Power Quality Improvements of Wind Farms: P Consumers

Power Quality Improvements
of Wind Farms
P
Consumers
Q
GTO-ASVC
of
Wind Farms
of
Wind Farms
Fredericia, June 1998

4 Kapitel 1
Kapitel 1 5
Towards
Sustainable Development
for the Next Generation
6 Power Quality Improvements of Wind Farms
ISBN No.:
87-90707-05-2
Type:
Formata
Excelsior
Layout and text:

Eltra . Information
Repro and printing:

Sndergaard Bogtryk og Offset
Binding:
J.P. Mller, Haderslev
Paper:
115 g mat offset
130 g silkmat coated offset
Impression:
200 copies
7
Preface
The project described in this book was carried out successfully, thanks mainly
to exemplary co-operation in the project team comprising participants from a
manufacturer of electrical equipment, the utilities and two universities:
Siemens, Power Transmission and Distribution, Erlangen

UMIST, Department of Electrical Engineering and Electronics, Manchester
National Wind Power, UK
The Technical University of Denmark
Snderjyllands Hjspndingsvrk (SH), Denmark
Eltra, Denmark
The project was supported financially by the European Commission under the
Fourth Framework, the Joule Non-Nuclear R&D Programme.
Siemens, Germany, was represented by Mr Ralf Stber as project manager,

Dr Michael Weinholdt, Mr Frank Schettler, Mr Heinz Tyll, Mr Klaus Bergmann
and Mr Reinhard Brieden. Besides, Mr Peter Jensen and later Mr Bjarne
Reeuwijk participated as the local project co-ordinators for Siemens in
Denmark.
Dr Nick Jenkins from the Manchester Centre for Electrical Energy contributed
to the system studies and Mr Magnus Davidson from National Wind Power
participated in the project as subcontractor.
The Technical University of Denmark was represented by three assistant pro-

fessors; Mr Kaas Pedersen, Mr Erik Andersen and Dr Knud Ole Helgesen, all
from the Department of Electric Power Engineering.
The participants from Eltra (formerly Elsam) were Mr J.J. Rype, who conducted
the civil works, and Mr Tonny Rasmussen, who was involved as part of his
Ph.D. study at the beginning of the project.
Finally, Mr Jacob Rath from SH and Mr Frede Nielsen from the local distribu-
tion undertaking of EASV contributed to the project.
Many people have been involved in the project. However, my task as project co-
ordinator was not a difficult one, thanks to the dedication to and involvement
of all project partners. Great efforts have been made to cope with the time
schedule whenever there were unforeseen delays because of uncertainties,
which cannot be avoided in a complicated development project like this.
The contract with the European Commission came into force on January 1,
1996, which was the official starting date of the project. However, intensive
negotiations with Siemens and the basic system studies were carried out well
before the EC contract came into force.
The first task was to determine the system design. Our first idea was to have
two 4 Mvar ASVCs, one connected to each of the two busbars in the switch-
gear. Preliminary studies indicated that with this solution the harmonic distor-
tion was acceptable.
However, detailed studies showed that under worst case conditions the distor-
tion was too high when all wind turbines were running on one busbar and only
one wind turbine was running on the other busbar. The harmonic studies
showed that shunt capacitors for no-load compensation in the wind turbines
function like a filter, which will absorb all harmonics and may cause overload
of the shunt capacitors.
The solution was to connect the two ASVCs in a 12-pulse configuration via a
three-winding transformer and to increase the series inductance between the
ASVC and the network. Also the ASVC pulse pattern was changed to reduce
the harmonic distortion.
Since the grass root movements put the environmental issue on the political
agenda, adoption of drastic measures will be required by the power industry.
Today, environmental issues and the Danish governments strategy for reduc-
tion of the emissions of carbon dioxide will eventually introduce large offshore
wind farms into the electrical power system.
To be ready for a deregulated marked, the former power pool of Elsam was
divided into two separate companies on January 1, 1998; an independent pro-
duction company, which will keep the name of Elsam, and a new transmission
company with the name of Eltra. Apart from being a transmission company
Eltra will bear the overall responsibility for the electrical system. Third-party
access to the transmission system, trade in electricity and public resistance to
new overhead lines are some of the challenges Eltra will be confronted with.
Therefore, from a Danish point of view this project may be seen as part of a
strategy for meeting the future challenges and for integrating a large number
of offshore wind farms into the Danish electrical power system.
Power Quality Improvements of Wind Farms 9
I wish to thank the European Commission for their financial support for the
project, and Mr K. Diamantaras, DG XII, for his help and guidance. I also give
my warmest thanks to all the people who have contributed to the project and
to the creation of this book. I feel that close relationships were established and
strong friendships were developed between the partners across the borders of
the European Union. Last, but not least, I also want to thank Mr Valdemar
Legaard from the graphical department of Eltra, who has made a very great
effort putting all the material together and creating the final layout of this
book.
Kent H. Sbrink
Project co-ordinator
Fredericia, June 1998
Kent H. Sbrink Jrgen Kaas Pedersen

I/S Eltra Technical University of Denmark
Fjordvejen 1-11 Department of Electrical Power
P.O. Box 140 Engineering
DK-7000, Fredericia DK-2800 Lyngby
Denmark
Knud Ole Helgesen Pedersen

Frank Schettler Technical University of Denmark
SIEMENS, Power Transmission Department of Electrical Power
and Distribution Engineering
Erlangen DK-2800 Lyngby
Germany
Zouhir Saad-Saoud,
Klaus Bergmann Maria Luiza Lisboa,
SIEMENS, Power Transmission Goran Strbac
and Distribution Department of Electrical
Erlangen Engineering and Electronics
Germany UMIST
PO Box 88
Manchester
Ralf Stber M60 1QD
SIEMENS, Power Transmission ENGLAND
and Distribution
Erlangen
Germany
Nicholas Jenkins
Department of Electrical
Engineering and Electronics
UMIST
PO Box 88
Manchester
M60 1QD
ENGLAND
Janaka Ekanayake
UMIST Visitor
Department of Electrical and
Electronic Engineering
Faculty of Engineering
University of Peradeniya
Peradeniya
Sri Lanka
Table of Contents
Chapter 1: Introduction 15
1.1 Introduction to the Book 17

1.2 Introduction to the Project 18
1.3 Introduction to the Project Partners 18
1.3.1 ELSAM - ELTRA 18
1.3.2 SIEMENS 19
1.3.3 Technical University of Denmark (DTU) 19
1.3.4 University of Manchester Institute of
Science and Technology (UMIST) 19
Chapter 2: Rejsby Hede Wind Farm 21
2.1 Introduction 23
2.2 The Rejsby Hede Wind Farm 25
2.2.1 Description of the Wind Farm 25
2.2.2 Description of the Turbines 28
2.2.3 Description of the Local Network 28
2.2.4 Need for Reactive Compensation 30
Chapter 3: Reactive Compensation for Wind Farms 35
3.1 Introduction to Reactive Power Compensation 37

3.2 Impact of Reactive Power Flows on a Simple Circuit 38
3.3 Conventional Methods of Shunt Reactive Power Compensation 39
3.3.1 Shunt Capacitors and Reactors 39
3.3.2 Synchronous Condenser 40
3.3.3 Thyristor Controlled Reactor (TCR) 40
3.3.4 Thyristor Switched Capacitor (TSC) 41
3.4 Advanced Static VAr Compensator (ASVC) 43

3.4.1 Introduction 43
3.4.2 Six-ulse Voltage Source Inverter Operating
with Fundamental Frequency Modulation 44
3.4.3 Principle of Operation of a Six-pulse VSI based ASVC 44
3.4.4 Multi-phase Configuration 45
3.4.5 Multi-level Configuration 46
3.4.6 PWM Technique to Improve the Performance
of the Six-pulse and multi-level Configurations 48
3.5 Power electronic Devices for ASVC Inverters 50
3.5.1 Insulated Gate Bipolar Transistor 50
3.5.2 Gate Turn-off Thyristor 50
Chapter 4: Basic Design of the ASVC 53
4.1 Introduction 55
4.2 Harmonic Study 56
4.2.1 General 56
4.2.2 Results of the Harmonic Study 58
4.3 Load Flow Study 61
4.3.1 General 61
4.3.2 Results of Load Flow Study 61
4.4 Transient Study 64
4.4.1 General 64
4.4.2 Results of the Transient Study 64
4.5 Basic Data of ASVC Design 66
4.5.1 General 66
4.5.2 Power Rating of the ASVC 66
4.5.3 Network Conditions 67
4.5.4 V/I Characteristic and Operating Points 67
4.5.5 ASVC Components 71
4.5.6 ASVC Losses 72
Chapter 5: Detailed Design of the ASVC 73
5.1 Component Arrangement 75

5.2 GTO-Converter Cubicles 77
5.3 Control Cubicle 78
5.4 Protection of Plant 79
5.5 Transformer 81
5.6 Iron Core Reactors 81
5.7 Auxiliary Supply 81
5.8 Control System Design and Implementation 82
5.8.1 Introduction 82
5.8.2 Interfaces 82
5.8.3 Control Scheme 84
5.8.4 TNA Testing 90
Chapter 6 : Installation at Rejsby Hede 95
6.1 Introduction 97
6.2 Civil Works 98
6.3 Interface with the Excisting Plant 102
6.3.1 Interface with the 15 kV Rejsby Hede Substation 102
6.3.2 ASVC Control 104
Chapter 7: Description of the Measuring Equipment 107
7.1 Introduction 111

7.2 Network Connection of the Wind Farm 111
7.2.1 The Signals 112
7.3 The Hardware 112
7.3.1 The Connection of the Transducers 112
7.3.2 The Capacitive Voltage Dividers 115
7.3.3 Signal Conditionering 115
7.3.4 The Personal Computers 115
7.3.5 Noise 116
7.4 Calibration of Measuring Equipment 118
7.4.1 The Capacitive Voltage Dividers 118
7.4.2 Calibration of the Signal Conditioning Unit 118
7.5 Software 120
7.5.1. Goals 120
7.5.2 Hardware Configuration 121
7.5.3 Basic Specifications 121
7.5.4 Programming Tools 122
7.6 Programs 122
7.61 Monitoring the Wind Farm 122
7.6.2 Monitoring the ASVC 124
7.6.3 Fast Data Acquisition 124
7.6.4 Conversion of Data 125
7.6.5 WEB-page 125
7.7 Measuring Results 125
Chapter 8: The Application of Advanced Static VAr

Compensators (ASVCs) to other Wind Farms 135
8.1 Introduction 137

8.2 Study Case 1 137
8.2.1 Steady State Performance 138
8.2.2 Dynamic Performance 142
8.3.1 20 kV Connection 151
8.3.2 66 kV Connection 151
8.3.3 20 kV Connection with Intermediate Load 153
8.4.1 The ASVC Characteristic 153
8.4.2 Application of the ASVC 156
8.5 Conclusions 159
Chapter 9: Conclusions 163

15
Chapter 1
Introduction
16 Introduction
Nicholas Jenkins
Department of Electrical Engineering and Electronics
UMIST
PO Box 88
Manchester
M60 1QD
ENGLAND
Introduction 17
1.1 Introduction to the Book
This book was written at the conclusion of a project to install and monitor the
performance of an 8 MVAr Advanced Static VAr Compensator (ASVC) at the
Rejsby Hede wind farm in Jutland, Denmark. The book is intended both to
record the main achievements of the project but also to illustrate how this new
type of power electronic equipment might be applied in other situations to
improve the quality of electrical power delivered to the distribution network
from renewable generation sources embedded within it.
Many changes occurred during the course of the project. A small change was
that it was decided internationally to refer to Advanced Static VAr Compen-
sators (ASVCs) as STATCOMs (Static Compensators). However as this change
occurred part way during the project the old name is retained in this book. Of
more consequence, the UN Climate Change Conference took place in Kyoto,
Japan and further increased the determination of European Governments to
reduce gaseous emissions. The installed capacity of wind turbines in Germany
exceeded 2000 MW and in Denmark 1000 MW. The installation of large arrays
of wind turbines offshore started to appear to be a practical proposition and
was encouraged by Danish Government policy. The separation of power utili-
ties into network and generation businesses continued world-wide and the
host utility of the project was split into two parts. Finally, progress was main-
tained in the technology of power electronic devices and systems.
All these changes served only to confirm the importance of the project. Wind
turbines and other forms of renewable generation embedded in the electricity
distribution networks will continue to play an important part in reducing the
CO2 emissions of European countries. The impact of wind farms on distribution
networks can be minimised and hence the connection costs reduced only if
effective reactive power schemes are in place. The separation of the power
utilities into a number of business units increases the requirements of the
transmission and distribution operator for independent control of reactive
power, while the continuing developments in power electronics offer the poten-
tial for cheaper and more effective compensators.
18 Introduction
1.2 Introduction to the Project
The main objective of the project was to investigate how the power quality of
the electrical output of wind farms could be improved by the use of modern
high power electronic technology. Although the research is of direct application
to wind energy it will also be relevant to many other types of small-scale gen-
eration embedded in utility distribution networks.
The operation of wind turbines with asynchronous generators requires reactive

power which, if supplied from the network, leads to low voltages and increased
losses. In order to improve the power factor of the generation, fixed capacitors
are usually used to provide reactive power. However, if they are sized for the
full requirement of the wind farm, they can cause self-excitation and poten-
tially damaging and hazardous overvoltages if the wind turbines connection
with the network is interrupted and they become islanded.
The reactive power consumption of wind turbines consists of a constant no-

load demand and a requirement which varies proportionally with the output of
the generator. In weak networks in particular this results in varying voltages
and poor power quality. The voltage quality can be improved considerably if
the wind turbines varying reactive power consumption is provided from a
dynamic compensator which supplies reactive power corresponding to the
varying consumption of the induction generators whereas only the constant
no-load consumption is provided from fixed capacitors. In this way the risk of
overvoltages due to islanding can also be eliminated.
An advanced Static VAr Compensator (ASVC) uses a power electronic convert-

er to generate or absorb reactive power. They can be used to provide reactive
power with rapid control and with only modestly sized passive components (i.e.
small capacitors and reactors). The objective of the project was to combine
research and development of this novel form of electronic equipment with its
application to increase the use of renewable energy, and wind power in partic-
ular, in the European Union.
1.3 Introduction to the Project Partners
The partners in the project were:
1.3.1 ELSAM ELTRA

At the start of the project ELSAM was the integrated Jutland-Funen power
pool and transmission authority. However, towards the end of the project the
organisation was split into its constituents parts and the project was completed
by the transmission utility which was renamed ELTRA.
Introduction 19
ELSAM was the initiator and co-ordinator of the project. Local support for the
project was provided by one of the 6 Jutland-Funen electrical utilities Snder-
jyllands Hjspndinsvrk.
1.3.2 SIEMENS
After extensive analysis of the wind farm system and associated network with
respect to impedans, harmonics and load flow, Siemens Power Quality Manage-
ment designed and manufactured the ASVC at their facilities at Erlangen,
Germany.
1.3.3 Technical University of Denmark (DTU)

The Technical University of Denmark designed and installed the measurement
equipment to monitor the performance of the ASVC.
1.3.4 University of Manchester Institute of Science and Technology (UMIST)

The role of UMIST was to undertake power systems analysis and simulations
in order to investigate the wider applications of ASVCs to wind farms. The
studies were carried out in collaboration with National Wind Power who pro-
vided details of typical wind farms connected to weak distribution networks.
20 INTRODUCTION
21
Chapter 2
Rejsby Hede Wind Farm

22 Rejsby Hede Wind Farm
Kent H. Sbrink
I/S Eltra
Fjordvejen 1-11
P.O. Box 140
DK-7000, Fredericia
Denmark
Rejsby Hede Wind Farm 23
2.1 Introduction
Wind power is one of the most promising sources of renewable energy, and the
application of wind turbines is one of the cheapest alternatives to reduce CO2
emissions from electrical power production. The disadvantages of wind tur-
bines are that it is impossible to control their power output and difficult to
predict their production of electrical power, as their production profile is only
determined by the wind speed. In addition, wind turbines with asynchronous
generators also have an impact on the voltage quality in the network to which
they are connected. This project addresses the last mentioned problem and
demonstrates how the power quality may be improved by using power elec-
tronic equipment Advanced Static VAr Compensators (ASVCs).
The goal of the Danish government is to reduce CO2 emissions by 20 % by the

year 2005 compared to the 1988 level and to halve emissions by the year 2030.
One of the important strategies to achieve this goal, is to promote electrical
wind power. Today, the total installed wind power capacity in the western part
of Denmark is 731.4 MW generated by more than 3,500 wind turbines, and the
yearly energy production is approximately 1,200 GWh. The total capacity in
2005 is estimated to be 1,500 MW and the potential capacity is estimated to be
4,000 MW in total mainly based on offshore wind farms.
Experience with offshore wind farms has shown that they are more efficient
than expected, and under certain circumstances, they are economically feasible
as a means to reduce CO2 emissions.
A commission with members from the Danish Ministry of Energy, the National
Forest and Nature Agency and the power industry proposed the erection of
4,000 MW of offshore wind farms within the next 25 to 30 years; 1,500 MW
before 2015 and 1,750 MW before 2030.
Therefore, the Danish electrical power industry recently announced their active
support for the Governments strategy concerning more wind farms. The
Commission has pointed out three offshore areas; two areas with a potential
capacity of 800 MW of wind power each, and a third area with a potential
capacity of 1,400 MW of wind power.
Norway
Sweden
150 kV
400 kV
HVDC
Zealand
Rejsby Hede
Germany
Figure 2.1: Map of the Danish Jutland peninsula.
2.2 The Rejsby Hede Wind Farm

The Rejsby Hede wind farm with a total capacity of 24 MW has been erected in
the southwestern part of Jutland by the local power production company
Snderjyllands Hjspndingsvrk as part of the Danish energy plan Energy
2000. Figure 2.1 shows the map of the Danish Jutland peninsula, showing the
main electrical circuits.
2.2.1 Description of the Wind Farm

The Rejsby Hede wind farm is located in a flat, high wind speed area near the
west coast. Because the energy content of the wind increases with the third
power of the wind speed careful location of wind turbines is vital for the eco-
nomical exploitation of the wind power. After a prolonged local planning
period, the 40 wind turbines were brought into service at Rejsby Hede in 1995.
No. of wind turbines 40

Wind turbine type M 1500 - 600/150 kW, MICON
Hub height 46 m
Rotor diameter/area 43.5m /1,452 m2
Blade type LM 19.1 m, with Vortex-generator, stall-
regulated
Gearbox FLENDER, PEAK 4280 - combined planet
wheel and cogwheel
Generator ELIN, MCT-445J21F9N, asynchronous
multiple-speed
Wind roughness class 1 - 2.5
Total installed power 24 MW
Expected yearly production 58,500 MWh
Table 2.1: Main data for the Rejsby Hede wind farm.
The wind farm consists of 40 Micon 600 kW turbines, of which one is privately
owned. The wind turbines are mounted on tubular towers at a hub height of 46
m. The three-bladed rotor is 43.5 m in diameter. The foundation of each turbine
consists of 70 m3 of concrete and 8 tonnes of steel reinforcement and takes up
an underground area of 100 m2. Table 2.1 gives details of the main data of the
wind farm.
The first wind turbine was erected during May 1995 and the last one was set
up during July. The period from July to October was used for testing, and the
production figures for this period were lower than expected for full operation.
The production figures from October to November 1995 were assumed to be
representative of normal operation. Based on commercial operating experience
in 1995, it was expected that the wind farm was to yield a total annual energy
production of 58.5 TWh. Table 2.2 shows the actual energy production of 48.7
TWh in 1996, which was lower than expected. Year 1996 was also less windy
than usual.
The wind farm output during the month of August of 1996 is shown in Figure
2.2. The curve is based on measurements on each quarter of an hour, and the
power output is shown in percentage of the installed capacity of the wind farm
(23.4 MW). Figure 2.3 shows the power duration and power loss curve for the
year 1996. The power duration curve is formed by normalising the power out-
put P/Pmax, i.e. dividing the measured power output values P by the installed
power capacity Pmax, and then by arranging all normalised power output values
in descending order.
The network losses increase with the square of the current (losses = R x I2),
which means that if the power factor is constant then the losses increase in
proportion to the square of the power. Therefore, the loss curve is obtained by
taking the square of the normalised power, i.e. the power duration curve.
The area below the power duration curve is equal the power duration time Td,
which expresses how long it takes to produce the actual yearly energy produc-
tion at full output power. Power duration time for 1996 was Td = 2080 h, which
corresponds to utilisation percentage of 23.7 % (= 2080 x 100 / 8760) of a max-
imum possible yearly production which is equal 8760 h multiplied by Pmax.
Similarly the area below the loss curve is equal to the loss factor or the loss
duration Tl, which expresses how long it takes to generate the actual yearly
losses at Pmax. The loss duration time for 1996 was 1062 h. The loss duration Tl
is less than the power duration Td, because the losses increase with the square
of the power, so the yearly power losses are generated in a shorter time at Pmax.
100
90
% of installed capacity
80
70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
Day of month
Figure 2.2: Wind farm power output in August 1996.

Month Energy production Peak power

(MWh) (MW)
January 5,210 23,621

February 5,317 23,394
March 5,285 21,607
April 1,859 18,953
May 4,143 23,454
June 4,182 22,038
July 4,071 19,778
August 2,826 22,117
September 3,211 22,107
October 4,015 23,949
November 5,277 24,387
December 3,280 22,979
Total 48,676 24,387
Table 2.2: Energy production 1996.
If the losses Plmax are known at Pmax and if Tl is known, then the capitalised
value of the losses may approximately be calculated as Tl x Plmax x Q, where Q
is the average marginal production price.
P/Pmax
1.0
0.9
0.8 Duration curve 1996

Loss curve 1996
0.7
0.6
0.5
0.4
0.3
0.2
0,1
Tl Td
0.0
0 1000 2000 3000 4000 5000 6000 7000 8000
Hour
Figure 2.3: The power duration and power loss curves, 1996.
2.2.2 Description of the Turbines

The turbines are up-wind, horizontal-axis, stall-regulated machines. Figure 2.4
shows components of the nacelle of each wind turbine. Each turbine drives a
dual-power, two-speed, asynchronous induction generator with an output of
150 kW at wind speeds of up to 7-8 m/s. The generator automatically switches
to 600 kW at higher wind speeds. For each turbine three 50 kVAr shunt capaci-
tors are provided for compensation of the no-load reactive consumption of the
asynchronous generators. When a turbine is operated with its 150 kW genera-
tor one capacitor (50 kVAr) is switched in, and when a turbine is operated with
the 600 kW generator three capacitors (total 150 kVAr) are switched in.
A purpose-made electronic unit controls each turbine and all the wind tur-
bines are connected to the local distribution companys control centre at
Skrbk. The system provides control for all the functions of the turbines,
records details of production and faults, and provides status reports.
Main shaft Radiator
Main bearing Gearbox Two-speed generator
Rigid body frame
Fail-safe hydraulic Flexible coupling

Bearing and gear ring for
disc brake system
the yawing system
Yaw motors
Figure 2.4: Components of the nacelle.
2.2.3 Description of the Local Network

The Rejsby Hede wind farm is connected to a 60 kV substation in a radial dis-
tribution network which is supplied from a 60/150 kV substation in Bredebro
as shown in Figure 2.5. Apart from the Rejsby Hede wind farm, the 60 kV dis-
tribution network supplies two villages Hjer and Ballum. Normally, the loads
in Hjer and Ballum are less than the wind farm production, which is why the
Hjer Ballum
Pmax 4.7 MW Pmax 3.5 MW

10 MVA 10 MVA
60/15 kV 6/15 kV Bredebro
Rejsby Hede 60 kV 150 kV
8.2 km 3.5 km 10.9 km
23 Mvar
15 kV 15 kV
15/0.69 kV ASVC 15/0.69 kV

8 Mvar
22 x 600 kW 18 x 600 kW
Figure 2.5: Single-line diagram of the 15 kV and 60 kV networks.
excess power production from the Rejsby Hede wind farm is fed into the 150
kV network in Bredebro.
Figure 2.5 shows that the 60 kV substation in Bredebro is provided with a 23

MVAr shunt capacitor. This means that if the wind farm together with the con-
sumers in Ballum and Hjer are isolated from the 150 kV network, i.e. if the
transformer breaker in Bredebro trips, the presence of the 23 MVAr capacitor
in Bredebro may give rise to self-excitation of the wind turbine generators.
Then, the wind farm becomes overcompensated, which leads to excessive over-
voltage and the danger that the consumers electrical equipment may be
destroyed.
In the wind farm, each wind turbine generator is equipped with a 0.69/15 kV
step-up transformer. All transformers are connected to the local 15 kV network
cable which consists of five radial feeders, of which two are connected to nine
generators each. Of the remaining three feeders, one is connected to six gene-
rators and the remaining two are connected to eight generators each. In order
to reduce the short-circuit level, the 15 kV switchgear is split into two busbars.
Two feeders with 18 generators are connected to one of the 15 kV busbars
while the remaining three feeders with 22 generators are connected to the
other busbar. The two 15 kV busbars are connected to the 60 kV substation
through a 15/15/60 kV three-winding step-up transformer.
2.2.4 Need for Reactive Compensation

The need for reactive power in electrical networks is due to energy stored in
electrical and magnetic fields in the network components such as overhead
lines, cables, capacitors, reactors, transformers and motors. Models of such net-
work components contain voltage and current sources and pure passive compo-
nents which are either Resistors (R), Inductors (L), Capacitors (C) or a combi-
nation of any of them.
Figure 2.6 shows the relationship between current and voltage for a resistive,
an inductive and a capacitive component. The current flowing through a resis-
tor and the voltage across it are in phase. Therefore, the electrical power in a
resistor, equal to current multiplied by voltage, is dissipated as heat.
In a capacitor, the current leads the voltage by 90 degrees. This means that
when the a.c current is at its maximum the voltage is zero, and when the volt-
age is at its maximum the current is zero. Therefore, no power is dissipated in
a pure capacitor. However, energy is stored in the electric field in the capacitor.
The capacitor alternately takes up energy and gives off energy as the a.c volt-
age across the capacitor alternates between plus and minus.
In a inductor, the current lags the voltage 90 degrees. As with capacitors, no

heat is dissipated in an inductor. However, energy is stored in the magnetic
field in the inductor. The inductor alternately takes up energy and gives off
energy as the a.c. current in the inductor alternates between plus and minus.
Inductive and capacitive currents are in opposite phases which means that the
electrical energy in capacitors and the magnetic energy in inductors are over-
lapping in an a.c network. When inductors deliver energy, capacitors consume
Ic
Ic IL IR
Ug U
IR
Xc XL R
IL
Figure 2.6: AC voltage and current in RLC components.

energy and vice versa. By definition, reactive power in capacitors is negative

and is calculated as Qc = CU2, whereas reactive power in inductors is positive
and calculated as
U2
QL = .
L
It is said that capacitors generate reactive power whereas inductors absorb
reactive power.
If there was no resistance in the network, this flow of magnetic and electric
energy or generation and absorption of reactive power would persist without
losses. However, except for super conducting material, all electrical compo-
nents have some resistance, which means that the transfer of reactive power
from capacitors to inductors causes losses in resistances in the network. If an
inductor is connected to an a.c. network, it will absorb reactive power from the
network, which will cause losses and voltage drops in the network. However, if
a capacitor is connected close to the inductor, the capacitor can deliver the
reactive power required by the inductor.
Figure 2.7 shows an example in which a reactive load (RL, XL) is compensated
with a shunt capacitor QC. If the reactive power from the shunt capacitor QC is
equal to the reactive power required by the load, the transfer of reactive power
from the generator (QL-QC) is zero, which means that losses and voltage drop,
in the line impedance (RL, XL), caused by this transfer of reactive power from
the generator are eliminated. The reactive power is transferred over a much
shorter distance from the shunt capacitor to the inductive load.
As mentioned above, models of network devices may be represented by RLC

components. The wind turbines are equipped with induction generators, which
Q L - Qc QL
PL PL
Xl Rl
Ug Qc
RL PL XL QL
shunt load
capacitor
Figure 2.7: Reactive power compensation with a shunt capacitor.

have a very simple, robust and reliable construction. An induction generator

consists of stator windings and rotor windings. The stator windings are embed-
ded in the stator, which is the non-rotating part of the machine, whereas the
rotor windings are wound on the rotor, which is the rotating part of the
machine. The rotor windings are short-circuited. If the stator winding is con-
nected to a balanced three-phase a.c. network, it creates a magnetic field
which rotates in synchronism with the network frequency. The rotating field
created by the stator induces voltages and currents in the rotor. The rotor cur-
rents react with the rotating field and create torque, which accelerates the
rotor in the direction of the rotating field. The induced voltage in the rotor
depends on the relative speed between the rotating field and the rotating rotor.
When the rotor accelerates, the relative speed between rotor and the rotating
field decreases and if the relative speed is zero no voltage and current are
induced in the rotor. In order to produce torque the speed of the rotor must
differ from the speed of the rotating magnetic field.
The rotational speed for a P-pole machine is given by

120
ns = ,
P
where ns is in rad/min, and P is the number of poles. The rotor speed nr must
be different from ns in order to create torque in the air gap between the rotor
and the stator. The per-unit slip of the rotor is defined as
ns nr
s= .
ns
If nr is less than ns, then the slip is positive and the induction machine is driven
as a motor. If ns is greater than nr, the slip is negative and the induction
machine is driven as a generator where external mechanical torque is applied
to the rotor, which is the case of wind turbine generators.
Figure 2.8 shows the equivalent circuit of an induction machine with all quan-
tities referred to the stator. Rs is the stator resistance, Xs is the stator leakage
reactance, Rr is the rotor resistance and Xr is the rotor reactance. Xm is the
magnetising reactance and s the rotational slip speed. Pa is the power trans-
ferred across the air gap and Ps is the mechanical power on the shaft of the
induction machine. As can be seen from the equivalent circuit, the induction
machine has a constant reactive power consumption which corresponds to Xm,
and a variable reactive power consumption which varies with the load.
The constant reactive power consumption is also known as no-load reactive

power consumption. The no-load reactive power depends only on the terminal
voltage and is therefore easy to compensate with a shunt capacitor connected
at the terminals of the induction machine. To prevent overcompensation, which
may lead to self-excitation and overvoltages, normally only 80 to 90 % of the
constant no-load consumption is compensated with shunt capacitors.
Rs Xs Pa Rr Xr Ps
XM Rr 1-S
S
Figure 2.8: Equivalent circuit of an induction machine.
The reactive power consumption corresponding to Xs and Xr varies with the

loading of the induction machine. This variable consumption is not easy to
compensate and so it is usually drawn from the network. This causes voltage
drops and losses due to the reactive power flow in the network. Therefore, it
would be an advantage if this reactive power could be provided from a variable
source such as an Advanced Static VAr Compensator (ASVC).
The main consumers of reactive power in the wind farm are the induction gen-
erators, whereas reactive losses in the lines are negligible. The reactive con-
sumption of each 600 kW generator is 219 kVAr at no-load and 372 kVAr at
full load, as given in Table 2.3. For the 150 kW generator the reactive consump-
tion is 87 kVAr at no-load and 124 kVAr at full load (see Table 2.3 for full
details of the machine). The total reactive power required by the wind farm is
8,760 MVAr at no-load and 14,880 MAVr at full load.
For no-load compensation, three 50 kVAr shunt capacitors are connected

across the terminals of each wind turbine generator. When the wind turbine is
operated with the 150 kW generator, only one 50 kVAr capacitor is connected.
When the wind turbine is operated with the 600 kW generator, all three 50
kVAr (150 kVAr) shunt capacitors are connected.
In total the wind farm is provided with 6 MVAr shunt capacitors for no-load
compensation. Compared with the total no-load consumption of the wind farm,
the compensation ratio is 68.5 per cent.
The wind farm operates with any mixture of 150 kW and 600 kW generators
and the variable reactive power requirement ranges from 0 to 8.880 MVAr.
Nominal power 600 kW 150 kW

Nominal apparent power 706 kVA 195 kVA
Number of poles 4 6
Coupling star star
Nominal voltage 690 V 690 V
No-load voltage 715 V 715 V
Frequency 50 Hz 50 Hz
Synchronous rev. 1500 rpm 1000 rpm
Full-load rev. 1605.7 rpm 1004.4 rpm
Efficiency 4/4 0.9732 ohm 0.9594 ohm
Slip 0.4 % 0.37 %
No-load consumption 219 kVAr 87 kVAr
Full-load consumption 372 kVAr 124 kVAr
No-load current 183 A 81 A
Full-load current 581.1 A 158.02 A
Cos phi 100 % 0.85 0.77
Table 2.3: Main data for the MICON 600/150 kV induction generator.
However, an 8 MVAr ASVC, which is a standard size, was chosen. At full power
output from the wind farm, there is a reactive power deficit of 880 kVAr. This
is drawn from the network with no real impact on the voltage quality.
Moreover, the wind farm is only operated for a few hours per year at full out-
put power according to the power duration curve depicted in Figure 2.3.
35
Chapter 3
Reactive Compensation
for
Wind Farms
36 Reactive Compensation for Wind Farms
Janaka Ekanayake
UMIST Visitor
Department of Electrical and Electronic Engineering
Peradeniya
Sri Lanka
Nicholas Jenkins
UMIST
PO Box 88
Manchester
M60 1QD
ENGLAND
Reactive Compensation for Wind Farms 37
3.1 Introduction to Reactive Power Compensation [1]
Reactive power compensation is the absorption or generation of reactive power

to maintain efficient and reliable operation of the power system and its loads.
It can be used to: (1) maintain load voltages and/or voltages of a transmission
network within maximum and minimum limits, (2) increase the power trans-
mission capability of lines while maintaining stability of the power system and
(3) minimise the reactive power flow in the circuits, by compensating large
reactive power consuming loads locally, and thus reduce the network losses.
Reactive power compensation has often been used in transmission systems to

improve network voltages and power transfer capability and also in distribu-
tion networks to compensate large reactive power consumers. Large reactive
power consumers are encouraged by the power utility to increase their power
factor, thereby reducing the reactive power demand on the network. The result
is a better voltage profile, reduced losses and increased stability of the system.
Wind farms are a rather unusual type of large reactive power consumer as they
import reactive power while generating active power. Many of the electrical
networks, to which wind farms are connected, are weak with high source
impedances. The output of a wind farm changes constantly with wind condi-
tions and so it draws varying amounts of reactive power from the utility net-
work resulting in variations in the voltage at the point of connection. A reac-
tive power compensator can be connected to the wind farm to supply this reac-
tive power locally and thus improve the steady state voltage profile at the point
of connection.
Another problem with wind farms is that, rapidly changing wind conditions
may create an objectionable voltage flicker condition. In this case the magni-
tude of the voltage variations are relatively small but occur at a frequency
which causes severe annoyance to the human eye. In such cases, a compensator
of appropriate rating may be employed to compensate the voltage changes and
hence eliminate flicker.
A reactive power compensator can also be connected to improve the steady

state stability limit of the network to which the wind farm is connected.
Effective voltage control through reactive power compensation enables the
connection of medium-size and large wind turbines to weak networks.
3.2 Impact of Reactive Power Flows on a Simple Circuit [2]
The basic circuit shown in Fig 3.1 can be considered to demonstrate the impact
of reactive power flow and how a reactive power compensator can be used to
minimise them.
Fig 3.1: Simple electrical power system.
Assuming the power flow in the line is P+jQ:
* = P + jQ
VRIR (3.1)
P - jQ (3.2)
IR =
V*R
Then
Vs = VR + (R + jX)IR (3.3)
P - jQ
= VR + (R + jX)
V*
R
According to the phasor diagram,
if VR = VR 0, then Vs = Vs
with usual complex notation. Then from equation (3.3), equating real and
imaginary parts the following equations can be obtained.
(3.4)
Vs cos - VR = PR + QX
VR
PX - QR (3.5)
Vs sin =
VR
As is very small, from equation (3.4)

PR + QX
Voltage drop across the line = Vs - VR = (3.6)
VR
Generally for transmission networks, R is much less than X and from equation
(3.5)
VsVRsin (3.7)
P=
X
For a circuit with a small resistance, R, it is clear that the voltage drop across
the line mainly depends upon the reactance of the line, X, and the reactive
power flow, Q. If the reactive power requirement of the load can be supplied
locally by connecting a reactive power compensator at the load bus, the voltage
drop across the line can be reduced, the steady state voltage at the load bus
can be raised and any flicker can be improved.
3.3 Conventional Methods of Shunt Reactive Power Compensation
Reactive power compensation can be achieved with a variety of shunt devices.

The simplest method is to connect a capacitor or an inductor in parallel with
the network. As the reactive power output of these component is not control-
lable, devices such as synchronous condensers and saturable reactor compen-
sators have also been used. However when a consumer creates rapidly varying
voltage fluctuations, it is not possible to improve the quality of power supply
with the aid of simple compensators. Therefore fast control reactive power
compensators, employing solid state devices, have emerged to supersede the
earlier technologies.
3.3.1 Shunt Capacitors and Reactors

Shunt capacitors and reactors may be used to supply or absorb reactive power.
In most cases they are connected to the power system by a circuit breaker
either directly or via a transformer and switched in and out when needed. As
the dynamic response of these shunt devices is determined by the time and per-
mitted frequency of operation of the circuit breaker, the usefulness of these
devices is limited. Another major drawback of them is that the reactive power
produced or absorbed by them falls as the voltage drops, leading to reduced
effectiveness when they are needed most.
3.3.2 Synchronous Condenser

A synchronous condenser is a synchronous motor running without a mechani-
cal load. Depending on the value of excitation, it can absorb or generate reac-
tive power. The synchronous condenser exhibits great advantages such as flexi-
bility of operation in all load conditions and an essentially inductive source
impedance that cannot cause harmonic resonances with the transmission net-
work. However it suffers from a number of shortcomings, such as slow response
(the average response time is around one second), requirement for significant
starting and protection equipment and high maintenance.
3.3.3 Thyristor Controlled Reactor (TCR)

The basic elements of a TCR are a reactor in series with a bi-directional thyris-
tor pair as shown in Fig 3.2(a). The thyristors conduct on alternative half-
cycles of the supply frequency. The current flow in the inductor (L) is control-
led by adjusting the conduction interval of the back to back connected thyris-
tors. This is achieved by delaying the closure of the thyristor switch by an angle
a, which is referred to as the firing angle, in each half cycle with respect to the
voltage zero. When =90, the current is essentially reactive and sinusoidal.
Partial conduction is obtained with firing angles between 90 and 180.
Outside the control range, when the thyristor is continuously conducting, the
TCR behaves simply as a linear reactor. The voltage current characteristics for
the TCR is shown in Fig 3.2(a).
It is important to note that the TCR current always lags the voltage, so that
reactive power can only be absorbed. However, the TCR compensator can be
biased by shunt capacitors so that its overall power factor can either be lag-
ging or leading. The voltage current characteristic of a TCR with a fixed or
switched capacitor is shown in Fig 3.2(b).
A slope characteristic, ks, shown in Fig 3.2(b), is used to control the voltage at
the point of connection (PC) of the TCR and any fixed or switched capacitor
arrangement (This combination is normally called as a static VAr compensator
- SVC). The electric power system when viewed from the PC busbar can be
represented by a Thevenin equivalent circuit with an equivalent reactance XTH.
The power system characteristics can be represented by a straight line with
slope XTH as also shown in Fig 3.2(b). To limit the voltage variation at the PC
with the set point voltage, Vo, the value of ks is selected to be less than XTH,
which results in voltage control with a droop [3].
Under the phase control mode, the TCR generates higher harmonic currents. In
three phase applications, the third harmonic components can be cancelled out
Fig 3.2(a): Output characteristic of TCR.

Fig 3.2(b): Output characteristic of a TCR with capacitors.
by connecting the basic TCR elements in delta through a transformer. The

transformer is necessary for matching the mains voltage to the thyristor valve
voltage. Further elimination of harmonics can be achieved by using two delta
connected TCRs of equal rating fed from two secondary windings of the step-
down transformer, one connected in star and other in delta. This forms a 12
pulse TCR as shown in Fig 3.3. Moreover the harmonics in the line current can
be reduced by replacing the fixed capacitors, associated with reactive power
generation, with a filter network. The filter can be designed to draw the same
fundamental current as the fixed capacitors at the system frequency and pro-
vide low-impedance shunt paths at the harmonic frequencies.
Fig 3.3: 12 - pulse TCR arrangement.
3.3.4 Thyristor Switched Capacitor (TSC)

The basic elements of a TSC are a capacitor in series with a bi-directional
thyristor pair and a small reactor. The purpose of the reactor is to limit switch-
ing transients, to damp inrush currents and may be used to form a filter for
harmonics coming from the power system. In three phase applications the
basic TSC elements are usually connected in delta. The susceptance is adjusted
by controlling the number of parallel capacitors connected in shunt. Each
capacitor always conducts for an integral number of half cycles. The total sus-
ceptance thus varies in a stepwise manner. The one line diagram of the TSC
scheme is shown in Fig 3.4(a). The output characteristic is discontinuous and
determined by the rating and number of parallel connected units. Therefore
the voltage support provided is discontinuous as shown in Fig 3.4(b).
The switching of capacitors excites transients and it is necessary to switch the

capacitors at a point where switching transients are minimum. This can be
achieved by pre-charging the capacitor to the crest value of the supply voltage
and by choosing the switching instant when the voltage across the thyristor
switch is minimum, i.e. at the crest value of the supply voltage. The switch off
period corresponds to a current zero after an integral number of half cycles.
More detailed analysis of transients associated with the TSC can be found in
[1 & 4].
Fig 3.4: TSC scheme and its output characteristics.
3.4 Advanced Static VAr Compensator (ASVC)
3.4.1 Introduction
Over the last few decades, the power handling capability of power electronic
devices has increased and new solid state switches with turn-off capability
have emerged. This has led to a reduction in the number of switching devices
required for a high power static VAr compensator and a corresponding
decrease in the cost of power electronic control [5]. This falling tendency in
specific price and introduction of new devices with turn-off capability is likely
to encourage a shift towards the increase use of solid state means for both the
generation (or absorption) and control of reactive power. As a results of this
trend all solid state reactive power compensators have started emerging. One of
the more developed of this type of reactive power compensator is the ASVC
based on the voltage source inverter (VSI).
3.4.2 Six-pulse Voltage Source Inverter Operating with

Fundamental Frequency Modulation
This is the simplest form of three phase dc to ac inverter and can be considered
as a basic building block of an ASVC. The power circuit of the six-pulse VSI is
shown in Fig 3.5 together with its output voltages. This circuit consists of six
power electronic switches with both turn-on and turn-off capability and six
diodes. In Fig 3.5, the power electronic switches are shown by an arrow with
arrowhead indicating the direction of current flow. The basic square wave out-
put, shown in Fig 3.5, can be obtained by operating the six switches in the
sequence, S1S2S3, S2S3S4, S3S4S5, S4S5S6, S5S6S1 and S6S1S2. From Fourier analy-
sis, it can be shown that, for an ideal wave shape, only harmonics of order of
6k1, where k = 1, 2, 3..., remain in the output waveform. For the ideal circuit,
the kth harmonic component has an amplitude 1/k relative to the fundamental
component. When the inverter is operating with this switching pattern, the dc
side current pulsates six times during one cycle of the alternative voltage and
therefore this circuit is called a six-pulse inverter.
3.4.3 Principle of Operation of a Six-pulse VSI based ASVC

The ASVC consists of a voltage source inverter whose DC side is connected to
a capacitor. The inverter is connected to the supply system via a transformer. A
basic ASVC circuit is shown in Fig 3.6(a), in which transformer leakage reac-
tance is represented by X. If the fundamental of the phase output voltages of
the inverter is in phase with the corresponding system voltage, the line current
flowing into or out of the VSI is always at 90 to the network voltage due to the
reactive coupling. When the fundamental of the inverter voltage is less than the
ac system voltage, reactive power is absorbed by the ASVC. On the other hand,
when the ASVC voltage is higher than the system voltage, reactive power flows
from the ASVC to the system. This is shown in Fig 3.6(b) and (c). The magni-
tude of the ASVC output voltage can be varied by controlling the dc capacitor
voltage. If the switching devices of the inverter are operated to obtain the fun-
damental of the ASVC output voltage leading or lagging the ac system voltage
by a small angle, a net amount of real power flows between the system and the
ASVC. This in turn decreases or increases the dc capacitor voltage.
The harmonic distortion at the output of this six-pulse inverter based ASVC is
very high and the circuit is not used for practical applications. Various other
inverters have emerged which operate either at the fundamental frequency or
at higher frequencies. In next few sections, these types of inverters are dis-
cussed.
Fig 3.5: Six-pulse voltage source inverter and its line voltages.
3.4.4 Multi-phase Configuration

A block diagram of this circuit configuration is shown in Fig 3.7. A 6P pulse
inverter (where P is an integer) can be obtained by connecting P six-pulse
inverters having a 2/6P phase displacement between each other, through a
transformer array. This transformer array has a complex winding arrangement
which is used to cancel the phase displacement of the inverter outputs.
In-phase transformed outputs of all the inverters are summed by the connec-
tion of the windings. Each of the P transformers required for this arrangement
is rated at 1/P of the total MVA output of the inverter.
If this 6P-pulse inverter is connected to a network having a frequency f, the

output frequency of each harmonic present in the ac side is (6Pk1)f, where k
is a positive integer.
Fig 3.6: Six - pulse inverter as an ASVC.
Examples of a 24-pulse inverter configuration and a 48-pulse configuration

can be found in references [5] and [6] respectively.
3.4.5 Multi-level Configuration [7]

A simplified schematic of a multi-level inverter having an arbitrary number of
levels is shown in Fig 3.8(a). Fig 3.8(b) shows one phase of the actual inverter
circuit. The number of levels in the n-level inverter is defined as the number of
positive voltage levels including zero which appear in the inverter line voltage
or by the number of DC voltage levels referred to the mid-point.
The output voltage of this configuration can assume any of the voltage levels
V1, V2........Vn+1 by selecting appropriate nodes. For an example, the voltage level
V45can be obtained at the output by turning the switches S1 to Sn on and turn-
ing all the other switches off. The voltage level V2 can be obtained by turning
switches S-1 to S-n-1 on and turning all the other switches off [8]. When the cur-
rent flowing out from the inverter leg is positive, only the switches correspond-
ing to that voltage level are conducting. On the other hand, when the current is
negative the anti-parallel diodes corresponding to the on-switches are con-
ducting.
Fig 3.7: Schematic diagram of the multi-phase inverter.
Fig 3.8: Equivalent circuit and the real arrangement of

one phase of a multi-level inverter.
The average voltage across each capacitor in the multi-level inverter should be
identical. However, when establishing sinusoidal current waveforms, the cur-
rents flowing out from the inner capacitors (towards the mid point of the in-
verter) are higher than that from the outer capacitors, causing voltage unbal-
ances in the string of capacitors. Various techniques, either based on a multi-
level PWM method [9] or employing a regulator circuit [8], are described in the
literature to overcome these unbalances.
An innovative approach is to use a cascaded inverter (sometimes known as a

chain circuit). Figure 3.9 shows a simple representation of a 5 level cascade
inverter. This uses a switching combination of the GTOs according to a preset
switching scheme to obtain an output voltage of five levels per half fundamen-
tal period. PWM techniques may also be used in the inverters switching
scheme to further reduced the harmonic content.
Fig 3.9: A 5-level cascade inverter.
3.4.6 PWM Technique to Improve the Performance of the

Six-pulse and Multi-level Configurations
The six-pulse inverter and multi-level inverter can be operated with a higher
switching frequency to eliminate some of the lower order harmonics appearing
in the inverter output voltage. Two common switching strategies which can be
used for ASVC applications are selective harmonic elimination modulation
(SHEM) and overmodulated or optimum pulse width modulation (PWM) tech-
niques.
In the SHEM technique, the basic square wave output produced by the six-
pulse or multi-level inverter is chopped a number of times to eliminate some of
the lower order harmonics. However the magnitude of some higher order har-
monics increases in this process. Fig 3.10 shows the voltage waveform corre-
sponding to this case with an arbitrary number of chops per half a cycle.
Assuming that the angle corresponding to the chop near to 90 is k, the
Fourier coefficient bn can be found as [10]:
2Vd
bn = [ k
n 1 + 2(-1)
i=l
i cos(n )
i ] for odd n (1)
Fig 3.10: Waveform for SHEM operation.
By introducing m number of chops into the inverter output voltage as shown in

Fig 3.10, any m number of harmonics can be eliminated from the output volt-
age. The angles corresponding to the m chops can be found by equating bn = 0
for the m harmonic to be eliminated and solving those m simultaneous equa-
tions for i (for i = 1 to m). As an example, if two chops were used with
1 = 16.3 and 2 = 22.1, the 5th and 7th harmonic components can be eliminated
from the output voltage waveform.
There are various PWM methods described in the literature [11, 12 & 13]. The
sinusoidal PWM technique is the basic one and most of the other techniques
are derivations of it. In the sinusoidal PWM technique, the angle of chops is
determined by the intersection between a sine wave and a triangular wave. To
obtain a three phase output the same triangular wave is compared with three
sinusoidal signals that are 120 out of phase. The sinusoidal waveform is at the
frequency which is required at the inverter output and the triangular wave-
form normally at a triplen multiple (mf) of that frequency. If the magnitude of
the sinusoidal waveform is less than that of the triangular waveform, then the
inverter output produces harmonics around a multiple of mf times fundamental
frequency and no harmonics appear below that. However, the magnitude of the
fundamental is very low in this case and it becomes a limiting factor in deter-
mining the size of the component needed for the converter and its control
requirements. Therefore overmodulation techniques, where the magnitude of
the sinusoid is higher than the triangular waveform, or some optimum PWM
pattern are often employed.
3.5 Power Electronic Devices for ASVC Inverters
The ideal power electronic device for an ASVC is a switch which can be turned
on or off at any instant of time by a simple control signal. Since voltage source
inverters are commonly used for the ASVC, these switches do not need to have
reverse blocking capability [14]. The commonly used switches for the ASVC
which satisfy the above specification are the Insulated Gate Bipolar Transistor
(IGBT) and Gate Turn-off Thyristor (GTO). GTOs are available up to very large
power levels such as 6 kV, 6 kA and IGBTs are available up to power level of
1500V, 1200A. The power rating of IGBTs is increasing constantly while their
cost is decreasing.
3.5.1 Insulated Gate Bipolar Transistor

The IGBT is a hybrid switch which consist of a MOSFET in its gate side and a
Bipolar Junction Transistor (BJT) in its conduction path as shown in Fig
3.11(a). Its circuit symbol is shown in Fig 3.11(b). This device combines the
characteristics of both the BJT and MOSFET. The conduction losses at on state
are less than that of a similar rated MOSFET and the switching on and off
times are lower than that of a similar rated BJT. Unlike the BJT, the IGBT is a
voltage driven device and can be easily switched on. The IGBT, turn-on process
is similar to that of a MOSFET, and in the turn-off process the BJT plays a
dominant role. Therefore during turn-off a long tail current flows, increasing
the switching losses.
The IGBT has become a popular choice for medium power applications,
because of its attractive characteristics. It would seem to be an appropriate
choice for the smaller ASVCs which might be employed in future wind farms
as their power level matches the reactive power requirement of the wind farm,
i.e. in the order of few MVArs.
3.5.2 Gate Turn-off Thyristor

The GTO is a switch having a thyristor like p-n-p-n structure but with turn-off
capability. In the case of the thyristor, once it is turned-on by a gate pulse, the
gate no longer has control over the conduction of the device and can not be
turned-off unless the conduction current is brought below a certain minimum.
Therefore various structural changes are introduced into the thyristor structure
to obtain the turn-off capability. First its p type anode is shorted by n+ regions
as shown in Fig 3.12(a) and cathode appears as n+ type islands on the p type
gate. The circuit symbol of a GTO is shown in Fig 3.12(b).
a) b)
Fig 3.11(a): Equivalent circuit of the IGBT.

Fig 3.11(b): Circuit symbol.
The GTO can be turned on by applying a positive voltage to the gate with
respect to the cathode. This creates a positive gate current pulse and the basic
turn-on process is similar to that of a thyristor. The device can be turned off by
applying a large negative gate current. This negative gate current must be very
large, in the order of 20 - 30% or even higher of the anode current, but is
required for only a short period of time. Therefore special drive circuits are
required to turn the device off. The turn-on and turn-off times are longer than
those of an IGBT.
This device is appropriate for large ASVCs whose rating is few MVArs to few
hundred of MVArs. The application of GTOs to ASVCs are described in detail
in reference [5].
a) b)
Fig 3.12(a): Structure of the GTO.

Fig 3.12(b): Circuit symbol of the GTO.
References
1. Miller, T.J.E., Reactive power control in electric systems., (Book), John

Wiley and Sons, 1982.
2. Weedy, B.M., Electric power system., (Book), John Wiley and Sons, 1991.
3. Erinmez, I.A. Ed., Static VAr compensators., Working group 38-01, Task
Force No. 2 on SVC, CIGRE, 1986.
4. Miller, T.J.E., Chadwick, P., An analysis of switching transients in Thristor
Switched Capacitor compensated systems., IEE International Conference
on Thyristor and Variable Static Equipment for AC & DC Transmission,
London, Nov/Dec 1981, pp 104-107.
5. Gyugyi, L., Hingorani, N.G., Nannery, P.R., Tai, N., Advanced Static VAr
compensator using Gate Turn-Off thyristors for utility applications.,
CIGRE 23-203, 1990.
6. Mori, S., Matsuno, K., Takeda, M., Seto, M., Development of a large static
VAr generator using self-commutated inverters for improving power system
stability., IEEE Tran. on Power Systems, Vol 8, No 1, Feb 1993, pp 371-377.
7. Newton C., Summer M., Multi-level convertors: a real solution to medi-
um/high voltage drives?, Power Engineering Journal, February 1998,
pp 21-26.
8. Choi, N.S., Cho, J.G., Cho, G.H., A general circuit topology of multi-level
inverter., IEEE Power Electronic Specialist Conference, USA, 1991,
pp 96-103.
9. Choi, N.S., Cho, G.C., Cho, G.H., Modelling and analysis of a static VAr
compensator using multi-level voltage source inverter., IEEE Industrial
Application Society 28th Annual Meeting, Oct 2-8, 1993, Canada,
pp 901-908.
10. Petal, H.S., Holt, R.G., Generalised technique of harmonic elimination and
voltage control in thyristor inverters, Part I - Harmonic elimination., IEEE
Tran. on Industrial Applications, Vol IA-9, No 3, May/June 1973,
pp 310-317.
11. Wuest, D., Static reactive power compensator with self-commutated invert-
er at a low switching frequency., European Power Engineering Conference,
Firenze, 1991, Vol 3, pp 70-75.
12. Boost, M.A., Ziogas, P.D., State-of-the-art carrier PWM techniques: A criti-
cal evaluation., IEEE Tran. on Industrial Application, Vol 24, No 2,
March/April 1988, pp 271-280.
13. Bowers, S.R., Midoun, A., Sub-optimal switching strategies for micro-
processor-controlled PWM inverter., IEE Proceedings, Vol 132, Part B, No
3, May 1985, pp 133-148.
14. Gyugui, L., Reactive power generation and control by thyristor circuits.,
IEEE Tran. on Industrial Application, Vol IA-15, No 5, Sep/Oct 1979,
pp 521-531.
53
Chapter 4
Basic Design of the ASVC

54 Basic Design of the ASVC
Frank Schettler
SIEMENS, Power Transmission and Distribution
Erlangen
Germany
Basic Design of the ASVC 55
4.1 Introduction
Figure 4.1 shows a single line diagram of the ASVC installed at the Rejsby
Hede wind farm. It consists of the following major components:
- one three winding transformer

- two reactors, one connected to each transformer secondary
- two converter units with capacitive energy storage
A converter unit converts the DC-voltage of the energy storage into a three
phase AC voltage. The converters used are of the three level type, i.e. the energy
storage is separated into two series connected capacitors and one phase mod-
ule of the converter connects sequentially the positive tapping, the negative
tapping or the mid-point to the AC terminal. Therefore the generated AC volt-
age has a stair-step wave form. A typical wave form of a line to line voltage is
shown in Figure 4.2. Wave forms like this are obtained if each GTO is switched
on and off once a period of the system frequency (Fundamental Frequency
Modulation - FFM).
Fig 4.1: Single line diagram of the ASVC.

The magnitude of the fundamental component of the converter voltage wave in

respect to the system voltage determines the reactive power output of the
ASVC. Moreover, due to the stair-step wave form the ASVC voltage contains a
higher harmonic voltage content. The ordinal numbers of the characteristic
harmonics generated are
n= 6 x k 1 (k=1, 2 . .n)
Besides these harmonics non-characteristic ones due to DC voltage ripple, sys-

tem unbalances and switching tolerances appear.
Fig 4.2: Line to line voltage at the converter terminals (Example using FFM).
Applying higher harmonic voltages results in voltage distortions in the sur-

rounding network. These voltage distortions can be influenced by varying the
pulse pattern of the converter and changing the transformer impedance.
Connecting filters was not considered to be a possible solution. However,
changing the harmonic content of the converter voltage as well as the variation
of the transformer impedance also has an effect on the fundamental rating of
the ASVC. That calls for a harmonic optimisation process.
4.2 Harmonic Study
4.2.1 General
The distortions in the system caused by the ASVC operation have to be below
specified limits. Figure 4.3 shows a single line diagram of the wind park and
the 60 kV network. The following harmonic requirements are to be met:
- Harmonics to be considered: 2nd to 50th

- Total Harmonic Distortion (THD): < 2.5 %
This values refer to the 15 kV bus where the ASVC is connected to (Point of
Common Coupling - PCC).
- Individual Harmonic Content at the PCC: < 1.5 %
- The ASVC must not cause overloading of the reactive power compensation
capacitors connected to the wind turbines and the 23 MVAr capacitor bank
connected to the Bredebro 60 kV bus. The pre-load due to other sources of
higher harmonics in the system is assumed to be zero.
Fig 4.3: Single line diagram of the wind park Rejsby Hede and
the 60 kV network. Only circuit breakers that have been important for
the design of the ASVC are shown.
Figure 4.4 shows the equivalent circuit for harmonic distortion calculation at
the PCC. The distortion depends on the impedances ZT representing the trans-
former and series connected reactors and ZN representing the system. Normally
the system impedance ZN can vary in a certain range due to various loading
and connecting or disconnecting of capacitors. The following cases had to be
investigated:
- The number of turbines in operation may vary between 0 to 40.
- Each turbine can operate either with its small generator (150 kW) or with
its large one (600 kW).
- The reactive power compensation capacitors installed locally at each tur-

bine are switched in 50 kVAr steps. Small generators are compensated by 50
kVAr, large generators are compensated by 150 kVAr. The total 150 kVAr are
switched on by two intermediate steps of 50 kVAr and 100 kVAr lasting 6
seconds each.
- The loads connected to the 60 kV system may vary between 0.1 MW and 4.7
MW connected in Hjer and between 0.1 MW and 4.7 MW connected in
Ballum.
- The 23 MVAr capacitor unit in Bredebro might be connected or disconnect-

ed according to the requirements of the 150 kV system.
The harmonic study included variation of load and wind park generation
based on small and large generators with the whole variety of operating
cases.
Fig 4.4: Equivalent circuit to calculate harmonic distortions at the PCC.
4.2.2 Results of the Harmonic Study

To calculate harmonic conditions the system as shown in Figure 4.3 was mod-
elled in a harmonic load flow program. Static components of the system like
transformers, transmission lines or capacitors were modelled by RX-branches.
For the generators the negative sequence equivalent impedance was used. To a
good approximation the R/X ratio of all inductive components can be assumed
constant over the frequency range considered, e.g. the damping increases in
direct proportion to the frequency.
The two GTO-converters are modelled by an equivalent voltage source.

Realistic values for the characteristic and non-characteristic harmonic voltages
generated were obtained from computer simulations using the Alternative
Transients Program (ATP).
In a first approach to an optimised ASVC design regarding harmonics, the two

GTO-converters were connected via separate transformers (one Y/Y, one Y/D) to
each of the 15 kV busbars. The GTOs were controlled according to Fundamen-
tal Frequency Modulation (FFM). Different areas of interest were recognised:
The total harmonic voltage distortion in the 60 kV network remained below

the specified value but single voltage distortion values were critical at the 11th
and 13th harmonic order. The harmonic loading of the fixed compensation
became critical when only a few wind turbines were in operation at the lower
rated output.
The harmonic problems were resolved by:
- Developing a pulse pattern which eliminates the 11th and 13th harmonic
voltages to the largest extent possible and minimises the 5th and 7th harmon-
ics using three pulse Selective Harmonic Elimination Modulation (SHEM).
Three pulse modulation means that each GTO is switched on and off three
times a cycle of the fundamental. A comparison of the harmonic voltage con-
tent using FFM and SHEM is shown in Figures 4.5 and 4.6. Figure 4.5 shows
the harmonic voltage content of the converter voltage at FFM. Figure 4.6 shows
the harmonic voltage content using the optimised pulse pattern described
above. The black bars show the theoretical values the white ones include
effects due to unbalances, DC-voltage ripple and switching tolerances obtained
by ATP simulations.
- Increasing the reactance between converter and 15 kV bus which changed

critical resonance conditions. The transformer impedance was approximately
doubled in respect to the optimum impedance calculated under reactive power
rating aspects.
- Connecting the GTO-converters via a three winding transformer to one 15 kV

bus. One converter is connected to a star the other one to a delta winding. In
the case of parallel operation such an arrangement reduces harmonic distor-
tion on the primary winding of a three winding transformer i.e. reducing six
pulse harmonic orders 5, 7, 17, 19, etc.
Extensive harmonic load flow calculations verified the above measures. The
results are presented in Figure 4.7. It shows the harmonic voltage distortion at
the 15 kV busbar, where the 12 pulse ASVC is connected in the final design
stage (see Figure 4.3).
Fig 4.5: Harmonic voltage content of converter voltage in % using FFM.
Fig 4.6: Harmonic voltage content of converter voltage in % using SHEM.

Fig 4.7: Harmonic voltage distortion of the 15 kV busbar (PCC)

obtained during the design steps as described.
4.3 Load Flow Study
4.3.1 General
As a result of the harmonic study both converters have to be connected to one
common 15 kV busbar forming a 12 pulse system.
Since the ASVC has to compensate the reactive power demand of the whole
wind park, the reactive power for the second 15 kV side has to be transferred
over the three winding wind park transformer (60/15/15 kV). Due to the leak-
age impedance of the transformer this causes a voltage drop from the 15 kV
bus where the ASVC is connected to the other 15 kV bus. In order to determine
the voltage conditions load flow calculations for the system shown in Figure
4.8 were carried out.
The ASVC is represented by a current source feeding in pure reactive current

(phase shifted by 90 el. with respect to the node voltage at WP2). The magni-
tude of the reactive current depends on the reactive power demand of the wind
park, measured at both 15 kV feeders of the three winding wind park trans-
former (60/15/15 kV).
For continuous operation the reactive current generated by the ASVC is limited
to the nominal value.
4.3.2 Results of Load Flow Study

Load flow calculations were carried out for machine numbers varying from 0
to 18 at WP1 (nWP1, see Figure 4.8) and 0 to 22 at WP2 (nWP2) respectively. Table
4.1 summarises characteristic cases including maximum and minimum voltage
values detected during the load flow calculations. The Table also shows the
Fig 4.8: General circuit diagram for fundamental load flow calculations.
active and reactive power flow and the voltage at the other points of interest.
With the exception of full load operation of the whole wind park (case 3), the
reactive power demand of the wind generators is fully compensated by the
reactive power output of the ASVC and the local phase shifting capacitors. The
remaining reactive power demand of the wind park as seen from 60 kV side in
Rejsby Hede (QWPK3) is determined by the reactive power consumption of the
60/15kV transformer. For case 3 the ASVC cannot cover the complete demand
for compensation and reactive power has to be taken from the grid system.
However, the cos at the 60 kV busbar is always very close to 1 (0.99).
The reactive power demand at WP1 is supplied by WP2, except for case 3 as
discussed above.
The voltage drop between WP2 and WP1 depends on the number of generators
connected to WP1. It is caused mainly by the transmission of reactive power.
case nM1 nM2 vWP1 vWP2 vK3 PWP1 QWP1 PWP2 QWP2 PWPK3 QWPK3 vM1 vM2 iASVC
[pu] [pu] [pu] MW Mvar MW Mvar MW Mvar [pu] [pu] [pu]
1 1 22 1.0 1.006 1.0 -0.6 0.218 -13.2 -0.218 -13.7 1.08 0.99 0.99 0.62
2 18 1 0.968 1.032 1.0 -10.8 4.063 -0.6 -4.063 -11.36 1.036 0.96 1.02 0.52
3 18 22 0.969 1.027 0.998 -10.8 4.058 -13.2 -3.547 -23.9 2.209 0.96 1.02 1.0
4 18 0 0.968 1.031 1.0 -10.8 4.064 0 -4.064 -10.77 1.05 0.959
5 18 11 0.97 1.033 1.0 -10.8 4.055 -6.6 -4.055 -17.35 1.125 0.961 1.026 0.77
6 9 11 0.988 1.018 1.002 -5.4 1.986 -6.6 -1.986 -11.98 0.427 0.98 1.011 0.53
nM1(,2) = number of large wind generators connected
vWP1(,2) = absolute value of node voltage at the 15kV busbar

vK3 = absolute value of node voltage at the 60kV feeder
P, QWP1(,2) = sum of the active / reactive power of one 15 kV busbar branch
P, QWPK3 = sum of the active / reactive power of the whole wind farm seen at the 60kV side
VM1(,2) = terminal voltage of the wind generators

63
Table 4.1: Results of load flow calculations.

The maximum voltage difference amounts to 6.4% (case 2). The highest voltage
occurs at WP2 (vWP2, case 5) with 1.033 pu (15.5 kV), the lowest voltage at WP1
(vWP1, case 2) with 0.968 pu (14.52 kV).
The voltages at the machines (vM1,2) vary between -1 % to -4.1 % at M1 and
-1 % to 2.6 % at M2.
This results in acceptable voltage conditions in the wind park.
4.4 Transient Study
4.4.1 General
Transient system conditions are a further important criteria for the design of
the ASVC components. One of the worst cases to be considered for Rejsby Hede
is islanding. Islanding means that the wind park becomes disconnected from
the transmission system. This can be caused by opening either the circuit
breaker connecting the 150 kV transmission system at Bredebro substation or
opening the circuit breaker connecting the wind park at Rejsby Hede (see
Figure 4.3).
Disconnecting the wind park means a sudden load rejection and change of the
reactive power flow. Due to the drop of active power the slip of the induction
machines becomes smaller. This results in higher system frequency immediately
after the breaker has opened. Furthermore the phase angle of the system volt-
age changes. The reaction of the system regarding the change of the reactive
power flow depends on the equivalent impedance of the remaining network as
seen from the generator terminals. An inductive characteristic will result in de-
excitation of the generators. In this case the terminal voltage will decrease, the
generator rotors will accelerate and the system frequency will increase. The
generator control will disconnect and brake the generators as soon as the cor-
responding setpoints are reached. In cases where the equivalent impedance is
capacitive, the machines will become self-excited. This will result in a rapid
voltage rise. Values of twice the nominal voltage could be reached within one
or two cycles of the system frequency if no effective actions to limit the voltage
are carried out. Effective actions would be disconnecting capacitors and
dynamic reactive power compensation. Besides the rapid voltage rise the gen-
erator rotational speed will slow down and the system frequency will decrease.
To examine the system behaviour under islanding conditions with and without
the ASVC, a transient study was carried out.
4.4.2 Results of the Transient Study

In an early stage of the project, the transient conditions were investigated by
modelling the wind park and the surrounding high voltage system in the
Alternative Transients Program (ATP). The results of these simulations formed
the basis for a first approach to the converter control design. As soon as the
real control cubicles of the converter became available the development of the
control algorithms and design of the control hardware continued in the
Transient Network Analyser (TNA).
ATP allowed for a very detailed representation of the wind park and the 60 kV
system. For the TNA-studies this model had to be simplified to reduce efforts to
an economical optimum while providing all the functionality necessary for
design and test of the control functions.
After the controller design was finished the ATP model was updated and the
response of the ASVC to reference value steps and system transients was veri-
fied. Hence, the ATP model includes all control functions relevant for transient
simulations up to a time constant of a few seconds in a manner corresponding
to a digital control system.
The following cases had to be investigated:
- Opening the circuit breaker at the 60 kV side of the three winding trans-
former in Rejsby Hede
- Opening the circuit breaker to the 150 kV transmission system in Bredebro.

The capacitor bank at Bredebro was taken into account. Large and small load
respectively connected at Ballum and Hjer were investigated separately.
The simulations have shown that there are no critical conditions to be expect-
ed in the case the wind park becomes disconnected from the 60 kV system
directly in Rejsby Hede. The voltage or frequency deviations respectively will
lead to braking and disconnecting the wind generators by the generator con-
trol if the corresponding setpoints are reached.
The behaviour of the island system after opening the circuit breaker at Brede-
bro mainly depends on whether the 23 MVAr capacitor bank is connected or
not.
If the capacitor bank is connected, unacceptable overvoltages under islanding

conditions will occur caused by self-excitation of the induction generators.
One of the important results of the ATP simulations is that the capacitor bank
has to be disconnected as fast as possible whenever the breaker to the 150 kV
system receives a trip signal. Therefore a fast relay that provides co-ordinated
tripping of both circuit breakers within 10 ms was installed at Bredebro.
Furthermore, it was seen that the ASVC would not react fast enough to limit
the rising system voltage within the first couple of milliseconds after islanding.
After the capacitor bank is disconnected the system behaves like it does with-
out the 23 MVAr capacitor bank: After islanding the system voltage decreases
and the system frequency increases. The load in Ballum and Hjer leads to
faster de-excitation of the generators.
The studies have shown, that islanding causes the ASVC to trip under worst
case conditions, where the remaining load in the system is small compared to
the generation of the wind park and the ASVC is at its maximum capacitive
output. It stays connected in cases, when the generation is low or the remain-
ing system is heavily loaded.
However, the transient simulations based on an equivalent circuit incorporat-

ing the fast relay for co-ordinated tripping of the capacitor bank in Bredebro
have not shown hazardous system conditions due to islanding. In all cases the
ASVC will be reliably protected by its protection schemes.
4.5 Basic Data of ASVC Design
4.5.1 General
As discussed in the sections above the basic design of the ASVC was deter-
mined by the results of the harmonic study. Keeping the harmonic distortions
in the system below specified limits requires:
- a special switching pattern of the converter (eliminating 11th and 13th

harmonics)
- a transformer impedance that is about twice the value calculated under

optimum reactive power rating aspects
- 12 pulse arrangement of the converters.
These measures have an impact on the wave shape of the converter currents
due to changes in their harmonic spectrum. Moreover, increasing the trans-
former impedance results in an increasing reactive power consumption of the
transformer. Thus, the capacitive reactive power output of the converters had to
be increased accordingly. Connecting both converters together to one of the 15
kV busbars has changed the voltage conditions in the wind park as calculated
in the load flow study. The voltage operating range of the ASVC had to be
extended towards higher values to cover all possible conditions of generation
in the wind park and to avoid interaction with the tap changer control of the
wind park transformer.
The results of the transient study are taken into account especially to define
the transformer saturation characteristics.
The following paragraphs show the final design of the ASVC power compo-
nents, that meets the requirements mentioned above.
4.5.2 Power Rating of the ASVC

The power rating of the ASVC for continuous operation at 1.0 pu is 2 x 4.0
MVAr capacitive and 2 x 4.0 MVAr inductive.
4.5.3 Network Conditions
Voltage
The voltage conditions at the 15 kV buses of the wind park are determined by
the voltage regulating relay of the 60 to 15 kV step down transformer in Rejsby
Hede. The design of the ASVC was based on the following assumptions:
- The relay controls the transformer tap changer to regulate the average volt-
age of both 15 kV systems to 1.01 pu.
- The relay will not change the tap changer position within a tolerance mar-
gin of 0.013 pu.
- The operating time is set to 30 sec.
For the load flow calculations described above a system voltage of 1.0 pu at the
Rejsby Hede 60 kV bus was assumed. This results in the following voltages at
the 15 kV side where the ASVC is connected to
Maximum continuous voltage: 15.5 kV (1.033 pu) (see Table 4.1, case 5)
Minimum continuous voltage: 15.0 kV (1.0 pu) (see Table 4.1, case 1)
Together with the settings at the voltage regulating relay the operating range
of the ASVC can be defined as follows
Nominal system voltage 15.0 kV (1.0 pu)

Maximum continuous voltage 15.85 kV (1.01+0.013+0.033=1.056 pu)
Minimum continuous voltage 15.0 kV (1.01-0.013=0.997 pu)
System Frequency
Nominal frequency 50.0 Hz
4.5.4 V/I Characteristic and Operating Points

The V/I characteristic of the ASVC as seen on HV side is shown in Figure 4.9.
Figure 4.10 shows the V/I characteristic as seen at the converter terminals.
Characteristic operating points are annotated. The corresponding electrical
data for one converter branch are given in Table 4.2, Figure 4.1 shows the mea-
suring points.
Fig 4.9: V/I characteristic as seen from HV-side.
Fig 4.10: V/I characteristic as seen from LV-side.

The ASVC provides VAr-control for all operating points on the area labelled
Continuous operation. This area is limited by the maximum converter cur-
rent, which is 770 A both for inductive and capacitive operation and by a sys-
tem voltage of 1.07 pu. Two control loops are incorporated to keep the operat-
ing point in the area for continuous operation: The current control limits the
converter current to 770 A, the voltage control limits within the power rating
of the ASVC the system voltage to 1.07 pu.
The setpoint chosen for voltage control (1.07 pu) is to prevent interference of
the voltage regulating relay of the 60 to 15 kV transformer and the ASVC by
means of voltage control.
The operating points given in Table 4.2 are now discussed in detail.
Operating points 1 and 2 show the voltages and currents at nominal inductive
or capacitive reactive power output at 1.0 pu system voltage.
Operating point 3 is the capacitive design point.
Operating point 4 marks ASVC operation at converter output of 4.8 MVA and
maximum average capacitor voltage of 5.0 kV.
Operating point 5 gives the maximum short time capacitive overload which is
restricted to 10 s (Starting from 0 MVAr pre-load). The average capacitor volt-
age is 5.0 kV. The converter current is limited by a maximum instantaneous
value of 2.7 kA. This overload range allows the ASVC control to reduce tempo-
rary over-currents caused by system transients to currents allowed during con-
tinuous operation. Instantaneous currents above 2.7 kA lead to pulse blocking
triggered by the converter protection.
Operating point 6 is based on maximum average capacitor voltage (5.0 kV) and
maximum converter rms current of 770 A. The converter output is 4.93 MVA.
Operating point 7 marks the inductive design point, i.e. maximum continuous
reactive power output.
Operating point 8 gives the maximum short time inductive overload restricted
to 3.0 sec (Starting from 0 MVAr pre-load). The converter current is limited by
a maximum instantaneous value of 2.7 kA. This overload range allows the SVC
control to reduce temporary over-currents caused by system transients to cur-
rents allowed during continuous operation. Instantaneous currents above 2.7
kA lead to pulse blocking triggered by the converter protection.
70
t (s) OP V1 Q1 V2 Q2 VCONV QCONV Vdc ICONV Remarks

(pu) (Mvar) (pu) (Mvar) (pu) (Mvar) (kV) (A)
1 1.0 -4.0 1.089 -4.36 1.157 -4.63 4.727 763.8 nominal reactive power capacitive
2 1.0 4.0 0.911 3.64 0.843 3.37 3.446 -763.8 nominal reactive power inductive
3 1.04 -4.16 1.129 -4.52 1.197 -4.79 4.89 763.8 capacitive design point.
4 1.07 -4.2 1.157 -4.54 1.224 -4.8 5.0 749.1
10.0 5 0.95 -6.64 1.105 -7.73 1.224 -8.56 5.0 1336.0 short time overload capacitive
6 1.066 -4.30 1.155 -4.66 1.224 -4.93 5.0 770.0
7 1.07 4.32 0.980 3.95 0.912 3.68 3.727 -770 inductive design point
3.0 8 1.07 7.83 0.907 6.64 0.783 5.73 3.201 -1397.3 short time oveload inductive
t time allowed for operation

OP number of operating point
V1 Voltage at HV-side (node WP2)

Q1 Reactive power output of one converter branch as seen from HV-side (node WP2)
V2 Voltage at transformer LV-side
Q2 Reactive power of one converter branch as seen from transformer LV-side
VCONV Voltage at converter terminals
QCONV Reactive power of one converter as seen from converter terminals
Vdc DC capacitor voltage of one converter (both capacitors connected in series)
Iconv Current flowing into one converter
Table 4.2: Operating points referring to one converter branch.

4.5.5 ASVC Components

The design requirements for the ASVC power components as shown in Figure
4.1 are given in detail in the following chapters.
- Converter data
Type of converter: Air cooled three level converter
AC Side
Rated voltage: 3.7 kV
Rated current: 763.8 A
Rated power: 4.9 MVAr
Terminal voltage ratio (Three pulse
modulation): 0.949
Switching angles: a1=12.2
a2=19.5
a1=23.9
Overload: according to Table 4.2, operating points 5 and 8
DC Side
Capacitors: 2 x 5 mF 10 %
Rated voltage: 2 x DC 2.5 kV
Max. voltage (instantaneous value) 2 x 2.9 kV
Rated current (rms value): AC 800 A
- Transformer data
Type of transformer: 3-phase, oil-insulated
No. of windings: 1 primary, 2 secondary
Insulation levels
Highest System Voltage, prim.: 24 kV
Lightning Impulse Withstand Voltage, prim.: 125 kV (Cable Connection Box)
Power Frequency Withstand Voltage, prim.: 50 kV
Highest System Voltage, secondary: 7.2 kV
Power Frequency Withstand Voltage second.: 20 kV
Nominal Power 2 x 4 MVAr at 1.0 pu system
voltage
Nominal Voltage, primary side 15 kV
Nominal Voltage, secondary side 3023 V
Nominal Frequency 50 Hz
Leakage Impedance 8.9 % (on 8 MVA basis)
Vector Group Ynyd5
Saturation knee-point above 1.3 pu
Maximum fundamental current for each
low voltage side (continuous) I(1)rms: 771,0 A
- Reactor data
Type of reactors: three phase iron-cored reactors
Inductance 0.471 mH
Tolerance on inductance 2%
Maximum total current (continuous) Irms: 771.0 A
Fig 4.11: Operating losses of the ASVC.
4.5.6 ASVC Losses

Losses depend on the operating point of the ASVC. Figure 4.11 shows the fun-
damental losses of the main components of the ASVC for the continuous ope-
rating range from 8 MVAr capacitive to 8 MVAr inductive referring to 1.0 pu
system voltage at the PCC.
73
Chapter 5
Detailed Design of the ASVC

74 Detailed Design of the ASVC
Klaus Bergmann
Ralf Stber
SIEMENS, Power Transmission and Distribution
Erlangen
Germany
Detailed Design of the ASVC 75
5.1 Component Arrangement
The ASVC is connected to one busbar of the wind farm. Potential transformers,
current transformers and the 15 kV circuit breaker are located in the switch-
gear building at Rejsby Hede Substation. From there 15 kV single core cables
lead to the three winding converter transformer which is positioned outdoors,
at one end of the ASVC building (Figure 5.1).
Figure 5.1: Component Arrangement.

To double the transformer short circuit impedance, one 3-phase iron core reac-
tor per secondary winding is connected via 3.6 kV cables. The reactors are
placed in a separate room for safety reasons, since their terminals are open.
Finally, the two air-cooled GTO-converter cubicles are connected to the power
circuit, again via 3.6 kV cables. They are also located in their own room to
reduce the noise of the cooling fans and to provide the necessary air flow
through the room for cooling purposes.
The ASVC Control Cubicle, the Uninterruptable Power Supply (UPS) System
and the AC-Distribution are located in the control room. Furthermore, the
measurement equipment developed by DTU was installed in this room.
To allow for remote control and monitoring, a selection of signals and alarms is
transferred from the Control Cubicle to the Remote Terminal Unit (RTU),
which is placed in the Rejsby Hede switchgear building and provides the par-
allel interface to the EASV signal transmission system.
The electrical interconnections are shown in the Single Line Diagram,

Figure 5.2.
Figure 5.2: Single Line Diagram.

5.2 GTO-Converter Cubicles
One converter cubicle, as shown in Figure 5.3, contains three vertically

arranged GTO-thyristor stacks with gate units and snubber circuits, the two
DC-link capacitors with short circuit device and pre-charge circuit, converter
control system, power supplies and grounding facilities. Since the converters
are air-cooled, two top mounted fans per cubicle have to provide the necessary
air flow. Cooling air is sucked in at the bottom front side of the cubicles and
blown out on top.
Figure 5.3: GTO-Converter Cubicle.

The converter used for the wind farm application is a Three-Level Converter
with a maximum continuous reactive power output at the cubicle terminals of
4.6 MVAr at 770 Amps.
Further technical details:

- Type of operation: three-pulse operation
- Converter connection voltage: 3 kV, 50 Hz
- GTO-thyristor:
reverse conducting (RCGTO)
electrical data: 3000 A / 4500 V
size (contact area / overall): 80 mm / 120 mm
- Size of one converter cubicle: L: 2100mm / H: 2800 mm / W: 1200mm
- Weight of one cubicle: 2700 kg
Further data is given in Chapter 4, Basic Data of ASVC Design.
Fast overcurrent and overvoltage protection functions are located in the con-
verter cubicle including the necessary current and voltage transformers. The
current transformers are also used for DC-current monitoring and control.
Furthermore, each converter cubicle contains a thyristor switched short circuit

device for the DC-capacitors, which is triggered by the fast overvoltage protec-
tion and absorbs the capacitor energy in a resistor.
The converter cubicles are equipped with the thyristor control set, which pro-
vides the interface between the converter and the ASVC controller. The control
set generates control pulses for the gate units which then produce the actual
firing pulses for the GTO-thyristors. There is one gate unit for each thyristor.
Before initial start-off, the DC-link capacitors have to be charged via the 230
V auxiliary supply system. During operation the capacitor voltages vary accord-
ing to the operating conditions.
5.3 Control Cubicle
The ASVC controller is based on SIMADYN D, the Siemens multiprocessor

digital control system. The system is fed with the actual voltage and current
values of the wind farm busbars. It contains all functions and software to
operate and protect the ASVC within the required operating range. A more
detailed description is given in the later section Control System Design and
Implementation.
Settings and parameters can be implemented and changed via the door mount-
ed Operator Interface OP2 which can also display several parameters, like bus-
bar voltages, reactive power generated, alarm messages, status of plant, etc.
Furthermore, there is a mimic panel at the front door of the control cubical
which indicates the status conditions of the 15 kV circuit breaker (CB), discon-
nector and ground switch (Figure 5.4). The 15 kV CB is switched by the ASVC
controller automatically, according to operational needs.
The controller also collects alarms and trip signals which are then shown on
the alarm panel at the cubicle front door. The detailed alarm statements are
available on the OP2 display or in form of a printout.
Various push buttons and current / voltage indicators provide the necessary
facilities for the start-up and close-down sequence of the ASVC.
Figure 5.4: Mimic Panel.
5.4 Protection of Plant
This paragraph describes the conventional protection system, i.e. protection

relays. There is one overcurrent relay installed in the switchgear building
which provides Overcurrent/Time and Instantaneous Overcurrent Protection
for the feeder bay and the 15 kV cables.
From the same set of current transformers one signal (3-phase) is taken for the
differential protection of the converter transformer. The other two current val-
ues (3-phase) for the differential protection relay are taken from bushing cur-
rent transformers at the secondary windings of the converter transformer
(please refer to attached Figure 5.5). Additionally, the differential protection
provides the back-up function for the overcurrent /time protection.
Furthermore, this relay contains an overload function for transformer protec-
tion which uses an integrated current technique.
The protection signals generated by the relays initiate the trip of the feeder
vacuum circuit breaker. The signals are also sent to the ASVC controller to ini-
tiate alarms. In addition to the conventional protection, the ASVC controller
can send trip signals to the circuit breaker, generated by overcurrents, over-
voltages or ground faults in the converter cubicles.
Figure 5.5: Protection Scheme.

5.5 Transformer
The natural cooled converter transformer is of double stack, three winding

design, with star-star-delta connection. The rated voltage at the primary wind-
ing is 15 kV, and at the secondary windings is 3 kV. The short circuit impedance
amounts to 8%. The transformer is designed for cable connection at the prima-
ry and secondary sides.
It weights 33 tons with an overall height / width / length of 3.8 / 3.6 / 4.5
meters.
Each phase of the secondary windings is equipped with one bushing current
transformer for connection of the differential protection relay.
The transformer protection consists of Oil Temperature and Oil Level

Monitoring and of Rapid Pressure Rise Detection. Furthermore a Pressure
Relief Device protects the sealed tank transformer against overpressure.
5.6 Iron Core Reactors
The first task of the iron core reactors is to approximately double the short cir-
cuit impedance of the converter transformer from 8% to 16%. To design the
transformer itself for 16% would have been very expensive and technically dif-
ficult.
The second task is fine tuning of the transformer plus reactor inductance with-
in a tolerance of 2%. This is achieved by manipulating the gap in the iron core
after having the measured values of the transformer inductance available.
The windings of the three-phase reactors are vacuum-pressure-impregnated
using Polyester resin.
5.7 Auxiliary Supply
The auxiliary transformer located in the 15 kV switchgear building also sup-

plies the ASVC equipment.
The auxiliary supply cable is connected into the AC distribution cubicle from
where the loads are fed. All loads are protected with miniature circuit breakers
(MCBs).
In case of loss of the AC-Supply the ASVC is switched-off. Then the control
and protection system will be powered from an Uninterruptable Power Supply
(UPS) for two hours, so that status and alarm signals can be transmitted to the
area control centre and a remote controlled re-start of the ASVC is possible
when the auxiliary supply voltage returns.
5.8 Control System Design and Implementation
5.8.1 Introduction
The following section gives a description the closed loop control of the Rejsby
Hede ASVC for Eltra/Denmark.
The control is based on the SIMADYN D digital control system. This control
system is a multiprocessor configuration that provides a sampling time of 1 ms.
For special applications, e.g. filtering, digital signal processors with shorter
sampling times are used. The digital control system SIMADYN D covers a
wide range of sophisticated power engineering applications, e.g. AC and DC
motor drives, HVDC, SVC and ASVC.
It offers the following main advantages:
- User-configurable and user-friendly operation

- Flexible configuration of control functions
- Expansible to suit growing user requirements
- Self diagnostic routines
- Remote diagnostics
- Error-free self documentation directly in block diagrams
- Worldwide support
- Normal production line equipment
5.8.2 Interfaces
The ASVC control has several interfaces for measurements and data exchange.
Figure 5.6 gives an overview of the layout and the main components of the
ASVC in conjunction with the interfacing of the control.
System Measurements
The controller takes voltage and current measurements from both 15 kV bus-
bars. The conversion of the CT and VT measurements into values suitable for
the controller is carried out in a number of steps. Beyond the terminals and
miniature circuit breakers (MCB) overvoltage protection devices and interfer-
ence suppression filters to guard against HF voltage components are fitted. The
signals are then converted by galvanically isolated V/V or I/V converters into
signals at control circuit potential with voltage limitation and interference
suppression. Calibration facilities are provided for each input.
The input interface and signal conditioning is realised with analogue devices.
This adapts the input signals to the subsequent digital and analogue control
circuits.
The following input signals are used:
- 15 kV line to line busbar voltages (VLine_1/2)
- 15 kV busbar currents (ILine_1/2)

60 kV
15 kV 15 kV
Busbar 1 Busbar 2
3 kV
IC NV_1 IC NV_2
C C
C V Line_1 C
U U
V C _1 V C _2
hardwired
V Line_1 seriel link
ILine_1
ASVC V Line_2
Printer Operating Panel Control ILine_2
PC
Remote Terminal
83
Figure 5.6: ASVC Layout.

Converter Measurements and Converter Control Signals
The converter currents (ICNV_1/2) and the DC-link voltages (VC_1/2) are measured
with electronic transducers and are digital processed in the converter control
unit (CCU), located in the DC-link cubicle. Additional the voltage measure-
ment of busbar 1 (Vline_1) and binary signals are transferred to the CCU. All
measured signals are digitally transferred to the controller (via a high speed
optical bus).
Converter control signals are digitally transferred via the same bus from the
ASVC control to the CCU. These signals are processed to generate gate signals
for Gate Drive Units (GDU).
Operating Panel
For operating purposes, service and diagnostics an operating panel is connect-

ed with a serial link to the control system. An optional printer can be connect-
ed to the operation panel.
Remote Terminal and Switchgear Control
Under normal condition the ASVC will be controlled from the control centre at
Skaerbaek. The data exchange for ASVC remote control and the swichgear
control are handled with an ET200 interface which is connected via an opti-
cal serial link to the ASVC control.
The trip circuits of the switchgear control are hardwired.
PC
For commissioning, service and diagnostics a standard PC can be connected to

the control system. This can be done directly in the local control room or from
remote using the remote diagnostic tool (Telemaster) and a standard phone
line.
5.8.3 Control Scheme

The control scheme is composed of several main function blocks shown in
Figure 5.7.
Double framed function blocks are indicate, that this function exists twice,
once for each converter.
Actual Value Sensing
This function processes the input quantities of the ASVC control and provides
calculated actual values for the subsequent space vector based control. The
main steps are:
- A/D conversion of the measured analogue values
- Transformation of 3-phase measurements into space vector quantities using
the rotating orthogonal a,b coordinate system
Capacitor
dC B
Balance
Control
VC
V Line |I C NV | Limitation
|V Line| Controls
IC NV Actual
Qmax Q min
QLine1,2 dQ d j
Value Converter Qref C NV
VC Q-Control Coordi-
Coordination nation
ILine Sensing Q act
IC NV dIDC
IDC
Control
Trigger Set
j PLL
V Line
PLL f
85
Figure 5.7: ASCV Control Scheme.

- Calculation of the actual instanteneous active and reacvtive power of both

busbars
- Calculation of the actual instanteneous reactive power of both converters
- Calculation of the phasor magnitude of the converter current and the line
voltage
- Single phase calculation of the magnitude of the 15 kV line to line voltages
- Filtering of measured and calculated values.
Converter Coordination
The converter coordination calculates the actual reactive power demand of the
wind park (QLine) and determines the reactive power reference value (Qref) for
both converters. The reference value is limited by the maximum capacitive
(Qmax) and inductive (Qmin) reactive power, which is dynamically determined by
the limitation control. The maximum rate of change of the reference value is
limited to 100 kVAr/ms. If, under extreme conditions, fast control action is
required the reference value will be modified instantaneously.
Q-Control
The Q-control determines the required reactive power output of the ASVC.
Input quantities to this control unit are the reference value Qref and the actual
value QCNV of the converter instantaneous reactive power. The difference
between both values is controlled to zero by a proportional/integral PI con-
troller. The output of the PI-controller defines the phase delta dQ between the
fundamental network voltage space vector and the fundamental converter volt-
age space vector.
Under steady state conditions this value is close to zero. Changing this value
causes an active power exchange of the converter with the power system and
an increase or decrease of the dc capacitor voltage. The capacitor voltage
determines the magnitude of the converter ac voltage and therefore the reac-
tive power exchange with the system (Figure 5.8).
To achieve a constant and operating point-independent step response the non-
linear control loop is linearized. This is covered with a linearization character-
istic forming the output signal dQ,lin.
The step response as shown in Figure 5.9 is approximately 20 ms.
Phased Locked Loop (PLL)
The PLL circuit calculates the phase and frequency information of the funda-
mental, positive sequence component of the 15 kV busbar voltage Vline_1.
Input quantities are the a,b-components of the 15 kV busbar voltage Vline_1.
Under steady state conditions the speed of response is very slow to get well fil-
tered and noise-free output signals. At system transients the PLL controller is
speeded up to achieve a fast resynchronization.
P P
ZS X Tr ICNV
VX V C NV VX V Line
Power VX
E
~ System
Equivalent
V Line
~ V CNV V Line V CNV
ICNV IC NV
-jQ -jQ
Capacitive Inductive
Operation Operation
Figure 5.8: ASCV Space Vectors.
Figure 5.9: Step Response Q-Control.
d-Coordination
The -coordination forms the angle jCNV of the required converter voltage space
vector for the trigger set by combining various input angles. These are:
- the actual power system phase angle jPLL given by the PLL
- phase angle difference dQ,lin between the converter and the power system
voltage
auxiliary modulation angles from capacitor balance control (150 Hz modula-

tion) and dc-current control (50 Hz modulation).
Trigger Set
Depending on the required converter voltage phase angle jCNV and the actual
system frequency f the switching signals for the GTOs are generated in the
trigger set. They are derived using an optimised pulse pattern with 3 pulse
modulation (150 Hz switching frequency), which is stored in a lockup table.
The off-line calculation of the pulse pattern has taken the basic design of the
ASVC as well as the harmonic requirements into consideration.
Since the trigger set operates with this fixed pulse pattern, the modulation
depth (ratio between dc capacitor voltage and the magnitude of the converter
voltage) is constant.
Limitation Controls
The maximum capacitive and inductive MVAr limits Qmax and Qmin are deter-
mined by dynamic limitation controls. These controls take the line voltage as
well as the internal dc capacitor voltage VC and the converter currents ICNV into
consideration to avoid overload situations of the converters and of the power
system.
VLine - Limitation
The VLine - limitation controls the 15 kV line voltage of busbar 1 to suitable val-
ues in case of system faults or extreme system conditions. Two levels are imple-
mented:
- VLine > 1.07 pu: Decrease of the capacitive MVAr output (or increase of the
inductive) to limit the system voltage to the threshold with a response time of
100 ms -200ms. The threshold is coordinated with the slow acting tap changer
control of the local 60 KV/15 kV transformer and should also be coordinated
with already existing overvoltage alarm levels.
- VLine > 1.20 pu: Instanteneously switchover of the Qref value to full inductive
operation (-8 MVAr) with a subsequent slow increase to the initial value.
Reduction of the first threshold level to 1.0 pu.
Capacitor Voltage Limitation
The capacitor voltage limitation protects the GTO-thyristors and the dc capa-
citors against excessive voltage stresses. Because the ASVC is operated with a
fixed pulse pattern, the dc capacitor voltages VC depend on the magnitude of
the system voltage and the amount of reactive power being exchanged.
Capacitive operation increases and inductive operation decreases the voltage
compared with no load conditions. Two levels are implemented:
- VC > 2.5 kV: The maximum value of the four capacitor voltages is controlled
to the 2.5 kV threshold by reducing the capacitive MVAr output (Qref reduc-
tion). Filtered VC measurements are used to suppress the harmonic ripple gen-
erated by the switching operation of the converter. The response time is 100 ms
- 200ms.
- VC > 2.7 kV: Instanteneously switchover of the Qref value to no load operation
(0 MVAr) with a subsequent slow increase to the initial value.
ICNV-Limitation
The GTOs can only carry a device dependent maximum current. The ICNV-limi-
tation lowers the amount of reactive power when excessive currents occur. The
ASVC converter can exchange twice the rated power for a few seconds,
depending on the pre-load conditions. Two levels are implemented:
- ICNV > 1pu: The maximum magnitude of the current space vectors of both con-
verters is controlled to the nominal current by reducing the MVAr output (Qref
reduction). The response time is 100 ms - 200ms.
- ICNV > 1.5pu: Instanteneously switchover of the Qref value to no load operation
(0 MVAr) with a subsequent slow increase to the initial value.
Capacitor Balance Control
In the case of 3-level converters, care must be taken that the individual capaci-
tor voltages of every converter are balanced. Otherwise capacitor overvoltage
and current distortion may result. If the voltages differ above a certain limit
(VC > 75 V), the control is activated and switching is modified such, that the
capacitors with the higher voltages inject energy to the network and capacitors
with the lower voltages receive energy at the same time and an equalisation is
thus affected.
For this the control provides an auxiliary modulation signal dCB which is added
to converter voltage phase angle via the d-coordination.
ID-Control
Non-ideal operation of the ASVC converter can cause a DC voltage in the line-
side converter voltages. This in turn causes a DC current and can lead to satu-
ration of the ASVC transformer resulting in highly distorted ASVC currents.
The dc components in the converter currents therefore is detected and con-
trolled to an acceptable minimum level. The control determines the modulation
of the phase angle by adding the auxiliary signal dIDC via the d-coordination.
Additional a dc current supervision circuit is implemented in the control, trip-
ping the ASVC in the case of unacceptable DC components.
Auxiliary Controls
The auxiliary controls have to detect extreme power system and ASVC condi-
tions and have to re-act rapidly to support the main controls and if possible
to avoid unnecessary tripping of the ASVC.
In the case of tripping, these circuits coordinate the shutdown sequence of the
ASVC due to the requirements (fast or slow shutdown), if possible the main
circuit breaker will be opened at 0 MVAr.
5.8.4 TNA testing

The AC/DC Simulator (TNA) is designed for transient real time AC and DC
system studies, as well as for tests of control and protection equipment for
HVDC, SVC and ASVC systems. The simulator components are high-quality
R-L-C elements combined with microelectronic components, a digital real time
simulator (RTDS) and a powerful digital data acquisition system.
RTDS models can be used both for fully digital real time simulations and in
combination with the analogue simulator as a hybrid simulator extension.
Controlled electronic sources can be used as system infeed. Any voltage, phase
angle or frequency variation including unbalances or harmonic injections can
easily be generated to test control systems even under extreme system condi-
tions. Thus, with a simple network representation, complex system distur-
bances can easily be simulated.
The ASVC representation includes a detailed modelling of all components

together with measuring circuits needed for the closed loop control and addi-
tional signals for the data acquisition system.
Figure 5.10 shows the test set-up for the TNA testing including a digital real-
time simulation of the windfarm.
For the verification of the ASVC performance under transient conditions, addi-
tional power system elements are provided. These are saturable transformers
for the simulation of inrush-effects, controllable variable loads and line models
for generation of higher frequency harmonics.
With variable line impedances tuned to any system fault level it is possible, to
test the dynamic response and the controller stability, even under extreme sys-
tem situations (weak network). By these means, the complete spectrum of sys-
tem transients from 0 Hz (TCR-DC currents causing transformer saturation
and second harmonic voltage) up to 1 kHz for line switching and above 5 kHz
for valve transients including protective firing can be simulated.
Simulator test were carried out to develop and optimise the ASVC control, to
verify the performance of the each individual control loop and of the complete
ASVC.
18 Wind Mills
Scalings: P max=10.8 MW
Qmax=3.8 MVAr
V Line: 2* 3 kV / V
Load
ILine: 120 A / V
Hojer/Ballum
ICNV : 250 A / V I Line_2 2x 4.3 MW/1.1 MVAr
V C : 400 V / V
V Line_2
745 MVA
P,Q 340 MVA
Y
60 kV-Line
Y
15 kV-Line ~
Y
P,Q CB_RH CB_BR
Converter 1 V 60 kV CB_C
I Line_1
V C1_1 I CNV_1 uk=6.46 % V Line_1
C-Bank
V C2_1 uk=8.9 % Bredebro
23 MVAr
Y
GTO-ASVC P,Q
Y
3 kV-Line
-S1 P,Q
Converter 2 CB_SVC
V C1_2 I CNV _2 -S2
V C2_2
uk=6.46 %
GTO-ASVC -5 MVAr
91
Figure 5.10 : TNA Set-up

The following test program was performed:
Verification of Controls
- Actual Value Sensing
- Phase Locked Loop (PLL)
- Reactive Power Control
- Capacitor Balance Control
- IDC Control
Verification of Protection Features

- Voltage and current limitation
- Converter crest-current protection
- Capacitor crest-voltage protection
- AC Under/Overvoltage protection
ON/OFF Sequence
- Pre/Discharging
- On/Off - Control
- Trip/Emergency Off
- Standby Mode
Load Variation at 60kV

- Symmetrical loads
- Unsymmetrical loads
- Energization of external transformers and capacitors
System Fault Behaviour

- Symmetrical faults
- Unsymmetrical faults
- System faults with load rejection (variation of system phase angle)
- System faults with variation of system frequency
Windfarm Load Variations

- Windmill startups
- Variation of power generation
- Windmill shutdown
Windpark Islanding
- Islanding at Rejsby Hede
- Islanding at Bredebro
Figure 5.11 gives an example of the TNA simulations. It shows the fast shut-
down (tripping) of 18 windmills at busbar 2 and the response of the ASVC.
93
Figure 5.11: TNA Simulation, Windmill shut-down.

95
Chapter 6
Installation at Rejsby Hede

96 Installation at Rejsby Hede
Kent H. Sbrink
I/S Eltra
Fjordvejen 1-11
P.O. Box 140
DK-7000, Fredericia
Denmark
Installation at Rejsby Hede 97
6.1 Introduction to the Rejsby Hede Wind Farm Site
The Rejsby Hede wind farm site is situated in the south western part of the
Danish peninsula of Jutland the mainland. It is a good windy site close to the
coast and it is a typical post-glacial flat area with sandy soil a moor which
has been reclaimed. The area has been protected from shifting sand by wind-
breaks of trees and in between the windbreaks the area is cultivated with corn.
Figure 6.1: Wind turbines (Micon 600/150 kW) in the Rejsby Hede wind farm.
It is a very sensitive and peculiar landscape, a part of the marshland which

covers several countries from Denmark in the north over Germany to the
Netherlands in the south. The location of technical equipment, such as a wind
farm and an electrical substation, has a great visual impact on the landscape
and is therefore a very delicate matter. Environmental considerations were
therefore important at the planning stage of the project in order to fit in the
equipment with as little impact on the landscape as possible.
6.2 Civil Works
The ASVC installation can be divided into four main parts; the converter cubi-
cles which contain the power electronic system, the control system, the output
reactors and an output power transformer. The transformer is an outdoor, oil-
cooled transformer, whereas the other equipment is installed indoors, in a
building close to the existing 15 kV substation in the wind farm.
The local authorities required that the building for the ASVC should look like
the existing switch-gear building. The building is therefore a red brickwork
construction with red tile roof, like the existing building, see Figure 6.2.
Figure 6.2: The ASVC building and the ASVC transformer.

Figure 6.3: Control room
The building is separated into three rooms; the control room, the converter
room and the reactor room, Figure 6.9.
The control room contains the control cubicles, the AC distribution system and
an Uninterruptible Power Supply (UPS) unit. The measuring system and an
event recorder are also placed in the control room as well the variable speed
control for ventilation systems in the converter and the reactor room.
The converter room contains the two 4 MVAr ASVC units which are encapsu-
lated in metal clad cubicles on top of which two radial ventilators are placed
for cooling the converter cubicles. Moreover, the converter room is provided
with a ventilation system in order to obtain an ambient room temperature
below 35 deg. C; the heat dissipation from the converters is up to 2 x 40 kW.
The converter room is depicted in Figure 6.4.
Figure 6.4: The converter room.
The reactor room contains the two output reactors, see Figure 6.5. The reactor
room is a restricted area with high voltage (3 kV), the door to the room is prop-
erly marked in accordance with the regulations and only authorised personnel
have admission to the room. The reactor room is also provided with a ventila-
tion system; because the heat dissipation from the reactors is up to 15 kW.
The three rooms are separated by fireproof walls which go up to the roof so
that roof fire cannot spread.
In the building an earthing system of copper bars (3 x 40 mm), for potential

equalisation and protection, goes through all three rooms in the cable trench.
Figure 6.5: The reactor room.
Metal parts of the equipment as well as the reinforcement steel in the building
are connected to the earthing system, which is connected to a grid buried in
the ground surrounding the building.
Besides, the electrical installation for the ASVC building is provided with ordi-
nary electrical wiring for socket outlets, light and heating.
At the end of the converter building, a concrete transformer foundation with

integral oil sump was cast. The oil sump is designed in such a way that it can
contain all the oil of the transformer in case of a leak. In order to damp the
transfer of vibration from the transformer, two planks of oak are placed
between the concrete foundation and the transformer.
The basic consideration for the civil work was:

Regulations
Requirements from the local authorities
Requirements from the distribution company
The visual impact of the building
The audible noise level (max. allowed is 45 dB at the boundaries of the lot)
The physical size of the electrical equipment
Requirements for interconnection of the electrical equipment
The ventilation requirement from the manufacturer
The physical size of the ventilation systems
Interface to the existing installation.
The design of the building was carried out in close co-operation with all rele-
vant parties.
An access road was established in connection with the construction work, and
after it was finished, plants similar to the existing vegetation were placed
around the building.
The building project was carried out on the basis of invitation to submit ten-
ders for three different contracts for the building, the electrical installation
and the ventilation systems respectively. Each of the different contracts was
awarded to the lowest bid.
The construction works started in August 1996 and were finished in January
1997, followed only by some minor follow-ups.
The ASVC building layout is depicted in Figure 6.9, and the site plan is
depicted in Figure 6.10.
6.3 Interface with the Existing Plant
The interface with the existing plant relates to the 15 kV system, the protec-
tion, measurements and to remote control. Besides, it was necessary to co-ordi-
nate the ASVC on one side with the capacitors for no-load compensation in the
wind turbines as well as a 60 kV, 23 MVAr shunt capacitor battery in the
Bredebro 150/60 kV substation from where a radial line supplies the Rejsby
Hede wind farm.
In order to limit the short-circuit level, the 15 kV switch-gear has been split up
on two busbars of which 22 wind turbine generators are connected to bus 1
whereas 18 are connected to bus 2.
6.3.1 Interface with the 15 kV Rejsby Hede Substation

The ASVC is connected to a vacuum circuit breaker on bus 1 (22 wind turbine
generators). The circuit breaker is provided with one shunt release and one
undervoltage release. In order to avoid improper switching of the ASVC circuit

breaker, it is not provided with a local on/off switching mechanism. The circuit
breaker can only be controlled electrically from the ASVC control panel in a
sequence determined by ASVC control system. The interface signals for operat-
ing the circuit breaker are provided with opto couplers with an insulation volt-
age of 3.5 kV.
For control and protection the two 15 kV busbars are provided with current
transformers (CTs) and voltage transformers (VTs) which are interfaced to the
ASVC control system. The ASVC is connected to bus 1, a voltage difference
therefore arises between the two buses, because of reactive power flow from
bus 1 to bus 2 via the 60/15/15 kV three winding infeed transformer. In order
to determine a mean voltage between the two busbar voltages, a special con-
nection between the VTs on bus 1 and bus 2 was implemented. The ASVC con-
trols the busbars voltages in such a way that they are lying on the opposite side
of the mean voltage. This means that if the voltage difference between the two
busbars is 1 kV, one busbar voltage is 0.5 kV above and the other busbar volt-
age is 0.5 kV below the mean value of the voltages between two busbars.
For overcurrent and transformer differential protection the ASVC circuit

breaker is provided with CTs.
The wind turbines are each provided with three 50 kVAr capacitors for no-load
compensation of the dual asynchronous generators (150/600 kW). When the
wind turbines are operated with their 150 kW generators, one capacitor (50
kVAr) is switched in, and when they are operated with their 600 kW genera-
tors, three capacitors (150 kVAr) are switched. The switching of the capacitors
is controlled by the wind generator control system. The capacitors are switched
in steps with a defined time span between each step. Harmonic analyses
revealed that the time span between switching the 50 kVAr capacitors was
critical and that it should be as short as possible. The time span of 200 ms was
found to be an acceptable compromise, which was implemented in the genera-
tor control system. The wind generator control cubicle at the bottom of a wind
turbine tower is depicted in Figure 6.6.
The wind farm is supplied from the Bredebro substation where a shunt capa-
citor battery of 23 MVAr for reactive compensation is provided. An islanding
situation may occur if the transformer breakers the in Bredebro substation
trip. Transient simulations revealed that overvoltage at islanding of the wind
farm is critical if the 23 MVAr shunt battery is switched in. In order to prevent
this critical islanding situation, a fast intertrip relay (10 ms) was installed with
the purpose of tripping the 23 MVAr shunt battery if the Bredebro transformer
trips. The Bredebro substation is depicted in Figure 6.7.
Figure 6.6: The wind generator control cubicles.
6.3.2 ASVC Control

Figure 6.8 shows a photo of the ASVC control cubicle front, which is provided
with push buttons for start and stop commands as well as buttons for confir-
mation or cancellation of the same commands. For shifting between
local/remote ASVC control a key lock switch is also provided. Furthermore,
there is a control panel with a display provided, where set points for automatic
start/stop can be changed and where the set points together with various mea-
sured values and fault messages can be displayed. On the front are meters
installed for showing the output currents and the 15 kV busbar voltages and
the total reactive power from the ASVC.
The ASVC can be operated in manual mode, remote mode and in standby
mode. In local mode the ASVC can only be operated locally, whereas in remote
mode it can only be operated from the regional control centre.
In standby mode the ASVC starts and stops automatically in accordance with
reactive power set points in the control system and the reactive power demand
of the wind farm. It has been experienced that proper set points for automatic
stop are 0.5 MVAr and that for automatic start it is 1.0 MVAr. A dead band of
0.5 MVAr between automatic start and automatic stop seems to be a reason-
able value in order to prevent on/off jitters.
Figure 6.7: 23 MVAr shunt bank in Bredebro substation.

A Remote Terminal Unit (RTU) for interface between the ASVC, the 15 kV
switchgear and the regional control room is installed in the wind farm substa-
tion. The interface signals comprise digital signals for control and check back
and an analogue signal representing the ASVC reactive power output. The vari-
ous types of control and status signals are summarised in Table 6.1.
The ASVC installation was successfully commissioned in August 1997.
Figure 6.8: The ASVC control panel.

Switch gear ASVC ASVC ASVC

control/status status local control remote control
Circuit Local/remote Local/remote

breaker on/of
Earth switch On/off On/off On/off
closed/open
Current relay Standby mode Standby mode Standby mode
indications
Trip/alarm Confirm/cancel Confirm/cancel
+/- 8 MV Ar Output control No output

(0 - 20 mA) by change of control
controller setting
Table 6.1: Summary of control and status signals.
Figure 6.9: The ASVC building layout.

60/15 kV substation
Transformer
Converter building
Figure 6.10: The substation site plan.

109
Chapter 7
Description of
the Measuring Equipment
110 Description of the Measuring Equipment
Jrgen Kaas Pedersen

Technical University of Denmark
Department of Electrical Power Engineering
DK-2800 Lyngby

Department of Electrical Power Engineering
DK-2800 Lyngby
Description of the Measuring Equipment 111
7.1 Introduction
It was the task of the Technical University of Denmark, DTU, to carry out the
measurements. The type of equipment which may be used for these measure-
ments was investigated and two alternatives were considered. The first alterna-
tive was to purchase ready-to-use components capable of producing, from
instantaneous signals, rms values of voltage and current, power and reactive
power to be logged into a PC (Personal Computer). However, this approach was
discarded as it produces a large number of measurements in a very short time
and requires additional sampling of the instantaneous signals in order to cal-
culate higher harmonics.
The second alternative was to sample all instantaneous signals with a suffi-
ciently high sampling rate before feeding them to a PC. The PC would then be
used to generate the desired rms values of voltage and current, power, reactive
power and to calculate higher harmonics. This type of equipment produces
large amounts of data and therefore reducing the amount of data without loos-
ing vital information is required. This option was chosen for the measurements,
using National Instruments equipment; even though it requires more hardware
and software equipment as compared to the first one.
7.2 Network Connection of the Wind Farm
A schematic representation of the wind farm is shown in Figure 7.1. The wind
farm consists of two groups of wind turbines (18 and 22 in number) connected
to different 15 kV busbars. Both busbars are connected to the 60 kV network
via a three phase three winding transformer. The Advanced Static VAr Com-
pensator (ASVC) is connected to the busbar with 22 wind turbines. The mea-
suring transducers are installed on both 15 kV busbars where the wind tur-
bines are connected.
7.2.1 The Signals

Instantaneous values of the networks voltages and currents were measured.
Figure 7.2 shows the connection of all 15 signals (2x3 phase voltage signals and
3x3 phase current signals) used for these measurements. A sampling rate of 256
per 50 Hz period was employed. This led to a tremendous amount of data gen-
erated, i.e. 192,000 values per second. Storing data sets of such size was a
Figure 7.1. Schematic drawing of the windfarm and its associated ASVC.
problem and a reduction in the amount of data was necessary during normal
operation. The full storage of data should only be performed for extreme con-
ditions and for very short periods of time. The following procedure was adopt-
ed for measurement and data storage.
Over each interval of 8 cycles, rms-values of all the 15 signals, the power, and
reactive power of each busbar and the compensators output are calculated.
After each 8 cycles interval, the harmonics and Total Harmonic Distortion
(THD) for every signal are determined. Then, every 2 minutes, the results are
reduced to mean, maximum and minimum value of each signal. This explains
why the time taken by calculations is much greater than that of the sampling
procedure. The sampling and calculations were simultaneous and therefore
only the results from two 8 cycle intervals per second were possible.
To limit the number and size of data, only the harmonics of interest are calcu-
lated. Harmonics which theoretically should not be in the networks voltages
and currents, such as even harmonics and third harmonics components, were
excluded. However, the second and third harmonic, which may be used for con-
trol purposes, were considered. The number of harmonics was initially chosen
to be 40. Following suggestions that harmonics greater than 40 were also
important, the number of harmonics has been increased to 80. Capturing all
this data during a 24 hour period would require 4.8 Mbytes and therefore a
software tool for analysing the data of interest was developed.
7.3 The Hardware
7.3.1 The Connection of the Transducers

Figure 7.2 shows the way in which the 15 transducers were connected. The
voltages are measured using a capacitive voltage divider which gives an alter-
native voltage to the signal which is otherwise be obtained by using a conven-
tional voltage transformer. Measurements carried out on a wound type voltage

transformer indicated a resonant frequency around 1.5 kHz in the transfer
function, which may lead to some difficulties when measuring higher harmon-
ics. Therefore, for a safe operation of the ASVC, the four wound type voltage
transformers were not used although they were already installed. Nine current
transformers were used to measure current signals. A short description of each
component in the measuring equipment is described in the following sections.
Figure 7.2. The connection of the transducers to the two 15 kV two busbars.
Figure 7.3. A photo of the three capacitive voltage dividers mounted

on one of the busbars. The top dark cylinders are high voltage fuses and
the light grey cylinders below them are the high voltage capacitors.
The low voltage capacitors are seen at the bottom.
7.3.2 The Capacitive Voltage Dividers

They consist of high voltage capacitors with oil/paper isolation with a maxi-
mum voltage of 42 kV rms. Under normal conditions, they operate at 9 kV
which gives a safety margin. The high voltage capacitors have a capacitance of
8 nF. The low voltage capacitors have a capacitance of 30 F plus some much
smaller capacitors for adjustments. All the six capacitive voltage dividers were
adjusted to give a voltage ratio of 1: 3733 0.2 %. The connection of each of
the capacitive voltage dividers to the signal conditioning circuit results in a
phase shift of 0.5 to 0.6 degrees (leading).
7.3.3 Signal Conditioning

The transducer signals were scaled to approximately 125 mV rms before being
fed to the signal conditioning unit, during normal conditions. The input ports
of the signal conditioning circuit were galvanically isolated from the output
circuit as shown in Figure 7.4. This was done both to minimise noise in the sig-
nal and to prevent earth currents from running into the unit. The digital filter
used was a 6th order Butterworth filter with a 3 dB cut off frequency at 5 kHz.
The sampling frequency was 12.8 kHz and the harmonics of interest were up to
80 which is equal to 4 kHz. The sample/hold circuit was used to allow all 15
channels to be sampled simultaneously as the data acquisition card can only
read one channel per 3 s.
Figure 7.4. Principle of the signal conditioning circuit.
7.3.4 The Personal Computers

Two PCs (HP 166 MHz, 64 Mbytes RAM and 2 Gbytes HD) were installed at the
Rejsby Hede wind farm site. One was used for receiving signals from the trans-
ducers whereas the other was employed for analysing data and capturing con-
trol signals from the ASVC. Both computers were equipped with a National
Instruments data acquisition cards NI-DAQ (type AT-MIO-64E-3) with 32
differential inputs. Both cards have a maximum sampling rate of 300 kHz
which is more than 15 channels multiplied 12.8 kHz. The A/D-converter used
measures each signal as a 12 bit digital value. One bit is reserved for critical
situations, another one is for the sign and the remaining 10 bits give the preci-
Figure 7.5. The signal conditioning unit with its auxiliary circuits.
sion of the A/D converter during normal operation. Figure 7.5 shows the signal
conditioning unit with its auxiliary circuits and the connections from the con-
trol cubicle of the ASVC. Figure 7.6 shows how the PCs are placed as near as
possible to the signal conditioning unit to avoid noise.
7.3.5 Noise
All the transducers were placed on busbars located in one substation building.
Figure 7.6. The location the PCs and the cabinet with
the signal conditioning circuit.
The ASVCs were located in another building together with all the control cubi-
cal and measurement equipment. This led to long cable connections, between
the transducers and the signal conditioning cards, which resulted in additio-
nal noise in the transducers signals. In an attempt to eliminate such noise, fil-
ters and galvanic isolation were installed both on the signal input side of the
signal conditioning unit and on the power supply side of network.
The uninterruptible power supply of the ASVC, also supplies power, containing
noise, to all the measuring equipment. A total of about 4 mV rms noise, which
was some 0.3 % of the signal, was present in the signal input ports of the DAQ
cards. However, the rms values of currents and voltages would be more accu-
rate than this since a noise contribution to an rms value averaged over 8 cycles
is negligible.
7.4 Calibration of Measuring Equipment
7.4.1 The Capacitive Voltage Dividers

These were calibrated with a voltage of 10 kV but were tested up to 20 kV. On
the high voltage side, the voltage was measured with a voltage transformer 20
kV/100 V, 50 VA class 0.5. Two voltmeters HP 3478A, with an accuracy of about
0.5 %, were used. The high voltage was generated by a 100 kV high voltage
transformer supplied from a variable transformer connected to the network.
Because of small variations in the network voltage and as it was difficult to
read two voltmeters at the same time, three measurements were taken and an
average value was calculated. Preliminary measurements were performed to
determine adjustments in the capacitance to be added, in parallel, to the low
voltage capacitor. Then, the voltage dividers were connected to the 20 kV cir-
cuit for 24 hours and measured again. These measurements were carried out to
determine the calibration constant. No work was done concerning the variation
of this constant with temperature as all voltage dividers operated indoors and
at room temperature. Figure 7.7 shows the results of calibration of the 6 capac-
itive voltage dividers.
0,2
Per Cent
0,1
0
-0,1
-0,2 1 2 3 4 5 6
Figure 7.7. Results of calibration of the voltage dividers.
7.4.2 Calibration of the Signal Conditioning Unit

A 50 Hz AC voltage, which is obtained from the networks voltage using a step
down transformer, was connected to the input ports of the signal conditioning
unit where the transducers were to be connected. The input signals were mea-
sured with a HP voltmeter 3457 with a 0.3 % accuracy. The instrument was set
up to calculate an rms value of the signal over 8 cycles after being triggered
each time. At the same time, a PC was used to calculate the rms value of the
signal taken over 8 cycles with a sampling rate of 256 per cycle. The triggering
of the voltmeter and the PC was done manually and simultaneously.
Calibrations were carried out for different input signals. The following three
figures show their corresponding results. The small variation in the network
voltage and the trigger time is not exactly the same and so may explain some
of the variations in these figures.
Figure 7.8 shows the measurement of the signals from the capacitive voltage
dividers. The bars show the deviations, in %, of the desired gain from input to
screen of the PC. Each grey scale represents a different input voltage. The
desired gain is 1.1126. The ratio from busbar to screen may be calculated as
1.1126/3733=3355.2 .
Figure 7.9 shows the current measurements of all 9 channels. The current was
determined as a voltage across a calibrated resistor with a resistance of
0.968555 . The gain from input to screen is fixed at 0.6800 V/A.
Figure 7.9 shows the deviation, in %, from this value. Each grey scale repre-
sents a different current. From busbar to screen, the ratio is equal to
(800/5)/06800=235.29 A/V.
Figure 7.10 shows measurements of the 4 channels for the existing voltage
transformers which were in use only for a very short time during the commis-
sioning stage. From input to screen, the gain is fixed at 0.023384 V/V. Figure
7.10 shows the deviation from this value, in %, and where each grey scale re-
presents a different input voltage. The ratio of voltage from busbar to screen is
5830.7 which is equal to (5000/110) / 0.023384.
0,2
0,1 1,058
1,554
0 1,965
-0,1 1 2 3 4 5 6 3,138
-0,2
Figure 7.8. Results of calibration of the total signal path

from transducers to reading on the screen for the voltage signals
from the capacitive voltage dividers.
0,1
1,363
0
1 2 3 4 5 6 7 8 9 2,713
-0,1 4,391
-0,2
Figure 7.9. Results of calibration of the total signal path from transducers to
reading on the screen for the voltage signals from the CTs.
0,1 41,26
0,05 90,74
110,07
0
1 2 3 4 132,73
-0,05
Figure 7.10. Results of calibration of the total signal path from transducers to
reading on the screen for the voltage signals from the VTs.
7.5 Software
7.5.1. Goals
The acquisition of measurement data from the network and the ASVC of the
Rejsby Hede wind farm, was performed using the following software devel-
oped by DTU for the project:
a) Continuous monitoring of the power quality and the power generation

from the wind farm.
b) Continuous monitoring of the operation of the ASVC combined with a
detection of any abnormal events.
c) Fast acquisition of measurements to be used for example in connection
with islanding tests.
d) Conversion of data into suitable formats, such as EXCEL and MATLAB,
to facilitate detailed analysis of the data.
e) Creating a WEB-page with an animated picture of the power production.
7.5.2 Hardware Configuration

The measurement system was based on PC-techniques. Two 166 MHz Pentium
PCs were installed in the control room at the wind farm site one used for
measurements of voltage and current in the network, and the other used for
measurements of control signals from the ASVC. Both PCs were equipped with
a 32 channels, 12 bit, 300 kHz ADC-board from National Instruments. Signals
from the measurement process were then connected to the ADC-boards
through transducers and the signal conditioning equipment.
The two PCs were connected to their own private LAN, which through an
ISDN-line was linked to the LAN at DTU as shown in Figure 7.11. This
meant that the two PCs were part of the DTU-LAN and therefore could be
accessed from DTU. This configuration was beneficial in many ways as listed
below:
a) The PCs in Rejsby Hede can be remotely controlled from DTU through
the program CarbonCopy, which has been a great advantage in program
development and program testing.
b) Data stored on disk in Rejsby Hede can be transmitted to DTU.
c) A WEB-page with automatic updating of production data can be
designed.
7.5.3 Basic Specifications

A complete picture of the state of the network and its associated wind farm
can be obtained from measurements of 15 signals, i.e the voltages and currents
Figure 7.11. PC configuration.

in all three phases of the two busbars where the wind turbines are connected,
and the currents in all three phases of the ASVCs, as was shown in Figure 7.2.
The harmonic content, up to a certain frequency, of these signals is also impor-
tant. The IEC-norms recommend calculation of harmonics up to the 40th, eval-
uated over a window width of at least 8 cycles (i.e 160 ms) at 50 Hz. Following
preliminary tests, it was decided that 40 was not enough and an increase in the
range of harmonics up to the 80th was considered. The cut-off frequency in the
signal conditioning equipment and the sampling rate were chosen accordingly.
The sampling rate was set to 12.8 kHz giving exactly 256 measurements per
signal at 50 Hz, which was suitable for performing FFT analysis. The 15 signals
sampled at 12.8 kHz, plus an FFT-analysis over 8 cycles (2048 points) required
a lengthy and complex process on the PC. Therefore, it was not possible to
carry out continuous real time analysis without missing any information. An
alternative approach was to use a computer platform with a DSP-based sys-
tem. However, due to time constraints on the project (WPQI-project), the PC
system combined with the available software tools were used.
A continuous monitoring of 5 out of the 23 different internal signals in the

control system of the ASVC device was required by Siemens. In addition, any
specific event on 4 of the control signals should be traced at any given time.
Furthermore, in case of such an event, all 23 control signals should be recorded
a few cycles before and after the event at a sampling rate of at least 6 kHz.
A sampling rate of 6.4 kHz was chosen.
A separate PC was installed specifically so that in case of any event, every sin-
gle sample in the 23 control signals is recorded. The use of this PC system was
satisfactory in capturing every single data needed but has put a severe limita-
tion on the calculations, which should be performed in real time.
7.5.4 Programming Tools

LABVIEW, which is a software package from National Instruments compatible
with the measuring equipment, was used. LABVIEW was chosen because of its
speed since it can acquire and store real time measurements on a disk very
fast. It works in a Windows environment and it is a multitasking system. It
includes a variety of drivers for PC peripherals and dedicated measurement
boards. LABVIEW is very flexible, it is equipped with a special graphical pro-
gramming language and has a number of effective algorithms for creating user
interfaces.
7.6 Programs
7.6.1 Monitoring the Wind Farm

The connection of an ASVC to a wind farm influences the power quality of its
associated network and introduces harmonics in the networks voltages.
Therefore, it was important to study and analyse the effects of the ASVC on
the network during data acquisition. A program has been specifically developed
for this purpose. Changes in the network were continuously monitored through
measurements of relevant 3-phase voltages and currents in all 15 signals
sampled at 12.8 kHz (see Figure 7.2). Parameters characterising each single sig-
nal and the power production were calculated every 160 ms, i.e. 8 periods at 50
Hz, and with a total of 2048 samples per signal. The voltage and current sig-
nals are described as follows:
1) RMS both the total RMS value with all harmonics and the
RMS value of the 1st harmonic
2) Phase of the 1st harmonic one signal is defined as reference,
and the phases of all other signals are specified as the
deviation from this reference
3) Harmonics up to the 80th or 4 kHz derived from a FFT over 2048
samples
4) THD calculated on the basis of the 80 harmonics
5) Frequency of the reference signal, estimated from detection of
phase changes between cycles of the 1st harmonic
The power is divided into the following categories:
1) Apparent power calculated from RMS values of voltage and current

signals in 3 phases.
2) Active power calculated by multiplying simultaneous measurements
of voltage and current signals in 3 phases.
3) Reactive power a) calculated as above, but with the currents displaced
a quarter of a period.
b) calculated using the space vector method.
Alternatively the same power categories may be determined using the positive
sequence components of the system found by a 1st harmonic analysis. This
analysis makes it possible to verify the symmetry of the network, which is a
basic assumption for the method used in controlling the ASVC. Thus, the intro-
duction of a final parameter defined as:
4) Asymmetry the ratio between the negative sequence component

and the positive system component.
The capacity of the PC is not sufficient for calculation of all these parameters
in real time without loosing parts of the signals, notably the FFT algorithm
which is a time consuming process. To reduce the load on the PC, the harmon-
ics of just one signal were evaluated each time a set of parameters is calculat-
ed, then at the next time another signal is evaluated and so on. In this way, new
parameters can be obtained twice a second, covering 16 cycles or around one
third of the content of the signals.
A huge amount of data was produced and therefore effective ways of reducing
this data into files of manageable sizes, to be stored on disk, was necessary. The
chosen procedure was based on storing data at 2 minutes intervals. At each

interval, the mean, maximum and minimum values of all parameters are
derived. A file, approximately 4.5 Mb for a 24 hours recording period, is then
written to the disk. A new file, with a date dependent name, is created each
midnight (i.e, at 00:00). The files may be transmitted, on demand, to DTU via
the ISDN link.
The performance of the Rejsby Hede wind farm-network was analysed using
these data files. Experience has shown that the data gave an adequate descrip-
tion of the operation of the network and allowed the effects of the ASVC on
the network to be investigated.
A simple user interface was build into the program. It presents the values of
most of the parameters on the screen and updates them each time new parame-
ters are calculated. It also allows plotting of pre-selected signals and parame-
ters.
7.6.2 Monitoring the ASVC

The internal functions of the ASVC were observed using the 23 signals, acces-
sible from the outside as +/- 15 V signals. Five of these signals have to be moni-
tored continuously. The mean, maximum and minimum values of each of the
signals are calculated every 2 minutes and saved on disk.
In case of specific events, the recording of all 23 signals was performed. The
recording consisted of a few cycles of pre-event and post-event data. This has
lead to a continuous scanning of every single sample measurements from all 23
signals without missing any measurements. A sampling rate of 6.4 kHz was
used, as suggested by Siemens.
A program performing both tasks have been developed. The 2-minute values
are stored in files on a daily basis whereas each event is stored in a separate
file. Since the recording function is very demanding on the PC, no further on-
line analysis can be carried out.
7.6.3 Fast Data Acquisition.

There was a need for a data acquisition system which can collect measure-
ments and store data on disk as fast as possible. Such a system was construct-
ed. Its performance was only limited by the capacity of the ADC-board ( i.e.
approximately 300,000 measurements a second).
An additional program which reads the measurements from disk and plots
selected signals has also been developed. It has been applied in an experiment
in Rejsby Hede, where the wind farm was disconnected from the network to
run in an islanding mode. The analysis of data from this experiment is not yet
completed.
7.6.4 Conversion of Data

LABVIEW stores data in a special format meaning that a conversion must be
made before the data can be used by standard PC-programs. A number of
DOS-programs, and which are compatible with all three types of data acquisi-
tion methods mentioned above, have been developed for this purpose. Data can
be translated into ASCII CSV-files, which is the format used in spreadsheets
(EXCEL), or can be transformed into binary MAT-files allowing the use of the
advanced mathematical analysis of MATLAB. To facilitate the data analysis, a
simple user interface was also made available, so that parameters of interest
could be selected and analysed.
7.6.5 WEB-page
A WEB-page showing the production of power in Rejsby Hede is featured on
the Internet with the address: http://www2.eltek.dtu.dk. The WEB-page is
accessible through standard Internet browsers from Netscape or Microsoft. The
system is set to manage a large number of Internet users simultaneously. The
Internet connection is established via a http-Server at DTU. The generation of
the data and plotting of curves are made by two JAVA-programs, one running
on the Server and the other running on the Internet user's the client's
machine. The client receives the JAVA program through a JAVA- applet includ-
ed in the html-page.
Three curves are plotted on this page. Two of the curves display the total active
and reactive power produced by the wind farm, and the third shows the reac-
tive power produced by the ASVC over the past 24 hours. The curves are
updated each 15 minutes. The power data from Rejsby Hede is created and
stored in a special data file by the PC, which monitors the wind farm perfor-
mance. The server reads the files and transfers the necessary data to DTU over
the ISDN link.
7.7 Measuring Results
After commissioning of the ASVC measurements were recorded.
Figure 7.12 shows the measurements of the mean values over a 2 minutes peri-
od of power, ASVCs reactive power and the resulting reactive power from the
network to the wind farm. The ASVC was started on the 22nd of October at
about 9 oclock. After a test with maximum VAr-production, it was capable of
supplying all the reactive power requirement of the induction generators in the
wind farm. On midday of the 23rd of October, the ASVC was operating at its
maximum reactive power output of 8 MVAr. This meant that some of the wind
farms reactive power demand was supplied from the network.
Figure 7.13 shows the measurements in more detail from the 23rd of October.
This figure shows the power, the reactive power exported to the network and
25
1997 October 22-24
Start of ASVC
20
Power Produced
reactive
MW, Mvar
15 power
10
5
Reactive
power
0
21
21
21
12
15
18
12
15
18
12
15
18
0
3
6
9
0
3
6
9
0
3
6
9
Time (hours)
-5
Figure 7.12. 72 hours measurements of generated power,

generated reactive power and resulting reactive power to the network.
All points are mean values over a 2 minutes period.
30
1997 October 23
ASVC is operating
25
20
P-max
MW, Mvar
15 P-mean
P-min
QC-max
10 QC-mean
QC-min
5 Q-max
Q-mean
Q-min
0
0
1
2
3
4
5
6
7
8
9
20
21
22
23
10
11
12
13
14
15
16
17
18
19
Time (hours)
-5
Figure 7.13. 24 hours measurement of generated power,

generated reactive power and resulting reactive power to the network
(Maximum, mean and minimum values over a 2 minute periods).
1,2
1997 October 28
ASVC out of operation
1
0,8
MW, Per Cent
11 Har. (%) Power/20 (MW)

0,6
0,4
0,2
13 Har. (%) 11 Har. (%)/13 Har. (%)
0
20
21
22
23
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
-0,2 Time
Figure 7.14. 11th and 13th harmonics in network voltage related

to the power output from the windfarm with the ASVC out of service.
the produced reactive power from the ASVC. The curves show the maximum,
the mean and the minimum value over a 2 minutes period for each of the vari-
ables.
During certain conditions, the wind farm was shown to be sensitive to the 11th
and 13th harmonics, as shown in Figures 7.14 and 7.15. In Figure 7.14, the
ASVC was out of operation and disconnected from the network. This figure
contains the power curve (red) which was conveniently scaled, and curves re-
presenting the 11th and 13th harmonics. Figure 7.15 shows the same plots as in
Figure 7.14 but with the ASVC in operation. The ASVC was carefully designed
to eliminate 11th and 13th harmonics and therefore it was concluded that these
harmonics may have been generated by the network.
It was also noted that 11th and 13th harmonics increase dramatically when the
power decreases to a level of 1 MW or less. When the power is less that 1 MW,
it means that all wind turbines are operated at low speed (6-poled state) with
the smaller or even no power factor capacitors. This condition leads to a reso-
nance condition for frequencies about 600 Hz in the wind farm. Unfortunately,
this condition could not be further investigated since it was not possible to
monitor the performance of each single wind turbine.
In Figure 7.16 the amplitudes of most of the largest harmonics are shown.
2,5
1997 October 24
ASVC in operation
2 P-mean/10 (MW)
11 Har. (%)
MW, Per Cent
P-mean/10 (MW)
1,5
1 11 Har. (%)
0,5 13 Har.
(%)
20
21
22
23
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Time (hours)
Figure 7.15. 11th and 13th harmonics in grid voltage related to

the power output from the windfarm with the ASVC in service.
There are no critical harmonics at higher frequencies in the data (recorded

between the 22nd and the 24th of October 1997). A comparison between this
figure and Figure 7.12 shows that there is no visible correlation between wind
farm generation and harmonic amplitude.
The data material is still to small for doing any conclusions about conditions
under which special harmonics occur, but the measurements continue.
Figures 7.17 and 7.18 show the deviation of the phase voltage in phase R on
bus 2 and bus 1 respectively.
The ASVC, connected to bus 2, was in operation from about 9 oclock. During
the first one and a half hour, the ASVC was working with full reactive power
which explains the increase in the voltage (4 %) at bus 2 and the decrease of
approximately the same percentage in bus 1 voltage. The power and reactive
power at each bus are plotted individually in two separate figures. To obtain
the total power and reactive power, the two curves from bus 2 and bus 1 must
be added.
Figure 7.19 shows the three phase voltages (designted R, S, T) on bus 1, which
is the bus without the ASVC. Figure 7.20 shows the three phase voltages on bus
2, which is the bus with the ASVC and this explains the variations observed in
these waveforms. The power generated when the measurements were carried,
1,4
59R
1997 October 22. Important harmonics
ASVC started 9:15
1,2
1
Per Cent of 1st Harmonic
0,8
61R
61R
0,6
47R
0,4
49R
0,2 49R
59R/61R 47R
Time (hours)
0
20
21
22
23
10
11
12
13
14
15
16
17
18
19
0
1
2
3
4
5
6
7
8
9
Figure 7.16. Harmonics 47, 49, 59 and 61 in the network voltage

before and after connection of the ASVC.
12
1997 October 22. Relative Phase Voltage and
MW, Mvar and Per Cent
10 the Powers on Bus 2.

ASVC started 9:15 Produced
8 reactive power
Power
6
0
0
20
21
22
23
10
11
12
13
14
15
16
17
18
19
1
2
3
4
5
6
7
8
9
-2 Time (hours)
Voltage (Deviation
-4 from 1 pu. in %)
Reactive power
-6
-8
Figure 7.17: Relationship between phase voltage and reactive power on bus 2.
10
1997 October 22. Relative Phase Voltage and Powers on Bus 1.
ASVC started 9:15
8
MW, Mvar and Per Cent.

Power
Reactive power
0
20
21
22
23
10
11
12
13
14
15
16
17
18
19
1
2
3
4
5
6
7
8
9
0
Time (hours)
-2
Voltage. Deviation from 1 pu. in %

-4
Figure 7.18. Relation between phase voltage and reactive power on bus 1.
approximately 17 MW, is far beyond the power level which may give rise to 11th
and 13th harmonics. The disturbance observed in these waveforms may be
caused by higher harmonics as shown in Figure 7.16.
Figures 7.21 and 7.22 show the currents at bus 1 and bus 2 respectively. The
currents from the ASVC are shown in Figure 7.23. This waveform contains
oscillations generated as a result of the switching of the power electronics. The
combination of delta/delta and delta/star for the connection of the two units of
the ASVC and the appropriate choice of the firing angle for the GTO-thyristors
have resulted in mitigating these oscillations and bringing the curve to a near
pure sinusoidal waveform.
Voltage on Bus 0 (1997 September 9)

15
kV
10 S T R
451
251
351
401
551
151
201
301
501
101
51
1
-5
-10
Samples
-15
Figure 7.19. Phase voltages on bus 1.
Voltage on Bus 2 with ASVCs Operating (1997 September 9)

15
kV
S T R
10
0
451
251
351
401
551
151
201
301
501
101
51
1
-5
-10
Samples
-15
Figure 7.20. Phase voltages on bus 2.

Current from Bus 1 with ASVCs Operating (1997 September 9)

500
A
400 R S T
300
200
100
451
251
351
401
551
151
201
301
501
101
51
1
-100
-200
-300
-400
Samples
-500
Figure 7.21. Current at bus 1.
Current from Bus 2 with ASVCs Operating (1997 September 9)

800
A
600
R S T
400
200
0
451
251
351
401
551
151
201
301
501
101
51
1
-200
-400
-600
Samples
-800
Figure 7.22. Current at bus 2.

Current from ASVCs (1997 September 9)

500
A
400
T R S
300
200
100
451
251
351
401
551
151
201
301
501
101
51
1
-100
-200
-300
-400
Samples
-500
Figure 7.23. Current from the ASVC.

134 DESCRIPTION OF THE MEASURING EQUIPMENT
135
Chapter 8
The Application of
Advanced Static VAr
Compensators (ASVCs)
to Other Wind Farms
136 Advanced Static VAr Compensators (ASVCs)
Zouhir Saad-Saoud, Maria Luiza Lisboa, Nicholas Jenkins, Goran Strbac

UMIST
PO Box 88
Manchester
M60 1QD
ENGLAND
Janaka Ekanayeka
Department of Electrical and Electronic Engineering
Peradeniya
SRI LANKA
Advanced Static VAr Compensators (ASVCs) 137
8.1 Introduction
The economic and technical aspects of connecting wind generation into distri-
bution networks are of significant interest to both wind farm developers and
network operators [1] and effective reactive power control and voltage regula-
tion can have an important influence on the level of embedded generation
which is acceptable and on the costs of its connection [2-4].
Therefore, a comprehensive set of studies was undertaken to investigate the use

of Advanced Static VAr Compensators [5-8] to improve both the steady state
and dynamic impact of wind farms on the networks to which they are connect-
ed. The application of ASVCs to three different wind farms was investigated
and their effects on networks' power losses, on the voltage profile and on the
amount of the active power that can be injected were examined.
8.2 Study Case 1
Study Case1 was that of a large (21.8 MW) wind farm connected to a rural 33
kV network with a low short circuit level. The size of the wind farm is similar
to that of Rejsby Hede but the distribution network is at a lower voltage level
(33 kV rather than 60 kV) and with a much lower short circuit level.
A simple representation of the electrical network is shown in Figure 8.1. The

wind farm consists of 36 x 600 kW fixed speed stall regulated wind turbines.
Each wind turbine is connected to the wind farm power collection network by
a 630 kVA, 0.69/33 kV transformer. Power factor correction capacitors of 175
kVAr are connected at the terminals of each wind turbine generator. The wind
farm is connected to the 33 kV network at B5 by mainly an overhead line with
a small section of underground cable. The reactive power demand of the induc-
tion generators at rated output is 10.5 MVAr. This leads to a reactive power
deficit of 4.2 MVAr which, unless supplied by an ASVC, will be obtained from
the network.
Under normal operating conditions, the three phase short-circuit level at the
boundary of the wind farm (B6) is 121 MVA with an X/R ratio of the source
impedance of 2.24. In the event of a fault on one of the 33 kV circuits between
B2 and B5, the fault level at the wind farm 33 kV connection will drop to 85
MVA with an X/R ratio of 2.18. Therefore, the ratio of fault level to embedded
generation rating for this network is 5.6 under normal conditions, dropping to
3.9 under outage conditions.
8.2.1 Steady State Performance

Unity Power Factor and Optimal Power Flow Control (UPF)
At Rejsby Hede, a Unity Power Factor based control strategy was adopted. The
UPF based control strategy makes the power factor of a wind farm equal to
one (cos = 1), resulting in a zero reactive power flow between the distribution
network and the wind farm at all times. With this control scheme, the wind
farm's reactive power requirement is locally supplied and so the impact of
embedded generation on the network losses is minimised.
In order to analyse the performance of the UPF control strategy, a simple load
flow simulation model was developed. The procedure was based on successive
iterations between the AC load flow and a simple algorithm that determined
the set voltage point of the ASVC at the wind farm busbar which leads to zero
reactive power flow between the wind farm and distribution system for a given
operating point. After the new ASVC voltage is determined, a load flow is per-
formed and a new overall system solution determined. This iterative process
between the two programs converges when the set point of the ASVC results in
a load flow solution with no reactive power flow between wind farm and the
distribution system.
The UPF control strategy does not explicitly take into consideration the net-
work's active power losses. Therefore, an Optimal Power Flow (OPF) model
which minimises losses explicitly was also developed and implemented. This
model optimises the ASVC's set voltage to minimise network power losses for a
given loading condition of the system. As minimisation of power losses in dis-
tribution networks is a non-linear optimisation problem, the OPF problem was
solved by implementing a successive linear programming (SLP) technique,
which iterates between an incremental load flow model and a linear program-
ming algorithm. The process converges when no further improvement (reduc-
tion) in active power can be achieved: power losses are minimised by adjusting
the voltage of the ASVC taking into consideration its reactive power capacity.
The UPF based control and OPF model were compared using the wind farm
illustrated in Figure 8.1. The simulation results show that the UPF control
leads to larger power losses than the loss minimisation based control (Figure
8.2(a)). However, the maximum difference of 51 kW, recorded for the maximum
power output of the wind farm, is not very significant. By minimising the reac-
tive power flow between the wind farm and the system, active power losses
may be close to minimum, particularly if there is no load connected close to the
wind farm. It is important to note that this small improvement in losses would
require a significant increase in ASVC capacity (an additional 5 MVAr) as
shown in Figure 8.2(b) which would be difficult to justify commercially.
B1
B2
L1
B4 B3
L2
B5
B7 B6
G L3
PFC ASVC
Fig 8.1 : Wind farm and network representation - Study Case 1.
For "weak" distribution systems, one alternative reactive power compensation

strategy would be to increase the amount of active power that can be injected.
The appropriateness of such strategy was examined by considering the voltage
profile and reactive power flows under extreme, but realistic, operating condi-
tions: light loading of the network with high levels of active power output of
Fig 8.2(a): Comparison between Unity Power Factor (UPF) and

Optimal Power Flow (OPF) control in terms of network power
losses (maximum network load).
Fig 8.2(b): Comparison between Unity Power Factor (UPF) and

Optimal Power Flow (OPF) control in terms of ASVCs reactive power
requirement (maximum network load).
wind farm. A maximum allowable voltage fluctuation of +/- 5 % was assumed.

The OPF was employed to compute the minimum reactive power that should
be imported by the ASVC to maintain the voltage within these limits, and was
compared with the UPF strategy.
When light loading of the distribution system coincides with the full active
power output of the wind farm, the voltage rise effect may become significant,
and voltages may increase above the maximum limit. The UPF control policy
is inappropriate for maintaining the voltage under such critical conditions as
shown in Tables 1 and 2. Instead of acting as a reactive source, the ASVC
should import reactive power from the system and reduce the voltage rise
effect. With the OPF control, the reactive power output of the wind farm is
adjusted to maintain the voltage in the system within the permissible range.
This is achieved by importing reactive power from the system (the ASVC acts
as a reactive power sink), and thus reduces the effect of voltage rise. The mini-
mum required reactive power import using this technique was determined for
various levels of active power generation as shown in Figure 8.3(a). A success-
ful voltage control was achieved only at the expense of an increase in network
power losses as shown in Figure 8.3(b). This situation is expected to arise only
occasionally and so the increase in network power losses would probably be
tolerable. The alternative solution of shedding generation would incur higher
overall costs.
voltage (pu)
Busbar no ASVC UPF OPF
B2 1.030 1.030 1.030
B3 1.051 1.067 1.038
B4 1.051 1.067 1.038
B5 1.054 1.067 1.038
B6 1.077 1.108 1.050
B7 1.067 1.099 1.040
Table 8.1 : The effects of various reactive power compensation schemes

on network voltage with rated wind farm output.
no ASVC UPF OPF

Losses (MW) 1.90 1.72 2.19
QASVC (MVAr) 0 +4.5 -3.5
Table 8.2 : Total network losses and ASVC rating under

various reactive power compensation schemes.
Wind farm output power (MW)

26
24
22
20
18
16
14
12
10
6
-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
ASVC's reactive power (MVAr)
Fig 8.3(a): ASVC reactive power requirement for different levels of

wind farm generation (minimum network load).
Wind farm output power (MW)

26
24
22
20
18
16
14
12
10
6
0 0.5 1 1.5 2 2.5 3
Losses (MW)
Fig 8.3(b): Network power losses for different levels of wind farm generation
(minimum network load)
8.2.2 Dynamic Performance

Dynamic studies are required to investigate the transient behaviour of the net-
work under various conditions and so appropriate dynamic analysis tools must
be used. The electromagnetic transient program, PSCAD/EMTDC [9], has been
used to investigate islanding phenomena and the mitigation of voltage fluctua-
tions caused by blade passing frequency. An EMTDC model of the network, the
wind farm and the ASVC was developed.
The ASVC is a three level design rated at +4.5/-3.5 MVAr and is connected to
busbar B6 through a three phase D-Y transformer. It is operated in a Selective
Harmonic Elimination Modulation (SHEM) mode to keep the total harmonic
voltage distortion within acceptable limits. The asymmetrical rating of the
ASVC is due to a 500 kVAr passive filter used to limit higher order harmonics.
Two control techniques for the ASVC, based on the various established control
strategies available in the literature [10-13], were implemented. A unity power
factor technique used the reactive power drawn by the wind farm as the con-
trol variable for the reactive power output of the ASVC. The ASVC continuous-
ly supplies VArs to balance the reactive power demand of the wind farm up to
its rating, beyond which a constant 4.5 MVAr is maintained. Alternatively, a
voltage control technique maintained the voltage at the wind farm connection
to a reference value by injecting or absorbing reactive power. The reference
voltage of the wind farm was maintained using a proportional and integral
control scheme.
Voltage Stability Performance
In a weak rural networks, where circuit impedance can be high, the voltage
can be depressed by the reactive power drawn by the wind farm. In particular,
when the real power produced by an induction generator exceeds its rated
value, there is the possibility of voltage instability as large amounts of reactive
power are absorbed [14]. This situation must be considered on wind farms as
the power control of wind turbines is often imprecise. The use of an ASVC to
actively control network voltage by locally supplying reactive power can
reduce the likelihood of voltage instability.
To study the improvement in steady state stability of the network with the
ASVC, investigations were carried out using the PSCAD/EMTDC model. The
machine torque was increased from its 1 pu value and the voltage at the wind
farm 33 kV busbar was measured. Figure 8.4 shows the wind farm 33 kV volt-
age plotted against the wind farm output power for the winter loads in the
network. That figure clearly indicates the improvement of the steady state volt-
age stability limit when the ASVC generates 4.5 MVAr of reactive power.
Without the ASVC With the ASVC

1.05
Wind farm 33 kV voltage / p.u
0.95
0.85
0.75
0.65
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
Output power / p.u
Fig 8.4: Wind farm 33 kV voltage versus wind farm output power
with and without the ASVC.
Islanding
A wind farm may be isolated from the network as a result of a post-fault

switching in the distribution network. This is generally referred to as islan-
ding which may give rise to self-excitation of the wind turbine generators due
to the presence of the PFC capacitors, thus increasing the voltage of the isolat-
ed network. Under an islanding condition, an ASVC can be used to absorb
reactive power from the wind turbines, thus reducing the magnetic flux which
supports the self-excitation [14,15].
In this study an islanding condition was studied by disconnecting the wind
farm at the 33 kV busbar (B6). When modelling the induction machines under
self-excitation conditions, the wind farm voltage increases and so it is neces-
sary to add the effect of saturation to the machine models. The machine satura-
tion was modelled with two piece-wise linear curves. The unsaturated mag-
netising reactance was set to magnetising reactance plus the stator reactance
and the saturated magnetising reactance was selected as 1/6th of the unsatu-
rated reactance. Figure 8.5(a) shows the voltages at the wind farm 33 kV bus-
bar and active and reactive power requirement of the wind farm when the
wind farm is disconnected from the 33 kV system without the ASVC in service.
Clearly the wind farm turbines are self-exciting and the voltage at the wind
farm busbar starts increasing. The ASVC, operated under voltage control, was
used to stabilise the wind farm voltage under this condition and Figure 8.5(b)
shows the results obtained with the ASVC. The 33 kV voltage is maintained
close to its set point value of 0.99 pu although there is a voltage transient at
the instant of islanding caused by the delayed response of the phase locked
loop control system of the ASVC.
Mitigation of voltage fluctuations
When a wind turbine blade passes its tower, it produces a pulsating torque in
the induction generators, of up to 20 % of rated output thus creating a voltage
fluctuation in the network around the wind farm. If the wind farm becomes
synchronised, i.e. the blades on different wind turbines all pass the towers at
the same time, then the power fluctuations at the blade passing frequency will
add together. The network voltage fluctuations can then become very high.
When a wind farm is synchronised, the synchronous torque is transferred

through the wind farm by voltage fluctuations on the circuit [16]. By stabilis-
ing the 33 kV voltage the synchronous torque can, no longer, be transferred.
Stabilising the wind farm 33 kV is a very interesting application of an ASVC
as both absorbing and generating modes of the ASVC can be used alternatively
in this case.
To study the possibility of applying an ASVC to mitigate voltage fluctuations

at blade passing frequency, a pulsating torque with magnitude variation 20 %
and frequency of 1.6 Hz was applied to the torque input of the wind farm gen-
erators. The ASVC was connected to the circuit using the voltage control tech-
nique described in Section 8.2.2. The results obtained with and without the
ASVC are shown in Figure 8.6. A similar result was obtained with the 33 kV
voltage signal passed through a washout filter or change function [11].
From Figure 8.6, it is clear that the ASVC can mitigate voltage fluctuations at
the wind farm 33 kV busbar effectively. Due to the active power variations
within the wind farm, the voltage at the wind farm 690 V busbar fluctuates
even though the ASVC is in operation. However, eliminating the voltage fluctu-
ations at the wind farm 33 kV is an important result as it should allow the syn-
chronised operation of the turbines to be disturbed and the synchronism bro-
ken.
Voltage at busbar B5
1.40
1.25
p.u
1.10
0.95
1 1.3 1.6 1.9 2.2 2.5
Windfarm 33 Voltage
1.40
1.25
p.u
1.10
0.95
1 1.3 1.6 1.9 2.2 2.5
Power output of the wind farm

5
0
-5
MW
-10
-15
-20
-25
1 1.3 1.6 1.9 2.2 2.5
Reactive power demand of the wind farm

16.0
10.5
MVAr
5.0
-0.5
-6.0
1 1.3 1.6 1.9 2.2 2.5
Turbine speed
1.16
1.12
1.08
p.u
1.04
1.00
1 1.3 1.6 1.9 2.2 2.5
Time
Fig 8.5(a): Islanding at the wind farm 33 kV busbar without the ASVC.
Windfarm 33 Voltage
1.10
1.07
p.u
1.03
1.00
0.96
1 1.3 1.6 1.9 2.2 2.5
Power output of the wind farm

10
3
MW
-4
-11
-18
-25
1 1.3 1.6 1.9 2.2 2.5
Reactive power demand of the wind farm

12.0
8.5
MVAr
5.0
1.5
-2.0
1 1.3 1.6 1.9 2.2 2.5
Turbine speed
1.16
1.12
p.u
1.08
1.04
1.00
1 1.3 1.6 1.9 2.2 2.5
Reactive power output of the ASVC

6
4
2
MVAr
0
-2
-4
-6
1 1.3 1.6 1.9 2.2 2.5
Time
Fig 8.5(b): Islanding at the wind farm 33 kV busbar with the ASVC.
Wind farm 33 kV voltage

0.995
0.988
0.981
pu
0.974
0.967
0.96
Wind farm 0.69 kV voltage

0.99
0.979
0.968
pu
0.957
0.946
0.935
Q_ASVC Q at the HV side of the wind farm tf

12
8
MVAr
-4
Machine output power Machine reactive power

15
6
MW/MVAr
-3
-12
-21
-30
1.5 1.75 2 2.25 2.5 2.75 3 3.25 3.5
Fig 8.6: ASVC used to mitigate voltage fluctuations created

by pulsating torque.
8.3 Study Case 2
Study Case 2 was a medium size wind farm (15 MW) sited in a remote area.
The available distribution voltage levels were 20 kV and 66 kV and there was
considerable interest in connecting the wind farm at 20 kV both to reduce
costs and to minimise the environment impact of the long overhead circuits
which would be required. However, this type of connection is characterised by
a rather low ratio of fault level to wind farm rating (i.e 3.5), under normal
operating conditions.
The main aspects of interest in the study case were: (1) steady state voltage, (2)
distribution network losses and (3) voltage stability.
A simple representation of a section of the network to which the wind farm is

connected, is shown in Figure 8.7(a). The wind farm is rated at 15 MW and was
represented by thirty 500 kW stall regulated wind turbines. Each wind turbine
is equipped with a local power factor capacitor of 187.5 kVAr. The connection
of the wind farm to the electrical network will be at the 66/20 kV substation
and so both 20 kV and 66 kV connections were considered. The connection cir-
cuits, 20 kV or 66 kV, would be approximately 15 km of overhead line with
short underground sections at each end close to the utility and wind farm sub-
stations.
Fig 8.7: Simple representation of the wind farm with various

options of connection - Study Case 2.
Fig 8.8: Voltage at 20 kV busbar with varying wind farm output.
Fig 8.9: Variation of network losses with wind farm output.

The application of an ASVC for the compensation of the reactive power

demand of this wind farm, during various operating conditions and with both
options of connection, was investigated. The ASVC was used to balance the
reactive power demand of the 20 kV or 66 kV circuit, up to its rating. These
studies were carried out using the IPSA [17] software.
8.3.1 20 kV Connection
The voltage at the 66/20 kV Substation is kept constant at 20.4 kV (1.02 pu)
using on-load tap changers of the 66/20 kV transformer. The power output of
the wind farm was varied from zero to 160 %, after which the system becomes
unstable. The corresponding voltages at the 20 kV busbar are shown in Figure
8.8. The kink in the voltage, at 60 % of mechanical power, is due to the opera-
tion of the transformer on-load tap changer. The 20 kV connection results in
high losses of nearly 1.5 MW at rated output as shown in Figure 8.9.
The effect of adding a +/- 4 MVAr ASVC at the wind farm (20 kV busbar),
using a unity power factor control scheme, was to increase the stability limit to
180 % and reduce losses by 168 kW at rated output as shown in Figure 8.9.
This also led to excessive voltage on the 20 kV circuit, 9 % at full generation as
shown in Figure 8.8, and therefore it would not be possible to apply the full
rating of the ASVC (i.e +/- 4 MVAr). A reduced ASVC rating of +/- 2 MVAr
was considered to supply part of the wind farm reactive power demand and to
keep the wind farm 20 kV voltage within limits (0.94-1.06 pu). This strategy
resulted in a stability limit of 180 % and a reduction in losses at rated wind
farm output of 65 kW.
Apart from the high losses in the connecting circuit, the 20 kV connection
without an ASVC appears to be acceptable. As the reduction in losses with the
ASVC was modest it would appear to be difficult to justify the application of a
compensator in this situation.
8.3.2 66 kV Connection
The wind farm is connected through a 66 kV circuit to the high voltage side of
the 66/20 kV transformer, as shown in Figure 8.7(b). Two transformers, 66/11
kV and 11/0.69 kV, were used to step up the voltage of the wind turbines. The
voltage at the 11 kV busbar is kept constant at 1.0 pu using the on-load tap
changers.
Figure 8.10 shows the variation of the wind farm voltage with different levels
of generation. The wind farm was shown to be stable up to 153 % of mechani-
cal input power under normal conditions. The kinks in the voltage at the 66 kV
busbar are due to on-load tap changer on the 66/11 kV transformers to keep
the voltage at the 11 kV busbar constant at 1 pu. The 66 kV voltage connection
circuit results in reasonable losses (750 kW at rated wind farm output) with
only a small change in voltage at the wind farm (i.e, 0.08 %) as shown in
Figures 8.10 and 8.11.
Fig 8.10: Voltage at 66 kV busbar with varying wind farm output.
Fig 8.11: Variation of network losses with wind farm output.

The effect of applying a +/- 4 MVAr ASVC at the wind farm was to increase
the stability limit to 173 % and to decrease the losses by 72 kW at rated wind
farm output as shown in Figure 8.10 and 8.11 respectively. The 66 kV connection
arrangement appears to be satisfactory without the ASVC and the reduction in
the network losses was small, it is likely to be very difficult to justify the addi-
tional cost of the compensator.
8.3.3 20 kV Connection with Intermediate Load

The 20 kV circuit is very resistive and so the losses are high. The only effective
way to reduce losses in this circuit is to connect a load to it. This is likely to be
desirable as 20 kV is used as a distribution voltage in the network, to which
the wind farm is connected, and consumers are supplied from fixed tap 20/0.4
kV transformers. In addition, the 20 kV overhead line is so long that it is likely
that loads would be connected along it. A 5 MW, 0.98 power factor load block,
representing peak loads, was connected at the mid-point of the circuit, as
shown in Figure 8.7(c). Summer loads were assumed to be 10 % of winter
loads.
The ASVC required to reduce losses and control the voltage of the 20 kV load
busbar to within 0.98-1.02 pu, especially during summer loads, was rated at +/-
2 MVAr. The control strategy of the ASVC is to keep the voltage at the load
busbar within the specified limits (i.e. +/- 2 %) by holding the voltage at the
wind farm busbar constant at 1.02 pu. This latter corresponds to the 20 kV
busbar voltage at rated output and maximum intermediate load when no
ASVC is connected. This control strategy was shown to be satisfactory in con-
trolling the load busbar voltage.
8.4 Study Case 3
Study Case 3 was that of a small wind farm connected to a rural 20 kV net-
work. The object of this study was to investigate possible control strategies for
the ASVC. Several of the voltage control techniques used on ASVCs applied to
transmission systems were investigated and modelled.
A simplified equivalent circuit representation of the network is shown in

Figure 8.12. The wind farm consisted of three 600 kW fixed speed wind turbine
generators. The ASVC is connected at the high voltage side of the wind farm
transformer and its role in controlling the reactive power requirements of the
wind farm for various loads was investigated. The Matlab Power System
Toolbox (PST) [18] was used for these studies.
8.4.1 The ASVC Characteristic

The ASVC characteristic, shown in Figure 8.13, was determined by the chosen
voltage reference value and the slope of the control characteristic. The maxi-
mum and minimum ASVC currents define the operational boundary. The ASVC
Fig 8.12: Simplified representation of the network

with the wind farm Study Case 3.
system characteristic
Fig 8.13: Output characteristic of the ASVC.

voltage for a particular operation condition is defined by the intersection of

the system and ASVC characteristics. For voltages around the reference volt-
age, the ASVC characteristic is similar if expressed either as a function of the
ASVC current or the reactive power capacity. The latter has been adopted in
this work.
ASVC model for load flow studies
An ASVC model [19] has been implemented in the Matlab-PST load flow pro-
gram. The ASVC busbar was represented as a special PQ node where the reac-
tive power is a function of VS, Q max , Q min (from the ASVC characteristic shown
in Figure 8.13) as follows:
(1)
Qs = Vs [ QV
max
min
(Vs - Vmax) +
Qmin
(V V)
Vmax min - s ]
In the load flow program, the terms of the Jacobean matrix, which correspond
to the ASVC busbar, were modified by adding the derivative of equation (1).
ASVC model for transient studies
A schematic diagram of the ASVC voltage control is illustrated in Figure 8.14.

This is a simplified version of the control circuit described in reference [10].
For voltage control, the slope corrected actual system voltage signal is com-
pared with the voltage reference value. The resulting voltage error signal
V (V = Vref - Vact) is fed to the compensator block, whose output signal repre-
sents the required ASVC susceptance Bref to correct and minimise the voltage
error V. The compensator block is represented by a lag function of the form
(k/1+sT) to improve the ASVC response by responding to sudden changes in the
network by smoothly changing to the appropriate voltage corresponding to the
slope control characteistic. A Proportional-Integral controller may also be
used, as suggested in reference [10]. This will allow a fast response to small
voltage variations (Proportional component) and a zero voltage error signal
control (Integral component). A further refinement is to use a wash-out filter
to compensate voltage and power oscillations.
A superimposed slow reactive power control mode was added to the voltage
control, once the ASVC currents or reactive power exceeded the minimum and
maximum limits (Imin and Imax or Qmin and Qmax). The difference between the
ASVCs actual reactive power value and the reference value (Qmin and Qmax) is
fed to a VAr regulator whose output, which is a voltage correction value, is
added to the voltage reference (Vref). This causes a shifting, along the ASVCs
voltage axis, of the voltage reference. Limits were imposed on the voltage cor-
rection value from the VAr regulator to avoid large deviations from the original
selected reference voltage.
Fig 8.14: Schematic representation of an ASVCs voltage control strategy.
The original Matlab Power System Toolbox (PST) code for a conventional SVC
device was based on the approach of correcting the voltage by varying the SVC
susceptance. Therefore, for computational convenience, the ASVC voltage con-
trol was also based on variations of the ASVC susceptance. However, Bmax and
Bmin were not fixed values, but calculated at each time step as a function of the
calculated voltage and maximum and minimum ASVC reactive power as fol-
lows:
Qmax Qmin (2)

Bmax = , Bmin =
V2 V2
8.4.2 Application of the ASVC

ASVC models were implemented in the load flow and transient Matlab PST
programs. Several load flow simulations were performed to analyse the steady
state response of the ASVC following changes in the reactive power demand of
the wind farm. The results, checked against the system characteristic, indicate
the adequacy of the ASVC model which was implemented.
The ASVC was also used to control the reactive power demand of the wind
farm dynamically for various levels of generation. Two control strategies for
the ASVC were implemented. The first used a voltage control scheme to keep
the voltage at the ASVC busbar within the specified limits (-/+ 5 %) and along
the slope characteristic. The second was to inject reactive power so that the
reactive power flow between busbars 1 and 2 was zero (see Figure 9.12).
Although a number of studies have been carried out, only the results of one
simulation are presented. Figure 8.15 shows changes in ASVC's voltage and
reactive power following a change from a full wind farm generation to 50 %
generation at time t = 1.3 sec (i.e., Q2 = 0). The change in the level of generation
causes a drop in the ASVCs voltage (from 1.0140 to 1.0120 pu) and reactive
power (from 378 to 320 kVAr). The simulations were carried out with a value
of gain k and time constant T chosen to be 50 and 0.5 respectively. The slow oscil-
lations in the voltage and reactive power waveforms are due to the dynamics
of the wind turbine induction generator as reflected in Figure 15(c) which shows
the wind farms generated power. The small stair-case variations in the wave-
forms are due to quantisation errors.
1.0158
1.0134
ASVC's voltage (pu)
1.0111
1.0088
1.0065
1.0041
1.0018
1.2 1.3 1.4 1.6 1.7 1.8
Time (sec)
Fig 8.15(a): Variation of ASVCs voltage with changes

in the wind farm generation.
-0.06
-0.12
ASVC's reactive power (MVAr)
-0.18
-0.24
-0.30
-0.36
-0.42
1.2 1.3 1.4 1.6 1.7 1.8
Time (sec)
Fig 8.15(b): Variation of ASVCs reactive power with changes

in the wind farm generation.
-0.04
-0.34
Wind farm generated power (MW)
-0.64
-0.94
-1.24
-1.54
-1.84
1.2 1.3 1.4 1.6 1.7 1.8
Time (sec)
Fig 8.15(c): Variation of wind farms generated power.

8.5 Conclusions
The application of Advanced Static VAr Compensators to different wind farms

was investigated. Several strategies for steady state voltage and reactive power
control of wind farms equipped with ASVCs were considered.
The unity power flow based strategy for controlling reactive power flows of a
wind farm is likely to lead to close to minimal losses in real distribution net-
works. However, the unity power factor based control strategy may lead to an
inefficient use of compensators in regulating the voltage in distribution sys-
tems in extreme circumstances, when the system is heavily or/and lightly
loaded. The unity power factor based control may limit the active power output
of embedded generators that can be accommodated, and therefore adversely
influence the level of penetration of embedded generation particularly that
connected to weak distribution systems.
The use of a unity power factor control technique and a voltage control scheme
for the ASVC was also examined using an electromagnetic simulation. The
ASVC was able to supply the reactive power requirements of the wind farm
under various operating conditions, to control the network voltage actively and
hence increase the permitted wind farm capacity and to improve the steady
state stability limit of the network. Its application to prevent damaging over
voltages which may occur under islanding conditions was demonstrated. The
ASVC was also shown to mitigate voltage fluctuations at blade passing fre-
quency successfully.
The operation of the ASVC with a unity power factor control scheme results in
a reduction in total network losses but no increase in the penetration of
embedded generation. The operation of the ASVC with a voltage control tech-
nique, operating on a conventional slope characteristic, would improve the
steady state voltage stability of the network and increase the capacity of
embedded generation that could be connected. This control technique can also
prevent large overvoltages due to self-excitation at islanding if a controller
with a fast enough speed of response is used. The control of transient voltage
fluctuations, which may be caused by synchronised wind turbines, is also a
natural consequence of a rapid voltage control scheme or may be implemented
by a supplementary control signal of the voltage variations only.
The costs of connecting wind farms to distribution systems are mainly driven
by the voltage level at which the connection is made and it is of considerable
interest to developers and operators of wind farms to connect the generation at
the lowest possible voltage level. On the other hand, in order to limit the effects
of connecting wind farms on the rest of the system, high voltage connections
are usually advocated by the utilities. The studies carried out on the wind farm
of study case 2 show that a high voltage connection (66 kV) is likely to be the
most suitable choice although a 20 kV connection is possible if the high losses
can be accepted. On the basis of the steady state performance of the system the
only justification of an ASVC would appear to be if customer loads are con-

nected to the 20 kV connecting circuit and then an ASVC would have a useful
role in controlling the voltage of the high voltage side of the fixed tap 20/0.4
kV transformers.
The control strategy chosen for an ASVC depends on the system requirements.
The use of the slope characteristic allows both voltage and reactive power con-
trol strategies to be performed simultaneously. The voltage control scheme is
usually fast as compared to the slow acting reactive power control mode. Using
a voltage control scheme, the voltage is controlled to within the specified limits
by varying the injected or absorbed ASVCs reactive power. The reactive power
control can be activated additional to the voltage control mode, to cater for
deviations in reactive power between a desired and actual value by changing
the reference voltage value. In this way, a unity power factor control scheme
may be implemented.
References
1. P Gardner, "Technical and Commercial Aspects of the Connection of Wind

Turbines to Electricity Supply Networks in Europe", Proceedings of the
American Wind Energy Association Conference, March 1996.
2. K Bergmann et al., "Application of GTO-Based SVCs for Improved Use of

the Rejsby Hede Wind Farm, 9th National Power Systems Conference, 19-21
December, Kanpur, India, 1996.
3. Z Saad-Saoud and N Jenkins, "The Application of Advanced Static VAr

Compensators To Wind Farms", IEE Colloquium on Power Electronics for
Renewable Energy, London, June 1997.
4. J E Hill, "A Practical Example of the Use of Distribution Static

Compensator (D-STATCOM) to Reduce Voltage Fluctuations", IEE
Colloquium on Power Electronics for Renewable Energy, London, June
1997.
5. J B Ekanayake, N Jenkins and C B Cooper, "Experimental investigation of

an Advanced Static VAr Compensator", Generation, Transmission and
Distribution, Proceedings of the IEE, Vol 142, No. 2, pp. 202-210, March
1995.
6. J B Ekanayake and N Jenkins, "Mathematical Models of a Three Level

Advanced Static VAr Compensator", Generation, Transmission and
Distribution, Proceedings of the IEE, Vol 44, No. 2, pp. 201-206, March
1997.
7. J B Ekanayake and N Jenkins, "A Three Level Advanced Static VAr

Compensator", IEEE Transactions on Power Delivery, Vol. 11, No. 1, pp. 540-
545, January 1996.
8. J B Ekanayake and N Jenkins, "Performance of a Three Level Advanced

Static VAr Compensator", International Symposium on Electric Power
Engineering, pp. 136-141, Stockholm Power Tech, June 1995.
9. Manitoba HVDC Research Centre, "EMTDC: Electromagnetic Transients

Program Including DC Systems", 1994.
10. H Tyll and K Bergmann, "Use of Static VAR Compensators (SVCs) in High
Voltage Systems", SRBE-KBVE, Les Compensateurs Statiques de Puissance
Reactive, 22 March 1990, Brussels, Belgium.
11. J D Ainsworth, "Phase-Locked Oscillator Control System for Thyristor

Controlled Reactors", Proceedings of the IEE, Vol. 135, Pt. C, No. 2, pp. 146-
156, March 1988.
12. CIGRE WG38-01 Task Force No. 2, "Static VAR Compensators", 1986.
13. J D Ainsworth, "The Phase-Locked Oscillator - A New Control for

Controlled Static Converter", IEEE Transactions on Power Apparatus and
Systems, PAS-87, pp. 859-865, 1968.
14. Z Saad-Saoud and N Jenkins, "The Use of Advanced Static Var

Compensators to Improve the Power Quality of Wind Farms", Proceedings
of the 19th British Wind Energy Association Conference, BWEA 19, July
1997, Mechanical Engineering Publications.
15. T W Rasmussen, "An EMTDC Model of A Three Level Four MVAr

Compensator", Proceedings of 32nd Universities Power Engineering
Conference, UPEC 97, September 1997, UMIST, Manchester.
16. A Stampa and F Santjer, "Synchronism of Grid Connected Wind Energy

Converters in a Wind Farm", DEWI Magazine, No. 7, August 1995 (in
German).
17. A E Efthymiadis, A J B Heath and R D Youssef, "Interactive Power System

Analysis (IPSA): User Manual", Version 9.3, Release 1.4, Power System
Laboratory of UMIST, September 1997.
18. Cherry Tree Scientific Software, "Power System Toolbox (PST)", Canada,
1997.
19. R D Youssef, "Implicit Generator and SVC Modelling for Contingency

Scheduling of Reactive Power Dispatch", IEE Proceedings on Generation,
Transmission and Distribution, Vol 142, No. 5, September 1985.
163
Chapter 9
Conclusions
164 Conclusions
Nicholas Jenkins
UMIST
PO Box 88
Manchester
M60 1QD
ENGLAND
Conclusions 165
Conclusions
The project successfully installed and commissioned an 8 MVAr GTO based

Advanced Static VAr Compensator at the Rejsby Hede wind farm. This was the
first application in the world of an ASVC used to improve the power quality of
a large renewable energy generation plant.
The subsequent operating experience demonstrated that the ASVC was able
dynamically to maintain the power factor of the wind farm at close to unity
over a wide range of wind farm output power. The converter had a satisfactory
harmonic performance although considerable studies were required at the
design stage in order to achieve this.
A comprehensive monitoring system was designed, and constructed and

installed at the wind farm and a large data set obtained for future analysis.
Various system studies were carried out to investigate the use of ASVCs on
other wind farm sites, particularly those with a low short circuit level relative
to the wind farm capacity. It was shown that, in a number of situations, an
ASVC is likely to be of significant benefit in improving various aspects of
power quality.
In summary it may be concluded that the project was successful in developing

and applying novel power electronic technology to a renewable energy system
and that the research which has been completed will allow similar schemes to
be implemented in the future.
Kent H. Sbrink
I/S Eltra
Ralf Stber
Frank Schettler
Klaus Bergmann
SIEMENS
Nicholas Jenkins
Janaka Ekanayake (Visitor)
Zouhir Saad-Saoud,
Maria Luiza Lisboa,
Goran Strbac
UMIST
Jrgen Kaas Pedersen

EU
ISBN No.: 8790707052

Power Quality Improvements of Wind Farms: P Consumers

Uploaded by

Copyright:

Available Formats

Power Quality Improvements of Wind Farms: P Consumers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Power Quality Improvements of Wind Farms: P Consumers

Uploaded by

Copyright:

Available Formats

Power Quality Improvements

Fredericia, June 1998

Layout and text:

Repro and printing:

Siemens, Power Transmission and Distribution, Erlangen

Siemens, Germany, was represented by Mr Ralf Stber as project manager,

The Technical University of Denmark was represented by three assistant pro-

Kent H. Sbrink Jrgen Kaas Pedersen

Knud Ole Helgesen Pedersen

1.1 Introduction to the Book 17

Chapter 2: Rejsby Hede Wind Farm 21

Chapter 3: Reactive Compensation for Wind Farms 35

3.1 Introduction to Reactive Power Compensation 37

3.4 Advanced Static VAr Compensator (ASVC) 43

Chapter 4: Basic Design of the ASVC 53

Chapter 5: Detailed Design of the ASVC 73

5.1 Component Arrangement 75

Chapter 6 : Installation at Rejsby Hede 95

Chapter 7: Description of the Measuring Equipment 107

7.1 Introduction 111

Chapter 8: The Application of Advanced Static VAr

8.1 Introduction 137

Chapter 9: Conclusions 163

1.1 Introduction to the Book

1.2 Introduction to the Project

The operation of wind turbines with asynchronous generators requires reactive

The reactive power consumption of wind turbines consists of a constant no-

An advanced Static VAr Compensator (ASVC) uses a power electronic convert-

1.3 Introduction to the Project Partners

The partners in the project were:

1.3.1 ELSAM ELTRA

1.3.3 Technical University of Denmark (DTU)

1.3.4 University of Manchester Institute of Science and Technology (UMIST)

Rejsby Hede Wind Farm

The goal of the Danish government is to reduce CO2 emissions by 20 % by the

2.2 The Rejsby Hede Wind Farm

2.2.1 Description of the Wind Farm

No. of wind turbines 40

Figure 2.2: Wind farm power output in August 1996.

Month Energy production Peak power

January 5,210 23,621

Table 2.2: Energy production 1996.

0.8 Duration curve 1996

2.2.2 Description of the Turbines

Main shaft Radiator

Main bearing Gearbox Two-speed generator

Rigid body frame

Fail-safe hydraulic Flexible coupling

Figure 2.4: Components of the nacelle.

2.2.3 Description of the Local Network

Pmax 4.7 MW Pmax 3.5 MW

8.2 km 3.5 km 10.9 km

15/0.69 kV ASVC 15/0.69 kV

Figure 2.5: Single-line diagram of the 15 kV and 60 kV networks.

Figure 2.5 shows that the 60 kV substation in Bredebro is provided with a 23

2.2.4 Need for Reactive Compensation

In a inductor, the current lags the voltage 90 degrees. As with capacitors, no

Figure 2.6: AC voltage and current in RLC components.

energy and vice versa. By definition, reactive power in capacitors is negative