Professional Documents
Culture Documents
LINE COMMUTATED
CONVERTERS
GEGridSolutions.com
DC Transmission Systems: Line Commutated Converters. © GE Vernova, 2023
ISBN: 978-0-9809331-2-3
Forth edition
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.
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
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.
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:
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.
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.
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
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
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
APPENDIX 564
APPENDIX
APPENDIX RELATED TO CHAPTER 2................................................................564
APPENDIX RELATED TO CHAPTER 4................................................................572
APPENDIX RELATED TO CHAPTER 6................................................................584
APPENDIX RELATED TO CHAPTER 9................................................................586
ACRONYMS..................................................................................................................596
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.
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.
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.
Fig.1.2a– One of the Kingsnorth valve halls showing the English Electric ARAG/4 4-anode mercury-arc valves in service
Fig. 1.2b– A controllable reactor used in a "self-tuning" filter, part of the Kingsnorth scheme
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.
HighÊcurrent HighÊvoltage
circuit circuit
V1
TestÊvalve
andÊlocal
circuit
Fig. 1.2i– H210 thyristor valves at the Sellindge converter station of the Cross-Channel HVDC scheme
Fig. 1.2j– Air-cooled thyristor valves for Les Mandarins converter station
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.
Fig. 1.2n– Suspended H400 quadrivalve for the Konti-Skan Pole 1 replacement project
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.
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.
Fig. 1.2q– A converter in the Cheju-Haenam HVDC refurbishment project using the updated H450 valve.
Bulk power transfer between independent systems of Sardinia, Corsica & Italy by overhead line (292 km) and submarine
cable (121 km)
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.
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).
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.
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.
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.
DC Reactor Rating 0.06H at 2000A
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
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
evaporative coolers.
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.
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.
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 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 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.
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.
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 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.
The Converter Transformers are 1-phase, 3-winding units, with the following rating data:
➙ 134.2 MVA, 315±6%/55/55 kV
➙ 134.2 MVA, 345±6%/55/55 kV
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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).
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.
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.
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.
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.
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.
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.
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.
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.
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.
Bulk power transfer from generation resources on the east coast of Korea into the Seoul metropolitan area by 220 km
of land cable.
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.
Fig. 2.1a– Comparison of an AC and a DC overhead line tower for comparable power transmission
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
AC SYSTEM
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
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.
BACK TO 74 | DC Transmission Systems: Line Commutated Converters
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
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
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
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).
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
Firing delay angle
ec
1 3
ec
ea
Step 1 Vd
ea
Vd
eb
ec
Giving A
ea
Vd
B
eb
A2 A1
ec
ea
Step 2 Vd
eb
1
2⋅π
A1 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫3
⋅ π3 [Eqn 2.2.1h]
1 3
π
A2 = 2 ⋅ 3 ⋅ Em ⋅ sin(ω ⋅ t ) d (ω ⋅ t )
π ∫α
⋅ [Eqn 2.2.1i]
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.
Ld → ∞ Id
ea T1 T3 T5
Lc ia
eb
Lc ib
Vd
ec
Lc ic
T4 T6 T2
Id
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
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
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:
As 1.0 p.u. current is the DC current, Eqn 2.2.2g can be re-written as:
2 ⋅ ω ⋅ Lc
µ = cos −1 cos(α ) − ⋅ Id − α [Eqn 2.2.2i]
2 ⋅ ELL
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
3
Vd = ⋅ 2 ⋅ 3 ⋅ Em ⋅ (cos(α ) + cos(α + µ )) [Eqn 2.2.2m]
2⋅π
1
Vd = ⋅ Vdi 0 ⋅ (cos(α ) + cos(α + µ )) [Eqn 2.2.2n]
2
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.
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.
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 ⋅ 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]
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.
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
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.
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
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
BACK TO DC Transmission Systems: Line Commutated Converters | 87
CHAPTER
2| HVDC CONVERTER THEORY
Firing
30° phase shift sequence
= 11 oʼclock
12
11 ea - 30° 1 Direction of rotating
phase vector
-eb ea
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.
L2
Vemf Xsys
L1
Vd
L2
Commutating
bus
Valve winding
impedance
AC harmonic
filter
Negative common
impedance
40 No common
impedance
20
-20
-40
-60
Positive common
-80 impedance
-100
-120
Negative common
impedance increases
valve voltage stresses
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,
Negative common
impedance increases
valve voltage stresses
γ > 15°
γ < 15°
γ = 15°
120
100
80
Valve voltage (%)
60
40
20
Time (electrical degrees)
0
0 50 100 150 200 250
-20
Negative common
-40 impedance
Positive common
impedance
No common
impedance
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
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°)
2 ⋅ ELL
Id = ⋅ [ sin(α + 60°) + sin(α + µ − 60°)] [Eqn 2.2.5h]
2 ⋅ ω ⋅ Lc ⋅ (1 + m )
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 ⋅ 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
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
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.
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.
Pdc (MW)
Fig. 2.3a– Calculated reactive power limits at a converter busbar for different system conditions [8]
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
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
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.
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
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.
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
Converter reactive power absorbtion (p.u.)
-0.4
-0.5
Xc = 12%
Xc = 12%
-0.6
Xc = 18%
-0.7 Xc = 18%
-0.8
DC real power (p.u.)
Fig. 2.3g– Reactive power absorption of a converter operating with constant valve winding voltage 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
-0.2
Converter reactive power absorbtion (p.u.)
Increasing γ
-0.3
Constant γ control
Increasing Xc
-0.4
-0.6
-0.7 Xc = 18%
-0.8
DC real power (p.u.)
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.
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]
0.6
AC harmonic filter
AC harmonic filter reactive switch point
0.4 power generation
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
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
100
Target DC Voltage (%)
100
DC power (%)
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
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.
3.1c
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.
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.
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
DCÊvoltage
measurement
3| HVDC STATION DESIGN
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
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
3.1.3.5. 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.
3.1.3.7. 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
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.
DCÊreactor
ACÊsystem ElectrodeÊline
Fig. 3.1p– CCC HVDC scheme with commutation capacitors on the valve side
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
1.0
MagnitudeÊ(p.u.)
0.9
0.5
0.3
T1
Tp
T2
Time
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
IEC standard wave shape
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)
IEC standard wave shape
Current 1/2 μs
Table 3.2a– Categories of electrical system overvoltage (IEC 60071-1)
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
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.
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
CREEPAGE CLEARANCE
VoltageÊriseÊtime
Precipitation
Contamination
AirÊdensityÊ(altitude)
Humidity
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.
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.
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
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
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.
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.
3.3.1.6. 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.
101.1ÊmH 5Êm
6Êm
4Êm
CapacitorÊC1 4Êm CapacitorÊC1
lowerÊpart upperÊpart
withÊassociatedÊ 6.56ʵF withÊassociatedÊ 6.56ʵF
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
117ʵF 0.42ÊmH 404 Ω
4Êm 4Êm
18Êm
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.
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.
Control BipoleÊ1
BipoleÊ2 building
ACÊfilters
ACÊfilters
CableÊroute
SVCÊ2 GIS SVCÊ1
2 x 3 ph/2 wdg
6 x 1 ph/2 wdg
Fig. 3.4b– Converter transformer arrangements
S
S + S
S
D +
D
D
S
+ VALVE
D D
HALL
S
S
+
D
D
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.
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
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
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
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
ValveÊhall
MVDCÊwallÊbushing HVDCÊharmonicÊfilter
}
DCÊswitchyard
HVDCÊwallÊbushing
DCÊsmoothingÊreactor
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
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.
ValveÊhall
DCÊsmoothingÊreactor
ControlÊbuilding
ControlÊroom
(i) (containingÊcontrolÊandÊprotectionÊpanels)
3.4mÊ(i)
(ii)
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
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)
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.
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.
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
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
4.1.2.5. Lighting
Many modern types of lighting devices such as fluorescent or discharge lights can generate harmonic currents.
5 5 2 3 4 2 2 1.8 1.4
7 4 2 9 1.2 1 4 1 0.8
ANSI/IEEE 519
Bus voltage at PCC Individual voltage distortion (%) Total voltage distortion (THD) (%)
69 kV and below 3.0 5.0
X
S = PÊ+ÊjQ
U1
InjectedÊharmonic
currentÊsource I2
Z1
SimplifiedÊACÊsystem
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.
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:
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 .
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:
Where:
n is the harmonic number.
BACK TO DC Transmission Systems: Line Commutated Converters | 167
CHAPTER
4| HARMONICS: CAUSES, CALCULATIONS AND FILTERING
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.
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)
XT
Rs = [Eqn 4.2.2r]
tan ϕ
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.
XÊ(ohms)
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.
-jX
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
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.
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
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:
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
0.07
0.06
0.05
11
HarmonicÊcurrentÊ(kAÊrms)
13
0.04 23
25
35
0.03 37
47
49
0.02
0.01
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]
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).
DCÊvoltageÊ(p.u.)
DCÊvoltageÊ(p.u.)
0.6 0.6 0.6 0.6
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
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).
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
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
N is the maximum harmonic number to be considered
U1 is the fundamental line to neutral voltage (RMS)
Where:
In is the single frequency RMS current at harmonic n
N is the maximum harmonic number to be considered
Wn = Cn ⋅ 5 ⋅ n ⋅ f0 is the single frequency TIF weighting at harmonic n
BACK TO
CHAPTER 4.7a Line Commutated Converters
DC Transmission Systems: | 185
4| HARMONICS: CAUSES, CALCULATIONS AND FILTERING
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
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]
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
Phase HarmonicÊnumber
Fig. 4.5d– Double-tuned band-pass filter, impedance characteristic
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
Phase HarmonicÊnumber
Fig. 4.5f– Triple-tuned band-pass filter tuned to 3rd, 11th and 24th harmonic, impedance characteristic
4.7f
BACK TO 188 | DC Transmission Systems: Line Commutated Converters
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
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
Phase HarmonicÊnumber
Fig. 4.5h– Single-tuned band-pass filter with parallel capacitor, impedance characteristic
4.7h
BACK TO DC Transmission Systems: Line Commutated Converters | 189
CHAPTER
4| HARMONICS: CAUSES, CALCULATIONS AND FILTERING
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
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
Phase HarmonicÊnumber
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
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
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
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
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
∆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
Sk min
n= [Eqn 4.5.4c]
Q
BACK TO DC Transmission Systems: Line Commutated Converters | 195
CHAPTER
4| HARMONICS: CAUSES, CALCULATIONS AND FILTERING
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.
4.5.5.1. 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.
However other approaches are also widely used for evaluation of rating.
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:
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
• 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)
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.
+ Z1
U SIPL CF
-
L CF
IA L R
CF - filter capacitor
LCF - inductance of the capacitor bank
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
IA - arrester (voltage dependent current source)
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.
CF
L CF
E SI
IA L R
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
FilterÊcapacitorÊunbalance
FilterÊcapacitorÊovervoltage
&ÊharmonicÊoverload
SETÊ2ÊPROTECTIONS
TOÊBUSBAR
PROTECTION FilterÊcapacitorÊunbalance
Metering
FilterÊovercurrentÊ&ÊearthÊfault
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.
Csh
2L3p
DCÊfilter DCÊfilter
Csl
L3p
Zel Zel
Cnb Cnb
Cnb Cnb
DCÊfilter DCÊfilter
Ylp Ylp
Csh
2L3p
DCÊfilter DCÊfilter
Csl
L3p
Zel Zel
Cnb Cnb
Yel Yel Yel Yel
Ylp Ylp
Zlp
Csh
2L3p
DCÊfilter DCÊfilter
Csl
L3p
Zel Zel
Cnb Cnb
L 3P = ξ ⋅ L C [Eqn 4.6e]
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
DCÊfilter DCÊfilter
LsmÊnb
Fig. 4.6d– Configuration with a split smoothing Fig. 4.6e– Configuration with a smoothing
reactor reactor split between pole and neutral bus
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.
• 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
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
c) Maximum installed reactive power per phase
N
Qinst = MAX ω 0 ⋅ C ⋅ U R2 , ∑ ω 0 ⋅ n ⋅ C ⋅ U n2 [Eqn 4.6l]
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
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
CF - filter capacitor
LCF - inductance of the capacitor bank
L, R - filter reactor and resistor
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
Pole
+
U SIPL CF Z1
-
L CF
CF - filter capacitor
LCF - inductance of the capacitor bank
L, R - filter reactor and resistor
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
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).
ACÊHARMONIC
ACÊNETWORK FILTERSÊAND PLCÊFILTERS CONVERTER/
SUBSTATION SOURCE
BUSBARS
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
Lrs
4.14c
Lru Re
Cs
LrsÊ-ÊAirÊcoreÊinductanceÊofÊtheÊsaturableÊreactor
Cd LruÊ-ÊMagnetizingÊinductanceÊofÊtheÊsaturableÊreactor
ReÊ -ÊEddyÊcurrentÊresistanceÊofÊtheÊsaturableÊreactor
CsÊ -ÊStrayÊcapacitanceÊofÊtheÊsaturableÊreactor
CdÊ -ÊDampingÊcapacitance
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
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.
ACÊlineÊnoiseÊwithÊPLCÊfilter
50
WithoutÊPLCÊfilters Limit
25
NoiseÊ(dBm)
-25
-50
0 50 100 150 200 250 300 350 400 450 500
FrequencyÊ(kHz)
4.14f
ACÊlineÊnoiseÊwithÊPLCÊfilter
50
WithÊPLCÊfilters Limit
25
NoiseÊ(dBm)
-25
-50
0 50 100 150 200 250 300 350 400 450 500
FrequencyÊ(kHz)
Converter ACÊsystem
L2A L4A
C1A C3A
C1B C3B
R1B R3B
L1B L3B
C1C C3C
R1C R3C
L1C L3C
Converter ACÊsystem
L2A L4A
C1A C3A
C1B C3B
R1B R3B
L1B L3B
C1C C3C
R1C R3C
L1C L3C
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.
ACÊHARMONIC
ACÊNETWORK FILTERSÊAND PLCÊFILTERS CONVERTER/
SUBSTATION SOURCE
BUSBARS
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.
The design requirement of the RFI shielding is that it should limit the interference appearing to the limits
specified.
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)
120 300
100 250
4.15b
RateÊofÊchangeÊofÊvoltageÊ(kV/μs)
80 200
VoltageÊ(kV)
60 150
40 100
20 50
➙ 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.
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
4.8.4.3. 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.
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.
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.)
3 m bonding to
Typically 100 mm bonding (from
Wire centers represent building
reference [23])
structure
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
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
TOC 230 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
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.
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
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
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.
(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)
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
DC Transmission Systems: Line Commutated Converters | 235
CHAPTER
5| AC/DC SYSTEM INTERACTIONS
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°.
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:
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
BACK TO 236 | DC Transmission Systems: Line Commutated Converters
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.
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
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:
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.
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.
df =
( ∆Pdc )⋅ fo ⋅ dt [Eqn 5.3c]
2 ⋅ H dc
or
H dc =
( ∆Pdc )⋅ dt ⋅ fo [Eqn 5.3d]
2 ⋅ df
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:
Idc
Xl
Xl
Vdc
Xl
[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.
αmin αmin
A A
B B
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.
B B
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)
Increasing γ Decreasing
E llÊ
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.
Vdc
OP1
OP2
OP3 CONVERTER
B
PdcÊ=Ê3P
PdcÊ=Ê2P
PdcÊ=Ê1P
CONVERTER ConstantÊÊ
A DCÊpowerÊlinesÊ
Idc
Iorder Ê1 Iorder Ê2 Iorder Ê3
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:
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
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.
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
A
CONVERTER
A
A1
A2 CONVERTERÊBÊ
ConstantÊreactiveÊ
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
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.
Vdc
CONVERTER
A
CONVERTERÊAÊ
ConstantÊreactiveÊ
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.
Vdc
CONVERTER
A A
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
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
Vdc
ConstantÊreactiveÊ
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
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
➙ 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.
Vdc
ConstantÊreactiveÊ
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
Vdc
ConstantÊreactiveÊ
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
Vdc
1.4
G ConstantÊreactiveÊ
powerÊlinesÊ
0.3Q
1.2 B
A
C 0.5Q
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
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
➙ 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.
Vdc
1.4 G
CONVERTERÊBÊ
ConstantÊreactiveÊ
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
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
5.8d
BACK TO 260 | DC Transmission Systems: Line Commutated Converters
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.
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
BACK TO 5.8e
DC Transmission Systems: Line Commutated Converters | 261
CHAPTER
5| AC/DC SYSTEM INTERACTIONS
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.
1.4 G1 CONVERTERÊAÊorÊBÊ
ConstantÊVacÊerrorÊlinesÊ
1.2 VacÊorderÊ=Ê1.2Êp.u.Ê
A B C VacÊsourceÊ=Ê1.0Êp.u.Ê VacÊconverterÊ=Ê1.0Êp.u.
1.0 G2 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
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
Vdc
5.8f
1.4
G CONVERTERÊAÊorÊBÊ
ConstantÊVacÊerrorÊlinesÊ
1.2 VacÊorderÊ=Ê1.2Êp.u.Ê
A B C
VacÊsourceÊ=Ê1.0Êp.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
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
5.8g
BACK TO DC Transmission Systems: Line Commutated Converters | 263
CHAPTER
5| AC/DC SYSTEM INTERACTIONS
Vdc
1.4
G CONVERTERÊAÊorÊBÊ
ConstantÊVacÊerrorÊlinesÊ
1.2 VacÊorderÊ=Ê1.2Êp.u.Ê
A B C VacÊsourceÊ=Ê1.0Êp.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
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
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.
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
+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
∆ 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]
(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.
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
TLineÊ1 TLineÊ1
2 2 2 TAPY
TÊ F12X TÊ F4X TÊ F8X TÊ TÊ TÊ
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
5| AC/DC SYSTEM INTERACTIONS
VDCX2
BLK
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].
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.
n+ p n+ G K
p
A
AluminiumÊmetallisation Anode
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.
p K
n G
p
PilotÊ
anode thyristor
A
Fig. 6.1 d– Thyristor with amplifying gate and cathode shorts
n n
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.
FullyÊfloatingÊsilicon BondedÊsilicon
Mo Mo Mo
Si Si Si
Mo Mo Mo
=ÊÊProtectiveÊsiliconeÊrubberÊcoating
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.
Molybdenum SiliconÊwafer
discs
Cathode
Copper
Gate
Copper
ProtectiveÊsilicone Porcelain
rubberÊcoating envelope
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.
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.
100%
DominatedÊby
leakageÊcurrent
50%
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
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.
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.
di/dtÊdeterminedÊby
IdealÊthyristor
externalÊcircuit
I RealÊthyristor
TypicalÊsplit:
40% 60%
0
t
Irr PeakÊreverse
ShadedÊareaÊ=Êstored
currentÊ(Irr)
chargeÊ(Qrr)
Reverse
recovery
timeÊ(trr)
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)
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).
ITSM
Current
0.7ÊVDRM
TypicalÊconduction
angleÊ=Ê330¡
Voltage
Fig. 6.1s– Idealized current, voltage and junction temperature for a single-cycle valve insulation fault
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.
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.
6.2.2.2 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.
di/dt TL
reactor
TCU
Valve
Cd TL
TCU
Rdc Rd
TL
TCU
TCU
di/dt reactor
TCU
Thyristor
TCU level
Valve
TCU
VS
TCU
TCU
TCU
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
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
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.
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.
VÊ(Rd)
4
Ñ2
Ñ4
Ñ6
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.
6.2.7.1. 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.
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.
FastÊ
transient
voltage
source RC
VÊ(thy) damping
circuits
0 1ʵs 2ʵs 3ʵs 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.
Thermal
compound CoolingÊduct
Primary I
FerromagneticÊcore conductor
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.
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
CouplingÊdiode CouplingÊdiode
EquivalentÊparallel External External External External
resistanceÊofÊcores damping damping damping damping
resistor resistor resistor resistor
Multi-wayÊconnector
toÊvoltageÊdivider
HeatsinkÊforÊmain
semiconductor Fast-chargeÊcapacitor
components
MainÊstorageÊcapacitor
Optical InterferenceÊshieldÊover
components FPGAÊandÊother
sensitiveÊcomponents
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
Voltage divider
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.
10%
15% PV1
PV6
PV7
60%
15% Others
Fig. 6.2.11a– Typical breakdown of valve losses at full load power, by category
PV3 Other conduction losses In the busbars and di/dt reactor primary conductors which cause them.
– 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.
7%
8%
Thyristor
15%
DampingÊresistor
di/dtÊreactor
70%
Others
Fig. 6.2.11b– Typical breakdown of valve losses at full load power, by component
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.
AluminumÊheatsink
PEXÊhose
StainlessÊsteelÊelectrode
ConverterÊtransformerÊvalveÊwinding
(shownÊwithÊstarÊconnection)
Xc
Xc
Xc
2 . VLLÊ(PK)Ê VLLÊ(PK)Ê
PeakÊcurrent*: IPK=Ê =
2 . XC XC
PeakÊrecoveryÊvoltage:ÊVrecÊ=ÊVLLÊ(PK)
*
*
neglectingÊdamping
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.
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.
ClampingÊforceÊ≈ 135ÊkN
forÊ125ÊmmÊthyristor
Thyristor
DiscÊspring
GRPÊtensionÊband
Heatsink
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.
ValveÊsupportÊexposedÊtoÊ50%
ofÊratedÊpoleÊDCÊvoltage
DCÊneutralÊ HVDC
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.
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.
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.
BACK TO 326 | DC Transmission Systems: Line Commutated Converters
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 StressÊshield
StressÊshield Module
Module Module
Module
Module
Module Module
Module Module
Module Module
Module
Module
Module Module
Module StressÊshield
StressÊshield StressÊshield
StressÊshield
Movement
Movement
Module
Module Module
Module ofÊvalveÊduring
ofÊvalveÊduring
earthquake
earthquake Module
Module Module
Module
StressÊshield
StressÊshield StressÊshield
StressÊshield Module
Module Module
Module
Module
Module Module
Module Module
Module Module
Module
Excessive
Excessive Module
Module Module
Module StressÊshield
StressÊshield StressÊshield
StressÊshield
bendingÊstress
bendingÊstress
onÊinsulatorsÊ
onÊinsulatorsÊ Module
Module Module
Module
ExcessiveÊdisplacement,
ExcessiveÊdisplacement,
potentiallyÊviolating
potentiallyÊviolating
StressÊshield
StressÊshield 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)
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)
DC3
+
STATCOM V2 V5
L1A V3 L2 L6
+ DC2
L3 +
C3
V1X V1
V4
DC1 C1A
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.
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
Fig. 6.2.13f– Schematic view of new test circuit for operational tests
Fig. 6.2.13g – Voltage and current waveforms from the new test circuit in inverter mode
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.
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.
6.3.1. 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.
6.3.6. 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
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:
6.3.11. Alarms
Individual alarms are displayed on the local control point. The following local alarms are typically provided
as a minimum.
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
-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
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
ShieldÊring 8.011e+006Ê:Ê9.156e+006
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
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
BACK TO 340 | DC Transmission Systems: Line Commutated Converters
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
ShieldÊring 5.254e+006Ê:Ê6.005e+006
4.504e+006Ê:Ê5.254e+006
3.753e+006Ê:Ê4.504e+006
ValveÊwinding LineÊwinding 3.002e+006Ê:Ê3.753e+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
DensityÊplotÊ:ÊIÊEÊI,ÊV/m
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
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
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.
BACK TO DC Transmission Systems: Line Commutated Converters | 341
CHAPTER
6| HVDC CONVERTER STATION EQUIPMENT
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
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.
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
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
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
DC Transmission Systems: Line Commutated Converters | 345
CHAPTER
6| HVDC CONVERTER STATION EQUIPMENT
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
DensityÊplotÊ:ÊIÊEÊI,ÊV/m
Fig. 6.4.3f– FEM analysis of DC stress transfer from paper taped cleat bar to DC bushing structure
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.
Fig. 6.4.9a– HVDC converter transformer, 3-phase, 385 MVA 400/96 kV unit, arranged for shipping
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)
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:
π π π π π
3 1 ⋅ 4 ⋅ cos ⋅ sin t⋅ ω i − − 4 ⋅ cos ⋅ sinγ 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 (total),
idr(t) + idi(t)
Vphase YY+Vphase YD
on the DC side of the
converter transformer
Ideal 12-pulse DC
voltage, inverter
(ground to pole), vi(t)
idr(t)Ê+Êidi(t)
RHV RHV
Rectifier Inverter
VL rÊ∠Êθr 4Ê.ÊLcr 2 2 4Ê.ÊLci VL iÊ∠Êθi
IdealÊtransformer IdealÊtransformer
InductancesÊofÊtransformer
andÊvalveÊreactorÊ(LcrÊandÊLci),
movedÊintoÊDCÊside
ForÊcalculatingÊidr(t)ÊandÊidi(t)ÊtheÊDCÊtotalÊresistance,ÊRHV,ÊisÊassumedÊzero.
idr(t)ÊandÊidi(t)ÊdefinedÊinÊEqnÊ6.4fÊandÊEqnÊ6.4gÊrespectively.
vt(t)ÊdefinedÊinÊEqnÊ6.4h.
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.
DCÊcurrentÊrippleÊFFTÊofÊtotalÊcurrent
150
120
AmpereÊ(rms)
90
600 720
60
30
0
400 560 720 880 1040 1200 1360 1520 1680 1840 2000
FrequencyÊinÊHz
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
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
DC Transmission Systems: Line Commutated Converters
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.
Idripple Idripple
contribution contribution
Operating Power level Idripple Idmin Idoffset
from from
conditions % (A) (A) (A)
rectifier inverter
(A) (A)
D 10 21 36 57 257 200
G 10 21 36 57 252 195
Idripple Idripple
contribution contribution
Operating Power level Idripple Idmin Idoffset
from from
conditions % (A) (A) (A)
rectifier inverter
(A) (A)
D 10 25 36 62 257 195
G 10 25 36 62 252 190
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.
BACK TO DC Transmission Systems: Line Commutated Converters | 359
CHAPTER
6| HVDC CONVERTER STATION EQUIPMENT
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)
1.5
1
1
ReductionÊfactor
0.5
0.5
0.2
0
0 1 2 3 4
TotalÊsmoothingÊreactorÊreactanceÊ(rectifierÊ+Êinverter)ÊinÊp.u.ÊofÊLTmin
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.
HV PrimaryÊcircuit/
terminal HVDCÊdivider
NominalÊoutput
BufferÊamplifierÊ1
Ramp1 Camp1
BufferÊamplifierÊ2
OV
OV
EarthÊpotential Ramp2 Camp2
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.
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
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
converter transformer
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
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.ÊofÊvoltageÊatÊ10ÊkAÊ8/20ʵs
1.3 DC
1.2
1.1 AC
1
0.9 30/60ʵsÊ
0.8
0.7 8/20ʵsÊ
10kA0.6 8/20µs
0.5 1/2ʵsÊ
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
100
50
0
0 50 100 150 200 250 300 350 400
kV
-50
-150
Peak Continuous
Operating Voltage, PCOV -200
Time (°)
Fig. 6.7.1d– Typical voltage waveform imposed on a valve arrester, shown for a rectifier
6.8.2. 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
1Ê-ÊFixedÊcontact
2Ê-ÊNozzle
3Ê-ÊMovingÊcontact
4Ê-ÊCompressionÊcylinder
5Ê-ÊFixedÊpiston
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.
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.
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.
ArcÊvoltage
ia ic
ArcÊcurrent
g(t)
i(t)
t
TRV
CBF
PoleÊ1Êconverter
ACÊfilters
6.8.5a CBP1
ACÊnetwork
CBP2
ACÊfilters
CBF
PoleÊ2Êconverter
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)
TRV
2
Vs
1
i Vc
-1
-2
Fig. 6.8.7b– TRV on capacitive switching
L
R
+
E ea
-
Fig. 6.8.8a– DC switching circuit
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).
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.
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
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
BPS BPS
HVDCÊlineÊ2
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.
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.
DG DG
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
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
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.
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
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.
TOC 390 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
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.
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
TOC DC Transmission
392 | Systems:
DC Transmission
Line Commutated
Systems: Line
Converters
Commutated 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.
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
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.
Vd
Tap VacÊ+ÊΔVac
position Id
αmeas+ÊΔα
IdÊ+ÊΔId
TapÊup α
+
Phase Ierror
VIIÊmax locked +
oscillator
Iorder
TapÊdown
+
Tap
VIIÊmin controller
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
Id
αmeas+ÊΔα
IdÊ+ÊΔId
TapÊup α
+
Phase Ierror
αmaxÊsÊ-s locked +
oscillator
Iorder
TapÊdown
+
αminÊsÊ-s Tap
controller
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.
PoleÊcontrol
Operator
power Current
demand Current order
control
Power FiringÊangle
Limits
control
Measured Voltage
Voltage
power order
control
1.0Êp.u.
DCÊvoltage
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).
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
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:
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.
Vd
Id
VdÊ+ÊΔVd γmeas+ÊΔγ
DCÊline
voltage
drop
compensation TapÊup
α
+
Vderror Phase γmaxÊs-s
+ locked
oscillator
Vdorder
TapÊdown
+
Tap γminÊs-s
controller
Id
γmeas+ÊΔγ
VdÊ+ÊΔVd
α TapÊup
+
DCÊline Phase γerror VdmaxÊs-s
voltage locked +
drop oscillator
compensation γorder
TapÊdown
+
Tap
controller VdminÊs-s
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).
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.
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
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].
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
Measured
frequency
Fig. 7.2i– Typical frequency controller
Measured
frequency
7.2j
P Operator
power
PoleÊcontrol
Current
demand Current order
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
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.
DCÊpowerÊorder
DCÊpowerÊorder
PÊorder
} } FrequencyÊrangeÊ
constraintÊofÊinverterÊ
ACÊsystem PÊorder } FrequencyÊrangeÊ
constraintÊofÊrectifierÊ
ACÊsystem
}
Frequency Frequency
NormalÊfrequencyÊrange NormalÊfrequencyÊrange
ofÊrectifierÊACÊsystem ofÊinverterÊACÊsystem
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
BACK TO DC Transmission Systems: Line Commutated Converters | 409
CHAPTER
7| CONTROL AND PROTECTION
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
ElectrodeÊline
5
4 6
1 HVDC
filter
8
3 3 3 TransmissionÊlink
9 FilterÊ FilterÊ FilterÊ
bankÊ1 bankÊ2 bankÊ3
2
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.
TransmissionÊlink
FaultÊ1 FaultÊ9
1 5 9
T1 FaultÊ4
FaultÊ2
FaultÊ5 7 11 3
FaultÊ8 12 4 8
T2
NBGS
FaultÊ6
ElectrodeÊline
FaultÊ7
Fig. 7.5a– Ground faults and short-circuits
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
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.
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
7.10a
BACK TO 422 | DC Transmission Systems: Line Commutated Converters
CHAPTER
DispatchÊcenterÊ/ÊSystemÊcontrol
StationÊcontrol
BipoleÊcontrol
PoleÊcontrol
ConverterÊgroupÊcontrol
Pole Pole
master slave
Filters Filters
ACÊsystem ACÊsystem
Pole Pole
master slave
Filters Filters
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
Valve
Protection M1
Voting
Protection M2 Trip
Protection M3
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.
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.
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.
7.13a
TOC 434 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
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.
Chapter contents
8. DC TRANSMISSION CIRCUITS................... 434
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)
In this section, unless otherwise stated, we will mainly be considering the bipolar line configuration.
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
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;
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
2 150 150 Average of yearly maximum values
3 500 500 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.
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
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
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)
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).
Where:
V50 is the insulation critical flashover voltage (50% probability), in kV
d is the gap distance (m)
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.
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)
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.
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)
BACK TO 444 | DC Transmission Systems: Line Commutated Converters
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)
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:
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
BACK TO DC Transmission Systems: Line Commutated Converters | 445
CHAPTER
8| DC TRANSMISSION CIRCUITS
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)
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 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.
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:
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).
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
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.
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.
q
AN = AN 0 + 86 log( g ) + k log ( n ) + 40 log ( d ) − 11.4 log( D ) + [Eqn 8.1.11b]
300
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).
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:
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].
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.
8.1.13.1. 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.
BACK TO 454 | DC Transmission Systems: Line Commutated Converters
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.
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).
25
µT
20
15
10
5
0
0 10 20 30 40 50
Distance from tower center (m)
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.
50
40
30
20
10
0
0 10 20 30 40 50
DistanceÊtoÊtowerÊcenterÊ(m) Electric field
40
30
10
equivalent, leading to similar conductor height at midspan.
8.1j 0
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.
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)
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
BACK TO DC Transmission Systems: Line Commutated Converters | 457
CHAPTER -80
-100
-60 -40 0 60
8| DC TRANSMISSION CIRCUITS
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.
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).
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:
Where:
Cp is the yearly cost of the power plant
Ce is the fuel cost
lf is the loss factor
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])
8.2.1. Introduction
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.
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:
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
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.
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.
8.2.3.1. 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.
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
•Ê ConductorÊ
•Ê ConductorÊscreenÊ
•Ê InsulationÊ
•Ê InsulationÊscreenÊ
•Ê LeadÊalloyÊsheathÊ
•Ê PolyethyleneÊsheathÊ
•Ê TransversalÊreinforcementÊ
•Ê OuterÊsheathÊ
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. 8.2e– 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.
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.
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.
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.
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
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)
Anode
electrode
Buried
pipe _
Cathode
electrode
AreaÊwhere
corrosionÊwillÊoccur LinesÊof
equipotential
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.
DCÊcomponent
t i
MagneticÊfluxÊ
withoutÊDCÊ
component
MagnetizationÊ
currentÊwithoutÊ
DCÊcomponent
t
MagnetizationÊ
currentÊwithÊDCÊ
component
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.
5
4.5
4 25ÊmÊdepth
MagneticÊcompassÊdeviationÊ(Deg)
3.5 30ÊmÊdepth
3 35ÊmÊdepth
2.5 40ÊmÊdepth
2 45ÊmÊdepth
1.5 50ÊmÊdepth
1
0.5
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.
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.
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)
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.
TOC 484 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
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.
Chapter contents
9. HVDC SCHEME PERFORMANCE............. 484
TOC DC Transmission
486 | Systems:
DC Transmission
Line Commutated
Systems: Line
Converters
Commutated 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.
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
BACK TO DC Transmission Systems: Line Commutated Converters | 487
CHAPTER
9| HVDC SCHEME PERFORMANCE
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.
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:
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.
0.6
0.5
(p.u.ÊofÊtotalÊschemeÊlossesÊatÊfullÊload)
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).
0.08
0.07
(p.u.ÊofÊtotalÊschemeÊlossesÊatÊfullÊload)
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
(p.u.ÊofÊtotalÊschemeÊlossesÊatÊfullÊload)
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
Vdr,Êldr Vdi,Êldi
ConverterÊsteadyÊstate
equations
∆ÊVd, ∆ÊVd,
α,ÊP,ÊQ,ÊId,ÊVd,ʵ ∆Êld, ∆Êld, γ,ÊP,ÊQ,ÊId,ÊVd,ʵ
Converter ∆ÊVac, ∆ÊVac, Converter
dynamic ∆γ, ∆γ, dynamic
equations CommÊfail CommÊfail equations
DCÊcontrol DCÊcontrol
∆α,ÊPdo,ÊIdo ∆α ,ÊPdo
algorithms algorithms
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).
DispatchÊcenterÊ/ÊSystemÊcontrol
StationÊcontrol
BipoleÊcontrol
PoleÊcontrol
ConverterÊgroupÊcontrol
Filters Filters
ACÊSystem ACÊSystem
Filters Filters
α<2¡ α>182¡
Logic
Vdc A(s)
Limits
Control
variables
Selector
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.
1 1
Tm (C 12) (C 13)
K 12 K 13
Kω [Eqn. 9.2a]
T = Msω + Dω +
s
This is equivalent to the time domain LCR circuit voltage equation:
9.2e
BACK TO 498 | DC Transmission Systems: Line Commutated Converters
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
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.
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
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
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
AtÊconverter
bus
Tap
changer
TgÊchange Qg/Qsystem
change
∆ÊTÊ(T+)
δÊchange
∆α,Ê∆γ,Ê∆ldÊinÊDCÊscheme
PloadÊchange Q loadÊchange
PdcÊchange QdcÊchange
Σ
ChangeÊinÊnetworkÊflows (ForÊconverterÊbuses)
Where:
α =ÊDCÊrectifierÊcontrolÊangle
δ =ÊÊMachineÊangle
γ =ÊÊDCÊinverterÊcontrolÊangle
ω =ÊMachineÊspeedÊdeviation
9.3d
BACK TO DC Transmission Systems: Line Commutated Converters | 503
CHAPTER
9| HVDC SCHEME PERFORMANCE
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.
Iac =
(P + jQ)conj
Vac conj [Eqn. 9.3e]
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
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
PmodÊhigh
Pmod
Low pass Differential Gain
filter pole
PmodÊlow
Fig. 9.3h– Simplified block diagram of power oscillation damping (POD) control
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
∆ω ∆δ ∆α ∆VdÊ,ÊId IfÊtotalÊphaseÊlag
betweenÊ∆αÊ&Ê∆Te
dueÊtoÊcontrolsÊis
suchÊthatÊreaction
torqueÊisÊinÊphase
withÊ∆ω,Êthen
∆Tm ∆Te ∆Pe ∆Pd instabilityÊisÊcaused
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
Shaft Shaft
torque torque
4 4
p.u.Ê(onÊmachineÊMVA)
p.u.Ê(onÊmachineÊMVA)
3 3
2 2
1 50ÊMW 1
0 0
-1 -1
-2 -2
4 4
p.u.Ê(onÊmachineÊMVA)
3
p.u.Ê(onÊmachineÊMVA)
3
2
2
1
300ÊMW 1
0
0
-1
-1
-2
-2
4
4
p.u.Ê(onÊmachineÊMVA)
3
p.u.Ê(onÊmachineÊMVA)
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
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
ControlÊangleÊ(degrees)
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)
9.7a
BACK TO DC Transmission Systems: Line Commutated Converters | 521
CHAPTER
9| HVDC SCHEME PERFORMANCE
FundamentalÊfrequencyÊRMSÊmodel Cycle-by-cycleÊmodel
1.2 1.2
ConverterÊbusÊACÊvoltageÊ(p.u.)
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
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
ControlÊangleÊ(degrees)
70 70
ControlÊangleÊ(degrees)
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)
BACK TO 9.7b
522 | DC Transmission Systems: Line Commutated Converters
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.
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
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
100% 100%
50%
100% 50%
50%
SwitchedÊredundantÊplant On-lineÊredundantÊplant
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.
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
TOC 532 | DC
HVDC: Connecting toTransmission
the future Systems: Line Commutated Converters
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.
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.
DC AC DC
Fig. 10.1b – Comparison between the power ratings of AC and DC overhead line towers
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.
7 3 3 5
BipoleÊI
BipoleÊII
+/-ÊÊÊ463ÊkV
+/-ÊÊÊ500ÊkV
1854ÊMV
2000ÊMV
Dorsey
230ÊkV
11ÊLinesÊto
ManitobaÊnetwork
500ÊkV
ToÊUSA
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.
10.2.1.3. 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].
10.2.1.5. 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.
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.
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.
.../...
• 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.
Ground region
Housing region
Industrial site
Foliage region
Point source
Grid
Grid point
Surface contour
Line source
GPS calibration point
Receiver
Base case
0m 70 m
scale = 1 : 2500
origin = 270, 0
200
0
400 600
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
10.3.1.11. Losses
Objective Calculate the ‘as manufactured’ losses of the converter station.
Design data expected
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.
Contents
APPENDIX RELATED TO CHAPTER 2 ................... 564
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]
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]
Fourier analysis
requires this reference for correct
calculation of phase shift
-60° 60°
0°
120°
30°
360° Equivalent
fundamental
frequency current
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:
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
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.
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
VLL − 0
ω ⋅ Lc = Xc p.u . ⋅ [Eqn A2.2.2f]
3 ⋅ Iac0
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
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
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
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
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.
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
10
C-message
Psophometric
0.1
0.01
100 1000 10000
FrequencyÊ(Hz)
A4.2a
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]
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]
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:
ω 2p1
x1 ⋅ x2 = [Eqn A4.3.1o]
L1C1
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
BACK TO DC Transmission Systems: Line Commutated Converters | 575
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
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.
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]
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
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 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:
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
BACK TO 578 | DC Transmission Systems: Line Commutated Converters
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
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.
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
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 )
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
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.
Filter Network
admittance admittance
IN
Ic IF
L1
YF YN
Fig. A4.5a– Circuit for the choice of worst network impedance
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)
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
DC ripple current
DC voltage, no overlap
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
2 I dc X cc
K1 = [Eqn. A9.1a]
Eacc
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)
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
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]
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 }
1 A 1 A 2Ê
µ1 BothÊelementsÊworking µ2
λ2 4 (1Ê-ÊA 1)(1Ê-ÊA 2) λ1
BothÊelementsÊfailed
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
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
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
Case 1 Case 2
Probability of being in the
P1
A9.3.3c P1 + P2
working state (availability)
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
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
The results obtained for the three different cases are as follows:
Case 1 Case 2 Case 3
Number of states in which
1 2 3
the system is working
Probability of being in the
P1 P1 + P2 P1 + P2 + P3
working state (availability)
Number of states in which
3 2 1
the system is failed
Probability of being in
P2 + P3 + P4 P3 + P4 P4
the failed state
1 A1A2A3Ê
µ1 Working Working Working µ3
µ2
λ2
λ1 λ3 λ1 λ2
λ3
Fig. A9.3.4c– Possible states of a three-element system consisting of three different elements
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.
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.
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