Modeling and investigation of the NorNed HVDC link with RTDS

MASTER OF SCIENCE THESIS

Electrical Power Systems

DELFT UNIVERSITY OF TECHNOLOGY

FACULTY OF ELECTRICAL ENGINEERING, MATHEMATICS
AND COMPUTER SCIENCE

Vikaash Sookha

# 1530291

Delft, May
M 2011

0
1
Modeling and investigation of the NorNed HVDC link with RTDS

MASTER OF SCIENCE THESIS

Electrical Power Systems

DELFT UNIVERSITY OF TECHNOLOGY

FACULTY OF ELECTRICAL ENGINEERING, MATHEMATICS
AND COMPUTER SCIENCE

Vikaash Sookha

# 1530291

Delft, May
M 2011

2
Thesis committee:

Prof. Ir. L. van der Sluis Delft University of Technology, Thesis supervisor
Dr. Ir. M. Popov Delft University of Technology, Daily supervisor
Ir. E. Wierenga TenneT TSO B.V., Daily supervisor
Dr. Ir. D. Djairam Delft University of Technology

3
Preface

As part of completing the Master of Electrical Power Engineering track at the Delft University of
Technology, the student must show his capacities and abilities gained during the period in
classrooms with the final objective: the thesis. The period spent on the thesis is not only a
period where the student spends time in gaining and applying knowledge, but it is a period
where personal development also is gained.
I have done my thesis at TenneT TSO in Arnhem, The Netherlands, at the department of
Transport & Infra in close collaboration with the department of Electrical Power Systems
(“Netmodellering”) of the TU Delft where the RTDS is installed. I did the project at TenneT with
much pleasure and I have tried my best to get good results. Being in a professional
environment, getting along with everyday activities in a company was very educative for me.
Being in such a professional company I learned a lot and I wanted to be of some value for the
company too by means of this thesis project.
The work carried out here is more than one persons effort, therefore I wish to express my
sincere gratitude to the ones who contributed to this task. I thank Dr. Ir. Marjan Popov for
bringing me in contact with TenneT and his supervision during the period of the thesis. From
TenneT I thank Ir. Frank Koers and Ir. Ernst Wierenga for giving me the opportunity to be at
TenneT. I thank Ir. Ernst Wierenga for his support and experienced way of supervision and
advising at points where I ran out of ideas and of course his professional guide concerning the
technical part. From the department Asset Management I thank Ir. Kees Koreman for the very
experienced and skillful guidance in the field of HVDC. I thank TenneT as a whole. I also thank
TU Delft as a whole.
This project would have been a lot more difficult without the fast and professional responses
and explanation about RSCAD of Onyinyechi Nzimako from RTDS, Canada. Thank you Onyi.

Finally, I thank my entire family for their enduring support and love. I thank all my friends for
their support and great moments during my study period. I thank my father Ivan and mother
Sabitrie, sisters and brother for always being there for me. Thanks to my uncle Rabin and
special thanks to my aunt Lalita who still encourages me. Last but not least I thank my girlfriend
Shariska for her constant support and love and believing in me.

Vikaash Sookha

Delft, Netherlands
May 2011

4
Summary

To be successful in operating and maintaining most challenges in electric power system,

understanding is important of the grid. Present-day the application of HVDC technology is
growing, because it proves to be beneficial for long distance and bulk transmission. This is also
the case in The Netherlands. Investigation of networks can be done with the aid of simulations.
The RTDS is a powerful tool to perform Electro Magnetic Transient simulations, the RTDS has
the strong property to carry out real-time simulations. For TenneT having a model which can be
used for investigation is beneficial. This report is the result of a thesis of which title is formulated
as: Modeling and investigation of the NorNed HVDC link with RTDS
After the short introduction, the fundamentals of HVDC technology is explored. Here the
important advantages of DC over AC are given, which include lower transmission losses but
higher substation costs which results from expensive filters and reactive power compensation
equipment. The operating principles of classic HVDC from a technical point of view are
analyzed. Further the developments of HVDC from Netherlands point of view is given.
Currently there are two HVDC connection in the Netherlands. One HVDC link between
Netherlands and Norway is operational: The NorNed, this is a 580 kilometers submarine cable
operating at +450kV/-450kV with a rated transport capacity of 600 Megawatts and the other is
the BritNed between Netherlands and the UK.
This HVDC connection is implemented in the RTDS. The steps needed to simulate a HVDC
connection are described in some detail to understand the operation and for future examination.
The implementation process begins with providing a description of the complete model. The
primary circuit is described, some detail is given on the controls as this is a key feature of fast
regulation for HVDC and finally the disturbances controls is explained. Because of some
shortcoming in the represented models in the RSCAD library, not all the actual equipments are
exactly modeled as installed. Some modifications are made.
Once the model has been created, simulations can be started. A number of simulations are
carried out showing the steady state performance of the HVDC model. With this model now
other cases can be studied to understand specific situations at points in the grid. Such an case
involved a single phase fault in April 2009 which resulted in the NorNed being offline for a
month. Some oscillogram recordings were made at the fault location. These are used to replicate
the fault in the RTDS with the model.
The task of replicating the responses of the actual recordings proved to be a difficult task. The
results in the simulations show differences. The effect however, of the events in the network is
clear. A possible solution to prevent the unwished phenomena after fault occurrence is to
disconnect the capacitor banks present in the filter building where the fault occurred. With this
action the magnitude of the voltage swings decrease.
The exact responses compared to the recording were not found during simulations. This is the
result of the model being modified because of the limitations of RSCAD. Nevertheless, an
important step was set for the investigation of HVDC links. This should be set forward in better
understanding and prevent interruptions. But with updated and enhanced models, and more
because of the strong feature of RTDS to perform real-time simulation which TenneT in
cooperation with the TU Delft can use in the future. In order to verify the models and the case
studied here, another simulation tool can also be used. This can be done with DIgSILENT
PowerFactory which is used at TenneT.

5
6
Contents

PREFACE ................................................................................................................................................... 4

SUMMARY ................................................................................................................................................. 5

1 INTRODUCTION ................................................................................................................................10
1.1 Background of the work ................................................................................................................................. 10
1.2 Definition and scope ...................................................................................................................................... 11
1.3 Approach ........................................................................................................................................................ 11

2 HVDC TRANSMISSION .....................................................................................................................12

2.1 Transmission by DC instead of AC................................................................................................................... 12
2.1.1 Justification ................................................................................................................................................... 12
2.1.2 Converter types ............................................................................................................................................. 15

2.2 Converter theory fundamentals ..................................................................................................................... 18

2.3 Reactive power and harmonics ...................................................................................................................... 25

3. HVDC DEVELOPMENTS ..................................................................................................................30

3.1 Practical HVDC connections ............................................................................................................................ 30
3.2 Trends in HVDC technology ............................................................................................................................ 32
3.3 HVDC from TenneT point of view ................................................................................................................... 34
3.4 NorNed at Eemshaven .................................................................................................................................... 36

4. NETWORK MODELING ...................................................................................................................43

4.1 The simulator ................................................................................................................................................. 44
4.2 Modeling the primary circuit .......................................................................................................................... 48
4.2.1 The components ............................................................................................................................................ 48
4.2.2 Transmission.................................................................................................................................................. 53

5 MODELING THE CONTROLS ..........................................................................................................55

5.1 Control properties of model ........................................................................................................................... 55
5.1.1 Master control ............................................................................................................................................... 56
5.1.2 Rectifier ......................................................................................................................................................... 57
5.1.3 Inverter .......................................................................................................................................................... 61

5.2 Full control properties .................................................................................................................................... 64

5.3 Disturbances controls ..................................................................................................................................... 68

7
6 SIMULATION ......................................................................................................................................73
6.1 Part I .............................................................................................................................................................. 74
6.1.1 Faults types ................................................................................................................................................... 74
6.1.2 Steady state ................................................................................................................................................... 74
6.1.3 Faults ............................................................................................................................................................. 79

6.2 Part II ............................................................................................................................................................. 80

6.2.1 Case AM......................................................................................................................................................... 80
6.2.2 Results studied case ...................................................................................................................................... 83

7. CONCLUSIONS & RECOMMENDATIONS ....................................................................................87

7.1 Conclusions .................................................................................................................................................... 87
7.2 Recommendations ......................................................................................................................................... 89

REFERENCES ..........................................................................................................................................90

APPENDICES ..........................................................................................................................................92
Appendix A. Comparison example AC and DC cable transmission ........................................................................ 92
Appendix B. Rectifier and Inverter operation signals ........................................................................................... 93
Appendix C. HVDC projects .................................................................................................................................. 95
Appendix D. Data for network model in chapter 4 ............................................................................................. 100
Appendix E. Network in RTDS ( RSCAD ) ............................................................................................................. 104
Appendix F. Full HVDC rectifier and inverter controls ........................................................................................ 106
Appendix G. Steady state properties NorNed..................................................................................................... 109
Appendix H. Oscillogram recordings of Case ...................................................................................................... 110
Appendix I. Case simulated in RTDS ................................................................................................................... 113

8
Terminology
ABB The ABB group- Automation and Power Technologies
AOR1 Alpha Order Rectifier- pole 1
CB Circuit Breaker
CINV2 Current Inverter- pole 2
Csn: Snubber Capacitance
DGE Delta Gamma Error
DIO Digital Input/Output card
DITS DIgital Time Stamp card
DSP Digital Signal Processor
EDC: DC Substation Eemshaven
EEMS substation Eemshaven (AC)
EMT Electro Magnetic Transient
FPG Firing Pulse Generator
GIS Gas Insulated Switchgear
GPC Giga Processor Card
GUI Graphical User Interface
HVAC High Voltage AC
HVDC High Voltage DC
Hz Hertz
IGBTs Insulated-Gate Bipolar Transistor
Iord I order
IRC Inter-Rack Communications Card
LCC Line Commutated Converters
PID Proportional Integral Derivative controller
PLL Phase Locked Loop
POW Point On Wave
PWM Pulse Width Modulation
RISC Reduced Instruction Set Computer
RMS Root Mean Square
RPC Risc Processor Cards
Rsn Snubber Resistance
RSCAD GUI through which the user is able to construct, run and analyze simulation cases
RTDS Real Time Digital Simulator
SCC Short Circuit Capacity
TSO Transmission System Operator
UCTE Union for the Coordination of Production and Transmission of Electricity
VDCOL Voltage Dependant Current Order Limiter
VGB Valve Group Block
VSC Voltage Source Converters
WIF Workstation Interface Card
XLPE Cross Linked Poly Ethylene

9
1 Introduction
1.1 Background of the work

Present-day the HVDC technology is applied and still developing because it proves to be
beneficial for long distance and bulk transmission. This is also the case in The Netherlands.
Examples are the already existing and operational HVDC link NorNed, connecting Norway and
Netherlands, BritNed between England and Netherlands, almost operational and more HVDC
connections planned to be built. Once a connection is made and implemented in the grid, it can
have failures, which in many cases need to be investigated and possibly mutated to prevent this
occurrence.
Investigation of networks can be done with the aid of simulations. But first the network needs to
be implemented in a proper simulation tool. There are several types of simulation tools to carry
out specific types of investigations, such as Electro Magnetic Transient, Power System Dynamic
and load flow studies.
The RTDS is such a simulator which is capable to perform dynamic simulations, the advantage
of RTDS being to do real-time simulations. To implement a real network into a model
knowledge is needed how the network is built and operated in practice, so theoretical
background is essential. Especially for HVDC, where the controls are an important tool for
regulation. Once the network is set up, simulations can be carried out and situation & cases
where interest is can be studied.
The NorNed is a 580 kilometers submarine cable operating at +450kV/-450kV with a rated
transport capacity of 600 Megawatts. The converter station is situated in Eemshaven. In 2009
NorNed went offline for a month as a result of a fault, this needed closer examination to
understand and possibly prevent such events.

10
1.2 Definition and scope

With the focus on the HVDC transmission the goal is to investigate what the effects of several
faults is at certain points in the network. These faults include: AC line-to-line/line-to-ground
faults at both ends of the DC cable connection, Converter faults and DC faults.
The subject is formulated as:

Modeling and investigation of the NorNed HVDC link with the RTDS

The report produced here describes the essential theoretical background of HVDC technology
with the developments from TenneT`s point of view. The modeling will be described in some
detail to understand the HVDC operation and for future examination. The HVDC controls also
have special attention which can be used to develop a model in other simulation tools.

Once the tool ready, investigations can be done at certain points in the network. TenneT will
have more insight in the network in relation to HVDC in the Eemshaven area.

1.3 Approach

The thesis work as discussed is divided in several parts which follow the sequence of:

• Theoretical background HVDC and developments

• Network and control modeling, followed by
• Some basic simulations showing the model functioning, and
• Investigation of an earlier occurred fault
• Finally the conclusions and some recommendations are given

11
2 HVDC transmission
“Why use direct current transmission?” This is a question often asked and often the response is
that losses are lower. But this is not correct. The level of losses is designed into a transmission
system and is regulated by the size of the conductor selected. DC and AC conductors, either as
overhead transmission lines or submarine cables can have lower losses but at higher expenses
since the larger cross-sectional area will generally result in lower losses but cost more.
This chapter deals with the fundamental aspects of HVDC transmission. The reasons behind
HVDC technology is briefly mentioned, followed by the operation of the converter and an
overview is given to the future of this technology further on.

2.1 Transmission by DC instead of AC

2.1.1 Justification
Since the first commercial HVDC link went in operation in 1954 connecting the island of
Gotland and the Swedish mainland with a 96 km long 100 kV submarine cable with a capacity of
20MW, it is fair to say that at present HVDC technology is booming worldwide with an installed
capacity of over 80 GW. This is 1.8 % of the worldwide installed generation capacity. This can
be explained by looking at two major trends in the development of power systems[1][2].
1. Transition of the traditional power systems, where generation is done by a small
number of large units to a new situation where a large amount of small(er) units are
responsible for generation of the power.
2. Increasing distance between sites where large scale generation is taking place and
the load centers where vast amounts of electrical energy is consumed.

Most striking is perhaps the difference between these two trends. The first requires a system that
is flexible in terms of control of both active and reactive power. Whereas for the second, the
ability of handling the high voltages and currents typically associated with large amounts of
power is much more of an issue. Thanks to advances in power electronic technology, HVDC can
provide solutions in both areas.

The main advantages generally claimed in favour of the DC alternative are:

• Economic considerations; DC transmission results in lower losses and costs than
equivalent AC lines, but the costs and losses at the DC terminal can be higher
An overhead dc transmission line with its towers can be designed to be less costly per unit of
length than an equivalent ac line designed to transmit the same level of electric power. However
the dc converter stations at each end are more costly than the terminating stations of an AC line
and so there is a breakeven distance above which the total cost of dc transmission is less than its
ac transmission alternative. For ac infrastructure it can be argued that the costs are mainly
determined by the lines and to a much less extent by the substations. AC lines usually consist of
at least two, three-phase circuits where often one phase if formed by a bundle of multiple
conductors. In all, many conductors are required to construct an AC connection. DC connections
on the other hand, usually consist of only two poles each of which is carried out as a bundle of
multiple conductors. Still, a DC connection requires much less conductors when compared to its
AC counterpart. Furthermore, due to the impedance of AC overhead lines the loadability of the

12
line decreases as the transmission distance increases. To avoid the loadability becoming small for
long(er) lines, expensive series- and shunt compensation have to be installed along the line.
Since for DC the impedance of the line equals the resistance, no such compensation is required.
However, the converter stations needed in the latter case for the conversion of AC into DC vice
versa, are due to their complexity very expensive. Substations found in AC networks on the other
hand, are less complex and as such, cheaper to built. The graph below gives an indication of the
economical benefits from a certain transmission distance.

Figure 2.1.1 AC-DC cost comparison [3]

The break-even distance is in the range of 500 – 800 km depending on a number of other factors,
like country-specific cost elements, rates for project financing, loss evaluation, cost of right way
etc. [3][4].

• Technical merits; AC transmission via cable is impractical over long distances, such a
restriction does not exist with DC, at DC the conductor cross section is fully utilized,
because there is no skin effect
If transmission is by submarine or underground cable, the breakeven distance is much less than
overhead transmission. The capacitance of an AC high-voltage cable gives rise to a charging
current. This charging current is effectively reducing the amount of power that the cable is able
to transfer. The capacitance of the cable, and thus its charging current is proportional to the
length of the cable. From this it follows that the longer the cable is, the more its ability to transfer
power is compromised. For a certain length the charging current equals the rated current and the
cable is no longer capable of transferring power. In practice this situation arises for cable lengths
of approximately 60 to 100 kilometers depending on the type of cable. Since for DC the

13
capacitive reactance of the cable is infinite, no charging currents exists. This makes the
transmission distance virtually unlimited[5].
See Appendix A for comparison [6].
Power transmitted over an AC line given by the power-angle equation:
VV
P = S R sin Θ SR
X SR (2.1)
P - Transmitted power
VS - Voltage sending end, with cos φ 1
VR - Voltage receiving end, with cos φ 2
X - Series resistance
Θ - Voltage phase shift(or load angle), Θ = φ 1 - φ 2

Figure 2.1.2 AC power transmission

• Interconnection; DC stations with or without transmission distance, can be justified for

the interconnection of AC systems of different frequencies or different control
philosophies.
Some ac electric power systems are not synchronized to neighboring networks even though their
physical distances between them is quite small. This occurs for example in Japan where half the
country is a 60 hz network and the other is a 50 hz system. It is physically impossible to connect
the two together by direct ac methods in order to exchange electric power between them.
However, if a dc converter station is located in each system with an interconnecting dc link
between them, it is possible to transfer the required power flow even though the ac systems so
connected remain asynchronous. This is used at several locations present day[7].

• Environmental aspects; The dc transmission line can have a lower visual profile than an
equivalent ac line and so contributes to a lower environmental impact. Its improved
energy transmission possibility contributes to a more efficient utilization of land coverage
for transmission towers. See figure 2.1.3 for a comparison There are also advantages to a
dc transmission line through the electric and magnetic fields being dc instead of ac.

14
Figure 2.1.3 Typical line structures for approximately 1000MW

Furthermore,
• DC constitutes an asynchronous interconnection and does not raise the fault level
appreciably

• The power flow in a DC scheme can easily be controlled at high speed, and thus with
appropriate controls a dc link can be used to improve AC-system stability

2.1.2 Converter types

Before moving forward, it is important to know what types of AC-DC conversion methods are
used today. Basically two methods are developed & used intensively, these are line commutated
converters and voltage source converters[8].
Line commutated converters (LCC) are based on current source conversion with naturally
commutated thyristors, so called Classical HVDC transmission systems. The name LCC
originates from the fact that the applied thyristors need an AC voltage source in order to
commutate. Figure 2.1.4 [9] shows the schematic of an LCC based HVDC transmission system.
Because classical HVDC transmission is a well established technology, it offers high reliability
and requires little maintenance. Compared to its counterpart, LCC based HVDC transmission has
much lower power losses (i.e. only 2-3% converter losses) and for high ratings it has comparably
low capital costs. However, based on the overall system economics, LCC based HVDC
transmission becomes only interesting for transmission capacities above the break-even distance
as shown earlier in figure 2.1.5. Since the first commercial LCC based HVDC link was installed
it has been installed frequently, primarily for bulk power transmission over long geographical
distances and for interconnecting non-synchronized or isolated power systems. Major
disadvantage of LCC based HVDC transmission system is that it cannot provide independent
control of the active and reactive powers. Instead, the LCC scheme consumes a certain amount
of reactive power. This is further explained in the next paragraph. Because of this drawback,
capacitor banks are needed for compensation when transporting energy. Also large amounts of
harmonics are produced which makes the use of large filters inevitable.

15
Figure 2.1.4 Line commutated converters and voltage source converters

Voltage source converters (VSC) based HVDC transmission is comparatively a new technology
(first commercial installation in 1999) and has only become possible by the development of the
IGBTs, which can switch off currents. This means that there is no need for an active
commutation voltage. Therefore, VSC based HVDC transmission does not require a strong AC
network and can even start up against a dead network (black-start capability). VSC based HVDC
transmission systems are gaining more and more attention. Today, VSC based solutions are
marketed by ABB under the name “HVDC Light” and by Siemens under the name “HVDC
Plus”. Figure 2.1.4 shows the schematic of a VSC based HVDC transmission system. The active
and reactive power can be controlled independently, which may reduce the need for reactive
power compensation and can contribute to stabilize the AC network at their connection points. In
addition, the IGBT semiconductors allow for much higher switching frequencies which reduces
the harmonic content of VSC based systems. Therefore, the filter requirements on the AC side
are considerably reduced compared to conventional HVDC converters. However, the high-
frequency PWM switching results in comparatively high converter losses. The total efficiency of
the two converter stations of a VSC based HVDC transmission system is therefore less than that
of an LCC based system(99.3% efficiency per terminal for LCC HVDC versus 98.2% efficiency
per terminal for VSC terminal[18]). Furthermore, the cost of VSC based systems is still high due
to the more advanced semiconductor valves required. In order to handle the high voltage,
multiple IGBTs have to be connected in series, which makes the valves expensive. Looking at
the overall system economics, VSC based HVDC transmission systems are most competitive at
transmission distances over 100 km or power levels of between approximately 200 and 900MW,
as depicted in figure 2.1.5 below, which shows a comparison of different transmission
technologies. However, the application of VSC based systems may already be advantageous for
shorter transmission distances depending on the specific project conditions. A list of HVDC
projects in the course of time is included in Appendix C.
In this study the VSC configuration will not be investigated, therefore only the LCC converter
operation is discussed[8].

16
Figure 2.1.5 Power transmission capacity with different technologies

17
2.2 Converter theory fundamentals

This section outlines the most important theoretical background of thyristor based, or classic
HVDC converters. It contains however by no means a full and in-depth description of such
converters; for this, the reader is referred to other literature on this topic. Figure 2.2.1 shows an
impression of the conversion procedure from ac to dc at rectifier, visa versa at inverter. At first
glimpse it seems a lot, but starting with the thyristor the entire process will be considered.

SIX-PULSE BRIDGE TRANSFORMER

REACTOR

THYRISTOR

Figure 2.2.1 Conversion procedure from ac to dc at rectifier, visa versa at inverter

The thyristor valve is the most important building block for LCC. Thyristor valves operate as
switches which turn on and conduct current when fired on receiving a gate pulse and are forward
biased. A thyristor valve will conduct current in one direction and once it conducts, will only
turn off when it is reverse biased and the current falls to zero. This process is known as line
commutation. An important (unwanted) property of the thyristor valve is that once its conducting
current falls to zero when it is reverse biased and the gate pulse is removed too rapid, an increase
in the magnitude of the forward biased voltage will cause the thyristor to inadvertently turn on
and conduct. The design of the thyristor valve and converter bridge must ensure such a condition
is avoided for useful inverter operation. This event called commutation failure is considered
later.
The six-pulse converter bridge of figure 2.2.1 as the basic converter unit of HVDC transmission
is used equally well for rectification where electric power flows from the ac side to the dc side
and inversion where the power flow is from the dc side to the ac side.
As figure 2.2.2 shows, two valves are connected to each phase terminal, one with anode
connected to it(shown on the upper side of the bridge) and the other with the cathode connected
to it(shown on the lower side of the bridge). The need for two valves conducting in series is not a
18
drawback in high-voltage applications, because of the need for many series- connected units to
withstand voltage levels used.

Figure 2.2.2 Operation six-pulse bridge

To understand the operation of a three-phase bridge rectifier let us consider the idealized case of
a converter bridge(i.e. source impedance is zero). Under these condition, the transfer of
current(commutation) between valves on the same side of bridge takes place instaneously. The
switching sequence and the rectified voltage waveform illustrated in figure x for the case of an
uncontrolled bridge rectifier(i.e. diode operation); valves1, 3, 5 at the top, and 4, 6, 2 at the
bottom are connected to phases R, Y and B(red, yellow & blue) respectively.
With reference to figure 2.2.2 a to g, and starting at instant A, phases R and Y are involved
through conducting valves 1 and 6. This operating state continues up to point B, after which
valve 2 becomes forward biased, since its anode, directly connected to that of valve 6 , is positive
with respect to its cathode(connected to phase B); therefore at point B the current commutates
naturally from valve 6 to valve 2 seen in figure 2.2.2b.
A similar argument applies at point C, with reference to valves 1 and 3 on the upper half of the
bridge. The anode of valve 3(connected to phase Y) begins to be positive with respect to its

19
cathode(connected to phase R through conducting valve 1) and a commutation takes places from
valve 1 to valve 3,figure 2.2.2c). This is followed from valve 2 to valve 4 at point D, valve 3 to
valve 5 at point E, valve 4 to valve 6 at point F, and valve 5 to valve 1 at point G. This completes
the switching cycle sequence, and the sequence is repeated. The output waveform in figure
2.2.2g shows the voltage variation of the positive(common cathode) and the negative (common
anode) poles with respect to the transformer neutral. Figure 2.2.2h shows the output voltage, i.e.
the voltage of the positive pole with respect to the negative pole. It is seen that the output voltage
has a ripple, or harmonic frequency, of six times the main frequency. Each valve carries the full
value of the direct current for one third of the cycle, and there are always two valves conducting
in series.

Gate control
Before explaining the effect of varying the firing angle, it is important to know what the
following angles mean:
• Firing delay angle α; is the interval between the moment the commutation voltage becomes
positive and the actual firing(R)
• Overlap angle µ; is the interval between the firing of the incoming valve and the cessation of
the current in the outgoing valve(R)
• Advance angle β; is the interval between the firing of the incoming valve and the moment the
commutation voltage is going negative(I)
• Extinction angle γ is the interval between cessation of the current in the outgoing valve and the
moment the commutation voltage is going negative(I).
Visualizing these angles can be done by figure 2.2.3.

Rectifier Inverter

Figure 2.2.3 Gate control angles

By delaying the firing instants of the valves with respect to the voltage crossings, the
commencement of the natural commutations described above can be delayed by a definite time
interval and the effect of this action on the direct-voltage waveforms is illustrated in figure 2.2.4a
and b. It is noticeable that the voltage area, and therefore the mean direct voltage, is reduced in
proportion to the magnitude of the delay. The voltage waveforms for a delay of 90º illustrated in
figure 2.2.4 shows equal positive and negative voltage regions, the mean direct voltage is

20
therefore zero with a delay of 90º. Beyond 90º the mean voltage is negative and bridge operation
can only be maintained in the presence of a DC voltage supply. This indicates the power being
supplied to the ac system, i.e. the converter is inverting. Figures 2.2.4 also illustrates the full
inversion at a delay angle of 180º.

Figure 2.2.4 Effect of delay angle α on the voltage

In the presence of a large smoothing reactor on the DC side, the voltage waveform of figure 2.2.4
will produce a constant direct current, the level of which depends on the mean voltages at both
ends of the link and the link resistance. For the idealized commutation the valve current will be a
rectangular pulse lasting 120º, its relative position with reference to the corresponding voltage
waveform being determined by the firing delay α.
From figure 2.2.4 the equation for the average DC voltage can be derived as follows.

π π
6 6
1
Vd = ⋅ [{ ∫ 2 ⋅ Vac ⋅ cos(ωt ) ⋅ d (ωt )} −{ ∫ 2 ⋅ Vac ⋅ sin(ωt ) ⋅ d (ωt )}] 
→
π 3 π π
− −
6 6
(2.2)
3 2
Vd = ⋅ Vac ⋅ cos α = 1.35 ⋅ Vac ⋅ cos α 
→Vdo ⋅ cos α [26] (2.3)
π
With Vd = DC output voltage
Vac = AC r.m.s. line-to-line voltage
ω = angular frequency (2πf)

21
 3 2
Vdo=ideal no-load direct voltage(with α=0) Vd = ⋅ Vac
π (2.4)
α = firing delay
Note the direct voltage being positive for 0º < α < 90º and negative for 90º < α < 180º.

Moving on to the practical commutation process, the zero-impedance supply required to produce
the voltage and current waveforms described, does not exist. Even if the AC system impedance
were negligible, there is considerable transformer leakage reactance between the converter and
the AC system. In theory, the converter transformer is not essential to the process of static power
conversion. However, there are practical reasons for using converter transformers, like the
possibility of phase shifting multiple bridges and the availability of on-load tap changing, which
will become apparent when discussing harmonics and reactive-power compensation later on. The
main effect of the AC system reactance is the reduction of rate of change of current, or in other
words, to lengthen the commutation time. During the commutation, the magnetic energy stored
in the reactance of the previously-conducting phase has to be transferred to the reactance of the
incoming phase.

To understand the entire conversion, let us first consider the commutation process between
valves 1 and 3 of the converter bridge, connected to a system with a source voltage vc, a
commutation reactance per phase Xc(and negligible source resistance). With reference to figure
3.2.5a, commutation from valve 1 to valve 3 can start(by the firing of 3) any time after the upper
voltage crossing between vCR and vCY(and must be completed before the lower crossing of these
two voltages. Since vCY > vCR, a commutating current ic(=ic3) builds up at the expense of i1 so
that all times i1+i3=Id.
The instaneous expression for the commutating current is:
Vac
ic = [cos α − cos(ωt )]
2 Xc (2.5)
The average direct current IdR at ωt = α + µ yields
Vac
Id R = [cos α − cos(α + µ )]
2 Xc (2.6)

22
Figure 2.2.5 Commutation proces with series reactance X per phase

Typical voltage and current waveforms of an entire six-pulse bridge operating as a rectifier with
the commutation effect included are shown in figure 2.2.6, where P indicates a firing instant(P1 is
the firing instant of valve 1), S indicates the end of a commutation(at S5 valve 5 stops
conducting) and C is the crossing(C1 indicates the positive crossing between phases B and R).
Figure 2.2.6a illustrates the positive(determined by the conduction of valves 1, 3 & 5) and the
negative (determined by the conduction of valves 2, 4 & 6) potentials with respect to the
transformer neutral. Figure 2.2.6b shows the direct voltage output waveform. The potential
across valve 1 depends on the conducting valves, is also shown in 2.2.6b. Figures c and d
illustrate the individual valve currents at the cathode(1, 3 & 5) and anode(2, 4 & 6), respectively.
From figure 2.2.6 the derived average output voltage or the mean direct voltage is:
1
Vd R = Vdo[cos α + cos(α + µ )] (2.7)
2
Where Vd0 is maximum average dc voltage(at no load without firing delay)
3 2
For three-phase bridge configuration Vdo = ( )Vac (2.8)
π
and Vac is the line to line r.m.s. commutating voltage referred to the secondary or the valve side
of the converter transformer. The value of the commutation angle is normally not available and a
more useful expression for the DC voltage, as a function of the DC current, can be derived from
(2.7) and (2.8), resulting in
3 Xc
Vd R = Vdo cos α − Id (2.9)
π
Where the last part of (2.9) compared to (2.3) represents the reduction due to commutation
reactance Xc.

23
Figure 2.2.6 Typical six-pulse rectifier operation signals
a. Positive and negative voltyages with respect to transformer nuetral
b. direct bridge voltage Vd and voltage across valve 1
c, d. Valve currents i1 to i6

The inverter operation signals are given in Appendix B.

For inverter operation the expression for the direct voltage Vd and currents Id in terms of β and γ
are:
1
Vd I = Vdo[cos β + cos γ ] (2.10)
2
Vac
Id I = [cos γ − cos β ]
2 Xc (2.11)

24
2.3 Reactive power and harmonics

Owing to the firing delay and commutation angles, the converter current in each phase always
lags its voltage, see figure 2.3.1. The rectifier therefore absorbs lagging current: consumes
VARs. In the presence of the perfect filters, no distorting current flows beyond the filtering
point, and the power factor can be approximated by the displacement factor(cos φ), where φ is
the phase difference between the fundamental frequency of the voltage and current. Under these
idealized conditions, with losses neglected, the active fundamental AC power is the same as the
DC power, i.e.:
Pdc = Vdc ⋅ Idc and (2.12)
Pac = 3Vac ⋅ Iac ⋅ cos ϕ (2.13)
Pac = Pdc ⇔
Vdc ⋅ Idc
cos ϕ = (2.14)
3 ⋅ Vac ⋅ Iac
Where Iac is the ac current without harmonics. This AC current can be obtained from the r.m.s.
π

1 3 2 2 ⋅ Idc
of the rectangular current waveform through [( ) ⋅ ∫ I dc ⋅ d (ωt )] = (2.15)
π π 3
−
3

Id 6
With Fourier analysis applied, the fundamental current is Iac = (2.16)
π
Substituting (2.7) and (2.16) in (2.14) gives cos ϕ = 1 [cos α + cos(α + µ )] (2.17)
2
The reactive power is often expressed in terms of the active power i.e.
Q = P ⋅ tan ϕ (2.18)
Similarly for (2.17) the following expression can be written for the power factor of the inverter
cos ϕ = 1 [cos γ + cos β ] (2.19)
2

Figure 2.3.1 Voltage and current displacement as a result of α

25
Referring again to the AC-voltage and valve current waveforms in figure 2.3.1 it is clear that the
current supplied by the inverter to the AC system lags the positive half of the corresponding
phase-voltage waveform more than 90º, or leads the negative half of the same voltage by less
than 90º . It can either be said that the inverter absorbs lagging current or provides leading
current, both concepts indicating that the inverter, like the rectifier, acts as a sink of reactive
power. This point is made clear in vector diagram of figure 2.3.2a. Equations (2.13), (2.17) and
(2.18) show that the active and reactive powers of a controlled rectifier vary with the cosine and
sine of the control angle, respectively.

Figure 2.3.2a Vector diagrams of current and P&Q at rectifier and inverter b. Q demand at
different P

Thus, when operating on constant current, the reactive power demand at low powers (φ= 90º)
can be very high. However, such an operating condition is prevented in HVDC converters by the
addition of on-load transformers tap changers, which try to reduce the steady-state control
angle(or the extinction angle γ) to the minimum specified. Under such controlled conditions,
figure 2.3.2b shows a typical variation of the reactive power demand against the active power of
an HVDC converter; the reactive power demand is shown to be approximately 60% of the full
active power transmitted.

In analogy with equation (2.18), for the reactive power that is consumed by the converter can be
written:
Q = 3 ⋅Vac ⋅ Iac ⋅ sin ϕ (2.20)
Q = three - phase reactive power

26
This would suggest that in case of no firing delay, the converter does not consume reactive
power. This suggestion will only hold for as long as the theoretical case of no source inductance
is considered. In practice, due to the presence of source inductance and the subsequent fact that
commutation takes a finite amount of time, the current waveform is no longer perfectly
rectangular but the front and tail are slanted, already shown in figure 2.2.6c and d. From this it
can be concluded that the converter will always consume a certain amount of reactive power,
even in the case of α = 0.
A realistic equation for the reactive power consumption is given by (2.23), which is derived
from:
Vdc ⋅ Idc
Pdc = Vdc ⋅ Idc = 3 ⋅Vac ⋅ Iac ⋅ cos −1 (ϕ ) → cos ϕ = (2.21)
3 ⋅ Vac ⋅ Iac
Id 6 Vdc ⋅ π Vd
Iac = → cos ϕ = = → (2.22)
π 3 ⋅ Vac ⋅ 6 Vdo
 Vd 
Q = 3 ⋅Vac ⋅ Iac ⋅ cos −1   (2.23)
 Vdo 

Equation (2.23) is valid regardless whether the bridge operates in rectifier or inverter mode.

The reactive power consumed at the converter is at expense of the voltage regulation for the AC
network, which can result in instabilities at AC side. To compensate the reactive power
compensation capacitor banks are used.
With the application of shunt capacitor banks for reactive power compensation or power factor
correction, there is always the inrush current issue during energization. Also, the outrush current
from the capacitor bank is a concern when a line circuit breaker closes in to nearby fault. In order
to limit both the inrush and outrush currents series reactors can be used.

Harmonics
The term harmonics is used to define the sinusoidal components of a repetitive waveform and
these consist exclusively of frequencies which are exact multiples( harmonic orders) of the basic
repetition frequency(i.e. the fundamental). HVDC converters generate harmonic voltages and
currents on the DC and AC sides, respectively. Since the commutation reactance is low in
relation to the DC smoothing reactance, an HVDC converter acts, from the AC point of view, as
a source of harmonic currents(high internal impedance) and from the DC point of view, as a
source of harmonic voltage(low internal impedance). Excessive levels of harmonic current must
be prevented as they will cause voltage distortion, extra losses and overheating, as well as
interference with external services(e.g. communication). The obvious place to eliminate the
harmonics is the source itself.
The line currents at the AC side of a six-pulse converter have a rectangular shape, figure 2.3.3a,
which results in a significant harmonic content. Since these harmonics are undesired, they should
be avoided as much as possible. By making a shift in the primary and secondary voltages of the
transformer, a more sinusoidal waveform is achieved as seen in figure b. This shows the AC line-
current drawn by the converter together with the momentary AC voltage has a more sinusoidal
shape and the ripple in the DC voltage is reduced. By putting two six-pulse bridges in series with
a 30° phase-shift between them(one YY, the other YD), a more sinusoidal current is drawn and,
less harmonics are produced(figure 2.3.4c). This is the 12-pulse converter bridge which is
covered in the next paragraph.
27
Figure 2.3.3 Voltage and current waveforms of a 12-pulse converter.

With the setting of figure 2.3.3c the 5th and 7th harmonics are effectively eliminated on AC side
and on DC side the 6th and 18th are eliminated.
The switching pattern of the converter is synchronized with the AC-side fundamental frequency,
and as such contains a large fundamental component. As switching is an on-off process,
harmonics of the fundamental are present as well. These manifest themselves as the
characteristic harmonics on both sides of 12-pulse converter bridge, i.e. on the DC side,
harmonics 12n, and on the AC side, harmonics 12n+1 in positive sequence, and 12n-1 in
negative sequence, where n is an integer. These components are always present, even under
ideal(undistorted AC voltage and DC current) operating conditions. Figure 2.3.4 shows the DC-
side voltages and AC-side currents that could be expected.

Figure 2.3.4 a. AC-side and b. DC-side harmonics

28
To eliminate the harmonics further filters can be used. Filters are networks of passive network
elements(RLC) tuned in such a way that only for specific frequencies they form a path of low
impedance to ground. Although many types of filters exist, the following are discussed in this
study:
• Single-tuned filter; sharply tuned to one of the lower order harmonic frequencies
• Double-tuned filter; nearly the same as the single-tuned filter, but tuned for two
frequencies
• High-pass filter; used to eliminate the higher-order harmonics.

Figure 3.3.5 Basic topology a. single tuned, b. double tuned and c. high pass filters

Commutation failure

Under normal circumstances, the voltage across the valve being turned off has to remain
negative for a certain period of time after the extinction of its current (after γ) so that it becomes
capable of blocking the forward voltage. Should the valve voltage become positive prematurely,
the valve may turn on even without a firing pulse, resulting in the failure of the commutation
process[22].
The application of forward voltage across a thyristor too soon after it has stopped conducting
will cause the thyristor to re−conduct and lead to a commutation failure. Commutation failures
normally occur at the inverter when there is a sudden drop in magnitude or phase shift in one or
more phases of the AC commutating voltage. Recovery from a commutation failure depends on
many factors. For long DC lines and cables the inverter dc current may become quite large due to
the stored charge in the line/cable capacitance. The rectifier current controller will increase the
rectifier firing angle in order to maintain the DC current at its set−point. To assist in the
recovery, the inverter DC control may include a circuit which transiently reduces the inverter
firing angle upon detection of a commutation failure. Reducing the inverter firing angle increases
the extinction angle and as such provides a longer time between turn−off and application of
forward voltage to the thyristors in the valve group. Detection of a commutation failure may be
done by comparing the DC current with the current flowing in converter transformer secondary
windings. During a commutation failure the DC current will be much larger than converter
transformer secondary windings.

29
3. HVDC developments
3.1 Practical HVDC connections
From recent paragraph it is seen which components build a HVDC connection, but still the most
important is the 6-pulse thyristor bridge(fig 3.1.1a). This 6-pulse thyristor bridge can be
connected in several ways to convert from ac to dc visa versa. Today's power systems are almost
exclusively AC systems. At places where HVDC is used, it is only because at that particular
place AC connections are technically not feasible or economical unattractive. Thus the DC
connection serves as a hub for ac networks, which means practical HVDC connections always
terminate at an AC bus. Taking this reasoning one step further it follows that for the integration
of a HVDC connection in an AC network, two converter stations are required: one that operates
as a rectifier and the other operates as an inverter. In this way, a power flow from rectifier to
inverter is established. With the 6-pulse bridge as main component, two basic configurations are
available to construct a HVDC link. Both configurations are shown in figure 3.1.1. Note also that
since thyristors cannot conduct reverse current, the direction of the current flow through the
connection is determined by the design (orientation of the bridges i.e.) of the connection and thus
is unalterable.
• Mono-polar system. This system consists of a single pole elevated at the DC voltage
with respect to ground, figure 3.1.1b. As a return in principle the ground or a metallic return
(conductor) may be used. In case of a ground return however, care should be taken since high
ground currents can have undesired side-effects like corrosion of nearby installed pipelines.
Because of this, the mono-polar system is not recommended.
• Bi-polar system. For each converter station in this system, two converters are connected
in the way depicted in figure 3.1.1c. With respect to ground the two poles are at a DC voltage
equal in amplitude, but opposite in polarity. This configuration offers numerous advantages
when compared to the mono-polar system. These include higher ratings while using the same
infrastructure, absence of ground currents under balanced conditions and the possibility to
continue operation at half of the rated power in case of a pole outage.

Figure 3.1.1 a. Six-pulse thyristor group b. mono-polar and c. bi-polar configuration

Depending on the requirements of a certain HVDC link, these two basic configurations can be
connected geographically in different ways, such as:
1. Back-to-back; for interconnections between power system networks of different
frequencies (50 and 60 Hz)

30
 2. Two-terminal; Transmitting power from a rectifier terminal to an inverter terminal,
typical of most HVDC transmission systems
3. Multi-terminal; When three or more HVDC substations are geographically separated with
interconnecting transmission lines or cables, the HVDC transmission system is multi-
terminal, which can further categorized as:
i. Series multi-terminal
ii. Parallel multi-terminal
iii. Hybrid
4. Unit connection; right at the point of generation, the converter transformer of the rectifier
is connected directly to the generator(for example hydro and wind).

Figure 3.1.2 Connection types HVDC link

A converter station can be roughly divided into two main sections, namely AC and DC part.
Figure 3.1.3 shows a HVDC converter station indicating the main components areas.

Figure 3.1.3 Typical HVDC converter station

31
The valve hall of a converter station actually forms the bridge between the AC and the DC
voltage since it houses the converter bridges. Up until now, a valve was treated as if it was just
one single unit
Because of the limited voltage rating of the individual thyristors, many of them must be
connected in series to constitute a HVDC valve. The series connection of thyristors requires
additional passive components to distribute the OFF state voltage uniformly between them and to
protect the individual thyristors from excessive rate-of-rise of voltage( dv dt ) and rate-of-inrush
current( di dt ). The thyristor, together with its local voltage-grading and thyristor-triggering
circuits, known as a thyristor level, is the building block of the valve architecture. The circuit of
a typical thyristor level is shown in figure 3.1.4a. It contains several parallel en series
components but for simplicity the thyristor level can be as given in figure 3.1.4b. By connecting
a suitable number of thyristor levels in series, a valve of the necessary voltage can be
constructed.

Figure 3.1.4 a. Thyristor level b. Simplified

3.2 Trends in HVDC technology

HVDC has become the dominating technology for long distance transmission of bulk power. The
use of 800 kV HVAC that was introduced in several countries during the 1960´s and 1970´s has
come to a halt [10]. The HVDC development started with the transmission of power in an order
of magnitude of a few hundred MW and was continuously increased to transmission ratings up to
3 - 4 GW over long distances. The rapid development and the increased confidence in the HVDC
technology have caused the transition from ac to dc as transmission alternative. Almost 80 GW
HVDC transmission capacities have been installed worldwide by 2005 and the expectation is that
the growing trend will continue. Over 104 GW are expected from China only[1].

32
Figure 3.2.1 Growth HVDC connections worldwide [11]

As mentioned earlier, several reasons are there to use the HVDC option. One of them worthwhile
repeating and going into some detail is the connection of different widespread small networks
into one large power system. This is important from every transmission system operator (TSO)
point of view. Such a large power system is the UCTE system in western Europe, which has
been extended step by step to the today very complex configuration. The UCTE(Union for the
Coordination of Production and Transmission of Electricity) is further connected to the
surrounding networks and discussions are in progress for further connections. It can easily be
concluded this extension surely will continue[1][13].
Other existing large interconnected systems are the western and eastern USA and USA with
Canada. Large blackouts in America and Europe confirm clearly, that the favorable close
electrical coupling by AC might include a strong risk of uncontrollable cascading effects in large
and heavily loaded interconnected systems. Here HVDC might play an important role. This
because of the major benefit of an HVDC link`s ability to control the power flow and its
flexibility to adapt to different AC system characteristics at both sides of the interconnection[1].

Among the development of HVDC systems in the last 10 years the main avenues compared with
the technology of 1990 is: The traditional classic HVDC technology(LLC) is still dominating but
with improved equipment and sub-systems(e.g. valves, dc-bushings, AC-filters, DC-filters etc.).
The new HVDC using (VSC) using IGBTs instead of thyristors is still in its research and
experimental phase with only few operational links[10]. The trend due to these developments, is
that LCC will still evolve and will be installed intensively[11]. Therefore the operation of HVDC
must be well understood.

At present, the largest thyristors available, can withstand voltages of > 7 kV and are able to
conduct currents well over 3 kA[5]. From this point of view it is not surprising that the thyristor
is used extensively in HVDC applications since whenever used, these installations are typically
33
designed for the transmission of bulk power at considerably high voltage levels. Appendix C
shows an overview of the HVDC projects operated and planned in the course of time.

3.3 HVDC from TenneT point of view

Presently one HVDC link is in operation in Netherlands, one is under construction and studies
are carried out to build more HVDC links connecting the Dutch grid with other distant grids. The
HVDC connection in operation is the NorNed, which connects Netherlands and Norway, the
connection under construction is the BritNed, between the Netherlands and Britain. The other
possible future connections are Cobra(Netherlands and Denmark) and a second link between
Netherlands and Norway, viz NorNed II.

NorNed
With a total length of 580 kilometers, the NorNed cable has a capacity to transmit 700 MW
electricity from the Netherlands to Norway, vice versa- enough to supply power to half of
Amsterdam or Oslo[12]. The cable was completed in May of 2008. The great advantage of the
NorNed cable for TenneT and Statnett(TSO of Norway) is the fact that the cable enhances the
security of supply of electricity in the Netherlands and Norway. This has strengthened TenneT’s
position in the Northwestern European electricity market and Statnett’s position in Scandinavia.

Figure 3.3.1 Dutch grid

(Red- 380kV;Green- 220 kV
Purple- HVDC)

34
The connecting points of the NorNed cable is situated in Feda, on Norway’s south coast, and
Eemshaven in the very north of the Netherlands. The Dutch electricity grid already had AC
interconnections to Germany and Belgium before the completion of the NorNed project, while
Statnett had cables linked to Sweden, Finland, Denmark and Russia. The NorNed cable
contributes to the overall security of supply. This point makes the Eemshaven area an import
node in the Dutch transmission grid.

Britned
The BritNed is a bipolar configuration, and is the first power transmission project establishing a
link between Great Britain and the Netherlands. The interconnection is rated to 1000 MW, but
capable to transmit 1200 MW power in both directions. The converter stations are connected by
a submarine DC cable of about 250 km length. The converter station in Great Britain is located at
the Isle of Grain near London and the converter station in the Netherlands is located at
Maasvlakte near Rotterdam. The project is a joint venture between National Grid (large
international electricity and gas company in Great Britain) and TenneT. From environmental
point of view, it is now possible to exchange green electricity between both power systems
depending on actual generation and demand. The link will go into commercial operation in the
first half of 2011. With the BritNed HVDC coming online it is shown, along the already existing
HVDC interconnection Cross-Channel between Great Britain and France, the interconnection of
different grids is continuing[10][12][14][15].

Cobra
A feasibility study has been carried out concerning another link connecting Denmark and
Netherlands, which`s converter station probably will be in Eemshaven. It is expected that a 700
MW interconnector between the Netherlands and Denmark will improve the accessibility of the
electricity market and the security of supply. Integration of additional renewable energy sources
(wind power) will be possible. The utilization of the transmission capacity is expected to be high.
Endrup in Denmark and Eemshaven in Netherlands are the preferred locations for the HVDC
converter stations. In the feasibility study both LCC and VSC have been considered from the
perspective of investment costs, capitalized losses, risks and the added-value of ancillary
services[16][24]. Results were slightly in favour of LCC. The Cobra Cable is scheduled to be
ready for operation in 2016.

NorNed II
Cobra is not the only link which is evaluated to be placed in Eemshaven. Another possible link to
enforce the exchange of power between Norway and Netherlands is the Norned II[17]. This link
might be placed in Eemshaven too. Recent developments are that large generation units are to be
placed at Eemshaven. Studies are still carried out.

With large generators situated in Eemshaven and yet more to come, together with planned
HVDC links, the Eemshaven area plays a significant role from stability point of view. Therefore
investigating in the already existing network, aiming at the existing NorNed HVDC link, is an
important tool to foresee congestions in the grid.

35
Using the fast controllability of the HVDC, it is possible to enhance the stability of both power
systems. As the need for more flexible networks grows, HVDC will play an important role from
a control point of view for stabilization and most importantly to connect widespread networks
economically advantageously. Therefore investigation is needed in HVDC systems already
existing, new more attractive DC technologies and of course the interaction between these two.

3.4 NorNed at Eemshaven

Some plots are given showing the actual situation at Eemshaven around the DC substation EDC.
Substation EDC in Eemshaven is situated in the North of Netherlands, also seen in figure 3.4.1.

Figure 3.4.1 Eemshaven area; EEMS and EDC

In the figure above it is seen two AC high voltage transmission lines end at substation
Eemshaven(EEMS), these are 220kV and 380 kV lines, green and red lines respectively. From
the 380kV AC substation the connection is made to the DC substation(EDC) with an
underground cable at a distance of 1500 meters, dashed red line. Here the conversion is made
from AC to DC(vice versa), and from here on the transmission is done by DC to(from) FEDA,
shown in purple line. See figure 3.4.1, here a floor plan is shown, in figure 3.4.2 the actual
situation is depicted.
36
Figure 3.4.2 Actual layout EEMS and EDC

In figure 3.4.2 the AC substation EEMS is seen at the near and at the far end EDC is seen. EDC
actually consists of two buildings. One building where the filters are situated, known as the filter
building and the other where the thyristors are for de AC-DC vice versa conversion known as the
thyristor hall. In figure 3.4.3 both buildings are shown at EDC, the flat low building is the filter
building, the higher building is the thyristor hall.

37
Figure 3.4.3 Filter building and Thyristor hall EDC

As the name indicates, in the filter building the filters for the elimination of the harmonics are
installed. Besides the filters, also the capacitor banks needed for reactive power compensation
are here. In the thyristor hall the thyristors are installed. These two buildings are separated by
some distance. The equipments in these two buildings are connected by a 50 meters underground
cable again, of the type used for connection between EEMS and EDC. The incoming point in the
filterhal of the three phase AC cables are seen in figure 3.4.7.

38
Filter building

Figure 3.4.4 Snapshot 1

Figure 3.4.5 Snapshot 2

39
Figure 3.4.6 Snapshot 3

Figure 3.4.7 Snapshot 4 Incoming three phase AC cables in Filter hall

40
Thyristor hall

Figure 3.4.8 Snapshot 1, Thyristor valves

41
Figure 3.4.9 Snapshot 2

Figure 3.4.10 Snapshot 3

42
4. Network modeling
The behavior of power systems is related to its ability of responding to the change in
energy/power balance of a system, i.e. the ability to store a fraction of energy change into its
electrical and mechanical components during power unbalance. Depending on the response time
for the power unbalance there are three ways in which the power system can store a fraction of
the energy; primary energy conversion, mechanical inertia and passive electrical components
(LC parameters of the grid). As a result of this energy storing and transferring, there are three
phenomena that occur in power system [19].
1. Electromagnetic phenomena
Electrical transient of the network are within time scale of 10-6 seconds to 10-3 seconds. In this
phenomenon the system tries to store the fraction of energy in its electrical components like the
inductors and capacitors. But the charging and discharging time constants of these components
are very small. Hence, the resulted transient phenomena can have large oscillation within this
time frame.
2. Electromechanical phenomena
Dynamic response of system generators due to their rotating inertia is within time scale of 10-3
seconds to 10 seconds. In this phenomenon the system tries to store a fraction of the energy into
its mechanical component (the rotor of the machines) in terms of its rotor angle δ. Compared to
the electrical components the mechanical component has larger time constant to store the energy
change.
3. Thermodynamic phenomena
The energy conversion process from primary energy sources within time scale of seconds to
hours. In this phenomenon the large fraction of energy change can be controlled by monitoring
the conversion of the primary energy source. The response time is longer compared to the other
phenomena.

Corresponding to the above there are at least three types of simulation methods to investigate
power system phenomena: electromagnetic transient, power system dynamic and power dispatch
(load flow) simulations.
1. Electromagnetic Transient (EMT) simulation
It is also called instantaneous value simulation, it is used to simulate an electromagnetic transient
phenomena. For this simulation method the model of the power system components is complex
as smaller time constant parameters are considered. To solve the equations of the network model
the required time step is in µs.

2. Power System Dynamic (RMS) simulation

This simulation method deals with electromechanical transients and neglects the electromagnetic
transients of the network. Hence, it considers only the fundamental frequency components of
voltage and current. The complexity of the component models is reduced by neglecting
differential equations that involve smaller time constant parameters. Hence, compared to EMT
simulation longer time step can be used to solve the network equation. A typical time step used
for power system dynamic simulation is 10 ms.

43
 3. Load flow and power dispatch simulation
This simulation method is used to solve the steady state power flow equations of the network.
The time constant for the steady state operation (thermodynamic process of primary energy
conversion) is in the order of several seconds to minutes. So all the differential equations
involved in the network model are assumed to be constant. Hence the power flow equations
become algebraic equations that can be solved using an iteration method like Newton-Raphson
method.

The understanding of electrical power systems in the µseconds area is still developing. With the
development of power electronics which are more and more incorporated in grids, the
investigation at smaller time scales becomes essential.
In this study the Real Time Digital Simulator-RTDS installed at Delft University of Technology
is used. The RTDS is a digital system built to perform real time simulations on electrical power
systems. This is a strong property compared to many other simulation tools which are only able
to perform off-line simulations. The simulation tool is enlarged in the following section together
with de operation. Followed by the modeling of the investigated network.

4.1 The simulator

The RTDS is a special purpose computer, initially designed to study Electro Magnetic Transient
(EMT) Phenomena in real-time [20]. This digital simulator is comprised of both specially
designed hardware and software. In other words, it is a sophisticated digital simulator both in
computing hardware and detailed models of power system components. RTDS hardware is
Digital Signal Processor (DSP) and Reduced Instruction Set Computer (RISC) based, and
utilizes advanced parallel processing techniques in order to achieve the computation speeds
required to maintain continuous real-time operation. The software called RSCAD has a power
system component model library and Graphical User Interface (GUI) environment to assemble a
given network.

Hardware
Unlike analogue simulators, which output continuous signals with respect to time, digital
simulators compute the state of the power system model only at discrete instants in time. The
time between these discrete instants is referred to as the simulation time-step (∆t). Many
hundreds of thousands of calculations must be performed during each time-step in order to
compute the state of the system at that instant. The temporary transients class of studies for
which the RTDS is most often used requires ∆t to be in the order of 50 to 75 µsec (frequency
response accurate to approximately 3000 Hz). By definition, the entire network calculations are
completed in less than 50 µsec of actual time[20].

The RTDS is a modular concept where multiple 19" processor cards are combined to form a
single rack. In turn, a complete simulator consists out of one or more racks. There are different
cards that contain the digital signal processors. These different types of processor cards are each
suited to perform specific tasks. Qua functionality the cards can be categorized in processing and
communicational card.

44
Processing cards
• 3PC- Triple Processor Cards; are used to run the simulation. The number of power
system and control system component models that can be included in the simulation is
determined by the available number of these processors. The processors need 25 ns to
perform each instruction cycle of a component model
• RPC-Risc Processor Cards; are only used to solve the network during simulation. They
need 1.67 ns to perform each instruction cycle of a component model.
• GPC-Giga Processor Cards; can be used to solve the network during simulation. Plus
they have analogue output channels for signal exchange with external equipment. They
need 1 ns to perform each instruction cycle of the component model, mostly these
processors are used for small time model of power system components that need small
time steps up to 2.5 µs[19].

Communication cards
There are two communication cards; that facilitate data exchange between racks and between
racks and the work station computer.
• Inter-Rack Communications Card (IRC): In multi-rack simulation cases the data
exchange between processors residing on separate racks is accomplished through high
speed communication channel mounted on the IRC.
• Workstation Interface Card (WIF): In addition to IRC, each rack has one WIF card that
facilitates the communication between the RTDS and the host computer workstation
through ethernet cable based local area network. WIF do not have any effect on network
solution, it regulates and facilitates data package exchange both between racks and
between racks and the workstation computer during simulation.
Furthermore there are cards for digital and analog input/output which are:
• Digital Input/Output card (DIO)
• Optical Analogue-Digital Converter card (OADC16).
• DIgital Time Stamp card (DITS)
But no real-time connection and simulation were done in this study, so the last category was not
used.

The specific functions of the cards can be found in reference [20].

45
Figure 4.1 RTDS installed at the Delft University of Technology

The RTDS at the Delft University of Technology that is used for the simulations consists out of a
total of eight racks. Each of these eight racks is composed of a WIF card, seven 3PC cards, a
RPC card and one DIO card. With this configuration each rack is capable to solve 54 single
phase nodes and 56 breakers.

The RTDS is operated by means of workstations on which the dedicated software package is
installed. For communication between the workstations and the RTDS a standard 10/100 Mbit
Ethernet connection in combination with the TCP/IP protocol are used.

Software
The RTDS software includes accurate power system component models required to represent
many of the complex elements which make up physical power systems. The overall network
solution technique employed in the RTDS is based on nodal analysis. The software package
developed to provide a fully graphical interface is RSCAD. The RSCAD software suit is built
around four separate modules that provide specific functionalities for modeling and simulation of
power system equipment.
• Draft module. The Draft module is very much comparable to other (off-line) simulation
tools. It consist of a sheet on which the model can be drawn and a library where a variety
of power system components are available. After having copied a component from the
library into the model, the user is then able to modify its parameters to match them with
those of the component to be modeled.
• Cable module. In contrast to other components, the parameters for high-voltage cables
used in the model are not entered directly in the draft module. Instead this is done in a

46
 separate cable module. The cable module outputs a file containing all the properties of
the cable per unit of its length. This file can then be linked to the connections in the
model where the type of cable defined by the output file is used. A single file can thus be
used for multiple connections.
• T-line module. Is essentially the same as the cable module except that it applies to
overhead lines instead of high-voltage cables. As such, in addition to the properties of the
conductors, also the geometry of the pylons must be defined.
• Runtime module. Serves as the link between RSCAD and the RTDS. Through the
Runtime module a compiled version of the model can be uploaded to the RTDS and data
produced by the simulation can be downloaded from it. The downloaded data is presented
to the user in a graphical way. Furthermore the user can interact with the model while it is
running by means of runtime controls. With software versions of buttons, switches and
sliders the user can simulate a whole array of events and observe their impact on the
simulated system in real time[5].

Once the network is assembled, compiled to map each component to the processors, the low
level machine language for the Runtime module is generated, the simulation can be started.

Before going to the modeling of the network a little clarification is needed concerning the time
step (∆t) of a simulation. In RSCAD there are two types of transmission line models; travelling
wave transmission line model and PI section models. For the travelling wave transmission line
model there is a separate data entry module in RSCAD. In this module it is possible to enter
either physical parameter data or equivalent RLC values.
One of the constraints using the T-Line module relates to the overall length of the line being
represented. When the modal propagation time (or “travel time”) of a line is less than the chosen
simulation time-step ∆t, the line cannot be represented using these general travelling wave
models. This limitation is a result of the calculation algorithm. The travel time of the line is
directly related to the line length, and hence it may found that, for short transmission lines, PI
section representation will be required to represent length accurately.
Normally as lines become shorter, the approximations resulting from using PI section modeling
become less significant. Although using PI sections made up of R,L and C components is not
normally the recommended method for line representation on the RTDS, there are instances
where length of the line of a network dictates that such modeling techniques must be used.

Generally, the propagation velocity of an electromagnetic wave through a transmission line or a

cable depends on the inductance and capacitance values. For a given value of L and C, the
propagation velocity of the electromagnetic wave[21] is given by:
1
vp =
L⋅C (4.1)
Where vp is the propagation velocity in m/s. Therefore, the travelling time tp of the
electromagnetic wave is given by
tp = L ⋅C
(4.2)
Assuming propagation velocity is equivalent to the speed of light, a 50 µsec time step would see
the waveform travel a total distance of approximately 15 km. This means that if the chosen
simulation time step ∆t is 50 µsec, any line of length less than about 15 km would have to be
represented using a PI section.

47
4.2 Modeling the primary circuit
The NorNed HVDC link is as depicted in figure 4.2.1. The model is based on documentation
available at TenneT. The HVDC transmission is designed to operate in the following operation
mode: The converter system is a bi-directional bipolar system without electrodes. Each converter
system is equipped with two times 6 pulse valve bridges and can be considered as a 12-pulse
system at the ac-side. The dc cable connecting the two converter stations is a 580 km submarine
cable at each pole. The system is designed to transport a nominal dc power of 600MW, with a
minimum of 60MW and maximum of 1000MW. These powers are reached with an operating dc
voltage of +450 and -450 kV, with currents ranging from 670A to 1252A for the earlier
mentioned powers, where the nominal current is 686A.
The basic control mode is constant power control, achieved by controlling the dc current to a
current order. The dc current is controlled by the firing angle α at the rectifier. The dc voltage is
controlled by the extinction angle γ of the inverter. When the dc voltage is not controllable by γ
(so called dead band), the tap-changer of the converter transformers are used to reach the desired
voltage range and γ is then used to fine-tune to achieve the steady state. The nominal firing
angles α and γ are 15º & 20º respectively.

Figure 4.2.1 The NorNed configuration from Eemshaven to Feda

4.2.1 The components

Source models
As with most simulation studies, the entire network cannot be included in the simulator. But the
impact the entire network has on the studied network, vice versa, is important and cannot be
neglected. Therefore the network part that is not included in the simulator, is modeled as an
infinite source with a certain short circuit capacity(SCC). This SCC is translated to an internal
source impedance Zi in the source(Zi series to the infinite source). The short circuit capacity of a
network is given by:
SCC = 3 ⋅ Vac ⋅ Isc (4.3)
And the short circuit impedance:
Zi = Ri + jXi (4.4)
(Vac) 2
Zi =
SCC (4.5)
The data which is used for the modeling, is included in Appendix D.
From the available data sheets, the normal short circuit capacities was used for substation
Eemshaven and substation Feda. The related X/R ratio were also given. See table 4.2.1.

48
The calculated internal impedance with the data of table 4.2.1 for Eemshaven and Feda are given
in table 4.2.2.

The problem encountered while modeling this directly in the RTDS was that a resistor series to
an inductor as internal impedance was not possible. Though it was possible to use a resistor
parallel to an inductor, a conversion was needed. Conversion from the series to parallel
impedance was done and the values of the parallel resistor Rp and inductor Xp for substations
Eemshaven and Feda are given in table 4.2.3. Now the parallel circuit of resistor Rp and inductor
Xp form the internal impedance Zi.

Valve group
For modeling classic HVDC converters in RSCAD, the valve group block is available in the
library of the Draft module. The valve group block comprises a converter transformer, six-pulse
bridge and an (optional) smoothing reactor. Although the user is able to define the parameters for
each of the three components individually, no changes whatsoever can be made to the way they
are interconnected, nor can components be added or be removed.

Converter transformer
Modeling the NorNed converter transformers as it is configured in real was not possible. Each
converter substation at both ends (rectifier and inverter) has one converter transformer. This
transformer consists of three single phase three-winding transformers. One single phase three-
winding transformer feeds one phase of each valve group. See figure 4.2.2a. Table 4.2.4
summarizes the nominal ratings of one single phase three-winding.

Figure 4.2.2 a. Real configuration- (one) single phase three-winding transformer b. Modeled
configuration- (two) three phase two-winding transformer

49
The hurdle encountered while building the network was that the converter transformer included
in the valve group block(VGB) model, cannot be changed from a three phase two-winding
transformer to a one single phase three-winding transformer. Therefore a three phase two-
winding converter transformer was modeled. So the parameters of the transformers in the VGB
were changed to match the real transformers configuration. Figure 4.2.1 shows the conversion
from real configuration to the model.
Another option tried during modeling, was to change the converter transformer parameters in the
VGB to an ideal transformer and use a separate three phase three winding transformer from the
RSCAD library to feed the ideal transformers which are connected to the converters. This option
was omitted because of limitations present in the model of the single phase three-winding
transformer where no tapchangers were present.
The converted ratings used for each of the converter transformers in the VGB is given in table
4.2.5.

It should be noted that the nominal rating given in table 4.2.5 is for one converter transformer.
Two valve group blocks were used in each converter station. The converter transformer of one of
the VGB is connected in Y-Y and the other in Y-D. See figure 4.2.3. Tap changers for controls
are also included in the model. The full control of tap changers is discussed in chapter 5.

Figure 4.2.3 Six-pulse bridge in each converter station

Valve group
As mentioned earlier, to reduce operational stresses, the so-called turn-off snubbers are placed
across each thyristor. In essence these snubbers are a series connection of a capacitor and a
resistor so that during turn-off, the voltage across the thyristor is clamped to the capacitor
voltage. In practice, a number of thyristors are placed in series and combined with the necessary
electronics, snubber circuit and grading capacitors to form a complete valve. A schematic
representation of a thyristor with its electronics, called thyristor level was shown in figure 3.1.4.

50
In the NorNed case 120 of these thyristor levels were connected in series to form a valve group.
Data was available of this valve group, given in table 4.2.6.

Unfortunately the VGB in the RSCAD library is modeled as a single thyristor level as shown in
figure 4.2.4. These thyristors are modeled as control switches which have a ON state resistance,
OFF state resistance and its snubber circuit, with its a resistance and a capacitance Rsn and Csn
respectively.

Figure 4.2.4 Valve group model as one thyristor level with Rsn and Csn in RSCAD

It proved these snubber circuit values have a great impact on the precision of the simulation. The
data provided in table 4.2.6 for the snubber circuit did not produce correct signals. Various
attempts to fit the real data into the model failed. Finally the choice was made to use values of
Rsn and Csn calculated from the RTDS tutorial case which`s rated DC voltage was 28.4 kV what
is used as the base. This is in fact an interpolation. Formulas[20] used for calculating Rsn and
Csn:

121.67 ⋅ Vdc _ rated

Rsn =
28.4 (4.6)
−6
(1.2 ⋅ 10 ) ⋅ 28.4
Csn =
Vdc _ rated (4.7)

Where Vdc_rated is the rated DC voltage of NorNed. The calculated values used in the model
for the thyristor level are presented in table 4.2.6.

The “gmmin” parameter defines the amount of time required for the thyristor to re-acquire
forward voltage carrying capability after commutating to the off state. This parameter is usually
only of interest at the inverter where the HVDC controller must ensure that valve firing is such
that this minimum extinction angle is maintained. If forward voltage appears too soon after turn-
off the valve will begin to re-conduct resulting in a commutation failure. If a positive voltage
appears across one of the thyristors in the valve group within the time specified by “gmmin” the
thyristor will put back into the on state by the model. The minimum gamma for NorNed is 435µs
(table 4.2.7), which is equal to 7.83º, calculated from seconds to degrees with
γdeg=γsec • f • 360 (4.8)

51
Reactor
The last component in the VGB is the optional reactor. To reduce the ripple component on the
DC bus voltages and currents, at each end of both DC cables a smoothing reactor is present.
These reactors serve the following purposes:
• Decrease harmonic voltages and currents in the DC line
• Prevent commutation failure in inverters
• Prevent current from being discontinuous at light load
• Limit the crest current in the rectifier during short-circuit on the DC line.

Neutral bus reactor

The midpoint between the two valve groups at Eemshaven is not directly grounded unlike the
midpoint at FEDA. Here a reactor is used at the neutral bus. The rating of the reactors are given
in table 4.2.7

AC Filters
For the Norned (12-pulse) converter bridge, only harmonics of the order 12 n ±1 (with n=
1,2,3...) are still present in the harmonic spectrum. It is desirable to further reduce the harmonic
content, this is done by the application of filters. Filters are networks of passive network
elements tuned in such a way that only for specific frequencies they form a path of low
impedance to ground. Two types of filters are used in the NorNed link:
• double-tuned band-filters
• high-pass band-filters

To remove the two harmonics with the highest amplitude from the spectrum (i.e. the 11th and the
13th), a double-tuned filter is used, at Eemshaven and Feda. The topology of the filter is depicted
in the figures of Appendix E. The values for the resistance, inductance, and capacitance were
drawn from data sheets, which are given for both stations in table 4.2.8

For the harmonics of the order 12 n ±1 (with n= 2,3,4...) high-pass band-filters are used. This
high-pass filter is tuned around the 24th harmonic and will filter out all harmonics from the 23rd
onwards. The topology of the high-pass filter is shown in the figures in Appendix E. The data is
given in table 4.2.9.

Because of the low short-circuit power (and hence high inductive impedance at 3rd harmonic), a
3rd harmonic filter bank is required at higher levels of DC power at Feda. The topology of this
high-pass 3rd harmonic filter is also given in Appendix E and table 4.2.10.

Reactive power compensation

Due to the excessive reactive power consumption by line commutated HVDC converters
explained in chapter 2.3, it is necessary to provide additional voltage-support by the local
injection of reactive power. Although there are several ways to achieve this, such as with
synchronous condensers or static compensators, in the NorNed case capacitor banks are used.
Furthermore, the capacitors which are part of the AC filters also deliver a certain amount of the
required reactive power. For redundancy each converter station has two sets capacitor banks. To
prevent high frequency transients currents when switching on the second bank, a current limiting

52
reactor is placed in one of the sets capacitor banks. At Eemshaven as well as at Feda. The ratings
of the capacitor banks used is given in table 4.2.11.

The total reactive power injected at the ac sides of the converter stations by all the capacitor
banks and the filters, is given in table 4.2.12.

Note that each converter station has two sets of filters and banks installed, except the 3rd
harmonic filter at Feda. The entire network is found in Appendix E.

4.2.2 Transmission

Cables and transmission line

To model a transmission line there are two possibilities in RSCAD, either the distributed
parameter (travelling wave) line model or the lumped (PI equivalent circuit) can be used. When
the travelling time of the signal over the transmission line is less than the simulation time step,
the PI equivalent circuit must be used. For example, for integration time step of 50 µs and
propagation velocity of 3•108 m/s, all transmission lines with a length of less than 15 km are
represented by the PI equivalent circuit. When the travelling time of the signal is larger than the
integration time step, it is better to use the travelling wave line model. Using the travelling wave
line model was not possible for two pieces of cables of 50 meters and 1500 meters. These are
two pieces of XLPE cables of type EYLKrvlwd 220/380kV 1x1600CuMil from Prysmian.

PI section
In the DRAFT the cable base frequency, the positive and zero sequence quantities of the cable
which represents the PI section must be provided. Parameters are entered as the total impedance
of the PI section. The positive and zero sequence impedances of the XLPE cable are Z1 = 0.015 +
j0.245 and Z0 = 0.070 + j0.032 per km respectively. Data entered for the PI sections is
summarized in table 4.2.13 below. The values of the zero sequence reactance’s Xz and Xcz were
modified. This was necessary for successful compilation of the circuit in RSCAD. While
compiling with the original data there was an error generated, which required the “Xz must be
larger than Xp” and “Xcz must be larger than Xcp”. So these values were made slightly higher
than the positive sequence reactance’s, as seen in the table.

Positive seq. series resistance (Rp)

Positive seq. series inductive reactance (Xp)
Positive seq. shunt capacitive reactance (Xcp)
Zero seq. series resistance (Rz)
Zero seq. series inductive reactance (Xz)
Zero sequence shunt capacitive reactance (Xcz)
Table 4.2.13 Rating of XLPE cable(Short and Long)

The situation where the two pieces of cables are used, is illustrated in the figures of chapter 3.4
and later on in figure 5.1.0. The DC converter station is located at some distance from the AC
substation Eemshaven. Here the 1500 cable is used to connect the AC network to the filterhal in
the converter station. In the filterhal the required filters and capacitor banks are connected. From
here another piece of 50 meters cable connects the filterhal with the converters.

53
Travelling wave model

DC cable
By using the T-Line software module, the DC cables connecting the two converter stations were
modeled with a frequency depended travelling wave model. Using travelling wave model
produces more accurate results, especially where transient situations are concerned. Table 4.2.14
gives an overview of the parameters that characterize the cable. Noteworthy is that the total
length of de DC transmission is 580.5 km from both converter stations, this length is a composite
of different types of cables. First both ends are land cables, with a length of 3 km and 1.5 km at
Eemshaven and Feda respectively. The centre and most important part is a submarine cable of
576 km. This submarine has two parts, one of type ABB and the other of type Nexan. As the
major part of the cable was of type ABB, the cable model was based on the ABB data. Two sets
of these cable model were used, one for the +450kV cable and the other for -450kV. The inner
distance between these two cables was averaged and chosen to be 2 meters.

54
5 Modeling the controls

In this section the HVDC controller used to control the HVDC and other controls in the RSCAD
is explained. The control of the HVDC system described here is based on a generic HVDC
system. This generic control scheme was derived from the case in the tutorial of the RSCAD
library(version 1.02). The HVDC case was based on the monopolar HVDC Blackwater system in
USA. The term generic is used since each manufacturer of HVDC controls apply their own
proprietary concepts for HVDC control functions. The controls used in the tutorial[20] was not
based on any particular manufacturer’s control scheme, but rather on well documented control
concepts. If there is access to detailed schematics of a particular HVDC controller then the model
can be built in the RSCAD. This was not available for the NorNed link.
From various literature[6][23] it was found that the chosen generic control scheme in the tutorial
is general enough to be applicable to other HVDC systems. And from the documentation of
NorNed it was found this control scheme was specific enough to use it for investigation.
Another study based on the HVDC tutorial for the HVDC mentioned here, was used earlier for a
study at the TU Delft[5].

5.1 Control properties of model

Key advantage of a HVDC transmission system is that the power flow from sending to the
receiving end is fully controllable. By assigning the proper firing delay angles to both of the
converters, a voltage difference is created across a resistance with a fixed value. Following
Ohm's law, this gives rise to a current, and thus a power flow as is illustrated by figure 5.1.1
given by equation (5.1)

(VdA − VdB )
Id = (5.1)
Rdc

Figure 5.1.1 Fundamental principle of power transfer using a HVDC transmission system

Control of the converters is done by setting the firing delay angle (alpha-α) of the thyristor
valves with respect to the AC system voltage. In most HVDC systems the rectifier’s firing angle
is chosen so that the DC current is maintained at a set-point. The inverter’s firing angle is chosen
so that its extinction angle (gamma-γ) is maintained at a set-point. Measured values of dc current

55
and extinction angle are provided to regulators whose output are the rectifier and inverter firing
angle respectively. The extinction angle set-point is normally fixed at a certain point. The dc
current set-point may be altered depending on the dc power requirements for the system. Instead
of directly entering the DC current set-point, sometimes the DC power order is entered. In this
case the measured DC voltage is used to compute the current set-point as
Pset
Iset = (5.2)
Vdc
The entire control scheme is based on per unit calculation of the variables (measured variables
and variables to be altered). The basic HVDC control structure is illustrated in 5.1.2, which is
further elaborated in this chapter.

Figure 5.1.2 Basic HVDC control structure

From figure 5.1.2 it is seen that there are three major parts in the controls circled in red, these
are:
1. Master controller,
2. Rectifier controller, and
3. Inverter controller.
Furthermore a division can be made regarding input and output of the controller. The important
variables are:
1. Input variables,
a. User input: power set point and gamma set point
b. Measured input: operated current and operated gamma, additionally operated dc
voltage
2. Output variables, Firing pulses for rectifier and inverter

5.1.1 Master control

Control functions that are related to both the rectifier and inverter are performed by the master
control. These functions comprise the blocking and de-blocking of the converters as well as
giving the current order to the current controller of either of the two station controls. Since it is
desirable to have the two poles operating in a balanced condition, the current ordered by the
master control is calculated in the following way:

56
 Pset
Iord = (5.3)
DCV 1 + DCV 2

Figure 5.1.3 depicts the master control scheme. The important input variable is the DC power
set-point entered by the user, from this the current order is calculated using DCV1 and DCV2
which are measured dc voltages at positive and negative polarity. Note that all the values are
converted to per unit values. Once the current order Iord is produced using by means of (5.3), it
is used in the entire control concept. The rate at which the current order is changed after a change
in the power set-point is limited by the rate-limiter to 0.1 p.u. per second (ramp up/ramp down
speed). This improves the system stability and reduces the stresses experienced by system
components.
Once the current order is determined, the steady state condition is maintained by keeping the
current constant by the rectifier translated by α, whereas the voltage is kept constant by the
inverter translated by γ.

Figure 5.1.3 Master controller

The option to control the current is also added. First the user switches with a control button from
power to current set-point, followed by setting the current set-point with a slider in the runtime
window.

5.1.2 Rectifier
Recalling the basic control structure from figure 5.1.2, the rectifier consists of two important
blocks, the current regulator and the firing pulse generator. Under normal circumstances the
rectifier is operating in a constant current mode, thus controlling the current through the DC link.

Current controller
The rectifier is equipped with a PI current controller which has the function to eliminate the
difference between the current order and the actual current. The input for this block is the output
of the master controller Iord, and the actual current Crect1. The actual current is measured
between converter and smoothing reactor. This signal is passed through a first-order low-pass
filter first to prevent erratic behavior by the controller, and also converted to a per unit value.
The integrator output is limited to a value between 5° and 90°. Moreover, the output of the

57
controller is also limited to these exact same values by the application of a limiter. See figure
5.1.4.
Note that the current order which is coming from the master controller is not directly compared
with the measured current signal to serve the PI controller. The current order is first compared
with a signal coming from the so called Voltage Dependant Current Order Limiter block
(VDCOL).

Figure 5.1.4 Rectifier current controller

The VDCOL block is added to the earlier mentioned two blocks for the rectifier controller. What
the VDCOL does is explained below.

Figure 5.1.5 Voltage Dependent Current Order Limiter, VDCOL

The voltage dependent current order limit block is shown in figure 5.1.5. As is indicated by the
name this control limits the current order in the event of low voltage levels. Reduced voltage on
the inverter terminals may lead to continuous commutation failure. As some valves stay in
continuous conduction it is important to reduce the current order so as not to overstress the
valves. Also, the voltage level at the inverter being low, the reactive power consumption by the
converter will be higher than at rated voltages. As the DC current increases the amount of
reactive power required also increases which in turn may lead to a further reduction of the
voltage. In addition, the reactive power compensation provided by the capacitor banks and AC
filters falls as a square of the voltage. When the inverter terminal voltage is greater than the rated
voltage, the VDCOL output is equal to 1.0 and thus does not affect the current order issued by

58
the master controller. But if the input to the VDCOL drops below the threshold (0.9 pu), the
output falls according to the characteristic shown in figure 5.1.6. The current order is now the
minimum of the value ordered by the master control and the output of the VDCOL. Appendix F
shows the full rectifier and inverter control characteristics.
The AC bus voltage at the inverter or the DC bus voltage can serve as input to the VDCOL. In
this study the AC bus voltage was used (N3, N4, N5). Again the rated values are converted to per
unit first to be used in the control scheme.

Figure 5.1.6 VDCOL characteristic

Finally the minimum of the VDCOL and the Iord is chose and given to de PI current controller.
The resulting signal is the alpha order rectifier: AOR1, where 1 is indicating the upper pole.
This alpha order is given to the next control block.

Firing pulse generator

The second block of the rectifier controller is the firing pulse generator. Within this block there
are two functions. The phase locked loop (PLL) and the actual firing pulse generator (FPG). To
ensure correct bridge operation the thyristors in the rectifier must be fired exactly at α degrees
after they have obtained a positive forward voltage. This is achieved by supplying the FPG with
a synchronization signal. This synchronization signal is provided by a control system that
generates a signal that has a fixed relation to the phase of a reference signal, called a phase
locked loop. A PLL automatically raises or lowers the frequency of a controlled oscillator until it
matches the frequency and phase of a reference signal. The AC bus voltages at the rectifier serve
as the reference signals for the three-phase PLL, in this case(N1, N2, N3), because the signals
within the valve group are highly distorted. The firing pulse generator is depicted in figure 5.1.7.

59
Figure 5.1.7 Firing pulse generator with a PLL to determine the actual firing pulse instants

FPG
With the firing angle for the rectifier known from the controls, it remains to generate the firing
pulses for the six thyristors in each valve group. The firing angle refers to the delay in firing the
thyristors relative to the time at which the voltage across the thyristor becomes positive. A firing
angle of zero is equivalent to diode operation. Normally two thyristors are conducting with firing
sequence 6,1 −1,2 −2,3 –3,4 −4,5 −5,6 −6,1 −1,2… During the commutation period three
thyristors conduct. For example, between the 6,1 and 1,2 conduction periods the three thyristors
6,1,2 conduct for a short period (overlap time µ). The current through thyristor 6 decreases from
Idc to zero while the current through thyristor 1 increases from zero to Idc. Depending on the
type of transformer YY or YD, the FPG is configured with or without a lag of 0° or 30°
respectively. Finally the output are the pulse words for the valve groups.
The full rectifier control scheme is given in Appendix F.

Figure 5.1.8 Firing sequence in steady state is 6,1 −1,2 −2,3 –3,4 −4,5 −5,6 −6,1 −1,2 . . .

The control described above is just for one pole of the rectifier. A copy of the same blocks and
functions is present for the other pole. The same variable from the master controller Iord is used
here as input. Furthermore the measured valve current of pole 2 is used here as comparison and
60
the output from the FPG is given to the valve group of pole 2. The scheme for pole 2 is also
included in Appendix F.

5.1.3 Inverter

In order to comprehend operation of HVDC valve groups in inverter mode, it is important that
the concept of extinction angle(γ) be well understood. The extinction angle is a measure of the
time between the current through a particular thyristor becomes zero and the time at which
forward voltage appears across the thyristor (i.e. voltage across thyristor > 0). If a forward
voltage is applied to the thyristor too soon after the thyristor current has become zero (i.e. the
thyristor stopped conducting) the thyristor may begin to re-conduct even though no firing pulse
is present. This event of commutation failure is already explained in section 2.3. The fact that the
thyristor may re-conduct when γ is too small is a physical characteristic of the thyristor. Of
course this is a unwanted property. Figure 5.1.9 shows the voltage across a thyristor within an
HVDC converter operating as an inverter. The value of gamma shown in the figure is measured
in seconds. To convert from seconds to degrees (4.8) is used.

Figure 5.1.9 Commutation failure

Note that the extinction angle is not a continuous value, but may only be measured whenever a
thyristor stops conducting. For one 6-pulse valve group, 6 distinct values of γ can be measured
each cycle. The minimum value for the six measured values of γ for a cycle may be monitored in
the VGB. Although modern thyristors may be able to operate with γ values of a few degrees the
operating set-point is usually set higher so as to have some margin of safety. A sudden drop in
AC voltage will tend to reduce the actual value of γ. There is a trade-off, however, since the
higher the value of γ the more reactive power consumed by the valve group.
The minimum value of the extinction angle required for the proper operation of a valve is
specified by the valve manufacturer; however, at the inverter side of HVDC systems the
extinction angle is regulated to a value higher than the valve specifications to allow control
adjustments and also leave an adequate safety margin for unforeseen events in the power
system, such as faults. Severe faults such as the voltage drops, phase shifts, or sudden increase in
dc current may cause the commutation process to fail[23]. Recall from table 4.2.6 the minimum
gamma for NorNed is specified to 7.83°. But the operational γ is kept above 20º for NorNed.

61
Inverter control
The inverter controls are somewhat more complex than the rectifier controls. In the illustration of
the basic HVDC control structure (like the rectifier) two important blocks are included for the
inverter. At the inverter the first block is the gamma controller whereas at the rectifier it was the
current controller. The second block is the same firing pulse generator. Like there was an
addition for stability reasons at the rectifier (VDCOL), at the inverter there are also additions.

Gamma controller
This block is called gamma controller because the basis for this inverter controller is the gamma.
The inputs are the gamma set by the user and measured gamma from the VGB. Within this block
two modification functions exist. But even before the control signal coming from the master
controller reaches the gamma controller, it is compared to the VDCOL.

Under normal operating conditions the inverter is controlling the voltage on the DC link. To keep
this voltage at its rated value the inverter is controlled on the principle of a constant extinction
angle γ. That is, the controller generates a firing delay angle such that the extinction angle
remains constant. To reduce the amount of reactive power drawn by the HVDC system a fairly
low value for the extinction angle is chosen. The minimum is set to 20° determined from
documentations of NorNed and the max set to 30° to limit the reactive power consumption. This
set-point for gamma is called GAMORDR, see figure 5.1.10. The difference between this signal
and the minimum of all of the extinction angles both in pole 1 and pole 2 is eliminated by a PI
gamma controller to reach the stable point, set by the user. The integrator as well as the output of
the controller is limited from 90° to 155°. A margin angle of 25° is thus maintained to avoid
commutation failures.

Figure 5.1.10 Gamma controller

In addition to the gamma controller, also a current controller is present at the inverter controller.
This is one of the earlier mentioned extra functions at the gamma controller block. Under normal
conditions the current controller’s output is not used. However, if the dc current error is greater
than a pre-determined amount (which is chosen to be 0.1 p.u. as in the base case) the inverter’s
current controller is used instead of the extinction angle controller. A current margin is created
by subtracting ten percent (CMARG=0.1 p.u.) of the rated current from the current ordered by
the master controls. The reason for this current controller with a decreased current order is to

62
ensure a stable operating point in case the inverter terminal voltage exceeds that of the rectifier.
Apart from the current order and limits, which in this case are 110° to 155°, it functions in the
same way as was described for the rectifier current controller. The output of the current
controller becomes active when the firing delay angle generated by it, is smaller than that of the
gamma controller. The minimum function selects this from the gamma PI and the current PI
controllers. Figure 5.1.11 illustrates the current controller of the inverter.

Figure 5.1.11 Inverter current controller

The second function added to the gamma controller block is the Inverter Delta Gamma Error.
Under certain conditions (caused by the gamma PI controller and the current controller PI,
explained later) there may exist two valid operating points, which can cause oscillations. To
eradicate this oscillation, the delta gamma error (DGE) is added before the Iord reaches the
gamma PI controller. The delta gamma error function is given in figure 5.1.12.

Figure 5.1.12 Delta Gamma Error controller

There is also a VDCOL included at the inverter. The Iord which was coming earlier from the
master controller, is first compared to the AC voltage at the inverter, with the same reason as at
the rectifier: to limit the current order in the event of low voltage levels at the inverter. See figure
5.1.13 This function is executed according to figure 5.1.6. Whenever the AC voltage falls below
0.9 pu, the current ordered decreases proportionally towards 0.5 pu, else the Iord from master
controller is used.

63
Figure 5.1.13 VDCOL at the inverter

Firing pulse generators

Now the inverter firing angle order (AOR2) known from the gamma controller(or inverter
current controller) the signal is transferred to the second block of the inverter controller(from
figure 5.1.2). This is the firing pulse generator. The firing pulse generator at the inverter is
exactly the same as it is at the rectifier. Within this block there are again two functions, the phase
locked loop (PLL) and the actual firing pulse generator (FPG). The AC bus voltages serve as the
reference signals for the three-phase PLL, and the FPG for the VGB with the YD transformer is
has a lag of 30°.

The entire control diagram is given in Appendix F inverter pole 1 pole 2. Note that there are two
complete sets of rectifier and inverter controls, each for one pole.
Before moving further, the full control properties is briefly explained using graphs.

5.2 Full control properties

Under steady-state conditions with rated AC voltages at both the rectifier and inverter
commutating buses the inverter firing angle is obtained using the PI regulator operating on the
error between the extinction angle set-point and the measured extinction angle. In this case the
HVDC system comes to a stable operating point defined by the intersection of the rectifier and
inverter characteristics as shown in figure 5.2.1a. The rectifier characteristic’s segment labeled
R1-R2 is due to the minimum firing angle. The vertical segment R2-R3 is due to the rectifier’s
constant current controller. The inverter segment I1-I2 represents constant extinction angle
control.

64
Figure 5.2.1 a. Rectifier characteristics b. Inverter characteristics

The rectifier R1-R2 characteristic portion is highly dependent upon the AC voltage available at
the rectifier commutating bus. If this voltage decreases so that the segment R1-R2 falls below
that of I1-I2 no stable operating point would exist and the dc current would fall to 0. In order to
avoid such a situation the inverter controls are equipped with both a constant extinction angle
controller and current controller. Under normal conditions the current controller’s output is not
used. However, if the dc current error is greater than the pre-determined amount (0.1 pu for
example) the inverter’s current controller is used rather than the extinction angle controller’s
output. Figure 5.2.1b shows the structure of the inverter extinction angle and current controllers.

Under normal operating conditions the current error signal (Inverr2 in figure 5.1.13) will be zero.
The input to the inverter’s current controller is offset by a constant value (CMARG 0.1 pu). The
PI current controller for the inverter thus sees a constant value of 0.1 as its input and the
integrator output ramps up to the maximum value. The minimum selection gate will then choose
the firing angle determined by the inverter’s gamma controller. So long as the measured DC
current(CINV2 in figure 5.1.13) is not less than 90% of the set-point current, the inverter’s
current controller output is not used. In the case where the DC current falls below 90% of the set-
point, the inverter’s current controller output will fall until its stable operating point is reached.
Such a new operating point is not optimum since the reactive power consumed by the inverter
will be greater than when operated in extinction angle control. Figure 5.2.2 shows the rectifier-
inverter operating characteristic with both the inverter’s current and gamma controllers included.
The new inverter segment labeled I1-I3 is due to the inverter current controller, figure 5.2.2a.
Under rectifier low AC voltage conditions the operating point will be set by the inverter current
controller as shown in the right hand side graph of figure 5.2.2b.

65
Figure 5.2.2 a. Current controller added b. Current controller active when AC voltage is low

If the AC voltage at the rectifier drops there is a potential situation where two valid operating
points may exist. In this case the operation of the HVDC system may oscillate between these two
points. Figure 5.2.3 shows the rectifier-inverter characteristic when the segments R1−R2 and
I1−I2 are near the same values.

Figure 5.2.3 Two valid operating points: oscillation risk

A modification to the inverter controls is used to change the operating characteristic of the
inverter so that there is a positive slope segment added between the constant current control and
constant extinction angle control. The positive slope segment is created by subtracting a signal
proportional to current error when creating the gamma controller error signal. The gain and
limits for this control section are chosen so that the controller is active when the current error is
in the range : 0,0 < CMARG < 0,1. Another approach that is often used to solve the two
operating point problem is to include the voltage control loop at the inverter. The voltage
controller adds a horizontal line segment to the inverter characteristic as shown in figure 5.2.4.

66
Figure 5.2.4 Alternate voltage control loop to eliminate oscillation risk

The VDCOL function was not shown in the preceding rectifier-inverter characteristics. This
function is applied to both the rectifier and inverter. The full rectifier-inverter characteristics are
shown in figure 5.2.5.

Figure 5.2.5 Full Rectifier-Inverter characteristic Idc vs Vdc

Tap-changers
Finally the tap-changers play an important role to reach a desired steady state. When the dc
voltage is not controllable by γ, the tap-changer of the converter transformers are used to reach
the desired voltage range and γ is then used to fine-tune to achieve the steady state. Also to have
adequate control over de voltage while keeping the firing delay angles at a value where the
reactive power drawn by the converters is acceptable, the tap-changers of the converter
transformers are used. Table 5.1 lists the data that was available on the tap-changers. To control
the tap-changers, a value corresponding to a percentage of the rated voltage of the transformer is
assigned to a specific variable of the valve group block. A tap position of +1 means that the rated
voltage of the secondary winding is 1.0125 times the value entered in Draft for the secondary
base voltage.

67
 Eemshaven Feda
Number of steps tap-changer +10 / -13 +11 / -9
Step size (% of Unom) 1.25 1.25
Table 5.1 Number of steps transformer tap-changer

To control the tap position during simulation, sliders were used. After having converted the real
values outputted by the sliders to integers it served as an input to a small control circuit. This
control circuit converted the desired tap position into a value relative to the transformer's rated
voltage so it could be assigned to the variable of the valve group block. Figure 5.2.6 shows this
circuit. Two sliders are included. Each slider controls both converter transformers at rectifier or
inverter.

Figure 5.2.6 Circuit for controlling the tap-changers

While modeling the circuit, it proved the sliders to adjust the ABC magnitude of the “sources” at
both sides of the HVDC link was important to reach a stable point of operation. So these sliders
were included in the runtime module. With the rated voltage at Eemshaven at 380kV and rated
voltage at Feda at 300kV it was difficult(not possible) to reach the rated operation conditions for
NorNed. Therefore the ABC magnitudes were altered and then other variables were tuned for a
desired steady state. Other control functions used in this study are explained in the next
paragraph.

5.3 Disturbances controls

Interaction with the network can be done from the runtime page/module. With switches, push
buttons and sliders the parameters of the circuit can be changed and modified, and if wished the
effects of these modifications can be observed graphically.
But first these switches, push buttons and sliders have to be modeled in the RSCAD module in
the network with specific functions.

In the three phase AC system different types of faults can be created. Such as single phase-to-
ground faults to phase-to-phase faults, in short all types of faults. In the network model, on three
locations AC faults were modeled. In RSCAD this is done by a fault branch. Known as “fault

68
point 1”, “fault point 2” and “fault point 3”, illustrated in figure 5.3.1. The exact location of the
fault points is discussed in chapter 6. The fault branch consists of a switch whose open resistance
is very large and whose closed resistance is specified by the user. In the menu of the fault branch
the value of the resistance of the single phase to ground can be changed by means of a
preprocessor variable slider, which was set to vary between 1µΩ to 10Ω. Furthermore each fault
type in the fault branch menu must be assigned a specified bit number which correspondents to
an integer from the logic. In this way from the logic(explained below) a specific fault type can be
triggered.

Figure 5.3.1 Fault branch at different locations

A fault can be triggered from the runtime page which correspondents to the specified type of
fault at a arbitrary moment, but during this study it was desirable to have the fault triggered at a
specific time or at a certain location on the sinusoidal voltage wave. This is realized by the logic
function of figure 5.3.2. This is the first part of the entire logic.
This circuit is used to control the point on wave(POW) upon which the fault is applied, or the
fault inception point. Node voltage N10 is used as a reference point for the point on wave delay.
An If-Then-Else logic gate with a positive edge detector determines when N10 voltage has
crossed the X-axis and is positive going (zero-crossing detector). The fault button when pressed
produces a 22ms pulse which is slightly longer than one cycle at 50 Hz base. A pulse is then
produced by the AND gate that combines the zero-crossing detector and fault button. The pulse
drives the point on wave logic, which is comprised of a slider, a gain block, and a pulse duration
timer set to detect a rising edge. When the pulse rises to logic one, the output of the duration
timer is set to logic one equaling the time it takes to rotate the number of degrees from the zero
crossing detection, set by the POW slider control. The pulse from the first duration timer is used
to drive the fault duration logic, which is comprised of a slider, a pulse duration timer set to
detect a falling edge. When the pulse falls to logic zero, the output of the duration timer is set to
logic one for the specified time which is determined again by a slider.

Figure 5.3.2 Fault control part I, Inception point

69
The second part of the fault control logic circuit is used to control the fault type and location.
Fault switches for the phase-to-phase and phase-to-ground fault types are combined to create the
necessary integer value. This value is multiplied by the pulse from the first part of the logic,
thereby creating a pulse width with an integer value that can control the fault branches. A dial
component is used to control selector control switches that will determine where the fault will be
applied: fault point. For example, the dial component is set to 1, and the B phase-to-ground fault
type is selected, the corresponding fault signal “FLTSIGI” will have an integer value of “2’. All
other fault signals will have an integer value of “0”. In this way the user can determine where
and when a fault is made in the network. The second part of the fault control is shown in figure
5.3.2.

Figure 5.3.2 Fault control part II, fault location and fault type

Also DC faults can be made in the network. This includes DC cable to ground faults and faults in
the valve group block. The faults in the VGB can be internal or external i.e. within the phases in
the block, or from the phases in the block to other point outside the block. These types need to be
selected first in the VGB and then be triggered from push buttons in runtime. The DC line to
ground faults are triggered by a push button and the duration of the fault can be varied. One more
additional fault type added is the voltage drop at the source. The duration and percentage of the
voltage drop can be selected and the effect can be monitored. The push buttons, sliders and
switches from the runtime page are shown in figure 5.3.3.

70
Figure 5.3.3 Control buttons, sliders and switches in Runtime

For a particular simulation case it was necessary to include a circuit breaker in the model. This
was easily used from the components library. But the logic to trip the circuit breaker was done
differently than in practice. In practice a fault is detected by a relay, which signals the circuit
breaker to trip and the network(part) is isolated by the breaker action. The relay/protection model
can be incorporated in the model from the RSCAD library.
Here a different way was used to trip the CB in that sense that there was no relay used. Instead a
small control logic was used. This control uses the signal to trigger the fault(given by the user),
but with a delay sent to the circuit breaker. This time delay can be specified. Also the time which
the breaker remains open can be specified. The option if the user wishes to open the breaker
manually, is possible by changing a switch from fault control to manual control. The logic
function is given in figure 5.3.4.

Figure 5.3.4 Control for the circuit breaker

Full control scheme and remarks

The full control scheme is presented in Appendix F. Regarding this scheme, some important
remarks have to be made.
• The scheme is based on general control principles applicable to two-terminal HVDC
transmission systems rather than that is based on manufacturer specific data used in practice.

71
• In practice, the controls are more elaborate then presented here. This due to the fact that,
amongst other things practical schemes comprise automated controls for the capacitor banks and
tapchangers. Something which in the model is done manually.

72
6 Simulation

In this part simulations with the model are carried out to investigate problems and results are
presented using graphs. The effect of certain action and or failures in the network is explored.
The main work until now was modeling the HVDC network. Of course the DC network can be
elaborated with addition of the AC part. With the addition of the AC network, combination of
faults are studied here. Desired situations can be studied and based on the responses statements
can be made. This is divided in two parts. In part I various basic simulations are carried out with
the AC and DC network to prove that the model is functioning as it should be at its rated
properties. In part II a case is studied which occurred in the past, here attention still was
necessary to investigate the cause of the occurrence and possibly enhance the network to prevent
the event.

The AC network is not extended at the Norwegian side (FEDA). Feda is represented as a bus
with a certain SCC. The AC network there was not important for this study. The AC network at
EEMS is extended from the DC converter till the 380 kV substation, which is further explained
in part II of this chapter.

In figure 6.1.0a the Valve Group Blocks in RTDS are shown, together with its controls in figure
b. These controls are very important for the HVDC operation. The elaborated controls are given
in Appendix F.

Figure 6.1.0 a. Valve Group Blocks in RTDS b. HVDC controls

73
6.1 Part I

6.1.1 Faults types

First the steady state situation must be reached from where on faults can be triggered. The steady
state properties of the NorNed HVDC link is given in Appendix G. Using the controls these
points are reached. Then the following faults are triggered. These faults are:

• AC Faults
§ Single phase to ground faults
§ Phase to phase
§ Three phase
§ Sudden voltage drop at rectifier/inverter
• DC faults
§ Positive polarity cable to ground
§ Negative polarity cable to ground
• Converter faults
§ Combination Several type faults in the Valve Group Blocks

6.1.2 Steady state

Steady state Rectifier mode
In steady state the ratings of the HVDC link is given in table 6.1.1. The following plots show the
steady state operating point.

Figure 6.1.1 DC power EEMS

and FEDA (rectifier and
inverter resp.)

Figure 6.1.2 AC power EEMS

and FEDA

74
 Figure 6.1.3 AC voltages EEMS
and FEDA

Figure 6.1.4 DC voltage at

EEMS and FEDA both at
positive and negative polarity

Figure 6.1.5 DC current in

valve group blocks EEMS and
FEDA

Figure 6.1.6 DC cable currents at positive(upper plot) and

negative polarity(lower plot)

75
Figure 6.1.7 Currents in capacitor banks, high pass filters and double tuned filters at EEMS

Figure 6.1.8 Control buttons and ABC voltage magnitudes EEMS and FEDA

Steady state Inverter mode

Figure 6.1.9 DC power EEMS

and FEDA (inverter and
rectifier resp.)

Figure 6.1.10 AC power EEMS

and FEDA

76
 Figure 6.1.2 AC voltages EEMS
and FEDA

Figure 6.1.11 DC voltages

EEMS and FEDA at positive
and negative polarity

Figure 6.1.12 DC currents in

valve group blocks at EEMS
and FEDA

Figure 6.1.13 DC cable currents at positive (upper plot) and

negative polarity (lower plot)

77
Figure 6.1.14 Currents in capacitor banks, high pass filter and double tuned filters at EEMS

Figure 6.1.15 Control buttons and ABC voltage magnitudes EEMS and FEDA

78
6.1.3 Results Faults

The faults mentioned earlier were triggered using the Runtime module. These fault types were
initiated from steady state. Once a fault occurred, the responses were recorded and presented
with graphs. The system kept on running on the background and returned to steady state.
Because of the many types of faults carried out, these are not given here. The system reacts well
and returns to the steady state after a fault is removed.
One certain type of fault is thoroughly investigated in the next paragraph.

79
6.2 Part II

6.2.1 Case AM
The network is considered ready now to do some other simulations and investigate practical
problems. Real cases and events can be studied now. One such an event occurred in the past
(April 11th 2009)[25] was present at the department Asset Management(AM). This case needed
attention in order to have better understanding concerning the effects of this case if reoccurred in
the future. This case was placed at department Transport & Infra (TI) for examination.

Case

Between the AC substation and DC converter station at Eemshaven is the Filterhal where the
harmonic filters and the capacitor banks are installed. At the end of the 1500 meter cable from
the 380AC substation EEMS a single line-to-ground fault occurred. Power systems are designed
to isolate the fault location or the faulted equipment as much accurately is possible with the
actions of circuit breakers. This is also the case with the mentioned line-to-ground fault at the
cable connection. After the single line-to-ground fault occurred, it was detected and the Gas
Insulated Switchgear opened and the AC substation was disconnected from the HVDC
transmission. The normal response of the fault current after disconnection is extinguishment,
because it is not anymore fed from “sources”. The term sources include the inertia of generators
still connected to the fault from either side.

The problem in this certain case is, that after switching the faulted part of the network, there was
a damage (caused by an explosion and fire) probably AFTER the disconnection. This event
should not have occurred. During the fault while the connection from phase to ground is
sustained, the fault current will be large, whereas the voltage is almost zero. After detection and
disconnection, there must have been some other reasons to cause the explosion. Oscillogram
records from this case of the actual fault for some of the currents and the voltage of the fault
point were available. Figure 6.2.2a shows the phase voltages of the three phases at the fault
location (filter building). In figure 6.2.2b the faulted phase is zoomed. Furthermore the available
current graphs at both ends of the 1500 meter cable are given in 6.2.3a and 6.2.3b. For the ease
of reading, the graphs will be mentioned as graph A, B, C and D further in the text. Another
current measurement was made at the far end of the 1500 meter cable, at the same location as
figure 6.2.3b. All these pictures are included in Appendix H in larger format.

The situation from the 380kV AC substation with the two pieces underground cables, filter
building and thyristor hall is shown in figure 6.1.1, this all together with converter transformers
and converters in Eemshaven is shown in figure 6.1.4, a view in RTDS.

80
Figure 6.2.1 Situation at Eemshaven-EDC, fault occurred at the end of the long cable

Figure 6.2.2 a. Three phase voltages at fault location-A b. Faulted phase voltage-B

The information extracted from the oscillogram above and below(voltage and current), is that at
a certain time, t = 0.24 seconds a short circuit occurs from phase B to ground. This fault is
sustained till the end of the recording. The fault is in the filter building, at the end of the long
cable. The voltage of phase B swings around zero while the current in the cable(fault current)
increases. At t = 0.29 seconds the detected fault is disconnected at the beginning of the long
cable by the action of the circuit breaker(GIS), this is approximately 50 milliseconds after the
ground fault occurs. After disconnection the voltages in the other phases have swings with higher
frequencies, figure 6.2.2a. The currents in the other phases also swing larger but not as large as
in phase B, see figure 6.2.3.

81
Figure 6.2.3 a. Current at the near end long cable-C b. Current at the far end long cable-D

Figure 6.2.4 Situation at Eemshaven-EDC in RTDS

Figure 6.2.5 Capacitor banks and Filters in Eemshaven

82
6.2.2 Results studied case
Starting point for this case study was a steady state situation. The chosen point was a DC power
transmission of 600MW at rated voltages +450/-450 kV. Snapshots of the responses at steady
state are already given in part I of this chapter. Because it was unknown in which mode the
substation EDC was operated when the fault occurred, both modes of operation were
investigated.

In the following text the concept of “Rectifier mode” and “Inverter mode” is used to indicate in
which mode the substation EDC in Eemshaven is functioning. In rectifier mode power is
transmitted from Netherlands to Norway through the HVDC link. In inverter mode from Norway
to Netherlands.

For determination of the mode of operation at the moment of fault occurrence the voltage and
current signals from the oscillogram at the fault were taken as reference. These were compared
with the voltage and current signals made in the simulation. The fault is made at the moment the
voltage is negative in the filter building, as seen in figure 6.2.2a. The controls were used for this,
the slider was set to 240º after zero crossing. This is seen in figures 6.2.6 in both rectifier and
inverter mode. The voltages just after the fault are more or less the same as the voltages from
plot A, but the emphasis is on the moment the fault occurs. For the current signals plot C(at near
end of long cable, figure 6.2.3a) was chosen as the reference. In figure 6.2.7 these currents of
both modes are given.

Figure 6.2.6a. Voltages Rectifier mode b. Voltages Inverter mode

Figure 6.2.7a. Currents Rectifier mode b. Currents Inverter mode

83
In plot C the current at the near end of long cable is positive at the moment of fault occurrence.
This is the case when the EDC is in inverter mode, that is seen in figure 6.2.7b. So the
simulations were carried out using inverter mode, but the rectifier mode was explored too, this
was done to know what the effect is of the entire system in both modes.

Once the determined mode of operation is determined, the goal is to replicate the plots A through
D. After this, a possible solution can be worked out. This can be incorporated in the network to
prevent the occurrence of fire/explosion after disconnection of the faulted part.

Using the Runtime in RTDS several signals were observed. Not all of them are included in the
results here. These can be found in Appendix I. Some of them however are shortly mentioned
and given here. Numerous simulations were carried out in the attempt to replicate the signals of
plots A, B, C & D.

Figure 6.2.8 a. Simulated plot A Rectifier mode b. Simulated plot A Inverter mode

The voltages in filter building achieved during simulations are as in figure 6.2.8. The first
difference between the actual recording and the simulation is that the magnitude of the voltage
swings of the unfaulted phases are not as large as in plot A after the CB has tripped.
Furthermore a distinct difference is the made in the mode of operation. In inverter mode, when
the main circuit breaker has disconnected the fault after 50 ms, after a period of 30 ms the swings
increase. This occurs possibly because the rectifier(FEDA) keeps operating, and the
inverter(EDC) is also active.
In practice whenever a fault occurs, the HVDC controls react fast on this by stopping the firing
of the valves even before the fault is disconnected by the main CB. This control function is
lacking in the present controls of the HVDC model.

The solution to eliminate the swings of phases A and B is to trip the capacitor banks together
with the main circuit breaker. With this action the phenomena of L-C swings between the
converter transformer and the capacitor banks is eliminated.

84
With the addition of circuit breakers to the capacitor banks and switching them together off at the
moment the main CB opens(50 ms after fault occurs), the results are as in figure 6.2.9

Figure 6.2.9 a. Cap. banks off- Rectifier mode b. Cap. Banks off- Inverter mode

It is clear that the swings are reduced. In both modes the swings after disconnecting the capacitor
banks are smaller than the 100 kV. This is a reduction of approximately 25%.

When the capacitor banks are tripped, the discharges of the charges over the ground fault are
stopped. This is seen in figures 6.2.12 & 6.2.13 in rectifier mode and in figures 6.2.18 & 6.2.19
in inverter mode.

The exact signals are not obtained with the simulations. This is a drawback. Other variables
altered in the course to get the replica of plots A and B are:

• Variation of the ground fault resistance

• Variation of Zero sequence reactance of Y-D transformer
• Saturation of converter transformer

In figure 6.2.22 and 6.2.23 the actual recording is compared with the simulation of the phase
voltage of the faulted phase. In 6.2.22 before and after the fault is cleared spikes appear in the
voltage signal. This is not found in the simulation 6.2.23. For the simulation a ground fault
resistance(Ron, as explained in chapter 5.3) of 0.01Ω was chosen. It was an fault from an
underground cable to ground fault, so the resistance was quite low. This Ron was varied from
1•10-6Ω to 10Ω. It is found when Ron is very low, the phase voltage stays constant approximately
zero. When Ron is high there is an sinusoidal response of the phase B voltage with a higher
amplitude.
When the fault occurred, the ground resistance varied in practice, which`s value is difficult to
know exactly, this is seen as the spikes in figure 6.2.2b. The model of the ground
fault(resistance) is quite simply modeled, getting the exact response as in6.2.2b was not
possible.

85
Another step undertaken to reach get the signals as in plot A is the variation of the zero sequence
reactance of the converter transformer. In the YY transformer it was not possible to change this
parameter. The YD transformer however had the option. The best results were found with 0.01
pu. The YY transformer model did not have the option to change/alter or even enter the zero
sequence reactance.

The model in the first step was modified here from one single phase three-winding transformers
to two three phase two-winding transformer.

Furthermore, the option to activate saturation of the converter transformer is not enabled in the
present model of the HVDC Valve Group Block.

All of these inadequacies of the present model and controls resulted in not replicating the exact
signals during simulation. But the results achieved above do have strong resemblance of the
responses.

86
7. Conclusions & Recommendations
Following the work done here during theoretical study, the modeling and the simulations,
several conclusions can be made. According to the hurdles and experiences some
recommendations are made.

7.1 Conclusions

• The HVDC model was created for the NorNed link;

This was done as much as possible through (control)standards in available software to carry out
dynamic simulations. The HVDC basics, procedure followed for modeling network and its
control are reported.

• Various simulations were carried out with this model, including AC 1-2-3 phase and
Ground faults, DC cable faults and converter faults;
From the results achieved during simulations, the model can be used for studies for more insight
in the NorNed HVDC link, because the response from basic fault simulations is as expected.

• The case which was to be investigated is discussed below;

The aim during the simulations was to replicate the signals from the recorded fault. To some
extend it is replicated. From the responses using this model it is clear what the effect is of the
events occurred in the past and solutions can be proposed using these results.

• The exact signals which were recorded at the moment of fault are not obtained with the
built model, this was because of some modifications while modeling the actual
equipment and some limitations in the simulation tool;
The parameters of these models were thoroughly altered during simulations to achieve the
response which were available from the fault recording oscillogram
o transformer configuration ; The change was made from one single phase three-winding
transformers to two three phase two-winding transformer model, the attempt to use a single
phase three-winding transformer to investigate the effect of it remained unsuccessful.
o pi-section cable models; With a simulation time step of 50 micro-sec no cables/lines can be
modeled using travelling wave model, the pi section was therefore used, which is limited in
its accuracy
o model which represents the ground fault; The actual fault was a single phase to ground fault
in the cable from which the oscillogram was present, this showed a response that is
unsteady/irregular. Various values for the ground fault resistance were simulated, this
resulted in changes in the phase to ground fault, but also in the other phases, which indicates
the right representation of the ground fault could have showed the desired response.

87
 • The necessary modification according to the responses is formulated;
Possible solutions were incorporated in the model and it is found from the responses of the
simulation a way of preventing what happened in the past, is to disconnect the capacitor banks
together with the main GIS which disconnects the HVDC link. With this action the swings of the
unfaulted phases is decreased.

88
7.2 Recommendations

Based on the hurdles encountered and results found the following recommendation is
formulated:

The results found here for the studied case are not convincing enough for adjustments in the grid,
the exact replica of recorded oscillogram were not obtained. Nevertheless, an important step was
set for the investigation of HVDC. TenneT is encouraged to investigate the problem of the fault
and if possible enhance the network to prevent such events, which results in long interruption of
the NorNed. This can be done with RTDS and/or other simulation tools. RTDS is limited in its
modeling which obstruct the exact modeling of actual equipment. Therefore the exact responses
were not found during the simulations. So using updated models in RSCAD is strongly
recommended. RTDS has the strong feature to perform real-time simulation which TenneT in
cooperation with the TU Delft can use in the future, where HVDC will play an important role for
bulk transmission. Fine-tuning is need for the present controls and model. In order to verify the
models and the case studied here, another simulation tool can also be used. This can be done
with DIgSILENT PowerFactory which is used at TenneT.

89
References

1. Role of HVDC and FACTS in future Power Systems, W. Breuer, D. Povh, D. Retzmann,
E. Teltsch 2004

2. UHV DC 800 kV Bulk Transmission, IEEE T&D Latin America –São Paulo,Brazil -
November 10th, 2010

3. High Voltage Direct Current Transmission- Proven technology for power exchange,
Siemens

4. High Voltage Direct Current (HVDC)Transmission Systems Technology Review Paper,

ABB

5. Dielectric stresses on 800 kV HVDC Converter Transformers and Thyristor Valves,

Mark Louwerse 2008

6. High Voltage Direct Current Transmission, 2nd edition, Jos Arrillaga

7. HVDC Transmission, Dennis A. Woodford, Manitoba HVDC Research Centre Canada,

18 march 1998

8. Novel Voltage Source Converter based HVDC Transmission System for Offshore Wind
Farms, Stephan Meier, Stockholm 2005

9. VSC-Based HVDC Power Transmission Systems: An Overview, Nikolas Flourentzou,

Vassilios G. Agelidis, and Georgios D. Demetriades, march 2009

10. Recent classic HVDC development, Lennart Carlsson and Gunnar Flisberg, ABB
Utilities

11. Recent and future trends in HVDC converter station design, Lennart Carlsson, Gunnar
Asplund, Hans Bjorlilund, Henrik Stomherg ABB

12. NorNed project Technische beschrijving, TenneT 2004

13. Trends for future HVDC applications, W. Breuer, D. Povh, D. Retzmann, E. Teltsch,
Siemens 2006

14. Future of hvdc power grid in Europe, Vojtech Straka, TU Delft

15. Modern HVDC: State of the art and development trends, Víctor F. Lescale, ABB

90
16. The Cobra Cable - A feasibility study regarding an HVDC submarine cable between the
Netherlands and Denmark, TenneT & Energinet.dk

17. http://www.statnett.no/en/

18. 8th AC/DC Conference, London IEEE, March 2006

19. Implementation of Noord-Holland Grid in RTDS, A. G. Ejigu 2009

20. Real Tim Digital Simulator power system users manual, February 2005

21. Transients in power systems, Lou van der Sluis 2001

22. Commutation Failure Analysis in HVDC Systems Using Advanced Multiple-Run

Methods, E. Rahimi, S. Filizadeh, A. M. Gole, IEEE

23. Power System Stability and Control, P. Kundur

24. The Cobra Cable -A feasibility study regarding an HVDC submarine cable between the
Netherlands and Denmark, TenneT-Energinet 2009

25. TenneT

26. Power electronics: Converters, Applications and Design, Mohan, Undeland and Robbins

27. HVDC Projects Listing Prepared for the DC and Flexible AC Transmission
Subcommittee of the IEEE Transmission and Distribution Committee, Working Group on
HVDC and FACTS Bibliography and Records, July 2009

91
APPENDICES
Appendix A. Comparison example AC and DC cable transmission

High voltage transmission by cable is rarely used because of the higher cost and longer repair
times; it is normally restricted to underwater crossings and infeed to urban centers. The high
voltage cables have a low series inductance and a large shunt capacitance. Moreover their
loading, owing to the lower surge impedance and thermal limitations, is usually below 0.3 times
the surge-impedance level. Therefore, high charging reactive powers are required, which
considerably limit the length of AC cable transmission. For instance, at 50 Hz the charging
current varies typically from 5.5 A/km for a 132 kV cable to about 15 A/km for a 380 kV cable.
With a 4.52 cm2, 380 kV cable of 600 A thermal limit, the charging current for a 40 km length
equals the thermal limit and no useful load can therefore be carried. Similarly, a 2.58 cm2, 450 A,
132 kV cable has a critical length of about 80 km.

These critical lengths may be extended by inserting shunt reactors. Even with 100%
compensation by means of two reactors, one at each end, the power transmission capacity is only
86.6% at critical length and reduces to zero at twice the critical length. With two intermediate
reactors, each providing 100% compensation, deciding the line in three equal parts, the critical
length will increase to three times. Moreover intermediate compensation is impractical in the
case of underwater links.[6]

92
Appendix B. Rectifier and Inverter operation signals

Typical six-pulse rectifier operation signals

a. Positive and negative voltyages with respect to transformer nuetral
b. direct bridge voltage Vd and voltage across valve 1
c, d. Valve currents i1 to i6
e. AC line current of phase R

93
Typical six-pulse inverter operation signals
a. Positive and negative voltyages with respect to transformer nuetral
b. direct bridge voltage Vd and voltage across valve 1
c, d. Valve currents i1 to i6
e. AC line current of phase R

94
Appendix C. HVDC projects
HVDC Projects Listing [27]

95
96
97
98
99
Appendix D. Data for network model in chapter 4
Empty

100
Empty

101
Empty

102
Empty

103
Appendix E. Network in RTDS ( RSCAD )

104
Filters and capacitor banks at Netherlands (EDC) and Norway (FEDA)

105
Appendix F. Full HVDC rectifier and inverter controls
Master controls

106
Rectifier controls

107
Inverter controls

108
Appendix G. Steady state properties NorNed

Main circuit parameters for the NorNed HVDC Cable transmission

109
Appendix H. Oscillogram recordings of Case

Voltages
filter
building

Voltage
phase B
filter
building

110
 Currents
short end
long cable

Currents
long end
long cable

111
 Currents
short end
long cable
(meting II)

Currents
long end
minus
short end

112
Appendix I. Case simulated in RTDS

Figure 6.2.10 Currents long cable, short cable, phase voltages Rectifier mode

113
Figure 6.2.11 Capacitor banks disconnected Rectifier mode

114
Figure 6.2.12 Currents Capacitor banks, High pass filters, Double tuned filters Rectifier mode

Figure 6.2.13 Capacitor banks disconnected Rectifier mode

Figure 6.2.14 Fault current phase B to ground Rectifier mode

Figure 6.2.15 Capacitor banks disconnected Rectifier mode

115
Figure 6.2.16 Currents long cable, short cable, phase voltages Inverter mode

Figure 6.2.17 Capacitor banks disconnected Inverter mode

116
Figure 6.2.18 Currents Capacitor banks, High pass filters, Double tuned filters Inverter mode

Figure 6.2.19 Capacitor banks disconnected Inverter mode

Figure 6.2.20 Fault current phase B to ground Inverter mode

Figure 6.2.21 Capacitor banks disconnected Inverter mode

117
Figure 6.2.22 Faulted phase voltage- plot B

b. Ron= 1•10-6Ω

Figure 6.2.23a. Simulated faulted phase voltage

c. Ron= 10Ω

118
119
120