
APPENDIX
APPENDIX RELATED TO CHAPTER 2................................................................564
ACRONYMS..................................................................................................................596

TOC
1 A BRIEF HISTORY
OF GE VERNOVA
AND HVDC
GE Vernova has a long history in HVDC, having been one of the first
manufacturers to enter the HVDC market in the early 1960s. In this
chapter you will learn how GE Vernova (as the then English Electric
Company) challenged and improved on early DC technology. The
chapter describes some of the many innovations in HVDC that have
been introduced by GE Vernova and its predecessor companies,
such as the phase-locked oscillator that now forms the basis of all
HVDC control systems. The chapter conclusion presents a round-up
of some of the important HVDC schemes built by GE Vernova up to
the time of publication.
The HVDC schemes developed since the early days of power
electronics, and the knowledge and experience that have been gained
from them, are the building blocks on which all future developments
have been based.
These innovations are the basis for the 21st century’s developments
in 600 kV and now 800 kV HVDC transmission.
TOC DC Transmission Systems: Line Commutated Converters | 15

1| A BRIEF HISTORY OF GE VERNOVA AND HVDC
Chapter contents
A BRIEF HISTORY OF
1.
GE VERNOVA AND HVDC .............................. 14
1.1. GE VERNOVA HVDC: PAST TO PRESENT.......................... 17

1.1.1. Early HVDC Transmission....................................................... 17
1.1.2. A New Industry is Born............................................................ 17
1.1.3. The DC Transmission Department .................................... 19
1.1.4. Testing HVDC Valves................................................................ 19
1.1.5. A Global Activity......................................................................... 20
1.2. GE VERNOVA INNOVATIONS................................................ 20

1.2.1. The Kingsnorth HVDC Transmission Project................. 21
1.2.2. The Nelson River HVDC Scheme......................................... 23
1.2.3. Synthetic Test Circuit for HVDC Valves............................ 24
1.2.4. Thyristor Valve Development............................................... 25
1.2.5. Multi-terminal HVDC................................................................ 31
1.2.6. McNeill Back-to-back Converter Station......................... 32
1.2.8. The H400 series HVDC Thyristor Valve............................ 34
1.2.9. The Hydro-Québec De-icer Project.................................... 34
1.2.10. The GCCIA 3 x 600 MW HVDC Project.............................. 35
1.2.11. Recent HVDC Projects............................................................. 37
1.3. GE VERNOVA HVDC EXPERIENCE...................................... 37

1.3.1. Mercury-arc Schemes.............................................................. 37
1.3.2. Oil-cooled Thyristors................................................................ 38
1.3.3. Air-Cooled Thyristor Valves................................................... 39
1.3.3.1 Eel River HVDC Link.................................................................. 40
1.3.4. Water-cooled Thyristor Valves............................................ 48
1.3.5. H400 Thyristor Valve Installations..................................... 55
BIBLIOGRAPHY ............................................................................................ 64
TOC
1.1. GE VERNOVA HVDC: PAST TO PRESENT
Since the birth of the electrical power industry in the late 19th century, alternating current (AC) quickly
became established as the preferred method for power transmission, principally because it enabled power
transformers to step-up and step-down the voltage to convenient levels, which was not possible with the
competing technology of DC (the technology championed by GE’s founder, Thomas Edison). However, DC
transmission was never entirely defeated in the ‘Battle of the Currents’, and it was realized at an early
date that DC had advantages over AC in some special circumstances. In the USA, GE built an experimental
HVDC scheme from a hydro-electric power plant at Mechanicville to its factory in Schenectady, New York
in the 1930s, but the system was ahead of its time and the technology was not commercialised until later.
From the 1950s onwards, interest in DC transmission started to increase in several countries, and one
of the ancestors of GE Vernova’s modern-day HVDC business, the English Electric Company in Stafford,
UK, was one of the first industrial companies to realize the potential of this new technology. This section
describes how English Electric first entered the HVDC market and how the business has grown, both
organically and by mergers and acquisitions, with CGEÉ Alsthom (France) and AEG (Germany) both joining
English Electric’s successor GEC as part of the Alstom organization before Alstom was in turn acquired by
GE Vernova (then operating as GE) in 2015.
1.1.1. EARLY HVDC TRANSMISSION

Early power generation and transmission systems used DC. These systems were constructed using
generators which produced DC voltage and current at their electrical terminals from a rotating mechanical
input. The first commercial scheme was introduced in 1882 [1] and many others soon followed. These
schemes were predominantly used for electric lighting, using the recently invented and improved electric
light bulb, which provided many hours of operation.
As soon as electric power became available, people started to find additional uses for it and therefore demand
grew. However, a disadvantage to early schemes was that the electricity had to be generated, transmitted
and used at the same voltage and thus transmission losses were relatively high, as the machine insulation
and end user applications tended to be designed for lower voltages.
Attempts to overcome the limitations of low voltage DC transmission were made by raising the DC voltage.
An early method of high voltage DC transmission used the Thury method [2] which used a number of series-
connected DC generators, electrically insulated from each other and from earth, driven off a common prime
mover: an example of which is the 57.6 kVdc, 75 A, 180 km, interconnection upgrade between Moutiers and
Lyon installed in 1905 [3].
In the 9 May 1924 edition of the Electrician, the development of the Highfield-Calverley Transverter by English
Electric Co., Ltd, in Preston, United Kingdom, was reported [4]. This was a mechanical rectifier specifically
designed for high voltage DC applications of around 100 kVdc. The machine consisted of a transformer with a
number of secondary windings each connected to a series of stationary contact plates. A set of rotating brushes,
operating at synchronous speed, swept across these contact plates connecting each secondary transformer
winding to the load only when the transformer’s EMF was in the appropriate direction. The decision of the new
nationalized Central Electricity Authority in the 1920s to standardize the grid voltage at 132 kVac ended any
future for the transverter.
Despite all of the methods invented for transmitting power as DC, the advantages of AC transmission, allowing
for power generation and usage at low voltage whilst transmitting power at higher voltages and hence
resulting lower losses by means of a simple transformer, along with ease of switching, meant that new DC
transmission was limited mainly to research projects.
1.1.2. A NEW INDUSTRY IS BORN

By the 1950s the industrial power conversion business was well established for rolling mills, railways,
aluminum smelting, mine winding etc. where accurate speed control was necessary, utilizing mercury-
arc valve technology. Many companies around the world were supplying and actively working on the
development of these products, including English Electric Co. Ltd.
BACK TO DC Transmission Systems: Line Commutated Converters | 17

CHAPTER
Fig. 1.1a– An English Electric 15 kV single anode glass bulb rectifier [5]. This was a rectifier developed by the
English Electric Rectifier Department, providing power for the BBC (British Broadcasting Corporation) radio
transmitters in association with the Marconi Company, which was a subsidiary of English Electric Co., Ltd.
English Electric’s Nelson Research Laboratories were actively developing higher voltage mercury-arc valves
and analytical methods of AC and DC system operation, including an AC/DC/AC analog simulator.
This simulator used passive components to represent a scalable version of a real power system along with
English Electric ‘mercury-pool’ valves rated at 1000 V, 2 A, connected in any arrangement through a plug-
board. The main purpose of the simulator was to investigate the interactions between AC and DC systems
and then to establish ways of optimizing the performance of the HVDC controller within the AC system.
Here, John Ainsworth, an engineer who was to have a major impact on HVDC in future years (see section
1.2), was able to develop English Electric’s understanding of how to control power conversion equipment.
Whilst English Electric and almost all other worldwide manufacturers had been involved in other activities
during World War II, ASEA of Sweden, under the guidance of the brilliant Uno Lamm, was heavily involved in
high voltage DC transmission research. The concept used by ASEA of applying power conversion equipment
to create DC from AC and vice versa, had the advantage of being able to be integrated into existing AC power
systems. By the early 1950s, ASEA had developed a 100 kV mercury-arc valve converter which they installed
in a 20 MW interconnection between mainland Sweden and the island of Gotland, a transmission distance of
96 km, which was too far for AC cable transmission.
In 1961 ASEA commissioned their second HVDC submarine cable link, which connected England with France.
This again operated at a voltage of 100 kVdc, but the power transmission capacity had risen to 160 MW.
English Electric’s management and engineers had closely watched ASEA’s progress. Whilst the UK’s Central
Electricity Generating Board (CEGB) was discussing the link to France with ASEA, they were also actively
in discussion with English Electric engineers. Discussions ranged from the technical issues of HVDC to the
schemes they were considering for the future.
In 1960 the English Electric representative in Italy heard that a mining company, named Carbosada, in Sardinia,
was proposing to build a power station and were in discussion with ASEA for a HVDC link in order to transmit
the power to mainland Italy. English Electric approached Carbosada, who were pleased to have competition for
BACK TO 18 | DC Transmission Systems: Line Commutated Converters
CHAPTER
ASEA, but had strong reservations over English Electric’s lack of proven technical experience. Circumstances
were about to change this.
Unknown to English Electric at the time, ASEA’s management had concluded that there was a great deal
of potential business for HVDC in the global market but, without the possibility of competitive bidding
many utilities would not consider it as an option. ASEA therefore approached English Electric to propose
a collaboration and, towards the end of 1961, English Electric reached an agreement with ASEA which
allowed for technical collaboration between the two companies whilst maintaining complete commercial
independence, thereby allowing the two companies to compete in the market-place. This agreement meant
that English Electric was able to reassure Carbosada with regard to its technical capabilities. For ASEA and
utilities around the world this ensured competitive bidding and for English Electric it was a technical insurance
policy. From this, a new business unit was born.
1.1.3. THE DC TRANSMISSION DEPARTMENT
5 January 1962
Stafford Works General Notice no 1/62
A new specialised activity to be known as the
DC Transmission Department, responsible for
the Commercial and Engineering aspects of DC
Transmission of Power, is to be created in Stafford.
This new department comprised four people: Tom Calverley, Unit Manager, Denys Montgomery, Manager
(Sales and Contracts), Aleksa Gavrilovic, Chief Engineer and Mrs Clarke, Tom Calverley’s Secretary.
Note: Tom Calverley was the son of the man who some years earlier had developed the Highfield-Calverley Transverter (see
section 1.1.1).
Work started in earnest on two fronts: firstly it was necessary to build up the resource capacity of this new
business unit and secondly, there was now a contract to be executed which had stiff penalties attached.
The work force was built up from existing employees of other business units in English Electric and those
completing their apprenticeship, all of whom wanted to be part of this new technological adventure.
ASEA may have thought that English Electric would simply act as a project supplier, purchasing valve and
control equipment from them. However, English Electric was determined to stand on its own and saw
the collaboration agreement as a way of accelerating their own development activities whilst, at the
same time, establishing itself in the market place. The research department focused on their own ideas
regarding the control of HVDC converters, something which would later revolutionize the whole HVDC
industry (see section 1.2.1.3). In parallel, developments in Stafford of the single-anode mercury-arc valve
technology, then state-of-the-art, was undertaken in a collaborative effort between the DC Transmission
Department and the Rectifier Department, leading to English Electric being the first to market with a 133
kVdc mercury-arc converter for the CEGB’s Kingsnorth project. They later developed the 150 kV valves for
the Nelson River project in Canada.
Over the next ten years, English Electric and ASEA engineering departments were to closely collaborate
on both developments and problem solving, with the technology transfer agreement only ending with the
emergence of thyristor valve technology.
1.1.4. TESTING HVDC VALVES

For English Electric to develop its own mercury-arc valve technology, it needed a testing facility. By
coincidence the Corporation Electricity Works in Stafford, an embedded coal fired power station in
the heart of Stafford, had, by 1959, reached the end of its economic usefulness. The power plant was
decommissioned, but the then MEB (Midlands Electricity Board) retained a substation in part of the building
and English Electric was able to lease the remainder of the building in order to create a new test facility. It

CHAPTER
is ironic to note that the voluminous space left by the removal of the generators and boilers, ideal for high
voltage testing, once housed a DC generating plant when the power station was originally built in 1895.
The local electricity supply could not provide enough power to allow a complete converter pole to be
tested in a method which represented the in-service conditions. To overcome this problem, a novel testing
method had to be developed. This test method was known as synthetic testing, a description of which is
provided in section 1.2.3.
1.1.5. A GLOBAL ACTIVITY

On the opposite side of the Atlantic, General Electric of the USA re-entered the HVDC market in the late
1960s as the overall project supplier for the Pacific Intertie project (which used mercury arc valves supplied
by ASEA) but thereafter concentrated its development on thyristor technology which, as a result of its
reduced maintenance, had become the preferred technology in the market by the early 1970s. In 1972, GE
built the world’s first thyristor-based HVDC scheme, the Eel River back-to-back interconnector and went
on to supply many thyristor-based HVDC projects across North America during the 1970s and 1980s. GE
also supplied valve and control equipment to CGEE Alsthom of France who then provided turnkey HVDC
solutions within Europe. CGEE- Alsthom was formed in 1971, following CGE’s acquisition of SGE, also of
France. CGEE (Compagnie Générale d’Entreprises Electriques) had been formed in 1913 by CGE of France
which, by 1967, was acknowledged as the largest European company.
In parallel, a joint venture was also set up between three German or Swiss companies; AEG, BBC (Brown
Boveri Company) and Siemens. This joint venture became known as the ‘Zamco Consortium’ or ‘German
Working Group’ and put a number of thyristor based HVDC schemes into service including the 1360 km
Cahora-Bassa interconnection between Mozambique and South Africa, rated at 1920 MW, ±533 kVdc,
commissioned in 1978. Various mergers and acquisitions took place between businesses: English Electric
merged with GEC (General Electric Company) UK and CGEE-Alsthom purchased General Electric’s HVDC
business in the USA forming a new division called CANA (CGEE Alsthom North America). CGE of France and
GEC of the UK merged forming GEC-Alsthom, later becoming ALSTOM. CGEE-Alsthom changed its name to
Cegelec, which was later acquired by Alstom. The German Working Group split up and GEC-Alsthom acquired
the AEG business relating to their HVDC activities. In 2004, AREVA purchased Alstom’s Transmission and
Distribution division, forming AREVA T&D. In 2010, Alstom reacquired the transmission business, creating
the company’s third sector: Alstom Grid, which was in turn acquired again by GE Vernova (then GE) in 2015.
GE Vernova’s HVDC heritage therefore draws upon three independent lines of ancestry:
➙ GE in the United States and its licensee, CGEE Alsthom in France
➙ AEG in Germany, using technology shared with Siemens and Brown Boveri
➙ English Electric/GEC in the United Kingdom
In this book, the original company names are used for activities that took place up until 2003, while for
activities from 2004 onwards (by which time only the former English Electric/GEC organisation, then called
AREVA T&D, was still active) the name "GE" or "GE Vernova" is used.
The worldwide increase in demand for HVDC over the last 10 years, together with new product
development activities, have provided a sustained period of growth for the company. This has allowed the
business to build on past successes and take a leading position in the HVDC market, and to create a wealth
of experience in HVDC schemes, technology innovation, design, and operation. From its HVDC Centre of
Excellence in Stafford, UK and from its local offices around the world, GE Vernova is now ideally placed to
provide optimum solutions to the technical challenges and needs of modern power systems.
1.2. GE VERNOVA INNOVATIONS

GE Vernova and its ancestor companies have been at the forefront of HVDC development ever since the
inception of English Electric’s DC transmission department in 1962 and had made significant advances
even before that with the experimental Mechanicville-Schenectady project. A significant number of major
innovations in HVDC, many of which have become industry standards, have originated from GE Vernova’s
center of excellence in HVDC based in Stafford, UK.

CHAPTER
This section describes some of the major innovations and developments introduced by GE Vernova in the
field of HVDC. As has been noted in the preceding section, Alstom Grid’s first commercial HVDC project
was the Italy-Sardinia 200 MW, 200 kV project. However, this section picks up the history from the next
HVDC project executed by Alstom (then called English Electric): the Kingsnorth HVDC transmission project.
1.2.1. THE KINGSNORTH HVDC TRANSMISSION PROJECT

The National Grid in the UK began in the 1930s, using a transmission voltage of 132 kV to interconnect the
hitherto independent regional electricity generating and supply companies. At first, the 132 kV lines were
lightly loaded, since most electricity continued to be generated close to where it was being used. Gradually
a shift towards larger, centralized power generating stations located close to the source of fuel (mostly
coal) began to occur. In the 1950s a higher transmission voltage of 275 kV was introduced to reinforce
the grid, and by the early 1960s the first components of the 400 kV transmission network had been built.
Each of the three largest cities (London, Birmingham and Manchester) was served by a 400 kV outer ring,
with radial connections inwards to an inner ring at lower voltage. In London, a new coal-fired power station
(Kingsnorth) was built during the 1960s on the Thames estuary to the East of the city, feeding into the outer
ring. However, significant planning problems were being encountered in injecting the power generated by
this station into the city center.
Accordingly, the Central Electricity Generating Board (CEGB), the then-nationalized electricity utility for
England and Wales, placed an order in 1966 for a 640 MW underground cable HVDC link from the Kingsnorth
power station to two sites: Beddington in South London and Willesden in West London.
The Kingsnorth station consisted of two poles, each with a power transmission rating of 320 MW at 266 kVdc
and consisting of two series-connected 6-pulse groups of mercury-arc valves. Both the Beddington and Willesden
stations were single poles rated at 320 MW and consisting of two 6-pulse groups. A neutral DC cable linked the
Willesden and Beddington stations and, together with changeover switchgear, permitted power to be transmitted
between these two stations in the event of the cable from Beddington to Kingsnorth being out of service.
The Kingsnorth scheme was commissioned in 1972 and contained many firsts for HVDC [6]. It was the first
HVDC scheme to be embedded in an existing AC system, as opposed to interconnecting two separate AC
systems. It was also the first HVDC scheme to use only land cables for power transmission.
Several other significant innovations in HVDC equipment and layout were introduced on this project.
1.2.1.1. ARAG/4 Mercury-arc Valve

The higher current rating of the Kingsnorth scheme in comparison with the Italy-Sardinia HVDC link required
a re-design of the mercury-arc valves. Mercury-arc valves, like the thyristor valves that were to replace them,
Fig.1.2a– One of the Kingsnorth valve halls showing the English Electric ARAG/4 4-anode mercury-arc valves in service

CHAPTER
normally conduct current in only one direction – from the anode terminal to the cathode terminal. However,
with high voltage mercury-arc valves, a limitation existed for the total current that could be carried by a
single anode connection. To achieve DC (bridge) current ratings in excess of 300 A, several anodes needed to
be connected in parallel, sharing a common cathode. The valves for the Sardinia-Italy project had used four
anodes in parallel, each rated at 250 Adc. For the Kingsnorth HVDC scheme, a total rating of 1200 A dc was
required so a new design of mercury-arc valve, designated type ARAG/4, was designed by English Electric.
The ARAG/4 (see Fig.1.2a) used four anodes of an uprated design with higher voltage and current ratings.
1.2.1.2. Station Layout and Main Equipment

Because of the relatively high level of atmospheric pollution existing in London at the time, the converter
stations were built indoors in order to minimize insulation requirements, particularly for the DC equipment.
Moreover, because of the high cost of land, both the Beddington and Willesden substations were built on
two levels.
The AC harmonic filters were located outdoors but were also novel. The filters were self-tuning [8], that is,
the inductance of the tuned filter changed such that the filter was always tuned to the exact harmonic to be
filtered, compensating for variations in AC system frequency and variations in capacitance due to changes in
temperature. Fig.1.2b shows one of these variable reactors.
Fig. 1.2b– A controllable reactor used in a "self-tuning" filter, part of the Kingsnorth scheme
1.2.1.3. Phase-Locked Oscillator Control System

The major innovation of the Kingsnorth HVDC link can still be found in all Line-Commutated HVDC links today.
Previous HVDC schemes had based the control timing signals on the measured AC system voltage. The
timing of grid pulses (for turning on the mercury-arc valves) was based on zero-crossing times of the
AC voltages for the relevant phases, delayed by an amount corresponding to the firing angle, a. This
approach was known as Individual Phase Control (see section 5.2 of [7]). Despite its apparent simplicity, it
suffered from a major shortcoming when applied on HVDC schemes where the AC network was very weak.
Because the timing of firing signals sent to the valves was based directly on the timing of the AC voltage,
any distortion in the AC voltage, for example due to low-order, non-characteristic harmonics such as 3rd
harmonic, would directly affect the timing of the turn-on pulses of the valves. Under some conditions, this
could generate more of the same harmonics which led to the problem in the first place. Consequently, on

CHAPTER
weak AC systems, the converter operation tended to amplify any distortions already existing in the AC
system voltage waveform, leading to harmonic instability. Although the problem could be mitigated by
introducing passive analog filters into the control system to attenuate the problematic harmonics, such
filters created problems of their own.
The Kingsnorth scheme was the first to use the phase locked oscillator [9], an innovative concept invented
by Dr John Ainsworth, one of seven engineers in GE Vernova and its ancestor companies to win the
prestigious Uno Lamm award for HVDC.
The phase locked oscillator removed the dependency of the converter control system on the AC system
voltage waveform, by deriving firing pulses from a voltage-controlled oscillator which delivers a train of equally
spaced firing pulses at a frequency which is directly proportional to the value of a control variable. The control
variable takes into account the various control loops required for the safe operation of the converter. The
voltage-controlled oscillator feeds the resulting train of firing pulses (each spaced at regular 60° intervals) into
a Ring Counter which distributes these firing pulses, in turn, to the six valves in a 6-pulse bridge.
This development, also known as the Equidistant Firing Control, significantly reduces the harmonic generation
from the converters, allowed future schemes to operate at much lower short-circuit levels than had previously
been possible and has become the industry standard for HVDC control systems.
1.2.2. THE NELSON RIVER HVDC SCHEME

In 1967, Manitoba Hydro of Canada issued an invitation to bid for a ± 400 kVdc, 1440 MW HVDC scheme
to transmit power from the new Kettle hydro-electric generating plant on the Nelson River in the north of
Manitoba, some 900 km south, to an inverter station at Dorsey substation near Winnipeg. From Dorsey, some
power would be consumed locally in the Winnipeg area but much would be exported to the Midwest USA.
Manitoba Hydro entered into negotiations with several suppliers, including some offering the new thyristor
based technology. However, soon the competition was between English Electric and ASEA for the contract.
By the end of 1967 English Electric was awarded the contract and, following discussions, the scheme rating
became 1620 MW at ± 450 kVdc.
Each pole of the HVDC scheme consisted of three 6-pulse bridges in series. Each 6-pulse bridge was rated at
150 kVdc, 1800 Adc. The scheme was built in stages, with the first stage using three valve groups to produce
+150 kV/-300 kV, commissioned in October 1972 [10], and the final stage completed in 1977.
This was to be the last HVDC scheme built with mercury-arc valves. In the early 1990s the mercury-arc valves
of pole 1 (valve groups VG11-VG13) were replaced by the new water-cooled thyristor valves of the H300 range
(see section 1.2.4.5).
1.2.2.1. The ARBJ/6 Mercury-arc Valve

The DC bridge current rating of 1800 A was 50% higher than for the Kingsnorth HVDC scheme and the project
worked at a higher DC voltage. This required further improvements to the mercury-arc valve technology.
The ARBJ/6 mercury-arc valves designed for this project used 6 anodes in parallel, as shown in Fig.1.2c, and
were the largest mercury-arc valves (in both voltage and current ratings) ever built by any manufacturer.
1.2.2.2. Damping Controls

One particular constraint of this project was that virtually the entire power output of the Kettle generating
station was to be exported by HVDC. This led to concerns about the stability of the remote (Kettle) AC
system in the event of a sudden reduction of power transmission caused, for example, by a loss of one
converter group or one generator.
The HVDC scheme, therefore, included special frequency control functions [11] which allowed the frequency
of either of the two AC networks to be controlled by power modulation in the HVDC line. In addition to
the normal steady-state power order, three additional transient inputs affected the rectifier current order:

CHAPTER
Fig. 1.2c– The ARBJ/6 mercury-arc valves for the Nelson River project
➙ Load area frequency changes, mimicking the action of governors with fixed frequency droop
characteristics
➙ Phase angle changes on the Dorsey 230 kVac bus
➙ Frequency changes at the rectifier station AC bus, to assist the governors on the generators at Kettle
generating station in their task of maintaining the frequency of the Kettle AC bus.
Whilst these principles had been used on some earlier HVDC schemes, the Nelson River scheme made greater
use of this technique than any scheme to date.
1.2.3. SYNTHETIC TEST CIRCUIT FOR HVDC VALVES

In parallel with developing its mercury-arc valves to higher and higher current and voltage ratings, English
Electric was also refining its techniques for testing these valves.
Testing HVDC valves under realistic voltage and current conditions is a challenge. The obvious solution of
building a complete 6-pulse converter and fitting the test valve into one of the positions in the converter
requires a very high power rating from the local supply. Even where a very strong supply is available, this is a
very expensive option.
The stresses on a HVDC valve consist of two parts; the conduction stresses resulting from the high DC current
and the blocked (non-conduction) voltage stresses. The testing method developed by English Electric was a
‘synthetic test circuit’ comprising a low voltage high current oscillator and a high voltage low current oscillator.
The operation of these two oscillators is carefully coordinated such that for a particular valve under test (the
test object), the operational wave shape can be reproduced; the load current during the on-state coming
from the high-current circuit and the off-state voltage coming from the high voltage circuit. One of many later
refinements was the addition of a Surge Injection Circuit, which permitted disturbances such as commutation
failures to be simulated artificially to check that the valve responded safely and correctly.
Until the mid 1990s, all other HVDC manufacturers used a different technique based on the 6-pulse back-
to-back test circuit, but the ever-increasing voltage and current ratings of power semiconductors began
to make this technique impractical and today all HVDC manufacturers use variants of the synthetic valve
test circuit originally developed by English Electric in the 1960s.
The synthetic test circuit concept pioneered by English Electric during the 1960s has been subject to a
number of refinements (for example, [12], [13]) and was still in use at the beginning of 2018. However, in

CHAPTER
Surge
injection
circuit
HighÊcurrent HighÊvoltage
circuit circuit
V1
TestÊvalve
andÊlocal
circuit
Fig. 1.2d– Basic principles of a synthetic test circuit
2018 GE Vernova commissioned a brand new test facility [14] based

on an entirely different principle, which has again lifted the bar on
the methodology for testing HVDC valves. The new test facility uses
1.2d
a high-power Voltage-Sourced Converter as part of the test plant to
directly recreate both the off-state voltage and on-state current of
the test object.
1.2.4. THYRISTOR VALVE DEVELOPMENT

Although mercury-arc valves were rugged and relatively immune
to permanent damage, they were complex pieces of equipment
containing many moving parts (vacuum pumps and so forth) and
consequently they required extensive maintenance. Mercury-arc
valves were also prone to a phenomenon known as arc-back, where
the valve transiently (and incorrectly) conducts current in the reverse
direction. Whilst generally not damaging to the valve itself, arc-back
was highly undesirable for the operation of the HVDC converter
because of the large transient fault currents it generated.
Therefore, as was mentioned in previous section, from the late
1960s onwards, efforts were made in many organizations to design
a solid-state replacement for the mercury-arc valve, using a new
semiconductor device, the thyristor (or SCR). Thyristors, as discussed
later in section 6.1, behave similarly to mercury-arc valves but without
the arc-back problem. However, because thyristors had (and still have)
voltage ratings that are very low compared with the voltage ratings
required for HVDC, new problems were introduced as large numbers
of thyristors had to be connected in series. However this problem was
also an opportunity, since it offered the prospect of extending the
range of operation to higher voltages than were previously possible
with mercury-arc valves. Fig. 1.2e– Internal structure of a H100 thyristor valve

CHAPTER
1.2.4.1. The H100 Oil-cooled Thyristor Valve
English Electric’s first thyristor valve design, now
referred to as the H100 valve, used 38 mm diameter,
4 kV-rated thyristors, with nearly 200 connected in
series in a helical arrangement [15] as shown in
Fig.1.2e.
The thyristors were triggered electro-optically via
a laser at ground potential, coupled via light-guides
to individual optical-electrical converters for each
thyristor.
In order to achieve sufficient electrical clearance
and creepage in a compact arrangement, and Fig. 1.2f– The H100 outdoor thyristor valves
to provide cooling for the thyristors and other located at the Lydd converter station in Kent, UK
components, the helical valve structure shown in
the photograph was immersed in insulating oil,
following the established practice for transformers
and other electrical equipment. External insulation
between the top and bottom of the valve assembly
was provided by an external porcelain insulating
housing. This porcelain housing remains, to date,
one of the largest such housings produced for any
item of electrical equipment.
The current rating of each of the helical valve
structures shown in Fig. 1.2e is relatively low, but
the construction lends itself quite well to connecting
more than one such structure in parallel, in a similar
way to paralleling of anodes on mercury-arc valves.
As a demonstration of this technology, three valve
units of this type were built. In cooperation with the
CEGB, these three valve units were connected in
parallel, installed and commissioned in 1972 at the
Lydd converter station (the UK end of the original
Fig. 1.2g– Valve module for H200 series HVDC valve
160 MW Cross-Channel HVDC scheme that had
been built in 1961) replacing one of the original
ASEA mercury-arc valves: see Fig. 1.2f.
1.2.4.2. The H200 Series Air-Cooled

Valve
The outdoor, oil-insulated valve concept reduced
the civil costs but made maintenance difficult as the
complete valve arrangement had to be dismantled
in clean room conditions. Therefore, for the next
generation of thyristor valves, the H200 series,
the more recognizable indoor, air-insulated valve
arrangement was used.
By this time, it was already becoming clear that
de-ionized water was the preferred method for
cooling thyristor valves.
However, at this time, the UK General Electric
Company, GEC (a company which was entirely
separate from GE Vernova (then operating as GE)
Fig. 1.2h– The prototype H200 thyristor valve
and which had acquired English Electric in the

CHAPTER
early 1970s) was also developing the solution for the UK end of the 2000 MW Cross-Channel HVDC link
interconnecting England and France (a project also known as IFA2000: Interconnexion France-Angleterre
2000 MW). Since there was at that time no operational experience of using water cooling for thyristor
valves, the customer for the converter station (CEGB) required that the valves be air-cooled.
Therefore, the H200 series of valves was developed using series-connected modules, with each module
comprising two forced-air-cooled 56 mm thyristors connected in parallel, as shown in Fig. 1.2g.
A prototype valve, type H200, was built and installed at the Willesden converter station (part of the Kingsnorth
scheme) in 1981 (see Fig. 1.2h).
The valve modules were installed in rows along the sides of an insulating plenum chamber, which acted as
the manifold assembly for supplying cooling air to the modules. Air was blown through the valve modules and
out into the valve hall air, to be cooled and re-circulated.
Fig. 1.2i– H210 thyristor valves at the Sellindge converter station of the Cross-Channel HVDC scheme

CHAPTER
The H200 valve was a prototype for the valves to be used at the UK-end (Sellindge) of the Cross-Channel HVDC
scheme. This project was a submarine cable scheme rated at 2000 MW, consisting of two bipoles of ±270 kV
and 1000 MW each. It was, and still remains, the largest-capacity submarine cable HVDC scheme in the world.
The Cross-Channel scheme was originally intended to facilitate bi-directional power trading between
England and France, but in practice for most of its life to date (and virtually all of its first 15 years of
operation) it has been importing full rated power from France to England, taking advantage of the
availability of low-cost nuclear power from Northern France. For this reason, this scheme has one of the
highest levels of utilization of any HVDC scheme in the world.
The thyristor valves for the Sellindge converter station [16] of the Cross-Channel link were of type H210,
a refinement of the prototype valve installed at Willesden. The valves were installed in the now standard
Fig. 1.2j– Air-cooled thyristor valves for Les Mandarins converter station

CHAPTER
quadrivalve arrangement where four valves corresponding to one AC phase are installed in a single tower as
shown in Fig. 1.2i.
1.2.4.3. GE Vernova-developed air-cooled thyristor valves

In parallel with the development of, first, mercury arc valves and then thyristor valves at English Electric and
GEC in the UK, GE in the United States was also developing thyristor valves. Indeed the thyristor itself was a
GE innovation and GE installed the world’s first thyristor-based HVDC scheme at Eel River in Canada in 1972
[17] and went on to supply similar technology for 8 (section 1.3) further HVDC schemes in North America,
culminating in the converter stations for Phase 1 of the Québec-New England HVDC scheme in 1986.
All of the thyristor valves supplied by GE during this time were air-cooled and used parallel-connected
thyristors, mounted on finned heatsinks attached to a vertically-mounted panel of Glass-Reinforced
Plastic (GRP). The GRP panels slid into apertures on the outside of a large GRP box which served as the
air manifold, and air was sucked from the valve hall into the valve structure and down into the air handling
plant in the basement. The distinctive red colour of the outside surface of the valve gave rise to the widely-
used colloquial name of “red-box valves” for this technology.
The same “red-box” technology was also licensed, and later sold, to CGEÉ Alsthom in France. In parallel
with the development of the H200 series valves in Stafford and the construction of the Sellindge converter
station for the Cross-Channel scheme, the converter station for the French end of the Cross-Channel
scheme, at Les Mandarins, was also being constructed by CGEÉ Alsthom. This project was the first HVDC
scheme to be executed with different manufacturers completing the two converter stations. Fig. 1.2j shows
the thyristor valves installed by CGEÉ Alsthom at Les Mandarins.
After completing the Cross-Channel project, CGEÉ Alsthom went on to use the same technology on
two further projects, both associated with the Sardinia-Italy project that had been English Electric’s first
commercial HVDC scheme.
1.2.4.4. The EPRI Gas-Insulated HVDC demonstrator

Air is a good insulation medium because it is so cheap and plentiful. This is the reason why, apart from a
handful of early HVDC projects using oil-insulated valves (including English Electric’s H100 demonstrator,
described in section 1.2.4.1 and the Cahora-Bassa project in Southern Africa, built by AEG and the “German
Working Group”), virtually all HVDC schemes have used air-insulated valves even until the present day.
However, at normal atmospheric pressures, air has a relatively low dielectric strength compared with most
insulating materials and, as a result, air-insulated equipment requires large clearances around it in order
to prevent flashovers. When the equipment needs to be mounted indoors, as is the case for HVDC valves,
the results is a large and expensive building.
The dielectric strength can be improved by operating at increased pressure and by using a high-performance
insulating gas such as SF6. GE was very early to realise the potential of this for HVDC, and in 1978 a gas-
insulated HVDC demonstrator installation was designed with EPRI in the United States and installed in
New York [18].
The demonstrator consisted of two 100 kV, 100 MW 12-pulse converters arranged as a short-distance
(600 m) transmission system between two nearby AC substations. The two most novel features of the
demonstrator system were the gas-insulated thyristor valves and the gas-insulated DC bus system. The
valve modules were based on those of the existing GE “red box” design but with the addition of plastic
coolant pipes through which Freon was pumped to cool the thyristors and other components. The valve
modules were installed in a sealed container filled with SF6 and capable of being pressurised at up to 2.6
bar. The system was equipped with a DC bias voltage supply to study the effects of a DC voltage of up to
±400 kV with respect to ground.
The technology was never commercialised, possibly because of the concerns that were already starting to
grow about the harmful effects of Freon on the ozone layer. Nevertheless it was a useful demonstration
of how much more compact an HVDC station can become if the key components can be gas-insulated.
There has been a resurgence of interest in this topic in recent years and we may yet see return to the use
of gas insulation for some critical components of an HVDC station.

CHAPTER
1.2.4.5. H300 Series Water-cooled Thyristor Valves
The air cooling of HVDC valves leads to a bulky solution involving expensive air-handling equipment and
does not lend itself readily to the use of large-diameter thyristors and hence to the attainment of high
current ratings. Therefore, in parallel with the execution of the Cross-Channel project, the R&D team in
Stafford continued with the development of a water-cooled range of thyristor valves, which was to become
the H300 series, forming the backbone of the HVDC solutions offered by GEC, GEC-Alsthom and ALSTOM
for almost fifteen years.
The H300 series valve was based on 5.2 kV, 100 mm diameter thyristors. The larger silicon diameter
compared with previous valves meant that parallel connection was no longer necessary, and instead
the thyristors were mounted in series-connected
banded pairs. Each banded pair included a novel
method for applying the clamping pressure to the
thyristors and incorporated the damping resistors
and di/dt-limiting reactor for each thyristor. The
banded pair construction (Fig. 1.2k) led to a very
efficient and compact design and contributed to
the attainment of the prestigious Queen’s Awards
for Technology and Export Achievement in 1991 and
1993 respectively.
Each module of the H300 valve was maintainable
in-situ and contained seven banded pairs, hence
fourteen series-connected thyristors along with the
Fig. 1.2k– H300 banded pair
associated damping circuits, inrush limiting reactors
and gate electronics as shown in Fig. 1.2l.
Fig. 1.2m shows a valve structure for the Nelson River valve replacement project completed in 1993. Valve
modules are vertically stacked and separated by floor-mounted composite insulators.
The use of water-cooling for thyristor valves introduced a new set of design challenges, as explained later
in section 6.2. One of the most complex challenges was the choice of materials for water-cooled heatsinks
and insulating pipe materials in order to avoid any problems of water leakage or electrochemical corrosion.
To mitigate this risk, GEC undertook a very comprehensive program of experimental evaluation [19] of
different insulating pipe systems, culminating in the choice of Cross-Linked Polyethylene (PEX) for the
cooling pipes. Almost 30 years after the first water-cooled H300 valves entered service, the reliability of
this system has remained excellent.
Fig. 1.2l– A H300 valve module

CHAPTER
Fig. 1.2m– H300 valves in valve group VG13 of the Nelson River Scheme
An additional objective of this development exercise was to ensure that the valve cooling system could
be cooled not only by pure de-ionized water, but also by de-ionized water/glycol mixtures, so that valve
cooling systems in cold climates could be provided with single-circuit cooling instead of the dual-circuit
cooling which was usual at the time.
1.2.5. MULTI-TERMINAL HVDC

The last air-cooled valves supplied by GE Vernova’s ancestors were built by CGEÉ Alsthom for the Sardinia-
Italy link – the same project that had given English Electric its foothold in the HVDC market. In 1988, CGEÉ
Alsthom built and commissioned a third converter station on the island of Corsica [20], through which
the HVDC transmission line already passed on its way from Sardinia to the Italian mainland. The scheme
henceforth became known as the SACOI project (SArdinia-COrsica-Italy). Although the power rating of the
Corsica converter station was small by today’s standards (50 MW), it was notable because the addition of
this third terminal turned the existing Sardinia-Italy link into the world’s first multi-terminal HVDC system.
Multi-terminal HVDC systems are challenging to design, particularly with Line-Commutated Converters,
because of the greater complexity of the controls and the need for the DC terminals of the intermediate
station to be reversible so that changes of power direction can be accommodated independently. The
Corsica station was specified to be capable of either importing or exporting power and thus needed to be
connected to the DC line via change-over switchgear so that the polarity of the DC voltage to which the
the converter valves were connected could be changed as necessary.
After completing the third terminal on Corsica, CGEÉ Alsthom’s last HVDC project to use the “red box”
GE valve technology was the replacement of the mercury-arc valves at the two ends of the scheme, which
was completed in 1992 [21].
1.2.6. MCNEILL BACK-TO-BACK CONVERTER STATION

The first commercial application of the new H300 valve was awarded to GEC in 1987. This was for the
McNeill back-to-back converter station which interconnects the Canadian provinces of Alberta and
Saskatchewan. This station, the first back-to-back converter built by GEC, is the most Northern of a series

CHAPTER
of back-to-back HVDC links which separate the Eastern and Western AC systems of North America (which
operate asynchronously).
The McNeill back-to-back converter [22] is rated at 150 MW (42 kV, 3600 A). The H300 HVDC valves each
contain 11 thyristor levels on the Alberta side and 12 on the Saskatchewan side, and are cooled directly by a
mixture of de-ionized water and ethylene glycol with a single-circuit cooling system – the first HVDC converter
station in the world to be cooled in this way.
In addition to the use of a single-circuit valve cooling circuit, the project incorporated a number of other firsts
in HVDC. Of particular note is that both AC systems are extremely weak, and under some conditions the
converter can operate with an Effective Short-Circuit Ratio (ESCR) of less than unity – the first HVDC scheme
ever to operate under such extreme conditions.
The very low AC system strength had a number of other consequences on the design of the converter.
Temporary overvoltages as a result of load rejection could be very high, and to mitigate this effect the
converter was designed to be able to operate in TCR Mode (Thyristor Controlled Reactor, by analogy with
the FACTS device of that name) for short periods of time. In TCR mode, the converter operates with zero
real power and a control angle close to 90°. Because this control action affects both sides of the HVDC link,
it is rarely used on HVDC transmission schemes, where telecommunications between the two ends cannot
be relied upon. On the other hand, for back-to-back converters, where the control equipment for both sides
is located in the same room, there is no reason not to take advantage of this additional controllability. This
TCR mode of operation has been carried forward to all other back-to-back converters built subsequently
by GE Vernova and its ancestors.
The very low AC system strength also imposed severe constraints on the bank size for the harmonic filters
in order to limit the size of voltage step imposed upon the AC system as a result of filter switching. Multiple,
switchable AC filters together with switchable shunt reactors and (on the Saskatchewan side) shunt
capacitors, of quite low MVAr ratings, were required in order to maintain the reactive power exchange between
the converter and the AC system within acceptable limits.
Because of the low MVAr ratings of each bank, it was decided to supply the AC harmonic filters and shunt
banks from a dedicated 25 kV winding on the converter transformer. As a result, the converter transformer
was of an unusual 3-phase, 4-winding construction.
However, probably the most important innovation introduced on this project was on the DC side of the
converter rather than the AC side, namely the absence of a DC smoothing reactor.
Every HVDC transmission (point-to-point) project contains a DC reactor whose functions are:
➙ Limitation of transient current due to DC line/cable faults or commutation failures
➙ Limitation of harmonic voltages on the DC line
➙ Protection of the converter from the effects of lightning strikes on the DC line.
However, for back-to-back HVDC converters, the reasons for including a DC reactor are less obvious. There is
no DC cable and hence no large DC-side capacitance to cause an unacceptably high transient current during
commutation failures. There is no DC line to be struck by lightning or from which telephone interference can
be induced. The only reasons for including a DC reactor on a back-to-back project are:
➙ To minimize cross-modulation of harmonics between the two AC systems
➙ To further reduce the transient current during a commutation failure.
Both of these can be rendered unnecessary with a well designed control system.
In choosing to omit the DC reactor from the McNeill back-to-back scheme, GEC was setting a trend which
would be continued by its successor companies to the present day. None of the turnkey back-to-back
projects executed by GE Vernova since McNeill have included a DC reactor.
1.2.7. Cheju-Haenam Submarine Cable HVDC Project

In 1991 GEC-Alsthom was awarded a turnkey contract to build the converter stations for the Cheju-Haenam
Submarine Cable HVDC scheme in South Korea. This was a bipole project rated at 300 MW,±180 kV [23].

CHAPTER
The island of Cheju (now written Jeju) is part of South Korea but separated from the mainland by about
100 km of sea. The island’s biggest industry was (and remains) tourism, with power generation provided at
the time mainly by diesel fuelled power stations. On the mainland however, South Korea had invested in
nuclear power, making the cost of generation much lower than on the island. KEPCO, the utility in charge
of power transmission, wanted to interconnect the island with the mainland power grid. Because of the
distances involved the solution had to be an HVDC submarine cable link. Two important criteria which
KEPCO stipulated were that they should be able to turn off the island generation at times of low load and in
the event of a blackout on the island, the island network should be re-startable using only the HVDC link. A
further requirement was that these functions had to be achievable without the benefit of telecommunications
between the two stations.
These criteria are difficult to meet with thyristor based HVDC schemes, where it is well known that a rotating
voltage reference must be available at all times in order to allow the converters to commutate – hence their
categorization as Line Commutated Converters. GEC-Alsthom overcame these obstacles with an innovative
solution involving a special control mode, modifications to the thyristor valve design and the provision of two
gas turbines at the island end in which the prime mover could be de-clutched and the generator continue to
spin, operating only as a synchronous compensator.
In order to operate the island AC network with the HVDC link providing the sole source of real power, the only
possible mechanism for controlling the island frequency was to modulate the power transmitted on the DC
link. This had to be achieved without the benefit of telecommunications, and with only the two quite small
(70 MVA) synchronous compensators operating. Moreover these two synchronous compensators had quite
low inertia, so in the event of a sudden load change on the island, the island frequency would tend to increase
or decrease very rapidly unless compensated by very fast control action of the HVDC link.
In a conventional HVDC converter, the rectifier controls the current and the inverter controls the voltage.
On this project, this traditional control mode was retained for the (unusual) operating mode where power is
transmitted from island to mainland. However, for the more usual operating mode where power is transmitted
from mainland to island, this traditional control mode was not fast enough to permit frequency control of the
island AC network in the absence of telecommunications.
The control of DC current can be executed very rapidly, whereas the control of DC voltage is slower because it
relies partly on the converter transformer tapchanger to obtain the correct transformer valve-winding voltage.
Consequently, it was decided to operate the island inverter such that it controlled current (and could therefore
provide fast power control), with the mainland rectifier controlling DC voltage. This was the opposite of the
usual approach and is believed to be the only HVDC scheme in the world configured in this way.
In steady-state, the two synchronous compensators at the island end provide the commutating EMF
(Electro-Motive Force) needed by the converters, even when all other generation on the island is shut
down. However, this left the problem of how to ‘black start’ the island AC system after a total blackout,
using only the HVDC scheme. The solution chosen was to equip the two synchronous compensators with
de-clutchable gas turbines. The gas turbines (rated at 55 MW) were provided solely for the purpose of
accelerating the synchronous compensators to 80% of rated speed and 80% of rated terminal voltage: once
these conditions are reached and the rectifier has been deblocked at reduced DC voltage, the gas turbines
can then be de-clutched. The inverter can then start-up using the (reduced) commutating EMF provided by
the synchronous compensators, and the HVDC scheme then transmits power to the island to accelerate the
synchronous compensators to full rated speed.
Another unusual feature of this start-up condition is that the rectifier station needs to operate at a DC power
of almost zero for some minutes. Most HVDC schemes are designed to operate only down to some minimum
power of typically 5-10% nominal power, since operation below this level leads to intermittent DC current.
Discontinuous (intermittent) DC current can result in dangerous thermal overload of some components in the
thyristor valve (mainly the damping resistors) under certain malfunction conditions. When discontinuous DC
current persists only for a few seconds, this can generally be tolerated, but on this project it could persist for
many minutes, so a novel switchable Breakover Diode (BOD) protection system was designed for the thyristor
valve to ensure that the damping resistors were not overloaded.

CHAPTER
Fig. 1.2n– Suspended H400 quadrivalve for the Konti-Skan Pole 1 replacement project
1.2.8. THE H400 SERIES HVDC THYRISTOR VALVE

After the Haenam-Cheju project, the H300 valve went on to be used in four more HVDC projects, all back-
to-back: three in India and one in Uruguay. However, by 2000 the H300 valve design was not as competitive
in the market-place as it had once been and the available thyristor technology had progressed. GE Vernova
therefore embarked on the development of the H400 valve which was to use the latest 125 mm 8.5 kV
thyristors. The H400 valve uses a simplified mechanical valve design in which the thyristors are grouped
into valve sections of five or six thyristors, each with a single ‘lumped’ di/dt reactor. The H400 valve was
the first GE Vernova valve to be designed primarily for a suspended arrangement. By suspending the
valve from the valve hall ceiling, the mechanical stresses on the valve during earthquakes are reduced
and lower-cost suspension insulators can be used. Most applications of the H400 valve are suspended,
but floor-mounting is still used in special cases, such as refurbishment projects (for example, when the
air-cooled valves of the Cross-Channel project were replaced by H400 valves in 2012) or where civil costs
favour floor-mounted valves.
The H400 valve was also the world’s first HVDC valve to be specifically designed to accommodate thyristors
made by more than one manufacturer. This flexibility has proven to be vital in adapting the valve to the rapidly
changing HVDC market of today.
The first scheme using the H400 valve was the Konti-Skan 380 MW Pole 1 replacement project, the valves
for which are shown in Fig. 1.2n.
The original Konti-Skan Pole 1 had been installed in 1965 by ASEA, but the owners wanted not only to
replace the converters, but to build two completely new converter stations, one in Sweden and one in
Denmark. The contract was awarded to GE Vernova and the project was commissioned in 2006.
1.2.9. THE HYDRO-QUÉBEC DE-ICER PROJECT

One of the most unusual applications for HVDC technology came in 2005 when Hydro-Québec of Canada
awarded GE Vernova (at the time, AREVA T&D) the contract to build the world’s first transmission line
de-icer at Lévis, near Québec City.
Hydro-Québec in Canada had suffered from severe power outages due to loss of parts of their power
transmission system during ice storms in January 1998, which, as a result of ice build up on the transmission
lines, had resulted in broken lines and damaged towers. Millions of consumers were left completely without

CHAPTER
Fig. 1.2o– View inside the valve hall at Lévis de-icer installation in Canada
power, some for several weeks in the middle of the Canadian winter. To prevent a recurrence of this problem,
Hydro-Québec embarked on an extensive program to reinforce their 315 kV and 735 kVac networks in the
crucial Montréal-Québec corridor.
One of the major measures taken by Hydro-Québec was to install a power electronics based transmission
line de-icer at Lévis substation. This equipment is, in principle, half a HVDC scheme (the rectifier only) and
during ice storms is connected to an AC transmission line, after removing the line from service. A high DC
current of up to 7200 A is circulated along the transmission line, through a deliberately-applied short at the
far end, and back again. The very high resistive heating of the transmission line conductors causes any ice
to melt and drop off. This process can be repeated in turn on up to five different transmission lines which all
radiate out from Lévis substation.
For this solution [17], GE Vernova (then operating as Alstom) chose its new H400 valve and installed two
6-pulse converter bridges to be connected in parallel, with a total power of 250 MW at ±17.4 kV. However,
the most novel aspect of this circuit was that, during the vast majority of its life, when not required to
operate as a de-icer, it operates instead as a Static VAr Compensator (SVC). The changeover between the
two modes uses motorized disconnectors to re-arrange the thyristor valves from de-icer mode (essentially
the standard HVDC configuration) to that of a delta-connected Thyristor Controlled Reactor (TCR).
HVDC converters (such as the McNeill back-to-back) had previously been used for short periods of time
in so-called TCR mode, but it is extremely inefficient to operate in this mode for long periods of time. The
re-arrangement of the thyristor valves at the Lévis de-icer installation allows TCR mode to be achieved with
much better efficiency than on other HVDC schemes, and close to the efficiency that can be obtained with
a purpose-built SVC.
Fig. 1.2o shows a view of the inside of the valve hall of the Lévis de-icer installation, showing the H400
valves and the disconnect switches used for the mode change.
1.2.10. THE GCCIA 3 X 600 MW HVDC PROJECT

In 2006 GE Vernova was awarded the contract to build three 600 MW back-to-back converters at the Al
Fadhili substation in Saudi Arabia. This HVDC scheme is part of the GCCIA interconnection project, which
interconnects Saudi Arabia (whose AC system operates at 60 Hz) with the 50 Hz Gulf states of Kuwait,

CHAPTER
Fig. 1.2p– The GCCIA interconnection project
Bahrain, Qatar, United Arab Emirates and Oman, as shown on Fig. 1.2p. Each of the three 600 MW back-
to-back converters is identical and completely independent, with a nominal DC voltage of 222 kV and a
nominal DC current of 2700 A – which is quite low by today’s standards, for reasons which will become clear.
At first sight this project appears to be a standard 50 Hz/60 Hz back-to-back frequency converter, similar to
the many such installations in Japan and South America. However, a number of special features of this project
[25] made it far from ordinary.
The converter station, the first HVDC installation in the Middle East, is located in the desert in Northern
Saudi Arabia, with intense dust storms and ambient temperatures of up to 55°C in summer. The very high
temperatures meant that the H400 thyristor valves needed to be operated at lower current than would be
usual, and the water coolant circulating system had to be dimensioned for very high water flow rates in order
to keep the thyristor valves cool. The cooling plants designed for this project were the largest (in terms of
flow rate) of any HVDC installation in the world to date. Being in the desert there is no access to water for
evaporative cooling, so only dry cooling was permitted.
However, the most unusual aspects of this project arose from the way in which the scheme was to be used.
Rather than simply being used for trading power between the 50 Hz and 60 Hz AC systems, the intention for
this converter station was that it would sit idle for much of the time, but be ready to transmit power in either
direction at very short notice (seconds) in response to the loss of a major generating station on either AC
system. The basic principle is for the station to be used to share the spinning reserve between the two AC
systems, a technique referred to as Dynamic Reserve Power Sharing (DRPS).
Normally, when a HVDC converter is required to transmit power at short notice, it is kept in the ‘energized
but blocked’ state, where the converter transformer is energized but the thyristor valves are not carrying
current. However, operating for long periods in this mode wastes a considerable amount of energy, so
GE Vernova’s engineers designed a special control strategy to allow the converters to be completely
de-energized, but automatically detect when one AC system suddenly needs power, and then go through
the normal energization and deblocking sequence much faster than normal.
Tests on the control system in the simulator laboratory confirmed that the scheme was capable of changing
from completely de-energized to the transmission of 600 MW (one converter pole) within one second and
the transmission of 1200 MW (two converter poles) within five seconds.

CHAPTER
1.2.11. RECENT HVDC PROJECTS
At the time of writing, the H400-series valve has been sold for a total of 17 commercial projects and the
versatility of the design is proving very valuable.
A variant of this valve, using 150 mm diameter thyristors [26] was used for the Lingbao2 back-to-back
project in China. This project has a nominal DC current rating of 4500 A (the highest nominal current
rating of any HVDC scheme until that time) and a nominal DC voltage of 167 kV. During the type tests in
GE Vernova's (then operating as Alstom Grid) synthetic test circuit in Stafford, a DC bridge current of 5600
A was successfully reached. The H400 valve was also used for the Ningdong-Shandong HVDC transmission
project in China, which was the world’s first HVDC project to operate at 660 kVdc and, when completed,
had the largest power rating of any single HVDC converter in the world (2000 MW per 12-pulse bridge).
Other notable recent projects have included Bipole 2 of the Rio Madeira HVDC transmission project in
Brazil (the world’s longest HVDC line) and the Champa-Kurukshetra 2 x 3000 MW, ±800 kVdc double-bipole
project in India.
Most recently, an evolutionary development of the H400 valve has been created, referred to as the H450,
using the same electrical components but a more efficient mechanical design and better external corona
shielding. To date the H450 valve has been sold for two HVDC projects in Korea, including the refurbishment
of the Cheju-Haenam HVDC link originally supplied by GEC Alsthom in the 1990s (Fig. 1.2q).
1.3. GE VERNOVA HVDC EXPERIENCE

GE Vernova's HVDC heritage draws upon three independent lines of ancestry as outlined in section 1.1.
The following sections list the HVDC projects installed by GE Vernova and all of its predecessor companies.
1.3.1. MERCURY-ARC SCHEMES

The first HVDC schemes, up until the mid 1970s, did not use thyristor valves but instead used mercury-arc
rectifier valves. English Electric built three HVDC schemes using mercury-arc valves, which are described
in the following section.
Fig. 1.2q– A converter in the Cheju-Haenam HVDC refurbishment project using the updated H450 valve.

CHAPTER
1.3.1.1. The Sardinia - Italy Mainland DC Link
Project Customer Country Type MW kV Date
Cable +
SACOI ENEL Italy 200 200 1967
Overhead Line
English Electric
Bulk power transfer between independent systems of Sardinia, Corsica & Italy by overhead line (292 km) and submarine
cable (121 km)
1.3.1.2. Kingsnorth HVDC Link

Kingsnorth National Grid England Cable 640 ±266 1972
English Electric
Bulk power transfer from Kingsnorth generating station to metropolitan London via an 82 km underground cable. The scheme
fed two receiving stations and supports the adjacent AC system without increasing short-circuit levels.
The main function of the link was urban network reinforcement within an AC interconnected system. Normal transmission
direction was from Kingsnorth to Beddington and Willesden, however transmission between Beddington and Willesden
was also possible (reversible) using an interconnection mode.
Many innovative features were incorporated into the design of this scheme, as described in section 1.2.1.
1.3.1.3. Nelson River Bipole 1 HVDC Transmission

1972 to
Nelson River Manitoba Hydro Canada Overhead Line 1620 ±450
1977
English Electric
Bulk power transfer by 930 km overhead line from remote hydro-electric generation on Nelson River to Winnipeg load centre.
The Nelson River flows north, down from the Canadian prairies to the Hudson Bay and this is where Manitoba Hydro
developed its hydro generation plants, to supply power to the entire Province and its neighbors, including the USA. As there
is a long transmission distance involved, the potential for seasonal bush fires and lightning strikes created the concern
that an AC transmission system would suffer disconnection problems causing poor reliability. DC transmission became
the preferred option.
The HVDC link was installed in 1972, with a rating of 1,620 MW, transmitted on the 890 km of HVDC overhead lines to
Winnipeg. At 450 kV, created from three 150 kV bridges in series per pole, which is the highest DC voltage ever used by
mercury-arc valves.
The scheme transmits half the total generated power in Manitoba and is controlled to assist AC stability.
The bipole 1 project was implemented in three construction stages (810 MW in 1972, 270 MW in 1973 and 540 MW in 1977).
The original installation included:

➙ Converter transformers: 341 MVA, 138 kV, and 323 MVA, 230 kV
➙ Synchronous compensators: six 160/-80 MVA at 17 kV
➙ AC filters: 230 MVA, tuned 2x5th, 2x7th, 11th, 13th and high-pass
➙ AC filters: 350 MVA, tuned 2x5th, 2x7th, 2x11th, 2x13th and
2 high-pass
➙ DC filters (both terminals): tuned 6th and 12th
➙ Smoothing reactors: two per pole per terminal each 0.5 H
➙ Land electrodes: each 305 m diameter ring of steel in coke, 0.16 to 0.60 ohms
1.3.2. OIL-COOLED THYRISTORS

From the late 1960s onwards, the advantages of using solid-state, semiconductor technology using
thyristors in place of mercury-arc valves were starting to become clear. The newer technology posed the
challenges of how to cool the thyristors and how to maintain electrical insulation between different parts
of the structure.

CHAPTER
In power transformers these challenges were solved by immersing the core and windings in mineral oil, and
early designs of thyristor valves followed the same principles, with the thyristors and associated components
being immersed in a tank of insulating oil. Several manufacturers, including English Electric (UK) and AEG
(Germany), produced designs of oil-insulated, oil-cooled thyristor valves, as described below.
1.3.2.1. Cross-Channel, UK-France Cable Link

Cross-Channel National Grid England Cable 80 100 1972
English Electric
Replacement of mercury-arc valve with a thyristor valve in England – France 64 km submarine cable scheme, as
described in section 1.2.4.1 above.
1.3.2.2. Cahora Bassa HVDC Link

South Africa
Cahora Bassa Eskom/HCB Overhead Line 1920 ±533 1979
Mozambique
AEG-Telefunken
Bulk power transfer interconnection between Songo in Mozambique and Apollo, near Johannesburg, South Africa, by
1420 km of overhead line.
The utilities involved were Hidroelectrica de Cahora Bassa, Mozambique (220 kV) and the Electricity Supply Commission,
Eskom, Johannesburg, South Africa (275 kV)
The suppliers were the ZAMCO consortium, consisting of AEG-Telefunken, BBC and Siemens AG, Germany.
Commissioning dates were
Stage I: March 1977 (4 bridges)
Stage II: April 1978 (2 bridges)
Stage III: June 1979 (2 bridges)
The configuration is a bipole, 1920 MW at ±533 kVdc and 1800 A/pole, with no overload capability, and earth return is
available for monopole operation.
Single-phase, 2-winding transformers are used at both terminals, rated at 96.7 MVA.
No DC filters were initially installed, but were added later by modifying the DC surge capacitors.
The thyristor valves were originally installed as a double valves, oil-insulated and oil-cooled
Stage I: (4 bridges) 280 thyristor levels in series connection per valve arm and two thyristors in parallel, giving a total for
one double valve of 560 thyristor levels or 1120 thyristors.
For a 6 pulse converter unit this makes 3360 thyristors for 133 kVdc and 1800 A.
Stages II and III: (4 bridges) 192 thyristor levels in series connection per valve arm, two thyristors in parallel, giving one
double valve a total of 384 thyristor levels or 768 thyristors.
For a 6-pulse converter unit this makes 2304 thyristors for 133 kV and 1800 A.
Each terminal has a total of eight 6-pulse bridges, giving a total of 11,328 thyristor levels or 22,656 thyristors.
AC harmonic filters are provided at each terminal, designed as two identical filters with total ratings of 210 MVAr at Songo,
and 195 MVAr at Apollo.
1.3.3. AIR-COOLED THYRISTOR VALVES

Oil-insulated thyristor valves proved to be difficult and inconvenient to maintain, requiring sophisticated
on-site facilities to remove the active components from the oil tank in order to affect repairs. The next
technological move was therefore to use air for cooling and insulation. GEC and CGEÉ Alsthom both
produced designs of air-cooled, air-insulated thyristor valves, as described in the following section.

CHAPTER
1.3.3.1 Eel River HVDC Link
Hydro Quebec /
Eel River New Brunswick Canada Double Back-to-Back 2 x 160 82 1972
Power
GE Vernova
This Back-to-Back HVDC interconnection is located in New Brunswick,

Canada, and forms an asynchronous link between the Hydro Quebec
and the New Brunswick networks to feed hydro generation from
Quebec.
The converters were designed with 10% continuous overload
capability. There are 2 independent back-to-back systems, each
rated at 160 MW. The converters consist of 2 thyristor valve structures
housed indoors in each valve hall, with each valve having 5 modules in
series connection, each module having 10 thyristor levels in series and
each thyristor level has 4 thyristors connected in parallel. This makes
a total of 1200 thyristors per 6-pulse converter unit.
The valves are air-insulated and air-cooled. The valve cooling system
consists of two closed loops and an evaporative-assisted outdoor
cooler. The heat from the air is moved from the lower plenum heat
exchangers by means of a pumped water-glycol loop to the heat-
exchanger in the outdoor evaporative cooler.
DC Reactor Rating 0.14H at 2000A
On the HQ side there are two 20.5 Mvar filter banks connected at
13kV, each bank having sub-banks tuned at 5th, 11th, 13th, HP. Also
on the HQ side there is a single 45Mvar HP filter connected at 230 kV
On the NB side there are two filter banks connected at 13kV, one bank
rated at 34.5Mvar having sub-banks tuned at 11th and HP, and the
second rated at 30.5Mvar having sub-banks tuned at 13th and HP.
Also on the NB side there is a single 45Mvar HP filter connected at 230 kV
There are also a number of Synchronous Condensers on this site to mitigate the weak systems on both sides, two rated
at 110Mvar connected at 13kV on the HQ side, and one rated at 110Mvar connected at 13kV on the NB side.
The Converter Transformers are 1-phase, 4-winding units, with the following rating data:
➙ 55 MVA, 230±12%/35.4/35.4/13 kV
➙ 55 MVA, 230±12%/35.4/35.4/13 kV
1.3.3.2 Square Butte HVDC Link

Square Butte Minnesota Power USA Overhead Line 2 x 250 250 1977
GE Vernova

CHAPTER
This 749-mile overhead line interconnection links the 230 kV AC system at Center, N Dakota and the 230 kV network at
Arrowhead, Minnesota. Its purpose is to transmit coal-by-wire, from the MP-owned coal fired generation plant in N Dakota.
The thyristor valves for this bipole system are air-cooled, housed indoors and in the form of quadrivalve structures, each
valve consisting of 12 modules. The valve modules consist of 12 thyristors in series, giving a total of 144 thyristors per
valves, and therefore 1728 thyristors per pole at each converter station.
The valves are cooled by a forced air primary cooling loop which transfers heat from the valve assemblies to a secondary
cooling loop employing a water-glycol solution. The secondary loop transfers heat to the atmosphere via outdoor
evaporative coolers.
The converter transformers at both stations are configured as 1-phase, 2-winding units, a total of 6 units per pole.
➙ N Dakota Terminal: 52 MVA, 235 kV {+15 %/ -5 %} / 110 kV
➙ Minnesota Terminal: 48 MVA, 234 kV {+10% / -10%} / 102 kV
The AC harmonic filters at each terminal are the same:
Two 11th and 13th filter sub-banks, each rated 39 Mvar
Two HP filters, each rated at 30 Mvar
There is one DC filter on each pole, and each is double tuned to 12th and 24th harmonic.
The DC Reactor is placed on the High Voltage circuit, and rated at 0.06 H, 250 kV, 1000A.
1.3.3.3 David A. Hamil HVDC Link

Tri-State
Generation &
David A. Hamil USA Back-to-Back 100 50 1977
Transmission
Association
GE Vernova
This Back-to-Back HVDC interconnection forms an asynchronous link between the WECC and Eastern Interconnection
networks. The link feeds energy to a relatively weak part of the Rocky Mountains area network.
The converters were designed with 10% continuous overload capability. The converters consist of 2 thyristor valve
structures housed indoors in one valve hall, with each valve having 3 modules in series connection, each module having 10
thyristor levels in series and each thyristor level has 4 thyristors connected in parallel. This makes a total of 120 thyristors
per valve arm or 720 thyristors per 6-pulse converter unit.
The valves are air-insulated and air-cooled. The valve cooling system consists of two closed loops and an evaporative-
assisted outdoor cooler. The heat from the air is moved from the lower plenum heat exchangers by means of a pumped
water-glycol loop to the heat-exchanger in the outdoor evaporative cooler.
There is 1x AC Harmonic Filter on each side, rated at 30Mvar, tuned to 11th harmonic.
In addition, there is 1x 30Mvar shunt capacitor on each side.
The Converter Transformers are 3-phase, 3-winding units, connecting to the 230kV AC network voltage on the line side
windings.
Transformer Data: 123.3 MVA, 230+16%-6%/23.8/23.8 kV
1.3.3.4 Vancouver Island HVDC Link

Vancouver Island
BC Hydro Canada Overhead Line / Cable 1 x 370 280 1979
Pole 2
GE Vernova

CHAPTER
This HVDC system connects the mainland converter station at Arnott, Delta,
Vancouver to Duncan, Vancouver Island, a total circuit length of 73.6 km,
comprising both overhead line and submarine cable sections. The scope
of this project was to add a second pole to the existing pole 1 mercury arc
converter.
The thyristor valves for the pole 2 converters are air-cooled, housed indoors
and in the form of double valve structures, each valve consisting of 15 modules,
and each module contains 12 thyristors. This gives a total of 2160 thyristors in
each converter.
The valves are cooled by a forced air primary cooling loop which transfers heat
from the valve assemblies to a secondary cooling loop employing a water-glycol
solution. The secondary loop transfers heat to the atmosphere via outdoor
The converter transformers at both stations are configured as 1-phase,
2-winding units, a total of 6 units per pole.
Vancouver Terminal: 83.1 MVA, 236 {+ 15% / - 13%} / 119.2 kV
Vancouver Island Terminal: 83.1 MVA, 236 {+ 15% / -13%} / 119.2 kV
The AC harmonic filters at each terminal are the same, with one 94 Mvar bank, with sub-banks tuned to 5th, 7th, 11th,
13th, HP.
There are no DC filters.
The DC Reactor is placed on the High Voltage circuit, and rated at 0.068 H, 280 kV, 1320A.
1.3.3.5. Kingsnorth HVDC Link Development

Kingsnorth National Grid England Cable 266 133 1981
GEC
Replacement of mercury-arc valve with an air-cooled thyristor valve in Kingsnorth underground cable scheme. This
was a prototype of the valve to be later installed at the Sellindge converter station of the Cross-Channel scheme, as
described in section 1.2.4.2 above.
The Kingsnorth scheme was decommissioned in 1987.
1.3.3.6. Eddy County HVDC Link

El Paso Electric,
Texas New
Mexico Power
Eddy County USA Back-to-Back 200 82 1983
Company,
Southwestern
Public Service
GE Vernova

CHAPTER
This Back-to-Back HVDC interconnection is located 9.5 miles east from Artesia, New Mexico, and forms an asynchronous
link between the WECC and Eastern Interconnection networks. The link feeds coal fired generation from the Southwestern
Public Service Company, in the Eastern Interconnect into the WECC network.
The converters were designed with 10% continuous overload capability. The converters consist of 2 thyristor valve
structures housed indoors in one valve hall, with each valve having 3 modules in series connection, each module having 24
thyristor levels in series and each thyristor level has 2 thyristors connected in parallel. This makes a total of 576 thyristors
per 6-pulse converter unit.
There are two AC Harmonic Filters on the 230kV side, each rated at 30Mvar, tuned to 11th harmonic.
In addition, there are two 30Mvar shunt capacitors on the 230kV side.
On the 345 kV side there are two High Pass filters.
➙ 273 MVA, 230/35.4/35.4 kV
➙ 273 MVA, 345/35.4/35.4 kV
1.3.3.7. Oklaunion HVDC Link

Public Service
Company of
Oklaunion USA Back-to-Back 200 82 1984
Oklahoma / West
Texas Utilities
GE Vernova
This Back-to-Back HVDC system forms an asynchronous interconnection between the ERCOT (Texas) and Eastern
Interconnection networks, both connections are at 345 kV. The main function of this system is to allow energy trading
between the two networks to take advantage of variation in generation and load patterns.
The valves are rated at 200 MW, and designed with 10% continuous overload capability, 25% for 1 hour. There are 8
thyristor levels in each module, and 2 parallel thyristors at each thyristor level. Three modules are connected make up a
complete valve, leading to a total of 1152 thyristors in the HVDC system.
The converter transformers are 3-phase 3-winding units, with a voltage rating of 345 / 35.4 / 35.4 kV.
The AC harmonic filters on the Oklahoma side comprises two 30 Mvar 11th and 13th HP filters, and a 30 Mvar shunt
capacitor. On the Texas side there are two 30 Mvar 11th and 13th HP filters, and a 30 Mvar shunt reactor.
1.3.3.8. Madawaska HVDC Link

Hydro Quebec /
Madawaska New Brunswick Canada Back-to-Back 350 130.5 1985
Power
GE Vernova

CHAPTER
This Back-to-Back system links the Hydro Quebec 315 kV network to the New Brunswick Power 345 kV network, allowing
an asynchronous interconnection to feed hydro generation into the NBP system.
The converters were designed with approximately 25% overload capability.
The converter consists of thyristor valve structures housed indoors in the valve hall, with each valve having 6 modules
in series connection, each module having 16 thyristor levels in series. This gives a total of 2304 thyristors in the system.
DC Reactor Rating 2 x 0.05H at 2500A
On the HQ side there are three 40.6 Mvar filter sub-banks connected at 315kV, each sub-bank tuned at 12th harmonic,
and three 30.1 Mvar filter sub-banks tuned at 24th harmonic, also connected at 315 kV. A total of 212.1 Mvar.
On the NB side there are two 55.6 Mvar filter sub-banks connected at 345kV, each sub-bank tuned at 12th harmonic, and
two 55.3 Mvar filter sub-banks tuned at 24th harmonic, also connected at 345 kV. A total of 221.8 Mvar. Also, on the NB
side there are two 55 Mvar shunt reactors connected at 345 kV.
➙ 134.2 MVA, 315±6%/55/55 kV
➙ 134.2 MVA, 345±6%/55/55 kV
1.3.3.9. Miles City HVDC Link

Miles City WAPA USA Back-to-Back 200 82 1985
GE Vernova
This Back-to-Back HVDC system forms an asynchronous interconnection

between the WECC and Eastern Interconnection networks. The main
function of this system is to allow energy trading between the two
networks to take advantage of variation in generation and load patterns.
The valves are rated at 200 MW and designed with 10% overload capability.
The valves are cooled by a forced air primary cooling loop which transfers
heat from the valve assemblies to a secondary cooling loop employing a
water-glycol solution. The secondary loop transfers heat to the atmosphere
via outdoor evaporative coolers.
The converter transformers are 3-phase 3-winding units, with the following
ratings:
➙ 239 / 119.5 / 119.5 MVA
➙ 230 / 34.13 / 34.13 kV
Tap range +12.5% / - 5.7%, steps 32
The AC harmonic filters are comprised of four double-tuned and two high-
pass filters on each of the east and west side 230 kV systems. The filters for
each side are identical and tuned to approximately the 3rd and 11th, 5th
and 13th, 7th and 23rd, 25th and 47th, high-pass 18th and high-pass 36th,
harmonics with a total reactive rating of 75 Mvar capacitive on each side.
The DC Reactor is an oil-filled unit rated at 0.074H at 2476A.

CHAPTER
1.3.3.10. Des Cantons - Comerford HVDC Link
Hydro Quebec
Des Cantons -
/ New England Canada - USA Overhead Line 2 x 345 ±450 1986
Comerford
Power Company
GE Vernova
The Phase 1 Bipole HVDC system links the Hydro Quebec and New England asynchronous AC networks between the Des
Cantons converter station at Sherbrooke, QC and the Comerford converter station at Monroe, NH. The purpose was to
transmit hydro energy from Quebec to the Boston load center. The total DC circuit length is 172km, including a buried
cable in a tunnel under the St Lawrence River.
The converters are connected to the AC systems on both sides of the link at 230 kV.
Converter Transformers
➙ The transformers are 1-phase, 3-winding units
➙ Des Cantons: 328.2 MVA, 230 kV {+12.4% / -7.4%} / 180.7 / 180.7
➙ Comerford: 167 MVA, 230 kV
AC Harmonic Filters
Des Cantons:
➙ Total filter rating 244.6 Mvar
➙ 2 Filter Banks, each with sub-banks tuned at 3rd (2 x 13.7 Mvar), 11th (2 x 30.2Mvar), 13th (2 x 29.5 Mvar), 24th (2 x
48.9 Mvar) harmonics
➙ 2 Capacitor Banks
Comerford:
➙ 4 Shunt Capacitor Banks, each rated at 31.5 Mvar
➙ 2 Shunt Capacitor Banks, each rated at 63 Mvar
➙ 2 Filter banks each rated at 63 Mvar, tuned to 2nd +25th, 23rd, 11th +37th, 3rd + 13th
➙ In addition, there are twelve 19 Mvar shunt reactors connected at 13.8 kV.
DC Filters
Des Cantons: Single Pole operation: Triple tuned filters, Bipole Operation: Double tuned filters
Comerford: One 6th harmonic, and One 12th harmonic shunt connected DC filter on each pole.
Thyristor Valves
➙ The air-cooled thyristor valves are arranged in quadrivalve structures.
➙ Each 12-pulse converter has 6 quadrivalve structures, with 16 modules per valve.
➙ Each module contains 8 thyristors, giving a total of 3072 thyristors for the bipole at each converter station.
Valve Cooling

CHAPTER
1.3.3.11. Cross-Channel, UK-France Cable Link
Cross-Channel
National Grid/EdF England-France Cable 2x1000 ±270 1986
(IFA 2000)
GEC and CGEÉ-Alsthom
HVDC was chosen as the preferred option as it links the networks asynchronously, and contracts were awarded for the
construction of HVDC converter stations on each side of the channel, rated at 2000 MW in a double bipole configuration
with eight submarine cables at 270 kVdc.
In both stations, the four poles, each of 500 MW rating (270 kVdc, 1852 Adc) arranged as two 1000 MW bipoles, are in
separate valve halls and linked by a central control block. Each pole has a single 12-pulse converter, with the twelve valves
housed in three indoor quadrivalve structures.
The UK converter station is situated at Sellindge, close to the Kent coast. The solution at Sellindge covers a 14 hectare
site, and includes 400 kV harmonic filters, two Static VAr Compensators (SVC), and an indoor, 400 kV GIS substation. The
Sellindge installation includes:
➙ 125 valve modules per valve, each module comprising two parallel-connected 56 mm air-cooled thyristors
➙ Two 3-phase, 2-winding, converter transformers per pole, rated at 316 MVA each
➙ Two C-type and two second order damped filters per bipole, each of 130 MVAr
➙ 15 GIS bays, 63 kA 400 kV arranged in double busbar configuration
➙ Two ±150 MVAr Static VAr Compensators, each with 500 MVAr 0.5 sec absorption capability to mitigate
overvoltages caused by faults on the France side.
The French converter station is at Les Mandarins, close to Calais, and includes:
➙ 12 valve modules per valve, each module comprising eight thyristor levels in series
➙ Each thyristor level has two parallel-connected 75 mm air-cooled thyristors
➙ Three single phase, 3-winding converter transformers per pole, rated at 206/103/103 MVA
➙ AC Harmonic Filters with a total installed rating of 1280 MVAr, divided into eight (high-pass, damped type) filters
of 160 MVAr each
➙ One DC reactor per converter, rated for ±270 kVdc, 370 mH at 1852 A.

CHAPTER
In the 1980s and 1990s the £500m project gave both the CEGB and EdF a power reserve of 1500 MW, assuming that three
of the four x 500 MW poles would be in service at any time, and was successfully used to exchange energy between the
two countries.
Today the link retains the accolade of being the most highly utilized and highest rated power submarine DC connection
in the world, with eight parallel 270 kV cables. It is in constant use as a power trading facility, being able to rapidly and
automatically switch in either direction.
The link delivers the equivalent power, in each country, of a large power station and is designed to have minimal impact
on the environment. No new overhead transmission lines were needed, and the only major construction in the UK is the
converter station, a fraction of the physical size of a power station of equal rating.

CHAPTER
1.3.3.12. Sardinia - Italy Upgrade
Cable + 1986/
SACOI ENEL Italy 300 200
Overhead Line 1992
CGEÉ-Alsthom
Bulk power transfer between independent systems of Sardinia, Corsica and Italy by overhead line (292 km)
and submarine cable (121 km). The first stage was completed in 1967 using mercury-arc valves (see
section 1.3.2.1). In 1986, a 50 MW tap, using air-cooled thyristor valves, was added in Corsica to make the scheme
the first multi-terminal HVDC link in the world. In 1992 the original mercury-arc valves at the two ends of the scheme
were replaced by thyristor valves and the scheme rating increased to 300 MW.
The interconnection is between Codrongianos on Sardinia Island, Lucciana (Bastia) on Corsica Island and Suvereto on the
Italian Mainland, with a total route length of 385 km.
The three stations are connected in parallel on a line which is operated at rated voltage of 200 kV: earth and sea return
were used.
The converter stations are rated and configured as follows:
Codrongianos: 300 MW at 200 kVdc and 1500 A. Anode situated on the Sardinian coast.
Suvereto: 300 MW, at 200 kVdc and 1500 A. Cathode situated on the coast.
Lucciana: 50 MW, at 200 kVdc and 250 A. Earth electrode on the shore used as an anode or a cathode.
The purpose of the link included:
➙ Frequency support for the Sardinian AC network by power/frequency control
➙ Energy supply, frequency control and static spinning reserve on Corsica
The power flow direction between the two main stations (Codrongianos and Suvereto) dictates the polarity of the DC
line voltage. The direction of the flow on the Corsican network was made independent of this polarity by reversing the
connection of the Corsican converters by high-speed isolating switches.
Automatic power reversal between the main stations is possible by a sequence eliminating the Corsican converters for
500 ms.
1.3.4. WATER-COOLED THYRISTOR VALVES

Although air-cooled, air-insulated thyristor valves solved the maintainability problems inherent with oil-
insulated, oil-cooled thyristor valves, they became considerably more bulky, as a consequence of both the
poorer dielectric withstand capability and poorer thermal properties of air in comparison with insulating oil.
Today, air remains the medium of choice for insulating between different parts of a thyristor valve, but as early
as the 1970s it was realized that de-ionized water would become the preferred method for cooling.
The development of water-cooled thyristor valves commenced in the late 1970s or early 1980s in both GEC
and AEG, and water-cooled valves have been applied in many projects since then, as described below.

CHAPTER
1.3.4.1. Nelson River Bipole 2 HVDC Link
1977 to
Nelson River Manitoba Hydro Canada Overhead Line 2000 ±500
1985
AEG-Telefunken
Bulk power transfer by 930 km of overhead lines from remote hydro-electric generation on Nelson river to the Winnipeg
load centre. Bipole 2 Project implemented in three stages.
Interconnection between Henday, near Gillam, and Dorsey, near Winnipeg, both in the Province of Manitoba, Canada.
The bipole 2 rating is 2000 MW at ± 500 kV / 2000 A with a total route length of 930 km.
Each valve contains 16 thyristor modules and eight valve reactor modules and each thyristor module has six thyristor
levels in series and two in parallel connection.
Each valve therefore contains 96 thyristor pairs in series, including five redundant levels.
The thyristor modules are air-insulated and water-cooled.
1.3.4.2. Dürnrohr Back-to-Back HVDC Link

Österreichische
Dürnrohr Austria Back-to-Back 550 145 1983
Elek. AG
AEG-Telefunken
Asynchronous interconnection between Austria and the Czech Republic

The interconnection is designed as an asynchronous HVDC coupling system, between the West Europe Network (UCPTE)
(420 kV/50 Hz) and the East Europe Network (CMEA) (400 kV/50 Hz) at Dürnrohr, near Vienna, Austria.
The function was for the exchange of energy between Eastern and Western Europe.
The link rating was 550 MW at 145 kVdc and 3790 A, with an overload capacity of 15 % at ambient temperatures below
10 °C.
Converter transformers: two 3-phase transformers are used for each 12-pulse converter, rated at 335 MVA, 401/62.5 kV,
50 Hz
The thyristor valves are water-cooled and air-insulated, with 44 thyristors in each valve.
This link was decommissioned following the re-integration of the eastern and western European networks in the 1990s.

CHAPTER
1.3.4.3. McNeill Back-to-Back HVDC Link
McNeill Alberta Power Ltd Canada Back-to-Back 150 42 1989
GEC
Asynchronous interconnection between Alberta and Saskatchewan.

The eastern and western power systems of Canada operated as completely separate entities and both systems were
independently connected to the corresponding power systems in the USA. In the 1980s, the potential for creating revenue
from trading energy and reserve power was growing and savings could be made by removing the need for maintaining a
reserve generation capacity.
These were prime movers for Alberta Power Limited (APL) and the Saskatchewan Power Corporation (SaskPower) to
instigate a link between the two Canadian power systems.
The McNeill HVDC link was awarded to GEC-Alsthom in December 1987 to provide an asynchronous power connection
between the Alberta and Saskatchewan provinces, requiring a 150 MW back-to-back HVDC converter station to be
built on the Alberta / Saskatchewan border.
To overcome difficulties with the weak systems on both sides of this link, the design incorporated many significant
innovations, which are fully described in section 1.2.5.
The full project specification included:
➙ 150 MW back-to-back mono-pole
➙ No DC smoothing reactor
➙ 3-phase 4-winding transformers
➙ 25 kV tertiary busbars for all filters and reactive power banks
➙ Water glycol mixture in a single loop cooling system, safe to -50ºC
➙ Transient overvoltage limited by multi-column surge arresters, permanently connected at 25 kV
➙ Equivalent Short-Circuit Ratio (ESCR) drops to 1.0, on Saskatchewan side when operating as an inverter.
1.3.4.4. Nelson River HVDC Link Pole 1 Refurbishment

Nelson River 2x334 1991 to
Manitoba Hydro Canada Overhead Line 500
VG11+VG12 in stages 1992
GEC-Alsthom
Replacement of two, 6-pulse mercury-arc valve groups with thyristor valves.
VG13 334 500 1993

GEC-Alsthom

CHAPTER
Replacement of the 3rd 6-pulse valve group mercury-arc valves with thyristor valves.
After 20 years of service the original mercury-arc valves were nearing the end of their operational life span and were
becoming expensive to maintain. In addition, pressure on utility companies to be more efficient and competitive, plus the
opportunity of increased power sales to the USA, drove Manitoba Hydro to upgrade pole 1 of bipole 1.
For the first upgrade GEC-Alsthom added automatic switching to the station controls between two parallel lines
for when a fault occurs. This was later extended to bipole 2 and was the first time four-terminal DC transmission
was achieved on one transmission line. This feature proved its worth in 1996 when strong winds blew down both
transmission lines and the repair of one was sufficient to quickly resume full power.
The refurbishment was carried out in stages between 1990 and 1993 and included refurbishment of the valves of pole
1 using 100 mm thyristors, and control system upgrades to achieve increased power rating of the three series valve
groups. This gave pole 1 of bipole 1 an increase in transmission ratings from 450 kV to 500 kV, and in power from 810 MW
to 1,000 MW.
1.3.4.5. Haenam-Cheju 300 MW HVDC Cable Link

Cheju Island KEPCO South Korea Cable 300 ±180 1994
GEC-Alsthom
Bulk power transfer by 100 km submarine cable from mainland grid at Haenam to Cheju island and replace existing local
generation.
The island of Cheju is 100 km off the coast of South Korea, it has no installed natural energy resource of its own, and the
energy costs are therefore very high. The expansion of tourism on the island caused a significant growth in power demands,
creating a need for additional power generation. However with tourism a key driver for development, the use of traditional
diesel and oil-fired steam generation did not match the environmental and aesthetic restrictions.

CHAPTER
This was the first time HVDC technology had been used in Korea. Two converter stations were needed for the 300 MW
HVDC interconnection: one at Haenam on the mainland and the other close to the coast on Cheju island.
As the HVDC link was the sole source of power to the isolated island, innovative design features were needed to ensure the
minimum risk to supply. Due to the converter station’s proximity to the sea, it was essential that the design of equipment
took into consideration the extreme levels of salt contamination. As a further precaution, at the Cheju terminal the AC
filters are accommodated indoors for added protection against the saline elements.
Since the aim of the project was to allow the local generation on the island to be shut down and the HVDC station to
supply the complete power consumption of the island, even in the absence of telecommunications between the two
ends, the scheme included a number of novel and innovative features, further details of which can be found in section
1.2.6.
HVDC technical data
➙ 300 MW ±180 kV bipole
➙ Two 101 km 800 mm2 sea cables, single wire-armored
➙3 -phase, 188 MVA star/star/delta converter transformers
➙ Four 27.5 MVAr 154 kV filters per terminal
➙ Two 70 MVA synchronous compensators at Cheju
➙S ea electrodes rated 834 A continuous and 1530 A for 10 seconds, 15 km from the converter stations
➙ Minimum ESCR (Cheju): 1.45 (1.10 monopolar)
➙ Valves: H300 type, 100 mm, 5.2 kV thyristors, 46 per valve (Haenam) or 48 per valve (Cheju)
➙ Thyristor valve cooling: 3:1 water/glycol.
1.3.4.6. Chandrapur Back-to-Back HVDC Link

Chandrapur PGCIL India Back-to-Back 2x500 205 1997
GEC-Alsthom
Interconnection between the Western and Southern electricity networks of India. India’s Southern and Western power
regions had originally operated as separate power networks.
The Western region’s power reserves come from coal sources and are drawn upon all year round for power generation.
The Southern region relies heavily on generation from hydro sources which are plentiful during the monsoon but leave
power shortages in other seasons.
The opportunity to trade energy and reserve power between the two regions would aid the South, by enabling them to
use the coal reserves of the West, and help the West conserve these valuable reserves during the generation surplus of
the South in wetter periods.

CHAPTER
The Power Grid Corporation of India Ltd. (PGCIL) was given the responsibility for establishing a connection between the
two networks and an AC tie was constructed.
This operated for some time but proved to be unreliable due to instability between the two differing transmission networks,
causing frequent trips of the interconnection and power outages.
In 1993, following a competitive tendering process, GEC-Alsthom was awarded the £140 million turnkey project to
design and build a new HVDC power connection between the two regions with a completion timescale of just three
years.
GEC-Alsthom designed and supplied all the elements for the HVDC converter stations and AC switchyards and managed
the overall project, whilst units in India were employed for transportation and installation.
The back-to-back HVDC link is situated at Chandrapur in the state of Maharashtra. The scheme comprises two 500 MW
poles, each operating at 205 kVdc, 2475 A, together with conventional AC switchyards at each end of the link.
The complete back-to-back HVDC link comprises:
➙ Two identical poles, totaling 1000 MW continuous rating
➙ Pole parameters: 500 MW, 205 kVdc, 2475 A
➙ 424 MVAr of shunt capacitance on each side of each 500 MW converter, configured as four switchable units of 106
MVAr
➙ Individual harmonic distortion limit of 1%
➙ Total harmonic distortion limit of 4%
➙ Overload capability of 10% for two hours and 33% for five seconds
➙ Additional overload capacity when ambient temperature is below maximum
➙ Rapid power reversal
➙ Power flow can be modified to enhance system operation. e.g. to damp frequency swings following a disturbance
in either AC system
➙ Thyristor valve cooling: closed loop pure water cooling system
➙ Transformers: single phase 400/93/93 kV 234 MVAr each
➙ Valves: H300 type, 54 100 mm thyristors per valve, each rated at 5.2 kV
➙ 400 kV switchyard
➙ SCADA system.
Experience gained through the design and installation of HVDC stations for similar applications around the world led to
GEC-Alsthom’s decision to omit the DC reactors, thus saving on project construction and operating costs.
Power equipment in India is subject to large fluctuations in voltage and supply frequency, so the high stability of the
HVDC converter and its control of DC power gives safe, reliable connection and operation of the existing 400 kV systems
on both sides.
Under transient operation, the converter operates in a reactive power control mode to limit overvoltages on the AC
systems.
Coordinated control of the converter and the harmonic filters enables effective, steady-state control of the AC voltage/
reactive power exchange across the two AC systems.

CHAPTER
1.3.4.7. Visakhapatnam Back-to-Back HVDC Link
Visakhapatnam PGCIL India Back-to-Back 500 205 1999
ALSTOM T&D
Interconnection between the Eastern and Southern electricity networks of India. The design details and function of this
link are very similar to those of the Chandrapur back-to-back installation, except that the Visakhapatnam (Vizag) link is
a single monopole installation.
1.3.4.8. Rivera Back-to-Back HVDC Link

Rivera U.T.E. Uruguay Back-to-Back 70 22 2000
ALSTOM T&D
Interconnection between Uruguay (50 Hz) and Brazil (60 Hz).

A major electricity transmission and development program initiated by Uruguay’s national power utility provided for the
interconnection of their 50 Hz power system with the 60 Hz system in Brazil.
The project was financed by the Inter-America Development Bank and was awarded on a turnkey basis by The National
Administration of Electricity Plants and Transmission in Uruguay (U.T.E.).
This HVDC link is very similar to the McNeill converter station (see section 1.3.4.3) using many common design features
including 4-winding transformers, creating a MV filter busbar using vacuum switchgear, to provide relatively fine control
of the AC voltages on each side of the link.
The project consists of a 50 Hz switchyard with harmonic filter bays connected to the Uruguayan 150 kVac system. The
60 Hz switchyard with harmonic filter bays is connected to the Brazilian 230 kVac system by way of a 13 km transmission
line. The two AC systems are interconnected by a HVDC back-to-back converter operating at 22 kVdc and 3300 Adc.
Both sides of the HVDC link utilize 15 kV busbars, supplied from an additional winding of the converter transformers, for
the connection of the harmonic filters and reactive power banks.
This arrangement has advantages in situations, such as in Rivera, where the AC systems are particularly weak. It also
permits the use of robust vacuum switchgear suitable for the frequent switching duties of the reactive power banks.
A key feature of the converter station is its ability to regulate the system voltage on the Uruguayan and Brazilian AC system
to a specific level. This is made possible by the coordinated use of the switchable banks of 15 kV reactive power equipment
and control of the reactive power absorbed by the HVDC converter.
Designed for remote operation from the U.T.E. national dispatch center in Montevideo, the control systems in the converter
station are fully automated, only requiring the operator to select the operating mode and set values for power and voltage.

CHAPTER
1.3.4.9. Sasaram Back-to-Back HVDC Link
Sasaram PGCIL India Back-to-Back 500 205 2002
ALSTOM T&D
Interconnection between the Eastern and Northern electricity networks of India

The project was the fifth in a series of initiatives set up to connect different regional electrical grids using HVDC back-to-
back connectors in India by the Power Grid Corporation of India Ltd (PGCIL).
The projects allow the hydroelectric-rich North and South regions to connect with the thermal-powered regions of the
East and West. Sasaram completed the loop between the regions, delivering a fully connected and integrated network.
The Sasaram HVDC project was a turnkey supply of a back-to-back link composed of a 500 MW pole operating at 205
kVdc, 2475 A, along with conventional switchyards on each side of the link.
The overall converter design was similar to that used for the Chandrapur and Vizag HVDC schemes. The thyristor valves
each contain fifty 100 mm 5.2 kV thyristors.
1.3.4.10. McNeill HVDC Link

McNeill Alberta Power Ltd Canada Back-to-Back 150 42 2014
Alstom Grid
The McNeill Back-to-Back system was first commissioned in 1989, and since that time the link has become a dependable
and essential part of the ATCO network. The system exchanges energy between the Alberta and Saskatchewan network
at the northern extremes of the WECC and Eastern Interconnected networks across the US and Canada.
The 2 AC networks are relatively weak at this location and the HVDC uses an extreme tap range and a large number of
low Mvar, AC harmonic filters, connected on a MV tertiary winding on the converter transformers to accommodate this.
After 20 years of service the availability of spare components for the control system was the motivation behind the plan
for replacement of the control system at McNeill. The scope of this contract was the turnkey replacement of the HVDC
Control & Protection System, and the MV Filter Circuit Breakers.
1.3.5. H400 THYRISTOR VALVE INSTALLATIONS

Until 2002, the thyristor valves installed by GE Vernova and its predecessors used relatively small thyristors
with silicon diameters no greater than 100 mm and rated voltages no greater than 5.2 kV. However,
semiconductor technology had moved on, and so in 2003 GE Vernova released its newest range of thyristor
valves, the H400 series.
Unlike any previous thyristor valve, this product range was specifically designed to accommodate thyristors
from more than one manufacturer and in a variety of sizes - a decision which was to prove highly successful.
At the time of publication, thyristor valves of the H400 series have been or are being manufactured for 11
HVDC projects, using thyristors of 7.2 kV or 8.5 kV voltage rating and silicon diameters of 125 mm or 150 mm.
1.3.5.1. Konti-Skan Pole 1 HVDC Link Refurbishment

Svenska Kraftnät
Sweden- Cable +
Konti-Skan 1 + 380 285 2006
Denmark Overhead Line
Energinet
ALSTOM T&D

CHAPTER
The contract was for the replacement of the mercury-arc valves of pole 1 with thyristor valves. This link is the
interconnection of two systems (Sweden-Denmark) by a 90 km undersea cable and 25 km transmission line.
Svenska Kraftnät is the owner and operator of Sweden’s transmission network, with responsibility for the national
electricity grid and the country’s 400 and 220 kV power lines. Energinet, its Danish counterpart, owns and operates the
400 kV transmission network for Jutland and Funen in western Denmark and is responsible for the overall security of the
supply, including the connections to neighboring countries.
Originally built in the 1960s, the mercury-arc HVDC system of pole 1 of the Konti-Skan HVDC undersea electricity
transmission link was nearing the end of its design lifetime and was scheduled for replacement and upgrading to match
the power rating of the conductor circuit (which had been enhanced over the years), and pole 2, built in the 1980s.
GE Vernova was selected to provide a cost-effective replacement, with the latest HVDC thyristor valve technology
replacing the old mercury-arc system. The submarine cable link, co-operated by Svenska Kraftnät and Energinet, spans
the Kattegat seaway between Sweden and Denmark and allows the two countries to exchange power.
When the Konti-Skan connection was originally built it put an end to independent, country-exclusive energy systems in
Scandinavia and reduced the two countries’ reliance on individual hydro or thermal sources.
As western Denmark is directly connected to the UCTE European network, Konti-Skan permits Sweden access to energy
from continental Europe, providing increased flexibility in dry and wet years and an increased security of supply.
The necessity to upgrade their grid management and trading system in a proactive, timely manner enables a more efficient
utilization of imported power - something that is of increasing importance to all energy managers.
This important re-investment program covered the complete renewal of the converter station in Vester Hassing, Denmark,
using state-of-the-art technologies. In Sweden, a new converter station with the same specification was built at Lindome
near Gothenburg, and required a 25 km extension to the overhead lines from Stenkullen to the new site. A key optimization
factor in the project specifications was the increase in the ratings of the new pole 1 equipment to utilize the unused
capacity in the cable/overhead line DC circuit, which had been separately modified and upgraded over the years.
The heart of the new installation is the latest version of GE Vernova's HVDC thyristor valve, the H400. These valves use
series-connected, fully protected thyristors, each with 8.5 kV rating and 125 mm diameter. The thyristors are controlled
by GE Vernova's industry leading Series V digital control and protection system, offering fully redundant operation,
including monitoring and alarm capabilities. Both the Swedish and Danish sites include two 203 MVA, 415/111.5 kV
HVDC converter transformers, one star-star connected and the other star-delta connected.

CHAPTER
1.3.5.2. Lévis De-Icer HVDC Project
Levis Hydro-Québec Canada De-icer 300 ±21 2007
AREVA T&D
Use of two HVDC converters connected in parallel to

provide a controlled DC current of up to 7,920 A for AC line
de-icing. Also functions as an SVC.
The Lévis De-icer is the world’s first HVDC-based
combination of a de-icing system and voltage controller,
and is located at a major connection point for some of
the 735 kV and 315 kV transmission lines of the Province
of Québec. The Static VAr Compensator (SVC) system is
flexible enough to be switched from SVC mode to de-icing
mode in less than one hour. When not de-icing, the SVC
functions as a reactive power compensator.
Further detail on this project may found in section 1.2.8.
1.3.5.3. GCCIA Back-to-Back HVDC Interconnection

Gulf
GCCIA Saudi Arabia Back-to-Back 3x600 222 2008
Interconnector
AREVA T&D
Interconnection of Saudi Arabia (60 Hz) into the 50 Hz Gulf AC Interconnector scheme
The Gulf Cooperation Council Interconnection Authority (GCCIA) was created to provide an effective means of exchanging
energy between the six member states in the region (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and the United Arab
Emirates):
➙ Interconnect the member states’ electrical power networks by providing the necessary investments for power sharing
to anticipate power generation loss in emergency situations
➙ Reduce the spinning reserves of each member state
➙ Improve the economic power system efficiency throughout the member states
➙ Provide cost-effective power sharing capabilities amongst the member states and strengthen collective electrical
supply reliability
➙ Deal with the existing companies and authorities in charge of the electricity sector in the member states and
elsewhere in order to coordinate their operations and strengthen the efficiency of operation with due regard to the
circumstances relating to each state
➙ Apply modern technological developments in the field of electricity.
Phase 1 of the project is focused on the northern portion of the interconnection.
Saudi Arabia runs its electricity transmission network at 380 kV, 60 Hz, whereas the other five countries use 400 kV, 50 Hz.
Based on the asynchronous nature of the states to be interconnected, the best solution which would allow Saudi Arabia
to participate in the exchange was to add a HVDC interconnection.
The Phase I system components linking the networks of Kuwait, Saudi Arabia, Bahrain and Qatar include:
➙ A double-circuit 400 kV, 50 Hz line from Al Zour (Kuwait) to Doha South (Qatar) via Ghunan (Saudi Arabia), with an
intermediate connection at Al Fadhili (Saudi Arabia) and associated substations
➙ A back-to-back HVDC interconnection to the Saudi Arabia 380 kV, 60 Hz system at Al Fadhili
➙ A double-circuit 400 kV interconnection comprising overhead lines and submarine link from Ghunan to Al-Jasra
(Bahrain) and associated substations.

CHAPTER
The control center located at Ghunan is linked with each member country’s national control center and will ensure security,
control, interconnection access, perform frequency and interchange regulation, coordinate interconnection operation and
facilitate transaction recording and billing.
The GE Vernova solution involved the creation of a 1,800 MW HVDC back-to-back link configured as three separate 600
MW substations. All three substations were built at the same location and constructed simultaneously. Each substation
can operate autonomously or in a coordinated manner. This 3-pole HVDC converter interconnection substation is
located next to Saudi Electric Company’s existing Al Fadhili 380 kVac substation.
One of the main functions of this HVDC facility is to constantly look for the occurrence of a power generation loss in the
interconnected networks. When a loss of generation is detected, the HVDC link injects power into the system and, through
the use of frequency control, restores the system to normal conditions.
The main equipment at the Al Fadhili converter station includes:
➙ 3 x 600 MW back-to-back HVDC links
➙ H400 thyristor valves, using 8.5 kV / 125 mm thyristors
➙ 12 converter transformers, three of each of the following ratings:
– 385.3 MVA, 380/97 kV, 60 Hz, Star/Star
– 385.3 MVA, 380/97 kV, 60 Hz, Star/Delta
– 380 MVA, 400/96 kV, 50 Hz, Star/Star
– 380 MVA, 400/96 kV, 50 Hz, Star/Delta.
The very high ambient temperature (up to 55°C) on this project posed a significant challenge. Because the temperature
of the valves’ active part (the silicon in the thyristors) needs to be limited to 90°C, the water-cooling plant required higher
coolant flow rates than a standard HVDC link. The cooling pipe arrangement within the valve was changed to a parallel
circuit to increase the total flow rate into the converter. This required the largest water-cooling plant ever built for a HVDC
installation.
HVDC converters need to be installed in a controlled environment with low levels of dust (converters have a tendency to
act as an electrostatic precipitator and to accumulate dust on insulating surfaces).
1.3.5.4. China-Russia Back-to-Back HVDC Interconnection

Sino-Russia XUJI/CEPRI China Back-to-Back 750 ±125 2008
AREVA T&D
Asynchronous interconnection of Northeast China with Russia.

The scope of supply on this project was for the valves and VBE of one converter only (other vendors provided the converter
on the other side of the link).

CHAPTER
1.3.5.5. Lingbao II Back-to-Back HVDC Interconnection
Lingbao II CEPRI China Back-to-Back 750 167 2009
AREVA T&D
Asynchronous interconnection of the Northwest China and North China power grids. The scope of supply on this project
was for the valves and VBE of one converter only, the other converter being provided by others. The converters on this
valve are the first use of 150 mm (6”) thyristors rated at 4500 Adc.
1.3.5.9. 3 Gorges-Shanghai HVDC Interconnection

3 Gorges-
CEPRI China Overhead Line 3000 ±500 2010
Shanghai
AREVA T&D
Sole supplier of thyristor valves and VBE cubicles at each station for bulk hydro power transfer on overhead lines from
Central China to Shanghai or the East Coast.
1.3.5.10. Ningdong-Shandong HVDC Interconnection

Ningdong-
CEPRI China Overhead Line 4000 ±660 2010
Shandong
AREVA T&D
Sole supplier of thyristor valves and VBE cubicles at each station for bulk hydro power transfer on overhead lines from
North Western China to the East Coast, close to Beijing.
1.3.5.6. IFA 2000 HVDC Cable Interconnection Refurbishment

National Grid England-
IFA 2000 Cable 2000 ±270 2011
+ RTE France
AREVA T&D
Refurbishment of the bulk power interconnection of English and French electricity systems by 45 km submarine cable
and 26 km underground cable. The original IFA2000 project was constructed in the mid-1980s interconnecting the UK
and France using air-cooled valves (see section 1.3.3.2).
The link is highly utilized and has become an essential part of the National Grid and RTE networks for energy trading
between the two countries. The equipment at each end of the link is different and the owners were experiencing increasing
failure rates and difficulty locating spares for obsolete components. They decided to replace the equipment in stages over
2 scheduled outages in 2010 and 2011, each outage with a scheduled duration of only 42 days. This requires that, as much
as possible, installation and testing work is carried out prior to the outage while the link remains in service.
The normal mechanical configuration of the H400 valve is ceiling-suspended, which requires the building structure for the
valve hall to be capable of supporting the load. In the case of this project however, the original valves were floor-standing,
so to avoid having to redesign and strengthen the building, a specially designed floor-standing version of the H400 valve
was created.
The equipment replaced under this contract includes:
➙ Thyristor valves
➙ Control equipment
➙ Cooling plant.
All other equipment including the HVDC cables, AC harmonic filters, converter transformers, switchgear, auxiliary power
supplies, etc remains unchanged.

CHAPTER
1.3.5.7. Melo Back-to-Back HVDC Interconnection
Melo UTE Uruguay Back-to-Back 500 ±79 2011
AREVA T&D
Asynchronous interconnection between Uruguay and Brazil, since Uruguay operates at 50 Hz and Brazil operates at
60 Hz. UTE already has the Rivera HVDC 70 MW link in operation (see section 1.3.4.8).
The turnkey scope for this project includes:
➙ 500 MW back-to-back Station at Melo (leaving provision for a possible expansion to 1000 MW in the future)
➙ 500 kV switchyard and substation on the Uruguay side
➙ Design and supply of 500 kV equipment for the extension of the substation in San Carlos on the Brazilian side, with
civil works and installation being carried out by the utility
➙ Design, supply and installation of the new and replacement control & protection system (SCADA) of the substation
in San Carlos.
The link is to be operated unmanned and under remote control similar to the previous Rivera HVDC link.
1.3.5.8. Jeju 2 Cable HVDC Interconnection

Jindo-Jeju 2 KEPCO South Korea Cable 400 ±250 2013
AREVA T&D
The second HVDC interconnection between the mainland of South Korea at JinDo and Jeju (formerly spelt "Cheju")
island.
The future development of the AC network on the island of Jeju is planned to include significant levels of renewable
generation, especially wind, and there is a need for more flexibility in the transfer of power between the mainland and
the island. This second HVDC link between island and mainland is at a higher rating than the first.
This new 400 MW bipole HVDC link was implemented using a third cable as metallic return conductor: there is no sea
electrodes in this installation. The total cable route is 120 km.
1.3.5.11. Rio Madeira HVDC Interconnection

Rio Madeira ANEEL Brazil Overhead Line 3150 ±600 2013
AREVA T&D
The transmission utility in Brazil (Furnas) already operates two 3150 MW / ±600 kV HVDC links. These take the hydro power
from the Itaipu generation plant in SW Brazil to the Rio de Janeiro and Sao Paulo load centers. The plan for developing
hydro generation and HVDC transmission is being replicated in the Rio Madeira project.
The project scope is for the turnkey supply of two converter stations rated at ±600 kVdc / 3150 MW (in 2 x 1575 MW
poles), for bulk hydro power transfer over 2375 km overhead lines. The link interconnects the Rio Madeira Hydro-plants
(at Santo Antonio and Jirau) in NW Brazil to the major load centers in South / Southeast Brazil, and as of 2018 was the
longest HVDC line in service in the world.
1.3.5.12. Champa - Kurukshetra 1 HVDC Link

Powergrid
Champa-
Corporation of India Overhead Line 2 x 1500 800 2017
Kurukshetra 1
India
GE Vernova

CHAPTER
The purpose of this thyristor-based HVDC project is for the connection of remote located generation in central India
to load centers in the north, through 1,365 km of 800 kV OHL. The overhead line includes a dedicated metallic return
conductor circuit.
The detailed design made allowance for the future addition of a second Bipole pair of converters in parallel using same
800 kV OHL.
1.3.5.13. Buk-Dangjin Godeok HVDC Links 1 and 2

Buk-Dangjin Submarine/
KEPCO S. Korea 3000 ±500 2018
Godeok Underground Cable
GE Vernova
Point to Point Interconnection between Buk-Dangjin and Godeok by Land and Submarine Cable, two cables rated
at 500kVdc, and the other at MV for the return current path. The purpose of the link is to allow the injection of large
amounts of energy into a rapidly growing commercial and industrial region south of Seoul. The scope of this project
was the supply of the first pole equipment, with provision for the future addition of a second pole to form a bipole.
Each Pole is rated at 500kVdc, 3000A, and the rating is capable of delivering 3000MW at the inverter end AC terminals.
The thyristor valves are arranged as water-cooled, ceiling-suspended quadrivalves, with 72 thyristors in each valve,
including 3 redundant levels. There are 6 valve modules in each valve, leading to a total of 72 x 12 x 2 = 1728 thyristors
at each converter station.
Reactive Power banks are provided at each converter station, which constitute the requirement for one pole, as follows:
Dangjin
➙ -4 x 238 Mvar
Godeok
➙ -9 x 136 Mvar
➙ -1 x 80 Mvar Shunt Reactor
The converter transformers are configured and rated as follows:
315/315 MVA, YN/y0 and YN/d11, 345/220 kV, 17% Impedance, +32% / -8% Tap range.

CHAPTER
1.3.5.14. Lower Churchill HVDC Link
Lower Churchill Overhead Line +
Nalcor Canada 2 x 450 ±350 2018
- Labrador Island Submarine Cable
GE Vernova
The function of this LCC Bipole HVDC link is for the transmission of hydro energy from the Lower Churchill dam in Labrador,
to the St John's area on Newfoundland. Converter stations at Muskrat Falls (Labrador) and Soldiers Pond (Newfoundland),
two transition compounds at the Strait of Belle Isle.
The converters are connected to the AC systems at Muskrat Falls at 315 kV and at 230 kV at Soldiers Pond. The Bipole link
is rated at 900 MW delivered at Soldiers Pond, and at 815 MW for power transfer in the reverse direction.
One key design consideration for this DC system is the relatively weak short circuit level at both ends of this link, going as
low as 1500MVA at the Muskrat Falls (rectifier) end, and around 2000 at the Soldiers Pond (inverter) end. Synchronous
compensators are being added at the inverter end, to support the converter and the AC network operation.
The low nominal current and high DC voltage rating of the converters in this system, has enabled the use of increased
current for overload (though only under specific DC line and cable configurations), and the converters are rated for 1.5pu
continuous overload, and 2pu for 10mins.
Reduced DC voltage operation is also possible down to 80% of the nominal 350 kVdc value.
The DC circuit consists of 2 HVDC conductors configured as a bipole circuit, and an electrode line from each converter
station to an electrode station. The electrode line from the Muskrat Falls converter is extremely long at 400km, while the
line at Soldiers Pond is 12km distant.
The DC smoothing reactors are located in the HV part of the circuit, rated at 150mH
The converter transformers are configured and rated as follows:
Muskrat Falls:
➙ 1-phase / 3-winding, YN/y0/d11, 393 / 196.5 / 196.5 MVA, 315 / 145 / 145 kV, +25 / -15 % Tap Range, 14% impedance.
➙ Soldiers Pond:
➙ 1-phase / 3-winding, YN/y0/d11, 348 / 174 / 174 MVA, 230 / 128 / 128 kV, +21 / -19 % Tap Range, 16% impedance.
The converters are thyristor-based, line-commutated converters, with 12-pulse valves. The valves are arranged as
water-cooled, floor-mounted quadrivalves, with 45 thyristors (44 at Soldiers Pond) in each valve, with 3 redundant
levels. There are 4 valve modules in each valve.
1.3.5.15. Champa - Kurukshetra 2

Powergrid
Champa-
Corporation of India Overhead Line 2 x 1500 ±800 2019
Kurukshetra 2
India
GE Vernova

CHAPTER
Phase II of Power Grid's plan to build a 6000 MW transmission corridor, connecting remote generation in Champa (Central
India) to load center in Kurukshetra (Northern India) via 1,365km of 800 kV OHL
The thyristor valves for this bipole system are air-cooled, housed indoors and in the form of ceiling-suspended, bi-valve
structures.
There are 2 parallel bipoles connected on the same HVDC Overhead Line circuit, and each bipole is rated for a significant
overload rating, to allow for contingencies such as an interruption on one bipole.
Under specific AC system conditions, the following approximate overload rating is incorporated:
➙ 110% continuous
➙ 120% for 30mins
➙ 140% for 5 secs
Reduced DC Voltage operation is also possible, operating at 640kVdc.
The valves are cooled by single circuit cooling loop which transfers heat from the valve assemblies using a water-glycol
solution, circulating the coolant to outdoor air-blast heat exchangers.
The converter transformers at both stations are configured as single phase, 2 winding units, each unit rated as follows:
296MVA, 400/334 kV, YN/y0 and YN/d11, 16% impedance, with +30%/-10% tap range.
A total of 6 units are provided per pole.
Reactive power banks are provided at each converter station as follows:
Champa
➙ 2 Filter Banks Type 1, each bank with the following sub-bank Mvar ratings: 125, 125, 157, 157, 157.
– Total Rating for each Type 1 Bank = 721Mvar
➙ 1 Filter Bank Type 2, with the following sub-bank Mvar ratings: 125, 165, 165, 165.
– Total Rating for Type 2 Bank = 620Mvar
➙ Total Installed Mvar rating at Champa = 2062Mvar
Kurukshetra
➙ 2 Filter Banks Type 1, each bank with the following sub-bank Mvar ratings: 125, 125, 165, 165, 165.
– Total Rating for each Type 1 Bank = 745Mvar
➙ 1 Filter Bank Type 2, with the following sub-bank Mvar ratings: 125, 165, 165, 165.
– Total Rating for Type 2 Bank = 620Mvar
➙ Total Installed Mvar rating at Kurukshetra = 2110Mvar
DC Reactors are placed in both the HV and LV DC circuits, rated at
HV = 140mH
LV = 70mH
1.3.5.16. East Power HVDC Link

East Power KEPCO S Korea Underground Cable 2 x 2000 500 2021
GE Vernova
Bulk power transfer from generation resources on the east coast of Korea into the Seoul metropolitan area by 220 km
of land cable.

CHAPTER
BIBLIOGRAPHY
[1] B. M. Weedy, B. J. Cory , “Electrical Power Systems”, John Wiley & Sons, 1999, Chapter 1.
[2] E.W. Kimbark ,“Direct Current Transmission, Vol. I”, Wiley-Interscience, 1971, Chapter 1.
[3] M. A. Rey, “Transport d’énergie Moutiers-Lyon par courant continu à 50 000 volts”, Communications
Techniques, Cinquième Année No. 56, December 1908.
[4] E. P. Hill , “Rotary Converters, Their Principles, Construction & Operation”, Chapman & Hall, Ltd, 1927,
p. 319.
[5] T. E. Calverley“The Flow of Power and Reactive Components in Rectifier and Inverter Equipments”, English
Electric Journal, Vol. XIII, Nos. 5 and 6, March & June 1954.
[6] T. E. Calverley, A. Gavrilovic, F. H. Last, C. W. Mott, “The Kingsnorth-Beddington-Willesden DC Link”, CIGRÉ
Paper 43-04, Paris, 1968.
[7] J. Arrillaga, “High Voltage Direct Current Transmission”, 2nd edition, ISBN 0 85296 941 4.
[8] C. D. Clarke, M. J. Johanson-Brown, “The Application of Self-tuned harmonic filters to HVDC Converters”,
IEE Conference Publication 22, HVDC Transmission, September 1966 at UMIST, p. 275.
[9] J.D. Ainsworth, “The phase-locked oscillator – a new control system for controlled static converters”, IEEE
Vol. Pas-87 No. 3, March 1968.
[10] D. S. Estey, J. W. Rolland, R. W. Haywood, D. B Willis, “Nelson River HVDC System Commissioning and
Initial Operating Experience”, CIGRÉ Conf. on Large HV Electrical Systems, August 1974.
[11] R. W. Haywood, K. J. Ralls , “Use of HVDC for Improving AC System Stability and Speed Control”,
Manitoba Power Conference on EHVDC Winnipeg, 1971, p. 780.
[12] R. Banks, B.A. Rowe, R.G. Noble, “Testing Thyristor Valves for HVDC Transmission”, CIGRÉ Session
paper 14-07, Paris (1978).
[13] M.L.Woodhouse, T.Simanwe, “A new facility for testing HVDC and SVC thyristor valves”, CIGRÉ Session
paper B4-309, Paris (2006).
[14] C C Davidson, J A Vodden, J J Snazell, “A new test circuit for operational testing of HVDC valves”, The
15th IET international conference on AC and DC Power Transmission, Coventry, UK, 2019.
[15] J. J. L. Weaver, A. M. Eccles, W. O. Kelham, “Development of a Thyristor Valve for HVDC Transmission”,
IEE Conference Publication No. 53, Power Thyristors and their Applications, London, May 1969, p.339.
[16] F. G. Goodrich, A. W. Tozer. “HV DC Converter Valves for the new Cross-Channel link”, Electronics &
Power, May 1982.
[17] C.M. Stairs, J.L. Fink, A.J. Molnar, “Eel River…vanguard of solid-state HVDC”, Transmission, February 1974.
[18] N.G.Hingorani, M.A. Dranchak, C. Flairty, A. Glassanos, R. Nakata, F.E. Fischer, N. Tai, “Compact Gas-
insulated HVDC Terminals – EPRI prototype DC link”, CIGRÉ Session paper 14-06, Paris (1978).
[19] D. M. Hodgson, “Qualification of XLPE tube systems for cooling high voltage high-power electrical
equipment”, IEE Power Engineering Journal, November 1991.
[20] V. Collet-Billon, J.P. Taisne, P. Charles, “The SACOI (Sardinia-Corsica-Italy) multiterminal link:
commissioning tests of the Corsican station Lucciana”, CIGRÉ Session paper 14-12, Paris (1988).
[21] V. Arcidiacono, S. Corsi, C. Pincella, E. Gasparini, A. Piffer, G. Toffolo, P. Ricci, S. Vascellari. G. Zafferani,
“System commissioning tests for the SACOI-2 HVDC three-terminal link”, CIGRÉ Session paper 14-107,
Paris (1994).
[22] R. P. Burgess, R. Kothari, “Design features of the Back-to-back HVDC Converter connecting the Western
and Eastern Canadian systems”, IEEE 1989 SM 793 PWRD.
[23] H. L. Thanawala, R. S. Whitehouse, G. O. Kwon, S. J. Lee, “Equipment and control features of Haenam-
Cheju HVDC Link in Korea”, CIGRÉ Session Paper 14-303, Paris, 1994.

CHAPTER
[24] C. C. Davidson, C. Horwill, M. Granger, A. Déry, “A Power-Electronics-based transmission line de-icing
system”, IEE 8th International Conference on AC/DC Transmission, London, 2006.
[25] B. T. Barrett, N. M. Macleod, A. A. Ebrahim, R. Azar, “GCC Interconnection HVDC Link, Novel features
and Exceptional Environmental Requirements”, GCC Power 2008 Conference, Manama, November 2008.
[26] C. C. Davidson, R. M. Preedy, J. Wen, “The Design and Type Testing of a New Range of HVDC Thyristor
Valves Using 6 inch Thyristors”, EPRI HVDC & FACTS Conference, Denver, November 2009.

CHAPTER
TOC 66 | DC Transmission Systems: Line Commutated Converters
2 HVDC CONVERTER
THEORY
It is important to understand why HVDC is applied and what benefits
can be obtained in electrical transmission networks by its use.
We describe the application of HVDC with overhead lines and cables
and explain the conditions in which HVDC systems are the optimum
solution. HVDC can minimize environmental impacts and short-
circuit capacity increases. HVDC controllability makes it a firewall
against faults when embedded in an AC transmission system.
This chapter illustrates a steady-state theoretical analysis of
a converter bridge, starting with an ideal 6-pulse bridge then
introducing the 12-pulse bridge. The discussion explains the
assumptions made in classical 6-pulse and 12-pulse bridge analysis
and reviews the non-ideal impacts of real circuit components. The
very important topic of reactive power - and how this is exchanged
between HVDC converters and AC networks - is fully explained, along
with a discussion on the reactive power sources in HVDC converter
stations. Different control methodologies for handling reactive power
exchange are highlighted and explained in detail.
By the end of this chapter, you will understand the operational
principles of a HVDC converter station.

2| HVDC CONVERTER THEORY
Chapter contents
HVDC CONVERTER THEORY........................
2. 66
2.1. DC TRANSMISSION APPLICATIONS................................... 69

2.1.1. A Historical Perspective.......................................................... 69
2.1.2. Benefits of DC Transmission over
AC Transmission......................................................................... 69
2.1.3. Bulk Power Transmission using Overhead Lines......... 69
2.1.4. Power Transmission via Cable............................................. 70
2.1.5. Transmission between Unsynchronized
AC Systems.................................................................................. 71
2.1.6. Parallel AC and DC Transmission........................................ 71
2.1.7. Minimizing Short-circuit Capacity Increases ................ 72
2.1.8. Minimizing Environmental Impacts................................... 72
2.2. STEADY-STATE ANALYSIS OF CONVERTER

BRIDGE OPERATION................................................................. 74
2.2.1. The 6-Pulse (or Graetz) Bridge............................................. 74
2.2.2. Overlap Modes............................................................................ 79
2.2.3. Inverter Operation.................................................................... 84
2.2.4. Real and Reactive Power........................................................ 85
2.2.5. 12-Pulse Converters................................................................. 86
2.2.6. Converters with Finite DC Reactance.............................. 92
2.3. REACTIVE POWER EXCHANGE BETWEEN HVDC

CONVERTER STATIONS AND AC SYSTEMS..................... 95
2.3.1. Reactive Power In AC Systems............................................ 95
2.3.2. Defining the Reactive Power Limits
at the Point of Connection.................................................... 96
2.3.3. Voltage Step Changes.............................................................. 97
2.3.4. The Reactive Power Load of a Converter........................ 99
2.3.5. The Impact of Control Methods on Converter
Reactive Power Absorption................................................... 101
2.3.6. Reactive Power Sources within
a Converter Station.................................................................. 103
2.3.7. Controlling Converter Reactive Power............................. 106
2.3.8. Using the Converter to
Control Reactive Power.......................................................... 108
BIBLIOGRAPHY ............................................................................................ 109
TOC DC TRAnsmission
68 | DC
Systems:
Transmission
Line Commutated
Systems: Line
Converters
Commutated Converters
2.1. DC TRANSMISSION APPLICATIONS
Although AC has long been the dominant method for electrical power transmission, there are certain niche
areas where DC transmission is preferable, as described below.
2.1.1. A HISTORICAL PERSPECTIVE

Although the earliest electrical power systems used DC for generation, transmission and loads, the
popularity of DC transmission was supplanted by AC transmission during the early part of the 20th century
as a result of the relatively simple means of changing between voltage levels using transformers [1]. DC
transmission was still perceived as having advantages over AC transmission in many applications, however.
The development of the high voltage mercury-arc valve in the early 1950s made it possible for certain
applications to economically convert AC power to DC power, transmit it as DC, then convert it back to AC.
With the development of the high voltage mercury-arc converter and, later, the high voltage thyristor
converter, the applications and usage of HVDC grew steadily in the second part of the 20th century. Today,
with the increasing demands on existing power systems of the world in terms of throughput and efficiency,
as well as meeting the ever increasing environmental concerns of the 21st century, the demand for HVDC,
UHVDC (Ultra High Voltage DC for voltages above 600 kVdc) and MVDC (Medium Voltage DC for voltages below
100 kVdc) is increasing. More than ever the economic and performance benefits of HVDC are being utilized
by power system planners: a trend which is only likely to grow with the increasing popularity of renewable
energy sources, coupled with increasing global electrical power demand.
2.1.2. BENEFITS OF DC TRANSMISSION OVER AC TRANSMISSION

Modern power electronic converters today convert from AC to DC and vice versa for electrical power
transmission purposes. These converters present clear technical and economic benefits over AC
transmission in a number of applications. Some, based on the present applications of HVDC, are listed
below [2]. However, with the ever increasing demands on power systems, the applications described below
cannot be considered exhaustive.
2.1.3. BULK POWER TRANSMISSION USING OVERHEAD LINES

HVDC interconnections require a conversion from AC to DC and vice versa at each connection point with
the AC system. This collection of equipment at each AC connection point is referred to as a ‘converter
station’. The converter station requires a higher capital outlay than would be required for an AC substation;
however, HVDC transmission towers are simpler and smaller than AC towers. For a monopolar HVDC link
(equivalent to a single AC circuit) only one high voltage conductor is required, as opposed to three in the
equivalent conventional 3-phase AC system. In some locations electrodes can even be inserted into the
earth to provide the low voltage return path.
In Fig. 2.1a, you can see that for a conventional 3-phase AC transmission system, each conductor must be
insulated for the peak of the AC voltage to earth. Also, to avoid corona discharge, its equivalent outside
diameter, that is the equivalent diameter created by a bundled conductor, must also be sized based on
the peak AC voltage. With DC transmission the DC voltage remains fixed, thus the insulation levels and
the equivalent outside conductor radius required for the transmission conductor are proportional to the
continuous voltage applied to the conductor.
The double-circuit 3-phase AC circuit, rated at a voltage of 400 kV, is able to transmit a power of 1850 MW
over some distance, utilizing twelve 282 mm2, conductors per circuit. As a result of the stray capacitance
and inductance of the transmission lines, the voltage will vary at the end of the transmission line furthest
away from the source of AC voltage control, typically the generator’s AVR (Automatic Voltage Regulator)
at the power station. This variation will be load dependent. Therefore, for a given range of AC load there
is a finite distance over which this double-circuit AC transmission line can transmit power without the
necessity for some form of voltage control to be added to the system, such as a Static VAr Compensator
(SVC): see section 2.3.

CHAPTER
The DC transmission solution in Fig. 2.1a shows a double bipolar arrangement, operating at ±250 kVdc. This
arrangement requires only four conductors per circuit, as the necessary equivalent outside diameter required
to control the corona of the high voltage line is much lower. In addition, the total conductor cross-sectional
area is reduced by over 30% and, as the voltage with respect to earth is lower, the support tower, or pylon,
can also be shorter, reducing the amount of steel required to construct it. In addition, the ground area (also
known as the Right-of-Way (ROW), wayleave or easement) occupied by a DC transmission interconnection
is less than that required for the equivalent AC power transmission.
The ability to directly and almost instantaneously control the real power transferred over a HVDC link can
be used to benefit the AC system, providing further direct and indirect cost benefits in DC’s favor. Therefore,
even though the capital cost of the HVDC terminal station can be greater than that of an equivalent AC
transmission substation, the per km distance of the transmission circuit will be lower for DC than for AC.
Hence, a break-even distance can be defined beyond which it is more economical to transmit bulk real
power as DC instead of AC. This break-even distance varies depending on the economic factors influencing
each case, but for overhead line transmission it is typically in the range of 600-800 km (section 8.1). See
also Fig. 2.1c.
AC 1850 MW per tower

400 kV 12 x 282 mm2
DC 1850 MW per tower

± 250 kV 4 x 644 mm2
Fig. 2.1a– Comparison of an AC and a DC overhead line tower for comparable power transmission
2.1.4. POWER TRANSMISSION VIA CABLE

The structure of a cable (a conductor surrounded by insulation and enclosed by an earth plane) means that
it has a much higher capacitance than an overhead line. When an alternating voltage is imposed on a cable,
there is a capacitive charging current in addition
to the load current flowing through the conductor.
This charging current changes direction every half Charges and discharges
AC
cycle, as the AC voltage polarity changes: charging, every half cycle
discharging and then re-charging the cable (Fig.
2.1b). Even for a cable of only moderate length, say
DC Only charges the cable once
40 km, (section 8.2), this reactive charging current
can utilize a major part of the total current carrying
capability of the cable, severely limiting the amount
of useful real power that can be transmitted. At
its limits, the capacitive charging current drawn
by the cable approaches its total current carrying
Fig. 2.1b– Electrical cable has a charging current
capacity, rendering the cable useless as a power requirement due to the inherent stray shunt
transmission medium. capacitance

CHAPTER
In the case of HVDC transmission, apart from an initial charging current, the cable draws no capacitive
charging current. Therefore, power can be transmitted via any length of cable, limited only by the cost of
the cable and the economics of the I2R losses incurred [3].
A ‘break-even distance’ can be defined for HVDC cable transmission in a similar way to overhead line
transmission, however the typical break-even distance is much shorter – typically 40-70 km.
2.1.5. TRANSMISSION BETWEEN UNSYNCHRONIZED AC SYSTEMS

When two adjacent AC systems operate at different frequencies such as 50 Hz and 60 Hz, the only practical
way to obtain the advantages of an interconnection between them is by means of a DC connection.
DC power is independent of the frequency and relative phase of the power systems. Therefore, power can be
transmitted between two independent AC systems without applying any operational restrictions to either
system. An asynchronous HVDC interconnection will not suffer from power swings and the risks of tripping
arising from overload, which may affect an AC connection between two AC systems in the event of a fault in
one of them. Therefore, even when AC networks operate at the same nominal frequency, an interconnection
by HVDC may be preferable to an AC interconnection. There are many examples of this in North America and
in India, often using back-to-back HVDC schemes in which both terminals of the scheme are located at the
same site.
Station Lines & stations

cost
DC Break even
distance
DC
AC
converter
stations
600 km – 800 km 40 km – 70 km
AC Overhead Submarine
stations line cable
Transmission
distance
Fig. 2.1c– HVDC versus AC transmission break-even distance
2.1.6. PARALLEL AC AND DC TRANSMISSION

When upgrading an AC system with additional transmission lines, it may be advantageous to make some
of them DC. The controllability of HVDC means that the power delivered can be modulated to provide
improved damping to the AC transmission, sometimes even allowing additional power to be transmitted
safely through the AC interconnection. One example
of this application of HVDC is the West Coast Pacific
Intertie in the USA. Here a DC line was installed, V1 XL V2
IL
enabling the maximum angular displacement of
voltage vectors between the ends of the parallel AC
line to be increased, thereby increasing the power Power
transmission capacity of the line (as the maximum Fig. 2.1d– Power through an AC transmission line
power transferable is proportional to the sine of the
displacement angle). See Fig. 2.1d.

CHAPTER
V1 ⋅ V2
Power = ⋅ sin(δ ) [Eqn. 2.1a]
XL
Where:
V1 = Sending end voltage
V2 = Receiving end voltage
XL = Reactance of transmission network
d = Angle between sending and receiving end voltages
2.1.7. MINIMIZING SHORT-CIRCUIT CAPACITY INCREASES

As power demand increases, additional AC infeeds are usually required. However, there may come a point
where any further infeeds will increase the short-circuit level in some part of the system beyond the rated
capability of the existing AC switchgear. By using HVDC to import the additional power, an upgrade of the
existing AC switchgear can be avoided. During a fault in the AC system, the infeed from the HVDC converter
is rapidly limited to a value no larger than the rated current, so the HVDC infeed does not contribute
significantly to the fault current.
AC SYSTEM
Fig. 2.1e– Adding generation to an AC system
2.1.8. MINIMIZING ENVIRONMENTAL IMPACTS

In many countries, rising demand for power can go hand-in-hand with increasing environmental objections
to the construction of new overhead lines. Modifying one or more existing AC lines to HVDC transmission
can provide a means of meeting rising power requirements while protecting the environment. For a given
level of environmental impact, the power carrying capacity of a DC line is significantly greater than that
of AC and, as already discussed, the addition of a parallel DC connection can often increase the maximum
safe power transfer through an existing AC line.

CHAPTER
By modifying the conductors, insulators and, for maximum benefit, the tower cross-arm structure, significantly
more electrical power can be transmitted through the same corridor or wayleave. See Fig. 2.1f [4].
Phase- Thermal DC DC DC PDC/PAC

3-phase
to-phase current voltage current power
3-phase AC power DC converted
voltage rating per pole rating (total)
AC line Line arrangement
(RMS) per phase per pole
(MW)
(kV) (A) (kV) (A) (MW) (Ratio)
Single-circuit (single- 220 630 240 Bipole (with single- 630 330
±260 =
conductor per phase) i.e. or or conductor per pole) or or 1.4
±2.0×127
Fig. 2.1f (ii) 127√3 1000 380 Fig. 2.1f (ii) 1000 520
Single-circuit (twin- 1260 480 Bipole (with three- 1900 1000

conductor bundle per 220 or or conductor bundle per ±260 or or 2.1
phase) Fig. 2.1f (iii) 2000 760 pole) Fig. 2.1f (iii) 3000 1560
Double-circuit (single- 630 480 Bipole (with three- 1900 1200

±320 =
conductor per phase) 220 or or conductor bundle per or or 2.5
±2.5×127
Fig. 2.1f (iv) 1000 760 pole) Fig. 2.1f (iv) 3000 1900
Double-circuit (twin 275

As above (six- ±370 =
conductor bundle per i.e. 2000 1900 6000 4440 2.3
conductor bundle) ±2.3×159
phase) 159√3
Double-circuit (twin- 400

±500 =
conductor bundle per i.e. 2400 3330 As above 7200 7200 2.2
±2.2×230
phase) 230√3
A B C + – A B C
+ –

(i) DC operation of an AC line (iii) Insulator rearrangement for conversion
without change to insulator arrangement of a multi-conductor AC line to HVDC
A B C + –
+ –

(ii) Insulator rearrangement for conversion (iv) Conversion to HVDC by
of a single-conductor AC line to HVDC use rearrangement of tower head and insulators
Fig. 2.1f– Potential increase in power transfer by conversion from AC (as thermal rating) to DC

CHAPTER
2.2. STEADY-STATE ANALYSIS OF CONVERTER BRIDGE OPERATION
To understand how HVDC works we must understand the operation of the converter bridge on which HVDC
is based, as described in this section.
2.2.1. THE 6-PULSE (OR GRAETZ) BRIDGE

The simplest type of converter bridge suitable for 3-phase applications such as HVDC is the 6-pulse bridge,
also known as the Graetz bridge. The analysis commences with an explanation of how the 6-pulse bridge
operates, initially with diodes and then with thyristors.
Diode
2.2.1.1. The Ideal Diode Bridge
Anode Cathode
A diode is a uni-directional device. It will allow current to flow in
one direction but not in the other. The ideal diode turns-on when
the anode voltage is more positive than the cathode and turns-
Fig. 2.2.1a– A diode
off when the cathode voltage is more positive than the anode
and current flow through the device has stopped. The symbol for
a diode is shown in Fig. 2.2.1a.
6-pulse converters are the building blocks of HVDC systems. An example of a 6-pulse converter, which
employs diodes is shown in Fig. 2.2.1b. Diodes conduct in the sequence 1, 2, 3, 4, 5, 6 - so the transitions
between one diode and the next occur alternately in the upper and lower half-bridges.
In order to simplify the analysis of the operation of a 6-pulse bridge, a number of assumptions are made.
These assumptions are:
➙ The DC side inductance is infinite, hence, the DC current is ripple free
➙ The applied 3-phase AC voltage waveforms are balanced in phase and magnitude
➙ Energy Losses in equipment (e.g. valves, converter transformers, etc) can be neglected
➙ The effect of the RC (Resistor Capacitor) circuits which would be connected across each diode in a real
circuit, along with any other stray capacitances within the circuit, can be neglected
Ld → ∞ Id
ea D1 D3 D5
Ia
eb
Ib
Vd
ec
Ic
D4 D6 D2
Id
Fig. 2.2.1b– 6-pulse converter
Each diode conducts for 120° in every 360° cycle, so that the successive conducting pairs of valves are 1 and
2, 2 and 3, 3 and 4, 4 and 5, 5 and 6 and 6 and 1.
The conducting pair is always that pair which exhibits the largest voltage difference compared to any of
the other five pairs of diodes, because the circuit imposes reverse voltage on all of the other diodes. With
time, the relative amplitudes of the converter’s three AC supply phases (valve-winding voltages) change,
so in Fig. 2.2.1d, the voltage B-C becomes greater than the voltage A-C and valve 3 takes over the current
which had been flowing in valve 1. This process is known as commutation. The phasor representation of
the AC voltage sequence and hence the conduction sequence is shown in Fig. 2.2.1c.
CHAPTER
1 3
Firing ec
sequence
1
ea
ea
Vdi0
6 -e
b
-ec 2
Direction of rotating eb
phase vector 2
ec eb
5 3
i1
-ea
4 i2
Fig. 2.2.1c– The phasor diagram of the conduction

sequence of a 6-pulse bridge i3
i4
1 3 i5
ec
i6
Id
ea
Vdi0
ia
eb
2 ib
ic
t
Fig. 2.2.1e– The average DC voltage during the
conduction period of one diode Fig. 2.2.1d– Switching pattern of a 6-pulse converter
Consider the interval when diode D1 is conducting:

As highlighted in Fig. 2.2.1e the applied voltage as seen from the DC side of the converter is initially the
difference between the instantaneous phase voltages ea and eb, then half way through the conduction
period of diode D1 the applied voltage becomes the difference between the instantaneous phase voltages
ea and ec.
Assuming a balanced, symmetrically applied AC voltage, then:
Ea = Eb = Ec = Em [Eqn 2.2.1a]
(where Em is the rms value of the applied phase voltage)

CHAPTER
From Fig. 2.2.1e it can be seen that the peak of the applied voltage to diode D1 (Vpeak) is the peak of the
line-to-line voltage, that is:
Vpeak = 2 ⋅ 3 ⋅ Em [Eqn 2.2.1b]
This voltage appears twice during the conduction period of D1 as a consequence of, first D1 and D6 being in
conduction, then D1 and D2 being in conduction.
It is therefore possible to further simplify the
waveform shown in Fig. 2.2.1e by realizing that the 1 3
ec
conduction period of D1 is symmetrical as indicated
in Fig 2.2.1f. In Fig. 2.2.1f it can be seen that the
area under the curve indicated as ‘A’ is the same A B
as that of ‘B’. ea
Vdi0
In this idealized case, the mean direct voltage, Vd,
can be calculated as a fixed value, determined
entirely by the transformer ratio. From Fig. 2.2.1f, eb
considering area ‘A’ this can be described as the 2
difference between the applied AC voltages ea and
eb, from 60° to 120° measured from the point ea = ec.
Fig. 2.2.1f– Symmetry of the conduction period
The average ideal DC voltage (Vdi0) over this period of a diode in an ideal 6-pulse bridge
of 60° can then be written as:
1 2⋅π
Vdi 0 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
3
π ∫π 3
⋅ [Eqn 2.2.1c]
3
Giving:
3
Vdi 0 = ⋅ 2 ⋅ 3 ⋅ Em [Eqn 2.2.1d]
π
As 3 ⋅ Em is equal to the line-to-line applied AC voltage (ELL) Eqn 2.2.1d can be rewritten to give the more
common form of the equation:
3⋅ 2
Vdi 0 = ⋅ ELL [Eqn 2.2.1e]
π
The RMS valve winding current (IRMS), as a function of the DC current (Id), can be shown (Appendix A2.2.1)
to be:
2
I RMS = ⋅ Id [Eqn 2.2.1f]
3
And the fundamental component of current (I1), Fig. 2.2.2c, can also be shown (Appendix A2.2.1) to be:
6
I1 = ⋅ Id [Eqn 2.2.1g]
π
2.2.1.2. The Thyristor

Like the diode described in section 2.2.1.1 the Thyristor is a uni-directional device, only permitting current
flow in one direction. However, the thyristor includes an additional terminal, known as a ‘gate’ terminal. In
order for a thyristor to enter conduction this gate terminal must also be energized. Therefore, by controlling
when the gate terminal is energized, the behavior of the circuit in which the thyristor is connected can be
influenced. It is important to note however, that the gate terminal only controls when the thyristor turns-on.
The thyristor can only be turned-off by the circuit connected between the anode and cathode of the device.
This can be explained by considering Fig. 2.2.1g.
CHAPTER
Firing delay angle VAC
L

t
1 3
iL
ec VTHY
VAC VTHY ig t
ea
Vd
ig
eb t
2 iL
t
i1
Fig. 2.2.1g– The gating and commutation of a thyristor
i2
Fig. 2.2.1g shows a simple thyristor circuit.
When a gate pulse (Ig) is applied while positive
i3
forward voltage is imposed between the anode
and cathode (VTHY), the thyristor will conduct
i4 current (I L). Conduction continues without
further gate pulses as long as current flows in
i5
the forward direction. Thyristor turn-off takes
place only when the current tries to reverse.
Hence, a thyristor converter requires an existing
i6 alternating AC voltage (Vac) in order to operate
as an inverter. This is why the thyristor-based
Id
converter topology used in HVDC is known as a
Line-Commutated Converter (LCC).
ia A more rigorous description of a thyristor is

provided in section 6.1.
ib
2.2.1.3. Introduction of Delay Angle
Control to a 6-Pulse Bridge
ic
Let us look again at the operation of a diode
bridge as shown in Fig. 2.2.1b. This time we will
t consider the use of thyristors instead of diodes
and we will use a time delay on the energization
Fig. 2.2.1h– Ideal thyristor controlled 6-pulse bridge of the gate terminal. As the 6-pulse bridge
operation operates cyclically, it is common to refer to this
time delay in terms of electrical degrees with 0°
being the point on wave where the ideal diode bridge would conduct. This delay angle is referred to as the
‘firing’ or ‘trigger’ angle and is given the symbol a.
Fig. 2.2.1h shows the impact of delaying the conduction of the thyristor by some arbitrary value between
0° and 90°.
From Fig. 2.2.1i it can be seen that by introducing a delay in firing the thyristors the area under the curve
(highlighted in green), representing the average DC voltage, has reduced – compared to what it would
have been without the delay. This waveform can be analyzed using the method previously described in
section 2.2.1.1 for the diode bridge. However, in this case it is performed more easily if the waveform is

CHAPTER

Firing delay angle

ec
1 3
ec
ea
Step 1 Vd
ea
Vd
eb
eb Move this part of the

2
waveform to give two
symmetrical areas
Fig. 2.2.1i– The average DC voltage during the

conduction period of one thyristor with the firing
angle (a) delay between 0° and 90°
ec
Giving A
ea
Vd
B
eb
A2 A1
ec
ea
Step 2 Vd
eb
Now separate out area ’A’ into

two parts ‘A1’ and ‘A2’
Fig. 2.2.1j– The manipulation of the 6-pulse thyristor bridge waveform in
order to perform a symmetrical analysis
manipulated in order to provide a symmetrical waveform to analyze, as shown in Fig. 2.2.1j.

The numeric value of the DC voltage can be calculated by analyzing areas ‘A1’ and ‘A2’ shown in Fig. 2.2.1j.
It can be seen that area ‘A1’ is a constant value for firing angles equal to, or less than, 60°, which can
therefore be described by the equation:
1
2⋅π
A1 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫3
⋅ π3 [Eqn 2.2.1h]

CHAPTER
Area ‘A2’ is a function of the firing angle and for a firing angle of less than, or equal to, 60°, which can be
described by the equation:
1 3
π
A2 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫α
⋅ [Eqn 2.2.1i]
Summing Eqn 2.2.1h and 2.2.1i and solving gives:
3
Vd = ⋅ 2 ⋅ 3 ⋅ Em ⋅ cos(α ) [Eqn 2.2.1j]
π
Again, as 3 ⋅ Em is equal to the line-to-line applied AC voltage Eqn 2.2.1j combined with Eqn 2.2.1e gives
the more common form of the equation:
Vd = Vdi0 cos(a) [Eqn 2.2.k]
A further assumption has been made here in order to simplify the steady-state theory of the ideal 6-pulse
bridge - and that is that the firing angle a, for each valve, is equal.
From Eqn 2.2.k, it can be seen that with a firing
angle equal to 90° the DC voltage will be zero. This
can be confirmed by examination of Fig. 2.2.1k. In = 90°
Fig. 2.2.1k it can be seen that with a firing angle of
90°, area A and area B above the zero axis equal
1 3
the equivalent areas below the zero axis, hence, the ec
average DC voltage is zero.
An important point to note from Fig. 2.2.1k is A B
that there is always a pair of thyristors within Vd
ea
the 6-pulse bridge turned on, hence, there is a
continuous current path through the converter. This
is important when considering the operation of a
HVDC scheme where two (or more) converters are
eb
connected in series: there is always a continuous 2
DC current path.
2.2.2. OVERLAP MODES Fig. 2.2.1k– An ideal 6-pulse bridge operating

with a firing angle of 90°
The review of the steady-state operation of the
ideal 6-pulse converter presented in section 2.2.1.3 assumed that there was zero inductance between
the applied 3-phase AC voltage and the converter bridge. In reality of course, there will always be some
inductance in the connection in order to limit the valve current during a fault. This phase inductance may
be in the form of dedicated coupling reactors (typically air-cored, air-insulated reactors) [5], but it is more
commonly a result of the leakage reactance of the converter transformer. In addition, other inductances
will contribute to the phase reactance, such as a series filter connected to the converter (see section 4.7)
or the series impedance within a valve (see section 6.2), but the inductance is predominantly that of the
coupling reactors or leakage reactance.
The total inductance in each phase connection is known as the commutating inductance and is commonly
given the symbol ‘Lc’. When referring to the AC system frequency, this is known as the commutating
impedance and given the symbol ‘Xc’. The commutating impedance is defined as the total series inductance,
per phase, between the ideal converter valves (see section 6.2) and the point in the AC system where the
AC voltage waveform is free from distortion, as a result of the operation of the converter. For convenience
this is normally identified as the point of common connection between the converter and the AC harmonic
filters, referred to as the commutation bus.
The ideal 6-pulse bridge shown in Fig. 2.2.1b can thus be modified to include this commutating impedance,
as shown in Fig. 2.2.1l.

CHAPTER
Ld → ∞ Id
ea T1 T3 T5
Lc ia
eb
Lc ib
Vd
ec
Lc ic
T4 T6 T2
Id
Fig. 2.2.1l– A 6-pulse bridge including phase reactance
Note that in the following analysis, a further assumption is made: that the commutating impedances in
each phase connected to the 6-pulse bridge are all equal. The commutating impedance will reduce the
rate of change of current through the valves, which is theoretically infinite in the simplified case, when
the current path through the 6-pulse bridge changes from one valve to another. This can be analyzed by
considering the circuit during the interval when the voltage of phase b, in Fig. 2.2.1m, is more positive than
that of phase a and a gate pulse is applied to T3, which now has a positive voltage between its terminals.
The equivalent circuit for this is shown in Fig. 2.2.2a.
Ld → ∞ Id
ea T1 T3 T5
Lc ia
eb
Lc ib
ec Vd
Lc ic
T4 T6 T2
Id
Fig. 2.2.1m– A 6-pulse bridge during a commutation event
2.2.2.1. Single Overlap

The DC current during the overlap can be calculated by considering the equivalent circuit shown in Fig.
2.2.2a, representing the current flow through valve T3 during a commutation from T1 to T3.
Given that:
di
v= L ⋅ [Eqn 2.2.2a]
dt

CHAPTER
The circuit shown in Fig. 2.2.2a results in a rate of
change of current of:
Lc Lc
di 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t )
= [Eqn 2.2.2b]
dt 2 ⋅ Lc
eb
However, more important than the di/dt is the
3 ⋅ Em i
overlap time, the time required for the changeover
from 1.0 p.u. current flowing in T1, to 1.0 p.u., the di
ea
current flowing in T3. dt
Considering that at the instant valve T3 is fired at
angle a, the T3 valve current is zero and the rate-
of-change of valve current is also zero, then from Fig. 2.2.2a– Equivalent circuit of the commutation process
Eqn 2.2.2b:
w t0 = a [Eqn 2.2.2c]
Hence:
α
t0 = [Eqn 2.2.2d]
ω
At some time later, ‘t1’, the current through valve T3 will equal 1.0 p.u. Therefore, integrating Eqn 2.2.2b
between t0 and t1, with respect to the final value t1, gives:
t1 3 ⋅ 2 ⋅ Em ⋅ sin(ω ⋅ t1 )
1.0 = ∫ dt1 [Eqn 2.2.2e]
2 ⋅ Lc
α
ω
2 ⋅ ELL ⋅ (cos(α ) − cos(ω ⋅ t1 ))

1.0 = [Eqn 2.2.2f]
2 ⋅ ω ⋅ Lc
The time taken for the current to change between zero and 1.0 p.u. in a valve is called the overlap time
and, as with the firing angle, it is normally referred to in electrical degrees. The symbol used for the overlap
angle is ‘m’. In Eqn 2.2.2f above the value of w t1 must equal the initial angle a, plus this overlap angle, m.
Hence, this equation can be re-written as:
2 ⋅ ELL ⋅ (cos(α ) − cos(α + µ ))

1.0 = [Eqn 2.2.2g]
2 ⋅ ω ⋅ Lc
As 1.0 p.u. current is the DC current, Eqn 2.2.2g can be re-written as:
2 ⋅ ELL ⋅ (cos(α ) − cos(α + µ ))

Id = [Eqn 2.2.2h] ea
2 ⋅ ω ⋅ Lc ea and eb shorted together
during commutation period
Re-arranging Eqn 2.2.2h to solve for m, gives:
 2 ⋅ ω ⋅ Lc 
µ = cos −1  cos(α ) − ⋅ Id  − α [Eqn 2.2.2i]
 2 ⋅ ELL 
Consider again the circuit shown in Fig. 2.2.1m. During the

commutation event, two of the AC phases are connected in
ec eb
parallel, therefore the equivalent AC voltage, as seen from
the DC terminals of the converter, can be calculated from Fig. 2.2.2b– Phasor diagram of a star connected
6-pulse bridge during a commutation
the equivalent phasor diagram shown in Fig. 2.2.2b.

CHAPTER
The calculation of the 6-pulse bridge average DC voltage can be calculated in a similar manner to that
described above in Fig. 2.2.1j and Eqn 2.2.1h, and 2.2.1i.
Eqn 2.2.1h remains the same:
1
2⋅π
A1 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫3
⋅ π3 [Eqn 2.2.2j]
However, area ‘A2’ is now a function of the firing angle and the overlap angle, and for a firing angle plus
overlap (a + m) of less than or equal to 60°, this can be described by the equation:
1 α +µ
A2 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫α
⋅ [Eqn 2.2.2k]
A third area ‘A3’ must now be added, representing the period of the overlap:
1 3 3
π
A3= 2⋅ ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫α + µ
⋅ [Eqn 2.2.2l]
2
Summing Eqn 2.2.2j, 2.2.2k and 2.2.2l and solving gives:
3
Vd = ⋅ 2 ⋅ 3 ⋅ Em ⋅ (cos(α ) + cos(α + µ )) [Eqn 2.2.2m]
2⋅π
Combining this with Eqn 2.2.1d gives:
1
Vd = ⋅ Vdi 0 ⋅ (cos(α ) + cos(α + µ )) [Eqn 2.2.2n]
2
Combining Eqn 2.2.2n and 2.2.2i produces:
3
Vd = Vdi 0 ⋅ cos(α ) − ⋅ ω ⋅ Lc ⋅ Id [Eqn 2.2.2o]
π
Eqn 2.2.2o is commonly used as a basis for converter bridge analysis and can be used in the form shown or
it can be converted to a p.u. base (Appendix A2.2.2), resulting in equations 2.2.2p and 2.2.2q. The derivation
of the p.u. current is given in Appendix A2.2.1.
 Xc 
Vd p.u . = k⋅V LL p.u.⋅  cos(α ) − p.u . ⋅ Id p .u .  [Eqn 2.2.2p]
 2 
Where:
Vdp.u. 1
k= ⋅
VLL p.u. Xc p .u . [Eqn 2.2.2q]
cos(α 0 ) −
2
Fig 2.2.2c shows the impact of the overlap angle on the valve and phase current. In addition, the voltage
developed across a valve, which includes ‘notches’ resulting from the commutation of other valves within
the bridge, is shown.
2.2.2.2. Double Overlap

π
In normal converter operation the overlap angle (μ) is less than 60° ( 3 ) and most converter analysis
can be done using the equations presented above in section 2.2.2.1 for the single overlap case. However,
under certain unusual operating conditions, such as when the DC current is high but the converter valve

CHAPTER
Overlap angle
Firing delay angle

1 3
ec
ea
Vd
eb
0.5 (eab-ebc)
eac eba ecb 0.5 (eab-eca)
0.5 (ebc-eca)
Valve 1 voltage
eab eac
Valve 1 current
Fundamental
component of current
‘A’ Phase current
t
Fig. 2.2.2c– Valve voltage and current considering both firing angle and overlap for rectifier operation
winding voltage is low, or when there is a short-circuit on the DC side, the converter can enter what is
known as double overlap mode.
In the single overlap mode, a commutation in the upper half of a 6-pulse bridge is always complete before
a commutation in the lower half starts. In the double overlap mode, the next commutation commences
before the previous one is completed. Hence, there are either three valves or four valves in conduction
during any time interval. Moreover, during the time interval when four valves are in conduction, the
converter appears as a 3-phase short-circuit to the AC system and the DC voltage will be zero. During the
interval when only three valves are conducting then, as with the single overlap mode, the DC voltage will
follow a sine wave with a magnitude of 1.5 × Em.

CHAPTER
A detailed analysis of the derivation of the equations of a 6-pulse converter is not presented here, but can
be found in other references [6]. The important derivation to note is that for the double overlap mode the
equation for DC voltage becomes:
9 9
Vd = ⋅ 2 ⋅ Em ⋅ cos(α − 30°) − ⋅ ω ⋅ Lc ⋅ Id [Eqn 2.2.2r]
π π
Comparing this to Eqn 2.2.2o, it can be seen that in double overlap mode the slope of the Vd – I d
characteristic (the voltage drop with increasing DC current) increases by a factor of three from the single
overlap mode.
From this, we can also derive that the maximum DC short-circuit current (Ids-c) that can flow equals:
2 ⋅ Em
IdS −C = [Eqn 2.2.2s]
ω ⋅ Lc
2.2.3. INVERTER OPERATION

By increasing the firing angle a beyond 90°, the voltage area of the phase-to-phase voltage connected to the
DC terminals via the conducting thyristors will be predominantly negative, hence the DC terminal voltage
will be negative. It is important to note however that the DC current flow through the 6-pulse bridge cannot
change, and therefore for current flow to exist through the 6-pulse bridge when operating as an inverter,
there must be a separate current source in the circuit. Whilst this may at first seem counter intuitive to the
idea of power flow, it must be remembered that a basic HVDC interconnection consists of a rectifier and an
inverter. The rectifier produces a DC voltage to drive the DC current around the DC circuit and the inverter
produces an opposing voltage which reduces the DC voltage developed across the DC circuit and hence
the DC current which circulates: refer to section 7.1.
6
2.2.3.1. Extinction Angle Definition α 2
C
Beyond 90° the firing angle of the converter
becomes large, so it is more common to refer to the
extinction angle or gamma g. This extinction angle
represents the time between the end of the overlap A - Vd
period and the time when the voltage across the
valve, which has just extinguished, becomes
positive again. This is mathematically expressed as:
g = 180° – m – a [Eqn 2.2.3a] B 3
γ
1
The position of g relative to a on the converter
waveform is shown in Fig. 2.2.3a. Fig. 2.2.3a– Effect of a firing angle of 140°
2.2.3.2. Single Overlap Operation

It is important to note in Fig. 2.2.3a that the wave shape is identical to that of Fig. 2.2.2c, except that here it
is inverted. This symmetry is important, as it means that the same analysis equations that were developed
in section 2.2.2 for rectifier operation can also be used for inverter operation, simply by replacing the firing
angle, a, with the extinction angle, g and noting the change in voltage polarity. Hence, for inverter operation
the converter equations become:
2 ⋅ E LL ⋅ (cos(γ ) − cos(γ + µ ))
Id = [Eqn 2.2.3b]
2 ⋅ ω ⋅ Lc
 2 ⋅ ω ⋅ Lc 
µ = cos −1  cos(γ ) − ⋅ Id  − γ
 2 ⋅ E LL 
[Eqn 2.2.3c]

CHAPTER
3 [Eqn 2.2.3d]
Vd = Vdi 0 ⋅ cos(γ ) − ⋅ ω ⋅ Lc ⋅ Id
π
 Xc 
Vd p .u . = k⋅V LL p.u.⋅  cos(γ ) − p.u . ⋅ Id p .u .  [Eqn 2.2.3e]
 2 
Where:
Vdp.u. 1
k= ⋅
VLL p.u. Xc p.u . [Eqn 2.2.3f]
cos(γ 0 ) −
2
It must be noted that the control of the output voltage of a 6-pulse bridge is only achieved by the firing
angle, a. The extinction angle, g, is a measure of the available turn-off time for the valve following the point
in time when the valve is fired. The extinction angle can therefore not be ‘controlled’, as it relates to events
which are still in the future at the time when the valve is fired: consequently, g can only be predicted.
When it is part of a HVDC interconnection, a 6-pulse bridge cannot operate as an inverter in the double
overlap mode because, with four valves in conduction, two of the valves turned on will be in the same
phase, resulting in a commutation failure.
2.2.3.3. Commutation Failure

A minimum extinction time is required in order for the valve in conduction to fully turn-off before its
terminal voltage becomes positive again. If the valve does not turn off then this valve will reconduct,
stopping commutation of the current to the next valve from occurring 60° later. When the other valve
Ld
6
α 2
2 6 4 LS A C
iA LS B
Vd Vʻ
d
iB LS E 1 C
iC A
5 3 1
ld
B
γ 3
1
5
Fig. 2.2.3b– Commutation failure
connected to the same phase in the 6-pulse bridge conducts, a short-circuit, known as a commutation
failure occurs. There is then a temporary cessation of power transfer until a control action has removed
the DC short-circuit. See Fig. 2.2.3b.
2.2.4. REAL AND REACTIVE POWER

In a purely resistive circuit the voltage and current are in phase and hence all of the apparent power, that
is the power which is apparently available as a resultant of the applied voltage and current, will provide
useful energy, known as real power. However, in any circuit where the voltage and current are not in phase,

CHAPTER
for example in an inductive or capacitive circuit, the apparent power will consist of both real power, and a
second component known as reactive power. This reactive power element does not transfer any energy
to the load. The relationship between apparent
power, real power and reactive power is commonly
represented by the ‘Power Triangle’, as shown in Apparent
Fig. 2.2.4a. power (MVA)
Consider the conventional definition of reactive Reactive
power (MVAr)
power, with reference to Fig. 2.2.4a:
Q = P tanf [Eqn 2.2.4a]
φ
P
cos φ = [Eqn 2.2.4b] Real power (MW)
S
Fig. 2.2.4a– The power triangle
Where:
Q = Reactive power (MVAr)
P = Real power (MW)
f =A C current phase angle
at fundamental frequency (°)
S = Apparent power (MVA)
2.2.4.1. Derivation of Real and Reactive Power

The real power supplied or consumed by a converter, along with the reactive power absorbed, is a function
of the phase angle between the converter transformer line side AC voltage and the fundamental component
of the converter phase current: see Fig. 2.3e.
The fundamental frequency component of the converter line current (Fig. 2.2.2c) can be calculated by
performing a Fourier analysis (Appendix A2.2.1). However, the results of this analysis are complex and
therefore it is typically more convenient to assume the maximum value of fundamental frequency current
given in Eqn 2.2.1g. To calculate the reactive power absorbed by the converter is also complex (Appendix
A2.2.1), hence it is common to use the ‘Uhlmann approximation’.
2.2.4.2. The Uhlmann Approximation for Reactive Power

A commonly used method for the calculation of the reactive power absorbed by a HVDC converter is known
as the Uhlmann approximation. This approximation assumes that the converter line winding currents are
trapezoidal, that is, that the DC side series reactance is infinite. This approximation provides a reasonable
estimate of the reactive power absorbed at high DC current, close to 1.0 p.u., but can become significantly
inaccurate if the AC voltage is low or if the overlap angle is large.
The Uhlmann approximation assumes that the real power (P) and the apparent power (S) can be defined,
neglecting converter and transformer losses. This gives:
Pd Vd ⋅ Id Vd
cos φ = = = [Eqn 2.2.4c]
3 ⋅ E LL ⋅ Iac fund 3 ⋅ E LL ⋅ π ⋅ Id Vdio
6
and rearranging (Eqn 2.2.2n) to give (Eqn 2.2.4c) results in:

1
cos φ = ⋅ [ cosα + cos(α + µ )] [Eqn 2.2.4d]
2
The reactive power loading of the converter can then be found from Eqn 2.2.4a.
2.2.5. 12-PULSE CONVERTERS

Although converter theory can be understood well by considering the 6-pulse bridge, most HVDC converters
use the more complex 12-pulse bridge which consists of two separate, phase-displaced 6-pulse bridges in
order to improve the converter harmonic performance.

CHAPTER
2.2.5.1. 12-Pulse Converter Configuration
The most common configuration of a converter bridge found in modern HVDC schemes is the 12-pulse bridge
arrangement, as shown in Fig. 2.2.5a. Each 6-pulse bridge is rated for half the DC voltage and the two are
connected in series in order to produce the full DC voltage. While such an arrangement could be constructed
with the supplied AC voltage being in-phase for each bridge (that is, ea, eb and ec for bridge 1 equal to ea, eb
and ec for bridge 2), there is no practical advantage of this arrangement unless a phase displacement of 30°
is introduced between the two bridges. In a 6-pulse bridge, each valve is fired every 60° and hence the DC
voltage of the 6-pulse bridge is the average voltage of the applied sinusoidal voltage over 60°.
Both the AC and DC harmonics produced by the converter are a function of the instantaneous variation
between the applied voltage and the average DC voltage. By introducing a second series-connected bridge
operating with an applied AC voltage 30° displaced from the first bridge, there will be a valve turning on
every 30° instead of every 60°, therefore reducing the instantaneous variation between the applied voltage
and the average DC voltage and thus the AC and DC harmonics produced by the converter. This is covered
in more detail in section 4.1.
Fortunately, a transformer with one winding connected in star and a second winding connected in delta
inherently introduces a 30° phase-shift between its valve side windings. Hence, the 12-pulse bridge
Ld → ∞ ID
ea D1 D5 D9
Ia
eb Ib Vd
BRIDGE 1 2
ec
lc
D7 D11 D3
Vd
ea - 30 ° D12 D4 D8
Ia
e b - 30 °
Ib Vd
BRIDGE 2 2
ec- 30 °
Ic
D6 D10 D2
Id
Fig. 2.2.5a– A 12-pulse converter bridge
arrangement shown in Fig. 2.2.5a can be realized by using a transformer with one line winding (AC system
side) connected in star and two valve side windings, one connected in star (that is, in-phase with the supply
voltage) and one in delta (that is, displaced from the supply voltage by 30°).
The normal notation used to denote the phase relationship between the valve side star and delta windings
is to relate the phase displacement to that of a 12-hour clock face (noting that a round clock face is 360°
and this is divided by 12 points, denoting the hours, hence each hour represents a 30° displacement). In
CHAPTER
Firing
30° phase shift sequence
= 11 oʼclock
12
11 ea - 30° 1 Direction of rotating
phase vector
-eb ea
10 -eb - 30° -ec - 30°

2
9 ec -ec 3
ec - 30° eb - 30°
8 4
-ea eb
7 -ea - 30°
5
6
Fig. 2.2.5b– The phasor diagram of the conduction sequence of a 12-pulse bridge (Note that this diagram has been
drawn with ‘12’ at the top to reflect the orientation of a clock face. Normally, firing instance ‘1’ is drawn at the top)
the arrangement shown in Fig. 2.2.5a, the applied voltage to Bridge 2 lags behind that applied to Bridge
1 by 30°, that is, it is connected as yd11: see Fig 2.2.5b. However, any arrangement could be used which
produces a phase shift of 30° between the bridges, for example yd1, yd5, yd7.
Consider the conduction sequence phasor diagram shown in Fig.2.2.1c. The equivalent diagram for this
configuration is shown below in Fig. 2.2.5b.
With a 12-pulse bridge the maximum overlap angle that can exist without there being more than one
commutation happening at any instant (operating in single overlap mode), is 30°.
➙ Between 30° and 60° of overlap, the 12-pulse bridge is in double overlap mode
➙ Between 60° and 90° of overlap, it is in triple overlap mode
➙ Between 90° and 120° of overlap, it is in quadruple overlap mode
However, for operation with overlap up to 60°, in the ideal case, the converter voltage can be considered
to be twice the resultant from the single overlap, 6-pulse equations. Equally, for overlaps above 60°, in
the ideal case, the converter can be considered to be twice the resultant from the double overlap 6-pulse
equations.
2.2.5.2. Mutual Commutating Impedance

The commutating bus is the busbar in the circuit that is considered, for all practical purposes, to have a
sinusoidal voltage waveform which is predominantly distortion free. The commutating busbar is therefore
normally considered to be the point on the AC side of the converters where the AC harmonic filters are
connected. For a 12-pulse bridge, two 6-pulse bridges are connected in parallel on the AC side as described
in section 2.2.5.1 above. Any series impedance between these two bridges and the commutating bus
is referred to as a common impedance as the current to both bridges must flow through this common
impedance. A typical example of this is shown in Fig. 2.2.5c.
There are several converter transformer arrangements that can be used to create a star and delta valve
winding connection: refer to section 6.4. Some of these arrangements tend to have more common impedance
than others. Consider for example the commonly used single-phase, 3-winding converter transformer
arrangement. Typically, the star and the delta windings are wound around separate limbs of a common

CHAPTER
Ideal
transformer 6-Pulse
bridge
Common
Vac impedance
L2
Vemf Xsys
L1
Vd
L2
Commutating
bus
Valve winding
impedance
AC harmonic
filter
Fig. 2.2.5c– A 12-pulse bridge with a common impedance
core, as shown in Fig. 2.2.5d. In this arrangement the

flux resultant from the harmonic currents 11, 13, 23, Transformer
Line side core Line side
25, etc. (refer to section 4.1) will be in-phase and winding 1 winding 2
circulate around the core. However, the harmonic
currents 5, 7, 17, 19, etc. will result in opposing fluxes
which must therefore circulate outside the core and,
consequently, have a higher reluctance.
As the path taken by the fluxes resulting from these
harmonics is dependent on the interaction between
the two windings (that is they are opposing each
other), this effect can be considered for analysis
purposes as a positive impedance common to both Valve side Valve side
winding 1 winding 2
windings, as shown in Fig. 2.2.5c. In practice, in
order to minimize this common impedance and to Fig. 2.2.5d– A typical single-phase, 3-winding converter transformer
reduce the effects of eddy current heating resulting arrangement (Note: ‘Line side winding 1’ and ‘Line side winding 2’ are
from these harmonic fluxes, a five-limb core is connected in parallel)
typically used. The five-limb core provides two half-
limbs with no additional windings in parallel, with each wound limb providing a flux return path within the
core. A five-limb core effectively eliminates the common impedance between the windings.
The consequence of this common impedance is that a commutation in one 6-pulse bridge, which is a
temporary line-to-line short-circuit, will result in a change to the applied voltage of the other 6-pulse bridge.
This change will therefore appear as another notch on the valve voltage waveform: refer to section 2.2.2.
When analyzing the impact of this common impedance it is convenient to reference it in terms of a mutual
factor, where this mutual factor can be defined as:
L1 ⋅ N 2
m= [Eqn 2.2.5a]
L2 + L1 ⋅ N 2
Where:

CHAPTER
m = mutual impedance factor
L1 = the common impedance between the two windings
L2 = the leakage impedance of each winding
N = the turns ratio of the converter transformer
Typically L1 and L2 are defined as X1 p.u. and X2 p.u. (when defined on a common base).
The presence of common impedance will mean that the 12-pulse bridge single and double overlap mode
operation, along with triple and quadruple overlap mode operation, cannot be analyzed using the same
algebraic methods as discussed in section 2.2.5.1 above.
For the 12-pulse bridge single overlap mode, that is where the overlap is less than 30°, there are no
commutations occurring in both 6-pulse bridges at the same instant and therefore the analysis of this mode
(which covers the majority of operating cases for a conventional HVDC interconnection) is as described
in section 2.2.2.
An important impact that the common impedance has in the single overlap mode is the position of the
consequential notch on the valve voltage waveform. As an example, consider the converter operating as
a rectifier where the common impedance, hence the coupling factor m, is positive. Assuming a converter
where X2 = 0.14 p.u. and X1 = ±0.014 p.u., operating with a firing angle of 15° and an overlap of 20°, this
Negative common
impedance
40 No common
impedance
20
Time (electrical degrees)

0
0 50 100 150 200 250
Valve voltage (%)
-20
-40
-60
Positive common
-80 impedance
-100
-120
Negative common
impedance increases
valve voltage stresses
Fig. 2.2.5e– Effect of common impedance on a rectifier valve winding voltage
will result in a valve voltage waveform as shown in Fig. 2.2.5e. This includes the effect of the notch as the
valve voltage crosses zero.
As can be seen from Fig. 2.2.5e the common impedance has little impact on the firing instance of the
valve, but a negative impedance may increase the peak voltage during part of the valve voltage waveform,
thereby increasing the voltage stress on the valve damping components (see section 6.2) and the valve
surge arrester (see section 6.7).
Now consider the same converter (X2 = 0.14 p.u. and X1 = ±0.014 p.u.) which is now operating as an inverter,

CHAPTER
again with an extinction angle g of 15° and an overlap of 20°. As can be seen from Fig. 2.2.5f the effective
extinction angle g is decreased with a positive common impedance, but is actually increased if the common
impedance is negative.
As discussed in section 2.2.3.3, the size of the extinction angle is important in order to minimize the
susceptibility of the converter to commutation failures. In practice, where the converter transformer
designer predicts a positive common impedance in the transformer design, this is accommodated by an
increase in the selected minimum extinction angle of the converter in order to compensate for the effects
of the positive going notch. Changing the minimum extinction angle changes all of the scheme parameters
Negative common
impedance increases
valve voltage stresses
γ > 15°
γ < 15°
γ = 15°
120
100
80
Valve voltage (%)
60
40
20
0
0 50 100 150 200 250
-20
Negative common
-40 impedance
Positive common
impedance
No common
impedance
Fig. 2.2.5f– Effect of common impedance on an inverter valve winding voltage
and hence the parameters of the converter transformer; consequently this is an iterative process between
the transformer designer and the scheme designer.
The change in effective extinction angle (Dg ) can be calculated as:
 1 
∆γ = 30° − tan −1   [Eqn 2.2.5b]
 3 ⋅ (1 − m ) 
The double, triple and quadruple overlap modes of operation cannot be analyzed using the same algebraic
methods as discussed in section 2.2.5.1 above, as interaction between the bridges will reduce the average
DC output.
It should be noted that where a 4-winding converter transformer is employed with the AC harmonic filters
connected to a transformer winding (see section 3.1.4.2), then the effective impedance between the
star, delta and filter winding can be configured to be negative and relatively large, thereby increasing the
effective extinction angle. This is particularly useful, as the 4-winding converter transformer configuration is
most applicable where the AC system Short-Circuit Level is low (see section 5.1). The converter transformer
tapchanger can be used to maintain the filter bus voltage constant in steady-state, compensating for
variations in the AC system voltage and thereby keeping the reactive power generated by the filters

CHAPTER
at a constant value (see section 2.3.4). As a low Short-Circuit Level AC system is more prone to voltage
disturbances, the added advantage of increased extinction angle is of benefit [7].
However, care still needs to be taken with this arrangement because although the effective extinction
angle may be extended by the notch, the notch will also introduce a fast, forward-direction voltage step
onto the valve voltage waveform at a time close to voltage zero. If not properly coordinated with the valve
dv/dt protection, this fast voltage step may still result in a commutation failure.
It can be shown that for operation in these modes the calculated DC voltage and current can be found as:
Double overlap mode (30° ≤ m < 60°):
1 2 ⋅ [ cos(α ) + cos(α + µ )] 

Vd = ⋅ Vdi 0 ⋅   [Eqn 2.2.5c]
( 2− m⋅ 3 ) − m ⋅ 3 ⋅ [ cos(α + 30°) + cos(α + µ − 30°)]
2 ⋅ [ cos(α ) − cos(α + µ )] 
2 ⋅ E LL  
Id = ⋅  − cos(α + 30°)  [Eqn 2.2.5d]
2 ⋅ ω ⋅ Lc ⋅ (2 + m ⋅ 3 ) + m ⋅ 3 ⋅  
  + cos(α + µ − 30°)  
Triple overlap mode (60° ≤ m < 90°):
3 (1 − m ) (2 + 3 ⋅ m ) ⋅ [ sin(α ) − sin(α + µ )] 

Vd = ⋅ ⋅ Vdi 0 ⋅   [Eqn 2.2.5e]
(
2 2− m⋅ 3 ) + (2 − m ) ⋅ 3 ⋅ [ cos(α ) + cos(α + µ )]
(2 + 3 ⋅ m ) ⋅ [ sin(α ) + sin((α + µ )]

2 ⋅ ELL  
Id = ⋅  cos(α )   [Eqn 2.2.5f]
4 ⋅ ω ⋅ Lc ⋅ (1 + m ) ⋅ (2 + m ⋅ 3 ) +(2 + m ) ⋅ 3 ⋅   
  − cos(α + µ )  
Quadruple overlap mode (90° ≤ m < 120°):
Vd = 3 ⋅ (1 − m ) ⋅ Vdi 0 ⋅ [ sin(α + 60°) − sin(α + µ − 60°)] [Eqn 2.2.5g]
2 ⋅ ELL
Id = ⋅ [ sin(α + 60°) + sin(α + µ − 60°)] [Eqn 2.2.5h]
2 ⋅ ω ⋅ Lc ⋅ (1 + m )
2.2.6. CONVERTERS WITH FINITE DC REACTANCE

All of the preceding analysis has assumed that the DC side reactance is infinite, that is, the DC current is
always ripple free and continuous. In the practical case, the DC side reactance always has a finite value.
This DC reactance comes from intentionally added reactance in the form of smoothing reactors plus a
contribution from the DC circuit and converter transformers. Even in the special case of the back-to-back
converter where no smoothing reactor is employed, there is still a considerable amount of inductance seen
by one converter as a consequence of the commutating impedance of the other converter (see section 6.5).
One important result of considering a finite DC reactance is that the DC current can now become discontinuous
at low DC current levels. Another change is that there will now be coupling between series converter bridges,
hence, there will be additional (typically small) notches in the valve voltage waveform. Consequently, the
harmonic voltages produced by the converters will no longer simply be the harmonics produced by one bridge
multiplied by the number of bridges.

CHAPTER
2.2.6.1. Operating Modes
The operating modes of a 12-pulse converter with finite DC reactance, in order of increasing DC current are:
Overlap angle Maximum number of valves
Mode
(m) conducting at any instant
Low discontinuous current - 1
Principal discontinuous current - 4
Single overlap 0° ≤ m < 30° 5
Double overlap 30° ≤ m < 60° 6
Triple overlap 60° ≤ m < 90° 7
Quadruple overlap 90° ≤ m < 120° 8
Short-circuit - 12
In the following sections, discontinuous current operating modes will be introduced and the consequence of
finite DC reactance will be introduced for the single overlap case. Further assessment of the impact of finite
DC reactance on operating modes at higher DC currents will not be reviewed because, as may be seen from
the single overlap case, the impact diminishes significantly with increasing DC current.
2.2.6.2. Single Overlap Mode

In order to assess the impact of a finite value of DC reactance it is necessary to introduce a new ratio,
referred to here as the finite reactance ratio (a) and defined as:
Ld
a= [Eqn 2.2.6a]
Lc
Where:
a = finite reactance ratio
Ld = the effective DC side impedance of the converter
Lc = the commutating impedance of the converter
From this, a set of equations can be developed to give the actual DC voltage and current for a given finite
reactance ratio.
 sin(α + µ ) − sin(α ) 
π ⋅ [ cos(α + µ ) + cos(α )] ⋅ (2 ⋅ a + 7 ) −
1  3−2 
Vd = ⋅Vdi 0 ⋅   [Eqn 2.2.6b]
2  6 ⋅ µ + π ⋅ ( 2 ⋅ a + 7 ) 
 
2 ⋅ ELL
Id = ⋅
4 ⋅ π ⋅ ω ⋅ Lc ⋅ (2 ⋅ a 2 + 15 ⋅ a + 28 )
  π  6⋅µ  
 ( 2 ⋅ a + 7 ) ⋅  + 12  +  ⋅ ( sin(α + µ ) + sin(α ))  [Eqn 2.2.6c]
 3−2  3 − 2 
 
− (cos(α + µ ) − cos(α )) ⋅  µ ⋅ (12 ⋅ a + 42 ) + π ⋅ (2 ⋅ a + 7 )2 − 12  
  3 − 2  
2.2.6.3. Principal Discontinuous Current Mode

With a finite DC reactance it is not possible to maintain the DC current constant at very low DC current
levels. Instead, the DC current separates into pulses of current through each valve during the conduction
period, hence this is referred to as discontinuous current operation. For a 12-pulse converter, discontinuous
current will always flow through two valves in series per bridge on different phases of the transformer. The
current in each phase will have four pulses as shown in Fig. 2.2.6a.
With a finite DC reactance, the DC voltage on the line side of the DC reactor may not be completely smooth
and there may be some ripple present instead. In the extreme case where there is no DC reactor, for example in
CHAPTER
the case of a back-to-back, the ripple voltage of one converter will be superimposed onto the other converter.
As the AC frequencies behind each converter will almost certainly not be synchronized, the pulses resulting
from one converter will sweep through those of the other converter as the angle between the AC frequency
vectors change. This can lead to a valve seeing up to eight current pulses during its conduction cycle.
An analysis of the threshold of the principal discontinuous current mode can be found in section 6.5.
DC current
Valve current
Valve winding current
0 50 100 150 200 250 300 350 400

Fig. 2.2.6a– Principal discontinuous current mode
2.2.6.4. Low Discontinuous Current Mode

Low discontinuous current mode is associated with the resistive – capacitive damping networks connected
in parallel with the thyristors in a valve: see section 6.2. It applies to an average DC current from zero up to
a value almost equivalent to the value of fundamental frequency AC current flowing in the valve-damping
network when the valve is blocked, typically at around 0.2% of rated DC current.
This mode is characterized by each valve conducting singly, once per cycle, with a short pulse of current,
typically around 4°, drawn by charge transfer in the valve damping network. Each valve winding current will
see two current pulses, one positive and one negative, 180° apart.
This operating mode is typically only of importance when the converter is operated with no DC load, for
example during open circuit testing. The main effect of this operating mode is that the DC voltage will be
higher than the rated DC voltage.
2.2.6.5. Errors in Classical Theory Resulting from Finite Reactance

In most practical cases it is sufficient to consider only the simple classical equations given in section 2.2.2,
assuming infinite DC reactance, for converter design purposes. In particular, under rated conditions the
errors introduced by the use of the classical equations are negligible, as shown in Fig. 2.2.6c. It is also
important to note that in most practical cases the value of the DC reactor is significant in relation to the
commutating impedance of the converter. Consider for example, a HVDC back-to-back converter with no
DC reactor. During steady-state operation, between the DC terminals of any one converter there will be
the leakage reactance of the other converter as shown in Fig. 2.2.6b.

CHAPTER
For a typical back-to-back converter such as [8] this gives a finite reactance ratio of approximately four and
for a typical transmission scheme such as [9], the ratio is approximately 10.
The equations presented in section 2.2.6.2 give the values of DC voltage and current for a finite DC
reactance and for a defined firing angle (a) and overlap angle (m). However, in most practical HVDC schemes
both the DC voltage and current are controlled in a closed loop, hence the operating conditions of the
converter (the firing angle) are adjusted to meet the target values of DC voltage and current. Errors resulting
from a finite DC reactance will manifest themselves in the actual operating angle and overlap angle of
the converter for a given target DC voltage and current, when compared with the operating conditions
predicted by the classical theory presented in section 2.2.2.1.
Side A Side B
Id Instantaneous current
path through converter
ea Lc la T1 T5 T9 T1 T5 T9 1 ec
eb Lc lb eb
ec Lc lc 2 ea
T7 T11 T3 T7 T11
T3
Vd
ea-30° Lc la T12 T4 T8 T12 T4 T8 4 ec-30°
eb-30° Lc lb 3 eb-30°
ec-30° Lc lc ea-30°
T6 T10 T2 T6 T10 T2
Id Effective DC reactance as seen from Side A
is 4 x leakage reactance of Side B
Fig. 2.2.6b – The effective DC reactance of a back-to-back converter without a separate DC reactor
Fig. 2.2.6c shows a graph of calculated errors in firing angle (a) and overlap angle (m) for a typical 12-pulse
converter operating at 1.0 p.u. DC voltage for currents between zero and 1.0 p.u., assuming a commutating
impedance of 0.12 p.u. and varying values of DC reactance. Note that the x-axis of the graph represents
the case of infinite DC reactance and that as the finite value of DC reactance increases with respect to the
commutating impedance of the converter, the operation of the converter becomes closer to the infinite
reactor case, that is, the classical theory.
In Fig. 2.2.6c it can be seen that the maximum error in overlap (m) is around 1.1° in the worst case, but when
considering a practical case the error is much smaller. The maximum error in firing angle (a) is around 0.5°
when the converter is operating in single overlap mode. Below a minimum DC current value the current
becomes discontinuous, hence overlap is irrelevant and the firing angle error increases rapidly.
2.3. REACTIVE POWER EXCHANGE BETWEEN HVDC CONVERTER

STATIONS AND AC SYSTEMS
HVDC converters absorb reactive power and therefore require reactive compensation. This is a significant
aspect of the design of a HVDC station, as discussed below.
2.3.1. REACTIVE POWER IN AC SYSTEMS

Reactive power is inherent within all AC power systems. It results from the stray capacitance and

CHAPTER
1.5 Ld
a=
Lc
Error in 1 a=0
overlap μ
a=1
angle μ 0.5
(degrees) a=3
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
a=3
Error in -0.5 a=1
DC current (p.u.)
firing a=0
angle α α
(degrees) -1
-1.5
Fig. 2.2.6c– Angle errors resulting from the use of classical theory with a finite DC reactor (12-pulse converter, Vdc =
1.0 p.u., Xc = 0.12 p.u.)
inductance within all elements of the power system, including transmission lines, transformers, machines,
cables, etc. Its effect is to shift, in phase, the current AC waveform with respect to the voltage AC waveform,
hence reducing the instantaneous value of voltage multiplied by current. In order to assess the effect of
this phase shift, the AC power is considered as two components: the ‘Real’ power which results from the
in-phase component of voltage and current and the 90° out-of-phase component of voltage and current
which is referred to as ‘Reactive’ power.
Reactive power can either be leading (current waveform is phase advanced with respect to the voltage
waveform) or lagging (current waveform is phase delayed with respect to the voltage waveform). In HVDC
systems it is conventional to consider leading reactive power as a source or generator of reactive power
and lagging reactive power as a load or absorber of reactive power. Conceptually speaking, reactive power
resulting from capacitance is generated and reactive power resulting from inductance and from the converter
is absorbed.
An AC system is composed of generators, VAr compensators, transmission lines and various inductive and
capacitive loads. Reactive power flow through the AC system results in voltage variation between busbars.
Part of the system operator’s function is to ensure that the power system operates in the steady-state,
with the voltages at the AC system busbars within their normal operating limits; for example maintaining
the voltage at each busbar to a range of 0.95 p.u. and 1.05 p.u. In order to achieve this voltage control with
changing system load conditions and configurations (for example, lines switched in or out), controllable
elements within the AC system are used.
Examples of such control elements are: switchable line-end shunt reactors, switchable shunt capacitors
(a special case of these is the Mechanically Switched Damped Capacitor Network, MSCDN, which provides
some harmonic damping on the AC system as well as fundamental frequency reactive power), transmission
transformer tapchangers and Automatic Voltage Regulator (AVR) controllers on generators. There may be
other reactive power devices within the AC system such as Static VAr Compensators and STATCOMs, etc., but
such devices are usually reserved by the System Operator for dynamic reactive power control (for example
to assist with fault recovery) and thus should not be considered as contributing to the steady-state reactive
power control of the AC system.
When any major new piece of equipment is added to the network, it is important to optimize the design of
that equipment and this includes the reactive power exchange limits within which the equipment is permitted
to operate in order to maintain the steady-state AC busbar voltages.
2.3.2. DEFINING THE REACTIVE POWER LIMITS AT THE POINT OF CONNECTION

As discussed above, the AC system busbar voltages are determined largely by the reactive power flow

CHAPTER
within the AC system. The voltages across the system change, depending on the reactive power loads
on the busbars and the length of the transmission lines interconnecting the busbars. Adding a new
reactive power load at a particular busbar, the additional reactive power loading on that busbar will change
the voltages at other busbars. By switching the switchable reactive power banks on other busbars and
operating the power system transformer tapchangers, the voltages may again be brought back into their
steady-state range.
Both the capacitive and inductive reactive power loading at a particular busbar can be found by simply
repeating the load flow analysis with varying sizes of reactive power load on that busbar until the busbar
voltage reaches the limit value. When no other switchable elements within the network can be switched on
or off, as appropriate, and no further power system transformer tapchangers can be operated, the limit of
reactive power exchange, for that particular operating condition is established.
One can easily determine the reactive power limits at the point of connection of a HVDC scheme based on
this method. Firstly, one should establish a loadflow representation of the network into which the HVDC
converter is to be connected. The number of busbars to include in this loadflow is dependent both on how
meshed the AC system is and on the length of the transmission lines: These will be different for each project.
Having established the loadflow, the value of real power (P) can be varied throughout the HVDC converter
capability range (from 1.0 p.u. rectifier operation to 1.0 p.u. inverter operation) and corresponding limits of
capacitive and inductive reactive power exchange with the AC system can then be calculated.
This method should be repeated for different AC conditions (for example, peak loading and light loading of
the network), as well as for changing system conditions (transmission lines and generators in or out). The
result will be a set of data points that represent the maximum absorption or generation from the HVDC
station busbar that can be accommodated by the AC system, under all considered conditions. An example
of such an analysis is shown below in Fig. 2.3a [8].
Fig. 2.3a shows the calculated reactive power exchange limits for a back-to-back HVDC interconnection.
From the figure, we can see that between an AC system busbar limit of 1.05 p.u. and another limit of 0.97
p.u. (selected by the client) there exists a band of reactive power exchange where the HVDC station can
operate.
2.3.3. VOLTAGE STEP CHANGES

Another requirement imposed on the reactive power design is that of limiting the AC voltage step change
Reactive power exchange with AC network, 50 Hz

Operating mode 1
600 Curve 1 - Net Q exchange
Curves 4 and 6 for LOW target Ud with one
500 set of extreme tolerances
400 Curve 2 - Net Q exchange

Curve 3 for LOW target Ud with
Reactive power (MVAr)
300 second set of extreme tolerances

Curve 3 - Net Q exchange
200 for HIGH target Ud with
Curve 1
same set of extreme
100 tolerances as curve 1
Curve 2 Curve 4 - Upper limit of Q
0
exchange C0:1.03 p.u.V 2008lt
-100 Curve 5 - Lower limit of Q
Curve 5 exchange C0:0.97 p.u.V 2008lt
-200
+ + + + + Curve 6 - Upper limit of Q
-300 exchange C3:1.05 p.u.V 2008lt
-1500 -1000 -500 0 500 1000 1500
Pdc (MW)
Fig. 2.3a– Calculated reactive power limits at a converter busbar for different system conditions [8]

CHAPTER
which will take place as a consequence of switching a filter bank (or any reactive power element). Many
utilities have their own rules concerning the magnitude of permissible voltage step changes and some
utilities apply a sliding scale: the more frequent a switching event takes place, the smaller the permissible
voltage step. It is therefore important when specifying a HVDC scheme that the expected load cycle of
the scheme is known, that is, how often will the DC power be ramped up and down, resulting in harmonic
filter switching?
Another factor when considering the maximum permissible voltage step is that of differentiating between
‘normal’ switching operations and reactive power equipment trips. As a tripping event will be rare, it may
be acceptable to permit a larger voltage change on tripping than on a normal, controlled, switching event.
Other actions can then be used to control the voltage magnitude on switching such as converter control
or binary switching as discussed in the following.
The magnitude of a voltage step change as a consequence of switching a filter can be approximated as:
QSWITCH
∆V = [Eqn 2.3a]
SCLmin − QTOTAL
Where:
DV = the change to AC voltage (p.u.)
SCLmin = the minimum Short-Circuit Level of the AC system on which the switching operation is to take
place (MVA)
QSWITCH = the reactive power step to be imposed on the AC system (MVAr)
QTOTAL = the total reactive power connected to the converter bus including the reactive power to be
switched (MVAr)
A typical value of DV imposed by clients is 3%.
A more exact voltage change can be calculated using the Thévenin equivalent circuit of the AC system
connected to the HVDC station busbar, Fig. 2.3b.
The magnitude of the AC system source voltage, ‘E’, shown in Fig. 2.3b can be derived from:
( Pac + jQac ) ⋅
E = Vac +
Vac
( 0 − jXSYS ) [Eqn 2.3b] AC system AC system HVDC
source voltage short circuit station
impedance
Where:
X sys
E = the Thévenin equivalent voltage of the AC
system
Vac = the HVDC station busbar voltage E Vac Pac + jQac
Pac = HVDC station real AC power
Qac = HVDC station reactive AC power
XSYS = the Thévenin equivalent short-circuit Fig. 2.3b– A simple Thévenin equivalent circuit of the
impedance of the AC system AC system connected to an HVDC station busbar
This AC system source voltage ‘E’ can first be

calculated for steady-state conditions at the HVDC station, that is before any reactive power elements
are switched. Hence, the calculated source voltage represents the steady-state AC system immediately
following any reactive power switching at the HVDC station, but before any of the generator’s AVRs,
transmission transformer tapchangers, reactive power flow control banks, etc., respond.
Using this value of ‘E’ and substituting the new value of reactive power following the switching operation
into Eqn 2.3b, the equation can be solved iteratively in order to find the new value of the HVDC station
busbar voltage, Vac. Hence the step voltage from the steady-state condition to this new, temporary
condition, (that is, before the AC system source voltage , ‘E’, has responded to the HVDC station change)
can be found.
Where the optimum size AC reactive power bank results in a step change in AC voltage which exceeds
the voltage change limit, it is possible to reduce the effective step change by temporarily imposing an
opposite change in reactive power at the converter busbar. As mentioned above, this opposite change can

CHAPTER
be achieved through binary switching of reactive power elements or by converter action.
When using the converter, a fast change to the DC voltage is applied whilst maintaining the DC power.
As an example, consider switching in a filter onto an AC system where the filter is too big in terms of its
fundamental frequency VAr rating, to meet the AC voltage step change limit. By increasing the DC converter
absorption at the same time as the filter bank circuit breaker closes, the net reactive power exchanged
with the AC system can be reduced and hence, the step change in AC voltage is also reduced. The converter
operating conditions can then be returned to steady-state over several cycles, thereby applying a gradual
change to the system voltage as opposed to a step. This gradual change will allow the AC system voltage
controllers to respond to the change in most power systems: see Fig. 2.3c.
The converse control action can also be achieved, as long as a margin is maintained within the converter
DC voltage control band. Alternatively, if a controlled switching operation is to take place, and the converter
controller ‘knows’ that it is about to switch a filter, the converter can pre-condition the AC system by
increasing its reactive power absorption several cycles prior to the reactive power bank being switched-off.
This allows the converter to step back to its steady-state condition: see Fig. 2.3d.
2.3.4. THE REACTIVE POWER LOAD OF A CONVERTER

Converters are a reactive power load as they operate with a delayed firing angle, thus the current always
lags the voltage. The converter transformer impedance (plus the small valve impedance) also introduces
an additional lag in the current, which is observed as the overlap angle. This is true for operation as either
a rectifier or an inverter.
Fig. 2.3e below shows the converter AC side current with respect to the AC system voltage, along with the
Capacitive reactive power Capacitive reactive power

from HVDC station switchable from HVDC station switchable
elements elements
AC harmonic filter de-energised
AC harmonic filter energised
Converter reactive power Converter reactive power

absorption Converter absorption absorption Converter absorption
rapidly increased
slowly increased
Converter absorption
Capacitive reactive power
exchanged between the HVDC
Converter absorption slowly Capacitive reactive power rapidly increased back
returned to steady-state exchanged between the HVDC to steady-state
station and the AC system
station and the AC system
} Voltage ʻstepʼ limited

} Voltage ʻstepʼ limited
Time
Time
Fig. 2.3c– Using the converter to minimize the Fig. 2.3d– Using the converter to minimize the
voltage step change when energizing a filter voltage step change when de-energizing a filter
associated phasor diagram. We can see that when the operating angle (firing angle, a or g, plus overlap
angle, m) equals 90°, the converter exchanges no real power with the AC system but still absorbs reactive
power. This is an important operating characteristic of line commutated converter HVDC, which prevents
voltage collapses in one AC system being exported into the interconnected AC system: see section 5.1.
The converter operating overlap angle is a function of the operating current and the converter transformer
leakage reactance, as described in section 2.2.1.

CHAPTER
From the equation given in section 2.2.1, the reactive power absorption of a converter at rated load can
be approximated as:
  X 
Qdc0 p.u. = tan  cos −1  cos δ − p   [Eqn 2.3c]
  2 
Where:
Qdc0p.u. = the reactive power absorption of the converter at rated DC current (p.u.)
(α + μ) = 0° IDEAL Case
Vc
Vb Converter line current
Ia Va
210 330 360
-30 0 30 60 90 120 150 180 240 270 300 390
Fundamental component of
μ = 16° converter line current
Φ
Va
Ia
(α + μ) = 68°
Va
Φ
Ia
(α + μ) = 90°
Va
Φ
Ia
(α + μ) = 165°
Va
Ia Φ
Fig. 2.3e– Lagging currents in a rectifier and an inverter

CHAPTER
Xp = converter transformer leakage reactance (p.u.)
d = converter control angle
= alpha (a) for rectifier operation (rad)
= gamma (g ) for inverter operation (rad)
This reactive power absorption by the converter is typically compensated for locally at the converter station by
capacitive reactive power generating components. The majority of this capacitive reactive power is normally
supplied by the switchable AC harmonic filters connected to the converter busbar.
It is important to note that, while the converter transformer is fitted with a tapchanger to compensate for
variations in AC system voltage, other reactive compensation equipment connected to the converter station
busbar, such as the AC harmonic filters, may not be compensated. Therefore, the reactive power these devices
generate will vary with AC system conditions. For example, the fundamental frequency capacitive reactive
power generated by the energized AC harmonic filters is defined by:
Vac 2 f
Qfilt = Qfilt 0 ⋅ ⋅ [Eqn 2.3d]
Vac02 f0
Where:
Qfilt = the actual reactive power supplied by the filters
Qfilt0 = the reactive power supplied by the filters at 1.0 p.u. voltage and frequency
Vac = the actual HVDC station busbar voltage
Vac0 = the 1.0 p.u. value of AC system voltage
f = the actual HVDC station busbar frequency
f0 = the 1.0 p.u. value of AC system frequency
Other factors will affect the actual reactive power generated by the AC harmonic filters such as component
manufacturing tolerances and ambient temperature. All of these variations, if required by the client, are taken
into consideration when calculating the net reactive power exchange with the AC system.
2.3.5. THE IMPACT OF CONTROL METHODS ON CONVERTER REACTIVE POWER

ABSORPTION
From the equations presented above it can be seen that the reactive power loading of a converter at rated
power is a function of the converter operating angle and the commutating impedance of the converter,
which is mainly the converter transformer leakage reactance. However, the reactive power absorbed by a
converter when operating below rated power will depend on the control method applied to that converter.
Let us consider Constant Extinction Angle (CEA) control compared to constant valve winding voltage control
(see section 7.1 for a description of these control methods).
Fig. 2.3f shows the reactive power increase with DC power for a converter operating in CEA control. It
can be observed that at low power the converter commutating impedance has a bigger influence on the
reactive power absorbed by the converter than the actual operating angle, as there is little divergence
between the two ‘12%’ or the two ‘18%’ curves. However at higher powers the operating angle becomes
the dominant factor in determining the reactive power absorbed. We can also observe that the reactive
power absorption with increasing DC power is non-linear and that a discontinuity of slope occurs at 1.0 p.u.
The discontinuity is a consequence of operating the converter in overload, which necessitates maintaining
the firing angle at, or above, the minimum value while reducing the DC voltage in order to maintain the
valve winding voltage constant - and not exceeding the converter voltage rating, as dictated by the number
of thyristor levels in the converter.
Fig. 2.3g shows a similar characteristic to Fig. 2.3f, but this time for constant valve winding voltage control.
Using this control method, the reactive power absorbed from the AC system changes linearly with changing
DC power up to rated operation, above which the converter operates in overload and follows the same

CHAPTER
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
-0.1
-0.2 Increasing γ
Converter reactive power absorbtion (p.u.)
-0.3 Increasing Xc
-0.4
-0.5
Xc = 12%
Xc = 12%
-0.6
Xc = 18%
-0.7 Xc = 18%
-0.8
DC real power (p.u.)
Fig. 2.3f– Reactive power absorption of a converter operating with CEA control
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
-0.1
Increasing Xc
-0.2
-0.3 Increasing γ min
-0.4
-0.5
Xc = 12%
Xc = 12%
-0.6
Xc = 18%
-0.7 Xc = 18%
-0.8
Fig. 2.3g– Reactive power absorption of a converter operating with constant valve winding voltage control

CHAPTER
characteristic as a converter in CEA control. In comparing these two controls, we can see that with reducing
DC power the reactive power absorbed using constant valve winding voltage control is higher than that of
CEA control. Hence, it may be possible to switch in AC harmonic filters earlier using constant valve winding
voltage control.
There is an important difference between CEA control and constant valve winding voltage control. Whilst
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
-0.1
-0.2
Increasing γ
-0.3
Constant γ control
Increasing Xc
-0.4
Constant valve winding voltage control Xc = 12%

-0.5
-0.6
-0.7 Xc = 18%
-0.8
Fig. 2.3h– Comparison of the reactive power absorption of a constant valve winding voltage control scheme and a CEA control
control action maintains the operating angle at a constant minimum value in CEA control, in constant
valve winding voltage control the scheme is designed so that, allowing for manufacturing tolerances and
measurement errors, the actual operating angle is at least the minimum value. Consequently, with constant
valve winding voltage control the actual operating angle will be greater than the minimum design limit and
hence the reactive power absorbed by the converter will also be greater.
2.3.6. REACTIVE POWER SOURCES

WITHIN A CONVERTER STATION
C1
The main sources of reactive power in a HVDC
station are the AC harmonic filters: refer to
section 4.1. Harmonic filters have two purposes:
L1 R1
reducing the harmonics injected into the AC system
and generating fundamental frequency reactive
power.
2.3.6.1. AC Harmonic Filters C2 L2 R2
An AC filter is composed of capacitance, inductance
Fig. 2.3i– The single line diagram of a typical AC

harmonic filter

CHAPTER
and resistance but at fundamental frequency the HV connected capacitor is the main contributor towards
the reactive power generated.
To a first approximation, the fundamental frequency reactive power generated by a band-pass filter
(section 4.5) is:
hn 2
Qfilt 0 = Qcap ⋅ [Eqn 2.3e]
hn 2 − 1
Where:
Qfilt0 = reactive power supplied by the filters at 1.0 p.u. voltage and frequency
Qcap = reactive power supplied by the filter’s main capacitor at 1.0 p.u. voltage and frequency
hn = harmonic number to which the filter is tuned
Therefore, for the 11th harmonic, the filter will generate more reactive power than would a plain capacitor
bank of the same value (Eqn 2.3f):
 112 
 112 − 1 − 1 ⋅ 100 = 0.8% [Eqn 2.3f]
For higher harmonics, the difference is smaller.
2.3.6.2. Switched Shunt Capacitors and Capacitor Damped Networks

Switched shunt capacitors and Mechanically Switched Capacitor Damped Networks (MSCDNs) are similar
to AC harmonic filters. However, while AC harmonic filters have the dual purpose of providing capacitive
reactive power and filtering unwanted harmonics from the converter busbar, switched shunt capacitors
and MSCDNs are included to supply capacitive reactive power only.
At a converter station, it is more common to use a shunt capacitor tuned to a known frequency - using a
small series reactor, or an MSCDN - than to use a simple shunt capacitor, as tuning the capacitor to a known
frequency avoids the risk of unpredicted resonances occurring with the power system, which could lead to
capacitor failure.
MSCDNs and switchable shunt capacitors are used where there is a requirement for additional reactive
power support in excess of that provided by the AC harmonic filters and where simply adding additional AC
harmonic filters is not cost effective.
2.3.6.3. Switchable Shunt Reactors

The reactive power generated by the AC harmonic filters required for filtering may exceed reactive power
exchange limits, particularly at low DC power. Under such conditions, it is sometimes necessary to include
one or more shunt connected reactors to provide additional reactive power absorption at the converter
station to compensate for the filters.
Typically, the shunt reactors used in a HVDC station are oil-immersed, iron-cored reactors connected to the
HVDC station AC busbar which could, for example, be at 400 kVac. The circuit breaker associated with switching
the reactor is not designed for frequent switching operations. Hence, shunt reactors are normally energized at
start-up, where the converter absorption is at its lowest, and kept energized until the DC power transmission
level exceeds some pre-determined level, at which point the reactor is de-energized. When DC power is
reduced, the shunt reactor is again energized but in this case the switching point for the reactor will be at a
different DC power level in order to provide some hysteresis in the switching operation of the shunt reactor.
Where it is possible to connect the shunt reactor to a medium bus voltage, for example, where the converter
transformer has four windings with the filters and shunt reactors connected to this fourth winding [7], then
air-insulated, air-cored shunt reactors can be used. As these reactors are single-phase units, the land area they
occupy - allowing for magnetic clearance (as they have no intrinsic magnetic shielding) - can be larger than
that of a tanked reactor, though their cost is significantly less. When connected to a medium voltage supply

CHAPTER
it is possible to use load-break switches instead of circuit breakers to control the reactors. This increases the
number of permitted switching operations, allowing binary switching to be considered as a reactive power
control strategy.
‘Binary Control’ is when two or more shunt-connected elements are switched on or off at the same time
(allowing for circuit breaker and load-break switch mechanical operation variations). As an example, assume
that the optimized AC harmonic filter solution suggests a 100 MVAr bank, but the reactive power limits will
not permit more than 50 MVAr to be energized at start-up. By switching on the 100 MVAr filter and, at the
same time, energizing a 50 MVAr shunt reactor, the net reactive power exchange with the AC system is 50
MVAr therefore meeting the limit value. As the power through the DC link is increased, the reactive power
absorbed by the converter will increase, and at some power level the converter will have increased by 50
MVAr. It will then be possible to de-energize the shunt reactor, again only applying a 50 MVAr step, leaving
the 100 MVAr filter connected.
Where there is a limit on the voltage step resulting from reactive power element switching, the technique of
binary switching is also useful, as it allows bigger and possibly more efficient and cost effective AC harmonic
filters to be used.
2.3.6.4. Static VAr Compensator (SVC)

A Static VAr Compensator (SVC) is a power electronic device which provides dynamic reactive power control
[10, 11]. The most common components for an SVC are a Thyristor Controlled Reactor (TCR) and a Fixed
Capacitor (FC). As the TCR generates harmonic currents, the FC is normally configured as a harmonic filter.
The TCR operates by controlling the AC current flow through a shunt-connected, 3-phase reactor set and
hence varying the effective reactance of the TCR as viewed from the AC system. The FC provides capacitive
reactive power and hence a TCR + FC SVC can provide a continuously variable reactive power output going
from fully capacitive (with the TCR blocked) to fully inductive (with the TCR operating at full current).
Where fast, capacitive reactive support is needed, Thyristor Switched Capacitors (TSCs) can be used. TSCs
allow capacitor banks to be connected or disconnected within one cycle.
Unlike the HVDC converter where the harmonic currents produced are approximately equal to 1/n, where
n is the harmonic number (n equals 12 × n ± 1 for a 12-pulse converter bridge), those produced by a TCR
are proportional to 1/n2 (where n equals 6 × n ± 1). Therefore, the harmonics generated and filtered by the
SVC can be considered independently of those of the HVDC converter, as the SVC makes little harmonic
contribution at the 11th harmonic and above.
2.3.6.5. STATCOM (STATic synchronous COMpensator)

As with the SVC, the STATCOM provides dynamic, reactive power to the AC system. However, whereas
the SVC uses passive components to provide the reactive power, the STATCOM generates a synthetic,
sinusoidal voltage waveform at its terminals using a power electronic converter [10, 12]. This synthetic
sine wave can be controlled in amplitude and phase with respect to the AC system voltage waveform. By
adjusting the phase of the STATCOM voltage to be above or below (and approximately in phase with) the
AC system voltage waveform, reactive power can be generated or absorbed.
The advantages of a STATCOM over an SVC are the reduced land area requirements (as the passive
components are eliminated), a faster response time and an ability to provide greater capacitive VAr output
at reduced AC system voltages. During undervoltages, the capacitive VAr output of an SVC falls in proportion
to voltage squared, whereas that of a STATCOM only falls in proportion to voltage.
2.3.6.7. Synchronous Compensator

A synchronous compensator is a synchronous machine which, when operating, has no mechanical power
input. The machine’s energy losses are provided by the AC system to which it is connected. A synchronous
compensator, as with a synchronous generator, has excitation control and controlling the excitation of a

CHAPTER
synchronous machine, the reactive power absorbed or generated can be controlled to within the ratings
of the machine, providing a linear reactive power output.
Synchronous compensators have an advantage over other passive or power electronic based means of
reactive power supply because, as they are rotating machines, they have inertia. This provides temporary
energy storage at the HVDC station busbar, which can have some practical advantages in terms of converter
interaction with the AC system. (See section 5.1.)
2.3.7. CONTROLLING CONVERTER REACTIVE POWER

In order to meet AC harmonic performance, each filter has to be switched in before a certain DC power
transmission level, as determined by the AC harmonic filter design. This method of control is known as
‘open-loop’ control, and is shown in Fig. 2.3j. The filter switch points are also dictated by the reactive power
exchange limits, as described in section 2.3.1.
The reactive power switch points may be open-loop, in which case they will be arranged to be the same as those
determined from the AC harmonic studies, or a ‘closed-loop’ control system will be implemented to determine
when to switch a reactive power element. A closed-loop controller will perform switching operations based on
0.6
AC harmonic filter
AC harmonic filter reactive switch point
0.4 power generation
Net reactive power exchange

0.2 with the AC system
Reactive power (p.u.)
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
-0.2
-0.4
Converter reactive power
absorption
-0.6
-0.8
DC power (p.u.)
Fig. 2.3j– Filters switched with changing DC power and resultant reactive power exchange with the AC system
the prevailing AC and DC conditions, as well as a set of rules established at the design stage.
However, in the case of a closed-loop reactive power controller, the open-loop filter switch point controller
always takes precedence. The closed-loop reactive power controller can energize an AC harmonic filter earlier,
or de-energize it later than demanded by the AC harmonic filters, but it is not permitted to energize it later
or de-energize it sooner than dictated by the AC harmonic filter open-loop controller. In this way, harmonic
performance is given priority over reactive power performance and filter operating parameters are maintained
within filter equipment ratings.
As can be seen from Fig. 2.3j there is almost always some reactive power exchange with the AC system.
The important design criterion is to ensure that the net exchange remains within the limits established as

CHAPTER
0.6
AC harmonic filter
AC harmonic filter reactive switch point
0.4 power generation
Net reactive power exchange

0.2 with the AC system
Reactive power (p.u.)
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
-0.2
-0.4
Converter reactive power
absorption
-0.6
-0.8
DC power (p.u.)
Fig. 2.3k– Filters switched with changing DC power and resultant reactive power exchange with the AC system.
Assuming two filters energized at start-up
per section 2.3.2. This usually means allowing for the complete range of steady-state AC system voltage
and frequency, as well as allowing for measurement errors on the DC side of the converter and equipment
manufacturing tolerances.
An important issue concerning the permitted reactive power exchange, particularly at low DC power
transmission levels, is that of compatibility with the AC harmonic filtering requirements. If the AC harmonic
constraints are too great, either because the limits are onerous or because the AC system is poorly damped,
it may be necessary to have multiple filters energized at start-up, as shown in Fig. 2.3k.
The reactive power exchange with the AC system shown in Fig. 2.3k results in a low power factor at DC
power transmission levels below about 0.4 p.u. The power factor, cosf, is defined by:
0.9
0.8
0.7
Power factor (cos φ )
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
DC power (p.u.)
Fig. 2.3l– The power factor (cos f) of the converter scheme shown in Fig. 2.3k

CHAPTER
  Qac   [Eqn 2.3g]
cos φ = cos  tan −1 
Where:   Pac  
cosf = HVDC converter station power factor angle
Qac = net reactive power exchange between the HVDC station and the AC system
Pac = net real power exchange between the HVDC station and the AC system
Therefore, at low power transfer levels, the converter may be operating with a low power factor (Fig. 2.3l).
2.3.8. USING THE CONVERTER TO CONTROL REACTIVE POWER

In a HVDC scheme the DC power is defined as:
DC Power = DC Voltage × DC Current [Eqn 2.3h]
For a given DC power level the voltage can be reduced and the current proportionately increased while
achieving the same DC power transmission level, but at the expense of additional I2R transmission losses.
Therefore, if the number of filters energized to meet AC harmonic filter performance exceeds the reactive
power exchange limits, the converter operating conditions can be changed to increase the reactive power
Typical operating area
100
Target DC Voltage (%)
100
DC power (%)
Fig. 2.3m– Typical operating range of DC voltage on a back-to-back scheme
absorbed by the converter, in order to reduce the capacitive reactive power being exported to the AC
system from the HVDC converter station.
The change in DC conditions is achieved by lowering the DC voltage which requires the firing delay angle to be
increased. With an increase in DC current (to maintain the DC power constant) the overlap angle increases,
hence the reactive power absorbed by the converter increases [13]. Note that as the DC side of the converter
is common to both the rectifier and inverter, changing the DC conditions will change the reactive power load
at both rectifier and inverter.
In Fig. 2.3m, the upper limit is defined by the minimum allowable operating angles of the converter and

CHAPTER
the voltage insulation level of the equipment. The lower limit is defined by the maximum voltage transient
that can be applied to the converter, resulting from either the firing voltage of a rectifier or the recovery
voltage of an inverter.
Such a reactive power control methodology is common in back-to-back schemes where the transmission
distance is zero, thus transmission losses can be neglected. Back-to-back schemes also have the advantage
that the control systems for the two ends are in the same building and communication between the two is
guaranteed.
This control action is less common in transmission schemes where the additional cost of transmission losses
may well pay for extra converter station equipment to meet the reactive power requirements: for example,
shunt reactors. Nevertheless, when designing a large, bulk power, transmission scheme, consideration should
be given to the amount of time the HVDC link is expected to operate at low DC power transmission levels. If
the scheme will only operate at low DC power for short periods (during start-up and shut-down), then there
may be some cost benefits to utilizing the converter inherent reactive power absorption capability at the cost
of the increase in transmission losses.
BIBLIOGRAPHY
[1] B. M. Weedy, B. J. Cory, “Electrical Power Systems”, John Wiley & Sons, 1999, Chapter 1.
[2] B. R. Andersen, C. D. Barker, “The application of HVDC and the benefits that the development of voltage
sourced converters could bring”, CEPSI 2000, Manilla, Philippines, October 23-27 2000.
[3] N. M. MacLeod, C. D. Barker, D. R. Critchley. “HVDC Submarine cable inter-connectors: A review of the
Socio-economic and technical issues of existing schemes”, AORC-CIGRÉ Regional Conference, Jakarta,
Indonesia, April 2008.
[4] H. L. Thanawala, M. H. Baker, G. R. Moore, “Discussion on embedding of DC transmission in AC networks”,
CIGRÉ Symposium, London, 1999.
[5] C. Horwill, C. C. Davidson, “A Power-electronics-based transmission line de-icing system”, IEE 8th ACDC
Conference, 2006.
[6] E. W. Kimbark , “Direct Current Transmission Vol. I”, Wiley-Interscience, 1971, Chapter 3.
[7] R. P. Burgess, J. D. Ainsworth, H. L. Thanawala, M. Jain, R. S. Burton, “Voltage/Var Control at McNeill
Back-to-Back HVDC Converter Station”, CIGRÉ Paper 14-104, Paris, 1990.
[8] B. T. Barrett, N. M. MacLeod, S. Sud, R. S. Al-Nasser, A. I. Al-Mohaisen, “Planning and Design of the
Al-Fadhili 1800 MW HVDC inter-connector in Saudi-Arabia”, CIGRÉ Paper B4-113, Paris, 2008.
[9] P. L. Sorensen, B. Franzén, J. D. Wheeler, R. E. Bonchang, C. D. Barker, R. M. Preedy, M. H. Baker, “Konti-
Skan 1 HVDC Pole Replacement”, CIGRÉ Paper, B4-207, Paris, 2004.
[10] Y. H. Song, A. T. Johns, “Flexible AC transmission systems (FACTS)”, Published by The Institute of
Electrical Engineers, 1999, Chapter 4.
[11] S. Sadullah, N. M. MacLeod, “Enhancement of Power Transmission through HVDC & SVC Technology”,
IEEE Pakistan, 21st Multi-topic International Symposium, Lahore, April 2006.
[12] D. J. Hanson, “A Transmission SVC for National Grid Company Incorporating a +/- 75 MVAr STATCOM”,
IEE Colloquium on FACTS, 23 November 1998.
[13] J. D. Wheeler, J. L. Haddock, “Chandrapur Back to Back HVDC Scheme in India”, ICPST Conference in
Beijing China, October 1994.

CHAPTER
3| HVDC STATION DESIGN
HVDC Station Design
TOC DC Transmission Systems: Line Commutated Converters

3 HVDC
STATION DESIGN
Once the type of HVDC scheme has been defined, the next step
is to design the most appropriate station design: a back-to-back
arrangement, monopolar or bipolar HVDC transmission scheme.
The main HVDC station equipment is introduced as well as typical
arrangements for connecting a HVDC system to an AC grid via an AC
switchyard.
In terms of design layout, this chapter will show that one of the most
important design variables in the HVDC station is the converter
transformers arrangement. Depending on the power rating required
and the physical size constraints of transportation, converter
transformers may be single-phase or three-phase and have either
two or three (or occasionally, even four) windings. The transformer
arrangement impacts the design of the main HVDC ‘valve hall’ which
houses the high voltage thyristor valves.
Insulation coordination, highly important in any high voltage
substation, but vital in a HVDC station design, is specified in
this chapter.

Chapter contents
3. HVDC STATION DESIGN ................................. 110
3.1. HVDC CONVERTER STATION CONFIGURATIONS........ 113

3.1.1. Back-to-back HVDC Scheme................................................ 113
3.1.2. Transmission Schemes........................................................... 114
3.1.3. HVDC Schemes and Main Components.......................... 121
3.1.4. Special HVDC Schemes........................................................... 124
3.2. HVDC STATION INSULATION COORDINATION............. 127

3.2.1. System Overvoltage................................................................. 127
3.2.2. Equipment Impulse Withstand Level............................... 129
3.2.3. Insulation Creepage Distance.............................................. 131
3.2.4. Insulation Clearance Distance............................................. 131
3.3. TRANSIENT OVERVOLTAGE STUDIES................................ 134

3.3.1. Switching Type Studies........................................................... 135
3.3.2. Lightning Type Studies............................................................ 138
3.4. CONVERTER STATION LAYOUTS ........................................ 141

3.4.1. The Converter Island................................................................ 142
3.4.2. AC Harmonic Filters................................................................. 153
3.4.3. AC Switchyard............................................................................. 153
3.4.4. Acoustic Noise Considerations........................................... 154
3.4.5. General Layout Information.................................................. 156
BIBLIOGRAPHY ............................................................................................ 157

3.1. HVDC CONVERTER STATION CONFIGURATIONS
HVDC schemes can be divided into two basic types:
➙ back-to-back schemes, where both converter stations are at the same location
➙ transmission schemes (sometimes also known as point-to-point or end-to-end schemes), where the
two converter stations are at different locations and connected by DC lines and/or cables
In both cases, the essential building block is the converter which can be a 6-pulse bridge but today is almost
always a 12-pulse bridge, see section 2.2.5.
3.1.1. BACK-TO-BACK HVDC SCHEME

The back-to-back configuration is a common type of HVDC scheme. A back-to-back converter has no
transmission line and both converters are located at the same site. It is normally used to create an
asynchronous interconnection between two AC networks, which could have the same or different
frequencies.
In the back-to-back configuration, the valves for both converters are normally located in one building, known
as a valve hall. The control system, cooling equipment and auxiliary system may also be integrated into
configurations common to the two converter ends. DC filters are not required, nor are electrodes or electrode
lines and the neutral connection is made within the valve hall.
Because there is no DC conductor, the scheme can be optimized to minimize the cost and losses of the
converter equipment. As a result, back-to-back converters are generally designed to operate at high current
(3000-4000 A) and relatively low DC voltage (20-200 kV).
Fig. 3.1a and Fig. 3.1b show two different circuit configurations for back-to-back links, each with 12-pulse
bridge configurations for the converter valves at each end.
The main advantage of the mid-point ground scheme shown in Fig. 3.1b is that for the same 12-pulse
bridge DC voltage, the DC stress on the transformer is reduced by a half. Hence, the necessary insulation
clearances between transformer windings as a result of the DC test levels will reduce, therefore possibly
reducing the overall size and weight of the transformer.
Fig. 3.1a– Back-to-back HVDC scheme

CHAPTER 3.1a
Fig. 3.1b– Back-to-back HVDC scheme with mid-point ground
3.1.2. TRANSMISSION SCHEMES

3.1b
Transmission schemes are commonly used for bulk energy transmission over long distances, or for
submarine cable connections. The two converter stations are connected together via DC overhead lines, DC
cables or a combination of the two. The cooling system, auxiliary system and control system are separate
for the two converter stations. DC filters may be required where DC overhead lines are running in parallel
with open wire telephone circuits in order to limit the harmonic currents that flow through the DC lines.
There are two common configurations, the monopole and the bipole, along with various more specialized
configurations.
3.1.2.1. Monopolar HVDC Scheme

The monopolar configuration is the simplest arrangement of a HVDC transmission scheme, where each
end of the scheme has just one converter. It normally has one HV conductor and a return conductor. The
return conductor could be either a dedicated metallic conductor (metallic return) or an earth and/or sea
conductor (ground return). See section 8.3 for more detail.
• Monopolar HVDC Scheme with Ground Return
This arrangement consists of one or more (normally two) 6-pulse converter units in series on the DC side
of the converter and in parallel on the AC side at each end, a single conductor (a cable or an overhead line)
and return through the earth or sea as shown in Fig. 3.1c.
At each end of the line, this arrangement requires an electrode line and a ground or sea electrode built for
continuous operation. The need for a continuously rated electrode can be a drawback to selecting this type
of scheme for transmitting power. As a result of cathodic reactions, every coulomb that flows through the
cathode electrode will cause the cathode material to erode, which means continuous operation using the
earth as the current path requires a very large cathode electrode to be installed or frequent maintenance or
replacement of that electrode.
• Monopolar HVDC Scheme with Metallic Return
This arrangement usually consists of one high voltage and one low voltage conductor as shown in Fig. 3.1d,
but could consist of two conductors at positive and negative half voltage in an arrangement similar to that
shown for back-to-back schemes in Fig. 3.1b.
CHAPTER
Fig. 3.1c– Monopolar HVDC scheme with ground return
3.1c
Fig. 3.1d– Monopolar HVDC scheme with metallic return
This configuration is used as the first stage of a bipolar scheme, avoiding ground currents when construction
3.1d
of electrode lines and ground electrodes is environmentally unacceptable (see section 8.3). It is also
used when the necessary length of electrode line, or the value of the earth resistivity, would result in an
uneconomic solution.
3.1.2.2. Bipolar Configuration

A bipolar configuration contains two independent poles, each consisting of an independent converter
(normally 12-pulse). There are two conductors, one with positive and the other with negative polarity with
respect to ground for power flow in one (or both) directions.

CHAPTER
The bipolar configuration is a combination of two monopolar schemes as shown in Fig. 3.1e, arranged so
that the neutral return currents for the two poles partly or completely cancel each other out. Consequently,
with both poles in operation and transmitting power in the same direction, the current flow in the ground
path (which is equal to the imbalance between the two pole currents) can be controlled to a very low value.
In this configuration each pole can be operated independently, known as ‘separate control’, or in a master/
slave type configuration, known as ‘balanced control’. The only advantage separate control has over balanced
operation is that the control of power flow, or even (in principle) the power direction, can be selected for each
pole irrespective of what the settings are on the other pole.
This is a very common arrangement with the following operational capabilities:
1) For an outage of one converter, in the case where long-term ground current flow is undesirable, the
bipole system could be operated in monopolar metallic return mode, if appropriate DC arrangements
are provided as shown in Fig. 3.1f
2) Transfer of the current to the metallic return path and back to ground return without interruption requires
a Metallic Return Transfer Breaker (MRTB) and other special-purpose switchgear in the ground path of one
terminal. When a short interruption of power flow is permitted, such a breaker is not necessary
3) During an outage of one pole conductor, the other pole could be operated continuously with ground return
if sufficient electrode material is included in the design
Fig. 3.1e– Bipole HVDC scheme

3.1e

CHAPTER
4) During maintenance of ground electrodes or electrode lines, operation is possible with connection of
neutrals to the grounding grid of the terminals, with the imbalance current between the two poles held
to a very low value
5) When one pole cannot be operated with full load current, the two poles of the bipolar scheme could be
operated with different currents, as long as both ground electrodes are connected
6) In place of ground return, a third conductor can be added end-to-end. This conductor carries unbalanced
currents during bipolar operation and serves as the return path, when a pole is out of service
3.1.2.3. Series Bridge Configuration
MRTB
Fig. 3.1f– Bipole HVDC scheme with metallic return for pole outage
The3.1f
configurations described in the preceding sections use a single converter unit per pole. However, a
HVDC scheme can also be built with two or more 12-pulse bridges connected in series per pole, in either
monopolar or bipolar configuration. This series bridge HVDC configuration is generally used where higher
transmission voltages are required in order to transfer bulk power across long distances.
Fig. 3.1g, 3.1h and 3.1i respectively show a monopolar scheme with earth return, a monopolar scheme with
metallic return and a bipolar scheme, all using the series bridge configuration.

CHAPTER
Fig. 3.1g– Series bridge connected HVDC scheme with earth return
3.1g
Fig. 3.1h– Series bridge connected HVDC scheme with metallic return
3.1h

CHAPTER
Fig. 3.1i– Series bridge connected HVDC bipolar scheme

CHAPTER
CHAPTER
BACK TO
120 |
ACÊharmonicÊfilter DCÊreactor
DCÊharmonic
filter DCÊPoleÊ1
Valve HVÊline
connection
DCÊvoltage
measurement
Quadrivalve DCÊcurrent ERTB

HF measurement
AC
system Converter
poleÊbreaker
NBS
NBGS Electrode
connection
Converter MRTB
AC transformer
system NBS
HFÊfilter
DC Transmission Systems: Line Commutated Converters

HF
DCÊpoleÊ2
HVÊline
connection
3.1j
Fig. 3.1j– Typical Single Line Diagram (SLD) for a bipole HVDC converter station
3.1.3. HVDC SCHEMES AND MAIN COMPONENTS
The following section reviews the major components which make up a converter station.
3.1.3.1. AC Switchyard
The AC system connects to a HVDC converter station via a converter bus, which is simply the AC busbar to
which the converter is connected. The AC connection(s), the HVDC connection(s), along with connections to
AC harmonic filters and other possible loads such as auxiliary supply transformer, additional reactive power
equipment, etc. can be arranged in several ways normally dictated by reliability/redundancy requirements,
the number of separately switchable converters and utility practices in AC substation design.
Fig. 3.1j to Fig. 3.1o illustrate a selection of typical AC connection arrangements that are used in HVDC
converter stations.
Fig. 3.1j below shows a typical Single Line Diagram (SLD) of one end of a bipole HVDC converter station.
Fig. 3.1k shows a simple, single, 3-phase busbar with one switchable connection to the AC system and the
switchable AC harmonic filters connected directly to it. In such an arrangement, it is not possible to use the
AC harmonic filters for reactive power support of the AC system without having the converter energized
(as the AC system connection is common).
Fig. 3.1l shows a scheme consisting of two converters and includes an additional circuit breaker dedicated
to each converter. In this arrangement, the AC harmonic filters can be used for AC reactive power support
without energizing the converter. However, a busbar fault would
result in the complete outage of the converter station.
Converter
AC
system
AC
Converter
system
Filter
Filter
Filter Filter
Fig. 3.1k– Simple single busbar scheme Fig. 3.1l– Single busbar scheme with separate
converter breaker
Fig. 3.1m shows a double busbar arrangement

where 3.1k
an AC busbar outage will result in Converter
those loads connected to that busbar being
disconnected until the disconnectors can be 3.1l
AC
arranged to reconnect the load to the remaining system
healthy busbar. Disconnector re-arrangement
typically takes many seconds to complete and in
Filter
some circumstances, such an outage would not
be acceptable.
Fig. 3.1n shows a double busbar arrangement Filter
with an AC breaker, which could be used
where an outage for even a few seconds is not
acceptable, as each load is connected via a
Fig. 3.1m– Double busbar scheme
dedicated circuit breaker to each busbar, allowing
for fast disconnection and reconnection in the

CHAPTER
event of a loss of a busbar (typically around 300 ms). A disadvantage of this arrangement is that a large
number of AC circuit breakers are required.
Fig. 3.1o shows a double busbar arrangement with a reduced number of circuit breakers and where two
loads can be individually switched between two 3-phase busbars via three circuit breakers, hence this
configuration is commonly known as a ‘breaker-and-a-half’ arrangement.
Converter
AC AC
system Filter
system
Filter
Filter Converter
Filter
Fig. 3.1n– Double busbar, double breaker scheme Fig. 3.1o– Breaker and a half scheme
3.1.3.2. AC Harmonic Filters

As described in detail in the next chapter, converter operation results in both the generation of AC current
3.1o
harmonics and the absorption of reactive power. In order to mitigate these undesirable effects, the converter
3.1n station normally includes shunt-connected switchable AC harmonic filters either connected directly to the
converter busbar or connected to a filter busbar, which in turn is connected to the converter busbar.
The AC harmonic filters are automatically switched on and off with conventional AC circuit breakers to meet
harmonic performance and reactive power performance limits. The AC harmonic filters are typically composed
of a high voltage connected capacitor bank in series with a medium voltage circuit composed of air-cored
air-insulated reactors, resistors and capacitor banks. These components are selected to provide the required
performance from the AC harmonic filter and to ensure that the filter is adequately rated.
3.1.3.3. High Frequency Filter

The converter operation will result in the generation of very high frequency interference which will
propagate out into the AC system from the converter bus. While the magnitude and frequency of this
interference is often of no importance to the safe operation of the AC system, there are some instances
where this high frequency interference may be undesirable, in particular when the AC system uses Power
Line Carrier (PLC) signaling.
PLC signaling is a system which transmits a communication signal as an amplitude-modulated signal,
superimposed on the fundamental frequency voltage signal. This system is used, in some power systems, as
a communication system between AC system protection devices. However, the high frequency interference
generated by converter operation can overlap with the frequencies used for PLC communications (typically in
the range of 40 kHz to 500 kHz). Therefore, it is sometimes necessary to include a High Frequency (HF) filter
(also known as a PLC filter) in the connection between the converter bus and the converter in order to limit
the interference that can propagate into the AC system.

CHAPTER
As with the AC harmonic filter, the HF filter is a combination of a capacitor bank and an air-cored air-insulated
reactor, along with an additional low voltage circuit composed of capacitors, reactors and resistors, which is
referred to as a tuning pack.
3.1.3.4. Converter Transformer

The converter transformer is the interface between the AC system and the thyristor valves. It is one of the
most important and specialized items of equipment in the station.
Converter transformers differ from conventional AC transformers in a number of important respects, as
discussed further in section 6.4. Typically, the HVDC converter transformer is subjected to a DC voltage
insulation stress as well as the AC voltage stresses normally experienced by a power transformer. These
stresses are fundamentally different with the AC voltage stress being predominantly in the insulating oil
and being defined by the geometry and permittivity of the materials, while the DC stress is governed by
the resistivity of the insulating materials, which in turn, vary with operating conditions. In addition, it is
important that the converter transformer be thermally designed to take into consideration the AC harmonic
currents that will flow from the converter through the converter transformer to the AC harmonic filters.
Typically, the converter transformer is arranged as an earthed star line winding with a floating star and delta
secondary windings. There is normally an on-load tapchanger on the line winding of the converter transformer
and the tapchanger frequently requires a wide tap range to optimize the converter design.
3.1.3.5. Converter
The converter provides the transformation from AC to DC or DC to AC as required. The basic building block
of the converter is the 6-pulse bridge, however most HVDC converters are connected as 12-pulse bridges.
The 12-pulse bridge is composed of 12 valves (see section 6.2), each of which may contain many series
connected thyristors in order to achieve the DC rating of the HVDC scheme.
3.1.3.6. DC Smoothing Reactor

DC smoothing reactors are required for HVDC transmission schemes. They are also used on some back-to-
back schemes, but the need to include them on back-to-back schemes is less clear. None of the back-to-
back schemes supplied by GE Vernova or its predecessors since 1989 have used DC smoothing reactors.
For a HVDC transmission scheme the DC smoothing reactor provides a number of functions but principally
it is used to:
➙ Reduce the DC current ripple on the overhead transmission line or cable
➙ Reduce the maximum potential fault current that could flow from the DC transmission circuit into a
converter fault, or into the converter as a result of a DC line or cable fault
➙ Modify the DC side resonances of the scheme to frequencies that are not multiples of the fundamental
AC frequency
The DC smoothing reactor is normally a large air-cored air-insulated reactor (although oil-insulated, iron-
cored reactors can also be used) and is principally located at the high voltage terminal of the HVDC converter
for schemes rated at, or below, 500 kVdc. For HVDC converters operating at higher voltages it may be
economically preferable to split the DC reactance between the HV and LV terminals.
3.1.3.7. DC Filter
Converter operation results in voltage harmonics being generated at the DC terminals of the converter,
that is, there are sinusoidal AC harmonic components superimposed on the DC terminal voltage. This
AC harmonic component of voltage will result in AC harmonic current flow in the DC circuit and the
field generated by this AC harmonic current flow can link with adjacent conductors, such as open-wire
telecommunication systems, and induce harmonic current flow in these other circuits.
In a back-to-back scheme these harmonics are contained within the valve hall with adequate shielding and
with a cable scheme the cable screen typically provides adequate shielding.
However, with open wire (overhead line) DC transmission it may be necessary to provide DC filters to limit
the amount of harmonic current flowing in the DC line if this harmonic interference will be detrimental to

CHAPTER
other services such as open-wire telephone lines. The DC filter is physically similar to an AC filter in that it
is connected to the high voltage potential via a capacitor bank. Other capacitors, along with reactors and
resistors, are then connected in series with the high voltage capacitor bank in order to provide the desired
tuning and damping.
3.1.3.8. DC Switchgear
Switchgear on the DC side of the converter is typically limited to disconnectors and earth switches for
scheme reconfiguration and safe maintenance operation. Interruption of fault events is achieved by the
controlled action of the converter and therefore, with the exception of the Neutral Bus Switch (NBS), does
not require switchgear with current interruption capability.
Where more than one HVDC pole shares a common transmission conductor (typically the neutral), it is
advantageous to be able to commutate the DC current between transmission paths without interrupting
the DC power flow. As a result, several items of DC switchgear are typically required. The following switches
can be identified from Fig. 3.1j.
NBGS - Neutral Bus Ground Switch
This switch is normally open but when closed it solidly connects the converter neutral to the
station earth mat. Operation with this switch can normally be maintained if the converter can be
operated in a bipole mode with balanced currents between the poles, that is, the DC current to
earth is very small. The switch is also able to open, commutating a small DC unbalance current
out of the switch and into the DC circuit.
NBS - Neutral Bus Switch
A NBS is in series with the neutral connection of each pole. In the event of an earth fault on one
pole, that pole will be blocked. However, the pole remaining in service will continue to feed DC
current into the fault via the common neutral connection. The NBS is used to interrupt the fault
current flowing through the blocked pole to ground.
ERTB - Earth Return Transfer Breaker
The connection between the HVDC conductor and the neutral point includes both a high voltage
disconnector and an ERTB, which is used as part of the switching operation to configure the HVDC
scheme as either a bipole scheme or a metallic return monopole. The ERTB is sometimes also
referred to as a Ground Return Transfer Switch (GRTS).
The disconnector is maintained open if the HV conductor is energized in order to isolate the
medium voltage ERTB from the high voltage. The ERTB is closed, following the closing of the
disconnector in order to put the HV conductor in parallel with the earth path. The ERTB is also used
to commutate the load current from the HV conductor, transferring the path to the ground return
path. Once current flow through the HV conductor is detected as having stopped, the disconnector
can be opened allowing the HV conductor to be re-energized at high voltage.
MRTB - Metallic Return Transfer Breaker
The MRTB is used in conjunction with the ERTB to commutate the DC load current between the
ground return and a parallel, otherwise unused, HV conductor (metallic return). The MRTB closes
in order to put the low impedance earth return path in parallel with the much higher impedance
metallic return path. The MRTB must also be able to open causing current flowing through the
earth return to commutate into the much higher impedance metallic return path.
3.1.3.9. DC Transducer
DC connected transducers fall into two types, those measuring the DC voltage of the scheme and those
measuring the DC current. They are explained in section 6.6 of this book.
3.1.4. SPECIAL HVDC SCHEMES

A number of HVDC schemes have been constructed which have a different circuit arrangement to that
introduced in section 3.1.3. The following section introduces some of these alternative arrangements.

CHAPTER
3.1.4.1. HVDC Scheme with De-icing Mode
An unusual application of HVDC involves providing one converter station (the rectifier) which injects a high
direct current into an AC transmission line in order to melt the ice on it. Such a scheme generally requires
quite low DC voltage (typically 20 kV) but may require very high direct current. GE Vernova has built one
example of this arrangement, which is located at the Lévis substation, near Québec City and is covered in
section 1.2 of this book.
3.1.4.2. HVDC Scheme with 4-winding Transformer

Normally the AC harmonic filters are connected to the same high voltage AC bus as the converter itself,
using one of the AC switchyard arrangements shown in section 3.1.3.1 above.
However, the AC circuit breakers that connect the AC harmonic filters to the AC bus can be very costly for high
AC voltages (e.g. 400 kV). For relatively small HVDC converters, an alternative can be to equip the converter
transformer with a fourth winding at a relatively low voltage, with AC harmonic filters connected to this fourth
bus via vacuum switchgear. In such an arrangement, the converter transformer would have a single-phase or
3-phase, 4-winding arrangement.
Two such schemes have been supplied by GE Vernova: the McNeill back-to-back scheme in Canada and the
Rivera back-to-back scheme in Uruguay. These two projects, which used 3-phase, 4-winding transformers,
are covered in section 1.2 and section 1.3 of this book.
3.1.4.3. Capacitor Commutated Converter (CCC) HVDC Scheme

The CCC HVDC scheme is characterized by the use of capacitors in series with the converter transformer.
These capacitors are referred to as commutation capacitors and can either be located between the
converter transformer and the thyristor valves or between the converter transformer and the AC filters,
as shown in Fig. 3.1p and Fig. 3.1q respectively.
When the capacitors are connected between the transformer and AC filters the capacitors may include
parallel reactors and TCSC thyristor valves and are known as a Controlled Series Capacitor Converter (CSCC).
When using a CCC, the series capacitor generates reactive power proportional to the converter load current.
The polarity of the voltage generated across the commutation capacitor is such that the voltage experienced
by the thyristor valves is bigger than, and has a phase shift with regard to, the AC source voltage. The phase
shift is such that the reactive power absorbed by a CCC is always lower compared to a conventional converter.
DCÊreactor
ACÊsystem ElectrodeÊline
Fig. 3.1p– CCC HVDC scheme with commutation capacitors on the valve side

CHAPTER
DCÊreactor
ACÊsystem ElectrodeÊline
Fig. 3.1q– CCC HVDC scheme with commutation capacitors on the line side
3.1q
Once the DC conditions are fixed, the reactive power absorbed by the converter at any power level is
determined by the capacitance of the commutation capacitor. Therefore, the size of the commutation
capacitor can be chosen so that at full load the reactive power consumption of the converter is small and
can be compensated for by the reactive power generated by one small AC filter. Therefore, the filtering and
reactive power supply functions are nearly fully separated and the need for switchable shunt capacitor banks
for reactive power consumption is reduced.
Another advantage of CCC is that Temporary OverVoltages (TOV) are lower because the converter requires
much less shunt reactive power compensation and the load rejection TOV is mainly determined by the amount
of shunt reactive power connected.
CCC has also the following disadvantages:
➙ Whilst for conventional HVDC schemes the only series component which can impact the overall
availability and reliability of a back-to-back is the converter transformer, in a CCC there is the additional
capacitor bank which will directly impact the failure rate of the converter station
➙ The series capacitor bank is under DC stress and therefore, if mounted externally, is subjected to
additional pollution risk
➙ The additional exposed conductors associated with the capacitor bank will increase the RFI radiated
from the converter station
➙ In order to avoid the risk of pollution build-up on the capacitor insulation and increased RFI generation,
the capacitor can be located indoors. However, the additional civil works will have an associated
additional cost and as the capacitor will be oil-filled, special fire protection methods must be employed
➙ The series capacitor forces the converter to operate at higher firing angles thereby increasing the
harmonic current generated by the converter
➙ The large amount of series capacitive energy can cause severe damage to the converter valve in the
event of bushing flashover; hence a large earthing resistor must be incorporated into the converter
DC circuit

CHAPTER
3.2. HVDC STATION INSULATION COORDINATION
Insulation is provided by materials (for example air, oil, ceramic material, resins/oil-impregnated paper)
which upon application of voltage (within limits) across them, results in no, or insignificant flow of current.
Air is the most prevalent insulation associated with electrical systems.
Insulation breakdown can occur on application of voltage across the insulation with a magnitude greater
than the insulation can withstand. The insulating property of an insulation medium can deteriorate if there is
an accumulation of conductive impurities. The insulation properties are also affected by ambient conditions
(altitude, humidity, pollution, etc.).
Insulation coordination is the procedure of selection of insulation strength of equipment in relation to the
voltages which can appear in the system for which the equipment is intended, taking into account the service
environment and the characteristics of the available protective devices [3].
Based on the capability of the insulation to recover following a discharge event, insulation material can be
categorized as self-restoring or non self-restoring. Self-restoring insulation is provided by air and is external to
the electrical equipment. Air (external insulation) recovers integrity following a disruptive discharge. Non-self
restoring insulation is generally provided by materials (solid, liquid, gaseous, or a combination) encapsulated
within the equipment and protected from most atmospheric conditions. Damage and impurities resulting from
a discharge within non self-restoring insulation irrevocably deteriorates the insulating properties.
There are two methods that can be applied for insulation coordination and the determination of insulation
withstand requirements and/or capability of equipment: statistical and deterministic. These are described
in sections 3.2.2.1 and 3.2.2.2 respectively.
For the high voltage, AC grid connection of a HVDC scheme the insulation coordination is based on the
established standard voltage levels included in the IEC guidelines [3] or an equivalent national industry
standard. The insulation coordination of the HVDC scheme (which includes the valve winding connections
of the converter transformer) is performed using the deterministic methodology. The protective and
withstand levels established by the empirical process are further confirmed by a detailed transient
overvoltage study (see section 3.3).
Overvoltage protection is provided by surge arresters. A surge arrester is a non-linear resistance. The
resistance reduces with the increase in the applied voltage. In the event that the voltage at the connected
bus exceeds the arrester threshold voltage, the arrester begins to conduct significant current, holding
down the voltage at the bus it is connected to and thus protecting the nearby equipment. Surge arresters
are discussed in detail in section 6.7.
The typical outputs of an insulation coordination study include the following:
➙ Surge arrester protective levels (section 6.7) – rating and position
➙ Equipment insulation withstand levels (section 3.2.2)
➙ Insulation creepages (section 3.2.3)
➙ Air clearances (section 3.2.4)
The following section describes the sources of system overvoltage and categorizes overvoltages based
on parameter wave shape.
3.2.1. SYSTEM OVERVOLTAGE

Various phenomena that can result in electrical system overvoltage include:
➙ Natural phenomena such as atmospheric discharge (direct or indirect lightning strike)
➙ Switching circuit components
➙ Initiation or clearing of fault events within the system
System overvoltages are categorized based on the resulting wave shape. Temporary overvoltages are
generally sinusoidal (frequency between 10 Hz to 500 Hz). Transient overvoltages are characterized by a
triangular envelope with typically a fast rise time and a slower fall time (Fig. 3.2a).

CHAPTER
1.0
MagnitudeÊ(p.u.)
0.9
0.5
0.3
T1
Tp
T2
Time
Fig. 3.2a– Typical transient overvoltage envelope (wave shape)
Event Wave shape Duration

3.2a
Low frequency – Power frequency 50 or 60 Hz ≥ 3600 s
Slow – Temporary 10 to 500 Hz 0.03 s to 3600 s
5000 μs ≥ Tp ≥ 20 μs
T2 ≤ 20 ms
(Fig. 3.2a)
Slow front – Switching
IEC standard wave shape
Voltage 250/2500 μs
Current 36/90 μs
20 μs ≥ T1 ≥ 0.1 μs
T2 ≤ 300 μs
(Fig. 3.2a)
Fast front – Lightning
Voltage 1.2/50 μs
Current 8/20 μs
2 μs ≥ T1 ≥ 0.1 μs
T2 ≤ 5 μs
Fast front – Front of wave (Fig. 3.2a)
Current 1/2 μs
Table 3.2a– Categories of electrical system overvoltage (IEC 60071-1)

CHAPTER
3.2.2. EQUIPMENT IMPULSE WITHSTAND LEVEL
The insulation of equipment is characterized by its impulse withstand level. The required withstand level
for particular equipment based on its location in the circuit can be determined based on two techniques
introduced above: statistical and deterministic. These are further elaborated in the following subsections.
3.2.2.1. Statistical Method

The statistical method for the identification of the impulse withstand requirements or capability is based
on probability analysis, which relies on the availability of large volumes of experimental data.
Identification of impulse withstand requirements is based on the following inputs:
➙ Frequency of occurrence of overvoltage events of a specific magnitude and characteristic
➙ Permitted insulation failure rate
➙ Established discharge probability for specific overvoltage (characteristic and magnitude) and
configuration of insulation
Considerable experimental data exists with regards to behavior of an air gap on application of voltage.
Historically a large number of experimental studies have been performed. Based on this experimental data the
flashover probability of an air gap (with different configurations such as rod-rod, rod-plane, conductor-structure,
etc.) against voltage has been found to conform closely to the normal or Gaussian distribution. The statistical
method can therefore be applied to self-restoring insulation design.
Confirmation of withstand capability of equipment by tests based on a statistical methodology is applicable to
self-restoring insulation and involves the application of multiple impulse overvoltages of relevant magnitudes.
There are three classes of statistical withstand tests according to IEC guidelines [5] and the results are
analyzed accordingly, applying appropriate probability and likelihood functions.
3.2.2.2. Deterministic Method

The deterministic method is applied when limited experimental data is available for statistical analysis of
behavior of the insulation. This method is specifically applicable for non-self restoring insulation.
For a HVDC scheme where all key equipment is protected either directly or indirectly against overvoltages
by a surge arrester arrangement, only the deterministic method is applied for the determination and
confirmation of insulation withstand capability of the scheme and the associated equipment [4]. Fig. 3.2b
presents a typical surge arrester arrangement for a point-to-point HVDC scheme and Table 3.2a indicates
the typical stresses imposed on these arresters for different events.
The following aspects are of key importance for determining the equipment impulse withstand voltages:
➙ Maximum voltage of a particular characteristic (waveform) that appears across the equipment and is
limited by a surge arrester, or a combination of surge arresters, for a transient or temporary event
The surge arrester protective level is identified in terms of maximum voltage that may appear across
the arrester during a transient event and the corresponding arrester current also known as the surge
arrester coordinating current.
➙ Industry standard safety factors that cover: [4]
– Ageing effect on the insulation
– Practical arrangement of the equipment on site
– Site conditions compared to test facility conditions
Typical safety margins applied for HVDC schemes based on IEC guidelines are as follows: [4] and [6]
Switching impulse safety Lightning impulse safety Front of wave impulse
margin (SSM) margin (LSM) safety margin (FSM)
Thyristor valves 15% 15% 20%
Other DC equipment 20% 25% 25%
AC equipment 20% 25% 25%
In order to identify air clearance from the calculated withstand voltage, environmental correction factors have
to be applied.

CHAPTER
DR
HVDCÊline
V DL
B
V
COMPONENTS
FD
AUXILIARY
DCÊFILTER
AFB AT* V
M
V
EL2
CN LVDCÊline
*ÊIfÊrequired
EL
Fig. 3.2b– Typical surge arrester arrangement for a point-to-point HVDC scheme
Switching type and temporary

Lightning and steep front type
Contingency overvoltage
3.2b Current Energy Current Energy
Earth fault, DC pole EL, CN, FD EL, CN, FD DL, DR, EL EL
Lightning from
DL, FD, DR, EL
DC line
Switching surges from
DL, EL, FD
DC line
Lightning from
EL
electrode line
Earth fault on bridge AC
V DR, V, EL, M V, EL, M
phase
Current extinction
V V
3-pulse group
Current extinction
M M
6-pulse group
Loss of return path,
EL EL
monopole operation
Earth faults and
V, AT, EL, FD
switching operations on AFB AFB V, M, AT, AFB, EL, FD, DL
AC side
Lightning from
AT
AC system
Lightning shielding
V, M
failure
Table 3.2b– Types of stresses on arresters for different fault events (based on Table 4.2 of CIGRÉ Publication 34 [8])

CHAPTER
3.2.3. INSULATION CREEPAGE DISTANCE
Insulation creepage is the shortest distance between two conductors at different potentials along the
surface of an insulation material (see Fig. 3.2c). It is an empirical factor (generally defined in the units
mm/kV) of the power system fundamental frequency voltage. Application of power frequency voltage
across an insulator causes leakage current to flow along the surface of the insulator. Increased magnitude
of this current will increase the probability of insulation breakdown along the surface of the insulator.
Apart from the power frequency, or RMS value of
the steady-state voltage, experienced across the
insulator, other factors that affect the creepage
distance include:
➙ Deposits of conducting pollution on the
insulator surface (the environmental
pollution level) [7, 13]
➙ Characteristic of the insulating material
(hydrophobicity, etc.)
➙ Diameter of the insulator: In accordance
Clearance
Creepage
with industry standards, the creepage
distance is corrected to a higher value if the
diameter of the insulator is typically beyond
300 mm [7]. k di is the creepage distance
multiplication factor used where
kdi = 1 for average diameter Dm < 300 mm
= 1.1 for 300 < Dm < 500 mm
= 1.2 for Dm > 500 mm
➙ Positioning of the insulators (the build-up
of dirt on the insulation and partial washing
effect of the rain is a function of the position
Fig. 3.2c– Creepage and clearance distance
of the insulation: vertical, horizontal, etc.).
This affects the creepage distance and also
the definition of the maintenance cycle.
Industrial experience and a large number of studies have culminated in IEC 60815, the standards which
define the applicable creepage factors for AC installations.
3.2c
For outdoor installation, the creepage distance per kV of DC voltage in the same environment is generally
higher than that of line-to-line AC voltage.
3.2.4. INSULATION CLEARANCE DISTANCE

Insulation clearance is the shortest distance
Dry air density (kg/cubic m)
between two conductors at different potentials via 1.4

the insulating medium e.g. air, oil, etc. 1.2
The factors that influence the clearance are:
1
➙ P hysical arrangement of the conductors
and conducting surfaces (configuration of 0.8
the air gap - rod-rod, rod-plane, conductor-
0.6
structure, etc. [4])
➙ Ambient environmental conditions 0.4
For external insulation where air as the insulating 0.2
medium, ambient conditions such as altitude, air 0 2 x 103 4 x 103 6 x 103 8 x 103 1 x 104
density, temperature, etc. affect the clearance d
Altitude above sea level (m) i
distance. The dielectric strength of air is a
proportional function of air density. Air density is an
Fig. 3.2d– Air density as a function of altitude above
inverse proportional function of both altitude above sea level
sea level (Fig. 3.2d) and ambient temperature.

CHAPTER 3.2d
Dielectric Strength ∝ Air Density
1
Air Density ∝
Temperature
1
Dielectric Strength ∝
Temperature
Insulation clearance in air is defined in mm. The clearance distance in air is identified based on empirical
equations arrived at by spark gap testing. For AC side clearances up to an altitude of 2000 m above sea level,
industry standards provide minimum clearance requirements based on the system voltage level [4].
For the DC side of the HVDC scheme, the following empirical equations can be applied:
• Clearance calculation based on switching impulse withstand level
1.667
U 1 
dSW =  50 S ⋅  [Eqn. 3.2a]
 E f k ⋅ 500 
Where: k = a factor based on ELECTRA29 and IEC 60071-2
Ef = atmospheric correction factor
U50S = 50% flashovervoltage for switching impulse
• Clearance calculation based on lightning impulse withstand level
U 50 L 1
d LW = ⋅ [Eqn. 3.2b]
E f 575
Where: U50L = 50% flashovervoltage for lightning impulse
The 50% flashovervoltage can be calculated from the estimated withstand voltage as follows:
Uz
U 50 = [Eqn. 3.2c]
1− σZ
Where: s = per unit value of the conventional deviation of an overvoltage distribution
Where: s = 0.06 for switching
s = 0.03 for lightning
Z = conventional deviation of the discharge probability function of a self-restoring insulation.
Z = 2 to give a probability of withstand to flashover of 97.7%
Z = 3 to give a probability of withstand to flashover of 99.87%
Uz = required withstand level
• T he environmental correction factor (Ef) can be determined based on IEC 60060. These methods are
recommended by industrial standards to be applicable up to 2000 m above sea level:
Ef = kd • kh [Eqn. 3.2d]
Where: kd = air density correction factor
kh = humidity correction factor
The air density correction factor is given by:
m
 b 273 + t o 
kd =  ⋅  [Eqn. 3.2e]
 bo 273 + t 
Where: to = 20°C (standard reference temperature)
bo = 1.013 • 105 N/m2 (standard reference pressure)
t = maximum ambient temperature
b = normal barometric pressure
m = a parameter calculated in accordance with section 11.2.3 of IEC 60060-1.
The normal barometric pressure depends on the site altitude as shown in Fig. 3.2d obtained from IEC 60664.
Atmospheric pressure for altitudes not given in the table are calculated using curve fitting techniques.

CHAPTER
The humidity correction factor kh is given by:
kh = k1w
Where:
k1 = a parameter that can be calculated based on section 11.2.2 of IEC 60060-1.
w = an exponent that can be calculated based on section 11.2.3 of IEC 60060‑1.
Since the function is non-linear and the parameters ‘m’ and ‘w’ are related to relative humidity, the
calculation process is iterative. The base data associated with the calculation of factors ‘m’ and ‘w’ are
available and valid up to altitudes of 2000 m. Extrapolation of this data for altitudes greater than 2000 m
will introduce errors.
• The environmental correction factor can be calculated based on IEC 60071-2:
One of the assumptions of this method is that the effects of ambient temperature and humidity cancel each
other out, and therefore only the air density (i.e. altitude) related factor is applicable.
Ef = 1/Ka [Eqn. 3.2f]
Where:
Ka = altitude correction factor
The altitude correction factor for altitudes below 2000 m is given by:
 H 
m
 8150 
Ka = e [Eqn. 3.2g]
Where: m = 1 for lightning withstand voltage. m is found from Figure 9 of IEC60071-2 for switching
withstand voltage. m has a value between 0 and 1 and is an inverse function of the impulse
withstand voltage
H = altitude in meters above sea level
There is also a modified IEC 60071-2 method, which considers the reference altitude as 1000 m above sea
level and the altitude correction factor equation is modified to:
 H −1000 
m
 8150  [Eqn. 3.2h]
Ka = e
This modified method relies on extensive industry experience of installations up to 1000 m above sea level
involving equipment such as transformers, bushings, breakers, instrument transformers, etc.
There is a very limited experience in the area of high altitude high voltage electrical installations (altitude
greater than 2000 m). Some of the highest installations around the world include:
➙ Colorado USA (up to 3333 m) – Rocky mountain region (≤ 345 kVac)
➙ Peru (up to 4330 m) – Andes mountain region (≤ 230 kVac)
Owing to the empirical nature of calculation of the clearance and environmental correction factors, extension
of environmental correction factors for altitudes greater than 2000 m above sea level is limited by the
availability of up to date research (experimental data followed by statistical analysis) on the assessment of
the dielectric properties of air at high altitudes. The earliest assessment of the dielectric properties of air at
high altitudes was by F.W. Peek in the 1920s. Recent contributions to this area of study include Chicoutimi
University – Québec, Canada and Chongqing University - China. [14 – 21]
Some key considerations for selecting the appropriate method of calculation of environmental correction
factors are:
➙ The IEC 60060 method is substantially accurate for altitudes less than 2000 m, for ambient conditions
within the range identified by the standard
➙ The IEC 60071-2 method provides safe clearance definitions up to altitudes of 2000 m, as long as the
base withstand voltage is identified appropriately
➙ The modified IEC 60071-2 method can be applied on equipment with substantial operating experience
for altitudes up to 1000 m

CHAPTER
CREEPAGE CLEARANCE
DefiningÊevents PowerÊfrequency SlowÊandÊfastÊfront
VoltageÊriseÊtime
Precipitation
Contamination
AirÊdensityÊ(altitude)
Humidity
Fig. 3.2e– Relative influence of various factors on creepage and clearance
3.2e
• For altitudes greater than 2000 m, the IEC 60071-2 method may be applied. At present, research entities
and industry are working together on defining a calculation methodology of clearance for high voltage and
high altitude installations (with due consideration for non-linear behavior of long air gaps on application
of impulses) for different conductor configurations.
3.3. TRANSIENT OVERVOLTAGE STUDIES

In transient overvoltage studies, transient electrical perturbations are simulated using an electrical
simulation program that is appropriate for modeling transient effects, for example PSCAD/EMTDC from the
Manitoba HVDC Research Center. Such a program will carry out a time domain simulation of the system,
where the system state is evaluated at each of a sequence of discrete instants in time. Typically, these
instants are equally spaced in time and the time interval is specified by the program user and is referred to
as the time step1.
The results from these simulations are analyzed with the objective of demonstrating that overvoltages
caused by such events will not cause damage to equipment. The transient overvoltage study plays a part in
specifying both equipment insulation level requirements and also the protective levels of surge arresters and
their energy absorption requirements.
There are two categories of system representation used in these studies: the name given to each of them
indicates the type of event that it is designed to examine. These categories are distinguished from each
other by the time scale (section 3.2.1) associated with the transient event studied. The two categories
are the following.
• Switching type: these have the longest time scale of the three, being typically in the range 0.1 to
1 second. Unlike the other two categories, these model the normal operation of the system including
the AC nature of the system voltage, valve firing controls, and the electromagnetic coupling of the
transformers. Their main purpose is to assess surge arrester energy absorption requirements. Transient
events studied in this way include routine energization events, faults2 on incoming transmission lines
1 - It is noted that this time domain solution method contrasts with the frequency domain solution method employed for the load flow
studies of section 5.11.1; the frequency domain solution method does not simulate transient states accurately, but is a much more
suitable tool for load flow studies.
2 - In the context of switching transient event studies, a fault is a short circuit; this may be between a conductor and ground, or, in the case
of an ungrounded fault, between two or more conductors. In the latter case, all three AC phase conductors may be shorted together to
form a 3-phase fault. In practice, a fault may be caused by a simple metallic short-circuit resulting from a mechanical malfunction, or it may
be caused by the formation of an arc of ionized air that provides a low resistance path between conductors – such arc formation is often
referred to as a flashover. Within a switching type simulation, the occurrence of any fault may be modeled as an instantaneous resistance
collapse.

CHAPTER
followed by their clearance, and faults within the valve hall. Circuit breaker Transient Recovery Voltage
(TRV) and Rate of Rise of Recovery Voltage (RRRV) capability requirements are also assessed by using
switching type simulations. Circuit breakers and their requirements are discussed in section 6.8. For
switching type studies, the time step employed in the simulations is typically around 50 μs, which is
approximately 1° electrically at the system frequency.
• Lightning type: these have an intermediate time scale of typically 100 μs. These simulations model
such phenomena as stray capacitances and busbar inductances because these phenomena have a
significant effect at the associated high frequencies involved. Electromagnetic models are not used for
the transformers. The time scale is too short for it to be necessary to model the normal operation of the
system. Instead, the pre-event state is a snapshot point on the AC cycle. Their main purpose is to assess
peak overvoltages. AC filter flashover events belong to this category. Transient events studied in this way
also include direct lightning strikes at a point in the system and also back flashovers after a lightning surge
has struck the guard wire of a transmission tower; the lightning strike itself is modeled as a current surge.
Direct lightning strike on equipment within the HVDC substation can only occur as a result of shielding
failure. The lightning shield for the substation is generally designed to limit the peak current of surges that
can penetrate the shielding tower when compared with the station equipment design. Filter bus flashovers
are more onerous than lightning strikes of this magnitude, both for assessing filter equipment voltage
withstand levels and for assessing surge arrester protective levels and energy absorption requirements.
For lightning type studies, the time step employed in the simulations is typically around 0.01 μs, in order
to model fast transients accurately.
There is a third category of transient overvoltage study, which is the fast front category. Studies in this category
simulate the effect of even higher frequencies than for lightning type studies; nonetheless, it involves similar
concepts. Fast front studies are employed to simulate the fast transient effects that are experienced by a
converter valve when a flashover occurs in its vicinity. In this situation, the only stored charges that are present
locally are those associated with stray capacitances, and so voltage tends to change much more rapidly than
in the case of a filter flashover. A fast front of wave overvoltage event is oscillatory in nature with a dominant
peak, sufficient to cause a single impulse current discharge in the arrester. The duration of the impulse is too
short, in order to be a definitive parameter for the surge arrester energy. Empirical calculations (based on stray
capacitance and inductance) can be utilized to confirm the peak surge arrester current.
3.3.1. SWITCHING TYPE STUDIES

The extent of the system that is usually modeled is illustrated in Fig.3.3a for one side of a HVDC link (the
other side is also modeled this way). This example shows one side of a HVDC line scheme that is fed by
two separate incoming transmission lines. Although the diagram is shown in single line form for clarity,
the model is fully three phase and simulates the normal operation of the system prior to the application
of a transient event. The frequencies of the voltage and current waveforms associated with these types of
event are such that it is not necessary to model the stray inductance or capacitance of busbars or other
electrical connectors: these are modeled as simple connections.
3.3.1.1. AC Sources
These may be modeled by an ideal voltage source, together with an RL impedance using the topology in
Fig. 3.3b.
All three single-phase representations are identical to one another, except of course for the sinusoidal voltage
waveforms of the idealized voltage sources being mutually 120° out of phase.
The formulae for calculating the values of the three components of the impedance are the following:
V2 ω LP
➙ LP = in henries ➙ Rp = wLpQ in ohms ➙ RS = in ohms
ω SCL Q
Where: w = 2pf, where f is the system frequency in Hz
Q is the chosen quality factor for the supply
V is the RMS line-to-line system voltage in kV
SCL is the chosen 3-phase short-circuit level (SCL) of the source in MVA.
In the model, if the converter and its associated filters are out of service, the SCL at the converter’s local

CHAPTER
Local
Remote switchyard LineÊside ValveÊside
substationÊ#1
ACÊsource
Ê#1
Line
breakers
DC
smoothing
Converter reactor
Line doubleÊcircuit breakers
breakers overheadÊlineÊ#1 HVDCÊline
PLC
filter 12
Filter pulse
bridge DC ToÊother
breakers filter sideÊofÊlink
Remote converter
substationÊ#2 Filter
main LowÊvoltage
ACÊsource capacitors DC
Ê#2 neutral lineÊforÊDC
Line filter returnÊpath
breakers Converter
AC AC transformers
filter filter
#1 #2
Line doubleÊcircuit
breakers overheadÊlineÊ#2
Fig. 3.3a– System model for switching type transient overvoltage studies
PhaseÊAÊ
ofÊidealisedÊ PhaseÊA
Rp
3-phaseÊAC connectionÊto
voltageÊsource systemÊnetwork
Rs
3.3a Lp
PhaseÊBÊ
ofÊidealisedÊ
Rp PhaseÊB
3-phaseÊAC
connectionÊto
voltageÊsource
Rs systemÊnetwork
Lp
PhaseÊCÊ
ofÊidealisedÊ PhaseÊC
Rp
3-phaseÊAC connectionÊto
voltageÊsource systemÊnetwork
Rs
Lp
Fig. 3.3b– AC source topology for switching studies
switchyard is the net effect of the AC sources and the impedance of the transmission lines that connect these
sources to the converter station. The converter station equipment is designed for operation with the actual
system SCL being between specified minimum and maximum values. Note that fault studies are normally
carried out at minimum SCL, as these give the highest overvoltages. However, routine energization studies
3.3b
are often carried out at maximum SCL, since these produce higher inrush currents into the equipment items
that are being energized.

CHAPTER
3.3.1.2. Transmission Lines and Cables
The input to the model for solution by a transient analysis program, such as PSCAD/EMTDC, is typically the
physical, geometric data of the transmission lines and their position on the towers. The program’s internal
routines will calculate the electrical properties from these. Cables also are modeled in a similar way: the
input is the physical, geometric data of the cables themselves and of the medium in which they are buried.
(An analysis of the actual modeling of transmission lines and cables is beyond the scope of this book.)
3.3.1.3. Filters
These are represented by their R, L and C component values, using the topology of the actual filter. Any filter
surge arresters are normally included in this representation. In the case of the PLC filter, the representation
may be simplified by including only the principal components.
3.3.1.4. Surge Arresters

These devices are described in detail in section 6.7. In a model for a switching type study, the arrester’s
switching characteristic (i.e. the relationship between the voltage across it and the current passing through
it) may be represented as a piecewise linear function, with the input to the transient analysis program being
a table of voltage and current coordinate pairs.
3.3.1.5. Converter Transformers

The switching type transient overvoltage category is the only transient overvoltage category for which
the ferromagnetic coupling between the line and valve windings is modeled. A transient analysis program
such as PSCAD/EMTDC will have its own library of transformer models, which include the simulation of
saturation effects and leakage inductance. Each transformer in the system representation is modeled using
the appropriate library model. (An analysis of the actual modeling of transformers is beyond the scope of
this book.) For transformer energization studies, it may be necessary to model the presence of remanence
in the transformer core. The reader is referred to the documentation of the particular transient analysis
program employed for details of the implementation of this.
3.3.1.6. Converters
In switching studies, every converter that is represented is modeled by its arrangement of individual
thyristor valves and surge arresters. The converter model includes a detailed control system for the
valves’ firing control actions. Each valve may be represented by using a model of a thyristor with snubber
components: a single thyristor model represents the series arrangement of actual thyristors in a physical
valve, and the snubber component values in the model are aggregated values to represent the combined
effect of these components in the physical valve.
3.3.1.7. Multiple Run Studies

For studies related to switching type events in order to obtain results close to maximum value the instance
of the transient event is required to be varied. A particular instance of transient can be referred to as a
run. There is not a fixed rule for the size of the time increment between runs; the time increment is a
compromise arrived at between (i) obtaining results that are very close to their overall maximum values
that would be obtained if the increment were infinitely fine and (ii) the practical consideration of avoiding
long run times.
A switching study of this type consists of two stages.
1. The initialization stage: the simulation is run from a cold start – where all voltages are zero and no currents
are flowing – to reach the required pre-event steady state operating condition. A total elapsed time of
typically 3 seconds from the cold start is usually adequate to achieve this.
2. T he event stage: the sequence of multiple event runs is performed, each one using as its starting point the
system state at the end of stage 1 of the study. In each run the event can be initiated at selecting equally
spaced points over a fundamental frequency cycle. A total elapsed time of typically 1 second from the time
of the last event in the chain of events simulated is usually adequate for the transient effects caused by
this chain of events to die down again.

CHAPTER
3.3.2. LIGHTNING TYPE STUDIES
While switching type simulations involve frequencies no higher than typically 3 to 4 kHz, require a study
duration of the order of 1 second, and do not model high frequency effects, lightning type simulations
involve high frequency effects with these frequencies being typically greater than 10 kHz. The simulations
include high frequency modeling of connection length inductance, skin effect resistance, and stray
capacitance. The values of these parameters in the actual system are a function of the geometry not only
of the connection length itself but also of the other conducting material in its vicinity: these parameters are
very complicated to evaluate accurately. Furthermore, transient overvoltage studies are often performed
before the converter station layout has been finalized. Therefore, a practical, approximate approach may
be adopted instead, by modeling the connection length as a T-section using discrete components.
Although lightning type studies tend to produce higher peak voltages and higher peak currents through the
surge arresters than switching type studies, the surge arrester energy absorptions tend to be lower because
of the shortness of the duration of the overvoltage. However, it may be that none of the overvoltages across
particular arresters arising during switching type simulations are sufficient to cause conduction: for surge
arresters it will be the lightning type simulations that give rise to the overall maximum energy absorption in
the transient overvoltage study.
DCÊsourceÊtoÊprecharge InductiveÊsource FilterÊbusÊflashover,

theÊfilterÊtoÊtheÊfilterÊbus impedance modelledÊbyÊtheÊcollapse
arresterÊSIPL correspondingÊtoÊthe ofÊtheÊresistanceÊacross
minimumÊsystemÊSCL theÊearthÊswitchÊgap
101.1ÊmH 5Êm
6Êm
4Êm
CapacitorÊC1 4Êm CapacitorÊC1
lowerÊpart upperÊpart
withÊassociatedÊ 6.56ÊµF withÊassociatedÊ 6.56ÊµF
internal CT internal
connection point connection
lengths 4Êm lengths 4Êm
3Êm
3Êm 3Êm 4Êm
Filter Filter
surge L1 R1
surge
arrester 15ÊmH 404 Ω
arrester
#1 #2
3Êm 4Êm
4Êm
4Êm 4Êm
C2 L2 R2
117ÊµF 0.42ÊmH 404 Ω
4Êm 4Êm
18Êm
Fig. 3.3c– Example system model for a lightning type study

CHAPTER
Unlike the system representation for switching impulse simulations (Fig. 3.3a), which includes all installed
equipment within the converter station, the representation used for a lightning type simulation needs to
model only the part of the system that is local to the transient event: this being the only part of the system
that is strongly affected by such high frequency transients. To model a greater extent of the system would
make the studies unduly time-consuming to set up and run, without having a significant effect on the results.
Separate system models are thus used for the simulation of different transient events. An example of a
system model for a lightning type study is shown in Fig. 3.3c. This system model is in fact for a filter flashover
study. Each connection length shown on this diagram will be modeled by a high frequency T-section of the
length indicated.
A lightning type transient event only takes place over a few degrees of a cycle of the system voltage. Therefore,
it is not necessary to simulate the normal operation of the system, which involves the use of sinusoidally
varying source voltages and also the use of valve firing controls. Instead, before the application of the transient
event, the system is energized using idealized DC sources in order to simulate a particular system state.
In the case of the modeling of lightning type events on the AC part of a system, only one of the three phases
is represented in the model. (The exception to this is the representation of any overhead transmission lines
within a model; for these all three phases may be represented if the simulation program’s internal algorithms
simulate the electromagnetic coupling between the phase conductors of a transmission line. These are
modeled in the same way as for switching type studies, described in the previous section.) The three phases
of items of AC equipment such as filters are sufficiently well separated physically that any electromagnetic
coupling between phases during the transient event itself can be neglected, and the time duration of the
simulation is too short for secondary phase imbalance effects to manifest themselves.
As with switching type simulations described in the previous subsection, a lightning type simulation consists
of two stages, the initialization stage and the event phase. Unlike switching type simulations however, point-
on-wave runs are not carried out as such.
The time step between calculations of the system state in the simulations is typically around 0.01 μs. This is
5000 times smaller than the time step for switching type simulations.
Other component parts for lightning type studies are listed in the following sections.
3.3.2.1. Surge Arresters

These are modeled in a similar way to those in switching type studies as described in section 3.3.1, but a
lightning characteristic is used (described further in section 6.7). Stray capacitances across the arrester
and to ground may also be modeled if values are available. Each surge arrester’s maximum tolerance
characteristic is used in order to establish the maximum overvoltages experienced. If the energy dissipation
in any arrester for a particular simulation is found to be close to the corresponding specified limit, this
simulation may be repeated using the minimum tolerance characteristic for that arrester, in order to predict
its overall maximum energy dissipation.
3.3.2.2. Transformers
For the high frequencies associated with a lightning surge, the stray capacitance effects dominate over
the electromagnetic coupling between the windings, so the electromagnetic model of the transformer
as employed in the switching type studies is never used in lightning type studies. Where a transformer is
represented in a lightning type simulation, it is often sufficient to represent it by its bushing capacitance
to ground. Additionally, HVDC converter transformers usually have a center entry line side winding. For
this design of transformer, the magnitude of the capacitive-transferred surge at the terminals of the valve
side tends to be insignificant, and so the transformer and the equipment on the DC converter side of the
transformers is not modeled.
3.3.2.3. Lightning Surge Representation

A lightning strike is normally simulated by the injection of a pulse of current from an idealized current source.
In the model, the current surge may have the following waveform (Fig. 3.3d). The current in the pulse firstly
commences from zero and rises linearly with time to its specified peak value in its specified rise time. It then
falls back to zero, again linearly with time; the rate of fall follows from the specified time interval after the
CHAPTER
current
Ipeak
Ipeak / 2
0
t1 time
t2
Fig. 3.3d– Current surge waveform used for lightning strike in simulations
pulse commencement at which the current falls back to half its peak value.
For systems protected against lightning strikes by guard wires, direct strikes only occur as the result of shielding
failure. The peak amplitude of strikes for which this can happen has an upper limit defined by the shield design.
In the event the surge arresters are coordinated for discharge currents greater than shield failure current (as
is often the case) and equipment withstand levels are appropriately coordinated, direct lightning type study
for arrester rating and equipment insulation rating is superfluous. Assessment of distance effect, i.e. distance
3.2i from the equipment is beyond the scope of this book. Empirical assessment of distance
of the surge arrester
effect is included in [4, 12]. For the simulation of lightning strikes whose peak current is 20 kAp or less, a rise
time of 3 μs is used and the time to fall to half peak value is 77 μs. These values follow from a statistical analysis
of lightning strikes [9].
In the case of a tower back flashover event, this commences with a lightning strike to the top of a transmission
tower. For a simulation intended to predict the worst-case overvoltages and surge arrester energies, the
maximum realistically possible value of peak current must be used. The IEEE statistical analysis [9] indicates
that the peak current is always below 200 kApk, therefore 200 kApk is a suitable, conservative value to use in
simulations as the peak amplitude if there is no justification for the use of a lower value in a particular project,
such as a specific peak value supplied by the customer. The rise time is normally taken as 8 μs, and the time
to fall to half peak value taken as 20 μs.
3.3.2.4. Tower Back Flashover Studies

For the tower back flashover studies, the flashover commences when the voltage across the insulator
reaches the breakdown voltage. The following procedure for determining the breakdown voltage taken
from IEEE studies [9] may be employed, since the breakdown voltage is a function of the time evolution of
the voltage across the insulator. The formula is:
K2
Vbd = K1 + [Eqn. 3.3a]
t 0.75
Where:
Vbd = the breakdown voltage, in kilovolts
K1 = 400 L
K2 = 710 L
L = the insulator length, in meters
t = t he elapsed time after lightning strike hits the shielding wire at the top of the tower, in microseconds
3.3.2.5. Flashover Arc

The representation of the flashover arc path by a direct short circuit is unduly onerous and unrealistic.
CHAPTER
The flashover path can be modeled as a non-linear resistor. A resistance value of 106 Ω pre-flashover is
sufficient for the conduction to be negligibly small during this stage of the simulation and a value of below
1 Ω is sufficiently low to simulate the arc path when the arc is fully established.
3.3.2.6. Circuit Breakers and Disconnectors in their Open State

Each of these is represented by a stray capacitance to earth. If manufacturer data is not available for
the actual devices to be installed in the system, suitable approximations can be made from the table
in section 5.14 of the PSCAD/EMTDC V3 Guide [10]. For a 400 kV system, the recommended value for a
disconnector switch is 200 pF and for a dead tank circuit breaker 150 pF. This table also gives values for (i)
a 115 kV system and (ii) a 765 kV system.
Control BipoleÊ1
BipoleÊ2 building
ACÊfilters
ACÊfilters
CableÊroute
SVCÊ2 GIS SVCÊ1
Fig. 3.4a– A 2000 MW HVDC converter terminal
3.4. CONVERTER STATION LAYOUTS

The actual site size and layout of a HVDC converter station is dependent on many factors ranging from the
rating of the scheme and the reactive power limits, both of which impact the number of filters, to the cost of
3.4a
land. Fig. 3.4a shows the station associated with one end of a 2 × 1000 MW double bipole interconnection
connected at 400 kVac. The overall site area for this example is a little less than 0.1km2 and this facility
includes two static VAr compensator systems. This compactness was mainly achieved through the use of
GIS AC connections.
There are three basic zones within a HVDC converter station:
➙ The converter island
➙ The AC harmonic filters
➙ The AC switchyard
Each of these zones contains equipment associated with a particular function. However, depending on the
site topology, the HVDC scheme (SLD), location of electrical connections (both AC and DC), the location of
existing equipment and environmental considerations, these zones often overlap in the overall site layout.

CHAPTER
3.4.1. THE CONVERTER ISLAND
The converter island is that part of the station layout which consists of all of the equipment associated
with DC power conversion and transmission. The converter island zone consists of the most varied amount
of equipment and buildings, but normally consists of:
➙ Converter transformer(s)
➙ Valve hall(s)
➙ DC area
➙ Service (control) building
➙ Auxiliary services
1 x 3 ph/3 wdg 3 x 1 ph/3 wdg
2 x 3 ph/2 wdg
6 x 1 ph/2 wdg
Fig. 3.4b– Converter transformer arrangements
3.4.1.1. The Converter Transformer(s)

A description of converter transformers along with the various winding configurations is given in
section 6.4. The 3-phase transformer bank can be made up of either single-phase units or 3-phase units.
The conventional arrangement of a star and a delta winding on the converter side of the transformer
can be achieved either by having the line side and both converter side windings in one tank or by having
separate tanks for the converter star and the converter delta windings. This gives a total of four possible
3.4b for a 12-pulse converter bridge as shown in Fig. 3.4b.
configurations
The selection of converter transformer arrangement is driven by two factors: size constraints (based on
whether the transformer can practically be manufactured and shipped to site) and spares requirements (if it
is required to provide one spare transformer of each type). The lowest cost arrangement for the transformer
will invariably be a 3-phase, 3-winding arrangement. However, such a transformer will also be physically the
largest design option and therefore shipping limitations may make it impossible to get the transformer to site.
In addition, the cost of providing a spare transformer (where required) could be prohibitive.

CHAPTER
If a 3-phase, 3-winding transformer cannot be shipped to site then the next alternative is to split the
transformer into two 3-phase, 2-winding units, one connected star-star and the other star/delta. With
this arrangement two spare transformers are needed since the transformers are of two different designs,
once again making the cost of spare transformers very high. By using single-phase, 3-winding converter
transformers the total 3-phase cost will be higher, but since all transformers are identical only one spare
would be needed. Hence, compared with two spare units when 3-phase, 2-winding transformers are used,
then this option will be at a lower cost. However, single-phase transformers will have an impact on the valve
hall layout as the star and delta connections now have to be made by busbar connections between the
transformers, as indicated in Fig. 3.4b and therefore require additional space and clearance considerations.
Where shipping constraints are significant, single-phase, 2-winding converter transformers can be used but
this will increase the space occupied by the converter transformers considerably.
For some converters, it has been preferable to locate the converter transformer(s) independently of the
valve hall, so that the connections between the converter transformer(s) and the valve hall are exposed
to the outdoor environment. This solution was commonly used in the past with mercury arc valves, which
used an outdoor damping circuit situated between the transformer and the valves. More recently, such an
arrangement can be cost effective where the project delivery schedule is short, as the civil structures on
which the converter transformer(s) sit, along with the oil catchment system, can be designed and constructed
physically separated from the valve hall.
However, such an arrangement requires the converter transformer valve-side bushings to be designed
with an extended creepage to account for the outdoor pollution, and additional through-wall bushings are
required to connect from the transformer(s) into the valve hall. More significantly, the connecting busbar
between the converter transformer and the valve hall is in air and hence the busbars will radiate radio
frequency interference. The magnitude of this radio frequency interference is proportional to the valve
winding voltage (refer to section 4.8) therefore, except for those HVDC schemes with relatively low valve
winding voltages and where the radio frequency interference limits are not too onerous, it is almost always
the case that the converter transformer bushings are designed to protrude into the valve hall. In this way all
of the connections between the converter transformers and the valves are within the valve hall enclosure
and hence within the Faraday cage (refer to Fig. 3.4d).
6ÊXÊ1ÊPH 3ÊXÊ1ÊPH 2ÊXÊ3ÊPH 1ÊXÊ3ÊPH

2ÊWDG 3ÊWDG 2ÊWDG 3ÊWDG
S
S + S
S
D +
D
D
S
+ VALVE
D D
HALL
S
S
+
D
D
Fig. 3.4c– Transformer site arrangements
BACK TO 3.3c DC Transmission Systems: Line Commutated Converters | 143

CHAPTER
Fig. 3.4c indicates the influence of the converter transformer arrangement on the size of the valve hall,
assuming that the converter transformer valve-side bushings must protrude into the valve hall.
The converter transformer, for most HVDC schemes, will be the heaviest single load that must be brought onto
the site. Hence, the route that the transformers will follow must be designed for this weight. Consideration
must also be given to the method of moving transformers on site, particularly, if spare transformers are
included. The two basic methods of moving the transformers on site are either by rail track or by ‘skidding’.
Where rail track is employed the transformers can be jacked-up and put on wheels that sit in rail tracks. Fixed
pulling points around the rail tracks can then be used to winch a transformer from one location to another.
Where there is a change of direction, that is going from forward/backward to left/right, the transformer is
normally jacked up and the wheels re-orientated and located into the rail tracks going in the new direction.
The alternative method of moving the transformer simply relies on sliding the transformer across a prepared
surface, for example across temporary steel beams with low friction mats between the transformer base
and the beams, or on greased plates. Again, the transformer is jacked up and either temporary steel beams
or plates are placed underneath it. Either pulling points strategically located around the transformer
transportation area or mechanical rams can be used to ‘slide’ the transformer. Where the cooling equipment
is located separately from the converter transformer tank the coolers can be moved by crane.
For some applications it is a requirement to have a spare transformer ready for rapid deployment in the event
of an in-service transformer failure. In such circumstances the spare is normally kept on a separate pad, full of
oil and with the line and valve side bushings in position. As the valve-side bushings are designed for in-door
use these are normally covered by some protective wrapping. It is important when arranging the layout to
ensure that the failed transformer can be removed and the replacement transformer put into its new location
with the maximum efficiency. It is necessary to ensure that the transformers do not have to cross during the
change-over and that there are no physical obstacles in the way, for example there must be sufficient space
to get the converter transformer bushings past any ‘fire walls’ associated with the converter transformers.
In order to stop the spread of fire from one transformer unit to another it is normal to build firewalls between
the transformer units, typically with a two-hour fire rating. Sufficient space must be made between the
converter transformer tank and the firewalls for access. When the cooler banks are mounted directly onto
the transformer tank (normally done when rapid replacement with a spare is demanded), space must also be
ensured for sufficient air circulation to allow the coolers to work efficiently.
The converter transformer tank contains oil, which in the event of a tank rupture, must be contained to
stop it contaminating the water table. The simplest method of containment is to built a ‘bund’ wall around
the transformer. However, considering the thousands of liters of oil normally contained within a converter
transformer tank, this arrangement would occupy a large area and hence is not commonly used. The more
common method of oil containment is to have an oil catchment pit under each transformer unit. The
catchment pits associated with several transformers then feed into an oil separator tank where the oil and
any water can be removed separately for disposal. The catchment pit must be adequately sized for not only
the oil content of a transformer, but also any water that may enter the catchment tank as a result of a water
‘deluge’ fire protection system or rain water.
3.4.1.2. Valve Halls

The valve halls are the buildings which contain the thyristor valves. Some early HVDC schemes utilized
outdoor valves, typically with the thyristor valve contained within a steel tank and insulated with oil, similar
to the converter transformer. However, maintenance required moving the tank to a clean, controlled
environment followed by the removal of the oil and then the reverse to put the valve back into service. This
is a highly labor-intensive activity and hence almost all modern HVDC schemes use indoor thyristor valves
where the valve is air insulated with a building protecting the thyristor valves, interconnecting buswork
and other equipment from the outside environment.
One of the major functions of the valve hall building is to minimize the pollution within the valve hall. Cooling
air is required as the thyristors, buswork, etc., will generate some heat into the valve hall air even though the
majority of the heat generated by the thyristor valves is removed by the water or water/glycol cooling circuit.
The air circulates around the valve hall, with additional air pulled in from outside for cooling. This outdoor air
is passed through filters in order to remove significant amounts of pollution before it enters into the valve

CHAPTER
hall. The air within the valve hall is also maintained at a slightly positive pressure with respect to atmospheric
pressure so that pollution is not drawn into the valve hall through any small gaps in the building fabric. As a
result of these clean conditions within the valve hall, it is possible to use specific creepages as low as 14 mm/
kV for equipment in the valve hall, irrespective of the outdoor environment.
The other important function of the valve hall is that of providing a radio frequency interference shield, also
known as a Faraday cage around the electrical equipment.
As explained in section 4.8, during every cycle the 12 valves comprising a 12-pulse bridge will turn-on and
turn-off. The turn-off is a relatively slow process but the turn-on, that is, the step change in voltage across
the valve at the start of conduction, is very fast. This fast change in voltage will generate radio frequency
interference which must be contained. The valve hall structure therefore includes a continuous electrostatic
shield to limit the radio interference outside of the valve hall. This screen can be in the form of a bonded
wire mesh across the walls, floor and roof, or it can be formed, at least for the walls and roof, using steel
Fig. 3.4d– Bushing turrets are bonded to the Faraday cage where they protrude into the valve hall
cladding. Where wire mesh is used, the mesh size dictates the amount of shielding, and for cladding the
spacing of screws connecting the cladding panels together dictates the shielding. In addition to the walls,
floor and roof, all openings need to be shielded, so the converter transformer turret, where it protrudes
into the valve hall, is also bonded to the shield as shown in Fig. 3.4d.
Observation windows in the valve hall must also have a wire mesh across them: doors, when closed, must form
part of the RFI shield and conduits out of the valve hall must be bonded to the RFI shield and be constructed
from a metal tube with sufficient length to attenuate any interference.
As described in section 6.2, the valves are typically arranged as stacks of four valves, physically arranged
one on top of another. This arrangement is known as a quadrivalve and a typical valve hall containing one
12-pulse bridge, arranged as three quadrivalves is shown in Fig. 3.4e, Fig. 3.4f and Fig.3.4g.

CHAPTER
ValveÊhall
ThyristorÊquadrivalves
MVDCÊwallÊbushing
ControlÊbuilding
ConverterÊtransformers
(single-phase,Ê3-winding)
HVDCÊwallÊbushing
ConverterÊtransformer
cooler
Fig. 3.4e– Typical thyristor valve hall containing three quadrivalves and with single-phase, 3-winding converter
transformers adjacent to the valve hall
3.4e
Fig. 3.4f– 285 kVdc thyristor valve hall containing three quadrivalves

CHAPTER
PhysicalÊpointÊofÊsuspension
HV/UHVÊconnection MV/HVÊconnection
Id Id
T1 T5 T9 T6 T10 T2
las lad
lbs BridgeÊ1 lbd BridgeÊ2
lcs lcd
T7 T11 T3 T12 T4 T8
T12 T4 T8 T7 T11 T3
lad lcs
lbd BridgeÊ2 lbs BridgeÊ1
lcd las
T6 T10 T2 T1 T5 T9
Id Id
MV/HVÊconnection HV/UHVÊconnection
(i) (ii) ThreeÊʻquadrivalvesʼ
PhysicalÊpointÊofÊsuspension
12ÊpulseÊmidpoint
connection
Id
T12 T4 T8 T7 T11 T3
lad las
lbd lbs
lcd lcs
T6 T10 T2 T1 T5 T9
Id Id
MV/HVÊDCÊconnection HV/UHVÊDCÊconnection
BridgeÊ2 SixÊʻdouble-valvesʼ BridgeÊ1

(iii)
Fig.3.4g– Typical thyristor valve arrangement
(i) Normal diagrammatic 12-pulse bridge
(ii) Quadrivalve arrangement
(iii) Double-valve arrangement

CHAPTER
Fig.3.4h– Typical thyristor valve hall containing six double-valves
In some applications, the physical height of the valve may preclude the use of quadrivalves because
of seismic constraints, building height limitations or simply because of access to the valve. In such
circumstances, it is possible to use a double-valve arrangement, that is two thyristor valves stacked on
top of each other, resulting in six structures within the valve hall for a 12-pulse bridge (see Fig. 3.4g and
Fig 3.4h).
As can be seen from Fig. 3.4e the quadrivalves are normally suspended from the roof with the lower
voltage connection (usually the DC neutral) at the top of the structure and the high voltage connection
at the bottom of the structure. Hence the highest insulation is provided by air between the bottom of the
quadrivalve and the valve hall floor and walls. Where double-valves are used, it is more typical to suspend
the double-valve structures at the mid-point voltage so that all structures are physically the same and the
buswork is simplified as shown in Fig. 3.4g.
Access for maintenance within the valve hall is achieved with the aid of an access platform. This is normally
a battery-powered motorized mobile platform (the more common fossil fuel powered platforms would
introduce pollution into the valve hall) which can elevate an operator platform in order to obtain access to
all levels of the valve for inspection or testing purposes, Fig. 3.4i.
As discussed in section 6.2, the valve is designed so that all the valve components are self-extinguishing in
the event of a fire. This principle is extended to the rest of the valve hall ensuring that the risk of fire within
the valve hall is minimized. Of particular concern is the valve hall wall adjacent to the converter transformers.
Typically this is designed with a two hour fire rating and any holes in the wall, for example where the converter
transformer bushings protrude through, are filled with a suitable fire retardant foam.
3.4.1.3. DC Area
The DC area refers to the DC switchyard, that is all of the switchgear, transducers, circuit protection,
DC harmonic filtering and the smoothing reactors required between the converters and the outgoing
transmission circuit. As discussed in section 3.1 for a back-to-back converter there is no DC area and hence
both sets of valves are located in the same valve hall, as shown in Fig. 3.4i. However, as shown in section 3.1,
whenever there is a transmission circuit there will always be some additional components associated with

CHAPTER
Fig.3.4i– An access platform used for maintenance access of a HVDC converter

CHAPTER
ValveÊhall
MVDCÊwallÊbushing HVDCÊharmonicÊfilter
}
DCÊswitchyard
HVDCÊwallÊbushing
DCÊsmoothingÊreactor
Fig. 3.4j– A typical DC outdoor area enclosed within a fenced compound
the DC circuit.
The most cost-effective DC area is typically outdoors as shown in Fig. 3.4j. Where practical, the components
within the DC area can be built on top of support structures to allow the DC area to be a ‘walk around’
switchyard.
Some equipment, such as the capacitor associated with a DC harmonic filter (Fig. 3.4k), is not suitable for
3.4j
Fig. 3.4k– A typical DC filter capacitor

CHAPTER
Fig. 3.4l– An outdoor DC smoothing reactor with the insulation mounted on the ground and with the DC through-
wall bushing from the valve hall in the background
location on top of a structure and therefore is either suspended from a frame or gantry, or located within a
fenced off compound. Where the environment in which the converter station is polluted or where there is a
seismic requirement which requires the insulation, and hence structure heights to be minimized, it may be
necessary to enclose the DC area within a building.
Even when enclosed within a building, some of the components are still mounted on top of structures in order
to allow personnel to walk through the indoor DC area, with other areas fenced off to reduce the insulation
and allow the equipment to be ground-mounted.
Where an enclosed DC area is used, air is circulated in order to remove the heat losses from the equipment, but
larger creepage levels than in the valve hall are used. Consequently, it is not necessary to maintain a positive
pressure with respect to the outdoors and less filtering is normally employed compared with the valve hall.
3.4.1.4. Service Building

The service building, sometimes also called the control building, contains not only the converter control
and protection equipment but also the services associated with the operation of the HVDC converter for
example:
➙ Main operator control facilities for the HVDC link
➙ Valve cooling plant
➙ DC batteries, chargers and distribution boards
➙ AC distribution boards
➙ Telecommunication equipment
➙ Workshop
➙ Storage
➙ Offices
➙ Document storage
➙ Lavatories
The service building can be either a single storey or a multi-storey building as demanded by the specific
project requirements. A simple single storey service building is shown in Fig. 3.4m.

CHAPTER
ValveÊhall
DCÊsmoothingÊreactor
ControlÊbuilding
ControlÊroom
(i) (containingÊcontrolÊandÊprotectionÊpanels)
3.4mÊ(i)
(ii)
Fig. 3.4m– A typical service building

(i) Computer Aided Design (CAD) image
(ii) Actual building

CHAPTER
3.4.1.5. Auxiliary Services
The auxiliary services area of the converter station may not be needed where there is an existing auxiliary
supply or may simply consist of auxiliary step-down transformers. Where a diesel generator or multiple
diesel generators are used for auxiliary supplies, these are also located in this area.
Fig. 3.4n– A typical AC harmonic filter compound
3.4.2. AC HARMONIC FILTERS

The AC harmonic filters are typically located adjacent to the converter island. Each filter is normally located
within an individually fenced compound and the components of the filter are ground mounted, as shown
in Fig. 3.4n. The individual compounds allow for isolation and access to individual banks for maintenance
activities without de-energizing the complete station. The filter components are located to give both
electrical clearance and maintenance access.
Although for most HVDC converter schemes the AC harmonic filters are located outdoors there are specific
cases where, because of the level of pollution, the AC harmonic filters have been located within buildings. An
example of such a project is the Cheju Island Bipole 1 scheme in South Korea, where the converter station on
the island of Cheju is located right on the shoreline [11].
3.4.3. AC SWITCHYARD
The AC switchyard is what connects together all of the elements of the HVDC converter station on the AC
side and is essentially the same technology as is used for conventional AC substations. There are two basic
types of switchyard that can be used: AIS (Air Insulated Switchgear) and GIS (Gas Insulated Substation). GIS
is essentially much more compact than AIS, but incurs an increased capital cost. GIS is therefore typically

CHAPTER
only used where land space is at a premium or when the outdoor pollution levels are very high making it
impractical to have exposed insulation. Often where GIS is used, interconnections between the switchgear
and the actual equipment will be made via HVAC cable, again to minimize the land area and to reduce the
exposed insulation.
The more common AC switchyard arrangement uses AIS because of cost. Here all AC connections are
typically exposed and therefore electrical and safety clearances have to be maintained throughout, which
will inevitably take up a considerable amount of space because the AC connection is 3-phase, particularly at
higher AC voltages.
3.4.4. ACOUSTIC NOISE CONSIDERATIONS

In many HVDC project locations there are requirements resulting from local environmental rules related to the
acoustic noise which any substation (or other industrial site) is permitted to generate at either its boundary or
at the nearest property. Much of the equipment that a HVDC converter station consists of generates acoustic
noise when operating and therefore careful consideration is required in terms of the layout of the equipment
in order to try to minimize the acoustic noise at the point of measurement. Typical acoustic noise sources
within a converter station (measured as sound power (Pw)) are:
➙ DC smoothing reactor (110 dB(A) sound power)
➙ Converter transformer(s) (105 dB(A) sound power)
➙ Valve cooling (100 dB(A) sound power)
➙ AC harmonic filter reactor (100 dB(A) sound power)
➙ Transformer cooling (105 dB(A) sound power)
➙ AC harmonic filter capacitors (80 dB(A) sound power)
As an approximation, the acoustic noise sound pressure (Lw(A)) from any individual point source at a
distance ‘x’ from the component is calculated from:
Lw(x) = Pw – 20 Log10(x) – 8 [Eqn. 3.4b]
Fig. 3.4o– Converter transformers within an acoustic noise enclosure

CHAPTER
Fig. 3.4p– AC harmonic filter reactor within an acoustic noise enclosure
Where:
Lw(x) = the sound pressure at a distance x (in meters)
Pw = the acoustic sound power of the point source (dB(A))
x = the distance from the point source at which the sound pressure is to be calculated (in meters)
Fig. 3.4q– DC smoothing reactor behind (silver-colored) acoustic noise enclosure

CHAPTER
Where the acoustic noise limits cannot be achieved through careful disposition of equipment, shielding
must be employed. This shielding can take several forms, the extreme being to totally enclose the equipment
generating the noise as shown in Fig. 3.4o and Fig. 3.4p. In Fig. 3.4o the converter transformer tanks have
been enclosed to limit the acoustic noise which can propagate from the tanks so only the AC 400 kV
connection bushings can be seen. The transformer coolers associated with the converter transformers can
also be seen. Whilst these cannot be enclosed, as this would restrict the airflow around them, the fans are
fitted with silencers.
In some instances total enclosure is not necessary but some screening is still required. An example of this
is shown in Fig. 3.4q, where the DC smoothing reactor is surrounded by an acoustic noise shield. The shield
has no roof, hence the DC connections are brought out from the top. There is also a gap around the bottom
of the shield to allow air to circulate.
An alternative measure for reducing noise propagating from a converter station is to use landscaping features
such as vegetation (although this is not always considered as a viable permanent solution) or a berm outside
or near the boundary. Such solutions are only practical where sufficient land exists around the station.
3.4.5. GENERAL LAYOUT INFORMATION

As a guide to the typical station footprint that is required for a HVDC installation, Table 3.4a lists a range
of typical converter station layouts, according to various criteria, including circuit configuration, AC and DC
voltage. All of these installations are based on the use of AIS equipment.
DC
HVDC Total HVDC Transmission AC swyrd AC filter Area Acres/
voltage m m Acres
rating MW configuration circuit rating voltage m2 MW
rating
230 kV / 345 25 kV / 25
200 MW 210 Monopole BTB 50 kV 150 120 18000 4.45 0.021
kV kV
400 kV / 400
500 MW 500 Monopole BTB 205 kV 400 kV 400 450 180000 44.5 0.089
kV
2×
500 Bipole Cable +/-250 kV 275 kV 275 kV 126 250 31500 7.78 0.016
250 MW
2×
1000 Bipole OHL +/-500 kV 500 kV 500 kV 300 325 97500 24.1 0.024
500 MW
2×
2000 Bipole OHL +/-500 kV 500 kV 500 kV 380 400 152000 37.6 0.019
1000 MW
2× 500 kV +
3000 Bipole OHL +/-500 kV 500 kV 400 600 240000 59.3 0.020
1500 MW 220 kV
Table 3.4a– A range of typical converter station land requirements, according to circuit configuration and AC and DC
voltages. All of these installations are based on the use of AIS equipment.

CHAPTER
BIBLIOGRAPHY
[1] C. Barker, “HVDC for beginners and beyond”, 1st edition, AREVA T&D publication, 2009.
[2] “Terminology for high voltage direct current (HVDC) transmission”, IEC 60633, 2009.
[3] “Insulation Co-ordination – Part 1: Definitions, principles and rules”, IEC 60071-1.
[4] “Insulation Co-ordination – Part 2: Application Guide”, IEC 60071-2.
[5] “High voltage testing techniques – Part 1: General”, IEC 60060-1.
[6] “Thyristor valves for high voltage direct current (HVDC) power transmission – Part 1: Electrical Testing”,
IEC 60700-1,
[7] “Guide for the Selection of Insulators in Respect of Polluted Conditions”, IEC 60815.
[8] “Guidelines for the Application of Metal-Oxide Arresters without gaps for HVDC Converter Stations”,
CIGRÉ Brochure 034, 1989.
[9] “Modeling Guidelines for Fast Front Transients”, IEEE Transactions on Power Delivery, Vol. 11, No. 1,
January 1996.
[10] Introduction to PSCAD/EMTDC V3 from Manitoba Hydro Research.
[11] J. L. Haddock, F. G. Goodrich, Kim Se Il, “Design Aspects Of Korean Mainland To Cheju Island HVDC
Transmission”, Power Technology International 1993, Sterling Publication Ltd., p. 125.
[12] W. Diesendorf, “Insulation co-ordination in high voltage electric power systems”, Butterworth, 1974.
[13] K. N. Mathes, “Performance of Simple Insulator Shapes Under Heavily Contaminated Conditions”, IEEE
Transactions on Electrical Insulation, Vol. EL-7. No. 2, p. 64-78.
[14] F. W. Peek Jr., “The Effect of Transient Voltages on Dielectrics”, Transaction A. I. E. E., 1915, p. 1857-1920.
[15] F. W. Peek Jr., “The Sphere Gap as a Means of Measuring High Voltage”, Transaction A. I. E. E., 1914, p.
923-949.
[16] F. W. Peek Jr., “The Effect of Transient Voltages on Dielectrics II, The Effect of Lightning Voltages on
Arrester Gaps, Insulators and Bushings on Transmission Lines”, Transaction A. I. E. E., 1919, p. 1137-1177.
[17] F. W. Peek Jr., “Effect of Altitude on the Spark-Over Voltages of Bushings, Leads and Insulators”,
Transaction A. I. E. E., 1914, p. 1721-1730.
[18] B. F. Jakobsen, “Corona test at High Altitude”, Transaction A. I. E. E., 1918, p. 91-122.
[19] P. A. Calva, F. P. Espino, “Correction Factors for Positive DC Voltages, IEEE Transactions on Dielectrics
and Electrical Insulation”, Vol. 5, No. 4, August 1998, p. 541-544.
[20] T. H. Frick, J. R. Stewart, A. R. Hileman, C. R. Chowaniec, T. E. McDermott, “Transmission Line Insulation
Design at High Altitude”, IEEE Transaction on Power Apparatus and Systems, Vol. PAS-103, No. 12, December
1984, p. 3672-3680.
[21] M. Ramirez, M. Moreno, A. Pigini, G. Rizzi, E. Garbagnati, “Air Density Influence on the Strength of
External Insulation under Positive Impulses: Experimental Investigation up to an Altitude of 3000 m a.s.l.”,
IEEE Transaction on Power Delivery, Vol. 5, No. 2, April 1990, p. 730-737.

CHAPTER
4| HARMONICS: CAUSES, CALCULATIONS AND FILTERING
TOC 158 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
4
HARMONICS:
CAUSES,
CALCULATIONS &
FILTERING
Unfortunately, harmonics exist in all AC power systems. If excessive,
these harmonics can cause undesirable distortion in the AC supply
voltage. In this chapter we will briefly describe sources of harmonic
distortion and their effects on equipment connected to the network.
As you might expect, various international bodies have introduced
limits to ensure that harmonic distortion does not cause a detrimental
effect on connected equipment.
As you read this chapter, you will be provided with extensive and
valuable information about the harmonics produced by HVDC
converters under all operating conditions and the types of filters used
to ensure that the various converters comply with their respective
harmonic limits.
We also present a filter design process giving the essential equations
to enable the main components of each filter type to be defined.
Additionally, the issue of interference radiated from the converters is
described, as well as how we achieve the required electromagnetic
compatibility for the whole converter station.
Detailed mathematical equations to support these calculations can
be found in the chapter’s appendice.

Chapter contents
HARMONICS: CAUSES,
4. 4.7. POWER LINE CARRIER FREQUENCY FILTERS................ 213
CALCULATIONS AND FILTERING........... 158 4.7.1. Power Line Carrier Communications................................ 214
4.7.2. PLC Studies and Filter Design.............................................. 214
4.1. SOURCES, EFFECTS AND LIMITS OF HARMONIC 4.7.3. Design Requirements.............................................................. 215
DISTORTION................................................................................. 161 4.7.4. Overview of the System......................................................... 216
4.1.1. Harmonics on AC Power Systems...................................... 161 4.7.5. Noise Source Model.................................................................. 217
4.1.2. Sources of Harmonic Distortion......................................... 162 4.7.6. The Thyristor Valves................................................................. 217
4.1.3. Effects of Harmonic Distortion............................................ 163 4.7.7. Converter Transformer............................................................ 217
4.1.4. Effects of Harmonics on DC Power Systems................. 163 4.7.8. The AC System............................................................................ 220
4.1.5. Harmonic Limits......................................................................... 163 4.7.9. PLC Filter Design........................................................................ 220
4.7.10. PLC Filter Components........................................................... 221
4.2. ARMONIC IMPEDANCE OF AC
H 4.7.11. Rating of PLC Filter Components........................................ 221
AND DC SYSTEMS...................................................................... 164 4.7.12. Site Measurements................................................................... 221
4.2.1. AC System Harmonic Impedance....................................... 164
4.2.2. Modeling AC Network Elements......................................... 166 4.8. RADIATED INTERFERENCE AND EMC............................... 221
4.2.3. Calculation of the AC Harmonic Impedance................. 170 4.8.1. Radiated Interference Studies............................................. 222
4.2.4. DC System Harmonic Impedance...................................... 172 4.8.2. Design Requirements.............................................................. 222
4.8.3. Overview of the System......................................................... 222
4.3. HVDC CONVERTER HARMONICS........................................ 173 4.8.4. Noise Source Model.................................................................. 224
4.3.1. Converter Characteristic AC Harmonics......................... 173 4.8.5. The Thyristor Valves................................................................. 225
4.8.6. Converter Transformer............................................................ 225
4.4. CONVERTER DC HARMONICS.............................................. 178 4.8.7. The AC Systems.......................................................................... 226
4.4.1. 3-pulse Model.............................................................................. 179 4.8.8. RFI Shielding Design................................................................. 226
4.4.2. Characteristic DC Harmonics............................................... 180 4.8.9. Radiation due to Noise Conducted
4.4.3. Non-characteristic DC Harmonics..................................... 181 to the Exposed Busbars.......................................................... 228
4.4.4. Combination of Characteristic and 4.8.10. Site Measurements................................................................... 228
Non-characteristic Voltages................................................. 182 4.8.11. Electromagnetic Compatibility (EMC) –
Immunity Issues......................................................................... 228
4.5. AC HARMONIC FILTERS........................................................... 182
4.5.1. Design Data.................................................................................. 182 BIBLIOGRAPHY ............................................................................................ 229
4.5.2. Filter Performance Requirements...................................... 183
4.5.3. Filter Types................................................................................... 185
4.5.4. Initial Design................................................................................ 193
4.5.5. Steady-State Design................................................................. 196
4.5.6. Transient Design........................................................................ 199
4.5.7. Protection .................................................................................... 202
4.6. DC HARMONIC FILTERS.......................................................... 203

4.6.1 Design Data.................................................................................. 203
4.6.2. Performance Requirements.................................................. 203
4.6.3. Filter Types................................................................................... 208
4.6.4. Steady-State Design................................................................. 208
4.6.5. Transient Design........................................................................ 212
4.6.6. Protection..................................................................................... 213

4.1. SOURCES, EFFECTS AND LIMITS OF HARMONIC DISTORTION
Harmonic distortion produced by HVDC converter stations represents a power quality issue to network
operators. Limits on acceptable harmonic distortion are normally defined by network operators or national
legislation. In this section the nature of harmonic distortion is briefly described, together with the common
sources and effects of harmonics.
4.1.1. HARMONICS ON AC POWER SYSTEMS

Harmonics within a power system are defined as the modulation of the voltage or current at an integer
multiple of the fundamental frequency. On a 50 Hz system for example, the presence of a 5th harmonic
voltage means that there is an additional 250 Hz component added to the voltage waveform which will
distort the voltage waveform as shown in Fig. 4.1a.
An example of excessive harmonic distortion is shown in Fig. 4.1b. The vertical scale in Fig. 4.1b is the
Total Harmonic Distortion (THD), one of the various measures used to quantify harmonic distortion, as
described later in section 4.5.
ElectricalÊdegrees
1.1
0.55
AmplitudeÊ(p.u.)
0
0 100 200 300 400 500 600 700
-Ê0.55
-Ê1.1
UndistortedÊ(fundamentalÊonly)
ElectricalÊdegrees
1.1
0.55
Amplitude
0
0 100 200 300 400 500 600 700
-Ê0.55
-Ê1.1
UndistortedÊ+Ê5%Ê5thÊharmonic
Fig. 4.1a– Example of 5% 5th harmonic distortion on a 3-phase AC waveform

CHAPTER 4.1a
VryÊTHDÊmaxÊ% VybÊTHDÊmaxÊ% VbrÊTHDÊmaxÊ% PlanningÊlimit

8
5
HarmonicÊdistortionÊ%
0
27/06/2004 28/06/2004 29/06/2004 30/06/2004 01/07/2004 02/07/2004 03/07/2004 04/07/2004 05/07/2004
00:00 00:00 00:00 00:00 00:00 00:00 00:00 00:00 00:00
Fig. 4.1b– An example of excessive background harmonic distortion on an 11 kV network
4.1.2. SOURCES OF HARMONIC DISTORTION

There are many sources of harmonic distortion on an AC network. The following list is not exhaustive, but
lists some of the most common sources which create the background harmonic pollution on a network.
4.1b Large Power Converters for HVDC and SVC

4.1.2.1.
HVDC stations generate harmonic distortion, even with filters connected. As will be discussed in detail in
this chapter, these are predominately at high harmonic orders, 11th and above. Static VAr Compensators
(SVC) generate harmonics at harmonic orders 5th and above.
4.1.2.2. Industrial Rectifiers

These can be controlled converters using thyristors, or uncontrolled converters using diodes. In normal
operation phase shifting transformer connections are used to increase the pulse number of the converters
from 6-pulse or 12-pulse to 24-pulse (lowest harmonic 23rd) or 48-pulse (lowest harmonic 47th) or higher.
4.1.2.3. Industrial Drives

Modern drives, whether AC/AC or AC/DC types, are normally harmonically compensated and contribute
little background distortion. Converters using Pulse Width Modulation (PWM) techniques will generate
harmonics, although the levels are small and the frequencies are high (around 1000 Hz).
4.1.2.4. Domestic Sources

Many domestic and commercial devices such as televisions and computers generate low levels of harmonic
currents from their power supplies. However, their vast numbers can lead to a significant level of low order
harmonic distortion, principally at 5th harmonic. In most countries the major contributor to background
distortion comes from these sources.
4.1.2.5. Lighting
Many modern types of lighting devices such as fluorescent or discharge lights can generate harmonic currents.

CHAPTER
4.1.2.6. Power Equipment
Under high voltage conditions (e.g. during light load), transformers and generators can generate harmonic
currents due to saturation effects in core steel.
4.1.2.7. Traction Loads

Single-phase traction loads can generate imbalances and harmonic currents onto the supply system.
4.1.2.8. Large Industrial Loads

Electric-arc furnaces can generate considerable levels of harmonic distortion and flicker. However, such
installations are normally connected at strong points of a network and have harmonic mitigation measures
in place.
The multiple sources of background harmonic distortion contribute to an ambient condition at the terminals
of the HVDC station. The harmonic performance limits will be dependent upon this pre-existing level of
distortion. The rating of the filter equipment will also need to take these harmonic levels into account when
determining individual equipment rating levels.
4.1.3. EFFECTS OF HARMONIC DISTORTION

The presence of high levels of harmonic distortion can have serious consequences on equipment connected
to a network. Some of the key problems are:
➙ Overloading of capacitor banks: as capacitors have an impedance characteristic inversely proportional
to frequency, they naturally act as filters and are exposed to overvoltage and overcurrent in the
presence of high levels of background harmonic distortion.
➙ Electrical machines can become overheated due to the presence of harmonics inducing eddy currents
in the windings.
➙ Interference with electronic circuits: however, this may be less of a problem with digital electronics.
➙ Interference with telecommunication systems: this is mainly the coupling of harmonic distortion from
the power lines to adjacent telephone lines.
Harmonic limits are set to minimize these and other effects caused by harmonic distortion on the power
network. In the specific case of DC transmission systems, as there are no consumers of power, the only effect
is the problem of interference with any adjacent open wire telephone circuits.
4.1.4. EFFECTS OF HARMONICS ON DC POWER SYSTEMS

Unlike AC systems, there are no consumers who take their power supply directly from the DC transmission
system. Thus, the only effect of harmonic distortion on a DC system is the potential for interference with adjacent
telecommunication systems. Electromagnetic coupling between the DC lines and open wire telephone lines is
quantified by evaluating the harmonic currents flowing on the DC lines. The level of equivalent disturbing current
flowing in the DC lines is the only parameter applied to DC side harmonic performance.
4.1.5. HARMONIC LIMITS

There are many harmonic limits used throughout the world, defined by international bodies such as IEC,
national bodies such as ANSI/IEEE, and by individual transmission or distribution operators. There are no
common harmonic standards: each is specific to a particular network and a particular system voltage level.
Harmonic limits tend to fall into four categories:
➙ Limits on voltage waveform distortion
➙ Limits on injected current
➙ Limits on telephone-weighted voltage distortion
➙ Limits on telephone-weighted current
Voltage waveform distortion is specified on almost all systems. Current limits are less widely used. Limits on
telephone weighted (or psophometric) values are only relevant where there is a long exposure of telephone
circuits to power circuits.

CHAPTER
Definitions of commonly used terms to define harmonic distortion are given in Appendix A4.1.
Two examples of voltage waveform distortion limits are shown in Tables 4.1a and 4.1b below.
IEC 61000-3-6
Odd harmonics non-multiple of 3 Odd harmonics multiple of 3 Even harmonics
Harmonic Harmonic voltage % Harmonic Harmonic voltage % Harmonic Harmonic voltage %

order order order
n MV HV-EHV n MV HV-EHV n MV HV-EHV
5 5 2 3 4 2 2 1.8 1.4
7 4 2 9 1.2 1 4 1 0.8
11 3 1.5 15 0.3 0.3 6 0.5 0.4
13 2.5 1.5 21 0.2 0.2 8 0.5 0.4
17 ≤ n ≤ 49 21 < n ≤ 45 0.2 0.2 10 ≤ n ≤ 50

Table 4.1a– IEC harmonic voltage limits
The IEC standard only provides limitation of voltage distortion; no limits are stated for current injection.
There are no telephone-weighted limits stated.
ANSI/IEEE 519
Bus voltage at PCC Individual voltage distortion (%) Total voltage distortion (THD) (%)
69 kV and below 3.0 5.0
69 kV to 161kV 1.5 2.5
161 kV and above 1.0 1.5

Table 4.1b– IEEE harmonic voltage limits
In addition to the voltage distortion limits, the IEEE standard also provides limits on harmonic current and
on telephone-weighted voltage and current.
4.2. HARMONIC IMPEDANCE OF AC AND DC SYSTEMS

In order to calculate the harmonic distortion which will be generated by the HVDC converter on the AC and
DC transmission systems, two key items of data are required: these are the harmonic currents and voltages
created by the converter and the harmonic impedance of the AC and DC systems.
4.2.1. AC SYSTEM HARMONIC IMPEDANCE

Information on the possible AC network impedance as seen from the converter terminals is required
for AC filter performance and rating calculations. This information is required not only for fundamental
frequency, where it is generally quite well characterized by the Short-Circuit Level, but also for harmonic
frequencies. For this reason, the AC network impedance is usually referred to as the harmonic impedance
of the network.
This AC system harmonic impedance data can be obtained either by measurement or by calculations. In
the case of large networks, the difficulty is that the calculated (or measured) impedance values are valid for
a specific system configuration only. The AC system impedance, as a function of frequency, might change
dramatically with the changed operating conditions. For the filter calculations however, it is necessary to
consider all system conditions that are possible today and in the future. Therefore, general models must be
used to characterize the system impedance.
For a simplified AC network, the harmonic impedance of the network, as seen from the terminals of the
HVDC converter station, is Zx as shown in Fig. 4.2a. The network (inside the circle) can be considered as
a general harmonic load flow model in a network analysis program (such as PSS/E), which consists of the
generator voltage U1 and its internal impedance Zu1 and the lumped impedance Z1 (including the transformer
CHAPTER
Zu1 Zx
Ux 1
X
S = PÊ+ÊjQ
U1
InjectedÊharmonic
currentÊsource I2
Z1
SimplifiedÊACÊsystem
Fig. 4.2a– A simplified AC network with an injected harmonic source
4.2a
impedances, shunt connected devices, system loads and transmission line impedances). The busbar voltage
Ux is obtained from the solved load flow solution.
Fig. 4.2.a shows a harmonic current source I2. In the harmonic impedance calculation, the injected harmonic
current source is automatically applied through a dummy generator by using a harmonic scanning program,
such as the IPLAN based GE Vernova program. The harmonic frequency can be changed from 1 p.u. to 70
p.u. with a step or interval (e.g. 0.1 p.u.) during the harmonic impedance calculation. In the program, the
harmonic impedance Zx of the network is calculated in three steps as follows:
Step 1: Calculation of the voltage Ux,bef :
The voltage Ux,bef , which is the voltage at node X before injection of the harmonic current (see Fig. 4.2a),
is given by:
Z1
U x ,bef = U 1 ⋅ [Eqn 4.2.1a]
Z 1 + Z u1
The voltage Ux,bef is calculated and obtained in the IPLAN program by solving the load flow of the AC network
without the harmonic current source. After retrieving the angle j and the magnitude |U|, Ux,bef is stored as
real and imaginary parts.
Ux,bef = |U| • cos(j) + j|U| • sin(j) [Eqn 4.2.1b]
Where:
j = phase angle of Ux,bef
Step 2: Calculation of the voltage Ux,aft :
After connecting the harmonic current source to the network (see Fig. 4.2a in the dashed lines) the overall
voltage Ux,aft is now composed of the voltage Ux,bef and a voltage Ux(I2) which is a function of the harmonic
current I2. By superposition, the overall voltage Ux,aft is given by:
Ux,aft = Ux,bef + Ux(I2) [Eqn 4.2.1c]
The voltage Ux(I2) is obtained as:

CHAPTER
Z u1 ⋅ Z 1 [Eqn 4.2.1d]
U x (I 2 ) = I 2 ⋅
Z u1 + Z 1
Replacing Ux,bef and Ux(I2) in Eqn. 4.2.1c with Eqn. 4.2.1a and Eqn. 4.2.1d, the overall voltage Ux,aft is expressed by:
Z1 Z u1 ⋅ Z 1
U x ,aft = U x ,bef + U x ( I 2 ) = U 1 ⋅ + I2 ⋅ [Eqn 4.2.1e]
Z 1 + Z u1 Z u1 + Z 1
This voltage Ux,aft is calculated and obtained in the program by solving the load flow of the AC network and
including the harmonic current source.
Step 3: Calculation of the harmonic impedance Zx :
To obtain the complex impedance Zx of the network viewed from the port defined between nodes 1 and 2
in Fig. 4.2a, it is necessary to derive the voltage Ux(I2) from the voltage Ux,aft and current. Using Eqn 4.2.1c
and Eqn 4.2.1d, the harmonic impedance Zx is obtained by:
U x ( I 2 ) U x ,aft − U x ,bef
Zx = = [Eqn 4.2.1f]
I2 I2
where I2 is the current supplied by the injected harmonic current source. This is calculated by measuring
the complex power S = P + jQ delivered from the current source to the network.
The current I2 to calculate Zx is then given by:
S*
I2 = [Eqn 4.2.1g]
U x*,aft
4.2.2. MODELING AC NETWORK ELEMENTS

Models of the network elements used for harmonic impedance purposes and simulation studies are shown
in the related documents [4, 5, 6, 8]. According to the CIGRÉ and IEEE guidelines for assessing network
harmonic impedance [4, 5, 6], the AC network element models are as follows:
4.2.2.1. Shunt Elements

Shunt capacitors and inductors are modeled as simple impedances. Their capacitance and inductance
respectively are calculated from the MVAr at fundamental frequency and nominal voltage.
A shunt element is represented by an admittance: Yshuntn = Gshuntn + j Bshuntn.
At any harmonic number n, with G1 = 0 at fundamental frequency (in the PSS/E load-flow program data files)
the conductance of practical power system elements is considered to be negligible, the shunt admittance
is calculated by:
Bshuntn = n • B1 for a capacitive shunt
and
Bshuntn = B1/n for an inductive shunt [Eqn 4.2.2a]
Here, G 1 + j B 1 is the admittance of the shunt at the
fundamental frequency and G1 is normally zero in the harmonic
load flow calculations.
4.2.2.2. Load Elements GshuntnÊ=Ê0 Bshuntn

Loads in the network are generally classified into two types.
One type is the linear load (e.g. residential and commercial
loads), induction motors, synchronous motors, etc. Another
type of load is the non-linear load. Non-linear loads include
power electronic converters, adjustable speed drives and
arc furnaces. Linear loads are a significant component of the
Fig. 4.2b– Circuit for the shunt model
network. Non-linear loads generate harmonics in the network

CHAPTER
4.2b
and are normally dealt with as harmonic sources rather than impedances.
Harmonics in the network are usually attenuated by the use of harmonic filters. In harmonic filter design,
the detailed AC network harmonic impedance is a key requirement. Studies by CIGRÉ [4] show that linear
loads play an important part in the harmonic impedance characteristic of the network. Linear loads
provide the main component of damping within the network, however they may affect the resonance
conditions of the network, particularly at higher frequencies. Calculation of network harmonic impedance
requires the definition of the impedances that describe the behavior at the harmonic frequencies of all
loads in the network. However, it is difficult if not impossible to obtain full information on this. Harmonic
impedance calculation of the network is therefore approximated by representing the loads in the network
as aggregated linear load models [4, 5].
The IEEE Task Force [6] shows that the load models [4, 5] are not accurate for harmonic orders below
5. These models neglect the harmonic attenuation of the motor load. They are only to be used if the
participation of the induction motors is moderate (K < 0.3, where K = percentage of total load). To consider
the harmonic attenuation (or damping) of the load, particularly if the load is dominated by motor loads (K
> 0.7), a more accurate representation of the load is given by Model 6, by including a resistor R1 in series
with X1. There is no further clarification in the reference of which model is most appropriate for 0.3 < K
<0.7. This indicates that Model 6 has to be used if the harmonic attenuation of the motor load needs to be
considered or a motor participation of K is larger than 0.7.
Considering motor participation in the load and harmonic attenuation, an aggregate load model (Model 6)
is presented and used, as shown in Fig. 4.2c.
Where:
Resistive Motive
V2
R2 = [Eqn 4.2.2b]
(1 − K ) P
jnX2 jnX1
X2 = 0.1R2 [Eqn 4.2.2c]
V2
X1 = X M [Eqn 4.2.2d] R2 R1
Km ⋅ K ⋅ P
R1 = X1/K3 [Eqn 4.2.2e]
Km is the ‘install’ factor, defined as Km = 1/power factor
(≈ 1.2) Fig. 4.2c– Representation of aggregate load
XM is the p.u. value of motor locked rotor reactance model (Model 6 in [6])
expressed on the motor rating (≈ 0.15-0.25)
K3 is the effective quality factor of the motor circuit (≈ 8)
The skin effect of the motive part R1 is considered in the same way as for generators [5] (i.e., by the
expression R1 _ skin = R1 n ). 4.2c
Similarly, for the resistive part, the skin effect of R2 is represented by R2 _ skin = R2 n .
4.2.2.3. Generator Elements

At harmonic frequencies, a generator is represented by its sub-transient reactance Xd″ and by a resistance
Rg [5]. Resistance Rg is derived from the machine power loss at fundamental frequency by the equation:
Rg = Kg Xd″ [Eqn 4.2.2f]
Where:
Kg is a factor corresponding to a sub-transient time constant of 32 ms [5]: Kg = 0.1 for a 50 Hz system, Kg
= 0.083 for a 60 Hz system.
Taking into account skin effect, the generator frequency dependent resistance Rgenn is given by:
Rgenn = Rg n [Eqn 4.2.2g]
Where:
n is the harmonic number.
CHAPTER
As the rotating magnetic field created by stator
harmonics rotates at a speed significantly higher
than that of the rotor, the harmonic impedance of
the generator approaches its negative sequence Rgenn
impedance. The IEEE Task Force on Harmonics
Modeling and Simulation [6] suggests the
synchronous machine harmonic inductance is
usually taken to be either the negative sequence Xgenn
impedance or equivalently the average of direct and
quadrature sub-transient reactances. The generator
reactance is represented by [8]:
( X d′′ + X q′′)
X genn = n ⋅ X 2 = n ⋅ [Eqn 4.2.2h] Fig. 4.2d– Representation of a synchronous generator
2
However, in PSS/E data sheets, if most generators are of the round rotor type (Xd″ = Xq″), their arithmetic
average = Xd″ can be used in the study. 4.2d
The harmonic impedance representation for a generator is shown in Fig. 4.2d.
4.2.2.4. Transmission Line Elements

Zʼ
While the simple classical lumped parameter
p-equivalent circuit with R-L series and C parallel
elements is often sufficient to represent short
transmission lines, a more exact representation is Yʼ/2
Yʼ/2
required for long transmission lines by employing
the distributed parameters taken from the
hyperbolic transmission line equations. The
distributed p-equivalent transmission-line model
is shown in Fig. 4.2e [5].
Fig. 4.2e– Equivalent circuit of the transmission
In Fig. 4.2e, the equivalent parameters Z ′ and Y ′
line model
of the line are defined by Eqn 4.2.2i and Eqn 4.2.2j.
Z sin n YZ
Z′ = [Eqn 4.2.2i]
YZ
4.2e
YZ
Y′ tan n
=Y 2 [Eqn 4.2.2j]
2 YZ
Where:
Z = R1 + jnX1 is the series impedance of the line at harmonic number n
Y = G1 + jnB1 is the shunt admittance of the line at harmonic number n
R1, X1, G1 and B1 are the lumped resistance, reactance, shunt conductance and shunt susceptance of the
line at fundamental frequency. Normally, G1 is zero in load flow data.
As no information on the types and configurations of lines and cables is available, the estimated factors for
skin effect of the overhead lines and cables are applied by increasing the line resistance R with harmonic
frequency number using the typical corrections [7].
For lines of voltage 200 kV and above:
➙ where n ≤ 4 (50 Hz system), n ≤ 3.3333 (60 Hz system)
 3.45 n 2 
Rn = R1  1 + [Eqn 4.2.2k]
 192 + 2.77 n 2 
➙ where 4 < n ≤ 8 (50 Hz system), 3.3333 < n ≤ 6.6667 (60 Hz system)

CHAPTER
(
Rn = R1 0.864 − 0.024 n + 0.105 n ) [Eqn 4.2.2l]
➙ where n > 8 (50 Hz system), n > 6.6667 (60 Hz system)

(
Rn = R1 0.267 + 0.485 n ) [Eqn 4.2.2m]
➙ For lines, voltage below 200 kV

 0.646 n 2 
Rn = R1  1 + [Eqn 4.2.2n]
 192 + 0.518 n 2 
➙ For cables, voltage above 200 kV, n ≥ 2 (50 Hz system), n ≥ 1.6667 (60 Hz system),
(
Rn = 0.74 R1 0.267 + 1.073 n ) [Eqn 4.2.2o]
➙ For cables, voltage below 200 kV, n ≥ 2 (50 Hz system), n ≥ 1.6667 (60 Hz system),
(
Rn = R1 0.187 + 0.532 n ) [Eqn 4.2.2p]
4.2.2.5. Transformer Elements

Complete representation of transformers including Xp
Rs
stray capacitances is not practical and cannot be
justified for harmonic frequencies. Experience
shows that stray capacitance starts to have some
effect at around 10 kHz. The maximum harmonic
frequency present in power systems due to power Rp
frequency effects is about 2 kHz [4]. Transformer
impedance is shown to be proportional to the Fig. 4.2f– Configuration of the transformer model
leakage reactance and is linear with harmonic
frequency.
CIGRÉ Working Group 36.05 [4,5] recommended that transformer models consist of a resistance Rs in series
with an assembly of reactance Xp in parallel with a resistance Rp, as shown in Fig. 4.2f.
In Fig. 4.2f, reactance Xp is represented by:
4.2f
Xp = nXT [Eqn 4.2.2q]
Where:
XT is the leakage reactance of the transformer at fundamental frequency.
Resistances Rs and Rp are constant at all frequencies. The estimated values of the resistances Rs and Rp
are [2]:
XT
Rs = [Eqn 4.2.2r]
tan ϕ
Rp = 10 ⋅ XT ⋅ tanj [Eqn 4.2.2s]

tanj is given by:
tanj = exp [0.693 + 0.796 lnSrating – 0.0421(lnSrating)2] [Eqn 4.2.2t]
Where:
Srating is the rating of the transformer, MVA.
The equivalent impedance of the transformer is:
jX p Rp jnXT Rp  n2 X 2 R   nX R 2 
ZT = Rs + = Rs + =  Rs + 2 T 2 p 2  + j  2 T 2p 2  [Eqn 4.2.2u]
Rp + jX p Rp + jnXT  R p + n XT   R p + n XT 
Eqn 4.2.2u shows the real part of the equivalent impedance is a function of harmonic frequency order n,
indicating that skin effect of the transformer resistance is also considered in the model.

CHAPTER
4.2.3. CALCULATION OF THE AC HARMONIC IMPEDANCE
The above modeling technique can be used in an AC system harmonic analysis tool to derive an impedance
plot of the AC system harmonic impedance, typically between 2nd and 50th harmonics. When the data points
are plotted on an impedance diagram a harmonic impedance locus is developed, such as that shown in
Fig. 4.2g.
XÊ(ohms)
Fig. 4.2g– Typical AC harmonic impedance plot for a real AC system

Note: harmonic number indicated alongside plot points
It is often advantageous to calculate harmonic impedances at intervals smaller than integer harmonics
to establish the entire range of harmonic impedance variation. Referring to Fig. 4.2g it is clear that these
intervals can often be quite small: for example, an interval of 0.1 time fundamental may be required to
4.2g
establish the detail of the impedance locus.
This impedance diagram represents one operating condition of the AC system. The study needs to be repeated
for all probable operating conditions of the system, for both present and future conditions. Clearly, such a
complicated impedance model would be difficult to use in a filter design program, especially as there may
be many hundreds of different operating conditions to be considered. Therefore it is necessary to define a
simpler harmonic model to represent the harmonic impedance of the AC system.
4.2.3.1. The Impedance Polygon Model

Based on a detailed study of the range of harmonic impedances which can exist in an AC system described
above, a simplified impedance representation can be created known as the Polygon model. In this case,
the impedance for each integer harmonic, or components close to the integer harmonic, are plotted on the
impedance plane and a polygon is drawn around the individual data points. This polygon, which is simple
to represent in a computer program, is used to define all possible impedances for a single harmonic, as
shown in Fig. 4.2h. Impedance polygons are drawn for each harmonic, or groups of harmonics of interest.
Typically, the points selected for a single harmonic should allow some tolerance to allow for uncertainties
in system parameters. At lower harmonics the uncertainties are smaller, so a range of ±5% or ±10% or ±1
harmonic is a commonly used approach. For higher harmonics there are more uncertainties, so the margins
are increased – commonly used is ±2 harmonics, ±10% up to even ±25%. As ever there is a balance between
robustness of filter harmonic performance against filter cost, complexity and vulnerability to transients.

CHAPTER
+jX
-jX
Fig. 4.2h– Impedance polygon
The polygon impedance model provides the most accurate harmonic impedance model of an AC system as
this method gives a realistic assessment of the system impedances and avoids any problems of over-designing
the filter. However, where sufficient AC system data is unavailable a more general AC system representation
can be used.
4.2h
4.2.3.2. The Individual or General Sector Model
The model is shown in Fig. 4.2i together with the typical parameter values. The boundary that encloses the
actual impedances is related to the minimum and maximum short-circuit impedances of the system. It
can be presented individually for each harmonic or more generally, one sector for a number of harmonics,
for example, one sector for harmonics n < 5, one for 5 ≤ n < 11 and one for n ≥ 11.
+jX
+jX Zmax
R2
Zmin Θ
Θ R1
R
R
ΘÊmaxÊ=Ê80¡ÊforÊÊÊÊÊÊÊÊÊÊnÊ<Ê5
Θ max = 75° for 5 ≤ n < 11
-jX Θ max = 70° for 11 ≤ n
Zmin = Zmin s.c. . √n -jX
ZmaxÊ=ÊZmaxÊs.c.Ê.Ên
Fig. 4.2i– The general sector model Fig. 4.2j– The general circle model
BACK TO 4.2j DC Transmission Systems: Line Commutated Converters | 171

CHAPTER
4.2i
4.2.3.3. The General Circle Model
The model is shown in Fig. 4.2j. Parameters R1, R2 and Q are determined from system characteristics. This
model has a close resemblance to system characteristics. However, since it is often not related to harmonic
order, it tends to cover too large an impedance area, especially for the low order harmonics. This can result
in filters that are unduly complex and excessively expensive.
4.2.3.4. The Infinite Impedance Model

This model, assuming an infinite impedance of the AC network, may be sometimes used for some
defined harmonics for performance calculations and is generally used for loss calculations [10] [11].
The performance calculations in this case will depend entirely on the filter impedance and no possible
resonances with the AC network will be simulated.
4.2.4. DC SYSTEM HARMONIC IMPEDANCE

The DC system is modeled explicitly as lines and ground impedances. This is necessary, as unlike AC filter
performance where the performance is generally assessed at a single point (or at defined points in the power
system), the requirement for DC filters is generally expressed throughout the length of the transmission corridor.
The nature of a DC transmission system (excluding back-to-back systems) is that it extends over a distance.
Consequently, the parameters of the DC system vary over this area and should be reflected in the DC system
model. The parameters affected may include the following:
➙ Ground impedance
➙ Tower configurations
➙ Line parameters
➙ Communications repeater stations
The DC system impedance is strongly dependent on the transmission line and/or cable configuration and
parameters such as the following need to be modeled appropriately.
For overhead lines:
➙ Tower configuration
➙ Conductors including shield wires
➙ Conductor bundle
➙ Line spatial arrangement including sag
➙ Ground resistivity/tower footing resistance
➙ Neutral/ground return conductors
For cables, the configuration of conductors needs to be considered:
➙ Twin conductor cable
➙ Integrated return cable
➙ Cable construction
DC filter and reactor arrangements also need to be considered and modeled. Traditionally DC reactors were
located in the HV connections to converters, but recently there have been a number of DC transmission
systems with substantial parts of the DC reactor located in the neutral connection.
On most modern DC transmission systems, the DC filters, if fitted, are connected from HV to neutral and from
neutral to ground. However, there are a number of schemes, mostly older schemes, with DC filters connected
from HV to ground. There may also be special circumstances that require filters connected from HV to ground.
The HVDC system may operate in a number of modes. These various modes result in significantly different
DC system impedances. Many modes are possible including the following:
➙ Bipole
➙ Monopole earth/ground return
➙ Monopole metallic return
➙ Parallel line/converter configurations for multiple bipole schemes
➙ Parallel monopole modes if available
CHAPTER
4.3. HVDC CONVERTER HARMONICS
At its AC terminals a HVDC converter can be considered as a source of harmonic current, with high internal
impedance. This section describes the generation of harmonic currents from the converter and the factors
which control the magnitude of these currents.
4.3.1. CONVERTER CHARACTERISTIC AC HARMONICS

For most HVDC schemes, the HVDC converter appears to the AC network as a current source, with high
internal impedance.
For back-to-back schemes with no or a very small DC smoothing reactor, the level of harmonic current
generated by the converter will be dependent on the impedance of the AC system as seen from the converter,
particularly for low order harmonics, such as 3rd harmonic.
A converter with pulse number p generates harmonic currents of order
n=p⋅k±1 k = 1, 2, 3, … [Eqn 4.3a]
These harmonics are called characteristic and they are a function of the main circuit parameters and the AC
voltage.
Harmonics of other integer orders, apart from those mentioned above, are called noncharacteristic. They
are generated mainly due to asymmetries of converter parameters and transformer reactances, as well
as unbalances in the supply voltage to the converters. These harmonics are normally of lower magnitudes
compared to the characteristic harmonics.
In some schemes, currents at frequencies that are not harmonics (not integer multiples of fundamental
frequency) may exist on the AC side of the converters. These may be caused by transfer of DC harmonics
created by one AC system, through a converter, to the other AC system. These effects are often termed inter-
harmonics or non-integer harmonics. The most common cause of such inter-harmonics is the inter-connection
of AC systems which operate at different fundamental frequencies, e.g. 50 Hz and 60 Hz.
In order to calculate the classical characteristic harmonics of the converter the following assumptions are made:
➙ The three-phase supply voltages are balanced: they are displaced exactly by one third of a cycle in time
from each other; the phases have equal magnitude, and consist only of fundamental frequency.
➙ The direct current is perfectly constant without ripple: an infinite smoothing inductance is assumed
in the DC circuit.
➙ The valves are fired at equal time intervals of one-sixth of a cycle: at a constant delay angle when
measured from the zero crossing of the respective commutating voltages.
➙ The commutation reactances are equal in the three phases: all commutation angles are the same.
Neglecting the effect of the commutating reactance, the current for a star/star connected converter
transformer feeding a 6-pulse group consists of rectangular pulses as shown in Fig. 4.3a.
The Fourier analysis of such a waveform is:
2⋅ 3  1 1 1 
i= ⋅ I d ⋅  cos ω t − cos 5ω t + cos 7ω t − cos11ω t + .... [Eqn 4.3b]
π  5 7 11 
Where:
Id = direct current.
The fundamental and the nth harmonic are then given by:
6
I1 = ⋅ Id [Eqn 4.3c]
π
I1
In = [Eqn 4.3d]
n
The current for a star/delta transformer feeding a 6-pulse group is shown in Fig. 4.3b.

CHAPTER
PhaseÊA
PhaseÊB
PhaseÊC
Magnitude
--240 -180 -120 --60 0 60 120 180 240 300 360 420 480
AngleÊ(¡)Ê➜
Fig. 4.3a– Phase currents for the star/star connected 6-pulse bridge, ignoring overlap
PhaseÊA
PhaseÊB
PhaseÊC
4.3a
Magnitude
--240 -180 -120 --60 0 60 120 180 240 300 360 420 480
AngleÊ(¡)Ê➜
Fig. 4.3b– Phase currents for the star/delta connected 6-pulse bridge, ignoring overlap

CHAPTER
The Fourier analysis of such a waveform is:
2⋅ 3  1 1 1 
i= ⋅ I d ⋅  cos ω t + cos 5ω t − cos 7ω t − cos11ω t + .... [Eqn 4.3e]
π  5 7 11 
A 12-pulse converter is built up from one star/star connected and one star/delta connected transformer,
each feeding one 6-pulse group. The transformers are connected in parallel on the AC system and the 6-pulse
groups are connected in series on the DC system. The harmonics of order:
n = 6 ⋅ k ± 1, where k = 1, 3, 5, …, [Eqn 4.3f]
are phase shifted 180° between the two transformers and are eliminated, giving:
4⋅ 3  1 1 1 1 
i= ⋅ Ιd ⋅  cos ω t − cos 11 ω t + cos 13 ω t − cos 23 ω t + cos 25 ω t ... [Eqn 4.3g]
π  11 13 23 25 
Thus, a 12-pulse converter will generate only the harmonics of the following order:
n = 12 ⋅ k ± 1, where k = 1, 2, 3, … [Eqn 4.3h]
The current for a 12-pulse group is shown in Fig. 4.3c.

A more detailed description of the calculation of characteristic harmonics can be found in reference [1].
The idealized waveforms shown above will, in reality, be modified by the reactance of the supply system:
mainly the transformer reactance. Due to this commutating reactance, the rapid transitions of current are
slowed down, making the waveform more sinusoidal. Thus, the harmonic current magnitudes are reduced
compared to those applicable to pure square wave pulses, as shown in Fig. 4.3d.
The equations given above are based on the following assumptions: firstly that the DC current is smooth,
PhaseÊA
PhaseÊB
PhaseÊC
Magnitude
--240 -180 -120 --60 0 60 120 180 240 300 360 420 480
AngleÊ(¡)Ê➜
Fig. 4.3c– Phase currents for the 12-pulse bridge, ignoring overlap

CHAPTER
4.3c
0.07
0.06
0.05
11
HarmonicÊcurrentÊ(kAÊrms)
13
0.04 23
25
35
0.03 37
47
49
0.02
0.01
0 0.2 0.4 0.6 0.8 1 1.2

DirectÊcurrentÊ(kA)
Fig. 4.3d– Typical level of characteristic harmonics for different DC power levels
that is, the DC reactor is infinite; and secondly that the AC system voltage waveforms are sinusoidal. Because
both of these assumptions are not valid for practical systems, more complex calculations are necessary and
purpose-built computer programs are used. A description of these tools is beyond the scope of this book.
4.3d
The usual published formulae and graphs for these currents give magnitudes only. For special purposes (e.g.
net harmonic contribution from two or more bridges of slightly different firing angles or reactances) both
magnitude and phase (i.e. vector solutions) are required. For a 6-pulse converter bridge having a transformer
of zero phase-shift, the harmonic current from Phase A at the AC busbar is given by:
In = An + j Bn [Eqn 4.3i]
Where:
Ι1
An = ( n sin (α + µ ) sin n (α + µ ) − sin α sin nα  +  cos (α + µ ) cos n (α + µ ) − cos α cos nα  )
nx (n 2 − 1) id
sin (α + µ ) sin n (α + µ ) − sin α sin nα  +  cos (α + µ ) cos n (α + µ ) − cos α cos nα  ) [Eqn 4.3j]
Ι1
Bn = ( n sin (α + µ ) cos n (α + µ ) − sin α cos nα  +  − cos (α + µ ) sin n (α + µ ) − cos α sin nα 
nx (n 2 − 1) id
sin (α + µ ) cos n (α + µ ) − sin α cos nα  +  − cos (α + µ ) sin n (α + µ ) − cos α sin nα  ) [Eqn 4.3k]
for n = 1, 11, 13, 23, 25, …

The sign of the arguments An and Bn is as above for a star/delta connected 6-pulse bridge for n = 5, 7, 17,
19 - but reversed for these values of n for a star/star connection.

CHAPTER
For reference:
cosa – cos (a + m) = id ⋅ x [Eqn 4.3l]
I1 = Id ⋅ √6/p (referred to valve winding turns) n = harmonic order
= fundamental RMS current magnitude (very nearly) x = per-unit commutation reactance
id = per-unit DC current a = firing angle
m = overlap angle
The currents in phases B and C are similar in magnitude, but shifted 120° or 240° at the particular harmonic
to form a positive-sequence current system for n = 1, 7, 13, etc. and a negative-sequence current system
for n = 5, 11, 17, etc.
The absolute phase of the above is based on the fundamental commutation EMF of phase A (phase-neutral)
being at zero reference angle.
As the converter firing and extinction angles affect the wave-shape of the converter current, both directly and
through their impact on the overlap angle, they have a very significant influence on the harmonic generation.
In general, higher control angles result in higher harmonic generation.
4.3.2. Converter Non-Characteristic AC Harmonics

The assumptions in the previous sections are never exactly fulfilled in reality. As a consequence, harmonics
of orders other than the characteristics will appear. These harmonics are normally of much lower magnitude
than the characteristic harmonics.
For example, the current waveform for a 12-pulse converter may display the following non-ideal conditions:
➙ Unbalance of the converter transformers
➙ Firing angle asymmetry
➙ AC voltage unbalance (negative sequence) and distortion
4.3.2.1. Reactance Imbalance between Phases

A difference in reactance between transformer phases will produce non-characteristic harmonics of odd
orders on the AC side and non-characteristic harmonics of even orders on the DC side. This phase reactance
difference is a result of manufacturing tolerances and can be of the order of 1%. The value is based on
measurements of converter transformers manufactured by GE Vernova. However, analysis will need to
consider the manufacturer’s guaranteed tolerance in calculating expected values.
4.3.2.2. Reactance Imbalance between Bridges

A difference in reactance between star/star and star/delta transformer groups will produce 6-pulse
characteristic harmonics, which in theory have been cancelled in the 12-pulse configuration.
On the AC side this means harmonics of 6 ⋅ k±1 where n is 1,3,5 – so this gives 5, 7, 17, 19 and so on.
On the DC side this means harmonics of 6 ⋅ k where n is 1,3,5 – so this gives 6, 18 and so on.
This reactance difference is a result of manufacturing tolerances and can be of the order of 1%. The value
is based on measurements of converter transformers manufactured by GE. However, analysis will need to
consider the manufacturer’s guaranteed tolerance in calculating expected values.
4.3.2.3. Difference in Converter Transformer Turns Ratio

The turns ratio of the star/star and star/delta transformers should ideally be in the ratio of 1/√3. However,
since √3 is an irrational number, this ideal can never be reached in practice. Any deviation from this ideal
1/√3 ratio will generate the theoretically cancelled characteristic 6-pulse harmonic currents on the AC
side of orders: 5, 7, 17, 19, …
On the DC side, harmonic voltages will be generated of the orders 6, 12, 18, 14, …

CHAPTER
4.3.2.4. Firing Angle Asymmetry
The deviation in firing angle between the valves is caused primarily by variations in the delay of the firing
pulse in the circuits between the control system and the valves. The level of the firing asymmetry is typically
0.01°, based on the existing control systems and valves manufactured by GE Vernova.
The firing angle asymmetry will give rise to harmonics of all orders.
4.3.2.5. AC Voltage Unbalance

Voltage unbalance produces harmonics by its influence on the firing angle and overlap in the different
phases. It will result in harmonics of order:
n=k⋅p±3 [Eqn 4.3m]
Where: k = 0, 1, 2, …
p = pulse number of the converter
The unbalance is introduced as a negative sequence voltage on the AC system.
On the DC side the effect of voltage unbalance will produce voltage harmonics of the order:
n=k⋅p±2 [Eqn 4.3n]
Where: k = 0,1,2, …
4.3.2.6. Inter-Harmonics
In schemes that connect two systems with different AC fundamental frequencies, the harmonics generated
on one side of the scheme will be transferred to the other side, creating sources of current at frequencies
that are not harmonics of the fundamental frequency. These frequencies are usually named ‘inter-
harmonics’ or ‘non-integer harmonics’. The presence of a large DC smoothing reactor or transmission line
can reduce, but not completely eliminate, these inter-harmonics. The order of these non-integer harmonics
may be calculated according to the following formula:
(nA ± 1) ⋅ fA ± fB
nB = [Eqn 4.3o]
fB
Where: nA = order of harmonic current generated in the AC system A
nB = order of harmonic current generated in the AC system B
fA = fundamental frequency of the AC system A
fB = fundamental frequency of the AC system B
These transferred characteristic harmonics can be of a significant magnitude (typically 10 – 20% of the
characteristic integer harmonics). The transferred non-characteristic harmonics are normally of much lower
magnitudes.
The same phenomena may occur when connecting two non-synchronous systems with the same nominal
fundamental frequency, when the actual frequencies of the two systems may differ substantially (by 1 to 2 Hz).
4.4. CONVERTER DC HARMONICS

For most HVDC schemes, from the DC side the converter can be seen as a voltage source with low internal
impedance.
The idealized voltage across an unloaded converter operating at a = 0° is shown in Fig. 4.4a for a 6-pulse
converter and in Fig.4.4b for a 12-pulse converter. The voltage is a mix of a direct voltage and harmonics.
Table 4.4a shows the DC voltage harmonics produced by a 6-pulse converter, relative to the no-load DC
voltage (Vdo).

CHAPTER
1 1 1 1
0.8 0.8 0.8 0.8

DCÊvoltageÊ(p.u.)
DCÊvoltageÊ(p.u.)
DCÊvoltageÊ(p.u.)
DCÊvoltageÊ(p.u.)
0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0 0 0 0
0 0 100 100 200 200 300 300 0 0 100 100 200 200 300 300
ElectricalÊdegrees
ElectricalÊdegrees ElectricalÊdegrees
ElectricalÊdegrees
Fig.4.4a– The idealized voltage across the DC terminals of a Fig. 4.4b– The idealized voltage across the DC terminals of a
6-pulse bridge 12-pulse bridge
Harmonic Frequency Voltage (p.u.)

4.4a4.4a 4.4b4.4b
0 0 1.0000
6 300 0.0404
12 600 0.0099
18 900 0.0044
24 1200 0.0025
Table 4.4a– Idealized DC voltage harmonics at the terminals of a 6-pulse bridge
4.4.1. 3-PULSE MODEL

A 6-pulse converter should be modeled as a harmonic voltage
source. Each 6-pulse bridge comprises two 3-pulse voltage L3p
sources phase shifted by 60°. The 3-pulse bridge generates
harmonics of order:
n = 3 ⋅ k [Eqn 4.4a] Csh
where k = 1, 2, 3, …
L3p
These are the characteristic harmonics for a 3-pulse group
and they may be calculated as a function of the main circuit
parameters.
A 12-pulse converter consists of two 6-pulse converters where
the supply voltages of the two bridges have been phase-shifted L3p
by 30° with respect to each other. If the supply voltage of the
upper 6-pulse bridge is chosen as a reference, the commutation
voltage of the lower bridge will be phase shifted by 30°
Csl
electrical. This is done by connecting one 6-pulse bridge to a star
(Y)-connected transformer, while the other bridge is connected
L3p
to the delta (Δ)-connected transformer. The complete 12-pulse
bridge is represented by the circuit according to Fig. 4.4c below.
This configuration can be reduced to the configuration shown
in Fig. 4.4d.
The following sections describe the main processes for the Fig. 4.4c– 3-pulse representation of
calculation and choice of the harmonic voltages used for the a 12-pulse converter (where L3p is
the equivalent 3-pulse commutating
calculation of performance and steady-state rating of the DC filters. inductance).

CHAPTER
4.4c
4.4.2. CHARACTERISTIC DC HARMONICS
In order to calculate the characteristic harmonics of a converter, the following assumptions are made:
➙ The supply voltages are symmetrical and with perfect sinusoidal shape
➙ The direct current is perfectly smooth, without ripple
➙ The valves are fired at equal time intervals: i.e. at a constant delay angle from the zeros of the
respective commutating voltages
➙ The commutation impedances are equal in the three phases
The 3-pulse voltage U3P(t) is given by:
U di 0 U ∞
U 3P (t ) = ⋅ ( cos α + cos δ ) + di 0 ⋅ ∑ (−1)k ⋅ ( a3 k cos 3kω t + b3k sin 3kω t ) [Eqn 4.4b]
4 4 k =1
Where: Udi0 = actual ideal no load direct voltage
a = firing angle
d = a + µ: where µ is the overlap angle
cos α (1 − 3k )  + cos δ (1 − 3k )  cos α (1 + 3k )  + cos δ (1 + 3k ) 
a3 k = + [Eqn 4.4c]
1 − 3k 1 − 3k
sin α (1 + 3k )  + sin δ (1 + 3k )  sin α (1 − 3k )  + sin δ (1 − 3k ) 
b3k = − [Eqn 4.4d]
1 + 3k 1 − 3k
The other 3-pulse voltages given in Fig. 4.4d can be calculated
directly from the formulae above.
The formulae above are valid both for inverter and rectifier L3p
operation.
The sum of the four 3-pulse voltage sources is equivalent to
a 12-pulse voltage source generating voltages of frequencies
that are integer multiples of the twelfth harmonic, as the other Csh
characteristic harmonics of the 3-pulse bridges (those of order
3, 6, 9, 15, etc.) will be cancelled if the impedances of each 2Ê⋅ L3p
3-pulse group are equal. When taking into consideration the
stray capacitances between the different valve groups and the
ground, the 3-pulse voltages created in the ground mode, will
not be cancelled but will have to be considered in the design of Csl
the DC filters.
The characteristic harmonic voltages for the performance L3p
calculations will be chosen differently for different modes of
operation.
4.4.2.1. Monopolar Operation

Consistent sets of harmonics for different main circuit parameters Fig. 4.4d– Simplified 3-pulse
are chosen in order to ensure that the performance will be met representation of a 12-pulse
with the harmonic set giving the highest harmonic currents on converter
the line. 4.4d
4.4.2.2. Normal Bipolar Operation
Under ideal balanced bipolar operation no ground mode current will be generated, giving no or very low
level of induced voltage in the test line. However, if there are differences in the main circuit parameters
in the two poles, an asymmetry between the poles is obtained and as a consequence, a non-zero ground
mode driving voltage is obtained. Where no electrode line runs parallel with the pole lines, the ground
mode driving voltage Ug is defined as:

CHAPTER
U p1 + U p 2 [Eqn. 4.4e]
Ug =
2
Where: Up1 = is the driving voltage of pole 1
Up2 = is the driving voltage of pole 2
The ground mode component of the driving voltage is the source of the ground current along the line and
consequently of the induced noise in the telephone lines.
The cases investigated in search of the highest ground mode component of the driving voltage should be
calculated by applying the biggest possible differences in main circuit parameters that can occur in the system
during the steady-state operation. The performance will be calculated for all the different main circuit sets of
parameters and for all the power levels.
4.4.2.3. Bipolar Operation with Reduced Voltage or with Increased Reactive

Power Absorption
For both these modes of operation, the firing angles will be higher than for the normal bipolar operation,
causing higher values of the generated harmonic voltages. In each of these cases, the possible differences
in main circuit parameters between the poles have to be investigated in order to find the highest ground
mode components of the driving voltages.
4.4.3. NON-CHARACTERISTIC DC HARMONICS

The assumptions in the previous section are never exactly fulfilled in reality. As a consequence, harmonics
of orders other than the characteristic will appear. These harmonics are normally of much lower magnitude
than the adjacent characteristic harmonics.
Considering, for example, the current waveform from a 12-pulse converter the following non-ideal issues
may occur:
➙ Unbalance of converter transformer
➙ Firing asymmetry
➙ AC voltage unbalance (negative sequence) and distortion
4.4.3.1. Reactance Imbalance between the Phases

A difference in reactance between transformer phases will produce noncharacteristic harmonics of odd
orders on the AC side. This difference is a result of manufacturing tolerances and can be of the order of 1%.
The value is based on measurements of converter transformers manufactured by GE Vernova.
4.4.3.2. Difference in Converter Transformer Turns Ratio

The difference in turns ratio between Y/Y and Y/Δ transformers will generate the theoretically cancelled
characteristic 6-pulse harmonics: 6,18,30, …
4.4.3.3. Firing Asymmetry

The deviation in firing between the valves is caused primarily by variations in the delay of the firing pulse
in the circuits between the control system and the valves. The level of the firing asymmetry is assumed to
be 0.01° according to the characteristics of the existing circuits and valves manufactured by GE.
The firing angle asymmetry will give rise to harmonics of all orders.
4.4.3.4. AC Voltage Imbalance

Voltage imbalance produces harmonics by its influence on the firing angle and overlap in the different
phases. It will result in harmonics of order:
n=k⋅p±2 [Eqn. 4.4f]
Where: k = 0, 1, 2, …
The imbalance is introduced as a negative sequence voltage.

CHAPTER
4.4.4. COMBINATION OF CHARACTERISTIC AND NON-CHARACTERISTIC VOLTAGES
For monopolar operation, the sets of harmonics are obtained by combining the characteristic harmonic
voltages together with the non-characteristic calculated as described above. These sets are calculated
separately for the rectifier and the inverter stations and applied for all the different combinations of main
circuit parameters and power levels.
For bipolar operation, a maximum possible inequality of the main circuit parameters will be applied in order
to create the highest possible ground mode generated voltage and consequently, the highest possible ground
mode current along the DC line.
For the non-characteristic harmonics, in addition to the differences in magnitude due to the different power
levels between the poles, a phase shift of 90° between the vectors of the two poles is introduced.
The only exception is the 2nd harmonic, generated mostly by the AC system unbalance which will influence
this harmonic in the same way in both poles. For this harmonic, a reasonable phase shift between the two
poles can be in the range 5 - 25°.
4.5. AC HARMONIC FILTERS

Harmonic filters are required at the AC terminals of the HVDC converter to limit harmonic distortion on
the AC transmission system. This section describes the design criteria for the performance of the filters
and equipment rating considerations. A variety of commonly used filters for HVDC projects are described
in detail.
4.5.1. DESIGN DATA

The following data is required to design the AC harmonic filters:
➙ The main circuit data for the calculation of the harmonic currents injected into the AC network
➙ Requirements for different operating conditions
➙ Information on the AC system harmonic impedance
➙ Performance requirements
➙ Evaluation of costs for losses
4.5.1.1. HVDC Scheme Operating Conditions

The filters must be dimensioned to operate according to the specifications for all possible operating
conditions of the HVDC converters. A possible requirement would be operation at reduced DC voltage or
operation at increased firing angles. These requirements may by imposed by the customer or chosen by
the main circuit designer. However, all credible operating conditions have to be investigated in order to
find the worst case with respect to filter performance and rating.
4.5.1.2. AC System Harmonic Impedance

As discussed in section 4.2 above, information on the AC network impedance at harmonic frequencies, as
seen from the converter terminals, is needed for AC filter performance and rating calculations.
Normally, the customer defines the range of network impedance to be used for the filter design, but in some
HVDC projects, the customer has left it to the supplier to make their own estimate.
When calculating the interaction between the AC filters and the AC network impedance, the requirement
from the customer is often to choose for each combination of filters, and at each frequency, the worst
network impedance in such a way that resonance between the filters and the network impedance is
created. See Appendix A4.5 for more details.
4.5.1.3. Loss Evaluation

A factor which may influence the filter configuration to some extent is the evaluation of the filter component
losses. If the losses are to be evaluated as comparatively high, the filter should be designed with high q
(quality)-factor (sharply tuned). If the loss evaluation rate is low, less sharp filters with lower q-factor can
be used. This could lead to a simpler filter configuration and therefore to a less costly overall design.

CHAPTER
4.5.2. FILTER PERFORMANCE REQUIREMENTS
The performance requirements for AC filters are usually specified by harmonic voltage distortion,
telephone interference and harmonic current criteria. The definitions of the most commonly used harmonic
performance criteria, together with the typical values of performance limits, are presented in the following
sections.
4.5.2.1. Voltage Distortion

There are a number of ways to calculate voltage waveform distortion and the relevant equations are
presented in this section.
• Individual Harmonic Distortion Dn (%)
This calculates the magnitude of a single harmonic component.
Un
Dn = ⋅ 100% [Eqn 4.5.2a]
U1
Where:
U1 = the line to neutral nominal fundamental frequency system voltage (RMS)
Un = the nth order harmonic line to neutral voltage appearing at the bus under consideration
A typical limit of Dn is 1%. Different limits may be applied for different types of harmonics, for example:
➙ 3rd, 5th, 7th harmonic 1.5%
➙ characteristic harmonics 1%
➙ other odd harmonics 0.5%
➙ other even harmonics 0.25%
• Total Harmonic Distortion THD or Deff (%)
This provides a measure of the effective harmonic distortion, by evaluation of the square root of the sum of
the squares (known as RSS) of the individual harmonic components.
N
THD = ∑D 2
n [Eqn 4.5.2b]
2
Where N is the maximum harmonic order considered

Typically specified limits of THD are in the range of 1.5% – 3%.
• Total Arithmetic Harmonic Distortion (D)
This gives an additional measure of the waveform distortion, where all of the harmonic components are
added arithmetically.
N
D = ∑ Dn [Eqn 4.5.2c]
2
Typically, specified limits of D are in the range of 3% – 5%.
4.5.2.2. Telephone Interference

The electromagnetically induced noise in telecommunication lines is estimated by means of weighted
quantity, taking into account the frequency sensitivity of the human ear and the telephone equipment,
but excluding the geometrical aspect of coupling (the study of the effect of the intersection of power
transmission lines and telecommunication lines).
The criteria are presented in two main categories: those commonly in use in countries following European
practices and those commonly in use in countries following North American practices.
• Criteria According to European Practices
a) Telephone Harmonic Form Factor, THFF, as defined by CCITT (Comité Consultatif International
Téléphonique et Télégraphique)

CHAPTER
2 [Eqn 4.5.2d]
U
N

THFF = ∑  Un ⋅ Fn 
1
Where:
Un is the component at harmonic n of the disturbing voltage
N is the maximum harmonic number to be considered
U is the line to neutral total RMS voltage and calculated by:
N
U= ∑U 2
n [Eqn 4.5.2e]
1
pn ⋅ n ⋅ f0
Fn = [Eqn 4.5.2f]
800
pn is the psophometric weighting factor (see Appendix A4.2 and [16])
f0 is the fundamental frequency
The required limit of THFF for HVDC schemes is typically around 1%.
It should be noted that THFF is the only voltage criterion that refers the harmonic voltage level to the level
of RMS voltage. All other criteria refer to the level of fundamental voltage.
b) Equivalent disturbing current Ip is defined according to CCITT, by:
1
Ip =
p800
⋅ ∑ (h f ⋅ p f ⋅ I f )2 [Eqn 4.5.2g]
f
Where:
If is the component at frequency f of the current causing the disturbance
pf is the psophometric weighting factor at frequency f
hf is a factor which is a function of frequency and which takes into account the type of coupling between
the lines concerned. By convention h800 = 1 (h800 is the coupling factor at 800 Hz)
• Criteria According to North American Practices
c) Telephone Interference Factor, TIF, as defined by IEEE 519 [2]:
2
N
U 
TIF = ∑  Un ⋅ Wn  [Eqn 4.5.2h]
1 1
Where:
Un is the single frequency RMS voltage at harmonic n
U1 is the fundamental line to neutral voltage (RMS)
Wn = Cn ⋅ 5 ⋅ n ⋅ f0 is the single frequency TIF weighting at harmonic n

Cn is the C-message weighting factor (see Appendix A4.2 and [15])
n is the harmonic order
f0 is the fundamental frequency
Typical requirements of TIF are between 15 and 50.
d) IT, a weighted current product, as defined by IEEE 519:
N
IT = ∑ (I n ⋅ Wn )2 [Eqn 4.5.2i]
1
Where:
In is the single frequency RMS current at harmonic n
Wn = Cn ⋅ 5 ⋅ n ⋅ f0 is the single frequency TIF weighting at harmonic n

CHAPTER
The IT criterion appears less frequently in specifications because of the difficulty in specifying reasonable
performance limits, especially where multiple transmission lines supply the converter station.
Typical requirements of IT are between 15,000 A and 50,000 A per line.
4.5.2.3. Other Performance Criteria

For converter stations with closely connected generators or synchronous machines, there may be a
requirement to limit the low order generator currents to a specific value in order to prevent overheating
of machine windings.
4.5.3. FILTER TYPES

This section relates to the design of passive AC filters. On rare occasions, there may be special reasons
such as lack of space or very limited system reactive power capability for active filters to be considered,
despite their added cost and complexity.
The harmonic filters at the AC side of a HVDC converter serve two purposes:
➙ To reduce the harmonic distortion appearing in the AC network to acceptable levels
➙ To supply all or part of the reactive power demand of the converters
Two main filter types are used today:
➙ The tuned filter or band-pass filter which is sharply tuned to one or several harmonic frequencies
These are filters tuned to a specific frequency, or frequencies. They are characterized by a relatively high
q factor: i.e. they have low damping (see Eqn 4.5.3d). The resistance of the filter may be in series with the
capacitor and inductor (more usually it is simply the loss of the inductor), or in parallel with the inductor,
in which case the resistor is of high value. Examples of tuned filters are discussed later in this section and
include single (e.g. 11th), double (e.g. 11/13th) and triple (e.g. 3/11/13th) tuned types.
➙ The damped filter or high-pass filter offering a low impedance over a broad band of frequencies
These are filters designed to damp more than one harmonic: for example, a filter tuned at 24th harmonic
will give low impedance for both 23rd and 25th harmonic and for most higher-order harmonics. These are
characterized by a relatively low q factor, i.e. they have high damping (see Eqn 4.5.3k). Damped filters always
include a resistor in parallel with the inductor, which produces a damped characteristic at frequencies above
the tuning frequency. Examples of damped filters are discussed later in this section and include single-tuned
damped high pass (e.g. HP 12) and double-tuned damped high pass (e.g. HP 12/24).
The distinction between tuned and damped filters may become
somewhat unclear depending on the choice of q-value for different
filter frequencies. A discussion on this will follow in subsequent
C
sections.
For a HVDC scheme with a 12-pulse converter the largest
characteristic harmonics will be the harmonics: 11th, 13th, 23rd,
25th, 35th, 37th, 47th and 49th. As the levels of the 11th and 13th L
harmonic are generally twice as high as the rest of the harmonics,
a normal practice is to provide band-pass filters for the 11th and
13th harmonics and high-pass filters for the higher harmonics.
Due consideration must be also taken to the possible low order R
resonance between the AC network and the filters and shunt
banks. When a large rating HVDC scheme is to be installed in a
weak AC system, a low order harmonic filter (most often tuned to
3rd harmonic) may also be needed. A discussion on this will follow
in subsequent sections.
Fig. 4.5a– Single-tuned band-pass
filter, circuit
BACK TO
CHAPTER 4.7a Line Commutated Converters
DC Transmission Systems: | 185
1.0E+05 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
0
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Phase HarmonicÊnumber
Fig. 4.5b– Single-tuned band-pass filter, impedance characteristic
4.5.3.1. Tuned Filters

This section considers a family of filters, which are sharply tuned to provide the maximum attenuation
of individual harmonic components. The design details plus the advantages and disadvantages of these
types of filters are described.
• Band-pass Filter
4.7b filter consists of an LC series resonance circuit shown in Fig. 4.5a, where the resistance R is
A band-pass
normally small and represents only the stray resistance of C and L. Its impedance characteristic in terms
of magnitude (solid line) and phase (dashed line) is shown in Fig. 4.5b, according to the formula:
 1 
Z BP = R + j  ω L −  [Eqn 4.5.3a]
 ωC 
The resonance frequency of this filter is:
1
ωn = [Eqn 4.5.3b]
LC
Defining a relative frequency deviation:
ω − ωn
δ= [Eqn 4.5.3c]
ωn
and the filter quality factor at resonance frequency:
L
ωn ⋅ L
= C
[Eqn 4.5.3d]
q=
R R
The filter impedance can be now expressed as:
 2 +δ 
Z BP = R  1 + j ⋅ q ⋅ δ ⋅  [Eqn 4.5.3e]
 1+δ 
For small frequency deviations, d << 1, the impedance can be approximated with:
ZBP = R(1 + j ⋅ 2 ⋅ d ⋅ q) [Eqn 4.5.3f]

CHAPTER
The band-pass of the filter is defined by:
1
PB = ± ωn
2⋅q C1
[Eqn 4.5.3g]
Typical values of q for band-pass filters are in the

range of 50 to 100.
L1 R1
The advantages and the disadvantages of a single-
tuned band-pass filter are as follows:
Advantages:
➙ Simple connection with only two
components
➙ Optimum damping for one harmonic
C2 L2 R2
➙ Low losses
➙ Low maintenance requirements
Disadvantages:
➙ Multiple filter branches may be needed for
different harmonics
➙ Sensitive to de-tuning effects
Fig. 4.5c– Double-tuned band-pass filter, circuit
➙ May require possibility of adjusting reactors
or capacitors
• Double-Tuned Band-pass Filter

A double-tuned band-pass filter has the equivalent function of two single-tuned filters. The configuration
is shown in Fig. 4.5c and the impedance plot in Fig. 4.5d. 4.7c
The advantages of the double-tuned filter are:
➙ At high system voltage, it is easier to optimize one larger main capacitor than two small ones.
➙ Only one reactor is exposed to the full impulse voltage as the lower reactor is connected in parallel
with the capacitor. In two single-tuned filters, the reactors in both filters are exposed to the full
impulse voltage.
1.0E+05 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5d– Double-tuned band-pass filter, impedance characteristic

CHAPTER
In a symmetrical double-tuned band-pass filter with
series resonance frequencies w1 and w2, the parallel
C1
resonance frequency can be calculated as:
ω r = ω1 ⋅ ω 2 [Eqn 4.5.3h]
By choosing the parallel resonance frequency L1 R1
closer to one of the series resonance frequencies
or by changing the MVAr split, any combination can
be achieved between the two series resonance
frequencies. This allows the possibilit y of
C2 L2 R2
incorporating a very low MVAr rated filter, which
on its own would be an uneconomical design, into
a larger filter to form a double-tuned filter. If two
single-tuned branches were used instead, there
could be a minimum filter size problem due to C3 L3 R3
connecting a possibly very low MVAr rated filter (as
required for one of the two frequencies) on to an
HV busbar.
The equations to solve the filter are given in
Appendix A4.3. Fig. 4.5e– Triple-tuned band-pass filter, circuit
The advantages and the disadvantages of a double-

tuned band-pass filter are as follows:
Advantages:
1.0E+05 4.7e 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
0
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5f– Triple-tuned band-pass filter tuned to 3rd, 11th and 24th harmonic, impedance characteristic
➙ Optimum damping for two harmonics

➙ Lower loss than for two single-tuned branches
➙ Only one HV capacitor and reactor needed to filter two harmonics
➙ Mitigates minimum filter size problem for a low magnitude harmonic
➙ Fewer branch types, facilitating filter redundancy
4.7f
CHAPTER
Disadvantages:
➙ Sensitive to de-tuning effects
➙ May require possibility of adjusting reactors
or capacitors C1
➙ Complex interconnection, with 4 or 5 C-L-R
components
➙ Requires two arresters to control insulation
levels
• Triple-Tuned Band-Pass Filter L1 C2
This type of filter is electrically equivalent to three
parallel-connected tuned filters, but is implemented
as a single combined filter. Fig. 4.5e shows the
circuit arrangement and Fig. 4.5f the impedance/
frequency response for a typical triple-tuned filter. R1
The use of triple-tuned filters could improve the

operational requirements for reactive power
control. This would be of particular importance at
low load conditions if a 3rd harmonic filter is needed
in the circuit from the beginning. Where low levels
of TIF and IT are specified these filters may achieve Fig. 4.5g– Single-tuned band-pass filter with
the required performance levels. As they are similar parallel capacitor circuit
in nature to double-tuned filters, their merits and
drawbacks are as described above.
1.0E+05
4.7g 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5h– Single-tuned band-pass filter with parallel capacitor, impedance characteristic
The equations to solve the filter are given in Appendix A4.3.

• Band-pass Filter with Parallel Capacitor
For each of the filter types (single, double or triple-tuned band-pass filters), a variation can be created by
connecting a capacitor in parallel with the bottom part of the filter (at the LV terminal of the main capacitor).
4.7h
CHAPTER
By doing this, a capacitive path will be created from the top to the
bottom of the filter, giving low impedance for frequencies above the
highest series resonance frequency. The disadvantage of this solution
is creating an extra parallel resonance frequency and around it a C
high impedance frequency span. Fig. 4.5g shows a circuit of a single-
tuned band-pass filter with parallel capacitor and Fig. 4.5h shows the
impedance characteristic of this filter.
As can be seen in the impedance diagram, the impedance of the
L R
filter is higher than the single-tuned 12th harmonic filter in harmonic
region 14th to 17th, but lower for harmonic frequencies > 17th.
The equations to solve the filter are given in Appendix A4.3.
The advantages and the disadvantages of a band-pass filter
with parallel capacitor in comparison to one without the parallel
capacitor are as follows:
Advantages:
➙ Low impedance for frequencies above the highest parallel
frequency
Disadvantages:
➙ More complex design with only one extra component Fig. 4.5i– High-pass filter, circuit
➙ Extra parallel frequency n p gives a high impedance
characteristic in the frequency range close to this parallel
frequency np
4.5.3.2. Damped Filters

This section considers a family of filters which are broadly tuned to provide attenuation over a range of
harmonic components. The design details plus the advantages and disadvantages 4.7i
of these types of filters
are described.
1.0E+05 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5j– High-pass filter, impedance characteristic

CHAPTER
• High-pass Filter
The high-pass filter provides low impedance over a wide band of frequencies. Normally, a 2nd order filter
is used (see Fig. 4.5i).
The impedance of such a filter is given by:
1 1
Z HP = +( )
jω C 1 1 [Eqn 4.5.3i]
+
R jω L
The resonance frequency is normally chosen close to the first pair of characteristic harmonics for which
band-pass filters are not provided, in order to optimize filter performance. For example, if there are band-
pass filters for 11th and 13th harmonics, a high-pass filter tuned to 24th harmonic will provide low impedance
for the characteristic harmonics 23rd and 25th.
The impedance characteristic of a high-pass filter, in this case at 3rd harmonic is shown in Fig. 4.5j.
The resonance angular frequency of the filter can be calculated by:
q2 + 1 1
ωr = ⋅ [Eqn 4.5.3j]
q2 LC
Where:
R
q= is the quality factor of the filter. [Eqn 4.5.3k]
ωr ⋅ L
High-pass filters are normally built with low q-values, typically in the range 2-10, in order to provide
damping for a large number of harmonics.
In some applications, only high-pass filters have
been used. The configuration would then be:
HP12 + HP24 (High-pass at 12th harmonic and high- C1
pass at 24th harmonic).

However, the attenuation achieved by a damped
C2
filter may be less than that achieved by two tuned
arms of the same total rating: for example, an 11th
and a 13th single-tuned filter. Thus, a larger installed L1 R1
MVAr rating of filter may be required to achieve the
same level of harmonic performance. By adding the
resistor, filter losses have been increased both at
harmonic frequencies where it is needed and at
fundamental frequency where it is not. These higher
losses can be prohibitive if the cost of losses is high,
especially for a filter designed to attenuate the 11th
and 13th harmonics.
The advantages and disadvantages of high-pass
filters are as follows:
Fig. 4.5k– 3rd order high-pass filter, circuit
Advantages:
➙ Provide attenuation over a spectrum of
harmonics
➙ Introduce damping in the AC network and filter circuit
➙ Are relatively insensitive to de-tuning effects
4.7k
➙ Because a wide spectrum of harmonics is filtered, it is not required to divide the filter bank into
several separate branches
➙ Non-characteristic harmonics are also filtered
➙ The need for tuning on site is reduced
➙ Maintenance is reduced, due to only three filter elements in comparison to multiple filter elements
in several tuned filter branches
CHAPTER
1.0E+05 90
1.0E+04
45
MagnitudeÊ(ohms) 1.0E+03
PhaseÊ(¡)
0
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5l– 3rd order high-pass filter, impedance characteristic
Disadvantages:
➙ May require larger installed MVAr rating than
multiple tuned branches
➙ Losses are higher than tuned filters
C1
4.7l
• 3rd Order Damped Filter
In this topology, an auxiliary capacitor (C2) is C2
connected in series with the resistor to act as a
blocking impedance, as shown in Fig. 4.5k. The main
application of such a filter would be at low harmonic L1 R1
orders where the losses in a 2nd order filter resistor
would be unacceptable. The impedance/frequency
response of such a filter at 3rd harmonic is shown
in Fig. 4.5l.
At fundamental frequency C2 has a high impedance,
thus reducing fundamental losses in the resistor
as current preferentially flows through the reactor.
At higher frequencies, as the impedance of bank
C2 decreases, harmonic current flows through R1,
providing the required damping. The choice of C2 is Fig. 4.5m– C-type high-pass filter, circuit
essentially a question of economics, as the reduced
power dissipation and hence lower cost of the resistor, plus the reduced capitalized losses, must cover the
costs of the C2 bank. The presence of the C2 bank slightly degrades the filter admittance characteristic,
thus a slightly larger MVAr rating may be needed to maintain performance.
To summarize, the advantages and disadvantages of the high-pass filter described above apply, plus:
Advantage: 4.7m
nd
➙ Lower fundamental frequency losses in the resistor than 2 order damped design

CHAPTER
1.0E+05 90
1.0E+04
45
MagnitudeÊ(ohms)
1.0E+03
PhaseÊ(¡)
0
1.0E+02
-Ê45
1.0E+01
1.0E+00 -Ê90
0 4 8 12 16 20 24 28 32 36 40 44 48
Magnitude
Fig. 4.5n– C-type high-pass filter, impedance characteristic
Disadvantages:
➙ Slightly poorer performance compared with 2nd order damped design
➙ More complex filter, with four C-L-R components
• C-type Filter
Here, an4.7n
auxiliary capacitor (C2) is connected in series with the reactor and is tuned to form a fundamental
frequency bypass of the resistor. Fig. 4.5m shows the circuit arrangement and Fig. 4.5n the impedance/
frequency response for a typical C-type filter.
By creating a tuned filter section C2-L1 within a 2nd order filter, virtually all fundamental current is excluded
from the resistor. At frequencies above the fundamental, harmonic current flows through R1, thus achieving
the desired damping. As C2-L1 is a tuned filter, de-tuning can occur due to variations in L or C values from
the rated values. However, in this case the effect of de-tuning is to increase the resistor rating rather than
degrade overall filter performance. The presence of the C2 capacitor has a small effect on the impedance
characteristic.
To summarize, the advantages and disadvantages of the high-pass filter described above apply, plus:
Advantage:
➙ Negligible fundamental frequency loss in resistor
Disadvantages:
➙ Resistor rating is very susceptible to de-tuning effects
➙ May require possibility of adjusting reactors or capacitors
➙ Is a more complex filter, with four C-L-R components
➙ Gives slightly poorer performance compared with 2nd order damped design
4.5.4. INITIAL DESIGN

In the initial stages of AC filter design, some basic information is collected and the main strategy for the
filters is determined.

CHAPTER
The requirements for reactive power compensation, together with the maximum allowed voltage step at bank
switching, decides the maximum bank size and the number of banks.
With the calculated harmonic currents and the harmonic performance requirements, the required filter
impedances can be calculated for major characteristic harmonics and for some typical operating conditions.
This information together with the information about suitable bank sizes is used to assume a filter solution.
The q-value of the filter is decided based on the maximum detuning of the filter and on performance
requirements.
Since the most expensive part of the filter is the main capacitor (sometimes called High Voltage capacitor or
HV capacitor), special care has to be taken to optimize the size of this capacitor with respect to the maximum
voltage to which it will be subjected.
During the final design stage the filter branch sizes, q-values and other characteristics may have to be changed
to some extent in order to fulfill the performance requirements for all the possible operating conditions.
4.5.4.1. Filter Bank Size

The sizes of the filter banks are determined by the reactive power requirements for the converter station
together with the maximum allowable voltage step at bank switching. This is treated in sections 2.3.1 and
2.3.3.
One filter bank can be divided into several branches built of tuned or damped filters tuned to different
frequencies. The minimum size of a filter branch is related to the possibilities of optimization of the HV
capacitor at a given voltage level. The distribution of MVAr between the different filter branches within
a bank is initially based on the estimation of the required filter impedance. For further information, see
Appendix A4.4.
4.5.4.2. Filter Detuning

The resonance frequency of the filter will deviate from the nominal value because of the following reasons:
➙ Variations of power system frequency resulting in proportional changes in the harmonic frequency
➙ Changes in filter component values due to temperature variations
➙ Changes in filter component values due to failures and aging
➙ Initial mistuning due to manufacturing tolerances
For the single-tuned band-pass filter, the detuning can be expressed as an equivalent frequency deviation
given by the following formula:
∆f 1  ∆C ∆L 
δ ekv = +  +  [Eqn 4.5.4a]
f0 2  C L 
Where:
∆f
is the AC system frequency deviation
f0
∆L1 ∆L1
∆LÊtot
Fig. 4.5o– Filter reactor with taps

CHAPTER
∆C
is given by the capacitance manufacturing tolerance, capacitance variation with temperature
C
changes and capacitance changes due to failed elements and aging
∆L
is given by the reactor manufacturing tolerance
L
In order to decrease the influence of manufacturing tolerance of both capacitors and reactors, the
possibility of adjusting reactors or capacitors may be introduced. The most commonly used solution
is to make off-line taps on one of the reactor terminals, to be able to choose the inductance giving the
correct value of resonance frequency for any filter branch. If the reactor is equipped with taps, then
∆C ∆L
will not include the manufacturing tolerance and will be half of the step of the taps, plus the
C L
error in the measurement of the correct resonance frequency.
This tuning of the filters, with the use of taps or other adjusting devices, will be a part of the pre-commissioning
process. The option of taps on the reactor is fairly inexpensive, if the reactor can be manufactured in one layer
– that is, if the reactor current does not exceed a value of approximately 100 – 200 A. The choice of equipping
the reactors with taps should be based on the optimization, taking into account on one side lower detuning
of the filters and on the other side the higher price of the reactor with taps.
Fig. 4.5oshows schematically the principle of using taps.
Where:
ΔL 1 is the difference of inductance between two taps of the reactor. The value used to calculate
∆L1
equivalent frequency deviation is: ∆L =
2
ΔLtot is the total range of the taps needed to compensate for the manufacturing tolerance for both the
capacitors and the reactors.
For other filter types, Eqn 4.5.4a above gives only an approximation of the equivalent frequency deviation.
The accurate value of the equivalent frequency deviation can be calculated for each specific frequency
using the following formula:
∆f ∆C ∆C2 ∆L ∆L
δ ekv = + δ C1 1 + δ C 2 + δ L1 1 + δ L 2 2 + ... [Eqn 4.5.4b]
f0 C1 C2 L1 L2
Where:
dC1, dC2, dL1 and dL2 are factors showing how the element changes influence the specific resonance frequency
of the filter. For the band-pass filter, these factors are equal to 0.5, for other filter types the factors may
vary for each element and resonance frequency.
4.5.4.3. Quality Factor

The choice of the quality factor of the filter should be made based on the following criteria:
➙ The quality factor of the filter should be high enough to achieve the required attenuation of the tuned
filter harmonic
➙ The quality factor of the filter should be low enough to avoid dangerous resonance with the AC network
impedance at maximum detuning of the filter
Typical ranges of quality factors for different filter types can be found in section 4.5.3.
4.5.4.4. Low Order Resonance

For HVDC schemes transmitting high power and connected to a weak AC network, the problem of
resonance at low harmonic order may occur. The low order impedance of the network can be estimated
by the short-circuit power Sk of the AC system. After connection of all or part of the AC filters, there will be
a magnitude of Q MVAr connected in parallel with the network impedance.
The resonance frequency can be estimated with:
Sk min
n= [Eqn 4.5.4c]
Q
CHAPTER
Where:
Sk min is minimum short-circuit power of the AC network
Q is maximum installed reactive power (filters and shunt capacitors)
Note that if n ≤ 3 – 3.3, there may be a need for a 3rd harmonic filter in the system.
4.5.5. STEADY-STATE DESIGN

Steady-state design is the dimensioning for steady-state events that can last longer than about one
minute. The filter branch sizes, resonance frequencies and q-factors are adjusted to achieve optimum filter
performance, minimum losses and an economically reasonable layout of the filter components (mainly HV
capacitors).
The filter components are rated with the purpose of finding the highest stresses that can occur in the filter
during steady-state conditions.
The losses, both fundamental frequency and harmonic, are calculated for the final solution.
4.5.5.1. Performance
The definitions of the performance quantities relevant for filter design together with typical requirements
are given in section 4.5.2. The filter performance has to be calculated for all specified operating conditions
and the complete load range, including any continuous overload conditions. This implies that several sets
of harmonic currents generated by the converter have to be calculated for different operating conditions
and loads. To improve performance, the branch sizes, q-factors and resonance frequencies of the filters
can be modified.
4.5.5.2. Steady-State Rating

When the final filter solution is obtained with respect to performance, the next step is to calculate the
rating of the filter components. The purpose is to find the decisive stresses on the filter components during
steady-state conditions. This is done by calculating the stresses of each component for various cases with
different conditions regarding operating mode, transmitted power, etc. For each individual component, the
highest decisive quantities found define the rating of the component.
The filter component stresses due to the fundamental system voltage are calculated by applying a voltage source
corresponding to the AC system phase-to-ground voltage across the filter. The harmonic current stresses are
found by applying a current source representing the harmonic current spectrum generated by the converters,
to the parallel impedance of the filters and the AC network. The presence of the AC network may increase the
current through the filter components due to the resonance between filter and network impedances.
Note that the AC parameters associated with rating may be specified by the customer to be more onerous
than those associated with performance calculations.
4.5.5.3. AC Network Impedance

It is common that the AC network impedance as defined for rating calculations differs from that for
performance calculations. The model for rating calculations is normally more pessimistic.
One conservative model used for the harmonic impedance is the X/R model, which defines impedance with given
value of the X/R ratio. The magnitude of impedance is defined as unlimited and the angle is defined as:
 X
Θ = ± arctan   [Eqn 4.5.5a]
 R
However other approaches are also widely used for evaluation of rating.
4.5.5.4. Detuning Conditions

The AC system frequency deviations used for rating are normally greater than those used for performance.
Even some short-term deviations have to be considered.
Component tolerances and possible tuning errors are generally the same as for the performance calculations.
CHAPTER
4.5.5.5. Rating Quantities
The following rating quantities are normally calculated and used in the specification of the filter components.
• Main (HV) Capacitors

a) Maximum capacitor phase voltage excluding harmonics
b) Maximum capacitor phase voltage including harmonics
This is calculated by:
50
U M = U1 + ∑U 2
n [Eqn 4.5.5b]
n=2
Where:
U1 is the fundamental harmonic voltage across the capacitor
Un is the nth harmonic voltage across the capacitor
c) Maximum installed reactive power per phase
N
 
Qinst = MAX  ω 0 ⋅ C ⋅ U M2 , ∑ ω 0 ⋅ n ⋅ C ⋅ U n2  [Eqn 4.5.5c]
 1 
Where: C is the capacitance of the main capacitor
d) Maximum capacitor thermal current
N
I CT = ∑I 2
Cn [Eqn 4.5.5d]
1
Where: ICn is the nth harmonic capacitor current

• Auxiliary (LV) Capacitors
a) Maximum capacitor phase voltage including harmonics
50
UM = ∑U 2
n [Eqn 4.5.5e]
n =1
Where: Un is the nth harmonic voltage across the capacitor
b) Switching impulse withstand level across the capacitor

The value of USIWL together with the value of UM will be used to define the rated voltage of the LV capacitor:
 U 
U R = MAX  k ⋅ U M , SIWL  [Eqn 4.5.5f]
 kSIWL 
Where:
k = 1.0 if the banks can be protected or fused units are used
k > 1.0 if the banks cannot be protected or fuseless units are used (a suitable derating factor used in a
number of projects is 1.25)
kSIWL = ratio of USIWL to rated voltage: typically 4 or another number as agreed with capacitor supplier

N
 
Qinst = MAX  ω 0 ⋅ C ⋅ U R2 , ∑ ω 0 ⋅ n ⋅ C ⋅ U n2  [Eqn 4.5.5g]
 1 
Where: C is the capacitance of the main capacitor

CHAPTER
d) Maximum capacitor thermal current
N
I CT = ∑I 2
Cn [Eqn 4.5.5h]
1
Where: ICn is the nth harmonic capacitor current

For double-tuned filters with high q-values, especially for 11/13th harmonic filter, the current of the LV capacitor
may be substantially higher than that calculated at fundamental frequency from the MVAr. The design of the
capacitor bank should take consideration of this current.
• Reactors
a) Maximum reactor thermal current
N
I LT = ∑I 2
Ln [Eqn 4.5.5i]
1
S
CF
L net
LC L CF
E ac
C
Eac - AC voltage source
Lnet - AC network inductance IA L R
C - capacitor already energized
LC - inductance of the energized capacitor bank
S - switch
CF - capacitor of the filter to be energized
LCF - inductance of the capacitor bank
L, R- filter reactor and resistor
IA - arrester (voltage dependent current source)
Fig. 4.5p– Typical energization circuit
Where: ILn is the nth harmonic reactor current

➙ Harmonic current spectrum for the case defined in a)
➙ Harmonic current spectrum for noise and loss calculations
4.10a
• Resistors
a) Maximum resistor thermal current
N
I RT = ∑I 2
Rn [Eqn 4.5.5j]
1

CHAPTER
Where: IRn is the nth harmonic resistor current
4.5.5.6. Losses
The filter losses are calculated for all the components. Both fundamental current losses and harmonic
current losses are determined. The losses are generally calculated at nominal conditions. This means that
the harmonic currents are to be calculated specially for this condition. It should be noted that the harmonic
losses of AC filters are inversely proportional to the q-factor of the filter. This should be taken into account
when choosing the optimal filter design.
4.5.6. TRANSIENT DESIGN

The transient rating of filter components serves the purpose of finding the extreme transient stresses
across each component when subjected to:
➙ Energization
➙ Ground faults
➙ Switching impulses
+ Z1
U SIPL CF
-
L CF
IA L R
CF - filter capacitor
L, R - filter reactor and resistor
Z1 - total impedance of the conductors and ground
connection to the fault location
USIPL - initial capacitor voltage equal to the SIPL value
Fig. 4.5q– Typical ground fault circuit
Transient rating calculations are generally made separately for each filter branch. If the impedance calculations
show that parallel resonances may occur between the filter branches that would influence the rating, then a
special investigation including the parallel branches has to be performed.
4.10b
4.5.6.1. Filter Representation
The HV capacitors are represented by their capacitance in series with an inductance corresponding to the
number of series and parallel connected units. The typical value of inductance value per unit is 1 – 1.5 μH.
Filter reactors and resistors are represented by their inductance and resistance respectively.

CHAPTER
The filter arresters are represented by a voltage dependent current source. Each arrester should be optimized,
taking into account the maximum operating voltage and maximum discharge currents and energies.
4.5.6.2. Energization Transients

During energization of the filter, the AC busbar voltage is assumed to be at its maximum peak value. The
AC network is represented by a sinusoidal voltage source with its peak voltage in series with an inductance
corresponding to the maximum short-circuit power. A typical circuit is shown in Fig. 4.5p.
4.5.6.3. Ground Faults

For all types of filter other than 3rd harmonic filters, the worst case of the ground fault is the phase-to-
ground short-circuit with the HV capacitor charged to the switching impulse level (SIPL) of the AC bus.
The stresses on the filter arrester will normally be highest when the fault is applied as close to the filter as
possible (minimum inductance in the arrester circuit).
For the reactor current, resistor current and resistor energy, more severe cases may appear if faults can be
applied at some distance from the filter. The reactor current may be determined from a sensitivity analysis.
The maximum value will be obtained when the voltage across the filter arrester is as high as possible without
any observable arrester current, just before the arrester starts to pass substantial amounts of current. This
case gives maximum voltage-time area across the reactor.
CF
L CF
E SI
IA L R
ESI - voltage source giving the switching impulse

CF - capacitor of the filter
Fig. 4.5r– Typical switching impulse circuit
For a series connected resistor, the resistor current is equal to the reactor current. For the resistor connected
in parallel with the reactor, the maximum resistor current will be:
4.10c
LIPLR
IR = [Eqn 4.5.6a]
Rmin

CHAPTER
SETÊ1ÊPROTECTIONS
FilterÊcapacitorÊunbalance
FilterÊcapacitorÊovervoltage
&ÊharmonicÊoverload
SETÊ2ÊPROTECTIONS
TOÊBUSBAR
PROTECTION FilterÊcapacitorÊunbalance
Metering
FilterÊovercurrentÊ&ÊearthÊfault
Fig. 4.5s– Typical filter protection circuit
Where:
LIPLR is the lightning impulse level for the arrester across the resistor
Rmin is the minimum resistor value (normally 90% of Rnom)
4.11a
The energy dissipation in the resistor may be determined from:
1
WR = ⋅ C ⋅ SIPL2bus [Eqn 4.5.6b]
2
Where:
C = capacitance of bank
SIPLbus = switching impulse protective level
A typical circuit is shown in Fig. 4.5q.
For the 3rd harmonic filter, the worst case is the single phase to ground fault creating a very high level of
negative-sequence fundamental voltage in the AC network. This high level of negative-sequence voltage
generates a very high level of the 2nd harmonic on the DC side and the 3rd harmonic on the AC side causing
high stresses in the filter tuned to the 3rd harmonic during the time before the fault is cleared.
4.5.6.4. Switching Surge

To determine the arrester protective levels for switching impulses, a switching impulse of typically
250/2500 μs waveform is applied on the AC bus. The crest value is assumed to be SIPLbus. A typical circuit
is shown in Fig. 4.5r.

CHAPTER
4.5.7. PROTECTION
There are three main types of AC filter protection:
➙ Unbalance current protection of the HV capacitor bank - used to detect element failures
➙ Overload protection - mainly to detect overloading of reactors and resistors
➙ Differential protection - to detect ground faults in the filter bank
4.5.7.1. Unbalance Protection

The two most commonly used designs of capacitor units for HV capacitors are the internally fused design
and the fuseless design.
In the internally fused design, the element failure leads to operation of a fuse connected in series with each
element and disconnection of the faulty element without disturbing the operation of the entire capacitor
bank. However, if many elements connected in parallel fail, the voltage across the remaining elements in the
same group may reach a very high level, causing a considerable reduction of the lifetime of the remaining
elements. As a final consequence, an avalanche condition may occur in this group, leading to a catastrophic
failure. To avoid this condition, an unbalance current protection is implemented in the HV capacitor. Typically,
three levels of protection settings are used:
➙ ALARM: when the expected time to filter trip is in the order of weeks, an alarm signal is given.
➙ DELAYED TRIP: when the expected time to filter trip is in the order of hours, an alarm signal is given,
and within two hours after the alarm signal, the bank is automatically disconnected.
➙ INSTANTANEOUS TRIP: when the risk of avalanche condition is imminent, the bank is disconnected
immediately.
In the fuseless design of the capacitor units, there is normally no danger of the avalanche condition.
Nevertheless, the unbalance protection is usually implemented with three levels of protection settings,
chosen to give action after failure of a similar number of elements as in the internally fused design.
Externally fused capacitors may be offered but are rarely competitive against fuseless or internally fused
designs. Additionally, there is a history of problems with the fuses of these capacitor banks.
4.5.7.2. Overload Protection

This facility is implemented for the protection of the filter capacitors, reactors and resistors.
The input signals for the overload protection are given by current transformers in the filter branches. There
is one current transformer in each filter phase measuring the current through each filter. In case the resistor
current may not be directly in proportion to the reactor current, an extra current transformer will be installed
in the resistor branch.
The characteristics of the protection system are coordinated with the maximum allowed continuous
fundamental and harmonic current IC (TRIP level), and the thermal time constant of the equipment. Both
the trip level and the time constant are based on the information from the suppliers of the reactors and
resistors.
4.5.7.3. Differential Protection

The primary objective of this protection is to detect and isolate ground faults within a filter bank. The input
signals for the differential protections come from current transformers installed at the HV connection and
neutral connection of each phase of the filter bank. The action of the protection is to provide an immediate
trip of the filter bank circuit breaker. The setting should be coordinated with the normal filter arrester current.
4.6. DC HARMONIC FILTERS

This section discusses the information necessary to design the DC harmonic filters.

CHAPTER
In broad terms, the information required is:
➙ The main circuit data for the calculation of the harmonic voltages injected into the DC line
➙ The requirements for different operating conditions
➙ Information on the DC line and electrode lines
➙ Performance requirements
During operation of the HVDC system, harmonic voltages are generated on the DC side of the system, injecting
harmonic currents into the DC line. The electromagnetic fields resulting from these harmonic currents may
induce voltages into adjacent telephone lines, causing interference problems.
The quality of the telephone transmission in these conditions depends on:
➙ Harmonic current level along the line
➙ The mutual coupling between the DC line and the telephone systems
➙ The susceptibility of the telephone facilities
To minimize the harmonic currents injected into the DC line, a filtering system containing the DC smoothing
reactor and the DC filters should be installed for projects where the DC link is established in whole or in part
as an overhead line.
The parameters of the filtering system should be chosen with care to fulfill the following requirements:
➙ Reduce the harmonic currents in the DC system to acceptable levels
➙ Avoid the resonance of the DC system at frequencies close to fundamental and second harmonic
➙ Limit the transient currents reaching the valve hall
➙ Optimize the design economically, including losses in the smoothing reactor and the filter
4.6.1 DESIGN DATA

This section discusses the information necessary for the design of the DC filters. The performance
requirements for different operating conditions define the maximum allowed currents injected into the
DC line.
The main circuit data are needed for the calculation of the harmonic voltages generated by the converters.
The DC circuit data, which include the impedance of the DC line and the impedances of the station equipment,
are needed to calculate the variation of current distribution along the line for each harmonic so that the
performance can be assessed at each point on the line.
4.6.2. PERFORMANCE REQUIREMENTS

The performance of the DC filter can be based on two different criteria: the sum of harmonic currents and
induced noise in the telephone line.
4.6.2.1. Sum of Harmonic Currents

The performance is based on the following formula:
I eq ( x ) = I e ( x )2R + I e ( x )2I [Eqn 4.6a]
Where:
Ieq(x) is the 800 Hz equivalent disturbing current in milli-amps, psophometrically weighted (mAp) at any
point along the transmission corridor
Ie(x)R is the magnitude of the RSS equivalent disturbing current due to harmonic voltage sources at rectifier
(mAp)
Ie(x)I is the magnitude of the RSS equivalent disturbing current due to harmonic voltage sources at inverter
(mAp)
The equivalent disturbing current at any point along the corridor due to harmonics from either the rectifier
or inverter is calculated as follows:
2
I e ( x ) =  I r (n, x ) ⋅ P (n ) ⋅ H f  [Eqn 4.6b]

CHAPTER
Where:
Ir(n,x) is the magnitude of the residual RMS current at each nth order harmonic, in milliamps, at distance x
km from the reference converter station along the transmission corridor
P(n) is the psophometric weighting factor at harmonic n
n is the harmonic number
Hf is the coupling factor, which represents the normalized frequency dependent effects of the typical
coupling impedance to open wire circuits
4.6.2.2. Induced Noise in the Telephone Line

The following formula applies:
2
 k
n 
U ind = ∑  ∑ I ij ⋅ Z  ⋅ Wi 2 [Eqn 4.6c]
i =1  j =1 
Where:
Z is the coupling impedance between the DC line and the telephone line
Wi is the weighting factor for the i-th harmonic (see Appendix A4.2)
Iij is the i-th harmonic current flowing into the j-th conductor of the DC line
Note: The coupling impedance Z is dependent on the distance between the lines, the ground resistivity in the
area and the length of the telephone line. For ground resistivity values below 60 Ω-m, an approximate formula
applies:
L⋅ρ
Z =K⋅
d2 [Eqn 4.6d]
Lsm Zlp Lsm
L3p Ylp Ylp
Csh
2L3p
DCÊfilter DCÊfilter
Csl
L3p
Zel Zel
Cnb Cnb
Yel Yel Yel Yel
Cnb Cnb
Ylp Ylp
Lsm Zlp Lsm
Fig. 4.6a– Scheme for bipolar operation

CHAPTER
Lsm Zlp Lsm
L3p Ylp Ylp
Csh
2L3p
Csl
L3p
Zel Zel
Cnb Cnb
Yel Yel Yel Yel
Ylp Ylp
Zlp
Fig. 4.6b– Scheme for monopolar operation with metallic return
Lsm Zlp Lsm

4.12b
L3p Ylp Ylp
Csh
2L3p
Csl
L3p
Zel Zel
Cnb Cnb
Yel Yel Yel Yel
Fig. 4.6c– Scheme for monopolar operation with ground return

CHAPTER
Where:
L = length of the telephone line
r = ground resistivity
d = distance between the DC line and the telephone line
K = proportionality factor
For the purpose of the performance requirements, a test line can be defined as a 1 km long open line parallel
to the DC line at a distance of 1 km from the centerline of the two pole lines of the DC line.
k
The sum of the harmonic currents ∑ I ij flowing in all conductors of the DC line tower is, for each
j =1
harmonic, equal to the ground return current. The level of this current is strongly dependent on the mode of
operation of the converters. For the balanced mode of operation, where the harmonic currents tend to cancel
each other, the ground return current will be low; for other operating modes, it will be higher.
4.6.2.3. Performance Levels

Typical values of performance levels are different for different modes of operation.
For bipolar balanced operation, which is the normal operation in a bipolar HVDC transmission, and for
monopolar operation with metallic return, typical values will be:
➙ Weighted sum of currents 1000 – 3000 mA
➙ Induced voltage 10 – 15 mV
For monopolar mode with ground return and for unbalanced bipolar modes, as well as for operation with the
reduced DC voltages, the acceptance limits are often doubled:
➙ Weighted sum of currents 2000 - 6000 mA
➙ Induced voltage 20 - 30 mV
4.6.2.4. HVDC Operating Conditions

The filter must be dimensioned to operate according to the specification for all possible operating
conditions of the HVDC converters. The following conditions are generally specified:
➙ Bipolar balanced operation
➙ Monopolar operation with metallic return
➙ Monopolar operation with ground return
➙ Bipolar operation with reduced DC voltage at one or both poles
➙ Bipolar operation with increased reactive power absorption
Figs. 4.6a - 4.6c below show the schematic arrangements for the first three operating modes mentioned above.
4.6.2.5. Circuit Parameters

The DC side circuit configurations shown in Figs. 4.6a - 4.6c above contain the DC line between the stations,
the electrode lines in each station, the harmonic voltage sources and the impedance in each station.
4.6.2.6. Overhead DC Lines and Electrode Lines

The overhead line is modeled in detail by its distributed parameters, i.e. the series impedance and shunt
admittance matrices per km for the conductor configuration defined by the geometrical data and the
material constants of the pole conductors, ground wires and earth. The parameters of the hyperbolical
p-sections are calculated and modeled in the computer program analyzing the DC circuit.
4.6.2.7. Converter Representation

A converter bridge can be represented by a Thévenin equivalent circuit with harmonic voltage sources
as calculated in section 4.4 and internal series impedance corresponding to the average value of the
commutating impedance per bridge.
The impedance to ground in the model described in section 4.4 comprises the total stray capacitances of
the valve winding and its connections to ground (Csh, Csl).

CHAPTER
The parameters of the model in this figure are given by:
L 3P = ξ ⋅ L C [Eqn 4.6e]
Where: LC is the commutating reactance of one 6-pulse converter
1 1.5µ 60 − µ
ξ= ⋅ + [Eqn 4.6f]
2 60 60
µ is the overlap angle (deg)
Csh is the sum of stray capacitances of the upper 6-pulse group
Csl is the sum of stray capacitances of the lower 6-pulse group
Lsm1 Lsm2 LsmÊpole
LsmÊnb
Fig. 4.6d– Configuration with a split smoothing Fig. 4.6e– Configuration with a smoothing
reactor reactor split between pole and neutral bus
4.6.2.8. Filter Arrangement

The arrangement for the filtering of the DC side harmonics comprises a DC smoothing reactor in series
with4.12d 4.12e
each pole line, a DC filter between each DC pole and neutral bus, and equipment installed between
the neutral bus and ground. This equipment could be a capacitor or a filter tuned to a certain frequency.
The DC reactor is located at the DC line pole in order to participate in filtering the harmonics in the most
efficient way. In addition, it forms part of overvoltage protection by preventing steep lightning surges from
the line to penetrate the converter station. In addition, it limits discharge currents from the DC line caused by
short-circuits in the converter and limits the current through the converter caused by DC line faults.
The choice of the parameters of the DC filters and the neutral bus arrangement is described below.
4.6.2.9. Choice of Smoothing Reactor Inductance

The smoothing reactor is located at the DC line pole between the converter and the DC filter. When
choosing the inductance value, the following aspects should be considered:
➙ The overall filter performance: a higher inductance reduces the current harmonics
➙ DC side resonances: the inductance should be chosen so that no resonance close to any multiple of
the fundamental frequency occurs between the filter arrangement and the converter
➙ The risk of a sudden voltage drop in the AC network causing a commutation failure of the inverter should
be minimized
CHAPTER
Further details of the choice of the DC smoothing reactor are described in section 4.2.
The resonance frequency between the filter arrangement and the converter is calculated as:
1
f = [Eqn 4.6g]
2 ⋅ π ⋅ LC
Where:
L = Lsm + Li Smoothing reactor inductance plus the average value of the 12-pulse commutating inductance
(see section 4.6.2.7)
C is the DC filter capacitance
As the average of the commutating inductance varies with the firing and overlap angles, different operational
conditions should be investigated.
When calculating the size of the smoothing reactor, the possibility of a split of this reactor should also be
taken into consideration. In some cases, it may be advantageous to connect the DC filter in between the
two parts of the smoothing reactor as shown in Fig. 4.6d below.
The reactor may also be split between the pole and neutral bus conductors as shown in Fig. 4.6e below.
4.6.3. FILTER TYPES

For modern HVDC stations operating with 12-pulse converters, the largest characteristic harmonic voltages
generated are the 12th,24th, 36th and 48th harmonic. Since the harmonic generation is generally inversely
proportional to the harmonic order, the usual practice is to provide a tuned filter for the 12th harmonic and
damped filters for the higher harmonics.
There may be two situations where a second harmonic filter may be needed in the filtering system:
➙ When the unbalance of the AC network is defined high, the high value of the negative sequence voltage
will generate a high value of the second harmonic voltage on the DC side
➙ When the impedance of the line is close to a resonance at the second harmonic
Because of the low weighting of the second harmonic for telephone interference, the need for a second
harmonic filter should be investigated carefully.
The different filter types used for the design of the DC filters are similar to the types described in
section 4.5.3 for AC filters.
As the 3-pulse generated harmonics occur mostly in a ground mode (the driving voltage occurs as a result of
the stray capacitances between different parts of valve and ground), the impedance between pole and ground
is important for the filtering of these. This impedance is made up of the impedance of the DC filter in series
with the impedance of the neutral bus arrangement.
4.6.4. STEADY-STATE DESIGN

The steady-state design is the design taking into consideration the operating conditions that may occur
for durations in excess of about one minute. This time may vary as different customers may have different
requirements. The filter branch sizes, resonance frequencies and q-values are determined to achieve
optimum filtering regarding both performance and economy of the filters. Since the most expensive part of
the filter is the main capacitor, special care should be taken to optimize the capacitance value with respect
to the maximum voltage across it.
The filter components are rated with the purpose of finding the highest current and voltage stresses during
steady-state conditions.
4.6.4.1. Filter Detuning

Filter detuning effects for DC filters are similar to those affecting AC filters, as discussed in section 4.5.3.1.
4.6.4.2. Performance Calculations

The definitions of the performance quantities relevant for the DC filter design together with the typical
CHAPTER
requirements are given in section 4.6.2. The filter performance has to be calculated for all operating modes
and the full load range. This implies that several sets of harmonic voltages have to be calculated for the
different operating modes including the variation of the direct voltage and firing angles.
When calculating the performance it should be verified that no resonance close to a multiple of the
fundamental frequency occurs in the system, even for a line length that may vary from the nominal specified
length by approximately +/- 5%.
4.6.4.3. Steady-State Rating

After the filter configuration fulfilling the performance C R
requirements is chosen, the rating of the filter
components must be calculated.
C R
The purpose is to find the decisive stresses on the
filter components during steady-state operation.
The calculations should be performed for each of C R
the possible operating modes, all the possible load
DCÊvoltageÊgrading
ranges and all the possible main circuit parameters, needsÊtoÊbeÊcarefully
in order to determine the highest voltages across, and C R
consideredÊbyÊthe
the currents flowing through the filter components. capacitorÊdesignerÊto
ensureÊthatÊindividual
Additionally, a margin of typically 10 – 20% is added unitsÊareÊnotÊsubjected
to all the calculated harmonic voltages. C R toÊexcessiveÊvoltageÊstress
• Detuning Conditions
The value of the AC system frequency deviation
C R
used for rating calculation is normally greater than
the value used for the performance calculations.
The component tolerances and the possible tuning
C R
errors are assumed to be the same as for the
performance calculations.
C R
4.6.4.4. Rating Quantities
The following rating quantities are normally
calculated and defined in the specification of the
Fig. 4.6f– Grading resistors of the HV DC capacitor
different filter components.
• High Voltage Capacitors 4.13a
The high voltage capacitor is the capacitor connected to the DC pole line. This capacitor is subjected to the
full DC voltage as well as to the AC harmonic voltages.
The maximum peak capacitor voltage is calculated as:
n
U C max = U DC max + 2 ⋅ ∑ U CRi
2
+ U CIi
2
[Eqn 4.6h]
i =1
Where:
UDC max is the highest DC voltage that can occur in the system during steady-state conditions
UCRi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages generated
by the rectifier
UCIi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages generated
by the inverter
The maximum thermal current is calculated as:
n
I CT = ∑ (I 2
CRi + I CIi
2
) [Eqn 4.6i]
i =1

CHAPTER
Where: ICRi is the capacitor current of order i caused by the harmonic voltages generated by the rectifier
ICIi is the capacitor current of order i caused by the harmonic voltages generated by the inverter
In addition, the harmonic current spectra for worst-case operation have to be specified for the rating of
the HV capacitor as well as the spectra for noise and loss calculations.
An important consideration in the design of a DC filter main capacitor bank is that this capacitor will be mainly
subjected to the applied DC voltage and hence the sharing of the DC voltage as well as the AC harmonic
voltage must be controlled. This means that resistive voltage grading needs to be installed in parallel with
each capacitor element inside the capacitor unit in order to control the DC voltage distribution amongst the
DC capacitors, as shown in Fig. 4.6f.
The DC voltage grading is also affected by leakage currents across the external insulation of the capacitor
bank. For this reason, it is common for DC filter capacitor banks to be constructed as one single tall bank as
opposed to any form of split bank where the split banks would have post insulators between the capacitor
racks that disturb the voltage distribution due to leakage currents across them.
• Low Voltage Capacitors
The low voltage capacitor is the capacitor connected in the bottom part of the filter, in series with a reactor.
This capacitor is stressed only by the harmonic voltages.
a) Maximum capacitor phase voltage including harmonics
n
U C max = ∑ (U CRi + U CIi ) [Eqn 4.6j]
i =1
Where: UCRi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages
generated by the rectifier.
UCIi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages
generated by the inverter.
b) Switching impulse withstand level across the capacitor (USIWL)
The value of USIWL together with the value of UM will be used to define the rated voltage of the LV capacitor:
 U 
U R = MAX  k ⋅ U M , SIWL  [Eqn 4.6k]
 kSIWL 
Where: k = 1.0 if the banks can be protected or fused units are used
k > 1.0 if the banks cannot be protected or fuseless units are used (a suitable derating factor used
in a number of projects is 1.25)
kSIWL = ratio of USIWL to rated voltage – typically 4 or such other number as agreed with capacitor
supplier
N
 
Qinst = MAX  ω 0 ⋅ C ⋅ U R2 , ∑ ω 0 ⋅ n ⋅ C ⋅ U n2  [Eqn 4.6l]
 1 
The maximum thermal current is calculated as:

n
I CT = ∑ (I 2
CRi + I CIi
2
) [Eqn 4.6m]
i =1
Where:
ICRi is the capacitor current of order i caused by the harmonic voltages generated by the rectifier
ICIi is the capacitor current of order i caused by the harmonic voltages generated by the inverter
In addition, the harmonic current spectra for worst-case operation have to be specified for rating of the LV
capacitors, as well as the spectra for noise and loss calculations.
• Reactors

CHAPTER
The maximum thermal current through the reactor is calculated as:
n
I LT = ∑ (I 2
LRi + I LIi
2
) [Eqn 4.6n]
i =1
Where:
ILRi is the reactor current of order i caused by the harmonic voltages generated by the rectifier
ILIi is the reactor current of order i caused by the harmonic voltages generated by the inverter
In addition, the harmonic current spectra for worst-case operation have to be specified for the rating of the
resistors as well as the spectra for noise and loss calculations.
• Resistors
The maximum thermal current through the resistor is calculated as:
n
I RT = ∑ (I 2
RRi + I RIi
2
) [Eqn 4.6o]
i =1
Where:
IRRi is the resistor current of order i caused by the harmonic voltages generated by the rectifier
IRIi is the resistor current of order i caused by the harmonic voltages generated by the inverter
• Neutral Bus Capacitors
The maximum peak voltage across the neutral bus capacitor is calculated as:
n
U C max = U DC max + 2 ∗ ∑ U CRi
2
i
+ U CIi
2
[Eqn 4.6p]
i =1
Where:
UDC max is the highest DC voltage that can occur on the neutral bus during steady-state conditions
UCRi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages generated
by the rectifier
UCIi is the RMS harmonic voltage of order i across the capacitor, caused by the harmonic voltages generated
by the inverter
Pole
+ Z1
U SIPL CF
-
L CF
Z1 - total impedance of the conductors
and ground connection between S
the converter station and the fault IA L R
location
USIPL - initial capacitor voltage equal
to the SIPL value Le
IA - filter arrester (voltage dependent NeutralÊbus
current source)
An - neutral bus arrester (voltage
dependent current source) Cnb
Cnb - neutral bus capacitor An
Le - electrode line inductance
Fig. 4.6g– Typical pole-to-ground fault circuit

CHAPTER
4.6.5. TRANSIENT DESIGN
The purpose of the transient design is to determine maximum transient currents and voltages on the DC
filter equipment and to design the DC filter arresters.
Pole
+
U SIPL CF Z1
-
L CF
Z1 - total impedance of the conductors and ground S
connection between the converter station and the fault
location IA L R
USIPL - initial capacitor voltage equal to the SIPL value
IA - filter arrester (voltage dependent current source)
NeutralÊbus
Fig. 4.6h– Typical pole-to-neutral fault circuit
The transient cases decisive for the rating of the DC filter components are ground faults and the short-circuits
between pole and neutral. These faults will cause a fast discharge of the main capacitor which will stress the
LV filter components. High transient stresses may also be caused by the switching surge overvoltages and
lightning overvoltages.
The most frequent faults will be faults along the DC line. In these cases, the impedance of the DC line between
the station and the fault location will contribute to reduce the stresses on the filter components. The decisive
faults for the filter component ratings will then4.13c
be the faults occurring in the station. The frequency of these
faults is very low; nevertheless, they should be considered for the design of the filter components.
CF
L CF
ESI - voltage source giving the switching impulse E SI

CF - capacitor of the filter IA L R
Fig. 4.6i– Typical switching impulse circuit

CHAPTER
4.6.5.1. Filter Representation
The HV capacitors are represented by their capacitance in series with an inductance corresponding to the
number of series and parallel connected units. The typical value of inductance per unit is 1 – 1.5 μH.
Filter reactors and resistors are represented by their inductance and resistance respectively.
The filter arresters are represented by voltage dependent current source. Each arrester should be optimized taking
into account maximum operating voltage and maximum discharge currents and energies.
The arrester characteristic is dependent on the current discharge waveform. The characteristic which best
approximates the discharge should be used. The maximum characteristics should be applied when calculating the
filter components and the minimum characteristics when calculating the maximum arrester current and energy.
The neutral bus is represented with the arresters and neutral bus capacitors, when applicable.
4.6.5.2. Pole-to-Ground Fault

The pole-to-ground fault is normally applied assuming the filter HV capacitor is charged to the switching
impulse voltage level of the DC bus (SIPL).
The stresses on the arrester are normally highest when the fault is applied as close to the filter as possible,
normally at the busbar insulator closest to the filter. For the filter component currents and energy, the worst
case may be when the fault occurs at some distance from the filter. The maximum value can be determined
after a sensitivity analysis. A typical circuit is shown in Fig. 4.6g.
The transient voltages across the filter components not protected by the arresters will also be defined in
this fault case.
4.6.5.3. Pole-to-Neutral Fault

The short-circuit should be applied assuming that the HV capacitor of the filter is charged to the maximum
operating voltage of the pole bus. A typical circuit is shown in Fig. 4.6h.
4.6.5.4. Switching Surges

To determine the filter arrester protective levels for switching impulse, a switching impulse of typically
250/2500 μs waveform with a crest value equal to SIPL of the pole bus is assumed to occur at the pole bus.
The surge is applied to one filter branch at a time. A typical circuit is shown in Fig. 4.6i.
4.6.6. PROTECTION
The only type of filter protection is the harmonic overload protection. The purpose is to detect steady-state
harmonic currents in excess of those specified flowing through the filter reactors and resistors.
In addition to this protection, the reactors and resistors will be protected against the transient overvoltage by arresters.
The capacitor units of the HV capacitor bank will be most often protected by the internal fuses which
disconnect the faulty elements in the bank without disturbing the voltage distribution and with only
minimal changes of the capacitor value (Celement< 0.02% ⋅ Cbank).
4.7. POWER LINE CARRIER FREQUENCY FILTERS

HVDC converters can produce electromagnetic noise on both the AC and DC transmission networks and
over a very wide range of frequencies. This noise has the potential to interfere with the customer’s Power
Line Carrier (PLC) communication systems, if suitable mitigation measures, in the form of high frequency
filters, are not installed at the converter station. This section describes the studies required to model the
electromagnetic interference as well as the design of the high frequency filters required to immunize the
transmission networks.

CHAPTER
4.7.1. POWER LINE CARRIER COMMUNICATIONS
In many power systems, the power system transmission lines and cables are used as communication media
[17, 18]. This Power Line Carrier (PLC) system provides an economical method of sending low bandwidth
information and even voice communications to distant parts of the power system without the need for
additional communication facilities.
The data or voice signals to be transmitted are coupled to the power system through coupling capacitors and
line tuning units which ensure the lowest possible level of signal loss. The data or voice signal is modulated
onto the carrier signal. In addition, line traps (air-cored reactors of low inductance) with or without tuning are
generally fitted to constrain the signal to the required transmission line.
PLC filters, also referred to as high frequency filters, are often required to protect PLC communications on
HV lines and cables. The nature of HVDC converter operation is to generate noise in the PLC frequency range
– typically considered to be 30 kHz to 500 kHz – or the subset of this range that is specified by the customer.
Even without HVDC converters present there are other sources of noise such as corona and switching
operations which interfere with the signal. However at the low frequency end of the PLC band the noise from
a HVDC converter is generally higher and needs special filtering.
PLC communications are still widely used on the AC system and occasionally on the DC lines of a HVDC system.
However the arrival of high bandwidth communications with optical fibers embedded in ground wires and in
cables is gradually superseding PLC communications for HV systems. Since the DC system is specific to a HVDC
converter and built alongside it, it is now rare to see PLC communications used on the DC system and in general
no PLC filters are required for the DC system. However, most HVDC schemes are connected to a pre-existing AC
system, usually built many decades earlier, and in many cases there are existing PLC communications systems,
so the requirement for considering PLC frequency interference from HVDC converters on the AC system is likely
to be required for some time.
ACÊHARMONIC
ACÊNETWORK FILTERSÊAND PLCÊFILTERS CONVERTER/
SUBSTATION SOURCE
BUSBARS
Fig. 4.7a– Block diagram of the HVDC converter system model
4.7.2. PLC STUDIES AND FILTER DESIGN

This section provides a description of the high frequency conducted interference (PLC, Power Line Carrier
Frequency) design procedure for a converter station.
4.14a The objectives of the studies described are to:
➙ Outline the design criteria
➙ Describe the system model
➙ Describe the design method
➙ Describe the method proposed to meet the performance requirements
Note: Several established documents such as the CIGRÉ and IEEE Guides on Power Line Carriers [17, 18], the EPRI report “Radio
Interference from HVDC Converter Stations”, EL-3712 [19], and CISPR 18 [20] are used as references.

CHAPTER
4.7.3. DESIGN REQUIREMENTS
High frequency noise from the HVDC converter is conducted along the connected power lines. This needs
to be limited to an acceptable level where PLC communication systems are used.
The actual level of interference permitted depends to a large extent on the configuration of the AC
system, the power of the PLC transmitters and the required signal to noise ratio. The customer’s technical
specifications concern noise on the power line systems and give the interference limit. References [17]
and [18] give considerable guidance on selection of limits. Typical limits are given below.
The design requirement of the PLC filters is that they should limit to an acceptable level the interference
appearing on the AC busbars at the point of connection to the power lines. The PLC interference limits given
by the technical specification for a nominal 3.1 kHz bandwidth, flat weighting, RMS measurement on a 50
Ω base, are:
Location Maximum Level
AC buses 20 dBm between 30 kHz and 500 kHz
C7 C8 C7
L3 L3 ÊÊÊÊL4Ê
L2 L2 L2 C6
C9
C1 L1 C3 C1 L1 C3 C1 L1
C3
C6 V1 C6 C6Ê V5Ê
C2 L1 C4 C2 L1 C2 L1
C4 C4
C B AÊ
C6 V4 C6 V6 C6 V2
L1
L3 L3 L4
C2 L1 C3 C2 L1 C3 C2 L1 C3
C9
C6
C6 V1 C6 V3 C6 V5
C2 L1 C2 L1 C2 L1
C4 C4 C4
C B A
C6 V4 C6 C6
VAa
C1 L1 C1 L1 C1 L1
C5 C5 C5
a
L2 L2 L2
L4
L3 L3
C7 C8 C7
C9
Fig. 4.7b– Model of 6-pulse bridge
BACK TO 4.14b DC Transmission Systems: Line Commutated Converters | 215

CHAPTER
4.7.4. OVERVIEW OF THE SYSTEM
The first step is to examine the PLC frequency noise
due to converter operation conducted along the
AC transmission lines connected to the converter Lrs Cs
stations.
The scheme can be divided into a number of
discrete elements: the thyristor valves, the
converter transformer, the AC systems (AC
harmonic filters and a simplified representation Fig. 4.7c– Non-commutating conducting valve
of the relevant AC system). Fig. 4.7a shows a block
diagram of the converter station model used to
analyze the PLC frequency interference.
Lrs
4.14c
Lru Re
Cs
LrsÊ-ÊAirÊcoreÊinductanceÊofÊtheÊsaturableÊreactor
Cd LruÊ-ÊMagnetizingÊinductanceÊofÊtheÊsaturableÊreactor
ReÊ -ÊEddyÊcurrentÊresistanceÊofÊtheÊsaturableÊreactor
CsÊ -ÊStrayÊcapacitanceÊofÊtheÊsaturableÊreactor
CdÊ -ÊDampingÊcapacitance
Fig. 4.7d– Non-commutating blocked valve
Veq Lrs Cs
C11
VeqÊisÊtheÊthyristor (30-500ÊkHz)Ê(PLCÊrange)
voltageÊbreakÊdown BeginningÊofÊtheÊcommutation
(valveÊstillÊfullÊconducting)
Turning-onÊvalve
4.14d Turning-offÊvalve
Fig. 4.7e– Turning-on and turning-off valves

CHAPTER
The converter transformers can be considered as single-phase units. Sensitivity studies have shown that
secondary effects such as interphase coupling do not significantly affect the high frequency characteristics
of the system. Therefore, because each phase is identical and their coupling is insignificant, a single-phase
model is appropriate.
The circuit model for each major element of the converter station contains components representing
busbar stray inductance and capacitance, which are calculated on the basis of 10 pF/m and 1 µH/m for all
station busbars. The following sections describe the major elements of the converter station in more detail.
4.7.5. NOISE SOURCE MODEL

There are two events during valve operation which cause high frequency noise; namely, turn-on and turn-
off. For PLC frequency purposes, only turn-on is significant, since turn-off is a comparatively slow event
(the rate of change of voltage is controlled by the valve damping network), so its influence on the frequency
range of interest is small.
4.7.5.1. Thyristor Valve Voltage

The turn-on event is modeled mathematically as an exponentially decaying voltage as in reference [21]
leading to the equation for equivalent noise voltage shown in reference [19] (equation 27) and repeated
as Eqn 4.7a below.
V 0 ⋅ K1
V eq ≈
ω 1 + (ω ⋅ T 0 )2 [Eqn 4.7a]
Where:
Veq = equivalent noise voltage at angular frequency w
V0 = breakdown voltage
K1 = bandwidth (Dw) for PLC frequency voltage measurement
w = angular frequency
T0 = breakdown time
The voltage between the terminals of the thyristor valve during turn-on is derived from the valve design
studies. For this purpose a detailed model of the thyristor valve and surrounding components and busbars
is used. From these studies, a worst-case thyristor valve turn-on case is found (with maximum switching
voltage and minimum switching time) to maximize Eqn 4.7a.
The turn-on event is brought about by a change of valve impedance from near open circuit to near short-
circuit. The sudden collapse of voltage brings about the discharge of any parallel-connected capacitances,
generating oscillations in circuits that are affected by it. The resulting high frequency currents are then
distributed via the busbars to the transmission lines.
4.7.6. THE THYRISTOR VALVES

The 12-pulse converter is modeled in detail for the thyristor valve design studies as shown in Fig. 4.7b.
Details of the valve models for the various valve states are given in Fig. 4.7c to Fig. 4.7e.
The component values for Fig. 4.7b are based on the arrangement of busbars in the valve hall - in terms of
length and diameter of busbars and spacing of busbars from other busbars and from earth planes.
In practice, the worst case for PLC frequency noise generation is when the valve turn-on occurs in the least
possible time (i.e. all levels together - ‘coherent turn-on’) and from the maximum continuous voltage.
4.7.7. CONVERTER TRANSFORMER

The converter transformer manufacturer provides detailed information regarding the winding arrangement,
interwinding capacitances, winding series capacitances, shunt capacitances and leakage inductances. This
is used to make a high frequency model of the transformer for the simulation.

CHAPTER
ACÊlineÊnoiseÊwithÊPLCÊfilter
50
WithoutÊPLCÊfilters Limit
25
NoiseÊ(dBm)
-25
-50
0 50 100 150 200 250 300 350 400 450 500
FrequencyÊ(kHz)
Fig. 4.7f– PLC frequency noise - without PLC filters
4.14f
ACÊlineÊnoiseÊwithÊPLCÊfilter
50
WithÊPLCÊfilters Limit
25
NoiseÊ(dBm)
-25
-50
0 50 100 150 200 250 300 350 400 450 500
FrequencyÊ(kHz)
Fig. 4.7g– PLC frequency noise - with PLC filters

CHAPTER
R2A C2A R4A C4A
C2B L2B R2B C4B L4B R4B
Converter ACÊsystem
L2A L4A
C1A C3A
L1A R1A L3A R3A
C1B C3B
R1B R3B
L1B L3B
C1C C3C
R1C R3C
L1C L3C
Fig. 4.7h– Arrangement of PLC filters
R2A C2A R4A C4A

4.14h
C2B L2B R2B C4B L4B R4B
Converter ACÊsystem
L2A L4A
C1A C3A
L1A R1A L3A R3A
C1B C3B
R1B R3B
L1B L3B
C1C C3C
R1C R3C
L1C L3C
Fig. 4.7i– Typical detailed PLC filter representation

CHAPTER
0.06
0.05
0.04
AdmittanceÊ(S)
0.03
0.02
0.01
0
0 100 200 300 400 500
FrequencyÊ(kHz)
Fig. 4.7j– Admittance of PLC filter
In general, a single-phase transformer model can be used since the interphase coupling of even a 3-phase
transformer at PLC frequencies is small.
4.7.8. THE AC SYSTEM

The AC system model includes major items of equipment connected to the AC busbars such as the busbars,
AC harmonic filters and the transmission lines.
4.14j
The capacitances and inductances of the interconnecting busbars are calculated from the dimensions of the
converter station layout. The values used are based on the parameters of the AC equipment and the effect
of the HV connections from the transformer to the measuring point.
A surge impedance of 400 Ω can be used to represent each transmission line based on IEC 60481 [22].
4.7.9. PLC FILTER DESIGN

The PLC frequency interference level is calculated and the PLC filters, if required, are designed by analyzing
the circuit at discrete frequencies over the specified range in order to identify resonant frequencies. An
important consideration is the potential for interaction with adjacent components – transformer, AC filter
and even busbars.
A combination of series and shunt elements are chosen to achieve the required level of interference reduction.
The preference for series or shunt elements will be affected by such factors as:
➙ Line current
➙ Line short-circuit current
➙ Connection voltage
➙ Availability of component values
➙ Effect on reactive power
➙ Effect on short-circuit level at the converter terminals
Higher connection voltages will act to minimize the capacitance of the coupling capacitor whereas high line
currents and short-circuit currents will act to minimize the series element.
Typical conducted noise, over the range 5 kHz to 500 kHz, transmitted onto the HV busbars without and
with PLC filters is shown in Fig. 4.7f and Fig. 4.7g.
Studies are performed to determine PLC filter parameters on which the details of the final filter design
are based. These details include component values, maximum permissible levels of stray capacitance and
inductance and tolerance limits. A typical PLC filter is shown in Fig. 4.7h.

CHAPTER
Fig. 4.7i shows the study representation of a typical final filter design including the maximum specified
stray inductance and capacitance of the components.
In normal operation the surge arresters present an equivalent capacitance at their terminals. This circuit
arrangement is connected in the position shown in Fig. 4.7a to complete the HV system model.
Fig. 4.7j shows the impedance and admittance characteristics of typical filter elements.
4.7.10. PLC FILTER COMPONENTS

The following sections describe the types of equipment used to form the PLC filters.
4.7.10.1. Coupling Capacitors

For low capacitance values, conventional coupling capacitors similar to those used in CVTs can be used. For
higher capacitance values, the capacitors will be of the type used for power factor correction.
4.7.10.2. Line Traps

A variety of line trap designs can be used. Typically line traps are low inductance and may be fitted with
tuning devices and surge arresters internally. A range of tuning arrangements may be considered:
➙ Untuned
➙ Single-tuned
➙ Double-tuned
➙ Wideband
The line traps may be supported on a dedicated insulator and structure, or they may be supported on the
coupling capacitor if it exists (provided that the mechanical forces are not excessive). Alternatively, line traps
may be suspended from a gantry.
4.7.10.3. Line Tuners

The line tuner units consist of a reactor, capacitor and resistor network protected by a surge arrester as shown
in Fig. 4.7i. Each unit is connected to the carrier terminal of its associated CVT or coupling capacitor.
The mechanical design is very similar to that used for the line tuner for the PLC communication facility but the
circuit is dedicated to the filtering function.
4.7.11. RATING OF PLC FILTER COMPONENTS

Power frequency ratings dominate and components must be able to withstand the full range of conditions
imposed by the power system.
The line tuner units are rated for the arithmetic sum of voltages and currents calculated in the range 30 kHz - 500 kHz.
4.7.12. SITE MEASUREMENTS

A program of site measurements is required to verify the achievement of the specified performance
requirements.
Measurements are made across a drain coil fitted to an appropriate CVT using a spectrum analyzer. Typically
measurements are made using an average detector but a peak detector may also be used to ensure there
are no excessive peaks in the interference.
4.8. RADIATED INTERFERENCE AND EMC

Radio communications are everywhere – radio, television, mobile phones, satellite communications, and
navigation systems. All equipment must be designed to coexist with such systems – both in terms of
immunity to interference and not interfering with such systems.
HVDC converters are complex systems with a range of sensitive control systems. HVDC converters also,
by their basic operation, generate electromagnetic noise. In this section, the concern is with radiated
electromagnetic noise.

CHAPTER
ACÊHARMONIC
ACÊNETWORK FILTERSÊAND PLCÊFILTERS CONVERTER/
SUBSTATION SOURCE
BUSBARS
Fig. 4.8a– Block diagram of the HVDC converter system model
In general HVDC converters generate radiated noise in the range from 150 kHz to approximately 10 MHz.
At higher frequencies other sources such as corona are more significant. At lower frequencies the radiation
process is not efficient and conducted noise is more significant.
4.15a 4.8.1. RADIATED INTERFERENCE STUDIES

Studies are performed for the calculation of radiated interference performance of a converter station as
part of the design process. This study is required to confirm that the valve hall screening is adequate to
ensure the converter stations are within the limits specified.
The purposes of radio frequency interference studies are to:
➙ Outline and describe the sources of the radio interference and their methods of propagation
➙ Describe the method and design used to limit the radio interference
➙ Describe the method proposed to meet the performance requirements
Several established documents such as the CIGRÉ “Guide on Power Line Carrier” [15], the EPRI report “Radio
Interference from HVDC Converter Stations”, EL-3712 [19], and CISPR 18 [20] can be used as background
information.
To limit the emission of radio frequency interference (RFI) from the converter station building, the walls, floor
and roof of the valve hall will be enclosed in a grounded Faraday cage made, where possible, from a wire
mesh screen. Other options for screening include the use of a bonded metal building, the use of building
reinforcement, or even sheet metal screening.
4.8.2. DESIGN REQUIREMENTS

The high frequency noise radiated from the converter station needs to be limited to an acceptable level.
Reference [20] provides useful guidance on appropriate levels of interference limits. Some typical limits are
as follows:
Location Maximum Level

Any point outside the boundary fence of the A maximum of 100 microvolts per meter.
station and at least 500 m from the nearest To be measured with quasi peak detector at 1 MHz.
bar which connects the valves with the
converter transformers
The design requirement of the RFI shielding is that it should limit the interference appearing to the limits
specified.
4.8.3. OVERVIEW OF THE SYSTEM

The radio frequency (RF) noise generated by the converter stations can be categorized as follows:
➙ Direct radiation from the thyristor valves
➙ Radiation from the converter station busbars due to high frequency currents circulating in the AC
busbars and associated stray capacitances and stray inductance
CHAPTER
VoltageÊ(kV)
ValveÊvoltage dv/dtÊ(kV/μs)
200 18
16
150
14
RateÊofÊchangeÊofÊvalveÊvoltage
100 12
ValveÊvoltageÊ(kV)
10
(kV/μs)
50
8
0 6
0 2 4 6 8 10 12 14 16 18 20
4
-50
2
-100 0
TimeÊ(μs)
Fig. 4.8b– Typical valve voltage waveform
ValveÊVoltageÊ(kV) RateÊofÊchangeÊofÊvalveÊvoltageÊ(kV/μs) Turn-onÊevent
120 300
100 250
4.15b
RateÊofÊchangeÊofÊvoltageÊ(kV/μs)
80 200
VoltageÊ(kV)
60 150
40 100
20 50
2660 2665 2670 2675 2680 2685 2690

TimeÊ(μs)
Fig. 4.8c– Typical valve voltage turn-on waveform
➙ RF noise due to corona from the high voltage equipment and busbars
The scheme can be divided into a number of discrete elements: the thyristor valves, the converter transformer,
the AC systems (AC harmonic filters and a simplified representation of the relevant AC system). Fig. 4.8a
shows a block diagram of the converter station model used to analyze the radiated interference.
Because of the range of operating states for the HVDC converters and the uncertainty regarding the duration
in each mode, any prediction regarding frequency and duration of particular operating conditions is difficult to
4.15c
establish accurately without the benefit of operational experience. Because the valve turn-on event is rarely,
if ever, completely ‘coherent’, interference levels will not reach the maximum levels described below even
when a HVDC converter is operating at high breakdown voltages.

CHAPTER
The following sections describe the major elements of the converter stations in more detail.
4.8.4. NOISE SOURCE MODEL

There are two events during valve operation which cause high frequency noise: turn-on (which is the
sudden collapse of voltage across the valve as the thyristors ‘turn-on’ (the ‘breakdown voltage’), bringing
about the discharge of parallel connected capacitances) and turn-off. For RFI purposes, only turn-on is
significant, since turn off is a comparatively slow event (the rate of change of voltage is controlled by the
valve-damping network), so its influence on the frequency range of interest is small.
The turn-on event is modeled mathematically as an exponentially decaying voltage [21] leading to the
equation for equivalent noise voltage shown in reference [19] (equation 27) and repeated as Eqn 4.8a
below.
V 0 ⋅ K1
V eq ≈ [Eqn 4.8a]
ω ⋅ 1 + (ω ⋅ T 0 )2
Where:
Veq = equivalent noise voltage at angular frequency w
V0 = breakdown voltage
K1 = 6320 × A(n) for RF quasi peak measurement
A(n) = pulse repetition rate weighting factor from ANSI C63.2 = 1.3
w = angular frequency
T0 = breakdown time
4.8.4.1. Calculation of Parameters for Noise Source Calculation

The valve voltage waveform used to establish the minimum turn-on time is for a coherent firing event when
all the thyristors are considered to ‘turn-on’ simultaneously, giving the fastest voltage collapse across the
valve and maximizing the equivalent noise voltage. Such coherence of firing is pessimistic, because practical
component tolerances will cause slight differences in firing instants between individual thyristors. This
incoherence slows the voltage collapse across the valve, increasing the valve turn on time and reducing
the equivalent noise voltage.
The voltage between the terminals of the thyristor valve during turn-on is derived from the valve design
studies. For this purpose a detailed model of the thyristor valve and surrounding components and busbars
is used. From these studies a worst-case thyristor valve turn-on case is found (with maximum switching
voltage and minimum switching time) to maximize Eqn 4.8a above [19].
Fig. 4.8b shows a typical valve voltage waveform.
Fig. 4.8c shows the detail of the turn-on event showing the valve breakdown voltage (V0), breakdown time
(T0) and rate of change of voltage from the converter design.
An analysis of the waveform gives values of V0 and T0 to maximize the equivalent noise voltage in Eqn 4.8a
above.
The noise source model outlined above is based on extreme conditions, but in reality the switching voltage
will be lower and the turn-on time longer.
In practice, the model described above will give pessimistically high values for very high frequency noise
because the stray capacitances and high frequency resistances of practical thyristor valves and the building
will attenuate such high frequency noise.
4.8.4.2. Radiated Noise due to High Frequency Busbar Currents

The turn-on event is brought about by a change of valve impedance from near open circuit to near short-
circuit. The sudden collapse of voltage brings about the discharge of any parallel-connected capacitances,
generating oscillations in circuits that are affected by it.

CHAPTER
Precise modeling of this noise source to evaluate radiated noise is extremely laborious requiring a great
deal of information regarding the high frequency impedances of components, their precise location and
their orientation in the substation. At a distance of approximately 40 m (assuming 30 m to the station
perimeter) from the busbars it is possible, with little error, to use an equivalent busbar loop as an antenna
with appropriate current values. These high frequency current values are derived from the aforementioned
thyristor valve design studies by applying the calculated equivalent noise voltage to a detailed high frequency
digital model of the converter station.
4.8.4.3. Corona
The corona noise from the HVDC converter is similar in nature to that of conventional substations. Care is
taken to control corona in the HVDC converters by paying close attention to layout and equipment design.
Reference [20] describes briefly the corona performance of high voltage lines and equipment.
4.8.5. THE THYRISTOR VALVES

The 12-pulse converter is modeled in detail for the high frequency busbar current studies. Details of the
valve models for the various valve states are given in Fig. 4.7c to Fig. 4.7e.
The component values for Fig. 4.7b are based on the arrangement of busbars in the valve hall - in terms of
length and diameter of busbars and spacing of busbars from other busbars and from earth planes.
RadiationÊfromÊthyristorÊvalveÊoperation
100.00
UnshieldedÊnoiseÊ(dB)
80.00 LIMIT
ShieldedÊnoiseÊ(dB)
60.00
40.00
ElectricÊfieldÊ(dB/1µV/m)
20.00
0.00
-20.00
-40.00
-60.00
-80.00
-100.00
0.1 1 10 100 1000
FrequencyÊ(MHz)
Fig. 4.8d– Typical RF noise level at the converter station perimeter due to radiation from the thyristor valves -
showing the effects of valve hall screening (mesh size typically 100 mm × 100 mm)
4.15h
4.8.6. CONVERTER TRANSFORMER
The converter transformer model is the same as used for the PLC frequency studies described in section 4.7.7.

CHAPTER
4.8.7. THE AC SYSTEMS
The AC system model is the same as used for the PLC frequency studies described in section 4.7.8.
4.8.8. RFI SHIELDING DESIGN

The requirement for RFI shielding is obtained by analyzing the thyristor valve at discrete frequencies over
the required frequency range, typically 150 kHz to 30 MHz.
4.8.8.1. Valve Generated Noise

In the frequency range of interest, the wavelength of the valve radiation is comparable with the valve
dimensions. The resulting equation for electric field strength (Eq) at distance r is:
h
120 ⋅ π ⋅ I ⋅ sin(θ ) 2  2⋅π⋅ x r + x ⋅ Cos(θ ) 

Eθ =
2⋅r ⋅λ ∫h  cos( λ ) × Cos(ω ⋅ (t − c
))  dx

[Eqn 4.8b]
−
2
Where:
r = distance from valve
I =d isplacement current at wavelength l in valve due to change in voltage
q = angle between measurement point and valve orientation (0° vertically above valve, 90° level with valve)
h = height of valve
c = speed of light
l = wavelength of interest
x = position on radiating structure
The radiated noise in the frequency range of interest is shown in Fig. 4.8d, with and without the shielding
described below for the electric field component. (The magnetic field in µA/m is equal to the electric field
in µV/m when multiplied by the impedance of free space, 120p W in the far field region i.e. r > 300 m.)
4.8.8.2. Attenuation of Valve Generated Noise

Calculations based on the equation given in section 4.8.8.1 above indicate that a maximum mesh size of
typically 100 mm provides adequate shielding to attenuate the RFI noise to the level given in section 4.8.2 at
the converter station perimeter. The natural construction of the building can be considered approximately
equivalent to a wire mesh with wire spacing of approximately 1 m.
The relative permeability of steel falls with increasing frequency [23].
The steel wire mesh requires not less than 1 mm diameter steel wire with the mesh-hole size not exceeding
100 mm. The mesh is calculated to give the attenuation listed in Table 4.8a.

CHAPTER
Relative permeability (see note 1) 1 Impedance of
Permeability of vacuum (H/m) 1.26e06 vacuum (W) 376.82
Conductivity (S/m) 5.82e+06 Wire diameter
Source screen distance (m) 1.5 (mm) 1
3 m bonding to
Typically 100 mm bonding (from
Wire centers represent building
reference [23])
structure
Magnetic shielding Electric shielding Electric shielding

Frequency (MHz) Wave length (m)
(dB) (dB) (dB)
0.15 2000.00 33.6 126.5 3.97
0.20 1500.00 33.6 121.5 3.97
0.30 1000.00 33.6 114.5 3.97
0.50 600.00 33.6 105.6 3.97
1.00 300.00 33.6 93.6 3.97
3.00 100.00 33.6 74.5 3.97
10.00 30.00 33.6 53.6 3.97
30.00 10.00 33.6 34.5 3.97
100.00 3.00 23.6 23.6 0.00
300.00 1.00 14.1 14.1 0.00
1000.00 0.30 3.6 3.6 0.00
Note 1: At high frequencies, e.g. above 1 MHz, the relative permeability of most magnetic materials approaches unity
Table 4.8a– Properties of typical RFI screen
RadiationÊfromÊbusbarÊcurrent
100.00
UnshieldedÊnoiseÊ(dB)
LIMIT
50.00
ElectricÊfieldÊ(dB/1μV/m)
0.00
-50.00
-100.00
0.1 1 10 100 1000
-150.00
FrequencyÊ(MHz)
Fig. 4.8e– Typical RF noise level at the converter station perimeter due to conducted interference on the high voltage busbars

CHAPTER
The equations for the shielding effectiveness (SE) of a wire mesh are [23]:
For: g (wire gap in meters) > 150/f (in MHz)
➙ SE (dB) = 0 [Eqn 4.8c]
For: g < 150/f and r >> g
• Far field - r (source-screen distance) > wavelength/(2 ⋅ p)
➙ SEE (dB) = SEH (dB) = 20 ⋅ log (150/(g ⋅ f)) [Eqn 4.8d]
• Near field - r < wavelength/(2 p)
➙ SEH (dB) = 20 ⋅ log (p ⋅ r/g) [Eqn 4.8e]
➙ SEE (dB) = 77 – 40 ⋅ log (f) – 20 log (r ⋅ g) [Eqn 4.8f]
To ensure effective shielding, aperture losses must be minimized. The overall attenuation introduced by
the building cannot be better than that of the worst aperture. Considerable attention is given to minimizing
the size and number of apertures in the building. All doors and windows are bonded to the building to form
an effective part of the shield.
m
0
4.8.9. RADIATION DUE TO NOISE CONDUCTED TO THE EXPOSED BUSBARS
0
5
Some high frequency noise can be conducted to

the outdoor switchyard through the converter
transformers. The appropriate noise voltage from
X
Converter
the thyristor valves is simulated and applied to station
500Êm
the models described above. The resulting circuit fenceÊ
is analyzed over a range of frequencies and the
results are plotted in Fig. 4.8e.
500Êm 500Êm
4.8.10. SITE MEASUREMENTS X X
A program of site measurements can be performed
to verif y the achievement of the specified
performance requirements.
500Êm
Measurements are performed using a spectrum

analyzer and a suitable antenna. The interference m
0
tests are scheduled to allow for measurements X = measurement point 0
5 X
with the converters shut down and with the
converters operating at rated output. Fig. 4.8f– RFI measurement points around the
converter station
Radiated interference measurements are made at
a number of locations around the converter station. Typical locations are shown in Fig. 4.8f.
4.8.11. ELECTROMAGNETIC COMPATIBILITY (EMC) –

IMMUNITY ISSUES
The complex electronic systems in the converter station need to be sufficiently immune to the interference
from converter station operation and maintenance. Specifically the installation needs to be sufficiently
immune to the following phenomena:
➙ Electrostatic discharge from personnel
➙ Electric surges
4.15j
➙ Bursts of conducted interference due to switchgear and relay operation
➙ Radiated interference from converter operation and also communications such as radio, television
and mobile phones. This requirement is particularly onerous for the equipment in close proximity
to the converter valves.
➙ Power supply interruptions and variations
These phenomena and appropriate test methods and levels where required are described in the
IEC 61000-4 series of standards.
CHAPTER
BIBLIOGRAPHY
[1] E. W. Kimbark, “Direct Current Transmission”, Vol. 1, John Wiley & Sons, 1971.
[2] “IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems”,
IEEE519.
[3] “Limits – Assessment of harmonic emission limits for the connection of distorting installations to MV, HV
and EHV power systems”, IEC 61000-3-6.
[4] “Harmonics, Characteristic Parameters, Methods of Study, Estimates of Existing Values in the Network”,
CIGRÉ Working Group 36.05, Electra, No. 77, July 1981, p. 35-54.
[5] “Guide for Assessing the Network Harmonic Impedance”, CIGRÉ Working Group 36.05/CIRED 2, Electra,
No. 167, August 1996.
[6] “Impact of Aggregate Linear Load Modelling on Harmonic Analysis: A Comparison of Common Practice
and Analysis Models”, IEEE Task Force on Harmonics Modelling and Simulation, IEEE Transactions on Power
Delivery, Vol. 18, No. 2, April 2003.
[7] J. Arrillaga et al., “Power System Harmonic Analysis”, John Wiley & Sons Ltd, 1997.
[8] “Modelling and Simulation of the Propagation of Harmonics in Electric Power Networks”, Task Force on
Harmonics Modelling and Simulation, (Part I & Part II), IEEE Transactions on Power Delivery, Vol. 11, No. 1,
January 1996, p. 452-474.
[9] “Test Systems for Harmonics Modelling and Simulation”, Task Force on Harmonics Modelling and
Simulation, IEEE Transactions on Power Delivery, Vol. 14, No. 2, April 1999, p. 579-587.
[10] “Recommended Practice for Determination of Power Losses in High Voltage Direct-Current (HVDC)
Converter Stations”, IEEE Standard 1158.
[11] “Determination of power losses in high voltage direct current (HVDC) converter stations”, IEC 61803.
[12] “Guide to the specification and design evaluation of AC filters for HVDC systems”, CIGRÉ Brochure 139, April
1989.
[13] “IEEE Guide for analysis and definition of DC side harmonic performance of HVDC transmission
systems”, IEEE Standard 1124.
[14] “High voltage direct current (HVDC) systems – Guidebook to the specification and design evaluation
of AC filters”, IEC/TR 62001.
[15] “The Telephone Influence Factor of Supply System Voltages and Currents”, Supplement to Engineering
Report No. 33, Joint Committee on Development and Research, Edison Electric Institute and Bell Telephone
System, E.E.I. Publication 60-68, September 12, 1960.
[16] Directives Concerning the Protection of Telecommunication Lines Against Harmful Effects from
Electricity Lines, C.C.I.T.T., published by International Communication Union, Geneva, 1963.
[17] “Guide on Power Line Carrier”, CIGRÉ Brochure 015, 1979.
[18] “IEEE Guide for Power-Line Carrier Applications”, IEEE 643.
[19] “Radio Interference from HVDC Converter Stations”, EPRI Report EL-3712.
[20] “Radio Interference Characteristics of Overhead Power Lines and High Voltage Equipment, Part 1:
Description of Phenomena”, CISPR Publication 18-1.
[21] Stig. A. Annestrand, “Radio Interference from HVDC Converter Stations”, IEEE PAS Vol. PAS 91, May-
June 1972, p. 874-882.
[22] “Coupling devices for power line carrier systems”, IEC 60481.
[23] White, R. J. Donald and M. Mardiguian, “Electromagnetic Shielding”, Interference control Technologies Inc,
1988.

CHAPTER
Y
AR
4| AC/DC SYSTEM INTERACTIONS
IN
IM
EL
PR
TOC 230 | DC
5 AC/DC SYSTEM
INTERACTIONS
It may not surprise you that interactions occur between a HVDC
converter and the AC network to which it is connected. Any HVDC
scheme which is introduced into an AC system will interact with that
system in terms of both real power exchange and reactive power
exchange. The impact of these interactions on both the steady-state
and dynamic behavior of the AC system and the DC system will be
influenced by the relative power transmission capacity of the HVDC
link with respect to the AC system Short-Circuit Level at the point
where the HVDC link is connected.
In this chapter you will gain a detailed understanding of these
interactions - together with the effect of AC network characteristics
on the control of the HVDC converter. We also consider the static
characteristics of the HVDC converter for stable operations under many
different AC and DC system operating conditions, for both rectifying
and inverting modes, as well as multi-terminal characteristics.
Methods of predicting and calculating the steady-state and temporary
expected performance are introduced along with the design techniques
which will influence this behavior.
The chapter concludes by illustrating how the HVDC system is
represented in system studies.

Chapter contents
AC/DC SYSTEM INTERACTIONS ........... 230
5. 5.8. BACK-TO-BACK HVDC SCHEMES
– STATIC CHARACTERISTICS................................................. 258
5.1. EFFECT OF THE AC SYSTEM SHORT-CIRCUIT 5.8.1.
Back-to-Back HVDC Converter
LEVEL ON AC / DC SYSTEM INTERACTION..................... 233 – AC Voltage Control................................................................ 261
5.8.2.
Back-to-Back HVDC Converter
5.2. P OWER TRANSMISSION INTO AN AC SYSTEM – TCR Mode................................................................................... 262
FROM A HVDC CONVERTER.................................................. 233
5.2.1. Maximum Power Curve .......................................................... 234 5.9. OTHER CONTROL CHARACTERISTICS.............................. 264
5.2.2. Minimum SCL Required for a 5.9.1. Rectifier Voltage Control........................................................ 265
Given Power Transmission Level........................................ 234
5.2.3. Alternative Calculations of Minimum SCR...................... 236 5.10. PROPOSED DEVELOPMENTS................................................ 266
5.2.4. Impact of CESCR on Power Controller Design.............. 237 5.10.1. Multiple HVDC Infeeds to AC Systems............................. 266
5.2.5. Multi-Infeed HVDC.................................................................... 238 5.10.2. Isolated AC Systems and Wind Farms............................. 266
5.3. SYSTEM INERTIA ....................................................................... 240 5.11. S TABLE OPERATION OF A HVDC CONVERTER
5.3.1. Interruptions in DC Power Transfer................................... 241 IN AN AC SYSTEM...................................................................... 268
5.11.1. Load Flow Studies..................................................................... 268
5.4. HVDC CONVERTER STATIC CHARACTERISTICS............ 242 5.11.2. Objectives and Criteria........................................................... 268
5.4.1. Basic Requirements of the 5.11.3. Balanced Load Flow
Converter Control System..................................................... 243 (Newton Raphson Algorithm) ............................................. 269
5.11.4. Unbalanced Load Flow............................................................ 271
5.5. EFFECTS OF AC SYSTEM VOLTAGE.................................... 248
5.5.1.
Effects of Sending End AC Voltage 5.12. SYSTEM REPRESENTATION................................................... 271
on Rectifier Operation............................................................. 248
5.5.2.
Effects of Receiving End AC Voltage on BIBLIOGRAPHY ............................................................................................ 275
Inverter Operation.................................................................... 251
5.5.3. Multiple Operating Points...................................................... 251
5.6. MULTI-TERMINAL HVDC SCHEMES.................................... 251
5.7. TYPICAL TRANSMISSION SCHEME

STATIC CHARACTERISTICS.................................................... 254
5.7.1. Rectifier Characteristics......................................................... 254
5.7.2. Inverter Characteristics.......................................................... 256

5.1. EFFECT OF THE AC SYSTEM SHORT-CIRCUIT LEVEL
ON AC / DC SYSTEM INTERACTION
The Short-Circuit Level (SCL) is a measure of the current that can be delivered into a solid 3-phase fault at a
particular bus in an AC system, assuming nominal driving voltage. SCL is defined by the following equation:
SCL = 3⋅ Vac ⋅ I ac [Eqn 5.1a]
Where:
SCL = the Short-Circuit Level at the bus under examination
Vac = the nominal phase-to-phase RMS voltage at the bus
Iac = the RMS solid 3-phase line fault current
Therefore, the AC system behind any particular
VAC ∠0 °
AC bus can be represented as a simple Thévenin
ZÊSYS ∠θ
equivalent circuit with the phase voltage derived
from Eqn 5.1b and the impedance derived from
Eqn 5.1c. The equivalent circuit is shown in Fig. 5.1a.
Vac
V1 = [Eqn 5.1b]
3
Vac2
Z SYS = [Eqn 5.1c]
SCL
Fig. 5.1a– Thévenin equivalent circuit
5.2. POWER TRANSMISSION INTO AN AC SYSTEM

FROM A HVDC CONVERTER Ê5.1a
A simplified representation of a DC link feeding 1.0 p.u. power into an AC system of impedance ZSYS is shown
in Fig. 5.2a(i). The AC system voltage Vac lags the converter bus voltage VCONV by a phase angle f, and VCONV
has a voltage magnitude of 1.0 p.u. at a phase angle of 0°. Assuming a converter absorbs around 0.55 p.u.
reactive power while transmitting 1.0 p.u. real power, the converter reactive power absorption will be
compensated for by shunt connected capacitor banks, which may be configured as AC harmonic filters.
The shunt connected capacitor banks are typically rated to give a power factor of 1.0 at rated power and at
a converter bus voltage of 1.0 p.u. Therefore, the shunt capacitance generates 0.55 p.u. reactive power to
fully compensate the reactive power being absorbed by the converter. The phasor diagram for the circuit
in Fig. 5.2a(i) is shown in Fig. 5.2a(ii) assuming that ZSYS is purely inductive, that is, q is 90°.
In Fig. 5.2a(i) the phase angle f of the AC system voltage Vac is a function of the current IL injected into
the AC system through the system impedance ZSYS. The phasor diagram in Fig. 5.2a(ii) corresponds to the
converter in combination with the shunt capacitor operating at unity power factor and hence IL is a purely
real current flowing from the HVDC inverter. AC system impedance voltage phasor IL ⋅ ZSYS leads the current
IL by an angle q, where:
X 
θ = tan −1  SYS  [Eqn 5.2a]
 RSYS 
• XSYS/RSYS is the inductive reactance/resistance ratio of the AC system impedance ZSYS.
The phasor diagram in Fig. 5.2a(iii) shows the influence on the network of an increase in the inverter
current, IC, beyond 1.0 p.u. The inverter absorbs more reactive power, creating a reactive power deficit at
the inverter station bus. The inverter bus voltage, VCONV, is no longer in phase with the increased current
IL flowing into the AC system. In order to maintain the inverter bus voltage VCONV at 1.0 p.u. under these
conditions, the AC system voltage Vac must be increased, otherwise VCONV will fall, reducing the power at
the converter bus, VCONV ⋅ IL.

CHAPTER
VAC ∠φ IL VCONV ∠0°

ZÊSYS ∠θ
IQ Ic
(i) +Ê1.0ÊP
-Ê0.55ÊQ
VAC
IQ
VAC
φ)
IQ (imaginary component)
VCONV
IL
Ic
IL (real component) VCONV
Ic' (imaginary component)
IC (ii) (iii)
Fig. 5.2a– Power transmission into an AC system

(i) simplified AC system
(ii) phasor diagram for (a)
(iii) phasor diagram when current Ic is increased
Ê5.2a 5.2.1. Maximum Power Curve

The Maximum Power Curve (MPC) represents the limit on power transfer from the inverter to the AC
system with respect to the DC current [1] [2] [3]. This is similar to the AC power system stability limit, which
occurs as the angle between the AC transmission
line voltage approaches 90o (Fig. 5.2b). The peak
value of power that can be transferred into the AC Pac MAP
system is known as the Maximum Available Power
(MAP) and is plotted in Fig. 5.2b for a constant MPC for γ CONSTANT
extinction angle, g.
We can see from Fig. 5.2b that prior to point
A an increase in the DC current Id will cause a
corresponding increase of the power flow into the Id
A
AC system. Any increase in Id beyond point A will
Fig. 5.2b – Maximum power curve
cause the converter bus voltage to fall and hence
the power into the AC system to reduce.
Ê5.2b
5.2.2. Minimum SCL Required for a Given Power Transmission Level
In order for a HVDC converter to be able to transmit power into an AC system at a given connection point
(or bus), the SCL at that point must be sufficient. It is conventional to express the relationship between
the SCL of a given AC system bus and the quantity of HVDC scheme power (Pdc) into the AC system as a
ratio, known as the Short-Circuit Ratio (SCR).
SCL
SCR =
Pdc [Eqn 5.2b]
CHAPTER
A strong AC System is defined as one with an SCR greater than 3, while a weak AC system is defined as
one with an SCR between 2 and 3 [2].
Consider an AC system: looking out from the converter station terminals the AC system appears as a
simple Thévenin equivalent circuit, comprising an AC voltage source behind an impedance as shown in
Fig. 5.2a(i). Under steady-state operation at 1.0 p.u. power transfer and 1.0 p.u. AC voltage, the passive
capacitive reactive power equals the reactive load of the converter, thus there is no reactive exchange
with the AC system.
Now, assume that the AC system voltage is reduced by 5% (where ±5% is a typical steady-state range
of operation for an AC system). If the converter is operating on a purely Constant Extinction Angle (CEA)
characteristic the change in the reactive load of the converter will be small. However, the change in the
capacitive reactive power provided by the passive shunt compensation will reduce by the square of the
voltage change, that is, the passive reactive power elements will now only be generating approximately 90%
of their rated value. There is, therefore, a shortfall in reactive power provision local to the converter station,
which must be supplied by the AC system. The reactive power flow through the AC system (as represented
by a Thévenin equivalent impedance in Fig. 5.2a(i)) will cause a voltage drop across the network, further
depressing the converter bus AC voltage, thereby further increasing the reactive power drawn from the AC
system. The magnitude of this voltage drop is dependent on the SCL of the AC system, that is, the lower the
AC system SCL, the higher the equivalent AC system impedance and hence the larger the voltage drop [4].
With constant DC current, a reduction in converter bus AC voltage will result in a reduction in the AC power
transferred into the AC converter bus from the converter. If the converter attempts to maintain constant
AC power then the DC current must increase, thus increasing the reactive power absorbed by the converter
and consequently further reducing the converter bus AC voltage, as discussed in section 5.2 above. This
describes a positive feedback effect, which could lead to instability.
Fig. 5.2c shows a plot of AC power (Pac) versus DC current for an inverter operating into an AC system for
various SCRs. We can see that the threshold of stability can be taken as an SCR of 2, though this has no
margin for disturbance. In practice, with weak AC systems, pure CEA control is not usually implemented,
instead other control techniques are used which can provide some negative feedback in terms of the
converter station response to AC voltage reductions and therefore, SCRs lower than 2 are achievable [5].
When the active power transmitted is reduced, the reactive power absorbed by the converters will reduce
proportionally if the extinction angle g is kept constant. Hence, there will be a surplus of capacitive reactive
power left connected to the converter busbar, which will lead to a high converter AC bus voltage as shown
in Fig 5.2c.
2
VacÊ(SCRÊ=Ê1.5)
1.8
1.6
VacÊ(SCRÊ=Ê2.0)
1.4 PacÊ(SCRÊ=Ê5.0)
VacÊ(SCRÊ=Ê3.0)
1.2
PacÊ(SCRÊ=Ê3.0)
VacÊ(SCRÊ=Ê5.0)
(p.u.)
1
0.8
0.6
PacÊ(SCRÊ=Ê2.0)
0.4
PacÊ(SCRÊ=Ê1.5)
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
DCÊcurrentÊ(IdÊp.u.)
Fig. 5.2c – Variation in injected AC power with DC current at various SCRs, assuming: commutating reactance (Xc) =
0.15 p.u., g = 18°, Qfilter = 0.55 . Pdc
BACK TO Ê5.2c
CHAPTER
5.2.3. Alternative Calculations of Minimum SCR
The SCR indicates the impact the transmitted power can have on the local converter bus voltage. However,
this relationship ignores the impact of the local, shunt connected, reactive power banks used to compensate
for the reactive power absorbed by the converter. In order to take the local shunt reactive power banks into
account, the SCR must be modified, giving the Equivalent Short-Circuit Ratio (ESCR) which is defined as:
SCL − Q f
ESCR = [Eqn 5.2c]
Pdc
Where:
Qf = MVArs generated by the shunt connected reactive power banks at rated Vac.
As discussed in section 5.2.2, beyond a certain level of power transmission, any further increase in the
DC current for constant extinction angle will result in a reduction in the converter bus AC voltage and
transmitted power. For a given AC system, the relationship between the SCL and the maximum DC power
that can be transmitted before the converter bus AC voltage falls can be defined as the Critical Equivalent
Short-Circuit Ratio (CESCR), which can be approximated by Eqn 5.2d [1] [6]. The equation for CESCR
assumes that the AC system is purely inductive, that is q in Fig. 5.2a equals 90°.
Pdc ⋅ cot (φ 2 ) − Qdc

CESCR = [Eqn 5.2d]
Vc2
Where:
f = 90° – g – m [Eqn 5.2e]
g = inverter extinction angle (°)
m = inverter overlap angle (°)
Pdc = active power supplied to the AC system (p.u.)
Qdc = reactive power consumed by the inverter (p.u.)
Vc = converter bus AC voltage (p.u.)
In order to understand the application of the equation for CESCR, consider an AC system into which it is
proposed to connect a 500 MW HVDC transmission system. The HVDC inverter operates in CEA control
(see section 7.2.3.1) with an extinction angle of 18° and a commutating reactance of 0.15 p.u. From Eqn
2.2.2i and Eqn 2.3c we calculate that the overlap (m) is 18.8° and the converter absorbed reactive power
(Qdc) is 0.55 p.u.
Assume that the AC voltage on the bus where the converter will be connected may transiently fall to 0.95
p.u. and the converter is expected to continue to transmit rated power. Here, a transient disturbance is
considered as the response during the 100 ms to 300 ms after the disturbance, when the converter has
had time to respond, but the converter transformer tapchanger and the AC generator automatic voltage
controllers, etc. have not had time to act. In the case of the example scheme above, the operating gamma
will remain at 18°, but because of the reduction in AC voltage and the resultant increase in DC current, in
an attempt to maintain 1.0 p.u. AC power the overlap will increase to 20.3°.
Assuming that there is sufficient shunt connected capacitive reactive power to achieve unity power factor
at 1.0 p.u. AC voltage, then the CESCR becomes:
 90 − 18 − 20.3 

1.0 ⋅ cot   − 0.55
 2 [Eqn 5.2f]
CESCR =
0.95 2
CESCR = 1.677
Therefore, the minimum SCL at the converter bus for the Eqn 5.2f is calculated as follows:
SCL = (CESCR + Qf) ⋅ MW
SCL = (1.677 + 0.55) ⋅ 500 MW
SCL = 1114 MVA
CHAPTER
Hence, if system studies show that under line or generator outage conditions the SCL at the proposed
converter bus can fall below 1114 MVA, attempting to operate with constant power control at or near to
rated power would not be possible.
Note: Fig. 7.2g illustrates how the minimum SCR necessary to avoid power loop instability varies with the maximum applied
voltage step.
5.2.4. Impact of CESCR on Power Controller Design

When designing a HVDC scheme there are a number of inverter control strategies which can be
implemented (see section 7.1 for a detailed discussion on control methods). These strategies broadly fall
into two categories:
➙ Constant extinction angle (constant gamma control): the inverter is operated with just sufficient time
for the thyristors to recover their blocking capability at all power levels
➙ DC voltage control: the operating DC voltage is maintained at a target value while the extinction
angle g is allowed to vary as the power transmitted varies
Analysis of the impact of these control schemes can be best understood by considering the simple circuit
shown in Fig. 5.2d(i) below, where a HVDC inverter is connected to a Thévenin equivalent network via two
parallel AC transmission lines. Each of the two parallel AC transmission lines has circuit breakers at either
end for protection purposes. In the event of the protection tripping one line, the converter station will then
be supplied via one transmission line only, reducing the SCL at the converter bus.
5.2.4.1. Constant Extinction Angle Control

As the current through the inverter increases from zero, the power transferred into the AC system will
increase along the line OABC shown in Fig. 5.2d(ii). If however, the current continues to increase beyond
the point corresponding to Idmax , the power into the AC system will decrease as shown by the line CD. The
inverter operation will remain stable at Id of 1.0 p.u. as long as this operating point is lower than Idmax.
In the event of a line tripping, the SCL at the converter bus will decrease and the Pd/Id characteristic will
change to OEAFG as shown in Fig. 5.2d(ii). This characteristic has a new maximum power transfer capability
point ‘E’ which is lower than 1.0 p.u. Consequently, the transferred DC power must be reduced to guarantee
stable operation.
x x
x x
C M
E D ConstantÊγÊminimum L
B
ConstantÊγÊminimum
Power
A K
Power
F
G
ConstantÊvoltage
0 Id Idmax Id 0 Id Idmax Id
Fig. 5.2d– The impact of ESCR on power transmission

(i) Simplified circuit
(ii) The impact of the loss of a line on a constant g controller
(iii) The operation of constant voltage
Ê5.2d

CHAPTER
5.2.4.2. Voltage Control
In a HVDC scheme where the operating voltage (either the DC voltage or the valve winding voltage) is
controlled at a target value, the extinction angle at which the converter operates will increase at power
levels below 1.0 p.u. Fig. 5.2d(iii) shows a Pd/Id characteristic similar to the curve in Fig. 5.2d(ii). However, it
can be seen in Fig. 5.2d(iii) that with constant voltage control the converter operates with a large margin
throughout most of the current control range:
➙ The power limit corresponding to point ‘M’ is observed to be of little consequence in constant
voltage control
➙ Point ‘L’ defines the ‘power limit’ for the constant voltage that can be seen to represent a sudden
change from a linear control characteristic to the unstable side of the maximum power transfer curve
➙ By rating the converter to operate at point ‘K’ under normal steady-state operating conditions the
inverter will be operating at an extinction angle gamma in excess of the minimum. This will increase
the rating and hence the capital cost of the equipment compared to a scheme designed for constant
extinction angle control
5.2.5. Multi-Infeed HVDC

With the proliferation of HVDC converters in power systems, one issue that cannot be ignored is that of
the interaction between converters which are electrically in close proximity to each other. By electrically
close, we mean that changes in the voltages at one converter busbar are reflected onto another converter’s
busbar, which can result in interactions between the converters.
The most notable interaction between converters connected to a common AC system is the risk of a
cascade commutation failure event (see section 2.2.3.3), where the recovery of one converter following a
commutation failure results in other converters also suffering a commutation failure.
When a converter suffers a commutation failure, the DC voltage collapses to zero. As the converter
recovers, the operating point will follow the static characteristic and during part of this recovery the
converter will in fact be absorbing more reactive power than is absorbed during steady-state (see Fig. 5.7a).
This means there will be insufficient reactive power generated by the capacitive reactive power banks to
compensate for the converter’s absorption, resulting in additional reactive power being drawn from the
AC system, thus depressing the AC voltage. This temporary depression of the AC voltage can lead to a
commutation failure on another converter, cascading the event between converters. This event is even
more obvious if the initiating event was a fault in the AC system (for example a single-phase to ground
fault) directly causing a commutation failure on multiple converters.
As these converters independently attempt to recover from the fault they will each follow their own static
characteristic. However, if they are all similar, then at around the same time they will each draw additional
reactive power from the AC system causing a depression of the AC voltage: where the magnitude of
the depression will be dependent on the strength of the AC system at each converter bus. Clearly, by
coordinating the converter station characteristics during recovery - and even possibly delaying the recovery
of some converters following an AC system disturbance - improvements in the recovery of the AC system
are possible. However, it may not always be feasible to coordinate the static characteristics and recovery
times in an AC system, particularly when these converters are owned by different parties.
Considerable work has been undertaken in order to establish a means of assessing the interaction between
converters in an AC system. This has led to the publication of CIGRÉ Brochure 364 [7], which introduces the
concept of the Multiple Infeed Interaction Factor (MIIF) and the related Multi Infeed Effective Short-Circuit
Ratio (MIESCR) calculation.
5.2.5.1. Multi Infeed Interaction Factor (MIIF)

As discussed above, variations in the AC bus voltage at one converter will result in variations in the AC
bus voltage at another converter. The level of interaction between these two AC voltages will be related
to the AC system impedance between the two converter buses. In order to establish a measurement of
this interaction, a 1% voltage step reduction is artificially induced at one converter bus in an AC system
model then the resultant step change on a second converter bus is measured. This is the Multi Infeed
Interaction Factor:

CHAPTER
∆Ve [Eqn 5.2g]
MIIF =
∆Vn
Where:
ΔVe = voltage change at observed bus
ΔVn = voltage step induced at initiating bus
Given the above equation, if the two converters are on the same AC busbar, that is they are both physically
and electrically adjacent to each other, the MIIF will be 1.0. However, if the two converter busses are
isolated from each other, or at least very remote, the MIIF will be 0.0.
In most practical cases the MIIF associated with converters in a common AC system will lie somewhere
between 1.0 and 0.0. Where multiple converters coexist in the same AC system, or are planned as additional
improvements to the AC system, the MIIF can be calculated for each converter AC bus to converter AC bus,
forming a matrix of values.
Observed AC bus
MIIF
Inverter 1 Inverter 2 Inverter 3
∆V1 ∆V2 ∆V3
Inverter 1 MIIF1,1 = MIIF2,1 = MIIF3,1 =
∆V1 ∆V1 ∆V1
AC bus at which ∆V1 ∆V2 ∆V3
voltage Inverter 2 MIIF1,2 = MIIF2,2 = MIIF3,2 =
∆V2 ∆V2 ∆V2
reduction is applied
∆V1 ∆V2 ∆V3
Inverter 3 MIIF1,3 = MIIF2,3 = MIIF3,3 =
∆V3 ∆V3 ∆V3
Table 5.2a– A three-element MIIF matrix
Consider the simple example of an AC system which contains three HVDC converter terminals. A matrix of
MIIFs can be constructed as shown above in Table 5.2a.
Having established the MIIF matrix in Table 5.2a, it is possible to look at the relative impact of each converter
on the other by multiplying the MIIF by the DC power rating of the converter at the observed bus, creating
Table 5.2b.
Observed AC bus
MIIF
Inverter 1 Inverter 2 Inverter 3
Inverter 1 Pdc1 Pdc1 MIIF2,1 ⋅ Pdc2 MIIF3,1 ⋅ Pdc3
AC bus at which
voltage reduction Inverter 2 Pdc2 MIIF1,2 ⋅ Pdc1 Pdc2 MIIF3,2 ⋅ Pdc3
is applied
Inverter 3 Pdc3 MIIF1,3 ⋅ Pdc1 MIIF2,3 ⋅ Pdc2 Pdc3
Table 5.2b– A three-element MIIF matrix comparing converter rating influence
As a guide to the amount of interaction, CIGRÉ [7] proposes that the resultant power of the observed AC
bus is compared to the rated power of the converter at the bus on which the voltage reduction is applied.
If the resultant of MIIF ⋅ Pdc is less than 15% of the local converter rating then there will be negligible
interaction between these two converters. However, if the resultant is between 15% and 40% there
will be some possibility of interaction and, if the value is above 40%, there will be a strong possibility of
interaction between the converters. It can be seen from this analysis that a large, remote HVDC converter
may influence the behavior of a particular converter far more than a closer, but smaller converter.
As an example, consider two inverters transferring power into an AC system: the first rated at 2500 MW
dc (for example a large bulk-power transmission line) and the second at 250 MW dc (for example a small
back-to-back tie). Assuming the MIIF from converter 1 and converter 2 (MIIF2,1) is 0.8, whilst the MIIF from
converter 2 to converter 1 (MIIF1,2) is 0.7, this gives:

CHAPTER
Observed AC bus
MIIF
Inverter 1 Inverter 2
Inverter 1 2500 MWdc 2500 0.8 × 250 = 200 (8%)

AC bus at which voltage
reduction is applied
Inverter 2 250 MWdc 0.8 × 2500 = 1750 (700%) 250
Table 5.2c– Two-element MIIF matrix comparing converter rating influence

We can see from the above table that in this example Inverter 2 would appear as a small, 8% converter
when referred to Inverter 1 converter AC bus. However, when referring Inverter 1 to the Inverter 2 converter
bus, Inverter 1 will dominate, with a 700% increase in injected power.
5.2.5.2. Multi Infeed Effective Short-Circuit Ratio (MIESCR)

The ESCR for one converter is described above in section 5.2.3. This measurement of a converter’s power/
voltage stability within an AC system assumes that only one converter is connected to that AC system,
or to put it another way, that all of the short-circuit power of the AC system is available to that converter.
If we consider multiple converters within an AC system, it can be seen that the total short-circuit strength
of the AC system must (in some way) be shared between the converters. Having established the Interaction
Factor MIIF in section 5.2.5.1, then following the CIGRÉ brochure 364 [7], we can use this to define the
total power at a given converter bus, i, considering all other converters connected in the AC system, giving:
Pdctotal = Pdci + ∑ j MIIFj ,i ⋅ Pdc j [Eqn 5.2h]
Where:
j represents all electrically connected converters. This gives:
SCLi − Q f i
MIESCR = [Eqn 5.2i]
Pdci + ∑ j MIIFj ,i ⋅ Pdcj
A critical value of MIESCR can be established for each converter in order to assess the power/voltage
stability limit. Unlike the simple evaluation of ESCR, in order to find the multi infeed quantities transient
stability analysis tools are required in order to find the MIIF figures. Taking the example shown in Table
5.2d, we can see the effect on the ESCRs when considering the effect of MIIFs.
P DC ESCR Pdci + ∑ j MIIFj,i + Pdcj MIESCR
Inverter 1 2500 MWdc 3.0 2700 2.3

Inverter 2 250 MWdc 30.0 2000 3.7
Table 5.2d– Two-element MIIF matrix comparing converter rating influence
The table shows that the ESCR for Inverter 1 falls from 3.0 to 2.3, therefore potentially changing the dynamic
stability of Inverter 1. However, the impact on the operating ESCR of Inverter 2 is much more significant
when the impact of being interconnected with Inverter 1 is considered.
5.3. SYSTEM INERTIA

As discussed in section 2.2.1.2 a line-commutated converter requires the presence of an existing AC
system in order to transmit power into a load. Where the HVDC infeed represents a major part of the supply
provision or even when the converter represents the sole infeed into an AC system [8], a synchronous
compensator will be required.
The synchronous compensator provides two basic functions. Firstly, the transient reactance of the
synchronous compensator will be electrically in parallel with the AC system impedance, reducing the
short-circuit impedance as seen from the converter station and hence improving the ESCR. Secondly, the

CHAPTER
inertia of the synchronous compensator will act as a temporary energy storage device for the AC system.
The inertia of a synchronous compensator is a function of both its mass and speed. It is worth noting that
larger machines do not inherently have a greater inertia, as for larger MVA ratings the number of poles in
the machine will increase and consequently the speed and hence inertia will reduce.
5.3.1. Interruptions in DC Power Transfer

The relationship between change of machine frequency (df), mechanical power input (Pm) and electrical
power output (Pe) for small changes of frequency can be represented by:
df =
( Pm − Pe )⋅ fo ⋅ dt [Eqn 5.3a]
2⋅H
Where:
fo = system nominal frequency (Hz)
H = the inertia constant of the machine (MW·s/MVA)
Pm = machine mechanical power (p.u.)
Pe = machine electrical power (p.u.)
The inertia constant (H) can be converted to the base of DC power to give an effective inertia constant Hdc.
MVA rating of the machine

H dc = H ⋅ [Eqn 5.3b]
MW rating of the DC system
Substituting eqn 5.3b into eqn 5.3a gives:
df =
( ∆Pdc )⋅ fo ⋅ dt [Eqn 5.3c]
2 ⋅ H dc
or
H dc =
( ∆Pdc )⋅ dt ⋅ fo [Eqn 5.3d]
2 ⋅ df
5.3.1.1. Faults at the Sending-end AC System

Consider a 100 MWdc interconnection. If the typical maximum AC fault duration is of the order
of 100 ms, the recovery time of the DC link is 200 ms and the maximum permissible decline in AC system
frequency is -0.05 p.u. The mechanical power into the synchronous compensator will be zero and the
electrical power out of the synchronous compensator will be 1.0 p.u., when the power is no longer being
supplied from the HVDC link, which gives:
H dc =
( 0 − 1)⋅(100 + 200 )⋅10 −3 ⋅1 [Eqn 5.3e]
2 ⋅ 0.05
= – 3 MW ⋅ s/MVA
A conventional H constant for a synchronous compensator is approximately 1.25. Therefore, from Eqn 5.3b:
MVA rating of the machine

3 = 1.25 ⋅ [Eqn 5.3f]
100 MW
This gives a minimum machine rating of 240 MVA.

It may be possible to reduce the rating of the machine by increasing its inertia by, for example, adding a
flywheel to the machine.

CHAPTER
5.3.1.2. Load Rejections at the Receiving-end of a HVDC Link
The equations used are the same as those previously applied above, except that the change in power is
positive and it is the increase in system frequency that is relevant.
Consider an interruption of transmission of 200 MW, a fault duration of 100 ms, a HVDC link recovery time
of 300 ms and maximum permissible increase in frequency of +0.05 p.u. Here, the mechanical power into
the synchronous compensator will be 1 p.u. while the electrical power out of the synchronous compensator
will be zero. This gives:
Idc
Xl
Xl
Vdc
Xl
Fig. 5.4a– Basic 6-pulse thyristor DC converter circuit ('Graetz bridge')
[Eqn 5.3g]
(1− 0 )⋅(100+200) 10−3 1 Ê5.4a
H dc =
2 ⋅ 0.05
= ⋅
= – 3 MW ⋅ s/MVA
and again, assuming a synchronous compensator H constant of approximately 1.25 and using Eqn 5.3f
gives a minimum machine rating of 240 MVA.
5.4. HVDC CONVERTER STATIC CHARACTERISTICS

The control of each converter during steady-state and transient conditions is carried out by the pole control
system. The mode of control depends on the AC and DC conditions prevailing at the time. The various
steady-state control modes are referred to as the static characteristics [9] or the station characteristics.
In Fig.5.4a a simple thyristor converter is connected to a variable DC voltage. By turning on the thyristors
at their minimum firing angle (amin), current (Idc) will flow from the converter into the DC voltage source.
The magnitude of the DC current is determined by the difference in magnitude of the AC and DC sources
and the circuit impedance, which in this case is simply the converter transformer reactance (Xl). By plotting
the DC current for different but static DC voltages, the converter’s natural steady-state minimum firing
angle characteristic can be determined.
If the thyristor firing is delayed for a given combination of AC and DC voltages, the DC current will be
reduced. By placing the thyristor delay angle (a ) within a closed loop control system the DC current can
be kept at a fixed value of DC current (Iorder) regardless of the DC voltage. While the DC voltage remains
positive the power flow is from the converter into the DC system and the delay angle will be less than or
equal to 90˚ (electrical). The converter is then said to be in rectification (it is a rectifier). The rectifier will
therefore have two basic characteristics: a constant minimum firing angle control (usually referred to as
amin) and constant current control. This is shown in Fig.5.4b, where AB represents the amin and BC the
constant current characteristics; these are the static characteristics for the converter.
The DC power transferred from the converter to the DC voltage source at any operating point is simply the
product of the DC voltage and the DC current. By reducing the DC voltage source to zero, an intercept on
the DC current axis (C) is reached and the DC power transfer is zero. The delay angle at this point will be
approximately 90˚ (electrical).

CHAPTER
Vdc Vdc
αmin αmin
A A
B B
Idc +Pdc Idc

(RECTIFIER)Ê
C C
Idc Idc
Iorder Iorder
Fig. 5.4b– Basic rectifier static characteristics
-Pdc
(INVERTER)Ê
Ê5.4b
If the DC voltage level is reduced further, making it negative,
the converter can continue to maintain the DC current at the
ordered value by further delaying the thyristor gating signal,
Y X
that is, the firing delay angle a is increased beyond 90˚. The γmin
DC power flow is now reversed, meaning that the DC source
is now supplying power to the converter and the converter is -ÊVdc
now an inverter.
Fig. 5.4c– Basic rectifier/inverter static characteristics
The DC voltage can be reduced further, becoming more
negative, until a point is reached at which practical
thyristors have inadequate time with negative voltage to properly turn off and they immediately re-conduct
when subjected to positive voltage. This is referred to as commutation failure. The duration of negative
voltage at the switching-off of the thyristors is called the extinction angle and is referred to as Gamma (g ).
Ê5.4c
To reduce the risk of commutation failure, another control characteristic can be added to ensure that the
switching-off time for the thyristors is a minimum (gmin) consistent with safe commutation of the thyristor
valves. Adding gmin line (XY) to the Vdc/Idc graph gives the converter static characteristics shown in Fig. 5.4c.
5.4.1. Basic Requirements of the Converter Control System

The converter control system has to meet several basic requirements. Firstly, it must allow the stable
controlled transfer of power under steady-state operation and secondly, under fault conditions, it must
control the response of the converters in a manner that is conducive to the recovery and operation of the
two interconnected AC systems.
In its simplest form a HVDC scheme consists of two converters where each one can, at a given point in time,
control one parameter of the scheme. As the most important parameter at the inverting converter is its
extinction angle (gamma) historically this has been chosen as the main operating mode for the inverter, that
is, the inverter valve firing instants will be controlled in such a manner to maintain the inverter extinction
angle at an ordered value. A secondary mode of operation has been the control of the DC current in the
event that the sending end converter (rectifier) is unable to deliver the required DC current.
As the primary objective of a HVDC scheme is power transmission, the rectifier is normally operated in
DC current control, that is its firing angle (a) is changed in response to the operating conditions in such
a manner as to keep the DC current at an ordered value. However to maintain the direct current at the
desired value, the rectifier must have sufficient valve winding voltage to overcome the DC voltage, that is
the back EMF presented by the inverter and the commutation voltage drop(s), created by the operation of
the rectifier. If the rectifier’s AC terminal voltage is insufficient, the rectifier will operate with a minimum
firing angle determined by the characteristics of the rectifier thyristors. This is usually referred to as the
minimum firing angle characteristic or simply alpha min (amin) control.

CHAPTER
Vdc Vdc
CONVERTER CONVERTER
A A
A B A B
X OP X
Y Y
∆I ∆I
W C C W Idc
Idc
Iorder Iorder
OP
CONVERTER CONVERTER
B B
-Vdc -Vdc
Fig. 5.4d– Basic two terminal static characteristics – Power Fig. 5.4e– Basic two terminal static characteristics – Power
flow A to B flow B to A
Ê5.4d
By placing both the inverter and the rectifier characteristics for both converters on the same Vdc/Idc graph,
Ê5.4e
the steady-state or static operating characteristics for the DC system are obtained. This is illustrated in
Fig. 5.4d for power flow from converter A to converter B.
The intercept of the two characteristics is the operating point (OP) for the scheme and determines the DC
current and DC voltage and thus the transferred DC power. Changing the current order changes the DC
current, and thus the DC power. However, as a consequence of the slope of the inverter g characteristic,
the DC voltage will also change with changes to the DC current.
If the current orders of the two converters are swapped over, that is converter B current order is greater
than that of converter A, the intercept point remains on the constant DC current line, but the DC voltage
polarity, and thus the power flow direction, reverses. This means that the power flow is now from converter
B to converter A as shown in Fig. 5.4e. [12].
In the diagrams Fig 5.4b to 5.4e, the DC voltage is not directly controlled: it is determined by the intercept
on the inverter extinction angle characteristic, which is in turn determined by the magnitude of the inverter
converter AC terminal voltage.
5.4.1.1. Firing Angle Characteristics

An equation (see Appendix A2.2.2 for more detail) showing the relationship between a and the various
AC and DC parameters is:
3⋅ 2  Xc V 
Vd = ⋅ VLL ⋅  cos(α ) − p.u . ⋅ LL − 0 Id p.u .  [Eqn 5.4a]
π  2 VLL 
It can be seen that a constant amin characteristic is dependent upon parameters other than just Vdc and Idc.
Fig. 5.4f illustrates amin lines for different values of amin drawn on the static characteristics at constant and
variable AC voltages. The constant current characteristic (BC) is the normal mode of control for a HVDC
rectifier and is set to a desired value (Iorder) which in turn is usually derived from a ‘slower’ outer power
control loop.

CHAPTER
Vdc Vdc
A Increasing A Decreasing
αmin Ell
B B
αminÊ=Ê2¡,ÊE ll =Ê1.0Êp.u. αminÊ=Ê2¡,ÊE ll =Ê1.00Êp.u.

αminÊ=Ê4¡,ÊE ll =Ê1.0Êp.u. αminÊ=Ê2¡,ÊE ll =Ê0.95Êp.u.
αminÊ=Ê6¡,ÊE =Ê1.0Êp.u. αminÊ=Ê2¡,ÊE ll =Ê0.90Êp.u.
αminÊ=Ê8¡,ÊE ll =Ê1.0Êp.u. αminÊ=Ê2¡,ÊE =Ê0.85Êp.u.
ll ll
C C
Idc Idc
Iorder Iorder
A B
Fig. 5.4f– A) Variable amin characteristics (constant AC voltage) and B) Constant amin characteristics (variable AC voltage)
Vdc Ê5.4f Vdc
Increasing γ Decreasing
E llÊ
γÊ=Ê12¡,ÊE llÊÊ=Ê1.0Êp.u. γÊ=Ê12¡,ÊE llÊÊ=Ê1.00Êp.u.

Idc Idc
A B
Fig. 5.4g– A) Variable g characteristics (constant AC voltage) and B) Constant g characteristics (variable AC voltage)
5.4.1.2. Extinction
5.4g Angle (g ) Characteristics
Similar to the rectifier constant amin characteristic, the inverter g characteristic is dependent upon both DC
and AC parameters. The relationship is defined by:
3⋅ 2  Xc V 
Vd = ⋅ VLL ⋅  cos(γ ) − p.u . ⋅ LL − 0 Id p .u .  [Eqn 5.4b]
π  2 VLL 
Fig. 5.4g (A) illustrates g lines for different values of g drawn on the static characteristics and Fig. 5.4g (B)
illustrates a constant g characteristic with different values of AC terminal voltages.
5.4.1.3. DC Power Characteristics

In the previous sections, the basic control modes of the rectifier and inverter were described; however,
the most important requirement is the control of DC power. DC power is the product of DC voltage and DC
current given by the following equation:
Pdc = Vd ⋅ Id [Eqn 5.4c]

CHAPTER
By plotting constant power curves and over-laying the static characteristics as shown in Fig. 5.4h, it can
be seen that moving the constant current line of the two converters changes the DC power flow. However
it can also be seen that due to the inherent slope of the inverter constant extinction line the DC voltage
varies with DC current. Consequently, a given increase in power order will have a disproportionate change
in DC current order.
Vdc
OP1
OP2
OP3 CONVERTER
B
PdcÊ=Ê3P
PdcÊ=Ê2P
PdcÊ=Ê1P
CONVERTER ConstantÊÊ
A DCÊpowerÊlinesÊ
Idc
Iorder Ê1 Iorder Ê2 Iorder Ê3
Fig. 5.4h– Constant DC power lines
For example, moving the operating point from OP2 to OP3 results in a decreased DC voltage and the DC
current has to be increased further to achieve the demanded power. For power reductions the reverse is
true, moving from OP2 to OP1 results in a rise in DC voltage and again a correspondingly smaller change
in DC current to achieve the new power order.
Ê5.4h
5.4.1.4. Reactive Power Characteristics
A HVDC converter not only exchanges real power with the AC system but also absorbs reactive power (Q)
from the system. That is, the converter appears to be an inductive element as seen from the perspective of
the AC system. The amount of reactive power absorbed depends upon the operating state of the converter.
The relationship (p.u. values) between the reactive power absorbed and the AC and DC parameters are
given by the following equations:
Q = Id p.u . ⋅ k 2 ⋅ VLL2 p .u . − Vd p2.u . [Eqn 5.4d]

CHAPTER
Where:
Vdop p .u .
k=
 Xc  [Eqn 5.4e]
VLL − op p .u . . cos(α op ) − p.u . ⋅ Idop p .u . 
 2 
Note that values are in per unit and the suffix ‘op’ indicates the values at the nominal operating point e.g.
OP1 in Fig. 5.4i.
By substituting values for either the DC current or voltage for a given scheme, the constant reactive power
lines can be generated and placed on the static characteristics as shown in Fig. 5.4i.
In all practical AC systems, the AC source impedance at any point in the network is inductive at fundamental
frequency. Assuming that the AC system is a simple inductor leads to the observation that increasing the
Vdc
CONVERTER
A
ConstantÊreactiveÊ
powerÊlinesÊ
OP1 0.1Q
0.3Q
0.5Q
0.7Q
OP2
CONVERTER
B
PdcÊ=Ê3P
PdcÊ=Ê2P
ConstantÊÊ
DCÊpowerÊlinesÊ
I Idc
Iorder Ê1 Iorder Ê2
Fig. 5.4i– Constant reactive power lines
reactive power demand on the AC system (reactive current flows from the AC system to the converter) will
cause the converter AC terminal voltage to decrease. Conversely decreasing the reactive power demand
on the AC system will cause the converter AC terminal voltage to increase.
Therefore the constant reactive power lines in Fig. 5.4i can, to a first approximation, also be viewed as
Ê5.4i lines where increasing values of Q relate to decreasing values of converter terminal
constant AC voltage
voltage as illustrated in Fig. 5.4j.

CHAPTER
Increasing Decreasing
reactive ACÊvoltage
powerÊ(Q) (Vac)
Vdc Vdc
CONVERTER CONVERTER
A A
0.1Q 1.2Vac
0.3Q 1.1Vac
OP1 0.5Q OP1 1.0Vac
0.7QÊ 0.9Vac
CONVERTER CONVERTER
B B
I Idc I Idc
IorderÊ1 IorderÊ1
A B
Fig. 5.4j– A) Constant reactive power lines and B) Constant AC voltage lines
5.5. EFFECTS OF AC SYSTEM VOLTAGE

Ê5.4j
Both the rectifier minimum firing angle and the inverter extinction angle are determined by the magnitude
of the AC terminal voltage of the respective converters and hence their position on the Vdc/Idc characteristic
varies with changing operating conditions.
The preceding sections assumed that the AC terminals of the converter were connected to ‘infinite
busbars’: there is no impedance between the converter terminals and the AC voltage sources of the AC
system, and consequently no interaction between the converters and the magnitude or phase of their
respective AC terminal voltages. In practice this is not the case: a typical AC system is complex and variable.
For the purposes of analyzing the basic operation of the HVDC converters, at fundamental frequency the
AC system is simplified into a single perfect AC voltage source and an inductance representing the short-
circuit level at the converter busbar.
A consequence of the operation of the converter in either rectification or inversion is the flow of a reactive
current from the AC system into the converter. The flow of the reactive current through the AC system
impedance gives a voltage drop between the AC system voltage source and the converter busbar. However
changing the converter terminal voltage directly interacts with the operating condition of the converter
and changes the converter’s reactive current component.
In practice the AC system also has a resistive element and the flow of the real power component of current
also affects the converter terminal voltage, however this generally has a much smaller effect and is ignored
in the following sections.
5.5.1. Effects of Sending End AC Voltage on Rectifier Operation

In Fig. 5.5a below, the line AB represents the rectifier converter constant minimum firing line a min for
operation with rated AC voltages. Small reductions in the rectifier AC system voltage will cause the amin line
to move downwards on the Vdc/Idc graph but will have no effect on the power transmission, as the operating

CHAPTER
Vdc
A
CONVERTER
A
A1
A2 CONVERTERÊBÊ
B powerÊlinesÊ
OP1ʼ 0.3QÊ
0.5Q
B1 0.7Q
A3 OP1 0.8Q
OP2 B2
CONVERTER
B
A4 OP3
B3
OP4
B4
C
Idc
Iorder
Fig. 5.5a– Reduced sending end AC busbar voltage
point OP1 is determined by the rectifier constant current characteristic BC. However if the AC voltage
reduces further to point A1B1, the rectifier will reach a point where it cannot develop sufficient voltage to
overcome the voltage presented by the inverter and consequently cannot maintain the DC current at the
ordered value. At this point the rectifier is in both constant current and amin mode of control and a further
Ê5.5a in the rectifier AC voltage could cause the operating point to move to OP1’.
very small reduction
Conversely, a very small increase in the rectifier AC terminal voltage will cause the operating point to move
back to OP1. Similarly small changes in the inverter AC busbar voltage will also cause the operating point to
move between OP1 and OP1’. However moving the operating point to OP1’ will reduce the reactive power
taken from the sending end AC system and cause an increase in the rectifier AC busbar voltage (see section
5.4.1.4).
Increasing the rectifier AC terminal voltage will move the amin line upwards and depending on the relative
slopes of the amin and g characteristics and the strength of the sending end AC system, the operating point
may move back to OP1. The process thus repeats and the operating point oscillates between the two
locations. This is a highly undesirable phenomenon giving rise to power oscillations in the both AC and
the DC systems. In practice this phenomenon is avoided by careful design of the inverter characteristics.
Further reductions in the rectifier AC system busbar voltage will cause the amin characteristic to move down
the Vdc axis and the operating point will move down the inverter constant current characteristic to OP2
with lower DC voltage and hence lower DC power transmission. Progressively greater reductions in the AC
busbar voltage will cause the operating point to move further down to OP3 and OP4 etc.

CHAPTER
Vdc
CONVERTER
A
CONVERTERÊAÊ
powerÊlinesÊ
0.3Q
OP1 0.5Q
X1
0.7Q
OP2 0.8Q
X2
Y1
X3 OP3 Y2
CONVERTER
B Y3
Idc
Iorder
Fig. 5.5b– Reduced receiving end AC busbar voltage
An important consideration other than simply the loss of DC power transmission is the increased reactive
Ê5.5b
power consumption of the inverter converter. Fig. 5.5a shows the rectifier amin characteristic sweeping
through increasing reactive power lines of the inverter. If the inverter is connected to a relatively strong
AC system, that is very low impedance between the inverter and its AC voltage source, then the increase
in reactive power consumption will be of little consequence. However if the inverter AC system connection
point is weak, the increased reactive power consumption of the inverter will result in the inverter busbar
voltage being reduced. In extreme cases, it can be seen that faults in the sending end AC system result
in significant and unacceptable reductions in the inverter AC busbar voltage. This means that the fault in
the sending end AC system is exported to the receiving AC system and will adversely affect consumers
connected to that system. This relationship between converter operation and busbar voltage gives rise to
adverse interactions and possible instabilities.
A further consideration is the reactive power exchange between the rectifier and its own AC system under
reduced voltage operation. At first glance the reactive power equation (see section 5.4.1.4) would suggest
that once the rectifier is operating on its amin characteristic, the reduction in AC system voltage would result
in reduced reactive power consumption by the rectifier. If this is the case, the reduction in the reactive
power supplied by the rectifier AC filters/capacitors is much greater, as this is related to the square of the
applied AC voltage. Consequently, the net reactive power absorbed from the AC system increases as the
AC system voltage falls. In any practical AC system the source impedance of the system is inductive at
fundamental frequency and thus any increase in reactive power consumption by the rectifier will cause the
AC terminal voltage to reduce further than caused by the initial fault. This is a positive feedback system
and, particularly with weak AC systems, can lead to instability and eventual collapse of the sending end
AC system voltage.

CHAPTER
5.5.2. Effects of Receiving End AC Voltage on Inverter Operation
In Fig. 5.5b above, the line X1Y1 represents the inverter converter constant extinction angle line (g )
for operation with rated AC voltage. The operating point (OP1) is determined by the intercept of this
characteristic on the rectifier constant current
characteristic. Reducing the inverter AC system Vdc
voltage will cause the line to move downwards on
the Vdc/Idc graph and the operating point will move
X
down the rectifier constant current characteristic to
OP2 and then OP3, etc.
A OPa B
A consequence of moving the operating point down OPb
the constant current characteristic is the increasing OPc
reactive power consumption at the rectifier and, if the
sending end AC system is particularly weak, this will Y
depress the rectifier AC terminal voltage.
CONVERTER CONVERTER
As in the previous section, the reduction in the inverter B A
AC terminal voltage will reduce the inverter converter
reactive power consumption, but the effective
reactive power support given by the inverter AC filters Idc
and capacitors will be reduced more significantly by
Iorder
the square of the AC voltage. Therefore, on balance Fig. 5.5c – Multiple operating points
the reactive power demand from the AC system will
be progressively increased with AC system voltage being reduced further as a consequence. This is again
Ê5.6a AC systems can lead to the collapse of
a positive feedback condition, which in particularly weak receiving
the AC busbar voltage and the loss of all power transmission.
5.5.3. Multiple Operating Points

In the preceding sections it was seen that the actual scheme operating point was determined by the
intersection of the two converters’ static characteristics. It was also shown that the position of the firing
angle (a) and extinction angle (g ) characteristics are determined by the magnitude of the converter AC
terminal voltages.
However, as:
 Xc 
Vd p.u . = k ⋅ VLL p .u . ⋅  cos(α ) − p.u . ⋅ Id p.u .  [Eqn 5.5a]
 2 
and
 Xc 
Vd p.u . = k ⋅ VLL p .u . ⋅  cos(γ ) − p.u . ⋅ Id p.u .  [Eqn 5.5b]
 2 
show, the slopes of the firing angle (a) and extinction angle (g ) characteristics at the two converters are
not necessarily the same. This can give rise to multiple operating points as shown in Fig. 5.5c.
This is a highly undesirable state as it is uncertain which of the operating points the scheme will operate at
and, in practice, small changes in the AC and DC conditions may cause the operating point to move between
any two or all three of the possible points. In practice, the static characteristics of the two converters are
designed to avoid multiple operating points as discussed in the following sections.
5.6. MULTI-TERMINAL HVDC SCHEMES

Section 5.4 dealt with a two terminal DC link. However, the principles can be extended to DC links with three
or more converters: these are referred to as multi-terminal HVDC schemes. At present there are only two
schemes designed and run as true multi-terminal HVDC schemes. These are Sardinia-Corsica-Italy (Corsica
tap – commissioned 1992) [11] and Hydro-Quebec New England Multi-terminal (Canada/USA – commissioned
1986-92).

CHAPTER
Vdc
CONVERTER
A A
OPB OPC CONVERTER

C
CONVERTER
B
Idc
Iorder ÊB Idc ÊBÊ Iorder ÊC Iorder ÊA
= Iorder ÊAÊ- Iorder ÊC
Fig 5.6a – Multi-terminal HVDC static characteristics
Vdc
CONVERTER
A Ê5.6b A
CONVERTER
B
OPB OPC
CONVERTER
C
Idc
Iorder ÊB Iorder ÊC Idc ÊBÊ Iorder ÊA
= Iorder ÊAÊ- Iorder ÊC
Fig. 5.6b– Multi-terminal HVDC static characteristics – reduced AC voltage at converter C

CHAPTER
Although the control of a multi-terminal scheme is more complex than for a two terminal scheme, the
principle is the same. Fig. 5.6a shows the simplified static characteristic for a basic three terminal HVDC
scheme consisting of a single rectifier (converter A) and two inverters (converters B and C).
Note that:
➙ The operating DC voltage is determined by converter B which is operating in constant extinction
angle (g ) control.
➙ Converter C is in constant DC current control, therefore the current available for converter B is the
difference between the current delivered by the rectifier (converter A) and that taken by converter C.
➙ At all times the sum of converter B and converter C current orders must be less than converter A
current order, or power transmission will be reversed.
In a practical system, current order C could be a control parameter derived from the actual DC current
flowing: this is referred to as floating current order. Note that increasing converter B current order does
not increase the DC current drawn by that converter and has only a secondary effect on the operating DC
voltage, therefore the DC power drawn by converter B is indirectly controlled.
In the above example converter C is operating in constant current control and thus it is operating with a
higher extinction angle than would normally be the case. This implies that the installed capacity and thus
the cost of the converter is increased.
If the AC terminal voltage at converter C is reduced significantly, current control moves to converter B with
converter C in g control and hence converter C determines the DC voltage of the scheme as shown in Fig. 5.6b.
A drawback to the characteristics as shown is that rapid changes in DC power drawn by one inverter will
directly affect the other. For example, the loss (blocking) of inverter C in Fig. 5.6b would result in converter B
being forced to operate in constant extinction angle control, taking all of the current supplied by the rectifier
unless a rapid and reliable telecommunications system is available to coordinate the power and current
orders between the various converters. The inference is therefore that both inverters need to be rated for
the full power flow of the DC system. Furthermore, this dramatic change in power flow into the inverter B AC
system would not only be undesirable for the AC system at converter B, but perhaps impossible to sustain
and a collapse of the DC transmission system would ensue.
Vdc
A B
OPB CONVERTER OPC CONVERTER
B C
CONVERTER
A
Idc
Idc ÊB Iorder ÊC Iorder ÊA
Fig. 5.6c– Alternative multi-terminal static characteristics

CHAPTER Ê5.6d
Variations to the basic static characteristics for multi-terminal schemes have been proposed to address
these issues, such as operating both inverters in constant current control and the rectifier in constant firing
angle control or (as shown in Fig. 5.6c) in constant DC voltage control.
Such methods give the inverters greater independence of operation but lead to higher overall ratings and
costs for all of the converters.
The simple idealized characteristics above describe the basics of multi-terminal operation, however such
simple characteristics do not address important practical issues relating to system stability, multiple
operating points, AC or DC system faults, or rapid power order changes at any of the converters.
5.7. TYPICAL TRANSMISSION SCHEME STATIC CHARACTERISTICS

An example of simplified static characteristics for a transmission line HVDC scheme is shown in Fig. 5.7a.
The characteristics are designed to maintain some power flow throughout all but the most severe of faults.
All of the values quoted below are typical and are very much scheme-dependent.
Vdc
powerÊlinesÊ
0.3Q
1.2
A B C 0.5Q
T
Y U
1.0 0.7Q
OP
X Z
0.8
∆I
W
0.6
0.4
V
E
D
0.2
F Idc
0.2 0.4 0.6 0.8 1.0 1.2
Fig. 5.7a– Typical HVDC transmission scheme static characteristics
5.7.1. Rectifier Characteristics

5.7a
The rectifier characteristics are bounded by the lines ABCDEF in Fig. 5.7b.
➙ AB is a maximum constant DC voltage characteristic which is used to limit the direct voltage should it
rise above a preset value. It is commonly used on transmission schemes for open circuit line testing,
but it is also used to prevent DC overvoltages in the event of an inverter control failure. It is typically
set to 1.05 p.u.
➙ BC is a minimum firing angle (amin) control characteristic, which is used to prevent the control
system issuing a firing pulse before a valve has sufficient forward voltage. The setting of this control
loop is higher than would be expected to fire the individual thyristors and is used to ‘mask’ the effect
of valves which have levels operating on Voltage Breakover (VBO) protective firing (see section 6.2).
The setting of this loop is scheme dependent, but is typically 5 degrees.

CHAPTER
Vdc
powerÊlinesÊ
0.3Q
1.2
A B C 0.5Q
T Y U
1.0 0.7Q
OP
X Z
0.8
∆I
W Dʼ
0.6
0.4
V
E
D
0.2
F Idc
0.2 0.4 0.6 0.8 1.0 1.2
Fig. 5.7b– Rectifier characteristics
➙ CD is a constant Idc characteristic. This is usually the main control mode for the rectifier. The control
5.7b
loop maintains the DC current at a value (Iorder), although this value is itself often derived from an
outer control loop, for example DC power control.
➙ DEF is the rectifier’s Low Voltage Current Clamp Characteristic (LVCC).This characteristic is only
invoked if the DC voltage falls below a preset level (typically 0.3 p.u.) as a result of receiving AC
system faults or DC line to ground faults. Historically the intercept with the DC current axis (F) was
chosen to be 0.3 p.u., so that the valves of a non-commutating inverter are typically subjected to
the same average current as normal operation and thus do not experience excessive heating. On
more modern control characteristics the current axis intercept is often a variable determined by the
DC current order, so as to minimize the change in reactive power consumption of both converters
between the fault and non-fault conditions.
There are many variations of these characteristics in use. A common variation is the use of a
Low Voltage Current Order Limit (LVCOL) rather than the current clamp characteristic (LVCC):
this characteristic (D’E) is intended to reduce the reactive power consumption at both converters,
in particular the inverter, during faults in the receiving AC system.
During such faults, the operating point moves down the rectifier constant current characteristic and
significantly increases the reactive power consumption of both converters. This in turn leads to further
voltage reductions and exacerbates the original problem, ultimately leading to an inverter commutation
failure. By reducing the reactive power consumption, the voltage reduction will be opposed and the inverter
commutation process will be assisted. In effect the characteristic holds the inverter busbar voltage in
negative feedback and is theoretically capable of stable steady-state operation.
A counter argument is that this characteristic reduces the difference between the effective current orders
of the two converters during recovery of the DC voltage following a fault. It is the difference between the
two converters’ DC currents which drives the recovery of the scheme. This current difference provides the
charging current for the DC system capacitance and the error signal which drives the converter control
system in returning the firing angles to their normal operating points. Note that the capacitance can be
considerable in a cable scheme. This latter consideration is present at all times and in the case of back-to-
back systems, is the major factor in determining the recovery time of the system following faults.

CHAPTER
5.7.2. Inverter Characteristics
The inverter characteristic is bounded by the lines VWXYZ in Fig. 5.7c:
➙ YZ is a constant extinction angle (gamma) characteristic. This limits the inverter extinction angle to
a pre-defined minimum (typically 15 degrees). This characteristic can be used as the main inverter
control mode in which case the scheme’s operating DC voltage is maintained within limits by the
transformer tapchanger.
➙ XY is the inverter current error characteristic. It is actually part of the inverter constant extinction
angle (gamma) control where the ordered value of gamma is increased with the increasing
discrepancy between the ordered current and the actual DC current. This characteristic is
encountered in the event of either small errors between the rectifier and inverter DC current
orders, or small reductions in the sending end AC system voltage, causing the rectifier to operate
on its amin characteristic. In the case of the latter, this characteristic avoids the interaction between
the rectifier amin and the inverter g characteristics, which can give rise to multiple or ambiguous
operating points and possibly instability (see section 5.5.1). In addition, by increasing the effective
gamma order the converter decreases its DC terminal voltage and ensures that the rectifier can
continue to deliver some DC current. The slope of the characteristic is typically chosen to be 70
degrees/ 1.0 p.u error, that is, the gamma order is increased by 70 degrees for every 1.0 p.u. of rated
current below the ordered value. This slope gives both converters an approximation to a constant
reactive power characteristic which limits the effect of faults in one AC system on the other. In many
examples of static characteristics this line is often ‘linearized’ (approximated to a straight line).
➙ WX is the inverter constant current limit. This is set to a value (typically 0.1 p.u.) below the rectifier
current order: this value is called the current margin. This characteristic is only encountered in the
event of significant reductions in the sending end AC system voltage and is intended to maintain
power flow. However, it should be noted that operating on this characteristic increases the reactive
power consumption by the inverter and reduces the receiving AC system busbar voltage. The value
of the current margin was historically chosen to be a compromise between minimizing the loss of
Vdc
powerÊlinesÊ
1.2 0.3Q
A B C 0.5Q
T Y U
1.0 0.7Q
OP
X Z
0.8
∆I
W
0.6
0.4
V
E
D
0.2
F Idc
0.2 0.4 0.6 0.8 1.0 1.2
Fig. 5.7c– Inverter characteristics

CHAPTER 5.7c
current for small voltage reductions in the sending end voltage and speed of recovery following
faults in the receiving AC system. The margin also had to have a minimum value to overcome both
measurement errors between the current transducers at both ends and the requirement to be able
to operate in floating current order control at the inverter with the loss of communications between
the two terminals of the scheme.
➙ VW is the voltage dependent current limit. It is essentially a constant DC current loop with an order
derived from Iorder and the measured DC voltage. The characteristic is only encountered during severe
faults in the sending end AC system and forms a limit on the inverter constant current characteristic,
thereby limiting the change in inverter reactive power consumption. Typically, the characteristic
intercepts the DC voltage axis (typical value 0.4 p.u.). The intercept on the voltage axis ensures that
in the case of DC line faults the inverter DC current is reduced to zero and does not contribute to
the fault current and thus allows fault removal (de-ionization).
On some schemes such as the 2000 MW Cross Channel Link [12], an additional low current
characteristic was added. This additional low current characteristic intercepts VW at approximately
0.1 p.u. DC current, the intention being to ensure cable discharge in the event of the unilateral
blocking of the rectifier converter. Characteristics used on other schemes have included a constant
minimum current characteristic at the inverter (typically set to 0.1 p.u.). This helps to mitigate
the increase in the inverter AC terminal voltage when the rectifier is not operating. However, on
overhead line transmission schemes, this characteristic has to be coordinated with the DC line fault
detection and protection strategy.
➙ TU is a constant DC voltage characteristic. If the main control mode for the inverter is constant
extinction angle control, then this characteristic is not normally encountered under steady-
state conditions. The control characteristic is intended to limit the DC terminal voltage of the
inverter below a predetermined value (Vorder) during transient conditions. On some schemes this
characteristic is the main operating mode and the g characteristic becomes a back stop for fault
conditions (see Fig. 5.7d). In such circumstances the inverter converter will operate with a higher
value of extinction angle (g ) and greater reactive power consumed from the AC system. This has
Vdc
powerÊlinesÊ
0.3Q
1.2
A B
C 0.5Q
T Y
1.0 U 0.7Q
OP
X Z
0.8
∆I
W
0.6
0.4
V
E
D
0.2
F Idc
0.2 0.4 0.6 0.8 1.0 1.2
Fig. 5.7d– DC voltage control

CHAPTER
5.7d
to be taken into account in the design of the converter and invariably involves higher equipment
and operating costs. However, despite the higher costs, constant DC voltage control does have
advantages; namely greater system stability and improved resilience to commutation failure
for small disturbances to the inverter AC busbar voltages. These advantages are of particular
importance when the inverter is connected to a weak AC system where the interactions between
the inverter and its local AC network are significant. When operating in this mode the transformer
tapchanger controls the inverter extinction angle to within a dead band of a few degrees of a
nominal value. The operating DC voltage order may be fixed or may be a variable determined by a
slower outer control loop. On the refurbished Konti-Skan I project the DC voltage order is set by
a slow outer control loop to obtain a fine trim on the inverter extinction angle, thus achieving the
benefits of greater stability without the costs of operating at large extinction angles.
5.8. BACK-TO-BACK HVDC SCHEMES – STATIC CHARACTERISTICS

The static characteristics for a back-to-back HVDC link are shown in Fig. 5.8a. Back-to-back links
are commonly used in conjunction with very weak AC systems and therefore their design reflects
the need to minimize the impact of faults and disturbances of either AC system on the other. In
addition, advantage can be taken of the fact that the two converters are on the same site and there
are no communication delays or risk of loss of communications between the two control systems.
The principal example of this is the use of the converters to operate at variable DC voltages and currents
to control the reactive power exchange between the converters and their respective AC systems. This
allows reactive power support and/or AC voltage control for either of the AC systems under steady-state
and transient conditions [5].
Vdc
1.4
G ConstantÊreactiveÊ
powerÊlinesÊ
0.3Q
1.2 B
A
C 0.5Q
1.0 Vorder X Y 0.7Q

OP
0.8 Z
H
0.6
0.4
V W
0.2
E
F D
Idc
0.2 0.4 0.6 0.8 1.0 1.2
Fig. 5.8a– Typical HVDC back-to-back scheme static characteristics

5.8a

CHAPTER
Many of these characteristics are the same as those described in section 5.7. The principal differences are:
➙ The inverter is usually operated in constant DC voltage control (XY), with the DC voltage order derived
from an outer control loop determined from either the AC voltage or reactive power control of one of
the AC systems. However, only one of the AC systems’ parameters may be controlled at any instant
(usually the weakest AC system) even though both converter AC terminals are affected. For example,
moving the operating point to a lower value of DC voltage and compensating with an increase in DC
current will give the same real power transfer, but increase the reactive power consumption of both
the rectifier and inverter (see Fig. 5.8b). Increasing the reactive power will reduce both converter
busbar voltages. Differential control of the two converters’ reactive power and firing angles is achieved
by their respective transformer tapchangers, which are very slow in comparison to the converter
control systems.
Vdc
1.4
CONVERTERÊBÊ
ConstantÊreactive
powerÊlines
1.2
0.3Q
1.0 Vo1 OP 1 0.5Q
Vo2 0.7Q
OP 2
0.8Q
0.8 Vo3 OP 3
Vo4 OP 4
0.6
ConstantÊDCÊ
powerÊlineÊ
0.4
0.2
E
F l o1 l o2 l o3 l o4
Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8b– Reactive power

PdcÊ=ÊÊcontrol
Vo1ÊxÊlo1Ê=ÊÊwith constant
Vo2ÊxÊl o2Ê DC
=ÊÊVo3ÊxÊl o3Ê=ÊÊVo4Êpower
xÊl o4
➙ Inverter Low Voltage characteristic (LVCL): if one or both of the AC systems are weak, the drawing
of excessive reactive power during faults in either system can result in the collapse of the healthy
or unfaulted system. By giving the inverter constant current characteristic a slope, the variations
in the inverter reactive power consumption can be minimized (see Fig. 5.8c).
In practice the slope is chosen for a particular scheme to give a small decrease in reactive power
consumption5.8bwith an acceptable rise (typically 10%) in the associated AC system voltages. A
constant low voltage limit (VW) is used to prevent inadvertent inverter operation in the rectification
region during major transients.
➙ Rectifier LVCC characteristic (DEF): this characteristic is intended to operate when the inverter
is suffering a commutation failure. With the converter operating range allowing reduced DC
voltage, the larger firing angles employed give rise to increased DC harmonic components. Without
significant signal filtering this could lead to false or intermittent triggering of the low voltage
characteristics. Such filtering would lead to control delays and affect the control system response
to transients. To avoid this, the characteristic is initiated by the inverter, which monitors its own
valve winding currents for early indications of commutation failure.

CHAPTER
Vdc
1.4 G
CONVERTERÊBÊ
1.2 powerÊlinesÊ
A
0.3Q
Vorder 0.5Q
1.0 X Y 0.7Q
OP 0.8Q
0.8 B Z
H
0.6 OPʼ
C
0.4
V W
0.2
E
F D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8c– Reduced sending end AC system voltage
Vdc
5.8c
1.4
CONVERTERÊBÊ
1.2 ConstantÊreactiveÊ
powerÊlinesÊ
A B C
1.0 0.3Q
0.5Q
Vorder X Y 0.7Q
OP 0.8Q
0.8
G Z
Yʼ
0.6 OPʼ
0.4
V W
Zʼ
0.2 H
E
F D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8d– Reduced receiving AC system voltage
5.8d
CHAPTER
➙ Rectifier Control of Inverter Extinction Angle (GH): the reduction in the receiving system AC terminal
voltage with the inverter in constant extinction angle control (XY) moves the Operating Point (OP)
down the rectifier constant current line (CD). Unfortunately this leads to an increase in reactive
power consumption by both converters. In particular, at the inverter the increased reactive power
consumption will cause further voltage reduction and can eventually precipitate commutation
failure of the converter. With weak AC systems, the converter will not be able to regain normal
steady-state commutation and repeated commutation failures will ensue with further disturbances
to both AC systems. However, with back-to-back HVDC schemes there is no difficulty in transferring
inverter control parameters to the rectifier. By allowing the rectifier to control the inverter extinction
angle to a predetermined minimum value, the operating point is forced down the inverter low
voltage characteristic (WX) and the reactive power consumption of the inverter remains either
unchanged or, in practice, decreases (see Fig. 5.8d).
Reducing the reactive power consumption of the inverter will oppose the initial busbar voltage
reduction and acts to stabilize the AC system. This control mode was first used on the Chandrapur
2 × 500 MW back-to-back HVDC scheme in India [13] and has been used on other schemes in India,
Uruguay and Saudi Arabia.
5.8.1. Back-to-Back HVDC Converter – AC Voltage Control

It has already been stated that with a back-to-back transmission scheme, the operating point can be changed
to achieve both controlled real power transfer and some limited control of AC reactive power consumption of
one of the converters. This is particularly useful where one or both of the interconnected AC systems are weak.
Due to the shape of the constant reactive power lines (see Fig. 5.4j), it is apparent that at, or near to rated
DC voltage, changing the operating DC voltage has a greater impact on the reactive power consumed by the
converters than does changing the DC current. As the inverter is usually directly or indirectly controlling the
DC voltage it follows that the use of the converters to control the AC system voltage(s) is best achieved by
the inverter. However during operation at reduced or zero DC voltage, the shape of the constant reactive
power characteristics shows that changing the DC voltage has very little effect. Therefore at low DC voltage,
reactive power control is best achieved by changing the DC current, and the rectifier converter would be the
Vdc
1.4
G CONVERTERÊAÊorÊBÊ
ConstantÊVacÊerrorÊlinesÊ
1.2 VacÊorderÊ=Ê1.2Êp.u.Ê
A B C VacÊsourceÊ=Ê1.1Êp.u.Ê VacÊconverterÊ=Ê1.0Êp.u.
1.0 Vorder X Y VacÊsourceÊ=Ê1.2Êp.u.Ê VacÊconverterÊ=Ê1.2Êp.u.
OP1 VacÊsourceÊ=Ê1.3Êp.u.Ê VacÊconverterÊ=Ê1.2Êp.u.
VacÊsourceÊ=Ê1.4Êp.u.Ê VacÊconverterÊ=Ê1.2Êp.u.
0.8 Z
OP2
H
0.6
OP3
0.4
V W
0.2
E
F D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8e– AC voltage control mode
BACK TO 5.8e
CHAPTER
optimum choice for such a control feature. In order to cover all operating conditions, it is necessary to involve
both converters in reactive power and/or AC voltage control.
With a back-to-back HVDC scheme, the ability to use measurements from both converters offers the
opportunity to extend the converter characteristics to the control of AC overvoltages in either AC system.
By comparing each AC system voltage (3-phase RMS) with a set ordered maximum (e.g. 1.2 p.u. but typically
a dynamic quantity), an AC voltage error signal can be derived and this can be drawn over the static
characteristics (Fig. 5.8e). This characteristic can be drawn for differing AC system voltages and hence
differing errors. In the following discussion, it is assumed that the rectifier is in DC current control and that
there is no outer DC power control loop.
In Fig. 5.8e, with the AC voltage at or below the ordered maximum value, the converters will operate normally
(OP1). However if the AC system voltage and thus the converter busbar voltage rises, the characteristic
moves to the right until the control loop error reaches zero and the operating voltage now equals the ordered
maximum while the converter starts to exert control (OP1). Continued increase in the source voltage will
move the error characteristic further to the right and the operating point will move down the rectifier constant
current characteristic (OP2). Further increases of the source voltage will require the operating point to move
to OP3 so that the converter reactive power consumption holds the AC terminal voltage to the desired value.
This process can continue, however due to the slope of the constant reactive power lines at lower DC voltage
there is a disproportionate change in DC voltage for a given change in the AC source voltage. At very low DC
voltages, the characteristic is practically vertical and the inverter has little control over the reactive power
and AC terminal voltages. Any further control must now be achieved by increasing the DC current, which is
controlled via the rectifier.
The use of the converters to dynamically limit AC overvoltages at either converter AC busbar should be seen
as a temporary measure to assist the AC systems and avoid tripping the scheme and/or AC filters. It is only
a control measure to provide time for the AC system(s) to re-establish ‘normal’ operation. When AC voltage
control is exercised, both converters are operating with excessive firing angles leading to increased losses
and harmonic generation. This is acceptable for short periods of up to a few seconds but not in steady-state
operation. If these extreme operating conditions persist, then the tripping of AC filters and capacitors will
be necessary and the operation of the scheme will be compromised. Eventually, tripping of the complete
scheme will be necessary.
When AC voltage control mode is invoked, both converters are operating with increased reactive power
consumption. While this will limit the AC voltage rise at one converter AC busbar, it will also result in a change
in AC voltage (potentially an undervoltage) at the other AC terminal. If the two AC systems are of considerably
different short-circuit levels (one is relatively strong and the other weak), the effects on the weaker AC system
from overvoltages in the stronger system will be disproportionately severe. In general it can be assumed that
overvoltages are more dangerous to equipment than undervoltages and therefore dynamic AC overvoltage
control can be tolerated only for short periods.
In this extreme condition, the inverter busbar voltage may be depressed to the point where the inverter valves
have insufficient voltage to commutate as the overlap angle approaches 60˚. To prevent the collapse of the
inverter operation the rectifier converter takes control of the inverter extinction angle. The practical effect of
this is for both the DC current and DC voltage to be reduced (see Fig. 5.8f).
The use of the converters to control either of the AC system voltages is possible with back-to-back HVDC
schemes where measurements from both terminals are readily accessible to both control systems. With
transmission schemes, the delays associated with communications between the two ends and/or the
requirements to operate with or without communications, means that the options for AC voltage control are
restricted. Nonetheless, the use of converters to control modest AC system voltage rises and/or mitigate the
switching or reactive elements at the AC terminals of the converter has been employed on several schemes
e.g. KEPCO, Cheju 1 HVDC link.
5.8.2. Back-to-Back HVDC Converter – TCR Mode

During an AC system fault, power flow across the DC link will be disrupted. In the extreme example of the
complete loss of all three phases of an AC system, the converter attached to that AC system will be unable
to operate and all power flow on the link will cease.

CHAPTER
Vdc
1.4 G1 CONVERTERÊAÊorÊBÊ
1.0 G2 X Y VacÊsourceÊ=Ê1.2Êp.u.Ê VacÊconverterÊ=Ê1.2Êp.u.
0.8 Z
G3 OP2
H1
OP2ʼ
0.6
OP3
0.4 OP3ʼ H2
V W
0.2
E H3
F D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8f– AC voltage control – weak inverter AC system
Vdc
5.8f
1.4
A B C
0.8 Z
H
0.6
0.4
V W
0.2
E
F OP2 OP3 OP4 D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8g– Rectifier TCR mode
5.8g
CHAPTER
Vdc
1.4
0.8 Z
H
0.6
0.4
V W
0.2
E
F OP2 OP3 OP4 D Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 5.8h– Inverter TCR mode
5.8hattached to the unfaulted AC system is a rectifier, it will now operate on its LVCC characteristic
If the converter
(see Fig. 5.8g) and the reactive power absorbed from the system should be sufficient to keep the AC terminal
voltages at a reasonable level.
If the AC system is exceptionally weak, has highly resistive (long) AC lines or a relatively small capacity
compared with the DC link rating, the AC terminal voltages may rise above acceptable levels. Under such
conditions the AC voltage control mode may be used to increase the DC current above the LVCC level (E)
and limit the AC terminal voltage to a controlled value: OP2, OP3, etc. Under these circumstances, the
converter is performing the same role as a Thyristor Controlled Reactor (TCR) [14]. It should be noted
however that the non-commutating thyristor valves at the inverter terminal are now subjected to greater
conduction losses than they would normally experience and care must be taken not to exceed their thermal
limits. If the DC current flow is excessive, or the duration of the fault is too great, then the removal of filters/
capacitors may be required and a return to the pre-fault power level may not be possible when the inverter
AC system is restored.
If the converter attached to the unfaulted AC system is an inverter, the normal inverter static characteristics
will bring the operating point onto the DC voltage axis at zero DC current. The inverter reactive power
consumption will be zero and the AC terminal voltage will increase. If the voltage rise is excessive, the inverter
can make use of an AC voltage control mode and, by removing the minimum DC voltage characteristic (VW),
reduce its firing angle (a < 90˚) and circulate a DC current through the faulted rectifier (see Fig. 5.8h).
In effect, the inverter is now operating just inside the firing angle region normally associated with rectifier
operation. This condition is no different to the one described above for the rectifier converter and the same
considerations regarding valve ratings apply.
5.9. OTHER CONTROL CHARACTERISTICS

There have been many variations on the standard control characteristics adopted by various manufacturers,
however the basic trend has been that under normal steady-state operation the rectifying converter is in
control of the direct current and the inverter is in control of the direct voltage or extinction angle. For most
schemes this is perfectly adequate, but in the case of HVDC links feeding an island, or isolated AC system,
the normal control modes have shortcomings which need addressing.

CHAPTER
Island AC systems which have little or no local power generation and rely mainly on the DC link for their power
supply are harder to control, in both steady-state and dynamic conditions, than conventional systems. The
limited number of AC generators means that there is reduced mechanical inertia in the system and as such,
minor fluctuations in power can result in significant changes in the island AC system frequency. To counter
this and maintain steady-state and dynamic system stability the HVDC link has to be rapidly able to change,
or modulate, the DC power transfer. Furthermore it is generally accepted that this should be achieved without
reliance on the telecommunications system between the two ends of the scheme. In the case of importing
power to the island this is not possible with the conventional characteristics as the DC current is controlled
by the remote rectifier. The island inverter could control the power feed by changing the DC voltage, however
while this is possible for reductions in power, rapid increases would be difficult to realize without adopting
excessive operating firing angles, with their attendant losses and AC system harmonics.
5.9.1. Rectifier Voltage Control

On the KEPCO Cheju 1 HVDC 300 MW bipole scheme [15] a novel approach was taken: the characteristics of
the mainland rectifier and the island inverter were reversed. The island inverter was operated in DC current
control with the current order derived from the island AC system frequency and the mainland rectifier was
operated in DC voltage control with the DC current control function retained for limiting DC current during
inverter commutation failures. The characteristics are shown in Fig. 5.9a.
The inverter constant DC current characteristic is controlled by a high-speed island AC system frequency
closed loop control system. The current order can vary from zero to full load under normal operation with
a 20% overload capability for several seconds to allow re-acceleration of the island following a major
disturbance to the island AC system. However if one of the poles is lost, the remaining pole current
maximum is raised to 1.8 p.u. to give the island generators time to take up the load or for a load shedding
strategy to be implemented.
The rectifier amin characteristic is chosen to ensure that it is not crossed by the inverter minimum operating
extinction angle (g = 18°). Unusually, there is no constant extinction angle (g ) characteristic at the inverter.
The reason for this is that, if the island AC system suffers an undervoltage such that the extinction angle
is controlled to, for example g = 18°, the operating point would move directly out to the rectifier constant
Vdc
Y1
1.4
Y2
1.2 A
OP B Z
C
1.0
0.8
0.6 X2
W
0.4 X1
E
D
0.2
F Io IoMAX
Idc
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Fig. 5.9a– Cheju 1 HVDC scheme – mainland to island power flow characteristics

CHAPTER
DC current characteristic at 1.9 p.u. This would dramatically increase the power infeed to the island
with disastrous consequences for the island AC system frequency. By removing the extinction angle
characteristic the inverter is deliberately allowed to fail commutation!
The resulting collapse of DC voltage signals to the rectifier that the inverter is unable to commutate
correctly then the rectifier assists the recovery process by reducing its voltage order from 1.0 p.u. to 0.6 p.u.
and then ramping it back up over a period of several seconds. This time delay gives the inverter adequate
extinction angle to commutate, and gives the island AC system time to remove the fault or the generators
and/or synchronous compensators and automatic voltage controllers (AVR) to compensate for the system
changes which precipitated the undervoltage.
5.10. PROPOSED DEVELOPMENTS

Historically most HVDC schemes have been point-to-point power transmission between two relatively
strong AC systems. However, schemes that propose multiple HVDC link connections to the AC systems
are under consideration. Other schemes will also connect to stochastic sources of generation such as
wind farms.
The traditional control characteristics are not necessarily ideally suited to such schemes and modifications
have been proposed to improve the performance under steady-state and transient conditions.
5.10.1. Multiple HVDC Infeeds to AC Systems

In many parts of the world there are AC systems which have several HVDC interconnections terminating
electrically close to each other. Generally the DC links operate independently and may even be operated by
different companies. However during faults, and more importantly during recovery from faults, there exists
the possibility of adverse interactions between the various HVDC stations. When a HVDC link recovers from
a minor fault (e.g. a commutation failure), the operating point sweeps from zero DC voltage through the
static characteristics back to the normal operating point. The time to complete this is scheme-dependent
with the AC and DC systems and the control system parameters all influencing the event. If several schemes
are ‘electrically close’ and experience the same disturbance, their recovery times may be different.
Studies have shown that the recovery of one HVDC scheme may adversely affect the recovery of an
adjacent HVDC scheme. If the adjacent converter(s) are inverter(s), this interaction can lead to further
commutation failures and prolong the overall recovery of the system(s). This has led to proposals for either
coordinating the recovery of the various schemes via extensive telecommunications systems and/or the
installation of power electronic stabilizers (FACTS) in the AC systems.
It has also been shown in studies that the conventional current margin characteristic employed in many
schemes is a contributory cause of the interaction between adjacent links, particularly in moderate to weak
AC systems. During the period that the DC link is operating at reduced DC voltage on the inverter constant
current characteristic, both converters of the link will be consuming more reactive power than normal from
their respective systems. For the converters in general, and the inverter in particular, this is not a problem
as they will be operating with large firing angles and there will be little risk of the commutation process
failing. However if the adjacent HVDC link has already recovered it will now be subjected to a reducing
AC terminal voltage caused by the increased reactive power consumption of the recovering scheme. A
combination of this and the general distortion of the AC busbar voltages following the disturbance can
lead to the recovered inverter suffering further commutation failures and a repeat of the recovery process.
Studies have also shown that by giving the rectifier and inverter constant reactive power characteristics
similar to those used on back-to-back HVDC schemes, the depression of the AC system networks during
the recovery phase can be minimized and the risk of consequential commutation failures reduced. This can
have the added benefit in negating the need for the installation of expensive stabilizing equipment (FACTS).
5.10.2. Isolated AC Systems and Wind Farms

At very low DC current, the current will become discontinuous and this may result in excessive power
dissipation in the damping resistors of any thyristor levels that are operating repetitively on their VBO
protection (see section 6.2). To avoid this problem most HVDC schemes operate with a minimum DC

CHAPTER
current order, typically 10% of rated current, although on some schemes such as Konti-Skan 1 this has been
reduced to 3%. For power reversal, where the power is taken to zero and reversed, the general practice
is to reduce the current order to 10%, block the converters, reverse the power flow order and deblock,
again with the minimum current order of 10%. A consequence of this is that the DC voltage has now been
reversed, as the DC current must continue to flow in the same direction.
In general this is acceptable for most schemes built to date. However on future schemes, particularly those
associated with wind-farms, there is a requirement that the scheme can not only operate smoothly down
to zero but can also reverse the power or even hover about zero power for prolonged periods. In addition,
isolated (island) systems with large amounts of wind generation will be subjected to a fluctuating supply
due to the stochastic nature of the wind. Maintaining control of the isolated system is therefore problematic
and necessitates a fast and reliable control of the power export to the main (remote) grid system. By basing
the control characteristics on the constant reactive power lines (Q-curves), an alternative set of control
characteristics has been proposed [16] which meets these conflicting requirements (see Fig. 5.10a).
In this scheme, DC current is never discontinuous and power transfer is primarily controlled by modifying
the DC voltage. If the isolated (island) system is relatively small, the installed system inertia will be small
and to maintain system stability and integrity, fast control of the AC system frequency will be the primary
concern. Therefore, whether exporting or importing power, island AC system frequency is the main high-
level mode of control and is used to determine the DC voltage order.
+Vdc
X2l X1l Y1l
+VoÊMAX
Z1l
AR BR
OP1* OP1
CR
PowerÊflowÊisolatedÊsystemÊtoÊgridÊsystemÊ
A1 R B1 R
C1R
GridÊsystemÊ
QmaxÊlimitÊ
OP2* OP2
Idc
W2 l W1l
IsolatedÊsystemÊ
ImaxÊlimitÊ
PowerÊflowÊgridÊsystemÊtoÊisolatedÊsystemÊ
C1l
A1 l B1l
Cl
OP3* OP3
-VoÊMAX Al Bl
ZR
X2 R X1 R Y1 R
-Vdc
IsolatedÊsystemÊ GridÊsystemÊ
characteristicsÊ characteristicsÊ
Fig. 5.10a– Isolated AC and grid system characteristics

CHAPTER
Considering power flow from the isolated system to the grid system, the frequency control would determine
the DC voltage order at a value (ARBR) up to (but not exceeding) +Vorder maximum. This value is necessarily
less than the maximum setting for the remote (grid) system inverter (X2I, X1I). The grid system inverter
is essentially in constant reactive power control, with a value determined to match the desired reactive
power exchange with the receiving AC system.
In Fig. 5.10a, the operating point for full export power is OP1. However, the inverter will control the reactive
power exchange with its AC system by modifying the position of its constant reactive power line (X1I, X1R)
between a maximum and minimum value. The minimum value (X2I, X2R) is determined by the onset of
discontinuous current, whilst the maximum value is set below the maximum current setting for the rectifier.
Hence power transmission can be smoothly changed from export (OP1), through zero power (OP2) to
import (OP3) without the need to block the converters.
Similarly the use of the tapchanger can be employed to reduce the converter firing angles (B1R, C1R),
however as tapchanger action is slow, this will impose limits on how quickly maximum power transfer of
the link can be achieved.
At all times, within the constraints of the receiving system, the control of the DC power transfer and hence
the AC system frequency is under the control of the isolated system converter, regardless of the presence
of a telecommunication system. It should also be noted that if these control characteristics were to be
used, the damping circuits and di/dt reactors within the valve would require rating for prolonged operation
at, or near to, a = 90°.
5.11. STABLE OPERATION OF A HVDC CONVERTER IN AN AC SYSTEM

Analytical studies form an important part of planning a HVDC scheme. Location and size of a proposed
scheme and the possible consequences of its connection can be explored at an early stage. Suitable
geographical locations must be aligned with their electrical locations in order to minimize additional
infrastructure that may be required to connect the HVDC scheme, such as new lengths of conventional
AC line. Assessments of interactions with the connected AC systems – both static and dynamic – are
necessary, particularly with weak contingent AC systems.
The above aspects including static (or quasi steady-state type) interactions are assessed by balanced AC/
DC system load flows which are key operational snapshots for the HVDC scheme in conjunction with the
adjoining AC systems.
5.11.1. Load Flow Studies

The main objective of the load flow study is to ensure good initial system conditions prior to dynamic
studies. Various system conditions may need to be set up in load flow form (e.g. peak load, light load,
directions of DC power transmission, levels of DC power transmission, contingencies, etc.).
Other studies may require load flows to be performed in order to derive operational characteristics.
An example may be exploring the limits of reactive power exchange at a converter bus with a range of
parameters, e.g. converter bus voltage limits.
5.11.2. Objectives and Criteria

An important objective is to calculate 3-phase symmetrical Short-Circuit Levels (SCL) which are frequently
of interest at the converter bus and a few key buses in its electrical proximity, and solved load flows are
required for their calculation. SCL calculations are performed to establish system strength under specified
network conditions and for validation of any reduced equivalent system that may need to be derived. Note
that if a DC link is represented by a machine, as some studies require, its impedance should be set to a
large value to avoid contribution to SCL.
A further study requiring load flows is that of harmonic impedance assessment. This study necessarily
investigates many operating points to ensure satisfactory operation of harmonic filters.
A load flow is the solution of a network for given loading and generation conditions, and represents a snapshot
of the system for one set of conditions. Consequently, convergent load flows including realistic DC operating

CHAPTER
conditions and giving acceptable bus voltages, transmission angles, line power flows and generator and
transformer operating parameters must be derived for the specified system conditions.
The criteria for acceptable load flows are generally as follows:
➙ No transgression of the AC voltage operating limits under steady-state or emergency operation
➙ No transgression of generator reactive power capabilities
➙ No overloading of any lines and transformers
➙ Achievement of realistic and acceptable DC operating parameters at converter busbars
➙ Observance of stated reactive power exchange limits at the converter buses
Power or load flow studies may be required for stand-alone investigations or as a prerequisite to dynamic
studies such as those for transient stability.
The studies require all the usual load flow type data of a single-phase representation of the 3-phase
balanced network. This includes impedances, loads, generators and operating levels. The DC link is modeled
by converter equations (see section 2.2). Such studies are usually done in a positive sequence RMS frame
of quantities, such as those used by the PSS/E program.
Reactive Power Exchange (RPE) control and AC Voltage Control (ACVC) within the Reactive Power Controller
(RPC) are features of voltage control during the quasi steady-state period and the load flow therefore uses
its settings. If their impact is to be assessed, filters may be adjusted for voltage or reactive power exchange
changes.
5.11.3. Balanced Load Flow (Newton Raphson Algorithm)

The balanced load flow algorithm used in software such as PSS/E (or similar programs) uses classical DC
equations explicitly. The computation process essentially involves the solution of algebraic (unified or
sequential) AC/DC system equations in discontinuous time using relatively large (≈0.25 ms to 10 ms) time
steps. These are incorporated with the AC power system’s real and reactive power balance equations in
the traditional Newton Raphson iterative algorithm.
5.11.3.1. System Equations

The general AC - DC load flow problem is summarized as the solution of:
 ∆ P(V ,θ ) 
 
 ∆ P term (V ,θ , x ) 
 
 ∆Q(V ,θ ) =0 [Eqn 5.11a]
 
 ∆Q term (V ,θ , x ) 
 
 R(Vterm , x ) 
Where:
V is a vector of the voltage magnitudes at all AC system busbars
q is a vector of the angles at all AC system busbars (except the reference bus which is assigned q = 0)
x is a vector of DC variables
The subscript ‘term’ refers to the converter AC terminal busbar
R (Vterm ⋅ x)k = 0 is a set of independent equations derived from the DC system conditions
The fundamental current on the converter side is related to the direct current (derived in Appendix A2.2.1)
and is given by the equation:
6
Is = k ⋅ ⋅ Id [Eqn. 5.11b]
π
Where:
Is is the fundamental frequency component of the current waveform on the secondary of the converter
transformer
Id is the converter direct current
k is the commutation overlap factor, typically k ≈ 0.995

CHAPTER
The DC model is summarized as follows:
R (x,Vterm)k = 0 [Eqn 5.11c]
Where:
R(1) = Vd − k1 ⋅ a ⋅ Vterm ⋅ cos φ
3
R(2 ) = Vd − k1 ⋅ a ⋅ Vterm ⋅ cos α + ⋅ I d ⋅ Xc
π
R(3) = f (Vd , Id)
R(4) = control equation
R(5) = control equation
x = [Vd , Id , a, cosa, f]T
Where:
Vd is average DC voltage
k1 = k ( 3 2 / π )
a is the transformer off-nominal tap ratio
Vterm is the busbar voltage on the system side of the converter transformer
(and thus the converter AC source commutating voltage)
f is the phase angle of Vterm
a is the firing delay angle
The variables Pterm(dc) and Qterm(dc) represent the injected real and reactive power at the terminal busbar
as a function of the DC system variables, as shown below:
Qterm(dc) = Vterm ⋅ Ip sinf

= Vterm ⋅ k1 ⋅ a ⋅ Id ⋅ sinf [Eqn.5.11d]
P term(dc) = Vterm ⋅ Ip cosf

= Vterm ⋅ k1 ⋅ a ⋅ Id ⋅ cosf [Eqn. 5.11e]
Pterm(dc) = Vd ⋅ Id [Eqn. 5.11f]
5.11.3.2. DC Control Equations

The following equations are examples of valid control equations, which are simple and can be easily
incorporated into the solution algorithm (with the superscript ‘sp’ meaning specified):
Specified converter transformer tap
a – asp = 0
Specified DC voltage
Vd – Vdsp = 0
Specified DC current
Id – Idsp = 0
Specified minimum firing angle
cosa – cosamin = 0
Specified DC power transmission
Vd ⋅ Id – Pdcsp = 0
Specified converter terminal voltage
Vtermsp – Vterm = 0 [Eqn. 5.11g]
Specified terminal reactive power
Qtermsp(dc) – Qterm(dc) = 0
Where:
Qtermsp(dc) is taken as the reactive power required to maintain the voltage constant, or to observe a Q
exchange (power factor) band
CHAPTER
5.11.3.3. Newton Raphson Algorithm
The Newton Raphson algorithm involving repeat solutions of the AC - DC system is:
 ∆ P (V ,θ )   
     ∆θ 
 ∆Pterm (V ,θ , x )     ∆θ term 
     
 ∆Q(V ,θ )  =  J   ∆V  [Eqn.5.11h]
     ∆V 
 ∆Qterm (V ,θ , x )     term

     ∆ x
 R(Vterm , x )   
Where:
J is the matrix of first order partial derivatives (Jacobian)
∆Pterm = Ptermsp – Pterm(ac) – Pterm(dc) [Eqn. 5.11i]
∆Qterm = Qtermsp – Qterm(ac) – Qterm(dc) [Eqn. 5.11j]
and as noted earlier:
Pterm(dc) = f (Vterm . x) [Eqn 5.11k]
Qterm(dc) = f (Vterm . x) [Eqn 5.11l]
5.11.4. Unbalanced Load Flow

The 3-phase, unbalanced load flow process is used in PSCAD/EMTDC (or similar programs)
and uses the AC to DC conversion process implicitly. The computation process essentially involves solution
by integration of the physics of actual AC/DC components in continuous simulated time, using a very small
(≈ 50 µs) time step. This form of physical simulation is incorporated with the DC control system’s real actions.
The basic process of the controlled DC scheme’s operation is depicted in Fig. 5.11a. It shows the
control of DC current at the rectifier and extinction angle at the inverter to produce the firing delays
at each end through an associated phase-locked oscillator. The firing instants produce DC voltage at
each end, and the difference of these voltages divided by the connecting DC resistance gives the steady-
state DC current.
5.12. SYSTEM REPRESENTATION

For balanced steady-state system considerations such as those in a PSS/E study, the system representation
includes data for line and cable impedances, transformers, loads, generators and operating levels:
➙ Loads may be entered as constant, real and reactive power, or constant current or constant
admittance.
➙ The HVDC link is modeled by converter equations as indicated in section 5.11.3.
➙ Generators target a bus voltage while delivering a scheduled real power (PV), or fixed P/Q, except for
the swing (slack or reference) bus.
➙ The action of an SVC may be modeled as a PV bus with zero real power.
➙ Transformers may be fixed ratio devices or use a tapchanger to target a bus voltage.
➙ Phase shifting transformers may also be modeled.
➙ Transformer saturation and other such non-linearities are not usually modeled.
➙ For HVDC schemes, a built-in HVDC line model is usually available. The scheme model typically has
two 2-winding converter transformers built in, so if a tertiary filter is used, two external 3-winding
transformers or the three winding equivalent impedance forms are used. The converter transformer
usually has tap changing which can be used to target valve-winding voltage. For scenarios where
the taps may be required to control another quantity, such as inverter gamma as in some HVDC line
schemes, an iterative method controlled by some external ‘wrapper’ may be required to aid the process
and this is indicated in the ‘control equations’ referred to in section 5.11.3.2.

CHAPTER
(a)
T1
T2 Vd Rect
Id Rect
α Rect_meas α Rect
Phase
locked Regulator
oscillator -
+ Id*Rect
Tap_Up
Tap_Down
Tap α up_threshold
controller
α dn_threshold
RectifierÊcontroller
(b)
T1
Vd Inv
T2
Id Inv
γ Inv_meas
α Inv Phase
locked Regulator
oscillator -
+ γ order
Tap_Up
Vd up_threshold Tap
controller Tap_Down
Vd dn_threshold
InverterÊcontroller
Fig.5.11a– DC scheme operation
➙ AC harmonic filters on a DC scheme are represented by shunt capacitor banks, which supply the
Ê5.11a
equivalent fundamental frequency reactive power.
➙ Surge arresters are not modeled, as they are dynamic protective models operating at the cyclic level.
➙ If the load flow is to be used in dynamic studies using the DC models, a dummy offline machine with
ID corresponding to the DC line number must be included at the terminal busbars of the DC line. This
is because the DC dynamic model is a current injecting machine model, so the dummy machine may
not be necessary for other non-machine type DC models.
A typical DC data set for such studies is shown below in Table 5.12a.

CHAPTER
Current Rated
Rdc- Rcmp- Scheduled Metered
Line Control margin in current Min DC DC V to switch
Ohm Ohm DC voltage end
number mode static char (amps or voltage (kV) mode (kV)
(ohms) (ohms) (kV) (Rect/Invr)
(p.u.) MW)
1 Power 0.4 0 0.1 -508 158.4 0 0 0
Max Min Primary
Commutating Commutating
Bus Bus firing firing Bridges in base Measuring
Type resistance reactance
number name angle angle series voltage bus
(ohms) (ohms)
(deg) (deg) (kV)
Rectifier 90480 X 25 20 2 500 0.078 2.722 0
Inverter 90008 Y 24 22 2 525 0.098 3.433 0
Trans ratio (p.u.) Tap setting (p.u.) Max tap setting (p.u.) Min tap setting (p.u.) Tap step (p.u.)
0.14 0.95 1.2966 0.9217 0.0125
0.1333 1.091 1.2964 0.9024 0.013
Table 5.12a– Typical set of DC scheme data for load flow
For 3-phase unbalanced AC systems, the study data will typically include all the items in Table 5.12a but
normally in full electromagnetic detail:
➙ Transmission lines, filters and transformers would be represented with electromagnetic data
including coupling where relevant, and including high frequency effects as appropriate.
➙ Lines may be represented as traveling wave models with their parameters calculated internally by
geometrical tower data.
➙ Transformer models would need to represent saturation in addition to air cored reactances.
➙ Filters would have all components represented, not just the fundamental frequency main capacitor,
which would suffice for the balanced case representation.
A typical PSCAD/EMTDC working file showing the DC system components needed for such studies is
shown below in Fig. 5.12a. It should be noted that the AC systems used in such investigations are invariably
reduced equivalent systems, which mimic static and dynamic behavior at low harmonic frequencies and
inertial oscillations in particular.

CHAPTER
CHAPTER
BACK TO
HVDC
connection
VDCX1 VDCY1
274
IDCX
|
0.2[ohm]Ê 0.2[ohm]Ê
2 2 2
TÊ F1X TÊ F5X TÊ F9X TÊ TÊ TÊ
2 F7Y2
IVWSAX F3Y2 F11Y IVWSAY
Main Converter IVWSBX IVWSBY
circuit transformer IVWSCX IVWSCY

2 2 2
breaker
F9Y2 F5Y2 F1Y2
1 TLINEÊ1 1
TLineÊ1 TLineÊ1
2 2 2 TAPY
TAPX Tap
A B C C B A Tap
A A F2Y 2 F10Y 2 F6Y2
A A
324.0Ê[MVA] 324.0Ê[MVA] BRKY1
SCL IVWDAX IVWDAY SCL
BRKX1 B #1 #3 B B #3 B
#1
A
IVWDBX IVWDBY
50ÊHz C C C C 60ÊHz
500Ê[kV] 70Ê[kV] 70Ê[kV]
A
V
IVWDCX IVWDCY 70Ê[kV] 70Ê[kV] 500Ê[kV]
Q=Ê390V
2 2 2
Q=Ê308.6
P=Ê-1.017
P=Ê-0.7568
F8Y2 F4Y2 F12Y2
1.0e6Ê[ohm]
1.0e6Ê[ohm]
IDCY
VDCX2
ShuntsÊX ShY4 ShuntsÊY

Shx4 LVDC
PROT_RFRX ValvesÊof connection PROT_RFRY
AC
network 12-pulseÊbridge
ILWX(1)
ILWX(2) 1
ILWX(3)
2
3
IVWSAX 2 F1X
FP1
Filters IVWSBX 1 FP2 2 F2X
IVWSCX 2 PoleÊ1 FP3 2 F3X
IVWDAX 3 2 F4X
FP4 Pole
4

IVWDBX FP5 2 F5X
IVWDCX 5 2 F6X
P1 24 FP6 control
VLWX(1) VLWX(1) 6 FPÊ FP7 2 F7X
VLWX(2) 1 VLWX(2) 7 2 FP8 2 F8X
2 VLWX(3) 8 FP9 2 F9X
VLWX(3) VDCX1 VDCX 9 RECX
3 VLWX + FP10 2 F10X
3 4ÊShÊX D - 10 3 TAPX
IDCX FP11 2 F11X
Master 2 P1 P1 F 11 ILW3X BLCKEDX
VLWY(1) VLWY3 FP12 2 F12X
1 control VDCX2 11 24
VLWY(2) 4ÊShÊY INX FP24t1Ê2x2
2 CONTROL
VLWY(3) IVWSAY 11 24 2
3 FP1 F1Y
IVWSBY 1 INY BLCKEDY
FP2 2 F2Y
IVWSCY 2 3 TAPY FP3 2 F3Y
IVWDAY 3 ILW3Y RECY 2
FP4 F4Y
IVWDBY 4 FP5 2 F5Y
IVWDCY 5 2 F6Y
FP6
VLWY(1) 6 FPÊ 2
FP7 F7Y
VLWY(2) 7 24 2 F8Y
FP8
VLWY(3) 8 FP9 2 F9Y
VDCY1 VDCY 9
Master + FP10 2 F10Y
D - 10
IDCY FP11 2 F11Y
control F 11 FP12 2 F12Y
VDCY2 ILWY(1)
1 FP24t1Ê2x2
ILWY(2)
ILWY(3)
2
3
BLCKEDX
BLCKEDX
BLK
Fig. 5.12a– PSCAD study system

BIBLIOGRAPHY
[1] A. GavriloviĆ, P. C. S. Krishnayya, C. A. O. Peixoto, G. Liss, C. V. Thio, J. D. Ainsworth, J. P. Bowles,
G. D. Breuer, A. Hammad, D. Povh, “Interactions between DC and AC Systems”, IEEE, Montec, 1986.
[2] “Guide to Planning DC Links Terminating at AC Locations having Low Short-Circuit Capabilities. Part I:
AC/DC System Interaction Phenomena”, CIGRÉ Brochure 68, June 1992.
[3] H. L. Thanawala, “Maximum available power features of ac and dc transmission systems”, Power
Technology International, 1990.
[4] C. D. Barker, N. M. Kirby, N. M. MacLeod, R. S. Whitehouse, “Renewable Generation: Connecting the
Generation to a HVDC Transmission Scheme”, CIGRÉ Toronto, October 2009.
[5] R. P. Burgess et al, “Voltage/Var Control at McNeill Back-to-Back HVDC Converter Station”, CIGRÉ Paper
14-104, Paris, 1990.
[6] J. D. Ainsworth, “Power Limit Instability (Voltage Instability) in an HVDC Link Connected to a Weak AC
System”, CIGRÉ, Colloquium on HVDC and Weak AC Systems, September 1985.
[7] “Systems with Multiple DC Infeed”, CIGRÉ Brochure 364, December 2008.
[8] J. D. Ainsworth, A. GavriloviĆ, H. L. Thanawala, “Static and Synchronous Compensators for HVDC
Transmission Convertors connected to Weak AC Systems”, CIGRÉ Paper 31-01, Paris, 1980.
[9] J. Arrillaga, “High Voltage Direct Current Transmission”, The Institution of Electrical Engineers,
1998, Chapter 5.
[10] E. Kimbark, “Direct Current Transmission”, Wiley-Interscience, 1971, Chapter 5.
[11] F. Mazzoldi, J. Taisne, C. Martin, B. Rowe, “Adaptation of the Control Equipment to Permit
3-Terminal Operation of the HVDC Link Between Sardinia, Corsica and Mainland Italy”, IEEE Summer
Power Meeting 1988.
[12] P. Adam, V. Collet Billon, J. Ainsworth, A. Jeunehomme, I. Whitlock ,“The 2000MW Cross
Channel Link Between France and England: Compatability of the Two Converter Stations”, IEE 4 th Int.
Conference on AC&DC Power Transmission, London, September 1985.
[13] B. Andersen, D. Monkhouse, R. Whitehouse, “Commissioning the 1000MW Back to Back HVDC Link at
Chandrapur, India”, CIGRÉ Paper 14-114, Paris, 1998.
[14] J. Ainsworth, L. Haddock, R. Whitehouse, “Comparison of Control and Protection Techniques for HVDC”,
CEA, March 1989, Toronto.
[15] H. Thanawala, R. Whitehouse, Kwon Goo Ourck, Lee Suk Jin, “Equipment and Control Features of
Haenam-Cheju HVDC Link in South Korea”, CIGRÉ Paper 14-303, Paris 1994.
[16] C. Barker, R. Whitehouse, “An Alternative Control Strategy for the Thyristor Based HVDC Interconnection
of Renewable Energy Supplies”, IEE 8th International Conference on ACDC Power Transmission, London, 2006.

CHAPTER
6| HVDC CONVERTER STATION EQUIPMENT

6 HVDC CONVERTER
STATION
EQUIPMENT
A HVDC station includes a variety of high voltage electrical equipment.
Some are relatively standard; others - such as the thyristor
valves - are highly specialized, being found only in HVDC schemes
or similar Flexible AC Transmission Systems (FACTS).
The chapter commences with an introduction to the properties and
complexities of the high-power thyristor, the most vital component
in the HVDC station. You will learn about the physics of operation
of this semiconductor device, how it is made and what its principal
characteristics are.
Next you will discover how these thyristors are assembled together
with other components to make up a ‘thyristor valve’. This is a high
voltage switch (with a voltage rating of hundreds of kilovolts) that
behaves as one very large thyristor. Thyristor valve control, protection,
communication and even heat removal are explained in detail.
An equally important element is the converter transformer.
This chapter discusses how it differs from conventional transformers
for AC transmission.
This chapter also includes a detailed explanation of DC smoothing
reactors, measurement transducers, surge arresters, switchgear, wall
bushings and auxiliary systems.

Chapter contents
6. HVDC CONVERTER STATION 6.4.6. Special Testing Features......................................................... 348
EQUIPMENT...................................................................... 276 6.4.7. HVDC Bushings........................................................................... 349
6.4.8. On Load Tapchangers.............................................................. 350
6.1. THYRISTORS FOR HVDC......................................................... 279 6.4.9. Protection Schemes................................................................. 350
6.1.1. What is a Thyristor?.................................................................. 279
6.1.2. The Two-Transistor Model..................................................... 279 6.5. DC SMOOTHING REACTOR.................................................... 352
6.1.3. Construction................................................................................ 280 6.5.1.
The Purpose of a DC Smoothing Reactor in a
6.1.4. Off-State Properties................................................................. 283 HVDC Transmission Scheme................................................ 352
6.1.5. Turn-On Properties................................................................... 286 6.5.2.
The Design of HVDC Back-to-Back Schemes without
6.1.6. On-State Properties................................................................. 289 Smoothing Reactors................................................................. 353
6.1.7. Turn-Off Properties................................................................... 290 6.5.3.
Eliminating Discontinuous Current under Steady-State
6.1.8. Surge Current Ratings............................................................. 293 Operating Conditions............................................................... 353
6.1.9. Light-Triggered Thyristors (LTT).......................................... 294 6.5.4. AC Harmonic Performance .................................................. 360
6.5.5. Peak DC Current due to Commutation Failure ............ 362
6.2. THYRISTOR CONVERTERS...................................................... 295
6.2.1. What is a Thyristor Valve?..................................................... 295 6.6. MEASURING TRANSDUCERS................................................ 362
6.2.2. Valve Electrical Circuit............................................................. 296 6.6.1. DC Current Measurement...................................................... 362
6.2.3. Valve Voltage Distribution..................................................... 297 6.6.2. DC Voltage Measurement...................................................... 363
6.2.4. Voltage Grading Circuits......................................................... 301 6.6.3. AC Current and Voltage Measurement
6.2.5. Damping Capacitor................................................................... 301 on the Converter AC Connections..................................... 364
6.2.6. Damping Resistor...................................................................... 303 6.6.4. Conventional AC Switchyard Measurements............... 367
6.2.7. DC Grading Resistor................................................................. 304 6.6.4.2. AC Measurements for Harmonic Filter protection..... 368
6.2.8. High-Frequency Grading Capacitors................................. 305
6.2.9. The di/dt Reactor....................................................................... 307 6.7. OVERVOLTAGE PROTECTION............................................... 368
6.2.10. Valve Electronics and Associated Components........... 312 6.7.1. Surge Arrester Construction and Characteristics ...... 368
6.2.11. Thermal Design........................................................................... 317
6.2.12. Mechanical and Seismic Design.......................................... 322 6.8. CIRCUIT BREAKERS................................................................... 370
6.2.13. Valve Testing................................................................................ 326 6.8.1. Purpose and Minimum Requirements.............................. 370
6.2.14. Other Design Issues.................................................................. 332 6.8.2. Types............................................................................................... 370
6.2.15. Installation, Commissioning and Maintenance............ 334 6.8.3. Arc Characteristics.................................................................... 373
6.8.4. Circuit Breaker Standards...................................................... 374
6.3. VALVE COOLING SYSTEM....................................................... 335 6.8.5.
The Operation of a Circuit Breaker
6.3.1. Coolant........................................................................................... 336 in an AC System........................................................................ 375
6.3.2. Filtering.......................................................................................... 336 6.8.6. Converter Pole Circuit Breaker Requirements............. 378
6.3.3. De-Ionization............................................................................... 336 6.8.7. AC Harmonic Filter Circuit Breaker Requirements..... 379
6.3.4. Main Pumps................................................................................. 336 6.8.8.
The Operation of Circuit Breakers for
6.3.5. 3-Way Valve................................................................................. 336 DC Interruption........................................................................... 379
6.3.6. Coolers........................................................................................... 336 6.8.9. DC Commutating Switches for HVDC Systems............ 382
6.3.7. Expansion Tank........................................................................... 336
6.3.8. Make-Up System........................................................................ 337 6.9. WALL BUSHINGS ...................................................................... 383
6.3.9. Control............................................................................................ 337 6.9.1.
Porcelain Oil-Filled HVDC Wall Bushing
6.3.10. Instrumentation......................................................................... 337 Performance................................................................................ 384
6.3.11. Alarms............................................................................................ 337 6.9.2. Composite HVDC Bushings................................................... 384
6.4. HVDC TRANSFORMERS........................................................... 338 6.10. AUXILIARY POWER SUPPLIES............................................... 384

6.4.1. Insulation Design....................................................................... 341
6.4.2. Harmonic Effects....................................................................... 343 BIBLIOGRAPHY ............................................................................................ 388
6.4.3. Connections................................................................................. 343
6.4.4. Special Manufacturing Features......................................... 347
6.4.5. Sound Level.................................................................................. 348

6.1. THYRISTORS FOR HVDC
The thyristor is the most critical component in the thyristor valve and hence one of the most important in
an entire HVDC station. Thyristors used in HVDC converters have some of the highest voltage and current
ratings of any semiconductor devices used in any industry.
6.1.1. What is a Thyristor?

A thyristor is a bipolar switching semiconductor device which behaves like a controllable diode. It has
two main terminals - the anode and the cathode - and a smaller control terminal, the gate. The thyristor
normally only conducts current from anode to cathode.
For reverse voltages (anode negative with respect to the cathode), the thyristor does not allow current to pass
and ‘blocks’ voltage. For forward voltages (anode positive with respect to cathode), the thyristor also blocks
voltage until a small current signal is injected into its gate terminal. Thereafter, the thyristor conducts current
until the current drops to zero, at which point the thyristor resets itself and turns off again.
These properties, plus its very high efficiency (the highest of any switching semiconductor device) make
the thyristor a very suitable component for line-commutated HVDC. The thyristor valves (see section 6.2)
are one of the most important items of equipment in a HVDC converter station and the thyristor is the
most important component in the thyristor valve.
Thyristors used for HVDC valves are amongst the largest semiconductors of any type produced for any
industry. These components are expensive and there may be many thousands of them in a HVDC station.
Moreover, they are quite delicate and require many additional components to control and protect them.
Fig. 6.1a shows an 8.5 kV thyristor with an active silicon diameter of 115 mm (which starts life as a silicon
ingot of 125 mm diameter, hence such thyristors are often referred to as 125 mm, or 5 inch thyristors).
Thyristors can be Electrically-Triggered (ETT) or Light-Triggered (LTT). Light Triggered Thyristors are
Approx.Ê125Êmm
Approx.Ê175Êmm
Cathode
Anode
Fig. 6.1a– Modern 8.5 kV 125 mm thyristor: silicon wafer (left) and complete capsule (right). © Infineon Technologies AG.
discussed in section 6.1.9 below. The rest of this chapter relates to the more common type, the Electrically
Triggered Thyristor.
In the following sections, a brief explanation will be given first of the basic physics of a thyristor, followed by
a short description of its construction and then a discussion of its two stable states (ON and OFF) and two
transitional states (turn-ON and turn-OFF).
Terminology for thyristors is covered in [1].
6.1a The Two-Transistor Model

6.1.2.
A thyristor is a latching device which can be turned on by control action, but once turned on can only be
turned off again by arranging for the external circuit to force the current to zero.
For the thyristor to be turned on, it must have positive voltage across it.

CHAPTER
The thyristor belongs to the bipolar family of semiconductor devices. It operates with charge carriers of
both polarities (holes and electrons). It has a four-layer structure of alternating p-type and n-type silicon
as shown in Fig. 6.1b.
The p-type and n-type regions are created by mixing tiny amounts of suitable impurities, or dopants into
the silicon. For p-type silicon, suitable dopants are those from Group III of the Periodic Table, such as gallium
or aluminum, while for n-type silicon, suitable dopants are those in Group V of the Periodic Table, such as
phosphorus or arsenic.
The method of operation can be understood by considering the thyristor as a complementary pair
of Bipolar Junction Transistors (BJTs) connected together: a PNP transistor and an NPN transistor.
The collector of the PNP transistor is connected to the base of the NPN transistor, and the base
of the PNP transistor to the collector of the PNP transistor.
ThyristorÊ= pnpÊtransistor + npnÊtransistor
Cathode Cathode
n n
gate p gate p p
n n n
p p
Anode Anode Cathode
Emitter
G K Gate
Collector Base
Collector
Base
A Emitter
Anode
Fig. 6.1b– The two-transistor model of a thyristor
To turn-on the thyristor, a pulse of current is injected into the gate terminal at a time when the anode is
positive compared to the cathode. The effect of this is to introduce current into the base of the NPN transistor.
Since bipolar transistors behave as current amplifiers, this causes the NPN transistor to pass a larger current
6.1b current passed into the NPN transistor results in base current being
from collector to emitter. The collector
drawn out of the PNP transistor. By the same current-amplifying action as for the NPN transistor, this causes
the PNP transistor to pass more current from its emitter to collector.
The collector current from the PNP transistor feeds back into the base of the NPN transistor, causing even
more current to be passed from its collector to its emitter, causing more current in the PNP transistor, and
so on. A positive feedback mechanism has been created and the thyristor ‘latches’ into the conducting state.
Once turned on, the thyristor remains on until the external circuit forces the current to drop below the
holding current, which for large thyristors is so much less than the normal rated current that it can be
approximated as zero.
6.1.3. Construction
In its most basic form, a thyristor consists of a simple four-layer p-n-p-n structure as shown in Fig. 6.1c.
The alternating p-n-p-n structure is apparent from this diagram. The n-layer used to form the cathode
connection is usually more highly doped than the other layers (for reasons that will be explained in section
6.1.3.1 below) and is referred to as an ‘n+’ layer.
However, for high-power thyristors suitable for HVDC there are two modifications that are commonly made
to this basic structure. First, an amplifying gate structure is normally used.

CHAPTER
Cathode Gate Cathode
n+ p n+ G K
p
A
AluminiumÊmetallisation Anode
Fig. 6.1c– Simple thyristor structure
The amplifying gate consists of a small pilot thyristor integrated into the main silicon slice and arranged so
that its cathode feeds into the gate of the main thyristor. The amplifying gate allows a large thyristor to be
turned on efficiently by a relatively small pulse of gate current. The amplifying gate structure is normally
6.1c with an inter-digitated gate arrangement, whereby the outer edge of the amplifying gate is
combined
extended via ‘fingers’ so that it gives access to a large proportion of the total area of the silicon, as can be
seen on the left side of Fig. 6.1.a. This is an efficient way of turning a large-area thyristor on rapidly and
reducing the spreading time, however the inter-digitated gate does use up some of the area available for
carrying on-state current.
Secondly, the cathode side of the thyristor is usually equipped with small regions known as ‘cathode shorts’.
These are small regions on the cathode side where the n+ layer is missing and the p base (gate) is directly
connected to the cathode. The effect of this is equivalent to connecting a resistor from the gate terminal
to the cathode. It makes the thyristor less sensitive to dv/dt and spurious triggering due to induced gate
currents (thyristors can be spontaneously turned on by excessive positive dv/dt as explained in section
6.1.4 below). Fig. 6.1d shows a thyristor with an amplifying gate and cathode shorts.
cathode amplifyingÊgate gate amplifyingÊgate cathode
p K
n G
p
PilotÊ
anode thyristor
A
Fig. 6.1 d– Thyristor with amplifying gate and cathode shorts
6.1.3.1. Diffusion and Metallization

To create a thyristor it is necessary to produce this four-layer structure of alternating p-type and n-type layers.
Although it is theoretically possible to deposit alternating layers of p-type and n-type silicon to build up
6.1d
a thyristor (a process known as epitaxy), it is much easier to start with silicon of one type and create the
other three regions within it, by diffusing impurities into it from the outside.
The technique that is almost universally used for power thyristors follows this principle and starts with
a disc of silicon that has been doped to make it (rather weakly) n-type. This n-type silicon is produced by
Neutron Transmutation Doping (NTD), which involves bombarding pure silicon with neutrons from a nuclear
reactor. Small proportions of the neutrons collide with silicon nuclei and are absorbed by them, converting
them to phosphorus nuclei. Phosphorus is a suitable n-type dopant and this technique is excellent for giving
a low but very uniform level of doping throughout a large silicon crystal.
The n-doped silicon is sliced into wafers of an appropriate thickness (the thickness depends on the required
voltage rating) and then begins the process of creating the two p layers and the cathode n+ layer.
CHAPTER
The p-type layers are created first (see Fig. 6.1e) by placing the silicon slice in a furnace with some suitable
p-type dopant material, such as gallium or aluminum. The p-type dopant gradually diffuses into the silicon,
displacing a small proportion of silicon atoms by gallium or aluminum atoms. The resulting concentration
of p-type dopant is highest at the surface of the silicon slice (much higher than the concentration of n-type
dopant in the bulk of the silicon). The concentration rapidly decreases away from the surface, so that at
some point the concentration of p-type dopant becomes equal to that of the n-type dopant that was there
to start with and the two cancel each other out. It is at this point that a junction is formed. For high voltage
thyristors, this typically happens at around 100 μm depth.
Next, the n+ region, which is to form the cathode connection, is created on one side of the silicon via a similar
process. The main difference is that this layer has to be much thinner than the p layer into which it is diffused
(otherwise it would simply cancel out the whole p layer), so this step uses a higher concentration of dopant
(typically phosphorus) applied for a shorter time. The n+ layer must not cover the entire cathode-side area of
the silicon – gaps need to be left, (a) for the cathode shorts and (b) for the spacing between the gate structure
and the main cathode area. These gaps are created by creating a protective layer of silicon dioxide (SiO2) over
the areas where doping is not wanted, and then machining off the SiO2 layer after doping.
Finally the anode side, and most of the cathode side, need to be metallized to make good electrical connections
to the silicon. This is achieved by evaporating aluminum on to the silicon surface. On the cathode side, part
of the metallization also provides the connection which feeds the cathode current of the pilot thyristor into
the gate of the main thyristor.
Step 1 -ÊStartÊwithÊdiscÊofÊrawÊNTDÊn-typeÊsilicon Step 2 Ê-ÊDiffuseÊp-typeÊregionsÊintoÊbothÊsides

p
n n
Step 3Ê-ÊDiffuseÊn+ÊregionsÊintoÊselectedÊpartsÊofÊcathodeÊside Step 4 Ê-ÊMetalizeÊanodeÊandÊselectedÊregionsÊofÊcathode

Cathode AmplifyingÊgate Gate AmplifyingÊgate Cathode
p p
n n
p p
Anode
Fig. 6.1e– Diffusion steps to produce thyristor wafer
6.1.3.2. Bonding
In the finished thyristor, the silicon slice is always sandwiched between two discs of a metal such as
6.1e
molybdenum, chosen to have a coefficient of thermal expansion similar to that of silicon (so as to alleviate
lateral stresses due to thermal expansion/contraction). In some designs of thyristor, the silicon is fully-
floating, with no permanent bond to the molybdenum discs on either side.
In other designs, the molybdenum disc is bonded or alloyed to the silicon slice on the anode side. Compared
with fully floating silicon, thyristors made by this process can have improved thermal impedance and better
surge current performance, but the bonding process, if not performed carefully, can lead to distortion of the
silicon and therefore to a greater risk of mechanical failure.
Normally the bonding process (where used) is performed after the diffusion and metallization.
6.1.3.3. Beveling and Edge Treatment

Fig. 6.1e shows an abrupt vertical surface at the edge of the thyristor. However, for high voltage thyristors

CHAPTER
this is not practical, as it would lead to excessive surface electric field. Consequently, the edges of high
voltage thyristors are always beveled, using one of a number of possible techniques, to reduce the surface
electric field. Beveling is achieved by a careful grinding and machining process applied to the silicon wafer
after diffusion and metallization. Fig. 6.1f shows the options for edge beveling.
For fully floating silicon, either the double-positive (pulley-wheel) or double-negative bevel designs can be
used. With bonded thyristors, the positive-negative bevel is used.
FullyÊfloatingÊsilicon BondedÊsilicon
Double-positiveÊbevel Double-negativeÊbevel Positive-negativeÊbevel
Cathode Cathode Cathode
Mo Mo Mo
Si Si Si
Mo Mo Mo
Anode Anode Anode
=ÊÊProtectiveÊsiliconeÊrubberÊcoating
Fig. 6.1f– Options for edge beveling of thyristor
After beveling, the high-field region at the edge of the device may also be ‘passivated’ in order to reduce its
vulnerability to surface trapped charges. Typically this is done with a thin semi-conducting coating (various
different techniques are possible to achieve this), which is then covered by a protective coating of silicone rubber.
6.1.3.4. Test and Encapsulation

6.1f
After metallization and (where appropriate) bonding, the silicon is fully functional and can be subjected
to most of its electrical tests.
Thyristors for HVDC applications are always subjected to quite rigorous testing on every device (routine
testing) to verify that all the key electrical parameters are within the specified limits. Thyristors that fail these
tests are not suitable for use in HVDC projects and are rejected. To minimize the cost impact of thyristors
which fail these tests, most of the electrical testing is performed before the final assembly stage.
In the final stage, the device is encapsulated in a ceramic-insulated housing where the silicon and molybdenum
discs are sandwiched between thick copper pole-pieces.
6.1.4. Off-State Properties

An ideal thyristor would pass no current and block any voltage applied to it in either direction. Real
thyristors fall short of this ideal in several respects, allowing small amounts of current to pass, and being
sensitive to the rate of change of voltage (dv/dt) applied to the thyristor. These aspects are described in
the following sections.
6.1.4.1. Physical Principles

Before it has been turned on, the thyristor blocks voltages of both polarities, that is to say it does not
CHAPTER
Molybdenum SiliconÊwafer
discs
Cathode
Copper
Gate
Copper
ProtectiveÊsilicone Porcelain
rubberÊcoating envelope
Fig. 6.1g– Cross-section through an encapsulated thyristor (fully floating silicon)
allow current to pass in either direction. It does so because there is always at least one reverse-biased p-n
6.1g
junction (a reverse-biased diode) in each direction.
When the thyristor is positively biased (anode is positive compared to the cathode), the voltage is supported
by the junction between the upper p layer and lower n layer. This is a p-n junction similar to that inside a
diode, and it behaves in the same way: for a short distance either side of the junction, the charge carriers
(electrons and holes) are cancelled out and a depletion region, also known as a space-charge region (Fig.
6.1h) is created, where the semiconductor behaves like an electrical insulator. Like any thin insulating layer,
the depletion region behaves as a capacitance; hence an off-state thyristor has a measurable capacitance
known as its junction capacitance. Because the width of the depletion region increases with the applied
voltage, the junction capacitance is non-linear, reducing as voltage is increased.
When the thyristor is negatively biased (anode is negative compared with cathode), two depletion regions
form, one at the junction between the cathode and gate, and one at the junction between the anode and the
lower n region (the so-called n-base). The latter takes up most of the voltage across the thyristor, the gate-
cathode junction being capable of withstanding only a quite low voltage.
6.1.4.2. Off-State Current

An ideal thyristor would behave as a perfect switch. This means that in the off-state it should carry no
current at all while blocking very high voltage.
A real thyristor does not achieve this ideal. During the off-state, there is always a small current that flows
through the thyristor, and the value of this current depends on temperature and voltage. Moreover, different
thyristors from the same production batch can have quite widely differing values of off-state current, which
can present some challenges in the design of HVDC valves.
Fig. 6.1i shows the V-I characteristics of an ideal and a real thyristor during the off-state.
There are several effects which contribute to the total off-state current through a thyristor:

CHAPTER
Cathode Cathode
n+ n+ n+ n+ n+ n+
p
p
DepletionÊregion
n
n
DepletionÊregion
p p
Anode Anode
(a)ÊForwardÊblocking (b)ÊReverseÊblocking
Fig. 6.1h– Depletion regions during forward blocking state (a) or reverse blocking state (b)
IdealÊthyristor
OFF-state
6.1h l l RealÊthyristor
ON-state (25¡C)
RealÊthyristor
(125¡C)
DominatedÊby DominatedÊby VBOÊself-triggering
avalancheÊcurrent leakageÊcurrent (usuallyÊdestructive)
V V
ReverseÊavalancheÊbreakdown
(safeÊifÊenergyÊlimited)
Fig. 6.1i– Off-state V-I characteristics of an idealized (left) and a real (right) thyristor
➙ L eakage current: this applies at all values of voltage and for any dv/dt (including the static case of
DC voltage). Leakage current arises from charge carriers (electrons and holes) that are thermally
excited (created by the random thermal vibration of the silicon crystal structure). This effect follows
6.1ian Arrhenius-type sensitivity to temperature, with leakage currents increasing exponentially with
temperature. On the other hand, leakage currents vary relatively little with voltage (a modest leakage
current appears at very low applied voltage and does not increase very much as the voltage is raised).
➙ Avalanche current: A thyristor subjected to off-state voltage will experience a high electric field
in its depletion regions, and this high field will cause charge carriers to be accelerated. Some of
these fast-moving charge carriers will collide with atoms in the crystal structure, causing further
generation of holes and electrons which then accelerate in opposite directions. A positive feedback
avalanche mechanism is created, giving rapid multiplication of current, which can be damaging if
the energy is not limited. The avalanche effect is similar, but not quite identical, to the Zener effect
used in Zener diodes. Unlike leakage current, avalanche current depends much more strongly on
voltage than on temperature. It increases with voltage raised to a power, but for a given voltage,
the avalanche current actually decreases as the temperature increases.
➙ Displacement current: this depends primarily on dv/dt, since it arises from the capacitive properties
of an off-state thyristor, where the depletion region(s) act as a voltage-dependent capacitance.

CHAPTER
6.1.4.3. dv/dt Capability
The thyristor exhibits a junction capacitance Cj when it is in its off-state, as a result of which an applied dv/
dt will cause a current to flow in the thyristor. This current is defined by:
i = Cj ⋅ dv/dt [Eqn 6.1a]
This current is superimposed on top of the leakage and avalanche currents that may already be present.
When the total current flowing through the thyristor, due to all these effects, exceeds a certain threshold, it
becomes sufficient to cause the thyristor to trigger.
In the right-hand curve in Fig. 6.1i we can see the combined effects of leakage and avalanche current.
Leakage current is the dominant source of off-state current at low to moderate voltages (up to around
70% of the nominal voltage rating of the thyristor). In this region, off-state current increases rapidly with
temperature. For voltages approaching the nominal voltage rating of the thyristor, avalanche current becomes
the dominant source of off-state current (except at very high temperature), which means that the voltage
rating of the thyristor (defined as the voltage at which the avalanche current becomes excessive) is lower
at low temperatures.
When the sum of leakage, avalanche and displacement currents becomes too high, the thyristor is at risk of
damage. When these currents flow in the reverse direction, the limiting factor is essentially thermal and related
to the temperature rise caused by the current (power dissipation is simply voltage multiplied by current, and
since the voltage can be quite high, it does not require much current to result in high power dissipation).
When these currents flow in the forward direction, the limit is generally reached at a lower voltage and
occurs when the total current becomes sufficient to cause the thyristor to turn itself on spontaneously.
Since this is usually destructive, it must be avoided at all costs.
Spontaneous triggering due to dv/dt in the positive direction is a particular risk in HVDC, and thyristors capable
of withstanding very high dv/dt without self-triggering are needed. In addition, it is usual to provide some form
of dv/dt protection for the thyristors.
6.1.4.4. Voltage Ratings at High Temperature

Thyristors can also be damaged by applying voltage at high temperature.
The headline voltage ratings normally quoted by the manufacturer are generally valid for temperatures between
room temperature and about 125°C. However, at higher temperatures, in the range reached during some fault
conditions in HVDC applications, the voltage withstand capability starts to fall off very rapidly with increasing
temperature. This is because the leakage current becomes the dominant source of off-state current in the
thyristor, and the leakage current increases exponentially with temperature.
Generally, the positive direction voltage capability falls off at a lower temperature than the negative
polarity. Fig. 6.1j illustrates a typical set of characteristics.
6.1.5. Turn-On Properties

A thyristor is turned on by passing a short pulse of current (a few amps) into the gate terminal of the device
for a few microseconds. When this occurs, the result is a positive feedback mechanism within the thyristor,
causing it to latch into the conducting state. This initial turn-on phase is fast – generally taking less than 1
μs from the appearance of gate current to the thyristor starting to conduct significant amounts of current.
However, the complete turn-on process takes several orders of magnitude longer than this (> 1 ms).
The initial turn-on phase is perhaps the most dangerous for a HVDC thyristor. This is because the thyristor
very quickly starts to conduct large amounts of current (rising to several thousand amps within just a few
microseconds) and all of this current goes through a tiny area of the silicon close to the gate terminal. It takes
much longer for the conducting area of the silicon to spread to cover the full area available.
The thyristor turn-on process is best considered by splitting the turn-on into two phases: the initial turn-on
(a few microseconds) and the spreading phase.

CHAPTER
DominatedÊby Forward
V avalancheÊcurrent
(%ÊofÊrated) Reverse
100%
DominatedÊby
leakageÊcurrent
50%
25¡C 75¡C 125¡C 175¡C

TÊjÊ(¡C)
Fig. 6.1j– Positive and negative voltage rating versus temperature
6.1.5.1. Initial Turn-On

After a pulse of current is applied to the gate terminal of the thyristor, initially (for a few hundred
nanoseconds) nothing happens. During this period, the positive feedback mechanism described in section
6.1.2 is starting to take6.1j
place, but initially the current that passes through the thyristor (from anode to
cathode) is not large enough to have any appreciable effect on the external circuit.
After a turn-on delay time, td, which depends on the anode voltage but is typically of the order of 1 μs,
the turn-on process finally starts to have a tangible effect, manifested by the voltage across the thyristor
starting to fall: see Fig. 6.1k. Turn-on delay time, td, is defined as the time from when gate current reaches
10% of its peak value (on the way up) to the time when the thyristor anode-cathode voltage falls to 90%
of its initial value.
After the turn-on delay time, the time taken for the anode voltage to fall from 90% to 10% varies from 100-200
nanoseconds at maximum voltage to a few microseconds at low voltage.
Once the anode-cathode voltage starts to fall, the current in the thyristor starts to rise very quickly. This
is where the most dangerous time begins for the thyristor, because all this current is confined to the tiny
region near the gate terminal, as mentioned in section 6.1.5. Moreover, there is no protection which can
be applied to the thyristor to prevent it from being damaged by excessive current during this period. The
only possible preventative measure is to ensure that the rate of rise of current is not excessive, by including
sufficient passive impedance (di/dt-limiting saturable reactance) in series with the thyristor.
During the first few microseconds after turn-on, the load current in the thyristor is negligible. Almost all of the
current that flows at this time arises from three sources:
➙ From external stray capacitances, chiefly the converter transformer windings and associated bushings.
This is limited by the di/dt reactor.
➙ From the damping (snubber) capacitor. This is limited by the damping resistor.
➙ From any undamped capacitances in parallel with the thyristor. Since the current from such
capacitances is limited only by the turn-on speed of the thyristor itself, it is important that these
undamped capacitances are kept to a minimum.
Taking these three factors together, the total current that flows in the thyristor in the first few microseconds
after turn-on looks like the solid line in Fig. 6.1l below.

CHAPTER
ShortÊriseÊtime
(egÊ<250Êns)
100%
DecayÊtime-constant
Gate typicallyÊaÊfewÊμs GateÊpulseÊtypicallyÊcontinues
current ofÊseveralÊtensÊofÊμs
10%
0%
t
Turn-onÊdelay IdealÊthyristor
timeÊ(td)
RealÊthyristor
100% RiseÊtimeÊ(tr)
90%
Thyristor
voltage
ʻTail voltageʼ - typically a few

hundredÊvolts
10%
0%
0 1 2 3 4
tÊ(μs)
Fig. 6.1k– Initial turn-on process Time
di/dtÊlimitedÊonlyÊby
saturatedÊinductanceÊof
reactor
6.1k
Reactor
Thyristor
saturates
current
di/dtÊlimitedÊby
unsaturatedÊinductance
ofÊreactor
DampingÊcircuitÊcurrent
≈(V/R)eÊ(-t/RC)
0 1 2 3 4 5 6
t (μs)
Time
Fig. 6.1l– Total thyristor inrush current
Typically, for turn-on from the full rated voltage of the thyristor, it will take 3-5 μs for the di/dt reactor to
saturate. During the unsaturated phase, the current will only have risen by a few hundred amps, but after
the reactor saturates, the rate of rise of current will be much greater. For a typical repetitive condition, the
6.1l
peak current reached might be around 1000-1500 A, at 6-8 μs after turn-on. However, if the valve turns on
while its surge arrester is carrying current, the saturated stage carries on for much longer and can reach a
peak of 6-8 kA after about 10 μs. This event does not occur repetitively but can occur for 3-5 consecutive
cycles (the detection and tripping time of the converter AC breaker) in cases of severe AC-side temporary
overvoltage. It represents the most severe turn-on duty a HVDC thyristor has to withstand.

CHAPTER
6.1.5.2. Spreading Phase
During the first few microseconds after turn-on, the thyristor is under great stress, since the current
flowing in it (from external capacitances discharging) is often very high and all this current flows through
a tiny area of the silicon.
Once conduction has been established, the conducting area starts to spread outwards from the area
initially turned on (a narrow region surrounding the gate structure). The speed at which it spreads is referred
to as the spreading velocity. The spreading velocity depends on the current density in the turned-on area,
but is generally of the order of tens of meters/second. Thus, it can easily take more than a millisecond (one
thousand times longer than the initial turn-on) for the conducting area to cover the entire silicon slice.
During this time, the current goes through an area smaller than the complete silicon slice, so the on-state
voltage is higher than it would normally be for a fully turned-on thyristor (Fig. 6.1m).
This gives rise to spreading losses, which are actually additional conduction losses over and above what are
obtained as a result of the normal on-state characteristics as described in the next section.
Valve
current
0
t
IdealÊthyristor
3
RealÊthyristor
Thyristor 2
voltage
(volts) 0
0
t
ʻSpreading lossʼ- typically a
ÊfewÊjoules
Thyristor
power
Fig. 6.1m– Illustration of the spreading (=V.I)
phase of a thyristor (note the much longer 0
time-scale in comparison with 6.1k) 0 1 2 3 4 5 6 7 t(ms)
Spreading losses generally amount only to a few Joules each cycle. Thus, they are not particularly important
in the overall losses of a thyristor. Nevertheless, they need to be calculated as part of the declaration of
power losses for the station.
6.1.6. On-State Properties

6.1m
VT
Once it has been fully turned on, a thyristor behaves (V) IdealÊthyristor
SimpleÊapproximation
as a very efficient switch. It is not perfect however, 4 RealÊthyristor
VTÊ=ÊVoÊ+ÊI.ÊRo
and there is a small on-state voltage drop across it.
This voltage drop, usually called VT, is a few volts 3
and is important because it gives rise to most of
the power losses that occur in the thyristor. The 2
Vo R0
so-called conduction loss in the thyristor is simply
1
current multiplied by on-state voltage drop.
The on-state voltage is not a constant but increases 0
with current and depends on temperature. I
ConductionÊlossÊ=ÊcurrentÊxÊon-stateÊvoltage
A simple approximation to the on-state voltage
drop of a thyristor is the straight-line approximation Fig. 6.1n– On-state voltage (straight-line
approximation)
shown in Fig. 6.1n. This approximation is generally

CHAPTER
6.1n
good enough for any steady-state calculations, up to the continuous rated current of the thyristor (a few
kA). It will be seen that the on-state voltage drop can be approximated as a fixed threshold voltage V0 and a
slope resistance R0. The threshold voltage V0 varies little between different thyristor types and is generally
about 1.2 V. The slope resistance R0 is a function of the silicon thickness (and hence voltage rating) and
area, larger areas and thinner silicon giving lower values of R0.
This expression is not exact, and at both very
low currents (< 10% of rated current) and high VT
(V) BetterÊapproximation
currents (significantly above rated current) this 4
VTÊ=ÊVo+ÊI.RoÊ+ÊEo.In(I)Ê+ÊFo.Ê√I
approximation is no longer accurate enough. Better
accuracy is achieved by adding terms proportional 3
to the log of current and the square root of current,
as shown in the equation below and in Fig. 6.1o: 2
VT = V0 + I ⋅ R0 + E0 ⋅ ln(I) + F0 ⋅ √ I [Eqn 6.1b] 1

The on-state voltage also depends on temperature.
0
As the temperature increases, the threshold voltage I
tends to reduce but the slope resistance increases. Fig. 6.1o– On-state voltage (more accurate
At low values of current, the on-state voltage drop approximation)
therefore reduces as the temperature increases.
However, above a certain value of current (the VT
cross-over current), the on-state volt drop will (V) VTÊ=ÊVoÊ+ÊI.RoÊ+ÊEo.In(I)Ê+ÊFo.Ê√I
increase as the temperature increases (see Fig. 4 Voʼ Roʼ EoÊandÊFoÊareÊtemperature-dependent
125¡ÊC
6.1p). 3
6.1o 25¡ÊC
Typically, the crossover current is a few kA for a
2
large thyristor. This means that at the rated current ʻCross-over currentʼ
at which the thyristor operates, the on-state voltage typically a few kA for
1
drop is relatively insensitive to temperature and large power thyristor
temperature effects can generally be neglected. 0
I
At high values of current, such as can occur during
Fig. 6.1p– On-state voltage (effects of
converter faults, temperature effects become temperature)
extremely important and it is vital for accurate
modeling that a proper temperature-dependent
on-state model is available.
6.1p
6.1.7. Turn-Off Properties
An ideal thyristor should cease conduction exactly at the instant where the current drops below the holding
current, and thereafter instantly recover its full ability to block voltages and dv/dt of either polarity.
Real thyristors fall short of this ideal in two important respects:
➙ The thyristor continues to conduct for a short period after current zero, resulting in a period of reverse
current and reverse recovered charge or stored charge, Qrr
➙ Even after the reverse current has decayed away, the thyristor remains vulnerable for some time to
excessive positive voltage or dv/dt and hence has a defined turn-off time, tq during which the voltage
must remain negative
6.1.7.1. Reverse Current and Stored Charge

During the on-state interval of the valve, the thyristor behaves almost like a short-circuit. Towards the end
of the on-state interval, the next valve in sequence is fired, and this causes the current in the valve that is
already conducting to decrease as the current commutates from it into the next valve (see section 2.2).
The rate of fall of current is determined by the commutating EMF and the total inductance around the
commutating loop, but is typically a few amps per microsecond.
An ideal thyristor should cease conduction exactly at the instant where the current reaches zero (orange
trace in Fig. 6.1q).

CHAPTER
However, real thyristors do not behave like this. For a short period after current zero, the current continues to
become more negative at the same rate as before, until a time trr, a few tens of microseconds after current zero.
At this point, the thyristor quite abruptly starts to increase its impedance and the reverse current decreases.
The result is a period of reverse current flow in the thyristor, comprising two phases (see Fig. 6.1q):
➙ Up to trr: where the current in the thyristor increases more or less linearly in the negative direction,
reaching a peak value Irr
➙ After trr: where the current in the thyristor decreases (becomes less negative) following approximately
an exponential relationship: I(t) = Irr ⋅ e–(t – trr)/t
di/dtÊdeterminedÊby
IdealÊthyristor
externalÊcircuit
I RealÊthyristor
TypicalÊsplit:
40% 60%
0
t
Irr PeakÊreverse
ShadedÊareaÊ=Êstored
currentÊ(Irr)
chargeÊ(Qrr)
Reverse
recovery
timeÊ(trr)
Fig. 6.1q– Reverse current in thyristor at turn-off
The total area under this curve represents the stored charge (Qrr) of the thyristor. Typically, about 40% of Qrr
occurs before the peak of reverse current and 60% after it, although this varies from thyristor to thyristor.
6.1q
Qrr and Irr are very important parameters that need to be known with a high accuracy so that the valve design
may proceed. They are important for two main reasons:
➙ Irr, and to a lesser extent Qrr, greatly increase the commutation overshoot which occurs after the
thyristor turns off compared to an ideal thyristor. This effect is discussed in section 6.2
➙ The spread (tolerance) of Qrr (referred to as ∆Qrr) amongst the different thyristors in a valve has a huge
adverse effect on voltage sharing in the valve after turn-off, an effect which can persist for seconds
(easily long enough to last until the thyristors are next turned on again)
6.1.7.2. Turn-Off Time

An ideal thyristor should be able to withstand any positive voltage with any dv/dt (up to the normal off-
state ratings as discussed in section 6.1.4 above) almost immediately after it has turned off. A real thyristor
cannot do this: it needs to have a recovery period (the turn-off time, tq – see Fig. 6.1r) of some hundreds
of microseconds, during which the voltage across it has to stay negative, before it is safe to apply any
positive voltage. Even then, a further period of time needs to elapse before the thyristor is safely capable
of withstanding high dv/dt.
If the voltage becomes positive too soon, or has too high a rate of change (dv/dt) after the thyristor has turned
off, the thyristor will spontaneously turn itself back on again, a process known as forward recovery failure.
Spontaneous self-triggering of the thyristor as a result of forward recovery failure can destroy the thyristor.
If the spontaneous self-triggering takes place at sufficiently low voltage or dv/dt, the thyristor will usually
survive, but at higher dv/dt there is a risk that the thyristor will be destroyed. The most dangerous zone is
where the thyristor nearly, but does not quite, withstand the re-applied voltage.
What is happening here is that, as the voltage across the thyristor changes from negative to positive, any
charge carriers still remaining in the silicon from the time when the thyristor was conducting, give rise to a
short pulse of forward recovery current in the thyristor. The more time that elapses after turn-off, the more
charge carriers recombine and hence the smaller this forward recovery current will be. Above a certain

CHAPTER
threshold value, the forward recovery current is high enough to cause spontaneous self-triggering via a
similar regenerative process to that seen during normal turn-on, but without involving the gate terminal.
At high dv/dt, the forward recovery current is superimposed on a displacement current flowing through
the junction capacitance of the thyristor (see section 6.1.4.3), hence making the thyristor more sensitive.
Similarly, at high temperatures the forward recovery current is superimposed on the leakage current. For
these reasons, turn-off time increases with both dv/dt and temperature.
Fig. 6.1r shows the voltage and current across a thyristor when subjected to an event which causes the
voltage to become positive at various times after current zero. For the blue, green and orange traces,
the voltage becomes positive after the turn-off time has elapsed and the thyristor safely withstands
the re-applied voltage. For the red trace, voltage is re-applied within the turn-off time and the thyristor
spontaneously re-conducts.
Because forward recovery failure can destroy the thyristor, it is important to provide protection against it.
This is normally performed by preemptively turning on the thyristor correctly, that is, applying a legitimate,
controlled gate pulse via its gate terminal, if the voltage across it appears likely to become positive within
the turn-off time.
Forward
recovery
failure
IÊ(thy)
Turn-off
timeÊ(tq) Externally
imposed
transient
VÊ(thy)
Fig. 6.1r– Voltage and current across a thyristor with re-applied forward voltage
Most of GE Vernova’s HVDC valves have included a sophisticated forward recovery protection in which the
TCU provides forward recovery protection locally at each thyristor level and the thresholds of this protection
depend on the 6.1r
junction temperature and rate of re-applied voltage (dv/dt).
6.1.7.3. Carrier Lifetime Control

The lifetime of the charge carriers (how long they last before re-combining) is a critical parameter for a
thyristor. Carrier lifetime is directly linked to the turn-off time of the thyristor and higher lifetimes result
in larger values of Qrr, ∆Qrr and Irr, all of which are undesirable. On the other hand, long lifetime is beneficial
for minimizing the on-state voltage drop, which is responsible for most of the power losses in the device.
Therefore, the carrier lifetime is an important compromise which must be made in the design of a thyristor.
The carrier lifetime can be made shorter by creating trapping sites in the silicon crystal structure, where
carriers can get trapped and are therefore more likely to recombine with a carrier of the opposite polarity.
There are various ways of creating these trapping sites and therefore altering the carrier lifetime of a thyristor.
In older thyristors, small amounts of gold were diffused into the silicon to achieve this. This technique is still
used today in some special devices, but for high power thyristors a better technique is electron irradiation,

CHAPTER
which involves bombarding the silicon with high-energy electrons, some of which collide with atoms in the
crystal structure and result in trapping sites. The big advantages of electron irradiation are that (a) it can be
applied at any stage prior to the final encapsulation of the thyristor and (b) it can be applied multiple times
in order to make fine adjustments (in an iterative process) to the performance of the thyristor. It is therefore
possible (although more costly) to perform multiple cycles of irradiation followed by Qrrmeasurement in order
to achieve very precisely controlled values of Qrr.
6.1.8. Surge Current Ratings

In HVDC valves, the highest surge current that can be obtained is due to a flashover across another valve in
the bridge. Here, the resulting fault current is limited by the leakage inductance of the converter transformer
and is a comparably modest value, typically only about 10 times the normal rated current.
This is lower than the fault current that is experienced for some applications of thyristors. However, the
thyristor is expected to be able to withstand voltage again after the fault current has stopped. This is a
very important point, since the combination of high voltage across the thyristor at very high temperature
(because of the current that has just passed) is very dangerous. HVDC thyristors must be able to block
voltage immediately after carrying a surge current.
As soon as such a fault is detected (which takes a few milliseconds) the HVDC control and protection
system immediately sends a trip signal to the AC circuit breaker, which takes about 3 cycles to come
into effect. It also sends a blocking command to the valve(s) which are carrying the surge current. What
normally happens then is that these valve(s) will carry a single cycle of fault current, then stop conducting,
and then will have to withstand the line-to-line AC voltage from the valve-winding side of the transformer
until the circuit breaker has actually opened. In this case, the valve has to be able to block several cycles
of both positive and negative voltage (Fig. 6.1s). The most severe case is the first cycle of positive voltage
after blocking, because (as noted in section 6.1.4 above) the thyristor voltage withstand capability at high
temperature is poorer in the positive than in the negative direction and the silicon is hottest immediately
after conduction.
Occasionally there may be disturbances to the voltage which is applied to the valve after blocking, which
make it unsafe for the valve to attempt to block positive voltage. In these cases, one of several different
protective functions may force the valve to re-conduct, after which it behaves as a diode and conducts
multiple cycles of fault current. In these cases, the valve no longer has to withstand any positive voltage,
ITSM
Current
0 360 720 1080

TimeÊ(electricalÊdegrees)
0.7ÊVDRM
TypicalÊconduction
angleÊ=Ê330¡
Voltage
Fig. 6.1s– Idealized current, voltage and junction temperature for a single-cycle valve insulation fault

CHAPTER
but still has to withstand the brief periods of negative voltage that occur between cycles (but not after the
last cycle), as shown on Fig. 6.1t. In this case, the most critical point is just at the end of the penultimate
cycle of current, where the valve sees a relatively high reverse voltage at very high temperature.
6.1.9. Light-Triggered Thyristors (LTT)

In an Electrically-Triggered Thyristor (ETT) the thyristor is turned on by injecting a pulse of current into the
gate terminal. The pulse of current has to come from a dedicated gate electronics unit located at the same
ITSM
Current
OpenÊbreaker
here
360 720 1080
TimeÊ(electricalÊdegrees)
TypicalÊconduction
angleÊ=Ê330¡
Voltage
0.7ÊVDRM
Fig. 6.1t– Idealized current, voltage and junction temperature for a multiple-cycle valve insulation fault
potential as the thyristor cathode. The Thyristor Control Unit (TCU) has to get its energy from somewhere,
which is not easy at the very high voltages at which the valve operates. The TCU also contains around 90%
of the total component count in the valve, and therefore has an impact on the calculated reliability of the
valve, which must be taken into consideration when calculating redundancy requirements to meet the
6.1trequirements.
overall system reliability
A Light-Triggered Thyristor (LTT) has no electrical connection to its gate, instead using a light guide to direct a
short pulse of infrared radiation from a laser onto a photosensitive region of the silicon in the gate area. The
optical pulse generates electron-hole pairs, which then perform the same function as the electrical current
pulse from a TCU (although a more sensitive amplifying gate is usually required, since the optical signal is
weaker than the current pulse from a TCU).
Since the LTT requires no electrical signal to turn it on, the laser and its power supply can conveniently be
located at ground potential in the Valve Base Electronics (VBE) cubicle.
Fig. 6.1u shows a LTT manufactured by Infineon Bipolar.
LTTs are not new; they have been in production for more than 20 years by several different manufacturers.
For high voltage applications such as HVDC, LTTs offer some advantages because of the great simplification
that is possible in terms of valve electronics. However, they still have a number of shortcomings. At the time
of writing, there are three principal shortcomings:
➙ In HVDC, the TCU does much more than just turn the thyristor on in response to a control command.
It also needs to turn the thyristor on autonomously during faults, in response to overvoltage, dv/dt
etc. There is really no point in providing an LTT if the thyristor level still needs a separate electronics
card to provide the protection functions. So in order to realise the full potential of the LTT, the device
needs to include in-built protection for these aspects (a so-called “fully Self-Protected LTT”).
➙ LTTs are significantly more expensive than ETTs. The differential cost of the LTT with respect to the
ETT may be more than the cost of the gate electronics that it makes obsolete.
➙ LTTs are at present only available from a very limited number of suppliers.

CHAPTER
Fig. 6.1u– Infineon light-triggered thyristor. © Infineon Technologies AG.
LTTs with in-built protection are available today. VBO and dv/dt protection have been commercially available
for more than 20 years but forward recovery protection has only become commercially available at a cost-
effective price in the last few years, and the price premium compared with ETTs remains high.
In addition, the in-built VBO level of an LTT is fixed, unlike the VBO level for an ETT which is determined by the
TCU and can therefore be adjusted. The adjustability of the VBO level with an ETT allows the valve to operate
in discontinuous current mode even when a thyristor level is operating on its VBO protection, a feature which
is not available with an LTT.
6.2. THYRISTOR CONVERTERS

The thyristor valves are amongst the most important items of equipment in a HVDC converter station.
They are a highly specialized type of equipment and, in the converter station, typically are second only to
the transformer in terms of cost.
6.2.1. What is a Thyristor Valve?

A thyristor valve is essentially just a switch that exhibits unidirectional (diode-like) conduction when turned
on. It is required to turn on when instructed by the control system, and then turn off again when the external
circuit forces the current to zero (which happens naturally each cycle in line-commutated HVDC installations).
The term valve is a throwback to the earlier days of HVDC, when mercury-arc valves were used
for this function. They operated in a totally different way (being essentially gas-discharge tubes, hence the
name valve) but did essentially the same job.
When thyristors were introduced, the name valve was retained. A Thyristor Valve is an assembly of series/
parallel thyristors that behaves like a single, very large thyristor.
Thyristor valves typically need to be able to withstand hundreds of kilovolts when they are in their off-state.
Since the thyristors from which they are constructed are only capable of withstanding a few kV each, the
thyristor valve needs to contain a large number of thyristors in series. Ensuring that these series-connected
thyristors share the applied voltage evenly is a significant challenge.
Many early designs of thyristor valve also used parallel connections of thyristors in order to increase the
effective current rating [3]. However, with the very large thyristors now available, with diameters of up to
150 mm or even larger (section 6.1), this is no longer necessary.
The correct design of a thyristor valve is a complex, multi-disciplinary process involving a range of engineering

CHAPTER
disciplines including power engineering, power electronics, analog electronics, semiconductor physics, high
voltage corona, materials science, heat transfer, fluid mechanics, mechanical and structural engineering.
6.2.2. Valve Electrical Circuit

The smallest building block of a thyristor valve is termed a Thyristor Level or often simply a level. A Thyristor
Level comprises one thyristor plus all the ancillary components associated with (and unique to) it.
6.2.2.1 Thyristor Level (Basic Circuit)

When Electrically-Triggered Thyristors (ETT) are used (see section 6.1), the simplest electrical circuit for
one thyristor level consists of the thyristor, its TCU, an RC damping circuit and a DC grading resistor (see
Fig. 6.2.2a).
The TCU issues the turn-on current pulse to the gate terminal of the thyristor when instructed to do so by the
control system, and under certain circumstances also as a protective action, in which case the action takes
place autonomously on that thyristor level.
The RC (damping) circuit controls the magnitude of the negative recovery (turn-OFF) voltage overshoot
and, together with the DC grading circuit, ensures uniform division of the valve voltage between the series-
connected thyristor levels. These two grading circuits cover different areas of the frequency domain: the
damping circuit covers frequencies from fundamental (f = 50/60 Hz) up to a few kHz, and the DC grading
circuit, as its name implies, deals with DC (f = 0).
A complete thyristor valve cannot be created simply
by a series connection of many thyristor levels of
the type shown in Fig. 6.2.2a. Certain additional Cd
components are needed, in particular a di/dt-limiting
reactor to control the stresses on the thyristor during
the early stages of conduction. However, the type
and arrangement of these other components depend Thy Rdc Rd
fundamentally on how the valve circuit is arranged.
There are basically two types of valve circuit – those
TCU
based on distributed di/dt reactors, and those based
on lumped di/dt reactors.
6.2.2.2 Valve Circuit with Distributed Fig. 6.2.2.a– Basic (minimal) circuit of one
di/dt Reactors thyristor level, with electrically-triggered
In this circuit, every thyristor level is an exact, thyristors
miniature replica of a complete valve. In particular,
each thyristor level has its own (small) di/dt reactor (Fig. 6.2.2b).
An advantage of this circuit is that, because each di/dt 6.2.2a
reactor is very small, the circuit lends itself to a
very simple construction method for the reactor, in which it is easy and cheap to provide a center-tap.
The center tap is very useful for connecting a fast-grading capacitor, as will be discussed in later sections.
The disadvantages are that the component count, and hence the overall cost, is high. In addition, it can
make the mechanical design of the valve more complex.
6.2.2.3. Valve Circuit with Lumped di/dt Reactors

In this circuit, certain components, chiefly (but not exclusively) the di/dt reactor, are no longer unique to
each thyristor level but are shared between several thyristor levels. The repeating unit within the valve is
then called a valve section and typically comprises between 5 and 15 thyristors connected in series, plus a
single lumped di/dt reactor. It may also have a valve section capacitor across the thyristors plus part, or all of
the reactor. A complete valve consists of several valve sections in series. Fig. 6.2.2c illustrates the principle.
The major advantage of this circuit is that it considerably simplifies the mechanical design of the valve.
In particular, all of the thyristors within the valve section can share the same clamping system. Through
economies of scale, it also allows for a more cost-effective design of the di/dt reactor.
CHAPTER
TL
Thyristor level (TL)
TCU
di/dt TL
reactor
TCU
Valve
Cd TL
TCU
Rdc Rd
TL
TCU
TCU
Fig. 6.2.2b– Valve circuit with distributed di/dt reactors
6.2.3. Valve Voltage Distribution

The voltage distribution (voltage sharing) within an off-state valve is complex and important. With perfect
voltage distribution, the number of thyristors required in a valve could be calculated very simply by just
6.2.2b
dividing the voltage rating required for the valve as a whole by the voltage rating of each thyristor.
However, the voltage distribution in a valve is never perfect and the various effects which cause unbalanced
voltage distribution need to be considered in the design. The consequence of this is that the valve always
needs more thyristors in series than would be predicted by the simple method outlined in the previous
paragraph. Since there is a clear economic incentive to use as few thyristors as possible, one of the
main objectives of the valve design is to make the voltage distribution as good as reasonably possible,
Valve section (VS) VS
di/dt reactor
TCU
Thyristor
TCU level
Valve
TCU
VS
TCU
TCU
TCU
Fig. 6.2.2c– Thyristor valve circuit with lumped di/dt reactors

CHAPTER
without incurring excessive costs for other components. The main mechanism for achieving good voltage
distribution is to fit voltage grading circuits as discussed in the next section.
Voltage imbalance errors occur as a result of a number of different effects, but they can be divided into
two main categories: ratio errors and offset errors.
6.2.3.1. Ratio Errors

In the off-state, every thyristor level in the valve can be expressed as an equivalent impedance. The
equivalent impedance arises from impedances which actually exist across that thyristor level, and
impedances from that thyristor level to ground (Z1 and Z2 in Fig. 6.2.3a).
These impedances are frequency-dependent and (except for DC) have both resistive and reactive terms.
Ratio errors relate to the extent to which the equivalent impedance of a particular thyristor level deviates
from the average of the impedances of all the thyristors in the valve. In other words:
Impedance of ‘worst’ thyristor level in the valve

Ratio error =
Average impedance of all thyristor levels Z2
Since we do not necessarily know what the impedance of the worst Z1

thyristor level in the valve is, we have to make certain assumptions
about how bad it could be. Z2
Ratio errors arise from a variety of causes such as:
➙ Leakage resistances of water pipes
➙ Parasitic (stray) capacitances between different parts of the Z2
valve and from the valve to its surroundings
Z1
➙ Tolerances of the voltage grading circuits in the valve: it is
easier to control these tolerances than the other effects. For
this reason, the impedance of the grading circuits is usually Z2
designed to be lower than those due to the other effects, Z1
so that the grading circuit tolerances dominate the overall
behavior
Z2
Ratio errors can be expressed as a percentage and are inherently Z1
linear. Thus, a 5% ratio error on a valve supporting 10 kV is 500 V and
the same ratio error on a valve supporting 100 kV is 5 kV.
For this reason, the term ‘Voltage distribution factor’ is often used Z2
to quantify the voltage imbalance in a valve. However, this approach
Fig. 6.2.3a– Impedances across
is misleading as it fails to recognize that not all sources of voltage each thyristor level (Z1) and from
imbalance are linear. those levels to ground (Z2)
6.2.3.2. Offset Errors

In addition to ratio errors, there are phenomena which result in a non-linear voltage imbalance. Such
effects cannot be expressed as a percentage and must be expressed as a voltage 6.2.3a
offset ∆V. For example if
the worst thyristor level in a valve has a voltage offset error of +500 V, then this is the voltage offset error,
irrespective of whether the valve voltage is –100 kV, +10 kV or +100 kV.
There are two main effects which can lead to voltage offset errors:
➙ Differences of leakage current between series-connected thyristors
➙ Differences of stored charge Qrr between series-connected thyristors
Both effects are described briefly in section 6.1, but the second is the most important.
When assessing the voltage imbalance within a valve, the effects of both ratio errors and offset errors must
be taken into account. By way of a simple example, consider a valve with 10 series-connected thyristor
levels. At a particular frequency, the average off-state impedance of the thyristor levels is 2 kW but one
level (the worst-case level) has an off-state impedance of 2.2 kW and a voltage offset error of +500 V.
By definition:
CHAPTER
Vvalve = 9 ⋅ Vaverage_level + Vworst_level
Where:
Vworst_level = Vaverage_level ⋅ (2.2/2.0) + 500 V
We can solve this equation to find Vworst_level and Vaverage_level for a given valve voltage.
➙ If the valve voltage is +50 kV, we find Vworst_level = 5891 V and Vaverage_level = 4901 V
Hence, the voltage distribution factor in this case would be 1.202
➙ If the valve voltage is only +10 kV, we find Vworst_level = 1535 V and Vaverage_level = 941 V
Hence, the voltage distribution factor in this case would be much higher, at 1.631
➙ If, in the same valve, the valve voltage is -50 kV, we find that both Vworst_level and Vaverage_level are exactly
equal to 5000 V
Hence, the voltage distribution factor in this case would be exactly 1.000.
These simple examples emphasize the fact that voltage distribution cannot meaningfully be quantified with
a simple, linear, voltage distribution factor.
Table 6.2.3a summarizes the main factors affecting voltage distribution in a HVDC valve under various
conditions.
Frequency range
Cause of voltage imbalance 50 Hz Switching Lightning
DC Steep front
60 Hz impulse impulse
Damping capacitor tolerance
Damping resistor tolerance
DC grading resistor tolerance
Fast-grading cap* ⋅ tolerance
Valve section cap* ⋅ tolerance
di/dt reactor tolerance
Inductance of damping circuit
Stray capacitances
Variations of water pipe length
ΔQrr - deblocked only
ΔID - blocked only
* Where fitted
= Very important effect / = Some effect / = Minor effect only
Table 6.2.3a– Factors affecting voltage distribution in an HVDC valve
6.2.3.3. Calculation of Number of Thyristor Levels in the Valve

Like most conventional high voltage electrical equipment, thyristor valves are protected from overvoltage
by zinc oxide surge arresters. The protective levels of the surge arrester define the maximum voltages that
the valve can ever experience – both in service (where the maximum voltage equals the arrester protective
level) and during type testing (where voltages some 15-20% higher than the arrester protective level have
to be applied with the arrester omitted).
The number of thyristor levels in the valve is therefore chosen to be the minimum safe number needed to
guarantee that no thyristor will be over-stressed when voltages as defined above are applied to the valve.
First however, we have to define the protective level of the valve surge arrester. Since the protective level
has a direct effect on the number of levels in (and hence the cost of) the valve, there is a strong incentive to
make the protective level as low as possible.
The protective level of the valve surge arrester is based on the peak value of the normal repetitive voltage
experienced by the valve between its terminals. The protective level is made as low as possible without
leading to a situation where the steady-state voltage across the arrester causes it to overheat. There are
two different definitions of peak voltage that have to be considered (see Fig. 6.2.3b).

CHAPTER
First is the peak of the fundamental component of voltage. In most cases, this is equal to the peak of the
valve-winding line-to-line voltage, and in the context of the valve surge arrester, it defines the Maximum
Continuous Applied Voltage (MCAV). Secondly, there are short-duration peaks that rise above this level,
due to commutation overshoots of the valve in question and other valves in the bridge. These define the
Peak Continuous Applied Voltage (PCAV) of the arrester. Because these commutation overshoot peaks
are of very short duration and come from a relatively high source impedance (which means that the surge
arrester actually clips them to a certain extent), the surge arrester is capable of tolerating a PCAV that is
higher than the MCAV.
So the valve surge arrester protective level is selected to be the lowest protective level which allows the
arrester to withstand both the MCAV and the PCAV to which the arrester is subjected in service.
Once the surge arrester protective level has been selected, the number of thyristor levels in series can be
chosen. Normally this is based on the Switching Impulse Protective Level (SIPL) of the valve arrester. The
Lightning Impulse Protective Level (LIPL) and Steep-Front Impulse Protective Level (STIPL) are less important
because although they may be higher than SIPL, they result in lower peak voltages across the thyristor,
because part of the applied voltage appears across the di/dt reactor.
During the valve type-testing, the number of levels should be high enough to avoid any risk of damage
to the thyristors during the reverse-direction switching impulse type-test, where the valve is subjected
to the Switching Impulse Withstand Level (SIWL) – normally 15% higher than the SIPL. The concern is to
ensure that the SIWL does not exceed the sum of the reverse avalanche voltages of all the thyristors in
the valve (see section 6.1).
The required number of thyristor levels is the minimum needed to withstand SIWL safely. It is then normal
to add a few more redundant thyristor levels as discussed in section 6.2.14.1 below.
6.2.4. Voltage Grading Circuits

The purpose of the voltage grading circuits is to provide a well-controlled parallel impedance across each
thyristor level so that the voltage distribution is as good as it reasonably can be in all relevant frequency
ranges. A thyristor valve may have several different grading circuits for different frequency ranges.
V1 V3 V5 V3ÊturnsÊon V4ÊturnsÊon
V4 V6 V2
MaximumÊContinuous
AppliedÊVoltageÊ(MCAV)
PeakÊContinuousÊAppliedÊ
VoltageÊ(PCAV)
Secondary
recovery
transients
Principal
recovery
transient
V4ÊturnsÊoff V5ÊturnsÊoff V6ÊturnsÊoff V1ÊturnsÊoff
Fig. 6.2.3b– Definition of MCAV and PCAV for valve surge arrester

CHAPTER
The most important grading circuit is the resistor-capacitor combination Cd – Rd. This circuit is often referred
to as the snubber circuit by analogy with the RC circuit used on some other lower-power converters, but
is better described by the term damping circuit because of its role in damping the oscillatory voltage
overshoot that occurs at thyristor turn-off. The damping circuit is very effective at (and is the main means
of) controlling voltage distribution at fundamental frequency and for moderately fast impulse waveforms.
However, it becomes less effective at very high frequencies (MHz) and is completely ineffective at zero
frequency (DC). To deal with these frequency régimes, all thyristor valves include a plain resistive DC grading
resistor and many designs also include some form of plain capacitive fast grading capacitor.
6.2.4.1. Explanation of Damping Function at Turn-off

The valve is connected to the AC system via series inductance, mainly arising from the leakage inductance
of the converter transformer. During the on-state period of the valve, the load current flows through this
series inductance and the thyristor. The thyristor provides a very low impedance path, which almost
completely prevents any voltage from appearing across the damping circuit. However, at turn-off, the
thyristor ceases to conduct current shortly after current zero, and at this time the damping circuit becomes
connected in series with the external inductance of the converter, forming a series LC circuit.
Any LC circuit subjected to a step-change of voltage will tend to oscillate. In a HVDC valve, this oscillation
results in the peak voltage across the thyristor reaching much higher values than the instantaneous value of
the fundamental frequency voltage as supplied by the converter transformer. This ‘commutation overshoot’,
in turn, contributes towards the steady-state heating of the valve surge arrester. The commutation
overshoot is exacerbated by the turn-off properties of the thyristor (Qrr and Irr – see Fig. 6.1g). With an
ideal thyristor (Qrr = 0) it would be possible to choose damping circuit components to give nearly critical
damping and achieve an overshoot of 10% or less. With real thyristors, overshoots of 50% or more are
common because of the effects of Irr and Qrr.
Unless the commutation overshoot can be damped sufficiently, the additional heating in the valve surge
arrester will result in the need to increase both the voltage rating of the arrester and the number of thyristor
levels in series in the valve. Therefore, the main function of the damping circuit is to provide damping of
the valve turn-off voltage transient (commutation overshoot – see Fig. 6.2.4a).
The spread (tolerance) of Qrr amongst the different thyristors in a valve (∆Qrr), has a very large adverse
effect on voltage sharing in the valve after turn-off, and here again the damping circuit plays an important
role. If two thyristors are connected in series, with different values of Qrr, then the thyristor with lower Qrr
will initially see a higher negative voltage than the other thyristor, but a lower positive voltage after the
voltage changes polarity (see Fig. 6.2.4b). The thyristor with the greater Qrr will experience a voltage offset
error equal to ∆V = ∆Qrr/Cd making its voltage more
I IdealÊthyristor
positive than the other thyristor.
RealÊthyristor
O
6.2.5. Damping Capacitor t
The damping capacitor (Cd) is the main mechanism

for enforcing voltage distribution between series- VÊ(thy)
connected thyristor levels, especially for the O t
fundamental component of valve voltage (50/60 Slight
Hz). With typical component values of 1.0-1.5 μF overshoot
(∼20%) Large
for C d and 40-60 W for R d, it is easy to see that
overshoot
the combined impedance of the damping circuit (50-100%)
is overwhelmingly capacitive at fundamental
frequency, and the effect of the resistor only Fig. 6.2.4a– Valve turn-off transient (commutation
overshoot) showing effect of thyristor stored charge
becomes significant at frequencies of several kHz.
6.2.5.1. Requirements
To be effective in its main role as a voltage divider, the damping 6.2.4a
capacitor must have an extremely stable
capacitance value – that is to say, the initial manufacturing tolerance must be very tight and there should
be little or no drift over its life. An overall end-of-life tolerance of ± a few percent is required.

CHAPTER
Other requirements are more concerned with
safety: in particular, it is vitally important that its LowÊQrr
failure mode is safe. The only safe failure mode is
for the capacitor to become open-circuit. When HighÊQrr
IÊ(thy)
this happens, the thyristor will very quickly be
destroyed, but because the thyristor always fails
O
to a short-circuit condition and therefore removes t
all stress from the thyristor level, this is a safe
outcome.
ΔtÊ=ÊΔV/Ê(dv/dt)
VÊ(thy)
If the damping capacitor were to fail due to a
O
short-circuit, then the impedance of the complete t
damping circuit would then just become the (low)
impedance of the resistor itself. This would result in
a very large power overload of this resistor, with the ΔVÊ=ÊΔQrr /Cd
risk that it could catch fire. It is equally obvious that Fig. 6.2.4b– Effect of variations of stored charge
the capacitor must not fail in an explosive manner. Qrr on voltage sharing
6.2.5.2. Types of Technology

In order to provide long life, high reliability and high stability, damping capacitors for HVDC valves are
normally made of metalized polypropylene film. The voltage stress on the dielectric film is chosen so that no
6.2.4b
internal partial discharges occur at the working voltage. This dielectric material can be made ‘self-healing’
against the consequences of dielectric failure during overvoltage so that small dielectric punctures naturally
isolate themselves by a fusing action and do not result in the entire capacitor becoming short-circuited.
Capacitors of a film/foil construction are dangerous and must be avoided, because they are not self-healing
and have the potential to become short-circuited.
One way of ensuring long life and high stability of a metalized polypropylene dielectric film capacitor is
to impregnate the capacitor with mineral oil; however, this creates a fire risk in the event that the oil
should leak out. This can be mitigated by insisting that the capacitor includes a bellows-type overpressure
disconnector (so that the gassing of the impregnating liquid cannot result in explosion), and by providing
an oil-containment system and fire barriers around the capacitors. These measures were taken on GE
Vernova’s (then Alstom Grid’s) H300 series HVDC valves.
On the newer HVDC valves, the oil has been eliminated altogether. However, a drawback of doing so is that
the resulting dry capacitor is physically larger than the equivalent oil-impregnated design. This is because
the dielectric has to be operated at lower voltage stress in order to realize a design that has comparably low
partial discharge performance.
6.2.5.3. Choice of Capacitance Value

The choice of capacitance value is a straightforward economic choice. Larger values of capacitor result in
better suppression of the commutation overshoot transient and better limitation of voltage imbalances
caused by variations of stored charge in the thyristors, which in turn allows fewer thyristor levels to be
used in series. However, the power losses of the damping circuit increase more than proportionally with
capacitance. This has two effects: firstly, the capitalized losses of the station increase, and secondly the
damping resistor has to be rated for higher power dissipation. These two effects, which can each be of
comparable importance, provide a powerful motivation to use the smallest possible capacitance that does
not result in excessive commutation overshoot.
With 8.5 kV, 125 mm thyristors, the economical optimum value of the damping capacitor is generally in
the range 1.0-1.5 μF per level.
6.2.6. Damping Resistor

The damping resistor performs two main functions:
➙ Damping of the turn-off voltage (recovery) overshoot of the thyristor

CHAPTER
If the values of commutating inductance (referred to one thyristor level) and damping capacitance
are known, it is a straightforward matter to calculate the value of damping resistance needed to give
critical damping, from the well-known formula:
L
Rcritical = 2 [Eqn 6.2a]
C
In practice, it is not possible to obtain critical damping in a HVDC thyristor valve, for two reasons:
1. Parasitic (stray) capacitances of equipment such as the converter transformer mean that there are
capacitances that cannot be damped.
2. The stored charge of the thyristors at turn-off causes the voltage overshoot to be much larger than
a conventional critically damped system.
It is therefore necessary to perform quite detailed simulations, including modeling accurately the
turn-off behavior of the thyristors, in order to optimize the value of damping resistor.
➙ Limitation of the step current into the thyristor at turn-on
When the thyristor switches on, the damping capacitor is already pre-charged to approximately the
same voltage as the thyristor. In the absence of a damping resistor, this capacitor would discharge
extremely quickly (< 1 μs) into the thyristor, giving a very short but extremely large pulse of current
which would destroy the thyristor.
The second function of the damping resistor is therefore to provide enough series resistance in
the circuit to limit this discharge current to a safe value. With a resistor in the circuit, the discharge
current then has a rapid (< 1 μs) rise to a plateau, or step current, from which the current decays
slowly with a time-constant of Cd ⋅ Rd (see Fig. 6.2.6a).
On the other hand, it is desirable to use as low a value of resistance as possible, since a high value of
resistance tends to increase the dv/dt experienced by the thyristor under certain abnormal operating
conditions.
Taking these requirements into account, the optimum value of damping resistor for a 125 mm, 8.5 kV
thyristor is typically of the order of 40-60 W.
6.2.6.1. Requirements +
The damping resistor is a demanding application for Cd
a power resistor, since not only is the total power
dissipation very high (several kW) but virtually all of Rd
this power comes in the form of short, high-energy TCU
pulses (Fig. 6.2.6b).
Note: the power dissipation is principally a function of the ∼1µs Vthy
V
damping capacitor value and is almost independent of the VRd
ohmic value of the damping resistor; the dissipation arising
mainly from the abrupt voltage steps which occur during the
off-state interval, as discussed later in section 6.2.11.1. Exponential decay
(time constant = Cd . Rd)
Like the damping capacitor, the resistor also
requires a close manufacturing tolerance and high O 10 µs 20 µs
t
stability. Although the damping capacitor plays the
main role in ensuring voltage grading under normal
conditions, for very fast transients the damping I
resistor becomes the dominant component and Peak current = Vthy /Rd
hence it too needs to be matched to ± a few percent
over its lifetime.
The damping circuit, comprising the damping
resistor, the damping capacitor and the associated
O 10 µs 20 µs
interconnecting wiring, must have low stray t
inductance. The typical requirement is for the L/R Fig. 6.2.6a– Role of damping circuit in limiting
discharge current at turn-on
time constant to be ≤ 0.1 μs.

CHAPTER
6.2.6a
6.2.6.2. Types of Technology
Several technologies are available for power resistors, but the two favorites for this application are (a)
thick-film and (b) wire-wound.
VÊ(Rd)
4
Ñ2
Ñ4
Ñ6
0 50 100 150 200 250 300

ωt
deg
Fig. 6.2.6b– Typical pattern of voltage pulses across a damping resistor in normal HVDC operation
Thick-film resistors allow very low inductance and facilitate good heat transfer from the resistive substrate to
the coolant, but have limited capability to handle the high-energy pulses of power inherent to this application.
Wire-wound resistors are much better at coping with the pulse energy demands, but can be more difficult
to cool and tend to be more inductive.
Cooling can be indirect (the resistor is mounted on a separate water-cooled heatsink) or direct (coolant is
passed through ducts or channels that are an integral part of the resistor). Direct-cooled resistors tend to
6.2.6b
give less short-time overload capacity (because the thermal capacity is lower) but are significantly more
compact and are therefore preferred in most modern HVDC valves.
Fig. 6.2.6c shows a design of a direct-cooled damping resistors used on recent HVDC valves.
6.2.7. DC Grading Resistor

The damping circuit is very effective at controlling the voltage distribution within the valve for all normal
operating conditions, and is reasonably effective even at the very high frequencies (MHz) involved in
steep-front impulses. However, the fact that it contains a capacitor makes it completely ineffective for any
condition when the valve is subjected to a DC voltage. This occurs during the type testing of the valve and
can occur on transmission HVDC schemes during some fault conditions.
To deal with voltage grading under these conditions, a resistive DC grading circuit is needed. This is normally
provided by fitting a resistor across each thyristor.
There are two principal effects which can lead to voltage imbalance under DC conditions which need to
be corrected by the DC grading resistor:
➙ Tolerances of off-state and reverse leakage current amongst series-connected thyristors
➙ DC grading errors resulting from the cooling pipework within the valve
With a water-cooled valve, the resistance of the water inside the coolant pipes, in particular the large
diameter inter-tier pipes, has a dramatic effect on voltage sharing under DC conditions. With a good
design, it is possible to arrange the coolant pipes so as to give equal resistances across each valve

CHAPTER
module. However, this is not what is required. What the valve needs is equal
(equivalent) resistance across each thyristor level. When all modules contain
the same numbers of thyristors, this occurs naturally, but there are several
reasons why the different modules that make up a valve may have unequal
numbers of thyristors.
The main function of the DC grading resistor is to correct these effects and
ensure that the resulting voltage distribution under DC conditions is adequate
to prevent any individual thyristor experiencing a voltage exceeding its DC
voltage rating.
A secondary function of the DC grading resistor can be to act as the top end of
a voltage divider, so that the TCU can measure the voltage across the thyristor
for control and protection purposes.
6.2.7.1. Requirements
As with the damping components, close tolerance and high stability are
important considerations. In its role of controlling DC voltage distribution
the tolerance requirements are somewhat less onerous than for the damping
capacitor but, when used as the top-end of a voltage divider, demands for
accuracy in voltage measurement usually override this.
In comparison to the damping resistor, the power dissipation rating of the DC
grading resistor is quite modest, although with high voltage thyristors it still tends
to be high enough to justify water-cooling. However, there is no pulse energy
duty, unlike the damping resistor. This allows for a variety of different resistor
technologies to be used.
Failure mode is an important consideration, since a failure of a DC grading Fig. 6.2.6c– Direct-cooled HVDC
resistor to a short-circuit condition would be extremely dangerous (it would damping resistors
short-circuit the thyristor externally, potentially creating a fire risk). It is
therefore essential that the DC grading resistor should fail only to an open-circuit condition.
6.2.7.2. Choice of Resistance Value

The choice of resistor value is an economic choice between a high value of resistance (which generates little
losses) and a low value, which is more effective at controlling the voltage distribution and hence permits
fewer thyristor levels for a given application.
Modern HVDC valves typically use values of the order of 100 kW per level.
6.2.8. High-Frequency Grading Capacitors

The DC grading circuit takes care of voltage distribution under DC conditions and the damping circuit takes
care of all normal repetitive conditions: in the frequency domain, from 50 Hz up to a few kHz, this being
the dominant frequency of a valve commutation overshoot transient. During fast-fronted voltage impulses
such as a lightning impulse or a steep-front impulse, the characteristic frequencies approach the 100’s of
kHz or even the MHz range. At such frequencies, the damping circuit becomes increasingly ineffective, the
impedance of the di/dt reactors becomes very important and the voltage distribution in the valve starts to
be dominated by stray inductances (in wiring, busbars, etc.) and capacitances (between tiers of the valve
and from the valve to earth).
For these reasons, some thyristor valves include some form of high-frequency grading circuits based on a
plain (undamped) capacitor. This capacitor is known as a fast grading capacitor when it is connected across
a single thyristor level and a valve section capacitor when it is connected across a group of thyristor levels
and associated di/dt reactor.
The main functions of the fast grading (Cfg) or valve section (Cvs) capacitors are as follows:
➙ To act as a terminating impedance at very high frequencies, helping to reduce the steepness of any

CHAPTER
high-dv/dt transients that can appear across the valve as a result of flashovers within the valve hall
(which are the origin of steep-front impulses)
➙ To buffer the thyristors from the effects of high dv/dt when a high dv/dt is applied to the complete
valve
➙ To improve the voltage distribution within the valve for high-dv/dt transients of external origin (such
as flashovers in the valve hall). This to ensure that maximum module and inter-tier voltages are not
exceeded
➙ To improve the voltage distribution within the valve for high-dv/dt transients of internal origin. The
principal concern here is cascade firing, when some or all thyristor levels in the valve fire (turn on)
by their VBO protection (see section 6.2.10.1 below).
The last of these functions can only be achieved with fast grading capacitors at individual thyristor levels
– the valve section capacitor is largely ineffective in this role.
6.2.8.1. Implementation Options

There are several options for connecting either the valve section capacitor or fast grading capacitor, as
shown in Fig. 6.2.8a.
The main advantages and disadvantages of each approach are outlined below:
When the capacitor is connected across thyristor(s) only, it is very effective at buffering the thyristors
from high-dv/dt (acting together with the di/dt reactor). The series combination of the di/dt reactor and
the capacitor forms a high-pass filter across the thyristors so that most of the high-dv/dt transient is
supported across the reactor. A fast grading capacitor connected directly across the thyristor can also be
quite effective at suppressing the high-dv/dt on that thyristor due to cascade firing.
On the other hand, the disadvantage of this option is that when the thyristor turns on, the capacitor discharges
straight into the thyristor, giving very high transient current. Therefore, with this option, the capacitor size is
limited to very small values (a few nF).
When the capacitor is connected across the thyristor(s) plus di/dt reactor, it can be very effective at acting
as a terminating impedance to reduce the total dv/dt across the valve. It is also very effective in enforcing
voltage sharing at high frequencies for externally
imposed transients. However, it provides no
buffering of the thyristors (in other words, it reduces
the total dv/dt across the valve but does not reduce Across Across Across
the proportion of that dv/dt seen by the thyristors). thyristors thyristorsÊ+ thyristorsÊ+Êpart
only di/dtÊreactor ofÊdi/dtÊreactor
It is also somewhat ineffective at limiting the dv/dt
on a late-fired level (cascade firing).
When the capacitor is connected across the
thyristor(s) plus part of the di/dt reactor, we have a
technical compromise which combines the greatest
advantages of the two preceding options. The part Fast-gradingÊcapacitorsÊ(distributedÊreactors)
of the reactor connected inside the capacitor/
thyristor loop protects the thyristor from excessive
current discharge at turn-on, while the part
outside the loop acts with the capacitor to buffer
the thyristor from high-dv/dt. This design works
particularly well when the outer half of the reactor
is ‘lossy’ (to provide damping) while the inner half
of the reactor is low-loss, so that the step-current
into the thyristors is minimized.
Valve-sectionÊcapacitorsÊ(lumpedÊreactors)
The disadvantage of this circuit is the extra
complexity introduced into the reactor design due
Fig. 6.2.8a– Options for connecting fast-grading
to the need for a center-tap. or valve section capacitors

CHAPTER
6.2.8a
6.2.8.2. General Requirements
Most of the requirements applicable to the damping capacitor (close tolerance, stability, safe failure mode,
etc.) apply equally to this component. However, an additional requirement is that the capacitor must have
very low inductance in order to function correctly, although it should be noted that the total inductance of
the circuit tends to be dominated by the inductance of the wiring, not of the capacitor itself.
6.2.9. The di/dt Reactor

As discussed in section 6.1, the thyristor is very vulnerable in the first few μs after turn-on, and requires
the rate of rise of current to be limited to a few hundred amps per μs. This is accomplished by connecting
a di/dt-limiting reactor in series with the thyristor.
In order not to increase the reactive power consumption of the converter, it is undesirable (and too expensive)
to use a linear reactor, so the magnetic material of the reactor is deliberately allowed to saturate each cycle.
Hence, the reactor is frequently called a saturating reactor.
As mentioned in section 6.2.8 the reactor also plays a useful role in limiting the rate of rise of voltage
(dv/dt) experienced by the thyristors in the off-state, as a result of certain fault events.
The design of the reactor is complex, because of compromises that have to be made between its ability to
limit di/dt in the very early stages of turn-on (which favor using low-loss magnetic material) and the need to
provide damping of the turn-on transient (for which a certain, known amount of core loss is desirable).
After the thyristor, the di/dt reactor is probably the most complex and important component in the valve, with
characteristics which influence almost every other aspect of the valve design.
6.2.9.1. Function of Reactor

The reactor performs many useful functions in the valve, but the two principal ones can be summarized as:
➙ Reduction of turn-on di/dt stresses (hence the name)
➙ Reduction of fast transient voltage stresses applied to the thyristor
• Reduction of Turn-on di/dt Stresses
Although modern thyristors have an active silicon diameter of up to 150 mm or even larger, the complete
thyristor does not turn on simultaneously. In the first few μs after turn-on, only a very small area, within a
millimeter or so of the edge of the gate terminal, initially conducts. Therefore, if the current rises too rapidly
after turn-on, this localized area may overheat and the thyristor will be destroyed. For this reason, the rate
of rise of current (di/dt) in the thyristor must be limited during these first few μs.
The main source of current at this time is from the unavoidable stray capacitance of the other converter
equipment, mainly the converter transformer. If not limited by a di/dt reactor, the discharge of these stray
capacitances when the thyristor turns on would result in a theoretically infinite current in the thyristor, which
the thyristor would not withstand.
The main function of the di/dt reactor is, therefore, to slow down the discharge of this stray capacitance,
so that it occurs sufficiently slowly for the thyristor to be able to withstand it safely (see Figs. 6.2.9a and
6.2.9b).
In addition to limiting the rate of rise of this initial current, and thereby the peak current, the reactor must
also provide damping to prevent the thyristor current from becoming too oscillatory. In particular, it is vital
to ensure that the current in the thyristor does not drop below zero after turn-on, otherwise the thyristor
could be destroyed.
When turn-on takes place from such a high voltage that the valve surge arrester is conducting, the inrush
current becomes considerably more severe. The current that was flowing in the valve surge arrester
commutates almost immediately to the thyristors and adds to the discharge current from the stray
capacitances. This event (commutation of arrester current) results in the worst single-shot inrush
current for the thyristors.
• Reduction of Fast Voltage Transient Stresses

CHAPTER
The reactor is in series with the thyristors and their associated damping circuits. The combination of reactor
and damping circuits forms a low pass filter which will attenuate fast voltage transients as shown in Fig.
6.2.9c, reducing both the amplitude and the steepness (dv/dt) of the voltage experienced by the thyristors.
This gives two important benefits to the valve design.
di-dtÊreactor ReactorÊmustÊ
ACÊsystem Stray Oscillatory
Thyristor provideÊenough
+Êconverter capacitance dischargeÊdueÊto dampingÊto
current
transformer Thyristor(s) externalÊstray preventÊthyristor
+
+ capacitance Damping currentÊfrom
RC circuitÊcurrent
reachingÊzero
damping ≈(V/R)eÊ(-t/RC)
OK
(snubber)
circuit time
Dangerous!
Fig. 6.2.9a– Discharge of external stray capacitance when valve turns on Fig. 6.2.9b– Effect of di/dt reactor on valve
current in first 10-20 μs
Firstly, the peak voltage reached is substantially reduced. This means that only the Switching Impulse
Protective Level (SIPL) of the valve arrester is relevant for calculating the minimum number of series
6.2.9b
connected thyristors required in a valve. Although the LIPL and STIPL are higher than the SIPL, the peak
voltages experienced by the thyristors are lower because of the buffering effect provided by the reactor.
6.2.9a Secondly, the dv/dt of the resulting voltage transient, as seen by the thyristors, is reduced. The
principal protection against such an event is for the TCU to include a dv/dt-sensitive protective firing
system which preemptively turns on the thyristor via its gate (which is safe), instead of allowing
it to turn on spontaneously. However, if the valve relied only on dv/dt protective firing (and the
di/dt reactor gave no buffering effect), spurious dv/dt firing could become a nuisance under some conditions.
The buffering effect provided by the di/dt reactor allows the dv/dt protective firing threshold to be set at a
level which does not interfere with the normal operation of the converter.
6.2.9.2. Desired Characteristics

To fulfill its primary requirement of limiting di/dt at turn-on, the reactor needs to have a high inductance
at low values of current (a few hundred amps). However, in order to avoid unnecessarily increasing the
reactive power consumption of the converter, it is undesirable for this inductance to be present at normal
load current. Hence, the di/dt reactor is nearly always of the saturating type, where the magnetic material
is deliberately allowed to enter saturation every time the valve is turned on. This also allows the reactor to
be much smaller than a linear reactor of the same unsaturated inductance.
The saturation current is normally a few hundred Amps, which is high enough to deal with the initial inrush
from stray capacitances when the valve turns on (and to act as a buffer for fast voltage transients). The
transition from unsaturated to saturated must be progressive and not exhibit an abrupt ‘knee’, as this can
result in excessively fast voltage collapse between the valve terminals, with a consequential increase in
electromagnetic radiation and dv/dt on the valve connected in series with the valve that is turning on.
The conduction losses (I2R in the primary) and hysteresis losses should be as low as possible, since these
give no benefit to the valve design. However, it may be advantageous for the core material to have some
eddy-current losses, since this is one (quite effective) method of providing damping of the turn-on current.
Many methods are possible for providing damping of the turn-on transient as discussed in section 6.2.9.4
below. However, most of these methods conflict with another very important design requirement for the
reactor - that it must give a low value of step-current when the thyristors turn on. The step-current from
the di/dt reactor adds to that from the damping circuit (see Fig. 6.2.9) and the sum of the two must be less

CHAPTER
Steep-front
impulseÊ(0.4/4Êµs)
di/dtÊreactor
FastÊ
transient
voltage
source RC
VÊ(thy) damping
circuits
0 1Êµs 2Êµs 3Êµs t
TotalÊappliedÊvoltage
VoltageÊacrossÊthyristors
Fig. 6.2.9c– Beneficial effect of di/dt reactor for suppressing dv/dt on thyristor
than the maximum step current that the thyristor can tolerate.
It is also important for the reactor to have a low value of equivalent parallel capacitance, as this capacitance
reduces the effectiveness of the reactor in suppressing dv/dt.
Lastly, the reactor must be designed to withstand short-circuit currents at least as great as those for
6.2.9c
which the thyristors are rated. The main consideration here is not the heating effect of the current but the
electromagnetic forces it generates.
Overall, the di/dt reactor is a difficult component to optimize, as there are many conflicting parameters.
6.2.9.3. Types of Construction

Many different types of construction have been proposed for the di/dt reactor, but they can be subdivided
into two main classes:
➙ Single-turn primary, where the primary conductor of the reactor is normally just a straight length of
busbar, with magnetic cores fitted over the outside of it
➙ Multi-turn primary, where the reactor resembles, to some extent, a classic wound component
In both cases, the primary conductor often includes a water cooling channel which is used for cooling of
the primary. Additional measures, for example extra heatsinks, are sometimes also included for cooling
the cores.
• Single-turn Primary
The single-turn primary has the great advantage that it permits a very simple construction, with minimal
electrical insulation requirements between the core and primary.
This in turn makes it easy to extract the heat from the cores. This type of reactor lends itself very well to
the distributed valve circuit, which requires large numbers of small di/dt reactors (see Fig. 6.2.9e).
• Multi-turn Primary
The main drawback of the single-turn reactor is that it does not use the magnetic material particularly
efficiently. In terms of the total mass of electrical steel required per valve, a more efficient design can be
achieved with the multi-turn primary approach. The ability to optimize the number of primary turns gives an
additional degree of freedom in the design and makes it easier to trade-off the unsaturated inductance versus
the saturation current. It generally results in a lighter and more compact reactor than would be possible with
the single-turn reactor. The multi-turn approach particularly favors the lumped main circuit.
Two variants of the multi-turn primary are possible. The first is relatively similar to a conventional
transformer or reactor, where the primary winding is effectively fitted around the magnetic core. This

CHAPTER
is illustrated on Fig. 6.2.9f below, with the inter-turn cast resin insulation omitted for clarity (left) and
included (right).
In the second variant, the reactor is effectively turned inside out and the cores are fitted around the
winding. This is illustrated in Fig. 6.2.9g below, with the inter-turn cast resin insulation omitted for clarity
(left) and included (right).
Thermal
compound CoolingÊduct
Primary I
FerromagneticÊcore conductor
Fig. 6.2.9d– Cross-section through single-turn reactor
The main disadvantage of the multi-turn reactor is that it requires substantial electrical insulation between
turns of the primary winding and from the winding to the cores. Hence, the construction is more complex
than for a single-turn reactor.
Fig. 6.2.9h shows a multi-turn reactor using this second approach, for the H450 valve design, in this
case using 6.2.9d
six, 8.5 kV thyristors per valve section. The reactor shown is equivalent to more than
10 of the reactors shown in Fig. 6.2.9d.
Fig. 6.2.9e– Single-turn di/dt reactor in H300 HVDC valve

CHAPTER
6.2.9.4. Damping of Turn-on Transient
In addition to limiting the rate of change of current, the di/dt reactor is required to damp the resulting current
oscillation and act to prevent, or minimize, current reversal in the main thyristors during turn-on.
Damping of the turn-on current oscillation can be achieved in a number of ways (see Fig. 6.2.9i), but
common to all approaches is the need for some form of damping resistor connected in parallel with
the inductance. Some designs utilize the naturally-occurring eddy-current and hysteresis losses of the
B B
I I
Cast resin
insulation
Ferromagnetic core Primary Ferromagnetic core Primary
conductor conductor
Fig. 6.2.9f– Multi-turn primary winding with winding fitted around cores
magnetic core(s) of the inductor to provide the required damping (Fig. 6.2.9i (a)). Alternatively, the resistor
may be a discrete, physical resistor connected externally to the inductor (Fig. 6.2.9i (b-e)). The resistor
may be directly connected to the inductor (Fig. 6.2.9i (b and c)) or transformer-coupled (Fig. 6.2.9i (d and
e)). In some cases the resistor may have a diode connected in series with it (Fig. 6.2.9i (c and e)) which
disconnects the damping resistor during the rising part of the current waveform (so that the resistor does
6.2.9f
not allow a damaging step-current into the thyristor at turn-on) and only provides damping on the falling
part of the current waveform.
Each of the options in Fig. 6.2.9i is possible and has both advantages and disadvantages. In the interests
of simplicity, the H300 and H400 valve series used the first option (using the naturally occurring damping
due to eddy current losses in the core material). It has the big advantage that it does not require additional
equipment which would need to be supported, electrically insulated and cooled.
Primary
FerromagneticÊcore conductor
B
I
CastÊresin
CoolingÊduct insulation
Fig. 6.2.9g– Multi-turn primary winding with cores fitted around winding

CHAPTER
6.2.9g
Fig. 6.2.9h– Multi-turn reactor in GE’s H450 HVDC valve
6.2.10. Valve Electronics and Associated Components

In addition to a thyristor and various voltage grading circuitry, a thyristor level also requires equipment
which is grouped under the heading of valve electronics. The main item of valve electronics is the TCU,
which communicates with the converter controls at ground level via optical fibers.
6.2.10.1. Thyristor Control Unit (TCU)

The Thyristor Control Unit (TCU), also known as the gate unit, basically has two essential functions:
➙ Turning the thyristor ON when instructed to do so by the control system at ground level (in this
respect, the TCU could be considered as merely an optical to electrical converter, although to do so
greatly oversimplifies its role)
➙ Monitoring the conditions at the thyristor level: in particular the state of health of the thyristor itself,
and reporting back to the controls
As well as being turned on by normal control action, the thyristor sometimes needs to be turned on as a
protective action. In principle, it would be possible for the protective turn-on action to be taken centrally
by the control system at ground level. In this case, the TCU could be completely ‘dumb’ and would only
require the two basic functions outlined above.
(a) (b) (c)Ê (d)Ê (e)

DampingÊresistanceÊusingÊ DampingÊresistanceÊwithÊ DampingÊresistanceÊwithÊ DampingÊresistanceÊwithÊ DampingÊresistanceÊwithÊ
lossesÊofÊcoreÊmaterial externalÊresistor diode-coupledÊexternalÊ transformer-coupled transformer-ÊandÊdiode-
resistor externalÊresistor coupledÊexternalÊresistor
CouplingÊdiode CouplingÊdiode
EquivalentÊparallel External External External External
resistanceÊofÊcores damping damping damping damping
resistor resistor resistor resistor
Thyristor Thyristor Thyristor Thyristor Thyristor

level(s) level(s) level(s) level(s) level(s)
Fig. 6.2.9i– Options for providing damping in di/dt reactor

CHAPTER
In practice, the delay time inherent in transmitting the necessary data to the controls followed by the
controls acting on this data and then transmitting a protective triggering command back to the relevant
TCU(s), makes it impractical to consider such an approach in a real HVDC valve. For this reason, most HVDC
valves delegate these protective actions to the TCU, which then performs them locally and autonomously.
Hence, there is a third essential function for a practical HVDC TCU:
➙ Turning the thyristor ON via autonomous protective action, when the local conditions demand it
This protective function demands that the TCU be ‘intelligent’ and is responsible for an appreciable
proportion of the total complexity of the TCU.
• Protection Requirements
Thyristors are quite delicate and can easily be destroyed in a number of ways. For some conditions, such
as excessive di/dt at turn-on or excessive reverse voltage, the only defense is to use passive external
components such as series-connected inductors or parallel-connected surge-arresters. However, there is
a range of conditions for which it is possible to provide a form of active protection by preemptively turning
on the thyristor. In all of these cases, the idea is to turn the thyristor on deliberately, via its gate, before it
spontaneously turns itself on. Spontaneous turn-on of a thyristor is often destructive unless the thyristor is
specifically designed to withstand this condition.
There are essentially three circumstances where pre-emptive turn-on of the thyristor can be valuable to
prevent thyristor failure:
➙ Excessive positive voltage
➙ Excessive positive dv/dt
➙ Positive voltage re-applied too soon after thyristor turn-off (forward recovery)
All of these depend on the thyristor Junction Temperature, Tj. As outlined in section 6.1, the thyristor
positive voltage rating decreases quite rapidly with increasing temperature above about 125°C. The dv/dt
capability also decreases steeply with increasing temperature, and the thyristor turn-off time increases
with temperature and re-applied dv/dt. The result is that these three protective functions are complex
and inter-linked.
The approach used by GE Vernova for many years has been for the controls at ground potential to calculate
the thyristor junction temperature in real time (from measurements of coolant temperature and valve
current) and transmit this information to all the TCUs in the valve. The TCUs then know how hot their local
thyristor is likely to be, and can adjust the protection thresholds accordingly. This allows the converter to
make maximum use of its inherent capabilities when the temperature is low.
In addition to knowing the thyristor junction temperature, the protective circuits also need a highly
accurate and high bandwidth measurement of voltage and dv/dt across the thyristor. The voltage
measurement is usually made with a voltage divider across the thyristor (normally using the DC
grading resistor as the top-end of the divider). Various techniques can be used for measuring the
dv/dt across the thyristor, and are discussed in section 6.2.10.2.
Because the protection characteristics are complicated (and vary between thyristor types) they are usually
encoded by using some form of programmable logic device such as a Field Programmable Gate Array
(FPGA), so as to give greater flexibility in adapting the protection characteristics to suit different makes
of thyristor.
• Output Stage
The output stage is the business end of the TCU and is the part of the circuit responsible for providing the
gate pulse to the thyristor.
To turn a thyristor on effectively, a current pulse of a few amps is required. Usually, the current pulse takes
the form shown in Fig. 6.2.10a. This comprises an initial peak of the order of 6 A, followed by a Back Porch
current of about 1 A, which is maintained for some tens of microseconds.
For the initial current peak, the main requirement is for the peak to be high and, even more important, for
the rise time to be as short as possible. This ensures a very rapid turn-on of the thyristor, giving the thyristor
the maximum capability to withstand di/dt from the main circuit.
The Back Porch current is mainly provided to allow enough time for the main circuit current to build up
to a value that is greater than the latching current of the thyristor. If the valve is fired at very low voltage,

CHAPTER
the di/dt of the main circuit current will also be very low. If the gate pulse does not include a Back Porch
current, the thyristor might extinguish immediately after the initial peak of gate current. The Back Porch
current also provides a measure of protection in the event that the main circuit current is highly oscillatory
and causes the thyristor current to drop below zero
shortly after turn-on.
6AÊnom
For a given gate current, the performance of the TCU
is better if the output stage is strong (that is to say,
it approaches an ideal current source). To achieve
this ideal, the gate pulse should be derived from a τ =Ê2Êµs
relatively high voltage supply.
1.0AÊ±Ê10%
• TCU Power Supply
0A
The TCU consumes a small amount of power (of 40Êµs+/-10%
the order of a few Watts), which has to be supplied
from somewhere. Because HVDC valves operate at Fig. 6.2.10a– Typical thyristor gate current
very high voltages with respect to ground potential,
there is no practical prospect of supplying this
power from ground potential via an isolating transformer. Therefore, the power has to be derived locally
by harvesting it from the other components in the thyristor level.
In theory at least, there are a number of possible methods of obtaining the required power from the other
thyristor level components:
6.2.10a
➙ From the displacement current (Cd ⋅ dV/dt) flowing in the main RC damping circuit
➙ From the displacement current (Cfg ⋅ dV/dt) flowing in the fast grading or valve section capacitors,
where these are used
➙ From the current (V/Rdc) flowing in the DC grading resistor
➙ From a Current Transformer (CT) in the main current circuit
➙ Via a secondary winding on the di/dt reactor
In practice most valves use the first method, because even when the valve is blocked there is always power
available, and when the valve is deblocked there is considerably more than is required. There are several
ways of obtaining power from the damping circuit, but in essence, they all involve diverting part or all of
the current that flows in the damping circuit into a regulating converter in the TCU.
During the positive-going parts of the valve voltage waveform, the displacement current Cd ⋅ dv/dt that
flows in the damping circuit is diverted into the TCU power supply. Initially most of it flows into a fast
charge capacitor with enough charge to allow for one or two emergency triggerings of the thyristor. The
main storage capacitor takes much longer to charge (typically tens of seconds), but thereafter holds its
charge for several seconds even if the input voltage is zero (which is important for providing fault ride-
through capability).
The main power supply storage capacitors are connected to the supply rail (typically 50 V) used for the
output stage. The lower voltage supplies (such as 5 V) needed by the logic and communications functions
of the TCU are derived by switching or linear regulators.
6.2.10.2. Voltage and dv/dt Measurement

The protections described in the preceding section require a very accurate and fast measurement of
thyristor voltage and (directly or indirectly) dv/dt.
The usual method for measuring thyristor voltage is to use the DC grading resistor as the top-end of
a voltage divider. This is not as straightforward as it sounds. Careful attention needs to be paid to the
frequency response so that the correct response can be obtained for the fastest protection functions
(where the voltage can be rising at 5 kV/μs or faster).
On the H400 valve shown above, the DC grading resistors are mounted on an aluminum water-cooled
heatsink.

CHAPTER
InputÊcapacitor
Multi-wayÊconnector
toÊvoltageÊdivider
HeatsinkÊforÊmain
semiconductor Fast-chargeÊcapacitor
components
MainÊstorageÊcapacitor
Optical InterferenceÊshieldÊover
components FPGAÊandÊother
sensitiveÊcomponents
Fig. 6.2.10b– H400 valve TCU
Since the resistors have a relatively high capacitance to the heatsink, a capacitor is connected in parallel
with the DC grading resistor, of a value large enough to dominate over these parasitic capacitance
effects. The high voltage capacitor and DC grading resistor together form a precision, high-bandwidth
voltage divider. The divider generates high-precision outputs representing voltage and dv/dt to the TCU
(Fig. 6.2.10c).
6.2.10b
6.2.10.3. Valve-VBE Communications

The high voltage at which the TCUs operate, with respect to ground, means that to date, the only practical
method for communicating between the TCU and VBE has been via an optical link consisting of a light
source (such as an LED), an optical fiber and an opto-receiver.
At its simplest, the communication protocol could consist of just a simple ‘light on = thyristor on’ signal on the
up channel (firing) and an equally simple ‘light on = thyristor supporting voltage’ signal on the down channel
(databack). However, such a system is prone to interference.
Therefore, some form of data encoding is generally used, both to improve the interference immunity and to
allow additional data to be transmitted. On the firing channel (VBE to valve), data can include the calculated
thyristor junction temperature. On the databack channel (valve to VBE) a number of status indications can
be included, for example, whether the thyristor is conducting, supporting voltage or triggering repetitively via
its VBO protection. Another benefit of using such an encoding system is that the duty cycle on the optical
components is much reduced, helping to increase their operating life.
There are two basic types of technology suitable for the optical fiber. Plastic (Poly Methyl Methacrylate: PMMA)
optical fibers using a visible red light (wavelength = 660 nm) are cheaper but have high optical attenuation.
Since HVDC valves are physically quite large and the fibers therefore quite long, the high attenuation can be
a significant problem.
An alternative is to use near-infrared radiation together with hard-clad silica glass fibers. This system has
much lower attenuation, so that there is essentially no practical limit to the fiber length.

CHAPTER
Voltage divider
Top-end Cd2 Cd1

capacitor
Rdc
Thy
Rd2 Rd1
dV/dt VAK
TCU
Firing &
databack
fibres Fig. 6.2.10c– Simplified circuit of thyristor level for H400 and H450 valves,
showing voltage divider
With Electrically Triggered Thyristors, relatively low-cost LEDs are suitable as the transmitting device, and
lasers are not required (although they were used on English Electric’s first thyristor valve in 1971). However,
this situation changes with LTTs.
6.2.10.4. Valve Electronics for Light Triggered Thyristors

6.2.10c
Because of the considerable complexity of the valve electronics for an HVDC valve, it has long been a goal
to eliminate it and use thyristors that are directly triggered optically.
The TCU does much more than simply turn the thyristor on when instructed to do so by the controls.
The protective turn-on function provided by the TCU can only be eliminated if the LTT can provide these
protective features itself. Producing fully self-protected LTTs has only become economically possible in the
last few years and they remain expensive.
Even with a fully self-protecting LTT, it would still not be possible to eliminate the valve electronics completely,
as there would still need to be a simple monitoring/interlock circuit. This is required for two reasons:
➙ To provide the databack function, indicating whether or not the thyristor is healthy (capable of
supporting voltage)
➙ To provide an interlock function, to prevent the controls from issuing an optical turn-on pulse when
the thyristor voltage is less than some critical minimum value such as +100 V (triggering an LTT at a
voltage below this is can lead to its failure).
A final complication associated with the use of LTTs is that the opto-transmitter in VBE needs to be
considerably more powerful than is the case for an ETT. With ETTs, the opto-transmitter is only required
to transmit information, for which an optical power of some tens of μW. Hence, it is possible to use a
relatively standard LED.
With LTTs, the opto-transmitter actually needs to transmit power - enough to generate electron-hole pairs
in the silicon of the thyristor. This requires several orders of magnitude more power, and optical powers
are generally quoted in tens of milliwatts, with the result that lasers, rather than LEDs, are needed in the
transmitter cards in VBE. This adds to the cost associated with using LTTs, although of course these costs
are offset by the costs saved on gate electronics.
The use of lasers also poses certain personnel safety risks as the laser radiation can result in eyesight damage.

CHAPTER
6.2.11. Thermal Design
HVDC converters are an extremely efficient method of power conversion, achieving a total efficiency of
better than 99% per station. However, even if the loss per station is only a fraction of 1% (typically 0.7%),
with a typical HVDC project rated at 3000 MW, 0.7% of 3000 MW is 21 MW of heat to get rid of – enough
to heat all the homes in a small town!). This waste heat has to be safely removed; otherwise critical
components will over-heat and will be damaged.
Power losses also represent lost revenue for most HVDC purchasers, so it is commonplace for purchasers
of HVDC systems to attach a financial weighting factor to the power losses, which is then added to the bid
price. At the project stage, financial penalties may be levied on the manufacturer if values guaranteed at the
tender stage are exceeded. Loss evaluation rates are typically $4,000-6,000/kW, which represents a strong
incentive to design for low losses.
6.2.11.1. Sources of Heat in a Valve

In the thyristor valve, the largest source of losses is normally the conduction loss in the thyristors. This arises
from the on-state voltage drop VT of each thyristor, which is normally in the region of 2 V at full-load current.
During the on-state interval, losses equal to I ⋅ VT are generated: for example if VT = 2 V and I is 3000 A, the
instantaneous power during the on-state interval is 6 kW. However, in HVDC, the duty ratio is approximately
1/3, so the average power (which is more important) is 2 kW. Other significant sources of losses include those
attributable to the stored charge of the thyristors at turn-off, and the ½ C ⋅ (ΔV)2 that is dissipated in the
damping resistors every time a valve in the converter turns on or off.
A full treatment of the sources of valve losses is given in IEC 61803 [4], but briefly – the valve losses are divided
into eight fundamentally different mechanisms, referred to as PV1 to PV8 respectively.
• P
V1: Thyristor Conduction Losses
Normally, by far the largest component of valve losses: this is simply the product of the current flowing
through the thyristor and the on-state volt drop VT of that thyristor.
• PV2: Thyristor Spreading Losses
PV2is the small additional loss (of the order of 1 J per cycle per thyristor) which arises from the
spreading time of the thyristor. During this time, the thyristor current is forced to flow through a smaller
area of silicon than it should, and hence the equation for on-state voltage drop used for PV1 is too
optimistic.
• PV3: Other Valve Conduction Losses
This small component of loss is simply the resistive (I2R) losses in current-carrying parts of the valve
other than the thyristor, occurring while the thyristors are conducting. Generally, this means busbars
that link together different sections of the valve, along with the primary conductor of the di/dt reactor.
• PV4: DC Voltage Dependent Losses
This small component of loss is the resistive loss that occurs when the thyristors are OFF, as a result of
the leakage current that flows through any shunt resistances in parallel with the thyristor. In practical
terms, the shunt resistances that contribute to this loss are (a) the DC grading resistors at each level
and (b) the water in the cooling pipes – mainly those that run from one tier of the valve to the next.
• PV5: Damping Resistor Losses (Resistor Dependent Term)
PV5 represents the losses in the damping resistors during the smoothly changing parts of the valve
voltage waveform (excluding the abrupt steps). This current is given by I = Cd ⋅ dV(t)/dt and results in
power losses given by I2 ⋅ Rd.
• PV6: Damping Resistor Losses (Change of Capacitor Energy Term)
PV6 represents the losses in the damping resistors due to the abrupt steps of voltage that take place
during the off-state interval. There are two large steps, caused by the turn-off and turn-on of the
valve in question, and at least six smaller ones caused by the turn-off and turn-on of other valves in
the bridge. Each time a step change of voltage occurs, energy equal to ½ Cd⋅ (ΔV)2 is dissipated in the
damping resistor. The average power dissipation due to this mechanism is then obtained simply by
summing the separate contributions of each voltage jump, during one AC cycle, and multiplying the
result by the frequency.

CHAPTER
• P
V7: Thyristor Turn-off Losses
PV7 represents the power losses arising from the stored charge Qrr of the thyristor. Each time the valve
turns off, energy is lost from the circuit, equal to the product of Qrr and voltage. The power loss is then
simply given by the energy per cycle, multiplied by the frequency.
V8: Reactor Hysteresis Losses
• P
PV8 represents the power losses due to hysteresis in the magnetic cores of the di/dt reactor. Hysteresis
losses per cycle depend on the amplitude of the magnetizing force H (and hence current) in both the
positive and negative directions. For HVDC valves, the assumption is that the current in the reactor is
sufficient to drive the core material fully into saturation once in the positive direction and once in the
negative direction (because of Irr) per cycle. Core losses due to eddy currents are not counted in PV8,
being considered part of PV6.
6.2.11.2. Typical Breakdown by PV1 - PV8 Category

Fig. 6.2.11a below shows a typical breakdown of valve losses by the categories PV1 - PV8. It should be pointed
out that this is for operation at full load current (for example 3000 Adc) and at relatively low control angles
(15-20˚).
The picture can be very different for low-power operation, especially if increased control angles are involved.
In such applications, the overall valve losses will be lower and the thyristor losses will be much lower, but the
damping losses will be higher.
6.2.11.3. Breakdown of Valve Losses by Component

In most cases, the component which causes the power losses is the component which suffers the
consequence (gets hot as a result). However, there are two exceptions: PV6 and PV7. Part of PV6 (due to
external, undamped stray capacitances) appears as eddy current losses in the di/dt reactor and turn-on
switching losses in the thyristor, and part of PV7 (the part due to the reverse current flow before Irr) appears
in the damping resistors and di/dt reactor. Table 6.2.11a summarizes where each of the loss mechanisms
PV1 to PV8 are dissipated.
10%
15% PV1
PV6
PV7
60%
15% Others
Fig. 6.2.11a– Typical breakdown of valve losses at full load power, by category

CHAPTER
Abbreviation Name Where dissipated
PV1 Thyristor conduction losses Thyristor
PV2 Thyristor spreading losses Thyristor
PV3 Other conduction losses In the busbars and di/dt reactor primary conductors which cause them.
In the DC grading resistors, water pipes and other shunt resistors

PV4 DC voltage-dependent losses
which cause them
Damping resistors (where there are two branches,

Damping resistor losses
PV5 the losses divide between them in the ratio of C2R).
(resistor-dependent term)
A tiny proportion also ends up in the TCU.
– Partly due to damping circuits: in the damping resistors.

Damping resistor losses
PV6 –P artly due to undamped capacitances: mainly in the damping resistors, but
(change of capacitor energy term)
some in the di/dt reactor cores and the thyristor.
– 55-65% (corresponding to the Qrr which occurs after peak reverse current) in
the thyristor.
PV7 Thyristor turn-off losses
– Remainder mainly in the damping resistors but with some in the di/dt reactor
cores.
PV8 Reactor hysteresis Losses di/dt reactor (cores)
Table 6.2.11a– Division of losses between components of the valve

Fig. 6.2.11b shows a typical breakdown of losses in a HVDC valve by component. It is based on the same
data as Fig. 6.2.11a (for full-load operation at low control angles). As with Fig. 6.2.11a, the picture is different
at reduced power.
From the above, it is clear that there are three main components which require liquid-cooling: the thyristor,
the damping resistor and the di/dt reactor. The DC grading resistor may also require liquid cooling.
6.2.11.4. Heatsink Design

A typical thyristor for HVDC can dissipate several kW of heat. This must be removed safely without allowing
an excessive temperature rise in the thyristor. The important parameter is the junction temperature, Tj
of the thyristor. For HVDC applications this normally needs to be limited to no greater than 90°C in order
to provide headroom for fault events (see section 6.2.11.6). In hot climates where the outdoor ambient
temperatures can be over 50°C, this is a big challenge.
7%
8%
Thyristor
15%
DampingÊresistor
di/dtÊreactor
70%
Others
Fig. 6.2.11b– Typical breakdown of valve losses at full load power, by component

CHAPTER
The valve design needs an efficient heatsink to be clamped on either side of the thyristor. The heatsink must
achieve the lowest possible thermal impedance (from the junction to the coolant) for a given coolant flow
rate, while at the same time not requiring an excessively high differential pressure.
The design of the thyristor heatsink for a HVDC valve is therefore quite sophisticated. Fig. 6.2.11c shows
two examples from HVDC valves. On the left, a counter-flow spiral arrangement welded to a central
aluminum body channels water over the entire contact face with the thyristor. On the right, a stack of
punched aluminum plates are bonded together. In both cases, the objective is to create a very large wetted
surface area so as to minimize the thermal resistance from the contact face to the water.
6.2.11.5. Cooling Circuit Design

Almost all modern thyristor valves use either pure de-ionized water or a mixture of de-ionized water and
ethylene glycol for cooling the valve. The use of water/glycol mixtures for cooling the valve, pioneered by
GE Vernova in the late 1980s with the H300 valve [5], allows the use of a single-circuit cooling plant, even
in climates where the ambient temperature can drop well below freezing.
StackedÊgrills
andÊcoverÊplate
Fig. 6.2.11c– Thyristor heatsink designs from H300 and H400 HVDC valves
Although it was not immediately accepted by the industry, water-cooling of high voltage equipment is perfectly
safe as long as the water is ultra-pure. This is because very pure water is a good electrical insulator and only
6.2.11celectricity to any appreciable extent as a result of ionic contaminants (salts etc.).
conducts
A separate coolant circulating and purifying system provides the valve with very low-conductivity coolant.
The waste heat is rejected in outdoor coolers (see section 6.3).
Water-cooling is always provided for the thyristors and damping resistors, and usually also for the di/dt reactor
and DC grading resistor. These components are at a multitude of different electrical potentials and hence the
distribution of the water coolant within the valve needs to be done with insulating plastic pipes.
The water-cooling circuit needs to be designed with care, in order to ensure that the system provides adequate
flow rates in all critical areas, avoiding excessively high flow rates that could cause erosion, or low flow-rates
that lead to accumulation of gas pockets.
The most important objective of the cooling system is to minimize the value of thermal impedance from
the thyristor junction to the coolant liquid. In this way, the thyristor junction temperature is minimized for
a given power dissipation. However, although minimizing thyristor junction temperature is important, it is
not sufficient. It is also important to ensure that all thyristors within a valve have similar values of junction
temperature. This is because thyristors have many temperature-dependent characteristics (not least, Qrr)
and matched temperatures mean more closely matched characteristics, which is an important requirement
when many thyristors are operated in series.
Ultra-pure water has very low conductivity, but the conductivity is never zero because the H2O molecules
dissociate into H+ and OH– ions at a rate dependent on temperature. There is therefore a small residual
conductivity, of the order of 0.05 μS/cm, which cannot be eliminated.

CHAPTER
Even such a small conductivity can cause electrochemical problems such as corrosion and gas generation.
Inevitably, some of the water is electrolyzed into hydrogen and oxygen (H2 and O2), and whilst the quantities
involved are small, the cooling system must be designed to sweep these gases out and remove them safely
at a low-pressure point in the system.
Electrochemical corrosion is a serious problem if it is allowed to occur, but it can be avoided by the right
choice of materials in the cooling circuit. In all of GE Vernova’s water-cooled valve designs to date, only the
following materials are permitted to be in contact with the coolant:
➙ AISI 316 stainless steel ➙ PVDF
➙ Aluminum (copper content < 0.1%) ➙ PEX (cross-linked polyethylene)
➙ ABS ➙ Polypropylene
➙ PTFE ➙ Neoprene/nitrile/EPDM
No other materials are permitted.
It is particularly important to ensure that the leakage currents that flow in the water pipes that span
from one potential to another, are not allowed to make contact directly with vulnerable materials such
as aluminum.
The valves use guard electrodes of 316-grade stainless steel to protect the vulnerable aluminum heatsinks.
The leakage currents flow safely through the guard electrodes and into the aluminum, rather than
contacting the aluminum directly. This system (illustrated in Fig. 6.2.11d) has been subjected to very long-
term experimental evaluation [5] to demonstrate its suitability for HVDC, and has been in service since
AluminumÊheatsink
PEXÊhose
StainlessÊsteelÊelectrode
ʻOʼ Ring NylonÊnut
Fig. 6.2.11d– GE Vernova coolant hose termination system
1989 where it has given excellent reliability.

The non-zero coolant conductivity also has another undesirable side effect – it causes the water in the
coolant pipes to act as shunt resistors from one part of the valve to another. These unwanted shunt
resistors result in additional valve losses and if not designed correctly, can seriously upset the voltage
6.2.11d
distribution in the valve under DC conditions. They are one of the major factors leading to the choice of DC
grading resistor value (see section 6.2.7).
6.2.11.6. Transient Thermal Design

Although steady-state considerations are important, it is equally important to ensure that the thermal design
of the valve has enough headroom to account for short-term increases of power dissipation. For this reason,
the thyristor junction temperature is normally limited to a steady-state value not exceeding 90°C.
For the thyristors, the most onerous thermal event that can take place in a HVDC valve is an insulation
failure within the valve hall. Several scenarios have to be taken into account, but usually the most onerous
is that caused by a flashover across (or through) a valve. This might be caused by the failure of a surge-
arrester, for example.
Rare though such an event is, the valves must not be damaged by it. The AC breaker is always tripped in
response to this type of event, but the valve has to be designed to withstand the very severe conditions that
occur during the few cycles needed for the protection to operate and the breaker to open.
If a flashover takes place across a valve while that valve is reverse-biased, then the next valve in sequence
on the same row of the converter (shown in red in Fig. 6.2.11e) will feed current into it. If the control and

CHAPTER
ConverterÊtransformerÊvalveÊwinding
(shownÊwithÊstarÊconnection)
Xc
Xc
Xc
2 . VLLÊ(PK)Ê VLLÊ(PK)Ê
PeakÊcurrent*: IPK=Ê =
2 . XC XC
PeakÊrecoveryÊvoltage:ÊVrecÊ=ÊVLLÊ(PK)
*
*
neglectingÊdamping
Fig. 6.2.11e– Fault across a valve
protection system does not respond sufficiently quickly, the third valve on the same row (shown in blue)
may also feed into the fault.
Under the worst possible conditions, this gives rise to a damped offset cosine-wave fault current whose
6.2.11eover 10 times the normal DC current Id (the amplitude of fault current is given
amplitude is typically
approximately by I f = 2 ⋅ Id/Xc, where Xc is the per-unit commutating reactance). The current can be
approximately modeled as a damped offset cosine function.
Normally, the protection system will try to suppress this fault by blocking the forward-feeding valve at the
earliest available opportunity (the end of the first cycle of current). This gives rise to a single-cycle fault
current followed by two or three cycles of AC voltage until the converter breaker opens. The most stressful
point for the thyristor is normally the first forward peak of recovery voltage, 1¼ cycles after fault initiation.
At this point the thyristor is still very hot and may have reduced forward voltage withstand capability.
Occasionally however, other protections may intervene and prevent the valve from blocking at this point.
The fault then becomes a repetitive event, with three or four cycles of fault current terminated by breaker
opening. The most stressful point for the thyristor is normally the commutation overshoot at the end of
the penultimate cycle of current. At this point the thyristor is still very hot (even hotter than for the single-
cycle fault) and may have reduced reverse voltage withstand capability.
6.2.12. Mechanical and Seismic Design

The design of a HVDC thyristor valve involves a number of substantial mechanical engineering challenges.
The principal challenges include the following:
➙ Design of clamping system for thyristors
➙ Design of valve sub-assembly (module)
➙ Structural design of multiple valve unit
➙ External profiling of valve
➙ Seismic design
6.2.12.1. Thyristor Clamping System

High power thyristors use a presspack construction, in which the silicon slice is not soldered or bonded to
the anode and cathode contacts, instead relying solely on a pressure contact. This means that to obtain a
good electrical and thermal contact, both for the various constituent parts within the thyristor and between
the thyristor and heatsink, the thyristor must be mechanically clamped at a very high force.
The required clamping force depends on the silicon diameter (being more or less directly proportional to
area), but for a 125 mm thyristor, a typical value is 135 kN (13.5 tonnes force).
CHAPTER
ValveÊcurrent
ValveÊcurrent
0 360¡ Maximum 720¡ 1080¡ 0 360¡ 720¡ 1080¡

Typical stress TimeÊ(¡) Typical TimeÊ(¡)
conduction Breaker conduction Breaker
ValveÊvoltage
opens angleÊ=Ê330¡ opens
ValveÊvoltage
angleÊ=Ê330¡
here here
Maximum
stress
JunctionÊtemperature
JunctionÊtemperature
Fig. 6.2.11f– Valve voltage, current and junction temperatures resulting from a fault across a valve: single-cycle (left)
and multiple-cycle (right)
The clamping system must apply this force without creating a short-circuit between the two sides of the
thyristor, so it is not possible simply to use a clamping bolt. The technique used on the H300 and H400
6.2.11f
valves is to use tension bands made of a high-performance composite material (glass-reinforced plastic)
with a load-spreader and disc springs at each end.
The disc springs, and the natural flexibility of the tension band itself, give the required compliance, which
allows the clamping system to keep the clamping force within the specified limits while allowing for thermal
expansion and contraction effects.
6.2.12.2. Design of Valve Sub-Assembly (module)

Usually, a number of thyristor levels are built in to a valve sub-assembly of a convenient size for testing
and transportation. The choice of the number of thyristor levels in this sub-assembly is a compromise
between the practicalities of transporting large assemblies and the economics of handling and installing
large numbers of smaller assemblies.
ClampingÊforceÊ≈ 135ÊkN
forÊ125ÊmmÊthyristor
Thyristor
DiscÊspring
GRPÊtensionÊband
Heatsink
Fig. 6.2.12a– Thyristor clamped assembly from H400 thyristor valve

CHAPTER
Fig. 6.2.12b below shows a H450 valve module. This module contains 12 thyristor levels of 8.5 kV thyristors.
Each valve module contains twelve thyristor levels electrically in series. The twelve thyristor levels are arranged
in a single stack of devices and the two di/dt reactors are positioned at the ends of the stack, as shown in
Fig. 6.2.12c.
The structural components in the module are either:
➙ Aluminum, where they are required to carry current
➙ Steel, where they are required neither to carry current nor to provide electrical insulation
➙ Fiber-reinforced composite, where they are required to be electrically insulating
The choice of composite materials must be made with care. Not only must they be strong, lightweight and
electrically insulating, but they must also be fire-retardant (self-extinguishing) and not adversely affected
by moisture. Certain types of composite material absorb water when it is spilt onto them (for example,
from a coolant leak) and then become sufficient electrically conductive to pose a risk of fire.
6.2.12.3. Structural Design of Multiple Valve Unit

HVDC valves are almost never installed as individual units. Nearly always, several valves are combined
together into a Multiple Valve Unit, or MVU. The MVU may either be mounted directly on the floor or, more
commonly today, suspended from the ceiling.
Fig. 6.2.12b– A H450 valve module
As a minimum, it is normal to fit two thyristor valves (from the same AC phase of the same 6-pulse bridge)
into an MVU, making a double-valve. In this way, the insulating support structure is shared between the
two valves. Each pole then contains six such double valves, three for the lower 6-pulse bridge and three for
the higher 6-pulse bridge, one possible arrangement of which is illustrated in Fig. 6.2.12d.
The double-valve structure is commonly used for very high voltage HVDC schemes where single-phase,
two-winding converter transformers are used, because the size of these transformers dictates the size of
the valve hall.
On lower-voltage HVDC schemes, a more common arrangement is with four valves associated with the
same AC phase stacked vertically into quadrivalve structures; three such quadrivalves being required at
each end of each pole (see Fig. 6.2.12e). The quadrivalve arrangement decreases the area of the valve hall,
but makes the valve hall taller.
The MVU structure can be very large; typically 6 m × 4 m × 10 m tall. Fig. 6.2.12f shows a typical suspended
Multiple Valve Unit. Suspended designs are common today because of the lower cost of structural
components2.

CHAPTER
Fig. 6.2.12c– Electrical circuit of H450 valve module
ValveÊsupportÊexposedÊtoÊ50%
ofÊratedÊpoleÊDCÊvoltage
DCÊneutralÊ HVDC
Fig. 6.2.12d– 12-pulse group of valves implemented as double-valves
6.2.12.4. External Profiling of Valve Structure

HVDC thyristor valves operate at very high voltages with respect to ground. This means that all external
surfaces of the MVU have to be carefully designed to ensure that the surface electric fields do not become too
6.2.12d
high. The breakdown strength of air at atmospheric pressure is about 3 kV/mm under uniform-field conditions,
but at field strengths above 2 kV/mm, corona discharges can occur. Corona discharge is very undesirable
because it generates ozone and UV radiation, both of which can damage plastic materials in the valve. It can
also cause electromagnetic interference and increase the risk of flashover.
For these reasons, the external surfaces of thyristor valves must be generously curved (see Fig. 6.2.12g),
with no sharp points. This is particularly true for the high voltage end of the MVU - normally at the top for a
base-mounted design or at the bottom for a suspended design - and becomes increasingly important for
higher-voltage HVDC schemes. The arrangement shown on the right hand side of Fig. 6.2.12g is designed
for operation at 800 kVdc and is not expected to require major changes at 1000 kVdc [7], [8].
6.2.12.5. Seismic Design

One of the major mechanical design challenges with thyristor valves concerns their behavior during
earthquakes. Many of the world’s existing or potential HVDC stations are in seismically active areas, and
since a thyristor valve is a very large and expensive structure, seismic design is a serious issue.
The choice of floor-mounted versus suspended design is a decisive factor here: floor-mounted valves, although
technically possible [9], are vulnerable to excessive mechanical (cantilever) stresses in the support insulators,
and to make these insulators strong enough can add to the cost.
The suspended design that is now common completely removes all the concerns about excessive stress
in the insulators, since these can now simply be very low-cost tension insulators similar to those used on
overhead lines. However, it introduces a completely new problem instead – the valve is now a pendulum, and

CHAPTER
ValveÊsupportÊexposedÊonlyÊ
toÊDCÊneutralÊpotential DCÊneutral TopÊstressÊshield
Module Module
DC
Neutral Module Module
AC
(Delta) Module Module
0.5ÊVdc
DCÊmidpoint CenterÊstressÊshield
Module Module
HVDC
Vdc Module Module
AC
(Star) Module Module
HVDC BottomÊstressÊshield
Fig. 6.2.12e– 12-pulse group of valves
implemented as quadrivalves Fig. 6.2.12f– Typical suspended MVU for HVDC
6.2.12e if it is excited by ground movements of sufficiently low frequency (<1 Hz) then very large displacements can
result. The problem is therefore one of ensuring that electrical clearances in the valve hall remain sufficient.
These issues are illustrated in Fig. 6.2.12h.
For either approach, a detailed Finite Element Analysis (FEA) of the valve structure is required.
6.2.13. Valve Testing

6.2.12f
Thyristor valves need comprehensive testing: both routine testing in the factory and type testing on new
designs. The type testing in particular is very specialized and complicated.
6.2.13.1. Routine Testing

All thyristor valves are subjected to a comprehensive routine test program in the factory. The purpose of this
test program is to prove that the thyristor valves have been correctly assembled. It aims to identify wiring
connections that have been incorrectly made, grading components that are out of tolerance, gate electronics
that are malfunctioning, blockages in the cooling circuit, poor joints between the thyristor and its heatsinks,
etc.
The relevant standards for routine testing of thyristor valves are outlined in IEC 60700-1 [10], although the
emphasis of this document is more on type tests.
6.2.13.2. Type Testing

When a new design of thyristor valve is produced, or where the valves for a particular project are based on
an existing design but are different in some way from valves that have previously been tested, a program
of type tests is performed. For these tests, the relevant standard is again IEC 60700-1 [10].
Type testing of thyristor valves is complex and time-consuming, with some type tests requiring highly
specialized test circuits.
Type testing falls into two broad categories: dielectric tests and operational tests.
• Dielectric Tests
2 – It is a misconception that suspending the valve from the ceiling solves all problems in seismic applications. The truth is that
the suspended design presents a different and generally more manageable set of challenges than a floor-mounted design.
CHAPTER
Fig. 6.2.12g– External corona profiling of H400 HVDC valves
StressÊshield
StressÊshield StressÊshield
StressÊshield
Movement
Movement
ofÊvalveÊduring
ofÊvalveÊduring Module
Module Module
Module
earthquake
earthquake
StressÊshield
StressÊshield Module
Module Module
Module
Module
Module Module
Module Module
Module Module
Module
Module
Module Module
Module StressÊshield
StressÊshield
Movement
Movement
Module
Module Module
Module ofÊvalveÊduring
ofÊvalveÊduring
earthquake
earthquake Module
Module Module
Module
StressÊshield
StressÊshield Module
Module Module
Module
Module
Module Module
Module Module
Module Module
Module
Excessive
Excessive Module
Module Module
Module StressÊshield
StressÊshield
bendingÊstress
bendingÊstress
onÊinsulatorsÊ
onÊinsulatorsÊ Module
Module Module
Module
ExcessiveÊdisplacement,
ExcessiveÊdisplacement,
potentiallyÊviolating
potentiallyÊviolating
StressÊshield
StressÊshield
electricalÊclearances
electricalÊclearances
Fig. 6.2.12h– Typical response of a floor-mounted (left) and suspended valve (right) during an earthquake event
Dielectric tests are aimed essentially at verifying that the thyristor valve can support the voltage (and dv/
dt) applied across it. They appear, superficially, to be quite similar to the dielectric tests performed on most
conventional electrical equipment. However, their apparent similarity to conventional equipment tests
belies the fact that they can still be quite difficult tests to perform on a thyristor valve.
6.2.12h
6.2.12h
Dielectric tests fall into three sub-categories:
(a) Tests on the valve support
(b) Tests between valve terminals
(c) Tests on a multiple valve unit (see Fig. 6.2.13a)

CHAPTER
a. Tests on the Valve Support
These tests do not test the active part of the
TestsÊonÊvalve
thyristor valve, but merely the structural insulators supportÊstructure
that support it: either from the ground (for a base-
mounted valve) or from the ceiling (for a suspended
valve). In many cases, insulation levels across the TestsÊonÊMVU
insulating support for the valve are quite modest,
but in some arrangements, for example where
the valves are implemented as double valves (see TestsÊbetween
valveÊterminals
Fig. 6.2.12d), the insulation levels can be quite high.
They represent probably the most straightforward Fig. 6.2.13a– Dielectric tests on a HVDC valve
of valve tests. The tests can normally be carried out
in any conventional HV test lab. However, since the
test arrangement is basically the same as is required for the valve dielectric tests between terminals, the
two groups of test are normally performed at the same time.
The dielectric tests performed are: 6.2.13a
➙ DC voltage
➙ AC voltage
➙ Switching impulse (250/2500 μs)
➙ Lightning impulse (1.2/50 μs)
Sometimes a steep-front impulse (0.4/4 μs) is also specified.
b. Tests between Valve Terminals
These tests involve applying very large voltages between the terminals of the thyristor valve, and as such,
they do stress the active part of the valve. They result in the highest voltages experienced by the thyristors
and other components under any conditions, and are a decisive factor in choosing the number of series-
connected thyristors.
Although the tests comprise the standard set of dielectric tests mentioned above, plus the steep-front
impulse, they are difficult to perform in practice because a valve has a very low off-state impedance compared
with most conventional AC test objects.
In addition, for impulse voltages in the forward direction, the valve can protectively fire, (as described in
Protection Requirements in section 6.2.10.1), thus short-circuiting the impulse generator.
Most items of electrical equipment would only draw a very small current of a few mA during dielectric
testing, but in the case of a thyristor valve the figure is much higher because of the relatively low
parallel impedance of the grading circuits, water cooling circuits and stray capacitances. As a result,
during dielectric testing between valve terminals it is normal to see currents of hundreds of mAdc,
several amps AC or tens of amps peak during impulse tests, even when the valve does not turn on. Those
tests which cause the valve to turn on (the forward-direction impulse tests) pose additional challenges, since
these tests effectively short-circuit the impulse generator. The circuit must be arranged so that the inrush
current from the impulse generator does not damage the valve (avoiding excessive di/dt which could destroy
the thyristors). For one particular test – the non-periodic firing test – the inrush current has to be tailored to
an exact amplitude and waveshape, based on studies, to confirm that the single shot inrush current capability
of the valve is adequate.
Hence, these tests require much stronger voltage sources and impulse generators than are available in most
HV labs.
c. Tests on a Multiple Valve Unit
As the name suggests, these tests are applied to assemblies containing more than one valve. In some
cases, they also involve applying voltages between the terminals of the respective valves in the MVU, but
this is not the primary purpose of the tests. The tests are mainly to prove the correct voltage withstand
between valves and from the outside of the MVU to its surroundings. One of the most important tests for
a HVDC valve is the DC corona test, which verifies that the external surfaces of the valve structure are
smooth enough to prevent corona discharge in normal operation [7], [8].

CHAPTER
The MVU tests can be quite difficult to perform because in some cases the current drawn from the supply,
like that during the valve dielectric tests between terminals, can be much higher than for a conventional test
object. In addition, the complete MVU is a very large structure, so it requires a correspondingly large test
laboratory, often including a suspension frame.
• Operational Tests
While the dielectric tests focus mainly on the stresses on the valve during rare fault conditions, the
operational tests are intended to mimic more closely the stresses experienced by a thyristor valve in
operation. Being a switch, a thyristor valve is subjected alternately to voltage (when it is off) and current
(when it is on), and as mentioned in section 6.1, the transitions between these states can be very stressful.
Operational tests fall into two basic sub-categories:
(a) Periodic firing and extinction tests: various tests involving repetitively turning the valves on and off
(b) Overcurrent tests: tests which aim to verify the short-time capability (typically for only a few cycles) of a
valve to withstand much larger values of current than normal
Because of the high powers involved, it is rarely practical to perform operational tests on a complete valve.
Instead, the tests are performed on valve sections comprising at least five levels and are repeated enough
times on different valve sections to ensure that the equivalent of a whole valve has been tested.
a. Periodic Firing and Extinction Tests
These are the most specialized of all valve tests because they aim to replicate the repetitive voltage and
current stresses experienced by a valve when it is functioning normally in the converter circuit. The tests have
to produce adequate current during the on-state and realistic voltages at turn-on and turn-off. Several other
conditions are also important, for example the di/dt at the instants of turn-on and turn-off, and the sum-of-
squares of the voltage jumps (∑(∆V)2) appearing during the off-state cycle (which have a significant influence
on the power dissipation in the damping circuits).
These factors are illustrated in Fig. 6.2.13b.
To reproduce all these conditions on a full valve at realistic power is impractical. There are two well-
established methods that can be used to perform these tests at scaled-down power levels: a 6-pulse
back-to-back test circuit and a synthetic test circuit (see Fig. 6.2.13c).
The 6-pulse back-to-back circuit has the advantage of inherently giving realistic voltage and current waveforms
and is intuitively easier to understand. However, with this circuit the voltages that can be generated with any
practical test set-up are very low, which limits the test circuit to very small numbers of thyristors in series.
The synthetic test circuit [11] avoids this problem by producing the high voltage and high current supplies
from separate circuits, each of modest power. It therefore allows much larger valve sections to be tested with
more moderate power consumption than is possible with the back-to-back circuit.
The synthetic test circuit was pioneered by GE Vernova but variations of it have now been adopted by all
major manufacturers.
A synthetic test circuit for performing the periodic firing and extinction tests is shown in simplified form
on Fig. 6.2.13d.
The high current circuit (left) consists of a low voltage, high current controlled rectifier (DC1) rated at
approximately 2 kAdc. This feeds a storage capacitor C3 and a small STATCOM, which controls the duty cycle
of the current fed into the test object V1. With the normal 1:3 duty ratio, the circuit can produce an equivalent
DC bridge current of up to 6000 A in the test object.
The high current circuit is isolated from the test valve by a thyristor valve V2, which must be able to carry both
the full current and full voltage of the test object. Another thyristor valve, V1X, is connected directly in parallel
with the test object and allows the circuit to be started up without firing the test valve.
The high voltage circuit is a resonant circuit in which energy is supplied from a low-power rectifier DC2.
DC2 in fact operates at a relatively low voltage but the amplifying nature of the circuit allows test voltages
approximately 20 times as great as this to be applied to the test valve. The circuit works by alternately firing
valves V5 (to charge up the capacitor C1A) and V3 (to discharge C1A into the test valve via the inductance
L1A). L1A is chosen to be equivalent to the commutating inductance of the HVDC scheme, so it allows the
circuit to generate the correct di/dt at turn-on and turn-off. Valve V4 is used to connect V1 to the capacitor
C1A, so that V1 experiences the correct turn-off voltage.
CHAPTER
On-state:
• Peak current
Turn-on:
• Rms current
• ∆V
• di/dt
Valve • Stray capacitance
voltage Turn-off:
• Prospective ∆V
Valve • di/dt
current • Stray capacitance
Off-state:
• Σ(∆V)2
Fig. 6.2.13b– Important factors for periodic firing and extinction tests
The surge injection circuit, consisting of an impulse generator and triggered spark gap, is used to perform the
‘positive voltage transient during recovery’ tests, which demonstrate the turn-off time of the thyristor and the
correct action of the forward recovery protection.
6.2.13b
Fig. 6.2.13e shows, in idealized form, the voltage and current waveforms obtained from the high current
and high voltage circuits. Although all the voltage applied to the test object comes from the high voltage
circuit, the current in the test object comes from both the high current and high voltage circuits.
Although the synthetic test circuit has enabled large sections of thyristor valves to be tested under
realistic voltage and current conditions, including at frequencies different to the local grid frequency, the
voltage and current waveforms it produces are not fully representative of those experienced by a valve
in service. In that respect, the 6-pulse back to back test circuit was better, although it was not perfect
and its limitations of power rating and the inability to operate at a frequency other than the local supply
frequency were major drawbacks.
Impulse
HI HV
Fig. 6.2.13c– 6-pulse back-to-back circuit (left) and synthetic test circuit (right)

CHAPTER
6.2.13c
SurgeÊinjectionÊcircuit
DC3
+
STATCOM V2 V5
L1A V3 L2 L6
+ DC2
L3 +
C3
V1X V1
V4
DC1 C1A
HighÊcurrentÊcircuit TestÊvalve HighÊvoltageÊcircuit

andÊlocalÊ
circuit
Fig. 6.2.13d– A synthetic test circuit for periodic firing and extinction tests
For this reason, when GE Vernova decided to build a brand new Valve Test Facility, a different path was
taken 6.2.13d
and GE Vernova once again found itself in the role of pioneer.
The new test facility inaugurated in 2019 operates on a quite different principle [12], which combines the
advantages of both of the previous methods: instead of using separate, interleaved high-voltage and high-
current sources, the new facility derives both the high voltage and high current from a single, high-power
Voltage-Sourced Converter (VSC). The VSC uses four parallel-connected branches of 48 “Full-Bridge” VSC
submodules connected in series as shown on Fig. 6.2.13f. the VSC submodules are operated, in effect, as
a “programmable voltage source”.
With the new test circuit it is possible to reproduce the off-state voltage and on-state current waveforms
of the valve with exceptional accuracy. Fig. 6.2.13g illustrates the valve voltage and current waveforms
that can be obtained with the test circuit. As can be seen, the results are very close to those applicable
to real operation in inverter mode.
The new facility can, in addition, also perform all the other required operational tests including operation
in intermittent direct current and the application of fast voltage impulses during the recovery interval of an
inverter valve.
b. Overcurrent Tests
For HVDC valves, the overcurrent test proves that the valve can safely survive the very large fault current,
and subsequent recovery voltage, that results from a short-circuit across one valve in the bridge (see
section 6.2.11.6). With large, modern thyristors, this typically gives a fault current of 40-50 kA peak. At
the end of the fault current, the test object may be required to block forward and/or reverse voltages of
several tens of kV.
For the overcurrent tests, the test circuit can be a fairly conventional high power test circuit similar to that
used for testing circuit breakers.

CHAPTER
V1(HIÊcircuit) V1(HVÊcircuit)
RealisticÊdi/dtÊat
I turn-onÊandÊturn-off
Êt
ÊV1 ÊC1A
Peak
firing
voltage
PFV
ÊV
Êt
Prospective
recovery
voltage
PRV
Fig. 6.2.13e– Voltage and current waveforms for synthetic test circuit
6.2.14. Other Design Issues

6.2.13e
HVDC installations are major capital projects that handle levels of power flow that are comparable with
the output of a large power station. Such installations require comparable levels of design life (30 years is
typical), extremely high levels of availability (approaching 99%) and low forced outage rate (no more than
a few times per annum). Achieving this is a challenge. Adequate and conservative rating of components,
rigorous testing to verify compliance with requirements and provision of redundant or back-up systems,
where possible, are amongst the methods that can be used to meet these objectives.
In the case of the thyristor valves, the large number of thyristor levels in a valve provides a large multiplying
factor leading to a high component count. All components exhibit the characteristic bathtub curve of failure
rate, with high initial ‘infant mortality’, declining to a long period of low but sustained random failures before
finally rising again as components start to wear out at the end of their usable life. HVDC projects typically
allow for an initial burn-in period of 3 to 6 months but thereafter, contractual penalties may be applied to
the reliability and availability performance of the equipment.
6.2.14.1. Life and Reliability

Occasional random component failures are inevitable. One action taken in valve design to deal with this
is to provide series redundancy: for example, by adding a few more thyristor levels to each valve so that
uninterrupted operation can continue in the presence of a small number of randomly distributed failures.
Redundancy is typically 3%, depending on the size of the valve and the forced outage rate specified for
the station.

CHAPTER
The principle of providing redundant series thyristor levels only works because thyristors always fail to a
short-circuit condition. In this state, they can safely conduct current for as long as is required. However, it
is important to ensure that:
➙ No component in parallel with the thyristor can fail to a short-circuit condition
➙ No component in series with the thyristor can fail to an open-circuit condition
Fig. 6.2.13f– Schematic view of new test circuit for operational tests
Both of these would lead to very dangerous conditions.

To maximize reliability, thyristor valves should be designed such that:
➙ Credible faults must affect no more than a single level
➙ All faults must be inherently safe: without risk of fire, explosion or other consequence that could
result in an immediate need to shut down. Fortunately, thyristors fail to a short-circuit so, where
possible, the failure of other components is arranged to result in the failure of the associated
thyristor
Fig. 6.2.13g – Voltage and current waveforms from the new test circuit in inverter mode

CHAPTER
➙ Failed thyristors must carry the normal load and overcurrent safely, until the next maintenance outage
➙ Failed thyristors are detected via the databack monitoring system. Individual failures will be flagged
and recorded for attention at the next maintenance outage. If the redundancy in any one valve is used
up, an alarm will be raised, giving the operators time to schedule an outage before another failure
occurs in that valve. If redundancy is exceeded, the converter will be automatically tripped
➙ Other faults have to be engineered to be so rare that they do not materially contribute to the forced
outage rate of the converters caused by the valves.
6.2.14.2. Operating Environment

HVDC valves require a very clean environment since they act as rather effective electrostatic precipitators
and tend to cause any dust present in the air to adhere to solid surfaces. For this reason, HVDC valves must
always be mounted indoors in a well-sealed building with only recirculating ventilation or air-conditioning, and
a high-degree of filtration of incoming air. It is also usual to pressurize the valve hall at a slight over-pressure
relative to atmospheric pressure, to minimize the ingress of dust through orifices such as door-surrounds, etc.
Temperature and humidity also need to be controlled, most importantly to prevent condensation forming
on insulating surfaces.
6.2.14.3. Fire
Great care must be taken to minimize the risk of fire since unless adequate precautions are taken thyristor
valves can be very vulnerable to fire. Between 1989 and 1993, three high-profile incidents on HVDC projects
built by other manufacturers [13] resulted in complete thyristor quadrivalves being destroyed by fire.
GE Vernova’s HVDC valve designs have always used materials of low flammability and care has been taken to
avoid the events which could lead to a fire in the first instance. This means generously rating all components
for the worst conditions they can be expected to encounter.
Choice of materials is also very important. Virtually any material will burn if given an adequate supply of oxygen
and sufficient heat input. However, the key requirement is that the material must be self-extinguishing, that
is to say that it may burn if subjected to heat input, but that the flames will extinguish when the heat source
is removed. The applicable standard is Underwriters Laboratory (UL) 94, which has several classifications, of
which the most onerous is V0.
Some older valve designs used oil-impregnated capacitors. The oil in these capacitors represents a source of
fuel which could leak out and, once ignited, continue to supply heat for many minutes. From the H400 valve
onwards, these oil-filled capacitors have been replaced by dry capacitors.
6.2.15. Installation, Commissioning and Maintenance

Normally, the factory-built sub-assemblies (see section 6.2.12.2) are shipped to site fully assembled, and the
remaining parts needed to complete the valve assembly such as inter-tier insulators, water-pipes, fiber-optic
trunking, busbars, etc. are shipped as separate items.
The installation of the valve at site therefore consists of building up each MVU from the ground up (for a
base-mounted design) or ceiling down (for a suspended design). For each tier of the structure, the structural
insulators are fitted first, followed by the valve modules, inter-tier water pipes, busbars, fiber-optic trunking
and finally the fibers themselves. The laying of the optical fibers is quite a delicate task, which has to be
coordinated with the installation of the VBE cubicles.
Thyristor valves do not require any commissioning as such, but prior to starting energized commissioning of
the complete scheme, the thyristor valves undergo a series of pre-commissioning tests. These essentially are
to confirm that the valve sub-assemblies (which are thoroughly routine-tested in the factory) have survived
the transportation to site. The pre-commissioning tests use a dedicated Valve Test Equipment (VTE).
The VTE permits a series of electrical checks to be made on each thyristor level. The tests performed typically
include forward and reverse voltage withstand, checks on the correct operation of VBO protection and of the
various protective features provided by the TCU, and checks of correct TCU power supply charging.
A simplified version of the VTE may be supplied with the maintenance equipment, to be used during outages
to verify all the functions of a thyristor level after replacing a component: for example, a failed thyristor.

CHAPTER
Coolers T P T
(1Êredundant) MainÊfilter
Dp
Thyristor
Cm Cm valves
Nitrogen
bottle(s)
Exp.Êtank
L
F
Main
deioniser Filter Cm
Make-up Cm
Deioniser Filter
pump
MakeÊup
tank
P P T
3-wayÊvalves
Degassing
MainÊpumps
tank
(1Êredundant) F
(1Êredundant)
KeyÊtoÊsensors
T Temperature Cm Conductivity
F Flow P Pressure
L Level Dp DifferentialÊpressure
Fig. 6.3a– Typical HVDC cooling plant flow diagram – main circuit
Thyristor valves require very little maintenance. Normally, a shutdown every year or two is acceptable, during
which time the operations that are typically performed are:
➙ 6.3a
Replacing failed components
➙ Cleaning external surfaces of highly-stressed insulation: for example the inter-tier insulators and water
pipes
When a thyristor needs to be replaced, a purpose-built thyristor changing kit is used. Typically this comprises
two tools: a primary hydraulic loading tool which pressurizes the clamped assembly, and a spreader tool to
separate the heatsinks either side of the failed thyristor.
6.3. VALVE COOLING SYSTEM

The high voltage part of the valve cooling system (the part contained within the valve itself) has already
been described in section 6.2.11.5 above. However, the complete system needs certain other equipment:
➙ Heat exchangers to reject the heat to the surroundings
➙ Pumping, filtering and purifying equipment (cooling plant)
➙ Pipework to interconnect the above to the thyristor valves
The detailed design of the cooling plant varies from project to project, depending on environmental conditions
and special requirements. However, typical characteristics of some of the main components are described
in the following pages.
A typical flow diagram of a single-circuit cooling plant used for HVDC thyristor valves is detailed in Fig. 6.3a
below. It shows the location of all key components in the cooling circuit between the connection of the valves
to the coolers. Some items in this diagram might be simplified or omitted in certain cases.

CHAPTER
The detailed design of the cooling plant varies from project to project, depending on environmental
conditions and special requirements. However, typical characteristics of some of the main components
are described below.
6.3.1. Coolant
The cooling system uses pure de-ionized water or (where freezing conditions can occur) a water/glycol
mixture, for the coolant. General-Purpose Reagent grade (GPR) ethylene glycol is specified with no additives.
It also has a very low conductivity (0.3 µS/cm or better) and a very low level of impurities (1% or less).
Propylene glycol can be used as alternative to ethylene glycol. Propylene glycol is less toxic than ethylene
glycol but is more difficult to use in very cold climates because the viscosity increases more rapidly at low
temperatures than that of ethylene glycol.
6.3.2. Filtering
A full flow filter is typically included in the main cooling circuit before the thyristor valves, to prevent debris
trapped in the system from entering the thyristor valves. The filter may be fitted with a full flow bypass
circuit to allow the replacement/cleaning of the filter basket while the system is in operation. A differential
pressure switch positioned across the filter will indicate a blockage in the filter basket.
6.3.3. De-Ionization
The main de-ionizer system is typically a secondary circuit off the main circuit. A small percentage of the
main flow is passed through the main de-ionizer cylinder to maintain the conductivity of the coolant below
a set limit. This circuit also contains a fine filter on the outlet side, a flow detection sensor and conductivity
meter to monitor the coolant leaving the de-ionizer system before it re-enters the main system.
6.3.4. Main Pumps

The main coolant circulation pumps generally consist of a duty pump and a standby pump. They are
normally hydraulically connected in parallel and are arranged so that maintenance work can be carried
out on one pump while the other pump is in operation. Usually the duty pump function is cycled between
the two main pumps to even out the average service hours. A non-return valve is positioned at the outlet
of each pump to prevent reverse flow through the standby pump.
6.3.5. 3-Way Valve

In areas where low ambient temperatures may be experienced, a 3-way valve is usually provided. The
coolant from the cooler bank circuit may be diverted to the cooler bank bypass circuit by using a 3-way
valve. The 3-way valve is used to limit the minimum coolant temperature entering the thyristor valves and
to prevent condensation forming within the thyristor valves when the thyristor valve inlet temperature is
low and the scheme power transfer level is low. The 3-way valve may be controlled by electrical actuators.
In locations with high minimum ambient conditions, a 3-way valve is unnecessary. Instead, only a bypass
with shutoff valve is used to quickly increase the coolant temperature.
6.3.6. Coolers
The coolant is generally circulated from the pumps to the cooler units located outside. These coolers can
be of a variety of different types, depending on environmental conditions and customer preferences:
➙ Dry air-blast coolers (the most common type)
➙ Air-blast coolers with supplementary spray cooling
➙ Evaporative coolers
➙ Water-to-water heat exchangers, where a secondary (raw) cooling water circuit exists

CHAPTER
6.3.7. Expansion Tank
The expansion tank accommodates the normal variation in volume of the coolant due to expansion over the
full working range of temperatures. The expansion tank has a pressurized nitrogen blanket to maintain the
static pressure required in the cooling system. A restricted flow of coolant from the main circuit is passed
through the expansion tank to keep the coolant in the tank in prime condition. Level switches are included
on the expansion tank to provide make-up control signals in the event of a minor coolant leak and alarm
and trip signal in the event of a major system leak.
6.3.8. Make-Up System

The risk of coolant escaping from the valves is very small but some coolant may be lost from pump seals or
during maintenance: for example, filter cleaning. Hence, a make-up system is included that adds coolant
to the main system to compensate for any loss. The system includes a tank, pump, de-ionizer, filter and
conductivity meter. The system is arranged so that the coolant in the tank can be circulated through the
make-up system filter and de-ionizer unit independently of the main system coolant, to condition the
make-up coolant quantity.
6.3.9. Control
The cooling plant controller is typically PLC based and controls the inlet temperature to be between 20°C
and 60°C, with a target temperature adjustable between 25°C and 55°C. The controller controls the fans
and the position of the motorized bypass valve so that this is achieved. The controller evens out the duty
of the pumps by rotating the duty of the pumps on a periodic basis.
A display is provided that enables the plant to be controlled automatically or manually and gives analog and
status information as well as alarm information. Typically, three different modes of control are provided: full
automatic, manual via the PLC and manual via control switches.
6.3.10. Instrumentation
Typically, the instruments, gauges, sensors and switches, listed in Table 6.3a are provided at the locations
shown in Fig. 6.3a:
Sensor Used for

Differential pressure sensor Total flow measurement, display and alarm
Flow switches Individual branch flow alarms and local display
Flow switch Total flow alarm
Temperature transmitter Display, alarm and control of 3-way valve
Temperature switch Alarm
Main loop conductivity sensor Display and alarm
De-ionizer loop conductivity sensor Display and alarm
Level gauge and switches on the expansion tank Coolant level alarm and contact output
Pressure switch at pump outlet Alarm and pump changeover
Pressure gauge at pump outlet Local display
Table 6.3a– Instrumentation sensors and switches
Two temperature transducers, mounted on the inlet to the thyristor valves, are provided. These outputs
are fed into the control system to calculate the thyristor junction temperature in real time.
6.3.11. Alarms
Individual alarms are displayed on the local control point. The following local alarms are typically provided
as a minimum.

CHAPTER
Idc
Upper
6-pulse bridge
AC system
Filter
DC system
Transformer
bank Lower
6-pulse bridge
Lower
6-pulse bridge
AC system
Filter
DC system
Transformer
bank Upper
6-pulse bridge
Idc
Fig. 6.4a– Typical 12-pulse bridge schematic, shown for a bipolar scheme
Non-trip alarms Trip alarms

➙ Low flow in each valve branch ➙ Very low flow in each branch
➙ Low total flow
6.4a ➙ Very low pressure at pump outlet
➙ Low pressure at pump outlet ➙ Very high conductivity
➙ High conductivity ➙ Very low level in the expansion tank
➙ Low level in the expansion tank ➙ Very high temperature
➙ High level in the expansion tank
➙ Falling level in the expansion tank The high conductivity alarm setting is typically 0.5
➙ High temperature µS/cm and the very high conductivity trip setting
➙ Low temperature is typically 1.0 µS/cm.
➙ Pump failure for each pump
➙ Fan failure for each fan
6.4. HVDC TRANSFORMERS

The HVDC converter transformer provides the connection between the AC system and the converter
station valve hall. The basic transformer function is the same as a normal AC power transformer: to
transform AC voltage between the different voltage levels of the AC supply system and the voltage
required by the converter design. It is conventional to refer to the transformer windings connected to
the AC system as the line windings and the windings connected to the converter as the valve windings.
A 12-pulse scheme will normally consist of a star/star and a star/delta transformer as detailed in
Fig.6.4a which shows a bipole scheme.
In most HVDC schemes, both star and delta arrangements are used for converter windings to provide
an effective 12-pulse operation. The transformer bank may be 3-phase or single-phase and there are
several options which may be adopted for each 12-pulse converter group.

CHAPTER
transformerÊsecondaryÊ(d)ÊL-NÊvoltage
2.0
1.0 a
0.0
p.u.
-1.0
b
-2.0
-3.0
-4.0
transformerÊsecondaryÊ(d)ÊL-LÊvoltage
2.0 a
1.5
1.0
0.5 b
p.u.
0.0
-0.5
-1.0
-1.5
-2.0
transformerÊsecondaryÊ(s)ÊL-LÊvoltage
2.0
a
1.5
1.0
0.5 b
p.u.
0.0
-0.5
-1.0
-1.5
-2.0
Fig. 6.4b– Valve winding voltage waveforms
6.4b
LineÊregulatingÊwinding
LineÊregulatingÊwinding
ValveÊwinding
ValveÊwinding
LineÊwinding
LineÊwinding
CoreÊlimb CoreÊlimb
ValveÊwindingÊnextÊtoÊcoreÊlimb LineÊregulatingÊwindingÊnextÊtoÊcoreÊlimb
Fig. 6.4c– Concentric winding cross section layouts
➙ One 3-phase, 3-winding transformer ➙ Three single-phase, 3-winding transformers

➙ Two 3-phase, 2-winding transformers ➙ Six single-phase, 2-winding transformers
Although the 3-phase 3-winding unit requires the minimum number of power transformers, (one per
12-pulse converter) the required internal layout of the windings, required to meet strict impedance
matching criteria between the star and delta line to valve winding pairs, limits the possible MVA rating of
6.4c
this arrangement, and as a result it is only used on a few, relatively small, HVDC installations.
The 3-phase, 2-winding and single phase, 3-winding options are widely used and require two or three
transformers per 12-pulse converter respectively. These arrangements tend to be the most economic for
system requirements on intermediate-sized HVDC systems, especially when the availability of a spare
unit is a factor.

CHAPTER
The single-phase, 2-winding option is more likely to be employed where either severe transport restrictions
have to be met or at larger MVA ratings or where very high DC voltage levels are required. As voltage and
power levels have increased in recent years, this option is becoming more common.
The transformer design will follow the same principles as a conventional HVAC transformer except that
there are additional effects of the HVDC application that must be considered. The most obvious of these
is the DC voltage stress, which has a major influence on the insulation structure of the valve windings. The
voltage waveforms that are applied onto the valve windings by connection to the converter bridge have to
be considered, with the transformer insulation being continuously stressed by a DC voltage component and
an AC voltage that is distorted by periodically recurring commutation steps as shown in Fig. 6.4b below.
The HVDC converter transformers manufactured by GE Vernova are conventionally of core type design:
the different windings required in the transformer design are arranged as successive concentric cylinders
over a central core limb (see Fig. 6.4c below).
The winding layouts applied to HVDC converter transformers are of two main types, with the valve
winding placed as either the inner winding or the outer winding. Which layout is the most cost effective is
governed by the insulation levels required for the line and valve windings. For conventional HVDC converter
transformers, with concentric windings, connected to line voltages above 230 kV, the valve winding will
more usually be the inner transformer winding on DC systems up to 300 kVdc. For higher DC system
voltages, it is difficult to design the valve winding lead exit insulation system between the windings and
the transformer bushing. Consequently, in these applications the valve winding is often arranged to be
the outermost winding.
For any HVDC converter transformer, however, the choice of inner or outer valve winding will be a balance between
the valve winding lead exit complexity and the line & regulating winding lead exit voltage insulation levels.
2.175e+007Ê:Ê>2.289e+007
2.060e+007Ê:Ê2.175e+007
WindingÊendÊpressboard 1.946e+007Ê:Ê2.060e+007
TopÊclampingÊplatform
insulationÊstructure 1.831e+007Ê:Ê1.946e+007
1.717e+007Ê:Ê1.831e+007
AngleÊring 1.602e+007Ê:Ê1.717e+007
1.488e+007Ê:Ê1.602e+007
BoxÊring
1.373e+007Ê:Ê1.488e+007
1.259e+007Ê:Ê1.373e+007
1.144e+007Ê:Ê1.259e+007
Core
1.030e+007Ê:Ê1.144e+007
ShieldÊring 9.156e+006Ê:Ê1.130e+007
6.867e+006Ê:Ê8.011e+006
5.722e+006Ê:Ê6.867e+006
ValveÊwinding LineÊwinding 4.578e+006Ê:Ê5.722e+006
3.433e+006Ê:Ê4.578e+006
2.289e+006Ê:Ê3.433e+006
1.144e+006Ê:Ê2.289e+006
<0.000e+000Ê:Ê1.144e+006
DensityÊplotÊ:ÊIÊEÊI,ÊV/m
Fig. 6.4.1d– DC voltage stress plot for inner valve winding
A wide variation of voltage output is required from a converter transformer to allow flexible control of power
flows and therefore regulating (tap) windings are an essential provision. The converter transformer may be
required to provide voltage ratio increments of less than 1.25%, with a total tapping range of up to 40%. In
addition to an AC voltage superimposed on a DC voltage level, other transient or short time stresses can be
imposed on the valve windings as a result of converter operations and faults.
6.4.1d These include:
➙ Normal converter control operations, such as blocking and starting of power transmission
➙ Polarity reversal of the converter bridge DC voltage during line faults
➙ External overvoltages such as lightning strikes to the DC line, induced overvoltages in the healthy
pole during ground faults on the other pole and from the AC side via transformed overvoltages
➙ Power reversal carried out by reversing the DC voltage for long time operation in this condition
CHAPTER
6.4.1. Insulation Design
The insulation structure must be designed to be suitable for both the combined AC and DC service voltages
and the various test voltages applied during Factory Acceptance Testing. Testing consists of both normal
AC testing of the transformers such as AC power frequency, lightning impulse and switching impulse, but
also includes additional testing with DC withstand voltage, polarity reversal voltage testing and extended
AC withstand voltage associated with the valve windings.
1.426e+007Ê:Ê>1.501e+007
1.351e+007Ê:Ê1.426e+007
1.276e+007Ê:Ê1.351e+007
1.201e+007Ê:Ê1.276e+007
1.126e+007Ê:Ê1.201e+007
1.051e+007Ê:Ê1.126e+007
9.758e+006Ê:Ê1.051e+007
9.007e+006Ê:Ê9.758e+006
8.257e+006Ê:Ê9.007e+006
7.506e+006Ê:Ê8.257e+006
Core
6.755e+006Ê:Ê7.506e+006
6.005e+006Ê:Ê6.755e+006
ShieldÊring
4.504e+006Ê:Ê5.254e+006
3.753e+006Ê:Ê4.504e+006
2.252e+006Ê:Ê3.002e+006
1.501e+006Ê:Ê2.252e+006
7.506e+005Ê:Ê1.501e+006
<0.000e+000Ê:Ê7.506e+005
Fig. 6.4.1e– AC field line plot for inner valve winding
2.175e+007Ê:Ê>2.289e+007
2.060e+007Ê:Ê2.175e+007
1.946e+007Ê:Ê2.060e+007
1.831e+007Ê:Ê1.946e+007
1.717e+007Ê:Ê1.831e+007
6.4.1e 1.602e+007Ê:Ê1.717e+007
1.488e+007Ê:Ê1.602e+007
1.373e+007Ê:Ê1.488e+007
1.259e+007Ê:Ê1.373e+007
1.144e+007Ê:Ê1.259e+007
Core
1.030e+007Ê:Ê1.144e+007
9.156e+006Ê:Ê1.030e+007
StressÊring 8.011e+006Ê:Ê9.156e+006
StressÊring
6.867e+006Ê:Ê8.011e+006
5.722e+006Ê:Ê6.867e+006
3.433e+006Ê:Ê4.578e+006
2.289e+006Ê:Ê3.433e+006
1.144e+006Ê:Ê2.289e+006
<0.000e+000Ê:Ê1.144e+006
Fig. 6.4.1f– DC field plot for inner valve winding
At the start of a DC voltage test, on applying DC voltage, the highest voltage stress is in the oil. With time,
charge migration transfers the stress concentrations to the solid insulation, and within a length of time
and certainly by the end of the 2 hour applied DC withstand test, the majority (approximately 95%) of the
voltage appears across the solid insulation. Considerably more insulation (paper and pressboard) is required
to accommodate the DC stresses than is needed for the equivalent AC insulation structure. Valve windings
have a large number of insulating angle rings and box rings at the winding ends to help distribute DC
6.4.1f
stresses evenly and reduce the voltage gradients at the termination points of solid insulation components
as shown in the cross section view of a typical 300 kVdc level winding arrangement in Fig. 6.4.1d.
CHAPTER
The minimum clearances between windings are generally set to give an intrinsically safe average equivalent
oil stress based on combined AC and DC working voltages. The test conditions are carefully checked and the
insulation structure is verified by a series of Finite Element Method (FEM) studies for AC, DC and polarity
reversal test conditions.
The inter-winding insulation structure typically consists of a series of radially disposed pressboard cylinders
interspersed with oil gaps, whilst the end insulation structure comprises alternate angle rings or box rings,
again interspersed with oil gaps.
20
StressÊlevelÊdueÊto HigherÊstressÊlevelÊon
initialÊcapacitive polarityÊreversalÊdueÊto
distribution existingÊspaceÊcharge
15
ElectricalÊstressÊlevelÊinÊkV/mm
10
StressÊmigratesÊinto Polarity
pressboard reversal
0
0 5 10 15 20ÊxÊ103
ElapsedÊtimeÊinÊseconds
Fig. 6.4.1g– Oil duct stress during polarity reversal testing
The voltage distribution in the transformer insulation differs significantly between AC/transient applications
and DC applications. The AC (power frequency and impulse) distribution is governed by the relative permittivity
6.4.1g
of the component insulation materials, which vary over only a narrow range of:
➙ 2.8 for the structure that makes up the oil gaps
➙ 3.6 for oil impregnated paper
➙ 4.2 for oil impregnated pressboard
➙ 2.2 for the oil
On the other hand, the steady-state DC distribution is governed by the ratio of resistivities of the component
insulation materials, which is of the order of 10:1 at 20°C and can increase ten or one hundredfold as the oil
temperature rises to 100°C or more. The DC stress is also sensitive to changes in the voltage stress, moisture
content, ageing and effects of impurities, in addition to the temperature effects.
The transient performance may be affected by pre-stress conditions and by any stored space charge,
particularly when polarity reversal occurs. Therefore the design of the insulation structure for a converter
transformer is much more complex than for a normal power transformer which is not subject to the DC
component of voltage.
The modulus of the stress distribution in an oil duct within a barrier insulation structure is shown in
Fig. 6.4.1g below for a standard double polarity reversal test. The stress rises to a value related to the
capacitive distribution during the application of the initial voltage but then gradually falls as the space
charge migrates from the oil into the solid pressboard insulation. It then increases again on each reversal
of the applied voltage, and because of the existing space charge, it actually attains a much higher value
(approaching twice) than that of the stress appearing at the initial application of voltage. The insulation
design must take these enhanced stresses into account.

CHAPTER
6.4.2. Harmonic Effects
The HVDC transformer experiences harmonic currents ranging from the 5th up to at least the 49th harmonic
of the supply frequency which will be present in the valve winding. If there are star and delta windings on
the same limb, then there will be some cancellation and less harmonic current in the HV (line) winding. The
levels of harmonics and their orders of magnitude are very dependent on the operating point of the HVDC
system control, particularly if the converter is being used for reactive power control.
An estimate of the worst case has to be made to allow an evaluation of the additional stray and eddy
losses and their effect on the hot spot in the windings. The eddy losses are proportional to the square
of the harmonic frequency whereas the stray losses are proportional to frequency to the power of 0.8.
Care must be taken in the thermal design to ensure that the harmonic effects are taken into account
when determining the local hot spot rises. It is not possible to simulate the actual harmonic flux during
transformer tests at the factory and therefore an allowance for the calculated additional losses due to the
harmonic current content is added for the thermal performance tests. This is not a true representation of
the service condition because the losses cannot be applied or distributed in the winding in the same manner
as will occur in service conditions. Detailed analysis of the thermal characteristics of the transformer
and windings is applied to predict the true hot spot temperature and compensation is required to the
temperature gradient setting of any analog type of winding temperature indicator. A fiber optic sensor
will, however, monitor the temperature at its point of location including the local harmonic effects and
will therefore require no compensation. The correct positioning of fiber optic sensors depends upon the
accuracy of the thermal calculation model and its analysis.
6.4.3. Connections
The connections between discrete parts of the windings out to the bushings, often referred to as cleat
bars, are critical to the operation of the HVDC converter transformer and, as with the winding insulation
structures - because of the presence of DC voltages - require more complex insulation structures than for
an AC power transformer.
The HV (line) winding connection, which could be 1550 kV BIL class for example, may have to emerge from an
inner winding of the assembly of windings. The electrode form presented by the connection is very important
and it is likely that for this application, a tube of about 100 mm diameter would be selected. This tube has
to be insulated with respect to the core and the metallic support framework for the core as it leaves the
LineÊwinding
HVÊleadÊexit
PaperÊcovered
copperÊtube
PressboardÊbox
ringsÊwithÊexitÊsnouts
RegulatingÊwinding
ValveÊwinding
LineÊwinding
PaperÊcovered
copperÊtube
Fig. 6.4.3a– Winding assembly cross section showing external valve winding lead exit

CHAPTER
Transformer ValveÊhallÊwall
tankÊwall
TransformerÊbushingÊturret
DCÊinsulationÊstructure
coronaÊshield Bushing
Transformer current
internalÊcleat transformers
barÊfrom
winding
ValveÊbushing
Pressboard
support
boards
DCÊinsulationÊstructure
(pressboardÊcylinderÊarrangement) CurrentÊtransformer
terminalÊbox
Fig. 6.4.3b– Typical DC structure for DC field transfer between transformer cleat bar and bushing (285 kVdc)
electrostatic protection afforded by the winding, and then must pass over or under other windings, including
the valve winding, before reaching the open oil volume of the transformer tank. Not only will this lead have
to 6.4.3b
be insulated for 1550 kV to earth (core, core frames, etc.) but it will inevitably cut through the various end
insulation structures that have to be adapted to permit passage of the connection without reduction of the
insulation integrity of either winding or the connection. In conjunction with the connection electrode, this is
accomplished by using a series of especially shaped insulating sleeves, snouts and elbows that interlace with
the angle and box rings that make up the winding end insulation structures.
The connection to the outer valve winding could be brought out from the face of the coil and use shaped
insulation pieces, as described above, that will interlace with the pressboard cylinders that form the HVDC
insulation structure between the windings and the tank. As noted previously, the DC stress is concentrated in
MainÊtankÊoil
PressboardÊFaltenbalg
structure BushingÊstem
connector
ValveÊbushing
PaperÊcovered
copperÊtube
Multilam BushingÊturretÊoil
connectors
Fig. 6.4.3c– Typical Faltenbalg layout to provide separation of turret oil from main tank oil

CHAPTER 6.4.3c
Pressboard
shields
Tap
changer
Paper
insulated
cleat bar
Core
Line regulating
windings
Tank base
Fig. 6.4.3d– Valve winding cleat bar for 3-phase, 2-winding transformer, delta connection
6.4.3d Tapchanger
diverter chamber
Regulating
windings
HV line entry
Regulating winding Tapchanger
connections selector
Fig. 6.4.3e– Line side cleat bar for 3-phase, 2-winding transformer, star connection
BACK TO 6.4.3e
CHAPTER
the cellulose insulation material and therefore a sufficient total thickness of paper and pressboard is required
to drop the voltage across the structure without exceeding safe stress criteria.
A further complication of the insulation design is that the windings are subjected to movements during factory
clamping and sizing processes. This means that the insulation structure along with any fixed connection
electrodes has to be designed and ‘toleranced’ to provide a degree of relative movement between its
components during the clamping procedures, while achieving the design intent position upon completion. It
may be necessary to provide some adjustable sections in the lead exit system that can be insulated during
the latter processing stages of the transformer after the main drying and clamping processes are complete.
Notwithstanding the requirement for adjustable components of the lead exit system, the overall lead exit
system design is optimized to minimize the number of joints and to keep these joints distant from areas of
higher stress.
The internal connection of the lead exit system to the valve bushing generally employs a series of pressboard
cylinders to smoothly transfer the DC stress between the field patterns produced by the lead exit and the
pattern generated by the bushing. Since it is desirable to minimize the risk of oil leakage through the DC
bushing into the valve hall, the economics of introducing a Faltenbalg, a pressboard barrier structure to
separate the local oil volume around the end of the bushing from the main transformer tank oil, may be
considered. This seal also allows the bushing connection to be made without exposure of the main insulation
of the transformer during the site erection works.
Currents flowing through the internal busbars produce losses, which require sufficient cooling to maintain
acceptable temperatures. Tubular electrode shields, even if they do not actually carry load current, are
influenced by the surrounding electromagnetic field, generating losses, enhanced and sometimes highly
localized by harmonics. The higher the DC voltage levels employed, the thicker the cellulose insulation is
required to be, however the thermal aspects of the cleat bar still have to be maintained. Where possible the
cleat bar tubes can be arranged to provide for some internal oil thermosyphonic flow to prevent stagnation,
and the current density may be kept low to avoid excessive general and local heating. This is an important
consideration with insulated high current conductors under DC voltage conditions because the resistivity
of the oil impregnated cellulose reduces significantly with rising temperature, whereby if the outer paper
insulation is significantly cooler than the internal paper, higher DC stress will be developed across the outer
insulation layers thereby reducing design margins. For converter transformers designed for DC voltages
higher than 500 kVdc, the lead exit busbars will more usually be of a hybrid paper and pressboard barrier
1.586e+007Ê:Ê>1.669e+007
Corona 1.502e+007Ê:Ê1.586e+007
shield 1.419e+007Ê:Ê1.502e+007
1.335e+007Ê:Ê1.419e+007
1.252e+007Ê:Ê1.335e+007
1.168e+007Ê:Ê1.252e+007
1.085e+007Ê:Ê1.168e+007
1.001e+007Ê:Ê1.085e+007
Pressboard 9.180e+006Ê:Ê1.001e+007
DC
8.345e+006Ê:Ê9.180e+006
insulation
structure 7.511e+006Ê:Ê8.345e+006
6.676e+006Ê:Ê7.511e+006
5.842e+006Ê:Ê6.676e+006
5.007e+006Ê:Ê5.842e+006
DCÊfieldÊlines 4.173e+006Ê:Ê5.007e+006
3.338e+006Ê:Ê4.173e+006
2.504e+006Ê:Ê3.338e+006
1.669e+006Ê:Ê2.504e+006
paperÊcovered 8.345e+005Ê:Ê1.669e+006
copperÊtube <0.000e+000Ê:Ê8.345e+005
Fig. 6.4.3f– FEM analysis of DC stress transfer from paper taped cleat bar to DC bushing structure

CHAPTER
type structure, as the simple paper covered copper tube becomes thermally inefficient because of the
excessive thickness of paper required for the DC test levels.
A typical arrangement of current carrying tubular busbars for HVDC applications is shown in Figs. 6.3.4d
and 6.3.4e.
As stated previously, the bushing to cleat bar interface design requires a suitable insulation structure to enable
satisfactory transfer of the DC stress. The internal field of the bushing is controlled by the strategically located
bushing foils within its condenser. The cleat bar field is controlled by the tubular pressboard sections of the DC
insulation structure. The junction between the two fields is critical, and the special insulating structure must
be designed, manufactured and once assembled, be maintained in its correct position. Complex shapes and
careful positioning of the insulation are required to maintain oil and solid insulation stresses within safe limits
during all conditions of AC & DC withstand and DC polarity reversal.
Finite Element Analysis (FEA) as shown below is required to verify the interface suitability.
6.4.4. Special Manufacturing Features

The AC short-circuit duty of a HVDC converter transformer is the same for any other power
transformer. It is a function of the AC system impedance and the impedance of the transformer,
with the exception that there is no infeed of fault current from the valve side since the converter
operation effectively blocks any flow of fault current from the converter side when operating normally.
The HVDC transformer can however be subject to very high currents flowing in the valve winding during
converter commutation failure conditions. The forces produced by these currents can be greater than for the AC
application and must be catered for in the design. The transformer will in any case be designed with sufficient
short-circuit withstand capability based on the peak fault current level for either of the above conditions.
The windings are subject to increased harmonics from the converter operation and a concentration of flux
at the winding ends. Accordingly, the windings are often manufactured from epoxy-bonded Continuously
Transposed Conductors (CTC) rather than strip conductors to minimize eddy losses and limit the effect of
the harmonic currents by keeping the individual copper strands of the CTC small. The epoxy bonding allows
the use of small copper strands whilst retaining the mechanical strength of the complete bundle of strands
to cater for short-circuit forces under fault conditions.
The rated AC voltage of the valve winding is usually significantly less than that of the line winding and for most
large HVDC transformers, the line and valve windings have very few turns per section. The impulse voltage
distribution control is therefore more usually achieved using specialized inter-shielding techniques; where
floating shield conductors are wound in with the main current carrying conductors close to the winding ends
to increase the series capacitance at the winding ends to give the requisite impulse withstand capability
for the winding. This intershielding technique can be employed in all high voltage transformers, however
for HVDC converter transformers the effects of current harmonics must be taken into consideration when
calculating the losses and determining hot spot temperatures in the non-current carrying inter-shielding
copper strip turns.
Converter transformers will usually have a large tapping range, to accommodate power flow control
requirements of the converter system. Tight tolerances on impedance variation across the tapping
range (nominal impedance +/- 7.5% across the range) and tight impedance tolerance between phases,
between star and delta and between transformer (+/- 2.5%) are required to minimize unbalance and
harmonic generation in the converters to reduce the need for and cost of harmonic filters. Very careful
dimensional and quality control checking procedures are required throughout the manufacturing
process to ensure that these tolerances are achieved, and the tolerances must be verified during
the Factory Acceptance Tests. Traditionally, this close matching of the impedance has been achieved by
designing both the star/star and star/delta transformer valve windings with the same dimensions and
interwinding clearances between line winding and valve winding. This approach means that in the star/delta
units the valve winding to line winding and the valve winding to earth clearances are excessive in relation
to the service and test voltages applied. As HVDC transmission voltage levels rise however, this approach
becomes less economic, and both the star/star and star/delta transformers have to be designed in accordance
with their specific dielectric withstand requirements, rather than dimensional matching.

CHAPTER
As a result of the large tapping range requirements on HVDC converter transformers, the transient voltages
during lightning and switching impulse testing may be high across the regulating winding, from winding to
ground and between individual tapping leads, therefore the On-Load Tapchanger (OLTC) insulation rating may
need to be higher than for a normal power transformer. Typically the insulation withstand of the tapping leads
may be increased by increasing the thickness of tap lead cable insulation employed, or possibly protected by
the use of non-linear metal oxide voltage limiting resistors connected across the regulating winding.
HVDC converter applications impose special needs upon the tapchanger, with valve switching introducing
considerable distortion in the sine wave and harmonic current content may have a de-rating effect on the OLTC.
Because of this distortion of the waveform, the rate of change of current is higher than for normal AC conditions
and this di/dt value has a significant effect on the contact wear within the OLTC and affects its ability to break
current. The di/dt is therefore always included in the OLTC specification for HVDC converter transformers.
6.4.5. Sound Level

During factory tests, it is normally only possible to use a sinusoidal voltage and current supply when
measuring no load and load noise. In service however, the HVDC converter transformer will experience
harmonics. Harmonic content in the load current can have a larger impact on the overall sound level
than expected because higher harmonics can constructively interfere with the power frequency current.
Frequencies of twice the harmonic frequencies and the sum and differences of all the constituent frequencies
may be produced. Thus, the highest contribution to the sound level due to harmonic current occurs when
the product of the load current and the harmonic current reaches a maximum. The resulting audible tones
are at the frequency of the harmonic current plus or minus the power frequency. The effect on the overall
sound level can be significant because the higher frequencies are attenuated less by A-weighting than the
100 or 120 Hz fundamental frequencies.
Harmonics in the excitation voltage may increase or decrease the transformer noise depending on the phase
angle, magnitude and frequency of these harmonics.
HVDC converter transformers are often subjected to a level of DC current due to converter operation
which can be typically in the order of tens of amps. This leads to DC magnetization of a transformer core
and will increase the audible sound level, even at moderate levels of DC magnetization. The reason is that
DC magnetization will add a tone at the power frequency and harmonics at the odd multiples of power
frequency to the sound. In addition, the sound at the normal even order harmonics of the power frequency
will all be increased.
6.4.6. Special Testing Features

The basic requirements for HVDC converter transformers are given in international standards IEC 61378-2
[16] and IEEE C57.129 [18]. These standards define the special tests to be performed on HVDC transformers
in addition to normal routine and type testing, including the following:
➙ DC applied voltage test – this test comprises a 2-hour test at positive polarity to establish a steady-
state field condition in the insulation with the number of partial discharge counts above a threshold
level (2000 pC) monitored over the final 30 minutes of the test.
➙ DC polarity reversal – this comprises 90 minutes at negative polarity, a polarity reversal within a
short time interval followed by 90 minutes at positive polarity, and then a further reversal to hold for
45 minutes at negative polarity. The number of partial discharge counts during the final 30 minutes
at each step is monitored.
➙ AC applied test – this is a 1-hour test at fundamental frequency in conjunction with partial discharge
measurement throughout the period.
The above dielectric test levels are applied to the valve windings with the other windings earthed.
The actual test levels for the above tests are dependent on the converter bridge position of the transformer
and the AC and DC voltages per bridge. Each bridge raises the DC to a new level: for example, to reach 500
kVdc there may be two or more bridges. The calculation of the test levels is outlined in the international
specifications, and is currently the same for IEC and IEEE testing of converter transformers. It is common to
have the delta connected valve winding attached to the lower bridge (1) and the star valve winding to the

CHAPTER
higher bridge (2) with resultant higher test voltages
on the star valve winding.
The partial discharge performance is critical to the
assessment of the HVDC transformer and only a very
low energy and number of partial discharge events
during the course of the specified measuring period
are permitted, although the specifications do allow
for extension of the test in cases where the number
of partial discharge counts is marginally above
the pass criteria. Partial discharge, external to the
transformer may be generated in the environment
of the test laboratory and detailed preparation of the
test plant and surrounding equipment is therefore
essential if multiple extended duration voltage
applications due to spurious external test circuit
discharges are to be avoided.
The test levels appropriate to the valve windings
may not be in line with the normal insulation
coordination rules. For example, the required Fig. 6.4.6a– HVDC transformer being prepared
switching impulse level may be abnormally high for test – 296 MVA, single-phase, two-winding,
in comparison with the assigned lightning impulse 800 kVdc system
level. The actual levels of impulse voltages selected
depend upon the overall station design, location and performance requirements of the surge arresters that
may be used to limit the potential overvoltage levels. Surge arresters are used to limit the voltage between
parts of the converter as well as the voltages to earth. Thus, the insulation on a particular transformer
terminal may be enhanced because it is connected via a surge arrester to a different terminal, which in turn
is connected to ground through another surge arrester. The effect is of the summated protective voltages
of the two arresters in series.
The switching impulse test on the valve windings is an applied test to demonstrate the voltage capability to
ground, as opposed to an induced switching impulse test, which tests between winding turns and sections.
This induced switching impulse is relevant to the line winding but for the valve windings, the voltage between
turns and winding sections capability is proved by the induced power frequency voltage tests and lightning
impulse tests. The switching surge test on the valve winding terminals is performed with the line winding
terminal earthed.
No load, sound level and induced overpotential testing is normally performed by induction
from a low voltage winding because of factory test equipment voltage limitations. The valve
winding voltages on many HVDC transformers tend to be higher than the maximum voltage of the testing
equipment that is available. An alternative, lower voltage supply point must then be provided. This can be
achieved by providing a separate test winding in the transformer, but it is more cost effective to make use
of the regulating winding and simply connect a
temporary test bushing to a suitable tapping point on
the regulating winding and supply between the test
terminal bushing and the permanent neutral bushing.
6.4.7. HVDC Bushings

Valve bushings are critical components of the
HVDC converter transformer and must be designed
to cope with both AC and DC stresses in normal
operation and at enhanced test levels. The bushing
must possess all the relevant performance
characteristics of an AC bushing, but the voltage
distribution within it is also affected by the DC Fig. 6.4.7a– HVDC SF6 insulated bushing being
stresses. The voltage distribution under AC tested

CHAPTER
conditions is controlled by the metallic foils in the bushing condenser body and the voltage is uniformly
distributed by selection of their spacings and heights to achieve the required capacitance grading. The
AC field distributions are not significantly affected by the oil end arrangements. The DC distribution is
controlled by the resistances between foils and to ground. This can be controlled by the inclusion of
higher resistance pressboard insulation in the oil, but the HVDC bushing must be designed in complete
conjunction with the cleat bar arrangements within the transformer as well as the insulation structure
that will interface the two. Because of the mismatch in the relative resistance of the oil and porcelain,
various alternative solutions have been considered to improve the control of the DC stress distribution.
One alternative is to use special porcelain with a higher conductivity that is similar to that of the oil, or the
porcelain itself can be completely removed to reduce changes in conductivity. Finite element analysis is
required to verify interface suitability.
The external part of the valve bushing must have extended creepage distance because of the pollution and
dust attracted by the DC field. The bushing may use a composite glass fiber insulator with silicone rubber
sheds to allow high creepage and take advantage of the reported improvements in DC performance. The
pollution may be minimized by arranging for the transformer bushings to protrude through the valve hall wall;
in this case, a dry type construction is preferred. Resin Impregnated Paper (RIP) bushings for instance can
have shedding machine-cut into the resin itself. At high voltages however, SF6 filled bushings have sufficient
technical advantages to justify their higher cost.
6.4.8. On Load Tapchangers

As mentioned previously, the main HVDC specific features to be catered for in the tapchanger design are the
di/dt and larger than normal tapping ranges. A further consideration is that of maintenance requirements.
Recent developments in tapchanger technology have incorporated the use of vacuum switching capability,
leading to a step reduction in inspection and maintenance requirements for the tapchanger. Whereas the
Fig. 6.4.8a– HVDC converter transformer installed at site, 385 MVA, 222 kVdc
previous conventional oil switching tapchanger diverter switches required inspection at intervals ranging
from 40,000 to 80,000 operations, (some 2-3 years in converter station terms) the newer vacuum switching
diverter technology requires inspection only after some 300,000 operations.
6.4.9. Protection Schemes

As with any high-value item of equipment, the converter transformer requires a comprehensive set
of overlapping protection functions to prevent minor failures from causing catastrophic damage. Both
mechanical and electrical protection functions are used, as described in the following section.

CHAPTER
6.4.9.1. Mechanical Protection
Mechanical protection of converter transformers is basically the same as for any large power transformer
and consists of:
➙ Oil level switches in the oil expansion vessels, to protect the transformer and tapchangers
from loss of oil
➙ Gas-in-oil relays in the main pipework between tank and expansion vessels, to detect any production
of gas or surge of oil within the main tank or tapchanger
➙ Pressure relief devices mounted on the main tank or tapchanger, to protect in case of overpressure
during an internal fault.
Converter transformers can be subject to a large number of tapchanger operations and it is desirable, where
a separate selector chamber with a barrier board is employed, to provide gas and oil-surge relay protection
on the tapchanger selector chamber as well as the normal Buchholz protection of the transformer main oil
volumes. Furthermore, it may be prudent and/or necessary to fit temperature monitoring to the tapchanger
diverter chamber, for two reasons:
➙ To restrict tapchanger operation when the diverter chamber oil is at very low temperature, usually
< -40°C. Under these conditions the oil temperature must be increased before tap changing can be
permitted. This is usually achieved by no load energization of the transformer, utilizing the core loss
to raise the oil temperature
➙ To restrict tapchanger operation when the diverter oil temperature is too high. If the tapchanger is
operated continuously, then the losses associated with the normal operation of the transition resistors
within the tapchanger diverter switch can raise the diverter chamber oil temperature considerably.
Under normal service conditions, the number of tap changes and the time interval between subsequent
changes is sufficient to allow dissipation of these losses and for the oil temperature to be maintained
at, or just above, that of the bulk oil of the transformer. Where tap changing can occur on a continuous
or extended basis, the operation of the tapchanger must be restricted when the diverter chamber oil
temperature rises above the tapchanger service limit (usually 115°C).
6.4.9.2. Electrical / Digital Protection Schemes

In modern HVDC systems, use of the 12-pulse bridge configuration is almost universal, and since the earth
connection on the valve side of the transformer is made via the converter bridge, the converter transformer
windings can be exposed to a continuous DC voltage. Thus, although the protection of a converter transformer
is very similar to that of a conventional power transformer, the windings may be exposed to DC currents. A
small DC current of a few amperes may continuously flow through the valve windings due to small variations
between phases of the converter firing angle or unbalances in the AC system. Additionally, under certain fault
scenarios, large DC currents may flow for short periods of time (up to a few seconds). Special precautions
therefore have to be made in the design of the Current Transformers (CTs) used for protection and measuring
on the valve winding terminal. If conventional CTs were used, any DC current would saturate the CTs causing
mal-operation of the protection. Specially designed CTs are therefore used which include an air gap within the
magnetic core circuit in order to minimize the flux density and provide DC ‘immunity’. Current transformers
are usually fitted into the line winding bushing turret to provide outputs for use with zone protection schemes.
With regard to general temperature monitoring of converter transformer windings, as in normal power
transformers, this is usually monitored using a thermal imaging device, which monitors the top oil temperature
in the transformer and adds an additional temperature. As outlined previously, although the Factory
Acceptance Testing includes additional current during the temperature rise testing to simulate the additional
losses expected in service due to harmonics, the actual in service temperature attained by the windings can
be best monitored by using fiber optic sensors placed in the winding at the time of manufacture to achieve
a direct measurement of winding temperature. Advances in fiber optic sensor manufacture have resulted in
increasingly robust installations of these sensors, which are usually fitted in pairs to provide duplication to
cover any loss of sensor function during the transformer manufacturing process.
Since HVDC converter schemes tend to be strategically important to the customer, thought should be given
to the use of on-line monitoring systems for both normal service monitoring and fault evolution within the
transformer, such as:

CHAPTER
Fig. 6.4.9a– HVDC converter transformer, 3-phase, 385 MVA 400/96 kV unit, arranged for shipping
➙ On-line service condition monitoring – continuous monitoring of load current, transformer

temperatures, bushing voltage, number of tapchanger operations, operating condition of cooling
pumps and fans etc., all of which can be collected and monitored at a single point to assess the current
condition of the transformer and assess the life usage.
➙ On-line gas-in-oil analysis – allowing on-line assessment of the level of combustible gases dissolved
in the transformer oil, facilitating early indication and diagnosis of internal transformer faults.
➙ On-line tapchanger monitoring – continuous monitoring and analysis of the power consumption of
the tapchanger drive, as well as vibration and acoustic monitoring to aid fault diagnosis and wear rate
assessment of the various tapchanger contacts.
➙ On-line bushing monitoring – on-line assessment of the bushing capacitance and dielectric dissipation factor.
6.5. DC SMOOTHING REACTOR

The smoothing reactor is a reactor connected in series between the DC valves and the DC circuit. It can be
located at the high voltage or low voltage part of the DC circuit.
Smoothing reactors can be either: air-insulated (or coreless) or iron-shielded and oil-filled. Most smoothing
reactors today are air insulated, see Fig. 3.4l.
6.5.1. The Purpose of a DC Smoothing Reactor in a HVDC Transmission Scheme

The classical literature in HVDC transmission systems assumes, for the sake of simplification, that the DC
current is perfectly smoothed by an infinite reactance - as presented in Chapter 2, Fig. 2.2.5a where an
ideal 12-pulse bridge is presented.
In reality, HVDC transmission schemes require the use of smoothing reactors to fulfill different purposes:
1. Elimination of discontinuous current under steady-state operating conditions
2. The smoothing reactor is part of the DC harmonic filtering scheme that limits the steady-state harmonic
current injection into the DC circuit.
In pure cable transmission the DC filters are normally not required, because the DC cable itself acts as a
high-frequency filter. Normally DC cables for conventional transmissions are robust enough to cope with
any residual steady harmonic current.
In pure DC overhead transmission or mix of DC cables and overhead transmission line systems the DC
filters are often included because of a more stringent harmonic current injection limit as presented in
section 4.5.

CHAPTER
3. To limit commutation failure currents: the smoothing reactor is required to be sized considering the
thermal effect of the resultant commutation failure current on the thyristor valves due to faults through
the valves and inverter AC network.
4. To limit the magnitude of transient current surges into the DC station due to lightning strikes on DC
transmission lines.
The classical HVDC literature presents the importance of smoothing reactors in HVDC schemes [19],[20].
In general, there is no discussion of the applicability of the smoothing reactors for transmission systems.
However other works discuss the theoretical and practical advantages and disadvantages of allocating
smoothing reactors in back-to-back schemes [21], [22], [23].
Many back-to-back HVDC schemes are in service today. Most of the schemes supplied by GE Vernova since
1989 have been operating without smoothing reactors as per the projects listed in section 1.3.
6.5.2. The Design of HVDC Back-to-Back Schemes without Smoothing Reactors

The absence of a DC smoothing reactor can be justified if the following criteria can be satisfied:
➙ Elimination of discontinuous current under steady-state operating conditions
➙ Acceptable harmonic performance of the scheme
➙ Peak current due to commutation failure is acceptably low
Because the DC lines are absent in back-to-back systems, it is not necessary to evaluate the smoothing
reactor regarding transient current surges due to lightning strikes on DC lines (see section 6.5.1).
6.5.3. Eliminating Discontinuous Current under Steady-State Operating Conditions

Operation of the HVDC converter in discontinuous current mode is undesirable and in some cases may be
dangerous. It is therefore necessary to understand and predict the operating conditions which can lead to
discontinuous current, as described below.
DC current (p.u.)
Id ripple
Id (p.u.)
Id total Id oﬀset
0.5
0
0 5 10 15 20
Fig. 6.5.3a– d(t), Idtotal , Idripple , Idoffset t (ms)

CHAPTER
6.5.3.1. Modeling the DC Current Ripple in Time and Frequency Domains
The DC current consists of two parts as indicated in the Fig. 6.5.3a and in Eqn 6.5a:
Idtotal = Idripple + Idoffset [Eqn. 6.5a]
Where:
Idtotal = Mean value of the DC current obtained assuming hypothetical infinite inductance on the DC side
of the circuit
Idripple = Mean value of the DC ripple
Idoffset = Part of the Id current not related to the DC current ripple
The existence of DC current discontinuities depends on the operating conditions, converter configurations,
system frequencies, etc. Normally Idtotal at minimum power is selected to give a positive value of Idoffset.
The classical expression for calculating the maximum Id ripple due to the influence of a single
12-pulse converter is found in Kimbark, E. W., 1971, ‘Direct Current Transmission’ [19]:
3 2 ⋅ E LL max pk
Idmin > Idripple = ⋅ ⋅ (0.023⋅ sin α max ) (rectifier or inverter) [Eqn. 6.5b]
π ω ⋅ L T min
Where:
a ≤ w t ≤ a + 30° = Interval where the expression is defined
Idripple = Maximum operating DC current ripple
ELL max pk = Maximum valve winding voltage, peak, 6-pulse converter
LT min = 4 × (Lcr +Lci) = Minimum value of the total inductance of the DC circuit
Lcr and Lci = 6-pulse commutating reactances of the rectifier and inverter respectively
The expression above is calculated by assuming no overlap. For this particular analysis, where Id is at its
minimum level, the overlap is normally very small and the approximation is quite acceptable.
Since for back-to-back converters the influence of the second converter cannot be disregarded, Eqn 6.5b
can be extended to:
[Eqn. 6.5c]
Where:
a r = Firing angle of the rectifier
a i = Firing angle of the inverter
On the inverter side, a i is defined as:
a i = p – (µi + gi) [Eqn. 6.5d]
Where:
μ i = Overlap angle of the inverter
g i = Gamma angle of the inverter
Since the overlap angle is assumed to be zero, sin (a i) = sin (p – g i) = sin (g i), Eqn 6.5c can be rewritten as:
[Eqn. 6.5e]
The equations presented until now describe the mean value of the DC current but are not sufficient to
explain the nature of the current ripple in HVDC schemes in the time and frequency domains.
The time domain contribution due to a single 6-pulse converter is calculated in [19]. For back-
to-back 12-pulse converters the ripple can be split into two contributions: one due to the
rectifier and one due to the inverter. Each individual contribution can be expressed in the range
a ≤ w t ≤ a + 30° using the following expressions for the rectifier and inverter respectively:

CHAPTER
 π   π  π   π   π   
3 1 ⋅ 4 ⋅ cos  ⋅ sin t⋅ ω r −   − 4 ⋅ cos  ⋅ sinα r −     
idr(t) = π ⋅  ⋅ ELL,r⋅  ⋅  6   12  12   12   12    [Eqn. 6.5f]
π LT  
 ωr 
 π   π  π   π   π  
3 1 ⋅ 4 ⋅ cos  ⋅ sin t⋅ ω i −   − 4 ⋅ cos  ⋅ sinγ i −     
idi(t) = π ⋅ ⋅ ELL,i⋅  ⋅  6   12  12   12   12    [Eqn. 6.5g]
π LT  
 ωi 
Eqn 6.5f and Eqn 6.5g represent the individual contribution to idripple(t) considering a single pulse from each
converter side. Switching functions are required to completely describe the idripple(t) for any time period.
This is presented in Appendix A6.5.
Fig. 6.5.3b shows an example of the idripple(t) and related DC quantities based on the time domain switching
functions described in Appendix A6.5. This example is for a back-to-back system connecting a 50 Hz system
αr+θr αi+θi
2. π fr 2. π fi
DC voltage pole to
ground, vt(t)
Ideal 12-pulse DC
voltage, rectifier (pole to
ground), vr(t)
Vphase YY+Vphase YD
on DC side of the
converter transformer
DC ripple current due to

rectifier, idr(t)
DC ripple (total),
idr(t) + idi(t)
DC ripple current due to

inverter, idi(t)
Vphase YY+Vphase YD
on the DC side of the
Ideal 12-pulse DC
voltage, inverter
(ground to pole), vi(t)
0 0.01 0.02 0.03
Time (s) Fig. 6.5.3b– DC quantities related to the

idripple(t) calculation

CHAPTER
idr(t)Ê+Êidi(t)
RHV RHV
Rectifier Inverter
VL rÊ∠Êθr 4Ê.ÊLcr 2 2 4Ê.ÊLci VL iÊ∠Êθi
VL rÊ/ÊVvr vr(t)p-gÊ vi(t)g-pÊ VviÊ/ÊVL i
VSrÊ∠Êθ vt(t)p-gÊ VSiÊ∠Êθ
IdealÊtransformer IdealÊtransformer
InductancesÊofÊtransformer
andÊvalveÊreactorÊ(LcrÊandÊLci),
movedÊintoÊDCÊside
ForÊcalculatingÊidr(t)ÊandÊidi(t)ÊtheÊDCÊtotalÊresistance,ÊRHV,ÊisÊassumedÊzero.
idr(t)ÊandÊidi(t)ÊdefinedÊinÊEqnÊ6.4fÊandÊEqnÊ6.4gÊrespectively.
vt(t)ÊdefinedÊinÊEqnÊ6.4h.
Fig. 6.5.3c– DC circuit related to the electrical quantities of Fig. 6.5.3b
6.5.3c
to a 60 Hz system. The plots presented are not to scale and are presented for facilitating the understanding
of the relationship between different electrical quantities.
Fig. 6.5.3c presents a diagram where the electrical quantities of Fig. 6.5.3b are identified in the circuit.
The rectifier is at 50 Hz, the inverter at 60 Hz. The complete set of switching functions for simulating the
conditions presented in this figure is shown in Appendix A6.5. In the figure, qr and qi are the AC system
phase voltage angles of the equivalent independent sources at the rectifier and the inverter. (qr – qi) can
be any number and will influence the id(t) wave shape as shown in Fig. 6.5.3e.
In Fig. 6.5.3b and Fig. 6.5.3c, the DC voltage pole to ground is calculated by the following expression:
Lci Lcr
vt ( t )p−g = vr ( t )p−g ⋅ − vi( t )g− p ⋅ [Eqn. 6.5h]
Lcr + Lci Lcr + Lci
Where:
vr(t)p–g = Ideal 12-pulse DC voltage, rectifier (pole to ground)
vi(t)g–p = Ideal 12-pulse DC voltage, inverter (ground to pole)
Based on equations presented in Appendix A6.5 and the graphical analysis of Fig. 6.5.3b it is possible to
draw the following conclusions:
➙ In Fig.6.5.3b and Fig.6.5.3c, qr and qi are the AC system phase voltage angles of the equivalent
independent sources at the rectifier and the inverter. (qr – qi) can be any number and will influence
the id(t) wave shape as shown later in Fig. 6.5.3e.
➙ The time domain analysis allows an immediate frequency domain evaluation as presented in Fig.
6.5.3d. A typical Fast Fourier Transform (FFT) of the idripple(t) is presented below for the (50/60) Hz case.
The contribution from the 50 Hz side has frequencies at 600 Hz, 1200 Hz etc. The contribution from the
60 Hz side has frequencies at 720 Hz, 1440 Hz etc.

CHAPTER
The following conclusions related to the mean value and maximum/minimum amplitude of idripple(t)can be
derived from Fig. 6.5.3b and Fig. 6.5.3d.
➙ For the 50/60 Hz case the harmonic content due to id ripple(t) due to one AC side appears in
frequencies that are different from the harmonic contribution due to the other AC side.
➙ For the 50/60 Hz converters the difference of AC voltage system angles, indicated in Fig. 6.5.3b as
(qr – qi), will impact on idripple(t) maximum and minimum amplitudes. The (qr – qi) values that will
generate the extreme values of idripple(t) are not so easy to predict. This is illustrated in Fig. 6.5.3d.
DCÊcurrentÊrippleÊFFTÊofÊtotalÊcurrent
150
120
AmpereÊ(rms)
90
600 720
60
30
0
400 560 720 880 1040 1200 1360 1520 1680 1840 2000
FrequencyÊinÊHz
Fig. 6.5.3d– Typical frequency spectrum for idripple(t)
6.5.3d
200
200
0
0.01 0.015 0.02 0.025 0.03 0.035
TimeÊ(s)
TotalÊDCrippleÊatÊminimumÊld MaximumÊldrippleÊdueÊtoÊrectifierÊ(orÊinverter),ÊatÊldmin
IdminÊ(meanÊvalue) MaximumÊldrippleÊdueÊtoÊtheÊinverterÊ(orÊrectifier),ÊatÊldmin
Fig. 6.5.3e– Typical idripple(t) waveshape for 50/60 Hz systems

CHAPTER
➙ For converters with identical frequencies, the AC voltage system angles will impact the harmonics
on idripple(t) as follows:
– For (qr – qi) = 0, idripple(t) is maximum
– For (qr – qi) = 7.5°, the 24th harmonic will be minimum or zero
– For (qr – qi) = 15°, the 12th harmonic will be minimum or zero
This can also be intuitively understood by the following analysis:
➙ For (qr – qi) = 0, if we assume perfect symmetry between the two sides, the harmonic spectra
generated by the rectifier and inverter will have the same phase angle and will add in phase.
➙ For (qr – qi) = 7.5°, if we assume perfect symmetry between the two sides, the harmonic spectrum
generated by the rectifier will be phase shifted by 7.5°. For the 24th harmonic current, the period
is 360°/24 = 15°. So if the rectifier and inverter contributions are phase shifted by 7.5° at the 24th
harmonic, they will cancel out when added.
➙ For (qr - qi) = 15°, if we assume perfect symmetry between the two sides, the harmonic spectrum
generated by the rectifier will be phase shifted by 15°. For the 12th harmonic current, the period
is 360°/12 = 30°. So if the rectifier and inverter contributions are phase shifted by 15° at the 12th
harmonic, they will cancel out when added.
Additional notes:
1. The AC system voltage assumed at the commutation busbar for this analysis is the positive sequence
fundamental frequency voltage. This is a reasonable assumption for the dominant harmonics since the
harmonic filter design will ensure that the harmonic content at the commutation busbar is minimal.
2. As an example, assuming that a + µ is limited to 45° during steady-state operation, the equations above
also represent a good approximation of the maximum harmonic ripple along the full power range. This is
based on the well-known behavior of the DC side ideal harmonic voltage source that varies as a function
IdealÊVdcÊ12-pulseÊharmonicÊ(p.u.)
alphaÊ=Ê43¡
alphaÊ=Ê30¡
10 alphaÊ=Ê20¡
alphaÊ=Ê10¡
alphaÊ=ÊÊ4¡
VdcÊ(12thÊharmonic)Êp.u.
0
0 10 20 30 40 50 60
OverlapÊ(¡)
Fig. 6.5.3f– 12 harmonic direct voltage, p.u. of 3/p ⋅ ELLpeak = Vdi0 (direct no load voltage)
th
BACK TO 358 |
6.5.3f
CHAPTER
of the firing angle and overlap as indicated in [19]. As an example, see the Fig. 6.5.3f below, where
the 12th harmonic of the ideal voltage DC source is calculated as a function of the overlap variation
considering different alpha values. Notice that the case with maximum harmonic occurs for the case
with minimum overlap and maximum firing angle.
6.5.3.2. Maximum Idripple Calculation for Defining Minimum Id

The maximum Idripple values are calculated using Eqn 6.5e, using sets of different of operating conditions
and manufacturing tolerances applied to the minimum power level. An example is presented in Table 6.5a.
Idripple Idripple
contribution contribution
Operating Power level Idripple Idmin Idoffset
from from
conditions % (A) (A) (A)
rectifier inverter
(A) (A)
A 10 63 81 144 278 134
B 10 101 126 226 321 95
C 10 101 126 227 315 88
D 10 21 36 57 257 200
E 10 71 89 160 315 155
F 10 56 75 132 257 125
G 10 21 36 57 252 195
Table 6.5a– Idripple calculated at Idmin , 60 Hz rectifier, 50 Hz inverter
Idripple Idripple
contribution contribution
Operating Power level Idripple Idmin Idoffset
from from
conditions % (A) (A) (A)
rectifier inverter
(A) (A)
A 10 76 71 147 278 131
B 10 119 106 225 321 96
C 10 119 106 225 315 90
D 10 25 36 62 257 195
E 10 83 75 158 315 157
F 10 68 68 136 257 121
G 10 25 36 62 252 190
Table 6.5b– Idripple calculated at Idmin , 50 Hz rectifier, 60 Hz inverter
The operating conditions A to G are calculated for the minimum power level, using extreme combinations of
measurement and equipment tolerances.
A PSCAD/EMTDC case is presented below in order to show the consistency of the results for the critical
case C of Table 6.5b, assuming 50 Hz as a rectifier and the 60 Hz as an inverter.
CHAPTER
0.50Ê
0.40Ê
0.30Ê
kA
0.20Ê
0.10Ê
0.00Ê
Time(s) 2.96Ê 2.97Ê 2.98Ê 2.99Ê 3.00Ê
CurrentÊorder DCÊcurrent
Fig. 6.5.3g– Id(t) current corresponding to case C in Table 6.5b, for Id = 315 A (50 Hz rectifier, 60 Hz inverter)
6.5.4. AC Harmonic Performance

The principles of harmonic current generation arising from the conversion of electrical power from AC to DC
6.5.3g
and vice-versa are well documented in Chapter 4. Most conventional analytical techniques assume that the
DC side impedance is extremely large, eliminating ripple on the direct current and effectively de-coupling
the rectifier and inverter stations. With no DC system impedance there is a considerable ripple on the direct
current. The resultant close coupling of the rectifier and inverter requires analytical techniques which do
not make simple assumptions about AC and DC system impedances. For this reason, all transmission HVDC
schemes and many back to back schemes include DC smoothing reactors to increase the impedance of the
DC system.
The harmonics on the DC side produced by one converter are integer multiples of the AC system frequency
of that converter. These harmonic currents flow to the other converter and are modulated at the frequency
of the AC system to which that converter is connected. Thus, at each terminal, there is a spectrum of
harmonic currents (IL) due to the local converter plus a separate spectrum of harmonic currents (IR) arising
from the remote converter.
If the two AC systems were to be completely synchronized, the resultant current for the nth harmonic at one
converter would be:
IT = IL ⋅ sin(wnt + fL) + IR ⋅ sin(wnt + fR) [Eqn. 6.5i]
Where:
IT = Total harmonic current in each terminal
IL ⋅ sin(wnt + fL) = Harmonic current due to the local converter
IR ⋅ sin(wnt + fR) = Harmonic current due to the remote converter
Normally the two systems operate asynchronously and the analysis of the harmonic conditions becomes
more complex.

CHAPTER
Consider the effect of the DC harmonics pw1 and pw2:
➙ At the local converter, the AC side harmonics associated with pw1 will be w1 ± pw1, which are integer
odd harmonics.
➙ However the harmonics caused by the demodulation of pw2 will be w1 ± pw2, which are not integer
harmonics of w1.
If, for example, we represent the (p + 1)th harmonic from the local converter as wL = (1 + p)w1 with magnitude
IL and the quasi (p + 1)th harmonic due to the remote converter as wR = w1 + pw2 with magnitude IR, Eqn 6.5i
for the (p + 1)th harmonic can be written as:
IT = IL ⋅ sin(wLt + fL) + IR ⋅ sin(wRt + fR) [Eqn. 6.5j]
This technique is essential in any design of a HVDC back-to-back system. This is because the coupling between
the ends needs to be considered even when a smoothing reactor is supplied. Typical smoothing reactor sizes
are in the range of 10 mH to 200 mH. Sensitivity studies have shown that adding smoothing reactors provides
some reduction in the cross-modulation of harmonic components; however they can be compensated in most
of the situations by the design of AC filters considering the correct amount of cross-modulated harmonics.
A sensitivity analysis can be done to show the mitigation effect of smoothing reactors on the reduction of
cross-modulation harmonic components.
The Inter-Harmonic Reduction Factor, IHRF is defined as:
4 ⋅ ( Lcr + Lci )
IHRF = [Eqn. 6.5k]
4 ⋅ ( Lcr + Lci ) + Ldr + Ldi
Where:
LT min = 4 ⋅ (Lcr + Lci) = Minimum value of the total inductance of the DC circuit
Lcr and Lci = 6-pulse commutating reactances of the rectifier and inverter respectively
Ldr = Smoothing reactor at the rectifier
Ldi = Smoothing reactor at the inverter
The IHRF can be plotted as function of the total smoothing reactance presented in the circuit, as presented
in Fig. 6.5.4a.
Fig. 6.5.4a shows that for reducing the cross-modulation harmonics by a factor of five, the total smoothing
reactance (Ldr + Ldi) in the circuit needs to be four times bigger than LT min.
This graphic explains the general view regarding the necessity to calculate inter-harmonics in HVDC
converters, since the inclusion of a typical small smoothing reactor does not really mitigate these harmonics
to a level that can be completely disregarded during the AC filter design.
1.5
1
1
ReductionÊfactor
0.5
0.5
0.2
0
0 1 2 3 4
TotalÊsmoothingÊreactorÊreactanceÊ(rectifierÊ+Êinverter)ÊinÊp.u.ÊofÊLTmin
Fig. 6.5.4a– Inter-harmonic reduction factor

CHAPTER
6.5.5. Peak DC Current due to Commutation Failure
At the inverter end, the valves could fail to commutate due to a step reduction in the AC system voltage.
In back-to-back schemes, since there is no DC cable or DC line to be discharged, the commutation failure
current contribution is only from the rectifier and persists until the rectifier control responds to reduce
the commutation failure current. The decision of whether smoothing reactor is required, or not, is made
by considering the thermal effect of the resultant commutation failure current.
Time domain studies are performed to verify the worst-case amplitude of the commutation failure current
assuming that no additional DC inductances are present.
The criterion for accepting a design without smoothing reactors is related to how the thyristor thermal protection
acts during the commutation failure. Typically, if the studies indicate that for the design reference condition of 1.0
p.u. load at the nominal inlet temperature, the thyristor thermal protection will not be invoked during the transient
periods, the converter design without smoothing reactor is acceptable.
6.6. MEASURING TRANSDUCERS

In a HVDC scheme, a variety of devices are used to measure the analog signals required for monitoring and
controlling the scheme. A combination of current transducers and voltage transducers is employed. The basic
performance and the required duty of the key devices are outlined in this section.
A number of measurement devices located in the HV plant are used to derive the appropriate measurement
signals for use by the control system for implementing the scheme. The measurement points and devices
used for a typical single 12-pulse converter transmission scheme are illustrated in Fig. 6.6a.
In Fig. 6.6a, the transmitted DC power is calculated from Vdc and Idnc, while the reactive and active AC
power is calculated from Vac (Vlw CVT) and Ilw. Frequency is derived from the zero crossings of voltages
measured by Vlw CVT.
A combination of high accuracy voltage and current transformers is required for the measurement of the plant
signals required for controlling and monitoring the operation of the scheme. These measurement devices
are located in the HV plant itself and usually convert the measured signal to a level and form suitable for
subsequent processing. Typically, CT signals are scaled down to 1 A or 5 A, while VT signals are scaled down to
110 V/√3 or a voltage in the range 0 to 10 V. Any major scaling of the measured signal is usually implemented
on the measurement device itself and/or additional associated equipment based on the measurement device.
The measurement transducers also afford some degree of electrical (galvanic) isolation between the raw HV
plant signals and the electrically clean environment of the control cubicles.
The DC power is determined from the product of the DC current measurement (see section 6.6.1) and
the DC voltage measurement (see section 6.6.2). The transducers required for active and reactive power
measurement are outlined in section 6.6.3. Additional measurement transducers are also required for
various monitoring and/or protection functions of the LCC scheme as outlined in section 6.6.4.
Traditionally, measurement devices produced an analog output at low voltage which was fed directly into
the control system via analog-digital converters. Increasingly, however, the trend is for the signal to be
digitized at the transducer itself and transmitted via a protocol based on the IEC 61850 series of standards.
This approach greatly reduces the amount of auxiliary wiring that must be fed into the control system.
6.6.1. DC Current Measurement

The measurement of the DC current in a scheme is carried out using a DC Current Transducer (DCCT).
Traditionally, Zero Flux Current Transformers (ZFCTs) are used to accomplish this, though optical current
transformers (OCT) are also increasingly used, especially in applications where high insulation levels are required.
For either type, the DCCT implementation is such that the HV plant current signal is converted to a low voltage
signal that has a magnitude in the range 0 to 12 V, compatible with the inputs to the control system.
DCCTs can achieve measurement accuracies of 0.2% or better and this accuracy is maintainable over a very
wide frequency range.
The DCCT is usually designed to have multiple outputs, which can be used to feed various control and

CHAPTER
Line
isolation
T1 Idl
llw1 lvw1
Vdl
Vlw
CVD T2
llw2 lvw2
AC
isolation Vdn Idne
Idnc
Vlw Idnc
CVT
Vdn
FilterÊbanks T1
llw1 lvw1
Vlw
CVD T2
llw2 lvw2 Neutral
Idl isolation
Vdl
ÊÊVlwÊCVT:Ê LineÊwindingÊvoltageÊCVT
VoltageÊdivider ÊÊVlwÊCVD:ÊÊ LineÊwindingÊvoltageÊCVD
ÊÊIvw1:ÊÊ StarÊvalveÊwindingÊcurrent
ÊÊIvw2:ÊÊ DeltaÊvalveÊwindingÊcurrent
CVD ÊÊIIw1:ÊÊ LineÊwindingÊcurrentÊ(T1)
FilterÊbanks ÊÊIIw2:ÊÊ LineÊwindingÊcurrentÊ(T2)
ÊÊVdl:ÊÊ LineÊwindingÊDC
CVT
ÊÊVdn:ÊÊ NeutralÊvoltageÊDC
ÊÊIdnc:ÊÊ HVDCÊneutralÊconverterÊcurrent
ZFCT/OCT ÊÊIdne:ÊÊ HVDCÊneutralÊearthÊcurrent
ÊÊIdl:ÊÊ HVDCÊlineÊcurrent
CT
Fig. 6.6a– Measurement points and devices for a typical HVDC scheme
6.6a
protection functions. In such cases, it is traditional to specify the outputs to have varied accuracies to suit
the intended application.
6.6.2. DC Voltage Measurement

Direct voltage on the HVDC busbar is measured by means of a DC resistive Voltage Divider (VD). The typical
locations of the VDs on an LCC scheme are shown in Fig. 6.6a and are labeled Vdl and Vdn.
The typical arrangement of a voltage divider is shown in Fig. 6.6.2a. The DC voltage divider is configured
such that it will step down the high voltage at its HV terminals to a much smaller output voltage
of typically up to 10 V, which is then used to supply a defined buffer amplifier load. The HV resistor part is
normally located in the DC yard while the low voltage divider connection box is located in a control interface

CHAPTER
cubicle. The HV resistor part of the VD is connected to the divider connection box via a purpose-designed
triaxial cable. The HV resistor, the triaxial cable and the divider connection box together form the voltage divider.
Some more modern voltage dividers digitize the measured signal directly at the base of the transducer and
transmit it in encoded form according to the IEC 61850 series of standards.
HV PrimaryÊcircuit/
terminal HVDCÊdivider
NominalÊoutput
TriaxialÊcable DividerÊconnectionÊbox ExternalÊburden
BufferÊamplifierÊ1
Ramp1 Camp1
BufferÊamplifierÊ2
OV
OV
EarthÊpotential Ramp2 Camp2
Fig. 6.6.2a– Typical arrangement of a voltage divider
The main resistor is a freestanding unit, which is insulated from earth and would additionally have a spark
6.6.2a gap and a non-linear resistor which connects the base of the resistor to earth.
The output is protected by suitable overvoltage limiting devices.
Measurement accuracies of 0.2% or better are normally achievable as standard. Any calibration checks must
include the correct length and type of triaxial cable as will be used in the actual scheme.
6.6.3. AC Current and Voltage Measurement on the Converter AC Connections

The converter control and protection system requires accurate measurements of both voltage and current
on the AC side of the converter, on both the Line winding and Valve winding. These measurements are
described below.
6.6.3.1. AC Current Measurements on Converter Transformer

For the correct implementation of the HVDC scheme, accurate measurements of the HV line winding
current (Ilw) and the valve winding current (Ivw) are required. The valve winding alternating currents are
measured using air-gapped current transformers that are suitably designed not to saturate when subjected

CHAPTER
to the DC offsets that are present in the currents. As an alternative, Optical Current Transformers (OCT)
may be used. These current transformers are usually located in the transformer bushing turret and are
also subject to the requirements in section 6.6.4.1.
6.6.3.2. AC Voltage Measurement on the AC Station Bus

Alternating voltage for conventional uses, such as calculating reactive power, is measured by means of a
Capacitor Voltage Transformer (Vlw CVT) located on the AC station bus (see Fig. 6.6a).
A typical line diagram of a Capacitor Voltage Transformer (CVT) is shown in Fig. 6.6.3a.
The magnetic intermediate transformer must be furnished with electrostatic shielding between the high
and low voltage windings to avoid transfer of transients from the high voltage side to the low voltage
windings.
The lower part of the CVT unit is fitted onto a metal base box containing the transformer, the series reactor,
the protective gap or varistors and the grounding switch.
The transformers are located in a hermetically sealed enclosure filled with mineral oil and the leads of the
secondary windings brought out to a waterproof terminal box.
Each CVT is formed by sections, depending on its voltage rating. Each capacitor section includes an
expansion chamber to allow changes in the fluid volume under varying temperature conditions and to
allow the fluid to escape outside the porcelain should there be an excessive pressure build-up due to
insulation break down.
CapacitorÊvoltageÊdivider
A HighÊvoltageÊterminal A
ElectromagneticÊunit
HighÊvoltage
capacitor C1
SeriesÊ/
Stray compensation
capacitance,ÊCs inductance
IntermediateÊvoltage
terminal
Intermediate Secondary
transformers windingÊ(s)
IntermediateÊvoltage
capacitor C1
Stray
inductance,ÊLs
Earth
terminal
LowÊvoltage
N
terminal
Fig. 6.6.3a– Typical line diagram of a capacitor voltage transformer

CHAPTER
6.6.3a
The CVT must be constructed to withstand without damage, the applicable mechanical, electrical and
thermal effects of an external short-circuit at the secondary windings and must be free of external corona
at the highest voltage for equipment.
The protective gap or device must not operate at a voltage lower than twice-normal operating voltage.
The CVT is essentially a tuned device with good voltage magnitude ratio accuracy (of typically 0.2%) at
rated frequency.
6.6.3.3. AC Voltage Measurement on the HV AC Bus of Converter Transformer

The line voltage can also be measured using a Capacitive Voltage Divider (CVD), which offers a wide
bandwidth frequency response.
Voltage measurements used by the HVDC control system for determination of firing and extinction
angles are obtained from the converter transformer line winding bushings. The bushing tap is effectively
a capacitive divider made up of the inherent bushing capacitance and additional low voltage capacitors.
The arrangement for deriving the required signals is as shown in Fig. 6.6.3b.
The converter transformer HV bushing incorporates concentric metal foil cylinders into its insulation. The
capacitance between these foils ensures correct voltage gradient across the insulation of the bushing.
However, the capacitance also provides a convenient means by which the bushing AC voltage can be
measured. To allow this, a tap connection from one of the inner foils is brought out and is terminated in a
housing mounted on the side of the transformer. This housing is identified as the bushing interconnection
box in Fig. 6.6.3b.
The bushing interconnection box components effectively implement capacitance matching with the
transformer bushing. The resultant capacitive voltage divider arrangement effectively reduces the source
voltage down to a low voltage signal, which is then passed through additional calibration stages that ensure
that both the magnitude and bandwidth performance are as required.
Capacitive voltage dividers can offer an essentially flat response in the frequency range 20 Hz to 20 kHz.
87,Ê87GÊ&Ê51
50Ê&
51N
MainÊandÊbackÊupÊbusbarÊdifferentialÊprotection
S1 S2 S2 S1
ConverterÊcontrolÊandÊprotection
P1 P2 P2 P1
E A M H G B
P1 S1
K
P2 S2
87,Ê87GÊ&Ê51
50Ê&
51N
S1 S2 S2 S1
P1 P2 P2 P1
F D N I J C
P1 S1
L
P2 S2
Fig. 6.6.4a– AC current measurement transducers on a

CHAPTER
6.6.4a
6.6.4. Conventional AC Switchyard Measurements
AC current needs to be measured at several locations in the switchyard and in the line-side bushings of
the converter transformer, as described in this section.
6.6.4.1. AC Current Measurements on the Converter Transformer Bushings

In order to derive the measurements required for converter control and the electrical converter transformer
protections described in section 7.3, the AC current measurement transducers shown in Fig. 6.6.4a will
be required.
These CTs are usually mounted inside the associated bushing turrets
of the converter transformer. Therefore, due consideration must Vlw
be given to the amount of space available in the bushing turrets
for installing the CTs, as well as the method and time required for C1
installation and for removal of a CT core assembly. It is also essential
BushingÊinterconnectionÊbox
that the CTs are appropriately designed and adequately rated for
normal operation in the high operating temperatures (typically up to C2 C3 C4
SG Vo
135°C) in the oil-filled bushing turrets. R1 C5
The primary function of these CTs is to measure accurately and
reliably the various currents resulting from fault and normal operating HV R2
conditions. During steady-state and various converter faults, a DC bushing LocatedÊcloseÊtoÊHVÊbushing
component of current can arise in the transformer windings. Thus it is
Fig. 6.6.3b– Schematic of the HV bushing
normal practice to adopt air-gapped CTs, such as TPS Class CTs, for the voltage measurement system
CTs installed on the valve winding bushing of the transformer (see the
6.6.3b
FilterÊdifferentialÊprotection CT
CT CT
FilterÊcapacitor
unbalance
protection
CT CT CT
FilterÊresistorÊoverload
protection
CT CT CT
FilterÊresistorÊoverload FilterÊreactorÊovercurrent
protection protection
CT CT CT
CTs CTs
CTs
MainÊcapacitorÊovervoltageÊprotection
&ÊfilterÊE/F
&ÊfilterÊoverloadÊprotection
&ÊfilterÊovercurrentÊprotection Fig. 6.6.4b– Typical disposition of measurement transducers
&ÊcircuitÊbreakerÊfailÊprotection for a harmonic filter

CHAPTER
CTs H, G, B, I, J and C in Fig. 6.6.4a). If a standard, non air-gapped CT is used, this DC component will cause
the CT to saturate, resulting in an erroneous output which could lead to incorrect converter action, spurious
trips and/or failure of the protection to operate.
6.6.4.2. AC Measurements for Harmonic Filter protection

An example of the protection scheme applied to a double damped filter is given in Fig. 6.6.4b.
This illustrative arrangement shows the measurement transducers that would be required to implement the
associated protective functions in an AC harmonic filter. The CTs in this application do not require air gaps.
As Fig. 6.6.4b shows, CTs are typically mounted in the common neutral end of each phase and in the center
limb of the main capacitor bank H connections to feed the capacitor unbalance protection.
Where appropriate, CTs are installed in the connections to the filter reactors and resistors to feed the reactor
and resistor overload protection.
If required, harmonic filter differential protection is fed from dedicated cores in the filter CB bushing CTs are
typically used in addition to CTs mounted at the common neutral end of each phase.
6.7. OVERVOLTAGE PROTECTION

Overvoltage protection for a converter station is provided in two forms:
➙ Dynamic overvoltage protection equipment - surge arresters (see section 3.2)
➙ Converter protection functions incorporated as a part of the dedicated digital HVDC control and
protection and/or conventional protection scheme (see chapter 7)
One of the primary objectives of insulation coordination (see section 3.2) of a HVDC converter station is
the design of the overvoltage protection scheme. This involves the rating and identification of appropriate
locations of surge arresters. This section details the functionality of overvoltage protection equipment.
6.7.1. Surge Arrester Construction and Characteristics

A surge arrester is a non-linear resistance which reduces its resistance with increasing applied voltage. In
the event that the voltage at the connected bus exceeds the break-over point of the arrester, the arrester
begins to conduct current, holding down the voltage at the bus it is connected to, thus protecting the
nearby equipment.
A surge arrester is typically composed of a number of cylindrical zinc oxide blocks (Fig. 6.7.1a) stacked
under pressure, generally enclosed by an insulator housing (Fig. 6.7.1b). A zinc oxide block is typically
characterized by voltage versus current curves, depending on the voltage waveshape (power frequency,
switching type, lightning type etc). The energy capability of a block is affected to some extent by the cross-
sectional area of the block: increased cross-sectional area increases the energy capability of a block. Typical
voltage versus current curves are included in Fig. 6.7.1c.
A surge arrester is typically characterized by the following inter-related quantities:
➙ Voltages (see section 3.2)
– Crest Continuous Operating Voltage, CCOV (at power frequency)
– Peak Continuous Operating Voltage, PCOV (applicable to the DC side of the converter station
Fig. 6.7.1d)
– Switching impulse voltage
– Lightning impulse voltage
– Fast front impulse voltage (in some applications on the converter DC side)
➙ Currents coordinated with the voltage levels (Fig. 6.7.1c)
➙ Energy capability for various voltage waveshapes
The number of series connected blocks is affected by the specified voltage levels. The number of parallel
columns is affected by the required current and/or energy capability. For steady-state operating voltage,
the arresters are designed to result in minimal leakage current so that the energy generated is dissipated
at an appropriate temperature difference with respect to ambient, maintaining thermal stability of the
surge arrester.

CHAPTER
The magnitude of voltage of different waveshapes and the corresponding coordinating currents and energy
capabilities for a selected block are fixed (within manufacturing tolerances, operating conditions and aging
effects). A combination of series and parallel block arrangements is utilized to obtain a surge arrester
configured to meet the defined requirements. The arrester block arrangement is optimized to meet the
defining parameters.
For a design consisting of more than one column in parallel, the arrester is derated for:
➙ Unequal distribution of current between the columns
➙ Energy dissipation capability
The details regarding the rating process applied to surge arresters and selection of appropriate circuit
configuration are discussed in section 3.2.
Fig. 6.7.1a– Typical surge arrester blocks Fig. 6.7.1b– Valve hall surge
arrester with the housing
1.6
1.5
1.4
ResidualÊvoltageÊinÊp.u.ÊofÊvoltageÊatÊ10ÊkAÊ8/20Êµs
1.3 DC
1.2
1.1 AC
1
0.9 30/60ÊµsÊ
0.8
0.7 8/20ÊµsÊ
10kA0.6 8/20µs
0.5 1/2ÊµsÊ
0.4
0.3
0.2
Residual
0.1 voltage in p.u. of voltage at
0
1E-05 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000
Fig. 6.7.1c– Typical surge arrester characteristics CurrentÊinÊA

CHAPTER
6.7.1c
100
50
0
0 50 100 150 200 250 300 350 400
kV
-50
Crest continuous -100

operating voltage, CCOV
-150
Peak Continuous
Operating Voltage, PCOV -200
Time (°)
Fig. 6.7.1d– Typical voltage waveform imposed on a valve arrester, shown for a rectifier
6.8. CIRCUIT BREAKERS

Circuit breakers are an essential component of electrical power networks. Their main purpose is to safely clear,
6.7.1d
when instructed by the protection, any fault that might occur on any part of the electrical power network in
order to protect the operational personnel, the connected equipment or the network itself. A typical HVDC
converter station uses a number of AC circuit breakers and may also include some specialised DC switchgear.
6.8.1. Purpose and Minimum Requirements

Apart from interrupting fault currents, an AC circuit breaker must, in the closed position, conduct the
maximum load current as well as the maximum steady-state and temporary overcurrent and maximum
transient overcurrent, without exceeding the thermal capability of the contacts or the insulating materials.
While closed, a circuit breaker must withstand the maximum steady-state voltage stress with regard to
ground, as well as the maximum specified system maximum temporary, switching or lightning overvoltages.
In the open position, a circuit breaker must also withstand the maximum steady-state and transient voltage
stresses across its open contacts.
The circuit breaker operates by rapidly separating two contacts surrounded by a liquid or gaseous insulating
medium in order to interrupt the current. However, current interruption does not occur immediately: there
is a short arcing period during which current continues to flow between the opened contacts. One of the
most critical aspects of the design of a circuit breaker is to ensure that this arc is extinguished as quickly
as possible.
After the circuit breaker has interrupted a current, there is an electrical separation between the two parts
of the system on either side of the breaker. Immediately after current zero, a transient recovery voltage
will appear across the open contacts. The circuit breaker should withstand this voltage without re-ignition
or restrike.
Apart from interrupting the rated fault current, a circuit breaker is also required to interrupt load currents.
These include inductive loads such as shunt reactors and unloaded transformers, and capacitive loads
such as shunt capacitors.
6.8.2. Types
Circuit breakers are generally classified according to the medium used to control the arc and provide
dielectric strength. Over the years, four main types have been used for HV transmission and distribution:
➙ Air-blast ➙ Sulfur hexafluoride (SF6)
➙ Oil ➙ Vacuum

CHAPTER
Nowadays, due to the high efficiency of the SF 6 in
quenching the arc and its high dielectric strength, SF6
circuit breakers are the most widely used for high voltage
transmission, although because of the high Global
Warming Potential of SF6, research and development work
is taking place into suitable alternative gaseous insulators.
Vacuum breakers are still popular, but are mainly used for
low or medium voltage applications.
6.8.2.1. Air-blast Circuit Breakers

Air-blast circuit breakers use air under pressure to extinguish
the arc. The compressed air is directed across the arc path to
cool the arc and remove the ionized gas, thus restoring the
dielectric strength. The compressed air is provided locally by
a system comprising an air tank and a compressor. Fig. 6.8.2a
shows a typical air-blast circuit breaker.
Air-blast circuit breakers are known to achieve very
rapid arc extinction, which is needed to ensure stability
of the electricity supply following a major disturbance.
Because of the low dielectric strength of air, a number
of interrupters must be connected in series in order to
achieve the necessary voltage capability. Typically, up to
12 interrupter heads may be required for a 420 kV system.
Under these circumstances, grading capacitors are used to
maintain equal voltage sharing between the interrupters.
Opening resistors may be applied where the disturbances
created after current interruption are not acceptable.
6.8.2.2 Oil Circuit Breakers

In oil circuit breakers, the electric arc is drawn through Fig. 6.8.2a– Typical air-blast circuit
breaker
mineral oil which decomposes it into a number of gases:
including hydrogen, acetylene, methane and ethylene. This chemical decomposition helps in removing
energy from the arc. Further cooling is provided by the diffusion of the gas (mainly hydrogen) and the fresh oil
surrounding the arc, which also provides the dielectric strength when the current is interrupted. The efficiency
of oil circuit breakers is improved by providing an arc control device which allows higher pressure around the
arc, thus helping with the extinction.
There are two forms of oil circuit breakers: bulk (dead tank) oil breakers, and minimum-oil breakers. The
dead-tank oil breakers comprise an earthed tank that contains the contacts and the arc control device.
Connection to the electrical system is made through the bushings. In minimum-oil breakers, the contacts
and arc control structures are surrounded by closely fitting casings, all of which are directly connected to
the electrical system. The main advantage of the minimum oil breaker is that it uses less oil, and is more
compact, compared to the bulk oil breakers.
6.8.2.3. SF6 Circuit Breakers

Most circuit breakers at transmission voltage levels use sulfur hexafluoride (SF6) both as an arc interrupting
medium and as a dielectric medium. SF6 circuit breakers are also used at distribution voltage levels, but
share this market with vacuum and oil circuit breakers. Fig. 6.8.2b shows a typical 420 kV SF 6 live-tank
circuit breaker.
SF6 is an inorganic, colorless, odorless, and non-flammable gas. It is gaseous at atmospheric pressure
and room temperature, but will liquefy at higher pressures or lower temperatures. Its dielectric strength
is several times that of air and is equivalent to oil when at a pressure of 1.9 Bar absolute. Within the high
temperature arc column, SF6 dissociates into large numbers of fragments composed of highly reactive ions,

CHAPTER
which contribute to its good arc-control, quenching
and dielectric recovery characteristics.
There are two main types of SF 6 interrupters,
namely the gas-blast interrupter and the puffer
interrupter.
➙ Gas-blast interrupters generally have a
separate tank with a compressor to store
SF 6 at high pressure, separate from the
main SF6 used for dielectric purposes. When
the breaker contacts separate, a valve is
opened to cause a flow of high-pressure
gas around the contacts to extinguish the
arc. The stored gas is then re-compressed
following the operation. One disadvantage
of gas-blast interrupters is that the high-
pressure gas liquefies at normal ambient
temperatures, requiring the use of heaters
to maintain SF6 in its gaseous state.
Fig. 6.8.2b– A typical 420 kV SF6 live-tank circuit
➙ The puffer interrupter resolves this problem
breaker
by compressing the gas during the initial
part of the opening stroke and prior to
separation of the arc contacts. To achieve this, a more powerful operating mechanism to pre-
compress the gas and a somewhat longer operating stroke are required. Despite this, the puffer
1Ê-ÊFixedÊcontact
2Ê-ÊNozzle
3Ê-ÊMovingÊcontact
4Ê-ÊCompressionÊcylinder
5Ê-ÊFixedÊpiston
Fig. 6.8.2c– A typical puffer interrupter
6.8.2c
interrupter is the most commonly used interrupter on transmission SF6 circuit breakers. Fig. 6.8.2c
shows a typical puffer interrupter.
The main disadvantage of SF6 is the very high Global Warming Potential (GWP) of the gas: 22,000 times
as powerful as CO2. Some countries are starting to restrict the use of SF6 and there are even discussions
about phasing it out altogether. For these reasons, GE Vernova and other industrial companies have been
investing in R&D activities to identify suitable alternative insulating gases.
6.8.2.4. Vacuum Circuit Breakers

The ambient gas pressure in a vacuum interrupter is typically in the range of 10-4 to 10-8 Torr. At these
pressures, the available ionizable molecules are reduced, rapidly increasing the dielectric strength for a
given gap. A 15 mm contact gap in vacuum interrupters would withstand continuous voltages up to 11 kV.
With current flowing through the breaker, arcing is established within the interrupter when the bellows
CHAPTER
MetalÊvapor
condensingÊshield
FlexibleÊmetallic
bellows
MovingÊcontactÊstem
FixedÊcontactÊstem
InsulatingÊenvelope
Fig. 6.8.2d– A typical vacuum interrupter construction
contact is withdrawn from the stationary contact. The arc burns in the metal vapor evaporated from the hot
spots on the contact surfaces. At6.8.2d
or near current zero, the arc is extinguished, the vapor production ceases and
very rapid re-combination and de-ionization of the metal vapor occurs. The metal vapor re-condenses on the
contact surfaces and on the surrounding metal vapor condensation shield. This ensures the clean conditions
within the interrupter required for withstanding transient recovery voltage across the open contacts.
6.8.3. Arc Characteristics

The unavoidable presence of inductance in electrical systems means that the current through the breaker
is generally maintained even after its contacts have separated, which leads to the creation of an arc. As a
circuit element, the arc behaves like a non-linear resistor. The arc voltage is generally a function of the arc
ArcÊvoltage
Fig. 6.8.3a– Arc static characteristic ArcÊcurrent
current and its shape and value are influenced by the length of the arc, the cooling and de-ionizing methods
used, and the heat transfer characteristics of the surrounding medium.
6.8.3.1. Static Characteristic 6.8.3a

Under DC current, the arc voltage exhibits a characteristic similar to that shown in Fig. 6.8.3a. At low
currents, the arc voltage is high but falls as the current increases. At high currents, the arc voltage becomes
almost independent of the arc current.

CHAPTER
ArcÊvoltage
ia ic
ArcÊcurrent
Fig. 6.8.3b– Arc dynamic characteristic
6.8.3.2. Dynamic Characteristic

Under changing current, the arc conductance and hence the arc voltage, does not necessarily respond
instantaneously to the change in current. This is due to the electrical inertia of the arc, characterized by
6.8.3b
its time constant. The conductance generally lags the current as illustrated in Fig. 6.8.3b. The degree of
this lag is frequency dependent. At low frequencies, the arc voltage is in phase with the arc current and
follows the static characteristic. At very high frequencies, the arc conductance cannot follow the current
but remains constant at the mid value.
6.8.4. Circuit Breaker Standards

Circuit breakers, like any other pieces of equipment, are designed and manufactured in accordance with
corresponding international standards, which set the minimum performance requirements that the
equipment must meet. Standards cover the design requirements that the manufacturers must meet, as
well as the application requirements that the end user must follow to ensure that the equipment will
operate satisfactorily under the specified system conditions. Standards specify the rating values, which are
used to express the capabilities of the equipment. The rating values for circuit breakers include quantities
such as maximum continuous voltage, insulation withstand levels, maximum continuous current, maximum
interrupting current, momentary current rating, maximum transient recovery voltage, maximum interrupting
time, etc.
Tables 6.8a and 6.8b list the most relevant IEC and ANSI circuit breaker standards respectively.

CHAPTER
Subject IEC standard number
Part 1: Common Specifications 62271-1
Part 2: Seismic Qualification for Rated Voltages of 72.5 kV and above 62271-2
Part 100: High Voltage Alternating Current Circuit Breakers 62271-100
Part 101: Synthetic Testing 62271-101
Part 104: Alternating Current Switches for Rated Voltages of 52 kV and above 62271-104
Part 109: Alternating Current Series Capacitor bypass Switches 62271-109
Part 110: Inductive Load Switching 62271-110
AC Metal-Enclosed Switchgear and Controlgear for Rated Voltages
Part 200: 62271-200
above 1 kV and up to and including 52 kV
AC Insulation Enclosed Switchgear and Controlgear for Rated Voltages above
Part 201: 62271-201
1 kV and up to and including 52 kV
Part 203: Gas Insulation Metal Enclosed Switchgear for Rated Voltages above 52 kV 62271-203
Part 205: Compact Switchgear Assemblies for Rated Voltages above 52 kV 62271-205
Part 207: Seismic Qualification for Gas Insulated Switchgear above 52 kV 62271-207
Part 303: Use and Handling of Sulfur Hexafluoride (SF6) 62271-303
Part 308: Guide for Asymmetrical Short-Circuit Test Breaking Duty 62271-308
Table 6.8a– Relevant IEC circuit breaker standards
Subject ANSI/IEEE standard number

Standard rating structure C37.04
Preferred ratings C37.06
Test procedure C37.09
Application guide:
• General C37.010
• Transient recovery voltage C37.011
• Capacitance current switching C37.012
• Shunt reactor switching C37.015
AC high voltage generator C37.013
Electrical control requirements C37.11
Guide specifications C37.12
Synthetic fault testing C37.081
Sound pressure level measurement C37.082
Synthetic capacitive switching tests C37.083
Standard definitions C37.100
Table 6.8b– Relevant ANSI/IEEE circuit breaker standards
6.8.5. The Operation of a Circuit Breaker in an AC System

In AC systems, the current naturally crosses zero twice in a cycle. When the current is zero, the magnetic
energy stored in the system is also zero. AC system circuit breakers take advantage of this property in their
attempt to interrupt the current. When the current is interrupted, the two sides of the circuit breaker will
re-adjust to their new operating conditions. The difference in voltage between the two sides is the transient
recovery voltage that will appear across the recently conducting arc space. For a successful interruption,
the arc space must withstand first the thermal stress, and later the dielectric stress, both imposed by the
transient recovery voltage.

CHAPTER
6.8.5.1. Thermal Stress
As indicated previously, when the arc current reaches zero, the arc conductance (g(t)) still has a finite
value because of its time constant. The transient recovery voltage will force a post-zero current between
the open contacts and therefore contribute a power input of i2/g. The corresponding energy input over
time may raise the gas plasma temperature, which will increase its conductance, while the arc cooling
effects are attempting to reduce the temperature by removing the arc energy. If the energy dissipated
by the cooling effects is higher that the energy input from the TRV, the arc temperature and hence the
conductance will continue to decline and eventually reach zero, as shown in Fig. 6.8.5a. The opposite will
lead to an overstress of the conductance, resulting in the increase of temperature, and the re-ignition of
the arc for another cycle.
6.8.5.2. Dielectric Stress

If the balance of the arc energy input and output is such that the current is interrupted during the thermal
stress zone, the recently conducting arc space must dielectrically withstand the peak value of the TRV,
which will appear across the open contacts from within tens to hundreds of microseconds after the current
interruption. For this to be achieved, the arc space must have been cooled enough, and the distance
between the two contacts must be large enough to guarantee the dielectric strength is not exceeded.
g(t)
i(t)
t
TRV
Fig. 6.8.5a– Effect of TRV on current interruption
CBF
PoleÊ1Êconverter
ACÊfilters
6.8.5a CBP1
ACÊnetwork
CBP2
ACÊfilters
CBF
PoleÊ2Êconverter
Fig. 6.8.6a– Typical HVDC converter power circuit arrangement

CHAPTER
6.8.6a
TRV:ÊconverterÊbreaker
EnvelopeÊ+Êve EnvelopeÊ-Êve PhaseÊa PhaseÊb PhaseÊc

1.5Êk
1.0Êk
0.5Êk
0.0
kV
-0.5Êk
-1.0Êk
-1.5Êk
EnvelopeÊ+Êve EnvelopeÊ-Êve PhaseÊa PhaseÊb PhaseÊc
1.5Êk
1.0Êk
0.5Êk
0.0
kV
-0.5Êk
-1.0Êk
-1.5Êk
0.220 0.230 0.240 0.250 0.260
TimeÊ(s)
6.8.6b
Fig. 6.8.6b– Typical TRV experienced by converter circuit breaker (500 kV system)

CHAPTER
6.8.6. Converter Pole Circuit Breaker Requirements
Fig. 6.8.6a shows a typical AC circuit arrangement for a HVDC converter. It is a simple single busbar
arrangement, with a bus coupler. The main converter station components which are connected to the AC
network include the HVDC converter (including the thyristor valves and the converter transformers) and
the AC filters. The converters are connected to the AC busbar through the converter pole circuit breakers
(CBP). The circuit breakers CBP are used to energize and de-energize the converters and to clear any fault
on the converter side.
These are standard AC breakers which must comply with all the performance requirements based on the AC
network to which they are connected. The presence of the HVDC converter makes two of these requirements
more severe, namely the rated interrupting time and converter fault clearance under TOV conditions.
6.8.6.1. Rated Interrupting Time

The fault current for any fault on the valve side of the converter transformer is mainly limited by the reactance
of the converter transformer. Therefore, the fault current may be minimized by increasing the reactance of
the transformer. However, doing so also increases the rating of the converter equipment, the converter losses
and the reactive power absorbed by the converter, thus increasing the converter station cost. The value of
the converter transformer reactance is generally a compromise to meet these conflicting requirements.
One of the most severe faults for the thyristor valves is the failure of insulation across a thyristor valve as
shown in Fig. 6.2.11e. The corresponding fault current through the healthy valve is shown in Fig. 6.2.11f.
Under these conditions, the protection will generally detect the fault before the first peak value is reached
and block the thyristor valves at the end of the first cycle of current: a thyristor valve, like a circuit breaker,
cannot interrupt current and needs to wait for the next naturally occurring current zero. However, the
thyristor valve that experiences the fault current may not be able to block at the end of the first cycle
because of the heating effects of the fault current during the first loop or disturbances on the AC network
resulting in non-sinusoidal voltages being applied to the valve. It is crucial that the converter circuit breaker
(CBP) clears the fault within a short time to avoid any damage to the thyristor valve. For high power HVDC
converters, the fault current must not last for more than three cycles.
6.8.6.2. Converter Fault Clearance under TOV Conditions

During normal operation, a HVDC converter always absorbs reactive power, regardless of whether it
operates as a rectifier or as an inverter. The absorbed reactive power is generally compensated for by
providing shunt reactive power elements, which are also used as harmonic filters, as shown in Fig. 6.8.6a.
These elements are switched as necessary to minimize the reactive power unbalance with the AC
network. Should a fault such as that described in section 6.8.6.1 above occur, the converter circuit breaker
will be instructed to trip. Since the converter circuit breaker is the first to operate when the current is
TRV
2
Vs
1
i Vc
-1
-2
Fig. 6.8.7b– TRV on capacitive switching

CHAPTER
interrupted, both the real and reactive power consumed by the converter will be reduced to zero, leaving
uncompensated reactive power being generated by the shunt reactive power elements. This will produce
a transient overvoltage on the network side of the converter circuit breaker, which, under very weak
AC system conditions, could reach the protective level of the filter bus surge arrester. This overvoltage
transient, combined with the fact that the currents through the circuit breaker phases will be interrupted
at different time instants, will produce a Transient Recovery Voltage (TRV) that the converter circuit breaker
should withstand without re-ignition or restrike. Fig. 6.8.6b shows typical transient recovery voltage
following a converter fault clearance.
6.8.7. AC Harmonic Filter Circuit Breaker i

Requirements
TRV
AC harmonic filter circuit breakers (CBF) are used to energize Vs Vc
and de-energize AC harmonic filters and to disconnect any
filter which has developed a fault, as shown in Fig. 6.8.6a.
AC harmonic filters generally comprise capacitive, inductive Fig. 6.8.7a– Simple capacitive circuit
and resistive elements; however, they are mainly capacitive
at fundamental frequency. Therefore, apart from meeting the standard requirements associated with the
AC network to which they are connected, the AC harmonic filter circuit breakers must switch capacitive
loads under normal and TOV conditions.
The simple circuit shown in Fig. 6.8.7a may be used to illustrate how the TRV develops when a capacitive
load is disconnected. Since the current leads the load voltage by 90°, when the current is interrupted, the
6.8.7athe breaker’s open contacts is
full voltage is trapped across the capacitive load and the voltage between
zero. However, as the network voltage continues to follow the sine wave, the voltage across the breaker will
reach 2 p.u. in half a cycle, as shown in Fig. 6.8.7b.
6.8.7.1. Capacitive Switching under Normal Operation

The normal AC filter load current is much smaller than the rated interrupting current of the associated circuit
breaker. As a consequence, it is likely for the circuit breaker to interrupt the current during the initial stage of
the contact opening process. This leads to the full (1-cos) TRV being applied to the breaker when the contacts
are not fully opened yet. The circuit breaker should be able to withstand this TRV without restrike.
6.8.7.2. Capacitive Switching under TOV Conditions

As explained in section 6.8.6.2 above, when a converter is blocked and tripped while the AC harmonic
filters are left energized, a load rejection TOV will occur. As part of TOV control, any unnecessary AC
harmonic filters will be subsequently disconnected. The circuit breaker for the first AC harmonic filter to
be disconnected will experience the highest TOV and should not suffer a restrike.
6.8.8. The Operation of Circuit Breakers for DC Interruption

Since DC current has no natural zero crossings, interrupting DC current is more difficult than interrupting
AC current. The only way that DC current can be interrupted is by creating current zero by artificial means.
To achieve this, a number of methods have been studied and tried. The most widely used methods include
the use of the breaker arc voltage, the use of passive oscillation circuit and the current injection method.
L
R
+
E ea
-
Fig. 6.8.8a– DC switching circuit

CHAPTER
ArcÊvoltage
E
ea1
ArcÊinterruption
ea2
StableÊarc
Fig. 6.8.8b– Criterion for DC current interruption
ArcÊcurrent
Even where one of these methods is used, it may not be economic (especially at high voltages) to achieve
a true interruption of DC current, and in HVDC applications the usual technique is instead to divert the DC
current into another parallel current6.8.8b
path. For this reason the DC switches used to perform this operation are
known as DC commutating switches, as distinct from true DC circuit breakers that are capable of interrupting
fault currents (which are generally not necessary on Line-Commutated Converter HVDC schemes).
6.8.8.1. Using the Arc Voltage

This method relies purely on the breaker arc voltage in order to interrupt the DC current.
Considering the simple circuit shown in Fig. 6.8.8a, the basic equation describing the circuit after the
breaker has opened is:
di
E=L + Ri + ea [Eqn. 6.8a]
dt
By rearranging the equation above, the rate of change of current can be obtained as follows:
di 1
= (E − Ri) − ea  [Eqn. 6.8b]
dt L 
From Eqn 6.8b above, it can be seen that in order to ensure that the system current keeps decreasing to
reach zero, the arc voltage ea must be larger than E – Ri all the time. If there is a point where ea becomes equal
to E – Ri, di/dt becomes zero and a stable point will be reached. This is illustrated in Fig. 6.8.8b.
This method is limited to low voltage applications, as it is not possible to obtain arc voltages above
approximately 3 kV at high currents.
6.8.8.2. Passive Oscillation Method

The basic circuit used for this arrangement is illustrated in Fig. 6.8.8c. This arrangement consists of the
following components:
➙ A current-carrying main switch (CB)
➙ A reactor (L) and a capacitor (C) in series to form the commutation circuit
➙ A surge arrester (SA)
The values of the commutation circuit components are selected such that when the main switch is opened,
an increasing current oscillation is created between the commutation circuit and the main switch due to
the negative resistance characteristic of its arc voltage, as shown in Fig. 6.8.8d. When the oscillation current
magnitude reaches that of the main DC current, the current through the main switch reaches zero and is

CHAPTER
6.0
5.0
4.0
3.0
ÊcurrentÊ(kA)
2.0
1.0
0.0
-1.0
-2.0
-3.0
sec 0.5075 0.5100 0.5125 0.5150 0.5175 0.5200
BreakerÊcurrent LCÊcircuitÊcurrent ArresterÊcurrent
Fig. 6.8.8d– DC commutating switch performance
interrupted. The DC current is then transferred to the commutation circuit until the voltage across the
capacitor C reaches the protective level of the parallel surge arrester.
6.8.8d
At this point, the current is transferred into the surge arrester. Besides SA
limiting the voltage across the DC breaker, the surge arrester also
absorbs the energy stored in the system. It is the voltage developed C L
across the commutation circuit capacitor and/or the surge arrester
that will eventually force the DC current into a parallel circuit.
CB
6.8.8.3. Current Injection Method
The basic circuit used for this arrangement is illustrated in Fig. 6.8.8e. Fig. 6.8.8c– Passive oscillation circuit arrangement
This arrangement consists of the following components:
➙ A current carrying main switch (CB1)
➙ A reactor (L) and a pre-charged capacitor (C) in series to form the commutation circuit
➙ An auxiliary switch (CB2)
➙ A surge arrester (SA)
6.8.8c
During normal operation, CB1 is closed while CB2 is open. C is pre-charged and is used to create current
zero through the main switch CB1. L is used to limit the di/dt prior to current interruption.
HVDCÊlineÊ1
BPS BPS
PoleÊ1
BPS BPS
NBS ERTB NBS
MRTB ElectrodeÊlines
NBS NBGS NBGS NBS

BPS BPS
PoleÊ2
BPS BPS
HVDCÊlineÊ2
Fig. 6.8.9a– Current commutating switches in HVDC system

CHAPTER
When the current must be commutated into a parallel path, CB1 is opened first and an arc is drawn between
its contacts. CB2 is then closed, establishing a circulating current in the opposite direction to the arc current,
forcing it to reach zero. At this point the arc is extinguished and the DC current is transferred into the
commutation circuit, until the voltage across the capacitor C reaches the same level as the protective level
of the surge arrester. The current is then transferred into the surge arrester.
6.8.9. DC Commutating Switches for HVDC Systems

Most HVDC transmission schemes require DC
SA
Commutating Switches. Fig. 6.8.9a shows a
simplified single line diagram of a bipolar HVDC
scheme, showing some of the typical DC switches CB2 C L
which may be required (although not all will be - +
required on every project and some projects may CB1
require other types of switch that are not shown).
These include: L
➙ Metallic Return Transfer Breaker (MRTB)
➙ Earth Return Transfer Breaker (ERTB) Fig. 6.8.8e– Current injection circuit arrangement
➙ Neutral Bus Switch (NBS)
➙ Neutral Bus Grounding Switch (NBGS)
➙ Bypass Switch (BPS) 6.8.8e
These devices perform a variety of duties, which include both reconfiguration and protection of the system.
The common duty for all these switches is to commutate direct current up to full load (and sometimes up
to the specified overload) from one circuit into an existing parallel circuit. They are therefore commonly
referred to as current commutating switches: what differentiates them are the parallel circuits involved.
6.8.9.1. Metallic Return Transfer Breaker and Earth Return Transfer Breaker
If one pole of a bipolar scheme is not available, the system is normally designed to continue operating in
monopolar configuration. The return current can flow either through the ground (via electrode lines leading
to ground electrode sites, normally remote from the converter station), or in “metallic return” mode, using
a conductor on the transmission towers. This conductor can be either a dedicated return conductor (the
so-called “Dedicated Metallic Return” system) or, in the case where the pole outage is caused by a converter
fault, it uses the healthy HV conductor on the out-of-service pole.
Ground return mode gives the lowest losses, but in many areas, it is not permitted to operate a HVDC
scheme continuously with high direct current through the ground because of the risk of corrosion of
underground metallic structures such as pipelines. The most cost-effective alternative is generally to use
the pole conductor of the out-of-service pole.
The Metallic Return Transfer Breaker (MRTB) and the Earth Return Transfer Breaker (ERTB) are used to
respectively allow the reconfiguration of the pole in service from ground return to metallic return, and
from metallic return to ground return, without interrupting the power. During monopolar ground return
operation, the MRTB is closed and carries the full load current, while the ERTB is open. During metallic return
operation, the ERTB is closed and carries the full load current, while the MRTB is open.
The transition between one configuration and the other is achieved by first closing the open switch (after
the associated disconnectors are already closed), thus putting the HVDC line associated with the out-of-
service converter in parallel with the ground return. The other switch is then opened to transfer the direct
current from one circuit to the other.
6.8.9.2 Neutral Bus Switch

The Neutral Bus Switch (NBS) allows the isolation of the faulty pole converter without interrupting the
power in the healthy pole, thus maintaining the power availability.
During bipolar operation the NBS is closed and carries the full load current. If a fault occurs within a pole
converter, the protection associated with that pole will detect the fault and block and trip the converter.
However, the healthy pole will continue to feed the load current into the fault since the fault path has lower

CHAPTER
(i) (ii)
Fig. 6.9a– HVDC valve hall wall bushing rated at 285 kVdc: inside view (i) and outside view (ii)
6.9a with regard to the electrode line. The NBS associated with the faulty pole converter is then
impedance
opened to force the current from the fault path into the electrode line.
6.8.9.3 Neutral Bus Grounding Switch

During normal operation, the NBGS is open and therefore carries no current. If there is an open-circuit fault
on the electrode line, the NBGS is closed to provide a temporary earth connection and protect the neutral bus
components from extended voltage stresses. The NBGS allows the scheme to continue transmitting power
and therefore increases its reliability. If the HVDC scheme operates as a bipole, the NBGS may be required to
commutate a small unbalance current back into the electrode line when the fault has been cleared.
6.8.9.4. Bypass Switch

The Bypass Switch (BPS) is required when there is a need to take a series-connected valve group out
of service while allowing the other valve groups in the same pole to continue transmitting power, thus
increasing energy availability.
During normal operation, the BPS is open, and hence carries no current. If a fault occurs within a valve group,
the protection will block and trip the valve group, while closing the corresponding BPS. The BPS will now carry
the full load current. To bring the valve group back into service, the valve group is deblocked and operated
such that it takes over most of the current flowing through the BPS. The BPS is then opened to transfer a small
residual current into the valve group.
6.9. WALL BUSHINGS

Wall bushings are applied in HVDC converter stations mainly to connect the converter valves through the
valve hall walls to both the AC side and the DC side as illustrated in Fig. 6.9a.
Historically, porcelain oil-filled bushings with an oil impregnated condenser core were applied in HVDC,
especially at voltages above 100 kVdc.
In the earlier HVDC systems especially in long distance applications where the DC voltage is higher than in
back-to-back applications, the converter transformer valve windings were connected to the valves via wall
bushings. This was because, on the mercury arc valves in use at the time, the valve damping components

CHAPTER
were physically separate equipment, which hence could be located outdoors, along with disconnecting and
earthing switchgear in order to minimize the size of the valve hall [27]. With thyristor valves, the motivation
to arrange the layouts in this way disappeared since the damping circuits are an integral part of the valve.
In the majority of the HVDC systems utilizing thyristor valves, the transformer valve winding bushings
penetrate through the valve hall wall and connect to the valves inside the valve hall. Such a configuration
is advantageous because it eliminates the need for separate wall bushings on the AC side, which would
have been typically exposed to the outdoor environment (see section 3.3.1.1). This arrangement certainly
improves the flashover performance of the HVDC link.
On the DC side, the wall bushings are typically used to connect the DC side of the converter valve group to
the air-cored, air insulated smoothing reactor.
6.9.1. Porcelain Oil-Filled HVDC Wall Bushing Performance

As mentioned earlier, porcelain oil-filled condenser core type HVDC wall bushings were normally used in
HVDC applications until the mid-eighties. Flashovers occurred on wall bushings, typically in HVDC systems
operating at 400 kV and higher. The causes of the flashovers were reported to be due to:
➙ Pollution
➙ Bad weather conditions
➙ Unequal wetting
The trend in the early days, based on the experience in operating HVDC systems, was to specify a higher
specific creepage as high as 60 mm/kV. However, such increases did not improve the flashover performance
of the wall bushings.
Utilities and manufacturers of HVDC equipment cooperated and the solutions adopted were:
➙ Frequent cleaning: this is costly and is only effective for a short period of time immediately following
the cleaning
➙ The application of silicone or petroleum based grease is effective and has been shown to reduce
the number of flashovers. However, grease application is labor intensive and the grease must be
reapplied periodically to avoid flashover and permanent equipment damage
➙ The application of RTV coating is certainly effective to avoid flashovers and has the advantage that
it is a hydro-phobic coating which is effective in preventing unequal wetting flashovers. However,
depending on the site pollution conditions it may periodically have to be removed and a new coating
applied
➙ The use of booster sheds, which have proven in laboratory tests and in the field to be effective in
reducing the number of flashovers.
6.9.2. Composite HVDC Bushings

Composite bushings typically include a resin impregnated graded condenser core, housed in a non-
porcelain housing, with externally applied silicone rubber sheds. In the higher voltage applications of 500
kV and above, the bushing is filled with SF6. However, in some of the recent designs, even at 500 kV, solid
composite bushings are available.
Composite silicone rubber type bushings have demonstrated good performance for both pollution and
unequal wetting type flashovers. Composite bushings are typically lighter in weight compared to an equally
rated porcelain bushing. They were introduced in HVDC applications up to 500 kV in the mid eighties.
Nevertheless, SF6-filled wall bushings at the highest transmission voltages in use today (800 kV) are very
long structures which require careful analysis to verify that the mechanical stresses are acceptable (see
Fig. 6.9.2a).
6.10. AUXILIARY POWER SUPPLIES

A HVDC installation requires both AC and DC power supply and distribution systems, to feed equipment
such as control and protection equipment, cooling of electrical plant (particularly the converter valves
and the converter transformers), HVAC (heating, ventilation and air-conditioning), as well as mains power
outlets, indoor and outdoor lighting and more.

CHAPTER
MA MC MB
DG DG
P1LA P1LB P1CN P2CN P2LB P2LA
OilÊfiltration
FireÊfighting
WaterÊtreatment
ExternalÊlighting
Non-essentialÊHVAC
LightingÊ&ÊsmallÊpower
Essential Essential 220ÊVdc 48ÊVdc 48ÊVdc 48ÊVdc 48ÊVdc 220ÊVdc Essential Essential
ACÊloads ACÊloads BatteryÊNo.Ê1 BatteryÊNo.Ê2 BatteryÊNo.Ê1 BatteryÊNo.Ê3 BatteryÊNo.Ê4 BatteryÊNo.Ê2 ACÊloads ACÊloads
chargersÊ chargersÊ chargersÊ chargersÊ chargersÊ chargersÊ
-Êstation -ÊpoleÊ1 &Êbusbar &Êbusbar &Êbusbar &Êbusbar &Êbusbar &Êbusbar -ÊpoleÊ2 -Êstation
ACÊbusbars
MA:ÊMediumÊvoltageÊAÊ-Ê(11ÊkV,Ê34.5ÊkVÊorÊsimilar) P1LA:ÊPoleÊ1Ê-ÊLowÊvoltageÊAÊ(600ÊV,Ê415ÊVÊorÊsimilar) P2LA:ÊPoleÊ2Ê-ÊLowÊvoltageÊAÊ(600ÊV,Ê415ÊVÊorÊsimilar)
MB:ÊMediumÊvoltageÊBÊ-Ê(11ÊkV,Ê34.5ÊkVÊorÊsimilar) P1LB:ÊPoleÊ1Ê-ÊLowÊvoltageÊBÊ(600ÊV,Ê415ÊVÊorÊsimilar) P2LB:ÊPoleÊ2Ê-ÊLowÊvoltageÊBÊ(600ÊV,Ê415ÊVÊorÊsimilar)
MC:ÊMediumÊvoltageÊCÊ(ifÊnecessary)Ê-Ê(11ÊkV,Ê34.5ÊkVÊorÊsimilar) P1CN:ÊPoleÊ1Ê-ÊLowÊvoltageÊcommonÊ(600ÊV,Ê415ÊVÊorÊsimilar) P2CN:ÊPoleÊ2Ê-ÊLowÊvoltageÊcommonÊ(600ÊV,Ê415ÊVÊorÊsimilar)
Fig. 6.10a– Auxiliary supplies - bipole diagram
AC and DC auxiliary supplies are normally duplicated (or even

triplicated), very often supplemented by standby generators. The
extent of redundancy which is incorporated into the complete AC and
6.10a
DC supply and distribution systems is determined by the reliability
and availability requirements for the scheme. The basic principles
of redundancy apply to the AC and DC auxiliary power supply and
distribution systems whether the HVDC scheme is a simple monopole
back-to-back, or a complex bipole scheme with series valve groups.
The intention is to secure continued transmission against any single
failure, and often to continue to provide some level of power transfer
in the event of two unrelated failures. Thus, as illustrated in Fig. 6.10a,
a power supply arrangement for the HVDC scheme will normally have
two independent sources, each capable of supplying the total load for Fig. 6.9.2a– Complete assembly of an 800 kV HVDC
the whole station via a sectioned distribution busbar. In some bipole bushing
schemes where maximum independence between the two poles is
required, duplicated independent sources are used for each pole.
In general, the loads in the HVDC substation are divided into AC and DC, and then sub-divided into essential
and non-essential. Essential loads are those which are needed to maintain safe operation without risk of
damage to equipment or injury to personnel.
The selection of the voltage level for the LV battery and distribution system depends on the particular

CHAPTER
requirements for the project, the utility or owner preference, and varies according to geographical location.
For example:
➙ 48 Vdc – commonly used for communications systems in many areas of the world
➙ 220 Vdc – commonly used in Europe, Asia and elsewhere for switchgear trip/open and close circuits
between control/protection and switchyard
➙ 125 Vdc – commonly used in North America and elsewhere for all auxiliary supply systems within
a substation
Therefore, for a HVDC installation in North America, the DC battery and distribution system would use 125
V, with duplicated systems and capacity to provide power to all circuits within the substation, as illustrated
in Fig. 6.10b.
P1LA
P2LA
P1LB
P2LB
125ÊV 125ÊV 125ÊV 125ÊV
batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger
1 2 3 4
125ÊV 125ÊV
battery battery
1 2
Interlocked
transfer
busbar
125ÊVdcÊdistributionÊ1 125ÊVdcÊdistributionÊ2
Fig. 6.10b– Auxiliary supplies - simple 125 Vdc distribution
This configuration of feeders, batteries, chargers and distribution is the same basic system which is
commonly used elsewhere. In a more complex bipole system, there may be both 48 V batteries supplying
the communications and control systems, and 220 V supplying the switchyard. This configuration is shown
in Fig. 6.10c.
6.10b
There may be situations where there is only one MV source to provide power to a HVDC station, such as in
remote locations with few MV distribution lines in the area. In this case, it may be appropriate to use the
HV power lines which feed the main converter circuit as a source of auxiliary power, by either:
➙ Using a step-down transformer directly on the HV busbar
➙ Putting an auxiliary winding on a line end reactor turning this into a step-down transformer (but
only when the reactor is connected to the AC bus in order to meet the reactive power exchange
limits of the converter station)
➙ Adding an auxiliary supply transformer to the tertiary busbar in situations where the AC harmonic
filters are connected to a bus on a tertiary winding on the converter transformer
Even if all auxiliary AC power is lost, sufficient energy is stored in batteries to ensure that safe shutdown
of the converter equipment can be accomplished and that the minimum station services including such
things as emergency lighting and operation of switchgear continue to be provided with operating power
for several hours after AC infeed has ceased.

CHAPTER
P1LA
P2LA
P1LB
P2LB
220ÊV 220ÊV 220ÊV 220ÊV
batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger
1 2 3 4
220ÊV 220ÊV
battery battery
1 2
Interlocked
transfer
busbar
220ÊVdcÊdistributionÊ1 220ÊVdcÊdistributionÊ2
StationÊequipmentÊsupplyÊ1 StationÊequipmentÊsupplyÊ2
P1LA
P1LA
P2LA
P2LA
P1LB
P1LB
P2LB
P2LB
48ÊV 48ÊV 48ÊV 48ÊV 48ÊV 48ÊV 48ÊV 48ÊV
batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger batteryÊcharger
11 12 13 14 21 22 23 24
48ÊV 48ÊV 48ÊV 48ÊV

battery battery battery battery
11 12 21 22
Interlocked Interlocked
transfer transfer
busbar busbar
48ÊVdcÊdistributionÊ1A 48ÊVdcÊdistributionÊ1B 48ÊVdcÊdistributionÊ2A 48ÊVdcÊdistributionÊ2B
PoleÊ1Ê PoleÊ1Ê PoleÊ2Ê PoleÊ2Ê

equipmentÊsupplyÊA equipmentÊsupplyÊB equipmentÊsupplyÊA equipmentÊsupplyÊB
Fig. 6.10c– Auxiliary supplies – 220 Vdc and 48 Vdc distribution
The need to keep DC power flowing through the link may be so critical on some HVDC schemes that
special provisions may be needed to maintain power to specific items of plant such as the thyristor valve
cooling. One example of this is the Chandrapur scheme (described in section 1.3.4.6), which was subject
to 6.10c
occasional AC supply interruptions, and the cooling plant was required to continue operation for up to
30 seconds to ride through such events without interrupting the main HVDC power flow and to provide
sufficient time for the back-up diesel generator to start. This was achieved by using a compressed air
storage system, whereby pumps maintained high-pressure air in a storage vessel, and when the main
AC supply failed the coolant circulation pumps were driven by release of the compressed air. On other
projects this continuous valve cooling facility has been provided by an auxiliary cooling pump powered off
a battery backed UPS.

CHAPTER
BIBLIOGRAPHY
[1] “Semiconductor Devices - Part 6: Thyristors”, IEC 60747-6.
[2] P. D. Taylor, “Thyristor Design and Realization”, ISBN 0 471 91178 X.
[3] M. L. Woodhouse, M. J. Boden, R. G. Noble, “Design proving Tests on a valve for the English terminal of
the Cross-Channel HVDC Scheme”, IEE Conference Publication No. 255, “AC & DC Power Transmission”,
September 1985.
[4] “Determination of power losses in HVDC converter stations”, IEC 61803.
[5] D. M. Hodgson, “Qualification of XLPE tube systems forcooling high voltage high-power electrical
equipment”, IEE Power Engineering Journal, November 1991.
[6] R. P. Burgess, I. E. Barker, “McNeill HVDC Convertor Station”, GEC ALSTHOM Technical Review No. 50 1991.
[7] N. M. Macleod, C. C. Davidson and M. L. Woodhouse, “Design and Testing of Thyristor Valves for 800 kV
HVDC Projects”, IEC/CIGRÉ UHV Symposium, Beijing, July 2007.
[8] J. Sturgess, A. Baker, N. Macleod, F. Perrot, “Development of Insulation Structures for thyristor valves for
use on 800kV HVDC transmission schemes”, IEEE.
[9] J. D. Wheeler, C. C. Davidson, J. D. G. Williams, A. K. Roy, “Design Aspects of the Chandrapur 2 x 500 MW
Back to Back HVDC Scheme”, CIGRÉ Paper 14-104, Paris, 1996.
[10] “Thyristor Valves for HVDC power transmission – Part 1: Electrical Testing”, IEC 60700.
[11] M. L. Woodhouse, M. J. Boden, “Tests to verify the correct operational characteristics of thyristor valves
for HVDC convertors”, IEE Conference Publication No. 255 “AC and DC Power Transmission”, London, 1985.
[12] C C Davidson, J A Vodden, J J Snazell, “A new test circuit for operational testing of HVDC valves”, The
15th IET international conference on AC and DC Power Transmission, Coventry, UK, 2019.
[13] “Fire aspects of HVDC thyristor valves and valve halls”, CIGRÉ Technical Brochure 136.
[14] “Power Transformers Vol. 1”, Fundamentals GE Vernova.
[15] “Power Transformers Vol. 2”, Expertise GE Vernova.
[16] “Convertor Transformers. Transformers for HVDC Applications”, IEC 61378-2 – (2001)©.
[17] “Bushings for DC applications”, IEC 62199 – (2004)©.
[18] “General requirements and test code for Oil-immersed HVDC Converter transformers”, IEEE C57.129™.
[19] E. W. Kimbark, “Direct Current Transmission”, John Wiley & Sons Inc., Vol. 1, 1971.
[20] J. Arrilaga, “High Voltage Direct Current Transmission”, Institution of Electrical Engineers, 2nd Edition.
[21] M. J. Luckett, N. M. MacLeod, D. J. Young, “Aspects of Filter Design for The Chandrapur 2 x 500 MW, HVDC
Back-to-back Converter Station”, ICE 6th International Conference on AC and DC Power Transmission,
London, 1996.
[22] L. Hu, R. Yacamini, “Calculation of Harmonics and Interharmonics in HVDC Schemes with Low DC Side
Impedance”, IEE Proceedings-C, Vol. 140, No. 6, November 1993.
[23] D. J. Melvold, W. F. Long, “Back-to-Back HVDC System Performance with Different Smoothing Reactors”,
IEEE Transaction on Power Delivery, Vol. 4, No. 1, January 1989.
[24] “Insulation co-ordination - Part 5: Procedures for high voltage direct current (HVDC) converter stations”,
IEC 60071-5, TC/SC 28.
[25] “Surge arresters - Part 4: Metal-oxide surge arresters without gaps for AC systems”, IEC 60099-4.
[26] Thomas E. Browne Jr, “Circuit Interruption – Theory and Techniques”.
[27] T. G. Tushingham, E. B. Everall, F. G. Goodrich, “Layouts Of Convertor Stations”, IEE Conf. Pub 22 - HVDC
Transmission, September 1966, p. 384.

CHAPTER
[28] H. M. Schneider, J. F. Hall, C. L. Nellis , S. S. Low and D. J. Lorden, “Rain and contamination tests on HVDC
wall bushings with and without coatings”, IEEE Transactions on Power Delivery, Vol. 6, No. 3, July 1991.
[29] P. J. Lambeth, “Laboratory tests to evaluate HVDC wall bushing performance in wet weather”, IEEE
Transactions on Power delivery, 1994.
“Experience with HVDC wall bushings associated with the Nelson River system” Second HVDC Operating
conference, Winnipeg, September 19895.
W. Lampe. K. A. Erikson, C. A. O. Peixoto, “Operating experience of HVDC stations with regard to natural
contamination”, CIGRÉ report 33-01, 1984.
[30] M. M. Rashwan, W. McDermid, F. Hammer and A. Küchler, “On the Design, Testing and Operating
Experience of Composite Dry Bushings in HVDC”, IEE 5th International Conference on AC and DC Power
Transmission, London, U.K., September 1991.

CHAPTER
7| CONTROL AND PROTECTION
TOC 390 | DC
7 CONTROL
AND PROTECTION
DC is inherently more efficient, transmitting up to three times more
power than AC in the same right-of-way, and reducing comparative
generation requirements.
One of the main differentiating factors of DC versus AC transmission
schemes is the controlability.
This chapter introduces the basic control concepts of a HVDC
converter scheme in terms of the manipulation of the controllable
elements in order to achieve the desired power flows.
As with any electrical system, asset protection in the event of mal-
operation or external faults is an important consideration and so the
basic protections applied to a HVDC scheme are introduced, along
with a description of their functions.
Finally, a typical control and protection scheme implementation is
presented, showing the segmentation of functions within the system.

Chapter contents
CONTROL AND PROTECTION.................... 390
7. 7.9. DC POLE PROTECTION............................................................ 416
7.9.1. DC Pole Protection types....................................................... 417
7.1. HVDC SCHEME CONTROL...................................................... 393 7.9.2 Converter Isolation Protection............................................ 419
7.1.1. Control Parameters of an AC/DC Converter.................. 393 7.9.3. Ultra High Voltage HVDC ....................................................... 420
7.1.2. Voltage and Current Control of a HVDC Link................ 394
7.2. Control of a HVDC Station..................................................... 396 7.10. HVDC CONTROL AND PROTECTION
7.2.1. Rectifier Control......................................................................... 397 ARCHITECTURE........................................................................... 420
7.2.2. DC Power Control...................................................................... 400
7.2.3. Inverter Control.......................................................................... 400 7.11. S TRUCTURE OF A TYPICAL
7.2.4.
The Effect of Constant Extinction Angle on HVDC CONTROL SYSTEM....................................................... 420
AC System Disturbance.......................................................... 402 7.11.1. Dispatch Center/ System Control ..................................... 421
7.2.5 Frequency Control..................................................................... 405 7.11.2. Station Control........................................................................... 421
7.2.6 Power Modulation Control.................................................... 406 7.11.3. Bipole Control ............................................................................ 424
7.2.7. Runback / Power Demand Override (PDO).................... 406 7.11.4. Pole Control................................................................................. 425
7.2.8 Limit Driven Control................................................................. 406 7.11.5. Converter Group Control....................................................... 426
7.3. CONVERTER STATION PROTECTION ................................ 408 7.12. DUPLICATION OF CONTROL EQUIPMENT....................... 427
7.3.1. Selectivity...................................................................................... 409 7.12.1. Selection of Duplication Boundaries................................. 428
7.3.2. Reliability....................................................................................... 409 7.12.2. Auxiliary Power Supplies........................................................ 428
7.3.3. Stability.......................................................................................... 410 7.12.3. Communications Systems..................................................... 429
7.12.4. Control System Inputs and Outputs................................. 430
7.4. PROTECTIVE ZONES................................................................. 410 7.12.5. Maintenance of the System ................................................. 431
7.4.1. Protective Actions..................................................................... 410
7.13. CONTROL FOR SERIES
7.5. F AULT CASES AND CORRESPONDING CONNECTED CONVERTERS ..................................................................... 431
PROTECTIONS............................................................................. 412
7.5.1. Flashover across a Valve – Fault 1...................................... 412 BIBLIOGRAPHY............................................................................................... 433
7.5.2. Flashover across a Bridge – Fault 2................................... 413
7.5.3. Flashover across a Converter – Fault 3............................ 413
7.5.4. Faults across Valve winding Phases – Fault 4............... 413
7.5.5. Converter Ground Faults – Faults 5 and 6...................... 413
7.5.6. Electrode Line Fault – Fault 7............................................... 413
7.5.7. AC System Fault – Fault 8...................................................... 413
7.5.8. Commutation Failure............................................................... 414
7.5.9. DC Line Fault – Fault 9............................................................. 414
7.5.10. Filter Fault – Fault 10............................................................... 415
7.6. OVERVOLTAGE............................................................................ 415
7.7. TELECOMMUNICATIONS REQUIREMENTS..................... 416
7.8. CONVERTER TRANSFORMER FAULTS .............................. 416
TOC DC Transmission
392 | Systems:
DC Transmission
Line Commutated
Systems: Line
Converters
7.1. HVDC SCHEME CONTROL
There are a number of operating conditions that must be closely monitored and which can be modified in
order to control the power flowing through a HVDC link. In this section, we will introduce these operating
conditions and present how to control them.
7.1.1. Control Parameters of an AC/DC Converter

A single, controlled converter as shown in Fig. 7.1a only provides control of its DC terminal voltage (Vd).
The value of the DC voltage with respect to the two input parameters: applied AC voltage and firing angle
can be expressed by Eqn 2.2.2p for a rectifier and Eqn 2.2.3e for an inverter.
 Xc 
Vd p.u . = k ⋅ VLL p.u.⋅  cos(α ) − p .u . ⋅ Id p.u .  [Eqn 2.2.2p]
 2 
The adjustment of DC voltage can, therefore, be achieved in two ways: the converter-firing angle (a or g ) can
be varied or the magnitude of the applied AC voltage can be varied.
Cathode
terminal
Vemf Xsys
Vd
Vac
Anode
FIRINGÊANGLE (α) terminal
CONVERTER
CONTROLLER
Fig. 7.1a– A single, controlled, converter
7.1.1.1. Firing Angle Control

7.1a the firing angle at which the converter operates the DC voltage at the terminals of the
By adjusting
converter can be controlled. From Eqn 2.2.2p we can see that the DC voltage at no load is proportional to
the cosine of the firing angle. Hence, adjusting the firing angle from 0° to 90° provides an output voltage
between 1.0 p.u. to 0.0 p.u. Moreover, by increasing the firing angle beyond 90° the DC terminal voltage
reverses polarity, increasing from 0.0 p.u. to -1.0 p.u. at 180°. (It must be noted that in practice a converter
is never operated at an angle as low as 0° or as high as 180°.)

CHAPTER
As the firing angle can be adjusted cycle-by-cycle,
very fast control responses can be achieved with LineÊwinding ValveÊwinding
firing angle control, and an output of any value, Connection Connection
within the operating range of the converter, can
be achieved.
7.1.1.2. Tapchanger Control

VacÊl-g
Each converter transformer is usually fitted with VLL
an on-load tapchanger. An on-load tapchanger is a 1
mechanical switching device that changes the turns- 2
ratio of the converter transformer over a range of 3
4
values. The operation of a tapchanger is illustrated
in Fig. 7.1b, which shows a simple four-position
TurnsÊratio
tapchanger connected to the neutral end of the
line winding of a converter transformer. Excluded PositionÊ1ÊÊÊÊÊÊÊÊÊÊÊÊÊÊ5Ê:Ê4
PositionÊ2Ê ÊÊÊÊÊÊÊ6Ê:Ê4
from Fig. 7.1b, for simplicity, is the diverter switch. PositionÊ3Ê ÊÊÊÊÊÊÊ7Ê:Ê4
The diverter switch ensures that the current path PositionÊ4Ê ÊÊÊÊÊÊÊ8Ê:Ê4
through the transformer is never broken (by initiating
a carefully sequenced make-before-break switching Fig. 7.1b– A simplified diagram of a converter
transformer tapchanger
operation). For further information on converter
transformer tapchangers refer to section 6.4.
Each tapchanger step has a finite value, consequently the tapchanger only provides a coarse control of
DC voltage. Smooth or vernier control is not achievable using a tapchanger. Typically, a tapchanger step
7.1b
adds or subtracts about 1.25% to the converter transformer turns ratio. The maximum number of steps
available on a tapchanger is usually around 32, giving a maximum achievable voltage change of around 40%.
As the converter transformer tapchanger is a mechanical device, its response time is relatively slow, typically
taking five seconds to complete one tap step. Also, in order to avoid spurious tapchange operations, a delay
of typically two seconds is usually applied to the first tapchange order during which the order must remain
active. Subsequent tapchanger operations resulting from the same order are not normally delayed.
7.1.1.3. Combining Firing Angle and Tapchanger Control

Normal control of a HVDC converter combines both firing angle control, to provide rapid responses to
changing system conditions or load demands, and a converter transformer tapchanger to provide steady-
state adjustments to the converter’s operating conditions.
The way that these two control mechanisms are used depends on the control scheme adopted for a particular
application: influenced by both performance requirements and cost.
7.1.2. Voltage and Current Control of a HVDC Link

This section presents the basic control parameters of a HVDC interconnection or link.
7.1.2.1. Basic Current Control

Current can only flow in one direction through a thyristor: from the anode terminal to the cathode terminal
(see Fig. 7.1a). If two or more converters are going to be connected together to create a HVDC link, two of
the converters must be connected through opposite terminals, that is, Converter 1 anode to Converter 2
cathode and Converter 2 anode to Converter 1 cathode, as shown in Fig. 7.1c.
In a multi-terminal HVDC scheme any converters whose anode and cathode connections are in parallel
must have power flow in the same direction, that is, both must be acting as a rectifier or as an inverter.
In Fig. 7.1c an earth reference has been made at Converter 1. Hence, in this example the Converter 1 anode
connection is connected to earth, whilst Converter 2’s cathode connection is connected to earth. The
location of the earth (anode or cathode terminal) has no impact on the operation of the scheme; it only

CHAPTER
Converter 1 Cathode Anode Converter 2
terminal Rd terminal
V1emf X1sys X2sys V2emf

Vd Vdʻ
V1ac V2ac
Anode Cathode
terminal terminal
Fig. 7.1c– A simplified point-to-point HVDC link (all DC side resistances are expressed as one element Rd)
affects which power flow direction is associated with positive DC voltage with respect to earth. Power
flow from Converter 1 to Converter 2 is associated with positive DC voltage (Vd) whilst power flow from
Converter 2 to Converter 1 is associated with a negative value of DC voltage (Vd’) with respect to earth.
For simplicity, all of the DC side resistances in Fig. 7.1c are represented as a single lumped resistor Rd.
7.1c
As discussed above, each converter can provide either a positive or a negative voltage across its anode-to-
cathode terminals. Therefore, if Converter 1 produces a positive DC voltage across its anode-to-cathode
terminals while Converter 2 produces a negative voltage across its anode-to-cathode terminals, then with
respect to earth, both converters are producing a positive voltage. As the rectifier voltage and the inverter
voltage are independently controlled, they can have different values and hence there can be a voltage
difference across the resistor (Rd ) in the DC circuit, which, as long as the rectifier voltage is larger than the
inverter voltage, will cause a DC current to flow. This can be expressed as:
Vd − Vd ′
I DC = [Eqn 7.1a]
Rd
Therefore, both DC voltage and DC current can be controlled in a HVDC link by the coordinated action of the
converters that compose the link.
7.1.2.2. The Compounding Point

The converters shown in Fig. 7.1c could be many hundreds or even thousands of miles apart and therefore
cannot share a common set of measured parameters. It is important from the point of view of controlling
the HVDC link to have one point on the DC circuit where the DC voltage is defined as a reference. This
reference point is called the compounding point. To illustrate the meaning of the compounding point, again
consider Fig. 7.1c. The DC voltage at the terminals of Converter 2 (Vd’) must be lower than that at Converter
1 by an amount equivalent to the voltage drop across the DC circuit resistance (Rd). Therefore, if we define
the DC terminals of Converter 1 as the compounding point, then the voltage as measured at Converter 2
must be compounded, that is, back-calculated to give the voltage at Converter 1. This back calculation is
made by measuring the DC current at Converter 2 and, from the known DC resistance (Rd ), calculating the
voltage drop across the DC circuit. By summing this calculated voltage drop with the measured DC voltage
at Converter 2, the DC voltage at Converter 1, the compounding point, is known. Hence, each converter in a
HVDC link is able to ‘know’ what the DC voltage is at one particular point in the circuit. The control of each
converter can therefore be coordinated on a common base.
The compounding point does not have to be at a converter station. For example, the compounding point for
the IFA2000 Cross Channel HVDC interconnection is located mid-way between the two converter stations
which interconnect England and France [1]. Nevertheless, the compounding point is most commonly defined
as the DC voltage at the terminals of the converter operating as a rectifier, which will also correspond to the
location of the maximum DC voltage in the circuit.

CHAPTER
Whilst the DC circuit resistance (Rd) can be found from calculating the resistance of each conductor under
a given set of conditions, this resistance will not remain constant when in service, but will change relatively
slowly as a result of, for example, varying conductor temperature due to load variation or external weather
conditions. Therefore, when telecommunications are available, the measured DC voltage at Converter 1
can be sent to Converter 2 and Converter 2 can then calculate the actual DC resistance from the measured
DC voltages and the DC current. This calculated DC resistance can be used to update the DC resistance
value being used in the calculation of the compounding point voltage. Communication of the DC voltage
between Converter 1 and Converter 2 does not need to occur frequently.
7.1.2.3. Steady-State DC Voltage and Current Control

Under normal, steady-state operation the control system of the converter operating as an inverter is set
to maintain the DC voltage at the compounding point at a targeted value. This target value is typically 1.0
p.u. for a transmission scheme, but for back-to-back schemes where the DC transmission losses can be
ignored, this value can be varied to provide a further degree of reactive power control (see section 2.3). As
noted previously, the compounding point is usually at the rectifier DC terminal. Hence, the inverter must
calculate this voltage based on the measured DC voltage and current at the inverter terminals, as well as
the known resistance of the transmission circuit. The rectifier normally controls the DC current flowing in
the circuit by adjusting its output DC voltage to give a current flow as described by Eqn 7.1a.
Whilst inverter DC voltage control and rectifier DC current control is the most common arrangement of
a HVDC scheme, there are some exceptions. As described in section 1.2.6, the first 300 MW, HVDC link
between the mainland of South Korea and the holiday island of Cheju was designed so that the HVDC link
could be the sole source of electrical energy on the island. The converter located on the island, therefore,
needed very fast control of the current in order to maintain the island AC system frequency. If a normal
converter scheme arrangement had been used then the island converter would have controlled the HVDC
link’s DC voltage when importing power onto the island. This means that without fast telecommunications
to the rectifier, poor frequency performance would have been achieved, as a change in island load would
have required a change in power order to be sent via telecommunications to the mainland rectifier. As
it was a project requirement that the scheme should be able to operate in the absence of inter-station
communications, the control scheme was arranged such that the inverter controlled the DC current, with
the rectifier controlling the DC voltage. The rectifier, therefore, behaves as a ‘battery’ and the inverter as
a variable load.
Operating the inverter as the DC current controlling converter does give very fast response to local inverter
load changes, however, in the event of an inverter commutation failure, the rectifier, as it is in voltage control,
will attempt to supply the current resulting from the sudden increase in difference in voltage between the
rectifier and inverter terminals. This will impact the ratings of some of the converter equipment. A maximum
DC current limit can be built into the controller at the rectifier station, but sufficient margin must be allowed
within the setting of this limit value in order to accommodate any legitimate sudden overloads that may be
demanded by the inverter.
In order to avoid large dynamic currents flowing though a HVDC link and any consequential rating
implications, it is, as described above, more common to arrange the HVDC scheme controls such that the
rectifier maintains control of the DC current under normal operation. This way, even if there is a sudden
drop in the inverter terminal DC voltage, the DC current remains at a defined value. The actual target values
of DC current and voltage during dynamic events are defined by the converter scheme static characteristics
which are discussed in section 5.2.
7.2. CONTROL OF A HVDC STATION

It has been shown that the operating values of DC voltage and DC current can be set by the input
parameters of the converters within a HVDC link. It has also been shown that the converter AC terminal
voltage can be controlled by the converter transformer tapchanger. These can be combined to give
coordinated control of the converter over a wide range of AC and DC system operating conditions.
There are two basic types of control associated with a rectifier and three with an inverter, each with their
own advantages and disadvantages. Whilst, in the following discussion each control method is treated

CHAPTER
independently, for a real HVDC scheme two schemes may be combined, with the priority given to one control
type for certain operating conditions (for example, at low power) and another control type given priority under
other conditions (for example, at high power).
The basic rectifier control schemes are:
➙ Constant valve winding voltage control
➙ Firing angle limit control
The basic inverter control schemes are:
➙ Constant valve winding voltage control
➙ Extinction angle limit control
➙ Constant Extinction Angle (CEA) control
7.2.1. Rectifier Control

This section presents the two basic methods of rectifier control.
7.2.1.1. Constant Valve Winding Voltage Control

Constant valve winding voltage (ELL) control, referred to in some texts as ‘constant Vdo control’ and also
known as constant Udio control, can best be described by examining Fig. 7.2a.
The DC current (Id) is measured via a transducer. The process of measuring the current will introduce a
measurement error (DId, where the symbol D represents an error term), where this error can be positive or
negative. Therefore, the DC current as seen by the control system can be referred to as Id ± DId.
Vd
Tap VacÊ+ÊΔVac
position Id
αmeas+ÊΔα
IdÊ+ÊΔId
TapÊup α
+
Phase Ierror
VIIÊmax locked +
oscillator
Iorder
TapÊdown
+
Tap
VIIÊmin controller
Fig. 7.2a– Constant valve winding control scheme at a rectifier

CHAPTER
The measured current is compared to the required DC current, known as the current order (Iorder). Any
difference between the measured and ordered current will cause an error (Ierror). The current error is fed into
a Phase Locked Oscillator (PLO) which will modify the control angle of the rectifier, alpha (a), in an attempt
to reduce Ierror to zero. Therefore, the operating current is controlled by the control angle of the converter.
However, the AC system voltages are not fixed, hence, from Eqn 7.1a, Vd and Vd’ are not fixed but, instead,
follow their associated AC system voltages. As discussed above, under steady-state conditions, the AC
system voltage will typically vary over the range ±5% to ±10% and it is usual to expect that the required
steady-state power transmission can be maintained over this range. To avoid having to allow for this AC
voltage variation in the rating of the converter, a converter transformer on-load tapchanger is used.
As the tapchanger is only a coarse controller of AC voltage, the valve winding cannot be kept truly constant but
must instead be allowed to vary within a deadband between an upper and lower limit (VLL max and VLL min). As it is
impractical to measure the valve winding voltage, the valve winding voltage is calculated from a measurement
of the AC voltage and the converter transformer tapchanger position.
As the tapchanger is a mechanical device, in order to minimize maintenance requirements, the number of
operations it is expected to perform should be kept to a minimum. It is therefore necessary to ensure that
following a tap operation the control system does not request a tap back to the initial position. Such situations
can lead to the tapchanger being in a constant state of movement and unable to find a steady-state value. This
is referred to as ‘hunting’. In order to avoid this situation the tapchanger deadband is designed to be greater
than one tap step (typically 1½ tap steps) in order to avoid small disturbances causing the tapchanger to
repeatedly switch between two tap positions.
The converter must be capable of transmitting 1 p.u. power with the worst case errors in both the
manufacturing tolerances of the components (for example variations in converter transformer impedance
variation) and in the tolerances of measurement systems, whilst operating with the valve winding voltage at
the bottom of the deadband (VLL min). However, the converter, including all associated equipment (e.g. reactive
power banks), must be rated to operate with the worst case errors and the valve winding voltage at the top
of the deadband (VLL max).
7.2.1.2. Alpha Limit Control

When using alpha limit control, as with Constant valve winding voltage control, the converter control
angle alpha (a) is used to control DC current. However, with alpha limit control the tapchanger is used
to keep the control angle between an upper and lower threshold. Variations in AC system voltage will
cause a change in the converter terminal voltage and thus changes in alpha, in order to maintain the DC
current at the ordered value. If the change in alpha causes a steady-state alpha threshold to be crossed,
the tapchanger will be used to change the applied valve winding voltage and bring the control angle back
into the steady-state range.
The deadband between the upper and lower steady-state limit of control angle is set, as with constant valve
winding control, in order to avoid excessive tapchanger movement. The value of maximum, steady-state
control angle, given the minimum steady-state angle, is:
 cos(α min s − s ) 
α max s − s = cos −1  [Eqn 7.2a]
 1 + DB 
Where:
DB = the deadband in p.u.
amin s-s = the minimum selected rectifier firing angle
amax s-s = the maximum rectifier firing angle resulting from the selected deadband
7.2.1.3. Comparison of Rectifier Control Methods

With constant valve winding voltage control the control angle alpha (a) will be large at power transfer levels
below 1.0 p.u. A large control angle results in increased reactive power absorption, increased harmonic
generation and higher converter valve losses. If losses are evaluated at power levels of less than 1.0 p.u.
then the higher converter valve losses can result in a significant penalty.

CHAPTER
Vd
Id
αmeas+ÊΔα
IdÊ+ÊΔId
TapÊup α
+
Phase Ierror
αmaxÊsÊ-s locked +
oscillator
Iorder
TapÊdown
+
αminÊsÊ-s Tap
controller
Fig. 7.2b– Alpha limit control scheme
Conversely, alpha limit control utilizes lower control angles across the power range and hence absorbs less
reactive power, generates less harmonics and incurs lower losses at power transfer level of less than 1.0
p.u. Under most circumstances the reduced reactive power absorption is an advantage, however at very low
power the reactive power absorbed by the converter may be so low that, with the minimum number of filters
7.2b purposes, the net reactive power generation may be in excess of the reactive power
energized for filtering
limits for which the scheme is designed.
Particularly when the converter is connected at the end of a long AC line, the maximum reactive power export
allowed may be small at low power in order to control the AC line voltage. In such circumstances it is necessary
to either add additional shunt inductive elements (switched shunt reactors) or increase the converter
operating angle in order to increase the reactive power being absorbed by the converter. An alternative to
operating with increased control angle could be the use of constant valve winding control. However, increasing
the converter control angle will increase the harmonic generation of the converter and may in turn increase
the amount of filtering required, thus increasing the shunt connected capacitance (in the form of more AC
harmonic filters) even further.
Alpha limit control uses the converter transformer tapchanger to compensate for variations in the applied
AC voltage magnitude and for the regulation of the converter itself. Hence, the number of tap steps required
for alpha limit control is greater than that required for constant valve winding voltage control. Moreover, a
change in steady-state power flow may require tapchanger operation when using alpha limit control, therefore
increasing the time required to move from one steady-state condition to another. For bulk power transmission
schemes this may be of little relevance, but for schemes where frequent changes to the power transmission
level are required this can be an important consideration in selecting the most appropriate control method.

CHAPTER
The most economic solution for a particular scheme is usually decided through an iterative design process
between the converter and the harmonic filters.
7.2.2. DC Power Control

As discussed previously, the converter in current control is typically the rectifier, which is normally under
the control of the DC power loop. The DC power loop will continuously calculate the DC power at the
compounding point and will adjust the DC current order to achieve the ordered power (see Fig 7.2c). This
is why the communications interface of the HVDC link for the system operators is in terms of power and
power ramp rates (that is, how fast the power changes from the present value to a new set value), and
not DC current and voltage. Typically the maximum power limit is defined by an overload controller, which
continuously calculates the thermal capability of the converter station equipment.
PoleÊcontrol
Operator
power Current
demand Current order
control
Power FiringÊangle
Limits
control
Measured Voltage
Voltage
power order
control
1.0Êp.u.
DCÊvoltage
Fig. 7.2c– Typical power controller
7.2.3. Inverter Control
7.2c
As with the rectifier, both Constant valve winding voltage control and extinction angle limit control (similar
to alpha limit control) can be used at an inverter to control the DC voltage. Fig. 7.2d and Fig. 7.2e show
the simplified control diagrams for these two control methods. The implementation of each control
scheme is as discussed above for the rectifier, but with the current measurement replaced by a DC voltage
measurement (typically obtained from a resistive voltage divider). In addition, there is a third method,
which is the most common for HVDC transmission schemes and is known as Constant Extinction Angle
control (CEA control).
7.2.3.1. Constant Extinction Angle (CEA) Control

Constant Extinction Angle control is best described by referring to Fig. 7.2f. The extinction angle is kept
constant in steady-state operation by allowing the DC voltage to vary with variations in power transfer and
AC system voltage. Therefore, it is not necessary to rate the converter for operation above the minimum
steady-state value of control angle, gamma (gmin). This leads to lower reactive power absorption, lower
harmonic generation, lower converter losses and a lower maximum valve winding voltage, hence, a lower
number of thyristor levels in each valve.
The DC voltage is controlled by the inverter tapchanger. However, the tapchanger is only a coarse controller
and therefore it is not possible to operate the inverter at a constant 1.0 p.u. DC voltage. A deadband must be
applied to the DC voltage and again this will be approximately equivalent 11/2 tap steps. With a total deadband
of say, 2%, the DC voltage will vary in the steady-state from 1.0 p.u. to 0.98 p.u.
If the steady-state DC voltage can change by 2% then, in order to maintain constant power transfer, the
rectifier must have enough firing angle margin to increase its DC voltage in order to maintain the DC current
if the DC voltage at the inverter increases.

CHAPTER
Vd
Tap VacÊ+ÊΔVac
Id position
VdÊ+ÊΔVd VIIÊ+ÊΔVIIÊ
DCÊline γmeas+ÊΔγ
voltage
drop
compensation
α TapÊup
+
Vderror Phase VIIÊmax
+ locked
oscillator
Vdorder
TapÊdown
+
Tap
controller
VIIÊmin
Fig. 7.2d– Constant valve winding control at an inverter
7.2d 1.0 p.u. power at the rectifier terminals with a DC voltage range of 1.0 p.u to 0.98 p.u, the
To maintain
DC current must have a respective range of 1.0 p.u. to 1.02 p.u. and this must also be considered in the
equipment ratings. If the DC power is defined at the inverter AC or DC terminals, then the losses in the DC
transmission system must also be taken into account. With a long distance transmission line, increasing
the DC current by 0.02 p.u. for a reduction in DC voltage of 0.02 p.u. will not maintain the received power
constant, as the I2R component of the the transmission loss will have increased by:
1.022 - 1 = 0.04 p.u. [Eqn 7.2b]
Therefore, the current must be increased still further in order to maintain constant power at the receiving end.
This can lead to both an increase in the rectifier rating and increased transmission losses.
Many HVDC schemes are required to be bi-directional at rated power, that is, they have to be designed to
transmit 1.0 p.u. power in either direction. Optimizing the rating of one converter for either rectifier or inverter
operation is therefore inappropriate: a suitable compromise between the two forms of operation must be found.
A further effect of CEA control to be considered is the impact of the constant extinction angle characteristic
during an AC system disturbance.

CHAPTER
Vd
Id
VdÊ+ÊΔVd γmeas+ÊΔγ
DCÊline
voltage
drop
compensation TapÊup
α
+
Vderror Phase γmaxÊs-s
+ locked
oscillator
Vdorder
TapÊdown
+
Tap γminÊs-s
controller
Fig. 7.2e– Extinction angle limit control scheme
7.2.4. The Effect of Constant Extinction Angle on AC System Disturbance

7.2e
Constant extinction angle control at an inverter is usually employed in order to minimize:
➙ Losses
➙ Reactive power demand
➙ Rating of converter transformers
➙ Rating of converter valves
The constant extinction angle should be small: close to the permissible transient minimum angle at which
commutation cannot take place (refer to section 6.1.7.2). This control method will almost always give rise
to the lowest capital cost converter equipment. However, this is at the penalty of stability when operating
into a weak AC system. In the event of an AC system disturbance, for example a step change reduction
in the converter bus AC voltage, the extinction angle will be held constant, hence, the reduction in AC
voltage will result in the reduction of DC voltage. In order to maintain the DC power at a constant value
the power controller must increase the DC current. Increasing the DC current will increase the reactive
power absorption of both converters and this increase in reactive power absorption may further reduce the
AC system voltage at the inverter, reducing the DC voltage still further. With a relatively weak AC system
(refer to section 5.1), the inverter side AC system voltage can fall by a significant percentage before a stable
operating point is found. With a very weak AC system, a stable operating point may never be found, causing
the converter AC busbar voltage to collapse (refer to section 5.1).
Fig. 7.2g shows the limits of converter AC bus voltage stability for different AC voltage step change
reductions for a converter operating on its constant extinction angle characteristic, that is, operating in

CHAPTER
Vd
Id
γmeas+ÊΔγ
VdÊ+ÊΔVd
α TapÊup
+
DCÊline Phase γerror VdmaxÊs-s
voltage locked +
drop oscillator
compensation γorder
TapÊdown
+
Tap
controller VdminÊs-s
Fig. 7.2f– CEA control
pure CEA control and assuming a constant DC power control loop is in operation. From the figure we can
see that below an SCR (refer to section 5.1) of around 2.0, pure CEA control combined with constant DC
7.2f
power cannot function without the risk of voltage instability at the inverter AC busbar. In the event of a
voltage step reduction of 10% in the converter AC busbar voltage, then from Fig. 7.2g we can see that for
pure CEA control this would typically require an SCR of around 2.5.
It must be noted that the reactive power switching at the converter busbar is not considered to constitute
a voltage step event. This is because the converter controller knows when it is going to switch a reactive
power bank and can therefore pre-condition the converter, that is, temporarily increase the extinction angle
to avoid inducing an instability in the control of the converter. Such pre-conditioning of the extinction angle
is known as a gamma kick (refer to section 2.3.3).
7.2.4.1. Cascaded DC Voltage/Constant Extinction Angle (CEA) Control

The disadvantage of the pure CEA control system described above is that every change in converter bus
AC system voltage will be reflected in the DC voltage and hence interact with the DC power controller,
changing the DC current and hence the reactive power absorbed by the converter. This step change in
reactive power load at the inverter AC bus will interact with the AC system, causing the inverter AC busbar
voltage to change. An alternative control scheme, which can be used when the SCR is low, is known as
cascaded DC voltage/Constant Extinction Angle (CEA) Control, shown in Fig. 7.2h.
With cascaded DC voltage/Constant Extinction Angle (CEA) Control a margin is built into the selected steady-
state extinction angle such that, in the event of a step reduction in the inverter busbar AC voltage, the

CHAPTER
10
8
StepÊreductionÊinÊconverterÊACÊbusÊvoltageÊ(%)
7
Unstable
6
4
Stable
3
0
2 2.1 2.2 2.3 2.4 2.5 2.6
ShortÊCircuitÊRatioÊ(SCR)
Fig. 7.2g– Typical voltage stability limit of pure CEA control and constant DC power
extinction angle is temporarily allowed to reduce towards the minimum steady-state value of control angle,
7.2g
gamma (gmin).
Allowing the extinction angle to reduce means that the DC voltage can be maintained at a constant value
for a given step reduction in inverter AC busbar voltage, thereby not changing the DC power and actually
reducing the reactive power absorbed by the converter, thus assisting the AC system in its recovery from the
disturbance. This response to maintain the DC voltage constant is given the highest priority in the controller,
that is, it is the fastest loop operating within a cycle.
A much slower loop keeps the extinction angle at a constant value, typically operating with a time constant of
around one second. Hence, in the steady-state the converter will operate with a constant or ordered extinction
angle gamma order (gorder) thereby achieving savings in terms of plant rating as discussed above, but to a lesser
extent, as the constant extinction angle now includes a disturbance margin.
7.2.4.2. Inverter Operation into a Weak AC System

As discussed in the preceding section, with weak AC systems there is a danger of creating a voltage
instability at the inverter AC busbar due to the operation of the HVDC constant power loop.
Whilst the inverter AC system may always be weak in some projects, this is not the case in others. In many
projects the normal system strength is more than adequate for stable operation of the HVDC link power infeed
and only under certain, specific, outages of the system (for example, the loss of certain lines and generators)
may the system strength fall - requiring special converter design considerations. Under such conditions, it
may be more pragmatic to optimize the cost of the converter for normal operating conditions and accept
a reduction in transmitted power under these low short-circuit strength conditions. This mode of control
is typically referred to as AC Voltage Control, where the converter reduces the power being transmitted if
necessary, in order to maintain the inverter AC busbar voltage. A special case of this AC voltage control for
back-to-back converters is known as the ‘gamma other converter’.
For a back-to-back converter, the rectifier and inverter are in the same location and hence so are the
controllers. As there is no telecommunications between the two converter controllers, a rapid and reliable

CHAPTER
Vd
Id
γmeasÊ+Ê∆γ
VdÊ+Ê∆Vd
DCÊline α Phase
voltage locked
drop oscillator
compensation
VdÊ+Ê∆Vd
+
Vderror TapÊup
+
+ γerror VdmaxÊs-s
Vdorder
1 +
s
γorder
TapÊdown
+
Tap
VdminÊs-s
controller
Fig. 7.2h– Cascaded DC voltage/Constant Extinction Angle (CEA) control
exchange of data can take place. The rectifier is able to receive a signal from the inverter with the value of
7.2h
the instantaneous extinction angle, gamma. If this extinction angle approaches a threshold then the rectifier
can automatically modify its operating condition in order to maintain the inverter extinction angle at a target
value, even at the expense of transmitted power. Consequently, voltage stability at the inverter AC busbar is
significantly improved. Such a scheme has been successfully deployed on a number of projects [2].
7.2.5 Frequency Control

A HVDC scheme can control the AC frequency of an AC system by automatically adjusting the power being
delivered into or out of that AC system in order to balance the load with the supply. The fast power control
by the HVDC scheme reduces the under-frequency or over-frequency which can result from a changing
load in a small power system such as an island.
Frequency limits can be applied to the power control function. For example, the sending end can be arranged
so that it will continue to supply power via the HVDC link to the receiving end as long as the sending end AC

CHAPTER
system frequency is above a certain threshold value. In this way, the sending end can be protected from a
severe system disturbance as a consequence of a disturbance in the receiving end AC system.
The controllability of a HVDC scheme is very important and is sometimes referred to as providing a fire-wall.
With a power system consisting of islands of AC interconnected with HVDC this fire-wall property will mitigate
the risk of cascade black-outs across multiple interconnected AC systems.
Other frequency limits can be applied. For example, the receiving end AC system could have an upper
frequency limit to automatically stop further increases in the power being delivered by the HVDC scheme.
Equally, the receiving AC system can have a lower frequency limit which, if reached, automatically increases
the power being delivered into the receiving AC system. This limit can normally be overridden by the sending
end minimum frequency limit described above, that is, the sending end system will assist the receiving end
AC system as much as possible without risking a cascade failure.
F Operator PoleÊcontrol
pe power Current
slo
ncyÊ demand Current order
que control
Fre Measured Power FiringÊangle
power control Limits
Voltage Voltage
control order
1.0Êp.u.
P DCÊvoltage
Operator StationÊcontrol Power

frequency order
andÊslope
demands
Frequency
control
Measured
frequency
Fig. 7.2i– Typical frequency controller
7.2.6 Power Modulation Control

The power being transferred through a HVDC link can be automatically modulated, typically in the range
0.8 to 5 Hz, to provide damping to low frequency power oscillations within either or both interconnected
AC systems7.2i
as determined by studies during the design phase of the HVDC scheme.
7.2.7. Runback / Power Demand Override (PDO)

In response to certain events, such as loss of an AC transmission line, loss of an AC generator or loss of
a major load, the HVDC interconnection can be programmed to respond in a pre-defined manner. For
example, if the loss of a line may result in instability within the AC system the HVDC interconnection can be
pre-programmed to reduce the power transfer at a pre-determined ramp rate to a safe value as established
by contract studies. Equally, the loss of a generator can be pre-programmed to automatically increase the
power flow through the HVDC interconnection.
7.2.8 Limit Driven Control

Immediately following an AC system fault, a HVDC scheme will try to continue to deliver ordered power
in the pre-defined direction. Therefore, generators which are isolated from the fault by the HVDC will not
‘see’ the fault and will not automatically contribute to the overall AC system frequency control. Hence,
the AC system frequency in the AC system(s) without a fault will remain unaltered whilst the frequency in
the area with the fault will reduce more, as there are fewer generators providing the short-circuit power.
This greater reduction in frequency will force the spinning reserve within the faulted network to make a
greater contribution to supporting the AC system frequency, thereby, not pulling additional power across
the interconnected network(s).

CHAPTER
P Operator PoleÊcontrol
power Current
control FiringÊangle
Measured Power Limits
power control Voltage
Voltage order
control
1.0Êp.u.
time DCÊvoltage
Operator StationÊcontrol Power
selects order
enableÊor
disable
PowerÊmodulation
control
Measured
frequency
Fig. 7.2j– Typical power modulation controller
7.2j
P Operator
power
PoleÊcontrol
Current
Power control
Measured Limits FiringÊangle
power control Voltage Voltage
control order
1,0Êp.u.
DCÊVoltage
time
StationÊcontrol
Digital/ Power
protection order
inputs
Runback/PDO
Analog control
meas.
inputs
Fig. 7.2k– Typical PDO controller
Automatic detection within the HVDC controller can detect the reduction in the system frequency in the
faulted area and 7.2k
can be pre-programmed to automatically increase the power being delivered into the
faulted network to help support the network frequency (see Fig. 7.2l). Importantly however, the HVDC
link can be programmed to only increase the power it is delivering into the disturbed AC network up to
a point that meets some threshold within the healthy network. For example, it can increase the power
delivered until the system frequency in the un-faulted network falls to some pre-programmed value, thus
the disturbance will not be propagated across HVDC coupled networks.
The converse is also true if, prior to the disturbance, the HVDC back-to-back was exporting power from the
area which is now disturbed. In this case the HVDC back-to-back can be pre-programmed to ramp down the
transferred power or rapidly reverse power to support the failed network.

CHAPTER
DCÊpowerÊorder
DCÊpowerÊorder
PÊorder
} } FrequencyÊrangeÊ
constraintÊofÊinverterÊ
ACÊsystem PÊorder } FrequencyÊrangeÊ
constraintÊofÊrectifierÊ
ACÊsystem
}
Frequency Frequency
NormalÊfrequencyÊrange NormalÊfrequencyÊrange
ofÊrectifierÊACÊsystem ofÊinverterÊACÊsystem
Fig. 7.2l– Typical frequency/power order automated control characteristics
7.3. CONVERTER STATION PROTECTION

This section attempts to lay down the principles underpinning a protection design and describes the
7.2l possible strategies to implement a protection scheme for various converter station configurations. The
objectives of the converter station protection scheme are not dissimilar to those of an AC protection
scheme. The objectives are to limit the damage to the faulted equipment, to isolate the faulted equipment
from the rest of the system in order to allow the system to continue to operate, to minimize fire risk and
to minimize hazards to personnel.
Contrary to the AC network faults, the majority of the practical fault cases on the DC side of the HVDC
scheme do not pose immediate danger to the equipment involved. However, the fast speed of response and
power transfer capability of the DC link renders it necessary to have an additional DC protection objective,
which is to protect the system from the consequences of HVDC control malfunction by prompt removal of
the faulty equipment. In any case, studies have to be carried out to ensure that the DC protection settings
are coordinated with the DC controls and the AC protections of the connected AC networks. Similar to the
AC networks, the two principal stresses on the equipment are overcurrents and overvoltages.
Surge arresters are installed across major circuit equipment such as the converter transformers and the
thyristor valves to protect the equipment insulation against transient and temporary overvoltages. The
control actions to limit long-term voltage stress are converter dynamic control, tapchanger control and
reactive power control. The high voltage operating condition of the equipment dictates that protection
of DC equipment against overcurrent using fuses is not a viable option, instead the transient overcurrent
withstand capability is accounted for in the main circuit design of the equipment ratings.
The nature of surge current due to a fault is determined by the scheme parameters, the prevailing system
conditions, the fault location, the pre-fault mode of operation and the subsequent control actions [3].
The HVDC control acts to limit DC side overcurrent and most of the ground faults do not give rise to huge
overcurrents as observed in AC networks. Thus, except for valve short-circuit or converter transformer
faults which may severely overstress the equipment, urgent circuit breaker tripping is not always required.
Whilst for AC protection the only action generally available is to trip the AC breaker, which usually results
in removal of transmission capacity, for DC protection, the system can make use of the control actions at
its disposal to clear the faults more quickly and with less disruption. The fast speed of response of the DC
link makes it possible to use valve firing sequences in addition to circuit breaker tripping and mechanical
switching sequences to implement fault clearing actions.
From a protection design point of view, the major differences between a point-to-point scheme and
a back-to-back scheme are the additional transmission, filtering and switching equipment on the DC
side to be considered. Moreover, the significant capacitance in the transmission circuit and the physical
distance between the two converters mean that the protective actions employed and the sequencing of

CHAPTER
these actions are different from that of a back-to-back scheme. Telecommunication is normally deployed
between the two AC stations for a point-to-point scheme, but the basic protection design must not be
reliant on the availability of a telecommunication link, that is, the protection scheme should function
without a telecommunication system and the inter-station communication should only be used judiciously
to enhance fault detection and discrimination.
HVDC schemes are normally operated in bipole or monopole power transmission mode. Other possible
configurations are mainly used for testing, such as the open line test configuration which is used where
the transmission line is connected at only one converter station and then energized to verify its voltage
withstand capability.
The converter station protection scheme is a set of functions which work in coordination to isolate promptly
any part of the equipment under extreme stresses and to prevent undue disturbances to the system due
to the maloperation of controls. The protection scheme is deployed on a per pole basis with clearly defined
protective zones and actions. The protection schemes for both ends of a pole are identical except for the filter
protection, and as such, only the protection scheme for one end of a pole is described in this section. Any end
or pole specific requirements will be highlighted when necessary. DC protections are primarily embedded
within the Converter Control and Protection suite (CCP) and the Valve Base Electronics (VBE).
The guiding principles for protection relays in AC networks are generally applicable to DC protection design.
7.3.1. Selectivity
Fault conditions or other abnormal conditions that might expose equipment to hazards, as well as
conditions that will cause unacceptable disturbances to operation have to be detected and the faulty
equipment has to be taken out of service or relieved of stresses in a controlled manner, so that disturbance
to the operation of the rest of the system is minimized. The aim of the protection philosophy is to limit the
amount of equipment removed when isolating a fault. Ideally, only the faulted equipment or the smallest
possible scope containing the fault is disconnected. Selectivity is achieved by dividing the protection
function into zones. The advantage of this approach is that the location of a fault can be determined and
it makes it possible to disconnect the faulted equipment while leaving the rest of the system in operation.
For example, in a multi-polar scheme the majority of faults are cleared by tripping one pole, leaving the
other poles in operation.
7.3.2. Reliability
Various measures can be deployed to improve reliability, that is, to minimize the impact of faulty protective
equipment. The primary measure is through redundancy. Redundancy should be applied to the entire
detection and tripping sequence so that failure of any single element will not prevent tripping. The
protection should have two independent tripping schemes and, where applicable, it is desirable to have
each fault scenario detected using different principles. The protections should be arranged into overlapping
protective zones and for each fault case there should be a fast primary protection with a restricted
protective zone, supported by a time-graded, less sensitive, backup protection based on a different
measurement principle and with a more extended protective zone.
As far as possible, these features are not allowed to depend upon common equipment, nor do they depend
upon the control system to detect maloperation. For cases where the primary/backup concept cannot be
practically applied, the protection function should be duplicated. A typical example is the valve short-circuit
case, where valve overcurrent limits dictate a short fault clearing time, which precludes the use of slow
backup protection.
As mentioned before, the provision of primary and backup protection is to avoid single point of failure
preventing tripping. Possible causes are due to failures in one of the following:
➙ Measurement equipment
➙ Power supply to the protection
➙ Power supply to the protective sequencing
➙ Protective sequencing hardware and software
➙ Protection hardware and software
➙ Circuit breaker tripping circuits or breaker mechanism
CHAPTER
To avoid single point failures primary and backup protections should be physically separated with independent
auxiliary power supplies and, where applicable, independent measurement transducers. The circuit breakers
should be provided with breaker failure protection and the tripping and blocking paths should be duplicated
and monitored by the control system.
The protection should, as far as possible, be independent of the control system, meaning separate power
supplies and separate/buffered measurements to avoid common mode failures. Control and protection
functions in practice cannot be totally separated. In AC networks, the protection action is restricted to circuit
breaker tripping, usually resulting in taking out the transmission capacity. Under some circumstances, the
controllability of HVDC systems makes it possible to use valve-firing sequences to avoid the permanent loss
of transmission capability.
7.3.3. Stability
The protection must be able to discriminate between external power system events and transient and
genuine internal faults, so that undue disruption or disturbance to DC transmission can be avoided. A typical
example is the requirement to avoid tripping on inverter commutation failure caused by AC network faults.
The DC link in this case is designed to recover from commutation failures. On the other hand, repeated
commutation failure caused by control maloperation must not be disregarded. Other protections may also
have to be coordinated under depressed AC voltage conditions to avoid maloperation. This is usually achieved
by setting the trip operating delay longer than the expected duration of an AC disturbance. Often such delays
are unacceptable and detection principles that can discriminate external and internal faults must be devised.
A protection must only act upon a specific type of fault within a designated zone and be inert to other types of
disturbances or faults external to the relevant zone. Stability and selectivity are achieved though time-grading
and a unit system. Protection settings and delays should be selected to avoid operation due to AC system
transient disturbances and recoverable faults. Time guards and (Inverse time with Definite Minimum Time)
IDMT characteristics are generally used for time grading. The unit system does not involve time grading and
therefore can be relatively fast in operation. This is achieved by means of a comparison of quantities at the
boundaries of the AC/DC or DC/DC system.
The AC network must also be protected from the consequences of DC control maloperation. The AC network
can, in most cases, tolerate the non-characteristic harmonics that are created during DC system disturbances.
In some cases protection operating times may need adjustment to meet AC network requirements.
7.4. PROTECTIVE ZONES

A HVDC facility is divided into a number of separately protected and overlapping zones. The protective
zones are illustrated in Fig. 7.4a below.
Fig. 7.4a shows the arrangement of one pole end. The arrangement is identical for all poles.
The protective zones are: Zone 1: busbar
Zone 2: harmonic filter interzone
Zone 3: harmonic filter
Zone 4: converter transformer
Zone 5: pole
Zone 6: converter
Zone 7: neutral
Zone 8: transmission link
Zone 9: lines and substation busbar
7.4.1. Protective Actions

The protective actions normally employed to clear a fault are:
7.4.1.1. Trip the Circuit Breaker

The objective of this action is to isolate the HVDC equipment from the AC system, thereby clearing the
fault and reducing the stress on the equipment. For an urgent converter fault such as a valve short-circuit
CHAPTER
8
5 TransmissionÊlink
6
HVDC
filter
7
ElectrodeÊline
5
4 6
1 HVDC
filter
8
3 3 3 TransmissionÊlink
9 FilterÊ FilterÊ FilterÊ
bankÊ1 bankÊ2 bankÊ3
2
Fig. 7.4a– Protection zones
where the converter is in immediate danger, the rated withstand of the valve should be greater than
the operating time of the circuit breaker, inclusive of the detection time. This is typically, three to four
fundamental frequency cycles.
7.4a converter fault where the converter is not in immediate danger, it is desirable to wait until
For a non-urgent
the control system has reduced the load current to a low level before tripping the feeder circuit breaker.
In the case of converter feeder circuit breaker tripping, the filters should be opened at the same instant, or
earlier, to assist the opening of the feeder circuit breaker.
7.4.1.2. Block
Protective blocking is used to stop the flow of both AC and DC current in order to limit the effect of the
fault. This is achieved by simply removing the firing pulses to all the valves in the converter. Normally a
protective block is followed by a trip of the circuit breaker, as only removing the firing pulses may not
always stop conduction.
7.4.1.3. Valve Refire

The entire valve should be refired to prevent possible valve damage caused by partial blocking if more than a
certain number of thyristors (taking into account the number of redundant levels) are protectively triggered.
7.4.1.4. Inhibit Raise

For moderate overvoltages the tapchanger is inhibited from tapping up to ensure that the overvoltage
condition is not worsened due to tapchanger action.
7.4.1.5. Force Lower

For more severe overvoltages the tapchanger is forced to tap down to alleviate the stress on the equipment
due to the overvoltage.
7.4.1.6. Block-and-Bypass (Bypass)

This action provides a DC short-circuit across the converter bridge. It consists of blocking four valves in
the 6-pulse bridge and firing the remaining two as a bypass pair. In each 6-pulse bridge there are three
possible bypass pairs. Under some circumstances, for example an external flashover across a valve, it is
CHAPTER
necessary to select the pair containing the flashover valve. Once the DC current has been stopped, the
converter valves can be blocked and the converter AC feeder breaker tripped.
7.4.1.7. Forced Retard

This action forces rectifier firing at a high firing angle into the inversion region, to extinguish the current
flowing on the DC side.
7.5. FAULT CASES AND CORRESPONDING PROTECTIONS

This section describes some of the most common fault cases that may severely compromise the integrity
of the equipment and which protection(s) normally operate(s) in response to the faults.
Fig. 7.5a shows some possible cases of short-circuits and ground faults. Faults are numbered and relate
to the following sections.
TransmissionÊlink
FaultÊ1 FaultÊ9
1 5 9
T1 FaultÊ4
FaultÊ2
FaultÊ5 7 11 3
FaultÊ8 12 4 8
T2
Filter Filter Filter

bankÊ1 bankÊÊ2 bank 3
FaultÊ3
FaultÊ10
6 10 2
NBGS
FaultÊ6
ElectrodeÊline
FaultÊ7
Fig. 7.5a– Ground faults and short-circuits
7.5.1. Flashover across a Valve – Fault 1

An internal valve short-circuit resulting from the catastrophic failure of all the series connected thyristors
7.5ais highly unlikely. A more probable, but nonetheless rare cause, is a flashover to ground across a lower
valve winding bushing, or a flashover across a valve surge arrester. The most onerous case is a flashover
across a valve at the rectifier, developed immediately after successful commutation to the next valve, which
experiences high surge current limited only by the converter transformer leakage impedance.
Immediate complete blocking of the converter is required to prevent the fault current from commutating to
another healthy valve. Fast differential detection is implemented by comparing the valve winding currents
with the DC current. A gross mismatch early in the current surge between the two quantities indicates a valve
short-circuit. The protection action is to apply a full block on all the valves. The converter circuit breaker is also
tripped, as under such severe conditions the valves may not block successfully, in which case a fast tripping
circuit breaker is required to limit the stress on the valves.

CHAPTER
In the case of flashover across a valve at the inverter, the probability of relatively high current in the other valves
is low, since this can only occur due to reverse voltage flashover early in the cycle (not in normal inversion).
7.5.2. Flashover across a Bridge – Fault 2

Flashover at the converter bridge is rare because the valves are housed indoors in clean conditions. The
resultant surge current is limited by the converter transformer leakage impedance and is usually lower
than in the case for flashover across a valve. The fault is detected by the DC differential protection, which
compares the DC current measurements between the line and the neutral.
7.5.3. Flashover across a Converter – Fault 3

Similar to the case of a flashover across a bridge, this fault case is extremely rare and is quickly detected
by the converter differential protection. However, the surge current is less than that of a flashover across
a bridge because the commutating reactance of both the star and delta transformers is in the current’s
path, while the driving voltage is less than twice that of a single bridge owing to the phase shift of the
commutating voltages between the star and delta bridges.
7.5.4. Faults across Valve winding Phases – Fault 4

As the valve winding phase busbars are housed in the valve hall, the occurrence of this fault is rare. In terms
of current stress on the valve winding phase buses, the most onerous case is a 3-phase short-circuit. The
current stress is slightly lower for the case involving only two phases.
A short-circuit across the valve winding phases results in a large valve winding surge current and is limited
by the phase impedance as seen from the valve side, which consists of the converter transformer reactance
and the AC network impedance. The fault is detected by looking for a gross mismatch between the AC and
DC currents. The protective actions are to block the converter and to trip the AC circuit breaker.
7.5.5. Converter Ground Faults – Faults 5 and 6

The primary protection for converter ground faults is the DC differential protection, while the DC
overcurrent protection serves as a backup.
An additional protection makes use of the zero sequence component of the valve winding voltage which
is measured at the bushing taps. If a fault is present prior to deblocking the converter, as indicated by the
measured valve winding voltage, deblocking is inhibited. However, once deblocked, this protection is inhibited
in order to avoid maloperation.
7.5.6. Electrode Line Fault – Fault 7

The electrode line connects the neutral bus to the remote ground electrode. The loss of the electrode line
due to an open circuit will result in a rapidly increasing neutral bus voltage. This increase in neutral bus
voltage triggers the electrode line open circuit protection.
In the event of an electrode line short-circuit to ground, the change in DC resistance may be negligible. Active
injection of high frequency current and monitoring the impedance change can be used to detect this condition.
7.5.7. AC System Fault – Fault 8

A DC link is designed to ride through temporary AC system faults. The DC protection should be designed
not to maloperate during AC system faults and to be able to discriminate between external AC system
events and internal faults. A prime example is a commutation failure induced by an AC disturbance, where
the DC protection should not act but allow the system to recover and to continue with the power transfer
operation. For a persistent commutation failure caused by control or transducer failure, the DC protection
should act to trip the scheme.
The DC protections have to coordinate with the auto-reclose time of the AC network for low AC network
voltage conditions. The operating time of the AC undervoltage protection should be set longer than the auto-
CHAPTER
reclose time. On the other hand, the valve gate unit power supplies are charged by the AC system voltage
and to guarantee correct operation of the converter valves the scheme should be prevented from operating
under low AC voltage conditions for a prolonged period.
7.5.8. Commutation Failure

A commutation failure is the result of a failure to commutate current from an outgoing valve to an incoming
valve before the driving voltage across the valves reverses its polarity, taking into account the need for
sufficient extinction time for charge recombination and carrier sweep-out in the outgoing valve in order for
it to regain its blocking capability. The failure causes a temporary collapse of normal operation: assuming
the cause of commutation failure is temporary, subsequent recovery is normally due to main control action
over a few cycles.
Commutation failures are mainly caused by substantial disturbances on the AC voltage, due for example
to local or remote faults in the AC system, and rarely due to internal faults in the converter equipment. For
a large AC voltage depression, even on one phase only, commutation failure is usually inevitable, and the
commencement of true recovery will only occur when the original fault is removed.
During commutation failure the control operates in Low Voltage Current Clamp (LVCC) or Voltage
Dependent Current Order Limit (VDCOL) due to collapse or depression in the DC voltage, in order to
minimize the stress on the conducting valves. The net DC line voltage in some cases might not fall by
a sufficient amount and the rectifier LVCC would not normally operate to reduce DC current to reduce
the stress on DC equipment or valves. The measure taken is to use the commutation failure detection
to increase the inverter-firing angle (gamma) and force the rectifier to reduce current until the fault is
removed. Should commutation failure persist due to a sustained AC system fault, persistent commutation
failure protection operates to cause permanent shutdown of the pole.
If the setting is very long, for example 3 seconds, the inverter may make repeated attempts at recovery
before the AC (and hence DC) system finally recovers. The auxiliary supplies necessary for each thyristor
are usually drawn from the AC voltage via the valve damping circuits and under a low AC voltage condition
the power supplies have to be designed to last for the duration of the fault. Therefore, the persistent
commutation failure time must be set judiciously and very often a compromise has to be made.
In small low voltage thyristors, a forward recovery failure (which implies premature forward conduction)
generally does not damage the thyristors. In large thyristors suitable for HVDC, damage can occur in this
condition and special protection is normally provided for this by deliberately re-firing each thyristor if it is
subject to more than a small premature forward voltage within a certain minimum time after current zero.
The re-fire time is determined in the thyristor firing controls, which is then the effective thyristor minimum
extinction time for control design and system performance studies.
7.5.9. DC Line Fault – Fault 9

Common causes of a line fault on overhead DC transmission lines are: lightning strikes, flashover due to
contamination, insulation failure and an extreme example is catastrophic mechanical failure resulting in fallen
towers due to, for example, the weight of the ice buildup during an ice storm.
DC line faults are usually line to ground. Faults across the DC lines of the two poles for a bipole scheme are
highly unlikely due to the physical separation of the two conductors. Control action on its own is insufficient
to extinguish the fault current so the rectifier has to be force-retarded to have both the rectifier and inverter
operating in inversion. Following the force retard attempt to extinguish the current, a restart will be made
after a pre-determined delay to allow the air surrounding the arc to de-ionize. If, following re-establishment of
DC current flow, the DC voltage cannot be raised to the rated voltage, the fault is assumed to still be present
and the protection cycle is repeated.
In subsequent attempts it is common to increase the delay period between cessation of DC current and
restart in order to increase the possibility for a temporary fault to clear. If after several attempts the DC voltage
cannot be re-established at its rated value, then a further attempt at recovery to a lower voltage, for example
0.8 p.u., may be attempted. If the fault is deemed permanent, shutdown is required.
Line fault detection is usually based on a combination of low DC voltage and the rate of change of DC voltage.

CHAPTER
Wavelet analysis can also be used to discriminate between a commutation failure and a DC line fault [4].
The backup for DC line fault protection is DC undervoltage protection. The operating time for undervoltage
protection must be sufficient to be able to discriminate between AC and DC line faults and is usually set to a
value above the AC fault clearance time to avoid inadvertent shutdown of the converters.
7.5.10. Filter Fault – Fault 10

A fault in a filter bank will result in the tripping of the filter circuit breaker to isolate the fault. A replacement
filter will usually be switched in by the filter controller and the rest of the scheme can normally carry on.
Ground faults in a filter bank are detected by the filter differential protection. This protection may be operated
from measurements between the high voltage connected current measurement and the neutral current
measurement of each filter, or it may be operated on a filter bus basis, comparing a group of filters’ neutral
current measurements with the AC filter bus feeder current measurement. In the latter case, a flashover will
result in the tripping of all the filters within the associated filter differential zone. The protection is backed-up
by the overcurrent protection, which is also used to protect against filter reactor overload.
The capacitors in the filter banks are normally made up of series and parallel connected capacitor units. An
unbalance protection is usually installed to detect failure of one or more elements.
In the event of failure to trip the filter breaker, the breaker failure protection will trip the main circuit breaker
or equivalent to isolate the fault.
7.6. OVERVOLTAGE
Overvoltages can be classified into 2 different types – transient and long term. HVDC equipment is
protected against transient and temporary overvoltages by surge arresters and converter control. It is also
designed for continuous operation within the AC system continuous steady-state voltage limits. The AC
overvoltage protection is used to protect the equipment against persistent AC system voltage excursions
beyond the steady-state limits resulting from a system disturbance.
Voltage stress on the valve winding side due to tapchanger control failure is detected by the tap limit
protection, which inhibits the tapchanger tap up action. For prolonged valve winding voltage stress, usually
in the order of tens of seconds, the DC link is tripped.
For the transformer valve winding side overvoltage protection, the characteristic with respect to time needs to
grade under the valve surge arrester power-frequency voltage/time characteristic in order to avoid excessive
energy dissipation in the arrester. The arrester characteristic is provided by the arrester manufacturer. The
valve surge arrester characteristic to be used is usually the MCOV curve (power-frequency voltage versus
time characteristic). For the transformer line winding side overvoltage protection, the characteristic with
respect to time needs to grade under the transformer line side surge arrester capability, to avoid excessive
energy dissipation in the arrester. It also needs to grade under the converter transformer overfluxing curve,
to protect the converter transformer.
The overvoltage protection of the transformer valve winding is usually profiled with time, to fall below the
conduction level of the valve surge arresters. In this way thermal stress to the arresters resulting from
prolonged dissipation can be avoided.
In both cases, the surge arrester characteristics to be used are the MCOV curve (power-frequency voltage
versus time characteristic) obtained from the manufacturer.
The converter transformer overfluxing characteristic used should ideally be the overfluxing curve corresponding
to the maximum tap number (minimum primary turns), as this gives rise to the worst fluxing condition for a given
applied primary voltage. Doing this ensures that the transformer is adequately protected under all operating
conditions. As the line winding overvoltage protection only responds to voltage and not volts per hertz (or
voltage-time area), an additional margin is required below the overfluxing characteristic shown to account for
low frequency conditions. To achieve this, use is made of the expected system frequency variations.
If possible, the line winding overvoltage protection should also provide some backup protection to the AC
filter bank capacitor overvoltage. This requires that the line winding overvoltage characteristic be set under

CHAPTER
the capacitor overvoltage withstand characteristic, which is derived from the applicable IEEE / IEC capacitor
standard, along with the capacitor rated voltage chosen for the project.
DC side overvoltage, although very unlikely, may stress the DC equipment, especially the insulation of the
transmission cable. The most likely cause is the failure of the inverter to deblock during startup resulting in the
rectifier deblocking to an open circuit. The magnitude of the overvoltage is limited by the rectifier control system.
The unilateral blocking of an inverter will impose a substantial fundamental frequency voltage component on
the DC system. If the combination of the DC capacitance and inductance creates a resonant condition at, or
near, fundamental frequency, the valve current waveforms may not experience an immediate current zero,
leading to repeated fundamental frequency current injection and a progressive build-up of DC voltage [5].
7.7. TELECOMMUNICATIONS REQUIREMENTS

It must be emphasized that the DC link should be designed to be able to operate without
telecommunications and telecommunications should only be used to enhance the performance of the
control and protection system. One possible enhancement brought by the use of telecommunications is to
transfer the block to the remote end in the event of a local converter terminal fault. For example, in the case
of a fault to ground at the inverter terminal, the fault can be fed from the rectifier AC system as well as the
inverter AC system. It will therefore be advantageous to send a transfer block via telecommunications to
the rectifier on detection of a ground fault at the inverter to speed up the clearance of the fault. However,
when telecommunications are unavailable, the rectifier protection system should still block, and in this
case, it should usually be implemented by the rectifier DC line fault or DC undervoltage protection.
Telecommunications can also be used to enhance the discrimination of DC line fault protection against
maloperation in the event of inverter commutation failure. On detection of an inverter commutation failure,
a signal is sent via telecommunications to the rectifier to block the DC line fault protection and allow normal
recovery to take place.
7.8. CONVERTER TRANSFORMER FAULTS

Converter transformers are protected primarily against faults and overloads. The type of protection
employed is to minimize the time of disconnection for faults within the transformer.
Failures in converter transformers fall into the following categories:
➙ Winding failures due to short-circuits (turn to turn, phase to phase, phase to ground) and open winding
faults
➙ Core faults (core insulation failure, shorted lamination, etc.)
➙ Terminal failures (open leads, line connections, short-circuits, etc.)
➙ On load tapchanger failures (mechanical, electrical, short-circuits overheating)
➙ Abnormal operating conditions (over fluxing, overloading, overvoltage)
➙ External faults
7.9. DC POLE PROTECTION

The pole protection scheme is designed to protect converter equipment on the DC side of the pole. In order
to meet reliability and security requirements set out earlier in this chapter, each pole should be equipped
with its own protection scheme which should work independently.
Pole protection for each pole equipment is in a form which contains duplication to provide the necessary high
level of redundancy, while at the same time avoiding spurious trips.
As described in the following sections, the protective functions are designed using the same principles of
overlapping zones, diversity in the ways in which faults are detected wherever possible, and otherwise
simply duplicated. The redundant protection channels are always active, ensuring continuous monitoring
operation, and safe and orderly shutdown in response to specific fault conditions. Any particular fault
requiring a trip or block action will always be detected by at least one protection channel.
The individual protection equipment is also provided with duplicated power supplies.
Further discussion on duplication of the control and protection equipment is provided later in this chapter
in section 7.12.

CHAPTER
Normally when a protection has operated, the pole is locked-out and requires a manual reset by the operator.
The protective zones for these protections are the pole zone and the converter zone. They include the
protection functions summarized below in section 7.9.1.
7.9.1. DC Pole Protection Types

This section describes various types of DC protections which can be employed within a HVDC converter pole.
7.9.1.1. Protection: Asymmetry Protection

Protective Zone: Pole Zone
Description: to detect the persistent presence of fundamental and 2nd harmonic voltages or current between
the DC terminals of the pole. Example: caused by a valve misfire or commutation failure.
7.9.1.2. Protection: Pole Differential Protection

Description: to detect ground faults on the DC side of the converter and the AC conductor between the
converter transformer and the valve hall, and in response, take the pole out of service.
This protection compares the pole direct current in the high voltage bus to the direct current in the neutral
bus. This protection requires coordination with DC line protection, as a ground fault on the HVDC bus may also
operate the DC line protection. This protection should block before the DC line protection can initiate a restart.
The protection has two levels of sensitivity. At the lower level, the delay time before the protection is triggered
is longer, at typically 30 ms. If the mismatch exceeds the higher level, the protection must act faster and the
delay before operating is typically brought down to 3 ms.
The delay is introduced to avoid spurious triggering if the protection detects mismatches during energization
caused by charging currents.
If the scheme is operating initially in bipole mode, the protection attempts to block and trip the pole and
allow the other pole to continue in monopole mode. If the scheme is initially in monopole mode, the pole will
block and trip immediately.
7.9.1.3. Protection: DC Overcurrent Protection

Description: to detect overcurrent in the HVDC link and take the pole out of service if a fault is detected.
This protection uses an IDMT (Inverse Definite Minimum Time) characteristic. This characteristic, commonly
used in protection systems, merges two characteristics:
➙ an Inverse Time characteristic, which means that the longer a fault is present the lower the magnitude
of the fault which will cause protective action.
➙ a Definite Minimum characteristic which will immediately initiate protective action if the magnitude
of the initiating signal is greater than the threshold.
7.9.1.4. Protection: Abnormal Firing Angle Protection

Protective Zone: Converter Zone
Description: to detect abnormal firing angle conditions that may result in thermal failure of valve damping
resistors or surge arresters connected between the thyristor valve terminals.
7.9.1.5. Protection: AC>DC Differential (Short-Circuit) Protection

Description: to detect valve short-circuits or other phase-to-phase short-circuits which give rise to high AC
currents and low DC currents. In response, take the pole out of service.

CHAPTER
The HVDC current is compared to the AC current and if the AC current is greater than the equivalent HVDC
current by a preset amount, the protection is initiated.
7.9.1.6. Protection: DC>AC Differential (Commutation Failure) Protection

Description: to detect commutation failures within a converter bridge and, if the fault persists, take the
pole out of service. Normally the system should be able to recover from a commutation failure, but if the
commutation failure persists indicating a control maloperation, the pole is blocked.
7.9.1.7. Protection: AC Overcurrent

Description: to detect overcurrents in any of the valve winding connections which can result from insulation
failures within the converter or as a consequence of a control system failure.
7.9.1.8. Protection: AC Overvoltage Line side

Protective Zone: Busbar Zone/Converter Transformer Zone
Description: to detect overvoltages in the line winding side that could stress the equipment and in response
take the pole out of service in case of persistent AC overvoltage.
Overvoltage is tolerated for a certain time, dependent on the voltage level. If the overvoltage is removed within
this time then the protection does not operate. If the overvoltage persists, or if the overvoltage is removed
but recurs within the cooling period allowed, then the protection operates.
The equipment must be protected for AC system voltage excursions beyond the specified range resulting
from a system disturbance. The overvoltage protection characteristic is defined by the equipment capability.
The line side overvoltage characteristic is based on the worst case of the converter transformer overfluxing in
the event of overvoltage. The converter transformer overfluxing characteristic associated with the maximum
tap at full frequency (60/50 Hz) is selected, as it encompasses the requirement of all the equipment connected
to the line terminal.
7.9.1.9. Protection: AC Overvoltage Valve Side

Protective Zone: Converter Zone/Converter Transformer Zone
Description: the voltage is measured on the line winding side of the converter transformer and calculated for
the valve winding using the measured tap position.
Overvoltage is tolerated for a certain time, dependent on the voltage level. If the overvoltage is removed within
this time then the protection does not operate. If the overvoltage persists, or if the overvoltage is removed
but recurs within the cooling period allowed, then the protection operates.
The equipment must be protected for AC system voltage excursions beyond the specified range resulting
from a system disturbance. The overvoltage protection characteristic is defined by the equipment capability.
For the transformer valve winding side overvoltage protection, the characteristic with respect to time is
defined by the valve surge arrester capability.
7.9.1.10. Protection: AC Undervoltage

Description: to monitor the line-to-line AC system voltage to ensure it is adequate to maintain the thyristor
gate unit supply.

CHAPTER
7.9.1.11. Protection: Tap Limits
Description: to prevent long-term voltage stress that may cause harm to the equipment if allowed to persist.
Valve line side voltages are measured and compared with pre-set thresholds. For moderate voltage stress, the
control is inhibited from raising the tapchanger position. For severe voltage stress, the tapchanger is forced
to tap down to acceptable levels.
7.9.1.12. Protection: Thermal Model

Description: to calculate the thyristor junction temperature and take protective action should the temperature
exceed preset limits.
The thermal model has two temperature levels:
➙ The Overtemperature ‘1’ level is deemed to be the maximum thyristor temperature for non-fault
conditions. Smoothing is introduced to avoid tripping on transient overtemperature.
➙ The Overtemperature ‘2’ level results in a firing sequence which forces continuous conduction in order
to prevent damage to the thyristors, which have a diminished forward voltage withstand capability
above this temperature.
The valve has a chance to recover during the reverse voltage period following the conduction period in which
the level was exceeded, but if this level is still exceeded at an instant shortly before the valve voltage goes
positive then the protection initiates a protective firing of the valve. A block command is issued to the other
valves and a trip is issued to the converter feeder AC breaker.
7.9.1.13. Protection: DC Undercurrent

Description: the low DC current protection prevents prolonged operation of either a rectifier or inverter
operating into an open circuit. Example: when one side fails to deblock.
The settings need to be coordinated with the minimum deblocking current for the scheme.
7.9.1.14. Protection: Converter Differential Protection

Description: to detect ground faults on the DC side of the converter and the AC conductor between the
converter transformer and the valve hall and, in response, take the pole out of service.
7.9.2 Converter Isolation Protection

The disconnection of a HVDC inverter from its AC system, commonly called islanding, would be expected to
result in the loss of the AC terminal voltage and DC power transmission would cease. However, this is not the
case. A combination of the converter transformer, the AC filters and any remaining AC lines would give rise
to a resonant circuit capable of supporting an AC voltage. The inverter is thus able to continue to commutate
and the DC link transfers ‘real’ power into a circuit consisting of almost entirely reactive components.
This condition cannot be permitted to exist and consequently the AC terminal voltages rapidly increase
until any connected surge arresters conduct and dissipate the transferred DC power. With realistic surge
arrester energy ratings, this level of dissipation is not sustainable for more than a few cycles and a detection
system and suitable control response is required to detect the condition and take appropriate action.
This control response uses a Converter Isolation Detector (CID) [6] which is primarily based on detection
of voltages which correspond to the conduction settings of the surge arresters. The control response to
the detection of a converter being isolated from its AC system has to be compatible with the particular
requirements of the scheme, for example automatic restarting of power transmission following re-connection
of the AC system. It is also important that the consequences of false operation of the detector are also
considered, especially if either or both of the AC systems are weak.

CHAPTER
7.9.3. Ultra High Voltage HVDC
For DC pole voltages up to 660 kV, each converter is normally a single 12-pulse bridge. However, for voltages
above 660 kV it is common for each converter pole to consist of two, series connected, 12-pulse bridges
(see section 3.1.2.3). In order to allow one 12-pulse bridge to be taken out of service or put back into service
while the other series 12-pulse bridge remains in service, a bypass breaker is required across each 12-pulse
bridge. To allow a 12-pulse bridge to be taken out of service for an extended period bypass isolators along
with converter isolators are also provided.
7.10. HVDC CONTROL AND PROTECTION ARCHITECTURE

The purpose of the HVDC system is to deliver a defined, stable and controllable power flow between at
least two AC substations, taking into account both the short and long-term capability of the equipment
and the characteristics of the AC networks.
The HVDC control and protection system is responsible for the operation of the main scheme equipment
(thyristor valves, breakers for AC filters and shunt reactors, etc.). It ensures that the equipment is operated
within its limits to achieve the desired power transfer within the specified control parameters. The HVDC
control and protection system provides protection for the DC equipment and DC line/cable. The protection
of the balance of the equipment is normally provided separately by conventional AC protection equipment.
The power flow in the HVDC system is affected by the interactions between the HVDC and AC systems,
particularly when the DC system terminates at a weak AC system. The design of the HVDC control system
must therefore account for the dynamics of the integrated AC/DC system and mitigate any adverse interaction
between them.
The control and protection systems required for back-to-back and transmission schemes are very similar. The
main differences relate to the end-to-end communication mechanism and the additional DC switchyard and
DC line protections required by a transmission scheme.
Compared with control systems for conventional AC substations, many aspects of the HVDC control system
require fast, real-time processing of large numbers of analog and digital input quantities. The HVDC control
and protection system is therefore a complex and specialized sub-system. The hierarchy of a HVDC control
system should be designed to meet the structure implied by IEC 60633:2009. [7] A compliant generic HVDC
control system hierarchy is shown in Fig 7.10a.
Historically these structural levels would have been housed in discrete analogue equipment cabinets.
However, in modern digital control systems all of these control functions are often combined into single
cabinets.
HVDC systems can vary in complexity, from relatively simple systems such as single pole back-to-back
schemes or a monopole transmission scheme, through to back-to-back schemes with multiple independent
poles and bipolar transmission schemes with series-connected converters. The functionality needed for the
control and protection system will naturally vary with the requirements of each installation, for example, the
bipole level of the hierarchy would not be required for a single pole system.
The architecture of a HVDC control and protection system should be designed:
➙ to provide a clear separation of the control and protection functionality.
➙ to be tolerant to at least one failure in any level of the hierarchy.
➙ for each level of the hierarchy to operate as independently as possible.
➙ to provide robust internal and external integrity checks.
Integrity checks vary from generic low-level communication check codes and hardware monitoring, to higher
level ‘coverall’ checks such as inter-component heartbeats1 and system performance observers, to specific
HVDC or application knowledge based checks such as firing sequence monitoring or breaker and disconnector
movement time monitoring.
7.11. STRUCTURE OF A TYPICAL HVDC CONTROL SYSTEM

A typical HVDC control system hierarchy is structured as illustrated in Fig. 7.11a. In general, a control
function at a given level generates output signals which coordinate the actions of the lower level functions.

CHAPTER
The HVDC pole is the basic element of a HVDC scheme. Each pole has two pole ends. At any one time, one
pole end will be operating as a rectifier and the other as an inverter.
There is always a need for a station control function to provide station level functionality. The scope and
location of this function changes to suit the system topology. For systems with more than one pole, station
control coordinates the distribution of the station level power orders to the next level down in the control
hierarchy: to one or more pole controllers for multi-pole installations, or one or more bipole controllers for
multi-bipole installations.
The intermediary level of bipole control is only required if a pair of poles are configured to operate as a
bipole. When a pair of poles is operated as a bipole, the DC terminals of their two converters at both ends
of the link are connected in series, with their midpoint being connected to a common DC neutral point
(see section 3.1.2.2).
The bipole controller controls both poles as one large center earthed bipolar link (with one high voltage
DC positive line and one high voltage DC negative line) and can balance the current between the poles
to almost eliminate any current flow to earth via the neutral point. Bipole control is normally used for
transmission schemes.
The hierarchy of the HVDC control system is divided into 5 main areas listed below in hierarchical order:
➙ Dispatch Center/System Control
➙ Station Control
➙ Bipole Control
➙ Pole Control
➙ Converter Group Control
7.11.1. Dispatch Center/ System Control

The top level of control is the Dispatch Center/System control function, which is normally located at some
considerable distance from the HVDC converter station. All other levels of the control hierarchy normally
reside at the converter station.
The AC network operators automated power network management systems and power network operator
personnel manage the overall power delivery network from one or more Dispatch Centers. The Dispatch
Centers send power demands to one or more HVDC converter stations (and other power network
equipment) to meet the currently prevailing requirements of their commercial power delivery schedules.
Although remote access to information from lower hierarchical levels of the HVDC converter station can be
provided at a Dispatch Center as a separate facility, the information passed to and from the Dispatch Center
control systems is normally restricted to a minimal or high-level summary of the HVDC station’s key control
and status information.
7.11.2. Station Control

The Station Control function implements the highest level functions that determine the influence of the
scheme on the connected AC networks. The main function of Station Control is to control the real and
reactive power exchange with the AC power networks.
The Station Control functions are concerned with the interactions between the AC and DC systems for the
complete station. The controls use AC bus measurements, inputs provided by the operators and power level
information from the lower level controls to optimize the reactive exchange at different power transmission
levels by switching reactive elements and modifying the converter firing-angle and transformer tap position.
The remote Dispatch Center typically communicates with Station Control using an established internationally
standardized or well-established protocol such as IEC 60870 [8], DNP 3 [10], IEC 61850 [12], etc. These
protocols are increasingly carried over, or embedded within the ubiquitous suite of IEEE 802 [9] protocols
that allow readily available commercial communications equipment to be used. Other physical methods
1 – A heartbeat is a supervised regular exchange of information between two different sub-systems that is used to infer that
the sub-system is still active and working correctly.

CHAPTER
DispatchÊcenterÊ/
AC/DC SystemÊcontrol
system
level
StationÊcontrol
BipoleÊpowerÊorder
Station FrequencyÊlimiting
FrequencyÊcontrol
level ACÊvoltageÊcontrol
ReactiveÊpowerÊcontrol
BipoleÊ1Êcontrol BipoleÊ2Êcontrol
Bipole PoleÊpowerÊorders PoleÊpowerÊorders
level PowerÊlimits PowerÊlimits
PoleÊcurrentÊbalancing PoleÊcurrentÊbalancing
PoleÊ1Êcontrol PoleÊ2Êcontrol
PoleÊpowerÊcontrol PoleÊpowerÊcontrol
Alpha Alpha
Pole Gamma Gamma
PhaseÊlimits
level PhaseÊlimits
StaticÊcharacteristics StaticÊcharacteristics
TapchangerÊcontrol TapchangerÊcontrol
SubÊsynchronousÊdamping SubÊsynchronousÊdamping
PowerÊswingÊdamping PowerÊswingÊdamping
PoleÊprotection PoleÊprotection
Converter
group VBE VBE
ThyristorÊfiringÊcontrol ThyristorÊfiringÊcontrol
level ThyristorÊstatusÊreporting ThyristorÊstatusÊreporting
ThyristorÊprotection ThyristorÊprotection
Fig 7.10a- Control system functional hierarchy nomenclature
7.10a
CHAPTER
DispatchÊcenterÊ/ÊSystemÊcontrol
StationÊcontrol
BipoleÊcontrol
PoleÊcontrol
ConverterÊgroupÊcontrol
Pole Pole
master slave
Filters Filters
ACÊsystem ACÊsystem
Pole Pole
master slave
Filters Filters
Fig 7.11a– Structure of a typical HVDC control system
of communication such as Asynchronous Serial Links and Power Line Carrier systems can also be used to
exchange information.
Customer-specific protocols and communications media may have to be accommodated at this level. This
is more common when the HVDC control system is being supplied as part of a HVDC station refurbishment.
Station Control comprises the following functions:
➙ Station Power Control
7.11a ➙ Power Demand Override Control
➙ Reactive Power Control
➙ AC Voltage Control
➙ Power Modulation Control
➙ Power/Frequency Control
7.11.2.1. Station Power Control

Scheme-specific station power control is implemented when the power transfer of the bipole(s) or pole
control has to be coordinated at station level, in which case the station power transfer is controlled in
the direction, and to the value demanded by, the operator and is distributed to the available lower level
control(s).
7.11.2.2. Station Power Demand Overrides

This is a station level function implemented to carry out pre-programmed automatic rapid emergency
changes to the power levels of the HVDC links in a station in response to specific events in the connected
AC system (see section 7.2.7).

CHAPTER
7.11.2.3. Reactive Power Control (RPC)
The control of the reactive power exchange at a DC converter station is critical to the interaction with
the connected AC network. The AC filters required to remove the harmonics generated by the conversion
process are normally implemented by adding tuning and damping components to shunt capacitor banks,
which also provide the required reactive power compensation. These banks are switched in or out as the
DC load varies: a Reactive Power Controller can perform this switching manually or automatically.
The Reactive Power Control function is normally deployed at station level. When the DC link is connected to
a weak AC network, static VAr compensators or synchronous compensators (condensers) may be required
in addition to the above strategy to achieve the reactive power regulation requirements.
7.11.2.4. AC Voltage Control Mode

The AC Voltage Control Mode (ACVCM) function may be included in, or be used as an alternative to, RPC
depending on the requirements of a particular project. When the ACVCM is selected, it attempts to control
the AC system voltage to the operator specified voltage target, within a fixed deadband, by switching reactive
elements in or out. This may require temporary operation of the converter at elevated control angles, subject
to the inherent capability of the converter equipment. The target set by the operator may not always be met,
due to insufficient reactive elements being available or limitations due to the requirement to control harmonics.
7.11.2.5. Power Modulation Control

The power modulation function modulates the HVDC link power transmission to provide damping of low
frequency power swings within one or both connected AC systems (see section 7.2.6).
7.11.2.6. Power/Frequency Control

When this function is activated, the frequency of one system can be controlled within the power capacity
of the HVDC link by variation of the transmitted power (see section 7.2.5).
7.11.3. Bipole Control

Bipole control functionality is only required for transmission schemes that are operated in bipolar mode.
It coordinates the power transfer of the two poles under its control.
The bipole power order and power order ramp rates are received from the higher levels of control and are
distributed to the poles managed by the bipole controller. Depending upon the scheme, the division of the
power order can be balanced (joint control) or unbalanced (separate control).
The bipole control function has the following functions:
➙ Pole power orders
➙ Power limits
➙ Pole current balancing
7.11.3.1. Pole Power Orders

Bipole controls receive the power demands and rate at which the power will be ramped from the higher
level controls and distributes orders to the individual poles.
7.11.3.2. Power Limits

The bipole power order will be subjected to minimum and maximum power limits for the available poles.
The individual pole power orders will also be subjected to limits as determined by the individual pole
power capability.
7.11.3.3. Pole Current Balancing

If required, the bipole controller will modify the two pole power orders to achieve (near) zero current in
the bipole DC neutral to earth connection. Under this operating mode the two poles will operate with the
same DC current but will not necessarily transfer exactly the same power since they may be operating
with different DC voltages.
CHAPTER
7.11.4. Pole Control
Pole Control implements the following functions:
➙ Power order interface
➙ Determine voltage order
➙ Pole Power control
➙ Power loss calculation
➙ Overload control
➙ Order coordination
➙ Sub synchronous damping
7.11.4.1. Power Order Interface

The Power Order Interface provides an interface to Pole Power Control from the higher level controls. The
received power order and power ramp rate are used to determine when to issue a block or deblock request
to Converter Unit Control. For an installation with more than one 12-pulse converter unit per pole end
(see section 7.13), higher level sequencing is needed to block/deblock the converters in the proper order.
7.11.4.2. Determine Voltage Order

This function is responsible for generating the voltage order for the pole.
If ACVCM is selected in RPC the DC voltage order acts to fulfill the AC voltage demand. When the DC voltage
order reaches upper or lower limits it will send a signal back to RPC, signaling that it cannot increase/decrease
the DC voltage order further to fulfill the AC voltage demand.
If Reactive Power Exchange Mode (RPEM) is selected a reactive power order for each pole is received and is
used to derive the voltage order to fulfill the reactive power demand.
If the converter is not operating in RPEM or ACVCM the voltage order is clamped to a set value.
7.11.4.3. Pole Power Control

The pole power control function is responsible for maintaining the transmitted DC power to a given order.
This is achieved by a combination of DC voltage and DC current control. The current order is determined
by dividing the power order by the voltage order, which in turn can be determined by a reactive power
controller, AC voltage controller or operator demand.
7.11.4.4. Power Loss Calculation

Subject to the project requirements, power loss calculation measures DC and AC power on both sides and
calculates the losses. The calculated losses are adjusted to reflect the selected point of reference.
7.11.4.5. Overload Control
The overload control function allows the thermal overload capability of the equipment to be safely utilized,
without risking an unnecessary shutdown.
7.11.4.6. Order Coordination
The current coordination function is responsible for coordinating the current order between the rectifier
and inverter and maintaining the current margin to avoid a power collapse.
Coordination of the current order between the two ends of the HVDC link is achieved via an inter pole end
communication link dedicated to the exchange of the required signals. A checkback signal indicates that a
new current order has reached its destination and is used to ensure that rectifier and inverter current orders
are coordinated. One pole end is selected as the master of this coordination and the other as the slave.
For HVDC transmission systems the communication link between the two geographically separated ends is
normally duplicated to increase its availability. In the event of a total end-to-end telecommunications failure,
the control system will continue to actively control the power flow using reduced dynamic performance
criteria. This means that maximum ramp rates may have to be limited and some of the more sophisticated
power control modes such as power oscillation damping may have a reduced functionality.

CHAPTER
7.11.4.7. Sub-Synchronous Damping Control
The Sub-Synchronous Damping Control (SSDC) function supplements the HVDC control system by
providing a positive damping to torsional oscillations in a sub-synchronous frequency bandwidth, typically
in the range of 15-40 Hz. It acts to modulate the pole current order according to the transfer functions
implemented within the controller. The output modulation is limited and contributes, cycle by cycle, to
damp any sub-synchronous oscillation detected.
The SSDC function has to be tuned to achieve the best overall damping performance for a range of worst-case
AC network operating conditions without adversely affecting the response of the control system. The need for
the SSDC function is normally assessed by studies during the design phase of the project.
7.11.5. Converter Group Control

The functions in Converter Group Control (CGC) deal with the operation of an individual converter. They
aim to provide a robust and efficient conversion process (rectification and inversion).
Converter Group Control contains the following functions:
➙ Voltage Stress limit
➙ Phase Control
➙ Block/Deblock Control
➙ Tapchanger Control
➙ Tapchanger Interface
➙ Open Circuit Test Mode
➙ Valve Base Electronics (VBE)
7.11.5.1. Voltage Stress Limit
In order to ensure that the valve winding voltage does not exceed the maximum value allowed by the valve
surge arrester, a limit is applied to the voltage order before it is presented to phase control.
7.11.5.2. Phase Control
Phase Control (PhC) is the core of the control system. It implements the fast control loops that act in
response to control orders and limits to determine the firing instants of each valve in the 12-pulse converter
in sequence. Collectively, these control loops implement the converter transient and static characteristics,
which also depend on the state of the AC systems. Phase control employs the phase locked oscillator
control method pioneered in the late 1960s (see section 1.2.1.3).
Phase control implements the fast DC voltage/DC current control loops of the converters, which have
been described above in section 7.1. Separate characteristics are implemented for rectifier operation and
inverter operation. The single crossing point of the two converter characteristics defines the working point
for the scheme for different operating conditions.
7.11.5.3. Block/Deblock Control

In response to demands from the higher levels of the control system, the block/deblock control function
issues block and deblock sequences in order to block or deblock the converters. Only the pole end currently
configured as the Master responds to higher level controls, unless Open Circuit test mode (section 7.11.5.6)
is selected for the converter. The Slave pole end only responds to requests to block/deblock commands
from the Master side.
The Block/Deblock function only handles blocking and deblocking for one converter. For an arrangement with
more than one 12-pulse converter, sequencing is needed at the higher levels of the control system before
issuing block/deblock commands to the individual converters.
7.11.5.4. Tapchanger Control

The tapchanger control function keeps the converter firing angles, DC voltage or valve winding
voltages within a limited steady-state range by controlling the tapchanger on the converter transformer.
Hunting is avoided by setting the limited steady-state range sufficiently wider than a one tap step size.

CHAPTER
7.11.5.5. Tapchanger Interface
The tapchanger interface function issues commands to the tapchanger to raise or lower the tap position
in order to maintain the converter valve winding voltage within a limited steady-state range. The interface
function checks that the operation takes place and if the operation fails, or an invalid tap position (not a
defined tap position) is reached, an alarm is raised.
7.11.5.6. Open Circuit Test Mode

When the system is set to this mode it is possible to apply a direct voltage to an open DC circuit. This is
used for commissioning, maintenance and testing of the converter.
7.11.5.7. Valve Base Electronics

The Valve Base Electronics (VBE) contains the following functions:
➙ Firing pulse encoding
➙ Optical transmission of firing information to the valves
➙ Optical reception of valve monitoring information
➙ Decoding of the monitoring information (databack)
VBE receives the firing information (‘firing word’) from Phase Control and the thyristor junction temperature
information from the thermal model.
The thermal model provides the VBE system with a calculated thyristor junction temperature that
is transmitted to all thyristor gate units. The thyristor gate units use this information to adjust junction
temperature dependent protection thresholds dynamically.
7.12. DUPLICATION OF CONTROL EQUIPMENT

A typical HVDC system can exchange several gigawatts of power between the source and destination
AC systems. The power transfer capacity of the HVDC system is often directly marketed by the scheme
owners to other power generation and transport companies. These companies generally impose onerous
penalties for unavailability of a contractually agreed time slot. The unexpected loss of transmission can
therefore affect huge numbers of people, cause power network stability difficulties and have a significant
financial impact for the scheme owners.
The HVDC power transmission system is normally specified and designed to meet or exceed certain
minimum performance criteria, such as availability (typically 99% to 99.5%) and reliability (typically less
than 0.5 forced outages per year).
The control system typically provides almost all of the functionality of the HVDC system and is therefore
the most complex individual element in the system. The current market trend is to increase the complexity
of the control system by demanding increasingly sophisticated control and maintenance functionality. The
hardware systems required to fulfill these expectations have high numbers of plant interface circuits and a
high core system part count. Consequently, these highly advanced control systems may become the reliability
bottleneck of the HVDC system.
As a key component of the overall HVDC scheme, the HVDC control and protection system must be designed
to exceed the specified reliability and availability guarantees. As a random component failure cannot be
eliminated and the cost of the control equipment hardware is a relatively small percentage of the cost of the
overall HVDC scheme, the recognized way to achieve and exceed the required performance guarantees is to
provide duplication of the control hardware.
The reliability and availability figures for a system can be calculated from the theoretical or actual failure
rates of all the components within the system. This is a complex calculation as it must take into account
the redundant and non-redundant failure paths within each system, the time required to repair the system
assuming a spare is immediately available, the time to replenish the spare, the number of components of a
particular type within the system, etc.
The results of this analysis are used to guide the overall control system design. Due to the complexity, the
number of components, wire terminations, solder joints, etc. within the control system, a full duplication of

CHAPTER
the system is normally recommended to achieve the required performance levels.
A duplicated control system has two identical hardware copies, each of which is connected to and capable
of controlling the HVDC plant.
The HVDC control and protection system must be designed to ensure that:
➙ At any instant, only one of these systems is actually in control of the scheme
➙ Control can be instantly transferred from one system to the other without a break or noticeable
disturbance in the power transmission
➙ Failed duplicated areas of the control system can be repaired while the other is in control
➙ Failures can be detected
➙ The impact of a failure on the scheme is minimized
7.12.1. Selection of Duplication Boundaries

The selection of the duplication boundaries and the level of functionality that needs to be duplicated
within these boundaries is very important [11]. As the main role of the protection system is to shut down
the HVDC system in the event of a serious problem, it directly affects the availability if it erroneously
requests a shutdown. For example, each duplicated ‘lane’ of Converter Control and Protection (CCP), as
illustrated in Fig. 7.12b, may contain duplicated DC protection hardware to allow one lane of converter
control and protection to be shut down for maintenance, while retaining adequate system protection. In
this case there will be four total sets of protection hardware overall. However, a more efficient solution
would be to provide three protections overall, such that there is one in each lane, and a third which may
be considered as 'floating'.
If more than two sets of protection are provided, a simple voting system can be used to increase the availability
of the overall control system, by assuming that the trip requests of less than 50% of the currently available
protection systems are erroneous.
The voting system would then wait for 50% or more protection system trip requests before the system is
shut down (see Fig. 7.12a for a typical system).
The choice of duplication boundaries dramatically affects the tolerance of the control system to multiple
failures. This is a cost/complexity trade-off with cheaper systems being designed to cater for a single fault
anywhere in the architecture and better-designed systems catering for a fault at each hierarchical level
of the design. The latter is more difficult to achieve, but will generate a system with much better RAM
(Reliability, Availability and Maintainability) figures despite the increased hardware count. Figure 7.12c
illustrates 2 different approaches to the provision of duplication in a hierarchical control system, namely
"simple" and "independent". The simple method maintains total isolation between the 2 systems, whereas
the more complex subdivision of each system into smaller, interconnected systems, ultimately yields a
more fault-tolerant system.
7.12.2. Auxiliary Power Supplies

Auxiliary power supplies are a key area for ensuring that duplication is adequate [10]. It is common practice
to supply control systems from a station battery, so that even if all AC auxiliary supplies fail, the control
system is unaffected.
What happens if the battery fails? This can be avoided by duplicating the battery and its associated equipment,
and providing two independent power feeds to each lane of control so that either battery can power the
complete control system.
The main disadvantage of this design choice is the cost of the duplication. Since the cost of the control/
protection hardware is only a small proportion of the overall system cost, it is generally preferable to reduce
the risk of tripping the HVDC equipment over its operational lifetime, rather than getting a slight reduction
in the initial investment cost.
The auxiliary distribution system within the control suite is even more important. All active power conversion
devices should be duplicated and each supply rail zoned to limit the scope of a fault. Each duplicate power
supply path should be effectively monitored to detect sleeping faults3 and all power conversion devices should

CHAPTER
Control Lane 1 VBE
Valve
Control Lane 1 VBE
Protection M1
Voting
Protection M2 Trip
Protection M3
Fig 7.12a– A duplicated HVDC control system
be used at 50% to 70% of their rated operating power to reduce their temperature rise and improve reliability.
Isolated current-limiting power supplies fitted with alarmed ground fault detection should be used for all
plant whetting supplies to allow correct operation to continue in the presence of a single earth fault in the
plant wiring.
7.12.3. Communications Systems

Most modern HVDC control systems make very effective use of high-speed, serial communications systems
to connect to remote control centers, local and remote HMIs, and to access the HVDC plant’s digital and
analog inputs and outputs [11]. It is also increasingly common for control systems to use either proprietary
or standardized serial communications systems to communicate between core control functions.
Communications between the different parts of the control and HMI systems has often been based on the
IEEE 802 [9] suite of protocols (100 base TX, 100 base FX, etc). Recently the use of the IEC61850 protocol
[12] has become more universal across many substations for both internal and external communications,
and this has allowed the more homogeneous integration of the HVDC control system, the converter station
equipment, and remote systems. Considerable care is required during the design of these networks (for
example, the choice between ring and radial structures) to ensure fault tolerance. The effect of the failure
of a physical device such as a router, a communicating node on the device it is communicating with and
the consequences and likelihood of a network-wide issue such as a packet storm must all be considered.
For systems with remote access, it is also important to consider the susceptibility of the system to both
deliberate or coincidental maloperation of the external network.
Many of these issues can be controlled quite simply by the correct selection of network hierarchy, duplication
and the use of high quality ethernet devices with hard-coded security mechanisms. As with any system, it
is important that the design engineers fully understand the potential failure modes of the communications
3 – A sleeping fault is a fault in a currently unused or duplicated path of functionality that does not affect the system
functionality until it is exposed by a mode change or a failure of its duplicated partner path. E.g. if two fully rated power
supply units are correctly connected in parallel, one unit can fail without affecting the operation of the system they supply.
The failed unit is said to be a sleeping fault.

CHAPTER
Fig 7.12b– eLumina Converter Control and Protection (CCP) cubicle - One lane
system and the protocols and equipment that are used. Each active device used should provide local status
information to the control system, to allow both early warning of imminent failures and the identification of
sleeping faults.
The communications systems and mechanisms used must support the use of redundant communicating
devices. The HMI system needs to accept information from the active lane of the control system, but must
also be capable of switching to the other lane should the active lane fail. This is often a many-to-many
relationship as the HMI system should be fully duplicated and be able to change over independently of
the control system.
7.12.4. Control System Inputs and Outputs

The consequence of the industry trend towards more detailed HVDC plant and control system self-
monitoring is to require the control systems to have an increased I/O count. While communications systems
assist greatly by reducing the physical I/O count and allow an increase in the logical signal count without
a linear reduction in availability, most HVDC systems still require significant quantities of physical coil and
contact I/O to interface with the HVDC plant.
The large number of interface devices used significantly affects the overall reliability, even when highly
reliable electronic relays are used. These devices must therefore be duplicated: for example, separate I/O
sub-systems, each with separate fault-tolerant 12 Mbit fiber-optical, Profibus™ rings, can be used for each
control rack and protection rack in each control lane at each hierarchical control level (pole, bipole, station)
of the HVDC control system.
If the duplicated interface devices need to be connected to a single physical plant contact, the design of
the contact whetting supply needs special care to avoid a single point of failure.

CHAPTER
The use of two physical signals to represent each
of the states of a physical device is also common:
‘not Open’ is not assumed to mean ‘Closed’, thus
both ‘Open’ and ‘Closed’ signals are provided. This
way the control system can detect either a failure
of the controlled device, or a failure of the wiring to
A Aʼ A Aʼ
the device, by monitoring the status of both signals
against a known good timing sequence. This not
only allows the identification of a fault, but can also
allow the control system to take remedial action B Bʼ B Bʼ
by using an alternative plant device (assuming one
is available), thus maintaining full availability of the
operational plant. C Cʼ C Cʼ
The mapping of the plant equipment to the
control system architecture can have an
important effect on the overall availability. If the D Dʼ D Dʼ
plant contains multiples of similar items such
as harmonic filters that are directly associated
(logically and physically) to one piece of
E Eʼ E Eʼ
equipment (for example, a pole control system),
then the filter cannot be used by another pole
control system should the need arise. This may
TRIP OK
require the station power to be limited and create
a consequential loss of availability, even though a
suitable filter was actually available. By choosing
a different architecture, the filters (including any
Fig 7.12c– Simple duplication (left) vs independent duplication at each
spare filters) could be controlled in a manner that level (right)
allows multiple poles to share the same filters. 7.12c
7.12.5. Maintenance of the System
A key aspect of a duplicated system is the ability to safely maintain the off-line part of the system while
the duplicated section continues to function. This functionality must be part of the inherent design of the
control system and carefully considered for each project-specific implementation. It typically requires the
control system to be physically, electrically and logically segregated, or perhaps even installed in different
rooms or buildings.
Clearly, the system is at a higher risk of loss of availability if one of the duplicated portions of the system is
already out of service. For example, one additional fault in the remaining (active) duplicated part of the system
might cause the HVDC power flow to stop. Hence, the time taken to repair (or service) the out-of-service
part of the system is critical to the overall availability. This actual time taken depends heavily on both the
availability of any replacement parts.
The control system itself can dramatically increase the effectiveness of the repairing engineer by providing:
detailed self-monitoring and reporting systems; the ability of both the customer and the manufacturer
specialist engineers to access the systems remotely; and by providing a suitable hierarchy of diagnostic detail.
The on-site spares manifest is also critical, as is the lead-time to replace any spares consumed. Careful
consideration of the effect the failure of each element of the control system has on the overall system
reliability (which is impacted by these lead-times), is required to assess the overall effect on the system’s
availability. A detailed RAM study is required to determine the most effective level of on-site spares holdings.
7.13. CONTROL FOR SERIES CONNECTED CONVERTERS

HVDC transmission systems have operated at voltages up to 500 kV for many years. Recently, transmission
voltages have increased dramatically in order to increase the transmitted power. UHVDC schemes at
±800 kV have been built and are now operational. An ±1100 kV scheme is presently being constructed,
which is the largest known UHVDC system. Such schemes can transmit many gigawatts of electrical power.
Clearly, the consequences of an unexpected interruption of this level of power are likely to be even more
severe than an interruption of power for a lower power HVDC link.
CHAPTER
As it is more efficient to provide very high power
HVDC transmission schemes as bipolar schemes,
+V dc1 I dc1
each pole end could have four distinct 400 kV bridges +800ÊkV
for a ±800 kV scheme as illustrated in Fig 7.13a.
Bridge x
In the event of a failure of a 12-pulse bridge or its
2
related control systems the high speed bypass +400ÊkV
breaker is closed allowing the pole to continue to
operate with the same current, but with the DC Bridge
voltage on that pole reduced to 50%. As a result, only 1
x
25% of the total bipole power is lost, compared with
50% when a single 12-pulse bridge is used per pole.
This very useful degraded operation mode requires
that the overall HVDC control and protection
system is able to operate with either one or two
large 12 pulse converters per pole, and that any Bridge x
1
common parts of the control system have an even
lower risk of failure. In addition, automated start-up -400ÊkV
and shutdown routines need to be developed to
Bridge x
allow series bridges to be inserted into or removed
2
from service with minimal interruption to power
-800ÊkV
flow. Taken to the extremes two completely
-V dc2 I dc2
separate, (but closely coordinated) individually
duplicated control systems are required per bridge.
Fig 7.13a– Bipole end with 4 bridges and bypass devices
In practice a detailed Reliability, Availability and
Maintainability study is required to determine the
optimum solution.
7.13a

CHAPTER
BIBLIOGRAPHY
[1] Y. B. Yates, “Power Engineering for the New Cross-Channel Link”, Electronics & Power, p. 77, January
1982.
[2] B. R. Andersen, D. R. Monkhouse, R. S. Whitehouse, J. D. G. Williams, V. K. Pasher, D. Kumar,
“Commissioning the 1000 MW Back to Back HVDC Link at Chandrapur, India”, CIGRÉ Paper 14-114, Paris,
1998.
[3] J. Reeve and S. Kapoor, “Analysis of Transient Short-Circuit Currents in HVDC Power Systems”, IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-90, Issue 3, May 1971, p. 1174-1182.
[4] Rashmi Aspi Keswani, “Identification of Fault in HVDC Converters Using Wavelet Based Multi-Resolution
Analysis”, Proceedings of the 2008 First International Conference on Emerging Trends in Engineering and
Technology, Vol. 00, 2008, p. 954-959.
[5] “Overvoltages on HVDC Cables”, CIGRÉ Brochure 086, August 1994.
[6] R. S. Whitehouse, “Protecting A HVDC Link Against Accidental Isolation From Its Receiving AC System”,
IEEE 92 SM 468-9PWRD.
[7] “Terminology for high voltage direct current (HVDC) transmission”, IEC 60633:2009.
[8] “Transmission protocols – Network access for IEC 60870-5-101 using standard transport profiles”,
IEC 60870-5-104.
[9] “IEEE Standard for Local and Metropolitan Area Networks: Overview and Architecture”, IEEE 802.
[10] “Distributed Network Protocol”, DNP3, DNP3 users Group.
[11] T. J. Stott, C. C. Davidson, “Structure and Functionality of an 800 kV HVDC Control system”, GridTech
2009.
[12] "IEC 61850" Communication networks and systems for power utility automation", and associated
sub-documents.

CHAPTER
8| DC TRANSMISSION CIRCUITS
TOC 434 | DC
8 DC TRANSMISSION
CIRCUITS
For HVDC converter stations to perform their role, they need to
be connected together. This chapter describes the different types
of HVDC scheme transmission connections and their particular
design, configurations and installation features.
You will learn about overhead lines, with all of their mechanical,
meteorological and insulation issues, together with how such
systems are designed to ensure that they are safe, economic and
have minimum right-of-way requirements. Topics include how
the lines can be designed to not interfere with radios and other
telecommunications equipment, or emit unacceptable levels of
electrical or magnetic fields and audible noise. The economics for
both capital equipment and losses are also discussed.
You will learn about the differences between AC and DC cables
and about different types of underground and submarine DC cable
technologies available to inter-connected converter stations. You
will also learn about the methods used for laying long undersea
cables using specialized high-tech equipment.
An important feature of DC transmission is its ability to use the earth
or the sea as a return current path using electrodes. This chapter will
cover this topic and explain this method’s environmental impacts to
provide a conduction path between converter stations.
TOC HVDC: Connecting to the future | 435

8|| DC
DC TRANSMISSION
TRANSMISSION CIRCUITS
CIRCUITS
Chapter contents
8. DC TRANSMISSION CIRCUITS................... 434
8.1. HVDC OVERHEAD TRANSMISSION LINES....................... 437

8.1.1. Overhead Configuration......................................................... 437
8.1.2.
Meteorological Conditions and
Mechanical Design ................................................................... 437
8.1.3. Electrical Design......................................................................... 439
8.1.4. Insulation Coordination.......................................................... 441
8.1.5. Pole Spacing ................................................................................ 444
8.1.6. Conductor Current Carrying Capability and Sags....... 445
8.1.7. Tower Height............................................................................... 447
8.1.8. Lightning Performance............................................................ 447
8.1.9. Right-of-Way Requirements for Insulation ................... 447
8.1.10. Corona Effects............................................................................. 448
8.1.11. Radio Interference (RI) and Audible Noise (AN)........... 451
8.1.12. Electric and Magnetic Fields................................................. 453
8.1.13. Health Effects.............................................................................. 454
8.1.14. Public Perceptions..................................................................... 456
8.1.15. Economics..................................................................................... 458
8.1.16. Other Designs.............................................................................. 460
8.2. HVDC SUBMARINE AND

UNDERGROUND CABLES....................................................... 461
8.2.1. Introduction ................................................................................ 461
8.2.2. Basic Thermo-Electrical Calculations............................... 462
8.2.3. Basic Cable design.................................................................... 464
8.2.5.4. Laying Submarine Cables....................................................... 470
8.3. EARTH/SEA ELECTRODES...................................................... 474

8.3.1. Environmental Effects............................................................. 476
8.3.2. Corrosion Effects........................................................................ 478
8.3.3. Direct Current in Power Transformer Neutrals............ 481
8.3.4. Gas Production........................................................................... 481
8.3.5. Compass Deviations................................................................ 481
8.3.6. Thermal Effects........................................................................... 482
BIBLIOGRAPHY ............................................................................................ 482

8.1. HVDC OVERHEAD TRANSMISSION LINES
This section presents an introduction to the design of overhead transmission line circuits used in HVDC
interconnections.
8.1.1. Overhead Configuration

For long distance overhead HVDC transmission, a bipolar configuration is the most common, although some
projects are constructed with the monopolar configuration. Bipolar HVDC schemes may be used in monopolar
mode, either as an intermediate stage in the development of the project, or during outages of one pole.
Consideration must be given to the basic configuration of the transmission line and the cost versus reliability
factors for the project. The various transmission line configurations and their reliability considerations showing
the transmission capacity after a permanent line fault are shown in the table below [1].
Remaining transmission capacity

Loss of one pole
Variant Tower configuration
Ground return Tower breakage
permitted not permitted
Single
0 0 0
monopolar line
Single
50 (100) 0 0
bipolar line
Double
100 100 0
bipolar line
Two
50 (100) 0 50 (100)
monopolar lines
Two lines
(bipolar or 100 100 100
homopolar)
Table 8.1a– Reliability of alternative configurations [1]

Note: Values between brackets apply to the case where the converters can be paralleled in the station and the remaining pole
has adequate current carrying capacity to take up the load lost as a result of the line fault.
In this section, unless otherwise stated, we will mainly be considering the bipolar line configuration.
8.1.2. Meteorological Conditions and Mechanical Design

For the design of transmission lines, the following meteorological conditions have to be taken into
consideration:
➙ Temperature: Every Day Stress (EDS), minimum, mean maximum, coincident with wind
➙ Wind: Gumbel distribution, defined by the mean value of the yearly maximum wind velocities and their
standard deviation. Values are measured at 10 m above soil every 10 minutes.
➙ Ice: Distribution of weight per unit length or uniform radial thickness around conductors and shield
wires
➙ Humidity: Pressure and air density

CHAPTER
The mechanical design consists of the evaluation of conductor and shield wire sag and tension, and of the
forces acting on the support towers.
Tensions are determined by applying the change of state equation below:
 S2   w  2  w  2   ( H − H1 ) 
 2  −  1   − ε (T2 − T1 ) −  2 =0 [Eqn 8.1.2a]
 24 
 H 2   H 1    E A 
Where:
w is the conductor weight per unit length or the equivalent vertical component due to wind and ice
H is the horizontal component of the conductor tension
T is the temperature
e is the length variation coefficient
S is the span length
E is the elasticity modulus
A is the conductor cross section
*Subscripts 1, 2 indicate the state.
In general, ‘state 1’ refers to the most frequent condition, called EDS- Every Day Stress (EDS temperature, no
ice, no wind). The value H1 is chosen as the design criterion (around 20% of UTS/RTS Ultimate/Rated Tensile
Strength for the conductor and 11% for Extra High Strength (EHS) steel shield wires to take vibration into
consideration).
‘State 2’ generally refers to: maximum and minimum conductor temperature, maximum wind or ice, and
the ice/wind combination [2]. As design criteria, tensions must not be above 30% in the case of minimum
temperature without wind, and not above 50 to 70% for extreme conditions combining wind and ice. [4]
Once the H values have been obtained, the corresponding sag can be calculated
  Sl  
s = c  cosh   − 1 [Eqn 8.1.2b]
  2 c 
H
c= [Eqn 8.1.2c]
w
Sl is the span length

For the calculation of the ice/wind equivalent weight that will lead to the evaluation of the tensions and also
the forces acting in the towers, the following methodology is recommended. [2]
Considering a Gumbel distribution (extreme values), the wind velocity to be considered, depending on the
return period, is determined by:
S
Vt = V + (Y − C2 ) [Eqn 8.1.2d]
C1
1
Y = − ln (− ln (1 − )) [Eqn 8.1.2 e]
T
Where:
Vt is the wind velocity (m/s) with return period T
V is the wind velocity - mean (m/s)
S is the standard deviation (m/s)
C1, C2 are coefficients dependent on the sample size: C1 = 1.11237 and C2 = 0.53622, for a sample of 30 years [5]
T is the return period (years)
CIGRÉ [4] uses the following sample values:
➙ Wind intensity: mean of the sample (10 min average wind) = 18.4 m/s
➙ Standard deviation = 3.68 m/s;

CHAPTER
This gives:
Return period (yr) Wind velocities (m/s)
50 29.52
150 33.175
IEC design criteria indicate the use of wind intensity with return periods of 50, 150, 500 years, for a reliability
index of 1, 2 or 3. In some countries, wind intensity with a return period of 150 years is used for the
determination of forces on the tower, whereas wind intensity with a return period of 50 years is used for
electrical design (conductor swing together with minimum clearances). [2]
Once the wind velocities have been found, the pressures on the conductors and shield wires can be
determined. [2] The type of terrain, conductor height, wind angle, air mass, span factor, etc. must all be
considered.
In regions subjected to icing conditions, design conditions that include only wind, only ice and a combination of
ice/wind must be considered. The return period of combined events of ice and wind are shown in Table 8.1c [2].
Return period
Return period T of the variable having Return period of remaining
Reliability level
(years) a low probability of variables (index H)
occurrence (index L)
1 50 50 Average of yearly maximum values
Table 8.1b – Ice data - return period
For any selected reliability level, three loading conditions are defined as shown in Table 8.1b.
Loading Effective drag

Ice weight Wind velocity Density
conditions coefficient
Condition 1 gL ViH CiH d1
Condition 2 gH ViL CiH d1
Condition 3* gH ViH CiL d2
* In practice, it was found that condition 3 is not critical for design purposes
Table 8.1c– Ice/wind combination
Finally, for tower design a certain number of loading conditions must be considered, including: high wind at
different angles, ice, wind/ice, rupture of the conductors in one pole or one shield wire, unbalanced loads
during construction, line angle at tower and conductor weights [4].
8.1.3. Electrical Design

Electrical design covers the following areas: overvoltages, insulation coordination, conductor temperature
and sags, lightning performance and corona and fields effects. These aspects determine the tower top
geometry and right-of-way (ROW) of the line.
8.1.3.1. Overvoltages
Line insulation levels are dependent on voltage stresses that reach the air gaps and thus are chosen to be
the best compromise between satisfactory electrical performance and reasonable costs.
The following voltage stresses must be considered to define the tower top geometry of the towers:
➙ sustained due to operating voltage
➙ transient due to lightning
➙ switching surge overvoltages
The switching surge overvoltages in a HVDC system occur in the DC as well as in the AC part of the system.
In the context of HVAC systems, overvoltages originate from switching operations: line energization, line
reclosing, load rejection, fault application, fault clearing and reactive load switching, and all should be

CHAPTER
evaluated. In the context of HVDC systems, the aforementioned overvoltages are also considered for the
converter station AC side insulation design: by the use of surge arresters, the overvoltages are limited to
values corresponding to the arrester maximum Switching and Lightning Impulse Protective Levels (SIPL and
LIPL). The surge discharge capability of the arrester needs to be verified as part of the overvoltage studies for
equipment specification.
Regarding switching surges on DC lines, fault application is the only type of the above listed switching
overvoltages to be considered, because of the intrinsic process of the HVDC system. For line energization
and reclosing, the DC voltage is ramped up smoothly from zero, and in the reclosing process the line
de-energization process eliminates the trapped charge by ramping down the voltage. Load rejection generally
does not transfer overvoltages to the DC line and DC filter switching does not cause overvoltages.
Lightning overvoltages may start a fault in the DC line, however their effect is smaller in comparison to AC
system faults because the fault current will be limited by the HVDC station controls. In these cases, the line
voltage is ramped down and after a sufficient time has elapsed for the trapped charge to discharge, the voltage
is ramped up again to the nominal value or to a reduced voltage value (around 80% for example).
Shield wires are normally installed above the DC poles to reduce the number of faults by providing appropriate
shielding against lightning. The major point in the design is then to place the shield wires in the right position.
Shield wires may also be used as a communication medium for system control, in which case their design
needs to take both functions into account.
Sustained overvoltages on the DC side of HVDC systems do not occur due to the intrinsic voltage control
process of the HVDC operation. It should be noted that overvoltages on the DC side may appear due to
harmonics/filter/smoothing reactor resonance. However, this is a problem to be solved by the correct design
of these sub-systems, so these kinds of stresses will not be considered for the insulation design of the DC line.
8.1.3.2. Overvoltage Resulting from the Simulation of a Fault

Overvoltages may be calculated with suitable circuit analysis software such as ATP (Alternative Transient
Program) or PSCAD/EMTDC (Power System simulation software supplied by Manitoba HVDC Research
Center), using models to take into consideration the AC in-feed, the converter stations, the smoothing
reactor and DC filters, and the DC line.
For the initiation of the fault in the negative pole, a positive surge of value equal to the pre-fault voltage is
2.1
2.0
1.9
OvervoltageÊ(p.u.)
1.8 FaultÊat
Mid
1.7
1.6
1.5
1.4
0 375 750 1,125 1,500
Rectifier Mid Inverter
TransmissionÊlineÊlengthÊ(km)
Fig. 8.1a– Maximum overvoltage in the healthy pole, due to a fault in the middle of the other pole
BACK TO 440 | 8.1a

CHAPTER
injected in the fault point, and the resulting surge travels in both line directions, reflecting at the line ends and
coming back to the fault point. The traveling wave is coupled to the positive pole resulting in an overvoltage
whose values are due to the combination of the DC voltage, the forward-travelling wave and the reflected wave.
The maximum overvoltage occurs for a fault initiated in the middle of the line, at a time close to the travel
time (of the first reflection) to the line end and back again to the midpoint. Faults in locations other than the
midpoint produce smaller overvoltages. Due to this behavior, the overvoltage profiles down the line are similar
for every line length. Line end equipment (DC filters, smoothing reactor and the source) play an important
role, as they define the traveling wave reflection coefficients. Line models using the parameters as functions
of frequency are of paramount importance.
Fig. 8.1a shows the maximum overvoltage profile along the line in the healthy pole for a fault initiated at
the midpoint of the other pole for a ± 500 kV, 1500 km line [4].
The maximum line overvoltage value is 2.03 p.u., however the overvoltages are above 1.8 p.u. (10% lower)
for only 1/4 of the line length. Taking into consideration that the standard deviation for insulation switching
surge withstands is 6%, one can conclude that the overvoltage in the major part of the line does not
contribute to the risk of failure and therefore the line is designed considering mainly the maximum value
and the line length close to the midpoint. If the line parameters are modeled as frequency dependent, then
the overvoltages are much lower with a maximum of 1.69 p.u. (see Fig. 8.1b).
Fig. 8.1b shows the overvoltage profile for faults initiated at other line positions. Frequency dependent
parameters of the line are used in these cases. It can be seen that very few values are all below 1.8 p.u. [4]
As just mentioned, the overvoltage values depend on the models of termination impedances (the source
and line).
8.1.4. Insulation Coordination

Insulation coordination refers to the determination of the safety clearances and the number and types of
insulators to be used in the insulator strings.
The number of insulators is initially selected based on the maximum DC voltage withstand and on the
assumption of a certain pollution level. The number of insulators obtained by these criteria is then verified
by considering the overvoltage values.
1.8
FaultÊat
1.7
Sending
1.6
1/8
1/4
OvervoltagesÊ(p.u.)
1.5
3/8
1.4 Mid
5/8
1.3
3/4
1.2 7/8
Receiving
1.1
1.0
0 375 750 1,125 1,500
TransmissionÊlineÊlengthÊ(km)
Fig. 8.1b– Overvoltage profiles: faults in different positions (line with frequency dependent parameter)

CHAPTER
8.1b
The clearances to be determined are conductor-to-tower cross arm, conductor-to-tower or objects
(lateral), conductor-to-ground or objects (at the ground), and conductor to guy wires. They are calculated
for switching surge overvoltage withstand. However, the clearance to tower and guy wires as well as to
the edge of right-of-way need to be verified in conditions of insulation string swing due to wind in order to
prevent flashovers and the contact of objects (such as trees) at the border of the right-of-way.
8.1.4.1. Number of Insulators

Anti-fog insulator designs are recommended for DC lines. The number of insulators in an insulator string is
defined by assuming a creepage distance compatible with the contamination in the line route. The creepage
distances recommended are:
Contamination Very light Light Moderate Heavy

severity (mg/cm2) < 0.005 0.005 – 0.02 0.02 – 0.05 > 0.05
Specific creepage
2.0 – 2.5 2.5 – 3.2 3.2 – 4.0 4.0 – 7.0
distance (cm/kV)
Table 8.1d– Creepage distances [7]
For agricultural areas and woodlands 23 mm/kV is recommended [6], and for outskirts of industrial areas 40
mm/kV is recommended. Some references recommend even lower creepage distances down to 20 mm/kV for
areas classified as ‘with very light pollution’. As a reference, the Itaipu lines (‘light pollution - agricultural area’)
were designed for 27 mm/kV and have shown adequate performance over more than 20 years of operation.
Considering a specific creepage distance of 30 mm/kV, the numbers of insulators for ± 600 kV and ± 800 kV
are respectively 36 and 46 insulators, each with creepage distance 508 mm: the insulator string lengths
are 6.2 and 8.2 m (considering also 165 mm for insulator pitch and 250 mm string hardware size).
Generally glass or porcelain insulators are used, however if the line crosses highly contaminated areas, then
composite insulators may be preferable to avoid large insulator string lengths. Insulators for DC applications
need to be specially designed because of electrochemical corrosion problems, especially on cap and pin
insulators and with a profile with longer creepage distance (anti fog type).
8.1.4.2. Air Clearances to Withstand Operating Voltages

In order to determine the minimum necessary conductor-structure clearances for operating voltage
insulation, the following principles must be considered:
➙ Withstand voltage for the most unfavorable condition: positive polarity, conductor-to-structure.
➙ Maximum operating voltage and correction for the atmospheric conditions: conductor swing due to
wind with established return period (e.g. 50 yr)
The conductor-to-structure distances are obtained from the Green Book, [6] (varying from 0.7 m for +300 kV
to 1.9 m for +800 kV).
The swing angle of the conductor due to wind is calculated using the adequate span factor, that is a function of
the wind intensity (effectiveness factor) and the minimum vertical/horizontal span ratio. [3]
Recommended swing angles for ACSR conductors vary from 44 to 57 degrees depending on the conductor
size: 1,274 mm2/ 2,515 MCM (thousand circular mils) to 403 mm2/ 795 MCM. Other important conditions are:
minimum ratio of vertical/horizontal span equal to 0.7, wind intensity equal to 29.52 m/s for 50 year return
and terrain type B. [4]
8.1.4.3. Clearances for Switching Surge Withstand

The withstand capability of the gaps is estimated by:
V50 = k ⋅ 500 ⋅ d 0.6 [Eqn 8.1.4a]
Where:
V50 is the insulation critical flashover voltage (50% probability), in kV
d is the gap distance (m)

CHAPTER
k is the gap factor: k = 1.15 conductor – plane
k = 1.30 conductor – structure under
k = 1.35 conductor – structure (lateral or above)
k = 1.40 conductor – guy wires
k = 1.50 conductor – cross arms (with insulator string)
The above equation applies to Extra High Voltage systems when 2 < d < 5 m. An alternative equation when
5 < d < 15 m, is:
3400
V50 = k . [Eqn 8.1.4b]
1+ 8 / d
Normally the clearances are determined based on the fault application overvoltage profiles, aiming at a
certain ‘flashover risk of failure’ target (design criteria) [4]. For DC lines the overvoltage to be considered is
the fault case described in section 8.1.3.2.. To fully evaluate the risks, the overvoltage profiles considering
faults in several positions along the line must be evaluated. The risk of failure is obtained by considering
the risk of the individual gaps at any position on the line and their parallel associations.
It should be noted that if the line is designed with I-insulator strings, then it is recommended to consider the
effect of possible winds simultaneously with the overvoltages in the risk calculation. There are two approaches
for taking this into account: first, by calculating the clearances for an established risk and admitting that such
clearances will be maintained with a certain swing due to wind [3]; or considering the simultaneous occurrence
of wind and overvoltage and calculating the composite risk.
8.1.4.4. Clearances for an Established Flashover Risk of Failure (no wind)

The following (Fig. 8.1c and Fig. 8.1d) figures show the clearances for some of the gaps mentioned above,
as a function of the line voltage. They were designed for a flashover risk of failure of 1/50 yr, and the
overvoltages were calculated using the frequency dependent parameter line model (J. Marti) [4].
8.1.4.5. Switching Overvoltages with Conductor Displacement due to Wind

CIGRÉ recommends the adoption of a swing angle caused by a wind intensity corresponding to 1% probability
of being exceeded in a year, together with the occurrence of switching surge overvoltages. [3]
Conductor-to-tower
8.0
7.0
6.0
5.0 750Êkm
ClearanceÊ(m)
1,500Êkm
4.0
2,250Êkm
3.0 3,000Êkm
2.0
1.0
0.0
300 400 500 600 700 800
VoltageÊ(kV)
Fig. 8.1c– Conductor to tower clearances
BACK TO DC Transmission Systems: Line Commutated Converters 443

CHAPTER 8.1c |
Conductor-to-groundÊ(object;Ê4.5Êm;Êunder)
8.0
7.0
6.0
5.0 750Êkm
ClearanceÊ(m)
1,500Êkm
4.0
2,250Êkm
3.0 3,000Êkm
2.0
1.0
0.0
300 400 500 600 700 800
VoltageÊ(kV)
Fig. 8.1d– Conductor to object clearance (add 4.5 m to get conductor to ground distance)
Using the wind distribution example above, the wind intensity is 13.54 m/s. The swing angles caused by this
wind varied from 13 to 19 degrees for ACSR conductors varying from 1,274 mm2 to 403 mm2. [4]
8.1d
It should be noted that considering both the conductor swing due to the wind, with 1% probability of being
exceeded in one year and the clearances corresponding to a risk of 1/50 years, that the final flashover risk
will be much smaller than 1/50, therefore the stated criteria are conservative. (In the case for a risk of 1/100
yr the swing angles varied from 6 to 9 degrees.)
An alternative approach would be to find a clearance considering the composite risk resulting from overvoltage
distribution and a swing due to the wind distribution.
8.1.5. Pole Spacing

To evaluate the pole spacing, the swing angles of I-insulator strings and clearances are used.
The minimum pole spacing DPTO is:
DPTO = (R + dmin + (L + R) sinq ) ⋅ 2 + w [Eqn 8.1.5a]
Where:
dmin is the operating voltage or switching surge clearance
a
R= is the bundle radius [Eqn 8.1.5b]
2sin( π / N )
a is the sub conductor spacing
N is the number of sub conductors in the bundle
L is the insulator string length
q is the swing angle for the wind speed as above
w is the tower width at conductor level and varies from 1.2 to 2.5 for voltages from ±300 to ±800 kV
The minimum pole spacing required for switching surges and operating voltage conditions for ±800 kV
bipole lines are shown in Fig. 8.1e. [4]
It can be seen from Fig. 8.1e that the operating voltage criterion governs the pole spacing for ±800 kV
voltages and of course for the other voltages as well.
In the case that V strings are used, there will be no swing angles due to wind at the towers and the clearance
requirements for switching surges will determine the pole spacing. However, the V strings having length (L)
CHAPTER
PoleÊspacingÊforÊ800ÊkV
22
OV:ÊOperatingÊVoltage
21 SS:ÊSwitchingÊSurge
20 OV
19 SSÊ750Êkm
PoleÊspacingÊ(m)
18 SSÊ1,500Êkm
17 SSÊ2,250Êkm
16 SSÊ3,000Êkm
15 OV : Operating Voltage
SS : Switching Surge
14
500 1,000 1,500 2,000 2,500 3,000
ConductorÊcrossÊsectionÊ(MCM)
Fig. 8.1e– Pole spacing (±800 kV, 750 to 3,000 km)
will be inserted in the tower, meaning that the minimum pole spacing (PSmin) for installation will be:
8.1e
PSmin = 2 ⋅ L ⋅ cos (45°) + w [Eqn 8.1.5c]
It is assumed here that the V string angle is 90 degrees, however this opening can be reduced.
The pole spacing requirement is otherwise calculated by:
DPTO = (dmin + R) ⋅ 2 + w (provided that DPTO > PSmin) [Eqn 8.1.5d]
8.1.6. Conductor Current Carrying Capability and Sags

The current carrying capability of ACSR conductors can be calculated based on CIGRÉ recommendations
[8] considering the aspects below:
Heat gain = heat loss
PJ + PM + PS + Pi = Pc + Pr + Pw [Eqn 8.1.6a]
Where:
PJis the Joule heating PM is the magnetic heating
PSis the solar heating Pi is the corona heating
Pcis the convective cooling Pr is the irradiative cooling
Pwis the evaporative cooling
The maximum temperature of an aluminum conductor is limited generally to 90 °C (a design criterion
commonly used in many countries) for steady-state and in emergency or short-term conditions. Temperatures
even above 100 °C could be accepted for non-special conductors (thermal resistant conductors can withstand
much more in steady-state conditions). However, the conductor is selected based on economic criteria (cost
of line plus losses) leading finally to a much lower maximum operating temperature in normal conditions (~55
to 60 °C). Therefore, in practice, 90 °C will usually only apply to pole conductors under abnormal conditions
as well as to electrode lines and metallic return conductors.
Fig. 8.1f shows the current capability for some conductors assuming the following data [4]:
Wind speed (lowest) 1 m/s
Wind angle related to the line 45 degrees
Ambient temperature 35 °C
Height above sea level 300 to 1,000 m
CHAPTER
ConductorÊcurrentÊcarryingÊcapability
2,500
2,000
1,500 90¡
ÊCurrentÊ(A) 70¡
60¡
1,000 50¡
500
0
0 500 1,000 1,500 2,000 2,500 3,000
ConductorÊcrossÊsectionÊ(MCM)
Fig. 8.1f– Conductor current carrying capability
Solar emissivity of surface 0.5

Conductor solar absorption coefficient 0.5
Global solar radiation 1,000 W/m2
8.1f
Increase in conductor temperature leads to a decrease in the horizontal tension and consequently an increase
in the sags.
ConductorÊSag
23
22
21 50¡
60¡
SagÊ(m)
20
70¡
19 90¡
18
17
500 1,000 1,500 2,000 2,500 3,000
ConductorÊcrossÊsectionÊ(kcmil)
Fig. 8.1g– Conductor sags
Fig. 8.1g shows the sags for conductor temperatures in the range of 50 to 90 °C, considering a span of 450
m. For ‘state 1’ condition, the EDS equals 20% of RTS with a temperature of 20 °C.
8.1g
It can be seen that the sags vary from 18 to 22 meters, depending on the conductor temperature and type of
conductor. It should be noted that the conductors considered in this graph are selected from manufacturer’s
catalogs and that conductors with the same aluminum but different steel contents will have different sags.

CHAPTER
8.1.7. Tower Height
The conductor height at the tower (hp) is calculated by:
hp = CS + sg + Ext + R [Eqn 8.1.7a]
Where:
hp is the distance from the center of the bundle to ground at tower
CS is the clearance to ground at mid-span determined by insulation or electric field criteria
sg is the conductor sag at maximum temperature
R is the bundle radius
Extis the height of any above ground extensions installed at the base of the tower
The shield wire height (hg) at the tower is:
hg = hp + R + dis + DG [Eqn 8.1.7b]
dis is the insulator string and hardware length

DG is the shield wire to cross-arm height (generally ~ 2.5 m)
In a CIGRÉ example [4], the hp obtained were 44.8 m and 50.8 m, whereas hg were 53.8 m and 61.8 m
respectively for ±600 kV and ±800 kV lines.
8.1.8. Lightning Performance

In order to get good performances under lightning strikes, the design of HVDC lines should include the use
of shield wires (one or two).
The shield wires reduce the incidence of direct strikes to the conductors. For the strikes that hit the shield wires
there will be an overvoltage that is coupled to the pole conductors, which may or may not cause flashovers.
For optimal design, certain conditions must be considered:
a) The current of the strikes that hits the pole conductors should not produce an overvoltage greater than
the insulation withstand of the line.
b) The closer the shield wires are to the pole conductor, the better will be the performance due to strikes
hitting the shield wires.
c) The tower footing resistance and the corresponding tower footing surge impedance should be low, therefore
requiring the use of an adequate grounding system, generally achieved with buried wires connected to
the tower.
In regions with ice, condition b) may conflict with the requirements of keeping a safe distance from the shield
wire to the pole conductors during icing events.
The clearances at the tower are designed to withstand the operating voltage or the switching overvoltages with
a pre-established risk of failure. Once the required clearances have been defined, the critical impulse flashover
capability of the insulation (50% probability) for lightning surges (fast front overvoltages) can be found.
Once the critical impulse flashover capability and the conductor surge impedance have been found, the
critical threshold current into the conductor for which a flashover will start can be determined, as well as
the corresponding striking distance [4]. The protection angles (shield-wire poles) are determined by drawing
circles parallel to the ground with radii equal to the striking distance centered on the conductors and the
shield wire. For voltages above 500 kV the minimum protection angles are larger than 10 degrees. The closer
the shield wires are to the conductors, the better the lightning performance is for back flashovers, due the
higher coupling factor. As a consequence, a protection angle of 10 degrees can be assumed when using two
shield wires.
8.1.9. Right-of-Way Requirements for Insulation

The Right-of-Way width (ROW) is defined considering the following variables:
➙ Conductor swing and clearances to objects at the border of the ROW
➙ Corona and field effects

CHAPTER
At this point, only the first condition is examined and the clearances for the operating voltage are used.
The swing angles are calculated using the parameters set down in CIGRÉ Brochure 48 [3], except that the
vertical and horizontal span ratio is equal to 1.0. Note that the wind intensity is selected to correspond to the
same return period as for clearances determined for operating voltage. The swing angles obtained in these
conditions varied from 34 to 48 degrees. [4]
The conductor sags are obtained by using ‘state 1’ as the EDS condition and considering the wind load with
the coincident temperature. In [4] the sags obtained varied from 36.5 to 33.6 degrees.
The minimum ROW when using I strings is determined by:
ROW = [(R + L + S) sinq + dmin] ⋅ 2 + PS [Eqn 8.1.9a]
Where:
dmin
is the operating voltage clearance
R is the bundle’s radius (m)
L is the insulators string length
S is the conductor sag
q is the swing angle due to wind
PS is the pole spacing
As an example, the ROW widths obtained in [4] were 66 to 70 m for ±600 kV, and from 73 to 88 m for ±800 kV.
The minimum ROW widths when using V strings are calculated according to the same equation, but
disregarding the insulator string length.
Note that the results (for I or V strings) are partial, as corona effects have not yet been considered. Also note
that only the horizontal design is considered (the vertical design will lead to smaller ROW).
8.1.10. Corona Effects

Corona effects to be considered for the design of HVDC transmission lines include corona losses (CL), radio
interference (RI) and audible noise (AN). [12]
8.1.10.1. Conductor Surface Gradient

For a bipolar HVDC transmission line with a single conductor, the average and maximum conductor surface
gradients Ea and Em, respectively, in kV/cm, are given as [9]:
V
E m = Ea =
 
 
2⋅H
r ⋅ ln   [Eqn 8.1.10a]
2 
 r ⋅  2 ⋅ H  + 1 
  S  
Where:
V is the voltage applied (actually ± V) to the conductors of the line (kV)
r is the conductor radius (cm)
H is the conductor height (cm)
S is the pole spacing (cm)
When bundled conductors are used, the Markt and Mengele’s method is used: the average (Ea) and maximum
(Em) bundle gradients of a bipolar HVDC line, with n-conductor bundles on each pole, are given as [9]:
V
Ea =
nr ⋅ ln 2⋅H
2
 2 ⋅ H
req ⋅   +1 [Eqn.8.1.10b]
S 

CHAPTER
 r [Eqn 8.1.10c]
Em = Ea 1 + ( n − 1) 
 R
Where:
r is the sub-conductor radius (cm)
R is the bundle radius (cm)
req is the equivalent bundle radius (cm)
a
and R= [Eqn 8.1.10d]
2 sin(π / N )
1
n . r n
and req = R⋅  [Eqn 8.1.10e]
R
Where:
a is the distance between adjacent sub conductors (cm)
The equations above give reasonably accurate results for the maximum bundle gradient, with errors not
exceeding 2%, for n ≤ 4 and for normal values of H and S. More accurate methods, such as the method of
successive images, are required for n > 4. For purposes of design and economic evaluations, these equations
are sufficiently accurate.
As an example, for the HVDC line built for the Madeira River power plant (Brazil) whose parameters are: S =
15 m; H = 14.5 m; 4 CAA conductors 2282MCM per pole, sub-conductor diameter 44.252 mm; the maximum
surface gradient is 20.2 kV/cm.
8.1.10.2. Corona Onset Gradient

When the electric field at the surface of a transmission line conductor exceeds a certain value, partial
electrical breakdown of the surrounding air takes place, giving rise to corona discharges. The occurrence
of corona discharges in the immediate vicinity of conductors leads to a number of corona effects that have
important influences on transmission line design.
Corona effects that are generally taken into account in the design of both AC and DC transmission lines
are Corona Loss (CL), Radio Interference (RI) and Audible Noise (AN) and visual effects. In the case of HVDC
transmission lines, the combined effect of DC electric fields and corona-generated ion currents at ground
level must also be taken into account as design considerations. Although corona on transmission lines also
generate ozone, studies on experimental as well as operating transmission lines have shown that contribution
to ambient ozone levels is almost negligible [9].
The conductor surface electric field at which the onset of corona discharges occurs is defined as the corona
onset gradient of the conductor. The corona onset gradient of a given conductor depends on many factors,
the most important being the conductor radius, surface conditions and ambient air density. It also depends
on the type of voltage applied to the conductor, AC or DC, and in the case of direct voltages, the polarity.
Corona onset gradients of cylindrical conductors have been determined experimentally in laboratory studies,
from which the following empirical formula for the corona onset gradient has been derived:
 K 
Ec = mE0 δ 1 +  [Eqn.8.1.10f]
 δ r 
Where:
Ec is the corona onset gradient, kV/cm
r is the conductor radius, cm
m is the conductor surface irregularity factor
E0 and K are empirical constants. According to [10], E0 = 33.7 and K = 0.24 for positive DC, and E0 = 31.0 and
K = 0.308 for negative DC
d is the relative air density
Since no significant differences have been observed between the corona onset gradients at positive and
negative polarities for practical conductors, the following formula, applicable at both polarities, is generally used:

CHAPTER
 0.301  [Eqn.8.1.10g]
Ec = 30 mδ 1 + 
 δ r 
Experimental studies show that for reasonably clean stranded conductors, m varies in the range of 0.75 to
0.85, depending on the relative diameters of the conductor and strands. Surface irregularities may reduce m
to values in the range of 0.6 to 0.8, while conductor surface deposits and precipitation (rain, snow etc.) may
further reduce it to values in the range of 0.3 to 0.6 [9].
It should be noted that although corona onset gradient may serve as a rough guideline, the selection of conductor
bundles for HVDC transmission lines is based mainly on criteria for corona performance defined in terms of CL,
RI, AN and visual effects.
As an example the value of Ec = 26.5 kV/cm (for d = 0.9) for the Madeira River power plant HVDC transmission line.
8.1.10.3. Corona Loss

There are basic differences between the physical mechanisms involved in AC and DC corona loss [9]. On
AC lines, the positive and negative ions created by corona are subject to an oscillatory movement in the
alternating electric field present near the conductors and are therefore confined to a very narrow region
around the conductors. On DC lines however, ions having the same polarity as the conductor move away
from it, while ions of opposite polarity are attracted towards the conductor and are neutralized on contact
with it. Thus, the positive conductor ‘in corona’ acts as a source of positive ions which fill the entire space
between the conductor and ground, and vice-versa for the negative conductor.
The case widely used is the bipolar HVDC transmission line, which produces corona of both polarities: positive
corona from the positive conductor, negative corona from the negative conductor. Unipolar space charges fill
the space between each pole and ground while ions of both polarities mix in the bipolar region between the
two poles and are subject to some amount of recombination. However, since both the corona onset voltage
and the mobility of the ions produced by the corona are polarity-dependent, the space charge regions may
be asymmetric.
Ambient weather conditions have a large influence on corona losses from the line. The losses are lower under
fair weather conditions than under foul weather conditions such as rain, snow etc. However, the ratio of foul
weather to fair weather CL on a DC line is much lower than in the case of an AC line.
Because of the complexity of theoretical calculations and the large number of factors influencing corona on
practical HVDC transmission lines, it is often preferable to obtain empirical formulas derived from a large
amount of data on long-term corona loss measurements made on experimental lines with different conductor
bundles and under different weather conditions.
For bipolar DC transmission lines, some empirical formulas have been developed [11].
 g  d  n  H . S
Pfair = P0 + 50 log   + 30 log   + 20 log   − 10 log  .  [Eqn 8.1.10h]
 g0   d0   n0   H0 S0 
 g  d  n  H . S
Pfoul = P0 + 40 log   + 20 log   + 15 log   − 10 log  .  [Eqn 8.1.10i]
 g0   d0   n0   H0 S0 
Where:
P is the bipole corona loss in dB above 1W/m,
d is conductor diameter in centimeters
g is the conductor surface gradient
n is the number of conductors
H is the height
S is the pole spacing
The reference values assumed are: g0 = 25 kV/cm, d0 = 3.05 cm, n0 = 3, H0 = 15 m and S0 = 15 m.
The corresponding reference values of P0 were obtained by regression analysis to minimize the arithmetic
average of the differences between the calculated and measured losses. The values obtained are P0 = 2.9 dB
for fair weather and P0 = 11 dB for foul weather.

CHAPTER
P(W/m) = 10P(dB)/1 bipole losses in Watts per meter [Eqn 8.1.10j]
8.1.11. Radio Interference (RI) and Audible Noise (AN)

The characteristics of corona-generated RI and AN on DC transmission lines differ significantly from those
on AC lines. Firstly, while all three phases of an AC line contribute to the overall RI and AN of the line,
only the positive pole of a DC line contributes to the RI and AN levels. Secondly, the RI and AN levels of
DC transmission lines under foul weather conditions are lower than those under fair weather conditions.
8.1.11.1. Radio Interference

Both analytical and empirical methods may be used for calculating the RI level of DC transmission lines.
Based on data obtained on experimental as well as operating lines, an empirical formula has been
developed [12] for predicting the average fair weather RI level for bipolar HVDC transmission lines:
g d 19.9 q
RI = 51.7 + 86 log + 40 log + 10{1 − [log(10 f )]2 } + 40 loog + [Eqn 8.1.11a]
g0 d0 D 300
Where:
RI is the radio interference level measured at a distance D from the positive pole with a CISPR instrument,
dB above 1 μV/m
g is the maximum bundle gradient, kV/cm
d is the conductor diameter, cm
f is the frequency, MHz
D is the radial distance from positive pole, m
q is the altitude, m
The reference values are g0 = 25.6 kV/cm and d0 = 4.62 cm.
Based on the results of some long-term studies, the maximum fair weather RI may be obtained by adding 6 dB
[13] and the average foul weather RI may be obtained by subtracting 5 dB from the average fair weather value.
Design criteria for RI from transmission lines are generally based on signal to noise ratios (SNR) for acceptable
AM radio reception. Studies carried out on corona-generated RI from AC and DC transmission lines [6] indicate
that the SNRs for acceptable radio reception are:
a) Background not detectable SNR >30 dB
b) Background detectable 20 dB
c) Background evident 8 dB
The minimum radio station signal requirement in many countries is 66 dB for cities with a population from
2,500 to 10,000 inhabitants.
In many countries the guidelines are for limiting the RI at the edge of the right-of-way to (66-20) = 46 dB or to
maintain a reception quality (b). The equation for calculating noise above the reception level gives the average
fair weather noise. For more stringent criteria, the noise should be below 46-4= 42 dB for 90% probability of not
being exceeded, meaning that 10% of the time the reception will be classified as between criteria b) and c)
above. The reference frequency is considered to be between 0.5 and 1 MHz.
8.1.11.2. Audible Noise

Based on measurements made on experimental as well as operating DC lines and the general characteristics
of corona-generated AN, an empirical formula has been developed [12] for the mean fair weather AN, in
dB(A), as:
q
AN = AN 0 + 86 log( g ) + k log ( n ) + 40 log ( d ) − 11.4 log( D ) + [Eqn 8.1.11b]
300

CHAPTER
Where:
AN is the A-weighted sound level above 20 µPa, on the ground at distance D, in dB(A)
n is the number of sub-conductors
k = 25.6 for n > 2 and k =0 for n = 1.2
AN0 = -100.62 for n > 2 and AN0 = -93.4 for n = 1.2
The fair weather AN, with 90% probability of not being exceeded [13] is calculated by adding 5 dB(A) to the
mean fair weather value obtained above, while the mean AN during rain is calculated by subtracting 6 dB(A)
from the mean fair weather AN.
The Environmental Protection Agency (EPA) in the US recommends that the day-night average sound level
Ldn [14] be limited to 55 dB(A) outdoors. The level Ldn is defined as:
 1  Ld Ln + 10
 
Ldn = 10 log  15⋅10 10 + 9⋅10 10   [Eqn 8.1.11c]
 24   
Where:
Ld and Ln are the day and night time sound levels, respectively.
However, since the highest level of AN from DC lines occurs in fair weather, it may be prudent to limit the Ldn
(10%) of the AN from HVDC transmission lines to 55 dB(A) and this corresponds to 50 dB(A) for Ldn (50%). The
variation in AN between day and night is limited to approximately 1.5 dB(A) [13].
Therefore assuming Ld = Ln = 42 dB(A), then Ldn~50 dB(A).
AudibleÊandÊradioÊnoiseÊAACÊ2282;ÊdÊ=Ê44.253Êmm;ÊPSÊ=Ê15Êm;ÊHÊ=Ê14,5Êm;ÊVÊ=Ê600ÊkV
50
45
40
ANÊorÊRIÊ(dB)
AN
35 RI
30
25
20
0 10 20 30 40 50 60
DistanceÊtoÊtowerÊcenterÊ(m)
Fig. 8.1h– RI and AN noise as function of the distance to tower center on positive polarity side
As a conclusion, the AN calculated by the equation above (average value) should be limited to ~42 dB(A) at
the edge of the right-of-way.
8.1h
8.1.11.3. RI and AN Calculations
Fig. 8.1h shows the RI and AN noise profiles for the Madeira River Power Plant HVDC lines.
As can be seen in the above figure, an ROW width of 20 m is enough to meet the AN criteria, however it should
be 55 m for RI criteria (the conductor diameter is relatively large).

CHAPTER
It should be remembered that an ROW width of 70 m is required to meet insulation coordination criteria
(see section 8.1.9).
8.1.12. Electric and Magnetic Fields

This section explains the calculations used for electric and magnetic fields, along with an overview of the
possible associated health effects.
8.1.12.1. Ground-Level Electric Fields and Ion Currents

Induction effects under AC transmission lines are defined mainly in terms of the magnitude and frequency of
the alternating electric fields at ground level. In the case of DC transmission lines however, the magnitudes
of both the electric field and the corona-generated ion currents at ground level are required to characterize
any induction effects.
Corona-generated ion space charge fills the entire space between the conductors and the ground plane. In
the cases of both unipolar and bipolar DC transmission lines, only positive or negative unipolar space charge
exists at ground level. The combined presence of a DC electric field and an ion space charge is generally known
as a space charge field [9].
• Calculation Method
An empirical calculation method called the ‘degree of corona saturation method’ has been proposed
for calculating ground-level electric fields and ion currents. The method is based on long-term test line
measurements, along with certain assumptions, and requires a number of empirical constants. It should be
noted that this method sometimes fails to reproduce measurements from existing lines. [15]
BPA (Bonneville Power Administration) developed a software tool based assuming that the pole is an
equivalent conductor. [16], [17]
The Finite Element Method [18] can also be used but it requires an iterative approach, some theoretical
assumptions and has difficulties in handling climatic conditions, thus becoming too complex for common
users. This means that there are no simple reliable tools to evaluate the phenomena.
In the absence of availability of experimental data for the proposed line design or access to numerical
calculation methods, an initial evaluation can be made using the degree of corona saturation method [15].
This method is mainly based on the concept that at a certain voltage the maximum amount of charge is
being released into space by corona on the conductor, called ‘saturated corona’. Above this value, there is no
increase in the charge released.
Empirical equations for the electric fields, ion currents and ion densities under saturated corona conditions in
the ground have been developed [15]: for the worst place under the line (below pole conductor) and for points
at distances varying from 1 to 4 times the conductor height from the conductor abscissa.
The saturated field and ion current values are functions of pole spacing, conductor height, distance
perpendicular to the line and line voltage.
The degree of saturation (saturation factors) is evaluated by:
s = 1 – e–k(G-G0) [Eqn 8.1.12a]
For a given electrical parameter (electric field, ion current, and ion density), Q the actual value is determined by:
Q = Qe + s(Qs – Qe) [Eqn 8.1.12b]
where Qe and Qs are the electrostatic and the saturated values of the parameter.
Values for k and G0 were determined by using measurements on test lines to get the values (L) with 50% and
95% probability, positive and negative polarity and for several weather conditions: summer foul with high
humidity/fog, summer fair, spring fair, fall fair, rain and snow.
In the absence of corona on the conductors, no space charges are created and the electric field under a DC
line may be calculated using the principles of electrostatics. The space-charge-free electric field Ee(x) at any
point P on the ground plane is obtained as:

CHAPTER
  [Eqn 8.1.12c]
V  2 H 2 H 
Ee( x ) = ⋅ − 
2H  S
2
 S
2
2 
ln   x −  + H
2
 x +  + H 
2 2
2
 2H   
req  +1
 S 
Where:
V = voltage applied to the bipolar line, kV
H = conductor height, m
S = pole spacing, m
req = equivalent radius of the conductor bundle, m
x = lateral distance of P from the center of the line, m
The presence of corona-generated space charge maintains the conductor surface electric field at the corona
onset value, but enhances the electric field at points away from the conductors, with maximum enhancement
occurring at ground level [9].
8.1.12.2. Magnetic Fields

The magnetic field can be calculated using the equations below.
I
H j = ∑ i i Φi j [Eqn 8.1.12d]
2π ri j

B = 4 ⋅ π ⋅ H ⋅ 10–7 [Eqn 8.1.12e]
Where:
➝
B is the magnetic field at point (x,y)
Ii are the pole currents (A)
rij is the distance from pole to the point (x,y) (m)
➝
F ij is the unit vector in the direction of the product of the vector current Ij and the vector segment rij
Note that vector B is in the direction of rij and that the current is a complex number (with modulus and phase).
The worst case is when there is current in only one pole.
8.1.13. Health Effects

In the context of HVDC lines and environment, the following aspects have to be examined:
➙ Electric field and ions
➙ Magnetic field
There have been many research programs on the effects of electromagnetic fields (EMF). They are listed
by internationally recognized entities like the ICNIRP (International Council on Large Electric Systems), a
non-governmental organization formally recognized by the World Health Organization (WHO). The ICNIRP
has published reference values for AC electric and magnetic fields which should be applied to HVAC lines.
Reference values are also given for HVDC line magnetic fields. [20], [21]
In order to understand the effects of EMF on health, it is important to know how these fields interact
with matter. “Exposure to Static and Low Frequency Electromagnetic Fields, Biological Effects and Health
Consequences (0 - 100 kHz)”, issued by ICNIRP, presents this subject. [23] Oak Ridge Laboratory has also
carried out many detailed analyses related to static EMF (aimed at HVDC line design), and has made
recommendations on this subject. [22]
8.1.13.1. Ions
Corona effects produce air ions of positive and negative polarity. Some of them are neutralized in the
opposite pole and some migrate to the soil. A HVDC line is a source of ions, but there are others that
generate much higher levels of ions.
CHAPTER
The interaction of air-borne ions with the body is through the skin and respiratory tract. The conductive tissues
at the surface of the body shield tissues below the body’s surface and organs. Most inhaled ions deposit in
the nose and bronchi, with none reaching the deep alveoli of the lung.
Research in animals has had a focus on behavior, metabolism of neuro-hormones, serotonin (one of the
chemicals that the brain and nerves release to affect adjacent cells), and the respiratory tract. Also, the
effects on reproduction, longevity and on farming (dairy cattle) have been studied. In the investigations the
ion concentration varied from 104 to 106 ions/cm3 (inside the ROW the maximum is 105).
Human studies examined mood and performance, serotonin, respiratory function and surveys on people living
near the lines as related to illness days, depression, drowsiness, and respiratory congestion.
The animal and the human studies provided no evidence of any harmful effects.
8.1.13.2. Electric Fields

The electric field intensity near HVDC transmission lines is due to two effects: the charge that appears in
the conductor when voltage is applied (nominal field) and corona effect, where a space charge appears
around the conductors thus enhancing the nearby field (electric field with space charge). In HVDC lines
the nominal field may reach values of 15 kV/m at ground, whereas the electric field with space charge may
reach 40 kV/m or more depending on weather.
Static electric fields can be perceived (movement of body hairs for instance) but cannot penetrate the
organism (skin is a conductive matter and provides a shield for inside the body).
A DC electric field can give rise to shocks to a person who touches large objects near a transmission line.
Three extreme cases are presented in [6]: a normally grounded person with ion current of I = 4 µA passing
through them, a highly insulated person touching a well-grounded object and a normally-grounded person
touching a large vehicle. The conclusion was that there were either no sensations or only minor irritations
similar to a ‘carpet type’ shock.
Research in animals focused on brain and behavior, respiratory function and reproduction/development.
Considered voltages were from 12 to 340 kV/m and short and long exposures were analyzed. Human studies
focused on heart rate, blood pressure and task performance during field exposure.
The conclusion was that there is no basis to conclude that electric fields pose health risks.
8.1.13.3. Magnetic Fields

The mechanisms of interaction are:
➙ Magnetic induction (electrodynamics interaction with moving electrolyte and Faraday currents)
➙ Magneto mechanical (magneto orientation and magneto mechanical translation)
➙ Electronic interaction [21]
The exposure limits indicated are:
Occupational exposure
Whole working day 200 mT
Ceiling 2T
Limbs 5T
General public
Continuous exposure 40 mT
The human body is always subject to the static magnetic field of the earth (which varies from 30 to 70 µT).
HVDC lines create a magnetic field that is below 70 µT, and therefore should not be a concern.
Research in animals has focused on genetic effects, cell growth, reproduction and development, directional
orientation and behavior (circadian rhythms and pineal gland). No major influence was found except as related
to navigation/orientation of certain bacteria, homing pigeons, honeybees and elasmobranches fish [22].
Human studies focused on inquiries with people working in industries with strong magnetic fields (aluminum
plants for instance).

CHAPTER
Magnetic field 500 kV, H = 12 m;

Inom = 2.7 kA
45
40 emergency 40% above nominal

35
steady state max-nominal
30
25
µT
20
15
10
5
0
0 10 20 30 40 50
Distance from tower center (m)
Fig. 8.1i– Magnetic field at ground level
Studies on humans and animals do not indicate that exposures to DC magnetic fields up to 2 mT (20 Gauss)
8.1i
would result in an adverse outcome. Avian or animal migration can be influenced however.
It should be noted that the magnetic field produced by HVDC lines adds to the earth’s magnetic field and
interferes with compasses. In this case, other navigation equipment may be used, like GPS.
Fig. 8.1i shows the magnetic field profile at ground level for a ±500 kV line.
The magnetic field is lower than 50 µTeslas (0.5 Gauss). The earth’s magnetic field is also approximately 0.5
Gauss, and this value is acceptable according to ICNIRP.
8.1.14. Public Perceptions

While research confirms no established proof that the environmental effects of EMF on biological
systems are harmful, there is public concern that arises when the field is perceived (body hair movement
for instance). Another aspect that stimulates complaints is the visual impact. Therefore a new HVDC
project has to start with public education and clarifications. Efforts have to be made to minimize such
misperceptions.
If at the edge of the ROW the field is below 10 kV/m, then there is a small probability of perception by the
public (lower than 10%). [19] Once adopted, the impact of this criterion on the choice of the ROW width is
small or nil. However, the electric field inside the ROW must be taken into account, where the maximum public
perception values occur under the poles at midspan.
The effects of statistical behavior and the time of the year on the electric field and ion current calculations
must also be considered. To understand this point we will look at the calculated values of the electric field at
midspan under the conductor for the Itaipu HVDC system (±600 kV): [15]
➙ During fall/spring the 50% value of the electric field is 28 kV/m (E(50) = 28 kV/m)
➙ During foul weather in summer (very high spread) the 95% value of the electric field is 46 kV/m (E(95)
= 46 kV/m)
Based on the example above, one may choose to design for:
• The worst position inside the line
➙ Worst conditions (worst weather and time of year- summer high humidity: 95% probability) where E
< 45 kV/m and J < 150 nA/m2
➙ Average conditions (fall 50% values) where E < 25 kV/m and J < 50 nA/m2
In the former case the probability of perception is 60% and in the latter 25%, but both criteria are almost

CHAPTER
ElectricÊfieldÊandÊlons
80 E+Ê95Ê%;ÊkV/m;ÊsummerÊfoul
70 J+Ê95%;ÊnA/m2;ÊsummerÊfoul
60 E+Ê50%;ÊkV/m;Êfall
J+Ê50%;ÊnA/m2;Êfall
kV/mÊorÊnA/m2
50
40
30
20
10
0
0 10 20 30 40 50
DistanceÊtoÊtowerÊcenterÊ(m) Electric field
40
30
Field with charge (kV/m)

Fig. 8.1j– Electric field and ion current profiles for the Rio Madeira power plant HVDC line
20
10
equivalent, leading to similar conductor height at midspan.
8.1j 0
• The edge of the ROW -10

➙ Worst conditions (worst weather and time of year: 95% probability)
-20 where E < 10 kV/m and J < 5 nA/m2
-30
These criteria are similar to the basic design proposed by EPRI. [24]
-40
From Fig. 8.1j it can be seen that the mentioned criteria are met with
-60a ROW-40
width of
-2080 m. [15]
0 20 40 60
There are different methods of calculation and no full agreement about what to use. AlsoDistance
theretoare
Center (m)
relatively
few consistent sets of field measurements to support a decision. The best set of measurements was obtained
with the Pacific Intertie project, including about two years of continuous measurements that allow a good
statistical evaluation of the phenomena to be made.
Electric field Ion current

40 80
30 60
Field with charge (kV/m)
20 40
ion current (nA/m^2)
10 20
0 0
-10 -20
-20 -40
-30 -60
-40 -80
-60 -40 -20 0 20 40 60
-100
Distance to Center (m) -60 -40 -20 0 20 40 60
Distance to Center (m)
Fig. 8.1k– Pacific Intertie values obtained with Anypole software
Another calculation method isIonthe one in the BPA software named Anypole developed from the measurements.
current
Using
80 this software with the default parameters leads to the electric field and ionic current values with a 90%
probability
60 of not being exceeded (according to BPA engineers).
The40results obtained with Anypole for the Pacific Intertie project are shown in the Fig. 8.1k below.
ion current (nA/m^2)
20
As the behavior of this line is considered fair, the results can be used for comparison with other lines (mainly
new0lines).
-20
-40
-60
CHAPTER -80
-100
-60 -40 0 60
LineÊcostÊasÊfunctionÊofÊvoltage
450,000
400,000
350,000 2;Ê300
CostÊ(US$/km) 3;Ê500
300,000
250,000 4;Ê600
200,000 5;Ê800
150,000
100,000
0 2,500 5,000 7,500 10,000 12,500 15,000
TotalÊcrossÊsectionÊ(MCM)
Fig. 8.1l – Adjusted line costs (Note: 2; 300: means 2 conductors and ±300 kV)
8.1.15. Economics
8.1k
The estimated transmission line costs, as well as the economic analysis considering staging, losses,
operation and maintenance costs and the financial parameters are presented in this section.
8.1.15.1. Transmission Line Costs (Cline)

Based on an approach taken by CIGRÉ, the following bipolar line cost equation was conceived [4]:
Cline = a + b V + S (c N + d) $US /km [Eqn 8.1.15a]
Where:
a, b, c, d are parameters obtained by curve fitting with a set of line cost estimates
V is the pole to ground voltage (kV)
N is the number of conductors per pole
S = N S1 total aluminum conductor cross section (MCM): S1 is one aluminum conductor cross section (not
including steel area).
Note: S (MCM) = (1/0.5067) ⋅ S (mm2 aluminum)
The values obtained were:

a = 69,950
b = 115.37
c = 1.177
d = 10.25
Where: a, b, c and d are empirical values obtained by the regression.
The bipolar line costs for 2, 3, 4 and 5 conductors per pole, for voltages of ±300, ±500, ±600 and ±800 kV,
are shown in Fig. 8.1l.
Cost of steel 1.7 $US /kg FOB
Cost of aluminum cable 3.5 $US /kg FOB
Cost of concrete 0.35 $US /m3 FOB
Exchange rate at the time:
$US = R 2.20 (Brazilian Real)
$US = €0.695
8.1.15.2. Joule Losses

In the context of transmission lines, the losses are due to Joule and corona effects. Joule losses (Lj) are

CHAPTER
calculated by:
2
1  P
Lj = r  MW / km [Eqn 8.1.15b]
2 V
Where:
P is the rated bipole power MW
V is the voltage to ground kV
r is the bundle resistance ohms/km (r = ro L / S)
ro is the conductor resistivity 58 ohms MCM/ km (or 58/0.5067 mm2/km)
L is the line length in km
S is the aluminum cross section in MCM
The economic basis for determining the cost of losses is that a thermal power plant is built at the load center
to supply the losses.
The cost of Joule losses (CLj) in one year will be:
Clj = (Cp + 8760 Ce lf) Lj = CI ⋅ Lj [Eqn 8.1.15c]
Where:
Cp is the yearly cost of the power plant
Ce is the fuel cost
lf is the loss factor
8.1.15.3. Operating and Maintenance Costs

Yearly operating and maintenance costs can be considered as a percentage of the total line cost: generally
about 2% per year.
8.1.15.4. Interest Payments during Construction

It is estimated here that two to four years are required for the construction of the transmission line,
depending on the length of the line and size of the construction crew. If the interest rate is estimated as
10% per year and the costs are allocated mid-year in equal payments, then at the end of the construction
period, the budgetary cost will be adjusted by a factor of: 1.10 (2 years), 1.16 (3 years) and 1.22 (4 years).
8.1.15.5. Most Economical Conductor

The yearly bipolar line cost is expressed by:
Cline = (k + 0.02) ⋅ 1.1 (A1 + B1 . S) = A + B . S [Eqn 8.1.15d]
Where:
S is the aluminum cross-section per pole,
k is the factor to convert Present Worth into yearly cost: k = j/[1 – (1 + j) – n], j being the interest rate and n
the period of amortization.
A, B, A1 and B1 are coefficients relating the cost of the line to the cross-section S (based on [Eqn 8.1.15a]).
0.02 is the factor to consider operation and maintenance costs and 1.1 is the factor related to the interest
during construction, assuming the construction period is 2 years.
Considering that CLj = C/S is the yearly cost of the Joule losses and (for the moment) neglecting the corona
losses, then the total yearly cost (Cty) is:
Cty = Cline + CLj = A + B . S + C/S [Eqn 8.1.15e]
Where:
C is a coefficient relating the cost of the losses to the cross-section S (based on [Eqn 8.1.15c])

CHAPTER
The minimum value of the function - which is the ‘most economical conductor cross-section’ - occurs for
d(Cty)/dS = 0, or:
C
Sec = [Eqn 8.1.15f]
B
Note: The Sec value does not depend on the line length, because it is a common multiplier in the Cty equation. By using Sec, the
transmission line investment and yearly cost as well as the yearly cost of losses are obtained.
8.1.15.6. Corona Losses

Corona losses in fair and foul weather can be estimated using the equations in section 8.1.10.3. By assuming
a fair/foul weather duration proportion, the total yearly corona loss is obtained as well as its cost using the
approach described for Joule losses (using Cp = 0 and lf = 1).
Starting from the Sec value above, the yearly cost of corona losses can be added to those of Joule losses and
the line itself to get the total cost. An iterative process is used to find the new economic conductor cross-
section, now considering corona losses.
Here is an example using the values from section 8.1.15.1.
MW 3,000 3,000 3,000 3,000 3,000

kV +800 +800 +800 +800 +800
conductor/pole 4 4 4 4 4
MCM 1,680* 1,800** 1,900 2,000 2,200
total $US/yr 54,789 54,700 54,730 54,839 55,251
line $US/yr 36,442 37,438 38,268 39,097 40,756
Joule $US/yr 13,970 13,039 12,352 11,735 10,668
Corona loss $US/yr 4,377 4,224 4,110 4,007 3,826
* optimal solution: calculated disregarding corona losses
** optimal solution: considering corona losses
Table 8.1e– Costs associated with overhead transmission lines

It is important to note that the converter station costs increase with voltage, whereas the line losses
decrease.
Finally it should be noted that the total system cost includes the cost of the converter station (function of
power and voltage) and line and Joule/corona losses (function of power and conductor aluminum cross
section). Introducing the optimal cross section (for a certain power), the total cost equation is a function of
the voltage, from which the minimum value (the optimum voltage) can be obtained.
8.1.16. Other Designs

It should be emphasized that most of the methodologies described for HVDC line electrical design are based
on empirical equations developed for the ‘horizontal bipolar line’. For other types of line like monopolar,
double bipolar or vertical bipolar, other procedures are necessary (the methodology here can present
approximate results for extrapolation).
An electrode line is a monopole with a conductor configuration of 50% to 100% of the pole cross section.
The electrode line voltage is low, consisting only of the line resistive voltage drop. Although a verification of
all aspects mentioned in this section has to be made, some of them lead to very low values (corona effect
for instance). This also applies to any metallic return assigned to the system.

CHAPTER
8.2. HVDC SUBMARINE AND UNDERGROUND CABLES
8.2.1. Introduction
8.2.1.1. Historical Note

The modern use of HVDC cables started with the 100 km long 100 kV cable between mainland Sweden and
the island of Gotland. This cable with mass-impregnated (MI) paper insulation was produced by Liljeholmen
Kabelfabrikk, Sweden in 1954 and was later upgraded to 150kV. The cable system was in service until 1983
when it was replaced by cables with much higher rating. In the early 1960s Cable de Lyon developed the
MI cable type up to 250 kV and Nexans had developed it to 500kV by the 1990s.
At present the MI insulation system is used for bulk power transmission at the 525kV level up to 840MW/
cable in submarine HVDC applications.
Also, Self Contained Fluid Filled (SCFF) cables have been developed up to the 500kV HVDC level. This is
the only cable type at present that may be used for both AC and DC transmission systems. However, the
fluid pressurizing system restricts the use of this cable type in length to approximately 100km and the
maximum depth to 1000m.
The first commercial underground DC cables with extruded insulation were commissioned in 1999 for
70kV in connection with a project on the island of Gotland by ABB. In 2016 it was announced that the
extruded insulation system had been developed to the 500kV level for underground cables although not
yet commercially available.
For submarine applications, the voltage level for extruded cables has reached 320 – 400kV. They are mainly
used in medium transmission level submarine connections between countries, such as the NEMO Link
(400kV 2x500MW) between Belgium and the UK and from offshore wind power parks to in-feed points far
inland such as in some German North Sea connections.
8.2.1.2. The Main Difference between AC and DC Cables

All insulation systems for HV cables are cylindrically
symmetric to minimise tangential variation in
electric field strength. The tangential breakdown
voltage is very much lower than the radial one.
Overall the insulation system consists of three layers
from the conductor outward:
➙ Conductor screen r
Êcscr
➙ Insulation r dr
➙ Insulation screen
In AC insulation the voltage distribution in the r
insulation is based on the geometric capacitances. Êis
Since the circumference increases with increasing
radius the capacitance increases per unit length in
the radial direction as well. Since the same current
flows through all the unit capacitors the voltage
across each unit decreases proportionally to the Fig. 8.2a– Definition of the geometry of the
insulation for calculations
increase of radius, the highest voltage appearing
across the innermost unit, with least capacitance,
and decreasing outwards. The capacitance of the insulating materials show almost no dependence on
change of temperature in the operational temperature range. The capacitance and hence the voltage
distribution per unit length, that is the electrical stresses, can be regarded as invariant, the highest stresses
occurring at the conductor screen and the lowest at the insulation 8.2e
screen, regardless of the load on the cable.

CHAPTER
In DC insulation the voltage distribution is DCÊgradients
governed by electrical resistivity per unit
length. The electrical resistivity for most
insulating materials is strongly dependent
30
on temperature. Almost all conventional
insulating materials (lapped paper and
28
e x t r u d e d p o l y m e r ) h a v e t e m p e ra t u re
GradientÊ(kV/mm)
dependence that is exponential, the resistivity
26
decreasing by approximately a decade for each
20 K increase in temperature. At no load, the
temperature is constant through the insulation, 24
consequently the voltage distribution is similar
to the AC case. However, when there is load 22
on the cable, there are losses, because of the
resistance in the conductor that generate heat.
0 4 8 12 16 20 24
This heat is conducted radially away from the
conductor. During stationary load conditions InsulationÊthicknessÊ(mm)
there will be a permanent temperature fall
across the insulation, the insulation being Fig. 8.2bExample – The electrical stresses at
no-load, (red) and at full-load, (blue)
warmest near the conductor and coolest at the
insulation shield. Therefore, the coolest outer
layers have the highest resistance and the highest electrical 8.2a stresses, see fig 8.2b.
8.2.2. Basic Thermo-Electrical Calculations

Based on the observed electrical and thermal behaviour of the insulation the maximum electrical stresses
are calculated as follows, see fig. 8.2a.
Basic Equations
Loss in the conductor:
ρ
W

A
( )
= c ⋅ I 2 ⋅ 1 + α c ⋅ θ c − 20 0 C  [Eqn 8.2.4a]
Where:
W = losses in the conductor, W/m
rc = specific resistivity of conductor at 20 °C, W.m
A = conductor cross-section, m2
I = current, A
ac = conductor temperature co-efficient of electrical resistivity, 1/K
qc = conductor temperature, °C
Temperature fall across the insulation:
Dqis = W ⋅ T1 [Eqn 8.2.4b]
Where: Dqis = temperature fall across the insulation, K

T1 = insulation thermal resistance, K ⋅ m/W
Thermal resistance of the insulation:
ρthis r
T1 = ⋅ ln is [Eqn 8.2.4c]
2⋅π rcsc r
Where: rthis = specific thermal resistivity of insulation, K ⋅ m/W
ris = insulation radius, m
rcscr = radius conductor screen, m

CHAPTER
The specific electrical resistivity of the insulation depends on the temperature and electrical stress and
according to experience can, at any radius r, be expressed by the following equation:
r(r) = r0 . e–a.(q(r)–q0) . e–g.(E(r)–E0) [Eqn 8.2.4d]
Where: a = insulation temperature co-efficient of electrical resistivity, 1/K
g = insulation stress co-efficient of electrical stress, m/MV
The values with subscript 0 are corresponding values at a starting point of the measurements, for example
at 20 oC and 0 kV/mm. Usually E0 = 0 kV/mm is used because of the very low voltage at the beginning of
the measurement.
The parts independent of radius may be gathered as a constant:
α ⋅ ∆θ is
β=
r [Eqn 8.2.4e]
ln is
rcsc r
The electrical stress as function of the radius r:
dU ( r ) r ( β −1) ⋅ exp ( −γ ⋅ E ( r ))
E (r ) = − = U0 ⋅ [Eqn 8.2.4f]
dr ris
r ( β −1) ⋅ exp ( −γ ⋅ E ( r )) ⋅ dr
∫rcsc r
The above equation is an implicit equation that can only be solved by numerical methods.
Since the value of the exponent, b, is proportional to Dqis the electrical stresses in the insulation are highly
dependent on temperature drop across the insulation, i.e. to the losses in the conductor. So if the current in
the conductor is increased above the rated current, the electrical stresses will increase very rapidly above
the design stresses in the insulation. See fig 8.2b.
For Mass Impregnated paper insulation and SCFF-low viscosity fluid impregnated paper insulation the
measured DC electrical properties are:
➙ resistivity co-efficient, a = 0.1 – 0.13 1/K
➙ electrical field co-efficient, g = 0.028 -0.035 mm/kV
The resistivity co-efficient implies that an increase of the temperature of approximately 20 K decreases
the electrical resistivity of the insulation by a factor of 10.
The electrical field co-efficient helps to distribute the voltage more evenly in the insulation. It partially
compensates for the adverse voltage distribution arising from the changes in resistivity generated by the
losses in the conductor, which would otherwise grow very rapidly.
In comparison to Eqn 8.2.4f, in AC cables the electrical stress is calculated from:
U0
E (r ) =
ris [Eqn 8.2.4g]
r ⋅ ln
rcsc r
8.2.2.1. Some Special Cases:

• The cable is unloaded/cold ⇒ b = 0, as Dqis = 0 ⇒ rb–1 = 1/r.
U0 exp ( −γ ⋅ E ( r ))
E (r ) = ⋅ [Eqn 8.2.4h]
r ris
r ⋅ exp ( −γ ⋅ E ( r )) ⋅ dr
−1
∫ rcsc r
The influence of the smoothing effect of the electrical field itself will cause the electrical stresses in the
insulation to be similar to the AC case, except that they are slightly lower at the conductor screen and
slightly higher at the insulation screen.

CHAPTER
• Assume that the specific insulation resistivity is independent of the electric stress coefficient, g = 0
r β −1 U0 ⋅ β
E (r ) = ⋅U0 = ⋅ r β −1
ris
r β −1
⋅ dr risβ − rcsc
β [Eqn 8.2.4i]
∫ rcsc r
r
Sometimes this equation is used to make a first assessment of the maximum electrical stress, as this
equation can be solved analytically. However, even at moderate electrical stress, the maximum electrical
stress at ris calculated at full load is too high.
However, this equation can be used as a comparative evaluation tool between designs, if a numerical
calculation tool is unavailable.
8.2.3. Basic Cable design
8.2.3.1. General
High Voltage DC Power Cables have some common features:
• T hey are built of concentric layers with a central conductor at high voltage. The conductor has to have
good conductivity to reduce the ohmic losses, good mechanical properties and flexibility. The materials
used are either Copper, Cu or Aluminum, Al.
• In all HV cables the insulation system consist of three layers:
– The Conductor Screen, CS
– The Insulation
– The Insulation Screen, IS
– The screens act as buffers and to separate the metallic materials from direct contact with the insulation.
The layers are neither a good conductors nor a good insulators, so in cable language it is often called a ”semi-
conductor” with specific resistivity, 0.01 – 1000 Ω.m. It also increases the withstand voltage level of the
insulation system against transients, such as lightning voltages.
• For DC long length high capacity energy transmission the mass-impregnated-paper-insulated system,
MI, has shown both longevity and reliability. It is now available commercially at 525kV with transmission
capacity of 840MW/cable. This insulation system tolerates polarity changes.
• The extruded XLPE system for DC was introduced 20 years ago at the 70kV level. It is a single layer
of insulation and as such it is a “weakest link” system. So to produce a reliable insulation the whole
length has to be tested. The test level has to be set to detect faults in the insulation, but not to induce
new faults.
8.2.3.2. The main cable features

Conductor
With DC cables the current flows through the entire conductor cross-section. There is no “skin effect” or
“proximity effect” to take into consideration as in AC cables. However, if cables are bundled or laid in close
proximity the mutual heating has to be taken into consideration.
Two metals are used for conductors materials Cu and Al. The conductivity of Cu is 63.9% higher than for Al.
So for the two metals to reach the same maximum temperature the diameter of an Al conductor must be
28% larger than for a Cu conductor. In the following table the properties of the two materials are compared.
Cu AI
1.7241 2.8264
Specific resistivity, Ω.m*108
Temperature coefficient of 3.93 4.03
resistivity, K-1*103
Specific gravity, kg/m3
8900 2700

CHAPTER
From the table above it can be calculated that the Cu conductor will, for the same rating, weigh 2.01 times
as much as the Al conductor. So for land cables, where transport aspects such as weight and diameter of
the transport drum are the main issues, the low conductor weight of Al is a big bonus.
In submarine cables, a lead sheath, Pb, is used as a radial water barrier and for longitudinal tensile strength
for laying in deeper water or pulling into long tubes steel wires are used. Both the lead and steel volumes
will increase by by at least 28% if Al is used as a conductor, and that will out-weigh the lighter conductor
weight in conventional cables.
A “Solidal” type of Al conductor that is a solid bolt of Al, is being introduced instead of the multi-stranded
normal conductor to save cost. Like any other cost cutting project it requires a “cultural” change in all parts
of the handling process to be an effective change. Otherwise, the failure rate may increase.
The Insulation systems
• There are two main insulation types in use, the lapped insulation and the extruded insulation:
– The lapped insulation consists of many layers of thin tape impregnated with insulating mass. The tape is high
purity Kraft paper tape. Since there are a large number of tape layers it is easy to adapt the tape properties
to the local electrical and mechanical requirements and to grade the insulation. Another benefit of the large
number of tapes is that the probability of having a number of defects radially at the same location is very low.
This leeds to the experience that if a 50 m piece of such cable is tested to satisfaction, a very long length, >500
km will exhibit the same properties with high confidence.
– The extruded insulation consists of a single layer of highly purified extrudable material, usually cross-linked PE
(XLPE). The cross-linking process takes place at high temperature, high pressure and with a cross-linking agent
such as peroxide. The cross-linked insulation is much more thermally stable than the basis material. However,
there are always some by-products generated in the process. The by-products have to be removed. Since
there is a single layer of insulation this leeds to a “weak-link” type of probability that weak points may exist in
the insulation that may lead to breakdown in operation. The probability increases with the extruded length/
volume. If a 50 m length of such cable is tested to satisfaction, it shows that the design criteria for the cable
are OK. However, to ensure the life expectation of the manufactured length a so-called “screening” test has
to be applied to each unit length of the cable to eliminate these weak points. The test shall have a level and
duration that exposes the weak points, but shall not initiate new ones.
Barrier against radial water penetration
The insulation system sustains high electrical stresses. If water enters the insulation system it leads to
electrical breakdown as the withstand properties are reduced. Only a metallic barrier has the ability to
hinder water ingress over a long time. Water will difuse over time through any non-metallic barrier.
Because of the load changes the barrier has to withstand mechanical cycling, so it has to have good fatigue
resistance. It is laid in a rather corrosive environment as well. It is usually transported so it has to have
good vibration resistivity. In addition it has to be extrudable onto the insulation at a temperature that does
not harm the insulation system. Low alloyed lead, such as Alloy E, has been the material that fulfils all
the requirement for submarine cables. To enhance the fatigue properties usually a PE-sheath is extruded
directly on the Pb-sheath.
On land, to reduce weight and cost, a layer of polyethylene (PE) foil with Al has replaced the lead sheath
for XLPE cables.
Armouring
In DC cables the armouring is to take care of the longitudinal mechanical tension during laying. Usually a
weldable quality of low carbon galvanised steel wires are chosen such as St 35. For high voltages and deep
crossings usually more than one layer is chosen to make the cable torsion free, a so called cross-wire-armour
design. The corrosion protection in the sea is usually asphaltic compound reinforced by PP-yarn. See 8.2.d.
On land, the conductor may lack the mechanical strength to withstand height differences or pull-in through long
pipes. For these reasons, the cable is produced with a layer of steel tapes known as longitudinal reinforcement.
Additionally, a tough extruded outer PE-sheath serves as corrosion protection. See 8.2.c.
DC Magnetic Field
Usually the magnetic field strength above a cable buried to 1m depth is approximately the same order of
magnitude as the Earth´s own magnetic field strength.

CHAPTER
•Ê ConductorÊ
•Ê ConductorÊscreenÊ
•Ê InsulationÊ
•Ê InsulationÊscreenÊ
•Ê LeadÊalloyÊsheathÊ
•Ê PolyethyleneÊsheathÊ
•Ê TransversalÊreinforcementÊ
•Ê OuterÊsheathÊ
Fig. 8.2c– Example cross-section of MI type Underground cable
8.2b
Ê
Ê
Ê
Ê
Ê• ConductorÊ
Ê
Ê Ê• ConductorÊscreenÊ
Ê Ê• InsulationÊ
Ê
Ê• InsulationÊscreenÊ
Ê
Ê Ê• LeadÊalloyÊsheathÊ
Ê
Ê• PolyethyleneÊsheathÊ
Ê
Ê Ê• TransversalÊreinforcementÊ
Ê Ê• ArmorÊ
Ê
Ê Ê• OuterÊservingÊ
Ê
Ê
Ê
Ê
Fig. 8.2d– Example cross-section of MI type Submarine cable

CHAPTER
8.2c
Ê Ê• ConductorÊ
Ê Ê• ConductorÊscreenÊ
Ê
Ê• InsulationÊ
Ê
Ê Ê• InsulationÊscreenÊ
Ê
Ê• LeadÊalloyÊsheathÊ
Ê
Ê Ê• PolyethyleneÊsheathÊ
Ê Ê• TransversalÊreinforcementÊ
Ê
Ê Ê• ReturnÊconductorÊ
Ê Ê• IRCÊinsulationÊ
Ê
Ê• ReinforcementÊ
Ê
Ê Ê• ArmorÊ
Ê Ê• OuterÊservingÊ
Fig. 8.2e– The Integrated Return Conductor. Cable with IRC, forming the complete circuit with no outside magnetic field
There is only one solution to reduce/eliminate DC magnetic field from the cable, which is to create an
opposite magnetic field by sending the return current through an insulated conductor laid near the main
cable or through a concentric metallic layer on the cable itself. Laying cables close together reduces the
8.2d
magnetic field, but also reduces the transmission capacity of the cable because of mutual heating. In
mono-polar circuits the return conductor can be designed as a concentric layer into the HV cable, a design
known as the Integrated Return Conductor (IRC) design. This design eliminates the magnetic field almost
completely, see fig. 8.2e. There is no evidence that time invariant, DC magnetic fields represent any type
of environmental or safety hazard.
8.2.4. Types of Underground Cables

The main requirement for underground/land cables is that they have low weight and volume so that they
may be transported on drums in the longest length possible. This is necessary to reduce the number of
jointing bays.
This requirement indicates the use of as few metallic materials as possible. However, the cable must still
be designed for longevity, so certain materials should be excluded, because of corrosion or thermal cycling
as described in 8.2.3.
8.2.4.1. The Underground Cable Types

At present three type of cables are available for underground installation:
• Mass Impregnated Paper Insulated cables
• SCFF pressure assisted cables
• XLPE insulated extruded cables

CHAPTER
Type MI SCFF XLPE
Conductor Cu/Al Cu/Al Cu/Al
Conductor water proofing Yes, mass Yes, fluid pressure Yes, powder*
Insulation system Paper & mass Paper & fluid Cross linked PE
Radial water barrier Lead alloy Lead alloy Metallized film
Extruded sheath Polymer Polymer -
Transversal reinforcement Steel tapes Steel Tapes** -
Outer protection Polymer sheath Polymer sheath Polymer sheath
Voltage, kV 525 500 400/500***
Capacity per cable, MW 840 600 500
Converter system* LCC, VSC LCC, VSC VSC
* powder ⇒ swelling powder that expands in contact with fresh water, thereby blocking further water ingress
** In case the cable is used both in AC and DC circuits non-magnetic tapes are used
*** Not yet in commercial use
LCC ⇒ Line Commutated Converter
VSC ⇒ Voltage Source Converter
The XLPE cable with its light weight and small dimensions may be the most suitable cable system for
underground installation. However, the system has been in operation only for a relatively short time so the
life expectancy of the system is not yet fully known.
8.2.4.2. The Submarine Cable Types

At present three type of cables are available for submarine installation:
➙ Mass Impregnated Paper Insulated cables
➙ SCFF pressure assisted cables
➙ XLPE insulated extruded cables
Type MI, Fig.8.2h SCFF, Fig.8.2f XLPE, Fig.8.2h

Conductor Cu Cu Cu
Conductor water proofing Yes, Mass Yes, fluid pressure Yes, compound*
Insulation system Paper&Mass Paper&Fluid Cross linked PE
Radial water barrier Lead alloy Lead alloy Lead alloy
Semi-conducting Semi-conducting
Extruded sheath Polymer
Polymer Polymer
Transversal reinforcement Steel tapes Steel Tapes -
Armour Steel wires Steel wires Steel wires
Outer protection Asphalt&PP-yarn Asphalt&PP-yarn Asphalt&PP-yarn
Voltage limit, kV 525 500 400
Capacity/ cable, MW 840 600 500
Converter system* LCC, VSC LCC, VSC VSC
* Compound ⇒ a highly viscous mass impregnating the conductor to prevent longitudinal water ingress
The number of factory joints on XLPE is far higher than for other types.
The MI and XLPE systems are non-pressurized systems, so there are no length limitations other than the
maximum acceptable transmission losses.
The SCFF system is a pressurized system and the length is limited by the fluid pressurization system to
about 100 km.

CHAPTER
Fig. 8.2f– SCFF submarine cable Fig. 8.2g– MI submarine cable
8.2.5. Transport, Installation and Protection of Cables
8.2.5.1. Transport of Underground Cables

During transport of cable to site the road transport restrictions decide the length between jointing bays. The
limitations are on the maximum drum diameter, width and weight. These limitations should be established
early in any project with a long land cable route.
If the land cable section is short (a few km from the landing site for a submarine cable) floating and winching
the underground cable may be a transport alternative.
8.2.5.2. Laying of Underground Cables

The terrain, the route details and the given constraints are
the main factors in choosing the laying method. Especially,
height differences on the route or long pull-in sections
through pipes, require longitudinal reinforcement and even
torsion balance for safe laying conditions.
The preferred method is to establish a roller guide way to
roll out the cable beside the trench, then transfer it into
the trench either by hand or by well proven mechanical
methods.
Depending on the thermal parameters of the local soil
and its consistency a thermally specified sand or cement-
bound sand, so called weak-mix, is laid on the bottom of
the trench, and than embedded to a certain level, according Fig. 8.2h– XLPE submarine cable
to the requirements. Often concrete slabs, brick layer and
warning tape are laid over the cable to hinder damage
caused by hand tools and to warn operators of mechanical excavators, such as backhoes, against digging
into the cable.
In deep loose soil with good thermal properties, direct trenching with plows may be used, instead of
excavating open trenches.

CHAPTER
Fig. 8.2i– Power cable transport, laying and repair vessel, (Nexans Skagerrak)
With open trenching the safety of workers and the requirements according to local regulations must be
engineered into the laying method.
At least the last 5 cm of the trench should be covered by native soil and reinstated to its former appearance.
8.2.5.3. Transport of Submarine Cables

Transport of submarine cables should be done on dedicated transport and laying vessels as any transfer
of long lengths of heavy cables poses additional risks. Although it is not always possible, it is preferable to
use a vessel capable of transporting the required length for the whole route.
There are two main modes of transport of long length:
• Spooling, where a rotating turn table or drum is used
• Coiling, where the cable is laid on a non-moving platform. The cable may be laid in a circle, skating rink
or figure of eight formation on the platform. In this case the cable is twisted 360o for each complete
rotation. This method is not recommended for high voltage cables as the twisting imposes torsional
mechanical stresses into the insulation. At joints where the torsional withstand moment changes,
damage may occur. When a cable is designed for coiling the armour does not take axial load after
coiling, therefore the core has to sustain all tension. So coiling may be used for medium voltage cables
laid in shallow water.
8.2.5.4. Laying Submarine Cables

Laying of HV submarine cables is a complex and
risky operation that has to be engineered and integr
ated in minute details and requires experienced
crew both onboard and on land. The operation
requires a number of steps:
• Perform a route survey, to establish the bottom
profile, depth, currents, sediment properties,
thermal properties and so on. A second survey
may even be necessary if the information in the
first one is not sufficient.
Fig. 8.2j– The complete navigation system during
• Perform a special survey of the landing sites and submarine cable laying
engineering the landing method and the pull-in
sequence. Predetermine the stationing position
for the vessel for the second end pull-in.

CHAPTER
• Acquire the meteorological data, including the special local
weather patterns, especially for the landing sites.
• Acquire as much knowledge as possible about the local fishing
activity in the route, and about marine activity across the route.
• Find if there are any crossings with pipelines or other cables.
• Inquire about any other restrictions or regulations regarding
environment, corrosion, magnetic fields.
• Based on the above, generate a route with geodesic coordinates
and establish the required cable length.
• From the coordinates and depth profile data, the master computer
program for the laying route for each cable is established.
• The navigational accuracy is established and the requirements
for both underwater and above-water navigational systems are
specified.
• If during the route survey debris is discovered, one or more grapnel
runs have to be carried out to clear the route.
• Usually immediately before laying, a trial run is made to check out
the route and that the navigational systems perform as specified.
• The landing of the cable requires much better conditions than
laying, so it starts in still water periods with low wind, usually
on the landing with the worst conditions. The cable is floated to
the landing point and winched up to the termination site. Then
the floats are removed and the cable lowered to the bottom. The
marine positioning system is deployed if necessary. Fig. 8.2k– HVDC cable termination
• The main laying proceeds until a predetermined stationing
position is reached on the other side.
• From the stationing position the necessary cable length is determined, it is measured onboard, the
cable is cut and sealed here. The length is floated out on water and pulled with boats and winch to the
termination site.
• The remaining length is floated ashore and stored as a spare/repair length.
If a single length is insufficient to cover the cable route, a number of lengths will be jointed together onboard,
by retrieving the end of the previously laid length and jointing it to the new length onboard. Thereafter the
laying continues. The largest DC cables today can be transported in 7,000 - 10,000 ton loads lengths by the
largest cable transport and laying vessels, such as the Nexans Skagerrak. Such vessels have jointing facilities
onboard with the same requirements as in the factory.
8.2.5.5. Termination of Cables

At the termination site (Figure 8.2k) the cable is terminated either against an overhead line or another cable.
If terminated to an overhead line, very steep-front impulse voltages will enter the cable. For DC termination
the creepage length of the insulator and the inner and outside voltage distribution both during normal
DC operation and during transient voltages are the important design criteria. The condenser cone based
terminations have proved to be a reliable solution for paper insulated HVDC cables. However, stress cone
based termination solutions are also feasible.
Normally such terminations, regardless of insulation type, are pressure assisted. This assures that the
pressure inside the termination is always above the outside pressure, so accumulation of air and water
because of lower than ambient pressure inside the termination, is eliminated.
The transient behaviour of long cables is covered by CIGRE Technical Brochure 268.
The installation of cable terminations is a speciality and should be carried out by specially trained and
certified jointers.

CHAPTER
8.2.5.6. Protection of Submarine Cables
CIGRE WG B1-21 (Third party damages on cables; TB 398) has instituted a Burial Protection Index that
is a structured way of evaluating the necessity and type of protection that should be applied. The fishing
activity and the maritime shipping activities are evaluated based on the activities and trends found from
dialogs, statistics and observations over time.
The necessary protection can be roughly put into three categories:
No fishing activity and no maritime activity, the route No protection is needed, the cable may be laid on the
section is deeper than 400 m bottom
Fishing activity with trawls, small commercial vessels The burial depth should be 0.5 – 1 m even in very soft
sediments
Heavy shipping traffic of tankers, large bulk carriers and It is almost impossible/too costly to bury the cables
large cruise ships deep enough, but the probability of anchor damage is
very low. So the burial depth is kept at 1 m so as not
to derate the cable, and in the unlikely event of anchor
damage, it is easier to recover the cable for repair.
There are many instances where the cable has to be buried deeper because of regulations or local
conditions.
Burial systems can be roughly
characterized:
Completely water jet based
systems, usually employed in
loose soil with < 200 kPa shear
strength. The system may be
employed on 95% of almost all
routes. Maximum commercial
burial depth is approximately
2.5 m. Such system as the
Nexans’ Capjet has an
excellent track record from
burying >5,000 km of pipes
and cables without damage.
See 8.2l. Fig. 8.2l– Water jetting based burial system
• Plows towed by high power
surface vessels, may bury
cables deeper and in harder soils than the jetting system. However, because of the high tow force (20
– 60 tons) it can damage the cable and has done so on many occasions.
• Various protection systems applied for short lengths. Rock installation, grout bags and gabion
mattresses are some of the systems used, for example in crossing pipelines, rocky areas, fishing gear
on the bottom and so on. Such systems require the mobilisation of special vessels and are very costly,
therefore should be employed only on short stretches.
8.2.6. Testing
A cable system must be tested in numerous ways before it can be offered commercially or installed. There
are a number of IEC international standards that form the basis of most testing regimes. However, for
HVDC cables and submarine cables there are the CIGRE recommendations that are the de-facto standards.
The recommendations are published in the bi-monthly publication, Electra and in the CIGRE Technical
Brochures, TB. The present recommendations for testing HVDC MI and SCFF cables may be found in Electra
No 189, with an addendum in Electra No 218. For extruded HVDC insulation the present recommendations
may be found in TB 496. The mechanical testing of submarine cables is covered in TB 623.
Some of the routine and sample tests for HVDC cables are based on IEC 60055-1 for MI cable and
IEC 60141-1 for SCFF cable. For extruded cables the tests depends on the voltage level: medium voltage

CHAPTER
(<72 kV) is covered in IEC 60502-2; HV (72 – 170 kV) in IEC 60840 and for EHV voltages in IEC 62067, as
far as relevant.
The following tests are made on DC cables:
• Pre-qualification tests/long time tests: tests to show the long time properties of the cable system used
on extruded systems because of the thermo-mechanical properties of the insulation. The duration is
usually a year with higher than nominal voltage, 100 m or more cable and all accessories to be used in
the cable system. The tests are found in the quoted recommendations above.
• Type tests are used to confirm that the specific cable to be delivered conforms to the test specification.
The mechanical and electrical tests are specifically designed for the delivery. They can be waived if
similar cable design has been tested to at least the same or higher mechanical and electrical levels.
• Routine tests/Screening tests are made to ensure that the manufactured factory lengths are according
to specifications. The screening tests apply specifically to extruded length to eliminate the statistical
weak spots.
• Factory Acceptance Test, FAT, is usually an electrical test to ensure the integrity of the complete
shipping length before transport and laying
• After installation test/Site Acceptance Test, SAT, is to assure the integrity of the complete cable system
after transport laying and installation.
Depending on the installation method or other special conditions and additional tests may be specified.
8.2.7. Temperature Monitoring Systems

As shown in Figure 8.2a. there is a direct connection between the ampacity of the cable and the maximum
electrical stresses at the insulation shield. So if the steady state load exceeds the maximal design load the
steady state electrical stresses will exceed the maximum design stresses.
However, if buried, the thermal time constant of the soil is very long, approximately two weeks for a burial
depth of 1 m. The cable’s own time constant is 2 – 3 hours. So if the cable has been running at lower than
full load, there would be a short time overload capacity.
To be able to utilise this overload capability without exceeding the design stresses the actual temperatures
of the cable, especially at the thermal bottlenecks, have to be continuously measured.
The measurement of changes in cable temperature can be made continuously by measuring the scattering
properties in the glass in the Fibre Optic (FO) cable. The monitoring system to measure such changes is
called the Distributed Temperature Sensing, DTS, system. At present such systems can monitor up to
approximately 70 km cable length with a resolution of 2 – 5 m and a temperature accuracy of 0.5 – 1.5 K,
depending on sampling time.
To make predictions on maximum allowable load, or on overload and time for which it can be sustained,
a transient computer program, utilising the temperatures from the DTS, the voltage and current
measurements from the cable and the cable properties, called Dynamic Rating System (DRS) has been
produced. This program may be interfaced directly with the SCADA control system for the transmission.
However, experience dictates that at present a human operator placed between the DTS and the SCADA
is a better solution.
8.2.8. The Future

In submarine applications the MI type of insulation has an excellent track record. It is still the most reliable
insulation system, especially for long, high-transmission-capacity links. It is a “solid” cable, therefore there is
no leakage of any kind even if it is cut. The electrical aging properties of the insulation has proved to be very
good: after 40 years in operation no recognisable aging has been recorded. As the reliability of the Submarine
cable is a crucial point, the MI type of submarine cable will be a preferred alternative for years to come.
In submarine applications the extruded cable has few advantages. In dynamic applications, for example
from offshore platforms, it will form a necessary part of the transmission system. However, as effective
screening tests are developed for the extruded cable it will see more deployment also in high capacity
transmission systems.

CHAPTER
At present no breakthrough for the application of new technologies is seen in the near future. If room
temperature Super-Conductors are discovered and can be reliably manufactured in long lengths a paradigm
shift will result.
If graphene-based conductors become a reality, a higher conductivity, low weight and mechanically superior
conductor material will be available. However, at present it seems that it will be many years, or even
decades, before the price and availability of this technology become suitable for use as a cable material.
Until such time, only a gradual development in transmission capacity, transport capacity and laying to
larger depths will happen.
8.3. EARTH/SEA ELECTRODES

As explained in section 3.1 the simplest form of a HVDC interconnection is a monopole and, as only one
conductor needs to be raised above earth potential (the HVDC conductor) the other current path (often
referred to as the return current path) can use the earth, sea or a combination of both as the conduction
medium between two electrodes. This can often provide a lower cost solution (in terms of both capital
cost and losses) to the DC circuit, when compared to the use of metallic conductors such as transmission
lines or cables (see sections 8.1 and 8.2). Although there are examples of such single-conductor-with-
earth-or-sea-return HVDC schemes successfully operating around the world, as will be described later,
environmental concerns and long-term impacts on buried infrastructure have reduced the attractiveness
of this as a solution.
CIGRE SC B4 and IEC TC115 have produced guides for the design of electrodes associated with HVDC
schemes [30], [31].
The general term for a current path based on the use of electrodes is ‘earth return’. The return path
resistance will be very low as the DC current will spread rapidly from the point of injection at the electrode
over a very large cross-sectional area within the earth. If the earth were perfectly homogenous then
Fig 8.3a– A lagoon electrode (rated for 800 Adc, located on the mainland of South Korea as part of the
Haenam-Cheju 300 MW HVDC interconnection)

CHAPTER
the current injected at the electrode would form
hemispherical lines of equipotential, centered on
the electrode. The equipotential lines would be
very close together in the vicinity of the electrode,
rapidly becoming further apart with increasing
distance from the electrode, as shown on Fig. 8.3c.
For the large transmission distances frequently
associated with HVDC transmission, particularly
when the sea is used as the return path, the
resistance of the path is effectively independent
of the distance between the electrodes. Where Fig. 8.3b– A typical land electrode with 2 concentric
rings (one each for bipole 1 and bipole 2), each
the return path comprises different media, in rated for 2000 Adc, located in Canada as part of the
particular when the earth is used as the return Nelson River 2 x 2000 MW HVDC project.
path, the actual current distribution will be complex.
Consequently, the actual prediction of the current
distribution will also be complex.
The earth is composed of many different strata of material ranging from soil to solid rock, whose resistance
can range from 103 Ω.m to 108 Ω.m. Further complications result from seasonal fluctuations in resistivity
in the surface layers associated with varying moisture content.
Sea return paths have a more predictable distribution, as a result of sea water resistivity typically only
varying from 0.2 Ω.m to 0.25 Ω.m. The actual current distribution will be affected by the contour of the
seabed and the land around the electrode which will, in general, remain consistent for the life of the project.
However, a less predictable influence on the current path is the effect of fresh water added to the sea from,
for example, rivers, particularly during periods of flooding as river water, unlike sea water, has a resistivity
in the range of 5 Ω.m to 300 Ω.m. It is therefore important to carefully consider the site of the electrode.
The final selection of an electrode site may be many kilometers from the associated converter station,
the location being dictated by geographic and environmental reasons. It is therefore necessary to have
an electrode line or electrode cable to connect the converter station and the electrode site. The electrode
line or cable need only be rated for a relatively low voltage as one end (the electrode end) is connected to
earth: the steady-state voltage rise along the line or cable therefore being dictated by the DC resistance of
the line or cables and the operating DC current of the converter. Typical lengths for electrode lines range
from 8 km for the New Zealand HVDC hybrid Inter Island Link, Benmore converter station, to 84.5 km for
the Intermountain Power Project (IPP) HVDC scheme, Adelanto converter station [32] although on the
Lower Churchill Project linking Labrador to Newfoundland in Canada, the electrode line on the Labrador
side is more than 400 km long. However, even with the use of electrode lines or cables, the earth return
section of the transmission system is likely to represent a significant portion of what would otherwise be
a metallic conductor. Therefore, as the losses within the earth return section are negligible the losses of
the overall scheme will be lower than a metallic return equivalent.
For a bipole HVDC scheme normal operation will be with the two pole currents balanced and hence the
current through the electrodes will be negligible (dictated by the measurement error in the transducer(s)
associated with measuring the earth return current). When one of the poles fails, the other pole can
continue to operate without interruption using the earth return as a current path. The direction of current
flow through the earth return will depend upon which pole remains in operation and hence the electrodes
have to be designed for bidirectional current flow. For monopole HVDC schemes on the other hand, the
DC current flow will always be in the same direction through the electrodes and hence one electrode
will permanently be an anode and the other a cathode. This has a major impact on the design of the
electrodes as only the anode electrode suffers from corrosion through electrolytic action. Hence, the
anode electrode(s) must be constructed from sufficient materials to allow for the expected operating life
of the HVDC scheme in terms of ampere-hours of operation, or alternatively facilities must be included
for periodic replacement of the electrodes: the cathode electrode on the other hand will not suffer from
electrolytic corrosion and can be designed as a non-maintained installation. As an example, the Italy –
Corsica – Sardinia monopole multi-terminal HVDC scheme was designed with the anode at the Sardinian

CHAPTER
terminal, which was constructed as a lagoon electrode similar to that shown in Fig. 8.3a. The cathode
electrode on the Italian mainland was constructed from two uninsulated steel-wire armored cables laid
on the seabed, approximately 2 km from the shore at a depth of 30 m to 35 m [33].
8.3.1. Environmental Effects

The most controversial topic surrounding the use of earth return for HVDC schemes is that of the impact
of the DC current within the earth or sea on the environment. Environmental concerns by third parties
may influence the decision as to whether to employ earth return or metallic return far more than any
technical reason associated with the design of the HVDC scheme. It is therefore important that, for a
given installation proposal, it can be shown that there will be no health hazards to human or animal life, no
detrimental effect on land or sea based equipment and that there will be no change to the local ecology.
In some circumstances, minor impacts may be acceptable when the earth return is to be used only for
short periods of time. This could occur due to an outage of one pole in a bipole scheme, with the earth
return providing a continuous current path until the DC circuit can be reconfigured to give metallic return
operation using the HV conductor of the other pole (see section 3.1).
Anode
electrode
Buried
pipe _
Cathode
electrode
AreaÊwhere
corrosionÊwillÊoccur LinesÊof
equipotential
Fig 8.3c– Interaction between earth current and structures
BACK TO 476 | 8.3c

CHAPTER
8.3.1.1. Visual Impact
The biggest influence on the visual appearance of an electrode site is that of the quantity of material
required to meet the ampere-hour rating of the electrode, as described above. The location of the electrode
site will also be important, and this must be dictated by the geographical and geological data associated
with the proposed earth return path, but the final site selection may be heavily influenced by visual impact.
The need to meet concerns over visual impacts has led to the design of many types of electrode
installations.
8.3.1.2. Effects on Human and Animal Life

As discussed in section 8.3 above, the injection of DC current can be considered to produce lines of
equipotential within the earth and/or sea. The spacing of these lines is a measure of the change in voltage
over a given distance: the closer the line spacing, the more rapidly the earth voltage is changing. With
HVDC applications, the electrode design requirements of a low resistance current path typically results
in there being a concentration of voltage gradient around the electrode(s), that is, around the electrode
there is a rapid change of the voltage with distance but, at some distance away, the change of voltage with
distance is very small.
If we know the resistance of a body which intersects these lines of equipotential, then applying Ohm’s
law the current flow through the body can be found. However, although the acceptable level of current
flow through various life forms is well documented, the prediction of current flow in a specific case is
very difficult as the skin resistance is difficult to predict, as is the contact resistance with the medium
with which the animal is in contact. Clearly, the longer the animal is, the greater the voltage differential
between the ends of the animal. Many countries have national standards with regards to the maximum
voltage gradient per meter. Many relate to electrical substation installations but are equally applicable to
land-based electrode sites. Step voltage and touch voltage are significant terms which form an important
part of the assessment of the potential impact on human and wildlife at a particular electrode site [31].
Typical limits for surface voltage gradients on pasture land are in the range of 13 V/m to 16.6 V/m, whilst
in the sea where fish of up to two meters have access, voltage gradient limits of 1.5 V/m to 2.5 V/m have
been used. This assumes some form of protective barrier around the sea electrodes to keep smaller aquatic
animals some distance from the electrodes. Where fish of 300 mm can approach the electrodes, a voltage
limit of 6.7 V/m has been applied.
The effect of electrical current and its associated electric field has an impact on the behavior of fish. Fish are
attracted to an anode but repelled by a cathode. Hence, for a monopole scheme where the anode and cathode
are fixed, this implies that a protective barrier is needed at the anode but not at the cathode.
Another important aspect of the design of sea electrodes is that of protecting the electrodes from
mechanical damage from the sea, such as from tidal damage or storms: this protection often also acts as
the barrier to aquatic life, see Fig. 8.3.a.
The actual size of a sea electrode is very much dependent on site conditions. As an example consider an
electrode injecting 800 Adc into the sea. Assuming the sea has a resisivity of 0.2 Ω.m and the land adjacent
to the electrode has a resisivity of 500 Ω.m, then the voltage gradient will be less than 6.7 V/m at 3 m from
the electrode and 1.5 V/m at 6 m.
For land-based electrodes, it is common practice to bury the electrodes deeply enough to ensure that the
surface step voltage is within safe limits for both humans and animals. The access of animals, in particular
large, four legged, animals such as cattle and horses, to the area around the electrode needs to be carefully
considered as the step voltage between legs will be higher than that experienced by a human. The actual
area of land occupied by a shallowly buried electrode can be as large as 2 km2. The land electrode shown
in Fig. 8.3b is a ring electrode, with a diameter of approximately 300 m. This does not have a fence around
it: if a fence was necessary then the clearance from the ring to the fence would be about 30 m.
Several studies have been carried out to investigate the use of deep earth electrodes to significantly reduce
the impact of earth current at or near earth’s surface level on the natural and man-made environments.
One example where a real comparison between alternative depths of electrode was carried out is in the
Baltic Sea [37]. There are other studies which have investigated the magneto telluric properties of earth

CHAPTER
3-PhaseÊearthed 3-Phase 3-PhaseÊearthed

starÊpowerÊ transmission starÊpower
transformerÊcircuit circuit transformerÊcircuit
SomeÊofÊtheÊearth
returnÊcurrent
flowsÊasÊaÊ
zero-sequence
componentÊinÊthe
transmission
circuit
DCÊearth
current
Fig. 8.3d– Zero sequence component of current flowing through the AC transmission circuit
strata down to depths of 100 km and more in the Southern African region, including detailed mapping of
8.3d
electrical resistivity (The Southern African Magneto telluric Experiment (SAMTEX)). However, in order to
get the benefits from the significantly lower resistivity of the deeper layers of the earth, in practical terms
this may mean reaching depths of 1 km or more (with a borehole of about 1 m diameter), which is obviously
more complex and costly than normal surface electrode installations.
Most HVDC electrode installations in service today are composed of either a single electrode which is
vertically mounted in the earth, or of multiple individual electrodes arranged in a ring or array of rods. These
arrays or rings are considered as horizontal electrodes, and it is also possible to install a complete loop of
copper (or another suitable anode/cathode conductive material) in the horizontal plane as an electrode.
Horizontal arrangements of individual electrodes are common, especially (a) in onshore installations where
a sufficiently large area of land is available with homogeneous ground characteristics close to the surface, or
(b) installations offshore where alternative materials and construction methods are used, such as titanium
mesh mats laid on the seabed secured in place by concrete or rocks.
8.3.2. Corrosion Effects

Corrosion within earth return systems falls into two categories: corrosion of the electrodes themselves
and corrosion of third party structures.

CHAPTER
MagneticÊfluxÊ
withÊDCÊ
component Magnetization
curve
DCÊcomponent
t i
MagneticÊfluxÊ
withoutÊDCÊ
component
MagnetizationÊ
currentÊwithoutÊ
DCÊcomponent
t
MagnetizationÊ
currentÊwithÊDCÊ
component
Fig. 8.3e– DC neutral current causing asymmetric magnetizing current
8.3.2.1. Electrode Corrosion

8.3e8.3 above, an electrode operating as an anode will be subject to a loss of material
As discussed in section
through electrolytic corrosion: the amount of material lost being proportional to the ampere-hour operation
of the electrode and the material from which the electrode is constructed. This loss of material can be
estimated from Faraday’s Law, which states that:
M t2
m= ⋅ ∫ i ⋅ dt [Eqn 8.3.2a]
n ⋅ F t1
Where:
m = mass of element removed (kg) t1, t2 = time (s)
n = number of electrons transferred in half reaction [34] M = average atomic mass of material [34]
F = Faraday’s constant = 96.485 C mol-1 i = current (A)
Considering iron as an electrode material: it has an average atomic mass of 55.847 and typically, a valence
of 2, hence operation at 1,000 Adc for 1 year (31,536,000s) would result in the loss of 9,125 kg. However,
in the case where the electrode is associated with a bipole HVDC converter, normal operation will be
in balanced operation, so the current flowing into the electrode will only be that resulting from the spill
current, that is, the current that cannot be measured because of the measuring inaccuracy of the DC
current transducers. If this spill current is approximately 5 Adc, then the loss of material per year falls to
45.625 kg.

CHAPTER
5
4.5
4 25ÊmÊdepth
MagneticÊcompassÊdeviationÊ(Deg)
3.5 30ÊmÊdepth
3 35ÊmÊdepth
2.5 40ÊmÊdepth
2 45ÊmÊdepth
1.5 50ÊmÊdepth
1
0.5
-100 -50 0 50 100

DistanceÊfromÊcenterÊofÊsingleÊHVDCÊcableÊ(m)
Fig. 8.3f– Magnetic deviation due to a single HVDC cable conducting 800 Adc and lying on a south/north axis for
depths of 25 m to 50 m.
In both cases, the loss of material through erosion is too high. It is therefore normal to surround the iron
electrodes with coke. If good surface contact between the iron and the coke is achieved, then most of
8.3f the current flow from the iron to the coke will be by electron flow, and not ionic, hence the iron will not
erode. The amount of coke eroded away each year will be much less than that of pure iron and coke is
a much cheaper material.
8.3.2.2. Third Party Structure Corrosion

An example of the type of corrosion that can be induced in a third party structure is indicated in Fig. 8.3c.
The figure shows that buried structures, which follow the lines of field between the converter electrodes,
may present a lower impedance path than the surrounding earth and hence current will flow through the
structure. Where current enters the structure no corrosion occurs, however, where the current exits the
structure corrosion will occur following the same process outlined above for the anode electrode [35].
Various methods of protecting buried pipelines and cables exist. The simplest method of protecting
buried structures is to locate the electrode site far enough away, in order to ensure that negligible current
flows through the structure. However, the necessary land distance is not always easy to predict due to
non-homogenous earth conditions.
The addition of some form of cathodic protection can be used to protect buried structures. Two basic
methods of cathodic protection exist: a passive method using sacrificial electrodes has been used in
some installations, but a more common approach is to use an active cathodic protection system.
Active cathodic protection induces a current to flow between a metal structure and a remote anode
electrode. Sufficient current must flow in order to ensure that the potential of the structure with respect
to the surrounding earth, or sea, is sufficient to stop any corrosion of the structure. Typically the voltage
considered safe is -0.85 V. If care is not taken in the application of cathodic protection other problems
may be induced: with a structure voltage in excess of -1.5 V damage may occur to any coating on the
outside of the structure [35].
It must be remembered that current flow through the earth is not only the result of HVDC electrode
schemes. Variations in the earth’s magnetic field can induce some current flow (known as telluric
current). One of the most notable causes of changing geomagnetic fields is sunspot activity, resulting
in solar flares hitting the earth’s atmosphere. The resulting induced currents are referred to as
‘Geomagnetically Induced Currents’.

CHAPTER
8.3.3. Direct Current in Power Transformer Neutrals
As described previously in section 8.3.2, the earth return current will always try to find the least resistive
path between any two points. This least resistive path may be an intentional earth or sea path, an
unintentional current flow in a buried structure, or a parallel AC system connection. As shown in Fig. 8.3d
the DC earth current can flow into the converter transformer neutral as a zero sequence component
and along the AC power conductors, re-appearing in the earth, via a transformer neutral, at a different
geographical location.
This neutral DC current will bias the transformer core, resulting in additional asymmetric magnetization
current being drawn from the AC system, as shown in Fig. 8.3e. This results in additional transformer core
heating, acoustic noise and AC harmonic currents.
8.3.4. Gas Production

There are two dangerous gases generated by the electro-chemical reaction of injecting DC current into the
surrounding media, with the majority of gases being generated when the surrounding media is seawater.
These two gases are chlorine and hydrogen.
The quantity of these gases generated in seawater by the injection of DC current is given by Eqn 8.3.2a,
simplified as:
m M ⋅i
= [Eqn 8.3.4a]
t n⋅F
Where:
i = DC current flowing through the electrode into the surrounding sea
t = time (hours)
m
t = mass of element released (grams per hour)
m
= 6.61 g/h for chlorine (n = 1, M = 35.45) [34]
t
m
t = 0.19 g/h for hydrogen (n = 2, M = 2) [34]
m
t = 1.49 g/h for Oxygen (n = 4, M = 32) [34]
As the water surrounding the electrodes is changed on a regular basis due to tidal action, the resulting
build-up of chlorine in the water is very low, and most of the chlorine will remain in solution. The very small
build-up of chlorine around the electrode in fact has the positive benefit of inhibiting marine organisms
from living on the electrodes, whilst being low enough not to endanger larger animals who are only exposed
to the electrode environment for a short amount of time.
The potential danger from hydrogen is that of a build-up of gas and the consequent risk of explosion. However,
as the quantity of hydrogen liberated is very small, this does not pose a problem in an open sea electrode.
Oxygen is also released through the electro-chemical reaction of injecting current into seawater. Whilst
this in itself does not produce a danger, it can combine with carbon in the surrounding water or from the
electrodes themselves, resulting in the production of carbon dioxide (CO2) which may give rise to some
environmental concerns.
8.3.5. Compass Deviations

Where an earth return system is employed, only one conductor is needed for each pole, that being the high
voltage DC conductor. Whilst the earth return current will be spread across many kilometers, the high voltage
connection concentrates the current in the cable. This cable current and earth return current will each
generate a magnetic field, but in the vicinity of the cable the two magnetic fields will not completely cancel
and there will be a residual, measurable, magnetic field associated with the HV cable.

CHAPTER
Where subsea cable routes are used, the magnetic field associated with a single HV conductor can cause a
change in the magnetic field as measured by a compass, deflecting the compass away from true magnetic
north. Fig. 8.3f shows how magnetic compass deviation varies with the distance from the cable, both
horizontally and vertically.
The deviations shown in Fig. 8.3f can be calculated from the equation [Eqn 8.3.5a] [36]:
Id 1
Tan D = ⋅ ⋅ cos λ
500 ⋅ S ⋅ He   X  2  [Eqn 8.3.5a]
1 +   
 S 
Where:
TanD = the variation from true magnetic north (degrees)
X = horizontal distance from the center of a single HVDC cable (meters)
S = depth of the cable below the compass (meters)
He = horizontal component of the earth’s magnetic field (Gauss)
He = 0.19 Gauss in the English Channel
l = angle between the cable and the earth’s polar axis (degrees)
Id = the DC current through the cable (Adc)
8.3.6. Thermal Effects

The current flow in the soil also has a thermal impact, and this will vary a great deal according to local
conditions throughout the conducting area. Higher currents will clearly result in greater increases in
temperature, (the equivalent of I2R losses in metallic conductors), as will the presence of moisture (through
the water table) and natural differences in thermal conductivity in soil, rocks, etc.
BIBLIOGRAPHY
[1] “Economic Assessment of HVDC Links”, CIGRÉ Brochure 186, June 2001.
[2] “Design Criteria of Overhead Transmission Lines”, CIGRÉ IEC 60826, 2003-10, 3rd edition.
[3] “Tower Top Geometry”, CIGRÉ Brochure 048, June 1995.
[4] “Impacts of HVDC Lines on the Economics of HVDC Projects”, CIGRÉ Brochure 388.
[5] “Probabilistic Design of Overhead Transmission Lines”, CIGRÉ Brochure 178, February 2001.
[6] “Transmission Line Reference Book HVDC to 600 kV”, EPRI, EPRI Report 1977.
[7] Hill Mc Graw, “Standard Handbook for Electrical Engineers”, 14th Edition.
[8] “Thermal Behavior of Overhead Conductors”, CIGRÉ Brochure 207, August 2002.
[9] P. S. Maruvada, “Corona Performance of High Voltage Transmission Lines”, Research Studies Press
Ltd., Baldock, Hertfordshire, U.K., 2000.
[10] J. B. Whitehead, “High Voltage Corona” in International Critical Tables, McGraw-Hill, 1929.
[11] U. Corbellini, P. Pelacchi, “Corona Losses on HVDC Bipolar Lines”, IEEE Trans., Vol. PWRD-11, No. 3,
July 1996, p. 1475-1480.
[12] “Interferencess Produced by Corona Effect of Electric Systems”, CIGRÉ Brochure 061, December 1996.
[13] V. L. Chartier, R. D. Sterns, A. L. Burns, “Electrical Environment of the Up rated Pacific NW/SW HVDC
Intertie”, IEEE PWRD, Vol. 4, No. 2, April 1989, p. 1305-1317.
[14] “Information on Levels of Environmental Noise Requisite to Protect Public Health and Welfare with
an Adequate Margin of Safety”, U. S. EPA., 550/9-74-004, 1974.
[15] G. B. Johnson, “Degree of Corona Saturation For HVDC Transmission Lines”, IEEE Transactions on
Power Delivery, Vol. PWRD-5, No. 2, April 1990, p. 695-707.

CHAPTER
[16] Maruvada P. Sarma, W. Janischewskyj, “Analysis of Corona Losses on DC Transmission Lines:
I – Unipolar Lines”, IEEE Trans., Vol. PAS-88, No. 5, 1969, p. 718-731.
[17] P. Sarma Maruvada, W. Janischewskyj, “Analysis of Corona Losses on DC Transmission Line. Part II
- Bipolar Lines”, IEEE Vol. PAS-98, No. 10, 1969.
[18] W. Janischewskyj, G. Gela, “Finite Element Solution for Electric Fields of Coronating DC Transmission
Lines”, IEEE Trans., Vol. PAS-98, No. 3, May/June 1979, p. 1000-1016.
[19] J. P. Blondin, D. H. Nguyen, J. Sbeghen, D. Goulet, C. Cardinal, P. S. Maruvada, M. Plante, W. H. Bailey,
“Human Perception of Electric Fields and Ion Currents Associated with High Voltage DC Transmission
Lines”, Bioelectromagnetics 17: 1996, p. 230-241.
[20] “Guidelines for Limiting Exposure to Time-varying Electric, Magnetic, and Electromagnetic Fields
(up to 300 GHz)”, ICNIRP.
[21] “Guidelines on Limits of Exposure to Static Magnetic Field”, ICNIRP.
[22] W. H. Bailey, D. E. Weil, J. R. Stewart, “HVDC Power Transmission”, Environmental Issues Review Oak
Ridge Laboratory report, 1997.
[23] “Exposure to Static and Low Frequency Electromagnetic Fields, Biological Effects and Health
Consequences (0 -100 kHz)”, ICNIRP, 2003.
[24] “DC Conductor Development and Transmission Line. Vol. I”, EPRI 2257, 1982.
[25] “Transient voltages affecting long cables”, WG B1.05 CIGRÉ Brochure 268.
[26] “Testing DC extruded cable systems for power transmission up to 250 kV”, WG B1.21 CIGRÉ
Brochure 219
[27] “Recommendations for testing of long submarine cables with extruded insulation for system
voltage 30 (36) to 150 (170) kV”, ELECTRA - CIGRÉ Document Ref ELT_189_1.
[28] “Recommendations for tests of power transmission DC cables for a rated voltage up to 800 kV”,
ELECTRA - CIGRÉ Document Ref. ELT_189_2 (rev. of edition 1980).
[29] “Recommendations for mechanical Tests on Submarine Cables”, ELECTRA - CIGRÉ Document Ref
ELT_171_3.
[30] “General Guidelines for HVDC electrode design”, CIGRÉ Technical Brochure 675.
[31] IEC/TS 62344 “Design of earth electrode stations for high-voltage direct current (HVDC) links - General
guidelines”.
[32] D. J. Christofersen (convenor), “Compendium of HVDC Schemes Throughout the World”, CIGRÉ SC14.
[33] S. D. Thorp, D. Macgregor, “Design of the Sea Electrode System, Sardinia – Italian Mainland 200 kV
Scheme”, IEE conference, 1966.
[34] www.environmentalchemistry.com/yogi/periodic <http://www.environmentalchemistry.com/yogi/
periodic>
[35] “HVDC Ground Electrode Design”, EPRI EL-2020, Project 1467-1, Final report, August 1981.
[36] M. W. Kennedy, “Earth Return for HVDC Systems”, Electrical Times, 2 May 1968.
[37] “Deep electromagnetic studies in the central Baltic Sea ”, Physics of The Solid Earth, English
Translation, Vol. 30, No. 12, July 1995.

CHAPTER
9| HVDC SCHEME PERFORMANCE
TOC 484 | DC
9 HVDC SCHEME
PERFORMANCE
The measure of any HVDC scheme is its ‘performance’, that is, its
ability to meet the requirements of the project. ‘Performance’ can be
measured in many ways from: can the power transmission required
be achieved under the appropriate AC system conditions, to what
energy is lost in the AC/DC conversion process and how reliable is
the equipment.
All of these issues need to be analyzed as part of the HVDC design.
Reliability analysis, for example, can affect the scheme design
from the point of built-in duplication and redundancy, as well as
establishing the spares holding.
Energy loss equates to lost revenue from the HVDC link and therefore
is important when considering the overall investment cost of the
project and the payback period.
Dynamic modeling of the HVDC link allows the main circuit and the
control system to be optimized in order to achieve the best overall
performance from the HVDC link. Modeling techniques have been
developed to simulate the behavior of the HVDC converter in terms
of both fundamental frequency and transient response.

9| HVDC
HVDC SCHEME
SCHEMEPERFORMANCE
PERFORMANCE
Chapter contents
9. HVDC SCHEME PERFORMANCE............. 484
9.1. CONVERTER STATION ELECTRICAL LOSSES.................. 487

9.1.1. What are Electrical Losses? ................................................. 487
9.1.2. Costs of Electrical Loss........................................................... 487
9.1.3. HVDC Scheme Electrical Loss.............................................. 488
9.1.4. Loss Components...................................................................... 489
9.2. MODELING OF HVDC CONVERTER SCHEMES.............. 493

9.2.1. Fundamental Frequency RMS (FFRMS) Approach...... 494
9.2.2. Cycle-by-Cycle Approach....................................................... 495
9.2.3. Modeling to Study Sub-synchronous Oscillations...... 498
9.2.4. Comparison of Models............................................................ 500
9.3. DYNAMIC PERFORMANCE STUDIES.................................. 501

9.3.1. Transient Stability..................................................................... 501
9.3.2. Dynamic Interaction................................................................. 508
9.4. TRANSIENT AND DYNAMIC

STABILITY INVESTIGATION.................................................... 512
9.4.1. Study Software and Methodology...................................... 513
9.4.2. Study Components................................................................... 513
9.5. CONTROLS DYNAMIC PERFORMANCE

ASSESSMENT INVESTIGATION............................................. 516
9.5.1. DC Link Dynamic Performance
Studies (DPS in PSCAD).......................................................... 516
9.5.2. DC Link Protection Coordination Studies....................... 517
9.5.3. DC Link Dynamic Performance
Studies (DPS in RTDS).............................................................. 517
9.6. FUNDAMENTAL FREQUENCY TOV STUDY...................... 518

9.6.1. System Representation.......................................................... 519
9.6.2. FFTOV Studies............................................................................. 519
9.7. TYPICAL STUDY RESULTS....................................................... 520

9.7.1. Rectifier Fault.............................................................................. 520
9.7.2. Inverter Fault............................................................................... 520
9.8. RELIABILITY, AVAILABILITY

AND MAINTAINABILITY........................................................... 523
9.8.1. What is RAM? ............................................................................. 523
9.8.2. Significance of RAM ................................................................. 524
9.8.3. Designing for RAM..................................................................... 525
9.8.4. Modeling Technique – Dependency Network............... 526
9.8.5. Calculation Techniques........................................................... 527
9.8.6. Typical RAM Performance
of LCC HVDC Schemes............................................................ 528
BIBLIOGRAPHY ............................................................................................ 529
TOC DC Transmission
486 | Systems:
DC Transmission
Line Commutated
Systems: Line
Converters
9.1. CONVERTER STATION ELECTRICAL LOSSES
In this section the concept of electrical loss, the sources of electrical loss in a HVDC scheme and the
significance of scheme losses for the owner/utility will be elaborated.
9.1.1. What are Electrical Losses?

In all practical electrical installations a small proportion of real power is lost (mostly dissipated as heat)
before it is actually used. This lost real electrical power is termed as electrical loss. The electrical loss
is a result of the inherent electrical resistance of the equipment. Depending on the electrical circuit of
the equipment, and its mode of operation, the real power lost can be represented in proportion with a
corresponding equivalent electrical resistance.
A HVDC scheme consists of a multitude of electrical equipment connected in the power circuit: converter
transformer, valves, filters, etc.; and a substantial auxiliary circuit consisting of electrical machines within the
valve hall ventilation system, valve cooling plant, etc.
Electrical loss in a HVDC scheme is therefore represented by both the real power loss of the DC portion, and
the AC portion of the HVDC installation, formed by the power circuit equipment and by the auxiliary circuits
on the system.
9.1.2. Costs of Electrical Loss

An electrical infrastructure represents two costs to its owner:
➙ The initial capital cost for the equipment design, manufacture, transport, installation and
commissioning, together with the balance of plant formed by the land acquisition, civil works and the
related engineering and project management expenditures
➙ The lifetime operational cost which includes operation manpower cost, cost of maintenance, cost of
electrical loss, cost of outage (scheduled/forced) and cost of decommissioning
The income from a utility-owned electrical infrastructure is related to the price per unit of real electric
energy consumed by the end client.
As discussed in section 9.1.1, electrical loss represents real power that is ‘lost’ between the sending end
and the receiving end of an electrical system. Electrical loss is thus a reflection of the efficiency of the
system. The upshot of improvement in efficiency, that is, reduction in electrical loss, is often a rise in the
capital investment for the equipment. The potential owner of an electrical infrastructure therefore requires
a mechanism to arrive at an optimum compromise between initial capital investment and operating cost
associated with electrical losses. One of the means of evaluating scheme losses prior to investment is
establishing a per unit price for the estimated/calculated losses. A cost can be associated with this loss
based on the average price per hour of electricity, the interest rate applicable on the capital investment or
the opportunity rate and the design life of the equipment. This relationship can be expressed as cost per
kilowatt-hour and can be derived from the following equation:
r yrs
NPV = C1× hrs × (1 + ) [Eqn. 9.1a]
100
Where:
NPV = the ‘Net Present Value’ of the cost of losses (US$/kW)
C1 = the cost of 1 kWh of electricity at the present value (US$)
hrs = the number of hours (in a year) the equipment will be in operation (hours)*
r = the interest rate applicable to the investment (%)
yrs = the life expectancy of the equipment (years)
* There are 8760 hours in a year but an allowance must be made for the outage time applicable to the maintenance of the equipment
in each year.
Such an analysis is important when the cost of operating the equipment will have a financial impact on the
owner. This can be illustrated by considering the tendering for a simple step-down power transformer. Assume
two quotes are received against the tender enquiry, one transformer quote is for US$3,000,000 and the second
CHAPTER
is for a transformer costing US$3,600,000. If only the initial or capital cost was to be considered, then clearly
the transformer costing US$3,000,000 would be selected.
However, let us assume that the bidders for these transformers have quoted guaranteed load losses and
that these losses are 2300 kW from the first transformer quotation and only 2000 kW from the second. If the
following data is assumed:
C1 = 0.10 US$/kWh
hrs = 8720 hours (assuming 40 hours scheduled maintenance outage per year)
r = 6.0%
yrs = 20 years
Then based on the above equation, the NPV of losses can be calculated as 2,797 US$/kW.
This operating cost of the transformers can then be added to the transformer’s capital cost in order to find the
capitalized cost of the transformers. Therefore, the capitalized cost for the two transformers is:
➙ First transformer US$ 3,000,000 + US$ 6,433,100 = US$ 9,433,100
➙ Second transformer US$ 3,600,000 + US$ 5,594,000 = US$ 9,194,000
When considering the capitalized cost of the two transformers, the second transformer proposal presents
much better value. It is therefore important for the purchaser to declare the cost of losses, known as the
loss capitalization figure, in the enquiry for any electrical plant item so that the manufacturer is able to offer
what it believes is the most economic proposal.
The loss capitalization figure can be further adapted to the particular application by developing weighting
factors depending on the expected utilization of the scheme and its cost to the owner, for example, the average
time for which the scheme will be:
a. Energized and in standby at no-load
b. Transmitting on average 1 p.u. power or another power level
This information enables the manufacturer to optimize the proposed solutions.
9.1.3. HVDC Scheme Electrical Loss

HVDC scheme equipment electrical losses are identified in accordance with applicable IEC and IEEE industry
standards [3, 4]. These standards are used by the manufacturers at the bidding phase for calculation of
electrical losses, and at the contract stage the standards are used to test key equipment to provide base
data for verification of the system losses previously identified at the bidding phase.
9.1.3.1. System Operating Conditions
The guaranteed converter station losses are required to represent average operating conditions of the
HVDC scheme and are typically calculated under the following operating conditions at specific power levels:
➙ Nominal converter transformer HV AC bus voltage and frequency
➙ Nominal DC voltage
➙ Nominal equipment tolerances
➙ Nominal measurement errors
➙ Ambient temperature of 20°C
➙ Transformer tap corresponding to the specific operating condition
The reactive power equipment is incorporated so that the exchange with the system is within the specified
operating limits. All equipment losses are identified for consistent operating conditions.
9.1.3.2. Contract Stage Loss Estimation
The contract stage losses are prepared in line with recommendations from the industry standards [3, 4]
mentioned in the preceding text, using both factory test report data and site measurements where these
are appropriate. Site measurements of power circuit losses in a converter station are not appropriate, as
the high efficiencies achieved by the power circuit equipment mean that there is a very small difference
between input and output powers at full rating and the accuracy of measurement equipment is too low
to achieve an accurate value. In addition, the losses are guaranteed for specified AC system and ambient
temperature conditions and such conditions, in practice, are difficult to reproduce. For these reasons,
industry standards [3] permit calculations from factory test results.

CHAPTER
It is possible to measure the loss corresponding to auxiliary power load in service, as it is a measure of the
consumed power only. However, it may be necessary to normalize these auxiliary power measurements to
the load demanded at the guaranteed ambient conditions.
9.1.4. Loss Components

HVDC schemes consist of a number of power circuit and auxiliary circuit components. Globally the losses
are segregated into no-load (equipment is energized but the system is on standby) and load losses; or fixed
(generally no-load losses) and variable losses (total load losses – no-load losses).
The following sub-sections identify some notable aspects related to loss estimation of the specific components
and groups of components in terms of no-load and load losses. The relevant industry standards [3, 4] provide
details regarding calculation and testing methods.
8% 1%
3% ConverterÊtransformer
5%
2% ConverterÊvalves
ValveÊcoolingÊplant
DCÊsmoothingÊreactorÊ
25%
56%
ACÊharmonicÊfilters
HFÊfilter
Auxiliaries
Fig. 9.1a– Typical split of losses for converter station of a point-to-point scheme at 1 p.u. load [1]
0% 5% 1% 5% ConverterÊtransformer
4%
9.1a ConverterÊvalves
ValveÊcoolingÊplant
DCÊsmoothingÊreactorÊ
35%
50% ACÊharmonicÊfilters
HFÊfilter
Auxiliaries
Fig. 9.1b– Typical split of losses for converter station of a back-to-back scheme at 1 p.u. load [1]
Fig. 9.1a and Fig. 9.1b show typical loss distributions for point-to-point and back-to-back schemes.
Additionally, Fig. 9.1c through Fig. 9.1f show the typical variation of electrical losses in the key components
of a point-to-point scheme.
9.1b
9.1.4.1. Converter Valves
As explained in section 6.2, the converter thyristor valve load losses consist of the following main
components:

CHAPTER
➙ Thyristor conduction loss – current dependent loss component
➙ Thyristor spreading loss – current and voltage dependent loss component related to the delay in
establishing full conduction after the thyristor is turned on
➙ Other conduction losses – current dependent loss associated with other circuitry in series with the
thyristors in the valve
➙ DC voltage dependent loss – voltage dependent loss component arising during the non-conducting
period of the valve: e.g. losses due to thyristor off-state and reverse leakage, losses in DC grading
resistors, other resistive circuits and elements connected in parallel with the thyristors (coolant pipes,
structure etc.)
➙ Damping loss – voltage dependent [2], resistor dependent term, capacitor energy term
➙ Turn-off loss – voltage dependent loss component
➙ Reactor hysteresis loss
Converter valve no-load losses (converter scheme is energized and in standby) consist of the following voltage
dependent components:
➙ Losses in the valve resistive components as a result of the current flowing through the circuit as a
result of the applied voltage
➙ Loss in the resistance parallel to the blocked valve
➙ Loss in the capacitive coupled resistances
Typically, when the converter is energized, but blocked (not transmitting power), the converter transformer
tapchanger maintains the valve winding voltage at a minimum value. The losses are mostly voltage dependent,
arising from the voltage grading circuit.
0.3
0.25
(p.u.ÊofÊtotalÊschemeÊlossesÊatÊfullÊload)
0.2
ValveÊlosses
0.15
0.1
0.05
0
0 20 40 60 80 100
DCÊpowerÊ(%)
Fig. 9.1c– Typical variation of the valve losses with load for a point-to-point scheme
Once the converter is de-blocked (transmitting power), the valve winding voltage will be set to an appropriate
level for the power transmission level. At full power, the converter losses are split between current dependent
9.1c losses and the split can vary significantly depending on the particular scheme
and voltage dependent
parameters. The valve losses are influenced slightly by the thyristor junction temperature. The guaranteed
losses are typically calculated for a junction temperature corresponding to an ambient temperature of 20°C
and the corresponding valve coolant inlet temperature. This is an iterative calculation.

CHAPTER
9.1.4.2. Converter Transformer
Converter transformer losses fall into two categories:
➙ Iron losses are those losses related to the magnetic energization of the transformer core and are
therefore not directly load dependent. There is a load dependency if the valve winding voltage is varied
with load and between blocked and deblocked operating conditions by means of tapchanger action.
➙ Copper losses are those losses related to the load current that flows through the transformer winding:
for example, the I 2 ⋅ R losses. Unlike a normal transformer, the converter transformer is also subjected
to a large amount of harmonic currents. The value of the winding resistance of the transformer must
be considered at the specific harmonics and the harmonic component of the losses is included in
order to establish the total copper losses of the transformer.
0.6
0.5
ConverterÊtransformerÊlosses
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100
DCÊpowerÊ(%)
Fig. 9.1d– Typical variation of the converter transformer losses with load for a point-to-point HVDC scheme
Losses in the transformer are dissipated as heat, resulting in increased temperature of the winding, core and
the insulating medium. The resistive losses (copper losses of the windings) are dependent on the winding
9.1d
conductor temperature. Typically for a converter transformer, the winding resistance for loss evaluation is
corrected to a winding temperature of 75°C ( where 75°C is based on a temperature rise of 55°C above an
ambient temperature of 20°C).
9.1.4.3. Other Power Circuit Components

Other main circuit components also have losses which have their calculation methods defined by relevant
industry standards [3]. The main additional loss components are:
➙ AC harmonic filters – based on nominal AC voltage, nominal fundamental frequency and harmonics
generated under nominal converter operating conditions. As filters are normally energized only
when power is being transmitted, these losses only contribute to load losses (Fig. 9.1e)
➙ DC smoothing reactor – as this is in series with the DC circuit, the losses in this device are dependent
on the DC current passing through the DC reactor (see Fig. 9.1f) and therefore only contributes to
the load losses
➙ High Frequency (HF) filter – the HF filter consists of two elements: a shunt capacitor and a series
reactor. As the HF filter is energized along with the converter transformer, the HF filter will contribute
to the no-load losses as a result of the nominal AC voltage at nominal frequency across the shunt
capacitance. There will also be a minor contribution to the no-load loss from the valve and converter
transformer no-load loss current passing through the HF filter series reactor, however, this is typically
negligible. Under load conditions, the main circuit current will flow through the HF filter series
reactance and so contribute to the load loss.
CHAPTER
0.08
0.07
0.06
0.05
ACÊfiltersÊlosses
0.04
0.03
0.02
0.01
0
0 20 40 60 80 100
DCÊpowerÊ(%)
Fig. 9.1e– Typical variation of the AC filters losses with load for a point-to-point HVDC scheme
9.1e
0.06
0.05
0.04
SmoothingÊreactorÊlosses
0.03
0.02
0.01
0
0 20 40 60 80 100
DCÊpowerÊ(%)
Fig. 9.1f– Typical variation of the smoothing reactor losses with load for a point-to-point HVDC scheme
➙ HVDC filter – point-to-point overhead transmission schemes may incorporate DC-side shunt filters.
The losses in these filters are voltage dependent and occur only when the scheme is on-load. Since
9.1f for a transmission scheme is fixed, the DC filter loss does not vary with load.
the DC voltage

CHAPTER
9.1.4.4. Auxiliary Load
A converter station requires many auxiliary services for its operation. As with other loss contributing
equipment, some of these losses are associated with no-load operation and others are associated with
load loss. However, the auxiliary losses generally vary little with the load level. Typical examples of auxiliary
load are identified in Table 9.1a.
No-load components Load related components

• Control equipment • Valve cooling fans
• Protection equipment • Valve cooling pumps (full speed or reduced speed)
• Valve cooling pumps (sometimes only at reduced speed) • Converter transformer cooling fans
• Converter transformer oil pumps • Valve hall ventilation/cooling
• Valve hall ventilation/cooling equipment
• Control room ventilation/cooling
• Other building services (fans, etc.)
Table 9.1a– Typical auxiliary power load of a HVDC converter station
9.2. MODELING OF HVDC CONVERTER SCHEMES

To help discover and solve potential problems with the design of a HVDC scheme prior to ordering
equipment, studies are performed in order to design and specify equipment and also to determine the
operating characteristics of the scheme.
For the electrical and control aspects discussed in this section, the studies require various types of models
and a selection of digital simulation tools.
The HVDC controls and electrical networks are modeled for digital studies, in two domains:
➙ Fundamental frequency RMS algebraic equations, with simulated time, used typically for studies
described in section 9.2.1
➙ On a cycle-by-cycle basis, physical equations in continuous time; used typically for studies described
in section 9.2.2
Fundamental frequency RMS studies investigate wider system dynamics with detailed models of the AC
networks and functionally accurate models of the DC controls. They model large AC networks accommodating
hundreds or even thousands of buses along with AC machine controllers and dynamic load models. DC controls
are modeled as functional blocks representing the proprietary controls in conjunction with standard DC
equations complementing algebraic AC system equations. Harmonics and most non-linearities are excluded.
Studies in the cycle-by-cycle domain use simpler AC system models, but more detailed models
of the HVDC controls. They use full models of the controls including phase locked oscillators, firing
pulse generation etc., driving 3-phase (explicitly represented) thyristor bridges. This type of
model enables a more detailed examination of the converter station behavior where harmonic effects may
be significant. The AC network may be up to a few tens of buses in size, with physical models including
non-linearities.
The final stage of investigating the DC controls for a project will involve testing of the actual control
software running on the real control hardware, in conjunction with a real-time digital simulator (RTDS) to
emulate valves and AC networks (see section 9.5.3). Usually, some sort of reduced equivalent AC network
has to be derived for use on the simulator, as the simulator has limits on network size governed by how
much simulator hardware is available.
It is from the full AC network data provided for fundamental frequency RMS studies that the equivalent is
derived, first as a static load flow equivalent, then finally including some actual machines closest to the
converter bus with aggregates of remote machines derived to form equivalent electro-mechanical responses.
The resulting equivalent network is compared with wider system representation for authentication in the
fundamental frequency domain, and then re-modeled in the cycle-by-cycle program (e.g. PSCAD/EMTDC)
where some initial performance checks may be carried out on the detailed model of the DC controls.

CHAPTER
Typically, a number of the test study cases are performed using a cycle-by-cycle program prior to simulator
tests in order to check that the proposed controls and settings are appropriate.
The equivalent network as used for the initial performance work is then finally transferred to the simulator
for testing of the actual control software.
9.2.1. Fundamental Frequency RMS (FFRMS) Approach

The software choice for this type of analysis is rather limited. Two examples of suitable software
are PSS/E ® and DIgSILENT PowerFactory. For project studies, a full AC system is generally
provided along with machine controllers and any other dynamic models they may have.
The HVDC model is then connected to the appropriate busbars within the network.
Studies such as Transient Stability, Fundamental Frequency Transient Overvoltage, and Dynamic Interaction
may be carried out to include the effects of the wider AC system.
The models written within GE Vernova for analysis of HVDC dynamic behavior simulate very closely, the
actual inner (fast) control functions of the actual (physical) controller in addition to the outer (slower)
power control functions. In other words, they are not merely a generic representation that provides some
degree of functional equivalence to key aspects of an overall physical controller.
This level of modeling uses actual translations of the transfer functions of the physical controller. These
transfer functions cover input signal measurement transducers, filters, feedback stabilizers, intentional delays
and a representation of the phase locked oscillator.
For a basic understanding of these, it is recommended that a textbook on controls theory be referred to [5].
Some functionality present in reality, and modeled in the cycle-by-cycle domain, is not possible to model in
this way: for example, balancing of power in the poles of a bipole to minimize mid-point return current. This
is because explicit modeling of the exact topology of a HVDC scheme in a load flow is neither possible nor
strictly necessary. The main concern is the terminal conditions at the converter busbars. This means that
the dynamic models produced for use in fundamental frequency RMS analysis do not reflect fully the control
hierarchy present in reality.
The required mix of DC equations and control algorithms may be conveniently split into three levels of
modeling:
➙ Converter dynamic equations model
➙ Converter control algorithms model
➙ Converter DC line model
The converter equations are similar to standard equations (see section 2.2). They may, however, be
extended to include transient DC effects within the bridge for adjacent faults, as the internal components
of the bridge are not modeled at this level (see Appendix A9.1).
The converter control algorithms cover proprietary methods, which vary between manufacturers, and will
usually include the static characteristics for the HVDC scheme (section 5.4), the fast control loops and an
approximate power control function.
The DC line model may be a simple R/L model suitable for a back-to-back HVDC scheme, or may allow for
cable/long lines including effects of charging current that may be relevant for transmission schemes. The
capability of placing a DC-side fault may also be included in this model, which may typically be a single
T-section.
The relationship between the three models is illustrated in Fig.9.2a, indicating the interactions between
themselves and the AC networks at each end.
A number of auxiliary models may be included as dictated by project requirements, some examples of
which are:
i. Power Oscillation Damping Controller (POD)
i. Frequency Controller (FC)
ii. Reactive Power Controller (RPC)
iii. Power Voltage Stability Controller (PVSC)
iv. Converter transformer tapchanger

CHAPTER
DCÊline DCÊline
model model
Vdr,Êldr Vdi,Êldi
ConverterÊsteadyÊstate
equations
∆ÊVd, ∆ÊVd,
α,ÊP,ÊQ,ÊId,ÊVd,Êµ ∆Êld, ∆Êld, γ,ÊP,ÊQ,ÊId,ÊVd,Êµ
Converter ∆ÊVac, ∆ÊVac, Converter
dynamic ∆γ, ∆γ, dynamic
equations CommÊfail CommÊfail equations
DCÊcontrol DCÊcontrol
∆α,ÊPdo,ÊIdo ∆α ,ÊPdo
algorithms algorithms
Fig. 9.2a– HVDC FFRMS level models interaction
The dynamic models take their initial operating conditions from a load flow containing the steady-state condition
of the HVDC link. A key consideration when writing models for this form of analysis is ensuring that models
calculate their internal initial conditions reasonably accurately, with special care where integrators are used.
The output of each (explicit and implicit [5]) integrator is its state, and its input is therefore {d(state)/dt}. This
9.2a
has to be a very small value at initialization (nearly zero).
9.2.2. Cycle-by-Cycle Approach

The model is built typically via a graphical user interface (GUI) that includes a solution engine. This enables
the user to schematically construct a circuit, run a simulation, analyze the results and manage the data
DispatchÊcenterÊ/ÊSystemÊcontrol
StationÊcontrol
BipoleÊcontrol
PoleÊcontrol
ConverterÊgroupÊcontrol
Filters Filters
ACÊSystem ACÊSystem
Filters Filters
Fig. 9.2b– Typical DC controls hierarchy

CHAPTER
in a completely integrated, graphical environment. For these studies, several well-known packages are
available such as PSCAD/EMTDC, EMTP, ATP, EuroStag, etc.
Power system equipment is interfaced with control elements to produce the model employing a building-
block approach. Certain functions of the model may be written as code where using the graphical elements
to construct a function is arduous. The building-block method lends itself to the hierarchical approach of
the controls already described in section 7.11, the key functions of which are described below and shown
hierarchically in Fig. 9.2b.
9.2.2.1. System and Station Control

System control (also referred to as master or station control) is responsible for interfacing possibly with
a Remote Dispatch Centre (RDC) and distributing power demands and ramp rates in accordance with the
remote operator’s demand. This may not be modeled as such, but certain orders may be set and considered
to be provided via this level. It also houses power demand override, AC voltage control, frequency control,
power modulation or Power Oscillation Damping (POD) control, Reactive Power Control (RPC) and filter
switching control (open loop vs. DC power order and TOV control).
The POD is implemented into the HVDC control system as determined by system studies to provide positive
damping to adjacent line real power. It usually operates to damp very-low-frequency inertial oscillations,
typically in the range 0.2 Hz to 2.0 Hz. The POD responds to changes in absolute phase of the converter busbar
voltages and damps using changes in DC current. It acts to modulate the pole power order and hence current
order (Iorder) within certain limits to damp electro-mechanical oscillations.
9.2.2.2. Bipole Control

Where applicable, bipole control is next in the hierarchy after the system control and its main purpose is to
distribute pole current orders to the two poles of the converter facility in response to operator demands.
The bipole control contains the Master Power Control (MPC), which is the recipient of operator or station
master power demands and provides the rate at which the power will be ramped. It is also responsible
for pole current balancing to minimize the ground return current. The sign of the master power demand
indicates the direction of power flow.
The main functions included in bipole control are:
➙ Master power control
➙ Power order distribution between poles
➙ Power limits
➙ Reactive power control (if not in station control)
➙ Bipole balancing
The reactive power control is responsible for switching the AC filters in or out, as per DC power levels. It also
controls reactive power exchange with the AC systems, and AC voltage control as required.
9.2.2.3. Pole Control

Pole control implements the higher level automatic functions, including voltage order, pole power control,
overload control, and power limits (if these are not implemented in bipole control). Also included are current
and voltage order coordination between the two ends and the Sub-synchronous Damping Control (SSDC).
The pole power control maintains the transmitted DC power to a given order, which is achieved by a
combination of DC voltage and DC current control. The current order is determined by dividing the power order
by the voltage order, which in turn can be determined by a reactive power controller, AC voltage controller or
operator demand. Voltage order is used instead of measured DC voltage to avoid unnecessary logic associated
with converter blocked and fault conditions.
The power order interface provides an interface to pole power control for either master control or independent
control mode orders. The interface receives a positive or negative signed pole power order and power ramp
rate and determines from the signed power order which side is operating as rectifier. The received power order
and power ramp rate is also used to determine when a block or deblock request is issued to the block/deblock
control residing under converter group control. For an installation with more than one 12-pulse converter unit
per pole end, sequencing is needed to block/deblock the converters in the proper order.

CHAPTER
The SSDC is occasionally employed, if needed. It damps torsional shaft oscillations of nearby generators in
a sub-synchronous frequency bandwidth, typically in the range of 15-40 Hz (see section 9.3.2.2).
9.2.2.4. Converter Group Control

The functions in Converter Group Control (CGC) deal with the operation of an individual converter. They
are aimed at providing a robust and efficient conversion process (rectification and inversion). The functions
contained are phase control including static characteristics, tapchanger control, block/deblock control and
interfaces with valve base electronics.
Vac Timing ValveÊwinding

V123 voltage PhaseÊlimits
ÊMeasurement di/dt
α<2¡ α>182¡
Logic
Vdc A(s)
Limits
Control
variables
Selector
Idc B(s) Vin PLO Foul FireÊpulse

andÊloop stream
orders
LoopÊcontrol Selector
Fig. 9.2c– Simplified diagram of the control loop within phase controls
➙ Phase control is the core of the control system. It implements the fast control loops that act in
response to control orders and limits in order to determine the firing instants of each valve of a
12-pulse converter in sequence. Collectively, these control loops implement the converter transient
and static characteristics which also depend on the state of the AC systems. Phase controls employ
9.2c oscillator control method (section 1.2.1.3), whose overall principle is shown in
the phase locked
Fig. 9.2c. In practice, there are a number of dynamic control loops which contribute to the selector,
whose decision logic selects the error that ramps at the fastest rate, thus ensuring the quickest
response by the PLO to create the optimum firing delay.
➙ The tapchanger control will send tap up/tap down commands to the tapchanger interface to implement
the tapping. Depending on scheme design, converter firing angles, DC voltage or valve winding voltages
may be controlled within limited steady-state ranges.
➙ The block/deblock control responds to high-level control orders and issues block and deblock
sequences to start or stop the converters. It observes a large variety of protection and safety
sequential schemes as well as logical state conditions.
For modeling purposes, Valve Base Electronics (VBE) simply contains the firing pulses to the valves. It receives
the firing information from phase control.
An overview of a HVDC model in the cycle-by-cycle domain is presented in the load flow (see chapter 5,
Fig. 5.12a) for a single pole HVDC scheme. The block labeled master control in this figure contains settings
for DC power order, rate of DC power ramping and the switching in of filters depending on DC power. The
block labeled control in the same figure contains pole control and converter group control.
The model in the form outlined in this section is used to perform various studies focusing on the HVDC scheme
and its controls. These studies may include dynamic performance, transient overvoltage and surge arrester
rating investigations. Through such studies, the controls can be subjected to various scenarios in which
harmonic and unbalance effects are present, and thus build confidence in the robustness of the controls and
settings for ultimate verification in the RTDS.

CHAPTER
LPÊturbine Gen. Exc.

Jii = Mi = Inertia values and IPÊturbine
Dii = C orresponding HPÊturbine
damping parameters DÊ12 DÊ23 DÊ34 DÊ45
Kij = Stiffness values and
Dij = Corresponding
damping parameters JÊ11 JÊ22 JÊ33 JÊ44 JÊ55
w i = Rotational speed
d ij = Shaft angular twist
between masses Mi KÊ12 KÊ23 KÊ34 KÊ45
and Mj
DÊ11 DÊ22 DÊ33 DÊ44 DÊ55
Fig. 9.2d– Schematic representation of a multi-mass model of a turbine-generator
9.2.3. Modeling to Study Sub-synchronous Oscillations

The phenomenon of Sub-synchronous Oscillations (SSO) of turbine shafts, excited by 15 to 30 Hz currents
from DC9.2d
rectifier current controllers is fully explained in section 9.3.2.2. Here the aspects of digital modeling
to assess their extent (once they are confirmed by initial screening analysis) are discussed.
A turbine-generator consists of a set of separate rotor masses which are joined together via shafts of different
mechanical stiffness as well as self and mutual damping. From the commonly known mechanical dynamic
equations for torque, it can be shown that each shaft section has a torsional mode, which can be excited
after a disturbance.
An essential power system component in SSO studies is the turbine-generator multi-mass system shown
schematically in Fig. 9.2d.
An electrical analog (Fig.9.2e) of the mechanical shaft equations is usually used to estimate resonant shaft
torsional oscillations:
Turb. Gen. Exc.
L2(M2) L1(M1) L3(M3)
1 1
Tm (C 12) (C 13)
K 12 K 13
Fig. 9.2e– Schematic representation of a multi-mass model of a turbine-generator
Kω [Eqn. 9.2a]
T = Msω + Dω +
s
This is equivalent to the time domain LCR circuit voltage equation:
9.2e
CHAPTER
di 1 [Eqn. 9.2b]
V=L + R ⋅ i + ∫ i dt
dt c
The torsional movements occur in the shafts between the different rotor masses. In the case of n rotor
masses, there are (n-1) shafts which results in (n-1) mechanical natural frequencies. Every torsional mode
has a certain natural frequency (eigenvalue) and a certain mode shape (eigenvector shape of torsional
vibration). The greater the mass of the generator compared to the mass of the turbine, the smaller the
influence of the electrical torque on the relative rotational displacement.
A transfer function block diagram can be constructed for the general arrangement of a Turbine Generator
(TG) connected to the rectifier end of a HVDC scheme. Fig. 9.2f represents such a block diagram where
the electrical torque experienced by the TG set is the variable of interest and for compactness the Laplace
operator ‘s’ has been dropped from the transfer functions.
Tm + - FTG FωT
+ Te
+
FΙT
Io
FωΙ
+ Idc - + -FREG
α
+
-FαI
Fig. 9.2f– Transfer function block diagram for turbine generator/HVDC SSO
There are essentially two paths which affect the electrical torque Te. The first is determined by the coupling
of the TG with the AC network. Here the TG and shaft dynamics are in series, with a term FwT (s) which
9.2fbetween changes in machine speed and the electrical torque experienced by
defines the relationship
the generator. This function FwT (s) is dependent on the AC network to which the generator is coupled:
the stronger the AC network, the greater the gain of this term will be. Hence, for AC networks with high
short-circuit levels, this path will have the major influence on the behavior of the TG. Conversely, if the TG
is connected radially to the HVDC scheme, then this term will be very small and may be negligible.
The second path is the influence of the machine speed variations on the electrical torque via the rectifier
and its DC current controller. Changes in w are manifested as changes in Idc through FwI (s) as the AC voltage
varies. Subsequently, Idc impacts on Te through FIT. However, the rectifier current regulator FREG modifies the
firing angle a in a closed loop attempting to maintain Idc at the current order Io. Thus, the rectifier current
control dynamics are embedded in this path of influence.
Legend:
Tm = Mechanical torque applied by steam turbine
Te = Generator electrical torque
w = Generator angular speed
a = Converter firing angle
Io = Rectifier current order
Idc = DC current
FTG = Turbine generator and shaft mechanical dynamics

CHAPTER
FwT = Turbine generator speed/electrical torque transfer function
FwI = Transfer function relating TG speed to converter current
FIT = Transfer function for effect of converter current on TG electrical torque
FREG = Converter current regulator loop transfer function*
FaI = Transfer function relating changes in converter valve firing angle to DC current*
* Note: a negative sign before the transfer function indicates that an increase in the input variable delivers a decrease in the output
variable.
The main parameters of influence are:

➙ Turbine mechanical parameters (natural frequencies and mode shapes)
➙ Strength of the electrical AC system
➙ DC power level
➙ Converter firing angle a
➙ Current control loop parameters
Low AC system short-circuit levels result in reduced damping of sub-synchronous oscillations. High values
of firing angle a and high DC power flows also reduce the damping capability.
FTG could thus contain the generator/governor/AVR model: FwT mirrors the machine swing equation (with
inertia H ∝ 1/SCL), FaI is the DC current controller model and FIT simply calculates the nodal electrical (real)
power corresponding to Idc.
Such equivalent modeling is usually not necessary, as the actual full system may be employed with all its
implicitly inherent interactions, even though that may imply too detailed a model for the purpose of SSO
assessments where only the dynamic behavior of w is to be assessed. This type of modeling may be done on
PSCAD/EMTDC, however, for ease of tuning of a damping controller (see section 9.3.2.2) such investigations
are best done on the RTDS.
9.2.4. Comparison of Models

It may be appreciated from the previous discussion, that there are some significant differences between
the modeling of the DC controls and AC networks in a cycle-by-cycle program such as PSCAD/EMTDC,
and a fundamental frequency RMS program such as PSS/E. This may show some differences in detailed
results even where the same study scenario is being explored between the two different software types.
The following table summarizes the key modeling differences:
Fundamental frequency
Cycle-by-cycle model
RMS model
Network representation Positive sequence full network 3-phase representation; reduced equivalent
Some adjacent machines as real machines
including their functional controllers; but includes
Functional controllers for AVRs,
Machine controllers equivalent generators tuned to give overall
governors and stabilizers
similar electro-mechanical response to full
network
No harmonics considered; sub- Includes harmonic effects; also sub-synchronous
Harmonics
synchronous oscillations only oscillations
DC equations; detailed
Full DC controls, transducers, necessary
DC controller control algorithms; functional
protection
representation of PLO
Fault application and line
Performed at designated time Current zero observed before breakers clear
switching
Table 9.2a– Comparison of fundamental frequency and cycle by cycle models
As an illustration, Fig. 9.7a and Fig. 9.7b provide plots from both types of software for a rectifier converter
bus fault and an inverter converter bus fault. In general, the correlation is good. The potential for harmonics
in the cycle-by-cycle model is obvious with more noisy traces.

CHAPTER
9.3. DYNAMIC PERFORMANCE STUDIES
Dynamic performance studies assess the performance of the DC link and its controls whilst interacting with
the rest of the system (which may be the adjoining AC systems perhaps containing other DC schemes). To
assess DC link behavior, two types of dynamic performance studies are normally performed:
1. Transient and dynamic stability study using the wider AC system and a functional DC model; this is
suitable for an algebraic equation-based simulated time fundamental frequency RMS application
2. Controls dynamic performance assessment study using a reduced AC system and a detailed DC model;
this is suitable for a physical equation-based continuous time cyclic application
However, it must be noted that both assessments A and B address issues within the remit of the subject
of transient stability, whose theoretical basis applies
to both types of studies.
Vr/Es
1.20
9.3.1. Transient Stability
In order to put the description of transient stability 1.00
STABLE
and dynamic performance studies into proper
context, it is helpful to have a brief recap of the
0.80 cosΦÊ=Ê1.0
conventional concepts of AC system transient and
dynamic stability. Then one can relate the beneficial cosΦÊ=Ê0.95
cosΦÊ=Ê0.90
impact of a typical conventional HVDC scheme to 0.60
the overall system’s transient and dynamic stability.
UNSTABLE
0.40
9.3.1.1. Traditional PV, QV and Rotor
Angle Stability in AC Systems
0.20
HVDC control systems are effectively a sub-set of
FACTS device controllers. The main requirements
of FACTS control in achieving transient stability are 0.00
as follows: [6] 0.00 0.20 0.40 0.60 0.80 1.00 1.20
Pr/Prmax
➙ The system should maintain stability after a
major disturbance Fig. 9.3a– Typical per unit PV curves across an AC transmission circuit
➙ The system should not remain near the for varying load power factor
maximum of the first swing
➙ The rotor backswing should be small
➙ Subsequent oscillations should be effectively
Qinj/Prmax
damped UNSTABLE STABLE

0.50
Basically, these considerations start from an
appreciation of the well known PV and QV stability Ê9.3a
limits (for quasi steady-state and dynamic stability) 0.30
and generator power-angle (P-d) considerations for
Pr/PrmaxÊ=Ê1.0
transient and dynamic stability. Here they may be Pr/PrmaxÊ=Ê0.9
0.10
considered for AC lines and generators in close
proximity to a DC converter station.
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20
Where: -0.10
Pr/PrmaxÊ=Ê0.75
CosF = Load power factor
Es = Transmission line sending end voltage Pr/PrmaxÊ=Ê0.6
-0.30
Pr = Real power of load
Pr/PrmaxÊ=Ê0.5
Pr max = PV stability limit real power transfer
Qinj = Reactive power of load at receiving end -0.50
Vr/Es
Vr = Transmission line receiving end voltage
Fig.9.3a shows the system is stable in the region Fig. 9.3b– Typical per unit QV curves across an AC transmission circuit
for varying transmitted power
where the derivative dP r /dV r is negative, and
unstable in the region with the positive derivative,

CHAPTER
9.3b
where increasing Pr results in increasing Vr. The critical operating condition corresponding to maximum power
represents the limit of the power transmission beyond which voltage will reduce rapidly and consequently,
so will power. The power transfer limit (critical operating point) is reached when the derivative dPr/dVr
equals zero.
Fig. 9.3b also shows the system is stable in the region where the derivative dQinj/dVr is positive while in
the negative derivative region, increasing Qinj results in decreasing Pr. The voltage stability limit (critical
operating point) is reached when the derivative dQinj/dVr equals zero.
Fig. 9.3c shows a typical power-angle (P-d) profile for a radially connected single generator radial supply.
The generator P delivery equation in the transient period may be written as: [6]
E ′ ⋅ Vt ⋅ sin δ  Vt 2   1 1 
Pg = +  ⋅ − ⋅ sin 2δ [Eqn. 9.3a]
Xd ′  2   Xq Xd ′ 
Where:
Pg = Generator real power delivered
E′ = Transient internal EMF of generator (air gap EMF)
X′ = Transient reactance on the stated axis (‘d’ or ‘q’)
Vt = Terminal voltage
aÊ(pre-fault)
XÊL XÊL Pmax
m cÊ(post-fault)
Amargin
1 2 A2
Fault Vr P1
A1
Vs bÊ(during-fault)
XÊT XÊL XÊL XÊT
3 4
0 δ1 δ2 π/2 δ3 δcrit π δ
Fig. 9.3c– P-d diagram for simple radial generator feed
9.3cÊ(01)d = Generator rotor angle (between E′ and Vt)

Assuming Xd′ = Xq′ (= Xq for salient pole generators), we have:
Pg
E Vt sin 9.3cÊ(partÊII) [Eqn. 9.3b]
Xd
This is also valid for round rotor generators and there is no loss of accuracy for transient stability
considerations.
Immediately after a disturbance, the above equation indicates a generator operating point on the power angle
diagram, which shows the behavior of generator real power versus rotor angle. The relationship is defined by
the machine inertial equation:
••
δ = k ′(Tm − Tg ) [Eqn. 9.3c]
Where:
Tm = Mechanical Torque
Tg = Electrical Torque
k′ = Proportionality Constant

CHAPTER
liÊchangeÊdueÊto: EiÊchange ∆ÊTÊ(T-) ωÊchange EfÊchange
YÊchange δÊchange
PgÊchange VacÊchange Machine/SystemÊ

angleÊchange TmÊchangeÊby EiÊchangeÊby
GOV AVR/STAB
AtÊconverter
bus
Tap
changer
TgÊchange Qg/Qsystem
change
∆ÊTÊ(T+)
δÊchange
RPC PVSC POD,ÊFC PDO
∆α,Ê∆γ,Ê∆ldÊinÊDCÊscheme
PloadÊchange Q loadÊchange
PdcÊchange QdcÊchange
Σ
ChangeÊinÊnetworkÊflows (ForÊconverterÊbuses)
Where:
α =ÊDCÊrectifierÊcontrolÊangle
δ =ÊÊMachineÊangle
γ =ÊÊDCÊinverterÊcontrolÊangle
ω =ÊMachineÊspeedÊdeviation
AVR/STABÊ=ÊMachineÊvoltageÊregulator/stabiliser RPCÊ=ÊÊ ReactiveÊPowerÊController

EiÊ=ÊÊ InternalÊvoltage Tg,ÊPm,ÊTmÊ=ÊGeneratorÊTorque,ÊPrimeÊMoverÊPowerÊandÊÊTorque
EfÊ=ÊÊ FieldÊvoltage VacÊ=ÊÊ ACÊsystemÊvoltage
GOVÊ=ÊÊ MachineÊspeedÊgovernor PVSCÊ=ÊÊ PowerÊVoltageÊStabilityÊController
IdÊ=ÊÊ DCÊcurrent YÊ=ÊÊ SystemÊadmittance
liÊ=ÊÊ InjectedÊcurrent
Pdc/QdcÊ=ÊÊ RealÊandÊreactiveÊpowerÊinjectionÊdueÊtoÊDCÊlink
PDOÊ=ÊÊ PowerÊDemandÊOverrideÊ(Runback)
Pi/QiÊ=ÊÊ RealÊandÊreactiveÊpowerÊinjection
POD/FCÊ=ÊÊ PowerÊOscillationÊDamping/FrequencyÊControllers
Fig. 9.3d– Computational AC-DC stability interactions
9.3d
CHAPTER
The generator’s dynamic interaction with the rest of the system producing system power oscillations and
other parameter changes can be calculated in a step-by-step digital computation, as depicted sequentially
in Fig. 9.3d, where the items in green relate to the AC system and those in red to the DC system. This
incorporates the DC system with the traditional AC system transient stability interactions and the impact
of the latter is described in the next section.
It should be noted that the machine torque Tm is itself determined by its governor responding to shaft
speed. Variations in the generator excitation system which responds to AC terminal voltage changes
(and control field voltage) lead to generator flux (and hence sub-transient voltage behind reactance E′′)
variations.
Post fault system and machine transient stability is governed by the generator first swing capability and by its
(transient) stability margin measured from the instant of fault clearance to that corresponding to the critical
angle (see Fig.9.3c). Beyond this, electrical power absorption area is less than mechanical power input area
and the generator accelerates into instability and has to be tripped by overspeed (overfrequency) protection.
Mathematically, this instability is explained with reference to the swing equation as follows:
••
Since δ = k ′(Tm − Tg )
T herefore, dd/dt = k′(Tm – Tg)Dt – and this gives a positive result beyond the critical angle, as d increases
with time and hence Tm > Tg.
Also dP/dd = k′′cosd and from the P–d diagram, this is clearly negative beyond the critical angle.
Hence dP/dt = dd/dt × dP/dd, is negative.
And since dP/dt = k′′cosd × {k′(Tm – Tg)Dt}.
Then Tm > Tg, since cosd is negative.
Thus, the mechanical input torque exceeds the generator electrical (or braking) torque, and the difference
continuously accelerates the machine into instability.
Fig. 9.3c demonstrates in general the concept of the prime mover accelerating area A1 and the load power
decelerating area A2. From a rotor angle starting point d1 on curve ‘a’, a fault reduces electrical power to
curve ‘b’, where the rotor angle moves from d2 (point of fault clearance) to d3 before settling. For stability,
angle dcrit cannot be exceeded, otherwise A1 > A2 which will cause acceleration instability. The area
available permitting safe recovery, Amargin is called the stability margin.
While first swing rotor angle transient stability defines the ability of a system to recover from a major
disturbance within the first few post transient cycles - dynamic stability recovery from a relatively minor
disturbance over a longer duration indicates the damping characteristics of the system: i.e. its ability to
damp oscillatory behavior over the longer term and hence the eventual maintenance of synchronism.
In order for such stability to be maintained, the power system should be operated with a reasonable margin
to its steady-state voltage stability limit, thus providing adequate margin to accommodate subsequent
dynamic power (rotor angle) swings. This level is shown as P1 in Fig. 9.3c above and its impact upon the
stability margin area (Amargin) can be readily appreciated.
The above explanations indicate the importance of bus voltage oscillations in the post transient period. These
are heavily influenced by the type of load models.
The total load power (MVA) takes the form:
SL = A + B ⋅ V + C ⋅ V 2 [Eqn. 9.3d]
The variables in Eqn 9.3d indicate constant power, constant current and constant admittance constituents,
with A + B + C = 1.0 per unit. If the trend is for A > B > C, then large post transient voltage oscillations may
result in decreasing voltage, causing increasing line currents and larger voltage drops. Conversely, if C > B
> A then decreasing voltage causes decreasing line currents and smaller voltage drops.

CHAPTER
9.3.1.2. Influence of DC Schemes on AC System Stability
Fig. 9.3d shows the AC/DC interactions during a transient stability computation. Note that the computations
basically use steady-state DC equations as developed in section 2.2. The firing angles are updated at each
computational time step.
During the transient and dynamic periods, the changing AC voltage and angle at the converter bus will
invoke controlled changes in converter firing angles, DC voltage and DC current at both ends of the DC
link. These are compounded by the various different control functions in the HVDC scheme as shown in
Fig. 9.3d. They will in turn, change the real power delivered and reactive power absorbed by the converter.
In the PSS/E dynamic model, this is equivalent to a change of bus-connected impedance or a change in
injected AC current at each converter bus:
Iac =
(P + jQ)conj
Vac conj [Eqn. 9.3e]
Where conj indicates complex conjugate.

In the PSCAD/EMTDC type of study, this is manifested as a change of DC voltage, current and hence real and
reactive power, which dovetails with the adjacent AC system commutations.
To restore transient and dynamic stability, fast control of ordered DC power is a very important feature. Rapid
control of rectifier firing angle and inverter extinction angle can provide the following features: [7]
➙ Respond to fast changes in ordered power to enhance stability following normal disturbances in both
separate (tie line) as well as AC/DC parallel systems
➙ Promote fast recovery from faults, thus enhancing transient stability
➙ Enable quasi steady-state AC voltage control by Q-control
➙ Maintain near constant Q at converter buses, thus stabilizing post dynamic AC voltages especially in
weak AC inverter systems
➙ Direct a near constant power/current supply corridor
➙ Provide reactive power exchange (power factor) control
➙ Follow AC line PV characteristics through power voltage stability control action
➙ Respond to changes in AC/DC interface power thus providing Power Oscillation Damping (POD)
➙ Respond to variations in converter bus frequency to provide drift control at the converter buses
The so-called inner level faster controllers actually modify the firing instant by feedback error control loops
and promote transient stability. The slower power level controllers contribute typically via: Power Voltage
Stability Controller (PVSC), Power Oscillation Damping and Frequency Controller (POD/FC), Reactive Power
Controller (RPC) and Tapchanger Controls (TC) and work towards improving dynamic stability.
The PVSC changes power order in a DC scheme in sympathy with voltage changes, thus promoting PV stability.
The POD/FC changes DC power order in accordance with converter bus voltage angle and frequency changes
respectively. The RPC targets power factor control or slow AC voltage drift control at the converter buses. Finally,
the TC can target valve side voltage or control angle, as a means to provide Q exchange trimming or coarse AC
system load angle control. These measures improve dynamic stability for all levels of power transmission.
An effective way of assessing the influence of a DC link on AC system recovery from a transient event is to
compare the abilities of a typical HVDC scheme against a pure AC connection. Improvement of both transient
stability and dynamic damping is best demonstrated by a set of plots from a dynamic simulation.
Transient voltage and adjacent line power following an AC system (rectifier) bus 3-phase solid fault is
shown in Fig. 9.3e. These responses show the case with a zero impedance AC tie line only and are overlaid
with those for a replacement DC link.
Without a DC link separating the two AC systems, the fault drives low inertia generating units into unstable
oscillations. With the DC link, the fault on one side is isolated from such generation on the other side and this
separation stabilizes overall system post fault oscillations.

CHAPTER
ÊACÊtie
350 ÊDCÊtie
Ê300
250
Ê200
Ê150
LineÊpowerÊ(MW)
Ê100
Ê50
0
Ê-50 Ê1 Ê2 Ê3 Ê4 Ê5 Ê6 Ê7 Ê8 Ê9 Ê10
Ê-100
Ê-150
Ê-200
Ê-250
Ê-300
Ê-350
TimeÊ(s)
Ê1.1 ÊACÊtie
ÊDCÊtie
Ê1.0
Ê0.9
Ê0.8
VoltageÊ(p.u.)
Ê0.7
Ê0.6
Ê0.5
Ê0.4
Ê0.3
Ê0 1 2 3 4 5 6 7 8 9 10
TimeÊ(s)
Fig. 9.3e– Transient recovery improvement with HVDC link spanning unstable AC tie
290 LineÊMWÊnoÊPOD
280
Ê9.3e LineÊMWÊwithÊPOD
270
260
250
LineÊpowerÊ(MW)
240
230
220
210
200
190
180
170
160
0 1 2 3 4 5 6 7 8 9 10
TimeÊ(s)
Fig. 9.3f– Adjacent line power oscillation damping using HVDC POD

CHAPTER
9.3f
IdcÊorder - IdcÊloop
+
R1
IdcÊmeasured
Minimum
VacÊmaxÊsetÊpoint +- VacÊloop value a
R4
VacÊmeasured
αmaxÊsetÊpoint +- αmaxÊloop
R5
αÊmeasured 179¡
Minimum k
γrecÊminÊsetÊpoint - γrecÊloop value s
+ f
R3
γrecÊmeasured
1¡
Minimum
value b
γinvÊminÊsetÊpoint +- γinvÊloop e
R6
γinvÊmeasured
Minimum Minimum
value c value
VdcÊmaxÊsetÊpoint - VdcÊloop
+
R7
VdcÊmeasured
d
αminÊsetÊpoint +- αminÊloop
R8
αÊmeasured
Fig.9.3g– Input-output block diagram for DC phase loop controls in fundamental frequency RMS (FFRMS) dynamics
application (rectifier end)
9.3g
θRef
θ Input Phase locked

θin Error filter oscillator
PmodÊhigh
Pmod
Low pass Differential Gain
filter pole
PmodÊlow
Fig. 9.3h– Simplified block diagram of power oscillation damping (POD) control

CHAPTER
9.3h
A typical plot of AC line power adjacent to a DC link whose dynamic oscillations are reduced by a POD is
shown in Fig. 9.3f. The POD is disabled for the early recovery transient period so that its differential pole
will not create any instability.
Recovery from the fault is instigated by a HVDC link, which reflects both the main HVDC control’s response
as well as its associated POD action. The recovery is a result of controller action transiently invoking
minimum alpha control at the rectifier and current control at the inverter. An example of HVDC dynamic
control loop functions that calculate firing delay is shown in generic form in Fig. 9.3g.
The damping action is a result of the response of the HVDC POD controller, a typical example of which
is shown in Fig.9.3h. This modulates DC power by operating on continuous change of bus voltage (i.e.
transmission) angle.
9.3.2. Dynamic Interaction

Dynamic interaction means the transient or longer-term destabilizing (or stabilizing) influence of one
dynamic state or parameter upon another, in the overall AC (or AC/DC) system.
The subject of dynamic interaction is closely related to transient stability estimates. However, there is the
need to identify and define these interactions with concrete indices and this is required in terms of:
➙ Actual numerical extents of such interactions
➙ Actual system states that participate in specific oscillatory interactions
➙ Factors that physically exacerbate or reduce such interactions
➙ Parameters that exert maximum influence upon such interactions
The factors that contribute to dynamic instability and hence adverse interaction, including those between
AC and DC systems, are: [8],[9]
➙ Electromechanically excited rotor angle driven voltage and power oscillations
➙ Medium order electromagnetic harmonic resonances (< 1 kHz)
➙ Temporary overvoltages including generator self excitation
➙ Series capacitor instigated (and DC controls initiated) sub-synchronous resonance leading to generator
shaft sub-synchronous oscillations
➙ Low frequency controller state interactions between AC/AC, AC/DC or DC/DC
➙ Low frequency transient interactions, instigated by large disturbances, including impact of load
behavior, protection, non-linearities and operating regimes
Most dynamic interactions that result from low SCL supply systems are of a destabilizing nature.
Electromechanical oscillations can be classified as inter-area swing modes (typically 0.1 to 0.8 Hz) or as
inter-generator modes (typically 0.5 to 2 Hz) and these can be assessed by small signal linearized techniques.
These rotor angle oscillations are addressed by the usual AC controllers (AVRs, governors and stabilizers).
The low to medium supersynchronous frequency electromagnetic oscillations arise from shunt capacitance
/ series inductance resonances in the AC system. In an AC/DC system, these can also arise from AC/DC
modulation or AC/DC/AC cross modulation effects. For the former modulation class, the DC switching function
creates 2nd harmonic if there is negative sequence and 3rd harmonic voltages in the AC side, and the DC
modulations are transformed to further 3rd harmonic back onto the AC side [10].
Cross modulation on the other hand, occurs when the DC is a frequency converter and 12 times frequency
difference between the AC systems occurs within the DC scheme, leading characteristically to 11 and 13
times frequency difference as the major component to appear on each AC side (see chapter 4).
9.3.2.1. Self Excitation

The phenomenon of machine self-excitation [11] can occur if a radially connected synchronous generator
is virtually isolated with a predominantly capacitive load: for example, filters left connected (which may
sometimes result from minimum filter requirements or a failed filter breaker) after a DC link is blocked,
or perhaps at minimum power. This occurs because the predominantly capacitive leading load current
is on the field circuit’s negative D-axis and hence excites negative D-axis flux and hence negative Q-axis
voltage. Normal rotor voltage is on its positive Q-axis due to lagging D-axis air gap flux, since V = d(flux)/
dt and it is this voltage that is typically controlled by the generator’s AVR, not the negative Q-axis voltage

CHAPTER
component. It would therefore be difficult for standard brushless regulators to control and limit such a rise
in generator internal quadrature voltage, if self-excitation conditions occur. This could potentially damage
the machine’s internal insulation. So it is not the DC link that directly causes this problem, it is the filters
that can remain connected once the link is blocked that can create these conditions. Even so, it needs a
radial connection to aggravate the condition and it is rare for generators to be so connected without any
inductive loading, other than in a black-start situation. Radial connection through long lines also causes
the possibility for this to be aggravated, due to the added duty on the generator to absorb additional line
capacitive reactive power.
Unlikely as such a configuration is in practice, this condition is sometimes specified and studied due to the
potential for uncontrolled insulation damage it presents if islanded operation is a possibility, so that operating
strategies and protective actions can be put in place as necessary.
One way of explaining this phenomenon conceptually, is to look at the generator as a natural receptacle of
inductive VAr absorption, which is being asked to absorb capacitive VArs under the above pseudo steady-state
condition. Then for a terminal voltage V, the approximate self-excitation occurrence condition is:
V2 V2
>
Xc Xd
Where Xd is its D-axis reactance and Xc is the total external capacitive reactance.
The condition for what is known as Immediate Self Excitation (ISE) then becomes:
Xd > Xc [Eqn. 9.3f]
Clearly, this is helped by having more generators in parallel. All this is, of course, only true if there is no other
inductive load/shunt considered present which reduces Xd and hence the risk of ISE.
If negative field current control is possible, the ISE condition to check becomes:
Xq > Xc [Eqn. 9.3g]
Due to (DC) load rejection conditions, the generator overspeeds and this increase in frequency slows the
(rate of) increase of Xc and accelerates that of Xd. Clearly then, if natural ISE is not to take place, the onset
of self-excitation may be delayed and depending upon the values of reactance, MVA rating, etc., such
delayed self-excitation is known as Non Immediate Self Excitation (NISE), for which a critical over-frequency
can be calculated or predicted. These phenomena are less likely for hydro turbines which have a low Xd
and are invariably large rated units.
If ISE has already occurred, then there is no question of NISE and mathematically, the critical frequency (fc)
will be less than synchronous frequency (f0) to indicate that:
fc = f0 ⋅ √(Xd/Xc), or fc = f0 ⋅ √(Xq/Xc) as applicable [Eqn. 9.3h]
frÊ23ÊHz
f0Ê= Rotor
23ÊHzÊioscln 50ÊHz
Stator
3-phase
winding
T ∝ Istator ⋅ y rotor
Components are (50-23) & (50+23) Hz (for
example)
27 Hz superimposed torque on shaft fo
Undamped due to effective negative slip
Fig. 9.3i– Basic machine SSO phenomenon

CHAPTER
9.3.2.2. Sub-synchronous Interaction
∆ω ∆δ ∆α ∆VdÊ,ÊId IfÊtotalÊphaseÊlag
betweenÊ∆αÊ&Ê∆Te
dueÊtoÊcontrolsÊis
suchÊthatÊreaction
torqueÊisÊinÊphase
withÊ∆ω,Êthen
∆Tm ∆Te ∆Pe ∆Pd instabilityÊisÊcaused
Fig. 9.3j– Simplified explanation of SSO interacting with DC quantities
It is well known that Sub-Synchronous Oscillations (SSO) can occur in series capacitance schemes.
[6],[7],[12],[13] These can also result from the current controllers at the rectifier end (only) in DC schemes
that can, under certain frequency response conditions, inject fairly large currents typically at around 20
Hz to 25 Hz, which is within the control loop’s operating bandwidth. In such an instance a nearby radially
connected and relatively small synchronous generator supplying nearly all the DC load and very little else,
will see the DC scheme working at a fairly large firing angle. The generator current may then be comparable
to the sub-synchronous injection. Sum and difference frequency air gap torques (due to the product of
9.3j
induced sub-synchronous currents and synchronous stator flux) will be incident on the rotor (Fig. 9.3i).
Adverse DC/SSO interaction may also be physically explained by Fig. 9.3j.
Here the interaction of the electrical torque and machine speed, influenced by the DC link, is shown graphically.
Thus large pulsating shaft torques of a damaging nature can result due to the difference component, which
may have a frequency near to the natural (mechanical) resonance of the shaft. Generally, steam turbine
generators are more susceptible to torsional interaction than hydro machines. [14]
The possibility of SSO occurring can be initially assessed by calculating the Unit Interaction Factor (UIF)
which, though only an indication, is nevertheless a strong pointer towards the likelihood of its occurrence
and dictates whether further time domain investigations through a spring-mass model are needed or not:
2
MVAdc  SCLg 
UIF = 1− [Eqn. 9.3i]
MVAg  SCLt 
Where:
MVAdc = MVA rating of the DC system
MVAg = MVA rating of the generator
SCLt = short-circuit capacity at the DC commutating bus (excluding AC filters) with the generator
SCLg = short-circuit capacity at the DC commutating bus (excluding AC filters) without the generator
If the UIF is less than 0.1, there is little interaction between the torsional oscillation and the DC controls
and no further study is necessary for this generator. Parallel AC lines tend to minimize the possibility of
ZÊAC1 ZÊAC2
TE ωG
Infinite
bus
ZÊFilters1 ZÊFilters2
Thermal
generator
Fig. 9.3k– Generator radial feed to HVDC rectifier

CHAPTER
torsional instability. Radial DC lines emanating from isolated thermal-generating stations are most prone
to unfavourable torsional interaction.
Where the UIF indicates significant interaction, a time-domain investigation may be carried out with
a detailed model of the DC controls and a generator model that includes some representation of its
mechanical make-up. The ideal vehicle for this study is at the testing of the actual controls in conjunction
with an RTDS simulator. Here the real controls action is combined with a model of the AC system, which
can include a detailed model of a radial generator. Fig. 9.3k shows schematically an HVDC system coupled
with an electrical AC system.
It has been determined that by designing a HVDC control system to ensure positive damping for the unit with
the greatest interaction, positive damping for all other units is ensured. This conclusion is based on the fact
that the magnitude of the interaction varies with the strength of the AC system between a given unit and
the HVDC line, but the phase remains relatively fixed. Thus designing a control system for the most extreme
cases of radial HVDC operation represents a complete solution with respect to changes in the AC transmission
network. Generator units near an inverter terminal do not suffer much destabilization: for some combinations
of torsional frequency and HVDC control characteristics, the inverter control action actually improves the
damping somewhat as compared to a pure AC transmission system. Thus, only units near a rectifier station
need to be considered.
An AC/DC converter will operate as a rectifier, or an inverter, depending on the value of the firing angle a. In
a typical HVDC system, the rectifier usually controls the current while the inverter controls the DC voltage.
With the HVDC scheme in constant power flow mode, the closed loop current control maintains the desired
DC current by varying the firing angle at the rectifier converter. Interaction between the HVDC rectifier
current controller and the torque/speed control of a turbine-generator can introduce negative damping
into the system at sub-synchronous frequencies (5 - 40 Hz). Sub-synchronous oscillation associated with
HVDC schemes is therefore determined by overall closed-loop stability rather than being a true resonance
phenomenon, as in series capacitor compensation schemes.
To enhance the damping capability of a HVDC scheme at sub-synchronous oscillation frequencies, a
supplementary controller can be introduced to the HVDC converter control system. Such a controller is
called a Sub-Synchronous Damping Controller (SSDC) which usually injects a very small supplementary signal
into the current control loop to counteract sub-synchronous oscillations over the relevant frequency band.
Thus additional damping over the sub-synchronous oscillation band is provided by means of an anti-phase
modulation on the converter firing angle. This anti-phase modulation can be derived from any signal in which
the sub-synchronous oscillation is manifested e.g. turbine angular speed, shaft torque, frequency of the AC
commutating bus voltage, etc. The requirement of such a supplementary control loop is of course that it should
not adversely interfere with the other control loops.
Generating stations are normally located quite a distance from the HVDC converter station and therefore it is
not practical to use signals from transducers on the turbine sets. A much more practical solution is to derive
the SSO measurements from variables local to the HVDC converter stations, which can provide broadband
information over the frequency range of interest.
Sub-synchronous damping controllers are designed to provide positive damping to potential sub-synchronous
oscillations in AC systems. An SSDC for HVDC schemes can be functionally divided into two sub-systems,
namely The Measurement System and The Shaping Function. The measurement system is implemented as
a Phase Measurement Unit (PMU) which detects the instantaneous phase angle change in the AC system
voltages and therefore is related to change in power flow. A shaping function is used to band limit the control
signal to the range of sub-synchronous frequencies of interest and thereby provide appropriate damping
signal through feedback into the main HVDC current control loop.
A graphic illustration of its influence is shown in Fig. 9.3l. The first set of graphs shows increasing shaft
oscillatory torque without SSDC, the second set shows a marked improvement with SSDC action.
9.3.2.3. Controller Interaction

Controller state interactions cannot be readily defined. Effectively, two states of a control system in
close electrical proximity could be construed to be exhibiting adverse dynamic interactions when a small
change in one produces an exacerbating effect on the other, in terms of instability. Thus if both states

CHAPTER
are contributing heavily towards an identified unstable mode of oscillation (eigenvalue), then it is a fairly
safe assumption that these two states are adversely interacting. Eigen analysis (see Appendix A9.2) and
controls-parameter sensitivity studies would then reveal the extent of such interacting behavior.
Transient interactions are also studied by the usual time domain transient stability or controls dynamic
performance type analyses, but the difference in emphasis is that interaction assessments set out to identify
specific localized dynamic interactions.
For example, its application could typically be aimed at assessing interactions between two or more DC
schemes in close proximity to each other. In this example, one scheme may be modeled as a benign load
to compare the transient/dynamic response with the study case where all the DC links are fully modeled.
Continued presence of comparable dynamic oscillations would indicate lack of interaction between the existing
DC link(s) and the one whose model was replaced by a simple (benign) load.
Shaft Shaft
torque torque
4 4
p.u.Ê(onÊmachineÊMVA)
3 3
2 2
1 50ÊMW 1
0 0
-1 -1
-2 -2
4 4
3
3
2
2
1
300ÊMW 1
0
0
-1
-1
-2
-2
4
4
3
2 3
1 2
500ÊMW 1
0
-1 0
-2 -1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
TimeÊ(sec) TimeÊ(sec)
Fig. 9.3l– Illustration of the effect of SSDC
AC/DC transient/dynamic interactions can be assessed in an electromagnetic or electromechanical study

environment. The general principles of such interactions in the computation of transient stability at an AC/
9.3l DC interface are shown in Fig. 9.3l.
There are a few specialized techniques used to quantify the indices that can more clearly define dynamic
interaction. These are described in Appendix A9.2.
9.4. TRANSIENT AND DYNAMIC STABILITY INVESTIGATION

The objectives of a transient stability study involving AC and DC systems are to assess rotor angle and
consequent voltage oscillation damping, and hence system stability after the application of a transient
event to the system, in close proximity to the equipment under consideration. Such oscillatory responses
are assessed under the influence of system controls, particularly those of any HVDC scheme being supplied.
The oscillations generally relate to time domain RMS fundamental frequency quantities: mainly in respect
to bus voltages and voltage angles, line real and reactive powers, machine powers, rotor angles and speed
deviation (or frequency deviation), control quantities and parameters concerning AC and DC systems.

CHAPTER
An important objective is the establishment of HVDC operating strategies in connection with transient events
that may cause voltage collapse instability. Included in these objectives are top-level control parameter setting
and their tuning. In this respect, DC power runback and frequency slope setting are assessed as a result of
these studies. Dynamic interaction between relevant controllers in the proximity of the HVDC installation
is also sometimes assessed with modal analysis resulting from the time domain transient stability studies.
9.4.1. Study Software and Methodology

Transient stability studies are carried out using programs such as PSS/E or similar. The program provides
flexible facilities for the representation of large systems, HVDC links and controls, generators (with their
excitation systems and prime movers), and loads - both static and variable.
These studies use the same models and programs as the FFTOV study (Fundamental Frequency Temporary
Over Voltage). For transient stability assessment including a HVDC link, in addition to the DC fast dynamic
control functions (see Fig. 9.3g), this study uses various additional auxiliary DC control features such as
power oscillation damping/frequency controls, reactive power/AC voltage controls, power voltage stability
controls, and tapchanger controls. These feature in Fig. 9.3d.
The basis of transient stability assessment methodology is to combine steady-state analysis (load flow with
voltage sources behind internal impedances) with a dynamic snapshot at time t = 0. The dynamic snapshot
contains the states of the overall system controllers in the s-domain, initialized at t = 0. After a successful
initialization in dynamics mode (the criteria being that dy/dt ≈ 0 (in practice, dy/dt < 0.1 y0, where y0 = initial
state)), a run can be performed in the time domain with transient events applied. Normally, the dynamic
simulation runs for about 10 to 20 seconds.
Transient stability studies complement the EMTDC/RTDS dynamic performance studies, which employ the
reduced equivalent networks (the EMTDC/RTDS studies are described in section 9.5).
9.4.2. Study Components

The stability investigations essentially fall into six key study groups:
➙ Steady-state system operation
➙ Transient stability and recovery following AC system faults and disturbances
➙ Frequency stabilization and power modulation for oscillation damping
➙ Establishment of run-back levels for AC system stabilization
➙ Quasi steady-state reactive power exchange and AC voltage control
➙ Dynamic interaction assessment including eigenvalue analysis
9.4.2.1. Steady-State System Operation

Before carrying out dynamic simulations on a system, it is always necessary to determine realistic steady-
state operating conditions, usually under a number of different prescribed regimes of generation and loading
within the framework for overall AC system operation. These are reflected in technical specifications and often
include the application of credible prior contingencies. Steady-state analysis essentially comprises load flow
results in a load flow output file that define the overall system operation prior to any transient disturbance.
For a study involving a HVDC scheme, load flow calculations are carried out on the complete AC system,
consisting of all applicable AC systems linked by the DC link(s). The simulation can start with any steady-state
DC power including zero, which is specified in the DC data part of the load flow file. This is followed by a short
steady dynamic run before the prescribed transient is applied. Some data aspects and representations will
always need to be resolved and confirmed, and these are usually identified as the system details are set up
for the digital studies.
9.4.2.2. Transient Stability & Recovery Following AC System Disturbances

In the dynamic studies, generators are represented by the d-q axis sub-transient models. Similarly,

CHAPTER
generator excitation systems, stabilizers and turbine generators are represented by detailed models.
Where such data is missing, some typical assumptions may be made.
A number of preliminary transient stability studies are carried out on the AC system, fully represented with
generator excitation control and prime mover action, including the DC schemes and other FACTS devices in the
AC system. One of the objectives, as with load flows, is to confirm system data including dynamic equipment
representations before proceeding further. This is done by observing the effects of a transient disturbance
and the system’s attempts to recover to a new acceptable and expected steady-state.
In some cases it is possible that all excitation systems, governors, stabilizers and DC models provided or
assumed may not yield acceptable non-oscillatory performance even in the absence of the HVDC link or
compensator. In such instances step response tests are carried out with a view to amending some models
or their data.
Once data issues are settled, transient stability of the AC/HVDC system involving its recovery following
application and clearance of faults or other prescribed transient events is carried out. These can include line
removal, auto-reclosure (where applicable), possible converter isolation for HVDC studies, DC power runback
for stability where applicable, etc.
9.4.2.3. Power Modulation for Oscillation Damping and Frequency Stabilization

This aspect of the transient stability study applies to a HVDC scheme assessment, which determines
required control parameters that will modulate the scheduled DC power to dampen prospective dynamic
frequency oscillation. This is termed Power Oscillation Damping (POD) control and is done around the base
frequency originating on either side of the link. In addition, frequency/power slope correction is applied
to immediate frequency drifts (Frequency Control or FC), whereas longer term drift of system frequency
would be corrected by governor control on generators. A few of the cases of the pure transient stability
study, which appear useful for specific assessment of frequency and modulation control, are used to carry
out further dynamic simulations in which system frequency oscillations are induced by the application of
suitable switching events. These are specific cases where POD/FC action may be pronounced, although
most of the other study cases may also invoke some POD/FC action.
The HVDC link controls are investigated to try to utilize the link’s capability to achieve appropriate damping
while meeting other control performance requirements, such as operating at power limits, etc.
9.4.2.4. Establishment of Run-back Levels for AC System Stabilization

This aspect of Transient Stability studies investigates the need for and nature of any reduced DC power
level to which the operation of the HVDC scheme should be adjusted following an outage of equipment or
lines directly adjacent to the converter stations. Technical specifications may also request a consideration
of prior contingencies and may indicate some specific credible outage contingencies near the DC link for
which the reduced (run-back) steady-state DC power level is to be found, at which stable power flow can
be maintained.
Such outages may be applied statically or dynamically, so any necessary DC power run-back is essentially
determined by control action after a transient event leading to a further line or generator outage that
specifically clears the transient event. Thus these studies address the transient behavior of the AC systems
following a disturbance, such as a fault in addition to the prior contingency, until a stable reduced run-back
power level is established. The rates of power reduction incorporated in the Power Demand Override (PDO)
function, which are necessary to achieve the run-back level without loss of stability, are also confirmed in
these studies using the PSS/E or a similar transient stability program. Any necessary signals from the converter
or adjacent substations to indicate the need for specific run-backs are considered.
The Fundamental Frequency Temporary Overvoltages (FFTOVs) corresponding to these outage conditions may
need to be evaluated for these reduced DC power levels. Those run‑back levels, which are not acceptable from
overvoltage considerations, are re‑evaluated and acceptable reduced power levels determined.
9.4.2.5. Quasi Steady-State Reactive Power Exchange and AC Voltage Control

Reactive Power Exchange (RPE) control and AC Voltage Control (ACVC) features are usually included as

CHAPTER
HVDC auxiliary controls, with the basic operation of the link remaining in constant power mode. The use of
the DC link and filter switching to control the net reactive power exchange and converter HV bus voltage
follows the DC scheme design, and is demonstrated by a few specifically designed studies of the transient
stability type. It should be noted that this is an auxiliary control feature which is active during steady-state
but only insofar as it does not limit the normal power order: i.e. power control has precedence over RPE
control. This feature is active only during the quasi steady-state period, that is, sufficiently into the post
disturbance recovery period (three to four seconds after removal of disturbance) which coincides with
slow system reactive power drifts.
It should be noted that the reactive power exchange and AC Voltage Control modes are mutually exclusive
features and both features are usually included in Transient Stability studies, with the basic operation of the
link remaining in constant power mode.
9.4.2.6. Dynamic Interaction Assessment (including eigenvalue analysis)

Stability studies may sometimes need to be performed where they focus on the topic of dynamic
interactions between adjacent DC schemes and with the nearby AC system controllers, or with SVC or
other FACTS device controllers. These studies specifically address the issue of dynamic interaction between
two or more states which contribute adversely (or are assistive) to the same unstable (or stable) mode.
Coordination of control and operation are assessed by dynamic interaction studies, which use both steady-state
and dynamic conditions and concentrate on controllers within, and in very close proximity to, the DC scheme(s).
9.4.2.7. Interaction Assessment by Dynamics

The essential effects investigated in these studies concern mainly the mutual state interaction of HVDC
schemes in close proximity to the converter bus in question, and also the state of interaction between
them and the nearby AC system generators and their controls. The most important elements in the
process are the controls of the HVDC links and AC generators alongside other operational features, such as
overall reactive power control, switching in and out of filters and power oscillation damping and frequency
controllers, if present in the scheme.
Information about the dynamic performance limitations and requirements of the DC schemes are useful in
order to assess any adverse interaction effects on them due to the dynamic control actions of the DC link.
Through this, the link controls may be tuned to meet performance requirements. In addition to the full AC
system details, the DC controls must also be appropriately represented and modeled as for the usual transient
stability studies.
The dynamic interaction studies involve system disturbances (such as load rejection and system short-
circuits) which are designed so that the resulting transients force specific features of the DC controls to react
to major external disturbances and system swings. They also involve alteration of DC operating regimes such
as runbacks, blocking, etc. Studies are often done successively with one of the DC schemes represented as
an AC load so as to judge whether the link’s controls are engaged in any dynamic interaction with the rest of
the nearby system.
9.4.2.8. Estimation of Adversely Interacting States

Those dynamic studies which show possible adverse DC interaction may be subjected to post transient
system eigenvalue analysis, in an attempt to identify adversely interacting DC states (if any) by quantifying
their participation in unstable modes of state oscillations by MPF estimation, as described in Appendix
A9.2. In such studies, linearized small perturbations application to the dynamic set-up calculates selected
state eigenvalues and eigenvectors, which result in state related modal participation factors. These lead
to assessments of adverse or assistive interactions between any two (or more) states.
9.4.2.9. Coordination of Controls

In carrying out the coordination study, the intention is to cover system disturbances which might cause
adverse interaction between integrated HVDC links in any single region. Local disturbances may cause
adverse oscillations at the far end of a long DC line and this may sometimes indicate the necessity of using
the full AC system in these digital studies.

CHAPTER
Whatever interactions are found to exist, attempts are made to identify and optimize the converter controls
through coordination between the DC controllers, so that adverse interactions do not occur and cases
requiring operating restrictions are identified. Possible modes of control to reduce undesirable interactions
are also proposed.
9.5. CONTROLS DYNAMIC PERFORMANCE ASSESSMENT

INVESTIGATION
The overall objective of the investigation is to demonstrate the robustness of the control system and its
dynamic performance, so as to establish stable operation of the HVDC link connected to the AC systems. In
addition, stable recovery from various fault scenarios within either AC system is demonstrated.
The specific objective of this investigation is to demonstrate that the dynamic responses of the HVDC
system to a variety of transient disturbances and changes will ensure stable operation of the link. The
complete electrical characteristics of the DC and AC systems including the controls and other converter
equipment in the HVDC installation are represented in the studies. Thus, the dynamic behavior of the controls
is assessed with more detailed and realistic controls modeling in software such as ATP or PSCAD/EMTDC.
This concentrates on the HVDC scheme and its control system with essential generic DC side protection. All
non-linearities that are not modeled in the transient stability type of work are represented here.
For the AC systems, reduced equivalent networks are required as these studies are meant to lead up to
the Real-Time Digital Simulator (RTDS) tests where the actual control (and protection) equipment dynamic
performance is to be demonstrated and verified. Full AC system networks are too large to be modeled on the
simulator and therefore reduced equivalent networks must be derived in a form suitable for simulator modeling.
A number of well-chosen steady-state and dynamic conditions are then applied to the equivalent network,
for a range of selected key events including energization of the converter transformer, step responses and AC
system faults. Additionally the static characteristics of the control system are verified. The settings derived
and any control adjustments found to be necessary during the course of this study are used in the acceptance
tests performed on the Real Time Digital Simulator (RTDS).
9.5.1. DC Link Dynamic Performance Studies (DPS in PSCAD)

The cases mentioned are typically carried out using a continuous time-domain simulation program (such
as PSCAD/EMTDC). The HVDC control system for these studies is composed of a comprehensive model
of the actual control equipment. All converter station equipment is modeled and interfaced with the AC
networks by power system equipment modeling facilities.
The study titles (A to G) listed and described below are for peak and light load conditions, thus covering
angular as well as voltage directed stability. For the study cases, only E, F and sometimes G need an equivalent
AC network at each end of the link. The other studies may employ a voltage source behind an impedance,
representing minimum or maximum SCL, depending on whether transient overvoltage or transient overcurrent
peak is being estimated.
AC networks can be represented for both peak and light conditions, though usually only the latter is needed
for DPS. The power flow in the DC link is considered in both directions.
9.5.1.1. A – Energization and Tripping of Filters

The effect of the inrush currents is analyzed in these tests at maximum and minimum SCL, following
energization of filters and transformers; their subsequent tripping is also done.
9.5.1.2. B – Step Change in Current Order

These cases check the response of the HVDC controls and the impact on the AC networks to a step change
in current order from steady-state. The change is applied in both directions.

CHAPTER
9.5.1.3. C– Step Change in Power Order
These cases check the response of the HVDC controls and the impact on the AC networks to a step change
in power order from steady-state. The change is applied in both directions.
9.5.1.4. D – Second Pole/Bipole Transformer Energization

For a bipolar or double bipole scheme, these tests will examine the interaction between the energization
of the second pole or bipole converter transformers and the first pole or bipole which has been running
in steady-state.
9.5.1.5. E – Transient Disturbances (Fault and Recovery)

In these tests 3-phase and single-phase faults are applied in close proximity to the converter station to
demonstrate the stability of the combined AC and DC system and the fault recovery capability of the HVDC
scheme. Considerations for auto-reclose facilities have to be applied. In these studies, aspects of post fault
recovery of the overall system are assessed, using the reduced equivalent AC systems. The AC system
faults include converter isolation on the inverter side. DC faults are studied to demonstrate converter
protective strategies.
9.5.1.6. F – Contingency Analysis (Transient Stability)

These studies look at the effect of line tripping on normal and prior contingent system stability and cover
emergency power/frequency control of the HVDC scheme (where present). Traditional aspects of transient
stability covering rotor angle and voltage stability as well as runback and power swing damping (low
frequency modulation) requirements are addressed in these studies.
9.5.1.7. G – Transient and Temporary Overvoltages

The overvoltages due to fault clearance with re-energization from a weakened source are fully covered
in study F above, whilst those solely due to (DC) load rejection are studied in these cases. Dynamic
overvoltages in fundamental frequency RMS terms due to these events can also be evaluated in the overall
system. Limitation of such overvoltages by control action is addressed in these investigations.
9.5.2. DC Link Protection Coordination Studies

The coordination of the DC protections from the AC line side to the DC line/cable usually forms a part of the
DC controls dynamic performance studies. The main objective of this study is to test the DC protections
of the HVDC system during a variety of fault conditions within the above boundaries of the DC scheme. An
EMTDC model of the HVDC scheme with controls, like the example in section 9.5.1, plus a comprehensive
set of DC protections is used to perform the coordination study.
Before each simulation, the EMTDC model parameters must be set correctly. These parameters include power
level and direction, short-circuit level (for each end), the location of the fault and the time duration that the
fault will be applied for.
The DC protections are discussed in section 7.3.
9.5.3. DC Link Dynamic Performance Studies (DPS in RTDS)

The overall objective of the study is to demonstrate the controls dynamic performance of the scheme with
the REAL PHYSICAL controller, using key settings from the results of the PSCAD/EMTDC DPS. Essentially,
tests that define the DC scheme’s dynamic performance, which are similar to those done during PSCAD/
EMTDC DPS, are repeated. The AC system test cases are fewer and more focused, while those for the DC
system include aspects such as starting, stopping, sequencing, etc., as these are more specific to the control
system’s operating strategy, but independent of its overall dynamic performance.
In addition to the tests (A-G) listed in section 9.5.1, the following DC system tests are usually done in the RTDS.

CHAPTER
9.5.3.1. H – Static Characteristics
Static characteristics are plotted to check that the software controls have been implemented correctly
and that the control settings and scaling are correct.
9.5.3.2. I – Ramping of Power and Reversal

These cases illustrate the performance of ramping power up to the continuous overload limit of the
converter and returning the power to the minimum power level. During the power ramp, tapchanger and
filter operation and the effect on the (simple) AC network is observed and assessed for suitability.
9.5.3.3. J – Starting and Stopping Scenarios

The start-up and shutdown scenarios are tested to ensure that methods and sequences provide
satisfactory operation as specified. The tests conform to the requirements of technical specifications and
DC scheme standard requirements.
9.5.3.4. K – Step Change in Extinction Angle

These cases check the response of the HVDC controls and the impact on the AC network to a step change
in extinction angle from steady-state.
9.6. FUNDAMENTAL FREQUENCY TOV STUDY

The objective of this study is to assess the performance of the HVDC scheme under consideration. Examined
are its controls, compensating equipment and voltage limiting control features in response to overvoltages
due to large load rejection and recovery from fault disturbances in the adjacent AC systems. This load rejection
may be a consequence of either a sudden blocking of the HVDC links due to a converter fault, or operation at
a reduced DC current following some AC system fault. The load rejection studies are conducted to determine
the resulting Fundamental Frequency Temporary (i.e. dynamic) OverVoltages (FFTOV).
Assessment must be made of the corresponding dynamic overvoltage factor, which is required to be limited
on the AC side, to determine the maximum overvoltage levels experienced at the converter station bus and
establish the performance of the HVDC scheme in respect of the FFTOVs. This is also important from the
point of view of voltage limiting surge arresters which should not be subject to long term (continuous) TOV,
as this may damage them.
Technical specifications generally give a TOV characteristic for the AC system or some limits which
overvoltages must be kept within.
FFTOV estimates generally use programs such as PSS/E, as this type of approach allows representation
of the overvoltage controllers in the wider AC system in the vicinity of a DC link. This type of
program provides flexible facilities for the representation of large systems, HVDC links and controls,
generators with their excitation systems, prime movers, and loads - both static and variable.
Stability investigations are preceded by calculations to establish system steady-states, with a
PSS/E-type load flow program being used for this purpose.
The studies require all the usual load flow and transient stability type data/models of a single-phase
representation of the 3-phase balanced network. This includes impedances, loads, generators/controls and
operating levels/features. The program incorporates all these together with HVDC link models and required
controls modeled with the necessary user-defined facilities of the transient stability programs.
Studies of partial and full-load rejection including AC system fault clearance scenarios and degraded system
conditions are used.
Appropriate cases will cover studies with and without a suitable simulation of the reactive power absorption
mode of control for voltage limitation, to determine its performance. This mode of control is a key feature
incorporated in the converter control system for AC overvoltage limiting. It forces the converter to consume
more reactive power temporarily in order to reduce any overvoltage to a set level, such as 1.3 p.u. for example.

CHAPTER
In addition to the converter dynamic control loops, Reactive Power Exchange (RPE) control and AC Voltage
Control (ACVC) within the Reactive Power Controller (RPC) are features of voltage control during the quasi
steady-state period (see section 9.4.2.5). However, if Vac control is present, then RPE and ACVC are not included
in these studies as their impact on dynamic overvoltage transients are negligible. Similarly, tapchanger controls
are neglected and it is not necessary to model Power Oscillation Damping (POD) and Frequency Control (FC)
for these studies as they feature more prominently in pure transient stability estimates.
9.6.1. System Representation

The analysis is made using fundamental frequency RMS currents and voltages, thus point-on-wave features
such as within-the-cycle valve firing and control operations are not possible in the simulation.
Filters are represented by shunt capacitor banks which supply the equivalent fundamental frequency
reactive power. Transformer saturation and surge arresters, both of which have non-linear characteristics,
are not modeled. In this respect, FFTOV studies differ from the transient overvoltage digital computer
studies performed using PSCAD/EMTDC: these involve instantaneous currents and voltages, assessing
overvoltages through a study very similar to DPS (section 9.5).
The digital computer stability (dynamic) studies utilize d-q axis models for generators, with their excitation
systems, turbines and governors. Nearby AVRs are key to the resulting FFTOVs.
The voltage dependency of the loads during machine swinging influences the magnitude of TOVs. Specified
load characteristics of MW varying with voltage magnitude V (i.e. real component is constant current) - and
MVAr varying with V2 (i.e. imaginary component is a constant admittance) - are typically employed in the
steady-state in the absence of explicit data.
In addition to the AC side busbar overvoltages on the converter station, the valve side overvoltages will also be
assessed. Due to the influence of the converter transformer operating tap position, the maximum voltages on
the valve side of the converter transformer do not necessarily correspond to those on the AC line side for load
rejection cases. When the link is blocked, the voltages on the valve-side transformer terminals are effectively
divided by √3 when seen between each valve terminal, thus they are much smaller than when the link is
deblocked and of no real consequence. The case of partial load rejection (i.e. sudden operation at reduced
DC current following some AC system fault) can, however, result in voltage stresses on the deblocked valves
and the studies will include an appropriate though fixed (i.e. non-dynamic) representation of the transformer
taps to enable the valve side (TOV) voltages to be assessed.
9.6.2. FFTOV Studies

The main objective of the FFTOV study is to determine the levels of temporary overvoltage arising from
disturbances on the AC system, with particular reference to the effect on the DC link converter busbars.
Conditions leading to the highest temporary overvoltages are those involving large load rejections and
studies normally concentrate on these, with a small number of cases of AC 3-phase faults (and consequent
interruption of DC power flow) also covered.
Line commutated converter HVDC links demand reactive power at both rectifier and inverter terminals.
The amount of reactive power demanded depends upon the firing delay angle a at the rectifier and the
extinction angle g at the inverter. The larger the a and g (the lower the DC voltage) the greater the amount
of reactive power consumed. In general it amounts to about 60% of the value of active power transferred
at each end of the link (see section 2.3). Blocking the link reduces the reactive power absorption to zero.
Thus, if one (say the largest) filter remains in service (e.g. due to a faulty filter breaker) on the respective AC
converter busbars, there is an excess of reactive power production, which causes elevated busbar voltages
around the converter station and beyond.
The situation of a temporary block by the DC link may also be considered, in which case no filters are tripped,
leading to higher TOVs.
The following load rejection situations are considered as broadly representative:
i. Sudden block of the DC link, with the AC filter(s) left in service. This represents a scenario in which the
filters remain connected at least for a short time, prior to some further action by auxiliary controls.
Thus persistent elevated voltages at and near the converter buses occur.
CHAPTER
ii. The situation in which an AC system 3-phase fault near the converter busbars causes complete
interruption of power transfer in the link, followed by clearing of the faulted line. This results in total load
rejection during the fault period, followed by system recovery aided by converter controls. AC filter(s)
remain in circuit. With filters in circuit, the largely capacitive recovery current passing through a mainly
inductive network (particularly at low SCLs) causes transient overvoltage at and near the converter buses.
These situations will be studied under pre-selected system operating states including the appropriate number
of converter filters in service for the active power transfer being considered in the DC scheme. It is expected
that the worst-case conditions will arise for the weaker AC system conditions, corresponding to the light load
network with maximum DC power flow, since the proportion of DC load rejected is higher for this condition.
9.7. TYPICAL STUDY RESULTS

This section discusses typical fault recovery characteristics of a HVDC link. There are usually specified
requirements on rate of recovery. Typically, this is between 100 ms to 300 ms after fault clearance, to
reach at least 90% of pre-fault DC power. A further requirement usually limits overshoot to around 1% to
2% of the change in power and settling time following changes in DC power order to a few hundredths of
milliseconds. Assuming the AC system itself has good recovery characteristics and the converter is working
with Short-Circuit Ratios (see section 5.2.2) above 3.0 as a minimum, achieving the specified recovery
criteria is down to the HVDC controls, which occasionally may require some tuning.
The scheme chosen to illustrate typical behavior connects two AC networks of different frequency, and also
of widely differing short-circuit levels and characteristics.
The Short-Circuit Ratio at the inverter end is around 4.0, while at the rectifier end this ratio is much greater
at 15.7. This assumes a rated DC power transfer of 1320 MW.
9.7.1. Rectifier Fault

Prior to the fault, the DC link is under current control at the rectifier and under voltage control at the inverter.
Referring to Fig. 9.7a, for a zero impedance 3-phase fault on the rectifier busbar, DC power is reduced to
zero for the duration of the fault, which is 100 ms here, as DC voltage is zero. The current is controlled to
a minimum as per the static characteristic.
Following removal of the fault, the AC voltage recovers and the DC voltage follows to the pre-fault operating
condition, as does the DC current. The line removed to clear the fault does not significantly alter system
strength at the rectifier busbar, hence the system returns quite close to the pre-fault operating condition.
The slow oscillatory nature of AC voltage during recovery is a characteristic of the AC network, even though
the SCR is quite high at 15.7.
The disparity in angle between the two software types is due to the way converter operating angle is
calculated. With the fundamental frequency program, algebraic equations are used throughout, while with
the cycle-by-cycle program actual phase-locked measurements are made which become indeterminate at
the faulted busbar, as there is no AC voltage present and no waveform from which to measure timings.
9.7.2. Inverter Fault

Prior to the fault, the DC link is under current control at the rectifier and under voltage control at the inverter.
Referring to Fig. 9.7b, for a zero impedance fault on the inverter busbar, DC power is zero for the duration of
the fault (100 ms) as DC voltage is zero. The current is controlled to a minimum on the static characteristic.
Here, DC current rises as this is an inverter end fault due to the stored energy in the transmission line/cable
and this is a notable (though expected) difference from the rectifier fault case.
Following removal of the fault, AC voltage recovers and the DC voltage follows to the pre-fault operating
condition, along with the DC current. The line removed to clear the fault for this case does alter system

CHAPTER
FundamentalÊfrequencyÊRMSÊmodel Cycle-by-cycleÊmodel
1.2 1.2
ConverterÊbusÊACÊvoltageÊ(p.u.)
1 1
0.8 Rectifier 0.8 Rectifier
Inverter Inverter
0.6 0.6
0.4 0.4
0.2 0.2
0 0
25 26 27 28 29 30 1 2 3 4 5 6
TimeÊ(s) TimeÊ(s)
1.2
1.2
1 1
0.8 0.8
DCÊvoltageÊ(p.u.)
DCÊvoltageÊ(p.u.)
0.6 0.6
0.4 0.4
0.2 VDC 0.2
0 0
-0.2 25 26 27 28 29 30
-0.2
1 2 3 4 5 6
TimeÊ(s) TimeÊ(s)
1.2 1.2
1 1
DCÊcurrentÊ(p.u.)
DCÊcurrentÊ(p.u.)
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2 IDC
IDC
0 0
25 26 27 28 29 30 1 2 3 4 5 6
TimeÊ(s) TimeÊ(s)
90 90
80 80
ControlÊangleÊ(degrees)
70 70
60 60
50 50
40 40
30 30
20 RecÊalpha 20 RecÊalpha
10 InvÊgamma 10 InvÊgamma
0 0
25 26 27 28 29 30 1 2 3 4 5 6
TimeÊ(s) TimeÊ(s)
Fig. 9.7a– Rectifier converter bus fault
9.7a
CHAPTER
FundamentalÊfrequencyÊRMSÊmodel Cycle-by-cycleÊmodel
1.2 1.2
1 1
0.8 Rectifier 0.8 Rectifier
Inverter Inverter
0.6 0.6
0.4 0.4
0.2 0.2
0 0
25 26 27 28 29 30 1 2 3 4 5
TimeÊ(s) TimeÊ(s)
1.2
1.2
1 1
0.8
DCÊvoltageÊ(p.u.) 0.8
DCÊvoltageÊ(p.u.)
0.6 0.6
0.4 0.4
0.2 VDC 0.2 VDC
0 0
-0.2 25 26 27 28 29 30
-0.2
1 2 3 4 5
TimeÊ(s) TimeÊ(s)
4 4
3.5 3.5
DCÊcurrentÊ(p.u.)
3 3
DCÊcurrentÊ(p.u.)
2.5 2.5
2 2
1.5 1.5
1 1
0.5 IDC 0.5 IDC
0 0
25 26 27 28 29 30 1 2 3 4 5
TimeÊ(s) TimeÊ(s)
90 90
80 80
70 70
60 60
50 50
40 40
30 30
20 RecÊalpha 20 RecÊalpha
10 InvÊgamma 10 InvÊgamma
0 0
25 26 27 28 29 30 1 2 3 4 5
TimeÊ(s) TimeÊ(s)
Fig. 9.7b– Inverter converter bus fault
BACK TO 9.7b
CHAPTER
strength significantly at the inverter busbar, and this is reflected in the slightly reduced AC system recovery
voltage at the inverter.
For both the above fault scenarios, the HVDC scheme has recovered as fast as the response of the AC system
voltage at the converter bus allows.
9.8. RELIABILITY, AVAILABILITY AND MAINTAINABILITY

In this section, the significance of Reliability, Availability and Maintainability (RAM) is established in a
generic context and the criteria that necessitate a focus on RAM are identified. Various HVDC scheme
design considerations, in order to achieve a specified RAM performance, are covered. The section concludes
with a descriptive analysis of different mathematical processes that can be applied to design a scheme for
a specific RAM, or to provide an estimate of scheme RAM. In addition, typical values for RAM accepted by
the industry for a line commutated HVDC scheme are included.
9.8.1. What is RAM?

RAM is employed to characterize system performance in the following terms:
➙ Readiness of the system for use
➙ System performance of the intended function
➙ Upkeep of the system in an operational state over its specified useful life
RAM requirements influence the design of a system, identification of spare requirements and definition of
the maintenance regime. Analysis of the RAM of a system involves investigation of probable random events.
RAM is thus suitably analyzed by utilizing the probability theory.
By its strict definition, Reliability is the probability of a system performing its intended function without fail
during a given period of time and is usually associated with non-repairable or one-off mission orientated
systems. In the context of repairable systems such as HVDC schemes, Reliability is measured by the average
Forced Outage Rate (FOR) - which is the failure rate per unit time (usually per year) over a definite life span, in
accordance with specified customer requirements and assuming that the system is appropriately maintained.
For a HVDC scheme, Reliability indicates the probability that the scheme is able to transmit a specified power
level.
In general, we can say that:
➙ Repairable systems are schemes where failure of a component part is rectified by replacement
with a serviceable spare. The failed component is repaired and returned to store or replaced by a
new component. There are two types of maintenance activities undertaken on a repairable system:
preventive maintenance and corrective maintenance. Corrective maintenance action is required in
the event of a failure resulting in a reduction in transmission capacity or a forced outage. Preventive
maintenance is a scheduled activity carried out in order to ensure availability of the system and its
reliable operation.
➙ Maintainability is the ability of a system to be retained in (ease of preventive maintenance), or restored
to (ease and rapidity of corrective maintenance), a specified condition when maintenance is performed
by personnel having specified skill levels, using prescribed procedures and resources.
➙ Availability is a measure of the system being in an operable state. It is the probability that the system
is operating satisfactorily at any time. Availability is dependent on its Reliability and Maintainability.
Unavailability of a system related to curtailed transmission capacity, forced outage and corrective
maintenance is identified as Forced Unavailability (FU), and that related to preventive maintenance
is termed as Scheduled Unavailability (SU). For a HVDC scheme, Availability is identified in terms of
Energy Availability (EA in percent) at specific transmission levels over a period of one year. Evaluation
of Availability requirements results in the definition of an appropriate maintenance regime and the
identification of the quantity of spare parts needed.

CHAPTER
System phase
System phase influencing RAM Examples
ClientÊrequirements
ClientÊrequirements
-ÊReliabilityÊandÊavailabilityÊguaranteesÊÊÊÊÊpowerÊtransferÊblocks
vs
-ÊOperational
-ÊEnvironmental ∝
-ÊSparesÊstrategyÊÊÊÊÊÊÊavailability
DesignÊphase ClientÊconstraints
ClientÊconstraints
-ÊOperational ∝
-ÊResourceÊskillÊlevelÊ&ÊfamiliarityÊwithÊsystemÊÊÊÊÊÊÊavailability
-ÊResources -ÊTimeÊrequiredÊtoÊassembleÊequipment
∝
&ÊskilledÊresourceÊforÊcorrectiveÊmaintenanceÊÊÊÊÊÊÊavailability
Manufacturing Component
phase manufacturingÊquality
∝
ComponentÊfailureÊrateÊÊÊÊÊÊÊsystemÊforcedÊoutageÊrate
1
∝ Reliability
Transportation
Logistics
Storage AppropriateÊtransportÊandÊstorageÊmeasures
1
OperatorÊtrainingÊand Skill ∝ MaintenanceÊtime
∝ availability
Operation skillÊlevelÊforÊmaintenance
phaseÊ AvailabilityÊofÊtoolsÊand 1
sparesÊforÊmaintenance AppropriateÊspares ∝ MaintenanceÊtime
∝ availability
Fig. 9.8a– Factors influencing RAM of a system
9.8.2. Significance of RAM

RAM is a measure of the performance of a system over time. It is a reflection of the quality of the system
and is therefore one of the key design parameters. Apart from the quality of the design and manufacture,
there are a number of environmental and operational criteria that can influence RAM. Fig.9.8a identifies
a few generic criteria that can influence RAM considering the time line from the inception of the design
activity to the operation of the system. These factors require due consideration during the system design
9.8a
and implementation phases in order to aim for the target RAM in practice. [18, 19, 20]
The achievement of a specified level of RAM for a system is significant for a number of reasons, some of which
include the following:
9.8.2.1. System Safety

Inadequate reliability and incorrect indication of the failure of safety-critical systems can lead to damage to
property and life. System safety is particularly critical to electrical systems such as HVDC schemes. The result
of the appropriate design of a scheme, with due consideration to RAM, must be a safe and secure system.
9.8.2.2. Lifetime Cost

Poor system-design focus on Reliability and Availability can result in system failures in excess of the
acceptable levels, thus incurring substantial operational and repair costs in terms of material, human
resources and loss of productivity. In the case of electricity infrastructure such as HVDC schemes, poor
RAM design can result in substantial additional costs to the consumers of electricity due to extended loss
of electricity supply. The cost accrued as a result of poor RAM design contributes towards the lifetime cost
of the system. During the conceptual and design phase it is thus necessary to evaluate and establish the

CHAPTER
RAM requirements in order to mitigate the lifetime cost of the system. The system should be designed to
be more reliable and maintainable, with a balanced view of RAM included in the initial capital cost.
9.8.2.3. Logistics Footprint

The operation of any system requires the arrangement of a minimum of logistics of operators, maintenance
crew and spares. Technological accomplishments have today resulted in unmanned operation of electrical
substations, including HVDC systems. Improved RAM reduces the size of the logistics footprint related to
the number of required spares, maintenance personnel, and support equipment, as well as the workforce
size needed to successfully accomplish a task.
9.8.3. Designing for RAM

Section 9.8.2 highlights the significance of RAM. The desirable RAM performance is achievable by integrating
the RAM requirements within the design process. The term design here encompasses:
➙ The equipment - system, subsystem, components
➙ The processes - design, manufacture, quality, logistics etc
Various measures to incorporate RAM into the design include, but are not limited to, the following (see
Fig. 9.8b):
➙ A comprehensive RAM-orientated program for designing and manufacturing components and
subsystems that comprise the scheme and review of the design
➙ Development of a detailed system model representing the components and subsystems, inclusive of
the associated RAM related parameters
➙ Modularize the system in order to simplify maintenance, fault diagnostics and component or
subsystem repair and replacement
➙ Determination of critical failure modes and possible degraded operation of the scheme to redress the
design
RAMÊorientedÊprogramÊfor
designÊandÊmanufactureÊof
componentsÊand
subsystems
DevelopÊconceptual ComponentÊlevelÊRAMÊdataÊfrom
systemÊmodelÊwith -ÊKnownÊdata
appropriateÊlevelÊofÊdetails Ê DataÊfromÊmanufacturers
(componentÊand Ê DataÊfromÊpastÊexperience
subsystemÊrepresentation) Ê PublishedÊdata
TargetÊRAMÊrequirements
forÊRAMÊanalysis
Review
IdentifyÊcriticalÊfailureÊmode -ÊAssumedÊdata
andÊmodesÊofÊdegraded Ê DataÊfromÊmanufacturers
operationÊÐÊredressÊin Ê ofÊsimilarÊcomponents
design
DesignÊfeatures
-ÊModularity
-ÊRedundancy
-ÊFaultÊdetectionÊsystems
-ÊFaultÊdiagnostics
Fig. 9.8b– Designing for RAM

CHAPTER
➙ Integrated monitoring and fault detection schemes within the system redundancy
➙ Appropriate design of various processes with focus on quality and RAM
➙ Incorporation of learning from past experience to achieve a RAM oriented design
Base data for the estimation of RAM for a HVDC scheme is available from a number of sources including:
equipment manufacturers, published industry standards and industry experience [21-28]. It is also vital to
monitor HVDC schemes in commercial operation for valuable feedback on RAM.
The following sections describe modeling and calculation techniques that can be used in defining the RAM
of a system.
9.8.4. Modeling Technique – Dependency Network

It is not realistic to analyze a complex engineering system in a single step. Usually the system has to be
considered as a network of smaller sub-systems that are first evaluated separately, followed by their
grouped effect in the system.
In this section a hierarchical technique for system analysis using block diagrams to represent sub-systems
within a dependency network that represents the complete system is described. The dependency network
defines the inter-relationships between the component blocks in the system in terms of series or parallel
dependency. The dependency network also defines the power capacity of each block, its contribution to the
100% 50% 50%
100% 100%
50%
100% 50%
50%
SwitchedÊredundantÊplant On-lineÊredundantÊplant
Fig. 9.8c– Dependency network – an example
system power capacity and the redundancy available within the network. The dependency network does
not necessarily relate directly to any electrical or physical representation of the system being analyzed. A
simple dependency network is shown in Fig. 9.8c.
The blocks within the network may be considered to represent either specific items of equipment (basic
elements) or sub-systems (network of basic elements).
9.8c
A basic element is defined as the item of equipment that is either working or not working: for example, a
battery charger may be considered as a basic element in a scheme, even though it is a complex system in its
own right. This implies that in the event of a battery charger failure, for whatever reason, the complete charger
is considered as failed until repairs have been made within a defined time.
The parameters that define basic elements differ from the ones that describe systems or sub-systems by
including aspects such as spares, replacement times, resource limitations and maintenance requirements.
A sub-system made up of identical basic elements is defined in terms of whether the elements have series
or parallel dependency within the sub-system.

CHAPTER
The following terms are used to describe the elements and sub-systems in the mathematical analysis:
➙ The number of identical elements in the sub-system (Ne)
➙ The number of redundant elements in the sub-system (Nr)
➙ Average failure rate of the element (le)
➙ The average replacement time of each element (Tr)
➙ The number of serviceable on-site spares (Ns)
➙ The on-site resource limitation on the number of simultaneous replacements that can occur (ERLimit)
➙ Whether the failed elements can be replaced with the sub-system on-line, if so there is an average
delay time before replacement (Td)
➙ The maintenance interval, after which time it is assumed full redundancy is restored (Ti)
➙ The average off-site repair time or lead-time for new (Tw)
➙ The supply limitation of new or repaired elements (EWLimit)
➙ The power rating of each element (Pe)
9.8.5. Calculation Techniques

As discussed in the preceding sections, the probability theory (measure of chance) is used to analyze the
RAM of a system. In this section, two mathematical theories are briefly described:
➙ Poisson distribution – simplistic calculation technique for Reliability
➙ Continuous Markov process – comprehensive technique to analyze Reliability and Availability of the
scheme
9.8.5.1. Poisson Distribution [16]

Failure rate or hazard rate l(t) is represented as a ratio of the number of failures per unit time and the
number of elements exposed to failure. The life of a component/equipment can be divided into three
periods:
a. Burn-in / debug period – where initially the hazard rate could be high until the system settles down into
steady operation
b. Normal operating period or useful life – where the hazard rate is substantially constant. This can be
achieved, for example, for mechanical systems by undertaking preventative maintenance
c. Wear-out period – where at the end of useful life, the hazard rate increases rapidly
At a fixed hazard rate, which can be assumed to be represented by the average hazard rate during normal/
useful life, the Poisson distribution represents the probability of an isolated event occurring a specified number
of times in a given interval of time.
Poisson’s expression for n failures during a time interval t is as follows:
(λ t ) n e− λ t
Pn (t ) = [Eqn. 9.8a]
n!
Where:
n = the number of occurrences of an event
e = base of natural logarithm
e = 2.71828
9.8.5.2. Continuous Time Markov Chain

RAM is normally concerned with a system that is discrete in space, that is, it can exist in one of a number
of discrete and identifiable states, and continuous in time. The system therefore exists continuously in one
of the states until a transition takes it discretely into another state.
Markov modeling essentially involves considering that the system under investigation may exist in a number
of well-defined, mutually exclusive states. A state is a particular condition of the system that represents its
capability to perform its intended function at full or partial capacity. The simplest system can exist in two
states - working or not working. The number of states increases for systems that can function at intermediate
power levels or have redundancy included. When redundancy is included, more than one state can exist
where the system has full capacity. The number of states also increases when spares are taken into account.

CHAPTER
Markov techniques require that: future states of the system are not dependent on any previous states; that
all possible states of the system are mutually exclusive and that the transition between states is constant
and random over the life of the scheme. The model considers every likely state in which the system can exist.
While suitable for systems that exist in a small number of states, it is impractical to apply the model directly
to large systems where a large number of states are possible.
Element reduction generally involves simple systems that may be dealt with by direct Markov techniques.
Sub-system reduction is not practical by the direct approach and is achieved instead by applying a set of
approximate equations that are derived from the Markov technique. This approximate model assumes that
each sub-system can only exist in two possible states, working or not working, and that the transition between
states is constant and random. This analysis is adequate for failure events, although care must be taken in
defining the failure criteria. To predict energy availability realistically, a second technique called capacity
probability analysis is used.
Both techniques are widely used in the electrical supply industry throughout the world and provide acceptable
results, providing a representative dependency network and valid base data are used. [16, 17]
Appendix A9.3 includes a few solved examples of RAM calculations, network reductions and the application
of the Markov chain technique to establish the reliability of a system.
9.8.6. Typical RAM Performance of LCC HVDC Schemes

The RAM performance target of a typical LCC HVDC scheme is as follows [27]:
Monopole arrangement
Forced outage rate 3 to 4 per year
Forced energy unavailability up to 0.5%
Scheduled energy unavailability up to 1%
Energy availability up to 98.5%
Bipole arrangement
Forced outage rate
100% power* 6 to 8 per year
50% power** 0.05 per year
Forced energy unavailability
100% power* up to 1%
50% power* up to 0.02%
Scheduled energy unavailability
100% power* less than 2%
50% power** less than 0.1%
Energy availability
100% power* up to 97%
50% power** up to 99.9%
Table 9.8a– Typical RAM performance requirements
Note:
* 100% power - both poles are available for power transfer
** 50% power - at least one pole is available for power transfer

CHAPTER
BIBLIOGRAPHY
[1] C. D. Barker, “HVDC for Beginners and Beyond”.
[2] S. B. Tennakoon, M. L. Woodhouse, “Calculation of Valve Damping Circuit Losses in 12-Pulse HVDC
Convertors”, IEEE Exposition & Conference, Dallas, Texas, 1991.
[3] “Determination of power losses in high voltage direct current (HVDC) converter stations”, IEC 61803.
[4] “Transformers for HVDC applications”, IEC/IEEE 60076-57-129.
[5] Katsuhigo Ogata, “Modern Control Engineering”, 3rd edition, ISBN 0-13-261389-1, chapters 7, 8 and 9.
[6] “Flexible AC Transmission Systems (FACTS)”, Y. H. Song & Allan T Johns (editors).
[7] “Guide for planning DC links at AC system locations having low short-circuit capacities (part 1): AC/DC
Interaction Phenomena”, IEEE WG 15.05.05, CIGRÉ WG 14.07.
[8] S. Lefebvre, A. M. Goole, J. Reeve, L. Pilotto, N. Martins, S. Bhattacharya, “Working Group on Dynamic
Performance & Modeling of DC Systems & Power Electronics for Transmission Systems: Report on Test
Systems for AC/DC Interactions”.
[9] C. Hatziadoniu, G. D. Galanos, “Interactions between the AC voltages and DC current in weak AC/
DC interconnections”.
[10] L. Bottacchiari, “Suppression of Third Harmonics in AC Systems by Second Harmonic Modulation
of Firing Angle in HVDC Schemes”, M.Sc Thesis, University of Bologna, Faculty of Engineering, Dept. of
Electronics & IT Systems.
[11] Y. Hu, P. G. McLaren, A. M. Goole, D. J. Fedirchuk, A. Castro, “Self Excitation Operating Constraints for
Generators Connected to DC Links”.
[12] C. S. Indulkar, B. Viswanathan, “Effect of Series Compensation on the Voltage Instability of EHV Long
Lines”.
[13] P. M. Anderson, B. L. Agarwal, J. E. Van Ness, “Sub-synchronous Resonance in Power Systems”.
[14] Chanki kim, “HVDC Transmission: Power Conversions Applications in Power Systems”, John Wiley and
Sons 2009, ISBN 0470822953:0470822951.
[15] T. Smed, G. Andersson, “Utilising HVDC to Damp Power Oscillations”.
[16] R. Billinton, R. N. Allan, “Reliability Evaluation of Engineering Systems: Concepts and Techniques”,
London, Pitman Publishing Ltd., 1983.
[17] R. Billinton, R. N. Allan, “Reliability Evaluation of Power Systems”, New York, Plenum Publishing Corp.,
1984.
[18] “Guide to Achieving Reliability, Availability and Maintainability”, Department of Defence (USA), August
2005.
[19] “Planning, Developing and Managing an Effective Reliability and Maintainability (R&M) Program”,
National Aeronautics and Space Administration, NASA-STD-8729.1, December 1998.
[20] “Reliability, Availability, Maintainability, and Cost Rationale Report Manual”, Department of Defense
(USA), June 2009.
[21] “Handbook of Reliability Data”, HRD4, January 1987, British Telecom.
[22] “Forced Outage Performance of Transmission Equipment for the period Jan.-1-1989 to Dec.-31-1993”,
Canadian Electrical Association.
[23] “Forced Outage Performance of Transmission Equipment for the period Jan.-1-1995 to Dec.-31-1999”,
Canadian Electrical Association.
[24] “Prediction of electronic component failure rates using the methods described by the MIL-HDBK-217”.
[25] “IEEE Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems”, IEEE
Std 493-1980.
[26] IEEE Std 500-1984 reaffirmed in 1991, Reliability Data.

CHAPTER
[27] “A Survey of the Reliability of HVDC Systems throughout the World”, CIGRÉ SC14 reports covering the
period 1991-2004.
[28] “An International Survey on Failure in Large Power Transformers in Service”, CIGRÉ WG12-05, August
1982.
[29] Sci-Tech Encyclopedia: http://www.answers.com
[30] H. A. Calderbank, J. L. Haddock, G. Mazur, “Availability And Reliability Considerations For Nelson River
Bipole 1”, CIGRÉ 14-304 30/08/1992.
[31] W. P. Williams, S. G. Mudge, “Reliability Assessment For Industrial Power Networks”, Conference:
Reliability in Power Systems, IEE, London, September, 1983, p. 107-111, 1983.
[32] R. P. Burgess, K. Rout, W. P. Williams, “Strategic Planning For The Analysis And Assessment Of Power
Systems Reliability Performance”, CIGRÉ Paper 37/38/39-01, 1986.
[33] K. J. Ralls, “The Reliability Of Transmission And Distribution Equipment”, IEE Power Engineering Journal,
June 1995, p. 109.
[34] C. D. Barker, A. Sykes, “Designing HVDC Transmission Schemes for Defined Availability”.

CHAPTER
CHAPTER
10| PLANNING A HVDC CONVERTER SCHEME
TOC 532 | DC
10 PLANNING A
HVDC CONVERTER
SCHEME
Installing a HVDC scheme is an expensive capital equipment project
and must therefore be well planned.
You may be surprised as to how much work is involved and how many
alternatives must be considered over a long period of time before
even a request for quotation is issued to build the desired project. In
this chapter, you will learn about all of the techno-economic studies
that are usually performed to ensure that the scheme, once it is
installed, is the optimum solution for the requirements and achieves
its desired objectives.
You will be introduced to the vast array of formal studies that are
carried out to assess the proposed scheme against environmental,
economic, operational performance and functional requirements.
Topics such as rights-of-way, visual impacts and land acquisition
are discussed, along with commercial, legal and operational lifetime
considerations.
You may also be interested to read about the different options for asset
ownership, operation and maintenance of the equipment once it is built.
At the end of the chapter is a detailed questionnaire used to define a
HVDC project.

Chapter contents
10. P
LANNING A HVDC
CONVERTER SCHEME.......................................... 532
10.1. PLANNING A HVDC PROJECT............................................... 535

10.1.1. Feasibility and Definition Studies....................................... 535
10.1.2. Project Timeline.......................................................................... 535
10.1.3. AC Network Studies and Other
Engineering Analyses............................................................... 536
10.1.3.1. Static Analysis............................................................................. 537
10.1.3.2. Dynamic Analysis...................................................................... 537
10.1.4. Environmental Studies............................................................ 538
10.1.5. Land Acquisition........................................................................ 538
10.1.6. Commercial, Financial, Legal, Schedule
and Operational Lifetime Considerations...................... 540
10.2. OPERATIONAL EXPERIENCE................................................. 544

10.2.1. Nelson River HVDC .................................................................. 544
10.2.2. Rivera Converter Station ....................................................... 550
10.3. STUDIES ASSOCIATED WITH

CONVERTER STATION DESIGN............................................ 551
10.3.1. Typical Contract Reports........................................................ 551
10.4. INFORMATION REQUIRED FOR

THE DESIGN OF A HVDC SCHEME..................................... 557
10.4.1. Questionnaire –
Data Requirements for a HVDC Scheme......................... 557
BIBLIOGRAPHY ............................................................................................ 561

10.1. PLANNING A HVDC PROJECT
The development of a HVDC project is a complex series of processes encompassing technical, commercial,
financial and legal activities involving many stakeholder organizations ranging from banks or other
financiers, government agencies, public bodies, private individuals, consultants, utilities, private developers
and manufacturers, architectural, engineering, construction, installation and other contractors.
Generally there is a need to get the project in service by a specific date, usually in order to meet either a
commercial commitment or electrical power network configuration (e.g. for reliability). The success of the
project in achieving this target date is then dependent on detailed planning and management of the many
individual sequential and parallel processes.
In this section we will discuss these various different processes, noting that the unique and special nature of
HVDC projects means that not all of these issues will be relevant to all projects.
10.1.1. Feasibility and Definition Studies

The initial part of the process is to carry out a range of studies, beginning with feasibility, then moving on
to definition. Generally, these studies are made up of the following broad stages:
➙ Stage 1
Identification of the study scope: includes assumptions, constraints, scenarios and alternatives. This is
a key stage, as it is here that there needs to be an agreement between the end-users and the groups
performing the studies.
➙ Stage 2
Initial feasibility analysis: having identified the different alternatives, there is a need to perform
several studies to determine the best alternative(s), taking into consideration financial, technical and
environmental issues.
The conclusion of the initial feasibility analysis may indicate that the project is not feasible: for example
it is too costly, not technically feasible or not environmentally acceptable. Alternatively, if the analysis
has determined that the project is practical, then the studies may go to the next level.
➙ Stage 3
Decision on the preferred alternative(s): in order to proceed to the detailed studies stage there is a
need to limit the scope to a manageable set of studies, which may be based on a single solution, or it
may include alternatives.
➙ Stage 4
Detailed studies to support the functional specification: includes general, environmental, geological,
meteorological and electrical project site characteristics. There will be a lot of information which needs
to be gathered which is not HVDC-specific, as well as the specific HVDC parameters, performance and
functionality which must be identified.
➙ Stage 5
Request for Quotation (RFQ) procurement process: the specification is packaged together with the
commercial and contractual documentation, which is issued to qualified manufacturers as an invitation
to bid.
10.1.2. Project Timeline

The scale of a HVDC project has a strong influence on its duration. Larger projects (bipole transmission
links above 1000 MW) can take more than 3 years until the actual project construction phase. The darker
brown durations in Fig. 10.1a and indicate the typical minimum project duration, while the lighter shades
of brown indicate the extended duration to a typical maximum range.
It should be noted that the project development phase, indicated as a single line in the timeline shown in
Fig. 10.1a, is actually made up of many sub-activities (see sections 10.1.4 to 10.1.6) necessary to bring the
project to a state where enough is known to create the detailed specification and to invite bids from the
specialist HVDC vendors.

CHAPTER
Years
Project development
Specification writing
Tendering
Evaluation
Negociation
Contract award
Construction phase
System design
Detailed design
Studies / Analysis
Manufacturing
Simulation / Testing
Shipping and installation
Commissioning
Fig. 10.1a – Indicative timeline for the complete cycle of an HVDC project
Like all complex studies, the planning and proper identification of a good road map is essential to ensuring
that10.1a
the key issues have been addressed to the full satisfaction of the stakeholders. Good quality and detailed
planning is vital to a HVDC project.
10.1.3. AC Network Studies and Other Engineering Analyses

An analysis of the AC network or networks is essential to determine the best ratings, the best connection
points for the link and the operational limits of the link. A load flow analysis and other detailed studies are
normally carried out at the initial stages of the project to begin the process of determining what the HVDC
interconnection will be required to do.
These studies will determine the functionality required, including:
➙ Ratings and operating range
– Configuration: bipole, monopole, cable, line, etc.
– Operating range: minimum and maximum MW
– Start-up/shut-down conditions
– MW ratings, overload
– Reactive power limits
– Power direction capabilities
➙ Operational performance
– Availability, maintenance and spares
– Losses
– Harmonics
– Audible noise
– Electrical noise, interference
– Earthquake ratings and continuity of operation
➙ Characteristics of the AC network(s) at the converter terminals
– Range of variation of voltage and frequency
– Positive, negative and zero sequence characteristics of AC voltage
– Minimum and maximum short circuit levels
– Harmonic impedances
– Resonance inception risk
➙ Identification of potential for interactions with other plants in the network, including any other HVDC
or power electronics facilities which may be in the area.
CHAPTER
10.1.3.1. Static Analysis
There are basically two types of studies carried out:
1. Determine the system strength of the existing network at the point of connection. This
is a study to determine the Equivalent Short Circuit Level (ESCL). The study helps to
determine if additional re-enforcements with new lines, generators, or FACTS devices
(Static VAr Compensators, STATCOMs, filter banks, series compensations, synchronous
compensators, etc.) are going to be needed.
2. Load flow analysis: to determine the limits of exchange of reactive power at the point of connection
- and if the overload limits of any of the neighboring busbars will be exceeded.
10.1.3.2. Dynamic Analysis

The objective of this analysis is to ensure network transient stability, or that the interconnected AC
networks will not suffer from issues such as:
➙ Voltage and frequency instabilities
➙ Voltage collapse
➙ Resonances leading to possible damage to generators
➙ Unacceptable operating conditions following a critical event, triggering contingency operations such
as load-shedding, islanding, etc.
These considerations amongst others would form the basis of the technical specifications for the HVDC project,
with additional supplementary information to be supplied where an owner has specific standards or requirements.
Specific functional requirements are often placed on major items of equipment, such as thyristor converters,
converter transformers, DC smoothing reactors, AC and DC filters, control equipment, control system structure
and functionality, local vs. remote control, etc. The specifications normally include sections on each of these items.
Many purchasers of HVDC links are transmission utilities with many existing installations and established
preferences or standards for specific equipment. This allows them to make their maintenance and spares
holdings activities more efficient. There are many areas of the HVDC plant which contain equipment similar
to AC substations, including circuit breakers and other switchgear, protection and monitoring equipment, etc.
Therefore, it is common for the requirements documentation for HVDC projects to incorporate a number of
standard equipment specifications normally used by the utility. This not only relates to equipment, but also to
building types and staff facilities.
CIGRE (International Council on Large Electric Systems), EPRI (Electric Power Research Institute in the USA),
IEEE and IEC issue guideline documents and current standards which may be adopted for specifying and
designing HVDC installations. The following is a non-exhaustive list of typical documents made available by
the respective organizations:
IEC 60700-1 Thyristor valves for High Voltage Direct Current (HVDC) power transmission
- Part 1: Electrical testing
IEC/TR 60919-1 Performance of Line Commutated Converter High Voltage DC (HVDC) Systems - Part
1: Steady-state conditions
2: Faults and switching
3: Dynamic conditions
IEEE 1124-2003 IEEE Guide for the Analysis and Definition of DC-Side Harmonic Performance of
HVDC Transmission Systems
IEEE 1240-2000 IEEE Guide for the Evaluation of the Reliability of HVDC Converter Stations
IEEE 1378-1997 IEEE Guide for Commissioning High Voltage Direct Current (HVDC) Converter Stations
and Associated Transmission Systems
CIGRE 388 Impacts of HVDC Lines on the Economics of HVDC Projects
CIGRE 370 Integration of Large Scale Wind Generation Using HVDC and Power Electronics
CIGRE 346 Protocols for Reporting the Operational Performance of HVDC Transmission Systems

CHAPTER
CIGRE 215 HVDC Converter Stations for Voltages above +/-600 kV
CIGRE 186 Economic Assessment of HVDC Links
CIGRE 130 Operational Guidelines and Monitoring of HVDC Systems
CIGRE 127 Guide for Upgrading Transmission Systems with HVDC Transmission
CIGRE 97 System Tests for HVDC Installations
EPRI EPRI HVDC Reference Book
Further information is available on their websites:
www.cigre.org www.iec.ch
www.epri.com www.ieee.org
10.1.4. Environmental Studies

The objective of this set of studies is to provide an evaluation of the impact of the different alternatives on
the environment. The environmental studies are normally often a legal requirement in most countries, in
order to achieve the permits for construction.
The information used for such a study is focused on:
➙ The required corridor for the HVDC overhead lines
➙ Layout of the HVDC cables (sea and underground)
➙ The possible locations of the electrodes (sea or ground)
➙ The location of the converter station(s)
➙ Conditions in the converter station area including: altitude, temperature range, pollution level,
contamination level, keraunic level, seismic activity, etc.
The proposed routes will be designed taking into account:
➙ The constraints of government legislation
➙ The existing topography and geological data
➙ Other existing and planned cable systems
➙ Other third party use of both the seas, the land and seabed
➙ Land use, soil/earth/rock removal restrictions,
➙ Populated areas, distance from urban or densely population centers
➙ Protected areas, wetlands, natural/national parks, marine habitats, etc.
➙ Water bodies: crossing of waterways may require cable or extended span towers.
➙ Existing infrastructure: re-using or extending existing pipeline/rail/power line, corridors, crossing large
roadways or railways.
A complete HVDC project consists of either a back-to-back converter station (with one or more AC lines on
either side of the station), or a pair of HVDC transmission converter stations (each with one or more AC lines
on the AC side and DC lines or cables on the DC side). All aspects of the proposed project must be taken into
account in the environmental assessment, including the impact of converter stations and the entire route of
the AC and DC lines or cables.
10.1.5. Land Acquisition

Once the best electrical points in the networks are identified, the next step is to find land on which to build
the converter stations (or station in the case of a back-to-back installation).
Land must be purchased or leased, including permits to build according to local regulations. In parallel with
this is the need for similar acquisition of land and permits for the construction of the overhead line, or cable.
In the case of the overhead line this is often the most contentious part of the process since it involves many
different stakeholders who may be directly or indirectly affected by the line, including landowners and local
authorities. The permitting process for overhead lines in some cases can take a number of years to complete.
The choice of circuit configuration will have an impact on the land requirements for not only the converter
stations and the line Right-of-Way (ROW) or wayleave, but an electrode station site may also be required at
each end of the link. An early resolution of issues such as the selection of cable, overhead line, a combination
of both, the use of sea return, ground return or metallic return is essential in the planning of the link.

CHAPTER
10.1.5.1. Environmental Issues
• Overhead lines, right of way, visual impact
Long distance transmission, whether it is by HVDC or AC, uses overhead lines, due to the relatively high cost
of underground cables for such distances.
There is inevitably a line Right-of-Way (ROW) or transmission corridor width and visual impact related to the
installation of new overhead towers and lines, whether it is HVDC or AC, but it is an established fact that the
dimensions and visual impact of HVDC towers are significantly lower than that of AC at equivalent power
transmission ratings. In other words, assuming that a certain ROW/visual impact is acceptable, then it is
possible to get much more power with that same ROW/visual impact using HVDC as opposed to using AC.
To look at this in more detail, please refer to Fig 10.1b. Consider the specific example of a line (AC or DC)
requiring the capacity to transfer 1850 MW of power. As Fig 10.1b shows, the tower profile for AC, using a
3-phase, double-circuit configuration is much larger than the DC-rated equivalent.
Taking this a step further, if the same AC tower profile is used and translated into a DC rating, it is
then possible to configure the tower to carry three HVDC bipolar circuits, which is able to carry up to three
times the power rating compared to the AC tower.
The selection of either AC or DC as the transmission medium is clearly going to be heavily influenced by
investment costs and the simple ratio [cost of transmission line (per km) to MVA transported] will give an
indication of viability.
The towers and conductors are often the most significant components of the overall project costs. In relatively
low rating transmission lines a single conductor is used in the circuit, but for higher rated power lines the use
of up to four conductors in parallel is common. Therefore the cost of the complete conductor circuit in both AC
and DC configurations must take into account the towers, the characteristics of the complete line, conductor
spacers, string insulators, the weight of the towers, etc.
There are many different tower profiles available, ranging from a single vertical pole with guy ropes, to
four-legged self-supporting towers (see section 8.1.1). It is necessary to carry out a full techno-economic
cost-benefit analysis of the desired power transfer rating and the alternative tower and line configurations
(both HVDC and AC), the consequent costs, and any other related factors which may have an influence on
the decision for the specific project.
• Electrodes (sea or land)
1850 MW 1850 MVA 5550 MW
DC AC DC
Fig. 10.1b – Comparison between the power ratings of AC and DC overhead line towers
BACK TO 10.1b DC Transmission Systems: Line Commutated Converters | 539

CHAPTER
A HVDC link requires a ground reference in order to operate effectively. Whether the HVDC link is configured
as a monopole or as a bipole, the system will require a ground electrode station unless an additional conductor
(i.e. cable or overhead line) is incorporated as a return path for the DC current.
There are several types of electrode available and the type of electrode used and the consequential impact
on the environment will vary. The main technical issues which arise due to continuous ground current use are:
a) Corrosive effects on buried metallic structures such as pipelines, foundations for buildings and other
structures which contain reinforced steel and the like: the impact here is obviously more of a concern
over the longer term, and is dependent on many factors including duration of usage, levels of DC current,
proximity of metallic structures and ground resistivity. Basically, the higher the current in the electrode,
the longer the usage, the lower the resistivity, and the closer the structures are to the electrode site, the
more impact there will be.
b) Ground currents in the converter station and nearby AC substation equipment: the DC current may be
relatively low (only the spill current while the HVDC link operates in balanced bipole mode which may
be in the order of only a few amps), or high under continuous duty in monopolar operation (where the
entire pole DC current rating is passing through the ground). In either case there is the possibility of
this DC current finding its way into the converter station or into nearby distribution or transmission
substation equipment (such as through grounded neutral connections in transformers). The objective
here is to ensure that the electrode site is remote from any such installations, as described in section
8.3.
There are over 200 working examples of HVDC schemes around the world, many of which either use a ground
electrode or sea electrode, and there is a great deal of operational experience on these systems demonstrating
no significant adverse impact.
Specifically in the case of sea electrodes there have been concerns over the impact on marine and botanical
life, though many studies have been carried out which have not indicated any definite or significant effects.
However, there are many local, national or regional regulations now in place which, as a precaution, prohibit
the permanent, continuous or long-term use of electrodes.
It is clear that there are local issues relating to the planning of specific HVDC projects which must be
adequately studied and the appropriate mitigation plans must be put in place in order to ensure that any
potential impact on the environment during the lifetime of the HVDC link is minimized.
Please refer to section 8.3 on Electrodes, for further details.
10.1.5.2. Other Issues

There are other environmental effects from the converter station and the overhead lines which may need
to be taken into account according to the local requirements. These include electromagnetic fields, DC
corona and audible noise, impacts on wildlife habitats, nature reserves and protected wetlands. These
apply equally to AC substations and lines.
It is not possible to say exactly what impacts must be studied for each HVDC project, since the rules of
engagement vary widely around the world. It is clear that there are many regulations in force, ranging from
those at international levels, down to national, regional and local authorities. These must all be taken into
account, according to the proposed converter station locations and the overhead line route, which may go
through many different jurisdictions.
10.1.6. Commercial, Financial, Legal, Schedule

and Operational Lifetime Considerations
Legal, commercial and financial processes and decisions all form a necessary part of the development
process. In the simplest cases, a single traditional transmission utility specifies, purchases, owns, operates
and maintains the plant. However, it is also increasingly common for multiple organizations to be involved
in a project at each stage. Therefore, in the development of a HVDC project there are many questions which
must be answered, including:

CHAPTER
• Who will ultimately own the asset?
The party who pays for the construction of the plant may be the ultimate owner, but it is equally possible for
someone to build the plant as a business and then sell it or lease it to another party to own and operate as
a separate business.
• Who will be operating the plant?
It could be the party who purchased the project, but not necessarily. It is sometimes the case that
an independent company is responsible for operating the plant on a day-to-day basis. The operational
responsibilities include supplying regular daily staff to manage the plant, including operators, administration,
and sometimes the maintenance activities.
• Who will be controlling / scheduling the power through the link?
In many cases the actual power transfer scheduling is carried out by someone who controls the overall
AC power network usually located in a regional or national dispatch control center (via telephone or other
manual communications, or as in most modern projects, via automated direct control).
• Who will maintain the plant?
The complexities of power electronics systems normally dictate the need for specialist support when
problems arise. The converter station consists of many items of equipment which may be called
conventional, such as transformers, switchgear, AC protection (lines, busbars, filters, etc.), AC and DC
auxiliary power supplies, and building services (including HVAC, lighting, etc.). The key areas which are
normally deemed special are the thyristor valves and the HVDC Control and Protection system. These
normally require the attention of specially trained staff able to diagnose a range of problems, which are
not part of a normal AC substation. Some HVDC utilities retain specialist staff to fulfill this role, but it is
increasingly the case that these specialist services are provided by third parties, or even the OEM or HVDC
vendor. Maintenance activities should include provisions for both scheduled maintenance (monthly, annual,
bi-annual) according to manufacturer’s instructions and breakdown support (on-site 24/7, available locally
or regionally, factory-based with telephone support or remote dial-in diagnostics, etc.).
• What is the revenue/income stream from the operation of the scheme?
The energy being transferred through the link is normally taken from one or multiple sources/generators
at the sending end and charged to Load-Serving Entities (LSE) at the receiving end. This can be one
utility in some areas of the world, where the operations for generation, transmission and distribution are
integrated within the same utility company. However, it may also be individual organizations, and they
may be in different regions, or even countries, since the use of HVDC in linking across borders is common.
Joint ownership of the complete systems, or individual ownership of the two ends of the systems are also
common.
• Who is going to pay for the project at the construction phase?
There is a range of possibilities including:
➙ Single utility entirely self-financed from internal funding
➙ Partnerships between utilities
➙ Private investment banks, or venture capitalists
➙ State governments
➙ International investment banks (World Bank, Asia Development Bank, etc.).
The construction costs for the converter stations (or station in the case of a back-to-back HVDC link) can
be from tens of millions of dollars at the extreme lower end of the scale, up to many hundreds of millions of
dollars for the higher HVDC link ratings. Project financing is discussed further in section 10.1.6.2.
•
W hat procurement model will be used for the contract (i.e. turnkey, design+supply+
supervision, equipment only)?
The selection of the interface points in a HVDC project is an important consideration which can have a
significant impact on the complexity and the successful implementation of an HVDC project.
Consider the scope of a typical complete converter station as consisting of two zones: a HVDC area and a
more conventional AC substation area. The outgoing connections on the AC side and DC side are treated
separately (see the Single Line Diagram in Fig. 3.2j):

CHAPTER
➙ AC switchyard - all the power and secondary equipment which would normally form part of a
conventional AC substation, up to the feeders for the HVDC-specific DC converter area. This includes
circuit breakers, disconnectors, earthing/grounding switches, surge arresters, CTs, VTs, busbars,
connectors, insulators, fittings, support structures, foundations, earth/ground grid, protection,
power and control cabling. The ground grid under the complete converter station, including under
the DC converter area, may be incorporated into this scope, subject to specific considerations for DC
applications. The outgoing AC connections (AC lines or busbars to other substations) are normally
either part of the utility scope or part of the AC substation scope. The protection panels for the AC
substation, lines, etc. would be installed within the service building in the DC converter area.
➙ DC converter area - including converter transformers, thyristor valves, AC harmonic filter
components (resistors capacitors, reactors, CTs, surge arresters, etc.), valve halls and service
buildings, HVDC control and protection, DCCTs and resistive voltage dividers, wall bushings, DC
smoothing reactors, DC surge arresters, DC switchgear (disconnects, earth/ground switches, Neutral
Bus Switch (NBS), Neutral Bus Grounding Switch (NBGS), Earth Return Transfer Breaker (ERTB) and
Metallic Return Transfer Breaker (MRTB), as detailed in section 6.8 of this book). In addition, the DC
filters on the HVDC and neutral busbars are included. The switchgear which feeds the converters
and the AC harmonic filters must be specially rated and be capable of direct control from the HVDC
control and protection systems, therefore they are normally part of the converter area scope. AC and
DC auxiliary supplies are also normally part of the DC converter scope. The same auxiliary supplies
for the DC converter area would also normally feed the AC substation areas.
➙ DC connection:
– Back-to-back - in this case, the DC connection is very short and indoors.
– Cable schemes - in the case of a HVDC cable link, the cable is normally brought into the converter
station, and the cable supplier or their appointed contractor would provide a cable sealing end
termination on which the HVDC connection would be made. This may be indoors or outdoors as
appropriate to the local environment. The supply of the HVDC cable would not normally be part of the
DC converter station scope of supply.
– Overhead line transmission schemes – similar to cables, in the case of an overhead HVDC
transmission line system the DC converter station supply limit would normally be at the dropper or
connection at the dead-end tower or gantry at the perimeter of the converter station. The supply of
the HVDC cable in most instances forms part of a separate contract.
Historically the HVDC vendor was usually expected to take on the full turnkey scope of supply, taking responsibility
for not only the HVDC project delivery during the construction phase but also, and potentially more significantly,
the performance guarantees which must be met once the link has entered commercial service.
There are situations where either a utility is well experienced in HVDC, or the utility chooses to enlist the
services of a preferred Engineer, Procure, Construct (EPC) contractor with HVDC experience (or individual
engineer with other procurement and construction subcontracts) to carry out the construction and installation
(i.e. the turnkey scope) work.
There are a number of alternative contracting strategies which may be adopted by the HVDC vendor and
an EPC contractor to deliver the complete turnkey scope, through either consortium or prime-sub contract
arrangements. Perhaps the more logical division of scope is for the EPC contractor to take the turnkey
lead, and allow the HVDC vendor to design the system, then manufacture and test the equipment on a
Design, Supply, Supervision, or Engineered Equipment Package approach. This allows the EPC contractor
to focus on the conventional equipment, and all local content including the civil works and construction.
The HVDC vendor may then in turn focus on the HVDC System design and manufacture, and support the
in-service performance warranties.
The utility may choose to install the HVDC equipment and test it under the supervision of the vendor in order
to satisfy the warranty terms. In this case, the contract model would be design supply and supervision only
(CIF + supervision according to Incoterms), the contract being between the utility and the HVDC vendor.
• What staffing level is to be assumed for the converter stations?

It is feasible to operate converter stations totally unmanned and under remote control, if sufficient and reliable
communications, monitoring, recording, and reporting facilities are incorporated into the design.

CHAPTER
At the other extreme, if the utility prefers to use manual methods for part or all of the station operation and
maintenance, then this is equally possible, although increasingly uncommon in today’s HVDC projects.
Any point in between these two extremes is also possible according to owner/utility preference.
10.1.6.1. Project Lifetime in Service

It is worth noting that the design life of a HVDC link is typically 30 years, and operational experience of
existing HVDC schemes indicates that this is generally achieved or exceeded. There are however some areas
(especially the control system) where equipment may need to be replaced due to component obsolescence
during this lifetime. Normally, a five-year component spares supply is identified and delivered on the initial
contract, this ensures that the required in-service reliability conditions can be achieved.
The required Reliability and Availability design targets are identified at the time of the initial specification
and tendering stage of the project. They are subsequently confirmed, and form an integral part of the design
phase, during the contract and the construction stage of the project. In practice this means that in order to
achieve a certain level of Availability, a certain design philosophy is adopted which includes: redundancy
in appropriate areas of the plant, spares-holdings and storage locations, manpower levels and locations,
proper assessment of the component failure rates, etc.
For a more detailed discussion of the Reliability, Availability and Maintenance (RAM) calculations and how
they affect the detail of the HVDC project design, please refer to section 9.8.
• Maintenance and Spares
The conventional areas of the plant obviously require maintenance, as do the more complex HVDC-specific
areas of the plant.
Transformers, circuit breakers, disconnectors, grounding/earth switches, CTs, VTs, batteries and chargers,
control, protection, monitoring and recording systems, HVAC systems, thyristor valve cooling plant, etc. all
rely on the Original Equipment Manufacturer (OEM) recommendations, and in this respect only the converter
transformer tapchangers are subjected to in-service duty that may be considered more onerous than
conventional AC duty.
In terms of maintenance on the HVDC-specific areas of the plant, this may be considered as only referring to
the thyristor valves and controls.
Historically, thyristor valves were supplied with the recommendation that annual maintenance shutdown work
was carried out, including invasive testing on the thyristor valves and controls. However, as the in-service
experience in reliability of thyristor devices and their associated converter circuits has evolved and improved
over the last 20 years or more, it has become clear that the on-line automated diagnostics, supplied as part of
the standard HVDC contract, is sufficient for the reliable reporting of the majority of the common valve failures.
Therefore, normally in modern HVDC systems there is no longer a provision for periodic thyristor level testing on
all locations of the valve, other than those thyristors which the monitoring system has reported as having failed.
The annual shutdown maintenance activity is therefore limited to changing any failed thyristors, cleaning,
vacuuming, and a general inspection. The valve components require checks on coolant leaks or other alarm
conditions not being picked up by certain conditions.
• Operational Performance Benchmarks
The CIGRE and IEEE organizations each maintain a worldwide list of HVDC projects in service.
In addition, CIGRE B4-04 also issues a report every 2 years, which details the performance of many HVDC
links around the world. The contributions are voluntary of course, and not all in-service schemes report to the
committee either for commercial confidentiality reasons, or they have no interest in the data. No other area of
the worldwide transmission community gathers this form of data.
However, it has been demonstrated that the information provided in this report is an invaluable guide to both
the owner and the broader industry:
➙ It indicates general and specific problems encountered on HVDC links in operation
➙ It allows owners/operators to identify common issues and trends in performance
➙ It illustrates the high levels of performance possible with such schemes.

CHAPTER
It is recommended that both existing and prospective owners and operators of HVDC participate in this
process, to benefit not only themselves, but also the broader HVDC community. For more information go to
www.cigre.org.
10.1.6.2. Financing the Project

In the traditional utility world, the national or regional utility would determine the need for a HVDC link.
The construction of the link would be carried out using their own budget funds and external finance, and
in many cases around the world the utility is a department of the national government.
If recovery of the construction cost is part of the overall plan then the recovery is often aggregated over a long
period through an increase in the amount billed to the customers. Depending on factors such as the utilization
of the link, the cost of energy to buy and to sell, the operating costs, etc, the period of time to actually recover
the construction cost may be as low as 2 or 3 years.
In the de-regulated market environment where the Regional Transmission Operator (RTO) determines the
need for the HVDC link, the RTO may dictate that the construction cost is to be incorporated into the rate-
base, meaning that the end-user pays.
In this situation it is common for the RTO to apportion the construction cost recovery across several areas
(utilities), which each benefit from the improvement in service in some way.
In a fully de-regulated environment there are opportunities for non-utilities to enter the market and build a
business around the use of a HVDC transmission link, by simply identifying a source of low-cost energy, a
buyer of that energy, and creating a project with construction and operating costs which result in a viable
business model.
Private or so-called merchant developers of transmission projects take many different forms, from single
entrepreneurs to large multi-national organizations. Venture capital funding is a major potential source of
money to support the development and construction phase of the project, with the promise of payback once
the scheme enters commercial service.
In some areas of the world other sources of finance are possible, from the European Bank, World Bank, Asia
Development Bank, etc., through to loans or grants to either fully or partially fund the project.
10.1.6.3. Legal and Regulatory Issues

Individual countries around the world implement their own legal and regulatory processes for the approval
of transmission projects. In some cases, even within different regions in the same country, there are
different processes to be followed to reach the stage where construction is approved.
In those countries where there is only one national utility and it is effectively a branch of the government,
the process to acquire the land and obtain all the necessary connection and construction permits may be
through various government agencies.
In other countries, there will be national, regional and local bodies responsible for the permitting of land for
construction or line/cable ROW use, who will need to be satisfied that there is sufficient justification for the
project, and that all viable alternatives have been fully explored.
It is clear that the process for each project will be unique and the legal and regulatory process to be followed
must be fully investigated and planned to ensure that the project goes forward according to schedule.
10.2. OPERATIONAL EXPERIENCE

The following projects are presented for illustrative purposes to provide insights into the particular
characteristics of HVDC projects.
10.2.1. Nelson River HVDC

The Manitoba (MB) Hydro electrical system supplies the province of Manitoba. The peak load of MB Hydro
in 2008 was approximately 4300 MW and the installed generating capacity was 5700 MW. Nearly 70% of
CHAPTER
the power is generated from three hydroelectric stations on the Nelson River in northern Manitoba. This
power is transmitted over a distance of 900 km via the Nelson River HVDC transmission to the Dorsey
converter station, located near the major load centre of Winnipeg.
MANITOBA Hudson Bay
Henday Nelson River

Converter
Station
Radisson Limestone 1330 MW
Converter Longspruce 980 MW
Station
Kettle 1272 MW
Lake Winnipeg
AN
EW
CH
AT
SK
ONTARIO
Winnipeg
SA
Dorsey
Converter
Station
10.2.1.110.2a1
. Nelson River HVDC History
The HVDC system is supplied today from three generating stations that total 3580 MW (in-service dates:
Kettle-1974, Long Spruce-1979, Limestone-1992) on the Nelson River AC collector system, which is
operated isolated from the remaining MB Hydro AC system.
The Nelson River HVDC transmission system consists of two bipoles (Fig. 10.2a). Bipole I is rated at +/-
463.5 kV, 1854 MW. Bipole I originally used mercury arc valves and Pole 1 was put in service in June 1972.
This has been recognized on the list of IEEE milestones in electrical engineering because at the time this
was the highest operating voltage in the world and transmitted the largest amount of power from a remote
site to a city. At 150 kVdc, the Nelson River mercury-arc valves were the highest voltage mercury-arc valves
ever developed. Stage 3 of Pole 2 of Bipole I was commissioned in 1977. Between 1954 and 1976, there
were eleven mercury-arc, HVDC systems that went in service and Nelson River was the last one to be
commissioned. Each pole had three 6-pulse converter valve groups in series, each rated at 154.5 kV and
2000 A. In 1992 and 1993, the Pole 1 mercury-arc valve groups were replaced with electrically triggered
thyristor valves and new valve group control and protection [11]. Pole 2 mercury-arc valve groups were
replaced in 2004 with light triggered thyristors.
In 1975, Manitoba Hydro awarded the Bipole II order to the German-Swiss HVDC transmission
working group formed by AEG-Telefunken, Brown Boveri and Siemens. Bipole II is rated at
+/- 500 kV, 2000 MW and each pole has two 12-pulse thyristor valve groups in series. Stage 1 (900 MW, +/- 250
kV) was put in service in October 1978 [4]. The final stage was put in service in June 1985 [16]. The scheme
was limited to 1800 MW at ambient temperatures above 26°C. A cooling upgrade was completed in 2001 to
allow operation at 2000 MW at any ambient temperature up to 36°C [16]. Increased power sales to the US
resulted in the need to operate the Bipole at its rated capacity of 2000 MW during summer peak conditions.

CHAPTER
Kettle LongÊSpruce Limestone

1272ÊMW 980ÊMW 1330ÊMW
7 3 3 5
138ÊkV Radisson 230ÊkV Henday
PoleÊ1 PoleÊ2 PoleÊ3 PoleÊ4
BipoleÊI
BipoleÊII
+/-ÊÊÊ463ÊkV
+/-ÊÊÊ500ÊkV
1854ÊMV
2000ÊMV
Dorsey
230ÊkV
11ÊLinesÊto
ManitobaÊnetwork
500ÊkV
ToÊUSA
Fig. 10.2a - Nelson River HVDC system
The Manitoba Hydro southern AC system is interconnected with Saskatchewan to the west via four 230 kVac
lines and with Ontario to the east via two 230 kVac lines via phase shifting transformers. There are three 230
kV and10.2a2
one 500 kVac interconnections between Manitoba Hydro and the U.S. power system to the south.
10.2.1.2. DC Line Faults

MB Hydro recently improved its line fault restart protection. The initial DC restart protection was replaced
by new protection that allows restarts following high ohmic resistance DC line faults as well as fast-front
(low ohmic) DC line faults. The modified DC line protection gives the flexibility of multiple restarts at full
or reduced DC voltage. The protection was designed based on offline studies that examined DC restart
conditions for various system outages and various voltage levels for restarts. Previously, only system intact
conditions had been studied and restarts were not enabled for high ohmic faults. The system was also
tested with actual staged DC line faults prior to being put in service.

CHAPTER
Shortly after installation, the design modifications and new settings that allowed multiple restarts at full and
reduced voltages were very helpful in the summer of 2008 when there was an unprecedented line fault season
with 46 DC line faults (normally about 20 [15]). It appears that a forest fire was the initial cause of many of the
DC line faults in 2008 and subsequently contamination due to smoke residue resulted in an increased number
of DC line faults that year. With the capability to be able to restart at reduced DC voltage and following high
resistance faults, Bipoles I and II were able to restart successfully for 43 of the 46 DC line faults.
An on-going issue with the DC line faults every year is that there are a number of DC line faults that cannot
be traced to smoke or weather in the area. It has been proposed that these anomalous line faults are due to
a combination of additional insulators being needed on the DC line and that the DC voltage surface gradient
is too high. This information will be used to better plan the new upcoming Bipole III DC line. The existing lines
have a bundle of two sub-conductors and suspension strings with 21 Sediver N18PDC glass insulators [2].
10.2.1.3. DC Controls
The Nelson River Bipole modulation controls and AC system interaction are complex and require proper
coordination [1], [6], [13]. A frequency-based capability controller of the collector system reduces DC
power when the frequency drops below 59 Hz at the rectifier AC busbar to prevent further frequency
decay. The rectifier frequency modulation control modulates DC power in relationship to the swing of the
AC frequency at any level in an attempt to damp oscillations. The rectifier is seen as a constant power load
by the northern AC generators, which results in negative damping. The DC power order can be modulated
according to the frequency deviation to damp AC system frequency swings.
At the receiving end (inverter), the frequency modulation controls can modulate the DC power in relationship
to the swing of the AC frequency at any level. In normal operation this controller provides little damping and
is intended for frequency control when the Manitoba Hydro system is isolated. In addition, the receiving end
of the system also has phase angle damping controls that modulate the DC power in relationship to swings
of the phase angle of the receiving bus. Since both the rectifier (sending) controls and the inverter (receiving)
controls are using the same DC power for modulation, coordination is critical. In general, the rectifier AC
system end frequency range is allowed to vary many orders of magnitude more than the receiving end since
the rectifier AC system is an isolated system with no customer load.
A Ud-hold control function was added to Bipole I and II in 1985 [9], which freezes the DC voltage to a preset
level in the current reference calculation. The Ud-hold sets the DC link to operate in a pseudo-constant current
control mode where power modulations are still allowed but otherwise the DC power current reference
remains constant. The modified control has proven to be very effective in stabilizing operation when operation
is close to the critical ESCR and there is risk of voltage collapse.
Other controls are provided for tripping of valve groups between 60.8 and 63.5 Hz for VAr relief and
frequency stabilization at the inverter.
Given the control complexity, it is important to have a detailed fully validated digital model for determining
control settings and for the analysis of disturbances: detailed models were developed for Bipole I and II and
have proven to be invaluable [14].
10.2.1.4. Special Protection

During hydraulic surplus conditions, power from MB Hydro is delivered to the interconnected systems
(primarily to the U.S.). The maximum power exported can be up to 2200 MW. During this time, the 500 kVac
line is loaded to approximately 1600 MW with the remainder on the 230 kV tie lines. Under these export
conditions, loss of any one of these lines without remedial action would result in cascade tripping of all
the remaining ties between MB Hydro system and the U.S., Ontario and Saskatchewan systems, resulting
in total isolation of the MB Hydro system.
In order to maintain the stability of the interconnected system and prevent system isolation, a
Special Protection Scheme (SPS) was designed [19]. The Special Protection Scheme continuously
monitors the power flow in each of the tie lines. Whenever a tie line trips, the special protection
system reduces the HVDC power order for the Nelson River bipoles by a specified amount to maintain
system stability and to prevent overloads on parallel lines. In addition, the SPS monitors the status

CHAPTER
of various components of the 500 kV line (e.g. series capacitor banks, transformers, SVC, south section
of 500 kV line, single-phase or 3-phase trip) and reduces the HVDC power order by a calculated amount based
on the identified condition.
Under some transmission system loading conditions the tripping of a tie line or other system
components can result in power/voltage instability in the AC network near Dorsey. This can arise
because a reduction in AC system voltage results in an increase in DC system current order to
maintain constant DC power schedules. This increased current order causes increased converter
station reactive power consumption, which tends to cause further degradation of AC system
voltage. This instability can result in complete shut down of the HVDC system if no action is taken.
The Ud-hold function mentioned earlier is one action taken to prevent this instability.
The system undervoltage controller [19] is another protection scheme. It detects an impending voltage
collapse condition by monitoring the level of the Dorsey 230 kV bus. If the voltage falls below 0.99 p.u., a 150
ms timer is initiated. If the voltage stays below that level for more than 150 ms, a 345 MW HVDC reduction is
initiated at the rectifier. This HVDC power order reduction reduces converter station reactive power demand,
thereby helping restore AC system voltage. The trip signal is inhibited if during the 150 ms time interval the
rate of voltage recovery exceeds a set level.
All reductions perform effectively due to the HVDC sending-end AC system being electrically isolated from
the southern AC system: power does not appear on any parallel AC transmission. As mentioned earlier, the
HVDC auxiliary controls also have sending-end frequency controls to provide damping in the isolated system.
These controls are inhibited for two minutes following the HVDC reduction to prevent an unwarranted HVDC
power deviation, which would counteract the desired power order reduction. Consequently, the frequency
control following the reduction is strictly by the governor action of the generators. The generators, the HVDC
system and all the auxiliaries were specified to stay in operation following full load rejection.
The installation of this Special Protection Scheme has allowed the maximum power transfer capability
between MB Hydro and the U.S. system to be increased from a few hundred megawatts to over 2000 MW. The
alternative would have been to build costly additional interconnected AC transmission to maintain system
stability under high exports.
10.2.1.5. Paralleling
MB Hydro has the ability to reconnect the DC poles of the same polarity from each bipole at high speed
and in parallel configuration on the same DC line. This critical operational mode proved itself during a
catastrophic failure of both Bipole I and Bipole II HVDC line towers in 1996. An extreme wind downburst
toppled both bipoles and restoration was greatly improved with the ability to send most of the power
of both Bipole I and II down just one of the normal two DC lines with the converter poles in parallel. The
original designers had the foresight to design the DC lines to be able to carry twice the normal rating of
current in order that in the future parallel operation could be part of one of the DC system operating modes
[1], [5]. The first low voltage paralleling and deparalleling tests reported by MB Hydro demonstrated the
practicality of future multi-terminal HVDC operation [5].
Since the storm of 1996, MB Hydro has been testing the parallel operation and associated controls on a regular
basis. One goal is to familiarize operators with the parallel operation mode so that during an emergency, where
parallel operation is needed, it will be performed smoothly. Another goal is to confirm guidelines for parallel
operation, which are for the most part more stringent in testing than in an actual emergency since the high risk
of losing normal export conditions arises in normal operation, whereas, in an emergency, any extra power is of
value even if it were to be subsequently lost. Parallel studies confirmed expectations of VAr consumption, voltage
excursions, frequency excursions, and were useful in defining permanent guides for emergency parallel operation.
Testing confirmed the models, and has given valuable operator experience, as well as confirmation of existing
programs presently implemented in relay logic. Tests have so far only been conducted at reduced load levels
(<1927 MW) and are low risk since there is the ability to recover any lost parallel poles with the remaining normal
poles in operation. It is hoped that at some time in the future MB Hydro will also be able to test at higher loads,
similar to emergency situations. MB Hydro also plans to test other programs in the future such as continuation
modes of deparallel operation, which allow more power to remain transmitted subsequent to a fault occurring
while in parallel operation. The continual evolution of the NERC transmission planning standards may result in
additional limitations being placed on paralleling operation.

CHAPTER
10.2.1.6. Converter Transformers
The original planning specifications did not require spare converter transformers on Bipoles I and II,
assuming that one pole spare could be maintained [1]. With the addition of Limestone generation in 1992,
the HVDC equipment began operating at high loads on a daily basis. A spare valve group can be maintained
over the winter but not the entire summer period.
Since 1992, several single-phase converter transformers failed at Dorsey on Bipole I. There were
a variety of causes including wet insulation, poor brazed joints and blocked cooling ducts.
However in Bipole II Stage 1 there were eight 3-phase converter transformers and most of these failed
in service [18]. The failures were thermal in nature as a result of stray flux heating of the conductors
of the first few discs at the end of the coils, where there was an oil duct only every second disc.
The stray flux heating was particularly severe in the Stage 1 transformers because of the tap leads that ran
parallel to the upper and lower core clamps. Because of the absence of oil movement at the hot spot, these
failures were not predicted by measurements of dissolved gas-in-oil or furans-in-oil.
For the Bipole II Stage 2 converter transformers, the tap lead arrangement was improved, but the winding
design was left unchanged. Only two of the eight Stage 2 transformers have failed in service (both at Dorsey).
Due to the many transformer failures, MB Hydro has changed its spares philosophy [16]. Now, one spare
transformer of each type is kept on site at both the rectifier and inverter stations. All converter transformers
have continuous on-line hydrogen-in-oil monitoring.
(As a result of the overall poor performance of the original transformers, which were not supplied by
GE Vernova, the converter transformers on Bipole 1 and Bipole 2 have been replaced by units manufactured
by GE Vernova.)
10.2.1.7. Bushings
When Bipole I began operating at full voltage in 1975, periodic flashovers of porcelain-clad, oil-filled
wall bushings and apparatus bushings occurred. Beginning in 1984, booster sheds were installed on
selected bushings in Bipole I. Other porcelains received a coating of RTV. These measures are described
in [20]. Similar measures had not been implemented in Bipole II when in 1987 a 500 kV wall bushing at
Dorsey punctured under conditions of non-uniform wetting, caught fire and burned. These bushings were
redesigned to reduce the radial electric stress and booster sheds were installed [8].
While performance had been greatly improved by the above-mentioned measures, there was concern about
the continuing risk of a fire in a valve hall. As the wall bushings in Bipole I had been in service for over 25 years
a decision was made to replace them with composite wall bushings with silicone rubber sheds. At 300 kV
and at 450 kV these consist of two resin-impregnated cores joined by a duct filled with a SF6/N2 mixture. The
air-end portions of the cores are installed within fibreglass cylinders equipped with silicone rubber sheds
[12]. Because of problems associated with transporting fully assembled wall bushings of this type, they were
assembled, processed and electrically tested at site.
10.2.1.8. Ferroresonance
The Dorsey HVDC converter station 230 kVac bus was composed of four bus sections on which the
converter valves and transmission lines are terminated. On May 20, 1995, bus A2 was removed from service
to commission replacement breakers. A potential transformer failed catastrophically causing damage to
nearby equipment. The switching procedure resulted in the de-energized bus and the associated VTs being
connected to the energized bus B2 through the grading capacitors (5061 pF) of nine open 230 kV circuit
breakers. A ferroresonance condition caused the failure of the VT [17].
All wound potential transformers concerned were replaced with capacitor voltage transformers. A third
bus has been added to enhance the reliability and reduce the amount of grading capacitance potentially
connected to a transformer bank. Permanently connected 200 Ω loading resistors were installed on the 4.16
kV secondary bus of two of the station service transformers.
The future Bipole III converter station will minimize the amount of circuit breaker grading capacitance and
include short bus lengths to maximize reliability and ensure ferroresonance does not occur. Only capacitor
voltage transformers will be used.

CHAPTER
10.2.1.9. Future Plans
MB Hydro plans to add a third Bipole in 2017 to enhance reliability and provide generator outlet
transmission for future Nelson River generating plants: Keeyask (630 MW) and Conawapa (1395 MW). The
DC transmission line will be sited on a right-of-way remote from the existing lines in order to eliminate the
probability of concurrent outages due to extreme weather events. The converter size currently planned for
reliability purposes is 2000 MW with at least 5% overload, which matches historic analysis [7].
With the addition of a third inverter in the Winnipeg area at Riel approximately 38 km from Dorsey, concerns
with HVDC multi-infeed effects have been raised [21]. Bipole I and II have been planned assuming a minimum
ESCR of 2.5, which has required the addition of 1860 MVAr of synchronous condensers [10] (3-160 MVAr
English Electric in 1971, 3-160 MVAr ASEA in 1971 and 3-300 MVAr by Marine Industries Limited in 1992).
In order to minimize interaction concerns between the Dorsey and Riel inverters, Bipole III will be designed
ensuring a minimum multi-infeed equivalent short-circuit ratio (MIESCR) of 2.5. The MIESCR estimates
the strength of a given inverter bus with respect to all HVDC systems sharing the common AC system. A
minimum of 1000 MVAr of new synchronous condensers will be installed to maintain this minimum level. The
synchronous condensers have the benefit of providing inertia to support the frequency in cases where all of
the inverters fail commutation simultaneously.
Transformer energizing caused concerns with commutation failures [3] and dynamic overvoltages [1]. This was
partly solved by using pre-insertion resistors and operating guides. Point-on-wave switching will be employed
for Bipole III due to advances in circuit breaker technology.
Due to the age of the thyristors and controls, Bipole II is scheduled to be upgraded in 2017,
Bipole I in 2027 (pole 1) and Bipole I (pole 2) in 2039. Rapid developments have been occurring in
Voltage Source Converter HVDC technology with overhead transmission. MB Hydro will be investigating
the feasibility and benefits of migrating to this new technology.
Second Edition Note : Since the printing of the First Edition of this book, Bipole 3 has entered service, and
planning for the refurbishment of Bipoles 1 and 2 is under way.
10.2.2. Rivera Converter Station

The Rivera converter station was commissioned in year 2000. It offers the following main benefits:
➙ Provides a bi-directional power interconnector between the 50 Hz Uruguay and the 60 Hz Brazilian
systems. Both countries benefit from an optimized energy balance and this interconnection has proven
to be a strategic support during the energy crisis experienced over the past 5 years, due to the lack of rain
and the subsequent drastic reduction of hydro-power resources.
➙ Enables the transfer of energy as assigned by the operator which is automatically controlled, steady
and independent of the phase-shifts of the respective national power systems.
➙ Enables the transmission of power through the interconnection while keeping the independence of
each power system under fault occurrences or electromagnetic transient phenomena, which may
strike either system.
➙ Enhances voltage stability owing to the installed reactive power compensation devices (harmonic
filters and shunt reactors), which provide a solid control of the bus voltage, under the conditions of
the 140 km long radial feeder and weak fault current.
➙ Enables the station bus voltage control within a very narrow bandwidth using the station as a Static
VAr Compensator (SVC), thanks to the control loop which controls the thyristor firing angles in function
of the reactive power balance variations of the grid.
The Rivera converter station is remotely controlled from the load dispatch center 500 km away, which controls
the various substations as well. The converter station is therefore unmanned (no staff continuously attend to
the station’s operation). It is operated in a simple and secure way, via the SCADA system delivered with the
converter system. The SCADA monitors the proper operation of all the equipment permanently, enables the
remote settings of the main parameters of the power transfer, displays the alarms and can switch to the local
control of all the switching components or automatic controls, such as the ramping-up or the interruption
of the power transfer.

CHAPTER
The periodic maintenance of the converter system is achieved once a year, during a 7-day total outage. This
limited maintenance period contributes to a high level of operational availability, with reduced financial losses
related to energy trading.
The global management of the engineering, operation and maintenance (whether scheduled or forced) is
under the responsibility of a 4-engineer team, having a solid background and operational experience of
high voltage installations and power electronics. The maintenance of the balance-of-plant of the station is
under the control of this 4-engineer team, in compliance with in-house service agreements and outsourced
maintenance contracts.
It should be emphasized that all benefits foreseen at the planning stage by UTE were achieved during the
execution of the Rivera Power Conversion Project:
➙ Power supply availability at an acceptable price via the connection to the Brazilian system, especially
during the critical period of the energy crisis combined with the high prices of the fossil resources
➙ Return-on-investment in less than three years from the power interchange agreement between
Uruguay and Brazil
➙ Low maintenance expenditures, compared to thermal or hydro generation
➙ Reduced unavailability resulting from the scheduled annual maintenance program
➙ High availability performance
➙ Improvement of the HV grid stability in the area
10.3. STUDIES ASSOCIATED WITH CONVERTER STATION DESIGN

The expectation of the prospective purchaser of an HVDC system, is that the HVDC vendor has the
capability to undertake a wide variety of power system and environmental studies in order to assist
in developing the project at various stages. Presented below is a description of recommended studies
performed during the contract phase of a HVDC converter scheme.
10.3.1. Typical Contract Reports

A typical HVDC system contract would include the vendor carrying out the following studies, and the results
of each study would normally be presented in a report.
10.3.1.1. Main Scheme Parameters

Objective Define the range of operating conditions of the HVDC scheme and the major component ratings.
Design data expected Consistent operating data for the HVDC scheme under different operating conditions, the
from study number of thyristor valve levels and the converter transformer rating.
Equipment affected Thyristor valves, converter transformers and reactive power banks (AC harmonic filters).
The steady-state operating parameters of the HVDC scheme across its operating power range
are calculated for coherent operating conditions.
Methodology The various operating parameters, equipment tolerances and measurement errors that are
applicable to the scheme are varied between studies in order to explore the boundaries of
the HVDC scheme’s operation.

CHAPTER
10.3.1.2. Reactive Power
Establish the necessary sub-bank rating and switching sequence to meet the reactive power
Objective
control requirements of the scheme.
Design data expected Reactive power bank/AC harmonic filter bank MVAr rating, switching sequence of banks under
from study different operating conditions and converter reactive power absorption capability utilization.
Equipment affected Reactive power bank/AC harmonic filters.

The HVDC converter absorption under all extremes of operating condition tolerances,
measurement errors and operating DC voltage are established as part of the main scheme
parameters reports. From this converter absorption, the total reactive power required, allowing
Methodology
for the appropriate tolerance conditions, is established. Using the reactive power exchange limits,
established switch points will be calculated which keep the net reactive power interchange of
the converter plus AC reactive power banks with the AC systems within the established limits.
10.3.1.3. Harmonic Filter

Evaluate the AC side harmonic currents generated by the converters as a function of DC power
Objective
and establish an AC harmonic filter solution which meets the harmonic limits of the project.
Design data expected
Filter design topology, component values and rating data.
from study
Equipment affected AC harmonic filters.
The analysis performed establishes the combinations of the AC system and converter conditions,
such as frequency, temperature, transformer impedance, etc., which would give rise to the
maximum levels of harmonic distortion at the terminals of the HVDC station.
The harmonic currents generated by the rectifier and inverter of the HVDC converter are
evaluated using a digital computer program called JESSICA. The JESSICA program calculates
the magnitude of the individual harmonic currents from a mathematical analysis of the frequency
domain behavior of the converter.
The performance of the AC harmonic filters and their operational losses are calculated using the
Methodology
network harmonic penetration program HARP. The program models the filters and the injected
currents from the converters.
A standard mathematical maximization technique is used to search each harmonic impedance
area to find the impedance which produces the maximum value of voltage at a chosen node, or
of current in a chosen branch. This system impedance is inserted into the impedance matrix of
the circuit being analyzed for the harmonic current penetration study. The program then solves
Ohm’s law, using standard matrix mathematics techniques. This procedure is repeated for each
harmonic of interest.
10.3.1.4. RI, TVI and PLC Filter

Evaluate the high-frequency interference generated by the converters and establish a PLC filter
Objective
and RF screening solution which meets the limits for the project.
Design data expected PLC filter design topology, component values and rating data.
from study RF shielding of the valve halls.
Equipment affected PLC filters, valve hall RFI screen.
The components of the converter station are modeled in the appropriate frequency range to the
necessary level of detail. Of most importance will be the converter transformers, the converters
and the PLC filters.
Methodology The PLC frequency noise will be calculated at the relevant busbars with a range of transmission
line impedances over the range of PLC frequencies of interest.
The radiated interference at 15 m from the substation fence will be calculated taking into
account practical levels of RF screening applied to the valve halls.

CHAPTER
10.3.1.5. Insulation Coordination
Establish the appropriate protective levels of station surge arresters and hence BIL, clearance
Objective
and creepage of station equipment.
Design data expected The study will yield protective levels of station surge arresters, equipment BIL and creepages
from study and clearances on the DC side of the converter transformer.
Equipment affected All insulation.
From the maximum valve winding and DC voltage along with the specified maximum AC system
operating voltage, the appropriate surge arrester protective levels are calculated based on
historical data. From this data, the insulation levels of primary equipment are calculated as well
Methodology as the insulation levels of insulators. The calculated insulation level for insulators is corrected
to select an insulator which provides the necessary withstand of flashover probability for the
DC side equipment.
Clearances between equipment on the DC side of the converter transformer are also calculated.
10.3.1.6. Transient Overvoltage

Determine the transient overvoltage and current stresses on major converter station equipment
Objective including surge arresters, which form a basis for their insulation coordination. Surge arrester
energy absorption requirements will also be determined.
Transient overvoltage levels, surge arrester protection levels and energy absorption.
from study
Equipment affected Converter station equipment, filters and surge arresters.
The disturbances considered can be categorized as:
a) Switching impulse studies involving fault application and clearance on the converter AC
busbars, simulation of transformer and other energization events and filter switching events.
For these studies, the transformers in the vicinity of the converter stations are represented
by electromagnetic models which include non-linear saturation effects. The close proximity
of the filters to the transformers can give rise to ferro-resonance effects, particularly if the
filters and transformers become isolated from the main system. Non-linear transient effects
of surge arresters are also represented as required.
b) Lightning on the incoming DC side lines and filter busbar flashovers to ground.
The insulation levels of the station equipment will be coordinated with the protective levels of
the surge arresters and the stresses on the latter determined to achieve suitable arrester and
Methodology equipment design.
The transient overvoltages to be studied under this report are faster than those of the
fundamental frequency temporary overvoltage type, occupying a much shorter time scale
with faster rise time. They must therefore be studied with a fast electromagnetic program. In
the calculations, all circuit components (R, L and C) of the converter and close-up equivalent
AC system are represented as necessary, including adjacent lines, modeled by distributed
parameter representations or equivalent Pi and T sections as appropriate. Stray capacitance,
transformer saturation and surge arresters are included where appropriate and all phases are
represented separately.
A range of operating conditions is investigated, together with events leading to transient
overvoltages, in order to determine worst overvoltages and energy duties of AC and DC side
surge arresters and other components.

CHAPTER
10.3.1.7. Control System Dynamic Performance
Optimize the control parameters, to provide stable operation and good dynamic performance of
Objective the HVDC scheme. This study will be primarily carried out using physical controls and the Real
Time Digital Simulator (RTDS).
Design data expected Static characteristics, control system functions and parameters, verification of stability, response
from study times and fault recovery.
Equipment affected Control equipment.
High voltage equipment in the respective converter stations (e.g. converter transformers, AC/DC
filters and DC filters) as well as relevant generators and step-up transformers will be represented
in the simulator model. Derived equivalent AC system network models will be used to represent
the AC systems.
This dynamic performance study will be used to achieve the following objectives:
• Evaluation of DC control and protection functions
Methodology
• Confirmation of stable operation
• Confirmation of static characteristics and control set points
• Evaluation of the performance of the AC/DC system for different DC system control modes
• Evaluation of DC system performance for DC-side disturbances such as converter blocking,
pole blocking, and valve winding faults, including demonstration of protective shutdown when
required .../...
.../...
• Demonstration of the DC system response in accordance with the specified response criteria,
including control system step responses and recovery from AC system faults
• Demonstration of the DC system transient response for reactive component switching
Methodology
• Demonstration of normal start-up and shutdown
• Study the interaction with local machines during disturbances
• Evaluation of the performance of the DC system during severe AC faults and subsequent to
fault clearing. This will include the evaluation of DC power run-backs, if necessary.
10.3.1.8. Audible Noise

Objective Investigate the acoustic noise levels at the site boundaries.
Design data expected Confirmation that the equipment design meets the maximum acoustic noise limits at the
from study boundary.
Acoustically active equipment: including converter transformers, converter transformer coolers,
Equipment affected
filter capacitors, air conditioning and valve cooling heat exchangers.
Individual equipment suppliers’ information giving the acoustic noise predicted for individual
items will be modeled in a graphical representation of the site layout. The predictor software
performs the acoustic noise calculations using the methodology set out in the following
standards.
Methodology • ISO 9613-1: Attenuation of sound during propagation outdoors, Part 1: Calculation of sound
by the atmosphere (first edition 1993-06-01).
• ISO 9613-2: Attenuation of sound during propagation outdoors, Part 2: General method of
calculation (first edition 1996-12-15).
• VDI 2571: Schallabstrahlung von Industriebauten (Sound emission from industrial buildings).
10.3.1.9. Fundamental Frequency Temporary Overvoltage

Assess the performance of the DC interconnection, its controls and compensating equipment
Objective and the overvoltage limiting feature in response to large load rejection disturbances in the
equivalent AC/DC system, in order to determine the resulting temporary overvoltages.
Design data expected Maximum fundamental frequency transient overvoltages experienced by system electrical
from study power equipment.
Equipment affected Electrical power equipment in the vicinity of the HVDC converter stations.

CHAPTER
600
LEGEND
Building
Ground region
Housing region
Industrial site
Foliage region
Point source
Grid
Grid point
Surface contour
Line source
GPS calibration point
Receiver
< 60.0 dB(A)
60.0 - 62.5 dB(A)

62.5 - 65.0 dB(A)
65.0 - 67.5 dB(A)
400 67.5 - 70.0 dB(A)
70.0 - 72.5 dB(A)
> 72.5 dB(A)
period: Day period
Base case
0m 70 m
scale = 1 : 2500
origin = 270, 0
200
0
400 600
Fig. 10.3a – Audible noise plot 10.3a

CHAPTER
Transient studies (for a short period of a few cycles) are carried out for specified base cases of
the system equivalent network. The studies include specific cases of load configuration, and for
both partial and full load rejections.
AC system outage condition situations are studied similarly. The appropriate study cases will be
repeated with and without the reactive power absorption mode (TCR mode) of control for voltage
limiting actions in order to determine its effects, characteristics and ratings.
The study requires all the usual load flow and transient stability type data/models of the network
impedances, loads, generators/controls and operating levels/features. The study does not
necessitate a very large, full AC network to be modeled in order to achieve its objectives of
converter station design/assessment.
Methodology
The maximum fundamental frequency TOVs occur when there is loss of power transmission in
the HVDC link, which can occur due to:
a) Blocking of the link: this causes the converter transformer circuit breakers to open, thereby
isolating the converter station. Any delay in tripping the filters will cause high overvoltages
due to the prior rejection of the HVDC link reactive power demand which affects both ends
of the link.
b) 3 -phase ‘close in’ solid faults to the HV busbars, which can also cause loss of DC
transmission, although the converter transformers remain connected for this scenario.
Following clearance of the fault, TOVs can be high in the period before full power transfer
is re-established
10.3.1.10. Reliability and Availability

Define the overall reliability and availability of the converter station equipment and confirm the
Objective
equipment design and the recommended spares to meet the specified requirements.
Design data expected The energy availability and the forced outage rate associated with different scheme
from study configurations.
Equipment affected All significant plant items are considered as part of the evaluation.
A reliability study model is built by grouping together smaller models, known as
‘sub-systems’. A sub-system is a collection of components (or smaller sub-systems) whose
individual reliabilities can be combined together on the basis of their inter-relationships
(dependencies) to provide an overall measure of sub-system reliability.
The sub-system is then treated as a single component with its own failure and repair
characteristics. In this way, a reliability study model can be simplified into a smaller quantity
of representative sub-system modules. There are no fixed rules regarding the way in which
Methodology components are combined together to form sub-systems: the choice is based on the nature of
the plant and the experience gained from existing installations.
Examples of sub-systems are:
a) Converter valve: thyristors, gate units, monitoring units, ground-level electronics, cooling
components, etc.
b) Harmonic filter: inductors, capacitors, resistors, CTs, isolators, AC circuit breakers, etc.
A reliability study model of a complete system is built up by relating all the sub-systems which it
contains in terms of the effect of their failures on the other sub-systems.
10.3.1.11. Losses
Objective Calculate the ‘as manufactured’ losses of the converter station.
Total operational loss from plant.
from study
Equipment affected All
The converter station losses for operation under nominal AC system voltage and frequency
conditions and with nominal equipment parameters at an outdoor ambient temperature of
Methodology 20°C will be presented. The study will compile results from equipment factory tests along with
the proposed nominal operating conditions and present the total converter station losses in
accordance with the formulae defined in IEC 61803.

CHAPTER
10.4. INFORMATION REQUIRED FOR THE DESIGN OF A HVDC SCHEME
This section presents a standard questionnaire prepared by GE Vernova for potential HVDC customers.
It highlights the specific information required by the HVDC supplier in order to prepare the design specifications
for the project.
10.4.1. Questionnaire – Data Requirements for a HVDC Scheme

The importance of each question is defined by the following categories:
A A list of the minimum information required to enable a formal quotation to be prepared
B A list of minimum information required to enable a budgetary proposal to be made, e.g. for feasibility study
C A list of the additional, minimum technical definitions required at the time of a tender award.
Please note that when specifying the requirements for a HVDC link, it is recommended that as much detail
as possible is provided in response to the questions here. This will:
➙ Allow a more optimized HVDC design to be developed, leading ultimately to a more accurate price
➙ Reduce the disparity between interpretations from different vendors in responding to the same
specification
Category
1. AC system for each terminal
1.1 Voltage
A B nominal
A B maximum continuous
A B minimum continuous
A maximum short time and duration
A minimum short time and duration
1.2 Frequency
A B nominal
A B maximum continuous
A B minimum continuous
A maximum short time and duration
A minimum short time and duration
A B 1.3 Short circuit levels, maximum and minimum, for each stage of the development
A B 1.4 Insulation levels
A 1.5 Creepage and clearance distances
A 1.6 Harmonic impedance
A 1.7 Distortion and/or TIF limits for AC system. Which harmonics are to be assessed?
A 1.8 Is the AC system solidly or resistance earthed?
A 1.9 What are the design constraints (number of feeders, security criteria, current user practice,
etc.) for the AC switching station? Are there other large transformers connected to the converter
station busbar? If so, give transformer ratings, reactances, tap range, etc.
A 1.10 Give details of undervoltages and durations for AC system faults, both during the faults and
during the fault clearance period for both main and back-up protection, for single-phase and
3-phase faults if they are different.
A B 1.11 Give results of AC system disturbance studies including transient voltage and frequency
variations, and details of any limits on acceptable VAr generation/absorption and switching during
and after the disturbance.
A B 1.12 a) What is the maximum permitted step voltage change arising from filter switching?
b) What is the maximum permitted ramped voltage change and over what time duration?
A B 1.13 What is maximum temporary (<1 sec) overvoltage that existing equipment can withstand?
Are there any other significant limits on permissible temporary overvoltage?
A B 1.14 How much reactive compensation is required (i.e. what power factor is to be achieved) at the
AC terminals at various transferred power levels up to full load?
A 1.15 Give details of outgoing AC lines from each converter station.

CHAPTER
A 1.16 Give negative sequence voltage on each converter station AC busbar, and existing harmonic
voltages.
2. DC system
A B 2.1 Is power flow required in both directions?
A B 2.2 What is the nominal power to be converted into DC? (It is usually convenient to define this at
the rectifier DC output terminals). Note that if the power flow is unidirectional, then the rating of
the inverter can be made less than that of the rectifier.
A B 2.3 Is an overload capability required? If so, how much and for how long and under what conditions
of ambient temperatures?
A B 2.4 What are the ratings and parameters of the DC line, or cable?
Where possible, include:
– Voltage rating and location at which this is to be defined
– Current rating
– Resistance of each line or cable
– Inductance of each line or cable with tolerances
– Capacitance of each line or cable
– Harmonic impedance of each line or cable
– Length of each line or cable
– Equivalent circuit for each pole
A B 2.5 What are the permitted limits for harmonic current injection into the DC line?
A B 2.6 Is monopolar operation required either in emergencies or continuously?
A B 2.7 Is ground current allowed either in emergencies or continuously?
A B 2.8 Are ground (return) electrodes to be provided? If so give:
– Details as in question 2.4 above for both electrode lines
– Electrode resistance (predicted)
– Electrode type (sea or land) and approximate location
A 2.9 Information relevant to any possible electromagnetic coupling to other adjacent circuits, and
the nature of the possible disturbances liable to be produced by such coupling.
A 2.10 Details of any requirements for switching DC lines
A 2.11 What is the time scale for any staged development?
A 2.12 What are the capitalized costs for fixed and variable losses and at what power loading do they
apply?
A 2.13 In order that the quantities of spares required may be determined, give details of the energy
availability and station reliability targets, preferred maintenance intervals and design criteria to be
adopted (information to include cost of loss of service for availability and reliability optimization).
3. Generators (if applicable)
A B 3.1 What are the ratings and parameters of the generators?
Include:
– MVA
– Voltage (including harmonic content)
– Power factor
– Reactances, both transient and sub-transient direct axis
A B 3.2 What are the ratings and parameters of the generator transformers? Include:
– MVA
– Voltage
– Percentage reactance
– Connection
A 3.3 Will the generators be designed to absorb harmonics from the converters?
A 3.4 What are the in-service dates for the generators?
A B 3.5 Will control be provided to limit AC busbar voltage variation? If so, what will the voltage limits be,
and what maximum reactive power can the machines safely absorb?

CHAPTER
A 3
.6 Is the generating station at the same site as the converter station? If not, give details of the AC
lines between generating station and converter station including:
– Length of lines
– Number of lines, voltage ratings
– Impedance and characteristics of each line
4. Auxiliary supplies, for each terminal
A 4.1 Give details of preferred voltages and frequencies to be used for the auxiliary power system
and sources of auxiliary power supply, i.e. will supply be provided from a generating station or will
contractor have to supply auxiliary power transformers connected to the AC busbars?
A 4.2 Define whether start up of auxiliaries is to be manual or automatic.
A 4.3 If auxiliary supply is being provided by the operator, give details of reliability of supply and
disturbances. If the auxiliary supply is to be provided by the contractor, define redundancy
requirements.
5. Controls and telecoms
A 5.1 Locations from which control instructions may be received
A
5.2 Control philosophy to be explained. To what extent is automatic control preferred to giving
detailed responsibility to the operator?
A 5.3 Any operation requirements to be defined, i.e. power and/or current control, rate of change of
power, etc.
C 5.4 Define the performance required during and after disturbances in either AC system.
C 5
.5 Define any restrictions on the recovery from DC disturbances arising from the requirements of
the AC system
C 5
.6 Define the disturbances liable to occur in either AC system and the control objectives required
of the DC system in such circumstances such as supplementary control signals in response to AC
frequency or voltage at one or more terminals.
For questions 5.7 to 5.14, consider the following aspects:
(a) What are the particular requirements for the project?
(b) What are the system practices which should be followed for operational convenience?
C 5.7 Control desk or panel requirements
A B 5.8 Protection requirements
A 5.9 Requirements for line fault protection
A 5.10 Requirements for line fault location
A 5.11 Requirements for alarm annunciation
A 5.12 Requirements for sequential event recording
A 5.13 Requirements for disturbance recording
A 5.14 Supervisory system requirements
A 5.15 Is a HVDC power line carrier (or fiber optic cable communications) system to be provided by
the contractor and, if so, what information will be carried on it?
A 5.16 Details of any interface between power line carrier (or fiber optic cable communications) and
any other telecommunication system(s).
A 5.17 Arrangements for communication during the construction/commissioning stage,
between sites and from each site to national and international circuits for telephone,
printer, fax, etc.
A 5.18 Requirements for permanent facilities to be provided by the contractor for telephone, telex,
fax, etc., circuits.
A 5.19 Map of route of DC line (if applicable) showing sites, and respective distances, for power line
carrier repeater stations, indication of any suitable auxiliary power supplies that may be available
at these sites.

CHAPTER
6. The following general information is required for each site:
A 6.1 Extent of supply to be clearly defined especially any services or equipment to be provided
outside the converter station.
A 6.2 Interfaces with equipment or services outside the scope of supply to be clearly defined
A B 6.3 Site conditions:
a) Ambient temperatures
(i) nominal
(ii) maximum
(iii) minimum
(iii) maximum wet bulb and coincident dry bulb
(b) Altitude
(c) Maximum wind speed
(d) Maximum depth of snow
(e) Rainfall
(f) Isokeraunic levels
(g) Range of relative humidity
(h) Incidence of air pollution (salt or industrial)
A 6.4 What are the RFI limits and where are they to apply?
A 6.5 What are the low frequency electromagnetic field limits?
A 6.6 What are the audible noise limits and where are they to apply?
A B 6.7 Are the sites in an earthquake zone; if so what horizontal/vertical ground accelerations do
structures and building have to be designed to withstand? Is there any reference elastic response
spectrum available? What are the the local regulatory codes which shall apply (if relevant)?
A 6.8 Local structural/building codes, defining what factors have to be applied to forces due to wind
and/or snow, when designing buildings and structures.
A B 6.9 What are the maximum loading gauges and weight restrictions at the ports and on the routes
to each site?
A 6.10 Maps of the areas of the sites showing the areas available for the converter station and those
available for ground or sea-shore electrodes.
A 6.11 Site surveys including soil analysis especially in the areas of the ground electrodes if these are
required.
A 6.12 Maps showing the location of the outgoing AC lines and DC lines from the converter stations.
A 6.13 Details of auxiliary supplies that will be available, during construction and installation.
A 6.14 Details of water supplies, available at the sites, including flow rates and chemical analysis.
A 6.15 What language should be used on drawings and instructions?
A 6.16 Details of any preferred materials that will utilize local resources, e.g. copper or aluminum,
brick or concrete.
A 6.17 Standards and specifications to be used, stating the order of precedence. This list should
also include specific standards relating to items of equipment including busbars, transformers,
switchgear, cabling and wiring, insulation oil, civil works and structures, equipment finishes,
painting, drawings and drawing symbols.
C If the operator has any standard specification for control and protection circuits, e.g. control circuits
for circuit breakers, then copies of these should be provided.
A 6.18 Do IEC test standards apply?
A 6.19 Define what office and workshop facilities are to be provided, e.g. should workshops be capable
of handling major items such as converter transformers?
A 6.20 Define any restrictions applicable to indoor, oil-filled equipment.

CHAPTER
BIBLIOGRAPHY
[1] “Nelson River HVDC Bipole-Two Part 1-System Aspects”, IEEE Trans. on Power Apparatus and Systems,
Vol. PAS-98, No. 1, p. 165-173, Jan./Feb. 1979.
[2] R. M. Morris, A. R. Morse, J. P. Griffin, O. C. Norris-Elye, C. V. Thio, J. S. Goodman, “The Corona and Radio
Interference Performance of the Nelson River HVDC Transmission Lines”, IEEE Trans. on Power Apparatus
and Systems, Vol. PAS-98, No. 6, p. 1924-1936, Nov/Dec. 1979.
[3] V. Koschik, and D. A. Woodford, “Impact of the Winnipeg-Minneapolis 500 kV AC Interconnection on
the Operation of the Nelson River HVDC System”, IEEE International Conference on Overvoltages and
Compensation on Integrated AC-DC Systems, Winnipeg, Canada, July 1980.
[4] G. Mazur, R. Carryer, S. T. Ranada, and T. Web, “Converter Control and Protection of the Nelson River Bipole
2 Commissioning and First Year of Commercial Operation”, IEEE Trans. on Power Apparatus and Systems,
Vol. PAS-100, No. 1, p. 327-335, January 1981.
[5] D. Brandt, I. McKay, M. M. Rashwan, and S. T. Ranade, “Paralleling and Deparalleling Tests on Nelson
River HVDC Bipoles 1 and 2”, IEEE Trans. on Power Apparatus and Systems, Vol. PAS-103, No. 4, p. 762-770,
April 1984.
[6] J. B. Davies, and D. G. Chapman, “Recent AC Control Enhancements of the Nelson River HVDC Links”,
Proceedings of IEEE/IREQ International Conference on DC Power Transmission, Montreal, Quebec, Canada,
June 1984.
[7] P. R. S. Kuruganty, and D. A. Woodford, “A Reliability Cost-Benefit Analysis for HVDC Transmission
Expansion Planning”, IEEE Trans. on Power Delivery, Vol. 3, No. 3, p. 1241-1248, July 1988.
[8] M. M. Rashwan, and W. McDermid, “Experience with DC Wall Bushings Associated with the Nelson River
System”, Proceedings of the Second HVDC System Operating Conference, Winnipeg, MB, Canada, p. 117-
125, 18-21, September 1989.
[9] M. M. Rashwan, G. B. Mazur, M. A. Weekes, and D. P. Brandt, “AC/DC System Power/Voltage Stability
Enhancement Using Modified Power Control”, Proceedings of the Second HVDC System Operating
Conference, Winnipeg, MB, Canada, p. 35-39, September 1989.
[10] C. V. Thio and J. B. Davies, “New Synchronous Compensators for the Nelson River HVDC System -
Planning Requirements and Specifications”, IEEE Trans. on Power Delivery, Vol. 6, No. 2, p. 922-928, April 1991.
[11] F. G. Goodrich, J. L Haddock, B. A. Rowe, H. L. Thanawala, D. B. Willis, “The Integration of New Valves
and Controls to Nelson River Bipole 1”, IEE 5th International Conference on AC and DC Power Transmission,
London, U.K., September 1991.
[12] M. M. Rashwan, W. McDermid, F. Hammer, and A. Küchler, “On the Design, Testing and Operating
Experience of Composite Dry Bushings in HVDC”, IEE 5th International Conference on AC and DC Power
Transmission, London, U.K., September 1991.
[13] J. Chand, “Auxiliary Power Controls on the Nelson River HVDC Scheme”, IEEE Trans. on Power Systems,
Vol. 7, No. 1, p. 398-402, February 1992.
[14] K. L. Kent, P. Kuffel and D. T. Y. Tang, “Investigation of Selected System Disturbances on the Nelson
River Bipole 1 HVDC System Using Detailed Control Model Digital Simulations”, Proceedings of the 3rd HVDC
Operating Conference, Winnipeg, Canada, May 1992.
[15] N. S. Dhaliwal, L. D. Recksiedler, and D. T. Y. Tang, “Operating Experiences of the Nelson River HVDC
System”, IEEE Transmission & Distribution Conference, p. 174-180, Sept 20, 1996.
[16] R. A. Valiquette, “HVDC Life Extension for Nelson River HVDC System”, IEEE PES Winter Meeting, p.
1059-1062, January 2002.
[17] D. A. N. Jacobson, “Examples of Ferroresonance in a High Voltage Power System”, Paper 03GM0984,
IEEE PES General Meeting, Toronto, ON, Canada, June 2003.

CHAPTER
[18] D. H. Grant and W. McDermid, “Assessment of Thermal Aging of HVDC Converter transformer Insulation”,
Conference Record of the 2004 IEEE International Symposium on Electrical Insulation, Indianapolis, IN, USA,
p. 230-232, 19-22 September 2004.
[19] N. S. Dhaliwal, J. B. Davies, D. A. N. Jacobson, R. Gonzalez, “Use of an Integrated AC/DC Special Protection
Scheme at Manitoba Hydro”, CIGRE Paper B5-206, Paris, 2006.
[20] W. McDermid, and T. Black, “Experience with Preventing External Flashovers in HVDC Converter
Stations”, Conference Record of the 2008 IEEE International Symposium on Electrical Insulation, Vancouver,
BC, Canada, p. 81-84, 8-11 June 2008.
[21] “Systems with Multiple DC Infeed”, CIGRE Brochure 364, December 2008.

CHAPTER
CHAPTER
TOC HVDC: Connecting
564 | DCtoTransmission
APPENDIX

APPENDIX
Contents
APPENDIX RELATED TO CHAPTER 2 ................... 564
A2.2.1 Derivation of RMS and Fundamental Frequency

Valve Winding Current .......................................................... 566
A2.2.2 Derivation of per-unit DC Voltage Equation ................ 570
APPENDIX RELATED TO CHAPTER 4 ................. 572
A4.1 BRIEF DEFINITION OF TERMS ............................................ 572
A4.2 TELEPHONE INTERFERENCE FACTORS ......................... 573
A4.3 MATHEMATICAL BACKGROUND OF FILTERS .............. 574

A4.3.1 Double-tuned Filter ................................................................ 574
A4.3.2 Triple-tuned Filter ................................................................... 578
A4.3.3 Single-tuned band-pass Filter
with Parallel Capacitor .......................................................... 580
A4.4 BANK SIZE OPTIMIZATION .................................................. 581
A4.5 NETWORK IMPEDANCE ........................................................ 582
APPENDIX RELATED TO CHAPTER 6 .................. 584
A6.5 SWITCHING FUNCTIONS FOR SIMULATING

THE OPERATING CONDITIONS PRESENTED
IN FIG. 6.5.3B ............................................................................. 584
APPENDIX RELATED TO CHAPTER 9 ................. 586
A9.1 CONVERTER EQUATION MODIFICATIONS

DURING TRANSIENT ABNORMAL CONDITIONS ........ 586
A9.2 SOME TECHNIQUES USED TO QUANTIFY

DYNAMIC INTERACTION ...................................................... 587
A9.2.1 Eigenderivative .............................................................................................................................................. 587
A 9.2.2 Frequency Domain Eigenanalysis .................................... 588
A.9.2.3 Prony Analysis .......................................................................... 588
A.9.2.4 Time Domain Full Model Analysis .................................... 589
A9.3 RAM ANALYSIS OF SIMPLE SYSTEMS ............................. 590

A9.3.1 Basis for Analysis .................................................................... 590
A 9.3.2 One-element System ............................................................ 590
A.9.3.3 Two-element System ............................................................ 592
A.9.3.4 Three-element System ......................................................... 593

A2.2.1 Derivation of RMS and Fundamental Frequency
Valve Winding Current
Fig. A2.2.1a shows an idealized valve winding
current waveform of a 6-pulse converter bridge.
30° 120° 60° 120° 30°
As an ideal waveform there is no overlap. We can
see from Fig. A2.2.1a that the positive and negative
half cycles of the waveform are symmetrical and
therefore we can limit the analysis to only the
positive half cycle for simplicity.
Firstly, consider the RMS equivalent current. The
RMS current of a symmetrical waveform can be
360°
derived from the general equation:
1
Fig. A2.2.1a– An idealized valve winding current
1 π
2

I RMS =  ⋅ ∫ f ( I )2 ⋅ dθ  [Eqn A2.2.1a] waveform
π 0 
For the ideal current waveform in Fig. A2.2.1a, this becomes:

1
 56⋅π
2

1 
I RMS =  ⋅ ∫ Id ⋅ dθ 
2
[Eqn A2.2.1b]
π π
 6

Giving:
2
I RMS = ⋅ Id [Eqn A2.2.1c]
3
By a similar process, the harmonic content (including fundamental) of the valve winding current can be
found from Fourier analysis as follows:
2
I1 = ⋅ (an2 + bn2 ) 2
1
[Eqn A2.2.1d]
2
where, an is:
1
π
an = ⋅ f (ω ⋅ t ) ⋅ cos(n ⋅ ω ⋅ t ) ⋅ d (ω ⋅ t )
π ∫0
[Eqn A2.2.1e]
and bn is:
1
π
bn = − ⋅ f (ω ⋅ t ) ⋅ sin(n ⋅ ω ⋅ t ) ⋅ d (ω ⋅ t )
π ∫0
[Eqn A2.2.1f]
n is the harmonic number of the of the component to be analyzed.

The phase shift in the fundamental current is given by:
bn
Φ = tan( ) [Eqn A2.2.1g]
an
Hence, the ideal fundamental component of valve winding current can be found by setting n = 1, to be:
6
I1 = ⋅ Id
π [Eqn A2.2.1h]

CHAPTER
APPENDIX
Fourier analysis
requires this reference for correct
calculation of phase shift
-60° 60°
0°
120°
30°
360° Equivalent
fundamental
frequency current
Fig. A2.2.1b– A practical valve winding current waveform
Now, consider the inclusion of firing delay angle and overlap angle, as shown above in Fig. A2.2.1b.
The equation for the rising edge of the valve current has already been derived in Eqn 2.2.2f and can be
written as:
2 ⋅ ELL ⋅ (cos(α ) − cos(ω ⋅ t ))

i= [Eqn A2.2.1i]
2 ⋅ ω ⋅ Lc
As the valve currents are assumed to be symmetrical, the falling edge of the valve current can be written as:
2 ⋅ E LL ⋅ (cos(α ) − cos(ω ⋅ t ))
i = Id − [Eqn A2.2.1j]
2 ⋅ ω ⋅ Lc
Hence, the RMS equivalent current of the practical valve winding waveform is given by:
1
2
  α + µ 2 ⋅ E ⋅ (cos(α ) − cos(θ )) 
  ∫ ( LL
)2 ⋅ dθ 
  α 2 ⋅ ω ⋅ Lc 
  120°+α 
1
I RMS =  ⋅  + ∫ Id 2 ⋅ dθ  [Eqn A2.2.1k]
π  
  α +µ 
  2 ⋅ ELL ⋅ (cos(α ) − cos(θ )) 2 
α +µ
  + ∫ ( Id − ) ⋅ dθ  
  α 2 ⋅ ω ⋅ Lc  

CHAPTER
The RMS value of the fundamental frequency component of the practical valve winding current is given by
Eqn A2.2.1d, where an and bn are given by:
an = a1 + a2 + a3 [A2.2.1l]
bn = b1 + b2 + b3 [A2.2.1m]
Where:
α + µ − 60°
1 2 ⋅ Ell ⋅ (cos(α ) − cos(ω t + 60°)
a1 = ⋅ ∫ ⋅ cos(ω t ) ⋅ d (ω t ) [A2.2.1n]
π α − 60°
2 ⋅ ω ⋅ Lc
1
α + 60°
a2 = Id ⋅ cos(ω t ) ⋅ d (ω t )
π α + µ∫− 60°
⋅ [A2.2.1o]
α + µ + 60°
1 2 ⋅ Ell ⋅ (cos(α ) − cos(ω t − 60°)
a3 = ⋅ ∫ Id − ⋅ cos(ω t ) ⋅ d (ω t ) [A2.2.1p]
π α + 60°
2 ⋅ ω ⋅ Lc
α + µ − 60°
1 2 ⋅ Ell ⋅ (cos(α ) − cos(ω t + 60°)
b1 = ⋅ ∫ ⋅ sin(ω t ) ⋅ d (ω t ) [A2.2.1q]
π α − 60°
2 ⋅ ω ⋅ Lc
1
α + 60°
b2 = Id ⋅ sin(ω t ) ⋅ d (ω t )
π α + µ∫− 60°
⋅ [A2.2.1r]
α + µ + 60°
1 2 ⋅ Ell ⋅ (cos(α ) − cos(ω t − 60°)
b3 = ⋅ ∫ Id − ⋅ sin(ω t ) ⋅ d (ω t ) [A2.2.1s]
π α + 60°
2 ⋅ ω ⋅ Lc
The resultant of this is typically only marginally lower than the resultant of Eqn A2.2.1h.
Also, from Eqn 2.2.4a the power factor angle of the converter can be calculated by substitution into
Eqn A2.2.1g.
For practical cases, the power factor found from these equations results in a converter reactive power
loading very close to that given by the Uhlmann approximation, Eqn 2.2.4d.

CHAPTER
APPENDIX
A2.2.2 Derivation of per-unit DC

Voltage Equation 3 . ω. L
π c
Id
Based on the model shown above in Fig. A2.2.2a
and using the mathematical identities below,
a per-unit expression for converter bridge DC
voltage with respect to DC current, operating angle
Vd1 Vd
and commutating impedance can be found (note,
subscript ‘0’ refers to rated conditions):
3
Vdi0 = ⋅ 2 ⋅ VLL [Eqn A2.2.2a]
π
Fig. A2.2.2a– A simplified electrical equivalent
3 model of a six-pulse converter bridge
Vd1 = ⋅ 2 ⋅ VLL ⋅ cos(α ) [Eqn A2.2.2b]
π
3 3
Vd = ⋅ 2 ⋅ VLL ⋅ cos(α ) − ⋅ ω ⋅ Lc ⋅ Id
[Eqn A2.2.2c]
π π
VLL2 − 0
ω ⋅ Lc = Xc p.u . ⋅
[Eqn A2.2.2d]
MVA0
MVA0 = 3 ⋅ VLL − 0 ⋅ Iac0

[Eqn A2.2.2e]
Substituting (A2.2.2e) into (A2.2.2d), gives:
VLL − 0
ω ⋅ Lc = Xc p.u . ⋅ [Eqn A2.2.2f]
3 ⋅ Iac0
Substituting (A2.2.2f) into (A2.2.2c), gives:
3 3 V
Vd = ⋅ 2 ⋅ VLL ⋅ cos(α ) − ⋅ Xc p.u . ⋅ LL − 0 ⋅ Id [Eqn A2.2.2g]
π π 3 ⋅ Iac0
2
Iac0 = ⋅ Id0 [Eqn A2.2.2h]
3
Substituting (A2.2.2h) into (A2.2.2g), gives:
3 3 V
Vd = ⋅ 2 ⋅ VLL ⋅ cos(α ) − ⋅ Xc p.u . ⋅ LL − 0 ⋅ Id [Eqn A2.2.2i]
π π 2 ⋅ Id0
Rearranging, gives:
3  1 VLL-0 Id 
Vd = ⋅ VLL  2 ⋅ cos(α ) − Xc p .u . ⋅ ⋅ ⋅ [Eqn A2.2.2j]
π  2 VLL Id0 
Assuming that the AC voltage applied to the converter is the rated AC voltage, that is, VLL equals VLL-0 and
then substituting (A2.2.2a) into (A2.2.2i), gives:
Vd Xc Id
= cos(α ) − p.u . ⋅ [Eqn A2.2.2k]
Vd0 2 Id0

CHAPTER
where Eqn A2.2.2k gives the per-unit relationship between DC voltage, DC current, firing angle and
commutating reactance (see Fig. A2.2.2b).
In Eqn A2.2.2k the operating DC voltage must always be less than the 1.0 p.u. no-load DC voltage. It is more
normal to refer to the nominal operating DC voltage as 1.0 p.u. Hence, a factor ‘k’ has to be introduced
into Eqn A2.2.2k, giving:
 Xc 
Vd p .u . = k ⋅ VL L p. u. ⋅  cos(α ) − p .u . ⋅ Id p.u . 
 2  [Eqn A2.2.2l]
Where: Vd p. u. 1
k= ⋅
VLL p.u. Xc
cos(α ) − p.u . ⋅ Id p.u.
2 [Eqn A2.2.2m]
Vd p.u.
Xc p.u. . Id
Regulation volage drop =
Vdi0 = 1.0 2
Vdi0 cos (α)
Vd
} cos (α)
1.0 Id p.u.
Fig. A2.2.2b– The per-unit relationship between no-load DC voltage, DC voltage and DC current as expressed by Eqn A2.2.2k

CHAPTER
APPENDIX
A4.1 BRIEF DEFINITION OF TERMS

Fundamental frequency
Frequency in the spectrum obtained from a Fourier transformation of a time function, to which all the
frequencies of the spectrum are referred. In general, the fundamental frequency is the same as the power
supply frequency.
Harmonic frequency
Frequency which is an integer multiple of the fundamental frequency. The ratio of the harmonic frequency
to the fundamental frequency is the harmonic order (recommended notation: “n”)
Interharmonic frequency
Any frequency which is not an integer multiple of the fundamental frequency.
Total Harmonic Distortion – THD

Ratio of the RMS value of the sum of all the harmonic components up to a specified order (N) to the RMS
value of the fundamental component.
2
Q 
N
THD = ∑  n 
Q 
n= 2 1
where:
Q represents either current or voltage
Q1 = RMS value of the fundamental component
n = harmonic order
Qn = RMS value of the harmonic component of order n
N = generally 40 or 50 depending on the application

CHAPTER
A4.2 TELEPHONE INTERFERENCE FACTORS
The performance requirements as defined in section 4.6.2 use the weighting factors p f and C n. The
description of these factors is provided in the following section.
The sensitivity of the human ear varies with frequency and standard weighting curves have been developed
to reflect this frequency dependence of the ear, the response of the telephone receiver and other factors,
and so to establish the effective noise level of the individual harmonics. In countries following European
practices, psophometric weighting (pf) is commonly used, while the C-message weighting curve (Cn) is
more in use in countries following North American practices, the difference between the two being quite
small. Fig. A4.2a shows the two curves.
10
C-message
Psophometric
0.1
0.01
100 1000 10000
FrequencyÊ(Hz)
Fig. A4.2a– C-message and psophometric weighting factors
A4.2a

CHAPTER
APPENDIX
A4.3 MATHEMATICAL BACKGROUND OF FILTERS

For the more complicated types of filters presented in this chapter, a full mathematical description is
provided in this appendix. In each case, the impedance of the filter is derived at any frequency based on
the individual filter component values: resistance, inductance and capacitance.
A4.3.1 Double-tuned Filter

The mathematical equations to describe an ideal, lossless
double-tuned filter are given in the following sections. The given
C1
parameters are:
➙ Q0 – reactive power generated by the filter at fundamental
frequency
➙ nS1 – first series resonance frequency given as a multiple L1
of fundamental frequency
➙ n S 2 – second series resonance frequency given as a
multiple of fundamental frequency
➙ nP1 – parallel resonance frequency given as a multiple of
fundamental frequency
C2 L2
The parameters to be evaluated are C1, C2, L1, L2.
The impedance of the whole filter may be expressed as:
ZF = ZC1 + ZL1 + Z(C2//L2) [Eqn A4.3.1a]
where: Fig. A4.3a– Circuit of an ideal
1 lossless double-tuned filter
Z C1 = − j
ω C1
ZL1 = jwL1
1 1 1 A4.3a
Z (C2 / / L2 ) = = = −j
1
+ jω C2 jω  C2 − 1   1 
ω  C2 − 2 
[Eqn A4.3.1b]
jω L 2  ω 2L2   ω L2 
 
1 1  1 1 
Z F = jω L1 − j −j = j  ω L1 − −  [Eqn A4.3.1c]
ω C1  1   ω C1  1 
ω  C2 − 2   ω  C2 − 2  
 ω L2    ω L2  
We define new values:
x1 = ω S21 = (2 ⋅ π ⋅ f0 ⋅ nS1 )2 [Eqn A4.3.1d]
x2 = ω S22= (2 ⋅ π ⋅ f0 ⋅ nS 2 )2 [Eqn A4.3.1e]

1
x p = ω 2p1= (2 ⋅ π ⋅ f0 ⋅ n p1 )2 = [Eqn A4.3.1f]
L2 C 2
ω s21 ⋅ ω s22 x1 ⋅ x2
k1 = = [Eqn A4.3.1g]
ω 2p1 xp
 
 1 1   1 ω2 
Z F = j ⋅  ω L1 − + ω L2 2
 = j ⋅  ω L1 − + ω L2 2 p 1 2 
 ω C1 ω   ω C1 ω p1 − ω 
 1− 2 
ω p1 
−ω 4 L1C1 + ω 2 L1C1ω 2p1 + ω 2 L2C1ω 2p1 + ω 2 − ω 2p1
= j
( )
ω 2p1 − ω 2 ω C1 [Eqn A4.3.1h]

CHAPTER
ZF = 0 if:
ω 4 L1C1 − ω 2 ( L1C1ω 2p1 + L2C1ω 2p1 + 1) + ω 2p1 = 0 [Eqn A4.3.1i]
We introduce parameter z:
z = w2
z 2 L1C1 − z( L1C1ω 2p1 + L2C1ω 2p1 + 1) + ω 2p1 = 0 [Eqn A4.3.1j]
L1C1ω 2p1 + L2C1ω 2p1 + 1 ω 2p1

z2 − z + =0 [Eqn A4.3.1k]
L1C1 L1C1
The impedance of the ideal lossless filter is equal to zero at both series resonance frequencies ns1 and ns2,
which can be expressed as:
(z – x1) ⋅ (z – x2) = 0 [Eqn A4.3.1l]
z2 – z(x1 + x2) + x1 ⋅ x2 = 0 [Eqn A4.3.1m]
Comparing equations [Eqn A4.3.1k] and [Eqn A4.3.1m], we get two equations:
L1C1ω 2p1 + L2C1ω 2p1 + 1

x1 + x2 = [Eqn A4.3.1n]
L1C1
ω 2p1
x1 ⋅ x2 = [Eqn A4.3.1o]
L1C1
From [Eqn A4.3.1o] and [Eqn A4.3.1g]:
1
k1 = [Eqn A4.3.1p]
L1C1
1
L1 = [Eqn A4.3.1q]
k1C1
From [Eqn A4.3.1n], [Eqn A4.3.1d], [Eqn A4.3.1e] and [Eqn A4.3.1g]:
1
L1C1ω 2p1 + L2C1ω 2p1 + 1 = ( x1 + x2 ) ⋅ [Eqn A4.3.1r]
k1
xp x1 + x2
C1
+ L2C1 x p + 1 = [Eqn A4.3.1s]
k1 k1
L1
x + x − x p − k1
L2 = 1 2 [Eqn A4.3.1t]
C1 x p k1
From [Eqn A4.3.1f]:
L2
1
C2 = [Eqn A4.3.1u]
x p L2
To calculate C 1 , we have to know the filter impedance at
fundamental frequency.
The filter circuit can be simplified as in Fig. A4.3b:
The reactance of this circuit is: Fig. A4.3b– Simplified circuit
for calculation of impedance at
1 1 x1 + x2 − x p − k1 fundamental frequency
X Fω 0 = − + ω + ⋅ ω 0 [Eqn A4.3.1v]
ω 0C1 k1C1 0 x p k1C1
A4.3b
CHAPTER
APPENDIX
  xx  
ω 02  x1 + x2 − x p − 1 2  
X Fω 0 =
1 
⋅  −1 +
ω
+
ω2
0
2
0 ( )
x1 + x2 − x p − k1 
=
1

ω x
2
⋅ 0 p +
 xp  
−1
ω 0C1  k1 x p k1  
 ω 0C1  x1 x2 x1 x2

 
[Eqn A4.3.1w]
 2 n2 n2   2 ns21ns22 
 n p + ns21 + ns22 − n 2p − s1 2 s 2   ns 1 + ns 2 − n 2
2

1 np 1
X Fω 0 = ⋅ − 1 = ⋅ p
− 1 [Eqn A4.3.1x]
ω 0C1  ns1ns 2
2 2
 ω 0 C1  ns1ns 2
2 2

   
1  1 1 1 
X Fω 0 = ⋅ + − −1 [Eqn A4.3.1y]
ω 0C1  ns21 ns22 n 2p 
And the impedance is:
1   1 1 1 
Z Fω 0 = − j ⋅ 1−  2 + 2 − 2   [Eqn A4.3.1z]
ω 0C1   ns1 ns 2 n p  
Fundamental frequency reactive power generated by the filter will be calculated according to:
U2 1
Q0 = = U 2C1ω 0 ⋅
Z Fω 0  1 1 1 [Eqn A4.3.1aa]
1−  2 + 2 − 2 
 ns 1 ns 2 n p 
and the capacitance C1 will be:
Q0   1 1 1 
C1 = ⋅ 1 −  2 + 2 − 2   [Eqn A4.3.1bb]
U ω 0   ns 1 ns 2 n p  
2
The equation [Eqn A4.3.1bb], together with equations [Eqn A4.3.1q], [Eqn A4.3.1t] and [Eqn A4.3.1u]
constitute the solution of the filter.
A4.3.2 Triple-tuned Filter C1
The mathematical equations to evaluate an ideal lossless

triple-tuned filter are given in the following section. The given L1
parameters are:
frequency
➙ nS1 – first series resonance frequency given as a multiple C2 L2
➙ n S 2 – second series resonance frequency given as a
multiple of fundamental frequency
➙ nS3 – third series resonance frequency given as a multiple C3 L3
➙ nP1 – first parallel resonance frequency given as a multiple
➙ n P 2 – second parallel resonance frequency given as a Fig. A4.3c– Circuit of an ideal
multiple of fundamental frequency lossless triple-tuned filter
The parameters to be found are C1, C2, C3, L1, L2 and L3.

CHAPTER
The impedance of the whole filter may be expressed as:
ZF = ZC1 + ZL1 + Z(C2//L2) + Z(C3//L3) [Eqn A4.3.2a]
where:
1
Z C1 = − j
ω C1
ZL1 = jwL1
1 1 1
Z (C2 / / L2 ) = = = −j
1  1   1  [Eqn A4.3.2b]
+ jω C2 jω C2 − ω  C2 − 2 
jω L 2  ω 2L2   ω L2 
1 1 1
Z (C3 / / L3 ) = = = −j
1
+ jω C3 jω  C3 − 1  
ω  C3 − 2 
1  [Eqn A4.3.2c]
jω L 3  ω 2L3   ω L3  
1 1 1  1 1 1 
Z F = jω L1 − j −j −j = j  ω L1 − − − 
ω C1  1   1   ω C1  1   1 
ω  C2 − 2  ω  C3 − 2   ω  C2 − 2  ω  C 3 − 2  
 ω L2   ω L3    ω L2   ω L3  
 
1  1 1 1 
−j = j  ω L1 − − −  [Eqn A4.3.2d]
  1   ω C1  1   1 
 ω  C3 − 2   ω  C2 − 2  ω  C 3 − 2  
 ω L3    ω L2   ω L3  
x1 = ω S21 = (2 ⋅ π ⋅ f0 ⋅ nS1 )2 [Eqn A4.3.2e]
x2 = ω = (2 ⋅ π ⋅ f0 ⋅ nS 2 )
2
S2
2
[Eqn A4.3.2f]
x3 = ω = (2 ⋅ π ⋅ f0 ⋅ nS 3 )
2
S3
2
[Eqn A4.3.2g]
1
k2 = ω 2p1= (2 ⋅ π ⋅ f0 ⋅ n p1 )2 = [Eqn A4.3.2h]
L2 C 2
1
k3 = ω 2p 2= (2 ⋅ π ⋅ f0 ⋅ n p 2 )2 = [Eqn A4.3.2i]
L3C3
ω s21 ⋅ ω s22 ⋅ ω s23 x1 x2 x3
k1 = = [Eqn A4.3.2j]
ω 2p1 ⋅ ω 2p 2 k2 k 3

Z F = j  ω L1 −
1 ω2 ω2 
+ ω L2 2 p 1 2 + ω L 3 2 p 2 2  = j
( )( ) ( )
ω 2 L1C1 ω 2p1 − ω 2 ω 2p 2 − ω 2 + ω 2 L2C1ω 2p1 ω 2p 2 − ω 2 + ω 2 L3C1ω 2p 2 ω 2p1 − ω (
 ω C1 ω p1 − ω ω p2 − ω  (
ω 2p1 − ω 2 ω 2p 2 − ω 2 ω C1 )( )
ω 2p1
+ ω L3 2 p 2 2  = j
( 2
)(2 2
) 2 2 2
( 2 2
) 2
( 2 2 2
) (
ω 2  ω L1C1 ω p1 − ω ω p 2 − ω + ω L2C1ω p1 ω p 2 − ω + ω L3C1ω p 2 ω p1 − ω − ω p1 − ω ω p 2 − ω [Eqn A4.3.2k]
2 2 2 2 2
)( )
2
ω −ω
p1
2
ω p2 − ω  2 2 2 2
(
ω p1 − ω ω p 2 − ω ω C1 )( )
ZF = 0 if:
( )
ω 6 L1C1 − ω 4 ( L1C1ω 2p1 + L1C1ω 2p 2 + L2C1ω 2p1 + L3C1ω 2p 2 + 1) + ω 2 L1C1ω 2p1ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2 − ω 2p1ω 2p 2 = 0
( )
C1ω 2p1 + L3C1ω 2p 2 + 1) + ω 2 L1C1ω 2p1ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2 − ω 2p1ω 2p 2 = 0 [Eqn A4.3.2l]
We introduce parameter z:
z = w2
( )
z 3 L1C1 − z 2 ( L1C1ω 2p1 + L1C1ω 2p 2 + L2C1ω 2p1 + L3C1ω 2p 2 + 1) + z L1C1ω 2p1ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2 − ω 2p1ω 2p 2 = 0
(
C1ω 2p1 + L3C1ω 2p 2 + 1) + z L1C1ω 2p1ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2 )−ω 2
p1 ω 2p 2 = 0 [Eqn A4.3.2m]

CHAPTER
APPENDIX
z3 − z2
( L1C1ω 2p1 + L1C1ω 2p 2 + L2C1ω 2p1 + L3C1ω 2p 2 + 1)
+z
(L C ω
1 1
2
ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2
p1 )−
L1C1 L1C1
C1ω 2p1 + L1C1ω 2p 2 + L2C1ω 2p1 + L3C1ω 2p 2 + 1)

+z
( −
)
L1C1ω 2p1ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2 ω 2p1ω 2p 2
[Eqn A4.3.2n]
L1C1 L1C1 L1C1
The impedance of the ideal lossless filter is equal to zero at all three series resonance frequencies ns1, ns2
and ns3 which can be expressed as:
(z – x1) ⋅ (z – x2) ⋅ (z – x3) = 0 [Eqn A4.3.2o]
z – z (x1 + x2 + x3) + z(x1x2 + x3x2 + x1x3) – x1x2x3 = 0
3 2
[Eqn A4.3.2p]
Comparing [Eqn A4.3.2n] and [Eqn A4.3.2p] we get three equations:
( L1C1ω 2p1 + L1C1ω 2p 2 + L2C1ω 2p1 + L3C1ω 2p 2 + 1)
x1 + x2 + x3 = [Eqn A4.3.2q]
L1C1
x1 x2 + x3 x2 + x1 x3 =
(L C ω
1 1
2
ω 2p 2 + L2C1ω 2p1ω 2p 2 + L3C1ω 2p1ω 2p 2 + ω 2p1 + ω 2p 2
p1 ) [Eqn A4.3.2r]
L1C1
ω 2p1ω 2p 2
x1 x2 x3 = [Eqn A4.3.2s]
L1C1
From [Eqn A4.3.2j] and [Eqn A4.3.2s]:

1
L1 = [Eqn A4.3.2t]
k1C1
From [Eqn A4.3.2h], [Eqn A4.3.2i] and [Eqn A4.3.2r]:

x1 x2 + x3 x2 + x1 x3 k2 k3 k2 k3 L2 k2 k3 L3
= + ⋅ + ⋅ + k2 + k 3 [Eqn A4.3.2u]
k1 k1 k1 L1 k1 L1
From [Eqn A4.3.2h], [Eqn A4.3.2i] and [Eqn A4.3.2q]:

L2 L
x1 + x2 + x3 = k1 + k2 + k3 + k2 + k3 3 [Eqn A4.3.2v]
L1 L1
L2k2 + L3k3 = L1(x1 + x2 + x3 – k1 – k2 – k3) [Eqn A4.3.2w]
From [Eqn A4.3.2u]:

L1
L2 + L 3 =
k2 k 3
( x1x2 + x3 x2 + x1x3 − k1k2 − k2 k3 − k3k1 ) [Eqn A4.3.2x]
From equations [Eqn A4.3.2w] and [Eqn A4.3.2x] we can find the relationship between L2 and L1 and also
between L3 and L1. After introducing equation [Eqn A4.3.2t], L2 and L3 can be expressed as:
x1 x2 + x2 x3 + x3 x1 − x1k2 − x2 k2 − x3k2 − k3k1 + k22

L2 = [Eqn A4.3.2y]
k1k2C1 ( k3 − k2 )
x1 x2 + x2 x3 + x3 x1 − x1k3 − x2 k3 − x3 k3 − k2 k1 + k32
L3 = [Eqn A4.3.2z]
k1k3C1 ( k2 − k3 )
The capacitances C2 and C3 can now be calculated from [Eqn A4.3.2h] and [Eqn A4.3.2i]:
1
C2 = [Eqn A4.3.2aa]
k 2 L2
1
C3 = [Eqn A4.3.2bb]
k 3 L3
CHAPTER
To calculate C1, we have to know the filter impedance at fundamental
frequency.
C1
The filter circuit can be simplified as in Fig. A4.3d:
In the same way as for the double-tuned filter, the impedance of the L1
filter for the fundamental frequency w0 will be:
1   1 1 1 1 1 
Z Fω 0 = − j ⋅  1 −  2 + 2 + 2 − 2 − 2   [Eqn A4.3.2cc]
ω 0C1   ns1 ns 2 ns 3 n p1 n p 2   L2
Fundamental frequency reactive power generated by the filter will be

calculated according to:
U2 1 L3
Q0 = = U 2C1ω 0 ⋅ [Eqn A4.3.2dd]
Z Fω 0  1 1 1 1 1 
1−  2 + 2 + 2 − 2 − 2 
 ns 1 ns 2 ns 3 n p1 n p 2 
and the capacitance C1 will be: Fig. A4.3d– Simplified circuit
for calculation of impedance
Q0   1 1 1 1 1  at fundamental frequency
C1 = ⋅  1 −  2 + 2 + 2 − 2 − 2   [Eqn A4.3.2ee]
U ω 0   n s 1 n s 2 n s 3 n p1 n p 2  
2
The equation [Eqn A4.3.2ee] together with equation [Eqn A4.3.2t], and equations [Eqn A4.3.2y] to [Eqn
A4.3.2bb], constitute the solution of the triple tuned filter.
A4.3.3 Single-tuned Band-pass Filter with Parallel Capacitor

The mathematical equations to evaluate an ideal lossless single-
tuned filter with parallel capacitor are given in the following
A4.3d
C1
section.
The given parameters are:
frequency L1 C2
➙ nS – series resonance frequency given as a multiple of
➙ nP – parallel resonance frequency given as a multiple of
The parameters to be found are C1, C2, L1.
Impedance of the whole filter may be expressed as:
ZF = ZC1 + Z(C2//L1) [Eqn A4.3.3a] Fig. A4.3e– Single-tuned band-pass
filter with parallel capacitor, circuit
where:
1
Z C1 = − j [Eqn A4.3.3b]
ω C1
1 1 1
Z (C2 / / L1 ) = = = −j
1
+ jω C2 jω C2 − 1
   1 
ω  C2 − 2 
[Eqn A4.3.3c]
jω L 1  
ω L1 
2
 ω L1 
 
A4.3e
1 1  1 1 
ZF = − j −j = − j +  [Eqn A4.3.3d]
ω C1  1   ω C1  1 
ω  C2 − 2   ω C −
 2 ω 2 L  
 ω L1   1
ω 2 L1C2 − 1 + ω 2 L1C1 ω 2 ( L1C2 + L1C1 ) − 1

ZF = − j = −j [Eqn A4.3.3e]
ω C1 (ω L1C2 − 1)
2
ω C1 (ω 2 L1C2 − 1)
CHAPTER
APPENDIX

x = ω S2 = (2 ⋅ π ⋅ f0 ⋅ nS )2 [Eqn A4.3.3f]
1
x p = ω 2p = (2 ⋅ π ⋅ f0 ⋅ n p )2 = [Eqn A4.3.3g]
L1C2
ZF is = 0 if:
w2(L1C2 + L1C1) – 1 = 0 [Eqn A4.3.3h]
1
ω s2 = [Eqn A4.3.3i]
L1 ( C1 + C2 )
To calculate C1, we have to know the filter impedance at fundamental frequency.

The filter circuit can be simplified as in Fig. A4.3f.
Impedance of this circuit is: C1
 1  1   1 1 
Z F = j  ω L1 −  = −j ⋅  1 −  2 − 2   [Eqn A4.3.3j]
 ω C1  ω 0C1   ns n p  
L1
Fundamental frequency reactive power generated by the filter will be
calculated according to:
U2 1
Q0 = = U 2C1ω 0 ⋅
Z Fω 0  1 1 [Eqn A4.3.3k]
1−  2 − 2 
 ns n p 
and the capacitance C1 will be: Fig. A4.3f– Simplified circuit
for calculation of impedance at
Q0   1 1  fundamental frequency
C1 = ⋅ 1 − −  [Eqn A4.3.3l]
U 2ω 0   ns2 n 2p  
The equation [Eqn A4.3.3l], together with the equations [Eqn A4.3.3g] and [Eqn A4.3.3i] constitute the
solution of the filter.
A4.3f

CHAPTER
A4.4 BANK SIZE OPTIMIZATION
The design of the minimum filter bank size that achieves an optimum cost is described in the following
section.
In the estimation of the cost of a filter bank, the following elements should be included:
➙ equipment: ➙ protection: ➙ other
– capacitors – arresters – foundations
– reactors – CT – erection
– resistors – protection cubicles – maintenance
Some of these costs are fairly constant, such as foundations, erection, CT and protection cubicles. Out
of all the others, the highest impact on the filter bank cost is the capacitors. The cost of the main (HV)
capacitor bank is generally about 70% of the total cost of the filter equipment, including filter arresters.
In the first approximation, the bank cost optimization is reduced to the optimization of the cost of the HV
capacitor bank.
Let us make following assumptions:
AC system voltage UN = 400 kV
Optimal minimum capacitor unit size Qu = 1 MVAr (normally, biggest unit is more optimal)
Maximum available bushing creepage Lcr = 560 mm
Creepage distance requirement Kcr = 43 mm/kV
UN = 400 kV gives UNmax = 420 kV, Uphasemax = 420/√3 = 242 kV
Maximum capacitor bank voltage in the filter is normally about 10-15% higher than Uphasemax:
URcap = 242 ⋅ 1.10 = 266 kV, hence URcap = 270 kV may be assumed.
The voltage for creepage distance calculations will be lower than this, because of the different way of
adding the harmonic voltages for rating and for creepage distance calculations. A realistic assumption on
Ucreep = 250 kV.
Calculate minimum number of capacitor units in series:
U creep 250
Smin = = = 19.2 = 20
Lcr 560
K cr 43
Minimum number of parallel strings of capacitors is two, because of the need of unbalance protection in
the HV capacitor bank.
The size of one phase is:
Smin ⋅ Pmin = 20 ⋅ 2 = 40
and the installed reactive power of one phase is:
Qinst = 40 ⋅ Qu = 40 ⋅ 1 MVAr = 40 MVAr
For three phases:
Qinst3 = 3 ⋅ 40 = 120 MVAr
This installed power will generate fundamental frequency reactive power at nominal system voltage:
UN 400 2
Qgen = Qinst ⋅ = 120 ⋅ = 120 ⋅ 0.73 = 88 MVAr
UR (270 ⋅ 3 )2
The result of this example shows that for the assumptions as specified, the minimum optimal 3-phase filter
bank should not be smaller than 88 MVAr.
With other assumptions e.g. different system voltage, different unit bushing, different optimal unit size etc.,
the calculation should be performed in similar way to achieve new optimal bank size.

CHAPTER
APPENDIX
A4.5 NETWORK IMPEDANCE

When calculating the interaction between the AC filters and the AC network impedance, the requirement
from the customer is often to choose for each filter and each frequency, the worst network impedance in
such a way that resonance between the filter and the network impedance is created. The way to find the
worst network impedance for each filter configuration and each frequency is given below.
Filter Network
admittance admittance
IN
Ic IF
L1
YF YN
Fig. A4.5a– Circuit for the choice of worst network impedance
The amplification of filter current can be expressed as:
IF YF
= [Eqn A4.5a]
I C YF + YN A4.5a
Fig. A4.5b below shows the vectors of the filter and network impedances.
IM Ê(Υ) ΥF
ΨN
ΥN
ΥN
Υ +FÊ Ê
ΨF
ReÊ(Υ)
Fig. A4.5b– The vectors of the filter and network impedances

CHAPTER
At given filter admittance, the network impedance to be chosen is the one that gives the lowest value of
vector sum: YF + YN . This vector is lowest when vector YF + YN is perpendicular to vector YN .
Then:
IF 1
= [Eqn A4.5b]
IC MAX
cosψ
ψ = ψ F + ψ N − 90O [Eqn A4.5c]

CHAPTER
APPENDIX
A6.5 SWITCHING FUNCTIONS FOR SIMULATING

THE OPERATING CONDITIONS PRESENTED
IN FIG. 6.5.3b
12-pulse DC voltage without commutation, window of 30 degrees
Ripple current, due to a single side, 12-pulse, single pulse window, p.u.
iripple12 pW ( j, ta, x, f , α ,θ ) =
    6     6  
  π   π x + θ ⋅ π  π   π π     π x +θ ⋅ π   
π ⋅ 4 ⋅ cos   ⋅ sin 2 ⋅ π ⋅ f ⋅  ta j − ⋅ −   − 4 ⋅ cos   ⋅ sin

 α −   − 12 ⋅ cos (α ) ⋅ 2 ⋅ π ⋅ f ⋅  ta j − ⋅  −α
   12    6 2 ⋅ π ⋅ f  12   12  12    6 2 ⋅ π ⋅ f ⋅  

6⋅2⋅π⋅ f
Auxiliary Switching Functions
Ripple current, due to a single side, 12-pulse, multiple pulses, p.u.
12-pulse DC voltage without commutation, multiple pulses, p.u.
DC ripple current
DC ripple current, rectifier

CHAPTER
DC ripple current, inverter
iripple12 p j = iripple12 prj + iripple12 pi j DC ripple current, total
DC voltage, no overlap
DC voltage, no overlap due to the rectifier
DC voltage, no overlap due to the inverter
Lci ⋅ 4 + Ldi Lcr ⋅ 4 + Ldr

vdc12 j = vdc12 prj ⋅ − vdc12 pi j ⋅
(Lcr + Lci)⋅ 4 + Ldr + Ldi (Lcr + Lci)⋅ 4 + Ldr + Ldi
DC voltage, no overlap, pole to neutral
Minimum Id to avoid discontinuity
Principal symbols:
j vector varying from 0 to N
N maximum number of point to plot
ta simulation time in s
x number of pulses of DC current ripple
F Heaviside step function
Lcr transformer inductance, rectifier
Lci transformer inductance, inverter
Ldr smoothing reactor inductance, rectifier
Ldi smoothing reactor inductance, inverter
ar firing angle, rectifier
ai firing angle, inverter, p–g, assuming overlap angle, μ = 0
g inverter extinction angle
ELLr, pk converter transformer peak secondary voltage, rectifier
ELLi, pk converter transformer peak secondary voltage, inverter
fr fundamental frequency of the AC system, rectifier
fi fundamental frequency of the AC system, inverter
qr phase angle of the fundamental frequency voltage at the PCC (Point of Common Coupling), rectifier
qi phase angle of the fundamental frequency voltage at the PCC (Point of Common Coupling), inverter

CHAPTER
APPENDIX
A9.1 CONVERTER EQUATION MODIFICATIONS

DURING TRANSIENT ABNORMAL CONDITIONS
On initialization the model sets the DC current to the value in the load flow and calculates ac based on
the load flow values of AC and DC voltage so that the model initial conditions are the same as the load
flow values.
Under certain abnormal operating conditions, the abnormal modes are separated by a current factor which
is defined as:
2 I dc X cc
K1 = [Eqn. A9.1a]
Eacc
This factor modifies the effective delay angle.

In normal operation it can be shown that:
K1 = cos(ac) – cos(ac + mc) = cos ac + cos gc [Eqn. A9.1b]
Commutation failure occurs when gc falls below gmin and will occur when:
K1 > cos ac + cos gmin [Eqn. A9.1c]
The limit of normal operation, or mode 1, occurs when mc reaches 60 degrees. If this limit is exceeded then
a commutation will commence before the previous commutation has ceased and periods of short circuit
will occur. Substituting this value in [Eqn. A9.1b] gives normal operation when:
K1 ≤ cos (60 – ac) [Eqn. A9.1d]
During a commutation overlap the voltage across the next valve to be fired is effectively delayed by
30 degrees. Thus it is possible, if ac is less than 30 degrees and mc is tending to greater than 60 degrees,
that the voltage across the next valve to be fired will not become positive until the previous commutation
is complete or 30 degrees after the expected point, whichever is sooner, effectively delaying the firing point.
If this is the case then the effective ac , up to 30 degrees, is that which satisfies the equality in [Eqn. A9.1c].
This is described as mode 2 operation and substituting this value in [Eqn. A9.1d] gives:
K1 ≤ √3/2 [Eqn. A9.1e]
If mc exceeds 60 degrees then periods of short circuit will occur and the above equations no longer apply.
In addition, if ac is less than 30 degrees then due to the above delay process the effective ac will be 30
degrees. If the periods of short circuit are continuous, then Vdc = 0.
LIST OF SYMBOLS
ac Converter firing angle (degrees)
gc Converter extinction angle (degrees)
gmin Converter minimum extinction angle (degrees)
mc Converter overlap angle (degrees)
Eacc Converter transformer open circuit secondary AC voltage (Volts)
Idc Direct current (Amps)
Xcc Converter transformer reactance (Ohm)

CHAPTER
A9.2 SOME TECHNIQUES USED TO QUANTIFY
DYNAMIC INTERACTION
A9.2.1 Eigenderivative [14]

This aims at evaluating the first partial derivative dl/dp = {f T ⋅ (dA/dp) ⋅ e}/f T ⋅ e, where:
For dx/dt = Ax, with respect to a specific parameter ‘p’, the variation of the eigenvalues (l) is calculated as
above; ‘e’ is the right eigenvector and ‘f’ is the left eigenvector element associated with a particular state,
with some dependency on the specified parameter (emanating from elements of the state matrix).
These computed values may generally be interpreted as shown in the table below:
Value Interpretation
Large +ve Potentially harmful increase of oscillatory instability & poor damping
Large -ve Good reduction of oscillations and positive damping
Small +ve Potentially harmless small oscillatory instability due to lack of adequate damping
Small -ve Small reduction of oscillations with inadequate damping
Table A9.2a– Eigenvalues
A9.2.2
Frequency Domain Eigenanalysis
(Modal Participation Factor)
State space representation of a dynamic system is:
•
x = Ax + Bu
[Eqn. A9.2a]
y = Cx + Du
Where: x = State variable matrix

u = Input matrix
y = Output matrix
Let x = Ux,x = U–1x = Vx,x is the modal variable that transforms x from the time to the modal domain.
U is the right hand eigenvector matrix (right eigenvector)
V is the left hand eigenvector matrix (left eigenvector)
Then x = UeltVx(0) is the state variable solution in the time domain.
Modal Participation Factors (MPF):
The time domain solution:
x = UeltVx(0) may be rewritten in expanded form:
 x1   U11 U12 U13   eλ1t   V11 V12 V13   x1 (0 ) 
 x =U U U   eλ 2 t  •  V V V   x (0 ) 
 2   21 22 23
    21 22 233   2 
• •
 x3   U 31 U 32 U 33   e λ 3t   V31 V32 V33   x3 (0 ) 

 U11eλ1t U12 eλ 2 t U13eλ 3t   V11 x1 (0 ) + V12 x2 (0 ) + V13 x3 (0 ) 
   V x (0 ) + V x (0 ) + V x (0 ) 
=  U 21eλ1t U 22 eλ 2 t U 23eλ 3t   21 1 22 2 23 3

•
 U 31eλ1t U 32 eλ 2 t U 33eλ 3t   V31 x1 (0 ) + V32 x2 (0 ) + V33 x3 (0 ) 

∴ x1 (t ) =
U11e λ1t V11 x1 (0 ) + U11e λ1t V12 x2 (0 ) + U11e λ1t V13 x3 (0 )
+U12 e λ 2 t V21 x1 (0 ) + U12 e λ 2 t V22 x2 (0 ) + U12 e λ 2 t V23 x3 (0 )
+U13e λ 3t V31 x1 (0 ) + U13e λ 3 t V32 x2 (0 ) + U13e λ 3t V33 x3 (0 )
and x2 (t ) = U 21e λ 1t V11 x1 (0 ) + U 21e λ 1t V21 x2 (0 ) + .... [Eqn. A9.2b]

CHAPTER
APPENDIX
The principal influence of state ‘i’ on mode ‘k’ is given by the term whose coefficient contains only (right
and left) eigenvector elements ‘ik’ and ‘ki’ (respectively).
Thus from the above, the principal influence of state 1 on mode 1 is indicated directly by the product (U11
⋅ V11), on mode 2 by (U12 ⋅ V21), of state 2 on mode 1 by (U21 ⋅ V12) and so on.
In general, the principal contribution of state ‘i’ on mode ‘k’ is given by the modal participation factor:
MPF(i,k) = Uik ⋅ Vki [Eqn. A9.2c]
The above per unit participation factors are normalized to the largest value. So if the natural participation
factors are p1, p2…pk, then:
p1 + p2 + … + pk = 1
Assuming p1 is the largest natural per unit participation factor, the normalized factors that appear above
are:
p1n = p1/p1 =1, p2n = p2/p1, … pkn = pk/p1
Hence p1 + p1(p2n + p3n + … pkn) = 1
giving p1 = 1/{1 + (p2n + p3n + … pkn)}
and therefore: p2 = p2n ⋅ p1, p3 = p3n ⋅ p1, … pk = pkn ⋅ p1. [Eqn. A9.2d]
A9.2.3 Prony Analysis

(a) Theory
A typical oscillatory system with a second order transfer function producing complex conjugate eigenvalues
may generally be expressed as:
R(s) = k/(s2 + as + b) [Eqn. A9.2e]
which may be rewritten as:
R(s) = k/(s2 + 2dws + w2)
where:
a = 2dwb = w2
d is the damping factor (or ratio),
w is the system natural (resonant) frequency.
The poles of a transfer function being the same as the eigenvalues of its state matrix, hence consider the
poles of s2 + as + b = 0, as being s = –s ± jb, b = 2pf.
Here, the complex conjugate pole (s) should be considered to be the dominant (potentially unstable)
eigenvalue for MPF > 0.5 (below).
It may be readily verified from the roots of the denominator of Eqn A9.2e:
d = σ 2 / (σ 2 + β 2 ) [Eqn. A9.2f]
Thus if d = 0.05, β / σ = (1 − d 2 ) / d 2 = 1 / d 2 − 1 = 19.975

Therefore, s/b = 0.05, and s × 2p/b = 0.3146
For oscillatory behavior, a parameter ‘Successive Positive Peak Ratio’ (SPPR) related to damping may be
defined:
SPPR = e–s ⋅ 2p/b = e–0.3146 = 0.73 [Eqn. A9.2g]
Obviously SPPR = 1 means s/b = 0 or s = 0
i.e. d = 0, indicating zero damping i.e., critical instability
SPPR > 1 means s/b < 0,
i.e. d < 0, indicating negative damping i.e., underdamped instability
(b) General Study Methodology
For the most severe steady state case without a damping controller (at perfect steady-state),

CHAPTER
perform a small perturbation analysis and assess Modal Participation Factors ( MPF) for key identified
states, to indicate their ‘contribution’ towards unstable or potentially unstable modes of oscillation.
Those unstable or potentially unstable (‘poorly’ damped) modes of oscillation will be listed, for
which the modal participation factors are > 0.5 p.u. (normalized to the largest). Thus, if state number
x contributes to unstable mode number y with an MPF > 0.5, then its eigenvalue is identified as
ly1 = sy1 + jwy1.
Next, repeat the exercise with a damping controller and again list the MPFs for the same state. The previous
poorly damped mode ly1 with MPF > 0.5 may now be identified as ly2 = sy2 + jwy2, with MPF > 0.5.
For damping assessments: calculate d1, SPPR1 (corresponding to ly1) and d2, SPPR2 (corresponding to
ly2); d2 should be greater than d1, SPPR2 should correspondingly be less than SPPR1.
A9.2.4 Time Domain Full Model Analysis

Certain aspects of interaction analysis may be deemed to involve steady-state and transient/dynamic
simulation of a wider nature. It is therefore appropriate to briefly list here the studies, modeling and study
methods that may generally be necessary to cover this broad area of investigation:
➙ Load Flow – full system representation, fundamental frequency RMS positive sequence only, PSS/E
or similar program
➙ Transient & Dynamic Stability – functional equivalent dynamic models of all AC and other (DC, SVC,
STATCOM, etc.) controllers to augment the above steady state AC system
➙ Controls Dynamic Performance – detailed controls and installation model on EMTP/EMTDC program,
with a reduced equivalent AC system capturing key electromagnetic and electromechanical
behavior, close to the installation only, including:
a) Controls Performance & Verification of Characteristics– EMTDC/RTDS studies
b) Controls Strategy Development – EMTDC/MATLAB with simple AC voltage source model
➙ Dynamic Interaction encompassing:
c) Time Domain full (wider) system analysis– PSS/E dynamic simulation
d) Eigenanalysis – PSS/E or similar Linear Systems Package operating on full system dynamic model
used typically for Transient Stability study
e) S ub Synchronous Oscillations – UIF calculations (see section 9.3.2.2) followed by RTDS
assessment and mitigation using a simple unloaded radial generator AC system model
f) Fundamental Frequency Temporary Overvoltages and Generator Self Excitation – PSS/E based
analysis, similar to Transient Stability study with simple radial modeling for self excitation
assessments
➙ Transient Overvoltages – EMTDC similar to (b) above but with surge arrester and TOV strategy
modeling

CHAPTER
APPENDIX
A9.3 RAM ANALYSIS OF SIMPLE SYSTEMS

This appendix provides some example solutions for RAM of systems composed of 1, 2 and 3 elements. The
elements are defined by their failure rate l and their repair rate m.
A9.3.1 Basis for Analysis

The states of the elements/systems will be
represented as follows: State ProbabilityÊPi ÊofÊbeingÊin
number theÊstateÊi
The idea is to determine the expression of the
availability A (or inversely, the unavailability U ), DescriptionÊofÊtheÊstate
failure rate l and repair rate m of the final system.
Fig. A9.3.1a– Representation of a particular state
A system can be either available (availability A) or
of a system
unavailable (unavailability U). We conclude that U
= 1 – A.
The availability A of the system is the probability of the system being in its working state and is given by
the expression:
A= ∑ Pi [Eqn. A9.3a]
i ∈{ working states } A9.3.1a
The failure rate l of the system is given by:
f
λ= [Eqn. A9.3b]
A
where: f = ∑ Pi × ∑ Rate of departure from state i across the failure booundary
i ∈{ boundary states }
The repair rate m of the system is given by:

f
µ= [Eqn. A9.3c]
1− A
Important note: the examples included here consider that the elements constituting the system are all
rated at the same power level.
A9.3.2 One-element System

A one-element system has only two possible states
1 A
(‘Working’ and ‘Failed’) as shown in the following
diagram. When the system is in the ‘Working’ state, SystemÊworking
it can fail with a failure rate l and go to the ‘Failed’
state. When in the ‘Failed’ state, it can be repaired µ
with a repair rate m and go to the ‘Working’ state.
λ
In this case, the system is available if in the
‘Working’ state. Its availability is equal to the 2 (1Ê-ÊA)Ê
probability of being in the ‘Working’ state. It can
be proved that for a one-element system having a SystemÊfailed
constant failure rate l and a constant repair rate m,
the availability is given by: Fig. A9.3.2a– Possible states of a one-element
system
µ
A= [Eqn. A9.3d]
λ+µ
λ
Consequently: U = 1 − A =
λ+µ
A9.3.2a
CHAPTER
A9.3.3 Two-element System
In a two-element system, the possible number of states is 22 i.e. 4 as shown in the following diagram:
1 A 1 A 2Ê
µ1 BothÊelementsÊworking µ2
2 (1Ê-ÊA 1)A 2Ê λ1 λ2 3 A 1(1Ê-ÊA 2)

ElementÊ1Êfailed µ2 µ1 ElementÊ1Êworking
ElementÊ2Êworking ElementÊ2Êfailed
λ2 4 (1Ê-ÊA 1)(1Ê-ÊA 2) λ1
BothÊelementsÊfailed
Fig. A9.3.3a– Possible states of a two-element system
where the availability of the elements 1 and 2 are defined as:

µ1 µ2 State number Probability Pi of being in state i
A1 = and A2 = [Eqn. A9.3e]
λ1 + µ1 λ2 + µ 2 1 A1A2
There are two possible operating cases: 2 (1 – A1)A2
➙ Case 1: the system fails if one of the two 3 A1(1 – A2)
A9.3.3a
elements fails. This case is equivalent to
4 (1 – A1)(1 – A2)
having the two elements in series or to
having the two elements in parallel, each Table A9.3.3a– State probabilities of the system
one rated at a power level less than 100%
(in which case, no element is redundant).
➙ Case 2: the system fails if and only if the
two elements fail. This case is equivalent to having the two elements in parallel with one being
redundant (classic definition of parallel systems). In that case, the two elements are rated at 100%
power or more.
Fig. A9.3.3b shows where the failure boundary is depending on the case.
CaseÊ1Ê CaseÊ2Ê
1 1
FailureÊboundary
2 3 2 3
FailureÊboundary
4 4
Fig. A9.3.3b– Possible positions of the failure boundary for a two-element system

CHAPTER
APPENDIX
The results obtained for the two different cases are as follows:
Case 1 Case 2
Number of states in which the
1 3
system is working
Probability of being in the
P1 P1 + P2 + P3
working state (availability)
Number of states in which
3 1
the system is failed
Probability of being in
P2 + P3 + P4 P4
the failed state
Availability of the system m1m2 m1m2 + l1m2 + l2m1

As = (l1 + m1)(l2 + m2) (l1 + m1)(l2 + m2)
Failure rate of the system (l1 + l2) P1 l2P2 + l1P3 l1l2 (m1 + m2)
= l1 + l2 =
l= As As m1m2 + l1m2 + l2m1
Repair rate of the system (l1 + l2) P1 m1m2 (l1 + l2) l2P2 + l1P3
= = m1 + m2
m= 1 – As l1l2 + l1m2 + l2m1 1 – As
Table A9.3.3b– Useful results for a two-element system
In the particular case where the two elements

are identical, it can be stated that states 2 and 2
1 AÊ
3 in Fig. A9.3.3a are identical. States 2 and 3 can
then be put together in one state and the entry BothÊelementsÊworking
and departure rates are then changed accordingly. µ
This method is very useful when the number of 2λ
elements in a system becomes high because it 2 2(1Ê-ÊA)AÊ
decreases the number of states and then simplifies
OneÊelementÊworking
the calculation. OneÊelementÊfailed
2µ
λ
2
3 (1Ê-ÊA) Ê
BothÊelementsÊfailed
Fig. A9.3.3c– Possible states of a two-identical

element system
Case 1 Case 2
P1
A9.3.3c P1 + P2
Availability of the system m2 m (m + 2l)

As = (l + m)2 (l + m)2
Failure rate of the system 2lP1 lP2 2l2

= 2l =
l= As As m + 2l
Repair rate of the system 2lP1 2m2 lP2

= = 2m
m= 1 – As 2m + l 1 – As
Table A9.3.3c– Useful results for a two-identical element system

CHAPTER
A9.3.4 Three-element System
In a three-element system, the possible number of
states is 23 i.e. 8. However, in order to simplify the 1 AÊ
3
calculation here, the three elements are selected 0ÊoutÊofÊ3Êfailed

to be identical and the number of states therefore µ
reduces to 4, as shown in the following diagram 3λ
(Fig. A9.3.4a). 2 2
3A Ê(1Ê-ÊA)Ê
Where the availability of the single elements are 1ÊoutÊofÊ3Êfailed
defined as: 2µ
2λ
µ
A= [Eqn. A9.3f] 3 3A(1Ê-ÊA) Ê
2
λ+µ
2ÊoutÊofÊ3Êfailed
3µ
State number Probability Pi of being in state i
λ
1 A3 4 3
(1Ê-ÊA) Ê
2
2 3A (1 – A) 3ÊoutÊofÊ3Êfailed
3 3A(1 – A)2
Fig. A9.3.4a– Possible states of a system
4 (1 – A)3 consisting of three-identical element
Table A9.3.4a– State probabilities of the system
There are three possible operating cases:

➙ Case 1: the system fails if one of the three elements fails. This case is equivalent to having the three
elements in series or to having the three elements in parallel, each one rated at a power level lower
A9.3.4a
than 50% (in that case no element is redundant)
➙ Case 2: the system fails if any two or more elements fail. This case is equivalent to having the three
elements in parallel with one being redundant. In that case, the three elements have to be rated at
50% power or more and less than 100%.
➙ Case 3: the system fails if and only if the three elements fail. This case is equivalent to having the
three elements in parallel with two being redundant (classic definition of parallel systems). In that
case, the three elements have to be rated at 100% power or more.
CaseÊ1Ê CaseÊ2Ê CaseÊ3Ê
1 1 1
FailureÊboundary
2 2 2
FailureÊboundary
3 3 3
FailureÊboundary
4 4 4
Fig. A9.3.4b– Possible positions of the failure boundary for a three-element system

CHAPTER
APPENDIX
The results obtained for the three different cases are as follows:
Case 1 Case 2 Case 3
1 2 3
the system is working
P1 P1 + P2 P1 + P2 + P3
3 2 1
the system is failed
Probability of being in
P2 + P3 + P4 P3 + P4 P4
the failed state
Availability of the system m3 m2 (m + 3l) m (m2 + 3lm + 3l2)

A= (l + m)3 (l + m)3 (l + m)3
Failure rate of the system 6l2 3l3

3l m + 3l m2 + 3lm + 3l2
l=
Repair rate of the system 3m3 6m2

3m2 + 3lm + l2 3m + l 3m
m=
Table A9.3.4b– Useful results for a three-identical element system

The state diagram for a three-element system with three different elements would be as follows.
1 A1A2A3Ê
µ1 Working Working Working µ3
µ2
λ2
2 (1-A 1 )A 2 A 3 Ê λ1 3 A 1 (1-A 2 )A 3 Ê λ3 3 A 1 A 2 (1-A 3 )
Failed Working Working µ2 Working Failed Working µ3 Working Working Failed

µ3 µ1 µ2 µ1
λ1 λ3 λ1 λ2
5 (1-A 1 )A 2 (1-A 3 )Ê λ2 6 (1-A 1 )(1-A 2 )A 3 Ê λ3 7 A 1 (1-A 2 )(1-A 3 )Ê
Failed Working Failed µ2 Failed Failed Working µ1 Working Failed Failed

µ3
λ3
λ2 8 (1-A 1 )(1-A 2 )(1-A 3 )Ê λ1

Failed Failed Failed
Fig. A9.3.4c– Possible states of a three-element system consisting of three different elements

CHAPTER A9.3.4c
CHAPTER
ACRONYMS
AC Alternating Current IRC Integrated Return Conductor

ACSR Aliminum Conductor Steel Reinforced ISE Immediate Self Excitation
ACVC AC Voltage Control
ACVCM AC Voltage Control Mode LCC Line Commutated Converter
AN Audible Noise LIPL Lightning Impulse Protective Levels
AVR Automatic Voltage Regulator LSE Load-Serving Entities
LTT Light Triggered Thyristor
BBC Brown Boveri Company LVCC Low Voltage Current Clamp
BOD Breakover Diode
BPS Bypass Switch MAP Maximum Available Power
MCAV Maximum Continuous Applied Voltage
CANA CGEE Alsthom North America MCM Thousand Circular Mills
CCP Converter Control and Protection MEB Midlands Electricity Board
CEA Constant Extinction Angle MIESCR Multi Infeed Effective Short Circuit Ratio
CEGB Central Electricity Generating Board (UK) MIIF Multiple Infeed Interaction Factor
CESCR Critical Equivalent Short Circuit Ratio MPC Master Power Control
CGC Converter Group Control MPC Maximum Power Curve
CGEÉ Compagnie Générale d’Entreprises Electriques MRTB Metallic Return Transfer Breaker
CID Converter Isolation Detector MSCDN Mechanically Switched Damped Capacitor
CIGRÉ International Council on Large Electric Systems Network
CL Corona Losses MVDC Medium Voltage DC for voltages at/or below 100
CT Current Transformer kVdc
CTC Continuously Transposed Conductor MVU Multiple Valve Unit
CVT Capacitive Voltage Divider
NBGS Neutral Bus Grounding Switch
DC Direct Current NBS Neutral Bus Switch
DCCT DC Current Transducer NISE Non Immediate Self Excitation
DPS Dynamic Performance Studies
DRPS Dynamic Reserve Power Sharing OLTC On Load Tapchanger
DRS Dynamic Rating System
DTS Distributed Temperature Sensing (system) PCAV Peak Continuous Applied Voltage
PDO Power Demand Override
EdF Electricité de France PEX Cross-Linked Polyethylene
EDS Every Day Stress PGCIL Power Grid Corporation Of India Ltd
EMF Electro-Motive Force PhC Phase Control
EPC Engineer, Procure, Construct POD Power Oscillation Damping
EPRI Electric Power Research Institute in the USA
ERTB Earth Return Transfer Breaker RAM Reliability, Availability and Maintainability
ESCR Effective Short Circuit Ratio RI Radio Interference
ETT Electrically Triggered Thyristor RDC Remote Dispatch Centre
ROW Right-Of-Way
FAT Factory Acceptance Test RPC Reactive Power Controller
FC Fixed Capacitor RPE Reactive Power Exchange
FEA Finite Element Analysis RTDS Real Time Digital Simulator
FFTOV Fundamental Frequency Transient Overvoltages RTO Regional Transmission Operator
FO Fiber Optic RTS Rated Tensile Stress
FOR Forced Outage Rate
FPGA Field Programmable Gate Array SAT Site Acceptance Test
SCADA Supervisory Control and Data Acquisition
GCCIA Gulf Cooperation Council Interconnection SCFF Self Contained Fluid Filled
Authority SCL Short Circuit Level
GE General Electric SCR Short Circuit Ratio
GEC General Electric Company (UK) SIPL Switching Impulse Protective Levels
GW GigaWatts SIWL Switching Impulse Withstand Level
SNR Signal to Noise (Ratio)
HVDC High Voltage Direct Current SPLTT Self-Protected Light Triggered Thyristor
SPS Special Protection Scheme
ICNIRP International Council on Large Electric Systems SSDC Sub Synchronous Damping Control
IDMT Inverse Definite Minimum Time SSO Subsynchronous Oscillations
IHRF Inter-Harmonic Reduction Factor STATCOM Static Synchronous Compensation

STIPL Steep-Front Impulse Protective Level
SVC Static Var Compensator
TCR Thyristor Controlled Reactor

TG Turbine Generator
TOV Temporary OverVoltage
TSC Thyristor Switched Capacitors
UHVDC Ultra High Voltage DC for voltages at/or above 600

kVdc
UTS Ultimate Tensile Stress
VBE Valve Base Electronics

VBO Voltage Breakover
VDCOL Voltage Dependent Current Order Limit
VSC Voltage Source Converter
VTE Valve Test Equipment
XLPE Cross Linked Polyethelyne
ZFCT Zero Flux Current Transformers
The authors also wish to acknowlege that the following acronyms are registered trademarks and in many cases utilized as
accepted industry standard names.
ATP®
PSCAD®
PSCAD/EMTDC®
RTDS®
PSS/E®
EMTP®
EuroStag®
ACKNOWLEDGEMENTS
The writers wish to express grateful thanks to:
Infineon Bipolar GmbH & Co Kg
Nexans
for authorizing us to use certain photographs in this book.
Jorge-Fidel Alvarez (photographer)
Other photographs: GE Vernova

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES
No information, recommendation, advice or other content included in
this book, shall constitute warranties of any kind (either expressly or
implied), or commitments that are not expressly stated in an applicable
written contract.
All information and technical data herein are believed to be valid at the
moment of the publication and its use is under the sole responsibility of
the reader and he should independently evaluate the accuracy and the
completeness of information and the usefulness to their particular needs of
any product or services. Technical data are often only approximate values.
GE Vernova reserves the right to modify, change and update the technical
data and information contained in this book without notice.
In no case whatsoever GE Vernova and the people that have participated

in the preparation of this book, shall be liable under, or in connection with,
this book or its contents, towards any person or entity in view of any
damages or losses. In particular GE Vernova shall not be liable for any loss
of profit, loss of production, loss of contracts, loss by reason of business
interruption, loss of revenue earnings, loss of data, cost of capital, special,
incidental, punitive, indirect or consequential loss or damages whatsoever
whether suffered by the readers or by any third party.
This book covers the history DC Transmission Systems:

and innovations in High Voltage Line Commutated Converters
D i re c t C u r re n t t e c h n o l o g y was designed and written to
since the early days of Power strengthen the knowledge base
Electronics in the 1960s. It of the men and women who
i s b a s e d o n t h e c o l l e c t i ve will be designing and building
experience and expertise of GE the transmission systems of
Vernova’s technical engineers, the future. Faced with the 21st
most of whom work at our century’s constantly evolving
worldwide competence centre energy challenges, High Voltage
in Stafford, UK. The book is Direct Current will play a major
complemented by writings role, permitting more efficient
from global experts at leading energy trading and transmission
companies and universities. and t he interconnection of
renewable energies into
networks worldwide.
TOC
GE Vernova
The Lord Nelson Building
Redhill Business Park
William Bagnall Drive
Stafford, ST16 1WS
ISBN 978-0-9809331-2-3
51150
GEGridSolutions.com
DC Transmission Systems: Line Commutated Converters.
9 780980 933123
© GE Vernova, 2023
GEA-33155(E) PRICE: $115 USD

HVDC LCC Reference Book

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HVDC LCC Reference Book

Uploaded by

Copyright:

Available Formats

DC Transmission Systems

All rights reserved. No part of this book may be reproducted or transmitted

4 | DC Transmission Systems: Line Commutated Converters

Dr. Mohamed Rashwan

DC Transmission Systems: Line Commutated Converters | 5

Mark A. Reynolds, P.E.

6 | DC Transmission Systems: Line Commutated Converters

DC Transmission Systems: Line Commutated Converters | 7

Carl BARKER Alban JENKINS

Georg BALOG (Nexans - retired)

The team wishes to thank the following people:

8 | DC Transmission Systems: Line Commutated Converters

Uno Lamm High Voltage Direct Current Award

GE Vernova Uno Lamm Award Recipients:

DC Transmission Systems: Line Commutated Converters | 9

10 | DC Transmission Systems: Line Commutated Converters

DC Transmission Systems: Line Commutated Converters | 11

12 | DC Transmission Systems: Line Commutated Converters

DC Transmission Systems: Line Commutated Converters | 13

TOC DC Transmission Systems: Line Commutated Converters | 15

1.1. GE VERNOVA HVDC: PAST TO PRESENT.......................... 17

1.2. GE VERNOVA INNOVATIONS................................................ 20

1.3. GE VERNOVA HVDC EXPERIENCE...................................... 37

1.1.1. EARLY HVDC TRANSMISSION

1.1.2. A NEW INDUSTRY IS BORN

BACK TO DC Transmission Systems: Line Commutated Converters | 17

1.1.3. THE DC TRANSMISSION DEPARTMENT

1.1.4. TESTING HVDC VALVES

BACK TO DC Transmission Systems: Line Commutated Converters | 19

1.1.5. A GLOBAL ACTIVITY

1.2. GE VERNOVA INNOVATIONS

BACK TO 20 | DC Transmission Systems: Line Commutated Converters

1.2.1. THE KINGSNORTH HVDC TRANSMISSION PROJECT

1.2.1.1. ARAG/4 Mercury-arc Valve

BACK TO DC Transmission Systems: Line Commutated Converters | 21

1.2.1.2. Station Layout and Main Equipment

1.2.1.3. Phase-Locked Oscillator Control System

BACK TO 22 | DC Transmission Systems: Line Commutated Converters

1.2.2. THE NELSON RIVER HVDC SCHEME

1.2.2.1. The ARBJ/6 Mercury-arc Valve

1.2.2.2. Damping Controls

BACK TO DC Transmission Systems: Line Commutated Converters | 23

1.2.3. SYNTHETIC TEST CIRCUIT FOR HVDC VALVES

BACK TO 24 | DC Transmission Systems: Line Commutated Converters

Fig. 1.2d– Basic principles of a synthetic test circuit

2018 GE Vernova commissioned a brand new test facility [14] based

1.2.4. THYRISTOR VALVE DEVELOPMENT

BACK TO DC Transmission Systems: Line Commutated Converters | 25

1.2.4.2. The H200 Series Air-Cooled

BACK TO 26 | DC Transmission Systems: Line Commutated Converters

BACK TO DC Transmission Systems: Line Commutated Converters | 27

BACK TO 28 | DC Transmission Systems: Line Commutated Converters

1.2.4.3. GE Vernova-developed air-cooled thyristor valves

1.2.4.4. The EPRI Gas-Insulated HVDC demonstrator

BACK TO DC Transmission Systems: Line Commutated Converters | 29

Fig. 1.2l– A H300 valve module

BACK TO 30 | DC Transmission Systems: Line Commutated Converters

1.2.5. MULTI-TERMINAL HVDC

1.2.6. MCNEILL BACK-TO-BACK CONVERTER STATION

BACK TO DC Transmission Systems: Line Commutated Converters | 31

1.2.7. Cheju-Haenam Submarine Cable HVDC Project

BACK TO 32 | DC Transmission Systems: Line Commutated Converters

BACK TO DC Transmission Systems: Line Commutated Converters | 33

1.2.4.2. The H200 Series Air-Cooled